<b>Bijsluiter</b>. De hyperlink naar het originele document werkt niet meer. Daarom laat Woogle de tekst zien die in dat document stond. Deze tekst kan vreemde foutieve woorden of zinnen bevatten en de opmaak kan verdwenen of veranderd zijn. Dit komt door het zwartlakken van vertrouwelijke informatie of doordat de tekst niet digitaal beschikbaar was en dus ingescand en vervolgens via OCR weer ingelezen is. Voor het originele document, neem contact op met de Woo-contactpersoon van het bestuursorgaan.<br><br>====================================================================== Pagina 1 ======================================================================

<pre>             Health Council of the Netherlands
          Mobile phones and cancer
             Part 1: Epidemiology of tumours in the head
2013/11
</pre>

====================================================================== Einde pagina 1 =================================================================

<br><br>====================================================================== Pagina 2 ======================================================================

<pre>    Health Council of the Netherlands
Mobile phones and cancer
    Part 1: Epidemiology of tumours in the head
</pre>

====================================================================== Einde pagina 2 =================================================================

<br><br>====================================================================== Pagina 3 ======================================================================

<pre></pre>

====================================================================== Einde pagina 3 =================================================================

<br><br>====================================================================== Pagina 4 ======================================================================

<pre>Gezondheidsraad
Health Council of the Netherlands
Aan de staatssecretaris van Infrastructuur en Milieu
Onderwerp              : Aanbieding advies Mobile phones and cancer.
                          Part 1. Epidemiology of tumours in the head
Ons kenmerk : U-7758/EvR/pm/673-M4
Bijlagen               :1
Datum                  : 3 juni 2013
Geachte staatssecretaris,
Hierbij bied ik u het advies Mobile phones and cancer. Part 1. Epidemiology of tumours in
the head aan. Het advies is opgesteld door de Commissie Elektromagnetische velden en
getoetst door de Beraadsgroep Gezondheid en omgeving.
Blootstelling aan radiofrequente elektromagnetische velden afkomstig van mobiele tele-
foons en andere bronnen in de leefomgeving is vrijwel onvermijdelijk. Mobiele toepassin-
gen bieden vele voordelen voor het dagelijks leven, maar leiden soms ook tot zorgen.
Sommigen zijn bezorgd dat de continue blootstelling aan elektromagnetische velden leidt
tot gezondheidsproblemen. Een belangrijke vrees in dat verband is, dat veelvuldig en inten-
sief gebruik van een mobiele telefoon de kans op tumoren in het hoofd, met name kwaad-
aardige hersentumoren, kan vergroten. In juni 2011 heeft het International Agency for
Research on Cancer (IARC) van de Wereldgezondheidsorganisatie op basis van een
evaluatie van de beschikbare literatuur radiofrequente elektromagnetische velden geclassi-
ficeerd als ‘mogelijk kankerverwekkend bij mensen’. Deze classificatie is vooral gebaseerd
op epidemiologisch onderzoek. Al voordat het IARC met zijn project startte, was de com-
missie Elektromagnetische velden van de Gezondheidsraad begonnen met een systemati-
sche analyse van de epidemiologische literatuur over dit onderwerp. De conclusies van de
commissie wijken enigszins af van die van het IARC. De commissie is van oordeel dat het
epidemiologisch onderzoek geen duidelijk en consistent bewijs levert voor een verhoogde
kans op tumoren in de hersenen of andere delen van het hoofd in relatie tot maximaal 13
jaar gebruik van een mobiele telefoon. Een klein risico kan echter ook niet met zekerheid
worden uitgesloten. Nader onderzoek gedurende een langere periode kan hierover meer
duidelijkheid geven. Dergelijk onderzoek wordt momenteel uitgevoerd, maar het zal nog
Bezoekadres                                                           Postadres
Rijnstraat 50                                                         Postbus 16052
2515 XP Den               Haag                                        2500 BB Den     Haag
E - m a i l : E . v a n . R o n g e n @ g r. n l                      w w w. g r. n l
Te l e f o o n ( 0 7 0 ) 3 4 0 5 7 3 0
</pre>

====================================================================== Einde pagina 4 =================================================================

<br><br>====================================================================== Pagina 5 ======================================================================

<pre></pre>

====================================================================== Einde pagina 5 =================================================================

<br><br>====================================================================== Pagina 6 ======================================================================

<pre>Gezondheidsraad
Health Council of the Netherlands
Onderwerp             : Aanbieding advies Mobile phones and cancer.
                         Part 1. Epidemiology of tumours in the head
Ons kenmerk : U-7758/EvR/pm/673-M4
Pagina                :2
Datum                 : 3 juni 2013
een aantal jaren duren voordat de eerste resultaten worden gepubliceerd. De Gezondheids-
raad zal de wetenschappelijke ontwikkelingen blijven volgen en daar zonodig over rappor-
teren.
Dit advies is het eerste in een serie van drie. De commissie werkt nu aan een systematische
analyse van de dierexperimentele gegevens over de kankerverwekkendheid van radiofre-
quente elektromagnetische velden. Dat advies zal naar verwachting nog dit jaar worden
uitgebracht. In een derde advies zal de commissie de gegevens uit de eerste twee adviezen
integreren en bespreken in het licht van de recent gepubliceerde evaluatie van het IARC.
Dat advies wordt begin volgend jaar verwacht.
Met vriendelijke groet,
prof. dr. W.A. van Gool
voorzitter
Bezoekadres                                                          Postadres
Rijnstraat 50                                                        Postbus 16052
2515 XP Den              Haag                                        2500 BB Den     Haag
E - m a il : E . v a n . R o n @g r. n l                             w w w. g r. n l
Te l e f o o n ( 0 7 0 ) 3 4 0 5 7 3 0
</pre>

====================================================================== Einde pagina 6 =================================================================

<br><br>====================================================================== Pagina 7 ======================================================================

<pre></pre>

====================================================================== Einde pagina 7 =================================================================

<br><br>====================================================================== Pagina 8 ======================================================================

<pre>Gezondheidsraad
Health Council of the Netherlands
To the State Secretary for Infrastructure and the Environment
Subject                   : Advisory report Mobile phones and cancer.
                            Part 1: Epidemiology of tumours in the head
Our reference : U-7758/EvR/pm/673-M5
Enclosure(s) : 1
Date                      : June 3, 2013
Dear State Secretary,
I have the pleasure of presenting you the advisory report Mobile phones and cancer. Part 1:
Epidemiology of tumours in the head. It has been drafted by the Electromagnetic Fields
Committee and reviewed by the Standing Committee on Health and the Environment.
Exposure to the radiofrequency electromagnetic fields from mobile phones and other
sources in the environment is almost inevitable. Mobile applications have brought many
benefits to our daily life, but also concerns. Some people are worried that the continuous
exposure to the fields may result in adverse health effects. A main fear in this respect is that
the frequent and intensive use of mobile phones may increase the risk of tumours in the
head, in particular malignant brain tumours. In June 2011 the International Agency for
Research on Cancer (IARC) of the WHO concluded on the basis of a review of the available
literature that radiofrequency electromagnetic fields should be classified as “possibly
carcinogenic to humans”. This classification is primarily based on evidence from
epidemiological studies. Even before the IARC started its project, the EMF Committee of
the Health Council initiated a systematic analysis of the epidemiogical literature on this
subject. Its conclusions are slightly different from those of IARC. The Committee
concludes that there is no clear and consistent evidence from epidemiological studies for an
increased risk for tumours in the brain and other regions in the head in association with
mobile phone use up to approximately 13 years. However, a slightly increased risk can also
not be excluded. This means that further studies with a longer follow-up period will need to
provide more clarity. Several studies are ongoing, but it will take a number of years before
results will be published. The Health Council will continue to monitor the scientific
developments and will report on them when relevant.
P. O . B o x 1 6 0 5 2                                                  Visiting Address
NL-2500 BB The Hague                                                    Rijnstraat 50
The Netherlands                                                         NL-2515 XPThe Hague
Te l e p h o n e + 3 1 ( 7 0 ) 3 4 0 5 7 3 0                            The Netherlands
E - m a i l : E . v a n . R o n g e n @ g r. n l                        w w w. h e a l t h c o u n c i l . n l
</pre>

====================================================================== Einde pagina 8 =================================================================

<br><br>====================================================================== Pagina 9 ======================================================================

<pre></pre>

====================================================================== Einde pagina 9 =================================================================

<br><br>====================================================================== Pagina 10 ======================================================================

<pre>Gezondheidsraad
Health Council of the Netherlands
Subject                   : Advisory report Mobile phones and cancer.
                            Part 1: Epidemiology of tumours in the head
Our reference : U-7758/EvR/pm/673-M5
Page                      :2
Date                      : June 3, 2013
This report is the first of three. The Committee is now preparing a systematic analysis of the
animal studies on the carcinogenicity of radiofrequency electromagnetic fields. This report
is expected to be published this year. In a third report the Committee will discuss the
observations from the first two reports in the light of the recently published IARC
evaluation. That report is expected early next year.
Kind regards,
(signed)
Prof. W.A. van Gool
President
P. O . B o x 1 6 0 5 2                                                  Visiting Address
NL-2500 BB The Hague                                                    Rijnstraat 50
The Netherlands                                                         NL-2515 XPThe Hague
Te l e p h o n e + 3 1 ( 7 0 ) 3 4 0 5 7 3 0                            The Netherlands
E - m a i l : E . v a n . R o n g e n @ g r. n l                        w w w. h e a l t h c o u n c i l . n l
</pre>

====================================================================== Einde pagina 10 =================================================================

<br><br>====================================================================== Pagina 11 ======================================================================

<pre></pre>

====================================================================== Einde pagina 11 =================================================================

<br><br>====================================================================== Pagina 12 ======================================================================

<pre>Mobile phones and cancer
Part 1: Epidemiology of tumours in the head
to:
the State Secretary for Infrastructure and the Environment
the Minister of Economic Affairs
the Minister of Health, Welfare and Sport
No. 2013/11, The Hague, June 3, 2013
</pre>

====================================================================== Einde pagina 12 =================================================================

<br><br>====================================================================== Pagina 13 ======================================================================

<pre>The Health Council of the Netherlands, established in 1902, is an independent
scientific advisory body. Its remit is “to advise the government and Parliament on
the current level of knowledge with respect to public health issues and health
(services) research...” (Section 22, Health Act).
     The Health Council receives most requests for advice from the Ministers of
Health, Welfare & Sport, Infrastructure & the Environment, Social Affairs &
Employment, Economic Affairs, and Education, Culture & Science. The Council
can publish advisory reports on its own initiative. It usually does this in order to
ask attention for developments or trends that are thought to be relevant to
government policy.
     Most Health Council reports are prepared by multidisciplinary committees of
Dutch or, sometimes, foreign experts, appointed in a personal capacity. The
reports are available to the public.
                 The Health Council of the Netherlands is a member of the European
                 Science Advisory Network for Health (EuSANH), a network of science
                 advisory bodies in Europe.
                 The Health Council of the Netherlands is a member of the International Network
                 of Agencies for Health Technology Assessment (INAHTA), an international
                 collaboration of organisations engaged with health technology assessment.
 I NA HTA
This report can be downloaded from www.healthcouncil.nl.
Preferred citation:
Health Council of the Netherlands. Mobile phones and cancer.
Part 1: Epidemiology of tumours in the head. The Hague: Health Council of the
Netherlands, 2013; publication no. 2013/11.
all rights reserved
ISBN: 978-90-5549-960-1
</pre>

====================================================================== Einde pagina 13 =================================================================

<br><br>====================================================================== Pagina 14 ======================================================================

<pre>    Contents
    Samenvatting 15
    Executive summary 27
1   Introduction 37
1.1 Why this report? 37
1.2 The research question 38
1.3 This report 38
2   Literature search 41
2.1 Method 41
2.2 Results 41
3   Methods of data analysis 43
3.1 Data extraction 43
3.2 Evaluation of the quality of studies 44
4   Study design and methods 47
4.1 Bias 47
4.2 Cohort studies 48
4.3 Case-control studies 50
4.4 Case-case studies 57
    Contents                                13
</pre>

====================================================================== Einde pagina 14 =================================================================

<br><br>====================================================================== Pagina 15 ======================================================================

<pre>4.5  Ecological studies 58
5    Evaluation of study quality 63
5.1  Results of the evaluation of study quality 64
5.2  Conclusion 71
6    Results: analysis of the data by disease 73
6.1  Issues to be considered 73
6.2  Brain tumours (not further specified) 75
6.3  Glioma 76
6.4  Meningioma 86
6.5  Acoustic neuroma 89
6.6  Parotid gland tumours 94
6.7  Pituitary tumours 99
6.8  Malignant melanoma of the eye 99
6.9  Intra-temporal facial nerve tumours 99
6.10 Neuroblastoma 99
7    Discussion 101
7.1  The research questions 101
7.2  Strengths and limitations of this analysis 101
7.3  Mobile vs. cordless phones 102
7.4  Strengths and limitations of the different study types and studies 104
7.5  Overall discussion per tumour type 108
7.6  The Bradford Hill considerations 117
8    Conclusions and recommendations 121
     References 125
     Annexes 139
A    The Committee 141
B    Search strategy and results 145
C    Data extraction 149
D    Evaluation of quality of the studies 151
E    Additional information for the publications used 155
F    Results of the data extraction 163
G    Results of the evaluation of quality of the studies 223
H    Results from the selected publications 225
I    Meta-analysis and forest plots 245
14   Mobile phones and cancer
</pre>

====================================================================== Einde pagina 15 =================================================================

<br><br>====================================================================== Pagina 16 ======================================================================

<pre>Samenvatting
Doel van het advies
Mobiele telefonie is gemeengoed geworden. Bijna iedereen in westerse landen
heeft een mobiele telefoon. Maar met de toename van het gebruik van mobiele
telefoons onstonden ook zorgen over mogelijke gezondheidseffecten van bloot-
stelling aan de radiofrequente elektromagnetische velden die deze apparaten uit-
zenden. Die zorgen zijn vooral gericht op een mogelijke relatie met
hersentumoren.
    In dit advies onderzoekt de Commissie Elektromagnetische Velden van de
Gezondheidsraad op basis van epidemiologische gegevens of er aanwijzingen
zijn voor een oorzakelijk verband tussen blootstelling aan radiofrequente velden
van mobiele telefoons en het optreden van tumoren in de hersenen en diverse
andere weefsels in het hoofd (zoals hersenvliezen, gehoorzenuw en speekselklie-
ren).
    De commissie heeft daartoe op een systematische wijze volgens een vooraf
vastgesteld protocol de relevante epidemiologische literatuur in kaart gebracht en
geëvalueerd.
    In een gerelateerd advies zal de commissie zich buigen over de resultaten van
dierexperimenteel onderzoek.
Samenvatting                                                                       15
</pre>

====================================================================== Einde pagina 16 =================================================================

<br><br>====================================================================== Pagina 17 ======================================================================

<pre>   Relevante typen onderzoek
   Alle relevante typen epidemiologisch onderzoek zijn gebruikt, voor zover
   beschikbaar: cohort-onderzoek, patiënt-controle-onderzoek, patiënt-patiënt-
   onderzoek en ecologisch onderzoek.
        Het enige relevante cohortonderzoek is een uitgebreid retrospectief onder-
   zoek uit Denemarken. Op basis van de gegevens van de mobiele telefonieaanbie-
   ders is daarbij bepaald of de deelnemers al voor 1996 een privé-abonnement
   hadden.
        Wat betreft de patiënt-controle-onderzoeken richt de analyse zich voor-
   nameljik op twee groepen onderzoeken. De eerste groep is een serie onderzoeken
   uitgevoerd door 16 onderzoeksgroepen in 13 landen, het INTERPHONE-consor-
   tium. Alle hebben ze gebruik gemaakt van een basisonderzoeksopzet die is ont-
   wikkeld in samenwerking met het International Agency for Research on Cancer
   (IARC) en die zich richt op verschillende typen tumoren in het hoofd-halsgebied,
   inclusief de hersenen.
        De tweede groep bevat publicaties over verschillende nauw met elkaar
   samenhangende onderzoeken van de onderzoeksgroep van Hardell uit Zweden.
   De leeftijdscategorieën zijn hier breder dan die in de INTERPHONE-onderzoe-
   ken. Omdat dit van invloed kan zijn op de uitkomsten (het vóórkomen van de
   meeste onderzochte tumoren is namelijk leeftijdsafhankelijk), heeft de commis-
   sie in haar analyse van deze gegevens zoveel mogelijk dezelfde leeftijdscatego-
   rieën gebruikt als in de INTERPHONE-onderzoeken.
        In verschillende landen zijn ecologische onderzoeken uitgevoerd naar de
   relatie tussen het vóórkomen van hersentumoren en de toename van het gebruik
   van mobiele telefoons. Gezien de lange latentietijd van hersentumoren, die waar-
   schijnlijk meer dan tien jaar duurt, is het mogelijk dat trends in het vóórkomen
   van tumoren die verband houden met het gebruik van mobiele telefoons nog niet
   te zien zijn.
   Methodologische kwaliteit
   De commissie heeft een scoringssysteem ontwikkeld om de methodologische
   kwaliteit van de geselecteerde publicaties te beoordelen. Toepassing hiervan
   brengt geen fundamentele verschillen aan het licht tussen de belangrijkste onder-
   zoeken: het Deense cohortonderzoek en de patiënt-controle-onderzoeken van
16 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 17 =================================================================

<br><br>====================================================================== Pagina 18 ======================================================================

<pre>Hardell en INTERPHONE. Op basis van de scoring is er geen reden om een van
deze onderzoeken meer gewicht te geven dan de andere.
Resultaten per type tumor
In de onderzoeken zijn verschillende blootstellingskenmerken gebruikt. De com-
missie heeft zich in haar evaluatie gericht op (a) het aantal jaren dat een mobiele
telefoon is gebruikt; (b) de totale blootstelling door het voeren van gesprekken
met de mobiele telefoon, bepaald aan de hand van het totale aantal gespreksuren
en (c) de zogenoemde lateralisatie; hierbij wordt gekeken of de telefoon voorna-
melijk gebruikt is aan de kant van het hoofd waar zich de tumor bevindt (ipsilate-
raal) of aan de andere kant (contralateraal).
Gliomen
Gliomen zijn kwaadaardige tumoren van het hersenweefsel. In het Deense cohort
werd geen verhoogde kans op het krijgen van een glioom gevonden bij een abon-
nement op mobiele telefonie gedurende maximaal 13 jaar. In de patiënt-controle-
onderzoeken is een vergelijkbaar criterium gebruikt: de tijd sinds het eerste
gebruik van een mobiele telefoon. In het INTERPHONE-onderzoek is daarbij
geen verhoogd risico gevonden. Dat was wel het geval bij de overeenkomstige
leeftijdsgroep in het onderzoek van Hardell.
     In de lateralisatie-analyse van de gegevens over de tijd sinds het eerste
gebruik vond Hardell voor zowel ipsilateraal als contralateraal gebruik een ver-
hoogd risico. In het INTERPHONE-onderzoek werd zowel voor ipsi- als contra-
lateraal gebruik geen verhoogd risico gevonden. De gegevens voor dit criterium
zijn dus niet consistent.
     Voor de totale beltijd werd in de onderzoeken van Hardell en INTERPHONE
in de hoogste blootstellingscategorie (1640 uur en meer) een verhoogd risico
gevonden, waarbij het risico in het onderzoek van Hardell hoger was dan in dat
van INTERPHONE. In het onderzoek van INTERPHONE werden in diverse
lagere categorieën, waaronder ook de op een na hoogste, juist verlaagde risico’s
gevonden. Er is dus geen duidelijke blootstellings-respons relatie.
     In de lateralisatie-analyse van de gegevens van de totale beltijd vond Hardell
een verhoogd risico voor zowel ipsilateraal als contralateraal gebruik. In het
INTERPHONE-onderzoek daarentegen werd alleen een verhoogd risico gevon-
den voor ipsilateraal gebruik in de hoogste van vijf categorieën (1640 uur en
meer), terwijl bij de laagste twee categorieën voor contralateraal gebruik de
risico’s juist verlaagd bleken te zijn.
Samenvatting                                                                        17
</pre>

====================================================================== Einde pagina 18 =================================================================

<br><br>====================================================================== Pagina 19 ======================================================================

<pre>        In de ecologische onderzoeken werd in de gegevens uit Scandinavische lan-
   den en Groot Brittannië geen toename gevonden in het vóórkomen van gliomen.
   In de Verenigde Staten werd een toename van 0,75% per jaar gevonden voor gli-
   omen in de temporaalkwab (het gedeelte van de hersenen dat zich het dichtst bij
   de mobiele telefoon bevindt tijdens het voeren van een gesprek). De Ameri-
   kaanse gegevens komen niet overeen met de relatieve risico’s uit de onderzoeken
   van Hardell. Als die relatieve risico’s echt zouden zijn, zou de toename van het
   aantal gliomen in de VS groter moeten zijn en zou die ook in andere landen
   zichtbaar moeten zijn. De Amerikaanse gegevens zijn niet strijdig met een kleine
   verhoging van het risico zoals dat in het INTERPHONE-onderzoek voor totale
   beltijd is gevonden, maar ze kunnen ook verklaard worden zonder de aanname
   van een verhoogd risico. De gegevens over het vóórkomen van gliomen in
   Nederland laten geen toename zien na de periode van snelle toename van het
   gebruik van mobiele telefoons in de leeftijdsgroepen die deze het meest gebrui-
   ken: die van 20 tot 29 en die van 30 tot 59 jaar.
   Meningiomen
   Meningiomen zijn tumoren van de hersenvliezen, die de scheiding vormen tus-
   sen het zenuwweefsel van de hersenen en de schedel. In het Deense cohortonder-
   zoek werd geen verhoogd risico voor meningiomen waargenomen. In de
   onderzoeken van Hardell werd een verhoogd risico gevonden in de hoogste cate-
   gorie voor verstreken tijd sinds het eerste gebruik (meer dan 10 jaar), maar alleen
   voor analoge en niet voor digitale mobiele telefoons. In het INTERPHONE-
   onderzoek werden in de twee middelste van vier categorieën juist verlaagde
   risico’s gevonden. Geen van de andere blootstellingsmaten was gecorreleerd met
   een risico voor meningiomen.
   Akoestische neuromas of brughoektumoren
   Brughoektumoren zijn tumoren van de gehoorzenuw. In het Deense cohortonder-
   zoek werd geen verhoogd risico voor brughoektumoren gevonden in relatie tot
   een mobiele telefonie-abonnement gedurende 11 jaar of langer. Hardell vond een
   verhoogd risico in associatie met het gebruik van analoge telefoons voor alle
   gebruiksduren, zelfs al bij een duur van één tot vijf jaar. Voor digitale telefoons
   werd alleen over kortere periodes een verhoogd risico gevonden, maar niet bij
   gebruik gedurende 10 jaar of langer.
        De lateraliteitsanalyse van de gegevens van Hardell gaf verhoogde risico’s te
   zien voor ipsilateraal gebruik van analoge en digitale mobiele telefoons bij een
18 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 19 =================================================================

<br><br>====================================================================== Pagina 20 ======================================================================

<pre>gebruiksduur van zowel meer dan een jaar als meer dan tien jaar. In het algemeen
werd geen verhoogd risico gevonden. Dat was ook niet het geval in de laterali-
teitsanalyse van het INTERPHONE-onderzoek en in een Japans patiënt-patiënt-
onderzoek.
     In de onderzoeken van Hardell werden verhoogde risico’s gevonden voor alle
typen telefoons bij een totale beltijd van meer dan 1000 uur. In het INTER-
PHONE-onderzoek werden daarentegen geen verhoogde risico’s gevonden bij
een totale beltijd van 1640 uur of meer, terwijl in verschillende lagere cate-
gorieën juist verlaagde risico’s werden gevonden. Voor ipsilateraal gebruik was
het risico verhoogd in de hoogste categorie (totale beltijd van 1640 uur of meer),
en verlaagd in de op een na hoogste categorie.
Parotiskliertumoren
Parotisklieren zijn de speekselklieren die het meest zijn blootgesteld bij het bel-
len met een mobiele telefoon. In de onderzoeken van Hardell werden geen ver-
hoogde risico’s voor tumoren in de parotisklier gevonden voor de tijd sinds het
eerste gebruik en voor de totale beltijd. Evenmin was dit het geval in de
INTERPHONE-onderzoeken. Het enige verhoogde risico werd gevonden in een
subgroep van een van de onderzoeken die volgens het INTERPHONE-protocol
zijn uitgevoerd, en wel in de groep patiënten met zowel goedaardige als kwaad-
aardige tumoren die de telefoon ipsilateraal gebruikten en die een totale beltijd
hadden van meer dan 266 uur.
     In slechts één ecologisch onderzoek werd over parotiskliertumoren gerappor-
teerd, waarbij het vóórkomen vrijwel constant was. Gegevens over het vóórko-
men van parotiskliertumoren in Nederland laten geen veranderingen zien over de
periode 1989-2010.
Overwegingen voor de evaluatie
Latentietijd
Bij onderzoek naar langzaam groeiende tumoren is het van belang rekening te
houden met de latentietijd, dat wil zeggen de tijd tussen het ontstaan van de
tumor en het moment dat deze klinisch aantoonbaar wordt. Er is echter nauwe-
lijks enige informatie beschikbaar over latentietijden voor de typen tumoren die
in dit advies worden besproken. De commissie acht het mogelijk dat een periode
van tien jaar te kort is om een toename in het vóórkomen van deze tumoren te
kunnen meten.
Samenvatting                                                                        19
</pre>

====================================================================== Einde pagina 20 =================================================================

<br><br>====================================================================== Pagina 21 ======================================================================

<pre>   Analoge versus digitale telefoons
   De eerste mobiele telefoons maakten gebruik van een analoog signaaltype, ter-
   wijl de latere GSMs een digitaal signaal gebruikten. Dit betekent dat de in de epi-
   demiologische onderzoeken opgenomen personen die het langst gebruik maakten
   van mobiele telefonie (tien jaar of langer) aanvankelijk belden met een analoge
   telefoon. De blootstelling bij gebruik van een analoge telefoon was hoger dan die
   bij gebruik van een digitale telefoon.
   Draadloze versus mobiele telefoons
   Draadloze telefoons zijn mobiele telefoons met een beperkt bereik voor gebruik
   binnenshuis, zoals DECT-telefoons. In verschillende onderzoeken heeft Hardell
   het gebruik van draadloze telefoons meegenomen, onder de aanname dat de
   blootstelling aan radiofrequente velden daarbij van vergelijkbare grootte is als bij
   mobiele telefoons. De blootstelling bij het gebruik van een draadloze telefoon is
   echter lager dan bij gebruik van een mobiele telefoon.
        Dit betekent dat het op grond van de daadwerkelijke blootstelling moeilijk te
   verklaren is dat Hardell ruwweg vergelijkbaar verhoogde risico’s vond bij
   gebruik van mobiele en draadloze telefoons. Het is niet bekend of het gebruik
   van beide typen telefoons gecorreleerd is, maar de commissie acht dit wel moge-
   lijk. Dat zou dan deels een verklaring kunnen zijn voor de verhoogde risico’s die
   zijn gevonden bij het gebruik van draadloze telefoons. De commissie is desal-
   niettemin van mening dat de gegevens over de draadloze telefoons vragen oproe-
   pen over de interne consistentie van de onderzoeken van Hardell.
   Sterke en zwakke punten in de onderzoeken
   Cohortonderzoeken
   Cohortonderzoeken leveren potentieel sterk bewijs, omdat de blootstelling her-
   haaldelijk en objectief gemeten of bepaald kan worden voordat de ziekte
   optreedt. Dergelijke onderzoeken hebben daarom geen last van vertekening op
   grond van foutieve herinnering. Wel kunnen er andere problemen zijn.
        In het Deense cohortonderzoek hebben de onderzoekers alleen gekeken of de
   deelnemers een privé-abonnement hadden dat was gestart voor 1996. Deze groep
   hebben zij vervolgens vergeleken met alle inwoners van Denemarken. Het is dui-
   delijk dat de tijd die is verstreken sinds het aangaan van een abonnement een
20 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 21 =================================================================

<br><br>====================================================================== Pagina 22 ======================================================================

<pre>minder relevante parameter is dan een schatting van de mate van daadwerkelijk
gebruik, die directer gerelateerd is aan de blootstelling.
     In de latere publicaties over dit onderzoek, die een langere periode bestrijken,
zal daarnaast in toenemende mate misclassificatie zijn opgetreden in de groep
niet-gebruikers. Daar zijn twee redenen voor: zakelijke gebruikers, die mogelijk
tot de meest intensieve gebruikers behoren, zijn niet opgenomen in de gebrui-
kersgroep, en het bezit van mobiele telefoons in de Deense bevolking is na 1996
sterk toegenomen. Het Deense cohortonderzoek is om deze redenen wel afge-
schilderd als een onderzoek van beperkte waarde.
     Ondanks het ontbreken van gegevens over de blootstelling beschouwt de
commissie het Deense cohortonderzoek echter als belangrijk voor de evaluatie.
Misclassificatie in de niet-gebruikersgroep heeft namelijk slechts een zeer
beperkt effect op het berekende risico, en van misclassificatie in de gebruikers-
groep is geen sprake.
Patiënt-controle-onderzoeken
De commissie beschouwt de INTERPHONE-onderzoeken als vatbaar voor ver-
tekening door selectie, vanwege de relatief lage deelnamepercentages. Omdat
deze bij de controles ook nog eens lager zijn dan bij de patiënten, kan er differen-
tiële misclassificatie optreden (dat wil zeggen dat de misclassificatie verschillend
is voor patiënten en controles). Dit versterkt vertekening door selectie.
     Deze vertekening is mogelijk de oorzaak van de verlaagde risico’s die in
sommige van de lagere blootstellingscategorieën zijn waargenomen; een
beschermend effect van mobiel bellen is namelijk niet waarschijnlijk. Maar dit
zou betekenen dat de verhoogde risico’s in de hoogste blootstellingscategorieën
ook te laag kunnen zijn als gevolg van vertekening door selectie. Anderzijds kun-
nen de risico’s door vertekening door selectieve herinnering juist weer hoger uit-
vallen. Het is niet mogelijk om de omvang van deze vertekeningen in te schatten.
     In de onderzoeken van Hardell worden hogere deelnamepercentages en klei-
nere verschillen tussen de deelname van patiënten en controles gemeld dan in de
INTERPHONE-onderzoeken. Deze onderzoeken hebben daarom waarschijnlijk
minder last van vertekening door selectie dan de INTERPHONE-onderzoeken.
De deelnamepercentages van met name de controles in de onderzoeken van Har-
dell zijn echter ongewoon hoog.
     Een ander punt bij de onderzoeken van Hardell is dat al na korte tijd sinds het
eerste gebruik verhoogde risico’s werden gevonden. Dat is onwaarschijnlijk in
het licht van de naar verwachting zeer lange latentietijden van de onderzochte
typen tumoren. Daarnaast zou, als deze verhoogde risico’s echt zouden zijn, in de
Samenvatting                                                                          21
</pre>

====================================================================== Einde pagina 22 =================================================================

<br><br>====================================================================== Pagina 23 ======================================================================

<pre>   ecologische onderzoeken gevonden moeten zijn dat het vóórkomen van deze
   tumoren toeneemt. Dit is echter niet het geval.
       Bij zowel de onderzoeken van Hardell als die van INTERPHONE kan er ook
   waarnemersvertekening optreden. Ondanks de training van degenen die de inter-
   views hebben afgenomen kunnen zij patiënten en controles ongemerkt net iets
   anders benaderen, al is niet bekend welke invloed dit kan hebben. Beide onder-
   zoeken zijn ook vatbaar voor vertekening door herinnering, omdat het vaststellen
   van de blootstelling in patiënt-controle-onderzoeken altijd achteraf gebeurt. Het
   is aangetoond dat vertekening door herinnering verschillend werkt voor patiën-
   ten en controles, en dat dit tot een overschatting van het risico kan leiden.
       Een ander punt dat in aanmerking moet worden genomen is dat de onderzoe-
   ken van Hardell in slechts één land zijn uitgevoerd (Zweden), terwijl de onder-
   zoeken van INTERPHONE 16 gebieden in 13 landen bestrijken, en dus ook een
   veel bredere populatie omvatten. Ook de totale aantallen patiënten en controles
   zijn in de onderzoeken van Hardell lager dan die in de INTERPHONE-onderzoe-
   ken.
    De moeilijk te verklaren verhoogde risico’s samenhangend met het gebruik van
   draadloze telefoons en korte latentietijden die zijn gevonden in de onderzoeken
   van Hardell, in combinatie met de geringere omvang van deze onderzoeken in
   verhouding tot het INTERPHONE-onderzoek, hebben de commissie doen
   besluiten de onderzoeken van Hardell minder gewicht te geven in de uiteinde-
   lijke evaluatie en conclusies dan de INTERPHONE-onderzoeken.
   Patiënt-patiënt-onderzoeken
   Onderzoeken met twee groepen patiënten zijn potentieel sterk, omdat ze minder
   te kampen hebben met vertekening door selectie en waarneming. Vertekening
   door selectieve herinnering kan natuurlijk nog wel optreden, maar deze zal niet-
   differentieel zijn, omdat alleen patiënten in het onderzoek zijn opgenomen.
   Ecologische onderzoeken
   Ecologische onderzoeken zijn per definitie van beperkte waarde, omdat de indi-
   viduele blootstelling niet wordt bepaald. Er kunnen hieruit dan ook geen bloot-
   stelling-effectrelaties worden vastgesteld. Hooguit kunnen ecologische
   onderzoeken laten zien dat er overeenkomsten zijn in trends van de toename van
   een ziekte en het gebruik van mobiele telefoons.
22 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 23 =================================================================

<br><br>====================================================================== Pagina 24 ======================================================================

<pre>     Als er na een toename van het bezit (en verondersteld gebruik) van mobiele
telefoons geen toename gevonden wordt in het voorkomen van een ziekte, is dat
echter nog geen bewijs dat er geen oorzakelijk verband is tussen blootstelling en
ziekte. Alleen als de latentietijd tien jaar of minder zou zijn, zou een verhoogd
risico in de trends nu al zichtbaar moeten zijn.
Conclusies over specifieke tumoren
Gliomen
De commissie concludeert dat er enkele zwakke en inconsistente aanwijzigingen
zijn voor een associatie tussen langdurig intensief gebruik van een mobiele tele-
foon en het vaker voorkomen voor gliomen. Verschillende vormen van verteke-
ning en toeval zouden een verklaring kunnen zijn voor deze uitkomsten, maar het
kan niet worden uitgesloten dat er een oorzakelijk verband is. De commissie
schat de kans hierop echter in als zeer klein.
     In de bevolkingsstatistieken is, ook in Nederland, geen toename te zien in het
vóórkomen van gliomen. Een toename kan echter ook nog niet zichtbaar zijn
geworden vanwege de waarschijnlijk lange latentietijd bij deze tumoren.
     De ecologische onderzoeken geven ook geen ondersteuning voor een ver-
hoogd risico. Als de door de groep van Hardell gerapporteerde risico’s werkelijk
voorkomen, zou in de recente kankerstatistieken een toename van gliomen zicht-
baar moeten zijn en zou de latentietijd veel korter moeten zijn dan de mogelijk
meer dan tien jaar die nu wordt vermoed. Een risicotoename zoals gerapporteerd
in de INTERPHONE-onderzoeken, die lager is dan die bij Hardell, zou in de sta-
tistieken nog niet te zien zijn.
     Op grond hiervan concludeert de commissie dat er een klein risico op het ver-
hoogd voorkomen van gliomen kan zijn in samenhang met mobiel telefoonge-
bruik, maar dat het ook mogelijk is dat er geen risico is.
Meningiomen
De commissie concludeert dat er geen duidelijke en consistente aanwijzingen
zijn dat het gebruik van een mobiele telefoon gepaard gaat met een verhoogd
risico voor meningiomen.
Samenvatting                                                                        23
</pre>

====================================================================== Einde pagina 24 =================================================================

<br><br>====================================================================== Pagina 25 ======================================================================

<pre>   Brughoektumoren
   De commissie oordeelt dat de gegevens over een associatie tussen langdurig
   gebruik van een mobiele telefoon en het vóórkomen van brughoektumoren niet
   consistent zijn en geen duidelijke aanwijzingen geven voor een verhoogd risico.
   Parotiskliertumoren
   De commissie concludeert dat er geen duidelijke aanwijzingen zijn dat gebruik
   van een mobiele telefoon een verhoogd risico op parotiskliertumoren oplevert. Er
   is slechts in één subgroep in één onderzoek met een beperkt aantal patiënten een
   verhoogd risico waargenomen. Dit zou door toeval kunnen worden verklaard. De
   bevolkingsstatistieken laten, ook in Nederland, geen toename zien in het vóórko-
   men van parotiskliertumoren.
   Andere tumoren
   Er kunnen geen conclusies worden getrokken over risico’s die samenhangen met
   het gebruik van mobiele telefoons met betrekking tot tumoren van de hypofyse,
   melanomas van het oog, tumoren aan andere zenuwen dan de gehoorzenuw en
   neuroblastomas.
   Eindconclusie
   De huidige systematische analyse laat zien dat er, ondanks uitgebreid onderzoek,
   nog steeds geen duidelijkheid is over een mogelijk verband tussen het gebruik
   van een mobiele telefoon en een verhoogde kans op het optreden van tumoren in
   de hersenen en andere delen van het hoofd.
        Er zijn enkele zwakke en inconsistente aanwijzingen voor een verband tussen
   langdurig intensief gebruik van een mobiele telefoon en een toename van het
   vóórkomen van gliomen. Die aanwijzingen kunnen verklaard worden door ver-
   schillende vormen van vertekening en door toeval, maar het kan ook niet worden
   uitgesloten dat er een oorzakelijk verband is. De aanwijzingen voor een ver-
   hoogd risico voor andere tumoren, waaronder meningiomen en brughoektumo-
   ren, zijn veel zwakker of ontbreken geheel.
        Op basis van de epidemiologische gegevens die in dit advies zijn beschreven
   en in aanmerking nemend de kwaliteit en de sterke en zwakke punten van de ver-
   schillende onderzoeken luidt de eindconclusie van deze systematische analyse
   daarom als volgt: er is geen duidelijk en consistent bewijs voor een verhoogd
24 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 25 =================================================================

<br><br>====================================================================== Pagina 26 ======================================================================

<pre>risico voor tumoren in de hersenen of andere delen van het hoofd gerelateerd aan
gebruik van een mobiele telefoon gedurende 13 jaar of minder; een dergelijk
risico kan echter ook niet worden uitgesloten. Over langduriger gebruik kan niets
worden gezegd.
Samenvatting                                                                      25
</pre>

====================================================================== Einde pagina 26 =================================================================

<br><br>====================================================================== Pagina 27 ======================================================================

<pre>26 Mobile phones and cancer</pre>

====================================================================== Einde pagina 27 =================================================================

<br><br>====================================================================== Pagina 28 ======================================================================

<pre>Executive summary
Why this report?
Mobile telephony has become an ubiquitous commodity. In Western countries
virtually everybody has a mobile telephone. But with the increase in mobile
phone use, also concerns developed on possible adverse effects of exposure to
the radiofrequency electromagnetic fields emitted by these devices. Much of this
concern focussed on a possible relation with cancer in the brain.
    In this report, the Electromagnetic Fields Committee of the Health Council of
the Netherlands investigates on the basis of the epidemiological evidence
whether there are indications for a causal relationship between exposure to
radiofrequency fields from mobile phones and tumours in the brain and various
other tissues in the head (e.g. meninges, acoustic nerve, parotid glands).
    To this end, the Committee has systematically searched and reviewed the
relevant epidemiological literature following an a priori defined protocol.
    In a related report the Committee will evaluate the results of animal studies.
Relevant types of studies
All available relevant types of epidemiological studies have been used: cohort,
case-control, case-case and ecological studies.
Executive summary                                                                  27
</pre>

====================================================================== Einde pagina 28 =================================================================

<br><br>====================================================================== Pagina 29 ======================================================================

<pre>        The only relevant cohort study is a very large retrospective study from
   Denmark, in which mobile phone company records were used to determine
   whether a private mobile phone subscription was started before 1996.
        Two groups of case-control studies are primarily used in the analysis. The
   first group is a series of studies from 16 research groups in 13 countries, the
   INTERPHONE consortium. They all used a core protocol developed in
   collaboration with the International Agency for Research on Cancer (IARC) on
   different types of tumours in the head and neck area, including the brain.
        The second group contains publications of several closely related studies of
   the Hardell group from Sweden. The age range covered in these studies is wider
   than that in the INTERPHONE study. This may affect the results, since the
   incidence of most tumours investigated is age-related. Therefore, whenever
   possible, the Committee used data for the same age range as used in the
   INTERPHONE study in the analysis.
        Ecological studies that study the incidence of brain cancers in relation to the
   increase in use of mobile phones, have been performed in various countries. In
   view of the long latency period of brain tumours of likely more than 10 years, it
   is possible that any trends in tumour incidence related to mobile phone use may
   not yet be visible.
   Methodological quality
   The Committee developed a scoring system to evaluate the methodological
   quality of the selected publications. This assessment did not result in major
   differences between the main studies, i.e. the Danish cohort study and the
   INTERPHONE and Hardell case-control studies. On the basis of this scoring
   system there is no reason to give one type of study more weight than the other.
   Results per tumour type
   Several exposure characteristics have been used in the studies. In this analysis
   the Committee focussed on (a) the duration of mobile phone use in years; (b) the
   cumulative exposure from mobile phone calls in hours over the respondents’
   lifetime and (c) the so-called lateralisation. Lateralisation addresses if the
   telephone predominantly was used at the side of the head where the tumour is
   located (ipsilateral), or on the other side (contralateral).
28 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 29 =================================================================

<br><br>====================================================================== Pagina 30 ======================================================================

<pre>Gliomas
Gliomas are malignant tumours of the brain. In the Danish cohort no increased
glioma risks were observed for having a mobile phone subscription for up to 13
years. The case-control studies investigated a similar endpoint: time since first
use. INTERPHONE found no increased risks, but in the corresponding age-range
in the Hardell studies an increased relative risk was found.
    In the laterality analysis of the time since first use data, Hardell found an
increased risk for both ipsilateral and contralateral use, while INTERPHONE
found no increased risks. So the data on this endpoint are not consistent.
    For cumulative call time, both groups found an increased risk for the highest
exposure category (1640 hours and more). In the Hardell studies it was higher
than in the INTERPHONE study. However, INTERPHONE identified decreased
risks in several lower categories, including the next-highest one, so there is no
obvious exposure-response relation.
    In the laterality analysis of the cumulative call time data, Hardell found an
increased risk for both ipsilateral and contralateral use, while INTERPHONE
found an increased risk only for ipsilateral use in the highest of five categories
(1640 hours and more), and decreased risks for contralateral use in the lowest
categories.
    In the ecological studies, no increase in glioma incidence was observed in the
Nordic countries and the UK, while in the USA a small increase of
approximately 0.75% per year was observed of gliomas in the temporal lobe (the
part of the brain closest to a mobile telephone when a call is made). These US
data are not compatible with the relative risks of the Hardell studies. If these
relative risks were true, the increase of the glioma rate in the USA should have
been much larger and an increased rate should also have to be visible in other
countries. The US data are consistent with a small increase in risk as found for
cumulative call time in the INTERPHONE studies, but also with no change in
risk. Brain cancer incidence data for the Netherlands indicate no increase in
gliomas following the period of rapid increase in mobile phone use in the age
groups that use them most: those of 20-29 and of 30-59 years.
Meningiomas
Meningiomas are tumours of the meninges, the membranes that separate the
nervous tissue of the brain from the skull. No increased risk for meningioma was
observed in the Danish cohort study. In the Hardell studies an increased risk was
found in the highest category for time since first use (more than 10 years), but
Executive summary                                                                  29
</pre>

====================================================================== Einde pagina 30 =================================================================

<br><br>====================================================================== Pagina 31 ======================================================================

<pre>   only for analogue and not for digital mobile phones. In the INTERPHONE study
   decreased relative risks were observed in the two middle of four categories. All
   other exposure metrics were not associated with risk for meningioma.
   Acoustic neuromas
   Acoustic neuromas are tumours of the acoustic nerve. In the Danish cohort study
   no increased risk for acoustic neuroma was found for having a mobile phone
   subscription for more than 11 years. Hardell found an increased risk associated
   with the use of analogue phones for all times since first use, even as short as
   more than 1-5 years. For digital phones an increased risk was found only for the
   shorter follow-up times, but not for more than 10 years use.
       In the laterality analysis of the Hardell data increased risks for both analogue
   and digital mobile phones were found for both more than 1 year and more than
   10 years ipsilateral use. No increased risks for time since first use were found
   overall and in the laterality analysis of the INTERPHONE study, nor in a
   Japanese case-case study.
       In the Hardell studies increased risks were associated with all types of phones
   for a cumulative call time of more than 1000 hour. No increased risks were found
   in the INTERPHONE study for cumulative call times of 1640 hours or more, but
   decreased risks were observed in several of the lower categories. For ipsilateral
   use the risk was increased for cumulative call times of 1640 hours or more, but
   decreased for the next-lower category.
   Parotid glands tumours
   Parotid glands are the salivary glands most exposed when making a call with a
   mobile phone. No increased risks associated with time since first use or
   cumulative call time were found for parotid gland tumours in the Hardell studies,
   nor in studies following the INTERPHONE protocol. The only increased risk
   was found in one subgroup in a study following the INTERPHONE protocol, in
   the group containing both benign and malignant tumours that reported ipsilateral
   phone use and a cumulated call time of more than 266 hour.
       Only one ecological study reported on parotid gland tumours and found a
   rather constant incidence. Incidence data for parotid gland tumours for the
   Netherlands do not show changes in the incidence of this tumour over the period
   1989-2010.
30 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 31 =================================================================

<br><br>====================================================================== Pagina 32 ======================================================================

<pre>Issues to be considered in the evaluation
Latency time
An important point to be considered in the study of slow growing tumours is
their latency time, i.e. the time between induction of the tumour and its clinical
manifestation. Hardly any information is available, however, on latency periods
for the tumours considered in this report. The Committee considers it possible
that a follow-up time of 10 years would not be enough to measure an increase in
tumour incidence.
Analogue versus digital phones
The first mobile phones used an analogue type of signal, while the later GSMs
used a digital signal. This means that the subjects in the epidemiological studies
that have been using mobile telephony for the longest time periods (10 years or
more) will initially have used analogue phones. The exposure from the analogue
phones was higher than that from the digital ones.
Cordless versus mobile phones
Cordless phones are wireless phones with a limited range used indoors, such as
DECT phones. In several studies Hardell also investigated the risks from the use
of cordless phones, under the assumption that the radiofrequency field exposures
from that type of phone is of comparable magnitude as that from mobile phones.
However, exposure from cordless phones is lower than that from mobile phones.
    Thus the grossly similar increased risks for the use of mobile or cordless
phones observed by Hardell are hard to explain on the basis of actual incident or
total exposure. It is not known, but considered possible by the Committee, that
there is a correlation between the use of both types of phones. This could in part
be an explanation for the increased risks found for cordless phone use.
Nevertheless the Committee feels that the cordless phone data challenge the
internal consistency of the Hardell studies.
Executive summary                                                                  31
</pre>

====================================================================== Einde pagina 32 =================================================================

<br><br>====================================================================== Pagina 33 ======================================================================

<pre>   Strengths and limitations of the studies
   Cohort studies
   Cohort studies generate potentially strong evidence, as the exposure can be
   repeatedly and objectively measured or assessed before the outcome occurs.
   These studies therefore do not suffer from recall bias, but they may suffer from
   other problems.
       The Danish cohort study merely considered whether or not subjects held a
   private subscription that was started before 1996, and compared this group to all
   other residents of Denmark. Clearly the time that passed since a subscription
   started is a less meaningful endpoint than an estimate of the actual amount of
   use, which is more directly associated with exposure.
       In the later publications of this study with longer follow-up there will be
   increasing misclassification in the non-users group. This is because holders of
   business contracts, who are possibly among the heaviest users, were excluded
   from the users group, and because mobile phone possession in the Danish
   population strongly increased after 1996. It has been argued that because of this
   the Danish cohort is of limited value.
       Despite the lack of actual exposure data, the Committee considers the Danish
   cohort important for the overall evaluation. This is because misclassification in
   the non-users group has only very limited effect on the calculated risk and there
   is no misclassification in the users group.
   Case-control studies
   From the case-control studies, the Committee considers the INTERPHONE
   studies to be prone to selection bias due to the overall relatively low response
   rates. Because these are also lower for the controls than for the cases, this might
   lead to differential misclassification (i.e. the misclassification is different for
   cases and controls). This reinforces the selection bias.
       This is possibly reflected in the decreased risks observed in some of the
   lowest exposure categories: a protective effect from mobile phone use is not very
   likely. But this would mean that the observed increased risks in the highest
   categories may also be too low due to selection bias, while on the other hand they
   also could be too high due to recall bias. It is not possible to assess the extent of
   these biases.
32 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 33 =================================================================

<br><br>====================================================================== Pagina 34 ======================================================================

<pre>    The Hardell studies reported higher response rates and smaller differences in
response rates between cases and controls than the INTERPHONE studies. So
they are less likely to suffer from selection bias than the INTERPHONE studies.
However, the response rates in especially the controls of the Hardell studies are
unusually high.
    Another issue with the Hardell studies is that increased risks were already
observed with short usage times. These are unlikely in view of the presumably
very long latency times of the tumours under consideration. Also, if these
increased risks were true, increased incidences in the ecological studies would be
expected, but these were not observed.
    In both the Hardell and INTERPHONE studies there is also the possibility of
observer bias. In spite of the training of the interviewers, they might in some way
have been unknowingly influenced by the case or control status of the subjects.
The direction of effect of this bias is unclear. Both studies are also inherently
prone to recall bias, as exposure assessment in case-control studies is always
retrospective. Recall bias has been shown to be different between cases and
controls and is expected to cause over-estimation of risk.
    Another point that is important to take into account is the fact that the Hardell
studies have been performed in only one country (Sweden), while the
INTERPHONE studies cover 16 areas in 13 countries, thus covering a much
broader population. The total numbers of cases and controls are also lower in the
Hardell studies compared to the INTERPHONE studies.
The increased risks associated with cordless phone use and short latency times
observed in the Hardell studies, that are difficult to explain, combined with the
smaller size of the Hardell studies compared to the INTERPHONE studies, made
the Committee to give the Hardell studies less weight than the INTERPHONE
studies in the overall analysis and conclusions.
Case-case studies
Case-case studies are potentially powerful, as they are less likely to suffer from
selection and observer bias. There will of course still be recall bias, but this will
be non-differential, since only patients are included.
Ecological studies
Ecological studies are inherently limited in their interpretation, since individual
exposure is not determined. Exposure-effect relationships cannot be derived
Executive summary                                                                     33
</pre>

====================================================================== Einde pagina 34 =================================================================

<br><br>====================================================================== Pagina 35 ======================================================================

<pre>   from ecological studies. At best, they can show a similarity in trends in increase
   of disease and phone use.
       Absence of an increase in disease incidence with a preceding increase in
   mobile phone possession (and presumed use) does not prove the absence of a
   causal relation between exposure and disease. If the latency would be a decade or
   less, an increased risk would have been expected in the trends by now.
   Tumour-specific conclusions
   Glioma
   The Committee concludes that there are some weak and inconsistent indications
   for an association between prolonged and intensive use of a mobile phone and an
   increased incidence of gliomas. These might be explained by various types of
   bias and chance, but it cannot be excluded that there is a causal relation.
   However, the Committee estimates the likelihood for a causal relation to be very
   low.
       The population statistics, also in the Netherlands, do not show an increased
   incidence of glioma. But since it is likely that the latency time for these tumours
   is very long, an increased incidence might not yet be visible.
       The ecological studies also do not support an increased risk. If the risks
   reported by the Hardell group were true, a clearly increased glioma rate should
   have been visible in recent cancer statistics and the latency time should have to
   be much shorter than the the currently assumed possibly more than 10 years. The
   increased risk reported in the INTERPHONE studies, that is lower than that in
   the Hardell studies, would not show up yet in the statistics.
       The Committee concludes that their may be a small risk for an increased
   glioma incidence in association with the use of mobile phones, but it is also
   possible that such risk does not exist.
   Meningioma
   The Committee concludes that there are no clear and consistent indications for an
   increased risk of meningioma from using a mobile telephone.
34 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 35 =================================================================

<br><br>====================================================================== Pagina 36 ======================================================================

<pre>Acoustic neuroma
The Committee feels that the data on an association between long term use of a
mobile phone and acoustic neuroma are inconsistent and do not really give an
indication for an increased risk.
Parotid gland tumours
The Committee concludes that there are no clear indications for an increased risk
of parotid gland tumours from using a mobile phone. Only one increased risk
estimate in one subgroup in one study with limited numbers of cases has been
observed. This could have been the result of chance. The incidence data,
including those from the Netherlands, also do not show an increase.
Other tumours
For pituitary tumours, melanoma eye tumours, intra-temporal facial nerve
tumours and neuroblastoma tumours no conclusions regarding risks associated
with the use of mobile phones can be drawn.
Overall conclusion
The present systematic analysis shows that, despite large research efforts, there is
still no clarity regarding a possible association between mobile phone use and an
increased risk of tumours in the brain and other regions of the head.
     There are some weak and inconsistent indications for an association between
prolonged and intensive use of a mobile phone and an increased incidence of
gliomas. These might be explained by various types of bias and by chance, but it
can also not be excluded that there is a causal relation. For the other types of
tumours, including meningiomas and acoustic neuromas, indications for an
increased risk are much weaker or completely absent.
     Based on the available epidemiological evidence described in this report and
taking into account the quality of the different studies and their strengths and
weaknesses, the final conclusion from this systematic analysis is then: there is no
clear and consistent evidence for an increased risk for tumours in the brain and
other regions in the head in association with up to approximately 13 years use of
a mobile telephone, but such risk can also not be excluded. It is not possible to
pronounce upon longer term use.
Executive summary                                                                    35
</pre>

====================================================================== Einde pagina 36 =================================================================

<br><br>====================================================================== Pagina 37 ======================================================================

<pre>36 Mobile phones and cancer</pre>

====================================================================== Einde pagina 37 =================================================================

<br><br>====================================================================== Pagina 38 ======================================================================

<pre>Chapter 1
        Introduction
1.1     Why this report?
        Since the allegation of a Florida inhabitant that his wife’s brain tumour was
        caused by excessive use of a mobile telephone, many studies have been
        performed into that hypothesis. When the first publication by Hardell et al. in
        19991 indeed suggested a relationship between the use of mobile telephones and
        brain cancer, this subject has become a matter of concern to the general public
        and to authorities.
            Despite the availability of quite some data, they do not present a clear-cut
        picture of the possible relationship between the use of mobile or cordless phones
        and tumours in the head and recent reviews have reached conflicting
        conclusions. In June 2011, the International Agency for Research on Cancer
        (IARC) classified radiofrequency electromagnetic fields as ‘possibly
        carcinogenic to humans’ (group 2B).2,3 When other recent reviews are
        considered, there is a lack of convergence into a common conclusion. Some
        reviews conclude, like IARC, that there are indications for an association
        between mobile phone use and an increased risk of brain cancers, and some call
        for (precautionary) measures.4-9 Others conclude that the data do not show such
        association.10,11 Many of these these reviews contain shortcomings and biases,
        but these will not be discussed.
            The Electromagnetic Fields Committee of the Health Council of the
        Netherlands (“the Committee”) performs its own analysis of the literature on this
        Introduction                                                                      37
</pre>

====================================================================== Einde pagina 38 =================================================================

<br><br>====================================================================== Pagina 39 ======================================================================

<pre>    subject. This report is the first of three to discuss this. The composition of the
    Committee is presented in Annex A.
1.2 The research question
    The basic question the Committee investigates is, whether there are indications
    for a causal relationship between exposure to radiofrequency electromagnetic
    fields (RF EMF) from mobile phones and tumours in the brain and various other
    tissues in the head (e.g. meninges, acoustic nerve, parotid glands). To this end,
    the Committee performs systematic analyses of the epidemiological and animal
    experimental literature. This report contains the results of the systematic analysis
    of the epidemiological evidence. Any associations observed in epidemiological
    studies may be indicative for a causal relation, but in general it is difficult to
    establish a causal relation from epidemiological evidence only, unless the
    association is consistently observed and the risk observed is high. Observing an
    exposure-response relationship is also an indication for a causal association. If
    this is not the case, additional evidence from experimental studies need to be
    investigated also. Therefore a second report will deal with the systematic
    analysis of animal experiments. The results of these two reports will be
    combined in a third report, that will present an overall evaluation.
1.3 This report
    The Committee has conducted an independent systematic search and review of
    the epidemiological literature on the relation between exposure to RF EMF from
    mobile phones and tumours in the brain and other tissues in the head, using
    objective methods. This report describes the methods used and presents the
    results of this study.
         Following an a priori defined protocol, all relevant studies, both case-control,
    cohort and other types of studies, were identified, extracted, selected for further
    analysis and evaluated for their quality.
         When analyzing epidemiological data, it is important to take into account a
    number of considerations formulated by Bradford Hill, in order to conclude on
    the possibility of a causal relation.12 These include strength, consistency,
    temporality, biological gradient (or exposure-response) and plausibility and will
    be discussed later.
    In Chapter 2 the Committee briefly describes the methods and results of the
    literature search; a full account of this process is provided in Annex B. In
38  Mobile phones and cancer
</pre>

====================================================================== Einde pagina 39 =================================================================

<br><br>====================================================================== Pagina 40 ======================================================================

<pre>Chapter 3 the methods of data analysis are given, including the extraction of the
data (with more details in Annex C) and the evaluation of the quality of the
studies (with more details in Annex D). The results of the literature selection
through the processes described in Chapter 2 are presented in Chapter 4,
organized by type of study (cohort, case-control, case-case and ecological), with
some remarks on strengths and weaknesses of the different study types. Annex E
to this chapter gives more details on the supporting literature, that is not included
in the main text. Annex F to this chapter gives the results of the data extraction
for core publications. The results of the quality evaluations of the selected studies
are given in Chapter 5, with more details on the qualitative evaluation in Annex
G. Chapter 6 presents a summary of the results of the selected studies, with more
details given in Annex H. In Chapter 7 the Committee discusses and integrates
the results (with detailed results of a meta-analysis presented in Annex I) and the
report is finalised with the conclusions and recommendations presented in
Chapter 8.
Introduction                                                                          39
</pre>

====================================================================== Einde pagina 40 =================================================================

<br><br>====================================================================== Pagina 41 ======================================================================

<pre>40 Mobile phones and cancer</pre>

====================================================================== Einde pagina 41 =================================================================

<br><br>====================================================================== Pagina 42 ======================================================================

<pre>Chapter 2
        Literature search
2.1     Method
        A systematic approach was followed to search for relevant publications. The
        search strategy and the methods of data analysis were determined before the start
        of the study. Using a combination of different keywords (cellular phone; mobile
        phone; cell phone; epidemiology; exposure assessment; dosimetry; radio waves;
        radio frequencies; electromagnetic fields; human; tumour; cancer; neoplasms),
        PubMed was searched, followed by hand-searching of reviews and other key
        papers. Initial searches were performed in the week of 20 July 2009 with a full
        repeat search on 15 August 2011, updated on 10 July 2012. A full account of this
        process is presented in Annex B.
2.2     Results
        There were 2083 publications identified in the final search. After the selection
        process, which is summarized in Figure 1 and Annex B, 85 publications
        remained that described original studies. These were subsequently analyzed as
        described in the next chapter.
        Literature search                                                                 41
</pre>

====================================================================== Einde pagina 42 =================================================================

<br><br>====================================================================== Pagina 43 ======================================================================

<pre>Figure 1 The selection of publications after the search.
42          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 43 =================================================================

<br><br>====================================================================== Pagina 44 ======================================================================

<pre>Chapter 3
        Methods of data analysis
        Prior to the analysis, literature searches were conducted on methods used for
        systematic review of observational research data. The results were used to
        develop methods for data-extraction and -evaluation.
3.1     Data extraction
        Searches in PubMed, Web of Science and Cochrane resources identified several
        systematic reviews that included elaborate descriptions of methods of data
        extraction.13-21 This material was used to check and expand the most extensive
        checklist identified. Particularly expansion for exposure assessment aspects was
        found to be needed as most checklists mainly focus on disease outcomes and the
        selection of the study populations.
            To ensure that important aspects were included, also a brief review of the
        epidemiology of relevant tumours was conducted. Relevant tumours were
        thought to be those related to the brain (including those of the acoustic system)
        and tumours of the parotid glands.22-36 Potential confounders (general risk
        factors for the relevant tumours) identified from this literature were age, sex,
        allergies and atopy, ethnicity (Caucasian vs. African or Asian) and a history of
        head irradiation. However, the associations were different for the different
        tumours.
            Taking these issues into account, the checklist presented in Table C1 in
        Annex C was developed and used in this study.
        Methods of data analysis                                                          43
</pre>

====================================================================== Einde pagina 44 =================================================================

<br><br>====================================================================== Pagina 45 ======================================================================

<pre>        The 85 selected publications were extracted independently by two trained
    epidemiologists. Conflicts of interpretation were resolved by discussion. There
    were no disagreements that necessitated third party arbitration. All extracted data
    were double checked for factual correctness (numbers in tables and graphs
    particularly) by a third party.
    Studies and publications
    In several cases, a single study or dataset was described in several publications.
    To evaluate all aspects of study design, studies have been extracted and evaluated
    as a whole. Not all publications contained all the information that was to be
    included in the extraction. Missing information could mostly be obtained from
    other publications on the same study. In some cases additional information was
    obtained by contacting the main author of the study. This will be specifically
    indicated in the presentation of results.
        The use of different publications to extract the results on different endpoints
    from a particular study carries the risk of double counting of data and therefore
    of overweighting the study. This has been carefully avoided. In the analysis by
    disease, the most recent and most complete publication from each study per
    tumour type was used. Only if a specific aspect was not described in the
    preferred publication, another paper was used to extract the data.
    Other information used
    Several papers generated Letters to the Editor. Most of these were identified in
    the literature search, others became available through other retrieval methods.
    They were used as supporting material together with the responses of the authors.
    This was also the case for editorials and commentaries.
3.2 Evaluation of the quality of studies
    The Committee thought it to be helpful for the interpretation of the data to
    consider some form of scoring or weighting of the evidence. Several of the
    reviews mentioned in 1.1 have actually applied some quality evaluation,
    although this was not11 or not clearly used5 in the overall analysis.
         A separate literature search for publications evaluating such scoring methods
    was performed and methods used by (collaborators of) Committee members
    were also considered. Several publications were identified that reviewed scoring
    methods for the quality of publications.37-39 One paper also validated domains
44  Mobile phones and cancer
</pre>

====================================================================== Einde pagina 45 =================================================================

<br><br>====================================================================== Pagina 46 ======================================================================

<pre>for assessment used in the various methods.40 However, various authors have
been quite strongly opposed to the use of scoring in general or of specific scoring
methods.41-44 As a compromise the Committee uses the scores only to summarize
the overall methodological quality of the selected papers and to present this in an
overview, but not as a numerical weight in an overall analysis.
     The Committee used the evaluation method of Monninckhof et al.45 as basis.
Since this method was originally developed for studies on physical activity and
breast cancer risk, slight modifications were introduced for the current purpose.
To evaluate the method, the opinion of external experts was sought regarding the
evaluation items themselves and the weights to be allocated. The Committee
further developed this into the detailed list of questions that is described in Table
D1 in Annex D. They are categorized into the main domains identified in the
literature on quality evaluation: selection of cases and controls, tumour
diagnosis, assessment of exposure, confounding and conflict of interest. Further
elaboration follows in Chapter 5.
Methods of data analysis                                                              45
</pre>

====================================================================== Einde pagina 46 =================================================================

<br><br>====================================================================== Pagina 47 ======================================================================

<pre>46 Mobile phones and cancer</pre>

====================================================================== Einde pagina 47 =================================================================

<br><br>====================================================================== Pagina 48 ======================================================================

<pre>Chapter 4
        Study design and methods
        The different studies identified are first described by study design: cohort, case-
        control, case-case and ecological studies. There are no intervention studies.
        Pooled analyses of studies from a particular study group are used preferentially.
        Meta-analyses are not included, since only primary studies were to be part of this
        review. Only data from the original publications are given here, additional
        information used in the evaluation is presented in Annex E and results of the
        extraction for selected publications as highlighted in the tables is presented in
        Annex F.
            But before giving the descriptions of the studies, it is necessary to discuss
        briefly different biases that may occur.
4.1     Bias
        Recall bias
        A major problem with many epidemiological studies is obtaining accurate
        information on past exposure. This is usually dependent on the memory of the
        subjects under study. Apart from the problem that the recollection of specific
        exposures in the sometimes distant past is generally inaccurate, memory may
        also be influenced if someone is aware that he or she has a particular disease:
        cases may report their exposure more accurately than controls, because the latter
        feel less involved in the study, or cases may overestimate their exposure, because
        Study design and methods                                                            47
</pre>

====================================================================== Einde pagina 48 =================================================================

<br><br>====================================================================== Pagina 49 ======================================================================

<pre>    they believe the exposure caused their disease. The accuracy of recollection of
    exposure may therefore differ between cases and controls. This is known as
    differential recall bias and can affect the outcome of the study.
         Since in particular brain tumours have a very poor prognosis, it is important
    that cases be identified and interviewed soon after diagnosis. The accuracy of
    information of mobile phone use provided by family members when the patient
    is too ill to be interviewed or deceased, is considered to be less than that provided
    by the cases themselves.46
    Observation bias
    If the researchers collecting the information via interviews or questionnaires are
    aware of the disease status of the study subjects, this may result in observation
    bias that may compound the recall bias. Both types of bias may result in
    differential misclassification, i.e. they affect case and control data differently.
    This usually results in overestimation of the actual risk, although
    underestimation is also possible.
    Selection bias
    An important issue with case-control studies is the selection of cases and
    controls. Ideally, the two groups should come from the same population and be
    sampled over the same period of time. If this is not the case, this may result in
    selection bias. Also, relatively high response rates are important, as these will
    reduce the risk of selection bias. This type of bias also may result in over- or
    underestimation of the actual risk.
4.2 Cohort studies
    In this study design a group of subjects that is initially free of the disease(s) of
    interest, the cohort, is followed over a certain period of time. During follow-up,
    the occurrence of the disease(s) of interest is registered and exposure to the
    factor(s) of interest is monitored or measured. At relevant follow-up times,
    disease incidence in different exposure groups can be determined and the risk of
    exposure calculated. Cohort studies can be either prospective or retrospective.
    The major advantage of prospective cohort studies is that the exposure is
    measured before the occurrence of the disease and that changes in exposure can
    be measured as they occur. This type of studies is thus not vulnerable to recall
    bias, i.e. misclassification of (past) exposure. In retrospective cohort studies the
48  Mobile phones and cancer
</pre>

====================================================================== Einde pagina 49 =================================================================

<br><br>====================================================================== Pagina 50 ======================================================================

<pre>              exposure took place in the past and is reconstructed using routine data such as
              employment records or e.g. subscriptions. Changes in exposure or precise
              estimates can be difficult to assess in retrospective cohort studies. Cohort studies
              examining brain cancers have some drawbacks. Because the disease is relatively
              rare, the cohort needs to be very large. The disease has a long latency period, i.e.
              it may take a long time after induction before the disease becomes manifest (see
              Chapter 6), therefore the follow-up period of the cohort needs to be long, up to
              several decades. Exposure assessment over such a long time period may give
              problems, particularly with changing exposures such as from (mobile) phone
              use.
                   Six publications on cohort studies were identified, based on two cohorts.47-52
              Table 4.1 presents the publications of the original studies. The studies selected
              for evaluation and final analysis are indicated in boldface type.
                   With mobile phone use, the exposure may fluctuate and change over time,
              which can lead to non-differential misclassification. It is therefore important to
              perform regular exposure assessments in cohort studies.
                   The first of the two cohorts discussed in this report is from the USA, with
              publications from Rothman et al. (1996)49 and Dryer et al. (1999).47 The main
              problems are the short period of follow up (the median duration of a mobile
              phone subscription in the highest category was 3.8 years) and the fact that
              mortality and not incidence was investigated. Therefore the results are only
              relevant for the question whether mobile phone use might act as a promoter
Table 4.1 Cohort studies.
Reference                       Type of tumour                    Exposure assessment          Country / time period / ages
Rothman et al. (1996)49         None; overall mortality           Length contract, type phone, Boston, Chicago, Dallas,
                                                                  duration calls               Washington DC/ USA/ 1994
                                                                                               ≥ 20 y at start
Dreyer et al. (1999)47          Brain cancer                      Idem                         Idem
                                                                                               ≥ 20 y at start
Johansen et al. (2001)48        Cancer, including brain &         Length of contract for those Denmark, 1982-1996
                                central nervous system            with contract before 1996    ≥ 30 y at start
                                tumours, parotid gland tumour
Schüz et al. (2006)50           Cancer, including glioma,         Idem                         Denmark, 1982-2002
                                meningioma, acoustic neuroma,                                  ≥ 30 y at start
                                parotid gland tumour
Schüz et al. (2011)51           Vestibular schwannoma             Idem                         Denmark, 1982-2006
                                (acoustic neuroma)                                             ≥ 30 y at start
Frei et al. (2011)52            Brain tumours, including          Idem                         Denmark, 1982-2007
                                glioma, meningioma                                             ≥ 30 y at start
 The publications indicated in bold were used for quality evaluation.
              Study design and methods                                                                                    49
</pre>

====================================================================== Einde pagina 50 =================================================================

<br><br>====================================================================== Pagina 51 ======================================================================

<pre>    for brain cancers, but even then the promoting effect would have to cause a
    considerable acceleration of tumour growth to result in measurable changes in
    mortality in such a short time period.
        The second cohort is from Denmark, with a maximum follow up time of 21
    years, with publications from Johansen et al. (2001), Schüz et al. (2006, 2011)
    and Frei et al. (2011).48,50-52 This is a very large retrospective cohort study, in
    which mobile phone company records were used to determine whether a private
    mobile phone subscription was started before 1996. These data thus do not
    provide information on actual exposure to radiofrequency fields such as number
    and duration of calls.
        A large multinational prospective cohort study (COSMOS) has recently been
    started, but it will take many years before results are available. The study design
    is described by Schüz et al. (2011).53
        There were 15 (invited) Letters to the Editor or Editorials and one supple-
    mentary publication concerning the cohort studies. They are listed in Annex E,
    Table E1. Results from the data extraction are presented in Annex F.
        The evaluation of the methodological quality of the cohort studies is
    presented in Chapter 5, Table 5.1, more details are given in Annex G, Table G2.
4.3 Case-control studies
    In this study design a comparison is made between a group of subjects with a
    given disease (cases) and a suitable control group of subjects without the disease.
    The past history of exposure to a suspected risk factor is determined and groups
    of cases and controls with similar exposures are compared. This allows the
    (relative) risk of exposure to be calculated, i.e. the risk of exposure to the factor
    under investigation relative to the combined risk of all other factors that are not
    studied.
        In the studies of cancer in relation to mobile phone use, the case-control
    studies come in three ‘clusters’ according to the study protocol used. The first
    group of studies contains those performed by Hardell et al.; these all used the
    same protocol. The second group of studies contains those from the
    INTERPHONE program; they all used a core protocol developed in
    collaboration with the International Agency for Research on Cancer (IARC). The
    third group contains all other case-control studies identified, that used a variety
    of protocols.
50  Mobile phones and cancer
</pre>

====================================================================== Einde pagina 51 =================================================================

<br><br>====================================================================== Pagina 52 ======================================================================

<pre>4.3.1 Case-control studies according to the Hardell protocol
      This group contains 18 publications of closely related studies, sometimes using
      the same data or combining the data used in previous publications.1,54-70 They all
      used the same study protocol and the same method of data collection, such as the
      same questionnaire and additional interview methods. Upon request, Hardell
      provided the questionnaire and informed the Committee that in all cases
      additional information was obtained by telephone interview, during which the
      interviewers were blinded for case-status. The Committee did not have access to
      the questions used in the telephone interviews.
          Hardell et al. performed three distinct case-control studies (designated 1, 2,
      and 3 in Table 4.2) that included only prevalent cases, i.e. respondents who were
      alive at the time of enrolment. A fourth study interviewed family members of
      deceased cases. Unfortunately there is some overlap in successive papers on the
      same studies, and many different subgroups are analysed. This makes it
      sometimes difficult to get a clear picture of the studies by the Hardell group and
      also increases the occurrence of significant results just by chance. Hardell et al.
      also performed several pooled analyses, in which the results from their studies 2
      and 3 and sometimes 4 are combined. The pooled study that included the
      deceased cases is the most complete one, because it addresses all incident cases.
      Table 4.2 presents the publications on the original and pooled studies. The
      studies selected for evaluation and final analysis are indicated in boldface type.
          There were two Letters to the Editor identified and one supporting paper and
      these are listed in Annex E, Table E2. Results from the data extraction are
      presented in Annex F.
          The quality evaluation of these studies is presented in Chapter 5, Table 5.1
      with details given in Annex G, Table G3.
      Study design and methods                                                            51
</pre>

====================================================================== Einde pagina 52 =================================================================

<br><br>====================================================================== Pagina 53 ======================================================================

<pre>Table 4.2 Case-control studies of the Hardell group.
Reference       Type of tumour Original / pooled Population /         Response (%)                              Time period /
                                   / study no.       hospital based /                                           place/ topic of
                                                     ages             Published              Recalculated a     analysis
Hardell et al.  Brain tumour       Original 1        Population       Cases: 90%             Cases: 77%         1994-1996/ 2 city
(1999)1         (incl. glioma,                       20-80 y old      Controls: 91%          Controls: 79%      regions Sweden
                meningioma,
                acoustic neuroma)
Hardell et al.  Brain tumour       Original 1        Population       Cases: 90%             Cases: 77%         1994-1996/ 2 city
(2001)54                                             20-80 y old      Controls: 91%          Controls: 79%      regions Sweden
Hardell et al.  Brain tumour       Original 2        Population       Cases: 88%             Cases: 72%         1997-2000/ 4 city
(2002)55        (incl. glioma,                       20-80 y old      Controls: 91% but      Controls: xx%      regions Sweden
                meningioma,                                           complete pairs 81%     (no details given)
                acoustic neuroma)                                     as used for analysis
Hardell et al.  Malignant brain Original 2           Population       Cases: 91%             Cases: 59%         1997-2000/ 4 city
(2002)56        tumour (incl.                        20-80 y old      Controls: 90% but      Controls: 90%      regions Sweden
                glioma)                                               complete pairs 82%
Hardell et al.  Brain tumour       Original 2        Population       Cases: 88% but is      Cases:72%          1997-2000/ 4 city
(2003)57        (incl glioma,                        20-80 y old      only 63% of cases      Controls: xx%      regions Sweden
                meningioma,                                           reported in cancer     (no details given)
                acoustic neuroma)                                     registry
                                                                      Controls: 91%
Hardell et al. Parotid gland       Original          Population       Cases: 91%             Cases: 64%         1994-2000/ 6 city
(2004)69 b      tumour                               20-80 y old      Controls: 92%          Controls: 90%      regions Sweden
Hardell et al.  Brain tumour       Original 2        Population       Cases:88%              Cases: 65%         1997-2000/ central
(2004)71                                             20-80 y old      Controls: 91%          Controls: xx%      region Sweden/
                                                                                             (no details given) Age
Hardell et al.  Brain tumour       Original 2        Population       Cases: 88% but is      Cases: 72%         1997-2000/ central
(2005)59                                             20-80 y old      only 63% of cases      Controls: xx%      region Sweden/
                                                                      reported in cancer     (no details given) Rural vs. urban
                                                                      registry
                                                                      Controls: 91%
Hardell et al.  Acoustic           Original 3        Population       Cases: 89% (but 18 Cases:59%              2000-2003/ 2 city
(2005)60        neuroma,                             20-80 y old      not incl. as deceased) Controls: xx%      regions Sweden
                meningioma                                            Controls:88%           (no details given)
Hardell et al.  Malignant brain    Original 3        Population       Cases: 88%             Not enough detail  2000-2003/ 2 city
(2006)61        tumour                               20-80 y old      Controls: 84%          for calculation    regions Sweden
Hardell et al.  Malignant brain    Original 4        Population       Cases: 75%             Cases: 65%         1997-2003/ 4 city
(2010)62        tumour                               20-80 y old      Controls 67%           Controls: xx%      regions Sweden
                                                                      (average)              (no details given)
                                                                      Controls cancer: 74%
                                                                      Controls other
                                                                      diseases 60%
Söderqvist et Parotid gland        Original          Population       Cases: 88%,            Cases: 75%         2000-2003 / 3 city
al. (2012) 72 tumour                                 22-80 y old      Controls: 83%          Controls: 83%      regions (9/21
                                                                                                                counties) Sweden
52             Mobile phones and cancer
</pre>

====================================================================== Einde pagina 53 =================================================================

<br><br>====================================================================== Pagina 54 ======================================================================

<pre>Hardell et al.   Malignant brain   Pooled 2+3       Population       Cases: 90% but this is Not enough detail 1997-2003
(2006)63         tumour                                              only 65% of cancer for calculation
                                                                     registry cases
                                                                     Controls: 89%
Hardell et al.   Benign brain      Pooled 2+3       Population       Cases: 88%             Not enough detail 1997-2003
(2006)64         tumour (incl.                                       Controls: 89%          for calculation
                 meningioma,
                 acoustic neuroma)
Hardell et al. Brain tumour        Pooled 2+3       Population       Cases:88%              Not enough detail 1997-2003
(2006)67         (incl. glioma,                                      Controls: 84%          for calculation     Mobile+cordless
                 meningioma,
                 acoustic
                 neuroma), parotid
                 gland tumour
Hansson Mild Brain tumour          Pooled 2+3       Population       Cases: 90%             Not enough detail 1997-2003
et al. (2007)65 (incl. glioma,                                       (malignant tum.);      for calculation
                 meningioma,                                         88% (benign tum.,
                 acoustic neuroma)                                   incl. meningioma,
                                                                     acoustic neuroma)
                                                                     Controls: 89%
Hardell et al. Brain tumour        Pooled 2+3       Population       Cases: 90%             Not enough detail 1997-2003
(2009)66         (incl. glioma,                                      (malignant tum.);      for calculation     Mobile+cordless
                 meningioma,                                         88% (benign tum.,
                 acoustic neuroma)                                   incl. meningioma,
                                                                     acoustic neuroma)
                                                                     Controls: 89%
Hardell et al. Malignant brain     Pooled 2+3+4     Population       Cases: 85%             Not enough detail 1997-2003
(2011)68         tumour                                              Controls: 84%          for calculation but
                                                                                            as includes
                                                                                            deceased expected
                                                                                            similar
a Recalculated by including excluded cases that were deceased or declared too ill by their physician. This was only done for the
studies where these subpopulations had been included in the response calculations.
b The publications indicated in bold were used for quality evaluation.
4.3.2          Case-control studies according to the INTERPHONE protocol
               In the INTERPHONE consortium 16 research groups conducted case-control
               studies on different types of tumours in the head and neck area, including the
               brain, in 13 countries using a common core protocol.73,74 Several groups
               published their data individually, several pooled assessments of a limited number
               of groups were made, and for glioma and acoustic neuroma pooled analyses of
               the data from all groups have been published (which was the initial objective of
               the INTERPHONE studies). In all the studies data were collected by computer-
               assisted personal interview.74
               Study design and methods                                                                                       53
</pre>

====================================================================== Einde pagina 54 =================================================================

<br><br>====================================================================== Pagina 55 ======================================================================

<pre>                   The 20 publications on original and pooled data are presented in Table 4.3.
              The studies selected for evaluation and final analysis are indicated in boldface
              type.
Table 4.3 Case-control studies of the INTERPHONE consortium.
Reference             Type of tumour        Original / Population /        Response (%)              Country a;
                                            pooled     hospital based /                              specific topic
                                                       ages
Christensen et        Acoustic neuroma      Original   Population          Cases: 82%                Denmark
al.(2004)75                                            20-69 y old         Controls: 64%
Lönn et al. (2004)76 Acoustic neuroma       Original   Population          Cases: 93%                3 cities Sweden
                                                       20-69 y old         Controls: 72%
Christensen et        Glioma, meningioma Original      Population          Cases: glioma 71%;        Denmark
al.(2005)77                                            20-69 y old         meningioma 74%
                                                                           Controls: 64%
Lönn et al. (2005)78 Glioma, meningioma     Original   Population          Cases: glioma 74%,        4 cities Sweden
                                                       20-69 y old         meningioma 85%
                                                                           Controls: 71%
Klaeboe et al.        Glioma, meningioma, Original     Hospital for cases, Cases: 74%                2 regions Norway
(2007)79              acoustic neuroma                 population          Controls: 69%
                                                       controls?
                                                       19-69 y old
Schüz et al. (2006)80 Glioma, meningioma    Original   Hospital for cases, Cases: glioma 80%,        4 cities Germany;
                                                       population          meningioma 88%            DECT base
                                                       controls?           Controls: 63%             stations
                                                       30-59 y old
Schüz et al. (2006)81 Glioma, meningioma    Original   Hospital for cases, Cases: glioma 80%,        4 cities Germany
                                                       population          meningioma 88%
                                                       controls?           Controls: 63%
                                                       30-59 y old
Lönn et al. (2005)82 bParotid gland tumour  Original   Population          Cases: 85% overall (79%   Denmark, 3 cities
                                                       20-69 y old         Denmark, 89% Sweden)      Sweden
                                                                           Controls:70% overall (60%
                                                                           Denmark, 72% Sweden)
Takebayashi et al.    Acoustic neuroma      Original   Hospital for cases, Cases: 84%                Greater Tokyo
(2006)83                                               population controls Controls:52%              area, Japan
                                                       30-69 y old
Hepworth et al.       Glioma                Original   Population for      Cases: 51%                South-east, north-
(2006)84                                               cases, GP’s for     Controls: 45%             east UK
                                                       controls
                                                       SE: 18-59 y
                                                       NE: 18-69 y
Sadetzki et al.       Parotid gland tumour  Original   Population          Cases: 87%                Israel
(2008)85                                               ≥ 18 y of age       Controls: 66%
Schlehofer et al.     Acoustic neuroma      Original   Hospital for cases, Cases: 89%                4 cities Germany
(2007)86                                               population controls Controls: 55%
                                                       30-59 y old
54            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 55 =================================================================

<br><br>====================================================================== Pagina 56 ======================================================================

<pre>Hours et al. (2007)87 Glioma, meningioma, Original          Population          Cases: glioma 60%,        Lyon, France
                      acoustic neuroma                      30-59 y old         meningioma 78%, acoustic
                                                                                neuroma 81%
                                                                                Controls: 75%
Takebayashi et al.    Glioma, meningioma, Original          Hospital for cases Cases: glioma 59%,         Greater Tokyo
(2008)88              pituitary adenoma                     estimated to        meningioma 78%, pituitary area, Japan
                                                            represent 75% of adenoma, 76%
                                                            total # of cases in Controls: 51%
                                                            area, population
                                                            controls
                                                            30-69 y old
Schoemaker et al.     Pituitary tumours       Original      Population for      Cases: 61% (calculated)   South-east UK
(2009)89                                                    cases, GP’s for     Controls 43%:
                                                            controls
                                                            18-59 y old
Schoemaker et al.     Acoustic neuroma        Pooled        Mixed               Cases: 83% (69-91%)       Nordic countries,
(2005)90                                                                        Controls: 51% (42-69%)    south-east UK
Lahkola et al.        Glioma                  Pooled        Mixed               Cases 69%:(37-81%)        Nordic countries,
(2007)91                                                                        Controls: 50% (42-69%)    south-east UK
Lahkola et al.        Meningioma              Pooled        Mixed               Cases: 74% (55-90%)       Nordic countries,
(2008)92                                                                        Controls: 50% (42-69%)    south-east UK
INTERPHONE            Glioma, meningioma      Pooled        Mixed               Cases: glioma 64% (36-    13 countries
study group (2010)93                                                            92%), meningioma 78%
                                                                                (56-92%)
                                                                                Controls: 53% (42-74%)
INTERPHONE            Acoustic neuroma        Pooled        Mixed               Cases: 82% (70-100%)      13 countries
study group (2011)94                                                            Controls: 53% (35-74%)
a Nordic countries: Denmark, Norway, Sweden & Finland;13 countries: Nordic, Australia, Canada, France, Germany, Israel,
Italy, Japan, New Zealand, north-east & south-east UK.
b The publications indicated in bold were used for quality evaluation.
              Fifteen supporting papers and 30 comments as well as Letters to the Editor and
              associated author responses were also considered in the context of these
              publications. These are listed in Annex E, Table E3. Results from the data
              extraction are presented in Annex F.
                  The quality evaluations of these case-control studies are presented in Chapter
              5, Table 5.1; details are given in Annex G, Table G3.
4.3.3         Case-control studies according to other protocols
              This group contains 14 publications on original investigations related to mobile
              phone use.95-108 A wide variety of methods for data collection and population
              sampling was used. The information on potential exposure (retrospectively
              gathered as in all case-control studies) was limited in most of these studies; also
              the number of cases and controls in the categories with longer exposure duration
              Study design and methods                                                                                    55
</pre>

====================================================================== Einde pagina 56 =================================================================

<br><br>====================================================================== Pagina 57 ======================================================================

<pre>               was extremely limited. For most of the studies the total number of participants
               was very small (often not more than 25), and the duration of exposure to mobile
               phones short (less than 5 years; see the details in the tables in Annex F).
               Therefore, the relevance to the interpretation of long-term effects is minimal.
               The details of the publications of case-control studies according to other
               protocols are presented in Table 4.4. The studies selected for evaluation and final
               analysis are indicated in boldface type.
Table 4.4 Other case-control studies.
Reference                Type of tumour            Original / Population /       Response (%)      Time period /
                                                   pooled     hospital based /                     place / country
                                                              ages
Inskip et al. (1999)95   Glioma, meningioma,       Original   Hospital           Cases: 92%        1994-1998
                         acoustic neuroma                     ≥ 18 y old         Controls: 86%     Phoenix, Boston,
                                                                                                   Pittsburgh, USA
Muscat et al. (2000)96 a Primary brain cancer,     Original   Hospital           Cases: 82%        1994-1998
                          incl. glioma                        18-80 y old        Controls: 90%     New York,
                                                                                                   Providence, Boston,
                                                                                                   USA
De Roos et al. (2001)97 Neuroblastoma              Original   Hospital           Cases: 73%        1992-1994
                                                              ≤ 19 y old         Controls: 71%     139 hospitals, USA
                                                                                                   & Canada
Stang et al. (2001)98    Uveal melanoma            Original   Population         Cases: 84%        1994-1997
                                                              35-69 y old +      Controls: 81%     Essen+ all of
                                                              Hospital                             Germany
                                                              35-74 y old
Inskip et al. (2001)99   Glioma, meningioma,       Original   Hospital           Cases: 92%        1994-1998
                         acoustic neuroma                     ≥ 18 y old         Controls: 86%     Phoenix, Boston,
                                                                                                   Pittsburgh, USA
Auvinen et al. (2002)100 Glioma, meningioma,       Original   Population         Cases: 100%       1996
                         parotid gland tumour                 20-69 y old        Controls: 100%    All Finland
                                                                                 as register-based
Muscat et al. (2002)101 Acoustic neuroma           Original   Hospital           Cases: 100%?      1997-1999
                                                              ≥ 18 y old         Controls: 100%?   New York, USA
Warren et al. (2003)102 Intratemporal facial nerve Original   Hospital           Cases: 100%?      1995-2000
                         tumours                              Cases: mean 47 y Controls: 100%?     Gainesville (Fl),
                                                              old                                  USA
                                                              Controls: mean
                                                              57.8, 52.6, 50.8 y
                                                              old
Gousias et al. (2009)103 Glioma                    Original   Population         Cases: 100%?      2005-2007
                                                              22-82 y old        Controls: 100%?   6 districts of Greece
Stang et al. (2009)104   Uveal melanoma            Original   Hospital           Cases: 94%        2002-2004
                                                              20-74 y old        Controls: 57%     Essen, Germany
                                                                                 (hospital) & 52%
                                                                                 (population)
56             Mobile phones and cancer
</pre>

====================================================================== Einde pagina 57 =================================================================

<br><br>====================================================================== Pagina 58 ======================================================================

<pre>Spinelli et al. (2010)105 Glioma                    Original      Hospital        Cases: 72%         2005
                                                                  ≥ 18 y old      Controls: 100%?    Marseille, Toulon,
                                                                                                     France
Duan et al. (2011)106     Parotid gland tumour      Original      Hospital        Cases: 78%         1993-2010
                                                                  7-80 y old      Controls: 62%      Beijing, China
Baldi et al. (2011)107    Brain tumours             Original      Population      Cases: 70%         1999-2001
                                                                  ≥ 15 y old      Controls: 69%      Gironde, France
Aydin et al. (2011)108    Brain tumours children    Original      Population      Cases: 83%         2004-2008
                                                                  7-19 y old      Controls: 71%      All of Denmark,
                                                                                                     Sweden, Norway,
                                                                                                     Switzerland
a The publications indicated in bold were used for quality evaluation.
              Two supporting papers were identified for these case-control studies. They are
              listed in Annex E, Table E4. Results from the data extraction are presented in
              Annex F.
                   The quality evaluation of the original studies is presented in Chapter 5, Table
              5.1 and details are given in Annex G, Table G3.
4.4           Case-case studies
              This study design compares two or more groups of cases that differ in a specific
              characteristic, such as exposure or location of the tumour. Comparison of for
              instance reported mobile phone use by cases with a tumour closer to the surface
              with that of cases with a tumour at more central locations in the brain, may
              provide supporting information on a possible causal relation between exposure
              and disease. Since differential recall bias is less likely (because only cases are
              included) the influence of recall bias in case-case studies is minimised. It is
              however not entirely gone, as people may be aware of the location of the tumour
              and thus report accordingly.
                    Six publications of this design were identified and details are presented in
              Table 4.5.109-113 The studies selected for evaluation and final analysis are
              indicated in boldface type.
Table 4.5 Case-case studies.
Reference                  Type of tumour         Exposure estimate              Original / pooled /  Response (%) /
                                                                                 meta analysis        ages
Ali Kahn et al. (2003)109 Glioma                  Handedness in phone users vs.  Original             100%
                                                  tumour location                                     20-81 y old
Salahaldin & Bener         Acoustic neuroma       Possession of phone (yes / no) Original             100%?
(2006)110                                                                                             34-66 y old
Hartikka et al. (2009)111  Glioma                 Distance phone - tumour        Subsample of         100% (published)
                                                                                 INTERPHONE           69% (calculated)
                                                                                                      20-60 y old
              Study design and methods                                                                                 57
</pre>

====================================================================== Einde pagina 58 =================================================================

<br><br>====================================================================== Pagina 59 ======================================================================

<pre>Sato et al. (2010)112       Acoustic neuroma      Intensity of phone use and         Original            51%
                                                  laterality vs. tumour location and                     ≥29 - ≤70 y old
                                                  size
Cardis et al. (2011)114     Glioma, meningioma Intensity of phone use; based on Subsample of             Cases: glioma 42%
                                                  calculated RF energy               INTERPHONE          meningioma 56%
                                                                                     Australia, Canada,  Controls: for
                                                                                     France, Israel, New glioma 36%, for
                                                                                     Zealand             meningioma 40%
                                                                                                         30-59 y old
Larjavaara et al. (2011)113 Glioma                Case-specular*                     Subsample of        63%
                                                  Based on calculated RF exposure INTERPHONE             18-59 y old
                                                                                     Denmark, Finland,
                                                                                     Germany, Italy,
                                                                                     Norway, Sweden,
                                                                                     Southeast England
* Simulated case
The publications indicated in bold were used for quality evaluation.
               No Letters to the Editor or supporting papers were identified. Results from the
               data extraction are presented in Annex F. The quality evaluation of these studies
               is presented in Chapter 5, Table 5.1, with details given in Annex G, Table G4.
4.5            Ecological studies
               These studies investigate the occurrence of disease at population level in relation
               to the prevalence of (a proxy for) exposure in the population. They may analyze
               for instance the pattern of tumour occurrence over time (either by incidence or by
               mortality) in geographic entities such as countries, to identify any trends and to
               see whether these could be explained e.g. by trends in possession or use of
               mobile phones. Individual data on mobile phone use are not used in these studies.
               Such studies will inherently be limited by the poor level of insight into trends and
               patterns of mobile phone use, and hence of actual exposure, particularly for
               specific age, sex and other population group definitions.
                   It should be noted that for many countries substantial and wide-spread
               mobile phone use is relatively recent (Figure 2).
                   In most Western-European countries approximately half of the population
               had a mobile phone subscription in the year 2000. In the Nordic countries
               (Norway, Sweden, Finland and Denmark) the increase started earlier, but was
               caught up by the other countries around the century mark. By 2005 most people
               in the countries presented (except France and the USA) owned a mobile phone,
               but the extent of use is much less certain.
58             Mobile phones and cancer
</pre>

====================================================================== Einde pagina 59 =================================================================

<br><br>====================================================================== Pagina 60 ======================================================================

<pre>Figure 2 Number of mobile phone subscriptions for some European countries and the USA. Panels B and C show the same data
as panel A, but separated for Nordic and other countries. Data from ITU (http://www.itu.int/ITU-D/ict/statistics/explorer/
index.html).
             Taking into account that the latency period of brain tumours is likely more than
             10 years (see Chapter 6), it is thus possible that any trends in tumour occurrence
             related to mobile phone use may not yet be visible in most countries, with an
             exception perhaps for the Nordic countries, since use started earlier there.
                  In analyzing ecological studies, it has to be realized that trends in mortality
             can also be influenced by the introduction of more effective treatments and that
             trends in incidence can be affected by changes in diagnostic techniques.
             Study design and methods                                                                                      59
</pre>

====================================================================== Einde pagina 60 =================================================================

<br><br>====================================================================== Pagina 61 ======================================================================

<pre>               Ecological studies identified in this search were performed in various countries
               and totalled 21 publications.115-121,58,122-125,125-133 A summary of the publications
               is presented in Table 4.6.
Table 4.6 Ecological studies.
Reference                     Tumour type, Endpoint   Exposure assessment                 Time period Country
Counsell et al. (1996)126     Brain tumours           Trend, not in relation to phone     1989-1990   Scotland
                                                      possession or use
Howitz et al. (2000)116       Acoustic neuroma        Trend, not in relation to phone     1977-1995   Denmark
                                                      possession or use
Gurney & Kadan-Lottick        Brain tumours           Trend, not in relation to phone     1975-1997   USA, 11 states
(2001)127                                             possession or use
Cook et al. (2003)115         Head and neck tumours   Trend, in relation to phone         1986-1998   New Zealand
                                                      possession and exposure (from
                                                      location of tumour)
Inskip et al. (2003)117       Ocular melanoma         Right- vs. left sided tumours       1974-1998   USA, 5 states & 4
                                                      (assuming predominantly right                   metropolitan
                                                      sided phone use): trends & contrast             areas
                                                      pre/post 1995
Hardell et al. (2003)58       Brain tumours, acoustic Trend, not in relation to phone     1960-1998   Sweden
                              neuroma                 possession or use
Lönn et al. (2004)119         Primary brain tumours   Trend, in relation to phone         1996-1998   Denmark,
                                                      subscriptions                                   Finland, Norway
                                                                                                      and Sweden
Muscat et al. (2006)120       Neuronal brain cancers  Contrast pre/post 1985, in relation 1973-2002   USA, 5 states & 4
                                                      to phone subscriptions                          metropolitan
                                                                                                      areas
Nelson et al. (2006)128       Acoustic neuroma        Trend, in relation to phone         1979-2001   England & Wales
                                                      subscriptions
Röösli et al. (2007)121       Brain tumour            Trend, in relation to predicted     1969-2002   Switzerland
                                                      phone use based on subscriptions
Deltour et al. (2009)122      Glioma, meningioma      Trends, in relation to general      1974-2003   Denmark,
                                                      mobile phone use pre/post mid                   Finland, Norway
                                                      1990s                                           & Sweden
Inskip et al. (2010)123       Brain cancer            Trends, in relation to phone        1997-2006   USA, 5 states & 4
                                                      subscriptions                                   metropolitan
                                                                                                      areas (10% USA
                                                                                                      population)
Lehrer et al. (2010)124       Primary brain tumours   Relation with subscriptions in      2007        USA, 19 states
                                                      2007; comparison of 19 states
Johansen et al. (2002)118     Ocular melanoma         Trend, in relation to phone         1943-1996   Denmark
                                                      subscriptions
Czerninski et al. (2011)129   Parotid gland tumour    Trends, in relation to increase in  1970-2000   Israel
                                                      phone use
De Vocht et al. (2011)130     Brain tumours           Trends, in relation to phone        1998-2007   England
                                                      subscriptions
De Vocht (2011)134            Parotid cancer          Trends, in relation to phone        1986-2008   England
                                                      subscriptions
60             Mobile phones and cancer
</pre>

====================================================================== Einde pagina 61 =================================================================

<br><br>====================================================================== Pagina 62 ======================================================================

<pre>Kohler et al. (2011)131       Brain tumours         Trends, not in relation to phone    1975-2007 USA, 46
                                                    possession or use                             population based
                                                                                                  cancer (93% USA
                                                                                                  population)
Larjavaara et al. (2011)132   Vestibular Schwannoma Trends incl. birth cohorts, not in 1987-2007  Denmark,
                              (acoustic neuroma)    relation to phone possession or use           Finland, Norway
                                                                                                  & Sweden
Deltour et al. (2012)135      Glioma                Trends, in relation to general      1974-2008 Denmark,
                                                    mobile phone use                              Finland, Norway
                                                                                                  & Sweden
Little et al. (2012)133       Glioma                Trends, in relation to results from 1997-2008 USA 12 SEER
                                                    INTERPHONE and Hardell studies                regions
               Only those investigations using both outcome and exposure are assessed and the
               most recent investigation of the same data is discussed. These publications are
               identified in bold type face in Table 4.6.
                   Two Letters to the Editor were identified and are listed in Annex E. Results
               from the data extraction are presented in Annex F.
                   The Committee deemed a quality evaluation of the ecological studies not
               meaningful.
               Study design and methods                                                                         61
</pre>

====================================================================== Einde pagina 62 =================================================================

<br><br>====================================================================== Pagina 63 ======================================================================

<pre>62 Mobile phones and cancer</pre>

====================================================================== Einde pagina 63 =================================================================

<br><br>====================================================================== Pagina 64 ======================================================================

<pre>Chapter 5
        Evaluation of study quality
        To prevent evaluation of multiple publications on the same study, only the most
        recent publication for each data set for a specific outcome was selected.
            The full list with items used for the evaluation is shown in Table D1 in Annex
        D. The items are divided into several domains: Selection bias, referring to the
        selection of cases and controls (scored out of 34); Misclassification of outcome,
        referring to the method of ascertainment of tumour diagnosis (scored out of 4);
        Misclassification of exposure, referring to the assessment and classification of
        exposure (scored out of 69); Confounding, referring to the possibility of other
        factors influencing the outcome (scored out of 16); and Conflict of Interest;
        referring to the possibility that the outcomes were influenced by (financial or
        other) interests (scored out of 5). The agreed evaluations for these domains for
        the two scorers are presented in Table 5.1 as percentage of the maximum score
        for each domain. A detailed listing of the scores for each individual question is
        given in Tables G1, G2 and G3 in Annex G.
            The Committee weighted the domains for the overall rating as 4 (Selection) :
        1 (Diagnosis) : 4 (Exposure) : 1 (Confounding) : 0 (Conflict of interest). The
        Committee considered Conflict of Interest to be important, but it could be poorly
        assessed due to missing information. The information that was used for scoring
        were the financial interests declared in the publications. In some cases, earlier
        publications about the same study revealed interests that were not declared later.
        This may be correct, as at the time of the later publication the funding may have
        ceased, but some level of conflict of interest could still be suspected. The
        Evaluation of study quality                                                        63
</pre>

====================================================================== Einde pagina 64 =================================================================

<br><br>====================================================================== Pagina 65 ======================================================================

<pre>               Committee felt that the impact of such financial ties can be widely different and
               there was insufficient information to take this into account. Also, non-financial
               interests and professional commitment to an opinion about an association
               between mobile phone use and brain cancer could also influence the presentation
               of the results. Again this could not be measured. Therefore the score for Conflict
               of Interest was not taken into account in the overall score but is only given for
               information.
5.1            Results of the evaluation of study quality
               The final rating is given in the last column of Table 5.1 as a number between 0
               and 10. To facilitate distinguishing higher from lower rated studies, they are
               colour coded, but without any particular meaning of the cut-off values. Ratings
               of 7.0 and higher are marked green, ratings of between 3.0 and 7.0 are marked
               yellow, and ratings lower than 3.0 are marked red.
Table 5.1 Results for the evaluation of selected cohort, case-control and case-case studies, grouped by tumour type.
                                      Domains:                 Selection Misclas- Mis-           Con-       Conflict Over-
                                                               bias        sification classifi- founding of interest all
                                                                           of outcome cation of                      score
                                                                                       exposure                      (0-10)
Reference                  Design Tumour                       % of maximum obtainable score
Dreyer et al.(1999)47      Cohort Brain cancer                 100.0          0.0      59.4      75.0        60.0    7.1
Baldi et al. (2011)107     Ca-co      Brain cancer              64.7       100.0       33.3      75.0       100.0    5.7
Aydin et al. (2011)108     Ca-co      Brain tumours children 76.5          100.0       66.7      75.0       100.0    7.5
Frei et al. (2011)52       Cohort Glioma, meningioma 100.0                 100.0       53.6      75.0         0.0    7.9
Muscat et al. (2000)96     Ca-co      Glioma                      0.0      100.0       53.6      75.0         0.0    3.9
Inskip et al. (2001)99     Ca-co      Glioma, meningioma,         35.3     100.0       46.4      75.0       100.0    5.0
                                      acoustic neuroma
Auvinen et al. (2002)100 Ca-co        Glioma, meningioma, 100.0            100.0       66.7      75.0        60.0    8.4
                                      parotid gland tumour
Gousias et al. (2009)103   Ca-co      Glioma                      0.0      100.0       27.5        0.0      100.0    2.1
Spinelli et al. (2010)105  Ca-co      Glioma                    14.7       100.0       27.5        0.0      100.0    2.7
INTERPHONE                 Ca-co      Glioma, meningioma        52.9       100.0       68.1      75.0        60.0    6.6
study group (2010)93
Hardell et al. (2011)68    Ca-co      Glioma, meningioma        76.5       100.0       63.8      75.0       100.0    7.4
Ali Kahn et al. (2003)109 Ca-ca       Glioma                   100.0       100.0       26.1        0.0      100.0    6.0
Hardell et al. (2009)66    Ca-co      Acoustic neuroma          76.5       100.0       63.8      75.0       100.0    7.4
Schüz et al. (2011)51      Cohort Acoustic neuroma             100.0       100.0       53.6      75.0         0.0    7.9
Muscat et al. (2002)101    Ca-co      Acoustic neuroma            0.0      100.0       42.0      75.0         0.0    3.4
INTERPHONE                 Ca-co      Acoustic neuroma          64.7       100.0       68.1      75.0        60.0    7.1
study group (2011)94
Salahaldin & Bener         Ca-ca      Acoustic neuroma         100.0       100.0        4.3        0.0      100.0    5.2
(2006)110
64             Mobile phones and cancer
</pre>

====================================================================== Einde pagina 65 =================================================================

<br><br>====================================================================== Pagina 66 ======================================================================

<pre>Sato et al. (2010)112       Ca-ca     Acoustic neuroma     100.0    100.0      63.8     75.0      100.0  8.3
Warren et al. (2003)102     Ca-co     Intratemporal facial 0.0      0.0        34.0       0.0     100.0  2.0
                                      nerve tumours
Hardell et al. (2004)69     Ca-co     Parotid gland tumour 52.9     100.0      63.8     75.0      100.0  6.4
Lönn et al. (2006)82        Ca-co     Parotid gland tumour 76.5        0.0     68.1     75.0       60.0  6.5
Sadetzki et al. (2008)85    Ca-co     Parotid gland tumour 47.1     100.0      68.1     75.0       60.0  6.4
Duan et al. (2011)106       Ca-co     Parotid gland tumour  0.0     100.0      63.8     75.0      100.0  4.3
Söderqvist et al. (2012)72 Ca-co      Parotid gland tumour 76.5     100.0      59.5     75.0      100.0  7.2
Takebayashi et al. (2008)88 Ca-co     Pituitary adenoma    23.5     100.0      71.0     75.0      100.0  5.5
Schoemaker et al. (2009)89 Ca-co      Pituitary tumours    64.7     100.0      68.1     75.0       60.0  7.1
Stang et al. (2001)98       Ca-co     Uveal melanoma       64.7     100.0      24.6       0.0     100.0  4.6
Stang et al. (2009)104      Ca-co     Uveal melanoma       76.5       50.0     79.7     75.0       60.0  7.5
De Roos et al.(2001)97      Ca-co     Neuroblastoma         0.0        0.0     18.8       0.0     100.0  0.8
Ca-co: case-control, Ca-ca: case-case
               Selection bias
               Selection biases are distortions that result from procedures used to select subjects
               and from factors that influence study participation. The common element of such
               biases is that the relation between exposure and disease is different for those who
               participate and for all those who should have been theoretically eligible for study,
               including those who did not participate.136
               Maximum scores in the selection bias domain are inherently generated for the
               cohort studies.
               A striking feature of the case-control studies in this domain is the generally high
               response rates of the Hardell studies. These are all population sampled for both
               cases and controls. Due to the early uptake of mobile phones in Sweden, the later
               studies have relatively high numbers of respondents with prolonged phone use.
               The high response rates make the studies adequately representative of population
               patterns for the exposures measured. However, some investigators have
               expressed concern that these high levels of response are virtually impossible to
               attain.137,138 In their response to a Letter to the Editor on this matter, Hardell et al.
               claim that they have obtained (very) high response rates in a number of earlier
               (non-mobile telephone) studies as well; these range from 90-100% for cases and
               83-100% for controls.137 Case-control studies on other topics performed in
               Sweden in the 1990’s using the same methods as Hardell et al., mailed
               questionnaires and telephone follow-up, obtained response rates of between 59%
               and 83% for cases and between 53% and 82% for population controls.139-142 In
               view of this, the response rates in the Hardell studies are rather high.
               Evaluation of study quality                                                                   65
</pre>

====================================================================== Einde pagina 66 =================================================================

<br><br>====================================================================== Pagina 67 ======================================================================

<pre>       However, in the calculation of the response rates for cases in their
   publications, Hardell et al. incorrectly did not include deceased cases and cases
   whose participation was denied by their physician, as was done in the
   INTERPHONE studies. In order to allow a better comparison with other studies,
   the Committee recalculated the Hardell response rates to include these cases as
   well. This led to the lower response rates shown in Table 4.3, ranging from 59-
   72% for cases and 79-90% for controls. The corrected response rates for cases
   are more in accordance with those from other Swedish studies mentioned above,
   but those for the controls are still high. This will be discussed later.
       The INTERPHONE studies score lower in this domain as they have rather
   low response rates and may thus suffer more from selection bias than the Hardell
   studies. For the INTERPHONE study with pooled glioma and meningioma data
   the overall response rates were 64% for glioma cases, 78% for meningioma cases
   and 53% for controls.93 In 6 of the 14 individual country studies control
   participation was less than 50% (on-line Annex in 93). These poor response rates
   for controls may have introduced selection bias, as only the more motivated
   subjects with potentially different mobile phone use characteristics may have
   participated. Indeed, a non-response analysis showed that both cases and controls
   that refused to participate in the main study in general had a lower use of mobile
   phones than participants.143 The underestimation of the risk due to selection bias
   in the INTERPHONE study is raised in the Letters to the Editor as an issue in the
   interpretation of differences between the findings of Hardell and INTERPHONE
   (see Annex E). According to the authors of the INTERPHONE study, non-
   participation bias may have led to a reduction in the odds ratios for regular use of
   5-15%.
       Comparison of the Hardell response rates with those from INTERPHONE
   should only be done for the Hardell study that pooled information of living and
   deceased cases, since that study is used in the analyses in this report. The
   response rate Hardell reported for cases was 85%, which does not fall into the
   range of the recalculated response rates of 59-72% indicated above. This might
   be the case because Hardell did not include those cases for which the physician
   refused participation as non-responders. The response rate for controls was 84%,
   which does fall in the range of recalculated values (79-90%). The response rates
   of INTERPHONE are markedly lower: 64-78% for cases and 53% for controls.
   The Hardell response rates might reflect a better representation of the population
   in question than in the INTERPHONE studies, and consequently a lower
   likelihood of selection bias, but this is challenged by the difference between the
   reported and recalculated response rates and by the fact that the response rates of
   the controls in the Hardell studies are much higher than those in other studies.
66 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 67 =================================================================

<br><br>====================================================================== Pagina 68 ======================================================================

<pre>This has no implications for the scoring in the present (methodological)
evaluation, but it does have consequences for the overall appraisal, as will be
discussed in later chapters.
    The strengths of the INTERPHONE studies are that large numbers of
respondents could be achieved by pooling of the results, and that the authors
went at great length to study various types of bias involved in these
studies.46,74,143-150 It should be mentioned, however, that there is also a limitation,
since the bulk of the data of the pooled INTERPHONE studies is coming from a
limited number of countries: for glioma 46.6% of cases come from the UK (with
two separate studies in the north and south of the country), Australia and
Germany, for meningioma 51.8% of cases come from these three countries,
while for acoustic neuroma 44% of cases come from the UK, Australia and
France. In the UK and Australia poor response rates were obtained. In Australia,
according to information obtained from the investigators, cases were ascertained
by hospital sampling and double checked in the cancer registry. Control selection
was from the electoral roll and in contrast to the case selection, this would not
fully include migrants, thus introducing potential selection bias. Altogether this
increases the likelihood of selection bias in the overall INTERPHONE results.
    Another limitation of INTERPHONE is that in many countries sampling of
controls from the population is difficult. As a result, the pooled database contains
a mix of respondents obtained by population sampling, hospital sampling or
other sampling methods (see Table 4.5). As such, it may not be fully
representative of the target population and therefore potentially biased. It is hard
to tell whether this would result in under- or overreporting of the risk. However,
the issue needs to be weighed with the relative contribution of certain countries
to the pooled results.
    Although the INTERPHONE protocol74 states that a complete population
sample for cases and controls was aimed for, case selection in some countries has
been incomplete: e.g. in Germany not all hospitals in the regions were covered.
Therefore further information was obtained from the authors of the German
study. They stated that all cases were referred to the tertiary hospitals for further
diagnostic procedures and consultation, even when primarily treated in local
hospitals. As a result all cases were seen, even when not admitted, and exclusion
of some local hospitals should not have caused selection bias.
For the case-case studies the domain of selection bias is inherently generating
maximum scores, since only cases are involved. For some of the studies it is
unclear if the included cases do represent a full or at least random selection of the
cases available in the target population. In the case-case study by Sato et al.
Evaluation of study quality                                                             67
</pre>

====================================================================== Einde pagina 68 =================================================================

<br><br>====================================================================== Pagina 69 ======================================================================

<pre>   (2010)112 the response rate is less than 100%, but the recruiting process seems to
   have been consistent and transparent, even though some hospitals have been
   missed. Therefore, there could still have been selection bias in this group, even
   though this does not show in the scoring.
   Misclassification of outcome
   In the domain of misclassification of outcome no problems were seen for most of
   the studies. The outcome is always reasonably well to very well defined and uses
   histology and location information. For those studies that separated the types of
   tumours, at least histological information had to be available to do so; this was
   missing in some publications.
   Misclassification of exposure
   In the domain of misclassification of exposure the items of interest are the bias
   resulting from the method of collecting the information on mobile phone use and
   the validity of the reported information.
   The most important cohort study, the Danish cohort study, used an objective but
   crude measure of exposure: time since first subscription as determined from
   provider records. The study compared a group of subjects that started a private
   subscription between 1 January 1982 and 31 December 1995 with the rest of the
   Danish population. This includes people not using a mobile phone, people that
   started owning and using a mobile phone after 31 December 1995, and people
   not owning a phone but having used one owned by others before and/or after 31
   December 1995. In the formation of the cohort, business contracts were
   excluded, because it was not possible to relate these to individual users. This
   means that a number of potentially heavy users was not included in the ‘exposed’
   group but in the ‘unexposed’ group (although some business users may have had
   a private subscription also). The advantage of considering the time since first
   subscription is that this is objective information. The disadvantage is that, since
   the subjects were not interviewed, no information is available on phone type or
   (intensity of) use, and hence actual exposure could not be assessed. In the first
   publication from 200148, which studied the cohort up to the end of 1996, the
   ‘unexposed’ population was probably mostly non-exposed, except for the
   relatively small group of business users. However, in the publications with
   follow-up up to 2002-200750-52, the originally defined group of early users is still
   being compared to the general population. In these later studies this ‘unexposed’
68 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 69 =================================================================

<br><br>====================================================================== Pagina 70 ======================================================================

<pre>group was clearly not unexposed anymore, as an estimated 100% of the Danish
population currently uses mobile phones and many have been doing so for over
10 years, since subjects included in the control group might have started their
subscription as early as 1996 (see also Figure 2 in 4.5).151 Since misclassification
in this cohort study is limited to the ‘unexposed’ group, it can be demonstrated
that the effect of misclassification on the calculated risk will only be minimal.152
In the Hardell case-control studies, the core information has been gathered with
observer blinding by using a mailed-in paper questionnaire. According to the
principal investigator, this was a larger questionnaire on environmental factors
that contained several questions on mobile telephone use. The Committee
received only those questions, that are not very detailed, in particular with regard
to assessment of mobile phone use. According to the principal investigator, in all
cases additional information was gathered by telephone interviews using a
protocol, but this is not available and therefore the validity of the data obtained
cannot be checked. This procedure may lead to misclassification bias. Although
the interviewers in the Hardell studies had no prior knowledge of the disease
status of the respondents, it is likely that the disease status was revealed during
the interview. This may have led to observer bias and, hence, differential
misclassification with potential overestimation of the risks. As a consequence,
the quality of the exposure assessment in the Hardell studies is difficult to judge.
    The INTERPHONE studies have the most detailed exposure assessment and
have spent much effort in validation of the questionnaire. The assessment of the
use of mobile phones in the INTERPHONE studies was done in person, showing
pictures of mobile phone models. This makes recall of the types of phone used,
and thus of the exposure, more accurate than when phone types are asked for by
mail or telephone interview, as in the Hardell studies. Since in the
INTERPHONE studies the data were mostly collected by personal (computer-
assisted) face-to-face interview, there may have been observer bias, as the cases
will have been notably ill. The protocol states that the observers were carefully
trained to reduce this effect, but still it is possible that also in the INTERPHONE
studies differential misclassification may have occurred.
    The INTERPHONE researchers performed several validation studies, such as
a separate study on recall bias using healthy volunteers.146 They used software-
modified phones that logged the time and duration of incoming and outgoing
calls. These data were compared with the data recalled by the subjects 6-12
months after the data logging period. They observed that the random error in
recall was larger for the duration of calls than for the number of calls. In another
study they compared call records of the operators with phone use reported by the
Evaluation of study quality                                                          69
</pre>

====================================================================== Einde pagina 70 =================================================================

<br><br>====================================================================== Pagina 71 ======================================================================

<pre>   subjects. They observed no difference between cases and controls, except that
   cases over-reported phone use 3-5 years back (but this was based on few cases
   with long-term data).46 It is likely that this effect will be stronger with reporting
   of phone use longer back. There was no operator information on phone use
   longer than 5 years back. There are no publications describing the validation of
   mobile phone use in the Hardell studies, but the INTERPHONE validation
   studies show that in case-control studies in general recall bias is potentially an
   important source of error. The possible differential long-term recall bias may
   have resulted in overestimation of the actual risks. This may then counteract the
   underestimation due to selection bias, as discussed above.
       As a result of all these considerations, the scores of the Hardell and
   INTERPHONE studies in the domain of misclassification are approximately
   similar.
   For the case-case studies the exposure assessment is generally very poorly
   described, resulting in low scores in this domain. An exception is the publication
   by Sato et al. (2010)112.
   Confounding
   A risk factor for brain tumours is a confounder when the exposure to that factor
   is associated with the exposure of interest, in this case exposure resulting from
   the use of mobile or cordless phones.
       The publications on the characteristics of mobile and cordless phone use
   among children and adolescents in the Nordic countries of Europe indicated a
   clear association between age and use and between gender and use.153-156
   However, the equivalent publications on adults did not provide adequate
   information.157,158 No information was found on the extent of use of cordless
   handsets for landline telephones. The age and sex distributions of cordless phone
   users are not necessarily similar to that of mobile phone users. It is clear,
   however, that there is an association between age and sex and mobile phone use,
   so they are to be considered confounders in the study of the association between
   mobile phone use and brain tumours. For cordless phones this is less clear, but it
   is assumed to be the case.
       In the domain of confounding all publications addressed these main
   confounders age and sex. There may be some confounding left, as little is known
   about the risk factors for brain tumours. The scoring for this domain does not
   distinguish between the different types of studies, since they all use conventional
   techniques for correction and all account for the standard confounders.
70 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 71 =================================================================

<br><br>====================================================================== Pagina 72 ======================================================================

<pre>5.2 Conclusion
    The evaluation of the methodological quality of the studies did not result in
    major differences between the main studies. The usefulness of several studies
    was very limited because of their short follow-up, but that does not necessarily
    mean that they were of low quality. In effect, however, only the case-control
    studies by Hardell et al. and INTERPHONE, and the Danish cohort study are
    useful for the current analysis that is aimed at an evaluation of long-term effects.
        The domains in the quality score that can best differentiate between the
    studies are those related to selection of the subjects and exposure assessment. In
    both domains and overall the Hardell studies had similar scores as the
    INTERPHONE studies. The Danish cohort scored the maximum for selection of
    subjects, but lower than the case-control studies on exposure assessment, while
    the overall score was similar to that of the Hardell and INTERPHONE studies.
    So on the basis of this scoring system for methodological quality there is no
    reason to give one type of studies more weight than the other.
    Evaluation of study quality                                                          71
</pre>

====================================================================== Einde pagina 72 =================================================================

<br><br>====================================================================== Pagina 73 ======================================================================

<pre>72 Mobile phones and cancer</pre>

====================================================================== Einde pagina 73 =================================================================

<br><br>====================================================================== Pagina 74 ======================================================================

<pre>Chapter 6
        Results: analysis of
        the data by disease
        In the different studies a number of exposure characteristics has been used in the
        analyses. Given the availability of the exposure characteristics across
        publications it was decided to focus on (a) the duration of mobile phone use in
        years and (b) the cumulative exposure from mobile phone calls in hours over the
        respondents’ lifetime in the analysis of effects. However, data on the estimated
        number of phone calls over lifetime will also be presented, even though the
        Committee considers the number of calls to be less relevant than the actual total
        call duration. On the other hand, total call duration carries a higher risk of
        overestimation than number of calls.46,145,146,159
6.1     Issues to be considered
        Lateralisation
        An important aspect considered in the studies was the so-called lateralisation:
        was the telephone predominantly used at the side of the head where the tumour is
        located (ipsilateral), or not (contralateral; this generally includes both use on the
        opposite side of the head from the location of the tumour and use on either side).
        These data definitely suffer from recall bias, as has been demonstrated by the
        INTERPHONE investigators.160 Cases tended to indicate more often that they
        used the phone on the side of the head where the tumour is located than they
        actually did. This was not the case with the controls, since they were allocated
        Results: analysis of the data by disease                                              73
</pre>

====================================================================== Einde pagina 74 =================================================================

<br><br>====================================================================== Pagina 75 ======================================================================

<pre>   the same hemisphere as their matching case after they had been interviewed. This
   means that there is differential recall bias, i.e. it is larger in the cases than in the
   controls, which will lead to overestimation of the risk.160 A clear indication for
   an overestimation of the ipsilateral risk is a concomitant decreased contralateral
   risk that seems to indicate a protective effect. This has been observed and
   discussed extensively in the INTERPHONE study.93
   Latency time
   Another important point to be considered in the study of slow growing tumours
   such as those considered in this report, is the latency time, i.e. the time between
   induction of the tumour and clinical manifestation. Hardly any information is
   available, however, on latency periods for these tumours. What is known comes
   from studies on secondary tumours after radiotherapy. However, since ionizing
   radiation is a known carcinogen, it is highly uncertain whether this information is
   in any way representative for the situation with exposure to RF EMF, which is at
   most, according to IARC, a possible carcinogen and for which, in contrast to
   ionising radiation, a carcinogenic mechanism of action is not known (see below).
   The Committee presents the data anyway, since it is all we know and it might be
   considered a worst-case situation.
       Two reviews present information on latency of gliomas after X-rays.161,162
   From these data, latency periods of 10.6 ± 9.2 years for cases up to an age of 19
   years at exposure, and of 11.6 ± 6.5 years for older cases can be derived. No clear
   relation between total X-ray dose and latency time was observed.
       For meningioma, a mean latency time of 20.8 years was calculated for
   patients aged ≤12 years and 21.8 year for >12 year-olds, with an overall range of
   1-63 years.163 Only 1.4% occurred within 5 years of treatment. Shorter latency
   periods were observed with increasing X-ray dose.
       A review on radiation-induced acoustic neuromas reported a latency time of
   38.3 ± 10.1 years, with an increasing risk with X-ray dose.164 All cases were <16
   years old at the time of treatment.
       So for all three major tumour types considered in this report, the latency time
   after X-ray exposure is very long, but with a considerable spread. The latency
   time is specific for a disease / exposure combination. If there would be a causal
   relation between RF EMF exposure and these tumours, the Committee considers
   it possible that the latency time will be longer than that after X-ray exposure.
   This would mean that a follow-up time of 10 years would not be enough to
   measure any increase in tumour incidence and, vice versa, that any increased
74 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 75 =================================================================

<br><br>====================================================================== Pagina 76 ======================================================================

<pre>    incidence observed with short follow-up times is not realistic and might indicate
    flaws in the study.
    Mechanism of action
    RF EMF such as generated by mobile phones do not act upon biological material
    in the way ionising radiation does. It is not known, but considered very unlikely,
    that RF EMF can cause direct damage to DNA that may lead to disruption of
    biological processes and the development of cancer (unless through thermal
    effects, but these do not occur when using a mobile phone). Animal studies also
    do not indicate that RF EMF exposure might influence the development of
    cancer that has been induced by another agent. This will be discussed in a
    separate report.
6.2 Brain tumours (not further specified)
    Cohort and case-control studies
    This includes the results from three publications.47,107,108 Aydin et al. (2011)108
    conducted a case-control study in children, the other two are studies on brain
    cancer in adults. The detailed results are presented in Tables H1, H2 and H3 in
    Annex H. Neither in the cohort nor in the case-control studies significantly
    increased risks were found. However, the cohort study from the USA47 only
    looked at duration of use of more or less than 3 years and had only one case in
    each category (Table H1). The case-control study of Aydin et al. (2011)108 on
    childhood brain cancers (which is the first report on the MOBI-KIDS study) had
    a maximum follow-up time of 5 years and maximum cumulative exposures of
    144 hours. The usefulness of these studies in this analysis is therefore limited.
    The study of Baldi et al. (2011)107 only made a distinction between use or no use
    of a mobile phone and did not register duration of use, or number or duration of
    calls. It can therefore not be used in the present analysis.
        The childhood brain cancer data108 allowed a laterality analysis (Table H3).
    Several increased odds ratios (ORs) were found, but with an inconsistent pattern.
    For time since first use an increased OR was found only in the middle category
    for contralateral use. For cumulative call time an increased OR was found only
    for the middle category of ipsilateral use and for the two highest categories for
    use on the contralateral side. For cumulative number of calls increased ORs were
    found in the highest category for ipsilateral use and in the two highest categories
    for contralateral use. Decreased ORs were found for the highest categories of
    Results: analysis of the data by disease                                            75
</pre>

====================================================================== Einde pagina 76 =================================================================

<br><br>====================================================================== Pagina 77 ======================================================================

<pre>    cumulative call time and cumulative number of calls for tumours with a central
    or unknown location. In all instances the number of cases and controls is very
    limited and this might be an explanation for not observing clear exposure-effect
    relationships, that would be expected in case there would be a causal relation.
    Ecological studies
    Kohler et al. (2011)131 investigated brain and other tumour incidences in the
    USA for the period 1980-2007, but did not link this to mobile phone use. They
    concluded that both malignant and non-malignant brain tumours demonstrate
    differing patterns of occurrence by sex, age, and race, and exhibit considerable
    biologic diversity.
        De Vocht et al. (2011)130 reported on brain tumour trends in England from
    1998-2007. They observed overall no statistically significant increases, but
    identified small but systematic increases in temporal lobe tumours in both men
    and women and of frontal lobe tumours in men, and decreases in tumours of the
    cerebrum, parietal lobe and cerebellum in men. Trends indicate a rapid increase
    in mobile phone use between 1998 and 2003, but this study cannot draw any
    conclusions when latency periods of 10 or more years are assumed.
        Röösli et al. (2007)121 analysed brain tumour incidence in Switzerland over
    the period 1969-2002. However, since mobile phone use was shown to be rapidly
    increasing in this time period, this study does not address any reasonable latency
    period.
        Cook et al. (2003)115 described trends for brain malignancies in New Zealand
    over the period 1986 to 1998 in relation to the prevalence of cell phones.
    However this prevalence was only slightly higher than 12.5% by the end of the
    observation period, so was still very low compared to later time periods. This
    study also does not address any reasonable latency period.
6.3 Glioma
    Nine different studies have been identified on gliomas, tumours of the brain
    nervous tissue.52,68,93,96,99,100,103,105,109 Figures 3-6 present the main outcomes of
    the studies. The more complete and detailed results are presented in Annex H;
    the data for duration of use are in Table H4, those for cumulative use in Table H5
    and the lateralisation data are in Table H6. All odds ratios presented in this
    report, both in the figures and in the tables, are the ones that are corrected for
    confounders, i.e. the adjusted odds ratios.
76  Mobile phones and cancer
</pre>

====================================================================== Einde pagina 77 =================================================================

<br><br>====================================================================== Pagina 78 ======================================================================

<pre>Cohort studies
The latest publication on the Danish cohort study by Frei et al. (2011)52 presented
overall results for duration of use. However, as has been mentioned earlier, the
associations reported in the later studies of this cohort are difficult to compare to
the results of the other studies. The cohort study compares a group of subjects
that started a private mobile phone contract before 1996 with the rest of the
Danish population, that includes people using phones through a business contract
and people that started a contract as of 1996. This is different from the case-
control and case-case studies that compared duration of mobile phone use with
no use or that compared start of regular use. Nevertheless, the cohort data are
included in Figure 3 and given in Table H4. They do not show any increased
risks for any of the durations of use, for either males of females.
Case-control and case-case studies
The case-control studies are more readily comparable, but some points need
attention. Hardell et al. (2011)68 presented glioma as such, but also made a
distinction between astrocytomas and ‘other malignant brain tumours’, which
included mixed gliomas and oligodendromas. They found the strongest
association for astrocytomas of the highest grade, i.e. the most malignant type.
However, as the INTERPHONE study (2010)93 does not present any
subdivisions of glioma, in the figures only a comparison with the overall glioma
results of the Hardell studies is made, while their astrocytoma data are presented
in the tables for completeness. It should also be noted that the age range of the
Hardell studies is wider (20-80 years) than that of the INTERPHONE studies
(30-59 years), which for strongly age-related illnesses such as glioma might give
different effects; also recall problems might be larger in the older age groups. In
order to allow a better comparison, Hardell et al. (2011)165 partially reanalyzed
their data to include only the 30-59 age categories. A third issue is that Hardell et
al. make a distinction between the use of mobile phones (such as GSMs) and
cordless phones (the wireless phones for indoor use, such as DECT). The
reanalysis they performed for the limited age range was done only for mobile
phone users, and not separately for cordless phone users.
    The Hardell data of the full age range for time since first use, show an
increased relative risk associated with mobile phone use for all gliomas in the
highest (>10 years) category (Figure 3, Table H4) and in the middle and highest
categories (>5-10 and >10 years) for astrocytoma (Table H4). For cordless phone
use increased relative risks were found only in the middle categories (Figure 3,
Results: analysis of the data by disease                                              77
</pre>

====================================================================== Einde pagina 78 =================================================================

<br><br>====================================================================== Pagina 79 ======================================================================

<pre>   Table H4). The recalculation for the limited age range was done only for the
   highest category (≥10 years) and resulted in a relative risk that was lower than
   for the full age range, but still increased (Figure 3, Table H4). It is puzzling that
   the OR of 2.26 for the full age range given in the reanalysis paper165 (95%
   Confidence Interval 1.60-3.39) differs from that in the pooled analysis paper68:
   2.6 (CI 1.7-4.1), while also the numbers of cases and controls differ: 88 / 99 in
   the reanalysis paper and 50 / 42 in the pooled analysis paper. Hardell et al. noted
   in the reanalysis paper that in their original analysis they used >10 years instead
   of the ≥10 years in the reanalysis, but then it would expected that the numbers of
   cases and controls would be lower in the reanalysis, while they are in fact higher.
   This is one of the inconsistencies of the Hardell papers. No increased risks were
   found by INTERPHONE for time since first use (Figure 3, Table H4), but for
   two categories, 1-1.9 and 5-9 years, decreased risks were found.
   Figure 3 Adjusted Odds Ratios with 95% confidence limits for glioma for years since first use of a
   mobile phone.
   • Data from INTERPHONE (2010)93; Hardell et al. (2011)68,165; Inskip et al.(2001)99, Auvinen et al.
      (2002)100 and Frei et al. (2011)52.
   • The midpoints of the ranges for years since first use are used, but for the highest category an
      arbitrary value has been chosen (7 yrs for the >5 yrs category of Inskip, 2.5 yrs for the >2 yrs
      category of Auvinen, 12 yrs for the >10 yrs categories of Interphone and Hardell, 14 yrs for the
      ≥13 yrs category for men and 12 yrs for the ≥10 yrs category for women of Frei). For the lowest
      category similarly arbitrary values were used (0.7 for the <1 yr for Auvinen). If necessary these
      values were slightly adjusted to show overlapping points.
   • The data point ‘Hardell 2011 mobile 30-59 y’ is a subset of ‘Hardell 2011 mobile’.
78 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 79 =================================================================

<br><br>====================================================================== Pagina 80 ======================================================================

<pre>The data of Frei et al. (2011) 52 are results from a cohort study, therefore the point
estimates refer to an Incidence Rate Ratio, not an Odds Ratio. This is a
comparison with the whole population, not with a group of subjects with no or
limited use.
For cumulative call time, Hardell et al. found increased risks for mobile phone
users in all categories (Figure 4, Table H5), and for cordless phone users in the
two highest categories, both for all gliomas and for astrocytomas. In the
recalculated data (Figure 4, Table H5) the risk was lower than in the full age
range data, but the category was also slightly different (>2000 h for the full age
range and ≥1640 h for the limited age range).165 The INTERPHONE data were
divided over 10 categories and an increased risk was only found in the highest
one (≥1640 h) (Figure 4, Table H5). In several lower categories, including the
next-highest one, decreased risks were found, so there is no obvious exposure-
response relationship. When the data from subjects who reported calls of on
average >5 h per day were excluded, because INTERPHONE considered those to
be unrealistically high usage data, the relative risk was not significantly
increased anymore (Table H5).
A validation study showed that the number of calls was slightly underestimated
and, as mentioned earlier, the random error in recall was larger for the duration of
calls than for the number of calls.146 This makes the number of calls potentially a
more reliable endpoint then duration of calls. Nevertheless, ORs for cumulative
number of calls were reported only by INTERPHONE (Table H5). In the two
lowest and fourth highest of ten categories the risk was decreased, in the others it
was not different from unity.
     The analysis of the data in terms of laterality is presented in Figure 5 and
Table H6. The Hardell publication68 from which the data for duration and
cumulative call time were derived did not present information on laterality. The
laterality data were obtained from another publication66 using the same data. For
the full age range, Hardell et al. observed increase risks for ipsilateral mobile
phone use already for a time since first use of >1 year (Table H6). Contralateral
use of >10 year also was associated with an increased risk. For ipsilateral
cordless phone use also increased risks were found already for a time since first
use of >1 year. The reanalysis for the limited age range165 was done for mobile
phone use only and resulted in a lower risk than for the full age range, but it was
still significantly increased (Figure 5, Table H6). No increased risks were found
by INTERPHONE for time since first use, but for contralateral use the risk was
decreased for the lowest and next-highest categories.
Results: analysis of the data by disease                                               79
</pre>

====================================================================== Einde pagina 80 =================================================================

<br><br>====================================================================== Pagina 81 ======================================================================

<pre>   Figure 4 Adjusted Odds Ratios with 95% confidence intervals for glioma for cumulative call time.
   • Data from INTERPHONE (2010)93; Hardell et al. (2011)68,165; Inskip et al. (2001)99.
   • The midpoints of the ranges for years since first use are used, but for the highest category an
      arbitrary value has been chosen (600 h for the >500 h category of Inskip, 2000 h for the ≥1640 h
      category of INTERPHONE and the reanalysis of the Hardell data and 2500 for the >2000 h
      category of Hardell). For the lowest category similarly arbitrary values were chosen (12 h for the
      <13 hr of Inskip and 4 h for the <5 h category of INTERPHONE). If necessary the values were
      slightly adjusted to show overlapping points.
   • The data point ‘Hardell 2011 mobile 30-59 y’ is a subset of ‘Hardell 2011 mobile’.
   For cumulative call time data are not available for the full age range in the
   Hardell et al. studies, but they are presented in the 30-59 y age range reanalysis
   for a cumulative call time similar to the highest category used by INTER-
   PHONE, ≥1640 h.165 Hardell found that for ipsilateral use the risk was increased,
   but not for contralateral use (Figure 6, Table H6). INTERPHONE found an
   increased risk for ipsilateral phone use in the highest of five categories (≥1640
   h), but a decreased risk in the one but lowest category (Figure 6, Table F6). For
   contralateral use the risk was decreased in the lowest, middle and next highest
   categories.
        Only INTERPHONE also reported data for cumulative number of calls. No
   increased risks were found, but decreased risks for the middle of five categories
   for ipsilateral use, and for the 2nd and 3rd category for contralateral use (Table
   H6). The effect of exclusion of cases and controls with unrealistically long daily
   call times on the risk estimate was not reported for the laterality data.
80 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 81 =================================================================

<br><br>====================================================================== Pagina 82 ======================================================================

<pre>Figure 5 Adjusted Odds Ratios with 95% confidence intervals for ipsilateral and
contralateral glioma for years since first use.
• Data from INTERPHONE (2010)93; Hardell et al. (2011)66,165.
• The midpoints of the ranges for years since first use are used, but arbitrary values of 12
   yrs for the >10 yrs categories of INTERPHONE and Hardell. The values were slightly
   adjusted to show overlapping points. Error bars were cut at an OR of 6.0.
• The data points ‘Hardell ipsi mobile 30-59 y’ and ‘Hardell contra mobile 30-59 y’ are
   subsets of ‘Hardell ipsi mobile’ and ‘Hardell contra mobile’, respectively.
Figure 6 Adjusted Odds Ratios with 95% confidence intervals for ipsi- and
contralateral glioma for cumulative call time.
• Data from INTERPHONE (2010)93; Hardell et al. (2011)165.
• The midpoints of the ranges for years since first use are used, but for the highest
   category an arbitrary value has been chosen (2,000 h for the >1,640 h category). For the
   lowest category similarly arbitrary values were chosen (4 h for the <5 h category). If
   necessary the values were slightly adjusted to show overlapping points.
Results: analysis of the data by disease                                                     81
</pre>

====================================================================== Einde pagina 82 =================================================================

<br><br>====================================================================== Pagina 83 ======================================================================

<pre>   Two groups of INTERPHONE researchers also independently assessed the
   relation between calculated energy uptake in brain tumours (and corresponding
   brain tissue in controls) and various endpoints. Cardis et al. (2011)114 calculated
   for a five-country subset of the INTERPHONE data (Australia, Canada, France,
   Israel and New Zealand) the total cumulated specific energy (in joules per kg,
   J/kg) in the tumours or brain tissue. The case-control pairs were selected on the
   basis of the estimation of the tumour centre by either a neuroradiologist or a
   computer algorithm. For comparison with the entire INTERPHONE dataset,
   risks were first calculated for cumulative call times. The risk for the highest of
   five categories (≥735 h) was not increased (Table H5), while it was for the
   highest of ten categories (≥1640 h) in the entire INTERPHONE study, as
   mentioned above. A decreased risk was calculated for the middle of five
   categories. When assessed for total cumulative energy, for the entire dataset no
   increased risks were found, but for the subgroup of use ≥7 years in the past an
   increased risk was found in the highest category (≥3124 J/kg cumulated energy)
   (Table H5).
   Cardis et al. (2011) also calculated risks in a case-case subset, where they
   compared cases with the centre of the tumour within the most exposed area of the
   brain with cases with a tumour outside that area. An increased risk was found for
   time since first use of >10 years (Table H4), but no increase risk for cumulative
   call time (highest category ≥1147 h) (Table H5). These data are based on low
   numbers of cases, however.
        The second INTERPHONE substudy was published by Larjavaara et al.
   (2011).113 They used another subset of the INTERPHONE data (from 7
   countries: Denmark, Finland, Germany, Italy, Norway, Sweden, and Southeast
   England) to calculate for cases the distance of the tumour midpoint to the source
   of exposure, where it was assumed that the mobile phone was always kept at the
   side of the head where the tumour was located, thus avoiding recall bias (but
   most likely introducing misclassification errors). A second approach was what
   they called a ‘case-specular’ analysis, where the actual cases were compared
   with hypothetical or ‘specular’ cases. The specular locations were constructed by
   mirroring the location of the tumour to the opposite side of the brain. These
   hypothetical cases thus represented the exposure that would have been incurred
   if the tumour had been located in another location. This counterfactual ‘control’
   was contrasted in the analysis with the actual case. Neither approach resulted in
   higher or lower percentages of tumours in locations receiving the highest
   exposure, in relation to time since first use, total duration of use and laterality.
   The data for the case-case analysis are given in Tables H4, H5 and H6. Those for
82 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 83 =================================================================

<br><br>====================================================================== Pagina 84 ======================================================================

<pre>the case-specular analysis are not given, since the Committee has doubts about
the usefulness of that analysis.
The case-control studies by Auvinen et al. (2002)100 and Inskip et al. (2001)99
each concern small numbers of cases, as well as relatively short durations of
phone use. Therefore the results from those studies, although presented in the
figures and tables, and although the study by Auvinen had a high score for the
quality evaluation, are not really useful for the current analysis. In general, no
increased risks were found. The only exception is an increased risk for 1-2 years
use of analogue mobile phones by Auvinen et al. (Table H4).
    The case-control study by Spinelli et al. (2010)105 also included only a small
number of cases. They presented the data as hour-years, based on the number of
monthly hours of call time available in the subscriptions (so not actual call time)
and years of subscription held. These data, that do not show increased risks, are
not readily comparable to the duration and cumulative call time data of the other
publications and are therefore not included in the figures, but only given in Table
H5. They cannot be used in the current analysis. The study by Muscat et al.
(2000)96 on unspecified brain tumours also contained data for specific tumour
types, including gliomas. However, these data were pooled for all follow-up
times and are therefore incomparable to the INTERPHONE and Hardell data.
These data are also not shown in the figures, but are given in Table F4. Gousias
et al. (2009)103 performed a case-control study in Greece and determined minute-
years of mobile phone use, but they only report on overall mobile phone use
without providing any number on minute-years. Data from this study cannot be
included in the figures and are only given in Table H5; they also cannot be used
in the current analysis.
Ali Kahn et al. (2003)109 investigated in a case-case study on glioma patients the
relation between the location of the tumour and handedness. They hypothesized
that handedness would be indicative of the preferred side of use of a mobile
phone and only included patients with a unilateral cortical glioma. However, in
later studies handedness has been shown to be a poor indicator of the preferred
side of phone use.84,166 No associations were observed. However, they did not
determine the duration of use or number of calls, therefore these data cannot be
compared to those of the case-controls studies and this study is therefore not
included in the figures and tables.
Results: analysis of the data by disease                                            83
</pre>

====================================================================== Einde pagina 84 =================================================================

<br><br>====================================================================== Pagina 85 ======================================================================

<pre>   Ecological studies
   Several recent studies investigated the incidence of brain tumours over time. The
   studies reporting on unspecified brain tumours have been discussed in 6.2. Here
   the studies reporting gliomas will be discussed.
        The publication by Little et al. (2012)133 is the most recent analysis of data
   from the United States of America (USA) Surveillance, Epidemiology and End
   Results (SEER) programme using population based cancer registries. The SEER
   data is generally held to be informative and trustworthy. The results show that
   both lower grade gliomas as well as those with poorly specified anatomical
   locations have decreased by 2.4-3.0% per year over the period 1997-2008.
   Gliomas with temporal lobe locations and other specified sites have increased by
   approximately 0.75% per year. This study uses mobile phone usage data to
   calculate scenarios of glioma incidence development, given the results of some
   important case-control studies. Assuming a latency time of 10 years and a
   relative risk of 1.5, the underlying glioma rate was expected to increase from
   17.7 per 100 000 people per year to 19.5 in 2008. When the relative risks of
   Hardell et al. (2011)167 were used, all predicted rates were substantially higher
   than the observed rates, i.e. if these risks were true, a clearly increased glioma
   rate should have been visible. However, using the (lower) relative risks from the
   INTERPHONE studies93 the predicted rates were within the observed patterns,
   i.e. the observed patterns are consistent with a small increase in risk, but also
   with no change in risk. It is unclear how much these calculations take the age
   range difference between the Hardell and INTERPHONE studies into account.
        Deltour et al. (2012)135 described the incidence data up to 2008 from the
   Nordic countries, that have the longest mobile phone use. They observed no clear
   increases in glioma incidence overall, but slight increases among the oldest age
   group of 60-79 year olds. They also carried out simulations for men aged 40-59
   and concluded that a relative risk of 1.5 should be visible in the incidence rates
   when a latency time of 10 years for all users is assumed, but when the latency
   time would be 15 years this should be less likely.
        In an editorial on the Frei et al. (2011) study, Ahlbom and Feychting
   (2011)151 presented brain tumour incidence data from the Swedish cancer
   registry over 1970-2009. They concluded that incidence has not changed, not in
   general nor for different age groups and genders. They argued that handheld
   mobile phones were introduced in Sweden in 1987 and that by 2002 87% of 16-
   75 year olds were mobile phone users. Since almost 90% of the population had
   been using mobile phones for at least seven years in 2009, and probably a
84 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 85 =================================================================

<br><br>====================================================================== Pagina 86 ======================================================================

<pre>significant proportion used them for 10 years or even 15 years, they state that any
increased risk should have shown up in the incidence rates by 2009.
The Committee has obtained brain cancer incidence data from the Netherlands
Cancer Registry (NCR) for the period 1989-2010. Incidences are reported in
Figure 7 for gliomas, including astrocytomas, oligodendromas,
oligoastrocytomas and malignant gliomas, for different age categories.
It is clear from this data that there is no increase in gliomas in the Netherlands
during the period of rapid increase in mobile phone use in the age groups that use
them most: 20-29 and 30-59 years. There is an continuous increase in the highest
age group of 60-79 years, but this started already before mobile phones started to
be used. These data correspond to those from other countries, for instance Inskip
et al. (2010)123 for the USA, De Vocht et al. (2011)130 for the UK and Deltour et
al. (2012)135 for Nordic countries.
Figure 7 Glioma incidence in the Netherlands for different age groups. Source: Netherlands Cancer
Registry managed by CCCNL.
Results: analysis of the data by disease                                                          85
</pre>

====================================================================== Einde pagina 86 =================================================================

<br><br>====================================================================== Pagina 87 ======================================================================

<pre>6.4 Meningioma
    Meningiomas are tumours of the meninges, the membranes that separate the
    nervous tissue of the brain from the skull, so they have no neurological origin.
    Five publications report on the association between mobile phone use and the
    risk of meningiomas.52,65,93,99,100
    Cohort studies
    The latest publication on the Danish cohort study52 reported for meningioma
    overall results for duration of use. As mentioned with gliomas, in this cohort
    study the definition of cases and controls is substantially different from that in
    the case-control studies. Nevertheless, they are included in Figure 8 and
    presented in Table H7. They do not show any increased risks for any of the
    durations of use, for either males of females.
    Case-control studies
    The publication of the Hardell group that was used to obtain the odds ratios for
    meningioma (Hansson Mild et al., 200765) does not present the numbers of cases
    and controls for the individual tumour types, but merely gives the total numbers
    for all types of brain tumours (2671 cases and 3723 controls). It could be derived
    from another publication that this study included 916 cases of meningioma.66
    Hansson Mild et al. made a distinction between analogue, digital and cordless
    phones. For time since first use an increased risk was found for analogue phones
    in the highest category of >10 years use, for digital phones no increased risks
    were found and for cordless phones the risk was increased in the middle category
    of >5-10 years use (Figure 8, Table F7). The results from the INTERPHONE
    study show decreased relative risks in the two middle of four categories.93 So no
    exposure-response relationships were observed. The studies by Auvinen et al.
    (2002)100 and Inskip et al. (2001)99 are not really useful for the current analysis
    for reasons mentioned with the gliomas (see 5.1.2). They are, again, presented in
    Figure 8 and Table F7 for completeness only.
    The Hardell group did not publish any data on cumulative call time for
    meningiomas. The INTERPHONE data for both cumulative call time and
    cumulative number of calls do not show any increased risks, but for several
    intermediate categories of both endpoints decreased risks were observed
86  Mobile phones and cancer
</pre>

====================================================================== Einde pagina 87 =================================================================

<br><br>====================================================================== Pagina 88 ======================================================================

<pre>Figure 8 Adjusted Odds Ratios with 95% confidence limits for effects on meningioma for years
since first use of a mobile phone.
• Data from INTERPHONE (2010)93; Hansson Mild et al. (2007)65; Auvinen et al. (2002)100; Inskip
   et al. (2001)99 and Frei et al. (2011)52.
• The midpoints of the ranges for years since first use are used, but for the highest category an
   arbitrary value has been chosen (4 yrs for the >3 yrs category of Inskip, 2.5 yrs for the >2 yrs
   category of Auvinen, 12 yrs for the >10 yrs categories of INTERPHONE and Hardell,14 yrs for
   the ≥13 yrs category for men and 12 yrs for the ≥10 yrs category for women of Frei). If necessary
   the values were slightly adjusted to show overlapping points.
• The data of Frei et al. (2011) 52 are results from a cohort study, therefore the point estimates refer
   to an Incidence Rate Ratio, not an Odds Ratio.
(Figure 9, Table H8). The data from Inskip et al. (2001)99 are presented for
completeness only.
     One publication from the Hardell group also presents analyses for duration of
use as continuous variable.65 No increased risks were observed for the analysis
per 100 h of use, but the analysis per 1 year of use an increased risk was found
for analogue and cordless phones (Table H9).
The publication of the Hardell group from which the data for exposure duration
was derived did not present information on laterality.65 The laterality data were
obtained from another publication on this study using the same data.66 For time
since first use, the Hardell group found an increased risk for ipsilateral use >1
year, but not for >10 years of use of mobile phones (only the latter data are
Results: analysis of the data by disease                                                                 87
</pre>

====================================================================== Einde pagina 88 =================================================================

<br><br>====================================================================== Pagina 89 ======================================================================

<pre>   shown in Figure 10). Also they observed an increased risk associated with >10
   years ipsilateral use of cordless phones (Figure 10, Table H10). In the
   INTERPHONE study no increased risks were observed, but in the two middle of
   four categories a decreased risk was observed for contralateral use (Figure 10,
   Table H10). For cumulative call time and cumulative number of calls decreased
   risks were observed for the 2nd and 4th of five categories for contralateral use
   (Table H10).
   Figure 9 Adjusted Odds Ratios with 95% confidence limits for effects on meningioma for
   cumulative call time.
   • Data from INTERPHONE (2010)97; Inskip et al. (2001)144.
   • The midpoints of the ranges for years since first use are used, but for the highest category an
      arbitrary value has been chosen (1200 h for the >1640 h category of INTERPHONE and 600 h for
      the >500 h category of Inskip). For the lowest category similarly arbitrary values were chosen (12
      h for the <13 hr of Inskip and 4 h for the <5 h category of INTERPHONE). If necessary the values
      were slightly adjusted to show overlapping points.
88 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 89 =================================================================

<br><br>====================================================================== Pagina 90 ======================================================================

<pre>    Figure 10 Adjusted Odds Ratios with 95% confidence intervals for ipsilateral and contralateral
    meningioma for years since first use.
    • Data from INTERPHONE (2010)93; Hardell et al. (2009)66.
    • The midpoints of the ranges for years since first use are used, but for the highest category an
       arbitrary value has been chosen (12 yrs for the >10 yrs categories of INTERPHONE and Hardell).
       If necessary these values were slightly adjusted to show overlapping points. Error bars were cut at
       an OR of 6.0.
    Data on cumulative call time were presented by Hardell et al. only for benign
    tumours, that encompassed both meningiomas and acoustic neuromas.65 No data
    for meningiomas from the Hardell studies can thus be presented for this
    endpoint. The INTERPHONE data show decreased risks in the 2nd and 4th of five
    categories for contralateral use (Table H10).
6.5 Acoustic neuroma
    Acoustic neuromas are tumours that originate from the nerve sheath of the
    acoustic nerve. Six studies on the association between mobile phone use and the
    risk of acoustic neuroma are reported.51,65,94,99,101,112
    Results: analysis of the data by disease                                                               89
</pre>

====================================================================== Einde pagina 90 =================================================================

<br><br>====================================================================== Pagina 91 ======================================================================

<pre>   Cohort studies
   Schüz et al. (2011)51 published results for acoustic neuroma from the Danish
   cohort study. As mentioned with gliomas, in this cohort study the definition of
   cases and controls is substantially different from that in the case-control studies.
   Nevertheless, they are included in Figure 11 and presented in Table H11. In
   contrast to the publications on gliomas and meningiomas, for acoustic neuromas
   only results for men were reported, and only for having a mobile phone
   subscription ≥11 years. No increased risk was found. In women, no acoustic
   neuromas were observed in the study period.
   Case-control and case-case studies
   In the acoustic neuroma data Hardell et al. again made a distinction between
   analogue, digital and cordless phones.65 An increased risk was found for all
   follow-up times for the older types of analogue phones (Figure 11, Table H11).
   For digital phones an increased risk was found only for the shorter follow-up
   times, but not for follow-up times >10 years. For cordless phones an increased
   risk was found only for follow-up times >1- 5 years, but not for >5 years. No
   increased risks were found in the INTERPHONE94 and other case-control
   studies99,101, but in the INTERPHONE study in the 7th and 9th of ten categories
   a decreased risk was observed (Figure 11, Table H11). Sato et al. (2010)
   presented the results of a case-case study of acoustic neuroma.112 They
   calculated risks for two groups: those cases that did not show acoustic neuroma-
   related symptoms at 1 or at 5 years before diagnosis. For each group risks were
   calculated for years since first use before the reference date. No increased risks
   were observed (Figure 11, Table F11).
   The publication of the Hardell group from which the data for duration of phone
   use were derived did not present information on cumulative call time.65 Another
   publication describing the same data was used to obtain data cumulative call
   time.64 Hardell et al. found increased risks associated with analogue, digital and
   cordless phone use for cumulative call times of >1000 h (Figure 12, Table H12).
   No increased risks were found in the INTERPHONE study for cumulative call
   times up to ≥1640 h and by Muscat et al (2002) for call times >60 h. Decreased
   risks were observed in the INTERPHONE study for the 6th and 8th of ten
   categories (Figure 12, Table H12).
90 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 91 =================================================================

<br><br>====================================================================== Pagina 92 ======================================================================

<pre>Figure 11 Adjusted Odds Ratios with 95% confidence intervals for acoustic neuroma for years since
first use.
• Data from INTERPHONE (2011)94; Hansson Mild et al. (2007)65; Inskip et al. (2001)99; Muscat et
    al. (2002)101; Schüz et al. (2011)51.
• The midpoints of the ranges for years since first use are used, but for the highest and lowest
    category an arbitrary value has been chosen (1.4 yrs for the < 0.5 yrs category of Inskip et al.
    (2001)99, 4 yrs for the >3 yrs category of Inskip et al. (2001)99, 4.5 yrs for the <5 yrs category of
    Sato et al. (2010112), 12 yrs for the >10 yrs category of INTERPHONE (2011)94 and for the >10
    yrs category of Hansson Mild et al. (2007)65 and 11 yrs for the > 10 yrs category of Sato et al.
    (2010)112 and 13 yrs for the ≥11 yrs category of Schüz et al (2011)51. If necessary these values
    were slightly adjusted to show overlapping points.
• The data of Schüz et al (2011)51 are results from a cohort study, therefore the point estimates refer
    to an Incidence Rate Ratio, not an Odds Ratio.
The INTERPHONE study group also analyzed the data on the basis of total
number of calls.94 No increased risks were found, but in the 6th and 8th of ten
categories the risk was decreased (Table H12).
      One publication from the Hardell group also presents analyses for duration of
use as continuous variable.65 Only for analogue phones an increased risk was
observed, both for the analyses per 100 h of use and per 1 year of use (Table
H13).
      The Hardell publication65 from which the data for duration of phone use were
derived did not present information on laterality for acoustic neuroma. Another
publication describing the same data was used to obtain these data.66
Results: analysis of the data by disease                                                                  91
</pre>

====================================================================== Einde pagina 92 =================================================================

<br><br>====================================================================== Pagina 93 ======================================================================

<pre>   Figure 12 Adjusted Odds Ratios with 95% confidence intervals for acoustic neuroma for cumulative
   exposure.
   • Data from INTERPHONE (2011)94; Hardell et al. (2006)64; Muscat et al. (2002)101.
   • The midpoints of the ranges for cumulative exposure are used, but for the highest category an
      arbitrary value has been chosen (70 h for the >60 h category of Muscat et al. (2002)101, 1700 h for
      the ≥1640 h category of INTERPHONE (2011)94, and 1200 h for the >1000 h category of Hardell
      et al. (2006)64. If necessary these values were slightly adjusted to show overlapping points. Error
      bars were cut at an OR of 6.0.
   The laterality data for years since first use of the Hardell group show increased
   risks for all mobile phones (analogue and digital) for both >1 year and >10 years
   ipsilateral use, and an increased risk for cordless phones only for >1 year
   ipsilateral use (Figure 13, Table H14). No increased risks were observed for
   contralateral use by Hardell. The INTERPHONE study did not find any
   increased risk for ipsi- or contralateral use when looking at time since first use,
   but for cumulative call time the risk was increased for ipsilateral use and
   exposure ≥1640 h (Figure 14, Table H14). For the next-lower category the risk
   was decreased. For cumulative number of calls, both for ipsilateral and
   contralateral use a decreased risk was found in the middle one of the five
   categories (Table H14). Hardell did not present data on laterality and cumulative
   call time or number of calls.
92 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 93 =================================================================

<br><br>====================================================================== Pagina 94 ======================================================================

<pre>Figure 13 Adjusted Odds Ratios with 95% confidence intervals for ipsilateral and contralateral
acoustic neuroma for years since first use.
• Data from INTERPHONE (2011)94; Hardell et al. (2009)66.
• The midpoints of the ranges for years since first use are used, but for the highest category an
   arbitrary value has been chosen (12 yrs for the >10 yrs categories of INTERPHONE and Hardell).
   If necessary these values were slightly adjusted to show overlapping points. Error bars were cut at
   an OR of 6.0.
Ecological studies
Larjavaara et al. (2011)132 addressed trends in the incidence of acoustic neuroma,
which is a very slow growing tumour. The results indicated a higher incidence
for later birth cohorts in practically all age groups. Patterns in trends were also
analysed, with widely differing results. The timing of some of the increased
incidences observed was thought to be linked to improvements in diagnostics
and registration or to increasing risk, but no relation with mobile phone use was
considered.
Incidence data for the Netherlands are not available, since registration of acoustic
neuromas is not complete.
Results: analysis of the data by disease                                                               93
</pre>

====================================================================== Einde pagina 94 =================================================================

<br><br>====================================================================== Pagina 95 ======================================================================

<pre>    Figure 14 Adjusted Odds Ratios with 95% confidence intervals for ipsilateral and contralateral
    acoustic neuroma for cumulative call time.
    • Data from INTERPHONE (2011)94.
    • The midpoints of the ranges for years since first use are used, but for the highest category an
       arbitrary value has been chosen (12 yrs for the >10 yrs categories of INTERPHONE). If necessary
       these values were slightly adjusted to show overlapping points.
6.6 Parotid gland tumours
    Parotid glands are the largest salivary glands and located below the ears. Thus
    they are the salivary glands most exposed when making a call with a mobile
    phone.
    Case-control studies
    Five publications report on parotid gland tumours.69,72,82,85,100 Since no pooled
    analysis of the parotid gland tumour data has been published at this time, these
    studies are presented separately. The publications from the Hardell group69,72
    only presented cumulative categories of exposure. As this did not allow a direct
    comparison with the other material, only the non-overlapping information is
    presented.
94  Mobile phones and cancer
</pre>

====================================================================== Einde pagina 95 =================================================================

<br><br>====================================================================== Pagina 96 ======================================================================

<pre>Figure 15 Adjusted Odds Ratios with 95% confidence intervals for parotid gland tumours for years
since first use.
• Data from: Sadetzki et al. (2007)85, Lönn et al. (2006)82); Hardell et al. (2004)69; Auvinen et al.
   (2002)100; Söderqvist et al. (2012)72.
• The midpoints of the ranges for years since first use are used, but for the highest category an
   arbitrary value has been chosen (1.4 yrs for the < 0.5 yrs category of Auvinen et al. (2002)100, 4
   yrs for the >3 yrs category of Auvinen et al. (2002)100, 12 yrs for the >10 yrs category of
   INTERPHONE, Hardell and Söderqvist). If necessary these values were slightly adjusted to show
   overlapping points. Error bars were cut at an OR of 6.0.
The analyses of the Hardell group data by duration of exposure69,72 did not result
in any increased risks for use of analogue, digital or cordless phones, and neither
did the data of Sadetzki et al. (2007)85 and Lönn et al. (2006)82 following the
INTERPHONE protocol, and Auvinen et al. (2002)100 (Figure 15, Table H15).
     The analyses of these studies by cumulative call time also did not result in
any increased risks (Figure 16, Table H16).
Results: analysis of the data by disease                                                              95
</pre>

====================================================================== Einde pagina 96 =================================================================

<br><br>====================================================================== Pagina 97 ======================================================================

<pre>   Figure 16 Adjusted Odds Ratios with 95% confidence intervals for parotid gland tumours for
   cumulative call time.
   • Data from: Sadetzki et al. (2007)85, Lönn et al. (2006)82; Hardell et al. (2004)69.
   • The midpoints of the ranges for cumulative exposure are used, but for the highest category an
      arbitrary value has been chosen (12 yrs for the >10 yrs category of INTERPHONE and Hardell).
      If necessary these values were slightly adjusted to show overlapping points.
   No data on lateralisation were presented by the Hardell group. Only the two
   publications according to the INTERPHONE protocol provide this.82,85 No
   increased risks were found for time since first use in either study (Figure 17,
   Table H17).
   The only increased risk was found for one subgroup analysis in the study by
   Sadetzki et al. (2007)85 in the group containing both benign and malignant
   tumours that reported ipsilateral phone use and a cumulated call time >266 h
   (Figure 18, Table H17).
96 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 97 =================================================================

<br><br>====================================================================== Pagina 98 ======================================================================

<pre>Figure 17 Adjusted Odds Ratios with 95% confidence intervals for ipsilateral and
contralateral parotid gland tumours for years since first use.
• Data from Sadetzki et al. (2007)85, Lönn et al. (2006)82.
• The midpoints of the ranges for years since first use are used, but for the highest category
   an arbitrary value has been chosen (12 yrs for the >10 yrs category). If necessary these values
   were slightly adjusted to show overlapping points. Error bars were cut at an OR of 6.0.
Figure 18 Adjusted Odds Ratios with 95% confidence intervals for ipsilateral and
contralateral parotid gland tumours for cumulative call time.
• Data from Sadetzki et al. (2007)85.
• Arbitrary values have been chosen: 200 h for the <266.3 h category and 350 h for the
   > 266.3 h category.
Results: analysis of the data by disease                                                           97
</pre>

====================================================================== Einde pagina 98 =================================================================

<br><br>====================================================================== Pagina 99 ======================================================================

<pre>   Ecological studies
   De Vocht et al. (2011) published a brief report on the trends in parotid gland
   tumours in England over the period 1998-2008, but this study cannot draw any
   conclusions when latency periods of 10 or more years are assumed.
       The brief report by Czerninski et al. (2011)129 described a quite steady
   incidence of most parotid gland tumours in Israel and a rapid increase in
   incidence of sublingual gland cancers. These data are not linked to mobile phone
   use.
       The Committee has obtained incidence data for parotid gland tumours from
   the Netherlands Cancer Registry for the period 1989-2010. These do not show
   changes in the incidence of this tumour (Figure 19).
   Figure 19 Parotid gland tumour incidence in the Netherlands for different age groups. Source:
   Netherlands Cancer Registry managed by CCCNL.
98 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 99 =================================================================

<br><br>====================================================================== Pagina 100 ======================================================================

<pre>6.7  Pituitary tumours
     The pituitary gland, or hypophysis, is an endocrine gland at the bottom of the
     hypothalamus at the base of the brain. It secretes important hormones such as
     growth hormone and thyroid stimulating hormone.
          Two publications reported on case-control studies on pituitary tumours.88,89
     The detailed results are presented in Annex G, Tables H18 and H19. No
     associations were found.
6.8  Malignant melanoma of the eye
     The structure giving rise to the colour of the eye is the uvea, which includes the
     iris. It contains pigment cells (melanocytes) from which cancer (melanoma) may
     arise.
          Also for this tumour results from two publications are available.98,104 The
     detailed results are presented in presented in Annex G, Tables H20 and H21. No
     associations were found
6.9  Intra-temporal facial nerve tumours
     This includes the results from one publication.102 However, the analysis
     presented in this publication does not allow any comparison with the other
     studies, as only the individual answers to a questionnaire are presented.
     Combinations of duration of mobile phone use while corrected for confounders
     are not given.
6.10 Neuroblastoma
     Neuroblastoma is a neuroendocrine tumour that originates in neural tissue
     outside the central nervous system. Only one publication reports the risk of this
     type of tumour in children in relation to mobile phone use by the parents.97 Also
     in this case, the analysis presented in this publication does not allow any
     comparison with the other studies, as only the individual answers to a
     questionnaire are presented. Combinations of duration of mobile phone use while
     corrected for confounders are not given.
     Results: analysis of the data by disease                                           99
</pre>

====================================================================== Einde pagina 100 =================================================================

<br><br>====================================================================== Pagina 101 ======================================================================

<pre>100 Mobile phones and cancer</pre>

====================================================================== Einde pagina 101 =================================================================

<br><br>====================================================================== Pagina 102 ======================================================================

<pre>Chapter 7
        Discussion
7.1     The research questions
        In this report, the Committee addresses the question whether there is evidence
        from epidemiological studies that exposure to radiofrequency electromagnetic
        fields (RF EMF) from mobile phones is associated with an increased risk of
        tumours in the brain and various other tissues in the head.
            In assessing the evidence for a causal association based on the
        epidemiological data discussed in the previous chapters, the Committee uses the
        considerations of Bradford Hill12 (see 7.6). A causal association is more likely
        when there is an exposure-response relationship, such that the risk increases with
        increasing intensity and/or time of exposure, and when there is consistency
        between the studies. These points will be explicitly discussed in this chapter.
            An important issue is the assumption that there is a long latency time
        between the induction and the clinical manifestation of tumours in the head. As
        was discussed in Chapter 6, for the main tumours considered in this report,
        latency times of 10-15 years are assumed.
7.2     Strengths and limitations of this analysis
        The strength of this investigation is that it has been systematic in both the
        identification of the information available through original study publications
        Discussion                                                                         101
</pre>

====================================================================== Einde pagina 102 =================================================================

<br><br>====================================================================== Pagina 103 ======================================================================

<pre>    and the way it has evaluated the methodological quality of the available
    information.
         A limitation is that there are only limited possibilities for pooling of the data
    from the publications selected, as the data were generated with very different
    protocols and are thus not always sufficiently compatible. Another limitation is
    that there are only few studies with long-term users.
7.3 Mobile vs. cordless phones
    An issue that needs to be discussed before going into detail on the strengths and
    weaknesses of the different studies, is the exposure from cordless phones versus
    that of mobile phones. Hardell claims in his studies that the RF EMF exposures
    from both types of phones are of comparable magnitude, and that the observed
    increased risks associated with cordless phone use he observed in his studies are
    consistent with this. But is this claim valid?
         Vrijheid et al. (2009)144 used software modified mobile phones used by over
    500 volunteers in 12 countries to measure the output power of mobile phones.
    The 900 MHz phones transmitted with an average power of 133.3 mW
    (maximum 250 mW, based on 46994 calls), and the 1800 MHz phones with an
    average of 64.2 mW (maximum 125 mW, based on 29505 calls).*
         The maximum power of a cordless DECT phone is 10 mW and during a call
    transmission is always at this maximum. There is no transmission in standby
    mode.** This means that exposure to radiofrequency electromagnetic fields from
    DECT phones is considerably lower than exposure from mobile phones.
         Some authors, however, have concluded otherwise. Redmayne et al.
    (2010)172 discussed the exposure by cordless phones and compared that with the
    data for mobile phones as assessed by Vrijheid et al. (2009).144 Vrijheid et al.
    state that “Analyses included data recorded during speech communication only.”
    This means: not during texting, but for the entire duration of a call, both during
    speaking and listening. However, Redmayne et al. (2010)172 erroneously inter-
    preted this statement that power was only registered during speaking and not
    during listening. They conclude from this that the average exposure from mobile
*   Maximum SAR values of mobile phones vary from 0.07 up to 1.59 W/kg (with similar models
    having sometimes different SAR values in different countries).168-170 This corresponds to 3.5-80% of
    the ICNIRP limit (=2 W/kg).Since the average power is just over half of the maximum, the average
    SAR will also be.
**  For one type of handset the manufacturer supplies the SAR10g, which is 0.06 W/kg. The maximum
    peak SAR for several types of DECT handsets was calculated at 0.00794 – 0.052 W/kg.171 This
    corresponds to 0.4 – 2.6% of the ICNIRP limit (=2 W/kg).
102 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 103 =================================================================

<br><br>====================================================================== Pagina 104 ======================================================================

<pre>phones is likely to be much lower than the levels given by Vrijheid et al .
(2009)144 and that the exposure from cordless phones during a conversation
might be considerably higher than that from a mobile phone. This incorrect
conclusion is adopted by Hardell et al. (2011)68,165 to explain the increased risks
observed with cordless phone use.
    Hardell et al. (2011)68 distinguished two exposure categories on the basis of
call time: below and above the median. The median for mobile phones is a call
time of 74 h, while for cordless phones it is 243 h. That means that at the median
the total ‘exposure’ (calling time x output power) is 9864 mWh for GSM 900
MHz phones (assuming the average power of 133.3 mW), 4751 mWh for GSM
1800 MHz (assuming the average power of 64.2 mW) and 2430 mWh for DECT
(assuming the output power of 10 mW). So there is a considerable difference in
‘exposure’ between especially the 900-MHz GSM and DECT phones. If only the
output power would be the relevant parameter this difference is even greater.
Another, related, issue is that of analogue versus digital phones. When mobile
phones were first introduced, the signal type was an analogue one, i.e., a
continuous signal that was amplitude and frequency modulated to transfer
information. Since the capacity and speed of data transfer using these signals
proved insufficient, a digital type signal was developed. This uses pulsed
transmissions with a complex modulation for speech and data transfer. The most
widespread type is the GSM standard, which is in use in most of the countries
included in the studies in this report.
    The output power of the (now outphased) analogue phones was higher than
that of the digital ones. While the digital phones have a facility called adaptive
power control, that regulates the output power according to need in order to
establish and maintain a connection with the nearest base station, analogue
phones did not, and the average distance to a base station was also higher than
with the digital systems. Kelsh et al. (2011)173 measured the output in various
types and models of mobile phones using 4 different operating systems,
including an analogue one and GSM 1900 MHz. They did not measure while
actual phone calls were made by volunteers, as in the Vrijheid et al. (2009)144
studies, but measured in a standardized setup while driving along fixed routes in
different environments (urban and rural). The mean output of the analogue
phones was 171.40 mW, while that of the GSM 1900 MHz phones was 25.76
mW. This value for the GSM phones is lower than the 64.2 mW measured by
Vrijheid et al. (2009).144 Apart from the differences in methods of data
collection, this difference may also have to do with the fact that the Kelsh et al.
(2011)173 study was performed in 2005/2006 in the USA, and the Vrijheid et al.
Discussion                                                                          103
</pre>

====================================================================== Einde pagina 104 =================================================================

<br><br>====================================================================== Pagina 105 ======================================================================

<pre>    (2009)144 study between 2001 and 2005 in 12 of the 13 INTERPHONE
    countries, so the results of both studies are probably not directly comparable. In
    any case, the Kelsh et al. (2011)173 study clearly shows that exposure from
    analogue phones is considerably higher than that of GSM phones.
        The GSM standard was first commercially introduced in Europe in 1990 (in
    Finland) and started being used at a large scale in the mid-1990’s. Recruitment in
    the case-control studies took place between 2000 and 2004 (INTERPHONE) and
    1997-2003 (Hardell), and in the Danish cohort study in 1997. This means that the
    subjects in the epidemiological studies that have been using mobile telephony for
    the longest time periods (10 year of more) will initially have used analogue
    phones. So the exposure in that period of use was likely to be considerably
    higher than that in the later period when GSM phones were used. This makes the
    difference in exposure with cordless phones even larger.
    These differences are at odds with the conclusion by Hardell et al. (2011)68 that
    exposure from both types of phones is of the same order. Hardell found grossly
    similar odds ratios for the use of mobile or cordless phones, that are thus hard to
    explain on the basis of actual incident or total ‘exposure’. It is not known, but
    considered possible by the Committee, that there is a correlation between the use
    of cordless phones and mobile phones. This could in part be an explanation for
    the increased risks found for cordless phone use. Hardell et al. did not clarify
    whether the risk estimates for cordless phones were adjusted for mobile phone
    use. The Committee feels that the cordless phone data challenge the internal
    consistency of the Hardell et al. studies.
7.4 Strengths and limitations of the different study types and studies
    Cohort, case-control, case-case and ecological studies all have different strengths
    and limitations.
    Cohort studies generate potentially strong evidence, as the exposure can be
    repeatedly and objectively measured or assessed before the outcome occurs.
    These studies therefore do not suffer from recall bias. However, it is often
    difficult to perform cohort studies in the optimal way, as the investment in
    (preferably longitudinal) exposure measurements can be high and the cohort will
    have to be followed for a long time, up to several decades. The main cohort study
    in this evaluation (the Danish cohort study48,50-52,174) merely considered the
    duration of the subscription for those people that started a private subscription
    before 1996, and compared that to all other residents of Denmark. No
104 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 105 =================================================================

<br><br>====================================================================== Pagina 106 ======================================================================

<pre>information has been gathered on the intensity and duration of use, such as the
number of calls and the total duration of calls, as has been done in the case-
control studies. Clearly the mere time that passed since a subscription started
(which was also assessed in the case-control studies) is a less meaningful
endpoint than an estimate of the amount of use, which is more directly associated
with exposure.
     Two other points need to be discussed with respect to the Danish cohort. The
first is that business contracts were excluded from the ‘exposed’ group, since
these subscriptions could not be related to individuals. This means that a number
of business users, who are possibly among the heaviest users in the period before
1996, are included in the control group. The second issue is that the mobile
phone use in the control group, the rest of the Danish population, also strongly
increased after 1996. This means that in the later publications with longer follow-
up there will be increasing misclassification in the control group. However, it can
be demonstrated that, because there is no misclassification in the ‘exposed’
group, any misclassification in the control group has only limited effect on the
calculated risk.152 Therefore the cohort study is potentially well suited to
examine risks also long after first use.
     So, because a cohort is a strong study design and the score of the Danish
cohort in the quality evaluation was good, the Committee considers the Danish
cohort, despite the lack of actual exposure data, important for the overall
evaluation.
     The Committee considers the other cohort studies identified of little value for
the overall data analysis, mainly because of the only short periods of follow-up,
which are not relevant for very slow growing tumours.
Case-control studies are very efficient in their data collection, since they focus
on new cases arising in a restricted time period. This has great advantages over
cohort studies, especially in case of relatively rare diseases such as brain
tumours, where large cohorts are needed to obtain sufficient cases. However,
exposure assessment in case-control studies is always retrospective, therefore
these studies potentially suffer from some major sources of bias, as discussed in
Chapter 4. The main case-control studies identified in this analysis are those
from the Hardell group and the INTERPHONE studies. The other case-control
studies have much lower overall ‘exposure’ and considered much shorter times
since first phone use. They are therefore not really relevant under the assumption
of the Committee that, if there would be a risk associated with mobile phone use,
it would be increasing with increasing exposure and usage time. They will not
further be discussed here.
Discussion                                                                           105
</pre>

====================================================================== Einde pagina 106 =================================================================

<br><br>====================================================================== Pagina 107 ======================================================================

<pre>        The Committee considers the INTERPHONE studies to be prone to selection
    bias due to the overall relatively low response rates. Because these are also lower
    for the controls than for the cases, this might lead to differential misclassification
    bias. The Hardell studies reported higher response rates and smaller differences
    in response rates between cases and controls than the INTERPHONE studies.
    These response rates were for the controls still higher than the response rates in
    the Swedish part of INTERPHONE (see 5.1). So also on the basis of the
    recalculated response rates, the Hardell studies are less likely to suffer from
    selection bias than the INTERPHONE studies, but the response rates in
    especially the Hardell controls are unusually high.
        In both study protocols there is also the possibility of observer bias. The
    interviews of the INTERPHONE studies were all done in person at the
    participants’ home. In spite of the training of the interviewers, they might in
    some way have been unknowingly influenced by the case or control status of the
    subjects. This is also the case with the Hardell studies. Although in these studies
    the initial information has been gathered by postal questionnaire, all participants
    received a follow-up interview by phone. The investigators state that this was
    conducted in a blinded fashion, but during the interview disease status may well
    have become known. So observer bias is not a likely explanation for any
    differences in outcomes between the two studies.
        Both the Hardell and INTERPHONE studies are also inherently prone to
    recall bias. A recent publication evaluated a subsample of the INTERPHONE
    study with the aim of improving the exposure assessment by taking the location
    of the tumour relative to the preferred position of the mobile phone into
    consideration.114 However, as this still relies on recall of both the position of the
    phone and the extent of its use, it is a refinement of the analysis but it does not
    solve the fundamental problem of recall bias.
    The Committee has spent a great deal of effort in systematically assessing the
    methodological quality of the various studies (see Chapter 5) and the issue of
    bias discussed above plays an important role in that analysis. It did not result in
    any major differences in quality between the two main research groups, Hardell
    and INTERPHONE. The overall rating of the Hardell studies was 7.6 for the
    glioma + meningioma and the acoustic neuroma studies, and 6.7 and 7.4 for the
    parotid gland tumour studies. For INTERPHONE the ratings were 6.7
    (glioma+meningioma), 7.2 (acoustic neuroma), and 6.7 and 6.5 (parotid gland
    tumours) (Table 5.1). However, this quality analysis has not taken into account a
    number of issues relating to internal and external consistency.
106 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 107 =================================================================

<br><br>====================================================================== Pagina 108 ======================================================================

<pre>                 The first issue is that of the cordless phones. In view of the lower exposure
            resulting from the use of these phones in comparison to mobile phones as
            discussed in 7.3, the Committee considers it highly unlikely that similar odds
            ratios would be observed, as was the case in the Hardell studies. But, as
            discussed earlier, an explanation for these findings might also be that there is a
            correlation between the use of mobile and cordless phones.
                 The second issue is that of the increased risks observed by Hardell et al. at
            very short usage times. These are unlikely in view of the presumably very long
            latency times of the tumours under consideration. Also, if these increased risks
            were true, increased incidences in the ecological studies would be expected, but
            these were not observed.
                  According to the Committee these issues cast some doubt on the validity of
            the Hardell et al. studies.
            Another point that is important to take into account is the fact that the Hardell et
            al. studies are performed in only one country (Sweden), while the
            INTERPHONE studies cover 16 areas in 13 countries, thus covering a much
            broader population. The total numbers of cases and controls are also lower in the
            Hardell et al. studies compared to INTERPHONE (Table 7.1).
                 Effectively this comparison can only be made for the glioma studies. It
            should also be born in mind that for the full data set the age ranges are dissimilar.
            As the incidence of brain tumours is very much age-dependent, this is a major
            issue and a direct comparison between the Hardell and INTERPHONE data
            should only be made with the age-limited Hardell dataset that has the same age
            range as the INTERPHONE dataset. This effectively limits this comparison to
            the highest categories for ‘Time since first use’ and ‘Cumulative call time’. In the
            studies on other tumours, Hardell et al. always make a distinction between users
            of cordless and mobile phones, with sometimes also a division between analogue
            and digital mobile phones. The numbers for these categories are sometimes
            provided, but there is overlap when subjects have used more than one type of
            phone, and the overall total numbers are not provided.
Table 7.1 Comparison of numbers of cases and controls in the Hardell and INTERPHONE studies.
                                                            Hardell66,165             INTERPHONE93
                                                            Cases / controls          Cases / controls
Glioma                Time since first use                  529 / 963 (20-80 yrs)     1666 / 1894 (30-59 yrs)
                      Cumulative call time                  529 / 963 (20-80 yrs)     1666 / 1894 (30-59 yrs)
Glioma 30-59 yrs      Time since first use ≥10 yrs          56 / 74                   252 / 232
                      Cumulative call time ≥1640 h          29 / 37                   210 / 154
            Discussion                                                                                        107
</pre>

====================================================================== Einde pagina 108 =================================================================

<br><br>====================================================================== Pagina 109 ======================================================================

<pre>           In summary, there is doubt on the internal and external consistency of the
      Hardell data on account of (1) the increased risk observed already with very short
      usage times; (2) the unusually high response rates in the controls; and (3) the
      increased risks observed for cordless phone use, again in some cases for very
      short usage time. For these reasons, in combination with the lower numbers of
      subjects, the Committee has given the Hardell et al. studies less weight than the
      INTERPHONE studies in the overall analysis and conclusions.
      Case-case studies are potentially powerful, as they are less likely to suffer from
      selection and observer bias. There will of course still be recall bias, but this will
      be non-differential, since only patients are involved. However, case-case studies
      are limited because they are often single-hospital based and thus will have very
      limited numbers of cases for rare diseases such as brain cancers. This applies to
      all the case-case studies discussed in this report, with the exception of the multi-
      hospital study by Sato et al. (2010)112 that included 1589 cases, and two
      subsamples of the INTERPHONE study that have been analysed in a case-case
      fashion, including respectively 2692147 and 888 cases.113
      Ecological studies are inherently limited in their interpretation, since individual
      exposure is not determined. Instead, these studies investigate trends in the
      incidence (or prevalence) of disease and, in this case, the development of the
      number of mobile phone subscriptions. However, as indicated, for the tumours
      considered in this report there is only limited information on the latency time.
      Exposure-effect relationships cannot be derived from ecological studies. At best,
      they can show a similarity in trends in increase of disease and phone use.
      Absence of an increase in disease incidence following an increase in mobile
      phone possession (and presumed use) does not prove the absence of a causal
      relation between exposure and disease, but might give support to it when the
      period of strong increase in phone use is a decade or longer in the past, if it is
      assumed that the latency period is more than a decade. If the latency would be a
      decade or less, an increased risk would have been expected in the trends by now.
7.5   Overall discussion per tumour type
7.5.1 Brain tumours not further specified
      It is not possible to draw any conclusions with respect to the relation between
      mobile phone use and the occurrence of brain tumours in general.
108   Mobile phones and cancer
</pre>

====================================================================== Einde pagina 109 =================================================================

<br><br>====================================================================== Pagina 110 ======================================================================

<pre>           No increased risks were found in the two case-control studies and in the
      cohort study. However, even though the quality of the case-controls studies is
      reasonable, the follow-up is too short for them to be meaningful. The cohort
      study has limitations with respect to exposure categorization.
7.5.2 Glioma
      The Committee concludes that there are some weak and inconsistent indications
      for an association between prolonged and intensive use of a mobile phone and an
      increased incidence of gliomas. These might be explained by various types of
      bias and chance, but it cannot be excluded that there is a causal relation.
      However, the Committee considers the likelihood for a causal relation very low.
      The population statistics also do not show an increased incidence of glioma. But
      since it is likely that the latency time for these tumours is very long, an increased
      incidence might not yet be visible.
      Time since first use, overall usage
      Most cohort studies had a follow-up period that is too short to show a possible
      increase in glioma risk and they are therefore not useful for the current analysis.
      The only long-term cohort study, the Danish cohort, gives no indication of an
      increased risk at follow-up times of ≥13 years for those who started to use a
      mobile phone before 1996.
           The measure of exposure used in this cohort study, length of subscription, is
      only a crude measure. It is also used in some of the case-control studies, mostly
      as time since first use. The Committee considers other endpoints used in the
      case-control studies that give a more direct measure of exposure – cumulative
      number of calls and, even more, cumulative call time – to provide the most
      relevant data, even though they are suffering from various types of bias, as
      discussed earlier.
           For time since first use the INTERPHONE study did not find any increased
      risks, only two decreased risks for intermediate follow-up times. The only
      explanation for this is bias (mainly selection bias) and chance. Case-case
      analyses of two subsets of the INTERPHONE data provided contradictory
      results: in one subset an increased risk was found for the highest category of time
      since first use, >10 years, while in the other study no increased risks were found.
      So nothing can be concluded from these case-case analyses.
           Hardell observed increased risks for all glioma in the highest category of >10
      years, and for aggressive brain tumours, astrocytomas, in the two highest
      Discussion                                                                            109
</pre>

====================================================================== Einde pagina 110 =================================================================

<br><br>====================================================================== Pagina 111 ======================================================================

<pre>    categories, >5-10 years and >10 years. This is a pattern that can be expected if
    there would be a causal relation between mobile phone use and brain tumours.
    However, Hardell also found an increased risk for cordless phone use, both for
    all gliomas and for astrocytomas, in the two highest categories. An increased risk
    with cordless phone use is not consistent with the lower exposure from cordless
    phones compared to that from mobile phones. It is also unlikely and not
    consistent with other data to observe an increased risk already after 5-10 years of
    phone use.
    A meta-analysis of the data from the longest usage time categories has been
    performed. A full description and all data are given in Annex I. The data were
    tested for heterogeneity and datasets for which the p-value was <0.05 were
    considered to be too heterogeneous for a meaningful meta-analysis and are not
    reported here. They are shown in Annex I for completeness, however.
         The data for time since first use >10 y using the Hardell data for the full age
    range of 20-80 y were too heterogeneous for a combined analysis (Annex I,
    tables I1). When the Hardell et al. data were not included, there was no
    heterogeneity (Annex I, tables I2), meaning that the Hardell data strongly deviate
    from the Frei et al. and INTERPHONE data. When the subset of the Hardell data
    for the age range of 30-59 y was used (which is similar to that of the
    INTERPHONE study) a non-significantly increased overall OR of 1.14 (95% CI:
    0.90, 1.45) was calculated (Figure 20; Annex I, Tables I3).
    Cumulative call time, overall usage
    For cumulative call time INTERPHONE found an increased risk only in the
    highest category (cumulative call time ≥1640 h), and a decreased risk in the next-
    highest (and several lower ones). So there is no obvious exposure-response
    relationship. Since it is not very likely that mobile phone use results in a
    protective effect, there should be another explanation for the decreased risks. The
    authors of the INTERPHONE publication conclude that the bias inherent to case-
    control studies could in part (thus not completely) explain their results.
         Especially selection bias associated with the poor response rates of the
    INTERPHONE studies could result in the decreased risks. But this would mean
    that the observed increased risk in the highest category may also have been
    underestimated due to selection bias, while on the other hand it also could have
    been overestimated due to recall bias. It is not possible to fully assess these
    differential effects. The INTERPHONE researchers also analysed the data for the
    highest category while excluding subjects that indicated a very high average
110 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 111 =================================================================

<br><br>====================================================================== Pagina 112 ======================================================================

<pre>Figure 20 Forest plot of the glioma data for use >10 y. Data from Frei et al. (2011)52, Hardell et al.
(2011)165, INTERPHONE (2010)93.
NB: this meta-analysis has only been performed on the data for the highest ‘exposure’ category. It
does not take into account any exposure-response relationships, and the possible influence of bias and
other factors that have been discussed in this report has not been accounted for. Therefore this
analysis does not provide an estimate of the true risk increase.
daily call time of >5 h (that was deemed implausible by some). This resulted in
no increased risk. It is questionable, however, whether this procedure is justified,
since it concerned quite some subjects: 41 of 210 cases (19.5%) and 20 of 154
controls (13%).
     Two subsets of the INTERPHONE study were also analysed for cumulative
call time. Neither found an increased risk. For one of the subsets the total
accumulated energy in the tumour was calculated also, and in the highest
category for the subset that had used the phone ≥7 years in the past the risk was
increased. Since the calculated cumulated energy still relies on reported phone
use, this method does not avoid the influence of recall bias.
     Hardell observed increased risks for both all glioma and astrocytoma in all
categories of cumulative call time (1-1000 h, 1001-2000 h and >2000 h) for
mobile phone use, and in the two highest categories for cordless phone use. The
latter is inconsistent with the lower exposure from cordless phones compared to
that from mobile phones. The Committee also considers an increased risk even in
the lowest category of 1-1000 h not very likely.
Discussion                                                                                             111
</pre>

====================================================================== Einde pagina 112 =================================================================

<br><br>====================================================================== Pagina 113 ======================================================================

<pre>    The heterogeneity analysis of data for cumulative call time for the full age range
    Hardell data (>2000 h) and INTERPHONE (>1640 h) resulted in a p-value
    <0.05, i.e. the data are too heterogeneous for a meta-analysis (Annex I, I4).
    When the limited age range Hardell data were used (in which they used a
    cumulative call time >1640 h to be more comparable to the INTERPHONE
    data), heterogeneity was less and the overall OR was significantly increased at
    1.48 (95% CI: 1.13, 1.93) (Figure 21; Annex I, I5).
    Figure 21 Forest plot of the glioma data for cumulative call time >1640 h. Data from Hardell et al.
    (2011)165 and INTERPHONE (2010) 93.
    NB: this meta-analysis has only been performed on the data for the highest ‘exposure’ category. It
    does not take into account any exposure-response relationship, and the possible influence of bias and
    other factors that have been discussed in this report has not been accounted for. Therefore this
    analysis does not provide an estimate of the true risk increase.
    Number of calls, overall usage
    INTERPHONE was the only study also to analyze the number of calls, a measure
    that was shown to be less prone to recall bias than duration of calls (and
    consequently total call time).146 No increased risk was observed in any of the
    categories (maximum > 27,000 calls). Decreased risks were observed in some of
    the lower categories, indicating some form of bias. No meta-analysis for this
    endpoint is possible, since there is only one study.
112 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 113 =================================================================

<br><br>====================================================================== Pagina 114 ======================================================================

<pre>Time since first use, laterality
In the laterality analyses of time since first use INTERPHONE did not observe
any increased risks with either ipsilateral (side of the head where the tumour is
located) or contralateral use (side of the head opposite from the tumour location).
Decreased risks were observed in two intermediate categories for contralateral
use. Again, bias and chance are the most likely explanations for this. Hardell on
the contrary observed increased risks for ipsilateral use for both mobile and
cordless phone use of even total usage times as short as >1 year, and for
contralateral use of a mobile phone for >10 years. The Committee considers it
highly unlikely that, with these slowly growing tumours, an increased risk would
be visible already after 1 year of phone use, and even more unlikely that this
could be the case after >1 year use of a cordless phone, which results, as
discussed earlier, in a considerably lower exposure than a mobile phone.
Moreover, the increased risk for contralateral use of a mobile phone is also
unlikely, since most of the energy of the phone that enters the head is deposited
within several centimetres of the antenna.114,147,175
The heterogeneity analysis indicated that the data for both ipsi- and contralateral
use with the Hardell data for the full age range had a p value <0.05 (Annex I,
I6, I7). They are therefore too heterogeneous for a meta-analysis. This is also
the case for the contralateral use data with the age-limited Hardell data set
(Annex I, I9). Only the ipsilateral data using this set had a p value >0.05; the
meta-analysis resulted in a non-significantly increased OR of 1.62 (95% CI:
0.87, 3.01) (Figure 22; Annex I, I8).
Cumulative call time, laterality
The only increased risk in the INTERPHONE glioma studies was found for
ipsilateral use and the highest category of cumulative call time (≥1640 h). In a
reanalysis of his data limiting it to the age range used by INTERPHONE (39-50
years), Hardell also found an increased risk for ipsilateral exposure (although
based on very few subjects). If there would indeed be an association between
mobile phone use and glioma risk, this is a situation where this might be
expected. However, in several other categories of both ipsi- and contralateral use,
INTERPHONE observed decreased risks. The INTERPHONE investigators
went at great length to find an explanation for this, but finally concluded that bias
(mainly selection bias) and chance were the most likely explanations. The
Committee agrees with that, and sees no reason why the only increased risk
Discussion                                                                            113
</pre>

====================================================================== Einde pagina 114 =================================================================

<br><br>====================================================================== Pagina 115 ======================================================================

<pre>    Figure 22 Forest plot of the glioma data for time since first use >10 y, ipsilateral use. Data from
    Hardell et al. (2011)165 and INTERPHONE (2010) 93.
    NB: this meta-analysis has only been performed on the data for the highest ‘exposure’ category. It
    does not take into account any exposure-response relationship, and the possible influence of bias and
    other factors that have been discussed in this report has not been accounted for. Therefore this
    analysis does not provide an estimate of the true risk increase.
    estimate could not also be explained by this. This point of view is supported by
    the fact that for cumulative number of calls, an endpoint that is closely related to
    cumulative call time, no increased risk was observed in the INTERPHONE
    studies, but again several decreased risks. It has been indicated earlier that this
    could be the result of selection bias.
    The heterogeneity analysis showed that the datasets including the age-restricted
    Hardell data for both ipsi- and contralateral use had p-values >0.05 and are
    therefore suitable for a meta-analysis. For ipsilateral use a statistically
    significantly increased OR of 2.03 (95% CI: 1.37, 3.00) was found, and for
    contralateral use the OR was not significantly increased: 1.32 (95% CI: 0.76,
    2.28) (Figure 23; Annex I, I10, I11).
114 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 115 =================================================================

<br><br>====================================================================== Pagina 116 ======================================================================

<pre>Figure 23 Forest plot of the glioma data for cumulative call time >1640 h; left panel: ipsilateral use, right panel: contralateral
use. Data from Hardell et al. (2011)165 and INTERPHONE (2010) 93.
NB: this meta-analysis has only been performed on the data for the highest ‘exposure’ category. It does not take into account
any exposure-response relationship, and the possible influence of bias and other factors that have been discussed in this report
has not been accounted for. Moreover, the INTERPHONE data include a number of subjects that reported an unlikely high daily
calling time. Therefore this analysis does not provide an estimate of the true risk increase.
              Ecological studies
              The ecological studies support the absence of an increased risk. Little et al.
              (2011)133 showed that if the risks Hardell et al. (2011)167 reported were true, a
              clearly increased glioma rate should already have been visible. If the increased
              risk reported in the INTERPHONE studies93 were true, that would be consistent
              with the observed incidence patterns. So there could be a small risk, but there
              could also be no risk at all. The incidence of brain tumours in the Netherlands
              has been constant over the period 1998-2010 in the age groups <60 y, supporting
              the absence of an increased risk from mobile phone use.
7.5.3         Meningioma
              The Committee concludes that there are no clear and consistent indications for an
              increased risk of meningioma from using a mobile telephone.
                  The INTERPHONE studies showed no increased risk at all for meningioma
              in any of the groups for any of the endpoints, only several decreased risk, but
              without a clear exposure-response relationship. These findings can be regarded
              as the result of (selection) bias and/or chance. For overall exposure Hardell
              observed an increased risk only for analogue phones and a time since first use of
              >10 years, and for cordless phone use for >5-10 years. The latter is inconsistent
              Discussion                                                                                                       115
</pre>

====================================================================== Einde pagina 116 =================================================================

<br><br>====================================================================== Pagina 117 ======================================================================

<pre>      with an exposure-response relation and with the lower exposure caused by
      cordless phones, but there might be a correlation between cordless and mobile
      phone use. The laterality data only show increased risks for ipsilateral use of
      mobile phones for >1 year (and not for >10 years) and for cordless phones of >10
      years. Again, these data from Hardell et al. are inconsistent and cannot logically
      be explained.
          The Danish cohort study did not show any increased meningioma risks. The
      other two case-control studies had too short follow-up times and too few cases in
      the highest duration category to be meaningful.
7.5.4 Acoustic neuroma
      The Committee feels that the data on an association between long term use of a
      mobile phone and acoustic neuroma are inconsistent and do not really give an
      indication for an increased risk.
           INTERPHONE found an increased risk only in the ipsilateral subgroup with
      the highest cumulative call time. This is not contradictory to expectations, but the
      lack of any exposure-response and even a decreased risk in the next-highest
      category are not supportive of a real increase in risk. In the data for cumulative
      number of calls, that are highly correlated to those of cumulative call time,
      decreased risks were found for both ipsi- and contralateral use in the middle one
      of five categories. The Committee feels that such results cannot logically be
      explained and that these data therefore should be regarded as being the result of
      bias and/or chance, as discussed before. It is likely that this is also true for the
      only increased risk, although this could work both ways: the actual risk could
      both be higher and lower than the observed one. No changes in risk were
      observed by INTERPHONE for time since first use for both ipsi- and
      contralateral use.
          The Hardell data show an increased risk for ipsilateral mobile phone use >10
      years, but also for ipsilateral use >1 year, for both mobile and cordless phones.
      The overall data (so including both ipsi- and contralateral use) for time since first
      use show increased risks for analogue phone use in all categories (>1-5, >5-10
      and >10 years), for digital phones only for >1-5 and >5-10 years and for cordless
      phones only for >1-5 years. For cumulative call time the Hardell data show
      increased risks only in the highest category, but for all three phone types. So
      these data are not really consistent. It is highly unlikely that any increased risk
      would show up already after >1 year of phone use.
116   Mobile phones and cancer
</pre>

====================================================================== Einde pagina 117 =================================================================

<br><br>====================================================================== Pagina 118 ======================================================================

<pre>          The data from the Danish cohort are very limited but also do not indicate any
      effect. The data from other case-control studies lack an adequate follow-up time
      and sufficient subjects, and are therefore practically of no value.
      A heterogeneity analysis was performed on the data for time since first use >10
      y, both for all use and for ipsi- and contralateral use, and on the data for
      cumulative call time >1000/1640 h (Annex I, I12-I15). In all cases the p-value
      was <0.05, indicating too much heterogeneity for a meta-analysis.
7.5.5 Parotid gland tumour
      The Committee concludes that there are no clear indications for an increased risk
      of parotid gland tumours from using a mobile phone. The data from the various
      studies on parotid gland tumours have shown only one increased risk estimate in
      one subgroup in one study with limited numbers of cases. This could have been
      the result of chance. The incidence data including those from the Netherlands
      also do not show an increase.
7.5.6 Other (pituitary, melanoma eye, intra-temporal facial nerve tumours and
      neuroblastomas)
      For pituitary tumours, melanoma eye tumours, intra-temporal facial nerve
      tumours and neuroblastomas tumours no conclusions regarding risks associated
      with the use of mobile phones can be drawn.
          In case of the studies on pituitary tumours and malignant melanoma of the
      eye, the numbers of cases and controls were very small in all exposure
      categories, and particularly in the groups with longer or heavier exposure. The
      studies on intra-temporal facial nerve tumours and neuroblastomas were of a
      nature that did not allow risks to be determined.
7.6   The Bradford Hill considerations
      The Committee has focused in this report on epidemiological studies. In such
      observational studies the quality of exposure assessment is crucial, especially in
      deriving exposure-response relations.176 Moreover, the extent of selection bias
      and the adjustment for confounding factors are important in assessing the
      evidence for causality of associations. A standard tool in assessing evidence for
      causality are Bradford Hill’s considerations.12 Of these, in more recent
      epidemiological literature, strength, consistency, temporality, biological gradient
      Discussion                                                                          117
</pre>

====================================================================== Einde pagina 118 =================================================================

<br><br>====================================================================== Pagina 119 ======================================================================

<pre>    (or exposure-response) and biological or physical plausibility are considered. It
    should be borne in mind that presence of these items is considered a contributing
    argument that causality is likely, but their absence does not prove that there is no
    causality.
    Strength
    A relative risk or odds ratio higher than 2 is usually considered to be a relatively
    strong association. Most relative risks observed in the studies discussed in this
    report are lower than 2. It is likely that in the studies described, misclassification
    of exposure occurs. This will mostly lead to underestimation of the odds ratio,
    thus decreasing the strength of the observed association. Nevertheless, an odds
    ratio of less than 2 could also be indicative of causality if it is consistently
    observed. This is not really the case in the studies described in this report.
    Consistency
    Consistency of results from different studies strengthens the causality argument.
    However, the consistency across and within the studies discussed here is not very
    high. In several studies some increased risks have been observed in subgroups,
    while in particular in the INTERPHONE studies many decreased risks were
    found. Mostly, however, no increased or decreased risks were observed.
    However, where one would expect the effect to occur if an effect exists, such as
    in the ipsilateral side of the exposure after longer or heavier exposure, some
    consistency might be perceived.
    Temporality
    This refers to the fact that the occurrence of the disease should always follow the
    exposure. In case-control studies exposure is always measured retrospectively, so
    temporality can never truly be addressed. Cohort studies could provide more
    insight into this, but the cohort studies described in this report do not report
    increased risks. So no conclusions on temporality can be made.
    Biological gradient or exposure-response
    Exposure-response relationships can only be assessed if exposure can be
    measured adequately and with sufficient precision.176 However, since the case-
    control studies used questionnaires to retrospectively assess exposures which
118 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 119 =================================================================

<br><br>====================================================================== Pagina 120 ======================================================================

<pre>often occurred long ago, recall bias will decrease the accuracy of exposure
assessment. Where in the INTERPHONE studies an increased risk was observed,
this was only in the highest out of 10 exposure categories for cumulative call
time. This does not constitute a clear exposure-response association. No
increased risks were found for cumulative number of calls. Hardell observed
several exposure-responses in the analysis of time since first use and cumulative
use for gliomas.
Plausibility
This refers to the understanding of the biological model underlying a true
association between mobile phone use and brain tumours. Many reviews have
concluded that there is no known biological model to explain a relation between
mobile phone use and an increased risk of cancer.7,11,177,178
In conclusion, application of the Bradford Hill considerations to the available
epidemiological data is not supportive of a causal relation between the use of
mobile phones and the occurrence of tumours in the head. This may be because
there really is no causal relation, but it may also reflect inadequacies of the
methods used in the studies up to date or in the ability to measure exposure and
outcome.
Discussion                                                                        119
</pre>

====================================================================== Einde pagina 120 =================================================================

<br><br>====================================================================== Pagina 121 ======================================================================

<pre>120 Mobile phones and cancer</pre>

====================================================================== Einde pagina 121 =================================================================

<br><br>====================================================================== Pagina 122 ======================================================================

<pre>Chapter 8
        Conclusions and recommendations
        On the basis of the data presented in this systematic analysis, the conclusions can
        only be based on the results of three groups of studies: the case-control studies of
        Hardell et al. and of INTERPHONE, and the Danish cohort. All three study
        groups scored approximately similar in the analysis of the methodological
        quality. But since there is doubt on the internal and external consistency of the
        results of Hardell et al. and since the numbers of subjects in these studies are
        much lower than in the other two studies, the Committee gives the studies by
        Hardell et al. less weight than the other studies in the overall analysis and
        conclusions.
        No proven risk
        Based on the available epidemiological evidence described in this report and
        taking into account the quality of the different studies and their strengths and
        weaknesses, the final conclusion from this systematic analysis is then, that there
        is no clear and consistent evidence for an increased risk of tumours in the brain
        and other regions in the head in association with up to approximately 13 years
        use of a mobile telephone. For longer term use, for which no data are available,
        such risk cannot be excluded at present. In general it can be stated that the use of
        mobile phones has considerably increased since the studies described in this
        report were conducted, but what the long-term health effects of this, if any, may
        be is impossible to predict. Currently ongoing cohort studies, that include a better
        Conclusions and recommendations                                                      121
</pre>

====================================================================== Einde pagina 122 =================================================================

<br><br>====================================================================== Pagina 123 ======================================================================

<pre>    characterization of exposure than in the studies described in this report, might
    allow more firm conclusions in due time. A challenge in these studies will be to
    take account of the rapidly changing intensity of use and patterns of exposure,
    due to the changing types and use of mobile phones.
    The present systematic analysis shows that, despite substantial research efforts,
    there is still insufficient clarity and consistency regarding a possible association
    between mobile phone use and an increased risk of tumours in the brain and
    other regions of the head. There is some weak and inconsistent evidence for an
    association between prolonged and intensive use of a mobile phone and an
    increased incidence of gliomas. This is most likely explained by various types of
    bias and chance, but it cannot be excluded that there is a causal relation. For the
    other types of tumours, including meningiomas and acoustic neuromas,
    indications for an increased risk are much weaker or completely absent. The
    Committee notes that the meta-analyses as presented in the forest plots have only
    been performed on the data for the highest ‘exposure’ category. They do not take
    into account any exposure-response relationships, and the possible influence of
    bias and other factors that have been discussed has not been accounted for.
    Therefore they do not provide estimates of the true risk increase.
         The case-control studies have severe limitations due to their inherent
    vulnerability to several biases. Any increased risks observed for long-term or
    extensive use might be related to use of the – now obsolete – analogue mobile
    phones. Since most studies did not make a distinction between exposures from
    analogue and digital phones it is not possible to conclude anything on this issue.
    It is also possible that the follow-up period in the available studies is too short for
    an effect on the slow growing types of tumours to become manifest. However, up
    to now there is no indication from cancer registry data, including those from the
    Netherlands, that the incidence of brain or other tumours in the head is
    increasing, despite the very fast and sharp increase in mobile phone use that
    occurred from the mid-1990’s onwards. But again, the time period for this might
    be too short, in view of the slow development of the types of tumour under study.
         With the currently available data, consideration of the Bradford Hill criteria
    is not supportive of a causal relation between the use of mobile phones and the
    occurrence of tumours in the head.
    Measures
    There are currently in the Netherlands no legally binding exposure limits, but the
    government policy is that the ICNIRP guidelines are observed. Without implying
122 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 123 =================================================================

<br><br>====================================================================== Pagina 124 ======================================================================

<pre>that either the exposures currently experienced in daily life or the exposure limits
such as those proposed by ICNIRP are too high, the Committee would like to
suggest that there is no reason not to apply the ALARA principle to exposure to
RF EMF, meaning that exposures should be As Low As Reasonably Achievable.
This is fully in line with the suggestions from the Health Council’s advisory
report ‘Prudent precaution’.229
    It is possible that some individuals would like to reduce their exposure,
despite the conclusion of the Committee that there is no consistent evidence for
an increased risk for tumours in the brain and other regions in the head associated
with mobile phone use. The Knowledge Platform Electromagnetic Fields
provides a number of suggestions for exposure reduction.179
Better focussed research
According to the Committee there still is a need for further, focused, research. A
large multinational prospective cohort study of mobile phone users (COSMOS)
has recently been started, but it will take many years before results are available.
Further results of the MOBI-KIDS case-control study investigating mobile
phone use and brain tumours in children are awaited. When necessary, the
Committee will report on new developments.
Conclusions and recommendations                                                      123
</pre>

====================================================================== Einde pagina 124 =================================================================

<br><br>====================================================================== Pagina 125 ======================================================================

<pre>124 Mobile phones and cancer</pre>

====================================================================== Einde pagina 125 =================================================================

<br><br>====================================================================== Pagina 126 ======================================================================

<pre>   References
1  Hardell L, Nasman A, Pahlson A, et al. Use of cellular telephones and the risk for brain tumours: A
   case-control study. Int J Oncol, 1999; 15(1): 113-116.
2  Baan R, Grosse Y, Lauby-Secretan B, et al. Carcinogenicity of radiofrequency electromagnetic fields.
   Lancet Oncol, 2011; 12(7): 624-626.
3  IARC - International Agency for Research on Cancer. IARC Monographs on the Evaluation of
   Carcinogenic Risks to Humans. Non-ionizing radiation, Part 2: Radiofrequency electromagnetic
   fields. Volume 102. Internet: http://monographs.iarc.fr/ENG/Monographs/vol102/index.php. Access
   date 23-4-2013.
4  Hardell L, Carlberg M, and Hansson Mild K. Epidemiological evidence for an association between
   use of wireless phones and tumor diseases. Pathophysiology, 2009; 16(2-3): 113-122.
5  Myung SK, Ju W, McDonnell DD, et al. Mobile phone use and risk of tumors: a meta-analysis. J Clin
   Oncol, 2009; 27(33): 5565-5572.
6  Kundi M. The controversy about a possible relationship between mobile phone use and cancer. Cien
   Saude Colet, 2010; 15(5): 2415-2430.
7  Kundi M. The controversy about a possible relationship between mobile phone use and cancer.
   Environ Health Perspect, 2009; 117(3): 316-324.
8  Levis AG, Minicuci N, Ricci P, et al. Mobile phones and head tumours. The discrepancies in cause-
   effect relationships in the epidemiological studies - how do they arise? Environ Health, 2011; 10: 59.
9  Dubey RB, Hanmandlu M, and Gupta SK. Risk of brain tumors from wireless phone use. J Comput
   Assist Tomogr, 2010; 34(6): 799-807.
10 Swerdlow AJ, Feychting M, Green AC, et al. Mobile phones, brain tumors, and the interphone study:
   where are we now? Environ Health Perspect, 2011; 119(11): 1534-1538.
   References                                                                                             125
</pre>

====================================================================== Einde pagina 126 =================================================================

<br><br>====================================================================== Pagina 127 ======================================================================

<pre>11  Repacholi MH, Lerchl A, Roosli M, et al. Systematic review of wireless phone use and brain cancer
    and other head tumors. Bioelectromagnetics, 2011.
12  Bradford Hill A. The environment and disease: association or causation? Proc R Soc Med, 1965; 58:
    295-300.
13  Boaz A, Ashby D, and Young K. Systematic reviews: what have they got to offer evidence based
    policy and practice? London: ESRC UK Centre for Evidence Based Policy and Practice, Queen Mary
    University of London, 2002.
14  Katrak P, Bialocerkowski AE, Massy-Westropp N, et al. A systematic review of the content of critical
    appraisal tools. BMC Med Res Methodol, 2004; 4: 22.
15  Kitchenham B. Procedures for performing systematic reviews. Keele, Staffs, UK: Keele Univerity,
    2004; Keele University TR/SE-0401; NICTA TR 0400011T.1.
16  Counsell C. Formulating questions and locating primary studies for inclusion in systematic reviews.
    Ann Intern Med, 1997; 127(5): 380-387.
17  Meade MO and Richardson WS. Selecting and appraising studies for a systematic review. Ann Intern
    Med, 1997; 127(7): 531-537.
18  van Leeuwen FE, Alers JC, Vlems FA, et al. De rol van lichaamsbeweging bij preventie van kanker.
    KWF Kankerbestrijding, 2005.
19  Vlaanderen J, Vermeulen R, Heederik D, et al. Guidelines to evaluate human observational studies
    for quantitative risk assessment. Environ Health Perspect, 2008; 116(12): 1700-1705.
20  Voskuil DW, Monninkhof EM, Elias SG, et al. Physical activity and endometrial cancer risk, a
    systematic review of current evidence. Cancer Epidemiol Biomarkers Prev, 2007; 16(4): 639-648.
21  van der Windt DAWM, Zeegers MPA, and Scholten RJPM. Systematische reviews van
    observationeel onderzoek. In: Inleiding in evidence-based medicine - klinisch handelen gebaseerd op
    bewijsmateriaal, Offringa M, Assendelft WJJ, and Scholten RJPM, Eds. Houten: Bohn Staafleu van
    Loghem, 2003.
22  Bondy ML and Wrensch MR. Epidemiology of primary malignant brain tumours. Baillieres Clin
    Neurol, 1996; 5(2): 251-270.
23  Carozza SE, Wrensch M, Miike R, et al. Occupation and adult gliomas. Am J Epidemiol, 2000;
    152(9): 838-846.
24  Chen P, Aldape K, Wiencke JK, et al. Ethnicity delineates different genetic pathways in malignant
    glioma. Cancer Res, 2001; 61(10): 3949-3954.
25  Connelly JM and Malkin MG. Environmental risk factors for brain tumors. Curr Neurol Neurosci
    Rep, 2007; 7(3): 208-214.
26  Wünsch Filho V. The epidemiology of laryngeal cancer in Brazil. Sao Paulo Med J, 2004; 122(5):
    188-194.
27  Hoffman S, Propp JM, and McCarthy BJ. Temporal trends in incidence of primary brain tumors in
    the United States, 1985-1999. Neuro Oncol, 2006; 8(1): 27-37.
28  Kademani D. Oral cancer. Mayo Clin Proc, 2007; 82(7): 878-887.
126 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 127 =================================================================

<br><br>====================================================================== Pagina 128 ======================================================================

<pre>29 Marur S and Forastiere AA. Head and neck cancer: changing epidemiology, diagnosis, and treatment.
   Mayo Clin Proc, 2008; 83(4): 489-501.
30 McKinney PA, Parslow RC, Lane SA, et al. Epidemiology of childhood brain tumours in Yorkshire,
   UK, 1974-95: geographical distribution and changing patterns of occurrence. Br J Cancer, 1998;
   78(7): 974-979.
31 McKinney PA. Brain tumours: incidence, survival, and aetiology. J Neurol Neurosurg Psychiatry,
   2004; 75 Suppl 2: ii12-ii17.
32 Ohgaki H and Kleihues P. Epidemiology and etiology of gliomas. Acta Neuropathol, 2005; 109(1):
   93-108.
33 Ohgaki H. Epidemiology of brain tumors. Methods Mol Biol, 2009; 472: 323-342.
34 Propp JM, McCarthy BJ, Davis FG, et al. Descriptive epidemiology of vestibular schwannomas.
   Neuro Oncol, 2006; 8(1): 1-11.
35 Sturgis EM and Cinciripini PM. Trends in head and neck cancer incidence in relation to smoking
   prevalence: an emerging epidemic of human papillomavirus-associated cancers? Cancer, 2007;
   110(7): 1429-1435.
36 Wrensch M, Minn Y, Chew T, et al. Epidemiology of primary brain tumors: current concepts and
   review of the literature. Neuro Oncol, 2002; 4(4): 278-299.
37 Deeks JJ, Dinnes J, D'Amico R, et al. Evaluating non-randomised intervention studies. Health
   Technol Assess, 2003; 7(27): iii-173.
38 Katrak P, Bialocerkowski AE, Massy-Westropp N, et al. A systematic review of the content of critical
   appraisal tools. BMC Med Res Methodol, 2004; 4: 22.
39 Sanderson S, Tatt ID, and Higgins JP. Tools for assessing quality and susceptibility to bias in
   observational studies in epidemiology: a systematic review and annotated bibliography. Int J
   Epidemiol, 2007; 36(3): 666-676.
40 Greenhalgh T. Assessing the methodological quality of published papers. BMJ, 1997; 315(7103):
   305-308.
41 Greenland S. Invited commentary: a critical look at some popular meta-analytic methods. Am J
   Epidemiol, 1994; 140(3): 290-296.
42 Greenland S and O'Rourke K. On the bias produced by quality scores in meta-analysis, and a
   hierarchical view of proposed solutions. Biostatistics, 2001; 2(4): 463-471.
43 Juni P, Witschi A, Bloch R, et al. The hazards of scoring the quality of clinical trials for meta-
   analysis. JAMA, 1999; 282(11): 1054-1060.
44 Stang A. Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of
   nonrandomized studies in meta-analyses. Eur J Epidemiol, 2010; 25(9): 603-605.
45 Monninkhof EM, Elias SG, Vlems FA, et al. Physical activity and breast cancer: a systematic review.
   Epidemiology, 2007; 18(1): 137-157.
46 Vrijheid M, Armstrong BK, Bedard D, et al. Recall bias in the assessment of exposure to mobile
   phones. J Expo Sci Environ Epidemiol, 2009; 19(4): 369-381.
   References                                                                                           127
</pre>

====================================================================== Einde pagina 128 =================================================================

<br><br>====================================================================== Pagina 129 ======================================================================

<pre>47  Dreyer NA, Loughlin JE, and Rothman KJ. Cause-specific mortality in cellular telephone users.
    JAMA, 1999; 282(19): 1814-1816.
48  Johansen C, Boice J, Jr., McLaughlin J, et al. Cellular telephones and cancer--a nationwide cohort
    study in Denmark. J Natl Cancer Inst, 2001; 93(3): 203-207.
49  Rothman KJ, Loughlin JE, Funch DP, et al. Overall mortality of cellular telephone customers.
    Epidemiology, 1996; 7(3): 303-305.
50  Schüz J, Jacobsen R, Olsen JH, et al. Cellular telephone use and cancer risk: update of a nationwide
    Danish cohort. J Natl Cancer Inst, 2006; 98(23): 1707-1713.
51  Schüz J, Steding-Jessen M, Hansen S, et al. Long-term mobile phone use and the risk of vestibular
    schwannoma: a Danish nationwide cohort study. Am J Epidemiol, 2011; 174(4): 416-422.
52  Frei P, Poulsen AH, Johansen C, et al. Use of mobile phones and risk of brain tumours: update of
    Danish cohort study. BMJ, 2011; 343: d6387.
53  Schuz J, Elliott P, Auvinen A, et al. An international prospective cohort study of mobile phone users
    and health (Cosmos): design considerations and enrolment. Cancer Epidemiol, 2011; 35(1): 37-43.
54  Hardell L, Hansson Mild K, Pahlson A, et al. Ionizing radiation, cellular telephones and the risk for
    brain tumours. Eur J Cancer Prev, 2001; 10(6): 523-529.
55  Hardell L, Hallquist A, Hansson Mild K, et al. Cellular and cordless telephones and the risk for brain
    tumours. Eur J Cancer Prev, 2002; 11(4): 377-386.
56  Hardell L, Hansson Mild K, and Carlberg M. Case-control study on the use of cellular and cordless
    phones and the risk for malignant brain tumours. Int J Radiat Biol, 2002; 78(10): 931-936.
57  Hardell L, Hansson Mild K, and Carlberg M. Further aspects on cellular and cordless telephones and
    brain tumours. Int J Oncol, 2003; 22(2): 399-407.
58  Hardell L, Hansson Mild K, Sandström M, et al. Vestibular schwannoma, tinnitus and cellular
    telephones. Neuroepidemiology, 2003; 22(2): 124-129.
59  Hardell L, Carlberg M, and Hansson Mild K. Use of cellular telephones and brain tumour risk in
    urban and rural areas. Occup Environ Med, 2005; 62(6): 390-394.
60  Hardell L, Carlberg M, and Hansson Mild K. Case-control study on cellular and cordless telephones
    and the risk for acoustic neuroma or meningioma in patients diagnosed 2000-2003.
    Neuroepidemiology, 2005; 25(3): 120-128.
61  Hardell L, Carlberg M, and Hansson Mild K. Case-control study of the association between the use of
    cellular and cordless telephones and malignant brain tumors diagnosed during 2000-2003. Environ
    Res, 2006; 100(2): 232-241.
62  Hardell L, Carlberg M, and Hansson Mild K. Mobile phone use and the risk for malignant brain
    tumors: a case-control study on deceased cases and controls. Neuroepidemiology, 2010; 35(2): 109-
    114.
63  Hardell L, Carlberg M, and Hansson Mild K. Pooled analysis of two case-control studies on use of
    cellular and cordless telephones and the risk for malignant brain tumours diagnosed in 1997-2003. Int
    Arch Occup Environ Health, 2006; 79(8): 630-639.
128 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 129 =================================================================

<br><br>====================================================================== Pagina 130 ======================================================================

<pre>64 Hardell L, Carlberg M, and Hansson Mild K. Pooled analysis of two case-control studies on the use
   of cellular and cordless telephones and the risk of benign brain tumours diagnosed during 1997-2003.
   Int J Oncol, 2006; 28(2): 509-518.
65 Hansson Mild K, Hardell L, and Carlberg M. Pooled analysis of two Swedish case-control studies on
   the use of mobile and cordless telephones and the risk of brain tumours diagnosed during 1997-2003.
   Int J Occup Saf Ergon, 2007; 13(1): 63-71.
66 Hardell L and Carlberg M. Mobile phones, cordless phones and the risk for brain tumours. Int J
   Oncol, 2009; 35(1): 5-17.
67 Hardell L, Mild KH, Carlberg M, et al. Tumour risk associated with use of cellular telephones or
   cordless desktop telephones. World J Surg Oncol, 2006; 4: 74.
68 Hardell L, Carlberg M, and Hansson Mild K. Pooled analysis of case-control studies on malignant
   brain tumours and the use of mobile and cordless phones including living and deceased subjects. Int J
   Oncol, 2011; 38(5): 1465-1474.
69 Hardell L, Hallquist A, Hansson Mild K, et al. No association between the use of cellular or cordless
   telephones and salivary gland tumours. Occup Environ Med, 2004; 61(8): 675-679.
70 Hardell L, Carlberg M, Ohlson CG, et al. Use of cellular and cordless telephones and risk of testicular
   cancer. Int J Androl, 2007; 30(2): 115-122.
71 Hardell L, Hansson Mild K, Carlberg M, et al. Cellular and cordless telephone use and the association
   with brain tumors in different age groups. Arch Environ Health, 2004; 59(3): 132-137.
72 Söderqvist F, Carlberg M, and Hardell L. Use of wireless phones and the risk of salivary gland
   tumours: a case-control study. Eur J Cancer Prev, 2012.
73 Cardis E and Kilkenny M. International case-control study of adult brain, head and neck tumours:
   results of the feasibility study. Radiat Protect Dosimetry, 1999; 83: 179-183.
74 Cardis E, Richardson L, Deltour I, et al. The INTERPHONE study: design, epidemiological methods,
   and description of the study population. Eur J Epidemiol, 2007; 22(9): 647-664.
75 Christensen HC, Schüz J, Kosteljanetz M, et al. Cellular telephone use and risk of acoustic neuroma.
   Am J Epidemiol, 2004; 159(3): 277-283.
76 Lönn S, Ahlbom A, Hall P, et al. Mobile phone use and the risk of acoustic neuroma. Epidemiology,
   2004; 15(6): 653-659.
77 Christensen HC, Schüz J, Kosteljanetz M, et al. Cellular telephones and risk for brain tumors: a
   population-based, incident case-control study. Neurology, 2005; 64(7): 1189-1195.
78 Lönn S, Ahlbom A, Hall P, et al. Long-term mobile phone use and brain tumor risk. Am J Epidemiol,
   2005; 161(6): 526-535.
79 Klaeboe L, Blaasaas KG, and Tynes T. Use of mobile phones in Norway and risk of intracranial
   tumours. Eur J Cancer Prev, 2007; 16(2): 158-164.
80 Schüz J, Bohler E, Schlehofer B, et al. Radiofrequency electromagnetic fields emitted from base
   stations of DECT cordless phones and the risk of glioma and meningioma (Interphone Study Group,
   Germany). Radiat Res, 2006; 166(1 Pt 1): 116-119.
   References                                                                                              129
</pre>

====================================================================== Einde pagina 130 =================================================================

<br><br>====================================================================== Pagina 131 ======================================================================

<pre>81  Schüz J, Bohler E, Berg G, et al. Cellular phones, cordless phones, and the risks of glioma and
    meningioma (Interphone Study Group, Germany). Am J Epidemiol, 2006; 163(6): 512-520.
82  Lönn S, Ahlbom A, Christensen HC, et al. Mobile phone use and risk of parotid gland tumor. Am J
    Epidemiol, 2006; 164(7): 637-643.
83  Takebayashi T, Akiba S, Kikuchi Y, et al. Mobile phone use and acoustic neuroma risk in Japan.
    Occup Environ Med, 2006; 63(12): 802-807.
84  Hepworth SJ, Schoemaker MJ, Muir KR, et al. Mobile phone use and risk of glioma in adults: case-
    control study. BMJ, 2006; 332(7546): 883-887.
85  Sadetzki S, Chetrit A, Jarus-Hakak A, et al. Cellular phone use and risk of benign and malignant
    parotid gland tumors--a nationwide case-control study. Am J Epidemiol, 2008; 167(4): 457-467.
86  Schlehofer B, Schlaefer K, Blettner M, et al. Environmental risk factors for sporadic acoustic
    neuroma (Interphone Study Group, Germany). Eur J Cancer, 2007; 43(11): 1741-1747.
87  Hours M, Bernard M, Montestrucq L, et al. [Cell Phones and Risk of brain and acoustic nerve
    tumours: the French INTERPHONE case-control study.]. Rev Epidemiol Sante Publique, 2007.
88  Takebayashi T, Varsier N, Kikuchi Y, et al. Mobile phone use, exposure to radiofrequency
    electromagnetic field, and brain tumour: a case-control study. Br J Cancer, 2008; 98(3): 652-659.
89  Schoemaker MJ and Swerdlow AJ. Risk of pituitary tumors in cellular phone users: a case-control
    study. Epidemiology, 2009; 20(3): 348-354.
90  Schoemaker MJ, Swerdlow AJ, Ahlbom A, et al. Mobile phone use and risk of acoustic neuroma:
    results of the Interphone case-control study in five North European countries. Br J Cancer, 2005;
    93(7): 842-848.
91  Lahkola A, Auvinen A, Raitanen J, et al. Mobile phone use and risk of glioma in 5 North European
    countries. Int J Cancer, 2007; 120(8): 1769-1775.
92  Lahkola A, Salminen T, Raitanen J, et al. Meningioma and mobile phone use--a collaborative case-
    control study in five North European countries. Int J Epidemiol, 2008; 37(6): 1304-1313.
93  INTERPHONE study group. Brain tumour risk in relation to mobile telephone use: results of the
    INTERPHONE international case-control study. Int J Epidemiol, 2010; 39(3): 675-694.
94  INTERPHONE study group. Acoustic neuroma risk in relation to mobile telephone use: Results of
    the INTERPHONE international case-control study. Cancer Epidemiol, 2011.
95  Inskip PD, Hatch EE, Stewart PA, et al. Study design for a case-control investigation of cellular
    telephones and other risk factors for brain tumour in adults. Radiat Prot Dosimet, 1999; 86: 45-52.
96  Muscat JE, Malkin MG, Thompson S, et al. Handheld cellular telephone use and risk of brain cancer.
    JAMA, 2000; 284(23): 3001-3007.
97  De Roos AJ, Teschke K, Savitz DA, et al. Parental occupational exposures to electromagnetic fields
    and radiation and the incidence of neuroblastoma in offspring. Epidemiology, 2001; 12(5): 508-517.
98  Stang A, Anastassiou G, Ahrens W, et al. The possible role of radofrequency radiation in the
    development of uveal melanoma. Epidemiology, 2001; 12: 7-12.
99  Inskip PD, Tarone RE, Hatch EE, et al. Cellular-telephone use and brain tumors. N Engl J Med, 2001;
    344(2): 79-86.
130 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 131 =================================================================

<br><br>====================================================================== Pagina 132 ======================================================================

<pre>100 Auvinen A, Hietanen M, Luukkonen R, et al. Brain tumors and salivary gland cancers among cellular
    telephone users. Epidemiology, 2002; 13(3): 356-359.
101 Muscat JE, Malkin MG, Shore RE, et al. Handheld cellular telephones and risk of acoustic neuroma.
    Neurology, 2002; 58(8): 1304-1306.
102 Warren HG, Prevatt AA, Daly KA, et al. Cellular telephone use and risk of intratemporal facial nerve
    tumor. Laryngoscope, 2003; 113(4): 663-667.
103 Gousias K, Markou M, Voulgaris S, et al. Descriptive epidemiology of cerebral gliomas in Northwest
    Greece and study of potential predisposing factors, 2005-2007. Neuroepidemiology, 2009; 33(2): 89-
    95.
104 Stang A, Schmidt-Pokrzywniak A, Lash TL, et al. Mobile phone use and risk of uveal melanoma:
    results of the risk factors for uveal melanoma case-control study. J Natl Cancer Inst, 2009; 101(2):
    120-123.
105 Spinelli V, Chinot O, Cabaniols C, et al. Occupational and environmental risk factors for brain
    cancer: a pilot case-control study in France. Presse Med, 2010; 39(2): e35-e44.
106 Duan Y, Zhang HZ, and Bu RF. Correlation between cellular phone use and epithelial parotid gland
    malignancies. Int J Oral Maxillofac Surg, 2011; 40(9): 966-972.
107 Baldi I, Coureau G, Jaffre A, et al. Occupational and residential exposure to electromagnetic fields
    and risk of brain tumors in adults: a case-control study in Gironde, France. Int J Cancer, 2011; 129(6):
    1477-1484.
108 Aydin D, Feychting M, Schüz J, et al. Mobile phone use and brain tumors in children and
    adolescents: a multicenter case-control study. J Natl Cancer Inst, 2011; 103(16): 1264-1276.
109 Ali Kahn A, O'Brien DF, Kelly P, et al. The anatomical distribution of cerebral gliomas in mobile
    phone users. Ir Med J, 2003; 96(8): 240-242.
110 Salahaldin AH and Bener A. Long-term and frequent cellular phone use and risk of acoustic
    neuroma. Int Tinnitus J, 2006; 12(2): 145-148.
111 Hartikka H, Heinavaara S, Mantyla R, et al. Mobile phone use and location of glioma: a case-case
    analysis. Bioelectromagnetics, 2009; 30(3): 176-182.
112 Sato Y, Akiba S, Kubo O, et al. A case-case study of mobile phone use and acoustic neuroma risk in
    Japan. Bioelectromagnetics, 2010.
113 Larjavaara S, Schüz J, Swerdlow A, et al. Location of gliomas in relation to mobile telephone use: a
    case-case and case-specular analysis. Am J Epidemiol, 2011; 174(1): 2-11.
114 Cardis E, Armstrong BK, Bowman JD, et al. Risk of brain tumours in relation to estimated RF dose
    from mobile phones: results from five Interphone countries. Occup Environ Med, 2011.
115 Cook A, Woodward A, Pearce N, et al. Cellular telephone use and time trends for brain, head and
    neck tumours. N Z Med J, 2003; 116(1175): U457.
116 Howitz MF, Johansen C, Tos M, et al. Incidence of vestibular schwannoma in Denmark, 1977-1995.
    Am J Otol, 2000; 21(5): 690-694.
117 Inskip PD, Devesa SS, and Fraumeni JF, Jr. Trends in the incidence of ocular melanoma in the United
    States, 1974-1998. Cancer Causes Control, 2003; 14(3): 251-257.
    References                                                                                               131
</pre>

====================================================================== Einde pagina 132 =================================================================

<br><br>====================================================================== Pagina 133 ======================================================================

<pre>118 Johansen C, Boice JD, Jr., McLaughlin JK, et al. Mobile phones and malignant melanoma of the eye.
    Br J Cancer, 2002; 86(3): 348-349.
119 Lönn S, Klaeboe L, Hall P, et al. Incidence trends of adult primary intracerebral tumors in four
    Nordic countries. Int J Cancer, 2004; 108(3): 450-455.
120 Muscat JE, Hinsvark M, and Malkin M. Mobile telephones and rates of brain cancer.
    Neuroepidemiology, 2006; 27(1): 55-56.
121 Röösli M, Michel G, Kuehni CE, et al. Cellular telephone use and time trends in brain tumour
    mortality in Switzerland from 1969 to 2002. Eur J Cancer Prev, 2007; 16(1): 77-82.
122 Deltour I, Johansen C, Auvinen A, et al. Time trends in brain tumor incidence rates in Denmark,
    Finland, Norway, and Sweden, 1974-2003. J Natl Cancer Inst, 2009; 101(24): 1721-1724.
123 Inskip PD, Hoover RN, and Devesa SS. Brain cancer incidence trends in relation to cellular telephone
    use in the United States. Neuro Oncol, 2010; 12(11): 1147-1151.
124 Lehrer S, Green S, and Stock RG. Association between number of cell phone contracts and brain
    tumor incidence in nineteen U.S. States. J Neurooncol, 2010.
125 Hardell L, Carlberg M, Söderqvist F, et al. Re: Time trends in brain tumor incidence rates in
    Denmark, Finland, Norway, and Sweden, 1974-2003. J Natl Cancer Inst, 2010; 102(10): 740-741.
126 Counsell CE, Collie DA, and Grant R. Incidence of intracranial tumours in the Lothian region of
    Scotland, 1989-90. J Neurol Neurosurg Psychiatry, 1996; 61(2): 143-150.
127 Gurney JG and Kadan-Lottick N. Brain and other central nervous system tumors: rates, trends, and
    epidemiology. Curr Opin Oncol, 2001; 13(3): 160-166.
128 Nelson PD, Toledano MB, McConville J, et al. Trends in acoustic neuroma and cellular phones: is
    there a link? Neurology, 2006; 66(2): 284-285.
129 Czerninski R, Zini A, and Sgan-Cohen HD. Risk of parotid malignant tumors in Israel (1970-2006).
    Epidemiology, 2011; 22(1): 130-131.
130 de Vocht F, Burstyn I, and Cherrie JW. Time trends (1998-2007) in brain cancer incidence rates in
    relation to mobile phone use in England. Bioelectromagnetics, 2011; DOI 10.1002/bem.20648.
131 Kohler BA, Ward E, McCarthy BJ, et al. Annual report to the nation on the status of cancer, 1975-
    2007, featuring tumors of the brain and other nervous system. J Natl Cancer Inst, 2011; 103(9): 714-
    736.
132 Larjavaara S, Feychting M, Sankila R, et al. Incidence trends of vestibular schwannomas in
    Denmark, Finland, Norway and Sweden in 1987-2007. Br J Cancer, 2011; 105(7): 1069-1075.
133 Little MP, Rajaraman P, Curtis RE, et al. Mobile phone use and glioma risk: comparison of
    epidemiological study results with incidence trends in the United States. BMJ, 2012; 344: e1147.
134 de Vocht F. Cell phones and parotid cancer trends in England. Epidemiology, 2011; 22(4): 608-609.
135 Deltour I, Auvinen A, Feychting M, et al. Mobile phone use and incidence of glioma in the Nordic
    countries 1979-2008: Consistency check. Epidemiology, 2012; 23(2): 301-307.
136 Rothman KJ, Greenland S, and Lash TL. Modern epidemiology. 3rd. Philidelphia: Lippincott,
    Williams & Wilkins, 2008.
132 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 133 =================================================================

<br><br>====================================================================== Pagina 134 ======================================================================

<pre>137 Ahlbom A and Feychting M. Re: Use of cellular phones and the risk of brain tumours: a case-control
    study. Int J Oncol, 1999; 15(5): 1045-1047.
138 Boice JD and McLaughlin JK. Epidemiologic studies of cellular telephones and cancer risk.
    Stockholm: Swedish Radiaition Protection Authority, 2002; 2002:16.
139 Michaëlsson K, Baron JA, Farahmand BY, et al. Influence of parity and lactation on hip fracture risk.
    Am J Epidemiol, 2001; 153(12): 1166-1172.
140 Riman T, Dickman PW, Nilsson S, et al. Risk factors for invasive epithelial ovarian cancer: results
    from a Swedish case-control study. Am J Epidemiol, 2002; 156(4): 363-373.
141 Olsson AR, Skogh T, and Wingren G. Occupational determinants for rheumatoid arthritis. Scand J
    Work Environ Health, 2000; 26(3): 243-249.
142 Spångéus A, El-Salhy M, Suhr O, et al. Prevalence of gastrointestinal symptoms in young and
    middle-aged diabetic patients. Scand J Gastroenterol, 1999; 34(12): 1196-1202.
143 Vrijheid M, Richardson L, Armstrong BK, et al. Quantifying the impact of selection bias caused by
    nonparticipation in a case-control study of mobile phone use. Ann Epidemiol, 2009; 19(1): 33-41.
144 Vrijheid M, Mann S, Vecchia P, et al. Determinants of mobile phone output power in a multinational
    study: implications for exposure assessment. Occup Environ Med, 2009; 66(10): 664-671.
145 Vrijheid M, Deltour I, Krewski D, et al. The effects of recall errors and of selection bias in
    epidemiologic studies of mobile phone use and cancer risk. J Expo Sci Environ Epidemiol, 2006;
    16(4): 371-384.
146 Vrijheid M, Cardis E, Armstrong BK, et al. Validation of short term recall of mobile phone use for the
    Interphone study. Occup Environ Med, 2006; 63(4): 237-243.
147 Cardis E, Varsier N, Bowman JD, et al. Estimation of RF energy absorbed in the brain from mobile
    phones in the Interphone Study. Occup Environ Med, 2011.
148 Berg G, Schuz J, Samkange-Zeeb F, et al. Assessment of radiofrequency exposure from cellular
    telephone daily use in an epidemiological study: German Validation study of the international case-
    control study of cancers of the brain--INTERPHONE-Study. J Expo Anal Environ Epidemiol, 2005;
    15(3): 217-224.
149 Samkange-Zeeb F, Berg G, and Blettner M. Validation of self-reported cellular phone use. J Expo
    Anal Environ Epidemiol, 2004; 14(3): 245-248.
150 Lahkola A, Salminen T, and Auvinen A. Selection bias due to differential participation in a case-
    control study of mobile phone use and brain tumors. Ann Epidemiol, 2005; 15(5): 321-325.
151 Ahlbom A and Feychting M. Mobile telephones and brain tumours. BMJ, 2011; 343: d6605.
152 Neubauer G, Röösli M, Feychting M, et al. Study on the feasibility of epidemiological studies on
    health effects of mobile telephone base stations - final report. Seibersdorf: ARC Seibersdorf Research
    GmBH, 2005; report nr ARC-IT-0124.
153 Leena K, Tomi L, and Arja RR. Intensity of mobile phone use and health compromising behaviours--
    how is information and communication technology connected to health-related lifestyle in
    adolescence? J Adolesc, 2005; 28(1): 35-47.
154 Schüz J. Mobile phone use and exposures in children. Bioelectromagnetics, 2005; Suppl 7: S45-S50.
    References                                                                                             133
</pre>

====================================================================== Einde pagina 134 =================================================================

<br><br>====================================================================== Pagina 135 ======================================================================

<pre>155 Söderqvist F, Hardell L, Carlberg M, et al. Ownership and use of wireless telephones: a population-
    based study of Swedish children aged 7-14 years. BMC Public Health, 2007; 7: 105.
156 Söderqvist F, Carlberg M, and Hardell L. Use of wireless telephones and self-reported health
    symptoms: a population-based study among Swedish adolescents aged 15-19 years. Environ Health,
    2008; 7: 18.
157 Söderqvist F, Carlberg M, and Hardell L. Mobile and cordless telephones, serum transthyretin and the
    blood-cerebrospinal fluid barrier: a cross-sectional study. Environ Health, 2009; 8: 19.
158 Söderqvist F, Carlberg M, and Hardell L. Use of wireless telephones and serum S100B levels: a
    descriptive cross-sectional study among healthy Swedish adults aged 18-65 years. Sci Total Environ,
    2009; 407(2): 798-805.
159 Schüz J and Johansen C. A comparison of self-reported cellular telephone use with subscriber data:
    agreement between the two methods and implications for risk estimation. Bioelectromagnetics, 2007;
    28(2): 130-136.
160 Schüz J. Lost in laterality: interpreting ''preferred side of the head during mobile phone use and risk
    of brain tumour'' associations. Scand J Public Health, 2009; 37(6): 664-667.
161 Enchev Y, Ferdinandov D, Kounin G, et al. Radiation-induced gliomas following radiotherapy for
    craniopharyngiomas: a case report and review of the literature. Clin Neurol Neurosurg, 2009; 111(7):
    591-596.
162 Salvati M, Frati A, Russo N, et al. Radiation-induced gliomas: report of 10 cases and review of the
    literature. Surg Neurol, 2003; 60(1): 60-67.
163 Paulino AC, Ahmed IM, Mai WY, et al. The influence of pretreatment characteristics and
    radiotherapy parameters on time interval to development of radiation-associated meningioma. Int J
    Radiat Oncol Biol Phys, 2009; 75(5): 1408-1414.
164 Schneider AB, Ron E, Lubin J, et al. Acoustic neuromas following childhood radiation treatment for
    benign conditions of the head and neck. Neuro Oncol, 2008; 10(1): 73-78.
165 Hardell L, Carlberg M, and Hansson Mild K. Re-analysis of risk for glioma in relation to mobile
    telephone use: comparison with the results of the Interphone international case-control study. Int J
    Epidemiol, 2011; 40(4): 1126-1128.
166 Inyang I, Benke G, McKenzie R, et al. A new method to determine laterality of mobile telephone use
    in adolescents. Occup Environ Med, 2010; 67(8): 507-512.
167 Hardell L, Carlberg M, and Hansson MK. Pooled analysis of case-control studies on malignant brain
    tumours and the use of mobile and cordless phones including living and deceased subjects. Int J
    Oncol, 2011; 38(5): 1465-1474.
168 What is SAR? Internet: http://www.mobile-phones-uk.org.uk/sar.htm. Access date 30-1-2012.
169 Mobile radiation - SAR value - Specific Absorption Rate. Internet: http://www.sardatabase.com/.
    Access date 30-1-2012.
170 Nokia. SAR information. Internet: http://sar.nokia.com/sar/index.jsp. Access date 30-1-2012.
171 Kramer A, Kühn S, Lott U, et al. Development of procedures for the assessment of human exposure
    to EMF from wireless devices in home and office environments. Zürich: IT'IS Foundation, 2005.
134 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 135 =================================================================

<br><br>====================================================================== Pagina 136 ======================================================================

<pre>172 Redmayne M, Inyang I, Dimitriadis C, et al. Cordless telephone use: implications for mobile phone
    research. J Environ Monit, 2010; 12(4): 809-812.
173 Kelsh MA, Shum M, Sheppard AR, et al. Measured radiofrequency exposure during various mobile-
    phone use scenarios. J Expo Sci Environ Epidemiol, 2011; 21(4): 343-354.
174 Schuz J, Waldemar G, Olsen JH, et al. Risks for central nervous system diseases among mobile phone
    subscribers: a Danish retrospective cohort study. PLoS One, 2009; 4(2): e4389.
175 Cardis E, Deltour I, Mann S, et al. Distribution of RF energy emitted by mobile phones in anatomical
    structures of the brain. Phys Med Biol, 2008; 53(11): 2771-2783.
176 Vlaanderen J, Vermeulen R, Heederik D, et al. Guidelines to evaluate human observational studies
    for quantitative risk assessment. Environ Health Perspect, 2008; 116(12): 1700-1705.
177 AGNIR - Advisory Group on Non-ionising Radiation. Health effects from radiofrequency
    electromagnetic fields. Documents of the Health Protection Agency, 2012.
178 SSM - Swedish Radiation Safety Authority - Independent Group of Experts. Recent research on EMF
    and health risk. Seventh annual report from SSM:s Independent Expert Group on Electromagnetic
    Fields, 2010. Stockholm: Swedish Radiation Safety Authority, 2011; SSM Report 2010:44.
179 Kennisplatform Elektromagnetische velden. Anders omgaan met mobiele telefoons. Internet:
    http://www.kennisplatform.nl/Onderwerpen/Mobieletelefoonsenzendmasten/omgaan-met-mobiele-
    telefoon.aspx. Access date 18-9-2012.
180 Greenhalgh T and Peacock R. Effectiveness and efficiency of search methods in systematic reviews
    of complex evidence: audit of primary sources. BMJ, 2005; 331(7524): 1064-1065.
181 Hardell L and Hansson Mild K. Re: Cellular telephones and cancer--a nationwide cohort study in
    Denmark. J Natl Cancer Inst, 2001; 93(12): 952-953.
182 Ahlbom A, Feychting M, Cardis E, et al. Re: Cellular telephone use and cancer risk: update of a
    nationwide Danish cohort study. J Natl Cancer Inst, 2007; 99(8): 655-656.
183 Charlier, P. Not enough data excluding cellphones' morbidity. Internet: http://www.bmj.com/content/
    343/bmj.d6387?page=1&tab=responses. Access date 19-1-2012.
184 Henshaw, DL. Mobile phone radiation could be detected by the human brain. Internet:
    http://www.bmj.com/content/343/bmj.d6387?tab=responses. Access date 19-1-2012.
185 Gujral, DM. Use of mobile phones and risk of brain tumours: update of Danish cohort study. Internet:
    http://www.bmj.com/content/343/bmj.d6387?tab=responses. Access date 19-1-2012.
186 Davis, DL. Re:Not enough data excluding cellphones' morbidity. Internet: http://www.bmj.com/
    content/343/bmj.d6387?tab=responses. Access date 19-1-2012.
187 Morgan, LL. The Danish cellphone subscriber study on the risk of cancer among subscribers is
    fundamentally flawed. Internet: http://www.bmj.com/content/343/bmj.d6387?tab=responses. Access
    date 19-1-2012.
188 Frey, AH. On the Safety of Cell Phone Radiation. Internet: http://www.bmj.com/content/343/
    bmj.d6387?tab=responses. Access date 19-1-2012.
189 Leszczynski, D. Re: Use of mobile phones and risk of brain tumours: update of Danish cohort study.
    Internet: http://www.bmj.com/content/343/bmj.d6387?tab=responses. Access date 19-1-2012.
    References                                                                                           135
</pre>

====================================================================== Einde pagina 136 =================================================================

<br><br>====================================================================== Pagina 137 ======================================================================

<pre>190 Glaser, MM. Re: Use of mobile phones and risk of brain tumours: update of Danish cohort study.
    Internet: http://www.bmj.com/content/343/bmj.d6387?tab=responses. Access date 19-1-2012.
191 Khurana VG. Questions about selection, exposure, and tumour incidence. BMJ, 2011; 343: d7893.
192 Philips A and Lamburn G. Updated study contains poor science and should be disregarded. BMJ,
    2011; 343: d7899.
193 Frei P, Poulsen AH, Johansen C, et al. Authors' reply to Khurana and to Philips and Lamburn. BMJ,
    2011; 343: d7912.
194 Kundi M. Failure to detect a link between mobile phone use and brain tumours in a large Danish
    cohort study: but findings may be due to bias. Evid Based Med, 2012.
195 Boice JD and McLaughlin JK. Concerning mobile phone use and risk of acoustic neuroma. Br J
    Cancer, 2006; 95(1): 130.
196 Hardell L and Hansson Mild K. Mobile phone use and risk of acoustic neuroma: results of the
    interphone case-control study in five North European countries. Br J Cancer, 2006; 94(9): 1348-1349.
197 Hansson Mild K, Carlberg M, Wilen J, et al. How to combine the use of different mobile and cordless
    telephones in epidemiological studies on brain tumours? Eur J Cancer Prev, 2005; 14(3): 285-288.
198 Tarone RE and Inskip PD. Mobile phone use and acoustic neuromas. Epidemiology, 2005; 16(3):
    414-418.
199 Stang A, Schmidt-Pokrzywniak A, and Jockel KH. Mobile phone use and acoustic neuromas.
    Epidemiology, 2005; 16(3): 414-415.
200 Hardell L and Hansson Mild K. Mobile phone use and acoustic neuromas. Epidemiology, 2005;
    16(3): 415-418.
201 Thomas BN, Flowers D, Caswell J, et al. Mobile phone use and acoustic neuromas. Epidemiology,
    2005; 16(3): 415-416.
202 Johnston SA and Scherb H. Mobile phone use and acoustic neuromas. Epidemiology, 2005; 16(3):
    416-417.
203 Savitz DA. Mixed signals on cell phones and cancer. Epidemiology, 2004; 15(6): 651-652.
204 Hardell L and Hansson Mild K. Re: "cellular telephone use and risk of acoustic neuroma". Am J
    Epidemiol, 2004; 160(9): 923-925.
205 Kundi M. Re: "cellular telephone use and risk of acoustic neuroma". Am J Epidemiol, 2004; 160(9):
    923-924.
206 Gale BD and Juran D. Cellular telephones and risk for brain tumors: a population-based, incident
    case-control study. Neurology, 2006; 66(5): 781.
207 Hardell L, Hansson Mild K, and Kundi M. Re: "Long-term mobile phone use and brain tumor risk".
    Am J Epidemiol, 2005; 162(6): 600-601.
208 Milham S. Re: "Long-term mobile phone use and brain tumor risk". Am J Epidemiol, 2005; 162(6):
    599.
209 Morgan LL. Re: "Cellular phones, cordless phones, and the risks of glioma and meningioma
    (Interphone Study Group, Germany)". Am J Epidemiol, 2006; 164(3): 294-295.
210 Schüz J, Böhler E, Berg G, et al. The authors reply. Am J Epidemiol, 2006; 164: 295.
136 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 137 =================================================================

<br><br>====================================================================== Pagina 138 ======================================================================

<pre>211 Hardell L and Hansson Mild K. Mobile phone use and risk of glioma in adults: results are difficult to
    interpret because of limitations. BMJ, 2006; 332(7548): 1035.
212 Kundi M. Mobile phone use and risk of glioma in adults: conclusions are questionable. BMJ, 2006;
    332(7548): 1035-1036.
213 Maier M. Brains and mobile phones. BMJ, 2006; 332(7546): 864-865.
214 Morgan LL. Mobile phone use and risk of glioma in adults: study has many flaws. BMJ, 2006;
    332(7548): 1035.
215 Hocking B. Japanese mobile phone study. Br J Cancer, 2008; 98(11): 1879.
216 Hocking B. Mobile phone use and risk of acoustic neuroma. Br J Cancer, 2006; 94(9): 1350-1353.
217 Milham S. Mobile phone use and risk of acoustic neuroma: results of the interphone case-control
    study in five north European countries [corrected]. Br J Cancer, 2006; 94(9): 1351-1353.
218 Schoemaker MJ, Swerdlow AJ, Auvinen A, et al. Reply: Mobile phone use and risk of acoustic
    neuroma: results of the Interphone case-control study in five North European countries. Br J Cancer,
    2006; 94(9): 1352-1353.
219 Noone P. Cancers and mobile phone use. Occup Med (Lond ), 2009; 59(4): 286-287.
220 Milham S. Meningioma and mobile phone use. Int J Epidemiol, 2010; 39(4): 1117.
221 Morgan LL. Reader's response: meningioma and mobile phone use--a collaborative case-control
    study in five North European countries. Int J Epidemiol, 2010; 39(4): 1117-1118.
222 Saracci R and Samet J. Commentary: Call me on my mobile phone...or better not?--a look at the
    INTERPHONE study results. Int J Epidemiol, 2010; 39(3): 695-698.
223 Clouston SA. Social and economic patterning in the Interphone study. Int J Epidemiol, 2011; 40(4):
    1122.
224 Behrens T, Terschuren C, and Hoffmann W. Limitations of interview-based risk assessment of RF
    exposure from appliances. Arch Environ Health, 2004; 59(6): 292-299.
225 Schoemaker MJ, Swerdlow AJ, Auvinen A, et al. Medical history, cigarette smoking and risk of
    acoustic neuroma: an international case-control study. Int J Cancer, 2006; 120(1): 103-110.
226 Berg G, Spallek J, Schuz J, et al. Occupational exposure to radio frequency/microwave radiation and
    the risk of brain tumors: Interphone Study Group, Germany. Am J Epidemiol, 2006; 164(6): 538-548.
227 Schmidt-Pokrzywniak A, Jockel KH, Bornfeld N, et al. Case-control study on uveal melanoma
    (RIFA): rational and design. BMC Ophthalmol, 2004; 4: 11.
228 Aydin D, Feychting M, Schüz J, et al. Impact of random and systematic recall errors and selection
    bias in case--control studies on mobile phone use and brain tumors in adolescents (CEFALO study).
    Bioelectromagnetics, 2011; 32(5): 396-407.
229 HCN - Health Council of the Netherlands. Prudent precaution. The Hague: Health Council of the
    Netherlands, 2008; publication nr 2008/18E.
230 Lönn S, Ahlbom A, Hall P, et al. (Authors reply to LTEs Tarone & Inskip, Stang et al., Hardell &
    Hansson Mild, Thomas et al. and Johnston & Scherb). Epidemiology, 2005; 417-418.
    References                                                                                            137
</pre>

====================================================================== Einde pagina 138 =================================================================

<br><br>====================================================================== Pagina 139 ======================================================================

<pre>231 Auvinen A, Lahkola A, Feychting M, et al. Response to commentary: Meningioma and mobile phone
    use – a collaborative case – control study in five North European countries. Int J Epidemiol, 2010;
    39: 1119.
138 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 139 =================================================================

<br><br>====================================================================== Pagina 140 ======================================================================

<pre>A The Committee
B Search strategy and results
C Data extraction
D Evaluation of quality of the studies
E Additional information for the publications used
F Results of the data extraction
G Results of the evaluation of quality of the studies
H Results from the selected publications
I Meta-analysis and forest plots
  Annexes
                                                      139
</pre>

====================================================================== Einde pagina 140 =================================================================

<br><br>====================================================================== Pagina 141 ======================================================================

<pre>140 Mobile phones and cancer</pre>

====================================================================== Einde pagina 141 =================================================================

<br><br>====================================================================== Pagina 142 ======================================================================

<pre>Annex A
      The Committee
      The membership of the Electromagnetic Fields Committee at the time of
      preparation of this advisory report was as follows:
      • Prof. G.C. van Rhoon, chair
         Professor of Physical Aspects of Electromagnetic Fields and Health, Erasmus
         University Medical Centre Rotterdam
      • Prof. A. Aleman
         Professor of Cognitive Neuropsychiatry, University of Groningen
      • Prof. H. Kromhout
         Professor of Epidemiology of Health Effects from Exposure to
         Electromagnetic Fields, Institute for Risk Assessment Sciences, University
         of Utrecht
      • Prof. F.E. van Leeuwen
         Professor of Cancer Epidemiology, Free University of Amsterdam,
         Epidemiologist, Netherlands Cancer Institute, Amsterdam
      • Prof. H.F.J. Savelkoul
         Professor of Cell Biology and Immunology, Wageningen University
      • Prof. W.J. Wadman
         Professor of Neurobiology, University of Amsterdam
      • D.H.J. van de Weerdt, MD
         Toxicologist and Specialist in Environmental Medicine, Central Gelderland
         Municipal Health Services (GGD), Arnhem
      The Committee                                                                  141
</pre>

====================================================================== Einde pagina 142 =================================================================

<br><br>====================================================================== Pagina 143 ======================================================================

<pre>    •   Prof. A.P.M. Zwamborn
        Professor of Electromagnetic Fields and Health, Eindhoven University of
        Technology, physicist, TNO (Netherlands Organisation for Applied
        Scientific Research), The Hague
    •   Dr. G. Kelfkens, advisor
        Physicist, Netherlands Institute for Public Health and the Environment,
        Bilthoven
    •   R.M. van der Graaf, observer
        Executive Director, Knowledge Platform Electromagnetic fields, Bilthoven
    •   Prof. E. Lebret, observer
        Professor of Environmental Health Impact Assessment, Institute for Risk
        Assessment Sciences, Utrecht University, and Chairman Science forum,
        Knowledge Platform Electromagnetic Fields, Bilthoven
    •   Dr. H.K. Leonhard, observer
        Physicist, Ministry of Economic Affairs, Groningen
    •   Prof. I.A. Kreis, scientific secretary
        Epidemiologist and Specialist in Social Medicine, Health Council of the
        Netherlands, The Hague
    •   Dr. E. van Rongen, scientific secretary
        Radiobiologist, Health Council of the Netherlands, The Hague
    Dr. M.C. Cardous-Ubbink, epidemiologist, assisted in the extraction and scoring
    of the data, and dr. W.L.J. van Putten, statistician, assisted with the forest plots.
    The registration teams of the Comprehensive Cancer Centre Netherlands and
    Comprehensive Cancer Centre South collected the data for the Netherlands
    Cancer Registry and the scientific staff of the Comprehensive Cancer Centre
    Netherlands provided the analysis of the data.
    The Health Council and interests
    Members of Health Council Committees are appointed in a personal capacity
    because of their special expertise in the matters to be addressed. Nonetheless, it
    is precisely because of this expertise that they may also have interests. This in
    itself does not necessarily present an obstacle for membership of a Health
    Council Committee. Transparency regarding possible conflicts of interest is
    nonetheless important, both for the chairperson and members of a Committee
    and for the President of the Health Council. On being invited to join a
    Committee, members are asked to submit a form detailing the functions they
    hold and any other material and immaterial interests which could be relevant for
142 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 143 =================================================================

<br><br>====================================================================== Pagina 144 ======================================================================

<pre>the Committee’s work. It is the responsibility of the President of the Health
Council to assess whether the interests indicated constitute grounds for non-
appointment. An advisorship will then sometimes make it possible to exploit the
expertise of the specialist involved. During the inaugural meeting the
declarations issued are discussed, so that all members of the Committee are
aware of each other’s possible interests.
The Committee                                                                   143
</pre>

====================================================================== Einde pagina 144 =================================================================

<br><br>====================================================================== Pagina 145 ======================================================================

<pre>144 Mobile phones and cancer</pre>

====================================================================== Einde pagina 145 =================================================================

<br><br>====================================================================== Pagina 146 ======================================================================

<pre>Annex B
      Search strategy and results
      A search strategy consists of the keywords and databases used. For this
      systematic review, a comparison of several important databases was carried out
      and the publications identified were evaluated for the relevance of the topics
      identified.
      Keywords
      Both intuitively relevant terms and MeSH (Medical Subject Headings) in
      PubMed were used as keywords. For exposure, the MeSH terms were “cellular
      phone”, “radio frequencies” and “electromagnetic fields”, but “mobile phone”,
      “radio waves” and “cell phone” were also used. To assess the impact of different
      words for telephone both “phone” and “*phone” were tried and gave identical
      numbers of hits. The keyword “telephone” give substantially fewer hits and was
      taken as included in “*phone”. As outcome parameter the MeSH was
      “neoplasms”, but “tumour” and “cancer” were also used. These keywords
      individually resulted in different numbers of hits, therefore they were all
      included in the search strategy. For methodology “epidemiology” (a MeSH term)
      and “exposure assessment” plus “dosimetry” were added.
      Search strategy and results                                                      145
</pre>

====================================================================== Einde pagina 146 =================================================================

<br><br>====================================================================== Pagina 147 ======================================================================

<pre>    Databases
    Initially both PubMed Central and PubMed were searched and compared for the
    number of hits. Since PubMed appeared a broader database then PubMed
    Central, only PubMed was used.
    Searches
    Initial searches were performed in the week of 20 July 2009 and fully repeated
    on 15 August 2011. The results of the search from August 2011 are presented in
    tables B.1 and B.2, where the number of hits for the different keywords and
    combinations of keywords is given.
    Using the combinations of the search terms that were evaluated, a combined
    search was conducted. The combined search used the terms:
    cellular phone* OR mobile phone* OR cell phone* OR radio waves OR electromagnetic fields OR
    radio frequency AND human AND (tumour OR cancer OR neoplasms) AND (epidemiology OR
    dosimetry OR exposure assessment).
    There were no restrictions on years, language or any other placed on this search.
    This resulted in 2083 hits.
         Based on title 420 papers were identified as possibly of interest. The rest was
    discarded as animal or cell studies (73), extremely low frequency fields (339),
    radio- or tv- or GSM masts (11), SAR (8), ionising radiation or therapy (583),
    using mobile phones as research tool (17), other (76), and tumours not in head or
    brain (537).
         The 420 remaining papers were evaluated using the abstracts. This resulted in
    76 publications on original studies of interest, 108 editorials, 68 reviews and 14
    of potential interest as theory forming papers. The rest was discarded as animal
    or cell studies (2), extremely low frequency fields (54), tumours not in head or
    brain (12), ionising radiation or therapy (27), language not English, French,
    German or Dutch (16), not mobile phones (43).
146 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 147 =================================================================

<br><br>====================================================================== Pagina 148 ======================================================================

<pre>Table B.1 PubMed search results in number of hits per single or two-term combination.
                                     solo            +H           +T       +C      +N      + Epi         + EA     +D
Cellular phone*                      2639            2247         314      343     2914    367           138      266
Mobile phone*                        1799            1366         203      230     1768    266           177      192
Cell phone*                          716             551          65       80      61      95            12       24
Radio waves                          15322           7098         2230     2228    2065    402           204      1279
Electromagnetic fields               14720           8381         1843     1890    1657    1044          557      1570
Radio frequency                      7779            4538         1743     1780    1644    1863          123      399
Human (H)                            12116038        -            -        -       -       -             -        -
Tumour (T)                           2568569         2111767      -        -       -       269665        5566     23874
Cancer (C)                           2537766         2075920      -        -       -       280169        6737     25520
Neoplasms (N)                        2274624         1918981      -        -       -       263885        5121     22233
Epidemiology (Epi)                   1391216         -            -        -       -       -             -        -
Exposure Assessment (EA)             37814           -            -        -       -       9300          -        -
Dosimetry (D)                        106491          -            -        -       -       5080          -        -
* operator term allowing for plural + : operator term “AND”;.?????
Table B.2 PubMed search results in number of hits per multiple term combination.
                             Cellular       Mobile        Cell phone* Radio waves Electromagnetic fields   Radio frequency
                             phone*         phone*
+ T/C/N                      362            249           84           2361        2040                    1854
+H+T                         276            178           60           1839        1565                    1591
+H+C                         314            202           72           1842        1598                    1612
+H+N                         266            167           59           1748        1446                    1545
+ H + T/C/N                  321            208           73           1915        1691                    1649
+ H + Epi)                   362            253           93           366         998                     1754
+ H + EA                     125            101           9            160         483                     113
+H+D                         198            151           18           670         940                     286
+ H + Epi + T                149            98            24           210         667                     979
+ H + Epi + C                157            101           29           213         692                     995
+ H + Epi + N                149            97            24           210         663                     973
+ H + Epi + T/C/N            157            101           29           213         694                     999
* operator term allowing for plural + : operator term “AND”, / : operator term “OR”, ????
              The resulting list of 76 publications was evaluated using full text publications.
              Thirty-three were set aside as validation studies (6), supporting papers (14),
              meta-analyses (2), not mobile phone studies (7), not tumour studies (3) and case
              study (1).
                  The remaining 41 publications were checked for completeness by an expert
              (EvR) and compared to the reference lists of recent reviews as well as searching
              for other publications by the main authors. This identified a further 27
              publications that were missing and 15 that were published in 2011 or later and
              probably missed for that reason. These experiences clearly indicate that
              Search strategy and results                                                                              147
</pre>

====================================================================== Einde pagina 148 =================================================================

<br><br>====================================================================== Pagina 149 ======================================================================

<pre>    searching needs to include a snowballing component and cannot solely rely on
    protocol-driven search strategies, as has also been observed by others.180
        This resulted in a total of 85 publications on original or pooled studies that
    were to be analysed. A complete list of all publications identified at any of the
    stages of the search is available upon request. The full flow of searches,
    decisions and numbers is presented in Figure 1 in the main text of this report.
    Duplicate publications
    It can be argued that overlapping publications should be excluded to avoid
    double counting and overweighting limited evidence. However, this would
    exclude potentially important evidence, so pooled and overlapping evidence is
    included in the extracted papers with due recognition of the problem. After the
    extractions were performed, a selection of papers was made that were used for
    presenting unique results. In some cases information on e.g. methods of numbers
    of cases and controls was obtained from related publications, but the data on
    odds ratios was taken from one single publication per study, to avoid
    overweighting.
    Updating search
    As there were many ‘later identified’ publications and the whole process took a
    long time, an update search was conducted on 10 July 2012 with a limited
    timeframe starting 01 January 2011. This confirmed the identification of 15
    ‘later’ publications which were all included in the evaluation process taking the
    total of publication evaluated to 85.
148 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 149 =================================================================

<br><br>====================================================================== Pagina 150 ======================================================================

<pre>Annex C
      Data extraction
      Table C1 was used for the extraction of data from the selected studies.
      Table C1 Data extraction items.
      Reference no. for extraction; file no. EMFcommittee, Reference no. document
      1st Author (Year)
      Title (short)
      GENERAL
      A        1 Why was the study done?
               2 What were the prior hypotheses, if any?
               3 What hypotheses were actually tested?
      B        1 What type of study was done?
               2 Was this design appropriate to the study question?
               3 How might some other design have been better?
               4 What was the follow-up period?
               5 Was the follow-up period relevant to the study questions?
      C        1 How was the size of the study population determined?
               2 How might some other size have been better?
               3 Was a power based assessment of adequacy of sample size done?
      D        1 How was the ratio case/controls or exposed/non–exposed determined?
               2 How might some other ratio have been better?
      E        1 What would be possible ethical issues in relation to the design and conduct of this study?
               2 Was the study cleared by an ethics committee?
      DATA COLLECTION
      F        1 What was the source of the subjects?
               2 How might another source have been better?
      Data extraction                                                                                       149
</pre>

====================================================================== Einde pagina 150 =================================================================

<br><br>====================================================================== Pagina 151 ======================================================================

<pre>    G       1 What were the response rates?
            2 Were the response rates adequate for interpretation of the results?
            3 What were the final numbers in the study?
            4 What was the percentage of follow-up?
            5 Was the follow-up percentage adequate for interpretation?
            6 What was the follow-up no. of years?
    H       1 Could there have been selection bias?
            2 What was the likely effect of selection bias on the data if identified?
    I       1 Could there have been responders bias?
            2 What was the likely effect of responders bias on the data if identified?
    J       1 Could there have been information bias?
            2 What was the likely effect of the information bias on the data if identified?
    K       1 Could there have been observation bias?
            2 What was the likely effect of the observation bias on the data if identified?
    L       1 What confounding bias was possible?
            2 Were the confounders measured?
    M       1 Could there have been misclassification bias?
            2 What were the sources of misclassification bias?
    N       1 How was exposure measured?
            2 Would other exposure measures have been better?
            3 Do exposure measures reflect person-dose or population-dose?
            4 What was the exposure among the controls?
            5 Can the exposure measures allow for a dose gradient measure?
    O       1 Was there a major influence of measurement error?
    P       2 Was there a major influence of random error?
    ANALYSIS
    Q       1 What were the methods used to control confounding bias?
            2 Would other methods have been better?
    R       1 What were the methods used to measure the association between exposure and disease?
            2 Would other methods have been better?
    S       1 What were methods used to measure the stability of the association between exposure and
               disease?
            2 Would other methods have been better?
    T       1 Was there internal consistency among the data presented in the paper?
    INTERPRETATION
    U       1 What were the major results of the study?
            2 What were the key results in numbers?
            3 Was the temporal relationship correct?
            4 Was there a dose-response gradient?
    V       1 How might bias including confounding have affected these results?
    W       1 How might misclassification have affected these results?
    X       1 Are the references up to date and relevant?
            2 Are there any glaring omissions in the references?
    Y       1 To whom may the results of this study be generalised?
    Z       1 Is the interpretation of the data conservative?
    note1
    note2
    post-hoc power calculation
150 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 151 =================================================================

<br><br>====================================================================== Pagina 152 ======================================================================

<pre>Annex       D
            Evaluation of quality of the studies
            Table D1 shows the method used to evaluate cohort, case-control and case-case
            studies. Ecological studies were not evaluated.
            Questions 1-4 are contributing to the domain of selection, with a maximum score
            of 34; question 5 contributes to the domain of diagnosis, with a maximum score
            of 4; questions 6-14 contribute to the domain of exposure, with a maximum score
            of 69; questions 15 and 16 contribute to the domain of confounding, with a
            maximum score of 16; and question 17 contributes to the domain of conflict of
            interest, with a maximum score of 5.
Table D1 Evaluation system used for cohort, case-control or case-case studies on mobile phone use and head and neck tumours.
nr      Question                                  Evaluation                         Score Remarks
SELECTION
1       Did cases & controls come from a          No or unknown                      0      Consider Berkson’s bias if
        the same source population?       b       Yes                                12     hospital based.
                                          c       Not applicable (cohort or case-    12
                                                  case)
2       Were the same inclusion/          a       No or unknown                      0
        exclusion criteria applied to     b       Yes                                6
        cases and controls?               c       Not applicable (cohort or case-    6
                                                  case)
            Evaluation of quality of the studies                                                                       151
</pre>

====================================================================== Einde pagina 152 =================================================================

<br><br>====================================================================== Pagina 153 ======================================================================

<pre>3     What was the % response of the   a < 76% or unknown or              0 Include deceased cases and
      cases?                             unclassifiable                     refusals by physician in
                                       b 76-90%                           4 (re)calculated response rates
                                       c > 90%                            8
                                       d Not applicable (cohort or case-  8
                                         case)
4     Was the absolute difference in % a No or unknown                    0
      response between cases and       b Yes                              4
      controls <20%?                   c Not applicable (cohort or case-  8
                                         case)
DIAGNOSIS
5     Was the cancer diagnosis valid?  a No or unknown                    0 If they use cancer registry they
                                       b Yes, but imaging only            1 probably have histology and
                                       c Yes, but imaging plus location   2 imaging but if they have glioma
                                         only                               vs. meningioma they certainly
                                                                            have histology
                                       d Yes, including histology         3
                                       e Yes, including histology and     4
                                         location
EXPOSURE
6     Could the type of administration a Participant or proxy, interview  0
      of the (exposure) questionnaire    (in person or by phone)
      lead to                            administered
      observer bias?                   b Participant or proxy, self       5
                                         administered
                                       c Register-based                   5
7     Were all cases and controls      a No or not provided               0 No is if there is clearly a different
      treated equally?                 b Yes                              5 data collection protocol or
                                       c Yes as is cohort study           5 people involved between the
                                                                            groups
8     Was there potential for non-     a Yes: register based data-        0
      differential misclassification?    collection
                                       b somewhat: self administered data 5
                                         collection
                                       c No: interview-based data         5
                                         collection
9     Completeness of type mobile      a Total of 2 points                2 Accumulate points for phone
      telephone history?               b Total of 3 points                3 type history
                                       c Total of 4 points                4 Mobile phone, non-specified
                                                                            analogue or digital: 3 points
                                       d Total of 5 points                5
                                                                            Mobile phone, specified
                                       e Total of 6 points                6 analogue or digital: 4 points
                                       f Total of 7 points                7 Cordless or DECT phone: 2
                                       g Total of 8 points                8 points
                                       h Total of 9 points                9 Change in phone type: 3 points
152       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 153 =================================================================

<br><br>====================================================================== Pagina 154 ======================================================================

<pre>10    Did the measure of exposure       a  No                              0
      include frequency and duration    b  Start date or call-duration or  4
      and start date?                      frequency
                                        c  Start date and call-duration or 6
                                           frequency
                                        d  All three, but no changes       8
                                        e  All three, including changes in 10
                                           use for all types
11    Did the exposure assessment       a  No                              0
      include lateralisation of phone   b  Indirectly via handedness       5
      use?                              c  Yes, directly via questions and 10
                                           allowing for combinations
12    Were changes over time            a  No                              0  If changes asked for and total
      considered in the analysis?       b  Yes                             5  hours called calculated: assumed
                                                                              changes incorporated
13    Was the exposure questionnaire    a  No or unknown                   0
      validated or was reliability      b  Validated in another (related)  5
      tested?                              study such as subsample
                                        c  Provider data verified          10
14    Was the exposure assessed         a  No (case-control)               0
      before the cancer diagnosis (thus b  Yes (cohort or nested case-     10
      avoiding recall bias)?               control)
CONFOUNDING
15    Were confounders adjusted in a    a  No or unknown                   0  Potential confounders: age, sex
      correct way?                      b  Yes                             8
16    Could residual confounding        a  Yes or unknown                  0  As little known about potential
      influence the results?            b  Partly                          4  confounders, this is likely to
                                        c  No                              8  always be partly true
CONFLICT OF INTEREST
17    Was there evidence of potential   a  Yes                             0
      conflict of interest?             b  Yes, but with firewall          3
                                        c  No                              5
           Evaluation of quality of the studies                                                              153
</pre>

====================================================================== Einde pagina 154 =================================================================

<br><br>====================================================================== Pagina 155 ======================================================================

<pre>154 Mobile phones and cancer</pre>

====================================================================== Einde pagina 155 =================================================================

<br><br>====================================================================== Pagina 156 ======================================================================

<pre>Annex E
      Additional information for the
      publications used
      In this Annex, all Letters to the Editors, Editorials and supplementary
      publications used in the evaluation of the original publications are listed.
      Cohort studies
      There were 15 (invited) Letters to the Editors and responses from authors and
      one supporting paper. These are listed in Table E1.
      The main issue identified was the possibility of socio-economic bias due to the
      selection of early adopters in the Danish cohort, but this was corrected for in the
      latest publication.52 Kundi (2012)194 pointed out that the total number of cases, in
      spite of the relatively large person-number of years in the publication, is small,
      much smaller than in the large case-control studies, and that although there is no
      recall bias, there still may be a substantial underestimation of risk. This is due to
      the fact that about 50% of the subjects labelled non-exposed have actually been
      exposed for over 10 years and because it is unclear if those labelled exposed
      were actually the ones using the phones. Rough calculations by Kundi showed a
      potential relative risk of 1.63 for the >10 years exposure category. Leszczynski
      (2011)189 also criticized the roughness of the exposure characterization, as
      people with widely different phone use would be grouped as exposed. This
      comment also addresses the small number of cases. Morgan (2011)187 felt that
      Additional information for the publications used                                      155
</pre>

====================================================================== Einde pagina 156 =================================================================

<br><br>====================================================================== Pagina 157 ======================================================================

<pre>Table E1 Supporting literature and Letters to the Editor for the cohort studies.
Reference                                 Supporting paper /                     Subject
                                          Letter to the Editor
Hardell et al. (2001)181                  Letter to the Editor                   Comment to Johansen et al. (2001)48
Ahlbom et al. (2007)182                   Letter to the Editor                   Comment to Schüz et al. (2006)50
Ahlbom et al. (2011)151                   Invited editorial                      Comment to Frei et al. (2011)52
Charlier (2011)183                        Letter to the Editor                   Comment to Frei et al. (2011)52
Henshaw (2011)184                         Letter to the Editor                   Comment to Frei et al. (2011)52
Gujral (2011)185                          Letter to the Editor                   Comment to Frei et al. (2011)52
Davis (2011)186                           Letter to the Editor                   Comment to Frei et al. (2011)52
Morgan (2011)187                          Letter to the Editor                   Comment to Frei et al. (2011)52
Frey (2011)188                            Letter to the Editor                   Comment to Frei et al. (2011)52
Leszczynski (2011)189                     Letter to the Editor                   Comment to Frei et al. (2011)52
Glaser (2012)190                          Letter to the Editor                   Comment to Frei et al. (2011)52
Khurana (2011)191                         Letter to the Editor                   Comment to Frei et al. (2011)52
Philips & Lamburn (2011)192               Letter to the Editor                   Comment to Frei et al. (2011)52
Frei et al. (2011)193                     Response from authors                  Reply to comments from Khurana (2011)
                                                                                 191 and Philips & Lamburn (2011) 192
Kundi (2012)194                           Letter to the Editor                   Comment to Frei et al. (2011)52
Schüz et al. (2011)53                     Original publication                   Study design COSMOS cohort
               the entire results had to be dismissed as the publications seemed to indicate for
               all cancers combined (the Danish cohort study not only considered cancers of the
               head but also other cancers) a protective effect of being a subscriber, thus
               indicating a clear healthy subscriber effect (i.e. the group of subscribers is not
               representative for the population as a whole, but has a better than average
               health). The comments by Davis (2011)186 and Gujral (2011)185 closely echo
               these points. Henshaw (2011)184 and Frey (2011)188 address the issue that there is
               no biological model that might explain any risk. Both argue that the current lack
               of an agreed model should not be used as an argument against the existence of a
               risk. They call for more well-designed studies that can actually address plausible
               effect models; the cohort study does not do this. The authors responded that there
               are indications from other sources that the early subscribers were on average
               heavier users than later subscribers.193 They agree that not incorporating the
               business subscriptions would not allow the detection of a small risk increase such
               as in subsets of the INTERPHONE study.
156            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 157 =================================================================

<br><br>====================================================================== Pagina 158 ======================================================================

<pre>               Case-control studies
               Case-control studies according to the Hardell protocol
               There were 2 (invited) Letters to the Editor and responses from authors and one
               supporting paper. These are listed in Table E2.
Table E2 Supporting literature and Letters to the Editor for the Hardell studies.
Reference                                 Supporting paper / Letter to the Editor Subject
Ahlbom & Feychting (1999)137              Letter to the Editor                    Comment on Hardell et al. (1999)1
Boice & McLaughlin (2006)195              Letter to the Editor                    Rebuttal to allegations in Hardell &
                                                                                  Hansson Mild (2006)196
Hansson Mild et al. (2005)197             Supporting paper                        Combining mobile and cordless phone
                                                                                  data
               The main issue on the Hardell case-control studies identified by Ahlbom and
               Feychting (1999)137 is a seeming discrepancy between the number of cases
               identified in the initial case-control study and those in the Swedish cancer
               registry for the same period, but this was refuted by the authors with substantial
               detail about the in- and exclusion criteria. The letter by Boice and McLaughlin
               (2006)195 mainly refutes perceived conflict of interest claims.
                    The supporting paper by Hanson Mild et al. (2005)197 analyses the likely
               contribution of different mobile and cordless phones to the total exposure and
               argues against simple cumulative measures, but proposes a weighting with
               exposure from GSM phones weighing 1/10th of that of NMT (analogue) phones
               and cordless (DECT) phones weighing 1/100th. However, such weighting has not
               been used in any of the publications used in this report.
               Case-control studies according to the INTERPHONE protocol
               Thirty (invited) Letters to the Editor and responses from authors and 15
               supporting papers were considered in the context of these publications. These are
               listed in Table E3.
Table E3 Supporting literature and Letters to the Editor for the INTERPHONE publications.
Reference                                 Supporting paper / Letter to the Editor Subject
Tarone & Inskip (2005)198                 Letter to the Editor                    Comment on Lönn et al. (2004)76
Stang et al. (2005)199                    Letter to the Editor                    Comment on Lönn et al. (2004)76
Hardell & Hansson Mild (2005)200          Letter to the Editor                    Comment on Lönn et al. (2004)76
Thomas et al. (2005)201                   Letter to the Editor                    Comment on Lönn et al. (2004)76
               Additional information for the publications used                                                        157
</pre>

====================================================================== Einde pagina 158 =================================================================

<br><br>====================================================================== Pagina 159 ======================================================================

<pre>Johnston & Scherb (2005)202                 Letter to the Editor                       Comment on Lönn et al. (2004)76
Lönn et al. (2005)230                       Response from Authors                      Comments on Lönn et al. (2004)76
Savitz (2004)203                            Commentary                                 Comment on Lönn et al. (2004)76
Hardell & Hansson Mild (2004)204            Letter to the Editor                       Comment on Christensen et al. (2004)75
Kundi (2004)205                             Letter to the Editor                       Comment on Christensen et al. (2004)75
Gale & Juran (2006)206                      Letter to the Editor                       Comment on Christensen et al. (2004)75
Hardell et al. (2005)207                    Letter to the Editor                       Comment on Lönn et al. (2005)78
Milham (2005)208                            Letter to the Editor                       Comment on Lönn et al. (2005)78
Morgan (2006)209                            Letter to the Editor                       Comment on Lönn et al. (2005)78
Morgan (2006)209                            Letter to the Editor                       Comment on Schüz et al. (2006)81
Schüz (2006)210                             Response from Authors                      Comment on Schüz et al. (2006)81
Hardell & Hansson Mild (2006)211            Letter to the Editor                       Comment on Hepworth et al. (2006)84
Kundi (2006)212                             Letter to the Editor                       Comment on Hepworth et al. (2006)84
Maier (2006)213                             Letter to the Editor                       Comment on Hepworth et al. (2006)84
Morgan (2006)214                            Letter to the Editor                       Comment on Hepworth et al. (2006)84
Hocking (2008)215                           Letter to the Editor                       Comment on Takebayashi et al. (2008)88
Hocking (2006)216                           Letter to the Editor                       Comment on Schoemaker et al. (2005)90
Hardell & Hansson Mild (2006)196            Letter to the Editor                       Comment on Schoemaker et al. (2005)90
Milham (2006)217                            Letter to the Editor                       Comment on Schoemaker et al. (2005)90
Schoemaker et al. (2006)218                 Response from authors                      Reply to comments on Schoemaker et al.
                                                                                       (2005)90
Noone (2009)219                             Letter to the Editor                       Comment on Lahkola et al. (2008)92
Milham (2010)220                            Letter to the Editor                       Comment on Lahkola et al. (2008)92
Morgan (2010)    221                        Letter to the Editor                       Comment on Lahkola et al. (2008)92
Auvinen et al. (2010)231                    Response from authors                      Reply to comments on Lahkola et al.
                                                                                       (2008)92
Saracci & Sammet (2010)    222              Commentary                                 Comment on INTERPHONE (2010)93
Clouston (2011)223                          Letter to the Editor                       Comment on INTERPHONE (2010)93
Cardis & Kilkenny (1999)73                  Supporting paper                           Feasibility study results
Berg et al. (2005)148                       Supporting paper                           German validation of exposure
Samkange-Zeeb et al. (2004)149              Supporting paper                           German self report validation study
Behrens et al. (2004)224                    Supporting paper                           Limits to exposure assessment
Lahkola et al. (2005)150                    Supporting paper                           Finnish selection bias study
Schoemaker et al. (2006)225                 Supporting paper                           Other determinants analysis
Berg et al. (2006)226                       Supporting paper                           Occupational exposure to RF
Vrijheid et al. (2006)146                   Supporting paper                           Mobile phone use recall bias validation
Vrijheid et al. (2006)145                   Supporting paper                           Recall and selection bias
Cardis et al. (2007)74                      Supporting paper                           Study design
Schüz & Johansen (2007)159                  Supporting paper                           Self-report versus subscriber data
Vrijheid et al. (2009)46                    Supporting paper                           Recall bias
Vrijheid et al. (2009)143                   Supporting paper                           Selection bias
Schüz (2009)160                             Supporting paper                           Laterality issues*
Vrijheid et al. (2009) 144                  Supporting paper                           Mobile phone output power
* These refer to a possible relationship between the location of the tumour in the head and the preferential side of use of the
mobile telephone
158            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 159 =================================================================

<br><br>====================================================================== Pagina 160 ======================================================================

<pre>The main issues identified in the supporting papers of the INTERPHONE study
are related to the possible effects of recall and selection bias. The publications by
Vrijheid et al. (2006)145,146, Schüz & Johansen (2007)159, Vrijheid et al.
(2009)46,143,144 and Schüz (2009)160 indicate the expectation of a considerable
effect of random error in recall of phone use, which might result in an
underestimation of the effect. Indications for differential recall bias (thus
different between cases and controls) were observed for recall periods of 4-5
years.46 This could result in overestimation of the effect, but it is not possible to
indicate to what extent this occurred in the main studies, where phone use up to
more than 10 years back was investigated. Selection bias, particularly the
measured selective non-response of non-phone users, also is expected to result in
lower risk estimates. Another finding was that possibly the number of calls
would be a better (more robust) measure of exposure than the cumulative hours
called.
    For the main INTERPHONE publications93,94 Clouston (2011)223 stated that
there was clear evidence of selection bias related to socioeconomic class and that
this in turn could have led to confounding, as socioeconomic class is closely
related to the survival related to glioma (if not the incidence also) which likely
results in underestimation of an effect. Saracci and Samet (2010)222 in an
editorial pointed out that even now widely established cancer risks such as from
tobacco smoking would not have been possible to be identified within the first 10
years or so after start of exposure. They also pointed at the high number of
significantly decreased relative risks, for which is it not realistic to assume a
protective effect, but for which in particular participation bias (i.e. differences in
participation between cases and controls, as reflected in the different response
rates) is the most likely explanation, as was also concluded by the
INTERPHONE authors. They therefore concluded that the question on effect
remains open and much more research is needed. Kundi (2006)212, in addressing
the paper by Hepworth et al. (2006)84, stated the same in view of the on average
short follow up period, as there are no occupational or other factors known that
are associated with effects in such a short period. Kundi considered only the
laterality analysis to be relevant and this showed a statistically significant
association. Hepworth et al. (2006)84 discussed that when the odds ratio for
contralateral tumours is lower than 1, this proves that the increased ipsilateral
risk is the result of recall bias. Kundi however considered this a consequence of
the method of analysis.
    The more methodological issues had often already been raised after the
publications of results from individual or small number of countries. The main
issues are:
Additional information for the publications used                                       159
</pre>

====================================================================== Einde pagina 160 =================================================================

<br><br>====================================================================== Pagina 161 ======================================================================

<pre>    • Since many odds ratios are statistically significant below unity, either mobile
      phone use protects (which is unlikely), or there is selection bias in the study
      population (Milham (2005)208, Noone (2009)219)
    • There is evidence of selection bias, as there is low response and the cases
      have higher affluence (Morgan (2006)214)
    • There is selection bias, as visible in the distribution of gender (Thomas et al.
      (2005)201)
    • The total number of cases was too small for realistic conclusions to be drawn
      (Morgan (2006)209, Johnston & Scherb (2005)202)
    • The interview method was too stressful for patients and there was possible
      exclusion of patients living in remote areas (Hardell et.al (2005)207)
    • There was recall bias resulting in underestimation of the risk (Hocking
      (2006)216)
    • There is a high non-response and the resulting bias leads to underestimation
      of the risk (Milham (2006)217)
    • There is a high non-response in the cases of the Japanese study88 and over-
      representation of the more affluent in the controls, resulting in substantial
      underestimation of the risk due to selection bias (Hocking (2008)215)
    • The method used for analyzing laterality in Lönn et al. (2005)78 is incorrect,
      as cases with contralateral use are labelled unexposed. The authors of this
      letter conclude on the basis of calculations that the risks reported in this study
      are substantially underestimated (Hardell et al (2005)207)
    • The laterality analysis indicates misclassification of exposure (Hardell and
      Hansson Mild (2004)204)
    • The laterality analysis is fundamentally flawed (Tarone and Inskip (2005)198)
    • Several odd hypotheses seem to underlie the analysis, such as the assumption
      that the effect of mobile phones should be associated with increasing
      aggressiveness of the tumour histology (Morgan (2006)209)
    • Odd data regarding histological verification of acoustic neuroma cases
      (Hardell & Hanson Mild (2005)200)
    • An extremely low cut-off for regular cell phone use (at least once a week for
      six months or more) (Morgan (2006)209).
    • Wrong assumptions about latency for the acoustic neuroma studies as ever/
      never would be better (Stang et al. (2005)199)
    • Not all wireless phones are considered, in particular cordless phones, and
      changes in phone use are insufficiently taken into account (change from
      analogue to digital phones) (Hardell and Hansson Mild (2004)204)
160 Mobile phones and cancer
</pre>

====================================================================== Einde pagina 161 =================================================================

<br><br>====================================================================== Pagina 162 ======================================================================

<pre>              •   There is potential conflict of interest as some investigations were (partly)
                  financed by the mobile phone industry (Hardell and Hansson Mild196,
                  Morgan (2006)214)
              •   Several inconsistencies in the publications were identified (Hardell et al
                  (2005)207, Morgan (2006)214, Hardell & Hansson Mild (2005)200) but these
                  were mostly accepted by the authors as typographical errors and corrected.
              Authors responses came from Schoemaker et al. (2006)218 who addressed the
              issue of conflict of interest and pointed at the firewall construction to prevent this
              and concluded that the biases that were elaborately discussed in the publication
              in their view did not amount to a likely substantial underestimation of risk. Schüz
              et al. (2006)210 replied to the comments on their publication by stating that they
              deliberately identified ‘regular users’ with a low cut-off, in order to get a
              reference group consisting of subjects with extremely low to no exposure; they
              discussed the issue of selection bias, thinking that their results are in line with
              others and thus not underestimated; and they addressed again potential conflicts
              of interest by stating that this is taken care of with a good firewall construction.
                  In a more general remark in relation to the partial INTERPHONE
              publications, Savitz (2004)203, reacting to the publication by Lönn et al. (2004)76,
              came to the conclusion that this publication shifted the likelihood of there being
              an effect from ‘highly unlikely to slightly more likely but still highly uncertain’.
              The publication by Hepworth et al. (2006)84 was discussed by Maier (2006)213 in
              an editorial concluding that, even though effects on tumours cannot be excluded,
              the most important effects of mobile phones are a positive one on the quality of
              people’s lives and a negative one as their use is dangerous while driving. Noone
              (2009)219 stated in relation to the publication by Lahkola et al. (2008)92 that there
              cannot be a conclusion yet, as too many widely different associations could still
              hold true.
              Case-control studies according to other protocols
              Two supporting papers were considered in the context of these publications.
              These are listed in Table E4. No Letters to the Editor were identified.
Table E4 Supporting literature and Letters to the Editor for the case-control studies according to other protocols.
Reference                                 Supporting paper / Letter to the Editor      Subject
Schmidt-Pokrzywniak et al. (2004)227      Supporting paper                             Study design of Stang et al. (2009)104
Aydin et al. (2011) 228                   Supporting paper                             Error issues for Aydin et al. (2011)108
              Additional information for the publications used                                                                 161
</pre>

====================================================================== Einde pagina 162 =================================================================

<br><br>====================================================================== Pagina 163 ======================================================================

<pre>               Schmidt-Pokrzywniak et al. (2004)227 described design issues related to the study
               by Stang et al. (2009)104. They mainly focused on the feasibility of case
               recruitment and concluded that cases can be recruited and exposure can be
               measured in the way proposed in the study design.
                    Aydin et al. (2011)228 published an evaluation of the errors in measurement
               related to the case-control study in children by Aydin et al. (2011)108. The paper
               concludes that there is overestimation of exposure, but that this does not differ
               between cases and controls but is associated with age and sex, making these
               factors clear confounders that need to be accommodated for in the analysis.
               Case-case studies
               No Letters to the Editors and no supporting papers were considered in the
               context of these publications.
               Ecological studies
               Two Letters to the Editor, no responses from authors and no supporting papers
               were found in the context of these publications. The letters are listed in Table E5.
Table E5 Letters to the Editor for the ecologic studies.
Reference                                  Supporting paper / Letter to the Editor Subject
Hardell et al. (2010)125                   Letter to the editor                    Comment on Deltour et al. (2009)122
Davis (2011)186                            Letter to the editor                    Comment on Ahlbom & Feychting
                                                                                   (2011)151
               Hardell et al (2010)125 commented on the publication by Deltour et al. (2009)122
               that their data collection stopped in 2003, while in any case in Sweden according
               to the information in the Hardell studies, the use of mobile phones sharply
               increased after 2003. Thus according to Hardell et al, the conclusions from this
               publication cannot be definitive. In the later publication by Deltour et al.
               (2012)135, the analysis extends to 2008, but this would still be not long enough to
               reflect any increase due to the increased mobile phone use indicated by Hardell,
               assuming an latency period of at least 10 years and a small relative risk. Hardell
               et al. concluded that to allow firm conclusions to be drawn, at least another 10
               years of observations is needed.
                   Davis (2011)186 challenged the conclusion of Ahlbom and Feychting
               (2011)151, by arguing that the Swedish cancer registry is not complete. In a
               personal communication to the Committee, Feychting denied this.
162            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 163 =================================================================

<br><br>====================================================================== Pagina 164 ======================================================================

<pre>Annex        F
             Results of the data extraction
             These tables show the results of the data extraction for the publications used in
             the evaluation of the quality of studies in Annex G.
             Cohort studies
Table F1 Extractions from Dreyer NA, Loughlin JE, and Rothman KJ. Cause-specific mortality in cellular telephone users.
JAMA, 1999; 282(19): 1814-1816.47
A1     concerns about potential biological effects, including brain cancers due to radiofrequency energy transmitted from
       mobile phones; additional to a previous study (1010), cause-specific mortality in expanded cohort
A2     not described, probably same as in 1010
A3     overall & specific mortality and length of mobile service contract
B1     cohort study
B2     design ok, detailed comparison with general population is missing
B3     design is ok, reasonably efficient because using registrations that have been linked, however many issues to make this
       method work for the study question; different time period would have been better (longer)
B4     not described, but highest category: >3 years of use
B5     too short to prove cancer, to not even consider prove cancer mortality
C1     based on registration of 2 US cellular telephone carriers; all subscribers to these
C2     size ok, very big cohort
C3     no
D1     all exposed to at least 1 phone type (cordless or mobile)
D2     include truly nonexposed
E1     that is done without consent, participants are not aware of all the privacy sensitive data that have been used for this
       investigation
E2     not mentioned
             Results of the data extraction                                                                                    163
</pre>

====================================================================== Einde pagina 164 =================================================================

<br><br>====================================================================== Pagina 165 ======================================================================

<pre>F1  all cellular telephone users covered by two US cellular telephone carriers serving several metropolitan areas
F2  source ok; but very restricted: only one contract per household, only households that clearly were not companies (might
    exclude single medical practices or trades people)
G1  not described, but in 1010 clear that substantial exclusion occurred: linking 2 registries, linking far from perfect; original
    cohort 770390 records (before which already excluded corporate users, multiple telephone users), finally after various
    eliminations and exclusions: 255868 records over (33,2%)
G2  no
G3  285561 records
G4  till category > 3 years nothing else described
G5  very short for interpretation cancer, to not even talk about cancer mortality
G6  highest category: > 3 years
H1  not described in this article but it is in 1010: yes but: in the early years who used mobile phones: predominantly working
    people, so healthy subpopulation, also various exclusions made that make it an indescribable study population, every
    exclusion factor probably results in selection bias
H2  can go either way
I1  no, because data linked without interference of research personnel
I2  n.a.
J1  yes, for information completely dependent on registrations, dependent on their quality, dependent if good data were
    delivered to them, missing information about duration of phone use, how much phone used etc., however in analyses
    only info used about which phone and this seems quite easily traced using an ESN, so for analyses not such a big
    problem
J2  so not much for analysis
K1  no, registrations used, at the most if extraction of data from the database not done well e.g. if someone has been
    selectively searching for people, however, this seems quite unlikely
K2  n.a.
L1  yes, sex, possibly particularly healthy working subpopulation that used mobile phones (and less risk of dying (early)
    than a ill non-working population), socio-economic status
L2  sex yes, nothing else
M1  everything if ESN does not lead to correct phone type
M2  n.a.
N1  see 1010, not given in this letter: only if used yes/no, and minimal 1 (or 2 or 3) years registered and 2 active accounts:
N2  your exact exposure, through duration of plan, duration and amount and frequently of phone calls, poss. Urban or rural
    etc.
N3  person-dose
N4  no controls, all exposed, 2 groups: portable (no risk to be expected) and mobile phone
N5  no, no information of how often, how long etc.
O   yes, limited info about measurement of exposure, so can contain all sorts of errors, really only known what type of phone
    is present and that has been used in the last 2 months
P   no, not to be expected, normal random error
Q1  none, except adjustment for sex and age and metropolitan area
Q2  correct for more variables (including SES), if need be stratify
R1  sex and age-specific mortality rates
R2  at least compare mortality rate with the general population, but really the mortality rate is not a good indicator with such
    a short latency time, incidence of specific health effects should be compared (e.g. cancer); compare with non-users and
    in a cohort logistic regression (outcome ill/not-ill and then mobile phone use in duration and frequency and type phone
    etc as variables in the model incl confounders and such to adjust for)
S1  95% CI
164        Mobile phones and cancer
</pre>

====================================================================== Einde pagina 165 =================================================================

<br><br>====================================================================== Pagina 166 ======================================================================

<pre>S2     ok
T      almost no numbers described, only tabular info, cannot be checked well, not all seems to add up
U1     no indication that risk is increasing with increasing minutes (except maybe for motor vehicle collisions)
U2     see Annex H
U3     yes because outcome is death so all exposure before
U4     no, not really, 2 and 3 years…. However, in this analysis this says nothing about dose
V      so much bias, can really go any direction, but anyway not the correct method and too short to say anything
W      everyone exposed, so at that level no misclassification, at the most in type of phone, so effect can go either way
X1     yes
X2     yes, only 4 references, this really is very limited even though at the time not much was known
Y      to no one as too much selection bias
Z      yes
Table F2 Extractions from Schüz J, Steding-Jessen M, Hansen S, e.a. Long-term mobile phone use and the risk of vestibular
schwannoma: a Danish nationwide cohort study. Am J Epidemiol, 2011; 174(4): 416-422.51
A1     to investigate cancer (acoustic neuroma) risk among Danish cellular telephone users who were followed for up to 21
       years
A2     none, before no clear hypotheses if there would be an increased or decreased risk, more general: is there a relation
       between use of cellular telephones and tumours of head and neck
A3     was cellular telephone use associated with increased risk of brain tumours?
B1     cohort study (combination of 2 cohort studies actually)
B2     yes
B3     n.a. (experiment, practically almost impossible)
B4     1987-2006
B5     yes, but still relatively short for largest part of cohort < 10 years and few people in the groups with the long follow-up,
       longer follow-up information simply does not exist given recent use of mobile phones
C1     cohort based on people who between 1982 and 1995 first used a mobile phone
C2     super big cohort, only suggestion for improvement: now many exclusions because professional connections could not be
       personalised; due to combination now more limited but more information on confounders
C3     no
D1     exposed: total cohort, unexposed: rest general population
D2     rest general population: assumed that they did not use telephone, not entirely correct even though the most recent users
       (1995-2006) have to short follow up for cancer
E1     no permission by members of the cohort, after announcements in the media, possibility for refusing participation
E2     yes
F1     cohort cellular telephone subscriptions
F2     good source
G1     420095 of the 723421 records received are included: 58%% but the paper does not say how many of these remain in the
       combined cohort which as no acoustic neuroma was observed in long-term exposed women (but almost as many casesd
       as in men!) was reduced (50%?) to men only; 404 cases
G2     no, certain subgroups now excluded which hinders interpretation, double addresses: use more than 1 phone? So higher
       exposure? Corporate subscriptions: those people are now in the rest general population but in reality they were exposed
G3     420095 exposed; rest of general population assumed as unexposed in general population, basis onto which expected is
       calculate as unexposed in Danish cancer registry
G4     no data given other than the conceptual that they would have been exposed since before 1995, how many stopped after
       1995 is not presented
             Results of the data extraction                                                                                    165
</pre>

====================================================================== Einde pagina 166 =================================================================

<br><br>====================================================================== Pagina 167 ======================================================================

<pre>G5  not really, follow-up should be longer due to the latency time for developing cancer
G6  max follow-up not really given but few person years in top follow-up
H1  yes, particularly group that uses mobile phones for work has been excluded, while this is possibly a highly exposed
    group
H2  effect in reality present, or in any way OR >1
I1  no
I2  n.a.
J1  no, because respondents were not approached themselves, but yes, information missing about exact use, time, frequency,
    preferred ear, however this is more a limit for the analyse than bias
J2  now all the use put together and no difference made in amount of use, only duration of use in years, so no analyse done
    on this and possible distortion of results, various levels of exposure now all on one heap
K1  no, cohort was not linked to cancer cases during phase of including, so blind for case and non-case; no interview or
    anything used, hard data from registrations used so no influence observer
K2  n.a.
L1  might still be possible that there is unknown or unmeasured confounder, because the cohort seemed to be a unique
    subgroup of persons with higher income and therefore risk profile; exposure to other factors that might cause brain
    cancer?
L2  income is measured, occupation and exposure to certain substances not measured
M1  yes
M2  regular use cellular telephone (compared with interphone case control data) also non regular subscribers are now in
    exposed group
N1  subscriber or not; duration use by cohort members compared to case control interphone, no information on frequency,
    duration of calls, preferred ear etc
N2  real use of received and send phones and their duration, number, how often, how long; via questionnaire such as this
    these cannot be traced at the telephone company
N3  person-dose
N4  n.a., no controls in cohort study rest of the general population assumed to be unexposed but isn’t as many have used mps
    for long time and also includes all non-personal subscriptions
N5  no, no information of how often, how long etc
O   yes, possibly, now frequent and less frequent users together and in ref population (to calculate expected) also
    professional users
P   no
Q1  corrected for confounding using regression as a lot of information available through linked cohort
Q2  despite it all being based on routine data this is quite elaborate
R1  person year analysis resulting in SIR but a lot of it is in %
R2  appropriate association for cohort, possibly regression analysis as addition and to control for variables in a multivariate
    analysis
S1  95% CI
S2  no, ok
T   can not really check as insufficient information given
U1  no evidence for association between acoustic neuroma risk and cellular telephone use among short and long-term users
U2  see Annex H
U3  yes, subscribers known and then checked is someone became a case, however not known if exposure and brain tumour
    not too close in time and so probably not associated given latency cancer
U4  not really, only years of having a cellular phone, with very few people in the long use group
V   some level of correction applied but always tricky as limited
W   unknown exposed from corporate subscriptions biased towards the null, so in this group possibly cases missed
166       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 167 =================================================================

<br><br>====================================================================== Pagina 168 ======================================================================

<pre>X1     as far as I can see yes
X2     as far as I can see no
Y      only to the included cohort members, rest population no good info about use mobile phone
Z      yes, effect would in reality only be bigger
Table F3 Extractions from Frei P, Poulsen AH, Johansen C, e.a. Use of mobile phones and risk of brain tumours: update of
Danish cohort study. BMJ, 2011; 343: d6387.52
A1     to investigate cancer (all central nervous tumours) risk among Danish cellular telephone users who were followed for up
       to 17 years (cancers had to occur between 1990 and 2007)
A2     none, before no clear hypotheses if there would be an increased or decreased risk, more general: is there a relation
       between use of cellular telephones and tumours of head and neck
A3     was cellular telephone use associated with increased risk of brain tumours?
B1     cohort study
B2     yes
B3     n.a. (experiment, practically almost impossible)
B4     claims all Danes born after 1925 but based on those exposed before 1995
B5     yes, but still relatively short for largest part of cohort < 10 years and few people in the groups with the long follow-up,
       longer follow-up information simply does not exist given recent use of mobile phones
C1     cohort based on people who between 1982 and 1995 first used a mobile phone; combined with cohort on those born in
       Denmark after 1925
C2     super big cohort, only suggestion for improvement: now many exclusions because professional connections could not be
       personalised; due to combination now more limited but more information on confounders
C3     no
D1     exposed: total cohort, unexposed: rest general population cohort
D2     rest general population cohort: assumed that they did not use telephone, not entirely correct even though the most recent
       users (1995-2007) have to short follow up for cancer
E1     no permission by members of the cohort, after announcements in the media, possibility for refusing participation
E2     yes
F1     cohort cellular telephone subscriptions
F2     good source
G1     358403 of the 723421 records received are included: 50% but the paper has 1853 glioma cases for men, 1455 for
       women; 429 meningioma cases for men and 1248 for women
G2     no, certain subgroups now excluded which hinders interpretation, double addresses: use more than 1 phone? So higher
       exposure? Corporate subscriptions: those people are now in the rest general population but in reality they were exposed
G3     358403 exposed; rest of general population cohort assumed as unexposed in gen population, basis onto which expected
       is calculate as unexposed in Danish cancer registry
G4     among glioma cases only 37 men had exposure over 13 yrs (37/324=11%) and no women; for meningioma cases only 65
       men (65/162=40%) were over 10 yrs exposed and only 12 women (12/35=34%),
G5     not really, follow-up should be longer due to the latency time for developing cancer
G6     max follow-up not really given but few person years in top follow-up
H1     yes, particularly group that uses mobile phones for work has been excluded, while this is possibly a highly exposed
       group
H2     effect in reality present, or in any way OR >1
I1     no
I2     n.a.
             Results of the data extraction                                                                                    167
</pre>

====================================================================== Einde pagina 168 =================================================================

<br><br>====================================================================== Pagina 169 ======================================================================

<pre>J1  no, because respondents were not approached themselves, but yes, information missing about exact use, time, frequency,
    preferred ear, however this is more a limit for the analyse than bias
J2  now all the use put together and no difference made in amount of use, only duration of use in years, so no analyse done
    on this and possible distortion of results, various levels of exposure now all on one heap
K1  no, cohort was not linked to cancer cases during phase of including, so blind for case and non-case; no interview or
    anything used, hard data from registrations used so no influence observer
K2  n.a.
L1  might still be possible that there is unknown or unmeasured confounder, because the cohort seemed to be a unique
    subgroup of persons with higher income and therefore risk profile; exposure to other factors that might cause brain
    cancer?
L2  income is measured, occupation and exposure to certain substances not measured
M1  yes
M2  regular use cellular telephone (compared with interphone case control data) also non regular subscribers are now in
    exposed group
N1  subscriber or not; duration use by cohort members compared to case control interphone, no information on frequency,
    duration of calls, preferred ear etc
N2  real use of received and send phones and their duration, number, how often, how long; via questionnaire such as this
    these cannot be traced at the telephone company
N3  person-dose
N4  n.a., no controls in cohort study rest of the general population assumed to be unexposed but isn’t as many have used mps
    for long time and also includes all non-personal subscriptions
N5  no, no information of how often, how long etc
O   yes, possibly, now frequent and less frequent users together and in ref population (to calculate expected) also
    professional users
P   no
Q1  corrected for confounding using regression as a lot of information available through linked cohort
Q2  despite it all being based on routine data this si quite elaborate
R1  person year analysis resulting in SIR
R2  appropriate association for cohort, possibly regression analysis as addition and to control for variables in a multivariate
    analysis
S1  95% CI
S2  no, ok
T   yes
U1  no evidence for association between glioma or meningioma risk and cellular telephone use among short and long-term
    users
U2  see Annex H
U3  yes, subscribers known and then checked if someone became a case, however not known if exposure and brain tumour
    not too close in time and so probably not associated given latency cancer
U4  not really, only years of having a cellular phone, with very few people in the long use group
V   some level of correction applied but always tricky as limited
W   unknown exposed from corporate subscriptions biased towards the null, so in this group possibly cases missed
X1  as far as I can see yes
X2  as far as I can see no
Y   only to the included cohort members, rest population no good info about use mobile phone
Z   yes, effect would in reality only be bigger
168       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 169 =================================================================

<br><br>====================================================================== Pagina 170 ======================================================================

<pre>             Case-control studies of the Hardell group
Table F4 Extractions from Hardell L, Hallquist A, Hansson Mild K, e.a. No association between the use of cellular or cordless
telephones and salivary gland tumours. Occup Environ Med, 2004; 61(8): 675-679.69
A1     to investigate the association between the use of cellular or cordless telephones and the risk for salivary gland tumours
       because the parotid gland is located in an area where some phones give a high exposure to microwaves
A2     use of cellular and cordless phones increases the risk for salivary gland tumours
A3     use of cellular and cordless phones increases the risk for salivary gland tumours
B1     case-control;
       population based
B2     yes
B3     observational cohort, so minimising all sorts of bias (experiment would be best but not feasible)
B4     no clearly described follow-up period. Info asked back to start use. In analysis latency period >10 years is used, but only
       for a very small subgroup more than 10 years follow-up
B5     longer follow-up better, but not available, as only since 1981 start first use analogue phones and since 84 use without
       fixed antennae, and particularly need to collect more cases and controls in longer follow-up group for stronger
       conclusions
C1     cases diagnosed in cancer registries between 1 Jan 1994 and 31 December 1999 for Stockholm and Linkoping; between
       1 Jan 1994 and 30 June 2000 for Uppsala- Orebro and between 1 Jan 1994 and 30 June 1999 for rest of Sweden (Umea,
       Goteborg, Lund); incident cases
C2     size ok
C3     yes, to detect OR >=1,4 (alpha 0,05 and beta 0,20)
D1     ratio 1:4
D2     ok, based on available cases and power calculation
E1     psychological burden for cancer patients
E2     yes
F1     cancer registry for whole Sweden and population registry for controls
F2     source ok
G1     415 cases, 293 included, 267 responders = 64,3%;
       815 from other study + 357 additional, 750+303 (1053) responded = 89,9%
G2     cases quite low, but that is due to the exclusion of the deceased (n=96), controls good response,
G3     415 cases invited; 96 dead, 26 excluded 16 refused, so 267 cases and 1053 controls
G4     n.a., only 6 cases had used a phone for more than 10 years
G5     duration of use is still relatively short, so nothing can be said over longer periods, but for short term use yes, however in
       this study unclear how many people exactly used mobile phones for more than 5 years
G6     highest category > 10 years, but only 6 cases in this group
H1     yes as a substantial group of deceased cases were not included (96/415= 23%)
H2     is the question, worst cases also the most exposed?, you do not know so effect could go either way
I1     yes, but limited as response around 90% (response 415 original cases: 267/415= 64.3%, so is quite low, mostly caused
       by death)
I2     could go either way: particularly users of mobile phones interested in participating, or particularly not as busy working
       population?
J1     yes, measurement errors in exposure variables due to recall bias
J2     possibly overestimation because memory in cases could lead to higher exposure
K1     yes
             Results of the data extraction                                                                                    169
</pre>

====================================================================== Einde pagina 170 =================================================================

<br><br>====================================================================== Pagina 171 ======================================================================

<pre>K2  minimal, questionnaire i.s.o. interview, in case of additional phone interview: blinded for case-control status and tumour
    details assessed without information about exposure data
L1  age, sex has been corrected, in 1 analysis also for study areas corrected for, also SEI, occupation or other exposures
    could be confounders
L2  yes
M1  yes misclassification exposure, due to recall bias, cases have been histologically verified so minimal chance for
    misclassification cases
M2  recall problems, different types of phones used, duration of use wrongly estimated, changes over time not correctly
    remembered (e.g. change from analogue to digital)
N1  questionnaire if needed with additional telephone interview
N2  phone habits prospectively monitored, phone habits traced at telecompany;
    additional questionnaire through telephone interview particularly for phone habits of early years where recall the biggest
    problem, not know how this exactly is asked
N3  person dose
N4  ever use analogue: cases 11,6% controls 13,0%;
    digital: cases 16,9% en controls 16,1%;
    cordless: cases 18,0% en controls 19,0%;
    overall: cases 34,1% en controls 33,4%
N5  yes
O   possible yes due to recall bias, however, probably particularly for the early years, recent memory possibly more reliable
P   not to be expected
Q1  all analyses adjusted for age, sex and 1 analysis also adjusted for region
Q2  ok
R1  unconditional logistic regression analyses for matched studies
R2  not described it incomplete pairs, than ok, if they are complete pairs, conditional logistic regression analysis would be
    better
S1  95% CI
S2  ok
T   yes
U1  no association between the use of cellular or cordless phones and salivary gland tumours was found, although this study
    does not permit conclusions for long term heavy use
U2  see Annex H
U3  debatable, only few >10 years exposure and > 5 years still relatively few cases digital, < 5 years almost certainly no
    temporal relationship in carcinogenesis
U4  no
V   most bias can go either way, but possibly the results are slightly overestimated due to overestimation of exposure by the
    cases and results not corrected for e.g. occupational exposures, however for important confounders corrected, sex and
    age, no correction for other exposures
W   can go either way, depends on errors in recall bias and accompanying misclassification of exposure, possibly to be
    expected overestimation of exposure by cases, so overestimation OR
X1  yes
X2  as far as I can see: ok, no major missing references
Y   Swedish population
Z   yes
170       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 171 =================================================================

<br><br>====================================================================== Pagina 172 ======================================================================

<pre>Table F5 Extractions from Hardell L and Carlberg M. Mobile phones, cordless phones and the risk for brain tumours. Int J
Oncol, 2009; 35(1): 5-17.66
A1     further and more detailed results of the pooled analysis of 2 case control studies
A2     is there an association between mobile phones and brain tumours (benign en malignant)?
A3     is there an association between mobile phones and brain tumours (benign en malignant)?
B1     population based case-control (in discussion talked about a hospital based study, but this is not correct as cancer registry
       is used?)
B2     yes
B3     observational cohort, so minimising all sorts of bias (experiment would be best but not feasible)
B4     no clearly described follow-up period. Info asked back to start use. In analysis latency period >10 years is used
B5     reasonably good, study long enough now to have sufficient numbers in the category > 10 years
C1     all incident cases aged 20-80 diagnosed between 1 Jan 1997 and 30 June 2000 in 4 regional cancer registries in 4
       medical regions & all living cases, aged 20-80 in time period 1 July 2000 and 31 deck 2003, living in Uppsala/Örebro or
       Linkoping region (recruited through cancer registry) but only if living so actually prevalent cases
C2     nationwide and including all cases, also the ill and dead ones (via proxy)
C3     no
D1     malignant ratio 1:2,4 ;
       benign ratio 1:1,7 (all controls used)
D2     ok, enough numbers
E1     psychological burden for cancer patients and people can become anxious about mobile phones
E2     yes
F1     1997-2000 for cases 4 Swedish medical regions (Uppsala/Örebro or Linkoping, Stockholm, Gothenburg) and for
       controls population registries and 2000-2003 region of Uppsala/Orebro or Linkoping, Sweden, for cases; cancer registry,
       for controls population registry; possible delay between diagnosis and notification, as living only: selecting out the early
       deceased
F2     national cancer registry, so all regions in Sweden;
       source controls is ok
G1     benign cases 88% and malignant cases 90% controls 89%
G2     quite high, however still chance of responders bias, 1 of the 10 after all refused to take part, note that selection based on
       inclusion criteria, is not included in the response %
G3     1254 benign cases and 905 malignant cases and 2162 controls
G4     n.a., in analysis a latency period of >10 years is used
G5     particularly for group with latency period > 10 year follow up time is long enough to develop cancer as thus make an
       inference
G6     highest category > 10 years
H1     yes, due to ill and dead cases and physician refusal, this can contain selection as these are particularly the cases with
       poor prognosis, other exclusion criteria will not generate selection;
       3729 cases cancer registry, exclusion metastases, misdiagnosis, deceased (748!), refusal via physician etc, ultimately
       only 2437 eligible (65%)
H2     if particularly the cases with poor prognosis were users of mobile phone the effect would be underestimated, as those
       people are now missing. However, if they are particularly non-users there is overestimation of effect
I1     yes, but limited as response around 90%
I2     limited effect expected as the response is quite high, no info about non responders, possibly particularly non-users of
       mobile phones so less interested in participating, or particularly not participated as it is the busy working population (so
       the users)?
J1     yes, measurement errors in exposure variables due to recall bias
J2     cases would probably refer to higher exposures than controls, which would lead to overestimation of effect
              Results of the data extraction                                                                                    171
</pre>

====================================================================== Einde pagina 172 =================================================================

<br><br>====================================================================== Pagina 173 ======================================================================

<pre>K1      questionnaire not, but additional interviews can, way of asking questions by interviewer can direct answer a certain way,
        particularly since this is often about the detail of the phone use
K2      possible so not applicable, otherwise a slight overestimation of effect as observer directed towards relation phone and
        tumour.
L1      age, sex and SES has been corrected for, also occupation or other exposures could be confounders
L2      yes, including socio-economic status and various occupational exposures
M1      yes, due to recall bias, cases are usually histologically verified, so minimal chance on misclassification of cases
M2      recall problems, different types of phones used, duration of use wrongly estimated, changes over time not correctly
        remembered (e.g. change from analogue to digital)
N1      questionnaire if needed with additional telephone interview
N2      for phone use: monitor calling habits prospectively or if possible ask about calling habits at telecompany, but
        questionnaire bar recall good method
N3      person dose
N4      exposures overall not described separately
N5      yes
O       possible yes due to recall bias
P       not to be expected
Q1      analysis adjusted for age, sex, SEI and year of diagnosis
Q2      ok, possibly correct for ionising radiation
R1      unconditional logistic regression analyses for matched studies
R2      ok, as not exactly 1 to 1 matching, all controls included
S1      95% CI
S2      ok
T       yes
U1      a consistent association between use of mobile or cordless phones and astrocytoma grade I-IV and acoustic neuroma,
        highest for ipsilateral exposure using > 10 year latency; especially high risk for persons that started use of mobile phone
        before the age of 20 years. Results are supported by increasing incidence of astrocytoma during 2000-2007 in Sweden
U2      see Annex H
U3      for longer latency group >10 years yes, for < 10 years this is debatable
U4      yes
V       adjusted for important confounders
W       can go either way, depends on errors in recall bias and accompanying misclassification of exposure and possibly
        overestimation
X1      yes
X2      no
Y       Swedish population (assuming the 4 regions representative for Sweden)
Z       unclear, probably not as not corrected for other variables (ionising radiation, other occupations exposures)
Table F6 Extractions from Hardell L, Carlberg M, and Hansson Mild K. Pooled analysis of case-control studies on malignant
brain tumours and the use of mobile and cordless phones including living and deceased subjects. Int J Oncol, 2011; 38(5): 1465-
1474.68
A1      to investigate the use of mobile or cordless phones and the risk for malignant brain tumours in a group of living and
        deceased cases
A2      not a clear hypothesis is described; there is an association between cellular and cordless phone use and malignant brain
        tumours
A3      Is there an association between cellular and cordless phone use and malignant brain tumours?
172           Mobile phones and cancer
</pre>

====================================================================== Einde pagina 173 =================================================================

<br><br>====================================================================== Pagina 174 ======================================================================

<pre>B1 population based case-control
B2 yes
B3 observational cohort, so minimising all sorts of bias
B4 n.a., not a clear follow-up period. Information starts at the beginning of the use of a cellular and/ or cordless telephone.
   In analysis the latency period >10 years is used, so a subgroup has more than 10 years of follow-up (of using a cellular
   phone)
B5 yes, all right, use of mobile phone long enough to have enough cases in the group of long-term users (>10 years).
C1 all living and deceased cases aged 20-80 diagnosed between 1 Jan 1997 and 30 June 2000 in 4 regional cancer registries
   in 4 medical regions (Uppsala-Örebro, Stockholm, Linkoping, Göteborg) & diagnosed in time period 1 July 2000 and 31
   deck 2003, living in Uppsala/Örebro or Linkoping region (recruited through cancer registry)
C2 nationwide
C3 no
D1 ratio 1:1 for living and deceased cases, but for living cases also controls of benign tumours are included, so ratio is 1:2
D2 enough cases and controls
E1 psychosocial burden for cancer patient and relatives of deceased cases and controls and possibility for anxiety for mobile
   phone use
E2 Yes
F1 1997-2000 for cases 4 Swedish medical regions (Uppsala/Örebro or Linkoping, Stockholm, Gothenburg) and 2000-2003
   region of Upssala/Örebro or Linkoping, Sweden, for cases; cancer registry, for controls population registry or death
   registry
F2 national cancer registry, so all regions in Sweden;
   sources controls are ok
G1 living cases: 90%, living controls 89%; deceased cases 75% and deceased controls 60%
G2 for living cases and controls good response; for deceased cases and controls moderate response
G3 905 living cases 2162 controls; 346 deceased cases and 276 deceased controls; total: 1251 cases and 2438 controls
G4 n.a., in analysis a latency period of >10 years is used
G5 especially for group with latency period > 10 years reliable conclusions possible, time is long enough for cancer to
   develop
G6 highest category: > 10 years
H1 yes, partly due to the fact that the physician could refuse participation of the cases
H2 if cases for who participation is refused by the physician are the most ill people who may be used mobile phone the
   most, the effect will be underestimated, but the physician probably did not know about mobile phone history. So the real
   effect is unknown
I1 yes, possibly, but not a large effect due to the relatively high response
I2 no information about non-responders, so the effect can go both ways, but little effect due to high response
J1 yes, recall bias especially for mobile phone use in the earliest years, so a long time ago
J2 cases possibly refer to higher exposure than controls, leading to an overestimation of the risk
K1 for questionnaires no observation bias, for the extra phone interviews this plays possibly a role
K2 small effect, if observer is focussing on phone and cancer relation possibly a little overestimation of the risk
L1 for age, sex, year of diagnosis and SEI is adjusted in analysis, but blue colour worker or radiation could be a confounder
L2 yes, including socio-economic status and several occupational exposures
M1 yes, due to recall bias, according to Hardell this effect is little, cases are histologically confirmed, so minimal chance of
   misclassification case
M2 recall problems (different phones used, lifetime use in wrong category, changes over time
N1 questionnaires, if necessary completed with interview over the phone
          Results of the data extraction                                                                                   173
</pre>

====================================================================== Einde pagina 174 =================================================================

<br><br>====================================================================== Pagina 175 ======================================================================

<pre>N2     collect data of use of mobile phone prospectively and/ or use data of phone company about phone use provided that
       these data can be connected with the correct persons. But questionnaire is good measurement, except for the recall
       problems
N3     person dose
N4     727 of 1251 cases exposed (58,1%) and 1267 of 2438 controls exposed (52,0%)
N5     yes
O      possible yes due to recall bias
P      is not expected
Q1     analysis adjusted for age, sex, SEI, year of diagnosis and vital status
Q2     appropriate method, possibly also adjustment for ionizing radiation
R1     unconditional logistic regression analyses for matched studies
R2     no
S1     95% CI
S2     ok
T      yes
U1     the risk for glioma increased with latency period and cumulative use in hours for both mobile and cordless phone and
       was highest in subjects with first use before the age of 20
U2     see Annex H
U3     for long latency period (> 10 year) the temporal relationship is correct, for < 10 year latency time the temporal
       relationship is doubtful, especially for < 5 year
U4     yes
V      possibly little overestimation due to overestimation use of mobile phones by cases
W      misclassification in recall bias? classification of categories of exposition can go both ways so leading to over and
       underestimation of the risk
X1     yes
X2     no
Y      Swedish population, provided that 4 regions are representative for Swedish population
Z      unclear
Table F7 Extractions from Söderqvist F, Carlberg M, and Hardell L. Use of wireless phones and the risk of salivary gland
tumours: a case-control study. Eur J Cancer Prev, 2012.72
A1     some indications of effect of mobile phones on parotid gland tumour risk
A2     association between having acoustic neuroma and reporting use of mobile phones
A3     as in A2
B1     case-control
B2     is ok
B3     cohort as exposure independently measured form outcome
B4     n.a.
B5     n.a.
C1     incident cases in designated area during designated period
C2     larger so more years or wider area
C3     not presented
D1     1 case : 4 controls
D2     is supposedly optimal
E1     burden for very ill patients
174          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 175 =================================================================

<br><br>====================================================================== Pagina 176 ======================================================================

<pre>E2 not mentioned but assumed to be yes
F1 patients with salivary gland tumours (ICD-7 142.0, 142.6 and 142.8) were recruited continuously between the years
   2000 and 2003 as reported by the Regional Oncology Centre of Uppsala / Örebro and Linkoping, including nine of 21
   Swedish counties.
F2 is ok but wider or longer would have been better
G1 88% of cases and 83% of controls responded with filled in questionnaire
G2 ok
G3 in total, 92 cases were reported and of these, six were dead, four had treating doctors who did not permit their patients’
   participation and an additional four cases had wrong diagnoses
G4 n.a.
G5 n..
G6 n.a.
H1 always some possible but response rates case/control very similar so not very likely
H2 n.a.
I1 yes as cases know they are ill so this is likely
I2 either direction
J1 yes somewhat
J2 other direction
K1 for certain
K2 either direction
L1 age, sex, sex
L2 yes as far as possible for SES
M1 some for mostly exposure
M2 mp use questionnaire
N1 questionnaire
N2 checking bills
N3 Person-dose
N4 57%
N5 yes
O  some
P  some
Q1 no association seen
Q2 see Annex H
R1 regression
R2 ok
S1 85% CI
S2 ok
T  seems ok
U1 the data presented in this short report do not support an association between the use of wireless phones (including both
   the mobile phone and the cordless desktop phone) and the risk for salivary gland tumours.
U2 see Annex H
U3 cannot tell
U4 no
V  could have underestimated
W  could cause underestimation
X1 ok
         Results of the data extraction                                                                                  175
</pre>

====================================================================== Einde pagina 176 =================================================================

<br><br>====================================================================== Pagina 177 ======================================================================

<pre>X2     no
Y      other similar countries
Z      ok
             Case-control studies of the INTERPHONE consortium
Table F8 Extractions from Lönn S, Ahlbom A, Christensen HC, e.a. Mobile phone use and risk of parotid gland tumor. Am J
Epidemiol, 2006; 164(7): 637-643.82
A1     potential concern of increased risk of acoustic neuroma due to its close position to the handset of a mobile phone
A2     is there an association between mobile phone use and acoustic neuroma?
A3     does mobile phone use increases acoustic neuroma?
B1     population-based case-control
B2     yes
B3     observational cohort, so minimising all sorts of bias (experiment would be best but not feasible)
B4     info mobile phone use retrospectively asked, highest usage category > 10 years
B5     not long enough, because the development of this tumour is slow and mobile phones are only recently used at a large
       scale, only few cases and controls have been using mobile phones for a long time
C1     all cases in specific area of cancer registry
C2     national registry, include all regions so more cases in general and more cases and controls that have been using mobile
       phones for a long time, now is a small subgroup
C3     no
D1     1 per brain tumour cases, 2 per acoustic neuroma case, 3 per parotid gland tumour, all controls included in this study.
D2     2 or 3 controls for all cases
E1     development of fear for mobile phones and burden for cases
E2     not mentioned
F1     residents of 3 geographical areas covered by the regional cancer registries in Stockholm, Gotenburg, and Lund; incident
       cases of an in 3 cancer registries (Stockholm, Gotenburg, Lund) Sept 2000 - Aug 2002, 20-69 yrs old, controls from pop
       register
F2     national registry, use all regions
G1     93% of 160 eligible cases: n=148; 72% of 838 controls: n=604
G2     cases yes, controls: relatively low response rate, information of some variables of the non-responders is necessary
G3     148 cases and 604 controls
G4     n.a.
G5     no, follow-up relatively short for developing cancer due to mobile phone use.
G6     n.a., highest category mobile phone use: > 10 years
H1     not likely
H2     n.a.
I1     yes, refusal and illness can generate selection in other variables, not reached is less of an issue; most non-response
       among cases possibly due to illness (too ill or dead): excluding the very ill if illness assoc with exposure causes
       underestimation. Among controls refusal very high so most motivated left in study, possibly overestimating control
       exposure.
I2     can go either way
J1     yes, measurement errors due to recall bias
J2     can be either under- or overestimation of the exposure
K1     yes, personal interview, so observer has much influence on the way the questions is asked and is not blinded for case/
       control status
176          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 177 =================================================================

<br><br>====================================================================== Pagina 178 ======================================================================

<pre>K2      overestimation because interviewer could (subconsciously) also be looking for effect higher phone use: greater risk
        cancer
L1      yes, e.g. no info known on occupations situations and exposure to other substances that can influence cancer
L2      sex, age, residential areas and education level yes, adjustment for hearing loss and tinnitus, use in rural or urban area
M1      yes, depending on memory phone use misclassified in wrong exposure group?; use different types of phones, which how
        long and when exactly used,
M2      recall problems, different type of phones used, duration of use wrongly assigned, recall bias due to occupation and other
        exposure factors?
N1      majority personal interview, 5% cases and controls interviewed by phone; 1% cases and 7% controls mailed
        questionnaire
N2      mailed questionnaire has advantage of minimising observer bias, but personal interview allows clarification of unclear
        questions and probing so hopefully you still get the right answer.
N3      person dose
N4      59% regular use mobile phone, comparable cases, but many fewer people with long use > 10 years
N5      yes
O       possibly yes (recall problems)
P       no, by laterality analysis yes, because controls randomised in different groups
Q1      analysis adjusted for age, sex, residential area and education
Q2      for large differences poss. stratify and present results per category
R1      unconditional logistic regression analysis
R2      ok, controls not matched
S1      95%CI
S2      ok
T       no, this is to mean total number of controls assumed, unclear how they got assembled precisely (which were matched to
        which diagnose e.g.), table 2 and 3 almost no summing of subcategories is right… How can this be?? missing values??
U1      no increased risk of mobile phone use and acoustic neuroma, however suggestion of increased risk > 10 year
U2      see Annex H
U3      cannot say, possible exposure before development of tumour, but given the short duration of phone use and long latency
        time for development of tumour possibly exposure only after start subclinical phase tumour
U4      possibly: longer use, so more exposure, higher risk
V       most bias can go either way, but most likely the results were overestimated due to overestimation of exposure by the
        cases (even so mobile phone), possibly also influenced by interviewer?
W       can go either way, depends on errors in recall bias and accompanying misclassification of exposure, possible
        overestimation
X1      yes
X2      probably not because at the time not much was known about this topic, did not do own literature search, however
        relatively few references in total, only 1 hard ell article in refs
Y       3 regions used in study, not clear if 3 regions are a good reflection of all of Sweden, e.g. for urban-rural and occupations
Z       too mild given results?
Table F9 Extractions from Sadetzki S, Chetrit A, Jarus-Hakak A, e.a. Cellular phone use and risk of benign and malignant
parotid gland tumors--a nationwide case-control study. Am J Epidemiol, 2008; 167(4): 457-467.85
A1      to assess the association between cellular phone use and development of parotid gland tumours
A2      is there an association between cellular phone use and development of parotid gland tumours
A3      do patients with meningioma, glioma, acoustic neuroma or parotid gland tumours have higher mp use
B1      nationwide population based case-control study in Israel
              Results of the data extraction                                                                                    177
</pre>

====================================================================== Einde pagina 178 =================================================================

<br><br>====================================================================== Pagina 179 ======================================================================

<pre>B2  yes
B3  observational cohort, so minimising all sorts of bias (experiment would be best but not feasible)
B4  info mobile phone use retrospectively asked, highest category of use > 10 year
B5  group > 10 year yes, but only relatively few cases in this group, but relatively many heavy users in <10 year group,
    possibly promotor function i.s.o. initiation
C1  all incident cases of PGT diagnosed in Israel at age 18 years or more, in 2001-2003, all 22 otolaryngology departments
    throughout the country participated, all Jewish (not Arab) patients with (confirmed) tumour aged 18-59 between
    jan.2001-dec.2003,
C2  ok, nationwide
C3  no
D1  all controls Interphone Israel used, resulting total ratio 1:3
D2  ok
E1  develop fear for mobile phone use and burden for cases
E2  yes
F1  all otolaryngeal units in Israel, all Jewish (not Arab) patients with (confirmed) tumour aged 18-59 between 2001-2003;
    controls from whole country from population registry, up to 7 controls potentially assigned to a case (?)
F2  all residents? Checking against the cancer registry for missed cases in e.g. mortality (inoperable so not referred to
    specialist unit?): this is probably marginal though
G1  cases 87%, controls 66%
G2  cases sufficient, controls much too few
G3  460 cases (58 malignant, 264 pleomorphic, 117 warthins tumour and 21 others) and 1266 controls
G4  n.a.
G5  highest category > 10 years, long enough, but still relatively small numbers
G6  n.a.> 10 years category
H1  unclear, all incident cases included, but not clearly how many e.g. deceased, n=531, is probably the group where cases
    that did not fulfil all inclusion criteria have been removed, how many deceased, how many too sick? Is this last group in
    the refusals?
H2  unclear, if 531 were all cases, than no selection bias; if underrepresentation of iterant workers than underrepresentation
    of heavy mobile phone users in controls so underestimation of effect.
I1  very high refusal rate; refusers that were interviewed seemed 'less connected': systematically different from total
I2  participating controls particularly users, gives underestimation
J1  the ill could be over representing their exposure plus proxy interviews and telephone interviews would be different also
    (more proxy for cases, more phone for controls and always in questionnaire research as people answer what they think
    you want to hear
J2  if cases report higher use, than overestimation of risk
K1  yes, personal interview, so observer has much influence on way of asking questions and has not been blinded for case/
    control status
K2  overestimation because interviewer possibly (subconsciously) is also looking for effect of higher phone use: greater
    chance of cancer
L1  sex, age, year of interview, ionizing radiation, SES
L2  yes
M1  yes, depends on memory of phone use thus allocated to wrong exposure group?; use of different types of phones, which
    how long and when precisely, cases have been histologically verified, so probably no misclassification in this aspect
M2  recall problems, different types of phones used, duration of use incorrectly allocated
N1  face to face interview
N2  questionnaire to avoid observation bias, but best will be to get information from registries about phone use from telecom
    companies
178       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 179 =================================================================

<br><br>====================================================================== Pagina 180 ======================================================================

<pre>N3      person dose
N4      regular use cases: 285 (62%) and controls 691 (55%)
N5      somewhat but limited as it relies on memory and personal estimation
O       possibly yes (recall problems)
P       not to be expected
Q1      adjustment for age, sex, year of interview (adjustment for ethnic origin did not influence the results, so not included)
Q2      ok, pos stratify if there are big difference, e.g. in sex
R1      conditional logistic regression analysis
R2      yes as individually matched
S1      95% CI
S2      ok
T       yes
U1      increased risk estimates were found for ipsilateral regular use 5 and 10 years in the past, although the latter was based on
        small numbers, significantly elevated odds ratios were observed consistently in the highest category of each of the
        measures of cellular phone use on the ipsilateral side, supporting a dose-response association.
U2      see Annex H
U3      for longer latency group >10 year yes, for < 10 year this is debatable
U4      yes
V       adjusted for important known confounders, but recall particularly for cases and non response for controls can
        respectively over and under estimate results underestimated were
W       can go either way, depends on errors in recall bias and associated misclassification in exposure, possibly overestimation
        in assessment of higher exposition cases
X1      yes
X2      no
Y       somewhat but exposure levels (and possibly output power levels) are higher in Israel than elsewhere
Z       yes, they seriously consider particularly a recall bias among the cases which might exaggerate the assoc
Table F10 Extractions from Takebayashi T, Varsier N, Kikuchi Y, e.a. Mobile phone use, exposure to radiofrequency
electromagnetic field, and brain tumour: a case-control study. Br J Cancer, 2008; 98(3): 652-659.88
A1      to investigate whether mobile phone use increased brain tumour risk in Japan
A2      mobile phone use increases brain tumour risk in Japan
A3      mobile phone use increases brain tumour risk in Japan
B1      population-based case-control (in several departments region Tokyo)
B2      yes
B3      observational cohort, so minimising all sorts of bias (experiment would be best but not feasible)
B4      info mobile phone use retrospectively asked, highest usage category > 10 years
B5      to short follow-up < 10 years and group > 10 years has only very few cases
C1      newly diagnosed meningiomas, gliomas, and pituitary adenomas aged 30-69 who were treated in the 21 participating
        hospitals between 1 Dec. 2000 to 30 Nov. 2004
C2      nationwide
C3      no
D1      ratio 1:4 according to text, but given very high non-response about 1:2
D2      1:3 for a bit more power
E1      development of fear for mobile phone use and burden of cases
E2      yes
              Results of the data extraction                                                                                    179
</pre>

====================================================================== Einde pagina 180 =================================================================

<br><br>====================================================================== Pagina 181 ======================================================================

<pre>F1  area of Tokyo, including 23 wards (metropolitan area) and 14 cities (municipal area) and 25 cities adjacent to Tokyo; see
    and c but not all cases histologically verified, some diagnosed more than 6 months before start of study (those were
    eliminated but that means some less than 6 months pre were still in the study)
F2  nationwide
G1  cases 58,7% glioma, 77,6% meningioma, 75,6% pituitary adenoma, controls 52,5% glioma, 51,6% meningioma, 49,4%
    pituitary adenoma
G2  glioma cases and all controls much too few, meningioma cases and pituitary cases just too few, but better than glioma
    cases
G3  83 glioma, 128 meningioma, 101 pituitary adenoma; 208 controls
G4  n.a.
G5  to short follow-up for development of cancer, to small numbers in long follow-up group
G6  n.a.> 10 years category
H1  yes, unclear of the mentioned wards cover the area of Tokyo, no check with e.g. a cancer registry (30 out of 172
    departements in Tokyo treated 90% of brain tumour in the area, only 21 participated)
H2  can go either way
I1  yes, particularly for glioma and controls low response, but also meningioma and pituitary somewhat marginal
I2  the questions whether for controls particularly those participated who are users, or particularly not as they were young
    workers who are possibly high users of mobile phones than non working people??
J1  yes, due to recall problems and incorrectly estimated SAR
J2  as cases report higher use, overestimation results of the risk, and if particularly non-respondent controls were users that
    would also result in overestimation
K1  yes, personal interview, so observer has much influence on way of asking questions (and is not blinded for case/ control
    status?)
K2  overestimation effect because interviewer is possibly (subconsciously) also looking for effect higher phone use: greater
    risk of cancer
L1  sex, age, ionizing radiation, SES, occupation marital status
L2  yes
M1  yes, depending on memory of phone use classified in the wrong exposure category?; use of different types of phones,
    which how long and when precisely, many cases have been histologically verified, so probably no misclassification here,
    and global estimated SAR values can contain much misclassification
M2  recall problems, use of different types of phones, duration of usage wrongly classified
N1  face to face interview
N2  questionnaire to avoid observation bias, but best will be to get information from registries about phone use from telecom
    companies
N3  person dose
N4  regular use cases: glioma 68%, meningioma 43% and pituitary 61%,
    controls glioma 65%, meningioma 52% and pituitary 65%
N5  yes
O   possibly yes (recall problems), and errors in SAR measurement
P   not to be expected
Q1  adjusted for educational level and marital status and matched on age, sex and residency
Q2  ok, assuming analyse stratified for matching variables, but has not been clearly stated
R1  conditional logistic regression analysis
R2  ok, because controls are matched
S1  95% CI
S2  ok
T   yes
180       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 181 =================================================================

<br><br>====================================================================== Pagina 182 ======================================================================

<pre>U1      no consistent increase was observed in the overall risk of glioma or meningioma among mobile phone users, nor
        increasing trend in risk in relation to cumulative length of use or cumulative call time; no substantial increase in risk was
        observed for glioma or meningioma
U2      see Annex H
U3      cannot know, possible exposure for development of tumour, but given short duration of use and long latency time for
        development of tumour probably exposure after start subclinical phase tumour
U4      no
V       it has been adjusted for confounding by education and marital status, but results can certainly be biased by recall,
        possibly resulting in overestimation, additionally high non-response for controls, can lead to over- and underestimation,
        cannot know, hardly info non responders
W       can go either way, depends on errors in recall bias and associated misclassification of exposure, possibly overestimation
        by estimation of the highest exposed cases
X1      yes
X2      no
Y       population of the area of Tokyo, or Japan if this area is representative for Japan
Z       they put a lot of store in it being similar results to the others (that are also too small to come to a conclusion)
Table F11 Extractions from Schoemaker MJ and Swerdlow AJ. Risk of pituitary tumors in cellular phone users: a case-control
study. Epidemiology, 2009; 20(3): 348-354.89
A1      specific tumour location, could be associated with mobile phone use
A2      association between having tumour and reporting mobile phone use.
A3      as in A2
B1      case-control
B2      ok
B3      cohort
B4      n.a.
B5      n.a.
C1      no of cases in study area
C2      larger or longer
C3      not presented in this paper but done earlier
D1      1:4
D2      is ok
E1      burden for patients
E2      yes
F1      cancer registry data
F2      ok
G1      63% of cases and 43% for controls
G2      ok but not lush
G3      317 cases and 630 controls
G4      n.a.
G5      n.a
G6      n.a
H1      good response rates so not overly likely
H2      n.a.
I1      yes
              Results of the data extraction                                                                                     181
</pre>

====================================================================== Einde pagina 182 =================================================================

<br><br>====================================================================== Pagina 183 ======================================================================

<pre>I2     could go either way
J1     yes some
J2     could go either way
K1     yes certainly
K2     could go either way
L1     age, sex and SES
L2     age/sex=x yes
M1     slightly
M2     anyways, most often underestimation
N1     questionnaire
N2     checking against bills
N3     person-does
N4     64%
N5     yes, theoretically
O      always some
P      slight
Q1     regression
Q2     ok
R1     regression coefficient or OR
R2     ok
S1     85% CI
S2     ok
T      ok
U1     no association seen
U2     see Annex H
U3     n.a
U4     no
V      underestimation
W      some is possible as always
X1     ok
X2     no
Y      similar countries
Z      ok
Table F12. Extractions from INTERPHONE study group. Brain tumour risk in relation to mobile telephone use: results of the
INTERPHONE international case-control study. Int J Epidemiol, 2010; 39(3): 675-694.93
A1     to determine whether mobile phone use increases the risk of these tumours and, specifically, whether RF energy emitted
       by mobile phones is tumourigenic.
A2     null hypothesis of no association would be expected to produce an approximately symmetric pattern of negative and
       positive log ORs.
A3     is there an (positive or negative) association between mobile phone use and brain cancer?
B1     international population-based case-control study in sixteen study centres from 13 countries
B2     yes
B3     observational cohort, so minimising all sorts of bias (experiment would be best but not feasible)
B4     info mobile phone use retrospectively asked, highest usage category > 10 years
182          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 183 =================================================================

<br><br>====================================================================== Pagina 184 ======================================================================

<pre>B5 reasonable, mobile phones only recently in wide spread use, only a few cases that have long (>10 years) mobile phone
   use, category > 5 years use relatively large number of cases and controls to come to conclusions
C1 all eligible cases with glioma or meningioma of the brain diagnosed in the study regions during study periods of 2-4
   years between 2000 and 2004, aged 30-59
C2 ok
C3 no
D1 ratio 1:1, and ratio Germany 1:2. 7 centres individual matching, frequency matching elsewhere
D2 ok, sufficient power due to large numbers
E1 development of fear for mobile phones and burden for cases
E2 not found in text but as far as I know the part studies all had
F1 16 study centres in 13 countries (Australia, Canada, Denmark, Finland, France, Germany, Israel, Italy, Japan, New
   Zealand, Norway, Sweden and the UK), aged 30-59 years, mainly large urban areas, all patients with glioma diagnosed
   2000-2004 (different years in this period for the study centres, 2-4 years for each centre), cases from all neurological and
   neurosurgical facilities (bar in Paris and Tokyo where some did not participate); controls selection as locally appropriate
F2 non-neurological clinics as some case may not have made it in? (checked in cancer registry so maybe not that much of
   an issue?) some cases are totally missed but no other strategy would have gotten those. Worry about catchment area of
   what are mostly tertiary clinics: unlikely to be a small geographic area so would need exclusions to match with possible
   control selection, main problem is with control selection is several areas. also note that problems are listed for Paris and
   Tokyo but the German study also had incomplete case ascertainment (see 8051)
G1 response meningioma cases: 78% (range 56-92), glioma cases 64% (36-92), controls, 53% (42-74) (analyse matched
   sets only, some smaller numbers used)
G2 no, not really, the lower ends of the ranges are much too low to ensure that there is no selection bias, the upper range
   would have been fine but the averages are not great
G3 2409 (i.s.o. 2425) meningioma, 2662 matched controls/ 2708 (i.s.o. 2765) glioma, 2972 matched controls
G4 n.a., highest category: > 10 year mobile phone use
G5 not even 5% of all meningioma cases and not even 10% of all glioma cases have > 10 years mobile phone use, so it stay
   relatively small numbers, group that has 5-9 year use of mobile phones is substantially larger
G6 n.a.> 10 years category
H1 yes given the very poor response rates this seems likely, also the results are mostly driven by 2 countries (UK and
   Australia) and the control selection there is highly selective for SEC particularly (the control selection for Australia is
   not described in a separate article and cannot be traced at this time but there is no proper control selection method in
   Australia)
H2 underestimation as it is likely to make cases and controls more alike; also worrying is the reasons for non-response as far
   as known: to ill (1-20%), refusal (11-30%) and not reached (5-15%)
I1 yes, some countries very low response
I2 can go either way, particularly users of mobile phones participated? Deceased, so worst glioma cases particularly the
   group most intensively and longest mobile phone use? Particularly working young population that uses mobile phones a
   lot in the non responders group?
J1 yes, measurement errors in exposure variables due to recall bias, all cases histologically verified or based on
   unequivocal diagnostic imaging, so chance of information bias here probably small
J2 could be either underestimation or overestimation of the exposure, for cases the expectation is for overestimation
K1 as the study used interviews at home, this could have been very substantial as interviewers were not blinded, even
   though there is no observation as such, the questions could have been given a leading tone, emphasis or more detail
   could have been sought of the cases than of the controls
K2 overestimation effect because interviewer possibly (subconsciously) also searching for effect higher phone use: greater
   chance of cancer
L1 age, sex, educational level, occupation
L2 yes
          Results of the data extraction                                                                                   183
</pre>

====================================================================== Einde pagina 184 =================================================================

<br><br>====================================================================== Pagina 185 ======================================================================

<pre>M1     yes, depends on memory phone use wrongly allocated to an exposure group?; use different types of phones, which how
       long and when exactly used
M2     recall problems, different types of phones used, duration of use wrongly classified, recall bias for occupational and other
       exposure factors?
N1     personal interview
N2     questionnaire to avoid observation bias, but best will be to get information from registries about phone use from telecom
       companies, than you will also avoid recall bias
N3     person dose
N4     regular use meningioma cases: 52,4% and controls 55,9%; glioma cases 61,5% controls 63,7%
N5     yes
O      possibly yes (recall problems), probably not for case ascertainment, mostly using (?) pathology reports
P      not to be expected
Q1     adjustment for sex, age, study centre, ethnicity in Israel, education
Q2     ok, possibly stratify if there are large differences for e.g. sex or centre
R1     conditional logistic regression analysis on the matched case-control datasets
R2     ok
S1     95%CI
S2     ok
T      yes
U1     quote: For meningioma, there is little evidence to counter a global null hypothesis, and we conclude that INTERPHONE
       finds no signs of an increased risk of meningioma among users of mobile telephones. For glioma, an increased OR was
       seen in analyses in the highest decile of cumulative call time, including tumours in the temporal lobe and subjects who
       reported having used the mobile phone mainly on the same side as where the tumour occurred. Still, the evidence for an
       increased risk of glioma among the highest users was inconclusive, as the increase could be due to one or more of the
       possible sources of error discussed ....
U2     see Annex H
U3     probably for a part of the cases, but for another part of the cases the latency period is too short
U4     no
V      most bias can go either way, but possibly the results are overestimated due to e.g. the overestimation of exposure by
       cases, possibly additionally influenced by the interviewer?, but also possibly underestimated by the very substantial non-
       response
W      can go either way, depends on errors in recall bias and accompanying misclassification of exposure, possibly
       overestimation effect due to overestimation exposure in cases
X1     yes
X2     no
Y      participating countries? Depends a bit on the size of the differences between the countries, whether you can generalise
       the overall results over the countries
Z      yes
Table F13 Extractions from INTERPHONE study group. Acoustic neuroma risk in relation to mobile telephone use: Results of
the INTERPHONE international case-control study. Cancer Epidemiol, 2011.94
A1     is AN caused by use of mobile phones
A2     is having AN associated with a history of using a mobile phone
A3     is having AN associated with a history of using a mobile phone
B1     case-control
B2     ok
184          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 185 =================================================================

<br><br>====================================================================== Pagina 186 ======================================================================

<pre>B3 n.a.
B4 n.a.
B5 n.a.
C1 available areas etc
C2 ok
C3 in a previous paper
D1 1:2
D2 ok given large numbers
E1 burden for patients
E2 in individual papers it did say so mostly
F1 16 sites in 13 countries
F2 wider region?
G1 82% for cases (70-100%) 53% for controls
G2 quite poor for the controls
G3 1105 cases 2145 controls
G4 n.a.
G5 n.a.
G6 n.a.
H1 certainly as poor response rates
H2 underestimation?
I1 yes
I2 can’t tell, either way
J1 yes
J2 can’t tell, either way
K1 yes
K2 can’t tell, either way
L1 age, sex, SES
L2 yes
M1 some
M2 various
N1 interview
N2 more checking with bills?
N3 person-dose
N4 1308 / 2145 = 61%
N5 theoretically
O  some as measurement imprecise
P  no
Q1 regression
Q2 ok
R1 regression coefficient / OR
R2 ok
S1 95% CI
S2 ok
T  ok
         Results of the data extraction      185
</pre>

====================================================================== Einde pagina 186 =================================================================

<br><br>====================================================================== Pagina 187 ======================================================================

<pre>U1     there was no increase in risk of acoustic neuroma with ever regular use of a mobile phone or for users who began regular
       use 10 years or more before the reference date.
U2     see Annex H
U3     unclear
U4     no
V      some left due to selection bias and observer bias
W      underestimation
X1     ok
X2     no
Y      quite widely
Z      ok
              Other case-control studies
Table F14 Extractions from Muscat JE, Malkin MG, Thompson S, e.a. Handheld cellular telephone use and risk of brain cancer.
JAMA, 2000; 284(23): 3001-3007.96
A1     public health concerns about the safety of cellular telephones
A2     using handheld cellular telephones is related to the risk of primary brain cancer
A3     is using handheld cellular telephones related to the risk of primary brain cancer
B1     case-control
B2     yes
B3     poss. cohort, however given low incidence one would need a long time to get enough cases
B4     highest category >= 4 years, in US start cellular phones in 1984
B5     still very short
C1     unclear description of how group exactly defined, eligible cases diagnosed as having primary brain cancer within the
       past year (which last year? Interviews have been conducted between '94 and '98) and spoke English
C2     deceased patients not in study, should include those, now exclusion worst cases, spoke English actually vague definition:
       how well?
C3     no
D1     ratio 1:1, frequency matched by age, sex, race, month of admission, hospital
D2     poss. ratio 1: 2 given relatively small numbers and now wide confidence intervals
E1     burden for hospital patients, both cases and controls (have another reason for hospital visit)
E2     not mentioned
F1     New York (Memorial Sloane Kettering) cancer centre, NY university medical centre and Columbia University
       Presbyterian hospital), Providence (Rhode Island hospital), Boston (Massachusetts General Hospital)
F2     population controls as the hospitals used were tertiary specialist units for all specialities so 'normal' cases might not be
       present as many more hospitals present in NY
G1     not presented, written as if 100% response rate in both cases & controls but Response rate cases: 82% (469/571; 2 dead,
       25 refused, 75 to ill), (97 not approached as to ill or do not speak English); response rate controls: 90%
G2     for cases certainly not, for controls ok
G3     469 cases and 422 controls
G4     17 cases (3,6%) and 22 controls (5,2%) >= 4 years use and 2-3 years follow-up: 6% cases and 5.7 controls
G5     very small numbers for long follow-up, cannot say anything about longer duration and short duration only effect on
       speeded up subclinical stages instead of development of new tumour
G6     ≥4 years highest category
186           Mobile phones and cancer
</pre>

====================================================================== Einde pagina 187 =================================================================

<br><br>====================================================================== Pagina 188 ======================================================================

<pre>H1 yes, 97 not approached, particularly the most ill and the group that does not speak English (so possibly the group with
   lower SES)
H2 can go either way, depends if the most ill particularly use or not use mobile phones, not speak English, lower SES
   probably lower use of mobile phones (particularly in the early years before the use became wide spread)
I1 yes 82% response cases not terribly bad, but still a substantial group non-responders that can differ systematically from
   responder cases, also for controls responders bias possible, chance smaller as response higher at 90%
I2 possibly particularly users of mobile phones participated? In that case overestimation risk
J1 yes, particularly due to recall bias, in text spearman correlation coefficients calculated and recall seems ok when
   compared to hours registered on accounts (is however also an estimate because accounts were not well traced at the
   telecom company), however recall will always play a role in retrospective investigations, also on the bills no info about
   call received, not described how all is comparable.
J2 can go either way but most likely overestimation of exposure by cases
K1 yes
K2 overestimation effect if interviewer convinced of existence of possible effects
L1 yes, (matching variables: age, sex, race, hospital) en potential confounders: SES, medical history, occupational exposure
L2 yes, except SES
M1 yes, due to recall bias
M2 all sorts of errors can occur in the measurement of the exposure (recall, type phone, how many minutes, how often,
   which ear used, how much with a 'cord', how much direct exposure to the head...)
N1 structured interview
N2 registrations via telecom companies, than no more information and observation bias
N3 person-dose
N4 cases: 14.1% user;
   controls 18,0% user
N5 yes, but with loads of issues attached
O  possibly yes due to information bias
P  not to be expected
Q1 multivariate analysis, in which adjustment for confounders is contained in the model and stratify
Q2 ok
R1 multivariate unconditional logistic regression analysis and test for trend and nonparametric regression analysis
   (alternative method assessing dose response relationship)
R2 ok, because frequency matched and not individually matched
S1 95% CI
S2 ok
T  yes, but some numbers missing, e.g. how many potential cases there were.
U1 use of handheld telephones is not associated with risk of brain cancer
U2 see Annex H
U3 is the question, because unclear when tumour developed exactly and if all types of exposure really did occur before
U4 no, still to small numbers and to wide confidence intervals to be able to say anything about this, effect cannot be
   excluded but these numbers of not indicate a dose-response relation
V  has been corrected for as analysis was multivariate, so in theory clean OR
W  can go either way depending on recall and allocation to user categories, possible overestimation exposure by cases so
   overestimation risk, but OR already below 1
X1 yes
X2 as far as I can see no
Y  to patients of the hospitals involved but too many problems to generalise
Z  reasonably as they consider the need for longer duration studies
         Results of the data extraction                                                                                  187
</pre>

====================================================================== Einde pagina 188 =================================================================

<br><br>====================================================================== Pagina 189 ======================================================================

<pre>Table F15 Extractions from De Roos AJ, Teschke K, Savitz DA, e.a. Parental occupational exposures to electromagnetic fields
and radiation and the incidence of neuroblastoma in offspring. Epidemiology, 2001; 12(5): 508-517.97
A1      determinants of neuroblastoma
A2      what is the (parental) mobile phone history In children with neuroblastoma
A3      as A2 but way too early for exposure to mps
B1      case-control
B2      ok
B3      cohort but would need to be extremely large
B4      n.a.
B5      n.a.
C1      total number of eligible pateints
C2      longer duration
C3      not presented
D1      1:1
D2      ok but generally assumed 1:4 better
E1      burden for patients
E2      not mentioned but assumed to be ok
F1      patients at 139 hospitals in the US, less than 19 yrs of age, 01/05/92-30/04/94
F2      newer as for mobile phones this is way too early
G1      73% of cases and 74% for controls
G2      yes
G3      n.a.
G4      n.a.
G5      n.a.
G6      n.a.
H1      very small number but good response rate so possibly not too bad . however, to be eligible many criteria were applied
H2      could go either way
I1      yes as ill
I2      could go either way
J1      yes as ill
J2      could go either way
K1      yes as interviews
K2      could go either way
L1      yes as always as poorly measured
L2      yes as far as possible
M1      yes
M2      many as poorly measured
N1      interview
N2      more elaborate but it was a tiny bit of many other interests
N3      person dose but poorly specified
N4      4 / 503 so minimal
N5      not really
O       as always
P       as always
188           Mobile phones and cancer
</pre>

====================================================================== Einde pagina 189 =================================================================

<br><br>====================================================================== Pagina 190 ======================================================================

<pre>Q1     regression
Q2     ok
R1     regression coeff / OR
R2     ok
S1     95% CI
S2     ok
T      numbers seem to add up
U1     overall, there was scant supportive evidence of strong associations between parental exposures in electromagnetic
       spectrum and neuroblastoma in offspring. (quote)
U2     see Annex H
U3     no
U4     no
V      poor measurements so could go either way
W      poor measurements so could go either way
X1     ok
X2     ok
Y      limited as study is limited
Z      ok
Table F16 Extractions from Stang A, Anastassiou G, Ahrens W, e.a. The possible role of radiofrequency radiation in the
development of uveal melanoma. Epidemiology, 2001; 12: 7-12.98
A1     interest in determinants of uveal melanoma and different sources of radiation
A2     what is the mobile phone (and other determinants) history in people with uveal melanoma
A3     A2 tested
B1     case-control
B2     ok
B3     cohort but would have to be extremely big
B4     n.a.
B5     n.a.
C1     total number of incident cases
C2     more hospitals, wider region?
C3     not presented
D1     1:12 for hospital study and ca 1:2 for population study
D2     1:4 is considered optimal
E1     burden for pateints
E2     not mentioned but assumed
F1     mixed model of hospital cases, hospital, family and populations controls for limited regions in Germany
F2     clearer choices and good population controls selection
G1     hospital based study: cases 84%, controls 48%; population based study cases 88% controls 79%
G2     population study yes, hospital controls response is poor
G3     57 cases and 699 controls in hospital study; 81 cases and 148 controls in population study
G4     n.a.
G5     n.a.
G6     n.a.
             Results of the data extraction                                                                              189
</pre>

====================================================================== Einde pagina 190 =================================================================

<br><br>====================================================================== Pagina 191 ======================================================================

<pre>H1     given poor response rates: yes
H2     could go either way
I1     yes
I2     could go either way
J1     yes
J2     could go either way
K1     as there were interviews: yes
K2     could go either way
L1     age, sex
L2     yes
M1     somewhat
M2     misunderstanding questions etc
N1     mostly interviews, some questionnaires
N2     all one or the other as this mix makes it hard to interpret
N3     personal dose
N4     person-dose
N5     12% (?)
O      as always
P      as always
Q1     regression
Q2     ok
R1     regression coeff/ OR
R2     ok
S1     95% CI
S2     ok
T      numbers do seem to add up
U1     we found an increased risk of uveal melanoma in relation to RFR as transmitted by radio sets and mobile phones. The
       association between electromagnetic fields and uveal melanoma was limited to RFR (quote)
U2     see Annex H
U3     some indications but unclear
U4     not clearly
V      could go eithr way
W      yes and could go either way
X1     ok
X2     ok
Y      limited as small scale study
Z      ok
Table F17 Extractions from Inskip PD, Tarone RE, Hatch EE, e.a. Cellular-telephone use and brain tumors. N Engl J Med,
2001; 344(2): 79-86.99
A1     because of concern about the risk of brain cancer associated with the use of hand-held cellular phones
A2     recent use of hand-held cellular telephones causes brain tumours
A3     does recent use of hand-held cellular telephones cause brain tumours?
190          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 191 =================================================================

<br><br>====================================================================== Pagina 192 ======================================================================

<pre>B1 case-control
B2 yes
B3 poss. cohort, however given low incidence one would need a long time to get enough cases
B4 n.a., cases and controls are at most allocated to use >= 5 years
B5 to short for the development of brain tumours, but assumed that if magnetic fields cause cancer, they act at a late stage in
   the process sand than it could potentially have an influence, exact mechanism unclear
C1 power calculation
C2 for subgroup analysis to small numbers still
C3 yes
D1 ratio cases: controls: 1:1
D2 power calculation done: sufficient power, so good ratio, however for subgroup analysis not sufficient power
E1 burden for hospital patients, however can refuse cooperation if they want, extra blood sampling for investigation
E2 yes, in this article can be found that institutional review boards approved the protocol
F1 2 hospitals in Arizona, 1 in Boston, 1 in Pennsylvania, all newly diagnosed cases over 4 years; controls admitted to same
   wards + general surgical, urology, cardiac, pulmonary, gastrointestinal & trauma
F2 nationwide, particularly include smaller hospitals, would give better reflection of population?, these 3 centres are truly
   referral hospitals
G1 cases: 92 %; controls 86 %
G2 yes, relatively high, although a non-response analyses would be preferable
G3 782 cases and 799 controls
G4 maximal category >= 5 years phone use, but all sorts of variables for exposure asked far back
G5 no, really too short for cancer to develop, possibly influence on the speeding up of a sub clinical state of the cancer
G6 category >= 5 years
H1 yes, but unclear which cases and controls did not participate and how many that were. E.g. why has not everyone been
   asked by a doctor? Also possibly selection bias because only large urban hospitals included. However, tight protocol
   about who is and who isn't included so hopefully no selection bias due to choice of doctors themselves if someone was to
   participate or not in the study
H2 possible so very small
I1 yes, 92 and 86% are quite high %, but also this can still contain bias, e.g. particularly people with an affinity with the
   topic are more prepared to participate
I2 relatively more people who use mobile phones may have participated? However, effect is not there so will not have
   changed this much
J1 yes, recall bias can be a big problem here, but is minimized by often having a partner present at the interviews, but
   maybe this was less often the case for controls.
J2 over reporting by cases for the various exposures assessed
K1 on the one hand yes because the interviewer has influence on the way of asking the questions and if needed explain
   them, on the other hand no because it is all according to a strict protocol
K2 overestimation of effect if interviewer convinced of existence of possible effects, however also taped interviews checked
   so probably no effect
L1 matching variables: (age, sex, race, hospital and distance to hospital) and education, self-reported income, date of
   interview, interview respondent
L2 yes
M1 yes
M2 limitations to capture historical changes, inaccuracies in recall, variations in levels of exposure, different types of
   telephones and different circumstances of use. (misclassification mainly in level of use than in use itself)
         Results of the data extraction                                                                                    191
</pre>

====================================================================== Einde pagina 192 =================================================================

<br><br>====================================================================== Pagina 193 ======================================================================

<pre>N1     interview and questionnaire
N2     very many exposures depend on memory, so use of registries would be better, e.g. phone companies and poss.
       Registration of exposure through occupation in registries? (e.g. dosimetry for people that worked with X-rays)
N3     person-dose
N4     358 of 799 used mobile phone, 172 regular use
N5     yes
O      possible as recall plays a role, but very detailed reconstruction asked for
P      no, not to be expected. Just normal random error
Q1     confounders included in model for logistic regression and thus adjusted OR´s
Q2     ok
R1     conditional logistic regression
R2     ok, because controls are matched
S1     95% CI
S2     ok
T      yes
U1     the study does not support the view that exposure to low-power microwave radiation from hand-held, analogue cellular
       telephones causes malignant or benign tumours of the brain or nervous system (note says nothing about long term and
       enormous increase in use in whole population)
U2     see Annex H
U3     is the question because unclear when tumour exactly developed and/if all sorts of exposure did occur before than
U4     yes
V      has been corrected for
W      can go either way depending on recall and allocation to user categories
X1     as far as I can see yes; ok (points to specific no-effect literature and wireless company literature)
X2     as far as I can see no
Y      urban US population of the three 3 regions
Z      no as they cannot prove or disprove the association given the lack to latency time in the study
Table F18 Extractions from Auvinen A, Hietanen M, Luukkonen R, e.a. Brain tumors and salivary gland cancers among cellular
telephone users. Epidemiology, 2002; 13(3): 356-359. 100
A1     possible health hazards of radiofrequency electromagnetic fields emitted by cellular phones
A2     not really formulated, but is about the question : increased risk brain and salivary gland tumour in cellular telephone
       users
A3     is the risk for brain and salivary gland tumour increased in cellular telephone users?
B1     case-control
B2     yes
B3     design is ok, reasonably efficient because uses registries that have been linked, however many issues with the method to
       answer the study questions, case control with individual exposure data or prospective cohort
B4     couple of years, average duration of subscription 2-3 year for analogue & less than 1 year for digital;
       highest category used > 2 years
B5     not at all
C1     all cases from population based Finnish cancer registry, 5 controls per case
C2     ok
C3     yes, to detect an OR of 1,4 or higher for brain tumours and 2,8 or higher for salivary gland cancers with α = 0,05, two-
       sided and 1-β = 0,8
192          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 193 =================================================================

<br><br>====================================================================== Pagina 194 ======================================================================

<pre>D1 ratio 1: 5 not described why (undoubtedly to do with power calculation but it is not presented)
D2 fine ratio, possibly 1:3 or 1:4 ok also good?? 5 controls is quite much
E1 that is done without consent, participants are not aware of all the privacy sensitive data that have been used for this
   investigation
E2 yes
F1 Finnish Cancer Registry (cases) and Population Registry Centre of Finland (controls); all cases in Finland in 1996,
   controls from population registry
F2 ok, however this way exposure cannot be asked back in detail (exposure now via subscribers list from the 2 cellular
   network providers)
G1 n.a., registry data
G2 n.a.
G3 cases: 432 (398 brain tumour and 34 salivary gland) controls: 2156 (1986 brain and 170 salivary gland)
G4 n.a.
G5 no highest category > 2 years, so much to short and very small numbers
G6 > 2 years
H1 no, all cases in registry included
H2 n.a.
I1 no, all cases in registry included
I2 n.a.
J1 yes, unclear if phone was really used by the case or control rather than e.g. a family member, also missing info about
   duration of use etc, maybe phone and phone plan was bought but was is hardly if ever used?; also very important that
   only private subscribers were included so no company subscriptions, these people are now if they are either a case or a
   control in the study taken as non-exposed
J2 overestimation exposure because you do not know for certain if the subscribers are users;
   underestimation because in unexposed group also people who do use mobile phones via a company plan and so are
   exposed
K1 no, not to be expected, all registry based
K2 n.a.
L1 overestimation because part effect due to other exposures
L2 yes some are, urban residence, SES, occupation farming or electromagnetic fields
M1 yes
M2 people labelled as exposed due to the phone provider data, while this may not be the person who actually uses the phone
   and users of company phones are missed and incorrectly labelled as unexposed
N1 subscription at telecom provider, and duration subscription; private subscription, little detail on non-private definition
   for exclusion (trades people etc?) duration was used for dose
N2 yes recall through questionnaire or interview, this information does not mean much
N3 should be person dose, but the question remains if it was the correct person
N4 13% brain cancer, 12% salivary gland and 11% controls ever had personal subscription to a cellular telephone
N5 no
O  yes, who really used the phone, case or control or maybe a family member and substantial measurement error because all
   company subscriptions are missing
P  not to be expected
Q1 not, only looked in the frequency tables if distribution for cases and controls comparable
Q2 correct in multivariate analysis
R1 conditional logistic regression
R2 ok
          Results of the data extraction                                                                                   193
</pre>

====================================================================== Einde pagina 194 =================================================================

<br><br>====================================================================== Pagina 195 ======================================================================

<pre>S1     95% CI
S2     ok
T      yes
U1     cellular phone use was not associated with brain tumours or salivary gland tumours overall, a weak association between
       gliomas and analogue cellular phones
U2     see Annex H
U3     no, exposure much too short to cause cancer
U4     no
V      overestimation because not corrected for confounders
W      can go either way, overestimation if users are not the actual users, and underestimation of unexposed people maybe use
       a company phone
X1     yes
X2     very limited number of references
Y      cannot be generalised, way too many shortcomings in this study to generalise conclusions
Z      yes as they themselves do not say they don't find an association, they realise you need better detailed data and longer
       period of observation
Table F19 Extractions from Muscat JE, Malkin MG, Shore RE, e.a. Handheld cellular telephones and risk of acoustic neuroma.
Neurology, 2002; 58(8): 1304-1306.101
A1     public health concerns about the safety of cellular telephones
A2     intracranial energy disposition from handheld cellular telephones causes acoustic neuroma
A3     intracranial energy disposition from handheld cellular telephones causes acoustic neuroma
B1     case-control
B2     yes
B3     poss. cohort, however given low incidence one would need a long time to get enough cases
B4     highest category 3-6 years use of cellular phone
B5     very short, particularly since acoustic neuroma has long latency time
C1     part of larger case-control study on brain tumour, form that this subgroup used
C2     small numbers, so use more than the indicated 2 hospitals as a source
C3     no
D1     ratio 1:1
D2     1:4 given small numbers
E1     burden for hospital patients, both cases and controls (have other reason for hospital visit)
E2     not described in the text
F1     18-80 yrs old, patients @ 3 NY, 1 RI, 1 Boston tertiary hospitals with brain tumours, diag 94-98; controls same hospitals
       daily admissions (benign illness other than 2 hospitals) excl leukaemia or lymphoma
F2     population controls as the hospitals used were tertiary specialist units for all specialities so 'normal' cases might not be
       present, in many places many more hospitals present
G1     only described that 90 cases and 86 controls were selected from a larger case control study (from 1020: Response rate
       cases: 82% (469/571; 2 dead, 25 refused, 75 to ill), (97 not approached because to ill or did not speak English); response
       rate controls: 90%)
G2     for cases certainly not, for controls ok
G3     90 cases, 86 controls
G4     only 11 (12,2%) patients and 6 (7,0%) controls have 3-6 years follow-up
G5     to begin with small numbers, miniscule small numbers in category with longest use
194          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 195 =================================================================

<br><br>====================================================================== Pagina 196 ======================================================================

<pre>G6 3-6 years highest category
H1 unknown, not described (see 1020)
H2 unknown, not described (see 1020)
I1 unknown, not described (see 1020)
I2 unknown, not described (see 1020)
J1 yes, particularly due to recall bias
J2 overestimation exposure by cases (so overestimation effect)
K1 yes because the interviewer's way of asking can influence
K2 overestimation effect if interviewer is convinced of the existence of possible effects
L1 yes, (matching variables: age, sex, race, hospital) and potential confounders: SES, medical history, occupational
   exposure
L2 yes, except SES
M1 yes, due to recall bias
M2 all sorts of errors can occur in measuring exposure and e.g. exposure to substances in occupation
N1 personal, structured interview
N2 use registrations of telecom companies and occupation related registries
N3 person-dose
N4 26,7% controls regularly using handheld cellular telephone versus 20,0 % cases
N5 potentially yes
O  bill seize is an approximation but reasonably close (not entirely matched as distance of call increases bill but not
   necessarily exposure)
P  not to be expected
Q1 multivariate analysis, with adjustment for confounders in the model
Q2 ok
R1 multivariate unconditional logistic regression analysis
R2 ok, because frequency matched and not individually matched
S1 95% CI
S2 ok
T  yes
U1 reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
U2 see Annex H
U3 reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
U4 reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
V  reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
W  reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
X1 reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
X2 reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
Y  reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
   analogue)
         Results of the data extraction                                                                                 195
</pre>

====================================================================== Einde pagina 196 =================================================================

<br><br>====================================================================== Pagina 197 ======================================================================

<pre>Z       reasonably as they consider the need for longer duration studies and the analogue/digital issue (use at the time mainly
        analogue)
Table F20 Extractions from Warren HG, Prevatt AA, Daly KA, e.a. Cellular telephone use and risk of intratemporal facial nerve
tumor. Laryngoscope, 2003; 113(4): 663-667.102
A1      to determine whether cellular telephone use is associated with an increased risk of intratemporal facial nerve tumours
A2      is cellular telephone electromagnetic radiation exposure a causative agent of facial nerve tumours?
A3      is cellular telephone use associated with an increased risk of intratemporal facial nerve tumours?
B1      hospital based case-control
B2      ok bar for inherent limitations
B3      poss. Cohort or case control over more years
B4      n.a.; cases and controls included from 1 July 1995 - 1 July 2000 and use phone , occupation, medical history, social
        habits etc retrospectively asked
B5      average number of years of use varies from 1-5,67, to short, long follow-up is for acoustic neuroma, but that is a slow
        growing tumour so also for this group to short follow-up
C1      all cases diagnosed with IFN between July 1995-2000 in the academic tertiary care medical centre
C2      small numbers, so if possible also include other hospitals or use larger region, are there other specialist centres?
C3      no
D1      1 to 12 for the non tumour controls, all acoustic neuromas (?)
D2      very few cases, so many controls needed to get some power, also included 3 different reference groups in controls
E1      burden for patients; too small so never a real result so unethical to conduct in the first place
E2      yes
F1      fiscal database at academic, tertiary-care medical centre: all newly diagnosed patients over 1 year in one (main?)
        hospital, controls from same department for both non-tumour and tumour controls, University hospital (unclear if based
        in Florida or in Minnesota, probably Florida)
F2      larger region, include more specialist hospitals
G1      not described, but all 18 cases have been included and 192/216 controls (88,9%), the intention was to use 12 controls per
        case and those cannot be traced in the tables
G2      if the numbers in the previous answer were right yes but unclear if and who were excluded
G3      18 cases, 192 controls (51 acoustic neuroma, 72 rhino sinusitis, 69 dysphonia or gastroesophageal reflux)
G4      n.a., use of mobile asked back, but only number of years of use described: (1 for cases and 1 for controls and 5,67 for
        acoustic neuroma patients), except for acoustic neuroma for the other tumours is the time to short anyway
G5      n.a.
G6      not real follow-up of course but time since first use: average 1 for IFN cases and controls and 5,67 years for acoustic
        neuroma
H1      if all 18 cases are included and if these were indeed the only cases, than not, for controls possible but unclear how people
        recruited (all people with named diagnosis of a selection?)
H2      probable so negligibly small
I1      for cases not as all 18 participated, controls only limited non-response so responder bias will be relatively small,
        however nothing presented about who the non responders are
I2      not really a large effect to be expected possibly overestimation for people with acoustic neuroma and underestimations
        rhinosinusitis (possibly also inclined to report higher exposure given illness history)
J1      yes, recall bias, although mobile phones have been used relatively recently only and memory might still be quite good
J2      possible over reporting cases INF and acoustic neuroma and rhinusitis (all in area head/ear)
K1      yes because interviewer can influence the conversation
K2      possible overestimation effect if interviewer convinced of the existence of a possible effect
196            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 197 =================================================================

<br><br>====================================================================== Pagina 198 ======================================================================

<pre>L1     many confounders measured (age, sex, occupation etc), SES,
L2     yes some have been, SES not, unclear how corrected for in multivariate regression: nowhere to be found which variables
       were included in the model
M1     yes
M2     exposure definition, duration, frequency use of phone, however in this analysis they were not used...
N1     structured interview about phone
N2     questionnaire to reduce observation bias or use telecom companies to get exact phone habits
N3     person-dose
N4     2 or 18 patients (11,1%) regular use (average 1 call a week), 11 of 51 acoustic neuroma (21,6%), 31 of 141 non-tumour
       control (22,0%)
N5     probably yes, given that many details were asked for of the various exposures, l however nothing said about in this
       article
O      possible due to recall
P      not to be expected
Q1     multivariate analysis, however nowhere to be found which variables corrected for
Q2     ok, assuming correct corrections, poss. Stratify
R1     multivariate unconditional logistic regression analysis
R2     conditional logistic regression analysis (because of matched data)
S1     95% CI
S2     ok
T      yes, except that it is unclear how they got the 192 controls with a ratio of 1 to 12 and in tables 2 and 3 the total number
       of non tumour controls doe not compute 5 times, 4 times 1 missing: probably a missing answer but also 1 time 3 controls
       to many??
U1     regular cellular telephone use does not appear to be associated with a higher risk of IFN tumour development
U2     see Annex H
U3     probably not as the period of phone use is very short and the tumour therefore probably existed a long time before the
       start of the phone use
U4     no
V      if the confounders have been corrected for in the multivariate analysis than the results are pure estimates, however
       cannot be traced if and how corrected for
W      misclassification in diagnose: nowhere described if diagnoses histologically verified and for the exposure:
       overestimation of use by the cases, so overestimation of effect?? However no effect found
X1     as far as I can see at this time yes
X2     as far as I can see not entirely, e.g. only 1 article by Hardell referenced
Y      patients of other academic tertiary care medical centres with 1 of the diagnoses used, unclear
Z      yes as they recognise that the numbers are too small for any conclusions and therefore do not present one
Table F21 Extractions from Gousias K, Markou M, Voulgaris S, e.a. Descriptive epidemiology of cerebral gliomas in Northwest
Greece and study of potential predisposing factors, 2005-2007. Neuroepidemiology, 2009; 33(2): 89-95.103
A1     the aim of the study was to investigate the epidemiologic and clinical characteristics of glioma patients in a defined area
       of northwest Greece with a total population of about 500,000 inhabitants
A2     not really stated but is about the question : cellular telephone use increases the risk of brain tumour? (and descriptive
       incidence rate)
A3     cellular telephone use increases the risk of brain tumour?
B1     case-control study
B2     yes
             Results of the data extraction                                                                                    197
</pre>

====================================================================== Einde pagina 198 =================================================================

<br><br>====================================================================== Pagina 199 ======================================================================

<pre>B3  observational cohort or maybe poss. experiment
B4  not described how long the phones were used for
B5  follow-up has not been used in this study
C1  all patients with newly diagnosed cerebral glioma during period 1 June 2005 and 31 May 2007, referred to the
    departments of neurosurgery and neurology of the university hospital of Ioannina as well as the other hospitals of the
    study area (6 districts, Ioannina, Arta, Preveza, Thesprotia, Corfu, Lefkada)
C2  nationwide
C3  no
D1  ratio 1:2
D2  ratio 1:3 of 1:4 for power
E1  burden on very ill patients and also on controls as they were neurologically assessed
E2  not mentioned
F1  the study area consisted of 6 districts: Ioannina, Arta, Preveza, Thesprotia, Corfu, Lefkada, source were hospitals within
    this area
F2  nationwide cancer registry
G1  first 41 of the 56 cases participated and 82 controls, nothing described about response
G2  not known what response rate is, possibly all included given small numbers? In that case good response
G3  41 cases and 82 controls (no drop-outs described but there very well might be non-response)
G4  nothing described
G5  unknown what follow-up is
G6  unknown how long people used their mobile phone
H1  unclear is this is complete, possibly some elderly rural areas missed due to wrong diagnosis stroke, however free access
    to all patients in hospital, so probably no or little selection bias
H2  probably so negligibly small
I1  unknown what response rate was, 100%?
I2  probably not or not much, but not described what response is
J1  yes, cases are likely to recall better and the controls are likely to have been the interested or SEC better off and thus more
    exposed+ yes as the mobile phone use had to be recalled as well as alcohol and tobacco which are always tricky
J2  possible overestimation use by cases, so overestimation effect??
K1  yes, because interviewer can influence conversation
K2  possible overestimation effect if interviewer convinced of the existence of an effect
L1  yes, age, sex, SES etc
L2  no, SES not and in analysis only alcohol consumption, smoking, use mobile phone and history severe cranial trauma
    included, matched op age and sex and district (unclear how this was included in the analysis)
M1  yes
M2  mobile phone use per minute years, very recall sensitive, so quickly wrong number of minute years
N1  interview
N2  questionnaire to reduce observation bias or use telecom companies to get exact phone habits
N3  yes, questions about time of start, minute-years and hands-free use
N4  not described
N5  could be
O   possibly yes due to recall
P   not to be expected
Q1  in analysis only alcohol consumption, smoking, use mobile phone and history severe cranial trauma included, matched
    op age and sex and district, but unclear how matching variables were included in the analysis
Q2  stratify for age, sex and district (probably to small numbers to do this?)
198        Mobile phones and cancer
</pre>

====================================================================== Einde pagina 199 =================================================================

<br><br>====================================================================== Pagina 200 ======================================================================

<pre>R1       logistic regression analysis
R2       conditional logistic regression analysis (due to matched data)
S1       95% CI and p-value
S2       ok
T        yes (not many absolute numbers given to check, cannot check in the table how many cases and controls per group)
U1       there is no significant association between glioma and mobile phone use
U2       see Annex H
U3       unclear, not described how long ago the use of the mobile phone started
U4       no that is to say not per minute -year, otherwise nothing investigated of a dose-response
V        unclear how exactly included in analysis so really not much to say
W        overestimation use due to cases, overestimation effect
X1       yes
X2       references particularly focussed on incidence rate, interphone references missing but some Hardell referred to and some
         others
Y        if 6 districts are representative for Greece, all Greece, otherwise only population 6 districts
Z        bit overstated given the very small numbers
Table F22 Extractions from Stang A, Schmidt-Pokrzywniak A, Lash TL, e.a. Mobile phone use and risk of uveal melanoma:
results of the risk factors for uveal melanoma case-control study. J Natl Cancer Inst, 2009; 101(2): 120-123.104
A1       recently reported increased risk of uveal melanoma now conducted with more valid exposure measurements more power
A2       mobile phone use increases the risk of uveal melanoma
A3       does mobile phone use increases the risk of uveal melanoma?
B1       hospital-based case-control study
B2       yes
B3       observational cohort or poss. experiment
B4       not described how long the phones were used for, highest usage category > 10 years
B5       no, only very small group >5 years use and only a few cases and controls in >10 years group
C1       subjects first diagnosed with uveal melanoma, aged 20-74, lived in Germany, proficient in German language between
         Sept. 25 2002 and September 24 2004 at University of Duisburg-Essen's referral centre for eye cancers
C2       bigger region, or nationwide
C3       not in this paper but in one referred: if achieving 380 cases & 760 controls an OR of 1.5 would be detectable
D1       3 control groups: population controls: 455 cases, 827 controls, ratio 1: 1,8
         ophthalmologist: 133 cases, 180 controls, ratio 1:1,4
         sibling controls: 187 cases, 187 controls, ratio 1:1
D2       all groups 1:2 for more power (here reasonable numbers of cases and controls in total, but subgroups to few people)
E1       burden for patients due to interview, fear for mobile phones
E2       not mentioned
F1       region of Duisburg/ Essen, Germany, University of Duisburg-Essen's referral centre for eye cancers; all newly diagnosed
         cases of uveal melanoma between 09/02 and 09/04 in the main tertiary clinic in one place supposedly missing 10 from
         another clinic in the state, controls form population census
F2       nationwide
G1       cases 94%, population controls 57%, sibling controls 57%, ophthalmologists controls 52%
G2       cases yes, controls: not at all
G3       455 cases, 827 population controls, 180 ophthalmologist controls (133 cases), 187 sibling controls (187 cases)
G4       n.a.
               Results of the data extraction                                                                                199
</pre>

====================================================================== Einde pagina 200 =================================================================

<br><br>====================================================================== Pagina 201 ======================================================================

<pre>G5  no, too few cases and controls > 10 years follow-up (also > 5 years relatively few people)
G6  highest category > 10 years
H1  yes, unclear if referral centre gets all cases, or e.g. only the worst cases
H2  can go either way, depends if the most ill particularly use or particularly do not use a phone
I1  yes for controls only a bit over 50% response, so substantial bias possible
I2  possible underestimation risk because for controls particularly users mobile phones participated
J1  yes recall problems
J2  possible overestimation use by cases, so overestimation effect??
K1  yes, in additional phone interviews possibly influenced by interviewer
K2  overestimation effect if interviewer "intend on proving" effect
L1  yes, age, sex, region of residence, SES etc
L2  yes, SES however unclear is this was measured
M1  yes
M2  recall bias, exposure definition, type, duration, frequency use phone
N1  questionnaire + additional phone interviews
N2  only questionnaire use to minimise observation bias, unclear what the aim was of the additional interviews (more details
    about exposure?)
N3  person-dose
N4  regular use (interphone definition) 36% of 827 population controls and 30% of 455 cases;
    30% of 180 opth. controls and 31% of 133 cases;
    35% of 187 sibling controls and 37% of 187 cases
N5  yes
O   possible, due to recall bias
P   not to be expected
Q1  log regression accounting for matching variables
Q2  multivariate analyse with also correction for e.g. SES
R1  conditional logistic regression
R2  ok
S1  95% CI
S2  ok
T   cannot be checked
U1  risk of uveal melanoma was not associated with regular mobile phone use, and no trend was observed for cumulative
    measures of exposure
U2  see Annex H
U3  no, much to small numbers > 10 years and also even > 5 years to small numbers to infer anything.
U4  no
V   can go either way, depends which bias has most influence
W   overestimation of use by cases, overestimation effect
X1  ok if a bit '1sided'
X2  no
Y   to population of Duisburg Essen
Z   reasonably as they recognise the time period problem
200       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 201 =================================================================

<br><br>====================================================================== Pagina 202 ======================================================================

<pre>Table F23 Extractions from Spinelli V, Chinot O, Cabaniols C, e.a. Occupational and environmental risk factors for brain
cancer: a pilot case-control study in France. Presse Med, 2010; 39(2): e35-e44.105
A1      what are deteminants of brain tumours
A2      in people with a brain tumour, what is the reported use of mobile phones (and other)
A3      ok A2 (note this is a pilot study)
B1      case-control
B2      ok
B3      cohort but would have to be very large
B4      n.a.
B5      n.a.
C1      all patients in 2 hospitals and very strict criteria
C2      larger but then this was a pilot
C3      not presented
D1      1:1
D2      1:4 considered better
E1      burden for respondents
E2      not mentioned
F1      patients in 2 hospitals, controls also hospitalised
F2      populations controls
G1      71% for cases and 90% for controls (?? unclear)
G2      would be ok
G3      122 cases and 122 controls
G4      n.a.
G5      n.a.
G6      n.a.
H1      yes
H2      could go either way
I1      yes
I2      could go either way
J1      yes
J2      could go either way
K1      yes
K2      could go either way
L1      some left
L2      could go either way
M1      yes
M2      could go either way
N1      interview
N2      yes, more detail, currently does have cumulative hrd
N3      hardly as too limited but if anything it is person
N4      cannot say
N5      potentially
O       as always
P       as always
              Results of the data extraction                                                                             201
</pre>

====================================================================== Einde pagina 202 =================================================================

<br><br>====================================================================== Pagina 203 ======================================================================

<pre>Q1     regression
Q2     ok
R1     regression coeff / OR
R2     ok
S1     95% CI
S2     ok
T      no seem to add up
U1     no effect seen but small numbers
U2     see Annex H
U3     cannot be assessed
U4     cannot be assessed
V      as always
W      as always
X1     ok
X2     ok
Y      very limited see next answer
Z      overstated as this is only a pilot so numbers are way too limited
Table F24 Extractions from Duan Y, Zhang HZ, and Bu RF. Correlation between cellular phone use and epithelial parotid gland
malignancies. Int J Oral Maxillofac Surg, 2011; 40(9): 966-972.106
A1     deteminants of parotid gland tumours including mobile phone use
A2     is there an association between having a parotid gland tumour and having a history of mobile phone use
A3     A2
B1     case-control
B2     ok
B3     cohort
B4     n.a.
B5     n.a.
C1     all cases in 1 hospital
C2     wider ranging?
C3     not presented
D1     1: 15 and 1: 30
D2     1:4 is considered optimal
E1     burden for the patients
E2     not mentioned
F1     all cases in 1 hospital (as all confirmed by 1 surgeon); controls in hospital too
F2     wider?
G1     62% for cases and 78% for controls
G2     yes but bit low for cases
G3     136 cases and 2051 controls (as only the living were on the study)
G4     n.a.
G5     n.a.
G6     n.a.
202          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 203 =================================================================

<br><br>====================================================================== Pagina 204 ======================================================================

<pre>H1      response rates are quite similar so possibly not too much?
H2      could go either way
I1      as responders are ill: yes
I2      could go either way
J1      as always
J2      could go either way
K1      as there were interviews yes
K2      could go either way
L1      age, sex, SES
L2      yes but SES poorly as always
M1      as always
M2      could go eitherway
N1      interview (face-2-face or telephone)
N2      more detail and verification
N3      person-dose
N4      57%
N5      potentially yes
O       as always
P       as always
Q1      regression
Q2      ok
R1      regression coeff / or
R2      ok
S1      95% Ci
S2      ok
T       seems ok, numbers add up
U1      the results suggest a possible dose–response relationship of cellular phone use with epithelial parotid gland malignancy
U2      see Annex H
U3      not obvious but maybe too limited range
U4      more consistently an association in highest exposure categories only
V       always
W       always
X1      ok
X2      ok
Y       to similar countries
Z       ok
Table F25 Extractions from Baldi I, Coureau G, Jaffre A, e.a. Occupational and residential exposure to electromagnetic fields
and risk of brain tumors in adults: a case-control study in Gironde, France. Int J Cancer, 2011; 129(6): 1477-1484.107
A1      widely assess possible determinants for brain tumours in adults
A2      do people with brain tumours have a different history of e.g.. mobile phone use
A3      A2 plus loads of other hypotheses
B1      case-control
B2      ok
              Results of the data extraction                                                                                203
</pre>

====================================================================== Einde pagina 204 =================================================================

<br><br>====================================================================== Pagina 205 ======================================================================

<pre>B3  cohort but would have to be extremely alrge
B4  n.a.
B5  n.a.
C1  all cases between 01/05/99 and 30/04/01 in one region
C2  larger?
C3  not presented
D1  1:2 matched
D2  ok but 1:4 would be better as numbers are not that large
E1  burden for patients to answer questions
E2  not mentioned but assumed
F1  incident cases Gironde, France, all cases between 01/05/99 and 30/04/01 in one region
F2  larger
G1  70% for cases and 69% for controls
G2  ok
G3  221 cases and 442 controls
G4  n.a.
G5  n.a.
G6  n.a. and duration of mp exposure is not given
H1  response rates are quite equal but there could be difference in non-response reasons and thus selection bias
H2  could go either way
I1  yes, they were ill and would have analysed for reasons
I2  could go either way
J1  yes
J2  could go either way
K1  yes, data collected by interview
K2  could go either way
L1  age, sex, SES
L2  yes but SES as that is always inadequately measured
M1  interview but limited on the phone information
M2  more detail
N1  person but limited on dose
N2  more detail at least
N3  not really
N4  112 / 441 = 25%
N5  barely probably not
O   yes as measurements are crude
P   yes as measurments are crude
Q1  regression
Q2  ok
R1  regression coeff / OR
R2  ok
S1  95% CI
S2  ok
T   seems ok
204       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 205 =================================================================

<br><br>====================================================================== Pagina 206 ======================================================================

<pre>U1     no stat sig association seen
U2     see Annex H
U3     not measured
U4     no but not measured
V      underestimation
W      underestimation
X1     ok
X2     ok
Y      limited
Z      ok
Table F26 Extractions from Aydin D, Feychting M, Schüz J, e.a. Mobile phone use and brain tumors in children and
adolescents: a multicenter case-control study. J Natl Cancer Inst, 2011; 103(16): 1264-1276.108
A1     several exposures such as mobile phones could be associated with brain tumours in children
A2     is there an association between having a brain tumour (as a child/adolescent) and having been exposed to mobile phones
A3     A2 was kind of tested
B1     case-control
B2     yes sort of with the inherent problems
B3     cohort but would have to be extraordinarily big
B4     n.a. but max exposure is 5 yrs
B5     n.a.
C1     total no of cases in region
C2     larger, longer
C3     not presented in this paper
D1     1:2
D2     1:4 is considered optimal
E1     burden for patients (and parents)
E2     not mentioned but assumed yes
F1     patients from Denmark, Sweden, Norway and Switzerland
F2     more countries?
G1     83% (68%-76%) for cases and 71% for controls (range for controls not given)
G2     ok good
G3     352 cases and 646 controls
G4     n.a. but longest exposure was 5 yrs
G5     n.a.
G6     n.a.
H1     yes as response rates are inequal
H2     could go either way
I1     yes as respondents were aware of hypotheses
I2     could go either way
J1     yes as cases will differ from controls in interest in the study questions
J2     could go either way
K1     yes as information gathered through interview so no blinding
K2     could go either way
             Results of the data extraction                                                                              205
</pre>

====================================================================== Einde pagina 206 =================================================================

<br><br>====================================================================== Pagina 207 ======================================================================

<pre>L1     age, sex and SES
L2     yes
M1     yes somewhat
M2     could go either way
N1     interview
N2     checking in bills (some of that was done but not always possible)
N3     person-dose
N4     317 / 636 = 50%
N5     theoretically yes
O      yes somewhat
P      yes somewhat
Q1     regression
Q2     ok
R1     regression coefficient /OR
R2     ok
S1     95% CI
S2     ok
T      yes numbers add up
U1     there was no consistent exposure–response relationship either in terms of the amount of mobile phone use or by the
       location of the tumor. In a small subset of study participants with operator recorded data (n = 163), however, time since
       the start of a mobile phone subscription was statistically significantly related to brain tumor risk.
U2     see Annex H
U3     unclear, too short
U4     unclear
V      underestimate
W      could go either way but ususally underestimate
X1     ok
X2     ok
Y      similar countries
Z      ok
             Case-case studies
Table F27 Extractions from Ali Kahn A, O'Brien DF, Kelly P, e.a. The anatomical distribution of cerebral gliomas in mobile
phone users. Ir Med J, 2003; 96(8): 240-242.109
A1     patients of the Beaumont neurosurgical unit have expressed concern regarding the possible role of mobile phones,
       concerns fuelled by various media reports on the subject
A2     were a cellular phone to cause a glioma, then it would do so on the dominant hand site
A3     correlation between handedness of patient and side of tumour and correlation between use, non use of mobile phone and
       location of tumour.
B1     unclear description, case-series of patients with supratentorial glioma
B2     no clear design
B3     cohort
B4     not described how long the mobile phones were used only category of how many minutes per day
B5     so unclear
206          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 207 =================================================================

<br><br>====================================================================== Pagina 208 ======================================================================

<pre>C1 study carried out between October 2000 and September 2001 of adult patients, histological diagnosis of supratentorial
   glioma at Beaumont neurological unit (diagnosed or interviewed between 2000-2001? unclear)
C2 more hospitals, larger region?
C3 no
D1 no controls, but also no comparisons with the general population e.g. pure handedness has been compared in this study
   with the location of the tumour
D2 n.a.
E1 burden for patients due to interview even though this seems to have been quite a short interview so probably not to bad
E2 not described in the text so possibly not
F1 all cases of glioma in 1 hospital 10/00 to 09/01; Beaumont neurosurgical unit (Ireland)
F2 all of Ireland??, maybe this is the only centre, unclear as not described
G1 response 100%, via case themselves otherwise via close family;
    73/92= 79%, due to exclusion of centrally located tumours or tumours of which the lateralisation was hard to establish
G2 80% is a bit low, however still acceptable to infer, reasons non-response would be interesting to trace
G3 73 cases
G4 nothing described
G5 therefore nothing to say
G6 unknown how long people used their mobile phone
H1 yes if people were excluded before the 92 that were left over, e.g. deceased patients who are not among the 92, and also
   because the centrally located tumours and those that were hard to localise were excluded, particularly that last group
   could have been influenced by phone use??
H2 unknown, possible underestimation effect, on the other hand possibly at the most some lack of power because some
   people were excluded who did not have a tumour localised in the part of the head that was exposed, while possibly this
   tumour had been located in the exposed part
I1 no, everyone participated and for those that did not give permission close family members did
I2 so no
J1 yes recall problems
J2 overestimation phone use, however that information is not used in the analysis and the location of the tumour and right-
   left handedness is not dependent on memory
K1 yes, because interviewer can influence the conversation
K2 possible overestimation effect if interviewer convinced of the existence of an effect
L1 relation location tumour and right or left handedness does not get influenced by confounders?,
L2 age , sex, clinical features have been measured, unclear if used in analysis, cannot trace id corrected for potential
   confounders
M1 for location tumour, theoretically yes, in practice this should be ok, and right or left handedness should not easily be
   misclassified either
M2 exposure mobile phones but that is not used in the analysis
N1 questionnaire, however also described that patients were visited, unclear if this was only to hand in the questionnaire or
   if the questionnaire was filled in using an interview
N2 right and left handedness and localisation and lateralisation tumour ok;
   exposure mobile phones via provider
N3 person-dose
N4 n.a.
N5 yes, but nothing done with the exposure as measured
O  possible recall, however concerns relatively short period of phone use and for localisation and lateralisation tumour and
   right /left handedness no misclassification to be expected. Cases have been histologically diagnosed, so also no
   misclassification to be expected
         Results of the data extraction                                                                                   207
</pre>

====================================================================== Einde pagina 208 =================================================================

<br><br>====================================================================== Pagina 209 ======================================================================

<pre>P      not to be expected
Q1     no correction for confounding done
Q2     ??, this is mostly an explorative article
R1     Fisher's exact test to test homogeneity of Odds ratios for case control comparing left and right sides cerebral gliomas
R2     logistic regression?
S1     p-value
S2     ok
T      yes, but few numbers mentioned and not clear how they got to the original 92 cases
U1     no statistical significance for glioma location based on the handedness of the patient in the mobile phone users group and
       location of the tumour in both user and non-user group
U2     see Annex H
U3     n.a.
U4     n.a.
V      can go either way, limited info on selection and such so little to say
W      misclassification will be limited. Cases have been histologically verified and use mobile yes/no and handedness should
       be answerable and correctly assigned
X1     yes
X2     very limited and short list
Y      Irish glioma patients
Z      bit overstated given the very small numbers
Table F28 Extractions from Salahaldin AH and Bener A. Long-term and frequent cellular phone use and risk of acoustic
neuroma. Int Tinnitus J, 2006; 12(2): 145-148.110
A1     interest in descriptive epi of acoustic neuromas, no clear exposure hypothesis but some mention of mobile phones
A2     not obvious
A3     not obvious
B1     case-series
B2     no control structure at all
B3     case-control over more years
B4     unclear
B5     n.a.
C1     unclear
C2     more years, controls etc: anything really
C3     not presented
D1     n.a.
D2     n.a.
E1     too small so never a real result so unethical to conduct in the first place
E2     not mentioned
F1     all newly diagnosed patients over an unspecified period in one (main?) hospital, no controls
F2     all cases in the country and population controls in 1 to 4 ratio over more years
G1     seemingly 100%
G2     n.a.
G3     13 cases
G4     n.a.
G5     n.a.
208          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 209 =================================================================

<br><br>====================================================================== Pagina 210 ======================================================================

<pre>G6      n.a.
H1      yes in the initial presentation and diagnosis of the patients
H2      n.a.
I1      unclear
I2      n.a.
J1      yes in answers to the mp questions but as they are not really used
J2      unclear
K1      yes as it is unclear if the interviewers were blinded for the location of the tumour
K2      could go either way
L1      age, sex, SEC
L2      nothing presented
M1      yes
M2      mp use
N1      questionnaire
N2      many ways
N3      personal
N4      n.a.
N5      not really
O       yes
P       yes
Q1      unclear, seemingly none
Q2      stratification, regression etc: anything really
R1      unclear
R2      anything really
S1      unclear
S2      anything really
T       there are always 13 patients in the tables
U1      incidence higher than expected no mention of mps
U2      see Annex H
U3      not addressed
U4      n.a.
V       unclear
W       n.a.
X1      bit sparse
X2      see previous
Y       limited
Z       bit overstated given the very small numbers
Table F29 Extractions from Sato Y, Akiba S, Kubo O, e.a. A case-case study of mobile phone use and acoustic neuroma risk in
Japan. Bioelectromagnetics, 2010.112
A1      literature on acoustic neuroma and mobile phones
A2      if there is an association there should be lateralisation
A3      is there an association between laterality of acoustic neuroma and reported mobile phone use
              Results of the data extraction                                                                          209
</pre>

====================================================================== Einde pagina 210 =================================================================

<br><br>====================================================================== Pagina 211 ======================================================================

<pre>B1  case-case or case-series
B2  yes as has non-differential recall bias
B3  is ok
B4  n.a.
B5  n.a.
C1  patients in a number of hospitals
C2  is ok as quite large
C3  not presented
D1  n.a.
D2  n.a.
E1  should be ok but burden to patients possible (likely even)
E2  not mentioned
F1  patients in named hospitals
F2  more hospitals?
G1  51%
G2  bit poor but given design not issue
G3  816 cases
G4  n.a.
G5  n.a
G6  n.a.
H1  no
H2  n.a.
I1  no, not differential
I2  n.a.
J1  no, not differential
J2  n.a.
K1  no, not differential
K2  n.a.
L1  age, sex, SES
L2  yes
M1  no, not differential
M2  n.a.
N1  questionnaire
N2  checking bills etc
N3  person-dose
N4  n.a.
N5  yes
O   yes
P   yes
Q1  regression
Q2  is ok
R1  regression coefficient
R2  is ok
S1  95% CI
S2  is ok
210       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 211 =================================================================

<br><br>====================================================================== Pagina 212 ======================================================================

<pre>T      ok
U1     so effect of lateralisation seen?
U2     increased risk of acoustic neuroma was observed in cases who reported having used mobile phones on the affected ear
       for >20 min/day on average. Risk ratio was 2.74 (95% CI, 1.18–7.85) for use until 1 year before diagnosis and 3.08
       (95% CI, 1.47–7.41) for use until 5 years before diagnosis.
U3     cannot be seen
U4     slightly
V      slightly
W      slightly
X1     ok
X2     no
Y      similar countries
Z      ok
             Ecological studies
Table F30 Extractions from Cook A, Woodward A, Pearce N, e.a. Cellular telephone use and time trends for brain, head and
neck tumours. N Z Med J, 2003; 116(1175): U457.115
A1     controversy about mp and tumours and now increasing use of mp
A2     if mp causes tumours we might start seeing it in cancer incidence
A3     what is the pattern in cancer incidence and what is the pattern in mp use
B1     ecological
B2     ok bar for inherent limitations
B3     ok but longer duration
B4     assumption was that as of 1987 mps started to be used so (theoretically) 11 yrs
B5     real use started up since 1995 so real fu was ca 3 yrs
C1     nationwide
C2     ok
C3     not presented
D1     n.a.
D2     n.a.
E1     identification of vulnerable subgroups but relatively minor issue, privacy as no consent possible
E2     not mentioned
F1     cancer registry (nationwide) and national data on mp subscriptions
F2     ok
G1     n.a.
G2     n.a.
G3     n.a.
G4     n.a.
G5     n.a.
G6     12 yrs: 1986-1998
H1     no
H2     n.a.
I1     n.a.
I2     n.a.
             Results of the data extraction                                                                              211
</pre>

====================================================================== Einde pagina 212 =================================================================

<br><br>====================================================================== Pagina 213 ======================================================================

<pre>J1      no
J2      n.a.
K1      no
K2      n.a.
L1      age, sex, SEC
L2      age and sex were
M1      no
M2      n.a.
N1      national data on mp subscriptions
N2      ok
N3      population
N4      9% at end of observation period
N5      not really
O       no
P       no
Q1      standardisation (unclear if direct or indirect)
Q2      ok
R1      trend & regression analysis
R2      ok
S1      not presented
S2      95%CI
T       probably ok given national registry data
U1      no evidence of an increase in brain tumour incidence in since with use of mps
U2      no increase in aa incidence (around10/100000 for bt's)
U3      as well as possible
U4      n.a.
V       has been taken care of
W       n.a.
X1      ok
X2      no
Y       somewhat
Z       yes
Table F31 Extractions from Röösli M, Michel G, Kuehni CE, e.a. Cellular telephone use and time trends in brain tumour
mortality in Switzerland from 1969 to 2002. Eur J Cancer Prev, 2007; 16(1): 77-82.121
A1      controversy about mp and tumours and now increasing use of mp
A2      if mp causes tumours we might start seeing it in mortality
A3      was the mortality for brain tumours higher in a period with (predicted use of ) mps than in a previous one without
B1      ecological
B2      ok bar for inherent limitations
B3      cancer incidence based study?
B4      assumption was that as of 1987 mps started to be used so (theoretically) 15 yrs
B5      real use started up since 1995 so real fu was ca 7 yrs
212           Mobile phones and cancer
</pre>

====================================================================== Einde pagina 213 =================================================================

<br><br>====================================================================== Pagina 214 ======================================================================

<pre>C1 nationwide
C2 ok
C3 not presented
D1 n.a.
D2 n.a.
E1 identification of vulnerable subgroups but relatively minor issue, privacy as no consent possible
E2 not mentioned
F1 national mortality data and national mp stats plus 2 surveys on mp use
F2 cancer registry data
G1 n.a.
G2 n.a.
G3 n.a.
G4 n.a.
G5 n.a.
G6 33 years: 1969-2002
H1 no
H2 n.a.
I1 n.a.
I2 n.a.
J1 no
J2 n.a.
K1 no
K2 n.a.
L1 age, sex, SEC
L2 age and sex were
M1 no
M2 n.a.
N1 national data on mp subscriptions and supporting surveys
N2 commercial verification but it is ok
N3 population
N4 around 0% till ca 1988, 20% in ca 1998, ca 70% in 2002
N5 not really
O  no
P  no
Q1 standardisation (unclear if direct or indirect)
Q2 ok
R1 predicted mortality rates vs. observed mortality rates
R2 this is ok
S1 95%CI
S2 ok
T  probably ok given national registry data
U1 no evidence of an increase in brain tumour mortality in since with use of mps
U2 no increase in aa rates in relevant ages for bt's
U3 as well as possible
U4 n.a.
         Results of the data extraction                                                              213
</pre>

====================================================================== Einde pagina 214 =================================================================

<br><br>====================================================================== Pagina 215 ======================================================================

<pre>V      has been taken care of
W      n.a.
X1     ok
X2     no
Y      somewhat
Z      yes
Table F32 Extractions from Czerninski R, Zini A, and Sgan-Cohen HD. Risk of parotid malignant tumors in Israel (1970-2006).
Epidemiology, 2011; 22(1): 130-131.129
A1     possibly ass parotid tumours and mp use
A2     if an increase in mps then an increase in parotid land tumours
A3     do trends seem to go in the expected direction but no trends for mps use given
B1     ecological
B2     somewhat but always limited as no association at personal level possible
B3     unclear
B4     unclear
B5     unclear but possibly too short
C1     whole population of Israel
C2     more countries
C3     not presented and n.a.
D1     n.a.
D2     n.a.
E1     involuntary
E2     as anonymous not an issue
F1     cancer registry of Israel
F2     is OK but more countries would be better
G1     n.a.
G2     n.a.
G3     n.a.
G4     n.a.
G5     n.a.
G6     n.a.
H1     no as all data
H2     n.a.
I1     no as routine data
I2     n.a.
J1     no as routine data
J2     n.a.
K1     no
K2     n.a.
L1     age, sex and SEC
L2     age and sex yes
M1     not really
M2     n.a.
214          Mobile phones and cancer
</pre>

====================================================================== Einde pagina 215 =================================================================

<br><br>====================================================================== Pagina 216 ======================================================================

<pre>N1      vague data on mobile phone subscriptions/plans/ownership
N2      actual data
N3      no
N4      n.a.
N5      no not really
O       no
P       no
Q1      age standardised rates
Q2      is ok
R1      not done
R2      many options
S1      not done
S2      many options
T       can’t be judged but assumed yes
U1      marked increase in incidence of parotid gland tumours
U2      as no association measured not relevant
U3      can’t be judged
U4      can’t be shown
V       some level of confounding by SEC possible
W       n.a.
X1      ok
X2      no
Y       Israel and similar countries
Z       is ok
Table F33 Extractions from de Vocht F, Burstyn I, and Cherrie JW. Time trends (1998-2007) in brain cancer incidence rates in
relation to mobile phone use in England. Bioelectromagnetics, 2011; DOI 10.1002/bem.20648.130
A1      aAssess if there are trends in incidence of brain tumours in association with trends in mp use
A2      if there is an association then incidence of brain tumours should be increasing (soon)
A3      is there a trend
B1      ecological
B2      as no association at personal level measured it is always limited
B3      cohort
B4      n.a.
B5      n.a.
C1      all country
C2      n.a
C3      n.a.
D1      n.a.
D2      n.a.
E1      involuntary participation
E2      as all information is anonymous limited problems so not relevant
F1      cancer registry
F2      is ok
              Results of the data extraction                                                                            215
</pre>

====================================================================== Einde pagina 216 =================================================================

<br><br>====================================================================== Pagina 217 ======================================================================

<pre>G1  n.a.
G2  n.a.
G3  n.a.
G4  n.a.
G5  n.a.
G6  n.a.
H1  no
H2  n.a.
I1  no
I2  n.a.
J1  no
J2  n.a.
K1  no
K2  n.a.
L1  age, sex and EC
L2  age and sex were
M1  limited
M2  n.a.
N1  trends in mp ownership
N2  actual use data
N3  population
N4  n.a.
N5  no
O   no
P   no
Q1  age and sex standardisation
Q2  ok
R1  not done
R2  regression
S1  no
S2  95% CI
T   should be ok as routine data but cannot be checked
U1  no evidence of an increasing trend
U2  reasonably stable numbers
U3  can’t tell
U4  n.a.
V   SEC and detection bias?
W   n.a.
X1  yes
X2  ok
Y   similar countries
Z   is OK
216       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 217 =================================================================

<br><br>====================================================================== Pagina 218 ======================================================================

<pre>Table F34 Extractions from de Vocht F. Cell phones and parotid cancer trends in England. Epidemiology, 2011; 22(4): 608-
609.134
A1      given use of mps is there a trend in parotid gland tumours
A2      if there is an association, trends should be starting to go up (if the effect is reasonably immediate)
A3      is there a trend
B1      ecological
B2      is inherently limited
B3      cohort
B4      n.a.
B5      n.a.
C1      n.a.
C2      n.a.
C3      n.a.
D1      n.a.
D2      n.a.
E1      involuntary participation s routine data used
E2      as is anonymous not much of an issue and often signed off by ethics committees without much of a problem
F1      cancer registry data
F2      is ok
G1      n.a.
G2      n.a.
G3      n.a.
G4      n.a.
G5      n.a.
G6      n.a.
H1      n.a.
H2      n.a.
I1      n.a.
I2      n.a.
J1      n.a.
J2      n.a.
K1      n.a.
K2      n.a.
L1      age, sex and SEC
L2      age and sex yes
M1      n.a.
M2      n.a.
N1      trend in mp ownership
N2      actual use
N3      population
N4      n.a.
N5      no
O       no
P       no
              Results of the data extraction                                                                           217
</pre>

====================================================================== Einde pagina 218 =================================================================

<br><br>====================================================================== Pagina 219 ======================================================================

<pre>Q1      age and sex standardised
Q2      is OK
R1      not done
R2      regression
S1      no
S2      n.a.
T       presumed ok
U1      there is increase
U2      2-fold increase in incident cases
U3      no: increase started before widespread use of mp
U4      n.a.
V       SES? detection bias?
W       n.a.
X1      ok
X2      no
Y       similar countries
Z       ok
Table F35 Extractions from Deltour I, Auvinen A, Feychting M, e.a. Mobile phone use and incidence of glioma in the Nordic
countries 1979-2008: Consistency check. Epidemiology, 2012; 23(2): 301-307.135
A1      if mobile phone use causes brain tumors, the marked increase in prevalence of use over a 20-year period will eventually
        influence the time trends of the incidence rates of these tumors. (quote)
A2      compare trends in glioma vs. trends in use of mobile phone
A3      as vague association it might see something, duration of mps is possibly long enough
B1      ecological
B2      reasonably
B3      cohort as it would actually measure use rather than compare groups
B4      1979-2008: only in last 10 years have almost all had mobile phones, in Nordic countries several years earlier than
        elsewhere: still bit short?
B5      getting there
C1      all cases in Denmark, Finland, Norway and Sweden
C2      should have been big enough
C3      not stated
D1      n.a.
D2      n.a.
E1      effectively involuntary participation as routine data are used
E2      if anonymous as is here, generally not considered a problem if no individuals can be traced
F1      cancer registries in all 5 countries
F2      good source
G1      n.a.
G2      n.a.
G3      n.a.
G4      n.a.
G5      n.a.
G6      n.a.
218           Mobile phones and cancer
</pre>

====================================================================== Einde pagina 219 =================================================================

<br><br>====================================================================== Pagina 220 ======================================================================

<pre>H1     no unless not diagnosed but mortality is often also included if post-mortem diagnosis
H2     n.a.
I1     n.a.
I2     n.a.
J1     n.a.
J2     n.a
K1     n.a.
K2     n.a.
L1     age and sex and possibly SEC
L2     age and sex were and were corrected for
M1     no not really
M2     n.a.
N1     as in population wide no. of subscriptions/plans etc
N2     this is quite crude: actual use would be better
N3     no
N4     n.a.
N5     not really
O      n.a.
P      yes in the mobile phone use data
Q1     age and sex standardised rates
Q2     is ok
R1     log linear model
R2     is ok
S1     95% CI
S2     is ok
T      yes
U1     there is no upward turn in the trends
U2     results compatible with those of studies showing no effect but INTERPHONE seize effects could still be true
U3     weak as it is possibly still too early
U4     n.a.
V      SEC is not controlled for as those higher up get diagnosed more and would have had phones earlier but this is all weak
W      n.a.
X1     ok
X2     no
Y      similar countries such as northern Europe
Z      ok
Table F36 Extractions from Little MP, Rajaraman P, Curtis RE, e.a. Mobile phone use and glioma risk: comparison of
epidemiological study results with incidence trends in the United States. BMJ, 2012; 344: e1147.133
A1     looking a trends in brain tumour incidence
A2     is mp use is associated with brain tumours the trends should be starting to go up
A3     is there a trend
B1     ecological
B2     kind of yes
             Results of the data extraction                                                                               219
</pre>

====================================================================== Einde pagina 220 =================================================================

<br><br>====================================================================== Pagina 221 ======================================================================

<pre>B3  cohort
B4  n.a.
B5  n.a.
C1  n.a.
C2  n.a.
C3  n.a.
D1  n.a.
D2  n.a.
E1  involuntary as routine data use
E2  anonymous data so ethics committee mostly say yes
F1  cancer registry data
F2  is ok, longer period would not have helped
G1  n.a.
G2  n.a.
G3  n.a.
G4  n.a.
G5  n.a.
G6  n.a.
H1  n.a.
H2  n.a.
I1  n.a.
I2  n.a.
J1  n.a.
J2  n.a.
K1  n.a.
K2  n.a.
L1  n.a.
L2  n.a.
M1  n.a.
M2  n.a.
N1  n.a.
N2  n.a.
N3  n.a.
N4  n.a.
N5  n.a.
O   n.a.
P   n.a.
Q1  age standardised rates
Q2  is ok
R1  regression
R2  is ok
S1  95% CI
S2  is ok
T   should be ok as routine data
220       Mobile phones and cancer
</pre>

====================================================================== Einde pagina 221 =================================================================

<br><br>====================================================================== Pagina 222 ======================================================================

<pre>U1 raised risks of glioma with mobile phone use, as reported by one (Swedish) study forming the basis of the IARC’s re-
   evaluation of mobile phone exposure, are not consistent with observed incidence trends in US population data, although
   the US data could be consistent with the modest excess risks in the Interphone study.
U2 n.a.
U3 not obvious
U4 not possible
V  n.a.
W  n.a.
X1 ok
X2 no
Y  similar countries
Z  ok
         Results of the data extraction                                                                              221
</pre>

====================================================================== Einde pagina 222 =================================================================

<br><br>====================================================================== Pagina 223 ======================================================================

<pre>222 Mobile phones and cancer</pre>

====================================================================== Einde pagina 223 =================================================================

<br><br>====================================================================== Pagina 224 ======================================================================

<pre>Annex          G
               Results of the evaluation of quality of
               the studies
               The results of the scores per question are presented in Tables G1, G2 and G3.
               These are the combined scores for the two evaluators (IK and MC). These final
               scores were the result of independent scoring, comparison and mediation (EvR).
Table G1 Results of the quality scores for the cohort studies.
                         Question
                         1    2     3      4     5      6     7      8 9 10 11 12  13  14   15 16 17
Dreyer et al. (1999)47 c      c     d      c     a      c     c      a b d  a  a   c   b    b  b  b
Schüz et al. (2011)51    c    c     d      c     e      c     c      a b b  a  a   c   b    b  b  a
Frei et al. (2011)52     c    c     d      c     e      c     c      a b b  a  a   c   b    b  b  a
Table G2 Results of the quality scores for the case-control studies.
                         Question
                         1    2     3      4     5      6     7      8 9 10 11 12  13  14   15 16 17
Hardell et al. (2004)69 b     b     a      a     e      a     b      c h e  c  b   a   a    b  b  c
Hardell et al. (2009)66 b     b     b      b     e      a     b      c h e  c  b   a   a    b  b  c
Hardell et al. (2011)68 b     b     b      b     e      a     b      c h e  c  b   a   a    b  b  c
Söderqvist et al.        b    b     b      b     e      a     b      c h e  b  a   a   b    b  b  c
(2012)72
Lönn et al. (2006)82     b    b     b      b     a      a     b      c f e  c  b   b   a    b  b  b
Sadetzki et al. (2008)85 b    a     b      a     e      a     b      c f e  c  b   b   a    b  b  b
Takebayashi et al.       a    a     b      b     e      a     b      c h e  c  b   b   a    b  b  c
(2008)88
               Results of the evaluation of quality of the studies                                 223
</pre>

====================================================================== Einde pagina 224 =================================================================

<br><br>====================================================================== Pagina 225 ======================================================================

<pre>Schoemaker et al.         b    b    a      b     e     a     b    c f e  c  b  b  a  b  b  b
(2009)89
INTERPHONE                b    b    a      a     e     a     b    c f e  c  b  b  a  b  b  b
(2010)93
INTERPHONE                b    b    b      a     e     a     b    c f e  c  b  b  a  b  b  b
(2011)94
Muscat et al. (2000)96 a       a    a      a     e     a     b    c f e  c  a  a  a  b  b  a
De Roos et al. (2001)97 a      a    a      a     a     a     b    c b a  a  a  a  a  a  a  c
Stang et al. (2001)98     b    b    b      a     e     a     b    c b b  a  a  a  a  a  a  c
Inskip et al. (2001)99 a       a    c      b     e     a     b    c c d  c  a  a  a  b  b  c
Auvinen et al. (2002)100 b     b    c      c     e     c     b    a f b  a  b  c  b  b  b  b
Muscat et al. (2002)101 a      a    a      a     e     a     b    c b c  c  a  a  a  b  b  a
Warren et al. (2003)102 a      a    a      a     a     a     b    c e d  c  a  a  a  a  a  c
Gousias et al. (2007)103 a     a    a      a     e     a     b    c b c  a  a  a  a  a  a  c
Stang et al. (2009)104 b       b    c      a     c     a     b    c f d  c  b  b  a  b  b  b
Spinelli et al. (2010)105 a    a    a      b     e     a     b    c b c  a  a  a  a  a  a  c
Duan et al. (2011)106 a        a    a      a     e     a     b    c h e  c  b  a  a  b  b  c
Baldi et al. (2011)107 b       b    a      b     e     a     b    c b a  a  a  a  b  b  b  c
Aydin et al. (2011)108 b       b    b      b     e     a     b    c e d  c  b  b  a  b  b  c
Table G3 Results of the quality scores for the case-case studies.
                          Question
                          1    2    3      4     5     6     7    8 9 10 11 12 13 14 15 16 17
Ali Kahn et al.           c    c    d      c     e     a     b    c b a  b  a  a  a  a  a  c
(2003)109
Salahaldin & Bener        c    c    d      c     e     a     a    a b a  a  a  a  a  a  a  c
(2006)110
Sato et al. (2010)112     c    c    d      c     e     b     b    b c e  c  b  a  a  b  b  c
224            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 225 =================================================================

<br><br>====================================================================== Pagina 226 ======================================================================

<pre>Annex H
      Results from the selected publications
      This Annex presents all the detailed results in tables, organized by tumour type.
      Statistically significant increased risks are in boldface type and highlighted in
      yellow, statistically significantly decreased risks are highlighted in light blue
      only.
      The publications of Hardell et al.64-66,68,69,72,165 and Stang et al.98 from which the
      data are obtained do not provide information on the numbers of cases and
      controls in the reference categories, nor can these be derived.
      Abbreviations used:
      Obs / Exp: observed and expected numbers of cases;
      SIR: standardized incidence ratio;
      SMR: standardized mortality ratio;
      CI: confidence interval;
      Ca / Co: numbers of cases and controls;
      OR: odds ratio.
      Results from the selected publications                                                 225
</pre>

====================================================================== Einde pagina 226 =================================================================

<br><br>====================================================================== Pagina 227 ======================================================================

<pre>              Brain tumours, not specified
Table H1 Brain tumours (not otherwise specified) and duration of use, results corrected for confounders.
Cohort                   Exposure             Person years        Obs / Exp             SMR                 95%CI
                         Time since 1st use
                         (yrs)
Dryer et al. (1999)47    ≤3                   88152               1 / --                1.4
(adults)
                         >3                   14447               1 / --                8.4
Case-control             Time since 1st use                       Ca / Co               OR                  95%CI
                         (yrs)
Muscat et al. (2000)96   0                                        403 / 306             1.0
(adults)                 1                                        21 / 30               0.7                 0.4 -1.3
                         2-3                                      28 / 24               1.1                 0.6 -2.0
                         ≥4                                       17 / 22               0.7                 0.4 -1.4
Aydin et al. (2011)108   0                                        158 / 317             1.0
(children)               ≤3.3                                     95 / 165              1.35                0.89-2.04
                         3.3-5.0                                  53 / 83               1.47                0.87-2.49
                         >5.0                                     46 / 81               1.26                0.70-2.28
Table H2 Brain tumours (not otherwise specified) and cumulative use, results corrected for confounders.
Case-control             Exposure                  Ca / Co                    OR                       95%CI
                         Cumulative call time (h)
Muscat et al. (2000)96   0                         403 / 306                  1.0
(adults)                 >0-≤8.7                   17 / 18                    1.0                      0.5-2.0
                         >8.7-≤60                  12 / 19                    0.6                      0.3-1.3
                         >60-≤480                  19 / 19                    0.9                      0.5-1.8
                         >480                      14 / 19                    0.7                      0.3-1.4
Aydin et al. (2011)108   0                         158 / 317                  1.0
(children)
                         ≤35                       94 / 162                   1.33                     0.89-2.01
                         36-144                    48 / 81                    1.44                     0.85-2.44
                         >144                      49 / 81                    1.55                     0.86-2.82
                         Cumulative number of calls
                         0                         158 / 317                  1.0
                         ≤936                      94 / 163                   1.34                     0.89-2.02
                         937-2638                  50 / 80                    1.47                     0.86-2.51
                         >2638                     47 / 79                    1.42                     0.79-2.53
226           Mobile phones and cancer
</pre>

====================================================================== Einde pagina 227 =================================================================

<br><br>====================================================================== Pagina 228 ======================================================================

<pre>Table H3 Brain tumours (not otherwise specified) in children and laterality, results corrected for confounders.
Aydin et al.         Ipsilateral                        Contralateral                          Central /
(2011)108                                                                                      unknown
                     Ca / Co      OR       95%CI        Ca / Co     OR            95%CI        Ca / Co    OR95%CI
Time since 1st use (yrs)
0                     146 / 267 1.0                     141 / 257 1.0                          147 / 257 1.0
≤3.3                  29 / 40     1.73     0.87-3.44 24 / 36        1.86          0.82-4.21 36 / 68       0.81         0.41-1.57
3.3-5.0               15 / 25     1.53     0.62-3.76 16 / 16        3.27          1.10-9.68 19 / 31       0.82         0.34-1.94
>5.0                  18 / 18     2.75     0.93-8.06 9 / 11         2.39          0.67-8.57 13 / 36       0.36         0.13-1.02
Cumulative call time (h)
0                     146 / 267 1.0                     141 / 257 1.0                          147 / 257 1.0
≤35                   28 / 48     1.46     0.74-2.91 19 / 35        1.65          0.73-3.74 40 / 59       0.97         0.50-1.85
36-155                17 / 17     2.66     1.05-6.71 13 / 17        4.14          1.25-13.7 15 / 37       0.43         0.18-1.03
>155                  17 / 18     2.64     0.92-7.59 16 / 9         6.19          1.57-24.4 12 / 36       0.24         0.08-0.73
Cumulative number of calls
0                     146 / 267 1.0                     141 / 257 1.0                          147 / 257 1.0
≤936                  30 / 46     1.59     0.81-3.12 22 / 38        1.74          0.78-3.90 37 / 57       0.98         0.51-1.92
937-2638              13 / 19     2.06     0.72-5.93 14 / 12        5.37          1.54-18.7 17 / 38       0.54         0.24-1.23
>2638                 19 / 18     2.91     1.09-7.76 12 / 11        4.82          1.21-19.2 13 / 37       0.31         0.11-0.87
               Glioma
Table H4 Glioma and duration of use, results corrected for confounders.
Cohort                       Gender            Exposure            Cases                  IRR                 95%CI
                                               Time since
                                               subscription (yrs)
Frei et al. (2011)52         Males             0                   4397                   1.00
                                               1-4                 85                     1.20                0.96-1.50
                                               5-9                 122                    1.05                0.87-1.26
                                               ≥10                 117                    1.04                0.85-1.26
                                               10-12               80                     1.06                0.85-1.34
                                               ≥13                 37                     0.98                0.70-1.36
                             Females           0                   5486                   1.00
                                               1-4                 8                      0.87                0.43-1.75
                                               5-9                 14                     1.02                0.60-1.72
                                               ≥10                 10                     1.04                0.56-1.95
Case-control                 Type of phone     Exposure            Ca / Co                OR                  95%CI
                                               Time since 1st use
                                               (yrs)
Hardell et al. (2011)68      Mobile            >1-5                250 / 571              1.1                 0.9 -1.4
All glioma                                     >5-10               156 / 286              1.3                 0.99-1.6
                                               >10                 123 / 106              2.5                 1.8 -3.4
                             Cordless          >1-5                205 / 463              1.2                 0.9 -1.5
                                               >5-10               152 / 244              1.5                 1.2 -1.9
               Results from the selected publications                                                                       227
</pre>

====================================================================== Einde pagina 228 =================================================================

<br><br>====================================================================== Pagina 229 ======================================================================

<pre>                                              >10                  45 / 55            1.7                1.1 -2.6
Astrocytoma (all)          Mobile             >1-5                 197 / 571          1.2                0.9 -1.5
                                              >5-10                132 / 286          1.4                1.04-1.8
                                              >10                  110 / 106          2.7                1.9 -3.7
                           Cordless           >1-5                 157 / 463          1.2                0.9 -1.5
                                              >5-10                135 / 244          1.7                1.3 -2.2
                                              >10                  41 / 55            1.8                1.2 -2.9
Hardell et al. (2011)165   Mobile (cordless = ≥10                  56 / 74            1.79               1.19-2.70
30-59 y old                unexposed)
INTERPHONE study           Mobile (excl.      0                    1042 / 1078        1.00
group (2010)93             cordless)          1-1.9                156 / 247          0.62               0.46-0.81
                                              2-4                  644 / 725          0.84               0.70-1.00
                                              5-9                  614 / 690          0.81               0.60-0.97
                                              ≥10                  252 / 232          0.98               0.76-1.26
Cardis et al. (2011)114    Mobile             0                    14 / 178           1.00
                           Case-case*         1-4                  12 / 133           1.37               0.59-3.19
                                              5-9                  7 / 147            0.72               0.27-1.90
                                              ≥10                  11 / 54            2.80               1.13-6.94
Larjavaara et al. (2011)   Mobile             0                    **                 1.00
113
                           Case-case          1.5-4                                   0.85               0.57-1.25
                                              5-9                                     0.71               0.43-1.18
                                              ≥10                                     0.85               0.39-1.86
                                              Duration of
                                              subscription (yrs)
Auvinen et al. (2002)100 Mobile analogue 0                         172 / 921          1.0
                                              <1                   4 / 13             1.6                0.5 -5.1
                                              1-2                  11 / 24            2.4                1.2 -5.1
                                              >2                   11 / 31            2.0                1.0 -4.1
                           Mobile digital     0                    188 / 938          1.0
                                              <1                   3 / 20             0.8                0.2 -2.6
                                              1-2                  7 / 25             1.4                0.6 -3.4
                                              >2                   0/6                0.0                -
                           Mobile all         0                    -                  1.0
                                              <1                   -                  1.2                0.5 -3.0
                                              1-2                  -                  1.6                0.8 -2.9
                                              >2                   -                  1.7                0.9 -3.5
Inskip et al. (2001)99     Mobile (excl.      0                    398 / 625          1.0
                           cordless)          <0.5                 24 / 56            0.6                0.3 -1.1
                                              0.5-<3               31 / 55            0.9                0.5 -1.6
                                              >3                   30 / 60            0.9                0.5 -1.5
                                              >5                   11 / 31            0.6                0.3-1.4
Muscat et al. (2000)96     Mobile             >1                   41 / 76            0.8                0.5-1.2
* Case-case study: cases with tumour within most exposed area vs. cases with tumour outside most exposed area
** Case-case study, no numbers provided
228            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 229 =================================================================

<br><br>====================================================================== Pagina 230 ======================================================================

<pre>Table H5 Glioma and cumulative use, results corrected for confounders.
                          Type of phone       Exposure           Ca / Co     OR   95%CI
                                              Cumulative call
                                              time (h)
Hardell et al. (2011)68   Mobile              1-1000             427 / 879   1.2  1.03-1.5
All glioma                                    1001–2000          44 / 51     1.8  1.2 -28
                                              >2000              58 / 33     3.2  2.0 -5.1
                          Cordless            1-1000             297 / 643   1.2  0.95 -1.4
                                              1001-2000          50 / 60     2.0  1.4 -3.1
                                              >2000              55 / 59     2.2  1.4 -3.2
Astrocytoma               Mobile              1-1000             346 / 879   1.3  1.1 -1.6
                                              1001-2000          42 / 51     2.2  1.4 -3.5
                                              >2000              51 / 33     3.4  2.1 -5.6
                          Cordless            1-1000             240 / 643   1.2  0.96-1.5
                                              1001-2000          45 / 60     2.3  1.5 -3.6
                                              >2000              48 / 59     2.4  1.5 -3.6
Hardell et al. (2011)165  Mobile (cordless = ≥1640               29 / 37     1.75 1.02-3.00
30-59 y old               unexposed)
INTERPHONE study          Mobile (excl.       0                  1042 / 1078 1.0
group (2010)93            cordless)           <5                 141 / 197   0.70 0.52-0.94
                                              5-12.9             145 / 198   0.71 0.53-0.94
                                              13-30.9            189 / 179   1.05 0.79-1.38
                                              31-60.9            144 / 196   0.74 0.55-0.98
                                              61-114.9           171 / 193   0.81 0.61-1.08
                                              115-199.9          160 / 194   0.73 0.54-0.98
                                              200-359.9          158 / 194   0.76 0.57-1.01
                                              360-734.9          189 / 205   0.82 0.62-1.08
                                              735-1639.9         159 / 184   0.71 0.53-0.96
                                              ≥1640              210 / 154   1.40 1.03-1.89
                                              ≥1640 (excl. >5 h/ 169 / 134   1.27 0.92-1.75
                                              d)
                                              Cumulative
                                              number of calls (x
                                              100)
INTERPHONE study          Mobile (excl.       0                  1042 / 1078 1.0
group (2010)93            cordless)           <1.5               147 / 182   0.74 0.55–0.99
                                              1.5-3.4            141/ 200    0.71 0.54–0.95
                                              3.5-7.4            161 / 201   0.76 0.58–1.00
                                              7.5-13.9           174 / 179   0.90 0.68–1.20
                                              14-25.4            180 / 206   0.78 0.59–1.02
                                              25.5-41.4          156/ 190    0.83 0.62–1.10
                                              41.5-67.9          163 / 194   0.71 0.53–0.94
                                              68-127.9           186 / 200   0.93 0.70–1.23
                                              128-269.9          193 / 180   0.96 0.72–1.28
                                              ≥270               165 / 162   0.96 0.71–1.31
               Results from the selected publications                                       229
</pre>

====================================================================== Einde pagina 230 =================================================================

<br><br>====================================================================== Pagina 231 ======================================================================

<pre>                                              Cumulative call
                                              time (h)
Cardis et al. (2011)114     Mobile            0                   196 / 617  1.00
                                              <13                 44 / 174   0.83 0.55-1.26
                                              13-60.9             68 / 223   0.93 0.65-1.32
                                              61-199.9            63 / 264   0.66 0.46-0.96
                                              200-734.9           90 / 237   1.07 0.76-1.50
                                              ≥735                90 / 205   1.25 0.88-1.77
                                              Total cumulative
                                              specific energy (J/
                                              kg)
                            Mobile, all users 0                   196 / 617  1.00
                                              <76.7               67 / 265   0.76 0.53-1.09
                                              76.7-284            68 / 227   0.94 0.66-1.35
                                              284.1-978.9         60 / 207   0.80 0.54-1.18
                                              979-3123.9          57 / 197   0.89 0.61-1.30
                                              ≥3124               103 / 207  1.35 0.96-1.90
                            Mobile, use ≥7 y  0                   421 / 1445 1.00
                            in past           <76.7               20 / 63    1.11 0.61-2.02
                                              76.7-284            23 / 53    1.53 0.85-2.78
                                              284.1-978.9         24 / 53    1.50 0.81-2.78
                                              979-3123.9          25 / 49    1.69 0.91-3.13
                                              ≥3124               38 / 57    1.91 1.05-3.47
                                              Cumulative call
                                              time (h)
                            Mobile            0                   14 / 178   1.00
                            Case-case*        <39                 6 / 65     1.19 0.40-3.51
                                              39-220              4 / 67     0.93 0.27-3.14
                                              220-520             5 / 68     1.38 0.42-4.53
                                              520-1147            10 / 66    2.55 0.94-6.91
                                              ≥1147               5 / 68     0.99 0.30-3.27
Larjavaara et al. (2011)113 Mobile            0                   **         1.00
                                              0.001-46                       0.82 0.51-1.31
                                              46-339                         0.97 0.60-1.56
                                              >339                           0.58 0.35-0.96
Inskip et al. (2001)99      Mobile (excl.     0                   398 / 625  1.0
                            cordless)         < 13                26 / 55    0.8  0.4 -1.4
                                              13-100              26 / 58    0.7  0.4 -1.3
                                              >100                32 / 54    0.9  0.5 -1.6
                                              >500                11 / 27    0.5  0.2-1.3
                                              Cumulative
                                              potential use
                                              (hour-years)
Spinelli et al. (2010)105   Mobile            0                   37 / 42    1.0
                                              ≤4                  8 / 11     0.86 0.30-2.44
                                              4-36                58 / 48    1.45 0.75-2.80
230            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 231 =================================================================

<br><br>====================================================================== Pagina 232 ======================================================================

<pre>                                                    ≥ 36                13 / 15                1.07                 0.41-2.82
                                                    Cumulative use
                                                    (minute-years)
Gousias et al. (2009)103       Mobile               Not provided        41 / 82                1.00                 0.99-1.01
* Case-case study: cases with tumour within most exposed area vs. cases with tumour outside most exposed area
** Case-case study, no numbers provided
Table H6 Glioma and laterality, results corrected for confounders.
                                                           Ipsilateral                            Contralateral
                             Exposure                      Ca / Co      OR       95%CI            Ca / Co       OR        95%CI
                             Time since first use (yrs)
Hardell et al. (2009)66 * Mobile > 1                       229 / 374    2.0      1.5 - 2.5        98 / 308      1.0       0.7 -1.4
             Astrocytoma Mobile >10                        50 / 45      3.3      2.0 - 5.4        26 / 29       2.8       1.5 -5.1
                             Cordless >1                   167 / 309    1.8      1.4 - 2.4        81 / 235      1.2       0.8 -1.6
                             Cordless >10                  19 / 15      5.0      2.3 -11          8 / 20        1.4       0.6 -3.5
Hardell et al. (2011)165 Mobile ≥10                        35 / 30      2.29     1.33-3.79        20 / 24       1.71      0.89-3.28
30-59 y old                  (cordless = unexposed)
                  Glioma
INTERPHONE                   0                             773 / 832    1.00                      721 / 718     1.00
(2010)93                     1-1.9                         69 / 91      0.77     0.49-1.20        24 / 58       0.34      0.20-0.71
                  Glioma 2-4                               261 / 300    0.80     0.62-1.04        145 / 178     0.81      0.57-1.14
                             5-9                           239 / 280    0.81     0.62-1.05        110 / 145     0.65      0.44-0.95
                             ≥10                           108 / 82     1.21     0.82-1.80        49 / 56       0.70      0.42-1.15
                             Cumulative call time (h)
Hardell et al. (2011)165 ≥1640                             20 / 18      2.18     1.09-4.35        8 / 11        1.48      0.57-3.87
30-59 y old
                  Glioma
INTERPHONE                   0                             773 / 838    1.00                      721 / 718     1.00
(2010)93                     <5                            64 / 76      0.83     0.53-1.31        23 / 50       0.43      0.22-0.84
                  Glioma 5-114.9                           253 / 321    0.75     0.58-0.97        135 / 170     0.74      0.53-1.03
                             115-359.9                     121 / 147    0.75     0.53-1.07        67 / 93       0.62      0.39-0.97
                             360-1639.9                    139 / 147    0.88     0.62-1.24        64 / 93       0.60      0.38-0.94
                             ≥1640                         100 / 62     1.96     1.22-3.16        39 / 31       1.25      0.64-2.42
Larjavaara et al.            **                            51 / 195     0.80     0.52-1.22        37 / 133      0.77      0.47-1.24
(2011)113
                             Cumulative number of
                             calls (x100)
INTERPHONE                   0                             773 / 838    1.00                      721 / 718     1.00
(2010)93                     <1.5                          61 / 71      0.66     0.41-1.07        26 / 44       0.61      0.32-1.17
                  Glioma 1.5-25.4                          263 / 318    0.80     0.62-1.04        138 / 179     0.69      0.49-0.96
                             25.5-67.9                     115 / 159    0.69     0.49-0.97        64 / 91       0.59      0.38-0.92
                             68-269.9                      164 / 145    1.09     0.78-1.52        72 / 86       0.81      0.51-1.28
                             ≥270                          74 / 60      1.51     0.91-2.51        28 / 37       0.61      0.32-1.18
* This publication groups ipsilateral and ipsi/contralateral, so the subjects that call at the side of the tumour and those who do
this and alternate it with the other side are grouped, the other publications do not group these.
** Case-case study; comparison only for ipsi- vs. contralateral use.
               Results from the selected publications                                                                             231
</pre>

====================================================================== Einde pagina 232 =================================================================

<br><br>====================================================================== Pagina 233 ======================================================================

<pre>               Meningioma
Table H7 Meningioma and duration of use, results corrected for confounders.
Cohort                   Gender              Exposure             Cases       IRR  95%CI
                                             Time since
                                             subscription (yrs)
Frei et al. (2011)52     Male                1-4                  15          0.92 0.55-1.56
                                             5-9                  14          0.56 0.33-0.96
                                             ≥10                  21          0.90 0.57-1.42
                         Female              1-4                  9           1.08 0.56-2.09
                                             5-9                  13          1.04 0.60-1.79
                                             ≥10                  8           0.93 0.46-1.87
Case-control             Type of phone       Exposure             Ca / Co     OR   95%CI
                                             Time since 1st use
                                             (yrs)
Hansson Mild et al.      Analogue            >1-5                 NR*         1.2  0.8 -1.8
(2007)65                                     >5-10                NR          1.2  0.8 -1.8
                                             >10                  NR          1.6  1.02-2.5
                         Digital             >1-5                 NR          1.0  0.8 -1.3
                                             >5-10                NR          1.1  0.8 -1.6
                                             >10                  NR          1.3  0.5 -3.2
                         Cordless            >1-5                 NR          1.0  0.8 -1.3
                                             >5-10                NR          1.3  1.01-1.8
                                             >10                  NR          1.6  0.9 -2.8
INTERPHONE study         Mobile (excl.       0                    1147 / 1174 1.0
group (2010)93           cordless)           1-1.9                178 / 214   0.90 0.68-1.18
                                             2-4                  557 / 675   0.77 0.65-0.92
                                             5-9                  417 / 487   0.76 0.63-0.93
                                             ≥10                  110 / 112   0.83 0.61-1.14
                                             Duration of
                                             subscription (yrs)
Auvinen et al. (2002)100 Mobile analogue 0                        121 / 615   1.0
                                             <1                   3/7         2.3  0.6 -9.2
                                             1-2                  3 / 10      1.6  0.4 -6.1
                                             >2                   2 / 11      1.0  0.2 -4.4
                         Mobile digital      0                    126 / 623   1.0
                                             <1                   1/9         0.6  0.1 -4.4
                                             1-2                  2 / 10      1.0  0.2 -4.6
                                             >2                   0/1         0.0  -
                         Mobile all          0                    NR          1.0
                                             <1                   NR          1.5  0.5 -4.6
                                             1-2                  NR          1.2  0.4 -3.6
                                             >2                   NR          0.8  0.2 -3.5
232            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 233 =================================================================

<br><br>====================================================================== Pagina 234 ======================================================================

<pre>                                            Time since 1st use
                                            (yrs)
Inskip et al. (2001)99    Mobile (excl.     0                    165 / 625   1.0
                          cordless)         <0.5                 6 / 56      0.5  0.2 -1.4
                                            0.5-<3               12 / 55     0.8  0.4 -1.9
                                            ≥3                   14 / 60     1.1  0.5 -2.5
                                            ≥5                   6 / 31      0.9  0.3-2.7
* NR: not reported
Table H8 Meningioma and cumulative use, results corrected for confounders.
                          Type of phone     Exposure             Ca / Co     OR   95%CI
                                            Cumulative call
                                            time (h)
INTERPHONE study          Mobile (excl.     0                    1147 / 1174 1.00
group (2010)93            cordless)         <5                   160 / 197   0.90 0.69-1.18
                                            5-12.9               142 / 159   0.82 0.61-1.10
                                            13-30.9              144 / 194   0.69 0.52-0.91
                                            31-60.9              122 / 145   0.69 0.51-0.94
                                            61-114.9             129 / 162   0.75 0.55-1.00
                                            115-199.9            96 / 155    0.69 0.50-0.96
                                            200-359.9            108 / 133   0.71 0.51-0.98
                                            360-734.9            123 / 133   0.90 0.66-1.23
                                            735-1639.9           108 / 103   0.76 0.54-1.08
                                            ≥1640                130 / 107   1.15 0.81-1.62
                                            Cumulative
                                            number of calls (x
                                            100)
INTERPHONE study          Mobile (excl.     0                    1147 / 1174 1.00
group (2010)93            cordless)         <1.5                 159 / 180   0.95 0.72–1.27
                                            1.5-3.4              136 / 182   0.62 0.46–0.83
                                            3.5-7.4              148 / 176   0.90 0.68–1.19
                                            7.6-13.9             176 / 173   0.80 0.61–1.07
                                            124-25.4             122 / 181   0.60 0.45–0.81
                                            25.5-41.4            111 / 126   0.81 0.58–1.13
                                            41.5-67.9            129 / 146   0.79 0.58–1.09
                                            68-127.9             134 / 126   0.92 0.67–1.26
                                            128-269.9            100 / 100   0.81 0.57–1.16
                                            ≥270                 80 / 98     0.80 0.55–1.17
                                            Cumulative call
                                            time (h)
Inskip et al. (2001)99    Mobile (excl.     0                    165 / 625   1.0
                          cordless)         <13                  8 / 55      0.7  0.3 -1.9
                                            13-100               13 / 58     1.1  0.5 -2.4
                                            >100                 11 / 54     0.7  0.3 -1.7
                                            >500                 6 / 27      0.7  0.2-2.4
               Results from the selected publications                                       233
</pre>

====================================================================== Einde pagina 234 =================================================================

<br><br>====================================================================== Pagina 235 ======================================================================

<pre>Table H9 Meningioma, analysis as continuous variables (Hansson Mild et al. (2007)65).
Variable                           Type of phone                   OR                                  95% CI
Per 100 h of use                   Digital                         0.99                                0.96-1.02
                                   Analogue                        1.02                                0.99 -1.05
                                   Cordless                        1.01                                0.997-1.02
Per 1 yr of use                    Digital                         1.02                                0.98-1.06
                                   Analogue                        1.05                                1.01 -1.09
                                   Cordless                        1.04                                1.01 -1.07
Table H10 Meningioma and laterality, results corrected for confounders.
                                                       Ipsilateral                              Contralateral
                            Exposure                   Ca / Co      OR           95%CI          Ca / Co       OR         95%CI
                            Time since first use (yrs)
Hardell et al. (2009)66 a Mobile >1                    167 / 374 1.3             1.01-1.7       125 / 308     1.1        0.8 -1.4
                            Mobile >10                 18 / 45      1.6          0.9 -2.9       12 / 29       1.6        0.7 -3.3
                            Cordless >1                134 / 309 1.2             0.9 -1.6       101 / 235     1.1        0.8 -1.5
                            Cordless >10               11 / 15      3.0          1.3 -7.2       7 / 20        1.1        0.5 -2.9
INTERPHONE study            0                          821 / 898 1.00                           832 / 841     1.00
group (2010)93              1-1.9                      54 / 79      0.71         0.44-1.15      41 / 59       0.67       0.38-1.20
                            2-4                        198 / 203 0.89            0.67-1.19      118 / 196     0.54       0.39-0.76
                            5-9                        132 / 155 0.87            0.63-1.21      100 / 126     0.64       0.44-0.94
                            ≥10                        40 / 42      0.88         0.52-1.47      20 / 25       0.58       0.29-1.16
                            Cumulative call time (h)
                            0                          821 / 828 1.00                           832 / 841     1.00
                            <5                         48 / 71      0.76         0.48-1.21      36 / 54       0.75       0.42-1.31
                            5-114.9                    185 / 209 0.86            0.65-1.15      125 / 190     0.55       0.40-0.75
                            115-359.9                  65 / 96      0.64         0.42-0.97      42 / 69       0.64       0.39-1.06
                            360-1639.9                 80 / 68      1.09         0.72-1.64      50 / 65       0.54       0.32-0.94
                            ≥1640                      46 / 35      1.45         0.80-2.61      28 / 28       0.62       0.31-1.25
                            Cumulative number of
                            calls (x100)
                            0                          821 / 891 1.00                           832 / 841     1.00
                            <1.5                       51 / 72      0.77         0.49-1.22      32 / 49       0.76       0.41-1.40
                            1.5-25.4                   187 / 229 0.80            0.60-1.05      131 / 191     0.59       0.44-0.81
                            25.5-67.9                  80 / 81      0.89         0.59-1.35      51 / 77       0.61       0.37-1.00
                            68-269.9                   76 / 61      1.22         0.77-1.95      49 / 66       0.39       0.23-0.68
                            ≥270                       30 / 36      1.01         0.56-1.82      18 / 23       0.66       0.30-1.46
a    This publication groups ipsilateral and ipsi/contralateral so the subjects that call at the side of the tumour and those who do
     this and alternate it with the other side are grouped, the other publications do not group these.
234            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 235 =================================================================

<br><br>====================================================================== Pagina 236 ======================================================================

<pre>               Acoustic neuroma
Table H11 Acoustic neuroma and duration of use, results corrected for confounders.
Cohort                                Gender                 Exposure           Cases     IRR  95%CI
                                                             Time since
                                                             subscription (yrs)
Schüz et al. (2011)51                 Men                    ≥11                15        0.87 0.52-1.46
Case-control                          Type of phone          Exposure           Ca / Co   OR   95%CI
                                                             Time since 1st use
                                                             (yrs)
Hansson Mild et al. (2007)65          Analogue               >1-5               NR        2.3  1.2 -4.1
                                                             >5-10              NR        3.4  2.1 -5.5
                                                             >10                NR        3.1  1.7 -5.7
                                      Digital                >1-5               NR        1.4  1.01-2.1
                                                             >5-10              NR        1.8  1.1 -3.0
                                                             >10                NR        0.6  0.1 -5.0
                                      Cordless               >1-5               NR        1.5  1.01-2.1
                                                             >5-10              NR        1.5  0.96-2.4
                                                             >10                NR        1.0  0.3 -2.9
INTERPHONE Study Group (2011)94 Mobile (excl.                0                  462 / 837 1.00
                                      cordless)              1-1.9              63 / 169  0.73 0.49–1.09
                                                             2-4                276 / 554 0.87 0.69–1.10
                                                             5-9                236 / 444 0.90 0.69–1.16
                                                             ≥10                68 / 141  0.76 0.52–1.11
Muscat et al. (2002)101               Mobile (excl.          0                  72 / 63   1.0
                                      cordless)              1-2                7 / 17    0.5  0.2 -1.3
                                                             3-6                11 / 6    1.7  0.5 -5.1
Inskip et al. (2001)99                Mobile (excl.          0                  74 / 625  1.0
                                      cordless)              <0.5               4 / 56    0.3  0.1 -1.3
                                                             0.5-<3             8 / 55    1.8  0.7 -4.5
                                                             ≥3                 10 / 60   1.4  0.6 -3.4
                                                             ≥5                 5 / 31    1.9  0.6-5.9
Case-case                             Reference date         Exposure           Ca        RRa* 95%CI
                                      (years before          Time since first
                                      diagnosis)             use at reference
                                                             date (yrs)
Sato et al. (2010)112                 1                      ≤5                 112       1.06 0.88-1.31
                                                             5-10               56        1.05 0.82-1.45
                                                             >10                12        1.62 0.79-4.77
                                      5                      ≤5                 123       1.11 0.92-1.38
                                                             5-10               21        1.56 0.90-3.34
                                                             >10                6         1.00 0.59-3.23
a    RR: risk ratio.
               Results from the selected publications                                                    235
</pre>

====================================================================== Einde pagina 236 =================================================================

<br><br>====================================================================== Pagina 237 ======================================================================

<pre>Table H12 Acoustic neuroma and cumulative use, results corrected for confounders.
Case-control              Type of phone    Exposure                     Ca / Co   OR   95%CI
                                           Cumulative call time (h)
Hardell et al. (2006)64   Analogue         1-500                        55 / 252  2.8  1.8-4.2
                                           501-1000                     7 / 29    3.3  1.3-8.0
                                           >1000                        6 / 16    5.1  1.9-14
                          Digital          1-500                        83 / 667  1.4  0.99-2.0
                                           501-1000                     10 / 64   1.8  0.8-3.8
                                           >1000                        12 / 45   3.1  1.5-6.4
                          Cordless         1-500                        60 / 502  1.3  0.9-1.9
                                           501-1000                     15 / 97   1.6  0.9-3.0
                                           >1000                        21 / 102  2.1  1.2-3.7
INTERPHONE Study          Mobile (excl.    0                            462 / 837 1.00
Group (2011)94            cordless)        <5                           58 / 144  0.77 0.52–1.15
                                           5-12.9                       63 / 129  0.80 0.54–1.18
                                           13-30.9                      80 / 136  1.04 0.71–1.52
                                           31-60.9                      66 / 131  0.95 0.63–1.42
                                           61-114.9                     74 / 137  0.96 0.66–1.41
                                           115-199.9                    68 / 128  0.96 0.65–1.42
                                           200-359.9                    50 / 144  0.60 0.39–0.91
                                           360-734.9                    58 / 126  0.72 0.48–1.09
                                           735-1639.9                   49 / 126  0.48 0.30–0.78
                                           ≥1640                        77 / 107  1.32 0.88–1.97
                                           Cumulative number of
                                           calls (x 100)
INTERPHONE Study          Mobile (excl.    0                            462 / 837 1.00
Group (2011)94            cordless)        <1.5                         59 / 135  0.76 0.51–1.14
                                           1.5-3.4                      60 / 137  0.68 0.45–1.03
                                           3.5-7.4                      73 / 135  1.11 0.76–1.61
                                           7.5-13.9                     87 / 138  1.22 0.84–1.77
                                           14-25.4                      79 / 132  1.11 0.75–1.64
                                           25.5-41.4                    55 / 137  0.64 0.42–0.98
                                           41.5-67.9                    50 / 133  0.74 0.49–1.12
                                           68-127.9                     62 / 133  0.65 0.43–0.98
                                           128-269.9                    56 / 115  0.67 0.44–1.02
                                           ≥270                         62 / 113  0.93 0.61–1.41
                                           Cumulative call time (h)
Muscat et al. (2002)101   Mobile (excl.    0                            72 / 63   1.0
                          cordless)        1-60                         9 / 11    0.9  0.3 -3.1
                                           >60                          9 / 12    0.7  0.2 -2.6
236            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 237 =================================================================

<br><br>====================================================================== Pagina 238 ======================================================================

<pre>Table H13 Acoustic neuroma, analysis as continuous variables (Hansson Mild et al. (2007)65)
Variable                   Type of phone                             OR                            95% CI
Per 100 h of use           Digital                                   1.03                          0.998-1.06
                           Analogue                                  1.05                          1.02 -1.9
                           Cordless                                  1.01                          0.997-1.02
Per 1 yr of use            Digital                                   1.06                          0.995-1.13
                           Analogue                                  1.12                          1.06 -1.17
                           Cordless                                  1.04                          0.99 -1.10
Table H14 Acoustic neuroma and laterality, results corrected for confounders.
                                                       Ipsilateral                            Contralateral
                      Exposure                         Ca / Co       OR      95%CI            Ca / Co        OR       95%CI
                      Time since first use (yrs)
Hardell et al.,       Mobile >1                        80 / 374      1.8     1.2 -2.6         48 / 308       1.4      0.9 -2.1
(2009)66 a            Mobile >10                       13 / 45       3.0     1.4 -6.2         6 / 29         2.4      0.9 -6.3
                      Cordless >1                      67 / 309      1.7     1.2 -2.5         28 / 235       1.1      0.7 -1.7
                      Cordless >10                     3 / 15        2.3     0.6 -8.8         1 / 20         0.5      0.1 -4.0
INTERPHONE            0                                416 / 615     1.00                     405 / 625      1.00
Study Group,          1-1.9                            23 / 62       0.42 0.22-0.81           32 / 51        1.75     0.90–3.42
(2011)94              2-4                              103 / 204     0.70 0.49-1.00           123 / 189      0.80     0.56–1.13
                      5-9                              101 / 153     0.95 0.64-1.41           89 / 120       0.96     0.64–1.43
                      ≥ 10                             44 / 52       1.18 0.69-2.04           17 / 30        0.69     0.33–1.42
                      Cumulative call time (h)
                      0                                416 / 615     1.00                     405 / 625      1.00
                      <5                               23 / 44       0.81 0.43-1.52           28 / 56        0.83     0.44–1.56
                      5.0-114.9                        108 / 200     0.71 0.50-1.00           131 / 151      1.28     0.90–1.83
                      115-359.9                        47 / 95       0.67 0.40-1.12           49 / 92        0.66     0.41–1.07
                      360-1639.9                       46 / 86       0.51 0.30-0.88           37 / 65        0.67     0.38–1.15
                      ≥1640                            47 / 46       2.33 1.23-4.40           16 / 26        0.72     0.34–1.53
                      Cumulative number of calls (x
                      100)
                      0                                416 / 615     1.00                     405 / 625      1.00
                      <1.5                             24 / 46       0.67 0.35-1.28           29 / 49        0.98     0.52-1.84
                      1.5-25.4                         108 / 193     0.81 0.57-1.14           143 / 158      1.36     0.96-1.93
                      25.5-67.9                        48 / 108      0.56 0.34-0.90           34 / 90        0.51     0.31-0.86
                      68-269.9                         50 / 81       0.68 0.40-1.13           44 / 66        0.67     0.39-1.14
                      ≥270                             41 / 43       1.67 0.90-3.09           11 / 27        0.52     0.21-1.26
a    This publication groups ipsilateral and ipsi/contralateral so the subjects that call at the side of the tumour and those who do
     this and alternate it with the other side are grouped, the other publications do not group these.
              Results from the selected publications                                                                            237
</pre>

====================================================================== Einde pagina 238 =================================================================

<br><br>====================================================================== Pagina 239 ======================================================================

<pre>               Parotid gland tumours
Table H15 Parotid gland tumours and duration of use, results corrected for confounders.
                   Type of tumour       Type of phone           Exposure           Ca / Co   OR   95%CI
                                                                Time since first
                                                                use (yrs)
Hardell et al.     All                  Analogue                 >1                31 / 137  0.92 0.58-1.44
(2004)69                                                         >5                17 / 88   0.78 0.44-1.38
                                                                 >10               6 / 35    0.71 0.29-1.74
                                        Digital                  >1                45 / 170  1.01 0.68-1.50
                                                                 >5                8 / 27    1.22 0.54-2.78
                                                                 >10               -         -    -
                                        Cordless                 >1                48 / 200  0.99 0.68-1.43
                                                                 >5                18 / 66   1.15 0.07-2.03
                                                                 >10               0/5       -    -
                                        All                      >1                91 / 352  1.02 0.75-1.38
                                                                 >5                32 / 145  0.90 0.58-1.39
                                                                 >10               6 / 38    0.65 0.27-1.59
Söderqvist et al.  All                  Analogue                 ≤ 52 h > 10 y     2/7       0.7  0.1-4.3
(2012)72                                                         > 52 h > 10 y     0 / 10    -    -
                                                                 All >10 y         2 / 17    0.3  0.1-1.7
                                        Digital                  ≤ 69 h > 10 y     0/0       -    -
                                                                 > 69 h > 10 y     2/5       1.3  0.2-7.4
                                                                 All >10 y         2/5       1.3  0.2-7.4
                                        Cordless                 ≤ 304 h >10y      1/4       1.0  0.1-9.6
                                                                 > 304 h >10y      3/8       1.1  0.2-5.2
                                                                 All >10 y         4 / 12    1.0  0.3-3.7
                                        Mobiles                  ≤ 66 h > 10 y     0/2       -    -
                                                                 > 66 h > 10 y     2 / 18    0.3  0.1-1.4
                                                                 All >10 y         2 / 20    0.3  0.1-1.4
Sadetzki et al.    All                  Mobile (excl.            0                 175 / 575 1.0
(2007)85                                cordless)                1-4.9             148 / 405 0.84 0.63-1.12
                                                                 5-9.9             124 / 264 0.92 0.67-1.27
                                                                 ≥10               13 / 22   1.0  0.48-2.09
                   Benign                                        0                 150 / 469 1.0
                                                                 1-4.9             127 / 351 0.79 0.54-1.08
                                                                 5-9.9             113 / 234 0.92 0.65-1.29
                                                                 ≥10               12 / 18   1.11 0.50-2.44
                   Malignant                                     0                 25 / 106  1.0
                                                                 1-4.9             21 / 54   1.25 0.58-2.68
                                                                 5-9.9             11 / 30   0.92 0.37-2.27
                                                                 ≥10               1/4       0.47 0.05-4.51
238            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 239 =================================================================

<br><br>====================================================================== Pagina 240 ======================================================================

<pre>Lönn et al. (2006)82 Benign              Mobile (excl.         0                  35 / 119 1.0
                                         cordless)             <5                 47 / 104 1.0    0.6 -1.8
                                                               5-9                23 / 76  0.8    0.4 -1.5
                                                               ≥10                7 / 15   1.4    0.5 -3.9
                     Malignant                                 0                  35 / 280 1.0
                                                               <5                 14 / 228 0.7    0.3 -1.3
                                                               5-9                8 / 128  0.7    0.3 -1.7
                                                               ≥10                2 / 36   0.4    0.1 -2.6
                                                               Duration of
                                                               subscription (y)
Auvinen et al.                           Analogue              0                  31 / 155 1.0
(2002)100                                                      <1                 0/3      -      -
                                                               1-2                2 / 11   0.9    0.2 -4.9
                                                               >2                 1/1      4.4    0.3 -71.6
                                         Digital               0                  33 / 167 1.0
                                                               <1                 0/2      -      -
                                                               1-2                1/1      5.0    0.3 -80.0
                                                               >2                 0/0      -      -
                                         All phones            <1                 -        -      -
                                                               1-2                -        1.7    0.4 -7.5
                                                               >2                 -        2.3    0.2 -25.3
Table H16 Parotid gland tumours and cumulative use, results corrected for confounders.
                          Type of tumour Type of phone        Exposure           Ca / Co     OR       95%CI
                                                              Cumulative call
                                                              time (h)
Hardell et al. (2004)69 All               Analogue            >1 y, >91 h        15 / 68     0.90     0.49-1.66
                                                              >5 y, > 91 h       10 / 52     0.78     0.38-1.61
                                                              >10 y, >91 h       4 / 25      0.66     0.22-1.95
                                          Digital             >1 y, >64 h        23 / 81     1.07     0.67-1.71
                                                              >5 y, >64 h        6 / 20      1.25     0.48-3.21
                                                              >10 y, >64 h       -           -        -
                                          Cordless            >1 y, >183 h       21 / 97     0.89     0.53-1.50
                                                              >5 y, >183 h       12 / 41     1.24     0.62-2.44
                                                              >10 y, >183h       0/4         -        -
                                          All                 >1 y, >182 h       42 / 175    0.94     0.63-1.39
                                                              >5 y, >182 h       21 / 100    0.86     0.51-1.44
                                                              >10 y, >182h       4 / 31      0.53     0.18-1.55
Söderqvist et al. (2012)72All             Analogue            1-1000 h           9/31        0.9      0.3-2.4
                                                              1001-2000 h        0/1         -        -
                                                              >2000 h            0/0         -        -
                                          Digital             1-1000 h           28 / 95     1.9      0.4-1.7
                                                              1001-2000 h        2/4         1.4      0.2-8.8
                                                              >2000 h            0/ 5        -        -
               Results from the selected publications                                                           239
</pre>

====================================================================== Einde pagina 240 =================================================================

<br><br>====================================================================== Pagina 241 ======================================================================

<pre>                                           Cordless                1-1000 h         17 / 80           0.6      0.3-1.3
                                                                   1001-2000 h      2/4               1.2      0.2-2.8
                                                                   >2000 h          0/9               -        -
                                           Mobiles                 1-1000 h         28 / 98           0.9      0.4-1.7
                                                                   1001-2000 h      2/8               0.86     0.1-3.6
                                                                   >2000 h          0/5               0.53     -
Sadetzki et al. (2007)85 All               Mobile (excl.           0                176 / 578         1.0
                                           cordless)               ≤266.3           121 / 390         0.82     0.62-1.09
                                                                   266.4-1034.9     80 / 155          1.03     0.72-1.47
                                                                   ≥1035            83 / 134          1.09     0.75-1.60
                           Benign                                  0                151 / 480         1.0
                                                                   ≤266.3           103 / 336         0.78     0.57-1.06
                                                                   266.4-1034.9     75 / 139          1.05     0.72-1.53
                                                                   ≥1035            73 / 117          1.08     0.72-1.62
                           Malignant                               0                25 / 107          1.0
                                                                   ≤266.3           18 / 54           1.21     0.58-2.53
                                                                   266.4-1034.9     5 / 16            0.67     0.19-2.38
                                                                   ≥1035            10 / 17           1.22     0.43-3.48
Lönn et al. (2006)82       Benign          Mobile (excl.           0                35 / 119          1.0
                                           cordless)               <30              20 / 45           1.1      0.6 -2.3
                                                                   30-449           34 / 92           0.9      0.5 -1.6
                                                                   >450             22 / 52           1.0      0.5 -2.1
                           Malignant                               0                35 / 280          1.0
                                                                   <30              7 / 110           0.7      0.3 -1.6
                                                                   30-449           11 / 184          0.7      0.3 -1.4
                                                                   >450             5 / 90            0.6      0.2 -1.8
Table H17 Parotid gland tumours and laterality, results corrected for confounders.
                                                             Ipsilateral                 Contralateral
                 Type of tumour Exposure                     Ca / Co      OR     95%C Ca / Co          OR         95%CI
                                  Time since first use (yrs)                     I
Sadetzki et al. All               0                          175 / 575    1.00                 175 / 575   1.00
(2007)85                          1-4.9                      84 / 220     0.88    0.63-1.24    53 / 166    0.82 0.56-1.21
                                  5-9.9                      83 / 148     1.13    0.78-1.64    45 / 118    0.96 0.63-1.46
                                  ≥10                        10 / 13      1.89    0.79-4.57    3 / 10      0.58 0.15-2.32
                                  Cumulative call time (h)
                                  0                          176 / 583    1.00                 175 / 578   1.00
                                  <266.3                     67 / 224     0.79    0.56-1.11    53 / 162    0.92 0.63-1.34
                                  >266.3                     115 / 158    1.49    1.05-2.13    48 / 129    0.84 0.55-1.28
                                  Time since first use (yrs)
Lönn et al.      Benign           0                          58 / 210     1.0                  74 / 209    1.0
(2006)82                          <5                         30 / 57      1.4     0.9-2.2      24 / 60     0.9    0.5-1.5
                                  5-9.9                      17 / 41      1.5     0.7-2.8      10 / 40     0.6    0.3-1.2
                                  ≥10                        4/8          2.0     0.5-7.0      1/8         0.3    0.0-2.6
240           Mobile phones and cancer
</pre>

====================================================================== Einde pagina 241 =================================================================

<br><br>====================================================================== Pagina 242 ======================================================================

<pre>               Malignant         0                         36 / 452      1.0                  45 / 460   1.0
                                 <5                        9 / 125       1.2      0.5-2.6     5 / 130    0.5    0.2-1.3
                                 5-9.9                     6 / 72        1.3      0.5-3.6     2 / 66     0.4    0.1-1.8
                                 ≥10                       1 / 23        0.7      0.1-5.7     0 / 16     -      -
            Pituitary tumours
Table H18 Pituitary tumours and duration of use, results corrected for confounders.
                                    Type of phone          Exposure             Ca / Co        OR          95%CI
                                                           Time since first use
                                                           (yrs)
Takebayashi et al. (2008)88         All                    0                    39 / 56        1.00
                                                           <2.4                 14 / 25        0.86        0.39-1.88
                                                           2.4-4.5              13 / 27        0.75        0.31-1.81
                                                           4.5-7.2              22 / 26        1.64        0.74-3.66
                                                           >7.2                 13 / 27        0.75        0.31-1.82
Schoemaker et al. (2009)89          All                    0                    116 / 545      1.0
                                                           1.5-4                89 / 197       1.0         0.7-1.5
                                                           5-9                  62 / 140       0.8         0.5-1.2
                                                           10-17                24 / 48        1.0         0.5-1.9
                                    Analogue               0                    116 / 245      1.0
                                                           1.5-4                2 / 13         0.4         0.1-2.1
                                                           5-9                  18 / 44        0.9         0.5-1.9
                                                           ≥10                  19 / 41        1.2         0.6-2.4
                                    Digital                0                    116 / 245      1.0
                                                           1.5-4                103 / 236      1.0         0.7-1.4
                                                           5-9                  53 / 120       0.7         0.4-1.1
                                                           ≥10                  10 / 6         2.5         0.7-9.1
Table H19. Pituitary tumours and cumulative use, results corrected for confounders.
                                Type of phone        Exposure                       Ca / Co         OR       95%CI
                                                     Cumulative call time (h)
Takebayashi et al. (2008)88     All                  0                              39 / 56         1.00
                                                     <39                            15 / 26         1.00     0.46-2.16
                                                     39-190                         14 / 26         0.97     0.40-2.32
                                                     190-560                        12 / 26         0.72     0.31-1.70
                                                     >560                           21 / 27         1.33     0.58-3.09
Schoemaker et al. (2009)89      All                  0                              116 / 245       1.0
                                                     <113                           79 / 190        0.9      0.6-1.3
                                                     113-596                        44 / 91         1.1      0.7-1.8
                                                     >596                           51 / 95         1.1      0.7-1.7
                                Analogue             0                              116 / 245       1.0
                                                     <96                            13 / 48         0.7      0.3-1.4
            Results from the selected publications                                                                      241
</pre>

====================================================================== Einde pagina 242 =================================================================

<br><br>====================================================================== Pagina 243 ======================================================================

<pre>                                                      96-371                      11 / 24         1.2 0.5-2.9
                                                      >371                        15 / 24         1.5 0.7-3.4
                               Digital                0                           116 / 245       1.0
                                                      <94                         75 / 178        0.9 0.6-1.3
                                                      94-453                      37 / 88         0.9 0.5-1.5
                                                      >453                        53 / 89         1.2 0.7-1.9
                                                      Cumulative number of calls
Schoemaker et al. (2009)89     All                    0                           116 / 245       1.0
                                                      <2203                       72 / 191        0.8 0.6-1.2
                                                      2203-8300                   45 / 94         1.1 0.7-1.8
                                                      >8300                       57 / 95         1.2 0.7-1.9
               Malignant melanoma of the eye
Table H20 Malignant melanoma of the eye and duration of use, results corrected for confounders.
                                                             Exposure                Ca / Co    OR     95%CI
Stang et al. (2001)98      Possible / probable / certain     Ever                    7 / 25     2.8    1.0-7.9
                           mobile phone exposure             ≥5 yrs before reference 4 / 10     4.1    0.7-24.0
                                                             date
                                                             ≥3 yrs                  6 / 16     3.0    0.9-9.7
                           Type of controls                  Duration of regular use Ca / Co    OR     95%CI
                                                             (y)
Stang et al. (2009)104     Population controls               0                       24 / 20    1.0
                                                             ≤4                      17 / 19    0.8    0.5-1.2
                                                             >5-9                    11 / 14    0.6    0.4-1.0
                                                             ≥10                     2/3        0.6    0.3-1.4
                           Ophthalmologist controls          0                       32 / 24    1.0
                                                             ≤4                      17 / 19    1.0    0.5-2.2
                                                             >5-9                    10 / 8     1.3    0.5-3.2
                                                             ≥10                     4/3        1.5    0.3-6.6
                           Sibling controls                  0                       14 / 17    1.0
                                                             ≤4                      21 / 18    1.4    0.6-3.3
                                                             >5-9                    13 / 13    1.1    0.4-2.8
                                                             ≥10                     2/3        0.7    0.2-3.0
242            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 243 =================================================================

<br><br>====================================================================== Pagina 244 ======================================================================

<pre>Table H21 Malignant melanoma of the eye and cumulative use, results corrected for confounders.
                           Type of controls              Exposure                 Ca / Co      OR  95%CI
                                                         Cumulative call time
                                                         (h)
Stang et al. (2009)104     Population controls           0                        24 / 20      1.0
                                                         ≤44                      11 / 15      0.6 0.4-1.0
                                                         >44-≤195                 9/8          0.9 0.5-1.5
                                                         >195                     10 / 12      0.8 0.5-1.3
                           Ophthalmologist controls      0                        23 / 24      1.0
                                                         ≤44                      14 / 13      1.2 0.6-2.8
                                                         >44-≤195                 7/8          0.9 0.3-2.4
                                                         >195                     10 / 8       1.2 0.4-3.6
                           Sibling controls              0                        14 / 17      1.0
                                                         ≤44                      12 / 16      0.8 0.3-2.1
                                                         >44-≤195                 11 / 8       1.7 0.7-4.5
                                                         >195                     13 / 11      1.5 0.5-4.3
                                                         Cumulative number of
                                                         calls
                           Population controls           0                        24 / 20      1.0
                                                         Sporadic                 47/44        0.9 0.7-1.3
                                                         ≤1176                    17 /19       0.8 0.5-1.2
                                                         >1176-≤4350              11 / 14      0.6 0.4-1.0
                                                         >4350                    2/3          0.6 0.3-1.4
                           Ophthalmologist controls      0                        23 / 24      1.0
                                                         Sporadic                 47/46        1.2 0.7-2.2
                                                         ≤1176                    17 / 19      1.0 0.5-2.2
                                                         >1176-≤4350              10 / 8       1.3 0.5-3.2
                                                         >4350                    4/3          1.5 0.3-6.6
                           Sibling controls              0                        14 / 17      1.0
                                                         Sporadic                 49/48        1.3 0.6-2.5
                                                         ≤1176                    21/18        1.4 0.6-3.3
                                                         >1176-≤4350              13/13        1.1 0.4-2.8)
                                                         >4350                    2/3          0.7 0.2-3.0
               Results from the selected publications                                                       243
</pre>

====================================================================== Einde pagina 244 =================================================================

<br><br>====================================================================== Pagina 245 ======================================================================

<pre>244 Mobile phones and cancer</pre>

====================================================================== Einde pagina 245 =================================================================

<br><br>====================================================================== Pagina 246 ======================================================================

<pre>Annex I
      Meta-analysis and forest plots
      Two models have been used to calculate the pooled estimates, using metaan.ado
      in Stata. The first is a fixed effects model, the second a random effects model
      (DerSimonian-Laird).The pooled variance includes the spread between the
      different studies and is therefore sometimes considerably larger than the variance
      of the individual studies. When there is no heterogeneity, the fixed and random
      effect estimates of the pooled effect are equal. Heterogeneity between studies has
      been determined using the Cochrane Q with p-value. A high Q and low p-value
      indicate heterogeneity between studies. When p was <0.05, heterogeneity was
      considered to be too large for a meaningful pooling of the data. This has
      nevertheless be done, but in those cases the data are only shown in this Annex in
      the tables, to show the differences. If p >0.05, heterogeneity was considered
      small enough to perform a meta-analysis. The figures drawn from the data in the
      tables are in those cases shown in the main text.
          OR, CI1 and CI2 are the odds ratio, lower and upper 95% confidence
      intervals, respectively, as reported in the papers. The log(OR) should be exactly
      the mean of log(CI1) en log(CI2). This is not always the case, as a result of
      rounding and reporting not enough decimal numbers.
      Meta-analysis and forest plots                                                     245
</pre>

====================================================================== Einde pagina 246 =================================================================

<br><br>====================================================================== Pagina 247 ======================================================================

<pre>Tables I1 Glioma, time since first use ≥10 years, Hardell 20-80 year.
Data
Study                                                   logOR         logCI1       logCI2         OR        CI1       CI2
Frei et al. (2011)52, females, ≥10 yr                   0.039         -0.580       0.668          1.04      0.56      1.95
Frei et al. (2011)52, males, ≥10 yr                     0.039         -0.163       0.247          1.04      0.85      1.28
Hardell et al. (2011)68, ≥10 yr                         0.916         0.588        1.194          2.50      1.80      3.30
INTERPHONE (2010)93, ≥10 yr                             -0.020        -0.274        0.231         0.98      0.76      1.26
Fixed-effects model
Study                                     Effect     [95% Conf. Interval]          % Weight          OR       CI1      CI2
Frei et al. (2011)52, females             0.039      -0.585         0.663          4.85
Frei et al. (2011)52, males,              0.039      -0.165         0.244          45.05
Hardell et al. (2011)68                   0.916      0.613          1.219          20.55
INTERPHONE (2010)93                       -0.020     -0.273         0.233          29.54
Overall effect                            0.202      0.065          0.339          100.00            1.22     1.07     1.40
Random-effects model
Study                                     Effect     [95% Conf. Interval]          % Weight          OR       CI1      CI2
Frei et al. (2011)52, females             0.039      -0.585         0.663          18.86
Frei et al. (2011)52, males,              0.039      -0.165         0.244          27.92
Hardell et al. (2011)68                   0.916      0.613          1.219          26.12
INTERPHONE (2010)93                       -0.020     -0.273         0.233          27.10
Overall effect                            0.252      -0.197         0.701          100.00            1.29     0.82     2.02
Heterogeneity
                    Value      df    p-value
Cochrane Q          27.00      3     0.000
Tables I2 Glioma, time since first use ≥10 years, without Hardell. This is the same analysis as the previous one, except without
the Hardell data.
Study                                                   logOR         logCI1       logCI2         OR        CI1       CI2
Frei et al. (2011)52, females, ≥10 yr                   0.039         -0.580       0.668          1.04      0.56      1.95
Frei et al. (2011)52, males, ≥10 yr                     0.039         -0.163       0.247          1.04      0.85      1.28
INTERPHONE (2010)93, ≥10 yr                             -0.020        -0.274       0.231          0.98      0.76      1.26
Fixed-effects model
Study                                      Effect     [95%Conf.Interval]           %Weight          OR        CI1      CI2
Frei et al. (2011)52, females              0.039      -0.585        0.663          6.11
Frei et al. (2011)52, males,               0.039      -0.165        0.244          56.71
INTERPHONE (2010)93                        -0.020     -0.273        0.233          37.19
Overall effect                             0.017      -0.137        0.171          100.00           1.02      0.87     1.19
246            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 247 =================================================================

<br><br>====================================================================== Pagina 248 ======================================================================

<pre>Random-effects model
Study                                     Effect      [95%Conf.Interval]    %Weight      OR      CI1    CI2
Frei et al. (2011)52, females             0.039       -0.585        0.663   6.11
Frei et al. (2011)52, males               0.039       -0.165        0.244   56.71
INTERPHONE (2010)93                       -0.020      -0.273        0.233   37.19
Overall effect                            0.017       -0.137        0.171   100.00       1.02    0.87   1.19
Heterogeneity
                    Value     df   p-value
Cochrane Q          0.13      2    0.935
Tables I3 Glioma, time since first use ≥10 years, Hardell 30-59 year.
Study                                               logOR         logCI1  logCI2     OR        CI1     CI2
Frei (2011)52, females, ≥10 yr                      0.039         -0.580  0.668      1.04      0.56    1.95
Frei (2011)\52, males, ≥10 yr                       0.039         -0.163  0.247      1.04      0.85    1.28
Hardell et al. (2011)165, ≥10 yr                    0.582         0.174   0.993      1.79      1.19    2.70
INTERPHONE (2010)93, ≥10 yr                         -0.020        -0.274  0.231      0.98      0.76    1.26
Fixed-effects model
Study                                    Effect     [95% Conf. Interval]    % Weight     OR       CI1   CI2
Frei et al. (2011)52, females            0.039      -0.585          0.663   5.35
Frei et al. (2011)52, males              0.039      -0.165          0.244   49.68
Hardell et al. (2011)165                 0.582      0.173           0.992   12.40
INTERPHONE (2010)93                      -0.020     -0.273          0.233   32.57
Overall effect                           0.087      -0.057          0.231   100.00       1.09     0.94  1.26
Random-effects model
Study                                    Effect     [95% Conf. Interval]    % Weight      OR      CI1   CI2
Frei et al. (2011)52, females            0.039      -0.585          0.663   11.42
Frei et al. (2011)52, males              0.039      -0.165          0.244   36.36
Hardell et al. (2011)165                 0.582      0.173           0.992   20.29
INTERPHONE (2010)93                      -0.020     -0.273          0.233   31.94
Overall effect                           0.130      -0.110          0.371   100.00        1.14    0.90  1.45
Heterogeneity
                    value     df   p-value
Cochrane Q          6.54      3    0.088
               Meta-analysis and forest plots                                                                247
</pre>

====================================================================== Einde pagina 248 =================================================================

<br><br>====================================================================== Pagina 249 ======================================================================

<pre>Tables I4 Glioma, cumulative call time, Hardell 20-80 year.
Study                                                 logOR       logCI1   logCI2 OR    CI1   CI2
Hardell et al. (2011)68, >2000 hr                     1.163       0.693    1.629  3.20  2.00  5.10
INTERPHONE (2010)93, >1640 hr                         0.336       0.030    0.637  1.40  1.03  1.89
Fixed-effects model
Study                                  Effect    [95%Conf.Interval]      %Weight   OR    CI1  CI2
Hardell et al. (2011)68                1.163     0.695         1.631     29.60
INTERPHONE (2010)93                    0.336     0.033         0.640     70.40
Overall effect                         0.581     0.327         0.836     100.00    1.79  1.39 2.31
Random-effects model
Study                                  Effect    [95%Conf.Interval]      %Weight   OR   CI1   CI2
Hardell et al. (2011)68                1.163     0.695         1.631     47.58
INTERPHONE (2010)93                    0.336     0.033         0.640     52.42
Overall effect                         0.730     -0.079        1.539     100.00    2.08 0.92  4.66
  Heterogeneity
                     value      df   p-value
  Cochrane Q         8.44       1    0.004
Tables I5 Glioma, cumulative call time, Hardell 30-59 year.
Study                                                 logOR       logCI1   logCI2 OR    CI1   CI2
Hardell et al. (2011)165, >1640 hr                    0.560       0.020    1.099  1.75  1.02  3.00
INTERPHONE (2010)93, >1640 hr                         0.336       0.030    0.637  1.40  1.03  1.89
Fixed-effects model
Study                                  Effect    [95%Conf.Interval]      %Weight   OR    CI1  CI2
Hardell et al. (2011)165               0.560     0.020         1.099     24.05
INTERPHONE (2010)93                    0.336     0.033         0.640     75.95
Overall effect                         0.390     0.126         0.655     100.00    1.48  1.13 1.93
Random-effects model
Study                                  Effect    [95%Conf.Interval]      %Weight   OR   CI1   CI2
Hardell et al. (2011)165               0.560     0.020        1.099      24.05
INTERPHONE (2010)93                    0.336     0.033        0.640      75.95
Overall effect                         0.390     0.126        0.655      100.00    1.48 1.13  1.93
Heterogeneity
                   value      df   p-value
Cochrane Q         0.50       1    0.480
248            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 249 =================================================================

<br><br>====================================================================== Pagina 250 ======================================================================

<pre>Tables I6 Glioma, time since first use ≥10 year, ipsilateral, Hardell 20-80 year.
Study                                                          logOR       logCI1       logCI2   OR      CI1    CI2
Hardell et al. (2009)66, ipsilateral, ≥10 yr                   1.194       0.693        1.686     3.30   2.00   5.40
INTERPHONE (2010)93, ipsilateral, ≥10 yr                       0.191       -0.198       0.588     1.21   0.82   1.80
Fixed-effects model
Study                                    Effect   [95%Conf.Interval]              %Weight       OR      CI1    CI2
Hardell et al. (2009)66                  1.194    0.697             1.691         38.52
INTERPHONE (2010)93                      0.191    -0.202            0.584         61.48
Overall effect                           0.577    0.269             0.885         100.00        1.78    1.31   2.42
Random-effects model
Study                                    Effect   [95%Conf.Interval]              %Weight       OR      CI1    CI2
Hardell et al. (2009)66                  1.194    0.697             1.691         48.81
INTERPHONE (2010)93                      0.191    -0.202            0.584         51.19
Overall effect                           0.680    -0.303            1.663         100.00        1.97    0.74   5.28
Heterogeneity
                   value       df     p-value
Cochrane Q         9.64        1      0.002
Tables I7 Glioma, time since first use ≥10 year, contralateral, Hardell 20-80 year.
Study                                                            logOR       logCI1      logCI2    OR     CI1   CI2
Hardell et al. (2009)66, contralateral, ≥10yr                    1.030       0.405       1.629     2.80   1.50  5.10
INTERPHONE (2010)93, contralateral, ≥10yr                        -0.357      -0.868      0.140     0.70   0.42  1.15
Fixed-effects model
Study                                    Effect    [95%Conf.Interval]              %Weight      OR      CI1    CI2
Hardell et al. (2009)66                  1.030     0.418            1.642          40.39
INTERPHONE (2010)93                      -0.357    -0.860           0.147          59.61
Overall effect                           0.203     -0.186           0.592          100.00       1.23    0.83   1.81
Random-effects model
Study                                    Effect    [95%Conf.Interval]              %Weight      OR      CI1    CI2
Hardell et al. (2009)66                  1.030     0.418            1.642          49.18
INTERPHONE (2010)93                      -0.357    -0.860           0.147          50.82
Overall effect                           0.325     -1.033           1.684          100.00       1.38    0.36   5.39
Heterogeneity
                   value       df     p-value
Cochrane Q         11.76       1      0.001
               Meta-analysis and forest plots                                                                        249
</pre>

====================================================================== Einde pagina 250 =================================================================

<br><br>====================================================================== Pagina 251 ======================================================================

<pre>Tables I8 Glioma, time since first use ≥10 year, ipsilateral, Hardell 30-59 year.
Study                                                          logOR       logCI1       logCI2    OR      CI1   CI2
Hardell et al. (2011)165, ipsilateral, ≥10 yr                  0.829       0.285        1.379     2.29    1.33  3.97
INTERPHONE (2010)93, ipsilateral, ≥10 yr                       0.191       -0.198       0.588     1.21    0.82  1.80
Fixed-effects model
Study                                     Effect  [95%Conf.Interval]              %Weight        OR      CI1    CI2
Hardell et al. (2011)165                  0.829   0.282            1.375          34.08
INTERPHONE (2010)93                       0.191   -0.202           0.584          65.92
Overall effect                            0.408   0.089            0.727          100.00         1.50    1.09   2.07
Random-effects model
Study                                     Effect   [95%Conf.Interval]             %Weight        OR      CI1    CI2
Hardell et al. (2011)165                  0.829    0.282           1.375          45.38
INTERPHONE (2010)93                       0.191    -0.202          0.584          54.62
Overall effect                            0.480    -0.142          1.103          100.00         1.62    0.87   3.01
Heterogeneity
                   value       df     p-value
Cochrane Q         3.45        1      0.063
Tables I9 Glioma, time since first use ≥10 year, contralateral, Hardell 30-59 year.
Study                                                            logOR       logCI1       logCI2   OR      CI1   CI2
Hardell et al. (2011)165, contralateral, ≥10 yr                  0.536       -0.117       1.188     1.71   0.89  3.28
INTERPHONE (2010)93, contralateral, ≥10 yr                       -0.357      -0.868       0.140     0.70   0.42  1.15
Fixed-effects model
Study                                     Effect    [95%Conf.Interval]            %Weight        OR      CI1    CI2
Hardell et al. (2011)165                  0.536     -0.116          1.189         37.36
INTERPHONE (2010)93                       -0.357    -0.860          0.147         62.64
Overall effect                            -0.023    -0.422          0.376         100.00         0.98    0.66   1.46
Random-effects model
Study                                     Effect    [95%Conf.Interval]            %Weight        OR      CI1    CI2
Hardell et al. (2011)165                  0.536     -0.116          1.189         47.20
INTERPHONE (2010)93                       -0.357    -0.860          0.147         52.80
Overall effect                            0.065     -0.809          0.939         100.00         1.07    0.45   2.56
Heterogeneity
                   Value        df     p-value
Cochrane Q         4.51         1      0.034
250            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 251 =================================================================

<br><br>====================================================================== Pagina 252 ======================================================================

<pre>Tables I10 Glioma, cumulative call time, ipsilateral, Hardell 30-59 year.
Study                                                          logOR       logCI1       logCI2      OR      CI1    CI2
Hardell et al. (2011)165, ipsilateral, >1640 hr                0.779       0.086        1.470       2.18    1.09   4.35
INTERPHONE (2010)93, ipsilateral, >1640 hr                     0.673       0.199        1.151       1.96    1.22   3.16
Fixed-effects model
Study                                     Effect    [95%Conf.Interval]           %Weight         OR       CI1     CI2
Hardell et al. (2011)165                  0.779     0.087         1.471          32.11
INTERPHONE (2010)93                       0.673     0.197         1.149          67.89
Overall effect                            0.707     0.315         1.099          100.00          2.03     1.37    3.00
Random-effects model
Study                                     Effect    [95%Conf.Interval]           %Weight         OR       CI1     CI2
Hardell et al. (2011)165                  0.779     0.087         1.471          32.11
INTERPHONE (2010)93                       0.673     0.197         1.149          67.89
Overall effect                            0.707     0.315         1.099          100.00          2.03     1.37    3.00
Heterogeneity
                   Value        df     p-value
Cochrane Q         0.06         1      0.804
Table I11 Glioma, cumulative call time, contralateral, Hardell 30-59 year.
Study                                                            logOR       logCI1       logCI2     OR      CI1    CI2
Hardell et al. (2011)\165, contralateral, >1640 hr               0.392       -0.562       1.353      1.48    0.57   3.87
INTERPHONE (2010)93, contralateral, >1640 hr                     0.223       -0.446       0.884      1.25    0.64   2.42
Fixed-effects model
Study                                     Effect    [95%Conf.Interval]           %Weight         OR       CI1     CI2
Hardell et al. (2011)165                  0.392     -0.566        1.350          32.53
INTERPHONE (2010)93                       0.223     -0.442        0.888          67.47
Overall effect                            0.278     -0.268        0.824          100.00          1.32     0.76    2.28
Random-effects model
Study                                   Effect     [95%Conf.Interval]           %Weight          OR       CI1     CI2
Hardell et al. (2011)165                0.392      -0.566        1.350          32.53
Interphone (2010)93                     0.223      -0.442        0.888          67.47
Overall effect                          0.278      -0.268        0.824          100.00           1.32     0.76    2.28
Heterogeneity
                   value       df     p-value
Cochrane Q         0.08        1      0.776
               Meta-analysis and forest plots                                                                           251
</pre>

====================================================================== Einde pagina 252 =================================================================

<br><br>====================================================================== Pagina 253 ======================================================================

<pre>Table I12 Acoustic neuroma, time since first use ≥10 years.
Study                                                    logOR       logCI2      logCI1      OR        CI1     CI2
Schüz et al. (2011)51                                    -0.1        0.38        -0.7        0.87      0.52    1.46
Hansson Mild et al. (2007)65, analogue                   1.1         1.74        0.5         3.10      1.70    5.70
Hansson Mild et al. (2007)65, digital                    -0.5        1.61        -2.3        0.60      0.10    5.00
Hansson Mild et al. (2007)65, cordless                   0.0         1.06        -1.2        1.00      0.30    2.90
INTERPHONE (2011)94                                      -0.3        0.10        -0.7        0.76      0.52    1.11
Sato et al. (2010)112                                    0.5         1.56        -0.2        1.62      0.79    4.77
Fixed-effects model
Study                                            Effect     [95% Conf. Interval]    % Weight      OR      CI1   CI2
Schüz et al. (2011)\51                           -0.139     -0.655       0.377      23.88
Hansson Mild et al. (2007)65, analogue           1.131      0.526        1.736      17.39
Hansson Mild et al. (2007)65, digital            -0.511     -2.467       1.445      1.66
Hansson Mild et al. (2007)65, cordless           0.000      -1.134       1.134      4.94
INTERPHONE (2011)94                              -0.274     -0.654       0.105      44.26
Sato et al. (2010)112                            0.482      -0.417       1.381      7.87
Overall effect (fe)                              0.071      -0.181       0.324      100.00        1.07    0.83  1.38
Random-effects model
Study                                            Effect     [95% Conf. Interval]    % Weight      OR      CI1    CI2
Schüz et al. (2011)51                            -0.139     -0.655       0.377      21.84
Hansson Mild et al. (2007)65, analogue           1.131      0.526        1.736      20.32
Hansson Mild et al. (2007)65, digital            -0.511     -2.467       1.445      5.93
Hansson Mild et al. (2007)65, cordless           0.000      -1.134       1.134      12.34
INTERPHONE (2011)94                              -0.274     -0.654       0.105      24.07
Sato et al. (2010)112                            0.482      -0.417       1.381      15.50
Overall effect (dl)                              0.178      -0.360       0.716      100.00        1.19    0.70   2.05
Heterogeneity
                    value    df    p-value
Cochrane Q          16.79    5     0.005
252            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 253 =================================================================

<br><br>====================================================================== Pagina 254 ======================================================================

<pre>Table I13 Acoustic neuroma, cumulative call time >1000/1640 h.
Study                                                          logOR      logCI2      logCI1   OR       CI1     CI2
Hardell et al. (2006)64, analogue, >1000 hr                    1.6        2.6         0.64     5.10     1.90    14.00
Hardell et al. (2006)64, digital, >1000 hr                     1.1        1.9         0.41     3.10     1.50    6.40
INTERPHONE (2011)94 >1640 hr                                   0.3        0.7         -0.13    1.32     0.88    1.97
Fixed-effects model
Study                                          Effect     [95% Conf. Interval]       % Weight      OR      CI1     CI2
Hardell et al. (2006)64, analogue              1.629      0.631         2.628        11.07
Hardell et al. (2006)64, digital               1.131      0.406         1.857        20.97
INTERPHONE (2011)94                            0.278      -0.125        0.681        67.97
Overall effect                                 0.606      0.274         0.938        100.00        1.83    1.32    2.55
Random-effects model
Study                                          Effect     [95% Conf. Interval]       % Weight      OR      CI1     CI2
Hardell et al. (2006)64, analogue              1.629      0.631         2.628        26.81
Hardell et al. (2006)64, digital               1.131      0.406         1.857        33.01
INTERPHONE (2011)94                            0.278      -0.125        0.681        40.18
Overall effect                                 0.922      0.102         1.742        100.00        2.51    1.11    5.71
Heterogeneity
                   value       df    p-value
Cochrane Q         8.60        2     0.014
Tables I14 Acoustic neuroma, time since first use ≥10 years, ipsilateral.
Study                                                    logOR        logCI2       logCI1     OR        CI1      CI2
Hardell et al. (2009)66, ipsilateral                     1.1          1.8          0.34       3.00      1.40     6.20
INTERPHONE (2011)94, ipsilateral                         0.2          0.7          -0.37      1.18      0.69     2.04
Fixed-effects model
Study                                   Effect     [95% Conf. Interval]          % Weight       OR       CI1      CI2
Hardell et al. (2009)66                 1.099      0.355           1.843         34.67
INTERPHONE (2011)94                     0.166      -0.376          0.708         65.33
Overall effect                          0.489      0.051           0.927         100.00         1.63     1.05     2.53
Random-effects model
Study                                   Effect     [95% Conf. Interval]          % Weight       OR       CI1     CI2
Hardell et al. (2009)66                 1.099      0.355           1.843         46.12
INTERPHONE (2011)94                     0.166      -0.376          0.708         53.88
Overall effect                          0.596      -0.316          1.507         100.00         1.81     0.73    4.51
Heterogeneity
                   value       df    p-value
Cochrane Q         3.95        1     0.047
               Meta-analysis and forest plots                                                                          253
</pre>

====================================================================== Einde pagina 254 =================================================================

<br><br>====================================================================== Pagina 255 ======================================================================

<pre>Tables I15 Acoustic neuroma, time since first use ≥10 years, contralateral.
Study                                                   logOR         logCI2    logCI1 OR    CI1  CI2
Hardell et al. (2009)66, contralateral                  0.88          1.8       -0.1   2.40  0.90 6.30
INTERPHONE (2011)94, contralateral                      -0.37         0.4       -1.1   0.69  0.33 1.42
Fixed-effects model
Study                                  Effect    [95% Conf. Interval]        % Weight   OR   CI1  CI2
Hardell et al. (2009)66                0.875     -0.097          1.848       36.00
INTERPHONE (2011)94                    -0.371    -1.101          0.359       64.00
Overall effect                         0.078     -0.506          0.661       100.00     1.08 0.60 1.94
Random-effects model
Study                                  Effect    [95% Conf. Interval]        % Weight   OR   CI1  CI2
Hardell et al. (2009)66                0.875     -0.097          1.848       46.53
INTERPHONE (2011)94                    -0.371    -1.101          0.359       53.47
Overall effect                         0.209     -1.010          1.428       100.00     1.23 0.36 4.17
Heterogeneity
                   value      df    p-value
Cochrane Q         4.04       1     0.045
254            Mobile phones and cancer
</pre>

====================================================================== Einde pagina 255 =================================================================

<br><br>====================================================================== Pagina 256 ======================================================================

<pre>Health Council of the Netherlands
Advisory Reports
The Health Council’s task is to       In addition, the Health Council
advise ministers and parliament on    issues unsolicited advice that
issues in the field of public health. has an ‘alerting’ function. In some
Most of the advisory opinions that    cases, such an alerting report
the Council produces every year       leads to a minister requesting
are prepared at the request of one    further advice on the subject.
of the ministers.
Areas of activity
Optimum healthcare                    Prevention                          Healthy nutrition
What is the optimum                   Which forms of                      Which foods promote
result of cure and care               prevention can help                 good health and
in view of the risks and              realise significant                 which carry certain
opportunities?                        health benefits?                    health risks?
Environmental health                  Healthy working                     Innovation and
Which environmental                   conditions                          the knowledge
influences could have                 How can employees                   infrastructure
a positive or negative                be protected against                Before we can harvest
effect on health?                     working conditions                  knowledge in the
                                      that could harm their               field of healthcare,
                                      health?                             we first need to
                                                                          ensure that the right
                                                                          seeds are sown.
www.healthcouncil.nl
</pre>

====================================================================== Einde pagina 256 =================================================================

<br><br>