Cellular phone use and brain tumor: a meta-analysis.

Abstract

The dramatic increase in the use of cellular phones has generated concerns about potential adverse effects, especially the development of brain tumors. We conducted a meta-analysis to examine the effect of cellular phone use on the risk of brain tumor development.
We searched the literature using MEDLINE to locate case-control studies on cellular phone use and brain tumors. Odds ratios (ORs) for overall effect and stratified ORs associated with specific brain tumors, long-term use, and analog/digital phones were calculated for each study using its original data. A pooled estimator of each OR was then calculated using a random-effects model.
Nine case-control studies containing 5,259 cases of primary brain tumors and 12,074 controls were included. All studies reported ORs according to brain tumor subtypes, and five provided ORs on patients with > or =10 years of follow up. Pooled analysis showed an overall OR of 0.90 (95% confidence interval [CI] 0.81-0.99) for cellular phone use and brain tumor development. The pooled OR for long-term users of > or =10 years (5 studies) was 1.25 (95% CI 1.01-1.54). No increased risk was observed in analog or digital cellular phone users.
We found no overall increased risk of brain tumors among cellular phone users. The potential elevated risk of brain tumors after long-term cellular phone use awaits confirmation by future studies.

Full text

Page 1

CLINICAL STUDY - PATIENT STUDIES
Cellular phone use and brain tumor: a meta-analysis
Peter Kan Æ Æ Sara E. Simonsen Æ Æ Joseph L. Lyon Æ Æ
John R. W. Kestle
Received: 3 May 2007/Accepted: 7 June 2007
? Springer Science+Business Media, B.V. 2007
Abstract
Background
phones has generated concerns about potential adverse
effects, especially the development of brain tumors. We
conducted a meta-analysis to examine the effect of cellular
phone use on the risk of brain tumor development.
Methods
We searched the literature using MEDLINE to
locate case-control studies on cellular phone use and brain
tumors. Odds ratios (ORs) for overall effect and stratified
ORs associated with specific brain tumors, long-term use,
and analog/digital phones were calculated for each study
using its original data. A pooled estimator of each OR was
then calculated using a random-effects model.
Results
Nine case-control studies containing 5,259 cases
of primary brain tumors and 12,074 controls were included.
All studies reported ORs according to brain tumor sub-
types, and five provided ORs on patients with ‡10 years of
follow up. Pooled analysis showed an overall OR of 0.90
(95% confidence interval [CI] 0.81–0.99) for cellular phone
use and brain tumor development. The pooled OR for long-
term users of ‡10 years (5 studies) was 1.25 (95% CI 1.01–
1.54). No increased risk was observed in analog or digital
cellular phone users.
Conclusions
We found no overall increased risk of brain
tumors among cellular phone users. The potential elevated
The dramatic increase in the use of cellular
risk of brain tumors after long-term cellular phone use
awaits confirmation by future studies.
Keywords
Brain tumors ? Cellular phones ? Radiation
Introduction
With the widespread use of cellular phones in the past
decade, human exposure to low-energy radiation in the
800- to 2,000-MHz range (microwave) has increased
dramatically. With more than a billion current cellular
phone users worldwide [1], this exposure could pose a
serious public health problem even if the radiation emit-
ted has only a small oncogenic effect. The risk of
developing intracranial tumors from cellular phone use is
of particular interest because of the proximity of expo-
sure. Although it is agreed that any carcinogenic effect
would have to be through a nonthermal, nonionizing
mechanism, the nature or the existence of this mechanism
remains unclear [2–5]. To date, most epidemiological
studies published on cellular phone use and brain tumors
have not demonstrated an increased risk with overall use
[6–13], but positive associations have been reported in a
few small subgroup analyses, such as with long-term
users (increased risk of acoustic neuroma), analog phones,
and ipsilateral use [14, 15]. The purpose of the study
was to provide a pooled estimate on cellular phone use
and the risk of brain tumor development using a meta-
analysis. In addition, by pooling available studies to-
gether, we hope to explore further how potentially
important clinical variables (e.g., duration of use and
phone type) can affect the risk of brain tumor develop-
ment in cellular phone users.
P. Kan ? J. R. W. Kestle (&)
Division of Pediatric Neurosurgery,
Department of Neurosurgery, University of Utah,
100 N. Medical Drive, Salt Lake City, UT 84113-1100, USA
e-mail: john.kestle@hsc.utah.edu
S. E. Simonsen ? J. L. Lyon
Department of Family and Preventive Medicine,
University of Utah, Salt Lake City, Utah, USA
123
J Neurooncol
DOI 10.1007/s11060-007-9432-1

Page 2

Methods
Literature review
A MEDLINE search was performed of literature dating from
1966 through April 2006 using the key words ‘‘cellular
phones,’’ ‘‘brain tumors,’’ ‘‘cancer,’’ ‘‘radiation,’’ and ‘‘micro-
wave’’ to identify all relevant articles on cellular phone use
and brain tumors. For each selected article, the bibliography
was searched to locate additional eligible publications.
Study selection and data abstraction
To be included in our review a study had to meet all of the
following criteria: 1) it was published in English (no case-
control studies were identified outside of the English liter-
ature); 2) it was case control in design (in the absence of
randomized-controlled trials and prospective cohort studies
for rare diseases like primary brain tumor, case-control
studies provide the next highest level of evidence); 3) it
presented sufficient data so that the crude odds ratios (ORs)
for cellular phone use and brain tumors could be derived;
and 4) the exposure to cellular phones was clearly defined
and evaluated to minimize misclassification. Published case
reports and studies with exposure other than cellular phones
(e.g., cordless phones) were excluded from the study.
In addition to the original data, other data abstracted
from the studies included the time period of study, country
of origin, study setting (population versus hospital),
information source for exposure measurement, adjusted
estimates of effect, and potential confounding factors ad-
justed for in their estimates.
Statistical analysis
For each study, summary results were tabulated for anal-
ysis. Crude ORs for the risk of brain cancer associated with
exposure to cellular phones and their standard errors were
calculated based on the abstracted data. When possible,
separate crude ORs were calculated for high-grade
gliomas, low-grade gliomas, meningioma, and acoustic
neuroma. In addition, when data were available, the results
were stratified by length of exposure (‡10 years or
<10 years), and by type of cellular phone (analog or digi-
tal). Pooled estimates of the crude ORs for all analyses
were obtained by weighting each study with the standard
error of the natural log of the OR. For our meta-analysis,
the random-effects method of DerSimonian and Laird was
used, with the estimate of heterogeneity taken from the
Mantel–Haenszel model. Statistical heterogeneity was as-
sessed using the chi-squared test; P < 0.10 was used as the
criterion to represent statistical heterogeneity. The random-
effects model incorporated both within- and between-study
variability. Because heterogeneity was present in some of
the comparisons (composite, high-grade glioma, analog
versus unexposed, and digital versus unexposed, Table 1),
it was more appropriate to use random-effects than fixed-
effects meta-analysis. In addition, the random-effects
method was selected because the results would be more
conservative.
Begg’s adjusted rank correlation test and Egger’s
regression asymmetry test were used to assess publication
bias. A P value of 0.05 or less was considered significant.
All analyses were performed using Stata 8.0 (release 8,
Stata Corporation, College Station, TX).
Results
We reviewed 48 articles and abstracts from our MEDLINE
search and identified 10 case-control studies published
between 2000 and 2006 that evaluated the association
between cellular phone use and brain tumors (Table 2).
One study was excluded from the meta-analysis because
there were insufficient data to calculate the crude OR [14].
Other articles excluded from the original search included
review articles, case reports, animal studies, or articles that
were not relevant to the research question (e.g., studies that
considered other forms of electromagnetic exposure or
studies that had tumors other than primary brain tumors as
their event of interest). In a further effort to be inclusive,
we also searched the bibliographies of the articles identi-
fied and found no additional studies.
The 9 studies included in the meta-analysis contained a
total of 5,259 cases of primary brain tumors and 12,074
controls. All studies reported ORs according to brain tumor
subtypes, and five provided ORs on patients with 10 years
or more of follow up. Five studies reported ORs according
to the type of phone used (analog versus digital). All
studies were performed in North America and Europe, and
6 of them [7, 8, 10, 12, 13, 15] were part of the Interphone
Table 1 Tests for heterogeneity between studies
Chi-squared test—P value
Composite0.052*
0.023*
High-grade glioma
Low-grade glioma0.178
Meningoma0.840
Acoustic neuroma0.813
Composite–10 years exposure0.833
0.001*
0.000*
Analog versus unexposed
Digital versus unexposed
Analog versus digital 0.452
*Statistically significant heterogeneity present
J Neurooncol
123

Page 3

Table 2 Description of studies included in meta-analysisa
Study
Study description
# cases, #
controls
Definition of exposure
Pooled OR
(95% CI)
Adjusted OR from
manuscript (95% CI)
Adjustments
Auvinen et al.
2002 [6]
Cases of brain tumors and salivary
gland cancers diagnosed in
Finland in 1996 from the
population-based Finnish
Cancer Registry. Five randomly
sampled, population-based
controls for each tumor case
matched on age and sex were
selected
796 cases, 3,972
controls
Regular cellular telephone use
(defined as having a
subscription to a cellular
telephone service)
1.27 (0.94–
1.72)
1.3 (0.9–1.8)
Adjusted for age and sex.
Occupations, socioeconomic
status, and places of residence
were similar for controls and
cases
Christensen et al.
2005 [7]b
Cases of glioma and meningioma
diagnosed in Denmark between
September 1, 2000 and August
31, 2002. Randomly sampled, population-based controls
matched on age and sex
427 cases, 822
controls
Regular cellular telephone use
(defined as at least once a weekfor 6 months or more) compared
with no regular use
0.73 (0.59–
0.90)
Not reported
Adjusted for sex, age, educational
level, use of hands-free devicesin cars, marital status, and
region of Denmark
Hepworth et al.
2006 [8]b
Cases of glioma (age 18–69)
identified between 2000 and
2004 from hospital records and
cancer registries in 5 areas of the
United Kingdom. Controls were
randomly selected from
practitioner’s lists and
frequency matched by age, sex,
and region
966 cases, 1,761
controls
Regular cellular telephone use
(defined as use for >6 monthsfor at least 1 year prior to
diagnosis)
1.01 (0.87–
1.19)
0.94 (0.78–1.13)
Adjusted for age, sex, region,
Townsend deprivation category,
and interview reference datecategory
Inskip et al. 2001
[9]
Cases of glioma, meningioma, and
acoustic neuroma diagnosed in
hospitals in 3 regions of the
United States. Hospital-based
controls with nonmalignantconditions frequency matched
by hospital, age, sex, race/
ethnicity, and proximity of
residence to the hospital
782 cases, 799
controls
Regular cellular telephone use
(defined as two or more calls per
week) compared to no use
0.80 (0.66–
0.98)
0.9 (0.7–1.1)
Adjusted for matching variables
(age, sex, race/ethnicity, and
distance from patient’s
residence) as well as education,
income, and date of interview
Lonn et al. 2004
[15]b
Cases of acoustic neuroma
diagnosed in regions of Swedenbetween 1999 and 2002.
Controls randomly selected
from population, frequency
matched on age, sex, and
residential area
148 cases, 604
controls
Regular cellular telephone use
(defined as at least once per
week during 6 months or more)
1.05 (0.73–
1.52)
1.0 (0.6–1.5)
Adjusted for age, sex, residential
area, and education
J Neurooncol
123

Page 4

Table 2 continued
Study
Study description
# cases, #
controls
Definition of exposure
Pooled OR
(95% CI)
Adjusted OR from
manuscript (95% CI)
Adjustments
Lonn et al. 2005
[10]b
Cases of glioma and meningioma
diagnosed in Sweden during2000–2002. Controls randomly
selected from population,
frequency matched on age, sex,
and residential area
644 cases, 674
controls
Regular cellular telephone use
(defined as at least once per
week during 6 months or more)
0.83 (0.69–
0.99)
Not reported
Adjusted for age, gender,
geographic region, and
education
Muscat et al. 2000
[11]
Cases with brain cancer selected
from 5 US academic medical
centers between 1994 and 1998.
Hospital-based controls with
benign conditions frequency
matched by age, race, and month of admission
469 cases, 422
controls
Regular cellular telephone use
(defined as having/having had a
subscription to a cellular
telephone service)
0.75 (0.52–
1.07)
0.8 (0.6–1.2)
Adjusted for age, education, sex,
race, study center, proxy
subject, month of interview, and
year of interview
Schoemaker et al.
2005 [12]b
Cases of acoustic neuroma were
identified by medical centers in
4 Nordic countries and the
United Kingdom between
September 1, 2000 and August
31, 2002. Controls were
randomly selected from the
population register and matchedon age and residence
678 cases, 3,553
controls
Regular cellular telephone use
(defined as use for >6 monthsfor at least 1 year prior to
diagnosis)
0.95 (0.81–
1.12)
0.9 (0.7–1.1)
Adjusted for center, region, age,
sex, education, interview year,
and interview lag time
Schuz et al. 2006
[13]b
Cases were diagnosed with glioma
or meningioma during 2000– 2003 in 3 regions of Germany.
Controls were randomly drawn
from population registries and
were matched on age, gender,
and region
747 cases, 1,494
controls
Regular cellular telephone use
(defined as at least one incoming
or outgoing call per week for6 months or more)
0.91 (0.75–
1.09)
Not reported
Adjusted for gender, study center,
age, socioeconomic status, and
living in a city
aAll studies included in the meta-analysis are case-control studies
bStudy is part of the Interphone Study
J Neurooncol
123

Page 5

Study, an international collaborative case-control study of
the relationship between mobile phone use and primary
brain tumors. All six studies shared the same methods to
avoid discrepancies between designs. Among the nine
studies, seven were population based [6–8, 10, 12, 13, 15]
and eight [7–13, 15] obtained exposure information from
direct interview of patients or their proxies. There was no
evidence ofstatistically significant
(Table 3).
publicationbias
Composite outcome
Associations between cellular phone use and brain tumor
development were evaluated in all 9 studies. Pooling of
data from all these studies resulted in an OR of 0.90 (95%
[confidence interval] CI 0.81 to 0.99) (Fig. 1; Table 4)
using a random-effects model. On the basis of the pooled
estimate, there was no increased risk of brain tumors (all
subtypes considered) associated with cellular phone use.
Brain tumor subtype
All studies evaluated the association between cellular
phone use and brain tumor subtypes. The pooled OR (and
95% CI) for high-grade glioma [7, 8, 10, 11, 13] was 0.86
(0.70 to 1.05), for low-grade glioma [7, 8, 10, 11, 13] was
1.14 (0.91 to 1.43), for meningioma [6, 7, 9, 10, 13] was
0.64 (0.56 to 0.74), and for acoustic neuroma [9, 12, 15]
was 0.96 (0.83 to 1.10) (Fig. 2A–D; Table 4) using ran-
dom-effects models. On the basis of the pooled estimates,
there appears to be no increased risk of any brain tumor
subtypes associated with cellular phone use.
Composite outcome with exposure stratified by length
of cellular phone use
Five studies [7, 8, 10, 12, 15] reported the association
between cellular phone use of 10 years or more and brain
tumor development. The pooled OR was calculated to be
1.25 (95% CI 1.01 to 1.54) (Fig. 3; Table 4) using a ran-
dom-effects model.
Analog and digital cellular phones
Five studies [6, 8, 10, 12, 15] evaluated the association
between the type of cellular phone used and brain tumor
development. The pooled OR (and its 95% CI) for analog
phone use compared with the unexposed group was 1.13
(0.83 to 1.54), whereas that for digital phone use compared
with the unexposed group was 0.86 (0.68 to 1.09) (Fig. 4A,
B; Table 4). When analog phone users were compared with
digital phone users, the pooled OR was 1.22 (95% CI 1.06
to 1.41).
Table 3 Tests for publication bias
Egger’s test—P
value
Begg’s test—P
value
Composite0.914 0.602
High-grade glioma0.4130.806
Low-grade glioma 0.1260.086
Meningoma0.6660.806
Acoustic neuroma0.9451.000
Composite—10 years
exposure
0.662 0.806
Analog verus unexposed0.546 0.462
Digital versus unexposed 0.5460.462
Analog versus digital0.2230.462
Fig. 1 Pooled odds ratio of brain tumor development associated with
cellular phone use
Table 4 Summary of pooled estimates
OutcomePooled
OR
95% CI
All brain tumors (with regular use as
exposure)
0.900.81–0.99
Low-grade glioma (with regular use as
exposure)
1.140.91–1.43
High-grade glioma (with regular use as
exposure)
0.860.70–1.05
Meningoma (with regular use as
exposure)
0.640.56–0.74
Acoustic neuroma (with regular use as
exposure)
0.960.83–1.10
All brain tumors (with 10+ years use as
exposure)
1.251.01–1.54
All brain tumors (with digital use as
exposure)
0.860.68–1.09
All brain tumors (with analog use as
exposure)
1.130.83–1.54
All brain tumors (comparing analog
with digital)
1.221.06–1.41
J Neurooncol
123

Page 6

Discussion
We found no increased risk of brain tumors associated with
cellular phone use when combining the findings from the 9
case-control studies. This finding is consistent with the
literature; most of the studies included or reviewed did not
report an overall OR that was significantly greater than
unity.
When the risk of tumor development was stratified
according to brain tumor subtype, we found no increased
risk for any subtype of brain tumor among cellular phone
users. Because exposure to low-energy radiation from
cellular phones is highest at the surface meningeal tissue
and the vestibulocochlear nerve closest to the handset, we
would predict an increased risk for meningioma and/or
acoustic neuroma (which was not observed in our meta-
analysis) if an oncogenic effect were present.
When we combined studies that involved patients with
10 or more years of exposure, we found a slightly increased
risk of tumor development among long-term users (OR
1.25, 95% CI 1.01 to 1.54). Ten years was chosen arbi-
trarily as the metric of exposure as some studies reported
an increased risk after 10 years [10]. Of course, the true
length of time for electromagnetic radiation to produce an
oncogenic effect (latency) remains unclear as this analysis
says nothing about latency but only about one particular
metric of exposure (10 years).
In the association between brain tumors and long-term
use of over 10 years, the type of phone used could be an
important confounder because analog phone use is asso-
ciated with both long-term use and an increased risk of
tumor development when compared with digital phone use
(OR 1.22; CI: 1.06–1.41). In fact, the similarity between
the two ORs (1.25 and 1.22) suggests that a large propor-
tion of the association observed between brain tumors and
long-term users may be explained by the confounding
relationship between duration of use and phone type (long-
term users were also analog phone users). As a result, the
Fig. 2 Pooled odds ratio of high-grade glioma (A), low-grade glioma (B), meningioma (C), and acoustic neuroma (D) associated with cellular
phone use
Fig. 3 Pooled odds ratio of brain tumor development associated with
cellular phone use of 10 or more years
J Neurooncol
123

Page 7

true risk of long-term cellular phone use remains incon-
clusive without the ability to control for confounding and
would need to be validated by future studies.
In the past decade, there has been a substantial change
in wireless technology. The earliest cell phones relied on
analog technology. These have been replaced by phones
using digital technology. Because analog cellular phones
operate at a higher power, giving off more electromag-
netic radiation, there has been concern that they may carry
an increased risk of brain cancer. Based on our pooled
estimates, although the risk of brain tumor development
may be slightly higher with analog phone use than with
digital phone use (OR 1.22, 95% CI 1.06 to 1.41), no
increased risk of brain tumors was found in either analog
or digital phone use when compared with the unexposed
group. However, in any study, one must interpret results
stratified by phone type with caution because almost all
analog users have now switched to digital phones and
therefore have been exposed to power settings of both
designs.
Selection bias, information bias, and confounding
interactions were potential limitations of our study. The
method we used to identify the articles included in the
study was comprehensive, which minimized selection bias
for articles with either a positive or a negative result.
Information bias, obtaining information from study sub-
jects that will misclassify exposure in a nonrandom man-
ner (often because of differential recall and intensity of
surveillance between cases and controls), unfortunately
cannot be controlled in a meta-analysis because it was
inherent with each of the studies individually. Finally,
without all of the original data, our study was also unable
to control for confounding factors. However, as is evi-
dent from Table 2, the crude and adjusted OR for each
study were very comparable, suggesting that confounding
did not represent a significant threat to the validity of our
results.
In summary, we found no overall increased risk of brain
tumors among cellular phone users. The potential elevated
risk of brain tumors after 10 years or more of cellular
phone use should be confirmed by additional data from
future studies.
Acknowledgements
her editorial assistance in preparing this manuscript.
The authors thank Kristin Kraus, M.Sc., for
References
1. National Radiological Protection Board (2004) Mobile phones
and health 2004. Report of the Board of NRPB. NRPB,
Oxon
2. Anderson V (2003) Comparisons of peak SAR levels in con-
centric sphere head models of children and adults for irradiation
by a dipole at 900 MHz. Phys Med Biol 48:3263–3275
3. Hermann DM, Hossmann KA (1997) Neurological effects of
microwave exposure related to mobile communication. J Neurol
Sci 152:1–14
4. Moulder JE, Erdreich LS, Malyapa RS et al (1999) Cell phones
and cancer: what is the evidence for a connection? Radiat Res
151:513–531
5. Valberg PA (1997) Radio frequency radiation (RFR): the nature
of exposure and carcinogenic potential. Cancer Causes Control
8:323–332
6. Auvinen A, Hietanen M, Luukkonen R et al (2002) Brain tumors
and salivary gland cancers among cellular telephone users. Epi-
demiology 13:356–359
7. Christensen HC, Schuz J, Kosteljanetz M et al (2005) Cellular
telephones and risk for brain tumors: a population-based, incident
case-control study. Neurology 64:1189–1195
8. Hepworth SJ, Schoemaker MJ, Muir KR et al (2006) Mobile
phone use and risk of glioma in adults: case-control study. BMJ
332:883–887
9. Inskip PD, Tarone RE, Hatch EE et al (2001) Cellular-telephone
use and brain tumors. N Engl J Med 344:79–86
10. Lonn S, Ahlbom A, Hall P et al (2005) Long-term mobile phone
use and brain tumor risk. Am J Epidemiol 161:526–535
11. Muscat JE, Malkin MG, Thompson S et al (2000) Handheld
cellular telephone use and risk of brain cancer. JAMA 284:3001–
3007
12. Schoemaker MJ, Swerdlow AJ, Ahlbom A et al (2005) Mobile
phone use and risk of acoustic neuroma: results of the Interphone
Fig. 4 (A) Pooled odds ratio of brain tumor development associated
with analog phone use compared with individuals not exposed to
analog phones; (B) pooled odds ratio of brain tumor development
associated with digital phone use compared with individuals not
exposed to digital phones
J Neurooncol
123

Page 8

case-control study in five North European countries. Br J Cancer
93:842–848
13. Schuz J, Bohler E, Berg G et al (2006) Cellular phones, cordless
phones, and the risks of glioma and meningioma (Interphone
Study Group, Germany). Am J Epidemiol 163:512–520
14. Hardell L, Mild KH, Carlberg M (2002) Case-control study on
the use of cellular and cordless phones and the risk for malignant
brain tumours. Int J Radiat Biol 78:931–936
15. Lonn S, Ahlbom A, Hall P et al (2004) Mobile phone use and the
risk of acoustic neuroma. Epidemiology 15:653–659
J Neurooncol
123