There is a concern that radiofrequency fields emitted from mobile phones may affect cancer risk. Exposure declines rapidly with distance to the source, and therefore, there is particular interest in acoustic neuroma (also known as vestibular schwannoma)—a benign nerve-sheet tumor of the vestibulocochlear nerve. Most epidemiologic studies on mobile phone use and acoustic neuroma have found overall risk estimates below or close to unity,1 with a few exceptions.2,3 The Swedish INTERPHONE study4 found an association (odds ratio [OR] = 1.9; 95% confidence interval [CI] = 0.9 to 4.1) between long-term mobile phone use (10 or more years) and acoustic neuroma—a result that was not confirmed in the larger pooled analyses of INTERPHONE studies.5,6 For slow-growing tumors such as acoustic neuroma, the evidence concerning long induction times is still insufficient.1
Energy absorption in the head from a mobile phone is highly localized. Higher risk estimates for mobile phone use on the same side as the tumor would therefore provide stronger support for a causal association. Studies on glioma often found increased risks on the side where the mobile phone was usually held, but these were accompanied with risk reductions on the other side.1 Corresponding results for acoustic neuroma have been mixed. Acoustic neuroma is a slow-growing tumor that could be present many years before diagnosis.7 Early symptoms are hearing loss and tinnitus. These symptoms may have led patients with acoustic neuroma to change the side where they hold their mobile phone long before the tumor was discovered, or to choose the unaffected side if the tumor was already present at the time mobile phone use began. To date, studies available have assessed only the side of mobile phone use at one point in time, usually at the time of data collection or diagnosis.
For the purpose of studying the association between long-term mobile phone use and acoustic neuroma, we conducted a case-control study covering a time period when hand-held mobile phones had been available for up to 20 years. In addition, we collected detailed information about the history of laterality of phone use to allow evaluation of potential reverse causality in laterality analyses.
We conducted a population-based, case-control study of acoustic neuroma in Sweden between 1 September 2002 and 31 August 2007. The material and data collection are described in detail in the study by Palmisano et al.8 In brief, incident cases of acoustic neuroma (International Classification of Diseases, ICD-10 C72.4 or D33.3 and ICD-O-2 9560.0) between 20 and 69 years of age at diagnosis were identified nationwide in collaboration with treating clinics. In addition, we searched the Swedish Regional Cancer Registers, which provide the basis for the nation-wide register.9 We also used local acoustic neuroma registries at the otorhinolaryngology clinics in the Uppsala and Linköping regions, which are more complete than the cancer registries because clinics sometimes wait for histological confirmation before reporting cases. The delay between radiologic diagnosis and a subsequent diagnostic surgical procedure can be substantial. Patients who undergo stereotactic radiation and patients with small nongrowing tumors that do not require treatment may not ever be entered into the cancer registers.
Information about tumor laterality, date of diagnosis, radiologic exams, and histologic confirmation was obtained from medical records. Date of diagnosis (reference date) was defined as the date of the first medical examination resulting in an acoustic neuroma diagnosis.
Two controls per case were randomly selected from the nation-wide Swedish population register, matched on age (5-year strata), sex, and residential region (six geographic regions). Controls were assigned a reference date and a fictive “tumor” laterality corresponding to their matched case.
Cases and controls diagnosed with neurofibromatosis (nine cases and two controls) or tuberous sclerosis (one case) were excluded. These are rare genetic conditions associated with nervous system tumors. Acoustic neuroma is associated primarily with neurofibromatosis type 2.
Information about exposure and potential confounding factors was collected through mailed questionnaires that sent simultaneously to cases and their matched controls starting from October 2007.
Participants were asked whether they had ever been regular mobile phone users, defined as having made or received a call on average at least once per week during at least 6 months. Regular users were asked the first year of regular use, whether they had stopped using mobile phones, and (if so) what year they stopped. Information about average number of calls, calling time, and hands-free use was collected in predefined categories every 3 years starting from 1987 (when hand-held mobile phones were introduced) (eAppendix, http://links.lww.com/EDE/A761, for exact phrasing of questions).
Cumulative call time and number of calls were calculated using the mid-point of the intervals, except for >60 minutes/day and >10 calls per day, for which the lowest value was used. This decision was on the basis of analyses of mobile phone-operator data available for a subset of participants (n = 394), among whom only one person had used a mobile phone on average >60 minutes/day, and 2% had made or received >10 calls per day. Use of hands-free devices lowers exposure to the head to <10%, and exposure from Bluetooth hands-free equipment is 100 times lower than from mobile phones.10 Therefore, for reporting “Always or almost always using hands-free device,” “more than half of the time,” “about half of the time,” and “less than half of the time” exposure was reduced by 100%, 75%, 50%, and 25%, respectively, for the specific time period. Exposure within 1 year of the reference date was not considered.
Participants were asked on which side of the head they usually hold the phone, and whether they had ever changed sides. If they had changed sides, they were asked the year and reason for the change, and how they had held the phone before the change. Exposure on the same side as the tumor (ipsilateral exposure) was defined as having held the mobile phone on the tumor side or on both sides during any period before the reference date. Contralateral exposure was defined analogously (holding the phone on the side opposite the tumor, or on both sides). Acoustic neuroma often causes hearing loss and tinnitus on the affected side long before diagnosis.7 Therefore, when defining laterality of phone use in the main laterality-specific analyses, we disregarded hearing-dependent side changes within 5 years before the reference date. Because previous studies determined laterality of phone use at only one point in time, we also performed laterality-specific analyses using information about preferred side of the head at specific time points, reflecting various ways of asking about laterality of phone use; at the time of data collection, at reference date, and at 1, 5, and 10 years before the reference date.
Information about use of cordless phones (ie, land-line phones with cordless hand-sets connected by radio link to a base station) was collected in the same way as for mobile phones, except that questions about laterality of cordless phone use and details about proportion of time using hands-free devices were not included.
Nonresponders who declined participation during telephone reminders were asked short questions about mobile phone use (ever use, regular use, and start of regular use) and education.
Associations between indicators of mobile or cordless phone use and acoustic neuroma were calculated as ORs with 95% CIs using conditional logistic regression. We adjusted for possible confounding from smoking, education, marital status, and parity, each of which has been reported to be associated with acoustic neuroma risk,8,11–14 and in the cordless phone analyses also for hands-free use. Adjustment did not have any noticeable effect on the risk estimates, and therefore, we present unadjusted ORs.
Cumulative call time and cumulative number of calls were categorized with cut points approximately at the quartiles of the exposure distribution among controls, decided a priori. We made a post hoc decision to analyze cumulative call time ≥1,640 hours, based on findings in the combined INTERPHONE analyses.6,15
Separate analyses were conducted for use of analog phones and digital phones (eAppendix, http://links.lww.com/EDE/A761, provides exact phrasing of questions about network), and for participants unable to report mobile phone type. We also performed analyses stratified on age, sex, and histological confirmation of cases.
In total, 542 eligible acoustic neuroma cases and 1095 controls were identified, of whom 451 cases (83%) and 710 controls (65%) participated. The most common reasons for nonparticipation were subject refusal (cases, 9%; controls, 31%) and physician’s refusal or illness/disability (cases, 6%; controls, 2%). Four cases and six controls had died before contact. Because of the matched design, analyses include only case-control sets for which the case and at least one control contributed exposure information, leaving 422 cases and 643 controls for analyses of mobile phone use and 417 cases and 635 controls for analyses of cordless phone use. Table 1 shows demographic characteristics of cases and controls. Of the cases, 47% were histologically confirmed.
The proportions of regular mobile phone users among cases and controls were similar (71% and 70%, respectively) although slightly lower (69%) among controls in complete matched strata. The prevalence of regular mobile phone use among controls increased over the study period, from 62% in 2002 to 81% in 2007.
The relative risk estimate for regular mobile phone use was close to 1.0 in the overall analysis (OR = 1.18 [95% CI = 0.88 to 1.59]) and also when restricting to histologically confirmed cases (0.99 [0.65 to 1.52]) (Table 2). Risk estimates did not increase with time since first regular use; the highest OR was found for intermediate-term exposure (5 to 9 years), whereas the OR with long-term mobile phone use (≥10 years) was 1.11 (0.76 to 1.61) overall, and 0.94 (0.55 to 1.62) for histologically confirmed cases. Results for at least 13 years of use were essentially the same as for ≥10 years. An increased risk estimate was observed in the highest quartile (≥680 hours) of cumulative calling time: OR = 1.46 (0.98 to 2.17) overall, and 1.14 (0.63 to 2.07) for histologically confirmed cases. The OR for reporting ≥1,640 hours was 1.51 (0.92 to 2.49) based on 45 exposed cases and 55 exposed controls, and 1.49 (0.71 to 3.10; 21 exposed cases, 22 exposed controls) for histologically confirmed cases (not shown). eTable 1 (http://links.lww.com/EDE/A761) shows stratified analyses and P values for heterogeneity of effects. None of the differences between histologically confirmed and nonconfirmed cases were greater than would be expected by chance (eg, P values were 0.27 and 0.40 for regular use of mobile and cordless phones, respectively).
Regular use of analog mobile phones was associated with an increased OR, but with wide CIs, and risk estimates decreased with time since first use (Table 2). For digital-phone use, the OR estimate was slightly higher than the overall OR, whereas “unknown network” was associated with a reduced OR, indicating that controls were less able to specify which network they had used.
ORs for ipsilateral use of mobile phones were generally lower than ORs for contralateral use (Table 3). Among cases who were regular users, 52% reported that they had changed their preferred side of mobile phone use. The proportion among controls was 8%. For cases, the most common reason for changing side of phone use was hearing loss (91%) and for controls “practical reasons” (40%) (eTable 2, http://links.lww.com/EDE/A761, shows reasons for side changes). Most changes took place within the 2 years before the reference date or after that date, but some changes were reported >10 years before reference date. Side changes were more common among cases regardless of time period (eTable 3, http://links.lww.com/EDE/A761, which shows distribution of time since side change). Laterality-specific analyses using information about the preferred side of the head at various time points show that ORs for ipsilateral use were reduced closer in time to diagnosis, and approached unity at 5 and 10 years before diagnosis, whereas ORs for contralateral use were increased closer to the reference date, became lower with increasing time before the reference date, but were above unity also for laterality information referring to 10 years before the reference date (Table 4).
Regular use of cordless phones was associated with an OR of 1.41 (95% CI = 1.07 to 1.86) overall, and 1.24 (0.83 to 1.86) for histologically confirmed cases. ORs did not increase with increasing time since first use; the lowest OR was found for ≥10 years since first use. The OR increased with cumulative number of hours of cordless phone use; 1.67 (1.13 to 2.49) overall, and 1.52 (0.87 to 2.66) for histologically confirmed cases in the highest quartile (≥900 hours). No similar pattern was found for cumulative number of calls (Table 5). Adjustment for hands-free use (32 cases and 38 controls) had no notable effect on the estimates.
There were no material differences in results between men and women or between age groups for any of the evaluated exposure indices (not shown).
The nonresponder questionnaire was answered by 93 (24%) of the nonparticipating controls and seven (8%) of the nonparticipating cases. Among them, 62 controls (67%) and four cases (57%) reported regular mobile phone use 1 year before the reference date.
This study, which includes the largest number of mobile phone users with >10 years of phone use reported to date, did not find results in support of the hypothesis that long-term mobile phone use increases the risk of acoustic neuroma. We observed ORs close to 1.0 for regular mobile phone use, for the longest induction times (≥10 years) and for ipsilateral mobile phone use. Risk estimates were generally higher for contralateral use than for ipsilateral use, and also higher when taking into consideration changes of the preferred side of phone use possibly caused by prodromal symptoms such as hearing loss. An increased risk in the highest quartile of cumulative calling time was found, but no corresponding increase in the highest category of cumulative number of calls. Increased risks were essentially confined to cases without histological confirmation of the tumor. Analyses of the complete history of laterality of mobile phone use revealed considerable potential reverse causality in laterality-specific estimates. For cordless land-line phones, increased ORs were found for regular use and in the highest categories of cumulative calling time, but not in the category with long-term use (≥10 years)—and again, mainly for cases without histological confirmation.
This is the first study to assess laterality of mobile phone use at various times before diagnosis, and also possible changes of laterality. As unilateral hearing loss is a common early symptom of acoustic neuroma, cases may have had to change their preferred side of mobile phone use long before the tumor was diagnosed. Previous studies have asked only for the preferred side of mobile phone use at one point in time, either at the time of the interview or “before diagnosis.” Cases might have been influenced by the knowledge about their tumor laterality when retrospectively reporting about preferred side of mobile phone use. The pattern of the results in previous studies5,6 indicates recall bias in reported side of use. Our results demonstrate that information about the side of mobile phone use at a single point in time is not sufficient to cover the etiologically relevant time period.
A large proportion of the cases reported changes of their preferred side, as did a small proportion of the controls. We found strongly reduced risk estimates for ipsilateral use, and increased risk estimates for contralateral use, when laterality was reported for the time of the data collection, at the time of diagnosis, or at 1 year before diagnosis. However, for laterality information referring to 5 and 10 years before diagnosis, the risk estimate for ipsilateral use was approximately 1.0 and for contralateral use slightly, >1.0. The laterality analysis is not compatible with an exposure-dependent causal relation between radiofrequency fields and the risk of acoustic neuroma.
We carried out sensitivity analyses restricting the material to histologically confirmed cases to reduce misclassification of the disease, which could have diluted risk estimates. We found that increased ORs were confined mainly to cases without histological confirmation although differences were small and may have been because of random variation. Another potential explanation for the unexpected pattern is that phone use may increase the probability of tumor detection by drawing attention to the unilateral hearing loss that is commonly caused by the disease. Such an effect would bias the risk estimates upward and mainly be present for the smallest and least symptomatic tumors, whereas people with more severe disease would have had their tumor detected anyway, and detection bias would be less likely. Patients whose tumors are surgically removed and therefore can be histologically confirmed generally have larger tumors than those treated with expectancy or radiotherapy.
It may be of interest to compare our findings with results for glioma, a severe and usually rapidly growing tumor, for which an almost total rate of detection is expected. The accumulating evidence on mobile phone use and glioma speaks against an association, including results from a recent prospective cohort study from the United Kingdom.16–18 After updating the follow-up to 2011, the United Kingdom cohort study found a slightly increased risk estimate for acoustic neuroma in the intermediate category of time since first use, but no increased risk among long-term users.
In the INTERPHONE study, selection bias caused by differential participation between cases and controls was estimated to have led to a 10% underestimation of risk.19 In our study as well, participation was higher among cases (83%) than controls (65%). Results of the nonresponder survey indicate that the difference in regular use between participating and nonparticipating controls was small. This suggests that any effect of nonparticipation on our risk estimates is likely to be small although caution is warranted when interpreting the nonresponder results, considering the small proportion of nonresponders reached.
We used mailed questionnaires for data collection, whereas all previous studies (except the three studies by Hardell et al2,3,20) used personal interviews. A mailed questionnaire allows greater differences in behavior between cases and controls when reporting about past exposures. Cases are likely to spend a longer time filling out the questionnaire and to a greater extent consult various sources of information (eg, old mobile phone bills). The median time used by cases to fill out the questionnaire was 50% longer than for controls. A mailed questionnaire immediately discloses all included questions to the participants, which may affect the answers, especially to questions where affirmative answers lead to extensive follow-up questions. However, a mailed questionnaire avoids potential interviewer effects. “Regular mobile phone use” was clearly defined in the questionnaire to avoid differences in interpretation between cases and controls. The prevalence of regular mobile phone use is largely as expected when compared with the 59% reported in the Swedish INTERPHONE study,4 ending data collection August 2002.
Validation studies have shown that self-reported mobile phone history is subject to recall errors, and that calling time tends to be more difficult to remember than number of calls.21,22 Nondifferential recall errors could potentially dilute the risk estimates and distort dose–response relations. Risk estimates in our study were somewhat higher during the first half of the study period compared with the risk estimates during the last half; thus, it seems unlikely that a more prevalent misclassification in the earlier period has diluted a true association (not shown).
Differential exposure misclassification is a potential source of bias when using self-reported exposure information collected retrospectively, especially when there is a debate and concern in society that draws attention to a possible association. In a validation study conducted within the INTERPHONE study, cases tended to overestimate the calling time more for more distant time periods—a pattern that was not seen for number of calls in either cases or controls.23 This indicates that results, especially for cumulative hours of use, may be affected by recall bias. No validation studies of recall of time since first use or of cordless phone use have been published.
Type of Network
Average output power is higher from analog than from digital phones.10 We observed increased ORs for analog phone use but a reverse dose–response pattern for time since first use. Moreover, the risk estimate for regular use of digital phones was also higher than the overall analysis, whereas the risk estimate was reduced for those who were unable to report type of network. This suggests that estimates for analog and digital phones have been biased because of more frequent missing data about network type among controls.
Cordless phone use seemed to be associated with an increased risk of acoustic neuroma. Risk estimates were, however, increased in all exposure categories, with the lowest risk estimate for the longest duration of use—in contrary to what would have been expected if cordless phone use was causally related to acoustic neuroma. In addition, the average output power emitted from cordless phones (10 mW) is considerably lower than that emitted from both analogue (900 mW) and digital (GSM900 240 mW, GSM1800 125 mW) mobile phones.10 The third generation of mobile phones (Universal Mobile Telecommunications System) became available only toward the end of the study period (ie, after any etiologically relevant induction and latency periods). Therefore, it seems unlikely that the reported exposure reflects a causal relation to risk of neuroma formation. Acoustic neuroma risk increased with increasing cumulative calling time with a cordless phone, with no consistent dose–response pattern for cumulative number of calls. Associations were much weaker in analyses restricted to cases with histologic confirmation, also without any consistent dose–response pattern.
Comparability with Previous Studies
Our results are broadly consistent with most previous studies of mobile phones and acoustic neuroma.1,6,16,17,24 Most epidemiologic studies have shown no increases in the overall risk estimates of acoustic neuroma risk and mobile phone use, with the exception of the studies by Hardell et al.2,3,25 The Hardell studies included only histologically confirmed cases reported to the regional cancer registers—the subgroup for which our study found no associations. The Swedish national INTERPHONE study4 found a slightly increased risk among participants who started to use a mobile phone at least 10 years before diagnosis (OR = 1.9 [95% CI = 0.9 to 4.1], 14 exposed cases). In the international INTERPHONE analysis, the corresponding result was OR = 0.76 (0.52 to 1.11), based on 68 exposed cases, whereas our result was OR = 1.11 (0.76 to 1.61), based on 103 exposed cases. The INTERPHONE study found an increased risk in the 10th decile of cumulative hours of use (≥1,640 hours), but with no dose–response pattern in the first nine exposure categories, and with the lowest risk estimate in the ninth decile. An increased risk for ≥1,640 hours of use was also observed in our study. In the INTERPHONE study, implausible numbers of hours of phone use were more commonly reported among cases than among controls, and recall bias seemed likely to be one explanation for the increased risk in the most extreme category of cumulative hours of use. In our study, we collected hours of use in categories, and can therefore not identify implausible hours of use. For cordless phone use, few studies are available on acoustic neuroma. The Swedish INTERPHONE study found no indication of increased risk related to cordless phone use,4 whereas Hardell et al25 reported a 50% increased risk after <5 years of cordless phone use. The German INTERPHONE study reported results related to cordless phone use only for glioma and meningioma, with no increased risks observed.26
This study, having more long-term mobile phone users (≥10 years) than all previous studies together, provides little support for the hypothesis that long-term mobile phone use increases the risk of acoustic neuroma. The study is the first to analyze a complete history of laterality of mobile phone use and shows that laterality analyses are prone to bias. The results suggest that detection bias may be present in studies of a slow-growing tumor such as acoustic neuroma. Results for cordless phone use are compatible with an increased risk although this finding is difficult to interpret as radiofrequency exposure is considerably lower than from mobile phones. An alternative explanation is recall bias when reporting historical cordless phone use although a causal association cannot be ruled out. Prospectively collected exposure information would be required to draw more firm conclusions.
We thank the Regional Cancer Registers for their collaboration.
1. Ahlbom A, Feychting M, Green A, Kheifets L, Savitz DA, Swerdlow AJICNIRP (International Commission for Non-Ionizing Radiation Protection) Standing Committee on Epidemiology. . Epidemiologic evidence on mobile phones and tumor risk: a review. Epidemiology. 2009;20:639–652
2. Hardell L, Carlberg M, 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;25:120–128
3. Hardell L, Hallquist A, Mild KH, Carlberg M, Påhlson A, Lilja A. Cellular and cordless telephones and the risk for brain tumours. Eur J Cancer Prev. 2002;11:377–386
4. Lönn S, Ahlbom A, Hall P, Feychting M. Mobile phone use and the risk of acoustic neuroma. Epidemiology. 2004;15:653–659
5. 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;93:842–848
6. Interphone Study Group. . Acoustic neuroma risk in relation to mobile telephone use: results of the INTERPHONE international case-control study. Cancer Epidemiol. 2011;35:453–464
7. Thomsen J, Tos M. Acoustic neuroma: clinical aspects, audiovestibular assessment, diagnostic delay, and growth rate. Am J Otol. 1990;11:12–19
8. Palmisano S, Schwartzbaum J, Prochazka M, et al. Role of tobacco use in the etiology of acoustic neuroma. Am J Epidemiol. 2012;175:1243–1251
9. Barlow L, Westergren K, Holmberg L, Talbäck M. The completeness of the Swedish Cancer Register: a sample survey for year 1998. Acta Oncol. 2009;48:27–33
11. Inskip PD, Tarone RE, Hatch EE, et al. Sociodemographic indicators and risk of brain tumours. Int J Epidemiol. 2003;32:225–233
12. Schüz J, Steding-Jessen M, Hansen S, Stangerup SE, Cayé-Thomasen P, Johansen C. Sociodemographic factors and vestibular schwannoma: a Danish nationwide cohort study. Neuro Oncol. 2010;12:1291–1299
13. Benson VS, Green J, Pirie K, Beral V. Cigarette smoking and risk of acoustic neuromas and pituitary tumours in the Million Women Study. Br J Cancer. 2010;102:1654–1656
14. 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. 2007;120:103–110
15. Interphone Study Group. . Brain tumour risk in relation to mobile telephone use: results of the INTERPHONE international case-control study. Int J Epidemiol. 2010;39:675–694
16. Benson VS, Pirie K, Schüz J, Reeves GK, Beral V, Green JMillion Women Study Collaborators. . Mobile phone use and risk of brain neoplasms and other cancers: prospective study. Int J Epidemiol. 2013;42:792–802
17. Benson VS, Pirie K, Schuz J, Reeves GK, Beral V, Green J. Authors' response to: The case of acoustic neuroma: comment on mobile phone use and risk of brain neoplasms and other cancers [Letter]. Int J Epidemiol. [Epub ahead of print], doi: 10.1093/ije/dyt186.
18. Swerdlow AJ, Feychting M, Green AC, Leeka Kheifets LK, Savitz DAInternational Commission for Non-Ionizing Radiation Protection Standing Committee on Epidemiology. . Mobile phones, brain tumors, and the interphone study: where are we now? Environ Health Perspect. 2011;119:1534–1538
19. Vrijheid M, Deltour I, Krewski D, Sanchez M, Cardis E. 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;16:371–384
20. Hardell L, Näsman A, Påhlson A, Hallquist A, Hansson Mild K. Use of cellular telephones and the risk for brain tumours: a case-control study. Int J Oncol. 1999;15:113–116
21. Inyang I, Benke G, Morrissey J, McKenzie R, Abramson M. How well do adolescents recall use of mobile telephones? Results of a validation study. BMC Med Res Methodol. 2009;9:36
22. Vrijheid M, Cardis E, Armstrong BK, et al.Interphone Study Group. Validation of short term recall of mobile phone use for the Interphone study. Occup Environ Med. 2006;63:237–243
23. Vrijheid M, Armstrong BK, Bédard D, et al. Recall bias in the assessment of exposure to mobile phones. J Expo Sci Environ Epidemiol. 2009;19:369–381
24. 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;174:416–422
25. Hardell L, Carlberg M, 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;28:509–518
26. Schüz J, Böhler E, Berg G, et al. Cellular phones, cordless phones, and the risks of glioma and meningioma (Interphone Study Group, Germany). Am J Epidemiol. 2006;163:512–520
Supplemental Digital Content
© 2014 by Lippincott Williams & Wilkins, Inc