Share this article on:

Mobile Phone Use and Incidence of Glioma in the Nordic Countries 1979–2008: Consistency Check

Deltour, Isabellea,b; Auvinen, Anssic,d; Feychting, Mariae; Johansen, Christofferf; Klaeboe, Larsg,h; Sankila, Ristoi; Schüz, Joachima

doi: 10.1097/EDE.0b013e3182448295

Background: Some case-control studies have reported increased risks of glioma associated with mobile phone use. If true, this would ultimately affect the time trends for incidence rates (IRs). Correspondingly, lack of change in IRs would exclude certain magnitudes of risk. We investigated glioma IR trends in the Nordic countries, and compared the observed with expected incidence rates under various risk scenarios.

Methods: We analyzed annual age-standardized incidence rates in men and women aged 20 to 79 years during 1979–2008 using joinpoint regression (35,250 glioma cases). Probabilities of detecting various levels of relative risk were computed using simulations.

Results: For the period 1979 through 2008, the annual percent change in incidence rates was 0.4% (95% confidence interval = 0.1% to 0.6%) among men and 0.3% (0.1% to 0.5%) among women. Incidence rates have decreased in young men (20–39 years) since 1987, remained stable in middle-aged men (40–59 years) throughout the 30-year study period, and increased slightly in older men (60–79 years). In simulations, assumed relative risks for all users of 2.0 for an induction time of up to 15 years, 1.5 for up to 10 years, and 1.2 for up to 5 years were incompatible with observed incidence time trends. For heavy users of mobile phones, risks of 2.0 for up to 5 years' induction were also incompatible.

Conclusion: No clear trend change in glioma incidence rates was observed. Several of the risk increases seen in case-control studies appear to be incompatible with the observed lack of incidence rate increase in middle-aged men. This suggests longer induction periods than currently investigated, lower risks than reported from some case-control studies, or the absence of any association.

Supplemental Digital Content is available in the text.

From the aSection of Environment and Radiation, International Agency for Research on Cancer, Lyon, France; bStatistics, Bioinformatics and Registry Unit, Danish Cancer Society Research Center, Copenhagen, Denmark; cDepartment of Epidemiology, Tampere School of Public Health, University of Tampere, Tampere, Finland; dResearch and Environmental Surveillance, STUK - Radiation and Nuclear Safety Authority, Helsinki, Finland; eUnit of Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden; fSurvivorship Unit, Danish Cancer Society Research Center, Copenhagen, Denmark; gDepartment of Radiation Protection and Nuclear Safety, Norwegian Radiation Protection Authority, Østerås, Norway; and hThe Cancer Registry of Norway, Oslo, Norway; and iInstitute for Statistical and Epidemiological Cancer Research, Finnish Cancer Registry, Helsinki, Finland.

Submitted 21 April 2011; accepted 20 October 2011; posted online 16 January 2012.

Supported by the Danish Strategic Research Council (grant number 2064-04-0010).

The simulations used data collected for the Interphone study in Denmark, Finland, Norway, and Sweden. The Interphone study was supported by funding from the European Fifth Framework Program, ‘Quality of Life and Management of Living Resources’ (contract QLK4-CT-1999901563) and the International Union against Cancer (UICC). The UICC received funds for this purpose from the Mobile Manufacturers' Forum and GSM Association. Provision of funds to the Interphone study investigators via the UICC was governed by agreements that guaranteed Interphone's complete scientific independence. The terms of these agreements are publicly available at

The authors reported no other financial interests related to this research.

Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article ( This content is not peer-reviewed or copyedited; it is the sole responsibility of the author.

Correspondence: Isabelle Deltour, Section of Environment and Radiation, International Agency of Research on Cancer, 150 Cours Albert Thomas, 69008 Lyon, France. E-mail:

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. The populations of the Nordic countries were among the first to adopt mobile phones extensively, and since 2005 have had more than one subscription per inhabitant.1 3 In these countries, high-quality nationwide population-based cancer registries have been maintained for the past half century,4 with mandatory and effective recording of all incident tumors including benign brain tumors. Hence, these registries provide excellent opportunities for surveillance of the occurrence of brain tumors. Glioma constitutes the most likely tumor type to show a risk increase associated with mobile phone use, according to some reviews5 7 and the monograph meeting of the International Agency for Research on Cancer.8

Up to 2003, time trends in the incidence of glioma among adults in the Nordic countries did not indicate a detectable increase paralleling the increasing prevalence of mobile phone use.9 Time trends in malignant brain tumors were similarly flat in the United States up to 200610 and 2007,11 and in the United Kingdom up to 2007.12 Before these surveillance studies had been reported, the association between mobile phone use and brain tumor was investigated in cohort13,14 and case-control studies.15 21 One Swedish case-control study reported markedly elevated risks of malignant brain tumors (odds ratios [ORs] = 2–4) associated with overall use of mobile phones,20 and increased risks were also found in the analyses22 24 pooling this dataset with a previous study of the same group; however, most other studies showed no consistent excess risks.16 18,25 The largest study was the international Interphone case-control study, in which the odds ratio for glioma was decreased for overall use but elevated for heavy users of mobile phones (the 10% of regular users who had talked on their phone,25 a lifetime total of ≥1640 hours).

Mathematically, the OR reported in case-control studies of the association between mobile phone use and cancer should correspond to the ratio of the incidence rate in the subgroups of the population using mobile phones to the incidence rate in the subgroups not using mobile phones, assuming a true effect and no change in other factors affecting the disease. Indeed, an elevated OR for a specific exposure group in a case-control study should correspond to an increase in incidence rates of similar magnitude in the population with similar exposure.

It has been argued that the proportion of the population using mobile phones in the Nordic countries in 2003 was still too small to detect a postulated effect of mobile phone use at population level,26 although a RR of about 2.0 after 10 years of use would have already increased the IR by about 20% between 1999 and 2003 among the 40–59-year-old men.27 It remains, however, relevant to evaluate how large risks and how short induction periods yield detectable increases in incidence rates; or conversely, if no increases in secular trends are detectable, what levels of risk and induction periods can be excluded. We conducted a simulation study to address this.

To update the previous publication9 with data from 2004 to 2008, we analyzed time trends in the incidence rates of glioma among adults aged 20 to 79 years of the Nordic countries from 1979 to 2008. In addition, assuming that the use of mobile phones caused glioma, we simulated a population similar to 40–59-year-old Nordic men, and computed the probability of detecting the postulated association. Our aim was to define which levels of risks and induction periods would be compatible with the observed time trends in this population, and which would not be compatible.

Back to Top | Article Outline


Study Population

We obtained the numbers of primary gliomas (International Classification of Disease for Oncology version 3, topographical codes C71 and morphologic codes between 938 and 948, and similar codes in the International Classification of Disease version 7 for the early period) in patients aged 20 to 79 years at diagnosis during 1979–2008, from the national cancer registries of Denmark, Finland, Norway, and Sweden. Sizes of the populations at risk by 5-year age groups were acquired from the national population registers for each calendar year.

Back to Top | Article Outline

Statistical Analysis of Observed Incidence Rates

Annual age-standardized incidence rates of gliomas per 100,000 person-years were calculated separately for men and women, standardized to the European standard population. The combined age range (20–79 years) and the 20-year age groups 20–39, 40–59, and 60–79 years were considered; country-specific and combined data were analyzed.

A piecewise log-linear model called joinpoint analysis (Joinpoint Regression Program, version 3.4.3 – April 2010; Statistical Research and Applications Branch, National Cancer Institute) without constraints on the positions of the nodes or joinpoints was used to identify trend changes and to estimate annual percent change in incidence rates over the period 1979–2008. The model was specified to include a maximum of 3 joinpoints, which could occur in the middle of a year or between 2 consecutive years. The model constrained the joinpoints to be at least a year and a half apart, and at least 2 years from the start and end of the study period. Under the assumption of heteroscedasticity and uncorrelated errors, the best-fitting model was searched on 4499 randomly permuted datasets using the grid search method. Tests, at an overall two-sided significance level of 0.05, were not adjusted for autocorrelation.28

Back to Top | Article Outline

Simulation Study

We generated datasets under the hypothesis that there is a risk related to mobile phone use, assuming various risk scenarios (eg, a 2-fold increased risk 10 years after first use of mobile phones). In these simulated datasets, the association between mobile phones and brain tumors shows up as increased number of cases with a random distribution around the expected value of excess cases. Next, we estimated the risk parameter and its 95% confidence interval (CI) from the dataset and the model, pretending the true value was unknown. The risk estimated for each dataset was distributed around the value used in the scenario, and its 95% CI indicated whether the increase was statistically significant. We considered the proportion of datasets with statistically significant increases as the measure of our ability to detect an increased risk caused by mobile phones in the observed incidence rates, for each simulated scenario.

Back to Top | Article Outline

Prevalence of Use of Mobile Phones Used in Simulation Study

Information on use of mobile phones was obtained from self-reports in a sample of the general population of the Nordic countries interviewed for the Interphone study (ie, controls).21 We abstracted the proportion of persons who used mobile phones regularly—defined as using a mobile phone on average at least once a week over a period of at least 6 months–and the proportion of heavy users, at least 1640 hours of lifetime cumulative call time; these prevalences were computed for each year from 1980 to 2002 by sex and by age group, with lags of 1, 5, 10, and 15 years accounting for aging during the lag period. We focused our simulation study on men aged 40 to 59 years who had the highest use of mobile phones between 1980 and 1993.27 Data collection for the Interphone study stopped in 2002–2003. For the period 2003–2008, we extrapolated use in 2002 as rising 3% annually, a slightly slower increase than the average observed over years 1990–2000. To account for possible selection bias (ie, lower participation in mobile phone studies among controls not using phones29), lower prevalences were used in a sensitivity analysis under conservative assumptions.

Back to Top | Article Outline

Generation of Datasets

We sampled the number of cases as a Poisson distribution in a population of the same size as the population of men aged 40 to 59 years of the Nordic countries (from 2,461,722 persons in 1979 to 3,439,372 persons in 2008) with the same prevalence of use, with baseline incidence rate for nonexposed subjects of 8.88 cases per 100,000 person-years at risk, and with the assumption that the use of mobile phones increased the risk of glioma after an apparent induction period during which no increase occurred. Each scenario combined a value of RR (0.8, 1.1, 1.2, 1.5, and 2.0) that multiplied the baseline incidence rate for all mobile phone users or for heavy users only, with an induction period (1, 5, 10, and 15 years).

The following equation describes our model:

where i denotes the calendar year (1979–2008); J, the presumed induction period (1, 5, 10, or 15 years); O, the observed number of cases; b, the baseline incidence rate; RR, the relative risk; E, the number of persons exposed (all users or heavy users); and N, the number of persons unexposed (the remainder of the population). We generated 10,000 datasets per scenario.

Back to Top | Article Outline

Estimation of Outcome

For each simulated dataset, we estimated the relative risk (RR) under the Poisson model (1), with known exposed and unexposed populations (E and N) and unknown RR, and baseline incidence rate (b), treated as a nuisance parameter. Because the likelihood of model (1) was not standard, we obtained the estimates of the parameters with a maximum likelihood procedure (ml function of Stata; StataCorp. 2007. Stata Statistical Software: release 10. College Station, TX: StataCorp LP). Of the 10,000 estimates, the proportion of lower bounds of the 95% CI above 1.0 was a bootstrap estimate of the probability of obtaining a significantly increased RR, conditional on the scenario. For comparison purposes, we fitted the same model (1) on the observed number of cases among men aged 40–59 years.

Back to Top | Article Outline


Glioma Incidence Trends

This study was based on 35,250 glioma cases diagnosed from 1979 to 2008; the annual average number of glioma was 1175 in a population of 17 million adults aged 20–79 years. Sweden accounted for approximately 40% of the population and of the cases; the remaining countries (Denmark, Finland, and Norway) each contributed roughly 20% of the population and cases. The age-standardized incidence rates were higher in men (8.6 per 100,000 person-years) than women (6.0 per 100,000 person-years), and increased with age (Fig.). All countries had similar rates (eTable 1,



Time trends did not strongly increase at any point during the period 1979 to 2008 in any country (data not shown), among men, women, or any age-group (Table 1, Figure). Incidence rates were generally stable over the whole period, and increased gradually among older persons. A slight decrease in incidence rates was observed after the late 1980s among the younger men overall (annual percent change = –0.7% [95% CI = –1.4% to 0.1%]) and in Denmark and Sweden, but not in Finland and Norway (data not shown).

Table 1

Table 1

Back to Top | Article Outline

Prevalence of Regular Use

The self-reported prevalence of regular mobile phone use among men 40–59 years of age ranged from 2% in 1980 (which was the earliest year on which data were collected in the Interphone questionnaire) to 79% in 2002; we extrapolated this to 98% in 2008. The prevalence of heavy users was 0.2% in 1988 (the first year above 0%) to 16% in 2002 and extrapolated to 33% in 2008.

These exposure prevalences, combined with the induction periods of 1, 5, 10, and 15 years, did not result in statistically significant increased risks when model (1) was fitted on the observed number of glioma cases. The RRs were below 1.09 for all induction periods, for both all users and heavy users, with the 95% CIs consistently including 1.0 (data not shown).

Back to Top | Article Outline

Simulation Study

Scenarios with shorter induction periods combined with higher RRs resulted in higher probabilities of detecting significant increases in estimated RRs (Table 2). For example, for the scenario in which all users of mobile phones were assumed to have 50% increased RR of glioma (RR = 1.5), 15 years after using their mobile phone for the first time (Table 2, row 2, column 6), 84% of the simulated datasets demonstrated a statistically significantly increased estimated relative risk. The remaining 16% had either nonsignificant or no increased risk for this scenario. Thus, if there was a true underlying risk increase 15 years after first mobile phone use, it would most likely be seen in the incidence rates (84%). When there was a doubling in risk among heavy users only, all datasets (100%) demonstrated a statistically significant effect of mobile phones with induction period of up to 5 years; this proportion decreased to 69% if induction was 10 years, and to 7% if induction time was 15 years (Table 2, row 5).

Table 2

Table 2

In scenarios with exposure defined as regular mobile phone use, statistically significantly increased estimated relative risks were obtained in all datasets for a risk of 2.0 and an induction period of maximum 15 years, for a risk of 1.5 and an induction period of maximum 10 years, and for a risk of 1.2 and an induction period of maximum 5 years. In scenarios where only heavy users were assumed to be at risk, statistically significantly increased risks were obtained in all or almost all the datasets for a risk of 2.0 and an induction period of maximum 5 years. All other combinations could be masked by the random fluctuations inherent in the occurrence of a rare disease in the population of middle-aged Nordic men. In particular, for a postulated risk of 1.5 among heavy users and an induction period of 1 year, statistically significantly increased risks were obtained in 98% of the datasets; thus, the probability of a true causal relative risk of 1.5 for the heaviest users of mobile phones, given the observed absence of an increase in incidence time trends, was estimated to be approximately 2%. When a true relative risk of 0.8 was modeled on all users, all datasets demonstrated significantly decreased risks for induction periods of up to 5 years. Essentially similar results were obtained in the sensitivity analyses under conservative assumptions on levels of use (eTable 2,

Back to Top | Article Outline


We detected no upward turn in the time trends of glioma incidence rates in the Nordic countries during 1979–2008, overall or in any subgroup by country, age, or sex among adults. Our results extend those of a previous study,9 of time trends in the Nordic countries up to 2003 by adding 5 more calendar years of follow-up. These analyses are based on the entire adult population of Denmark, Finland, Norway, and Sweden (17 million people) and are strengthened by the comprehensive high-quality cancer registration in these countries.30

If mobile phone use causes brain tumors, the change in prevalence of use from zero to nearly 100% over a 20-year period would eventually influence the incidence rates of these tumors. Conversely, a lack of change in the incidence time trends, at any point in time, would constitute evidence against this association. Given that the incidence time trends were stable up to 2008, several risk scenarios could be discarded with confidence, based on our simulations. The risk (should one exist) would have to be lower, or occur after a longer induction period, or act on a smaller population, than the scenarios that we discarded.

Several case-control and cohort studies have evaluated the association between mobile phone use and glioma risk in adults, covering cases ascertained up to 2004. The results of our simulation study are compatible with studies showing no increased risks of gliomas.16,17,31 In the Interphone study,25 the OR for glioma related to mobile phone use overall was statistically significantly decreased among regular users (OR = 0.8 [95% CI = 0.7 to 0.9]). This OR is also incompatible with the observed time trends and therefore, appears due to biases and errors.29,32 34 An elevated risk (1.4 [1.0 to 1.9]) among the 10% heaviest regular users was also reported, although bias was discussed as an alternative explanation. We evaluated the probability of a true OR of this magnitude (1.5) as approximately 2%, given the observed incidence trend. Therefore either the observed time trends reflect the lack of a moderate effect on the occurrence of glioma among heavy users, or these results are within the 2% margin of error that was the estimated probability for failing to detect such an effect in the current setting.

The association between mobile phone use and malignant brain tumors in adults has also been evaluated in 3 Swedish case-control studies.15,19,20 The results of the first 2 studies were compatible with our simulations, as no statistically significant increased risks were reported for overall use of mobile phones15,19 (OR of malignant tumors for use of analog phone = 1.1 [95% CI = 0.8 to 1.6]; OR for use of digital phones = 1.1 [0.9 to 1.5]). The third study, with case ascertainment in 2000–2003, yielded markedly elevated risks: OR for all analog phone use = 2.6 (1.5 to 4.3) and OR for all digital phone use = 1.9 (1.3 to 2.7), with elevated risks for all latency periods, but with even higher OR for a latency of >10 years.20 We found that such highly increased risks are strongly inconsistent with observed time trends in risk in the Nordic countries up to 2008. This may indicate the presence of biases and errors in self-reported mobile phone use, resulting in unrealistically high ORs. In the Interphone study, analyses restricted to the subpopulation who used mobile phones regularly were performed as a way to correct for participation bias. In these, the risk of glioma was increased for both short- and long-term use (OR for use 2–4 years = 1.7 [1.2 to 2.4]; OR for use 5–9 years = 1.5 [1.1 to 2.2]; OR for use 10 or more years = 2.2 [1.4 to 3.3]).25 Our simulation study indicated that these results were also unrealistically high. Several meta-analyses have been published,35 37 but such pooling of evidence does not remove systematic biases. Incidence rates of brain tumors in Swedish children and adolescents (5–19 years old) have recently been published for the period 1990 to 2008.38 Children and adolescents adopted mobile phone use later than adults, and so follow-up may be too short to observe an effect specific to this age group.

Our simulation study is based on certain assumptions. The induction period relating mobile phone use and glioma risk, if such an association exists, is unknown, as is the magnitude of risk, and the real patterns may be more complex than the scenarios we simulated. In addition, there are several factors that we did not account for. The coverage of the Nordic cancer registries is not complete; 1.5% to 6% of the malignant tumors are missed among men aged 40 to 59 years,39 41 and although cancer registration is thought to have improved, we have no information on possible changes in completeness over time. We assumed that other risk factors of the disease, as well as its detection, had no major impact over the period 1979–2008 in this age group, which is consistent with the observed stable rates. Some increase was present in the older age group, as reported in other studies,10 which could reflect improved imaging techniques and increased diagnostic activity. Our main results are based on the group 40–59 years of age and therefore should not have been affected by this phenomenon. We used self-reported use of mobile phones to evaluate the prevalence of use and heavy use up to 2002, and extrapolated the prevalence for the period 2003–2008 based on documented overall trends. The extrapolation over time affected only the results for the induction period of 1 year. The use of hands-free devices was not accounted for, but this was not frequent in these populations (data not shown). In sensitivity analyses, we investigated conservative assumptions on prevalence of use, with similar results. Indeed, Interphone controls were likely to overestimate prevalence and amount of mobile phone use in the entire population due to selection and information bias.29,32 34

Because we focused our simulation study on men aged 40–59 years, we did not account for the other age and sex subgroups that, independently, also did not show any change in their incidence time trends.

Our data indicate that, so far, no risk associated with mobile phone use has manifested in adult glioma incidence trends, although the induction period, if any, is unknown. Case-control studies conducted in 2000–2003 were not able to evaluate risks for exposures longer than 10–12 years. Our simulations show that many increased or decreased risks reported in case-control studies are implausible, implying that biases and errors in the self-reported use of mobile phone have likely distorted the findings.

Back to Top | Article Outline


1. Official Statistics of Finland (OSF). Telecommunications [e-publication]. 2006, Table 6. Number of mobile phone subscriptions and number of subscriptions per 100 population in 1980, 1985 and 1990–2006. Available at: Accessed March 1, 2011.
2. IT- og Telestyrelsen, Post- og Teletilsynet, Finnish Communications Regulatory Authority, Post & Telestyrelsen, The Post and Telecom Administration in Iceland. Telecommunication Markets in the Nordic countries. Available at: Accessed March 1, 2011.
3. IT- og Telestyrelsen, Post- og Teletilsynet, Post og Fjarskipastofnun, Finnish Communications Regulatory Authority, Post & Telestyrelsen. Competition and regulation in the Nordic mobile markets. Available at: Accessed March 1, 2011.
4. Engholm G, Ferlay J, Christensen N, et al.. NORDCAN: Cancer Incidence, Mortality, Prevalence and Survival in the Nordic Countries, Version 4.0. Available at: Accessed March 1, 2011.
5. Ahlbom A, Feychting M, Green A, Kheifets L, Savitz DA, Swerdlow AJ. Epidemiologic evidence on mobile phones and tumor risk: a review. Epidemiology. 2009;20:639–652.
6. Hardell LO, Carlberg M, Soderqvist F, Mild KH, Morgan LL. Long-term use of cellular phones and brain tumours - increased risk associated with use for > or = 10 years. Occup Environ Med. 2007;64:626–632.
7. Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR). Health Effects of Exposure to EMF. Available at: Accessed March 1, 2011.
8. Baan R, Grosse Y, Lauby-Secretan B, et al.. Carcinogenicity of radiofrequency electromagnetic fields. Lancet Oncol. 2011;12:624–626.
9. Deltour I, Johansen C, Auvinen A, Feychting M, Klaeboe L, Schuz J. Time trends in brain tumor incidence rates in Denmark, Finland, Norway, and Sweden, 1974–2003. J Natl Cancer Inst. 2009;101:1721–1724.
10. Inskip PD, Hoover RN, Devesa SS. Brain cancer incidence trends in relation to cellular telephone use in the United States. Neuro Oncol. 2010;12:1147–1151.
11. 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;103:714–736.
12. de Vocht F, Burstyn I, Cherrie JW. Time trends (1998–2007) in brain cancer incidence rates in relation to mobile phone use in England. Bioelectromagnetics. 2011;32:334–339.
13. Dreyer NA, Loughlin JE, Rothman KJ. Cause-specific mortality in cellular telephone users. JAMA. 1999;282:1814–1816.
14. Schuz J, Jacobsen R, Olsen JH, Boice JD Jr, McLaughlin JK, Johansen C. Cellular telephone use and cancer risk: update of a nationwide Danish cohort. J Natl Cancer Inst. 2006;98:1707–1713.
15. Hardell L, Nasman A, Pahlson A, Hallquist A, Hansson MK. Use of cellular telephones and the risk for brain tumours: A case-control study. Int J Oncol. 1999;15:113–116.
16. Muscat JE, Malkin MG, Thompson S, et al.. Handheld cellular telephone use and risk of brain cancer. JAMA. 2000;284:3001–3007.
17. Inskip PD, Tarone RE, Hatch EE, et al.. Cellular-telephone use and brain tumors. N Engl J Med. 2001;344:79–86.
18. Auvinen A, Hietanen M, Luukkonen R, Koskela RS. Brain tumors and salivary gland cancers among cellular telephone users. Epidemiology. 2002;13:356–359.
19. Hardell L, Mild KH, 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;78:931–936.
20. Hardell L, Carlberg M, Mild KH. Case-control study of the association between the use of cellular and cordless telephones and malignant brain tumors diagnosed during 2000–2003. Environ Res. 2005;100:232–241.
21. Cardis E, Richardson L, Deltour I, et al.. The INTERPHONE study: Design, epidemiological methods, and description of the study population. Eur J Epidemiol. 2007;22:647–664.
22. Hardell L, Carlberg M, Mild KH. 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;79:630–639.
23. Mild KH, Hardell L, 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;13:63–71.
24. Hardell L, Carlberg M. Mobile phones, cordless phones and the risk for brain tumours. Int J Oncol. 2009;35:5–17.
25. 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.
26. Hardell L, Carlberg M, Soderqvist F, Mild KH. Re: Time trends in brain tumor incidence rates in Denmark, Finland, Norway, and Sweden, 1974–2003. J Natl Cancer Inst. 2010;102:740–741.
27. Deltour I, Johansen C, Auvinen A, Feychting M, Klaeboe L, Schuz J. Response: Re: Time trends in brain tumor incidence rates in Denmark, Finland, Norway, and Sweden, 1974–2003. J Natl Cancer Inst. 2010;102:742–743.
28. Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med. 2000;19:335–351.
29. 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;19:33–41.
30. Engholm G, Ferlay J, Christensen N, et al.. NORDCAN–a Nordic tool for cancer information, planning, quality control and research. Acta Oncol. 2010;49:725–736.
31. Frei P, Poulsen AH, Johansen C, Olsen JH, Steding-Jessen M, Schuz J. Use of mobile phones and risk of brain tumours: update of Danish cohort study. BMJ. 2011;343:d6387.
32. 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.
33. 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;63:237–243.
34. Vrijheid M, Armstrong BK, Bedard D, et al.. Recall bias in the assessment of exposure to mobile phones. J Expo Sci Environ Epidemiol. 2009;19:369–381.
35. Hardell L, Carlberg M, Soderqvist F, Hansson MK. Meta-analysis of long-term mobile phone use and the association with brain tumours. Int J Oncol. 2008;32:1097–1103.
36. Kan P, Simonsen SE, Lyon JL, Kestle JR. Cellular phone use and brain tumor: a meta-analysis. J Neurooncol. 2008;86:71–78.
37. Myung SK, Ju W, McDonnell DD, et al.. Mobile phone use and risk of tumors: a meta-analysis. J Clin Oncol. 2009;27:5565–5572.
38. Aydin D, Feychting M, Schuz J, et al.. Mobile phone use and brain tumors in children and adolescents: a multicenter case-control study. J Natl Cancer Inst. 2011;103:1264–1276.
39. Larsen IK, Smastuen M, Johannesen TB, et al.. Data quality at the Cancer Registry of Norway: an overview of comparability, completeness, validity and timeliness. Eur J Cancer. 2009;45:1218–1231.
40. Teppo L, Pukkala E, Lehtonen M. Data quality and quality control of a population-based cancer registry. Experience in Finland. Acta Oncol. 1994;33:365–369.
41. Barlow L, Westergren K, Holmberg L, Talback M. The completeness of the Swedish Cancer Register - a sample survey for year 1998. Acta Oncol. 2008;1–7.
Back to Top | Article Outline


We thank the Cancer Registries of Denmark, Finland, Norway, and Sweden for providing the incidence data, Monika Moissonnier (IARC) for her help in running the simulations, and John Daniel (IARC) for careful editing of the manuscript.

Supplemental Digital Content

Back to Top | Article Outline
© 2012 Lippincott Williams & Wilkins, Inc.