The INTERPHONE Study Group. Brain tumour risk in relation to mobile telephone use: results of the INTERPHONE international case–control study. International Journal of Epidemiology. EPub ahead of print, May 18, 2010.
Detailed information on mobile phone use was collected during face-to-face interviews with the study subjects or, if the study subject had died or was too ill, with a proxy respondent. The questionnaire used for the interviews also included questions about socio-demographic factors, occupational and medical exposures to ionizing and non-ionizing radiation, medical history and smoking. Information on anatomical location of the tumors was obtained from MRI reports or images, surgical or clinical records.
The analyses were conducted using conditional logistic regression for matched sets. The reference category for computation of odd ratios (ORs), “never regular users”, included subjects who had never used a mobile phone and subjects who used mobile phones occasionally, but never as much as one call a week for ≥6 months. Cumulative number and duration of calls were analyzed by categories based on deciles of the distribution of these variables among controls and regular mobile phone users.
For meningioma, the OR for tumors in the temporal lobe was significantly below 1.0 (see table). The risk of meningioma located in the temporal lobe was reduced (though not always significantly) in all categories of duration of use, cumulative call time and cumulative number of calls. For glioma, the risk of tumors in the temporal lobe was reduced, but not significantly. The OR for temporal lobe glioma was above 1.0 in the highest categories of duration of use (≥10 years), cumulative call time (≥1640 hours) and cumulative number of calls (≥27,000); in the highest category of cumulative call time this increase was significant (OR=1.87; 95% CI 1.09-3.22).
The ORs for ipsilateral mobile phone use (i.e. use on the side on which the tumor occurred) for meningioma were below or around 1.0 in most categories of duration of phone use, cumulative call time or cumulative number of calls; a slight non-significant increase was observed in the highest category of cumulative call time and second highest category of cumulative number of calls. For glioma, the OR for ipsilateral mobile phone use was significantly increased in the highest category of cumulative call time (OR=1.96; 95% CI 1.22-3.16); a non-significant increase was observed in the highest category of duration of use and the highest category of cumulative number of calls. For both meningioma and glioma, the OR for ipsilateral mobile phone use was higher than OR for contralateral use, and this was the case for most categories of duration of use, cumulative call time and cumulative number of calls. For meningioma, the ipsilateral/contralateral ratio of the ORs was the highest in the two highest categories of cumulative call time and in the second highest category of cumulative number of calls. For glioma, the ipsilateral/contralateral ratio of the ORs was the highest in the highest category of cumulative number of calls.
*Reference category - “never regular users” as described in the “Methods” section; adjusted for sex, age, study center, ethnicity in Israel, and education
Regarding the apparently decreased risk among regular mobile phone users, the authors reject a genuine protective effect as implausible and explore a number of alternative explanations, such as possible sampling bias, relatively low level of participation (78% among meningioma cases, 64% among glioma cases and 53% among controls), timing of interviews, and confounding. Based on the results of previously conducted work to characterize possible sources of bias and on the results of sensitivity analyses presented in this article, the authors have concluded that these potential sources of error cannot fully explain the observed reduction in risk. The possibility of mathematical correction of the OR estimates for the effects of these errors are now being explored.
Regarding the apparently elevated risks of glioma among heavy users, some subjects (more cases than controls) reported implausible values of reported use, such as >5 hours a day and even ≥12 hours a day. Previously conducted validation studies indicated that heavy users tended to over-report their phone use (errors were larger for duration of calls than for number of calls), and that cases tended to overestimate their mobile phone use more than controls. These errors could contribute to the apparently increased risk of glioma in the highest category of cumulative duration of calls.
At all levels of exposure, the ORs for ipsilateral mobile phone use were greater than the ORs for contralateral use, even though there was a tendency towards higher ipsilateral/contralateral ratio at higher levels. It is possible that at least part of this effect results from over-reporting by cases of phone use on the side of the tumor. However, true effect cannot be ruled out. The results for tumors in different lobes are less susceptible to reporting bias, and it was demonstrated that, for glioma, the ORs in the highest exposure categories were higher for tumors in the temporal lobe (where the exposure is generally highest) than in other lobes. However, the lobe-specific OR estimates were imprecise (with wide confidence intervals).
Despite similar patterns in ORs for glioma and meningioma, the ORs for meningioma were generally lower than those for glioma, and, unlike glioma, there was no evidence of increase in risk of meningioma in the temporal lobe. The authors have concluded that their study provides no evidence of an increased risk of meningioma among users of mobile telephones. The similar patterns in ORs for these two types of brain tumors “could indicate shared etiology or shared bias”.