Epidemiologic Assessment of Cancer Risk from Mobile Phone Use: Where are We

Abstract

More than a dozen epidemiological studies have addressed the possible risk of cancer associated with mobile phone use. Overall, the evidence is reassuring, as risk estimates are close to unity and confidence interval relatively narrow. However, most studies have been based on relatively small number of long-term users. When the analysis was restricted to long-term use of mobile phones, some indication of increased risk was found for acoustic neurinomas. Also, effect related to use on the same side as where the tumor was diagnosed could not be excluded. Despite the substantial volume of research some increase in risk cannot be ruled out at the moment. Knowledge could be further advanced by improving exposure assessment rather than increasing the number of case-control studies. Prospective cohort study is a gold standard in epidemiology and would substantially advance our understanding of the possible health effects of radiofrequency electromagnetic fields emitted by mobile phones. 1. Introduction When new factors (exposures) are introduced or identified that have the potential to affect human health, multidisciplinary evaluation of possible health impact is required. Risk assessment involves hazard identification, exposure assessment and risk estimation. Hazard identification entails discovery of harmful potential, with its possible target for toxicity. Exposure assessment includes describing the occurrence of the agent, pathways and distribution in the population. Risk estimation comprises identification of mechanism of effect and evaluation of dose-response. In this review, we summarize the findings from epidemiological studies. In addition, weaknesses in published studies are considered and some suggestions for improved assessment given. 2. Methods We review the epidemiological evidence regarding cancer risk from mobile phone use. The evidence from studies conducted at individual level is summarized by means of meta-analysis, i.e., quantitative synthesis of results by obtaining a pooled estimate from published results. The pooled results is obtained by weighting the individual estimates with the inverse of the variance (obtained from confidence intervals), which is a measure of precision (amount of information). Consistency of results is evaluated by tests for heterogeneity. When heterogeneity is present, a random effects model is used. If no heterogeneity is found, a fixed effects model is used, assuming that all results represent the same global distribution of values. No such assumption is involved in random effects model. 3. Results In ecological studies, brain tumor incidence and mortality have been related to mobile phone use at popula- tion level, without being able to assess if tumors have occurred in mobile phone users or not. Analyses regarding four Nordic countries showed no obvious increase in benign (1) or malignant intracranial tumors (2) parallel with increasing mobile phone coverage. However, in some subgroups including the oldest age groups and incidence of glioblastoma increase during the late 1990's was reported. A total of 14 epidemiological studies on mobile phone use and cancer have been published by late 2005. Twelve have been case-control studies and they have included a total of more than 5000 cases with intracranial tumors. The total number of exposed cases is more than 1800 (corresponding to exposure prevalence of 1/3). In the two cohort studies the total number of brain tumor cases is much smaller, only 160. A further limitation of the latter has been relatively short follow-up, only one year in the US cohort and three years on average in the Danish study. This review will therefore focus on case-control studies, which also have an additional strength in more detailed exposure assessment.