Study of mobile phone use and glioma risk was fatally flawedBMJ 2012; 344 doi: http://dx.doi.org/10.1136/bmj.e3078 (Published 01 May 2012) Cite this as: BMJ 2012;344:e3078
- Michael Kundi, head1
- 1Institute of Environmental Health, Centre for Public Health, Medical University of Vienna, Kinderspitalgasse 15, A-1095 Vienna, Austria
The article by Little and colleagues tries to show that the trend in mobile phone use and risk of glioma in the United States is incompatible with the findings of an increased risk in Sweden.1 But the methods are flawed and the results are either faulty or do not show what they are purported to show.
The legend of fig 3 reads: “Observed and projected rates (95% CI) of malignant glioma in non-Hispanic white people” but shows the estimated glioma rate in men aged 60-64 from Los Angeles. The background to the authors’ methods, contained in an appendix, shows (besides an error in formula A2) that they have simply projected glioma rates from the odds ratios reported in the studies of the Hardell and INTERPHONE groups.2 3
Applying the same rates of mobile phone use to all members of the population whether 18 or 80 years old, male or female, or from urban or rural regions is worse than naive. Indeed, early studies by two of the authors showed the huge difference in the proportion of mobile phone users in subgroups of the population.4 Modelling these differences in projections of incidence trends from the odds ratios of the Hardell group results in a small increase very close to the rates reported by CBTRUS, the Central Brain Tumor Registry of the United States, for 2004-7.5 Unfortunately, these data cannot easily be used to estimate long term trends because the coverage rate increased strongly over past decades.
Cite this as: BMJ 2012;344:e3078
Competing interests: None declared.