The authors respond:
Drs. Tarone and Inskip1 raise some interesting questions regarding our laterality analyses.2 End points and exposures that lend themselves to laterality analysis always engage methodologically interested epidemiologists, and there is more than 1 analytical approach. Our method is basically straightforward and builds on standard case–control methodology. The only deviation is that we randomly divide controls into 2 groups. This adds an extra amount of random variability to the odds ratio estimate, but a very small amount. The approach by Inskip et al3 has the disadvantage mentioned by Tarone and Inskip in their letter—that the estimation of the usual odds ratio requires an assumption about tumor laterality in the population. Thus, with an uneven lateral distribution of tumors, as is the case in our study, it is not possible to correctly quantify the effect using the approach of Inskip et al.3 Unfortunately, we cannot compare the results of the 2 methods because reanalysis of our long-term (>10 years) mobile phone use data with the Inskip et al method would yield zero in 1 of the denominator cells.
In addition to this basic analytical issue, there is a second question concerning how to handle individuals who cannot unequivocally be assigned left or right side with respect to exposure. Our approach allows for the classification of people using mobile phones on both sides of the head as exposed, whereas the Inskip et al method requires that these people be omitted from the analysis.
Stang et al4 mention some of the studies of mobile phone use and cancer. It should be clarified that mobile phones have not been used long enough to adequately evaluate their possible carcinogenic effect because of the presumingly long induction and latency period of most solid tumors. This is particularly relevant because the majority of the available results are negative.
Drs. Hardell and Hansson Mild5 have identified 2 typographic errors in our manuscript2; controls were selected from 3 geographic areas, not 4 as was erroneously stated within parentheses on page 654. The number of regular analog phone users among controls in Table 2 should be 86, not 85. In addition, a footnote in Table 3 might have been informative stating that for 9 cases, information about tumor side was not available from medical records. Among regular mobile phone users, 11% of the cases and 10% of the controls used the phone on both sides of the head.
Acoustic neuroma is often detected several years before the histologic verification, and therefore a restriction to only histologically verified cases will inevitably lead to exclusion of a large proportion of the acoustic neuroma cases, which would increase the potential for selection bias. Confirmation through imaging techniques such as magnetic resonance imaging or computed tomography is generally regarded as sufficient for this tumor type.
It is very common in an epidemiologic study that participants fail to answer some of the questions included in the interview, and this is the reason why the number of subjects differs slightly when summed over exposure categories for different variables.
Thomas et al6 discuss the possibility that recall bias in reporting side of the head where the phone was used might explain the results. However, if such recall bias is the only explanation for the findings, then a decreased relative risk would be expected on the opposite side of the head, which was not evident in our results. Confounding requires that the studied exposure is related also to the potential confounding factor, which seems unlikely for the few known or suspected risk factors for acoustic neuroma.
Drs. Johnston and Scherb7 suggest that the findings for long-term ipsilateral mobile phone use in our study are caused by chance, an explanation that we find unlikely. The issues they raise are common to all epidemiologic studies, and all other empiric research for that matter, and are covered in basic epidemiologic textbooks.
It is clearly of interest that a laterality analysis of subjects with more than 5 years of mobile phone use in the Inskip et al data1,3 does not give a stronger association. We now eagerly await the next Interphone Study results to see whether the Swedish Interphone results are replicated.
Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden, Stefan.Lonn@imm.ki.se
Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden
1. Tarone RE, Inskip PD. Mobile phone use and acoustic neuroma [letter]. Epidemiology
2. Lönn S, Ahlbom A, Hall P, et al. Mobile phone use and the risk of acoustic neuroma. Epidemiology
3. Inskip PD, Tarone RE, Hatch EE, et al. Cellular-telephone use and brain tumors. N Engl J Med
4. Stang A, Schmidt-Pokrzywniak A, Jöckel KH, et al. Mobile phone use and acoustic neuroma [letter]. Epidemiology
5. Hardell L, Mild KJ. Mobile phone use and acoustic neuroma [letter]. Epidemiology
6. Thomas BN, Flowers D, Caswell J, et al. Mobile phone use and acoustic neuroma [letter]. Epidemiology
7. Johnston S, Scherb H. Mobile phone use and acoustic neuroma [letter]. Epidemiology