Comments on “Mortality by neoplasia and cellular telephone base stations in the Belo Horizonte municipality, Minas Gerais state, Brazil” by A. C. Dode et al. Science of the Total Environment 409 (2011) 3649–3665

Dode et al. (2011) report an apparent increase in “accumulated incidence rates” of cancer deaths in residents living within 500m from a cellular base station (BS). This report has provoked considerable discussion by the public and media, if not by health agencies themselves. This letter calls attention to major weaknesses in the study that prevent any conclusions about possible health effects from living near a cellular base station. The authors conducted a statistical analysis of locations of cellular base stations in relation to residences of individuals who had died from any form of cancer during 1996–2006 in Belo Horizonte, Brazil. This city of 2million residents is divided into nine sanitary districts of roughly similar area but with greatly varying numbers of cellular base stations. For each “eligible death by neoplasia” during 1996–2006, the authors determined the distance from the decedent's home to “the location of the first transmitter antenna of the mobile phone network to which the resident was possibly exposed”, and the total population in the same area, and aggregated the results in 100meter intervals from the transmitter (Table 1, from Table 5 of Dode et al.). “Compared to the total population mortality rate”, the authors conclude, the relative risk in this “area [closest to the transmitter] was 1.35.” The authors do not clearly describe how they obtained the results shown in Table 1 (their Table 5), how they avoided double-counting residents in overlapping 1-km radii surrounding different base stations, or even what is “the base station” to which they refer in the caption to their Table 5, nor do the authors provide a rationale why distance to the “first licensed transmitter” within 1km of a house should be a useful proxy measure for radiofrequency field exposure to an individual, given the presence of many other sources of radiofrequency energy in the environment (including many base stations erected subsequent to the first transmitter within 1km of a decedent's residence and not considered in their analysis). In an attempt to shed light on Table 1, we listed the same data in Table 2 showing mortality rates per area section, which is a more useful measure of risk than a cumulative distribution over the entire population. Table 2 shows a very rapid falloff in population and population density with distance from “the base station”. This cannot plausibly represent the population in any specific location in the city. More likely, Table 1 shows the cumulative distribution of distances from the decedents' homes to the “first transmitter antennas”within 1km of each home, not (as the text suggests) the numbers of decedents within a given distance of any particular station. If so, each row in the table includes individuals that are widely scattered throughout the city. The closer-in rings would over-represent individuals living in the central area (with by far the highest density of base stations) and the outer segmentswould over-represent individuals living inmore remote districts with fewer and more widely spaced base stations.