Mobile Phone base Stations and Health

GUNNHILD OFTEDAL / COST 281 Workshop in Dublin, 15may03

Abstract:

May epidemiological studies be used to study health effects related to RF fields from base stations? Methodological considerations and reported studies as bases for discussion.

Gunnhild Oftedal, Sør-Trøndelag University Collage, Trondheim, Norway.

Some few epidemiological studies investigating health effects of RF exposure from base stations have been accomplished so far. These studies have focused on subjectively reported symptoms and on cognitive performance. Cross-sectional surveys have been used. In general, this method may be useful to raise questions about an association, rather than to test hypotheses. In order to test hypotheses, case-control studies or cohort studies may be relevant. The choice of method also depends on the nature of the endpoint and of the exposure, for instance whether the exposure/health effect is rare or common, and whether the effect is acute or delayed relative to the time of exposure.

In this presentation methodological aspects of epidemiological studies will be discussed using the reported studies as examples. First, brief summaries focusing on the methods of the studies will be given.

In France Santini et al. (2002) obtained information from 530 people responding to a questionnaire about non-specific heath symptoms and about the distance to base station. Respondents living various distances (less than 300 meters) from base stations were compared with a reference group living more that 300 meters from base stations. A Spanish study conducted by Navarro et al. (2002), used the same list of symptoms that was used in France. The 89 respondents included in the study were grouped according to measured electric field in the bedroom. The questionnaire also included question about the time of exposure, the use of mobile phone, and presence of other sources of electromagnetic fields.

In Austria Hutter et al. (2002) investigated symptoms, sleep quality and cognitive performance of 336 people living in the proximity of base stations. The subjects were told that the investigation was about environmental health problems. Tests and questions, presented by the use of computers, proceeded RF emission measurements in the subject’s bedrooms. Also information about air quality and noise was sampled. Age, sex, the subjective rating of health consequence of base stations, and the use mobile phone were included in the analysis as potential confounding factors.

Some of the aspects that need to be considered from a methodological point of view are statistical power, bias, confounding and the exposure assessment. With respect to the exposure, here it will just be mentioned that the distance to a base station can not be regarded as a relevant surrogate for the RF exposure since the correlation between distance and indoor exposure level is low (Schutz and Mann, 2000).

When the significance level is determined (usually α = 0.05), the statistical power depends on different factors. Among these the difference in health effect between the exposed group of people and the reference group, and the number of people in the groups that are compared, may be controlled by the design of the study. The challenge will be to identify at least a group of people that is sufficiently strongly exposed to maximise the effect of the exposure (if there is an effect) and a group with a low exposure level as a reference group. Furthermore the number of people within each group should be sufficiently high. In Austria, the number of subjects in each of the three exposure groups varied from about 90 to 150, while in the Spanish study almost all subjects fell within the lowest exposed group leaving just a few subjects in each of the three groups with higher exposure levels. Using three or more categories with different exposure levels or doses is preferable if possible. Then a potential dose – response association may be revealed. The number people that is required in each group to obtain a sufficient statistical power should be estimated when planning the study.

Avoiding bias is essential for the interpretation of the results. Bias may for instance be caused by a low participation or response rate if selected people that do not participate differ from the participants with respect to the statistical relation between exposure and health condition. Efforts to increase the percentage of participants are important, and if there is a low participation rate, a non-participation analysis may be useful to estimate the impact of the non-participants. The Austrian report describes in details the selection procedure. The Spanish study lacks information about how many who were not included due to neurological and psychological records.

Information bias may in particular be a problem when the subjects are aware of the exposure under consideration. An exposed person may tend to over-report symptoms compared to one that is not exposed. The subjects of the Austrian study were not informed about the reason for the investigation at the stage of the test and response to the questionnaire. However, the report does not state whether the persons who presented the tests and questionnaires knew whether the house was located in the presumably most intensely exposed area or not. A double blind design should be used to avoid possible unconscious influences. Any bias has to be avoided by a careful design of the study.

Confounding may occur when a factor is associated with the exposure and with the health effect independently of the influence of the exposure. The authors of the Austrian study included various factors as covariates in the statistical analysis to adjust for potential confounding effects. Concern about negative health implications of the base station was statistically related to some symptoms. The crude analysis suggested a statistically significant relation between base station RF exposure and sleep quality, but this relation did not remain when adjusting for the effect of concern about base station. This implies that taking into account concerns for health effects and possibly other potential confounding factors, is crucial. Otherwise the results may be misinterpreted.

What about exposure from other RF radiation sources, including the use of mobile phones? If the RF radiation from base stations causes negative health effects, a health implication of the radiation from other sources can not be ruled out. Thus, the study should be designed so that signals from radio and television transmitters are as weak as possible in the area of the study population, or the field strengths from these sources should be measured and treated as potential confounding factors. Also RF exposure at working places should be taken into account. The use of mobile phones was recorded in the Austrian and Spanish studies, and in Austria it was treated as a potential confounding factor. It should also be considered whether the amount of mobile phone use should be included in the analysis since epidemiological studies concerning non-specific health effects suggest a statistical correlation between minutes of use per day and frequency of symptoms (Sandström et al. 2001).

These methodological considerations are important, but are just a few of those that are needed when deciding whether or how an epidemiological study should be accomplished.

References:

Hutter H-P. Moshammer H, Kundi M. Mobile telephone base stations: Effects on health and wellbeing. Proceedings of Biological Effects of EMFs 2nd Int. Workshop, Rhode, Oct. 2002.

Navarro EA, Segura J, Gómez-Perretta C, Portolés M, Maestu C, Bardasano JL. Exposure from cellular phone base sations: A first approach. Proceedings of Biological Effects of EMFs 2nd Int. Workshop, Rhode, Oct. 2002.

Sandström M, Wilén J, Oftedal G, Mild KH. Mobile phone use and subjective symptoms. Comparison of symptoms experienced by users of analogue and digital mobile phones. Occup Med 2001;51:25-35.

Santini R, Santini P, Seigne M, Danze JM. Symptoms experienced by people in the vicinity of base station: I/Incidences of distance and sex. Pathol. Biol. 2002;50:369-73.

Schuz J, Mann S. A discussion of potential exposure metrics for use in epidemiological studies on human exposure to radiowaves from mobile phone base stations. J expo Anal Environ Epidemiol 2000;10(6 Pt 1):600-5.