Non-specific physical symptoms in relation to actual and perceived proximity to mobile phone base stations and powerlines

Technological development does not only improve people's quality of life but is often accompanied by increased worry about potential health effects related to environmental exposures [1]. A considerable part of the general population does not only express serious concerns but also attributes various health complaints and symptoms to relatively low-level exposure to Electromagnetic fields (EMF), emitted by sources such as mobile phone devices, base stations and powerlines [2–5]. This phenomenon of symptom attribution to EMF exposure is often referred to as "Electromagnetic Hypersensitivity" (EHS) and more recently as "Idiopathic Environmental Intolerance attributed to Electromagnetic Fields" (IEI-EMF) [6].

According to the World Health Organization (WHO) IEI-EMF is characterized by physical symptoms such as redness, tingling and burning sensations in the face, fatigue, tiredness, lack of concentration, dizziness, nausea, heart palpitation and digestive disturbances [7]. These complaints are estimated to be prevalent in 1.5% of the general population in Sweden [2], 3.2% in California [8], 5% in Switzerland [3], 3.5% in Austria [4] and 10.3% in Germany [5] and seem to be frequently accompanied by occupational, social and mental impairment [9, 10]. Age, gender, education, occupational status and ethnicity have been recognised as stable predisposing factors for the NSPS attributed to EMF [2, 3, 5, 11].

Results from well-designed epidemiological studies indicate no consistent associations between various symptoms and residential EMF exposure [12–16]. Recent reviews strengthen the aforementioned evidence, concluding that a causal relationship between health complaints and exposure to EMF cannot be adequately and consistently supported [17–19]. Additionally, the need of improvement in major methodological aspects such as exposure characterization, symptom assessment, study design, population selection, sample size and the investigation of possible confounders has been highlighted. Since the causes of EMF-attributed symptoms are unspecified and so far there is a lack of objective findings that could support a causal mechanism, these subjective complaints belong to the domain of the so-called "Non-specific physical symptoms" or "Medically Unexplained (Physical) Symptoms" which are often attributed to environmental exposures [20]. In the current paper the term "Non-specific physical symptoms" (NSPS) is used to refer to the symptoms, as a broader and more neutral term which does not imply a link with particular etiologic agents, especially since similar symptoms are very common in the general population [21].

The most recent systematic review focusing (exclusively) on experimental evidence was based on the examination of 46 studies involving 1117 subjects [19]. It was suggested that symptoms attributed to EMF might be a result of underlying psychological processes related to the nocebo effect. The latter reflects the triggering of symptoms under blind experimental conditions, due to individual's expectations of harmful health effects produced by a sham exposure source. Perceived exposure to EMF sources such as BS might be associated with elevated symptom scores [22] and could comprise an important element in this process; the subjective belief of being exposed to a hazardous source can reinforce the alertness for the presence of potential exposure indicators, the expectations of symptom occurrence and consequently the development and report of symptoms [23].

Although a number of studies have accentuated the role of psychological factors in unexplained environmental intolerances [23–28] evidence regarding a psycho-physiological process underlying this phenomenon is still scarce and consensus on a conceptual framework is lacking. In view of the possible overlap between diverse environmental sensitivities [29], it is also questionable whether IEI-EMF constitutes a unique condition or should be considered as a part of a broader syndrome. It has been shown that subjects with IEI-EMF report increased self-reported sensitivity to several other environmental stressors apart from EMF [2].

Approaches from the area of health psychology support the notion that investigation of both the individual and environmental context can elucidate the mechanisms behind the occurrence of ill health, including socioeconomic, geographic, demographic and psychological components [30]. In line with this perspective, research in environmental epidemiology has indicated that NSPS attributed to environmental exposures might be the result of an interaction between biological, psychological and social pathways [31].

This exploratory study aims to a better understanding of the pathways through which exposure to EMF could be associated with increased report of non-specific physical symptoms, by introducing potential determinants and moderators of this relationship. More specifically, adjusting for demographic, home and area characteristics, the present analysis was performed to subsequently test:

The results of this study show that the actual distance to mobile phone base stations and powerlines did not predict non-specific physical symptoms, while socio-demographic and psychological factors have a significant effect on symptom report. Higher self-reported environmental sensitivity, perceived proximity to base stations and powerlines, lower perceived control, increased avoidance, living in a rented home, female gender, lower educational level and incapacity for work were significantly associated with increased NSPS report.

Comparing the symptom frequency in our sample with previous studies using the somatization scale of 4DSQ in the working population [39], we observed an average increase between 3%-6% (per symptom) for people reporting symptoms "regularly or more often" ("Fainting" was the only exception, reported almost in the same frequency). This increase can be explained if we take into account that in the current study more demographic categories are included (such as people being unemployed/retired or unable to work who are prone to symptom report). Therefore we consider these symptom rates as representative for the general population. This can be also supported by the fact that the mean symptom scores in the current sample (83% scored between 0 - 10, 14% 11 - 20 and 2.8% 21 - 32) were lower compared to general practice patients [40] and higher compared to the working population [40].

Previous cross-sectional studies investigating the link between actual distance to an EMF source and NSPS, showed inconclusive results due to methodological differences. A study solely based on female participants didn't detect any effect of distance from powerlines on the report of NSPS [41] while a recent epidemiological study determining actual distance from BS using geo-coding, demonstrated a statistically significant but very small impact of actual distance on NSPS [5]. A possible explanation could be that in our study the association between actual distance and symptoms was tested for a greater range of other possible determinants than in the earlier studies. In addition, in the current analyses we adjusted for area effects (PC4 and PC6) and SES levels. Still, the effect of actual distance in our study was increased considerably in the fourth model and almost reached borderline significance. This is unlikely to be caused by collinearity among the examined variables, since the Variance Inflation Factor (VIF) indicated a low possibility for multicollinearity. Nevertheless, an effect overestimation due to overadjustment for (similar) socio-demographic characteristics cannot be ruled out. It is notable that after adjustment for house characteristics, the effect of "full-time" employment (> 20 hours/week) was no more significant. Additionally, the unadjusted effect of actual distance to BS (measured per meter) on NSPS is negligible compared to the unadjusted effects of the other examined variables (data are not shown). The fact that we found strong determinants of NSPS in the analyses, especially in the last model, reduces the possibility of residual confounding. However, other potentially strong determinants of symptomatology such as obesity and smoking habits were not taken into account.

A main outcome was the significant effect yielded for perceived proximity to both BS and powerlines on NSPS, which was stronger for powerlines compared to BS. This might be partly explained by the visual aspects of powerlines. Even though previous findings have suggested a relation between NSPS and self-reported distance/proximity [42], the latter was not examined as a psychologically-oriented determinant but rather as a proxy of the actual exposure and there was a lack of proper confounding investigation.

Another important finding was the contribution of psychological characteristics to symptom report; increased perceived environmental sensitivity, lack of perceived control and an avoidant coping style were associated with elevated report of NSPS even after adjusting for actual distance and perceived proximity to BS and powerlines, demographic, home and area characteristics. The role of these psychological factors as determinants of NSPS related to EMF has to date not been extensively investigated in epidemiological studies therefore there are no previous results for comparison. However, there is some evidence that IEI-EMF samples tend to report also other sensitivities [2]. In addition, avoidance behavior has been suggested as a possible characteristic of sensitive to EMF people [6] and perceived control as a determinant of subjective pain experience [43]. No effect was observed in the current study for the problem oriented coping strategy, the improvement of which comprises an important element in psychological treatments of NSPS [44]. Possibly, this does not hold for environmental stressors which are typically outside the control of individuals [45].

This is the first study in which the possible relation between actual distance and perceived proximity to BS and powerlines, perceived environmental sensitivity, coping strategies, perceived control and NSPS was investigated in a relatively large population sample.

An important strength is the limited possibility of awareness bias, since the sample was not originally derived from subjects residing in varying vicinities of BS but was stratified based on areas with contrasting risk of environmental problems such as air and noise pollution and limited availability of green. Apart from the two questions on perceived proximity to BS and powerlines, the issue of EMF exposure was not addressed in the original study nor included in the questions regarding environmental sensitivities. The limited possibility for such bias could be also supported by the non-significant association between actual and perceived proximity for both BS and powerlines, although this association could be influenced by the definition used to describe "vicinity", which leaves some room for subjective interpretation.

Besides the cross-sectional nature of the present study, further limitations should be acknowledged. One weakness is related to the utilization of actual distance to BS as a proxy for exposure. Geo-coded distance might be a useful component in an EMF exposure prediction model but it is moderately correlated with residential exposure from fixed transmitters [46]; it is considered as a too simplistic proxy of the actual exposure level [46, 47] which is a function of the square root of the Equivalent Isotropic Radiated Power divided by the distance. In a better approximation the power level and the antenna characteristics, e.g. the direction of the main beam of the transmitter, as well as the reflections and absorptions along the path from antenna to the home of the participant, as the housing parameters should be taken into account. Also the contribution of other EMF sources is of prior importance [46, 47].

Another limitation of the study is the relatively low response rate which could increase the risk for non-response bias. Possible reasons could be the length of the study questionnaire and the small reward for participation. Non-response analysis however did not reveal large differences. Finally, at the time of this study only data on BS location as far back as June 2008 were available. Therefore the sample addresses in 2006 could only be compared with the base stations of 2008. This implies that for some addresses the closest base station did not exist yet or was not yet operable in 2006. More specifically, in June 2006 the total number of base stations (GSM900, GSM1800 and UMTS) amounted to 20.821; for June 2008 this number was 24.240 indicating an increase of 16% (data derived from the website of the Dutch 'Antennebureau', (http://www.antennebureau.nl/antenneregister) consulted on March 15 2011). We judged this 16% mismatch in the number of base stations as acceptable and had no means to reduce it. Thus, we realize that this mismatch resulted in an underestimation of the distance of the sampled addresses to the base stations.

Bearing these limitations in mind, this analysis has laid the ground for future studies into the effects of actual and perceived exposure to EMF by pinpointing the influence of individual and environmental factors when examining the link between environmental risks and health. The findings suggest that the report of NSPS in EMF studies should be approached as the outcome of a complex interaction between aspects such as actual exposure to environmental factors, the perception of being exposed and psychological factors.

Definition and outcome measurement issues are still under debate, such as the consideration of IEI-EMF as syndrome, disorder or set of symptoms, and its differentiation from somatoform disorders and NSPS. Under a common conceptual ground in terms of diagnostic criteria, future studies have to target on the reduction of recall and selection bias in EMF studies by the combination of the electronic medical records of general practitioners and self-reported health data, and the separate examination of actual and perceived exposure. Appropriate methods for rating symptoms as EMF-related are required, taking into consideration measurement determinants that have been proposed by the broader research field of medically unexplained symptoms such as population type, use of validated symptom checklists and frequency, severity and duration of the symptoms [48]. This should be accompanied with testing the significance of psychological variables that have been proposed as relevant to the report of NSPS while adjusting for psychiatric comorbidity.

The possible role of external influential factors such as media in the perception of risk and the magnification of related worries can additionally be a dimension of research on EMF and NSPS. It is also necessary to conduct more longitudinal and prospective research to address which variables constitute stable determinants of NSPS.

The present cross-sectional epidemiological study in the Netherlands is an exploration of potential determinants of symptom report related to distance to mobile phone base stations and powelines. It shows no relation between actual distance to these EMF sources and NSPS. Perceived environmental sensitivity, perceived proximity, lower perceived control, increased avoidance behavior and particular demographic characteristics and home aspects were significantly associated with increased symptom report. Further analyses showed a trend towards a moderator effect of perceived environmental sensitivity on the relation between perceived proximity to BS and NSPS. These components should be introduced in future epidemiological studies as potential moderating factors in order to comprehend the causal pathways that lead to the activation of somatic responses and subsequent symptoms.