Objective: The aim of the cross-sectional study was to test the hypothesis that exposure to continuous low-level radio frequency electromagnetic fields (RF-EMFs) emitted from mobile phone base stations was related to various health disturbances.
Methods: For the investigation people living mainly in urban regions were selected from a nationwide study in 2006. In total, 3526 persons responded to a questionnaire (response rate 85%). For the exposure assessment a dosimeter measuring different RF-EMF frequencies was used. Participants answered a postal questionnaire on how mobile phone base stations affected their health and they gave information on sleep disturbances, headaches, health complaints and mental and physical health using standardised health questionnaires. Information on stress was also collected. Multiple linear regression models were used with health outcomes as dependent variables (n = 1326).
Results: For the five health scores used, no differences in their medians were observed for exposed versus non-exposed participants. People who attributed adverse health effects to mobile phone base stations reported significantly more sleep disturbances and health complaints, but they did not report more headaches or less mental and physical health. Individuals concerned about mobile phone base stations did not have different well-being scores compared with those who were not concerned.
Conclusions: In this large population-based study, measured RF-EMFs emitted from mobile phone base stations were not associated with adverse health effects.
Statistics from Altmetric.com
Radio frequency electromagnetic fields (RF-EMFs) are a ubiquitous exposure that have increased continuously for many years. At the same time the public complaints about a loss in general well-being from RF-EMFs give rise to public controversy. A scientific report by an EU expert group1 states that there are no established health disturbances related to exposures to RF-EMFs, as long as the respective guidelines for public protection are not exceeded. The guidelines suggested by the International Commission on Non-Ionising Radiation Protection (ICNIRP)2 are the basis of the EU Council Recommendation of 12 July 1999 for the limitation of exposure of the general public to electromagnetic fields. These limits are based on the thermal effects from RF-EMFs. Even in buildings close to mobile phone base stations, exposures to RF-EMFs emitted from these antennas hardly exceed a few per cent of the ICNIRP protection limits.3
Our knowledge about risk perception of RF-EMFs emitted by mobile phone base stations is based on manifold anecdotal reports from persons attributing adverse health effects to various sources of electromagnetic fields. Whereas it seems to be established that the reported health disturbances exist,4 5 the evidence that these effects are causally linked to the RF-EMF exposure is very weak.1 6 7 This assessment is, however, mainly based on the absence of consistent effects in human provocation and sleep studies and on the lack of a biologically plausible explanation.1 5 8 9 Very little is known about how these health complaints relate to true RF-EMF exposure and/or to the proximity to mobile phone base stations outside laboratory environments.1
We performed a comprehensive population-based cross-sectional study in Germany. In phase 1 the prevalence of health complaints in the general population and its association to the vicinity of base stations was investigated.10 In this article we report results of phase 2 with the aim to investigate whether there is an association between health disturbances and the exposure to RF-EMFs as measured in the participants’ residences.
The study was conducted in two phases. In phase 1, a nationwide survey of 51 444 persons who belonged to a panel population of the Test Panel Institute (TPI) was carried out by the Taylor Nelson Sofres (TNS) Institute. The panel bases on a large household sample of persons who agreed to participate in future ones. Thirty thousand and forty-seven persons (58%) of the panel who were selected from the German population of the age of 14 to 71 years filled in the questionnaire, which consisted of a health complaint list and of items about the perception of risks of mobile phone base stations.10
Phase 2 started in December 2005 when personal dosimeters for RF-EMF measurements became available.11 From the study base of phase 1, participants of a sub-sample were particularly chosen from urban regions (Berlin, the western part of the Ruhr district, Hamburg, Cologne/Bonn, Munich, Dresden, Hanover, Stuttgart) because estimates in phase 1 had shown that the RF-EMF exposure was higher in urban regions than in rural areas and because the access to the participants was easier. In these selected regions 4150 persons who had all participated in phase 1 were re-contacted, and asked to fill in a detailed questionnaire about health disturbances and risk perception. Those who participated were asked for written consent whether they were willing to have RF-EMF measurements performed in their homes. The participants were informed that filling the questionnaire and answering individual questions is voluntary. The topic of the research, and the participating institutes were named in the letter of introduction. As is usually the case with this type of panel surveys conducted by marketing research companies and in the case of social science, ethical approach was not sought.
The postal questioning of this sub-sample occurred in February and March 2006. Dosimetric measurements in the households of people willing to participate occurred between March 2006 and August 2006. Parallel to the dosimetric measurements, by a computer-assisted personal interview (CAPI), information was obtained about the location of the bedroom and the visibility of mobile phone base stations. In addition, two questions were asked about sleep disturbances in the night before the measurements and headaches during the day of the measurement.
During the data collection in phase 2, all 4150 participants were asked to fill in several standardised health questionnaires, namely the 18-item Pittsburgh Sleep Quality Index (PSQI),12 13 the six-item Headache Impact Test (HIT-6),14 the 24-item list of psychosomatic complaints (von Zerssen list),15 the 36-item profile of mental and physical health (SF-36),16 and a short version of the Trier Inventory of Chronic Stress (TICS — 12 items).17 Scores were calculated in line with the manuals of these scales (a higher score indicates an increased level of complaints for PSQI, HIT-6, von Zerssen, and TICS, and a decreased level of complaints for SF-36). Additionally, participants were asked to fill in a questionnaire on whether they were concerned about mobile phone base stations and whether they attributed own adverse health effects to mobile phone base stations, as well as questions on residence characteristics, the use of mobile phones and sociodemographic factors. Participants were also asked whether they would accept the visit of an interviewer to perform RF-EMF measurements. In total, 3526 (85.0%) subjects replied to the questionnaire and 1808 (51.3%) agreed that measurements could be taken in their homes. The study procedure guaranteed data security and anonymity of participants during analyses. The postal questionnaires were sent and collected by the holder of the TPI. All data were forwarded to the TNS Institute and further to the University of Bielefeld and Mainz as anonymous datasets. For the RF-EMF measurements a written consent of the participants was necessary. The TPI forwarded the addresses of the participants who were willing to participate in this study to the TNS Institute. No personal information was inputted in the datasets. At the University of Mainz and Bielefeld, where the final analyses were carried out, only anonymous data were available. All requirements of the German Data Protection Act were adhered to.
Exposure assessment of RF-EMF
For the exposure assessment RF-EMF dosimeters from Antennessa were used, which measure 12 frequency ranges: FM radio (88–108 MHz), analogue TV and DVB-T (174–223 MHz), TETRA (380–400 MHz), analogue TV and DVB-T (470–830 MHz), DECT (1.88–1.9 GHz), wireless LAN and Bluetooth (2.4–2.5 GHz), and uplinks and downlinks of the three different frequency ranges used for mobile communication. Downlink frequencies are used for the communication between the mobile phone base station and the handset (GSM900: 925–960 MHz, GSM1800: 1805–1880 MHz, UMTS: 2110–2170 MHz). Uplink frequencies are used vice versa for the communication between the mobile phone and the mobile phone base station (GSM900: 880–915 MHz, GSM1800: 1710–1785 MHz, UMTS: 1920–1980 MHz). The uplink frequencies were not used for further analyses. Measurements of 5 minutes each were carried out on four different positions on the beds of the participants: in the middle of the pillow, 10 cm to the left, 10 cm to the right of the middle position, and finally, on the blanket 50 cm down from the middle of the pillow. On each position, 75 measurements were obtained for each of the 12 frequency ranges. The measurements took place during daytime. The earliest measurement started at 7 am and the latest ended at 8 pm. To standardise the procedure, bedroom windows were closed, and no one was allowed to enter the bedroom during the measurements. All special events occurring during the measurements (such as opening the door or using a mobile phone) were documented. The dosimeter software was linked to the CAPI program to check the battery level, the number of measurements during each cycle and the correct procedure of the measurement. The dosimetric measurements as well as the checks in the CAPI program were continuously monitored.
At the beginning, in the middle and at the end of the exposure assessment period, each dosimeter was checked by an engineer. Except for one dosimeter, all the dosimeters proved to show very repeatable measurements in all the 12 frequency ranges. Before the repeatability test took place, a tolerance threshold for the deviations to the original state was defined as ±3 dB. The repeatability measurements showed in all frequencies (except for the one defect dosimeter) deviations of less than ±1 dB. For the three downlink frequency ranges used for mobile communication the variation was less than ±0.5 dB. Measurements from the defect dosimeter have been excluded from further analyses.
For the exposure to RF-EMFs of the mobile phone base stations, a dichotomised variable was created: the first five measurements on each of the four positions of the dosimeter on the bed were excluded to avoid any interference in the readings which could have occurred when the interviewers had to start and/or change the position of the dosimeters. Thus, for each of the four positions 70 measurements were used for further analysis, that is, 280 measurements in total.
For each measurement, a total field value was calculated for the three downlink frequency ranges (GMS900, GMS1800 and UMTS) by adding up the squares of the electric field strengths of the three downlink frequency ranges and calculating the square root of this sum afterwards (according to the International Commission on Non-Ionizing Radiation Protection2). Then, all the 280 remaining measurements on the four positions of the dosimeters were combined by calculating the root mean square of the 280 total field values for each person (“mean total field value”). As cut-off point for categorisation of the exposure the 90th percentile of the distribution of the mean total field values was chosen. Thus, a person was defined as exposed when the mean total field was above 0.1 V/m (0.029 mW/m2).
Additionally, the total RF-EMF was calculated in the same manner using the remaining nine frequencies without the three uplink frequencies. These uplink frequencies were excluded as the study aim was the background RF-EMFs from mobile phone base stations and other external sources as stated above and not the RF-EMFs from the mobile phone itself.
Risk perception in the context of mobile phone base stations was defined as a trichotomous variable:
Attribution of adverse health effects to the RF-EMFs of mobile phone base stations.
Concern about mobile phone base stations, but no attribution of adverse health effects to the RF-EMFs of mobile phone base stations.
Neither concern about nor attribution of adverse health effects to the RF-EMFs of mobile phone base stations.
The following five outcome variables were considered: the complete scores from the PSQI, the HIT-6, the von Zerssen list and the physical and mental health summary scores of the SF-36. In the univariate analysis, the median values of the scores were calculated. Additionally, two multiple linear regression models were calculated for each outcome variable, with RF-EMF exposure as independent variable (only mobile telecommunication downlink frequencies in one model and total exposure in the other model). Age, gender, rural or urban area, educational level, and mobile phone use, risk perception and stress (TICS) were considered as potential confounding factors in these models. The normality of the residuals was tested. In all models there was a significant deviation from the normality. We tested different transformations of the health scores like log-transformation or square root transformation but in all cases the normal distribution of the residuals was not improved. We also tested for collinearity. In all models the variance inflation factor for linear variables varied between 1.0 and 1.1. Therefore, collinearity can be excluded. We decided to present the results for the linear regression model without having normal distributed residuals, as the numbers of observation (n = 1326) were large enough for this analysis. For the purpose of visualisation and comparing the results between the different dependent variables, the estimator for the regression parameter (ß) in the multiple linear regression analyses was calculated as percentage of the mean score measured in the corresponding reference categories.
The analyses for the PSQI and HIT-6 were repeated using the questions on sleeping problems during the night before or headaches during the day the measurement took place as dichotomous outcome variables.
Four thousand one hundred and fifty persons had been asked to fill in the postal questionnaire about health disturbances and risk perception of mobile phone base stations. Three thousand five hundred and twenty-six persons responded to the questionnaire (response rate 85.0%). Among these 3526 persons, 1808 (51.0%) had agreed to dosimetric measurements in their sleeping room. Eighteen persons could not be contacted at home, and 290 persons refused to participate in the measurement procedures although they had agreed to them in the postal questionnaire. In total, measurements took place in the households of 1500 persons (fig 1). Eighty-three persons whose measurements had been carried out with the defect dosimeter had to be excluded from analyses as well as an additional 27 persons with disagreements in personal characteristics between questionnaire and measurements. These discrepancies were not possible to be corrected, as personal information on names and addresses was not available at the TNS Institute or at the University of Bielefeld. Finally, data for the exposure to RF-EMFs from base stations were available for 1390 persons. Persons with at least one missing value for any of the variables of the regression models were also excluded to assure that all the analyses were carried out with an identical dataset of 1326 persons.
Among the total of 3526 participants, 7.5% attributed adverse health effects to mobile phone base stations while 20.0% were concerned but did not attribute own health complaints to EMFs of mobile base stations. Restricted to 1326 participants with valid RF-EMF measurements, these figures were 8.8% (attribution) and 27.1% (concern) (table 1). In 65.8% of the households a mean total field value below the sensitivity limit of the dosimeters of 0.05 V/m was measured for the downlink frequencies. The 90% percentile of the downlink frequencies was 0.1 V/m (0.027 mW/m2) and this was lower in the rural area than in the suburban and urban areas. The maximum value was 1.141 V/m (3.452 mW/m; table 2).
For all five health scores, no differences in their medians were seen for exposed versus non-exposed (mobile phone downlink frequencies). However, we observed differences across the three groups of risk perception (table 3). Participants attributing adverse health effects to mobile phone base stations had more health disturbances on all scores, compared with those who were concerned and those who were neither concerned nor attributed own health complaints to the EMFs of mobile phone base stations. Multiple linear regression adjusting for age and gender only, confirmed both the lack of associations with RF-EMF exposure and the associations with attribution (data not shown). Additionally, including type of environment, school degree, mobile phone use and TICS score (stress) in the model, did not alter the results for RF-EMFs (table 4). However, the associations between attribution and the health disturbances decreased, being statistically significant only for the PSQI (p = 0.0003) and the complaints list (von Zerssen) (p = 0.0009). This statistical significance is given when the confidence interval of the beta-coefficient did not include zero. Most of the variance in the models was explained by the TICS score suggesting an impact of stress on health and a relationship between this factor and the attribution of adverse health effects to mobile phone base stations. The partial r2 from stress was most pronounced in the multiple linear regression model on SF-36 mental (delta r2 = 0.42) followed by the regression model on the von Zerssen list (delta r2 = 0.33). The partial r2 for the PSQI model was 0.21 and for the HIT score 0.19, respectively. The partial r2 was less pronounced for the physical SF-36 score (delta r2 = 0.05). Figure 2 summarises the results of the multiple linear regressions. For the different health disturbances, the ratios of the ß values and the mean scores of the respective reference groups were calculated. For sleep disturbances, for example, the ß value of attribution was 1.08 (95% CI 0.50 to 1.66), and this was 21.4% (95% CI 9.9% to 32.9%) of the mean score of the reference group which was 5.04 (table 4).
When total RF-EMF was used instead of exposure only to mobile telecommunication frequencies, the overall picture was very similar (data not shown). In particular, there was no association between total RF-EMF exposure and the five health disturbances.
In this part of our large cross-sectional survey, including 1326 subjects, measurements of RF-EMFs showed very low exposures in most homes. We did not observe any association between RF-EMF exposure and five different health disturbances, measured with standardised questionnaires, namely sleep disturbances (PSQI), headaches (HIT-6), health complaints (von Zerssen list), and mental and physical health (SF-36). This was true both for RF-EMFs at mobile telecommunication frequencies (GSM900/1800, UMTS) and for total RF-EMFs, including also exposures from TV and radio broadcast towers, cordless phone base stations, and Wi-Fi. Variation of the health disturbances was, however, observed by degree of stress (TICS). Persons concerned about mobile phone base stations did not have any different well-being scores compared with persons not being concerned.
Our study is so far the largest investigation on the possible relationship between RF-EMF exposure and adverse health disturbances. In a previous smaller cross-sectional study in Austria, the two main findings were a positive association between concerns and sleep disorders and between measured RF-EMF fields and headaches.18 The first finding was confirmed by our data, while the latter was not. In the Austrian study each item from the health complaints list (von Zerssen) was evaluated separately while we calculated the total score. No adjustment for multiple testing was carried out in the Austrian study. Hence finding may have occurred by chance.19 In addition, the Austrian study18 applied a random selection of participants within zones that were estimated to differ considerably in exposure. Therefore it is difficult to compare it with the present study. Another cross-sectional study which took place in South Germany used a personal dosimeter for RF-EMF measurements.20 In total no overall association between exposure to RF-EMFs during a day and general health disturbances was found in this study. Further previous cross-sectional studies have severe design limitations,1 especially due to the selective ascertainment of participants, and are therefore difficult to interpret.21–23 The findings of our study are in line with the overall evidence from experimental studies, indicating no adverse health effects from exposure to low-level RF-EMFs.1
The strength of our survey was that for study participants it was not immediately obvious that the questions on health disturbances and mobile phone base stations were related to each other, as they were part of a much longer panel questionnaire. Also, our study was not restricted to areas where persons were already actively involved in actions against the erection of mobile phone base stations.
A limitation is the confinement of the measurement to the beds. Possibly different exposure levels during the day (eg, at work) were not considered. The measurements were carried out in the sleeping rooms, as the participants stayed there for at least several hours and a continued exposure was guaranteed in the bedroom. The personal RF-EMF dosimetric measurements during a day cycle of persons is limited by the shielding of the body and up to now, the whole body exposure is still difficult to estimate.11 24 In this study, the exposure to RF-EMF on the bed was measured during the day and not at night. However, a study showed that the RF-EMF exposure from mobile phone base stations between day and night did not show great differences.25 The exposure measurements in the bedrooms showed that the exposure to RF-EMFs was quite low even in urban areas. The measured maximum of the total field value (1.141 V/m) was still far below the guidelines for limiting exposure of the public to time-varying electromagnetic fields, that is, for example, 41 V/m at 900 MHz.2 When planning the study design, we decided to choose particular urban areas to select areas with higher exposure.
In this study, the selection of participants was not completely at random, as urban areas were preferred to increase the proportion of higher RF-EMF exposures. Hence, the distributions of RF-EMF exposures and health scores derived from this group cannot be utilised as figures representative for the entire German population. However, the prevalence of persons with concerns and attributions is very similar in both phases of our study. The prevalence of concerns was somewhat higher only among those that participated in the RF-EMF measurements. A further limitation of our investigation was that RF-EMF measurements were only performed once and only in one place of the residence. This could introduce exposure misclassification if the spot measurements were not representative of the individuals’ exposure in the long run. Moreover, the measurements were performed closely (between 1 and 5 months) after the health information was collected. However, using the information on sleeping problems the night before and headaches the day the measurement took place as outcome variables yielded the same patterns for sleep quality and severity of headaches derived from the questionnaire (data not shown). This implies that the impact of this time lag was presumably minor.
A general limitation of a cross-sectional study is that exposure and outcome are assessed at the same time and it is therefore difficult to draw conclusions on the temporal relation of cause and effect. Hence, the underlying hypothesis tested in this study is that the health disturbances are an immediate effect following exposure; investigating adverse health effects caused by continuous long-term exposure requires a different study design, like case-control, cohort studies, or cross-over studies.26 For the observed association between attribution of adverse health effects to the mobile phone base stations and the actual occurrence of health disturbances, it is difficult to say whether persons with persisting health disturbances attribute these symptoms to the mobile phone base station in an effort to identify a cause, or whether high levels of stress together with regarding the mobile phone base stations as a hazard promotes health disturbances, or a mixture of both.
Overall, measurements of RF-EMF exposure from base stations were far below the guidelines for limiting exposure of the public to time-varying electromagnetic fields.
In this large population-based study, measured RF-EMFs emitted from mobile phone base stations were not associated with health disturbances. However, this cross-sectional study shows that sleep disturbances and health complaints are related to the attribution of adverse health effects to mobile phone base stations.
We thus believe that the worries and health complaints of people living close to mobile phone base stations need to be taken seriously, and that improved risk communication with concerned persons is required. This however is a difficult task, as these fears are often associated with political sentiments or opinions.
There is no scientific evidence that radio frequency electromagnetic fields emitted from mobile phone base station antennas are associated with the occurrence of adverse health effects. There are however numerous reports of patients suffering from headaches, sleep disturbances, reduced life quality, or poor mental and physical health, who attribute these adverse effects and their symptoms to the base station exposure.
Until now no large-scale field study addressing this topic has been conducted.
Based on a sample of more than 1300 participants, there was no association between exposure to measured radio frequency electromagnetic fields (RF-EMFs) and sleep disturbances, headaches, health complaints, and mental and physical health. This applies to both, RF-EMF above 0.1 V/m at mobile phone base station frequencies only and for total RF-EMF exposure above the 10th percentile of the total RF-EMF.
There was an association between attribution of adverse health effects to mobile phone base stations and sleep disturbances and health complaints, demonstrating that symptoms reported by people living close to mobile phone base stations have to be taken seriously.
We would like to thank Ernst Schroeder and Claudia Barthold (TNS Healthcare, Munich), and Jörg Riedel (University of Bielefeld) for their support. We would also like to thank Hans-Peter Neitzke and his colleagues from the Ecolog Institute, Hanover, for helpful discussions about how to design exposure metrics.
Funding: The study received its funding entirely from the German Federal Ministry for the Environment, Nuclear Safety, and Nature Protection, within the context of the “Deutsches Mobilfunkforschungsprogramm” (German Mobile Telecommunication Research Programme). The study sponsor had no involvement in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.
Competing interests: None.
Ethics approval: Was not required for this type of study (a population-based cross-sectional survey without intervention). The study was approved by the Data Protection Authority from the University of Bielefeld.
Patient consent: Obtained.
We declare that we participated in the study and have seen and approved the final version. MB, GB-B, BS, PP and JS conceptualised the study and developed the study protocol. GB-B and MB were responsible for the conduction of the study. PP was responsible for the survey, which was managed by UR. PP and UR analysed the survey data. GB-B, BK and JB were responsible for data collection and, together with SS, for the analysis. CB was responsible for the quality control of dosimeters. JS, GB-B and MB prepared the manuscript which was jointly finalised by all authors. GB-B is the guarantor of the work.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.