Kühnlein A, Heumann C, Thomas S, Heinrich S, Radon K. Personal exposure to mobile communication networks and well-being in children – A statistical analysis based on a functional approach. Bioelectromagnetics 29 jan 2009 Ahead of print
Background and Objective
Although public exposure to high-frequency electromagnetic fields (EMFs) from mobile communication networks is at levels far below the legislated limits, there is recognition that children are of special concern, possibly for a higher vulnerability based on the potential for greater cumulative lifetime exposure. The present study assesses health events in children in relation to exposure to mobile communication networks, using EMF dosimetry. The study also explored the challenge of analyzing complex data typical of EMF studies, using a functional data approach which models the data as mathematical functions rather than as a sequence of individual measurements.
The study population was selected from the MobilEe-study, which involved children aged 8–12 years. The MobilEe-study was a population-based cross-sectional study that took place in four Bavarian cities (Germany) with different population sizes. The children were randomly selected from the registration offices of these cities (response rate 53%). The analyses presented here were based on 1,433 children who had complete exposure measurements. Outcome measures included various chronic symptoms, such as headache, irritation, nervousness, dizziness, fatigue, fear, and sleeping problems. Exposure was measured using a personal dosimeter. The overall simultaneous exposure to multiple frequency bands was calculated for every second by summation of the squared field strengths per second (covering GSM900, GSM1800, DECT, WLAN2400 frequency bands). Functional Data Analysis (FDA) of the exposure data was implemented by creating combinations of B-Spline basis functions, which allows complex time-dependent data to be modeled (and “smoothed”).
Maximum exposure in children was only 0.35% above the reference level. Using functional data analysis, and other classification methods for comparison, the children were split into two somewhat homogeneous exposure groups, one at a higher level (12% of children) than the other (88% of children). The different exposure classification schemes showed about 94% overall agreement. Relating exposure to well-being, there was no statistically significant association between the categorized exposure to EMF and any of the considered chronic symptoms (regardless of which classification scheme was used), though most odds ratio (OR) estimates were below 1.0, with the exception of dizziness.
The authors discussed some limitations of their study, including the restriction to one 24-hour measurement period and the lower validity of personal dosimeters during nighttime measurements when the subject is stationary. Levels of overall exposure to EMF from mobile communication networks were very low, being on average less than 1% of the reference level and being highest in the largest city, Munich. The authors concluded that displaying functional data as smooth functions in epidemiological studies can help to visualize the underlying structure of measurement data, providing a helpful tool for explorative analysis. The findings suggested that about 10% of the children had EMF exposure from mobile communication networks that differed noticeably from the rest of the study population, but no association between EMF exposure and well-being was observed in the children.