Personal exposure to mobile communication networks and well-being in children--a statistical analysis based on a functional approach

The MobilEe-study was the first cross-sectional population-based study to investigate possible health effects of mobile communication networks on children using personal dosimetry. Exposure was assessed every second resulting in 86,400 measurements over 24 h for each participant. Therefore, a functional approach to analyze the exposure data was considered appropriate. The aim was to categorize exposure taking into account the course of the measurements over 24 h. The analyses were based on the 480 maxima of each 3 min time interval. Exposure was classified using a nonparametric functional method. Heterogeneity of a sample of functional data was assessed by comparing the functional mode and mean of the distribution of a functional variable. The partition was built within a descending hierarchical method. The resulting exposure groups were compared with categories derived from a standard method, which used the average exposure over 24 h and set the cut-off at the 90th percentile. The functional classification resulted in a splitting of the exposure data into two groups. Plots of the mean curves showed that the groups could be interpreted as children with "low exposure" (88%) and "higher exposure" (12%). These groups were comparable with categories of the standard method. No association between the categorized exposure and well-being was observed in logistic regression models. The functional classification approach yielded a plausible partition of the exposure data. The comparability with the standard approach might be due to the data structure and should not be generalized to other exposures.