Topic

Validation of the exposure surrogate of the cross-sectional study on base stations

Start

01.08.2007

End

31.01.2008

Project Management

University of Bielefeld

Objective

Within the DMF-project "Cross-sectional study on subjective symptoms in relation to base stations" an exposure surrogate based on data of the site of the stations and other parameters was developed. Previous evaluations of this surrogate showed some misclassification. Due to a lack of comparison data, no external validation has been possible up to now. Meanwhile measurements by means of personal dosimeters are available for 1.390 members of the cohort of the cross-sectional study (DMF-project "Addendum to the cross sectional study on base stations"). For this group an external validation of the exposure surrogate by means of measurements will be performed.

Methods

Information based on the following sources were used for the calculation of the exposure surrogate of ECOLOG:

1. Database of the Federal Network Agency
   • Geo-coordinates (WGS84-coordinate system) of the antennas
   • Mounting height of the antennas
   • Main beam direction of every antenna
   • Safety distances of the antennas
2. Questionnaire (self-reported by participants)
   • Urbanity
   • Surrounding properties
   • Buildings/vegetation in front of the bedroom windows
   • Height of flat (bedroom) above ground (estimated via the reported floor)
   • Direction of the windows in the bedroom (defined by the time of the day the sun shines through the window)
3. Interviewer protocol
   • Height of the flat (bedroom) above ground (estimated via the documented floor)
   • Direction of the windows in the bedroom (determined via compass)

For 1.390 individuals data on both exposure variables had been available, the calculated exposure-surrogate as well as the measurements by personal dosimeters of the company Antenessa, that had been performed in the bedroom of the participants. Individuals with estimated distances between base station and home of more than 500 m, were excluded from the analysis, because it was assumed that the exposure is practically zero. Thus 1.132 data pairs remained for the validation analyses. Both exposure variables were dichotomized by the 90 % percentile. In order to evaluate the validity of the exposure surrogate the correlation coefficient between surrogate and measurements as well as sensitivity and specifity were calculated.

Results

The Kappa-coefficient of 0,17 as well as the sensitivity of 25,4 and specifity of 91,7 showed little correlation between exposure surrogate and estimated field intensity by measurements. Problematic were inaccuracies of geo-coordinates of the sites of the base stations as well as of the residence and self-reported data on the surroundings of the home and other relevant variables. In the former internal validation study of ECOLOG the correlation between the measured values and those estimated by the exposure surrogate was appreciably higher, because the entry data for model had been much more accurate (e.g. exact coordinates of the base station and the residence, expert rating of surrounding, etc.).

Conclusion

Without major improvements in the entry data for the exposure surrogate the use of this surrogate for risk assessment in epidemiological studies cannot be recommended.