Comparisons between magnetic field exposure indices in an automobile transmission plant

Personal monitoring of extremely low frequency magnetic fields was conducted at a large automatic transmission plant for a case-control study of primary brain cancer. Current workers were selected to represent the jobs most commonly held by study subjects. Several exposure indices, corresponding to different plausible biological mechanisms, were computed for each of 81 workers who wore the monitoring instrument for one-half shift. Average exposures covered a range from 0.16 to 46 mG; median exposure was 1.3 mG. Nonparametric correlations were estimated to learn whether all of these indices rise and fall together. Results were mixed, in that indices sensitive to high values showed correlations above 0.7, but other correlations were between 0.4 and 0.6. Different indices may thus identify different groups as "highly" exposed. The authors also tested whether indices based on the fraction of time spent above hypothesized thresholds were accurately predicted by a lognormal model. For 47% of the workers, the observed indices significantly exceeded those predicted by such a model, suggesting that lognormality is not a good model for distributions of individuals' short-term exposures.