Reliability and agreement of ELF magnetic field dosimeters: An analysis based on calibration data

In this paper we describe an analysis of reliability and agreement of two types of 50-Hz magnetic field dosimeters, the same instruments which are currently in use in a nation-wide case-control study conducted in Italy to ascertain etiologic factors for childhood cancer. Our analysis is restricted to the calibration setting: we estimated instrument agreement and reliability by utilizing repeated calibration data obtained for the two types of instrument over several values of the reference magnetic field generated in the calibration experiment. The statistical analysis is based on a hierarchical model and is discussed in detail in a previous paper. Here, we summarize our modeling approach and we use a larger set of calibration data to confirm our previous findings. We confirm a substantial agreement between the two meter types and a lack of bias of any practical meaning. We also confirm our finding for one of the two meters of a rather strong dependence of the instrument’s measurement error on orientation within the reference magnetic field. It is unclear what the implications of our results may be for epidemiologic studies, particularly in the presence of other sources of error or distortion that may be far more evident in those studies.