Sampling design and field methods of a large, randomized, multisite survey of occupational magnetic field exposure

Large, randomized assessments of occupational exposure remain unusual in epidemiologic studies because of logistical barriers. Our study of workers in five U.S. electric companies requires estimates of full-shift time- weighted average (TWA) magnetic field exposure for 28 job groups, with the ability to separate wlthin-worker and between-worker variability. Valid exposure assessment for this study is challenging because workers are dispersed over wide geographic areas. We used an innovative random monitoring strategy that may be useful in future studies. A target sample size of 4000 measurements was chosen based on considerations of precision and feasibility. Jobs were aggregated into three levels of presumed exposure. The number of measurements for each job was weighted such that the presumed medium- exposure and high-exposure groups were respectively sampled with three and five times the frequency of the low-exposure group, and measurements were distributed within these levels in proportion to the relative sizes of the five companies. Workers were selected randomly for monitoring: Those in the two highest exposure levels were measured twice on random days. An easy-to- use integrating personal exposure meter was mailed to the selected workers, who wore it unsupervised for a full shift and then returned it. An employee at each company coordinated distribution of the meters and recorded data. After 11 months of full operation, 4259 monitoring attempts had been made on eligible workers, yielding 2885 usable measurements of 2196 workers. The low-exposure, medium-exposure, and high-exposure sampling groups had mean TWAs of 0.26, 0.53, and 1.02 μT, respectively. Appropriate instrumentation and assistance from workers and management facilitated this large randomized survey, despite substantial challenges.