A Study of Beryllium Exposure Measurements, Part 1: Estimation and Categorization of Average Exposures from Daily Weighted Average Data in the Beryllium Industry

Publisher Website
Google Scholar
Abstract
Time-weighted exposure measurements are frequently available for use in occupational epidemiology studies of exposure–response relationships. Generally, these measurements must be combined and extrapolated over time to develop job-specific exposure estimates. A stepwise strategy for the development of exposure estimates from time-weighted data is described. Elements include describing the distribution of the data, exploring trends over time, calculating estimates, and methods of categorizing the estimates. The process is applied to more than 2200 time-weighted exposures from the beryllium processing industry collected from 1950 through 1978. The data were approximately lognormally distributed, requiring the calculation of the mean of the log-transformed data, and subsequent estimation of the arithmetic mean exposure for each job title at the five facilities. Few trends in exposure estimates over time were detected in the data, which may be a result of the small number of data points for each job. Exposure estimates were categorized by several methods; criteria for selecting a specific scheme included (1) having one group with mean exposure at or near the current allowable exposure level, (2) ensuring that the groups were statistically separable, and (3) minimizing the gaps between confidence intervals. The resulting exposure estimates were generally consistent with the range of previously published values describing industry-wide exposures, or those at a single plant. Thus the estimates derived from the uniform procedures described do not appear to be biased, compared with estimates calculated historically. For a specific job title, exposures varied across each facility. Acquisition of additional data from existing company and archive sources would probably increase the number of job titles for which exposure data are available and improve the utility of these data for studies of potential health effects.

This publication has 6 references indexed in Scilit: