Group Outcomes for Time-Weighted Averaging in WBGT-Based Heat Stress Exposure Assessment

Abstract The wet bulb globe temperature (WBGT)-based occupational exposure limits (OELs) were developed from steady exposures to heat stress at constant WBGT and metabolic rate (M). The exposure limits were based on compensable heat stress exposures at the upper limit of the prescriptive zone for most healthy people. Professional practice allows for using time-weighted averages (TWAs) of WBGT and M to account for heterogeneous heat stress exposures. The purpose of the current paper was to report on the effectiveness of time-weighted averaging to assess occupational heat stress using published studies. Our hypothesis was using TWA-WBGT and TWA-M was as protective as the recommended OELs for steady exposures. The current paper reports on 62 observations of work that alternate between at least two heat stress conditions (usually work and recovery) reported in 16 papers. The TWA-WBGT and TWA-M were determined for all observations. ΔLimit was the observed TWA-WBGT minus the exposure limit at the TWA-M based on acclimatization state. The observations were then classified as above or below ΔLimit = 0. Each observation was also classified as uncompensable if the mean core temperature for the group was greater than 38°C or a less tolerant individual was above 38.5°C. When comparing exposure classifications to outcome classifications using 2 × 2 tables, the sensitivity and specificity for all observations were 0.72 and 0.73, respectively. The sensitivity was much less than the expected value near 1.0, and the large difference called into question the ability of TWAs to represent actual heat stress. There was some suspicion that there were differences between acclimatized and unacclimatized observations. Before any of these findings are embedded in policy or practice, a more careful evaluation of TWAs is required. In conclusion, we believe that the use of TWAs for heat stress analysis was not fully evaluated, and we proposed a framework for evaluation.