| Data related to individual worker exposures are used to determine compliance with workplace standards, are subsequently warehoused and thereafter rarely used as an information resource. This research applies two statistical models to data from the Occupational Exposure Database of the United States Navy. The Within-Between Lognormal Model quantifies between-worker variability within a selected worker group: BR.95 = exp[3.92sigma B], the ratio of arithmetic mean exposures received by workers in the 97.5th and 2.5th percentiles; sigmaB is the standard deviation of the distribution of each worker's average exposure value. The Proportional Odds Model, a generalization of the logistic model to ordinal data, identifies probabilities for worker group exposure above the Occupational Exposure Limit (OEL), or the Action Level (AL), which is one-half of the OEL. Values for BR.95 at or near 10 support selection as a Similar Exposure Group, and further evaluation of other available variables can then be done using the Proportional Odds Model. Potential Similar Exposure Groups have been identified for Asbestos Workers removing friable asbestos (BR.95 = 11.0) and non-friable asbestos (BR.95 = 6.5); Metal Cleaning Workers sanding specialized equipment (BR.95 = 11.3), and Workers at Target Shooting Ranges cleaning up lead debris (BR.95 = 1.0). Estimated probabilities for the categories <AL, AL-OEL, and >OEL support current understanding of work processes examined, and show that the variables Task and Ventilation are most useful for identifying groups with potential overexposures. Comparisons of Similar Exposure Group and Proportional Odds Model results from 1987--1997, with independent data from 1997--1999, indicated a trend toward lower exposure values for asbestos workers, while similar probabilities of exposure were observed for supervisors at Target Shooting Ranges in both data sets. The Proportional Odds Model serves as a useful sorting tool when applying several categorical variables to a worker group. The model can help predict probability of membership in categories that may further define Similar Exposure Groups, and helps to identify which are the most likely determinants of excessive exposures. Such analyses of retrospective exposure data can identify work site and work practice factors for efficient targeting of resources, and for continued study. |
