Abstract
Job-exposure matrices (JEMs) are being used to organize occupational exposure assessments for a nested case—control study of British Columbia pulp and paper workers. To validate the methodology used to create the JEMs, the estimates from the first mill studied were compared with personal measurements subsequently collected from that site. In total, 1678 samples were collected from workers to assess their exposure to five chemicals; these samples characterized both shift-long average and short-term high level exposures. The comparison indicated that the estimates were effective at predicting which job titles were exposed to carbon monoxide, calcium oxide, and wood dust, but less so for chlorine dioxide. Estimates were also effective at predicting which job titles were unexposed to calcium oxide, but less so for carbon monoxide, chlorine dioxide, and wood dust. The differing degrees of agreement between the estimates and measurements can be partly explained by the physical characteristics of each chemical and process changes at the mill. This validation generated refinements to the JEM methodology. Originally there were two parameters for each exposure estimate: concentration and proportion of time exposed. The short-term high level exposures that were originally incorporated into the concentration parameter are now assessed independently. Also, proportion of time exposed was revised from four categories to three to reflect the time spent in exposed environments by the three categories of workers: operators, field workers, and laborers. The sampling results identified that factors such as process upsets, tasks, and location can affect the variability of exposures. The necessity of validation and the benefit of mill-specific over industry-specific retrospective exposure assessments are clearly demonstrated.