Article Text
Abstract
Introduction One of the principal challenges in community-based occupational studies is retrospective assessment of exposure. Job-exposure-matrices (JEMs) have been proposed as a cost-effective tool. However, most JEMs are built from the assessment of jobs held by men, with few studies assessing the applicability of those JEMs to the same jobs when held by women.
Objective To compare within-occupation exposure assessments for jobs held by men versus women in a Canadian JEM (CANJEM).
Method Two sex-specific JEMs were created using data from CANJEM, a JEM based on the expert assessment of exposure to 258 chemicals in >30,000 jobs held during 1933–2005, by participants in five Montreal-based case-control studies. Each cell in the JEMs provided the probability, intensity, frequency, and frequency-weighted intensity (FWI) of exposure to a selected occupational agent, for a specific combination of a 4-digit Canadian Classification and Dictionary of Occupations code and one of four time periods. We used intra-class correlation coefficients (ICC) to compare the probability, frequency and FWI between cells considered exposed in the two JEMs; ‘exposed’ cells were defined by probability ≥ 5%. Given the semi-quantitative nature of intensity, Kendall’s Tau (τ) was used. We further compared differences between the JEMs using empirical cumulative distribution functions (ECDF).
Results In total, 1,488 cells considered exposed in both JEMs were included in the analysis. ICC (95% confidence interval) of 0.58 (0.55–0.61), 0.55 (0.52–0.58) and 0.12 (0.07–0.17) were found for probability, frequency and FWI of exposure, respectively. For intensity, a τ of 0.24 (0.20–0.29) was found. The ECDFs showed a tendency for frequency to be higher in the female-JEM, but for intensity and FWI to be higher in the male-JEM. No clear trend was observed for probability of exposure.
Conclusion Our results suggest that CANJEM assessment of jobs held by men was inadequate to estimate exposure in women’s.