Introduction The Canadian job-exposure matrix (CANJEM) is a general population JEM built from expert assessment data of 31,673 jobs held by 8,760 participants from four Montreal case-control studies.

Objective To examine the validity of CANJEM for jobs held by women, by comparing exposure assessments using CANJEM and our expert assessment method to a selected list of 69 agents.

Methods We compared the exposure estimates for 69 agents within a population-based case-control study of lung cancer assigned by expert assessment to those derived from CANJEM. We linked the job histories of 998 women (3403 jobs) to CANJEM and thereby, derived probability of exposure to each of the 69 selected agents in each job. To create binary exposure variables (exposed/unexposed), we dichotomised probability of exposure using two cutpoints: 25% and 50% (referred to as CANJEM-25% and CANJEM-50%). Using the 3403 jobs as units of observation, we estimated the prevalence of exposure to each selected agent using CANJEM-25% and CANJEM-50%, and using expert assessment. Further, using expert assessment as the gold standard, for each agent, we estimated sensitivity, specificity and Kappa.

Results CANJEM-based prevalence estimates correlated well with the prevalences assessed by the experts. Sensitivity, specificity and Kappa varied greatly among agents, and between CANJEM-25% and CANJEM-50% probability of exposure. For some agents such as fabric dust and cooking fumes, the concordance between CANJEM-based and expert-based assessments was high and inspired confidence that CANJEM-based assessments will be adequate; however for many other agents, the concordance was low. We present concordance estimates for 69 agents.

Conclusion Exposure concordance measures between CANJEM and expert assessment differed greatly by agents. The results of this study could guide users of CANJEM as to which agents are most likely to provide results that mimic those that would be obtained with expert assessment.