Objective To evaluate the performance of a hybrid expert approach for retrospective occupational exposure assessment.
Methods PROtEuS is a population-based case-control study of prostate cancer in Montreal, Canada, comprising approximately 4000 subjects. Semi-structured interviews and questionnaires were used to collect lifetime occupational histories. Experts evaluated exposure to 345 agents, assigning semi-quantitative levels (3 categories) by certainty, intensity and frequency of exposure. Towards this, they used job-exposure profiles (JEPs) developed using traditional expert ratings from a case-control study of lung cancer conducted in Montreal in the 1990s. JEPs present summaries of ratings across agents by job title, as well as guidelines helping experts adjust their ratings based on specific exposure circumstances.
We compared expert ratings with JEP distributions to evaluate how experts modulated their assessments beyond those proposed by the JEPs while taking into account the information provided by study subjects. We restricted comparisons to blue collar occupations, and to agents and 7-digit 1971 Canadian Classification and Dictionary of Occupation codes common between the two sets.
Results A total of 279 agents and 667 blue collar occupations were retained for analysis. Experts rated exposures with higher certainty (OR 1.18, 95% CI: 1.16–1.21) and intensity (OR 1.07, 95% CI: 1.04–1.10) relative to JEPs, while the opposite trend was found for frequency (OR 0.75, 95% CI: 0.73–0.76). Weighted Kappa values for the agreement between the ratings assigned by experts with those with the highest proportion of jobs in JEPs was highest for frequency (0.70) and intensity (0.66) compared to certainty (0.44).
Conclusions PROtEuS experts provided with JEPs rated exposures with higher certainty and intensity, and with lower frequency, compared to the ratings from the JEPs themselves. This suggests that experts can use subjects’ job descriptions to modulate the ratings proposed by JEPs towards an overall greater confidence in their occupational exposure assignments.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.