Background Gender differences in jobs and tasks within jobs have been identified as a major challenge when assessing exposure in occupational cancer research. Increasingly used, general population job exposure matrixes (JEM) are nevertheless limited when accounting for intra-job variability. The objective was to assess a JEM performance against an expert case by case evaluation for typically men and women jobs.

Methods Data were generated by an ongoing case-only cancer study in Seine-Saint Denis, France (Giscop93). Between 2002 and 2014, 1,253 cancer patients (1,029 men and 224 women) were recruited and detailed occupational history was collected through a semi directed interview. On the corresponding 5,951 jobs (4,975 carried by men, and 976 by women), exposure to chlorinated solvents as assessed by the experts’ team (serving as standard) was compared to Matgene evaluation, a French population-based JEM. Indices of exposure (probability and intensity) were contrasted using Cohen’s κ coefficient of agreement separately for jobs traditionally held by men and those traditionally held by women.

Results The prevalence of exposure to chlorinated solvents was greater for the expert assessment as compared to the JEM. Agreement in classifying exposed jobs was fair to high according to the JEM cut-off points (52-80%) for the prevalence, but rather poor when exposure intensity was taken into account.

Conclusions Chlorinated solvents exposure estimates based on JEM were overall (e.g. for men and women’s jobs) consistent with the expert evaluation. Preliminary results suggest however a poor agreement beyond the binary cut-off. Further analyses are being carried on benzene and asbestos exposure.