Objectives To construct a cohort-specific job-exposure matrix (JEM) using surrogate metrics of exposure for a cancer study on career firefighters from the Chicago, Philadelphia and San Francisco Fire Departments.

Methods Departmental work history records, along with data on historical annual fire-runs and hours, were collected from 1950 to 2009 and coded into separate databases. These data were used to create a JEM based on standardised job titles and fire apparatus assignments using several surrogate exposure metrics to estimate firefighters’ exposure to the combustion byproducts of fire. The metrics included duration of exposure (cumulative time with a standardised exposed job title and assignment), fire-runs (cumulative events of potential fire exposure) and time at fire (cumulative hours of potential fire exposure).

Results The JEM consisted of 2298 unique job titles alongside 16 174 fire apparatus assignments from the three departments, which were collapsed into 15 standardised job titles and 15 standardised job assignments. Correlations were found between fire-runs and time at fires (Pearson coefficient=0.92), duration of exposure and time at fires (Pearson coefficient=0.85), and duration of exposure and fire-runs (Pearson coefficient=0.82). Total misclassification rates were found to be between 16–30% when using duration of employment as an exposure surrogate, which has been traditionally used in most epidemiological studies, compared with using the duration of exposure surrogate metric.

Conclusions The constructed JEM successfully differentiated firefighters based on gradient levels of potential exposure to the combustion byproducts of fire using multiple surrogate exposure metrics.