Using job exposure matrices to estimate an individual's exposure for compensation?

Job-exposure matrices (JEMs) are being used to assign (quantitative levels of) exposure to individuals based on their job history. In human observational studies, a group-based approach in which every individual with a similar job will be assigned similar exposure will not bias exposure-response associations but will result in loss of precision. However, since JEMs do not consider between-worker differences in average exposure, some individual workers' cumulative exposures will be underestimated. This may affect their chances of compensation when a minimal (cumulative) exposure threshold is applied.

We analysed more than 80 000 repeated exposure measurements from a variety of industries and consequently combined variance components of location and worker (within a location within a job) to estimate the bandwidth of individual average exposures within a job. This allowed estimating percentiles one and two standard deviations (SD) above the median of workers' exposure distribution within a job (across locations/companies).

The bandwidth factor appeared to be larger for exposures to particulates than for gases. It was also larger for biological agents. For exposure to particulate matter, the bandwidth factor varied slightly between industries (₈₄BW_factor range 1-4) with a median 2.5.

By applying a default bandwidth factor to an average exposure estimate resulting from a quantitative JEM, the Dutch occupational disease compensation scheme has chosen for an approach that recognises between-worker differences in exposure. This approach, in addition to considering uncertainty in exposure-response associations, addresses another important factor of uncertainty in ascertaining occupational disease based on the 'presumably plausible' principle.