Background Job Exposure Matrices (JEM) are rarely used to estimate work-related physical exposures; few studies have compared JEM exposures to other methods. Using existing data from a prospective cohort study of carpal tunnel syndrome (CTS) in the US, we compared associations between physical factors and incident CTS using different exposure estimates: individual level job observations, a JEM that pooled these observations at the job level, and a JEM based on O*NET, a publicly available US dataset describing the physical requirements of different occupations.

Methods Workers were recruited from >50 US workplaces and followed from baseline examination for incident CTS. Direct observations and video analysis were used to estimate hand force and activity. We studied 4 observed exposures associated with incident CTS in previous publications from this cohort, and selected a priori from O*NET 5 exposures corresponding to these same domains. Cox proportional hazards models included age, gender, body mass index, and study site. Exposure variables were dichotomized at the median of their respective distributions.

Results We analysed 2,393 workers, representing 3617 person years and 140 incident cases of CTS. Statistically significant HRs were found for the individually observed exposures of hand exertion, time in forceful hand exertion, and Hand Activity Level (HAL). Analysis using exposures derived from both JEMs demonstrated similar associations, although the confidence intervals were wider, and effect sizes using the O*NET data (HRs from 1.3 to 1.8) were somewhat smaller than those calculated using individually observed exposures (HRs from 1.5 to 1.9).

Conclusions In a large population study of CTS, exposures assigned via a JEM derived from publicly available physical exposure data found similar though somewhat weaker associations than exposures assessed via direct observation of individual workers. Use of JEM allows studies that are otherwise infeasible, and is an appropriate study design for some studies of MSD.