Background: When conducting large scale epidemiologic studies, it is a challenge to obtain quantitative exposure estimates, which do not rely on self-report where estimates may be influenced by symptoms and knowledge of disease status. In this study we developed a job exposure matrix (JEM) for use in population studies of the work-relatedness of hip and knee osteoarthritis.
Methods: Based on all 2227 occupational titles in the Danish version of the International Standard Classification of Occupations (D-ISCO 88), we constructed 121 job groups comprising occupational titles with expected homogeneous exposure patterns in addition to a minimally exposed job group, which was not included in the JEM. The job groups were allocated the mean value of five experts' ratings of daily duration (hours/day) of standing/walking, kneeling/squatting, and whole-body vibration as well as total load lifted (kg/day), and frequency of lifting loads weighing ≥20 kg (times/day). Weighted kappa statistics were used to evaluate inter-rater agreement on rankings of the job groups for four of these exposures (whole-body vibration could not be evaluated due to few exposed job groups). Two external experts checked the face validity of the rankings of the mean values.
Results: A JEM was constructed and English ISCO codes were provided where possible. The experts' ratings showed fair to moderate agreement with respect to rankings of the job groups (mean weighted kappa values between 0.36 and 0.49). The external experts agreed on 586 of the 605 rankings.
Conclusion: The Lower Body JEM based on experts' ratings was established. Experts agreed on rankings of the job groups, and rankings based on mean values were in accordance with the opinion of external experts.