Application of job-exposure matrices in studies of the general population: Some clues to their performance

A study of the effects of occupational exposures in the general population through the use of job-exposure matrices was reviewed step-by-step. Recall of occupational history appeared to be reliable, even after a considerable number of years. Coding of occupational histories appeared to be less reliable, especially when more than two coders were involved, and coding was done at the job level. Translation of codes in order to be able to apply alternative general population job-exposure matrices seemed to be possible for related coding schemes. Differences in assigned exposure were generally less than differences between codes, indicating that errors in coding did not necessarily lead to errors in exposure assignment. Stringent assignment of exposure in general population job-exposure matrices appeared to be beneficial since most occupational exposures have a low prevalence in the general population. Alternative methods for exposure assessment in the general population are available, but all demand more effort and costs. General population job-exposure matrices will, therefore, continue to be applied in the future. Knowledge of the possible pitfalls and sources of bias related to their use will be crucial to detect the relatively moderate-to-low occupational risks within the general population.