[Retrospective exposure assessment in occupational epidemiology: principles and methods]

Occupational histories in case-control studies typically include a variety of past exposure circumstances and no monitoring data, posing serious challenges to the retrospective assessment of occupational exposures. METHODS. I will use examples from the EPILYMPH case-control study on lymphoma risk to introduce principles and methods of retrospective assessment of occupational exposures.

Exposure assessment consists in several indicators, such as frequency and intensity of exposure, as well as a confidence score, expressing the occupational expert own judgement on the reliability of the assessment itself. Testing the null hypothesis from multiple perspectives allows boosting inference: while trends by the individual exposure indicators were all of borderline statistical significance, testing the association between CLL risk and exposure to ethylene oxide with the Fisher's test for combined testing of multiple probabilities yielded a p-value of 0.003. Using the occupational expert assessment as the gold standard, the specificity of a prior job-exposure matrix for benzene was 93%, and its sensitivity 40%., with a positive and negative predictive values ranging 71-77%.

Once bias can be excluded, assuming a true association between exposure and disease, retrospective exposure assessment only under estimates the true risk, which size also depends on frequency of the exposure itself.