Assessing exposure misclassification by expert assessment in multicenter occupational studies.

BACKGROUND: In a multicenter case-control study of lung cancer in central and eastern Europe and in Liverpool, exposure to occupational agents was assessed by teams of local experts. We performed an interteam agreement study to estimate the levels of exposure misclassification and the expected attenuation of the risk estimate. METHODS: Eight teams of experts and a reference rater assessed exposure to 70 putative lung carcinogens for 19 jobs. Agreement among teams was calculated through Cohen's kappa, sensitivity, and specificity. RESULTS: Each team showed an overall fair to good agreement with the reference (kappa between 0.53 and 0.64). The agreement among teams in the presence of exposure was excellent for 9 agents, fair to good for 16, and poor for 29. For all agents the specificity was high (average 0.94), although sensitivity varied considerably. CONCLUSIONS: This study of expert exposure assessment showed a small range in reliability among teams of experts, but large differences among agents. This paper presents the range in levels of misclassification that can be expected using experts for assessing occupational exposure to different agents, and the attenuation of the odds ratio that can be expected to result from this misclassification.