[An epidemiologic research design using the Annual of the Pathological Autopsy Cases in Japan]

Epidemiologic research designs using the Annual of the Pathological Autopsy Cases in Japan (APAC hereinafter) are discussed in this paper. The APAC data base has been recommended for use in epidemiologic research. However, it has not often been utilized for such research, even though it covers all pathological autopsy cases in Japan, perhaps because of different sampling proportions of each disease and because of the dead control series in APAC. First, we present epidemiologic measures in using the APAC. We show that the data base can be treated as a case-control design and that the magnitude of the exposure effect should be estimated by an odds ratio. Next, selection bias and information bias in using the APAC are discussed. The independence of the control series from the exposure is important in the determination of the control disease. Because this design is based on internal comparison, non-differential misclassification should be stressed more than differential misclassification.