Autopsy diagnosis versus clinical diagnosis, particularly in malignant disease. Comparison of two periods: 1961-70 and 1978-87.

We present a comparison of autopsy records in the Institute of Pathology of Münster University during 1961-70 with those from 1978-87. The study was based on 11,716 autopsy cases in which reliable documentation of clinical diagnosis could be adequately correlated and compared with the respective autopsy diagnoses. In the decade 1961-70, 34% of the clinical diagnoses were correct, 15% were nearly correct and 23%, incorrect (24% had to be supplemented without relevance to therapy or prognosis, 4% side diagnoses). In the decade 1978-87, 50% of the clinical diagnoses were correct, 15% nearly correct and 18% incorrect (17% supplemented). With respect to malignant tumours, in 1961-70 clinical diagnoses were correct in 37% of cases and incorrect for 26%; in 1978-87, 47% were correct and 15% incorrect. This comparison covered all major tumour locations. The difference in incorrect clinical diagnoses between the two periods is statistically significant. With respect to infectious diseases, the concordance between clinical and autopsy diagnoses is even poorer. Lues (syphilis) is now rarely recognized in clinical diagnoses, even though it may be fatal; 50% of tuberculosis patients die from miliary tuberculosis without a correct diagnosis. Endocarditis in all its forms was underdiagnosed clinically in 75% of cases. These data provide substantial arguments in favour of autopsy control of clinical diagnosis, also including histological findings in biopsy specimens.