Autopsy as clinical quality control: a study of 15,143 autopsy cases.

The results of autopsy records at the Institute of Pathology of Münster University documented between 1961-70, 1978-87 and 1988-92 were compared with the documented clinical diagnoses of these cases. In the decade 1961-70, 23% of clinical diagnoses were found incorrect (with therapeutic relevance). In the decade 1978-87, the error rate was 18% and in 1988-92, 12%. The difference between the three groups was statistically significant (p < 0.01). Specific data were assembled for the main malignant tumor types. With respect to infectious diseases, the agreement between clinical and autopsy diagnosis is even poorer. Lues (syphilis) is now rarely recognized in clinical diagnosis, even if it is fatal. Merely 50% of all patients with active tuberculosis as primary disease and cause of death, had been diagnosed clinically. Endocarditis in all its forms was underdiagnosed clinically in 75% of the cases. In view of the widely ranging discussion about the value of autopsy today, and in view of the fact that in epidemiological studies the correctness of in vivo diagnosis is tacitly presumed, it has to be stated that autopsy is the best quality control for progress in clinical medicine.