Continual audit of clinical diagnostic accuracy by computer: a study of 592 autopsy cases.

In a prospective study of 592 autopsy cases, correlations between clinical and pathological data were made by computer from a novel protocol on which morphological and functional diagnoses were separated. The accuracy of diagnosis was analysed with respect to time in hospital, specialty, consultant in charge, and other criteria. Overall, there were 168 clinical 'overdiagnoses', not confirmed at autopsy; 492 'underdiagnoses', not made clinically; and 432 'agreements' between clinical and pathological diagnoses. The computer program allows the data to be analysed at any time as the number of cases increases. Entry of information entails comparatively little extra work and a similar system could be incorporated into the service commitment of any pathology department with a suitable microcomputer and access to a mainframe computer.