Abstract
Context.—With the advent of modern diagnostic technology, use of the autopsy as a means of assessing diagnostic accuracy has declined precipitously. Interestingly, during the same period, the rate of misdiagnosis found at autopsy has not changed.
Objectives.—To ascertain why an autopsy was requested, whether or not questions asked by clinicians were specifically addressed, and what types of misdiagnoses were found.
Design.—One hundred forty-two consecutive autopsy records from the University of Arkansas for Medical Sciences Hospital were reviewed. In the same period, 715 deaths occurred, giving an overall autopsy rate of 20.14%.
Results.—Of the 125 autopsies in which the problem-oriented autopsy request was available for review, a reason for the autopsy was given in only 69 cases (55%). One hundred three clinical questions were asked, and of these, 81 were specifically addressed in the final anatomic diagnosis, 10 were addressed in some part of the autopsy report but not in the final anatomic diagnosis, 10 were not addressed at all, and 2 could not be answered by the autopsy. Sixty-one autopsies revealed 81 misdiagnoses: 47 class I (missed major diagnosis that, if detected before death, could have led to a change in management that might have resulted in cure or prolonged survival) and 34 class II misdiagnoses (missed major diagnosis in which antemortem detection would have not led to a change in management).
Conclusions.—The autopsy continues to be a vital part of medical education and quality assurance. It is important for the clinician to provide a clinical summary and specific clinical questions to be addressed or to speak directly with the pathologist and for the pathologist to provide answers that are easily accessible within the autopsy report. In this way, a problem-oriented autopsy can be performed based on questions raised by the clinician and the pathologist as a result of the gross dissection and microscopic evaluation.
A Better Learning Community: Mixed-Methods Reveal Medical Student Preferences with Implications for Learning Community Design and Implementation
