Autopsy 2018

Abstract

The average autopsy rate in US hospitals was ≈50% in the 1940s and 41% in 1970, just before the Joint Commission on the Accreditation of Hospitals eliminated the requirement for a 20% autopsy rate. Since that time, autopsy rates have been in free fall, with estimated rates currently ≈8% overall, including forensic cases, but only 4% among in-hospital deaths. About 700 000 Americans die in acute-care hospitals each year, so these percentages translate into ≈28 000 hospital autopsies annually. Three explanations are commonly proposed for these falling rates: the lack of reimbursement for autopsies, the fear of disclosing mistakes that would lead to malpractice suits, and the belief that advances in medical technology, including but not limited to computed tomographic scans and magnetic resonance imaging, have made autopsies obsolete. The first 2 rationales could easily be resolved by effective legislation, such as Medicare reimbursement for autopsies, or regulation, such as reinstatement of a minimum required autopsy rate. From a scientific perspective, however, the key issue is whether autopsies remain as critical for measuring the quality of care and advancing medicine as they were 80 or even 50 years ago. AUTOPSY AS A DIAGNOSTIC TEST The autopsy is the ultimate diagnostic test, typically the gold standard, with an assumed 100% sensitivity for finding causes of death and 100% specificity for excluding them. Of course, no test is perfect; any gold standard is simply the cur-rent best. Diagnosis in isolation, although interesting, is not sufficient to justify the ex-pense and inconvenience of a test. Diagnostic tests are most useful when they result in a change in treatment. For example, my colleagues and I showed that cardiac nuclear medicine scans in 1980 and echocardiograms in 1994 provided information that led to an appropriate change in treatment, which would not have been clear without the test's result, after ≈10% of examinations. Since then, however, the landscape has changed dramatically. Cardiac nuclear medicine circa 1980 was of 3 types: gated radionuclide scans to assess ventricular function, tech-netium pyrophosphate scans to diagnose acute myocardial infarction, and thallium scintigraphy to diagnose transient ischemia or prior myocardial infarction. Gated radionuclide scans were made obsolete by 2-dimensional echocardiography. Py-rophosphate scans, which at best probably confirmed large enough infarcts to be found by enzymatic or other criteria anyway, are rarely used today except for the occasional diagnosis of amyloidosis. By comparison, more sophisticated versions of myocardial ischemia scanning still remain useful, despite new technologies such as coronary calcium scoring by computed tomography.