Autopsy Rates and Diagnosis

To the Editor: Autopsies could have new and important roles in monitoring quality and improvement in addition to the roles enumerated in JAMA.^{1 - 2} First, increasing proportions of Americans die outside of the hospital and most are old, multiply ill, and have had limited diagnostic interventions.³ Without autopsies, no one can know how often medical mistakes are buried.

Second, measures of function, symptoms, and costs in a health care system may appear to improve when practices are adopted that generally yield a slightly shorter life. Better rehabilitation or prevention services might appear to cause a healthier functional status in one population compared with another. This conclusion could be wrong if the time spent with severe disability was generally curtailed in the healthier population by earlier death.

Third, autopsies could be indicators of overall performance of care systems compared over time or among systems.⁴ Methods could be generated to sample deaths in health care systems with autopsies and chart reviews to assess quality management, reasonable decision making, and appropriateness of aggressive and palliative care. This report card could be uniquely illuminating for purchasers and potential patients.

From about 1988 to 1993, the geriatrics service at George Washington University, Washington, DC, had a nursing home, home care, and hospice population with about 300 deaths per year. We encouraged the families of patients to permit a cost-free autopsy and had a more than 10% acceptance rate. Autopsy was effective in keeping the clinicians alert and thoughtful and led to a number of quality improvements.

Our experience begins to illuminate a method for evaluating care practices. Perhaps one could express the comparable descriptors of 2 care systems in a city, for example, in terms of their incidence of clinically significant missed diagnoses, the incidence of defensible decision-making practices, the likely effects of a care plan on length of life, and other outcomes. That would enable a life-span correction in the measurement of symptoms and function, such as the Health of Seniors measurements being implemented by Health Plan Employer Data and Information Set (HEDIS).⁵ In addition, findings would better inform health care purchasers and system reformers about these issues.

Of course, substantial methodological issues need to be addressed, particularly adjusting data obtained to allow comparisons across populations. However, demand for quality comparison and for life-span adjustments to other measures could provide quite a force for reviving the autopsy, and yielding all of the traditional benefits that Dr Lundberg enumerated.²