10 Year Autopsy Experience at a Tertiary Care Hospital.
Barbra Y Bradshaw, Todd P Chaba. University of Alberta, Edmonton, Canada
Background: The autopsy has long been accepted as the gold standard of diagnosis, playing a valuable role in both clinical accuracy and medical education. The error rates shown in 1980s autopsy series have remained relatively constant, despite speculation that the advent of modern imaging techniques would improve antemortem diagnostic accuracy. The aims of this study were to examine ten years of autopsy data at a tertiary care university hospital, looking at autopsy rates, diagnostic accuracy, and the potential influence of modern imaging techniques on error rates.
Design: Consecutive autopsy reports performed at the University of Alberta Hospital during three years of a ten-year period were obtained, from January 1 to December 31 of 1997, 2003 and 2006. Pediatric and neurologic-only autopsies were excluded. The data were analyzed to determine the numbers of autopsies performed, the hospital autopsy rate, the major error and class I error rates, and the concordance between radiologic and autopsy diagnoses.
Results: The data showed that hospital autopsy rates declined from 11.7% to 5.6%. The most common diagnostic categories causing death in 1997 and 2003 included cardiovascular (33%) and infection (20.4%). In 2006 a shift was seen, with malignancy accounting for the highest proportion of deaths (29.2%). The most common anatomic cause of death was coronary artery disease in 1997 and 2003, with a shift to gastrointestinal and lung malignancies in 2006. The overall concordance rate increased from 39.7% in 1997 to 45.9% in 2006, while the overall discrepancy rate decreased from 60.3% in 1997 to 54.1% in 2006. The major error rate did not change throughout the study period, but the class I error rate decreased from 13.1% in 1997 to 7.3% in 2006. The most common class I errors were myocardial infarction/coronary artery disease (29.6%), followed by pneumonia and pulmonary thromboembolism. Factors such as age, sex, length of hospital stay, as well as antemortem radiologic investigations had no impact on major error rates.
Conclusions: The autopsy data show that the major error, class I error rates and overall discrepancy rates are similar to those previously described in the literature. They have remained relatively stable over a ten year period, and do not seem to have been changed by the use of ante mortem radiologic investigations. This study supports the ongoing belief that continued performance of the hospital autopsy is important with respect to clinical accuracy and medical education, as there remains a significant rate of diagnostic error despite the use of modern imaging techniques.
Monday, February 28, 2011 1:00 PM
Platform Session: Section G, Monday Afternoon