The autopsy in pediatrics and pediatric oncology: a single-institution experience.

Autopsy rates and clinicopathologic correlations for pediatric autopsies and the subgroup of pediatric oncology autopsies in a large teaching hospital were studied to evaluate the utility of autopsy in these populations. Autopsy records of the University of North Carolina hospitals from 1982 to 2001 were reviewed for all patients less than 18 years of age. Autopsies performed during 1982 to 1991 (decade 1) were compared to those from 1992 to 2001 (decade 2) with respect to absolute numbers of autopsy and rates of unexpected postmortem diagnoses. Postmortem diagnostic discrepancies were subclassified into major and minor categories. The mean number of autopsies per year for decade 1 was 110+/-24.5, compared with 77.5+/-40.9 for decade 2 (P<0.001), a change largely due to a decline in fetal/perinatal (patients<7 days of age) autopsies. Of 533 pediatric autopsies, 43 were in patients with a primary diagnosis of a neoplasia. At least one antemortem misdiagnosis and/or clinically occult process was identified in 20.5% of all pediatric autopsy cases, and in 25.6% of pediatric oncology cases. These rates did not change significantly over time. In 10 of 43 pediatric oncology cases (23.3%), autopsies provided diagnostic information that was previously unknown to the clinicians. Three of these patients died shortly after presentation. Although autopsy rates for fetal/perinatal cases have declined, those for non-perinatal and pediatric oncology patients remain stable. Autopsy continues to provide diagnostic information that is unknown during the antemortem period, and remains a valuable tool for pediatricians and pediatric oncologists.