Objective: To determine the degree of concordance between clinical cause of death and autopsy diagnosis in a medical ICU (MICU) setting.
Design: Retrospective medical chart and autopsy report review.
Setting: MICU in a tertiary referral hospital.
Patients: Consecutive admissions to an MICU over a 2-year period from January 1, 1994, to December 31, 1995.
Interventions: None.
Measurements and main results: One thousand eight hundred patients were admitted to the MICU during the study period. There were 401 in-ICU deaths (22.3%). The autopsy rate was 22.7% (91 of 401). The mean +/- SD age of the autopsied patients (55.1 +/- 13.5 years) was lower than those without autopsy (62.4 +/- 15.2 years; p < 0.001). The two groups were otherwise similar with regard to sex, race, APACHE (acute physiology and chronic health evaluation) III scores, and lengths of stay in the MICU and hospital. The discordance between clinical and postmortem diagnoses was 19.8% (95% confidence interval, 12 to 29%). There were no differences in age, sex, APACHE III scores, predicted mortality, and lengths of stay in MICU hospital between patients with concordant and discordant diagnoses. In 44.4% (8 of 18) of the discordant cases, management would have been modified had the autopsied diagnosis been made premortem. Seven of the autopsied patients had organ transplantation. Three of the patients who had organ transplantation had discordant diagnoses, including two patients with disseminated fungal infection that was not diagnosed clinically. Although the observed discordance in transplant patients (43%) was higher than in those without transplant (19%), the difference was not statistically significant (p = 0.15).
Conclusion: Younger patients tended to have a higher autopsy rate than older patients. The discordance between the clinical cause of death and postmortem diagnosis was 19.8%. In 44.4% of the discordant cases, knowledge of the correct diagnosis would have altered therapy.