The purpose of this study was to determine the accuracy of clinical diagnoses compared to autopsy findings in critically ill patients in the current medical era.Design:
We conducted a retrospective, blinded review of matched medical records and postmortem findings in patients who died between June 2006 and June 2011.Setting:
An ICU of a major university teaching hospital in Dublin, Ireland.Intervention:
None.Measurements and Main Results:
A modification of the Goldman criteria was used to classify diagnostic error. There were 629 ICU deaths during the study period. Two hundred and seven patients underwent autopsy and 204 records were available for review. The mean age was 59 ± 18.1 years, 62% were male, 70% were postoperative patients, and median length of ICU stay was 3 days. Admission diagnosis, admission source, and admission specialty were similar between autopsy and nonautopsy patients. Five patients (2.4%; CI, 0.8–5.6%) had a class I discrepancy and 11 patients (5.4%; CI, 2.4–9.7%) had a class II discrepancy. Minor missed diagnoses were present in 31 patients (15.2%; CI, 4.5–12.4%). There was complete concordance between clinical and postmortem findings (class V) in 161 patients (79%; CI, 72.7–84.3%). In more than half the cases of discrepancy, it was not possible for physicians to make the diagnosis antemortem in the time available, despite appropriate investigations.Conclusions:
We detected a lower rate of clinicopathological discrepancy in critically ill patients than previously reported. Potential reasons for such findings include advances in diagnostic techniques and the use of a more robust definition to classify diagnostic discrepancies. Autopsy can still identify discrepancies in diagnosis even in patients who have undergone appropriate investigations. Prospective research is required to accurately define discrepancy rates in the critically ill population and to identify the patient subgroups in whom autopsy will continue to yield valuable information.