A comparison of post mortem findings with post hoc estimated clinical diagnoses of patients who die in a United Kingdom intensive care unit.

OBJECTIVE: To determine the accuracy of clinical diagnosis compared to post mortem findings in intensive care patients.
DESIGN: A retrospective and blinded review of medical records.
SETTING: A nine-bed combined high dependency and intensive care unit in a district general hospital in Gloucester, England. PATIENTS AND PARTICIPANTS: Ninety-seven patients who died on the intensive care unit and subsequently underwent post mortem examination.
RESULTS: Suspected cause of death and main clinical diagnoses were determined and compared with findings at post mortem examination. All patients in whom a discrepancy was found had their cases reviewed to determine the significance of the discrepancy using the Goldman classification of post mortem discrepancies. Between June 1996 and May 1999 there were 1718 admissions with 252 deaths (14.7%), where 102 post mortem examinations were performed (40.5%). Five patients were not studied. Of the 97 patients, 4 (4.1%) had Goldman I discrepancies and 19 (19.6%) had Goldman II discrepancies. Complete agreement between pre and post mortem diagnosis was found in 74 (76.3%). Discrepancies fell into four main groups; unrecognised haemorrhage (7 patients), myocardial infarction (5), thromboembolic disease (5) and infectious complications (4).
CONCLUSIONS: This study reveals that in an intensive care unit important diagnostic discrepancies were found in 19.6% of patients who underwent a post mortem examination. In a fifth of these (4.1%), survival may have been adversely affected. Haemorrhage was the most commonly missed diagnosis. Despite technological advances in intensive care medicine the post mortem examination continues to have an important role in auditing clinical practice and diagnostic performance.