UCLA

Study objective

To determine, in medical patients who received mechanical ventilation, the frequency and types of major unexpected diagnoses at autopsy that, if known before death, would probably have led to improved survival (class I errors) or substantively changed management but not survival (class IIB errors).

Design

Retrospective cohort study.

Setting

Six medical intensive care units in a Department of Veterans Affairs Cooperative Study.

Patients

One hundred seventy-two autopsied patients of the 401 veterans who received mechanical ventilation and died in the hospital.

Results

The class I error rate was 12%. Abdominal pathologic conditions--abscesses, bowel perforations, or infarction--were as frequent as pulmonary emboli as a cause of class I errors. While patients with abdominal pathologic conditions generally complained of abdominal pain, results of examination of the abdomen were considered unremarkable in most patients, and the symptom was not pursued. Six percent of patients had extensive malignant neoplasms (class IIB errors).

Conclusions

Atypical presentation of potentially treatable abdominal pathologic conditions is a common cause of class I errors in veterans who receive mechanical ventilation. Conversely, several patients with unrecognized terminal conditions underwent intensive intervention. If the information gained at autopsy had been known before death, management would have probably changed substantively in 18% of patients.