Can we rely on computed tomographic scanning to diagnose pulmonary embolism in critically ill surgical patients?

Spiral computed tomographic pulmonary angiography (CTPA) is gaining an increasing role in pulmonary embolism (PE) diagnosis because it is more convenient and less invasive than conventional pulmonary angiography (PA). Encouraging reports on the reliability of CTPA for medical patients have prompted widespread use despite the fact that its value in critically ill surgical patients has been inadequately explored. Hemodynamic and respiratory issues of critical illness may interfere with CTPA's diagnostic accuracy. The objective of this study was to compare CTPA with PA for the diagnosis of PE in critically ill surgical patients.

Over 30 months (August 1999-February 2002), 37 critically ill surgical patients (28 trauma and 9 non-trauma patients) wiith clinical suspicion of PE were enrolled prospectively. CTPA and PA were independently interpreted by four radiologists (two for each test) blinded to each other's interpretation. Clinical suspicion for PE was classified as high, intermediate,or low on the basis of predetermined criteria. PA was considered as the standard of reference for the diagnosis of PE.

PE was found in 15 (40%) patients by: central PE in 8 and peripheral PE in 7. CTPA and PA findings were different in 11 patients (30%): CTPA was false-negative in 9 patients and false-positive in 2. Its sensitivity and specificity were PE 50% and 100%, respectively, for central PE; 28% and 93% for peripheral PE; and 40% and 91% for all PE. There were no differences in risk factors or clinical characteristics between patients with and without PE. The level of clinical suspicion was identical in the two groups. The independent reviewers disagreed on CTPA or PA interpretations in 11% and 16% of the readings, respectively.

PA remains the "gold standard" for diagnosis of PE in critically ill surgical patients. CTPA should be explored further before being universally accepted. Clinical criteria are unreliable for detecting PE in this population and therefore a high index of suspicion should be maintained.