BACKGROUND: An estimated 175,000 Americans have a pulmonary embolism (PE) each year. Pulmonary angiography is the accepted gold standard for diagnosing PE, but it is invasive, expensive, and causes cardiopulmonary complications in 3% to 4% of patients. A ventilation-perfusion (V/Q) scan is less invasive, but also less accurate. Used in combination with clinical assessment, it fails to find 20% of PEs.1 Recent studies evaluating the use of spiral computed tomography (CT) have reported favorable results in diagnosing PE. However, the role of CT for this use is not yet fully defined.
POPULATION STUDIED: In this systematic review, neither specific patient characteristics nor exclusion criteria were mentioned. Enrollment criteria were described as inconsistent.
STUDY DESIGN AND VALIDITY: The authors conducted a systematic review of the literature evaluating the use of spiral CT in diagnosing PE. They searched MEDLINE and Current Contents through July 1998 and reviewed pertinent references. Eleven articles met their preset inclusion criteria. The articles were rated by using a set of 11 basic methodologic standards for addressing diagnostic test research. None of the 11 studies met all of the criteria; only 5 studies met 5 or more criteria. All studies compared CT with either pulmonary angiography or another reference standard (high-probability V/Q scan plus high clinical suspicion) to confirm the diagnosis of PE. The studies were not methodologically similar enough to perform a meta-analysis.
OUTCOMES MEASURED: The primary outcome was the presence of a PE.
RESULTS: Compared with the gold standard of pulmonary angiography, the sensitivity of spiral CT ranged from 64% to 93%. If a PE is present, the probability of a positive CT scan is 64% to 93%. That means up to one third of PEs could be missed. The reported specificity ranged from 89% to 100%, which corresponds to a false-positive rate of 0% to 11%. These results are similar to those of another recent systematic review in which the authors reported a sensitivity range of 53% to 100% and a specificity range of 81% to 100%.1 Nine of the studies differentiated between large central and small subsegmental vessel embolism. When stratified by site, the sensitivity for spiral CT was much higher for central vessel PE (83% to 100%) than for subsegmental vessel PE (29%).
A review of the current available literature does not support the use of spiral CT for diagnosing PE. Although it appears that CT is better for identifying larger vessel PEs, the high false-negative rate prohibits its routine use as a rule-out test. In addition, many of the currently available studies employ methods that do not answer questions about the role and cost-effectiveness of spiral CT. More information is needed before we can recommend the routine use of spiral CT for the diagnosis of PE in clinical practice.