CT Angiography and Clinician Behavior in Suspected Pulmonary Embolism

Abstract & Commentary

Synopsis: In this retrospective comparison of patients suspected of having pulmonary thromboembolism who underwent V/Q scanning vs. CT angiography, subsequent management was more likely to be concordant with the study findings for the latter group, perhaps reflecting greater clinician confidence in the newer procedure.

Source: Crawford T, et al. The effect of imaging modality on patient management in the evaluation of pulmonary thromboembolism. J Thoracic Imaging. 2001;16:163-169.

Crawford and associates from ucla sought to determine how the results of pulmonary CT angiography (CTA) were related to subsequent management of patients suspected of having pulmonary thromboembolism (PE). They reviewed the charts of 138 consecutive patients who underwent ventilation-perfusion (V/Q) scanning for investigation of suspected PE during a 6-month period in 1994-1995. They compared the findings with data from 149 patients in whom CTA was used for the same indication during the same calendar months in 1996-1997, after the latter procedure had largely replaced V/Q scanning for this indication at Crawford et al’s institution. CTA interpretations were made according to a fixed protocol and required the agreement of 2 thoracic radiologists; how the V/Q scans were read is not stated.

From the information recorded in the patients’ charts, the patients who underwent CTA for suspected PE had tachypnea, auscultatory abnormalities, and normal ECGs more often than those undergoing V/Q scanning, while the latter group had a higher incidence of recent surgery and right heart strain by ECG (all, P < 0.05). Crawford et al do not state how many of the 287 patients in this report were in the ICU or on the wards, nor do they provide in-hospital mortality or other outcome data aside from the duration of hospitalization following the index diagnostic procedure (7 vs 9 days for V/Q scans and CTA, respectively).

The V/Q scans were interpreted as normal to low probability for PE in 64% of patients, intermediate/indeterminate in 25%, and high probability in 11%. Corresponding results for CTA were negative in 81%, indeterminate in 5%, and positive in 15%. The diagnosis of PE was excluded on the basis of a negative CTA in 79 patients (53%) without further investigation. The corresponding numbers for patients with normal or low-probability V/Q scans were 56% and 41%. All 22 patients whose CTA results were read as positive were treated for PE without further study. Additional studies in patients with intermediate or indeterminate results on the index study were done in 6 CTA patients and 20 V/Q scan patients. Only 3% of CTA patients and 4% of V/Q patients underwent catheter angiography.

Clinicians subsequently managed patients with respect to PE in concordance with the study procedure’s findings 69% of the time with V/Q scanning and 94% of the time with CTA. Crawford et al conclude that pulmonary CTA results are more directly reflective of physician management, are less frequently indeterminate, and reduce the proportion of patients in whom management is initiated despite inconclusive diagnostic information.


Few diagnostic procedures have so taken the medical world by storm as CTA for suspected PE during the last 5 or 6 years. Two big reasons for this are the notorious ambiguity of V/Q scan results, particularly when the patient has underlying lung disease, and the ease with which clinicians can obtain a CTA at any time of the day or night—often in stark contrast to the situation with V/Q scans. Laying aside for a moment the issue of whether the present study contributes anything in the ongoing debate about CTA’s diagnostic accuracy, Crawford et al demonstrate that clinicians have embraced the procedure and tend to act on its results without seeking further confirmation. In particular, when CTA was negative, the doctors managing the patients concluded that PE had been excluded, and neither anticoagulated them nor pursued further workup for PE.

For telling us anything beyond that, there are several major problems with this study. One is its retrospective, open design, and the challenge of obtaining unbiased data in one of the most controversial areas of clinical medicine. This problem is compounded by the 2-year interval between sampling periods, which strains the credibility of the implicit assumption that the diagnostic investigation of first choice was the only thing different in the patients’ management. Crawford et al’s finding that there were multiple significant differences between the clinical characteristics of 2 patient groups reinforces this concern.

Perhaps more worrisome, however, is the lack of use of a reference standard for the diagnosis of PE. Fewer than 4% of the patients underwent traditional pulmonary angiography, and only a minority had Doppler ultrasonography or any other procedure to detect thromboembolism. Thus, the fact is that we have no way of really knowing what the patients in this study actually had—and especially how many of those who were not treated for PE because their CTAs did not show clots actually had PE. All that we can say for sure is that the physicians at Crawford et al’s institution appear to have placed greater reliance in the results of CTA than in the findings of V/Q scanning. Whether that greater reliance is justified remains to be seen.

The results of a recently published study from Geneva underscore the importance of not placing too much faith in a negative CTA.1 In that study, Perrier and colleagues prospectively evaluated 299 patients suspected of having PE, using a validated algorithm that included clinical assessment, D-dimer levels, V/Q scans, and pulmonary angiography. Each patient also had CTA but this was not read until 3 months later and thus was not used in making patient management decisions. Of the 299 patients, 118 (39%) had proven PE. Sensitivity and specificity of CTA for the diagnosis of PE were 70% (95% CI, 62-78%) and 91% (95% CI, 87-95%), respectively. The false-negative rate for CTA when used as the primary determinant of PE was 30%—meaning that 30% of the patients whose CTA results would have "excluded" PE in the Crawford study actually had it. Perrier et al echo the conclusions of other investigators in recommending that CTA not be used as the sole diagnostic study for suspected PE.

Considering these findings in the context of the study by Crawford et al discussed above, the diagnosis of PE should not be ruled out solely on the basis of a negative CTA, particularly in patients with moderate-to-high clinical suspicion of PE.


1. Perrier A, et al. Performance of helical computed tomography in unselected outpatients with suspected pulmonary embolism. Ann Intern Med. 2001;135(2):88-97.