| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Medical University of South Carolina Charleston, SC 29425
We read with interest the article by Eng et al. [1] titled "Accuracy of CT in the Diagnosis of Pulmonary Embolism: A Systematic Literature Review." Unfortunately, the review provides little new information from prior systematic reviews comparing conventional angiography with CT and we believe that it sends the incorrect message. In light of current technical developments since the introduction of MDCT, it seems inappropriate to pool recent data with results that were generated in the early 1990s, shortly after the inception of helical CT. More importantly, the comparison with conventional angiography, while on the surface desirable, is probably not the best reference standard.
Conventional angiography established itself as a gold standard on the basis of negative outcome trials—that is, if a conventional pulmonary angiogram was interpreted as showing no evidence of pulmonary embolism, there was a negative predictive value of at least 98% for future thromboembolic events [2]. Unfortunately, this result is not the same as a true-negative for the absence of pulmonary embolism. Therefore, even if it was performed, pulmonary angiography should not necessarily be used as an accurate and objective test for the verification of findings from other imaging techniques.
Animal experiments that use artificial emboli as an independent gold standard indicate that 4-MDCT is at least as accurate as invasive pulmonary angiography for the detection of pulmonary emboli [3]. Furthermore, conventional angiography has limited accuracy and interobserver agreement for detection of subsegmental emboli [4, 5], one of the main criticisms of CT.
Given the limited interobserver correlation of pulmonary angiography for subsegmental emboli, it appears doubtful that conventional pulmonary angiography, even if it were performed, would provide useful and adequate correlative proof for findings at MDCT. Finally, conventional angiography is not frequently performed anymore for suspected acute pulmonary embolism. This practice trend means that radiologists interpreting the examination now have limited experience, particularly in contrast to CT, and the test characteristics establishing it as a gold standard may no longer be valid.
We believe, therefore, that negative outcomes studies are the most appropriate means of evaluating the efficacy of CT for the exclusion of pulmonary embolism. Although it has the disadvantages as noted for conventional angiography, CT pulmonary angiography provides appropriate clinical validity to the test when interpreted as negative. A large number of studies, in fact, confirm the high negative predictive value of CT with both single-detector CT and MDCT.
When summed together, as with conventional pulmonary angiography, data indicate that the negative predictive value for CT pulmonary angiography exceeds 98% [6]. The fact that the rate for clinically evident thromboembolic events within 3 months of a negative CT study is the same as for conventional angiography provides compelling evidence that the test characteristics are more similar than a direct comparison provides and should establish without question that CT is in fact the most appropriate current standard for the detection and exclusion of pulmonary embolus.
 Top References References |
|---|
Johns Hopkins University School of Medicine Baltimore, MD 21287
In conducting our systematic review [1], we aimed to summarize the best available evidence regarding the diagnostic accuracy of CT in the diagnosis of pulmonary embolism. We believed further review was necessary because of the significant improvements in CT technology that have occurred since CT was first studied for the diagnosis of pulmonary embolism [2]. Previous reviews have reported a wide range of sensitivities, which we thought may have been the result of including studies with significant methodologic flaws. Therefore, we used more stringent inclusion criteria, including one that required uniform application of conventional pulmonary angiography as the reference standard.
Our review of the evidence showed no clear upward trend in accuracy over the reviewed time period, despite the simultaneous improvements in single-detector scanner technology that have occurred. Although we would have liked to include MDCT scanners in our review, we identified no MDCT studies that met the inclusion criteria for either primary studies or systematic reviews. However, lack of evidence concerning MDCT is not the same as lack of value. A multicenter clinical trial, the Prospective Investigation of Pulmonary Embolism Diagnosis II (PIOPED II), was designed to assess the efficacy of MDCT, and we await publication of that trial's results [3].
We acknowledge that conventional pulmonary angiography may well be an imperfect reference standard for a number of reasons. However, a standard—even if imperfect—must be chosen to perform a meaningful review. Conventional angiography is a reasonable choice because many good diagnostic studies use it as a reference standard. It is also the diagnostic reference test that comes to the minds of most clinicians when they consider the diagnosis of pulmonary embolism.
In the care of patients, clinical outcome is certainly the primary concern and supersedes intermediate outcomes such as test accuracy. In pursuit of this principle, a number of published studies have reported very high clinical negative predictive values for CT in patients suspected of having pulmonary embolism [4–7]. In these studies, if the CT findings were negative for pulmonary embolism and anticoagulation was not prescribed, clinical follow-up identified low probability of deep venous thromboembolism. These results appear to make a compelling case for the use of CT in ruling out pulmonary embolism.
However, the actual evidence may not allow us to attribute this high negative predictive value to CT alone. In each of these studies [4–7], for example, the decision to withhold anticoagulation therapy was not based on the CT results alone. Many, if not all, of the participants in each study underwent additional diagnostic evaluation, including lower extremity Doppler sonography, ventilation–perfusion scintigraphy, conventional pulmonary angiography, serum D-dimer testing, or formal scoring of clinical suspicion. The involvement of additional diagnostic tests could have inflated the apparent negative predictive value of CT by removing false-negative cases from the study population. Nevertheless, the good news is that a combination of clinical judgment and diagnostic testing, including CT, is effective in ruling out pulmonary embolism.
Even if it turns out that CT alone has a high clinical negative predictive value, this may not tell the whole story. Some authors have proposed that the reason CT has a high apparent negative predictive value—even though many reviews have reported variable or inadequate sensitivity—is that small peripheral emboli missed on CT have no clinical consequence [8]. As CT technology continues to improve, its ability to detect these clinically insignificant emboli may concomitantly improve. If such patients are placed on anticoagulation therapy, the clinical negative predictive value would be unaffected, but the clinical positive predictive value—and therefore the overall clinical accuracy—would be diminished because such patients would be receiving clinically unnecessary treatment.
Although continuing improvements in imaging technology always hold the promise of improved diagnostic accuracy and patient outcomes, we hope these improvements will continue to be accompanied by well-designed studies and systematic reviews.
|
TopReferencesReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
