AFP Journal Club
The Story Behind the Study
The Role of Chest CT in Diagnosing Pulmonary Embolism
Am Fam Physician. 2008 Sep 1;78(5):554-556.
Each month, our three presenters review an interesting journal article in a conversational manner. These articles involve “hot topics” that affect family physicians or “bust” commonly held medical myths. The presenters give their opinions about the clinical value of the studies discussed. The opinions reflect the views of the presenters, not those of AFP or the AAFP.
This Month's Article
Stein PD, Fowler SE, Goodman LR, et al., for the PIOPED II Investigators. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med. 2006;354(22):2317–2327.
How good is chest CT at identifying PE?
Bob: Pulmonary embolism (PE) accounts for 300,000 deaths per year. Because the signs and symptoms are inconsistent, the diagnosis is often missed. In the hope of catching this elusive diagnosis, many physicians have turned to computed tomography (CT). Some physicians have become so enamored with this technology that they now indiscriminately order chest CTs for every case of pleuritic chest pain. The phrase “just spin them” echoes in many hospital departments, suggesting that when in doubt about whether a PE is present, a spiral chest CT should be obtained. But before ordering a CT on every patient with the slightest suggestion of PE, two questions need to be asked: (1) how good is chest CT at identifying PE?; and (2) how often is the CT positive when, in reality, there is no PE?
What does this article say?
Bob: This study was performed at eight different centers and enrolled adult patients with suspected PE. Overall, 824 patients underwent all of the following tests: CT angiography of the chest and CT venography of the abdomen, pelvis, and lower extremities, using a four-, eight-, or 16-row scanner; a ventilation/perfusion scan; ultrasonography of the lower extremities; and if necessary, a digital subtraction angiography. Chest CT was then compared with a composite reference standard, which was a positive result on any one of the other tests. If the CT and reference standards were all negative, follow-up at three and six months was performed to confirm that no thromboembolic event had occurred.
How did the chest CT compare with the reference standard? Of 824 patients, 51 (6 percent) had a CT that could not be interpreted. Of the remaining 773 patients, chest CT identified 150 of 181 patients with PE. That is a sensitivity of 83 percent (95% confidence interval [CI], 76 to 92 percent).
Andrea: A sensitivity of 83 percent? That means that chest CT misses one in six cases of PE!
Mark: And with a 95% CI of 76 to 92 percent, the sensitivity could be as low as 76 percent, meaning the CT could miss nearly one in four cases.
Bob: The authors noted that sensitivity could be improved to 90 percent if CT venography was performed in every patient in addition to CT of the chest.
Andrea: That is a 7 percent gain in sensitivity, but at what cost? Not just financial, but also to the long-term health of my patients. If CT venography were to be added in every patient, just imagine how many women's pelvises and ovaries would be needlessly irradiated. So, the sensitivity is not so great—what about the specificity?
Bob: Chest CT correctly noted that no PEs were present in 567 of 592 cases (96 percent specificity). This means that 4 percent of the cases are false-positives.
Mark: In clinical terms, chest CT results in one of 25 patients being given an incorrect diagnosis of PE and getting placed on long-term anticoagulation when no PE is present. Is there any silver lining?
Bob: I'm not sure there is a silver lining, but there is an important concept illustrated in this study. The researchers assigned a pretest probability of PE (low, indeterminate, or high) to every patient based on the Wells Criteria (Table 1). They found that when the chest CT result was concordant with the pretest probability, the CT often confirmed the clinical suspicion. However, when the chest CT result was opposite the clinical suspicion (discordant), the CT result was unreliable.
In patients with a high risk of PE and a positive chest CT for PE (i.e., the clinical impression and test are concordant), 96 percent of the CT results are true-positives. However, if the clinical suspicion is high, but the CT is negative, the chest CT is wrong (i.e., it misses the PE) 40 percent of the time. In this situation, do not abandon your clinical impression—the patient needs further evaluation.
Table 1 Model for Determining the Clinical Probability of PE According to the Wells Score
Model for Determining the Clinical Probability of PE According to the Wells Score
Clinical signs and symptoms of DVT (objectively measured leg swelling and pain with palpation in the deep-vein system)
Heart rate more than 100 beats per minute
Immobilization for at least three consecutive days (bed rest except to go to the bathroom) or surgery in previous four weeks
Previous objectively diagnosed PE or DVT
Cancer (with treatment within past six months or palliative treatment)
PE likely or more likely than alternative diagnoses (on the basis of history, physical examination, chest radiography, electrocardiography, and blood tests)
note: A score of less than 2.0 = low probability; 2.0 to 6.0 = moderate probability; and more than 6.0 = high probability.
DVT = deep venous thrombosis; PE = pulmonary embolism.
Adapted with permission from Stein PD, Fowler SE, Goodman LR, et al., For the PIOPED II Investigators. Multidetector computed tomography For acute pulmonary embolism. N Engl J Med. 2006;354(22):2319.
At the other end of the clinical spectrum, in patients with a low probability of PE, but a positive chest CT, the CT result is false-positive 42 percent of the time. Blindly accepting a positive CT result in the case of low clinical suspicion could lead to more than 40 percent of these patients being needlessly treated with anticoagulants.
Mark: This is a great example of Bayesian analysis. How well a test performs is dependent on the population you are testing. In this study, chest CT is a great test if the clinical probability of disease is high (good sensitivity, good specificity), but there are a lot of false-positive results (42 percent) in patients who are low risk.
Bottom line: chest CT for PE is useful when it is concordant with your clinical impression (essentially confirming your clinical impression) or when you have an indeterminate or intermediate risk of disease (then the test can point you in the right direction). But if a chest CT result is opposite your clinical impression, you cannot rely on it; further testing is needed.
Should we believe this study?
Bob: I do. Some have argued that pulmonary angiography, the accepted confirmatory test for PE, was not used in this study. Because pulmonary angiograms are invasive, using this test would have been very difficult and dangerous, and would likely not have passed investigational review boards. Some believe that even pulmonary angiography is not the perfect test. I think the authors' composite reference standard was about as good as they could do.
Mark: One other point—most of the chest CTs performed in this study used four-slice CT machines. Many institutions are now employing eight-, 16- or 64-slice CT machines. It is possible that the accuracy will improve with the newer-generation machines.
What should the family physician do?
Bob: Be careful when ordering this or any test. Although the chest CT can be useful when you are uncertain of the diagnosis, at times it can lead you astray.
Andrea: Because of the radiation-associated risks, CTs should not be ordered indiscriminately. A recent article noted that 1.5 to 2.0 percent of all cancers in the United States may be attributable to the radiation from CT.1
Mark: And if your clinical judgment and the test results are at odds, do not assume the test is right; sometimes your clinical skills are better than a test.
Chest CT used to diagnose the presence of PE may have a higher false-negative or false-positive rate than previously suspected.
When the chest CT result is inconsistent with the clinical probability, additional testing is warranted.
Physicians must consider the potential long-term cancer risks associated with CT when ordering this test.
All tests follow a Bayesian model. They are more likely to be true-positive in a sick patient and false-positive in a well patient.
Sensitivity and specificity only tell part of the story. Always look at the false-positive and false-negative rates of a test.
False-positive studies: how many patients who really don't have the disease are you treating simply because the test was positive? In the case of the low-risk patients in this study, 42 percent had a positive chest CT, but no evidence of PE on any confirmatory test. If the chest CT result had been accepted, a significant number of these patients would have been needlessly exposed to anticoagulation.
False-negative studies: how many patients with true disease are you missing and not treating because the test was negative?
1. Brenner DJ, Hall EJ. Computed tomography—an increasing source of radiation exposure. N Engl J Med. 2007;357(22):2277–2284.
Copyright © 2008 by the American Academy of Family Physicians.
This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP. Contact firstname.lastname@example.org for copyright questions and/or permission requests.
Want to use this article elsewhere? Get Permissions