Correspondence
Computed Tomography for Pulmonary Embolism
N Engl J Med 2006; 355:955-956August 31, 2006
- Article
To the Editor:
In an important recent study, the Prospective Investigation of Pulmonary Embolism Diagnosis II (PIOPED II) trial, Stein et al. (June 1 issue)1 found that multidetector computed tomographic angiography (CTA) had excellent specificity and good sensitivity for the detection of pulmonary embolism. The PIOPED II trial also replicates the findings of the PIOPED I trial2 (which used ventilation–perfusion scanning as the imaging technique) that the a priori — that is, before imaging — clinical suspicion of whether or not a patient has a pulmonary embolism is crucial in making the diagnosis. But one issue I have not seen addressed is, whose clinical suspicion? Consider a patient seen in the month of July. Can the diagnosis of pulmonary embolism be made on the basis of the clinical suspicion of a new medical intern who has been a physician for only days or hours or of a resident similarly just removed from internship? For a patient at a rehabilitation facility, is a physiatrist's suspicion sufficient? Or in these cases is the suspicion of a board-certified internist or pulmonologist necessary?
2 ReferencesEric L. Altschuler, M.D., Ph.D.
University of Medicine and Dentistry of New Jersey, Newark, NJ 07101
eric.altschuler@umdnj. edu 1
Stein PD ,Fowler SE ,Goodman LR , et al. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med 2006;354:2317-2327
Full Text | Web of Science | Medline2
The PIOPED Investigators
CrossRef | Web of Science
To the Editor:
The PIOPED II study confirms that helical computed tomography (CT) can help diagnose or rule out pulmonary embolism when the results are concordant with clinical suspicion. However, the study's generalizability may be limited in that the scan interpretations were determined by the consensus of two expert readers — a setup that may not mirror real-world practice. Perhaps the authors can provide some measure of interreader agreement (or even a sensitivity analysis that defines the CT result on the basis of the interpretation of a single radiologist as opposed to the final consensus interpretation)?
Whereas individual radiologists will have varying interpretive accuracy, which may or may not correlate with their degree of confidence, quantifying interrater agreement in the PIOPED II trial may clarify the extent to which interpreting helical CT scans remains an art, even in the best of hands, particularly when one is attempting to detect or exclude smaller emboli.
Daniel J. Brotman, M.D.
Johns Hopkins Hospital, Baltimore, MD 21287
brotman@jhmi.edu Author/Editor Response
As Altschuler indicates, an assessment of the clinical probability of pulmonary embolism is important. In the PIOPED I trial, the assessment of clinical probability was based on the physician's judgment.1 In the PIOPED II trial, an objective assessment of clinical probability based on the Wells score was used,2 but other methods for objective clinical assessment could be used.3,4 Objective assessment may be more robust when applied by nonexperts.
Regarding Brotman's questions, for readers of CTA, the single unweighted kappa statistic5 for agreement between the first two readers was 0.73 (95 percent confidence interval, 0.68 to 0.78). For CT venous-phase imaging (CTV), the kappa statistic for acute deep venous thrombosis was 0.73 (95 percent confidence interval, 0.67 to 0.78). For digital subtraction angiography, the kappa statistic was 0.66 (95 percent confidence interval, 0.54 to 0.78), and for ventilation–perfusion lung scans, the kappa statistic for agreement across three categories (high, nondiagnostic, or very low or normal) was 0.54 (95 percent confidence interval, 0.48 to 0.59). Thus, there was substantial interobserver agreement with CT.
To determine whether having two readers influenced our results, we looked at the sensitivity and specificity of the first central reader of scans from each patient. The results were similar to the results of the consensus readings. The sensitivity of CTA based on interpretations of the first central reader was 80 percent, and the specificity was 95 percent. The sensitivity of CTA combined with CTV on the basis of interpretations of the first central reader was 89 percent, and the specificity was 94 percent.
The PIOPED II trial did not assess the performance of nonsubspecialist radiologists. Art and skill, we agree, play an important role. We believe, however, that many expert clinical radiologists outside of academic centers can achieve high accuracy in CT interpretation.
5 ReferencesPaul D. Stein, M.D.
St. Joseph Mercy Oakland Hospital, Pontiac, MI 48341-5023
steinp@trinity-health.org Lawrence R. Goodman, M.D.
Medical College of Wisconsin, Milwaukee, WI 53226-3596H. Dirk Sostman, M.D.
Methodist Hospital, Houston, TX 770301
The PIOPED Investigators
CrossRef | Web of Science2
Wells PS ,Anderson DR ,Rodger M , et al. Excluding pulmonary embolism at the bedside without diagnostic imaging: management of patients with suspected pulmonary embolism presenting to the emergency department by using a simple clinical model and d-dimer. Ann Intern Med 2001;135:98-107
Web of Science | Medline3
Wicki J ,Perneger TV ,Junod AF ,Bounameaux H ,Perrier A . Assessing clinical probability of pulmonary embolism in the emergency ward: a simple score. Arch Intern Med 2001;161:92-97
CrossRef | Web of Science | Medline4
Le Gal G ,Righini M ,Roy PM , et al. Prediction of pulmonary embolism in the emergency department: the revised Geneva score. Ann Intern Med 2006;144:165-171
Web of Science | Medline5
Landis RJ ,Koch GG . The measurement of observer agreement for categorical data. Biometrics 1977;33:159-174
CrossRef | Web of Science | Medline
