The Pulmonary Embolism Rule-Out Criteria without subsequent testing in a low-risk population may not increase the risk of undiagnosed symptomatic thromboembolism
EBM Focus - Volume 13, Issue 8
- The American College of Physicians suggests using the Pulmonary Embolism Rule-Out Criteria (PERC) to rule out pulmonary embolism (PE) in low-risk patients, with subsequent D-dimer testing only in patients who do not meet PERC criteria. However, direct comparison of this PERC-based strategy with a D-dimer testing-based strategy has not been performed.
- The PROPER cluster-randomized noninferiority trial randomized 14 emergency departments to a PERC-based strategy (D-dimer testing only if PE was not ruled out by PERC) vs. the usual diagnostic strategy (D-dimer testing in all patients). The 1,916 included patients had a low risk of PE as estimated by the treating physician’s clinical impression.
- The PERC-based strategy was noninferior to the usual diagnostic strategy in the per-protocol analysis, with diagnostic failure (symptomatic thromboembolism within 3 months in patients for whom PE was ruled out) in 0.1% vs. 0%. However, noninferiority was not met in an intention-to-treat analysis, and the possibility that this trial is underpowered cannot be excluded.
In patients with suspected pulmonary embolism (PE) but with a low clinical probability of PE, the American College of Physicians recommends that PE be ruled out without subsequent D-dimer testing in patients who meet all of the Pulmonary Embolism Rule-Out Criteria (PERC) (age < 50 years old, pulse < 100 bpm, oxygen saturation [SpO2] > 94%, no unilateral leg swelling, no hemoptysis, no surgery or trauma within the past 4 weeks, no previous deep venous thrombosis or PE, and no oral hormone use) (Ann Intern Med 2015). Such patients are thought to be at a low enough risk of PE that the potential benefits of additional testing are outweighed by the risks of a false positive D-dimer test, subsequent computed tomographic pulmonary angiography (CTPA), and false positive PE diagnosis (Ann Intern Med 2015). However, the PERC-based strategy has not been evaluated prospectively, and two retrospective studies suggest that it may lead to unacceptably high rates of false negatives (J Thromb Haemost 2011, 2005). Of note, these two studies were conducted in Europe, which has a higher prevalence of PE than the United States. In addition, the retrospective nature of the two studies and how they selected patients make interpretation of their results unclear.
To directly evaluate the consequences of a PERC-based strategy, the PROPER cluster-randomized noninferiority trial randomized 14 emergency departments (EDs) in France to a PERC-based strategy vs. the usual diagnostic strategy for 6 months. After a 2 month washout, the EDs were crossed over to the alternate strategy for an additional 6 months. The trial included 1,916 patients with new-onset presence or worsening of chest pain or shortness of breath and who were at a low risk of PE based on the treating physician’s clinical impression. In the PERC-based strategy, PE was ruled out if all 8 criteria of the PERC rule were met, and subsequent testing was not conducted. Otherwise, the usual diagnostic strategy was followed, which consisted of D-dimer testing and, if positive, CTPA, with PE being ruled out if either test was negative. PE was diagnosed in the ED in 1.5% of patients following the PERC-based strategy and 2.7% following the usual diagnostic strategy. The primary outcome of this trial was the rate of diagnostic failure, defined as a symptomatic thromboembolic event occurring within 3 months of the ED visit in patients for whom PE was ruled out in the ED. The criterion for noninferiority of the PERC-based strategy was a diagnostic failure rate ≤ 1.5% higher compared to the usual diagnostic strategy at the limit of the 95% CI for difference in the per-protocol analysis.
The per-protocol analysis excluded 167 patients (8.7%) who were lost to follow-up, deviated from the protocol, or wrongly included. Noninferiority of the PERC-based strategy was met: the rate of diagnostic failure was 0.1% (1 patient) with the PERC-based strategy vs. 0% with the usual diagnostic strategy, with 0.8% as the upper limit of the 95% CI for difference in rates. However, noninferiority was not met in an intention-to-treat analysis in which diagnostic failure was assumed in patients with missing data, with rates of 3% vs. 3% and 1.6% as the upper limit of the 95% CI for difference. Analyses of secondary outcomes in the intention-to-treat population favored the PERC-based strategy, with CTPA being performed during the ED visit in 13% vs. 23% (p < 0.001) and the median duration of the ED visit 4.6 vs. 5.2 hours (p < 0.001). All-cause mortality was similar between the two groups (0.3% vs. 0.2%, not significant).
The patient sample size of this trial was based on an estimated diagnostic failure rate of 1.5% among patients with usual diagnostic strategy, but the lower rate in the per-protocol analysis (0%) leaves open the possibility that the trial is underpowered to detect a difference in rates. The low event rate, and the wide confidence interval for difference in rates in the intention-to-treat analysis, add some uncertainty to the results. Also, it is important to emphasize that the prevalence of PE was not high in this patient population (2.7% who had the usual diagnostic strategy) and that only patients who were first estimated by the physician’s clinical impression to have a low risk of PE were included. The results of this trial do not extend to other situations. Nevertheless, this randomized trial provides prospective evidence that following a PERC-based strategy may not increase the risk of undiagnosed symptomatic thromboembolic events compared to a D-dimer testing-based strategy in patients with a low risk of PE.