|Pulmonary embolism rule-out criteria (PERC) in pulmonary embolism — revisited: a systematic review and meta-analysis
|Singh B, Mommer SK, Patricia EJ, Mascarenhas SS, Parsaik AK
This review concluded that because of their high sensitivity and low negative likelihood ratio, the pulmonary embolism rule-out criteria could be used confidently in settings where patients have low clinical probability. Given the potential overestimation of accuracy, in the included studies, and the calculation of pooled estimates of accuracy in separate analyses, the conclusions should be treated with caution.
To determine the diagnostic performance of pulmonary embolism rule-out criteria (PERC), to rule out pulmonary embolism, without the need for D-dimer testing, in the emergency department.
EMBASE, MEDLINE, MEDLINE In-Process and Other Non-Indexed Citations, PsycINFO, Scopus, Web of Knowledge, Cochrane Central Register of Controlled Trials (CENTRAL), and Cochrane Database of Systematic Reviews, were searched, without language restrictions, to May 2012; the search strategy was reported. Bibliographies of retrieved articles, and conference proceedings, from the previous two years, from two relevant emergency medicine organisations, were searched.
Studies that evaluated the diagnostic performance of PERC to rule out pulmonary embolism, and were set in the emergency department, were eligible for inclusion. The primary outcome was the diagnosis of pulmonary embolism, or venous thromboembolism or death caused by venous thromboembolism within 90 days of initial emergency department evaluation.
Across the included studies, the mean age was 56.7 years, and 63% of participants were female. The prevalence of pulmonary embolism ranged from 1.74% to 29.8%. The selection criteria and the prior tests conducted on recruited patients, varied considerably across studies. Most studies used a ventilation or perfusion scan, computed tomography, or both to determine the presence of pulmonary embolism.
Two reviewers independently selected studies for inclusion; disagreements were resolved by consensus.
Assessment of study quality
Two reviewers independently assessed study quality, using a checklist based on published tools, including QUADAS. The criteria covered: patient selection; patient spectrum; blinding of the assessors of the predictor variables; blinding of outcome assessors; accuracy of the definition of outcomes; loss-to-follow-up (<10%); and explicit interpretation of the risk score by clinicians, who were blind to the outcome. Disagreements were resolved by consensus.
Two reviewers independently extracted the data to produce 2x2 tables of test performance, which were used to calculate the sensitivity, specificity, and positive and negative likelihood ratios, with 95% confidence intervals. Disagreements were resolved by consensus.
Methods of synthesis
Pooled estimates of sensitivity, specificity, positive and negative likelihood ratios, and diagnostic odds ratios, with 95% confidence intervals, were calculated using a random-effects meta-analysis. Heterogeneity was assessed using Ι²; less than 25% was minimal, 25 to 50% was moderate, and over 50% was substantial.
Studies reported only as abstracts, and those scoring less than 50% on the methodology checklist, were excluded from the quantitative synthesis. Subgroup analysis and the Moses-Littenberg linear meta-regression model were used to investigate the impact of the prevalence of pulmonary embolism.
Results of the review
Twelve studies, with 13 data sets, met the inclusion criteria (14,844 patients); nine were prospective and three were retrospective. The quality results were reported for nine studies (10 data sets), all of which used consecutive or random selection, blinded assessors of the predictor variables and reference standard tests, predefined the outcomes, and had adequate follow-up. Seven studies recruited a representative sample; none explicitly stated that the interpreters of the PERC were blinded to the results of the other tests or the outcomes.
For PERC alone (13 data sets), the pooled sensitivity was 97% (95% CI 96 to 98; Ι²=23.7%), the specificity was 22% (95% CI 22 to 23; Ι²=97.4%), the positive likelihood ratio was 1.22 (95% CI 1.16 to 1.29; Ι²=89.1%), and the negative likelihood ratio was 0.17 (95% CI 0.13 to 0.23; Ι²=0). For PERC combined with clinical gestalt probability (one data set), there were small clinically insignificant reductions in the pooled specificity and likelihood ratios.
The pooled diagnostic odds ratio was 7.4 (95% CI 5.5 to 9.8). The overall proportion of missed pulmonary embolism was 0.3% (48 of 14,844 cases) with PERC alone, and 0.296% (44 of 14,844) with PERC with clinical gestalt. The estimates of diagnostic accuracy were similar for high (four data sets) and for low (nine data sets) prevalence groups.
Because of their high sensitivity and low negative likelihood ratio, the PERC could be used confidently in settings where patients have a low clinical probability of pulmonary embolism.
The review addressed a clear question and used reproducible inclusion criteria. The search was extensive for published studies, and attempts were made to reduce any language and publication bias. Diagnostic filters were included in the search strategy, which may have resulted in some studies being missed. Each stage of the review process was conducted by two people, reducing the risk of error and bias. Appropriate criteria were used to assess study quality, and the results were given, in full, but only for nine of the studies.
Pooled estimates of sensitivity and specificity were derived separately, using standard frequentist meta-analysis. This means that the within-study relationship between sensitivity and specificity was not retained. The reliability and generalisability of the estimates, derived in this way, for such clinically different studies, is uncertain. More robust analytic methods were available to derive these estimates.
Given the uncertainty over whether the interpreters of the PERC were blind to the results of other tests or outcomes (which could overestimate accuracy), the clinical variation between studies, and the separate analyses for sensitivity and specificity, the conclusions should be treated with caution.
Implications of the review for practice and research
Practice: The authors stated that the PERC could avoid expensive diagnostic imaging that frequently occurred with positive D-dimer results.
Research: The authors stated that the validation of the PERC with gestalt clinical probability, in various settings, with large numbers of patients, would be useful.
Singh B, Mommer SK, Patricia EJ, Mascarenhas SS, Parsaik AK. Pulmonary embolism rule-out criteria (PERC) in pulmonary embolism — revisited: a systematic review and meta-analysis Emergency Medicine Journal 2012: doi:10.1136/emermed-2012-201730
Other publications of related interest
Singh B, Parsaik AK, Agarwal D, Surana A, Mascarenhas SS, Chandra S. Diagnostic accuracy of pulmonary embolism rule-out criteria: a systematic review and meta-analysis. Annals of Emergency Medicine 2012; 59(6): 517-520.e4.
Subject indexing assigned by CRD
Diagnosis, Differential; Pulmonary Embolism; Humans
Date bibliographic record published
Date abstract record published
This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.