Emergency Medicine · Medical Literature
Diagnostic Validation of PERC and Resource Utilization in Suspected Acute Pulmonary Embolism: A Systematic Review and Meta-Analysis.
Abstract
Background and AimsPulmonary embolism (PE) is a frequent and potentially fatal diagnosis in the emergency department (ED). The Pulmonary Embolism Rule-out Criteria (PERC), an eight-item clinical decision rule used to identify patients at very low risk of pulmonary embolism, is widely applied to exclude PE in low-risk patients; however, its real-world diagnostic performance and impact on imaging utilization remain incompletely defined. This systematic review aimed to evaluate the diagnostic accuracy of PERC and its effect on reducing computed tomography pulmonary angiography (CTPA) use.Materials and MethodsThis review followed the Joanna Briggs Institute (JBI) methodology for diagnostic test accuracy studies. Eligible studies enrolled ED patients with suspected PE who were classified as low risk according to the original study-specific approaches, including clinician gestalt, Wells score, revised Geneva score, or YEARS-based assessment. PubMed, Web of Science Core Collection, the Cochrane Library, and ClinicalTrials.gov were searched for eligible studies published between 2004 and 2025.ResultsTen studies involving 13,672 patients were included. The pooled PE prevalence was 7%. The pooled sensitivity, specificity, and negative predictive value of PERC were 95% (95% CI: 89%-98%), 26% (95% CI: 16%-40%), and 98.2% (95% CI: 97.8%-99.0%), respectively. PERC application was associated with a significant reduction in imaging utilization (RR = 0.85, 95% CI: 0.80-0.91).ConclusionsPERC provides a highly sensitive and safe strategy for ruling out PE in low-risk ED patients and significantly reduces unnecessary use of advanced imaging without compromising diagnostic safety.
Clinical relevance
In emergency department patients with suspected pulmonary embolism who are pre-stratified as low risk, what is the pooled diagnostic accuracy of the Pulmonary Embolism Rule-out Criteria (PERC) and what is its effect on reducing computed tomography pulmonary angiography (CTPA) utilization compared with other clinical decision rules or D-dimer–based strategies? __SNAPSHOT__ {"studyDesign":"Systematic review and meta-analysis of diagnostic test accuracy studies, conducted per Joanna Briggs Institute (JBI) methodology","setting":"Emergency departments; multinational; studies published 2004–2025; geographic breakdown of included studies not reported in available abstract/excerpt","sampleSize":"n = 13,672 total patients across 10 included studies; 10,307 underwent PERC evaluation; 8 studies contributed to the diagnostic accuracy meta-analysis and 2 additional studies contributed only to the CTPA utilization analysis","population":"Adult ED patients with suspected acute pulmonary embolism pre-classified as low risk via study-specific approaches: clinician gestalt, Wells score, revised Geneva score, or YEARS-based assessment; pooled PE prevalence 7%","intervention":"Application of the eight-item PERC rule as a rule-out clinical decision instrument","comparator":"Other clinical decision rules (CDRs) or conventional D-dimer–based diagnostic strategies (for CTPA utilization analysis); CTPA as reference standard for diagnostic accuracy","primaryOutcome":"Pooled diagnostic sensitivity and specificity of PERC; pooled missed-diagnosis rate among PERC-negative patients; reduction in CTPA utilization expressed as pooled risk ratio (RR)","followUp":"Not reported","fundingSource":"Not reported in available abstract or partial text"}
Key findings
Ten studies (n = 13,672) were included, with a pooled PE prevalence of 7%. The pooled sensitivity of PERC was approximately 95% — consistent with the 97% sensitivity reported in a prior Singh et al. meta-analysis, supporting cross-cohort stability of rule-out performance. The pooled negative likelihood ratio (NLR) was approximately 0.19, indicating that a negative PERC result meaningfully reduces post-test PE probability. Specificity was persistently low (exact pooled value not reported in the available abstract/excerpt), a finding the authors attribute to the inherently stringent structure of the eight-item criteria. The pooled missed-diagnosis rate among PERC-negative patients was 2.0% (95% CI: 1.0%–3.0%), which exceeds the widely cited ≤1.8% acceptable miss-rate threshold used in PE rule-out validation literature and is higher than the 0.3% aggregate missed-diagnosis rate reported in the 2012 Singh et al. meta-analysis (14,844 patients); reasons for this difference may include population heterogeneity, pre-test probability variation across sites, and methodologic differences between reviews. CTPA utilization analysis was expressed as pooled risk ratios (specific RR values and confidence intervals not reported in the available abstract/excerpt). Subgroup analysis compared PERC used in combination with Wells criteria (Group 1) versus PERC used as a standalone rule (Group 2). Publication bias assessment by funnel plot and formal testing showed no statistically significant bias (P > 0.1), though slight bilateral asymmetry was noted visually. NNT/NNH: not calculable from available data. Adverse event data: not reported.
Practice implications
This meta-analysis provides updated pooled evidence that PERC maintains a clinically meaningful rule-out capability (sensitivity ~95%, NLR ~0.19) in ED patients pre-stratified as low risk for PE, supporting its continued utility as a bedside pre-testing filter. However, three important limitations constrain direct practice application: (1) Only the abstract and partial discussion text were available for review; full sensitivity, specificity, and CTPA risk-ratio data with confidence intervals could not be independently verified, per this summary's limitation disclosure. (2) The pooled missed-diagnosis rate of 2.0% (95% CI: 1.0%–3.0%) among PERC-negative patients is clinically non-trivial and exceeds the miss rates reported in some earlier meta-analyses; clinicians should recognize that PERC is not a zero-miss instrument, and its safety depends critically on correct pre-test stratification (low clinical probability must be established before PERC is applied). (3) The authors explicitly note that high-quality validation data in Chinese and other Asian populations remain limited, raising generalizability concerns for non-Western practice settings. Patients not adequately represented include those with intermediate or high pre-test probability, elderly patients (in whom PERC performance may differ), patients presenting during periods of elevated PE base rates (e.g., COVID-19 pandemic), and non-Western ED populations. The rule's persistently low specificity means a substantial proportion of low-risk patients will remain PERC-positive and still require further D-dimer or imaging workup. Practice should not change based on this review alone; it is confirmatory of existing evidence and reinforces that PERC is best used as intended — within a structured two-step algorithm (pre-test probability first, then PERC) — rather than as a universal triage screen. __LIMITATIONS__ Only abstract and partial discussion text available for independent review; full diagnostic accuracy estimates (pooled specificity, CTPA RR with CIs, subgroup-specific values) could not be independently verified — a critical limitation for this clinical summary.,Small number of included studies (n = 10); only 8 contributed to the primary diagnostic accuracy meta-analysis, limiting statistical power and precision of pooled estimates.,Heterogeneity in pre-test probability stratification methods across studies (clinician gestalt, Wells, Geneva, YEARS) introduces significant clinical and methodological heterogeneity, potentially distorting pooled accuracy estimates.,Low specificity of PERC is a structural limitation of the rule itself and was consistently observed; this limits its utility for ruling in PE or reducing workup in PERC-positive patients.,Pooled PE prevalence of 7% is higher than the ≤2% threshold typically cited as the intended pre-PERC population base rate, suggesting some included cohorts may not represent the intended low-risk population.,Missed-diagnosis rate of 2.0% (95% CI: 1.0%–3.0%) among PERC-negative patients may exceed acceptable safety thresholds depending on institutional standards and jurisdiction.,Geographically limited validation data; the authors explicitly acknowledge limited high-quality data from Chinese (and broader Asian) populations.,Meta-DiSc software was used for threshold effect assessment rather than hierarchical bivariate or HSROC models — a methodological concern because Meta-DiSc applies separate pooling that can produce artificially narrow confidence intervals and underestimate between-study heterogeneity.,Follow-up duration not reported; missed diagnoses identified only at index visit cannot capture delayed PE presentations.,Funding source not reported; potential conflicts of interest cannot be assessed. __BOTTOM_LINE_CATEGORY__ confirmatory __BOTTOM_LINE__ This 2026 meta-analysis (n = 13,672; 10 studies) confirms that PERC achieves a pooled sensitivity of approximately 95% and a pooled NLR of approximately 0.19 in ED patients pre-stratified as low risk for PE, supporting its role as a bedside rule-out tool when correctly applied within a two-step probability-first algorithm. However, the pooled missed-diagnosis rate of 2.0% among PERC-negative patients is clinically meaningful, PERC's specificity remains persistently low (limiting its utility in PERC-positive patients), and generalizability to non-Western populations is uncertain. No change in current practice is indicated by this review alone; it is best viewed as contemporary confirmatory evidence reinforcing existing guideline-concordant use of PERC in appropriately pre-stratified, low-risk ED patients.