PERC Rule: If all 8 criteria are negative AND pre-test probability is < 15%, PE is ruled out WITHOUT further testing.
Guidelines & Evidence
Clinical Details
Section 1
When to Use
When to Use
Suspected PE where the treating clinician has already established LOW pre-test probability (< 15%) by clinical gestalt or Wells score ≤ 2
To safely avoid D-dimer testing and further workup in genuinely low-risk patients — PERC is applied before any lab test is ordered
Emergency department and general medicine settings where over-investigation is a recognised concern
When CTPA use needs to be rationally curtailed — approximately 1–2% of all ED visits in the US involve CTPA for PE, with fewer than 5% of those scans confirming the diagnosis
As part of a consistent, pre-committed diagnostic strategy — do not apply PERC selectively or intermittently
Critical Prerequisite
PERC is ONLY valid when the treating clinician has independently established LOW pre-test probability (implicit belief that PE prevalence is < 15%) before applying the criteria. It is not a standalone diagnostic test and cannot be used to first establish probability. If any clinical doubt exists about whether PTP is truly low — apply Wells or Revised Geneva Score first. PERC applied without gestalt assessment is unsafe and has produced false-negative rates > 6% in high-prevalence European populations.
Do Not Apply PERC In
Pregnancy — PERC has not been validated in pregnant patients; use modified YEARS or clinical specialist input
Haemodynamic instability — this is high-risk PE by definition; proceed directly to echocardiography or CTPA
Patients already on therapeutic anticoagulation — clinical scenario has already been adjudicated
Patients with symptom onset > 3 days before presentation — D-dimer sensitivity decreases; the full PERC model was derived from acute presentations
Moderate or high pre-test probability patients — applying PERC outside the low-PTP window produces an unacceptable false-negative rate
Section 2
Formula & Logic
The Test-Treatment Threshold
PERC is built on the concept of the "test-treatment threshold" — a disease prevalence below which the harms of diagnostic testing outweigh the benefits. Kline et al. derived this threshold at approximately 1.8% for PE. Below this prevalence, the probability of a false-positive D-dimer leading to unnecessary CTPA, contrast nephropathy, radiation exposure, or iatrogenic anticoagulation injury exceeds the probability that testing will identify a clinically meaningful PE. PERC identifies patients whose post-test probability — after applying 8 objective criteria — falls below this threshold.
The 8 PERC Criteria
01
Age < 50 years
02
Heart rate < 100 bpm (resting)
03
SpO2 ≥ 95% on room air
04
No unilateral leg swelling (visual assessment of calf asymmetry)
05
No haemoptysis
06
No exogenous oestrogen use (OCP, HRT, testosterone — any hormonal therapy)
07
No prior venous thromboembolism (DVT or PE, objectively documented)
08
No recent surgery or trauma requiring hospitalisation or endotracheal intubation within 4 weeks
Criteria Logic
01
ALL 8 criteria must be ABSENT (each = 0) to apply PERC rule-out — PERC score of 0
02
A single positive criterion (any PERC item present) mandates D-dimer testing — do not interpret partial negativity as reassuring
03
PERC score of 0 in a low-PTP patient: post-test PE probability ≈ 1.0% — below treatment threshold, no further testing warranted
04
PERC score > 0: proceed to D-dimer with age-adjusted threshold (age × 10 µg/L if > 50 years)
Statistical Performance
Pooled across 12–13 cohorts (n = 13,885–14,844): sensitivity 97% (95% CI 96–98%), specificity 22–23%, negative LR 0.17 (95% CI 0.13–0.23), positive LR 1.22–1.24. The low specificity is expected and acceptable — PERC is a rule-out tool only. The negative LR of 0.17 represents a meaningful reduction in post-test probability when pre-test probability is already low (< 15%). In the Kline 2008 validation (n = 8138), low suspicion + PERC negative had a false-negative rate of 1.0% (15/1666 patients), within the pre-specified < 2% safety threshold.
Section 3
Pearls/Pitfalls
When PERC Fails — European Prevalence Problem
PERC was derived and validated in US emergency departments where PE prevalence among tested patients is approximately 8–10%. In European populations — where the threshold for ordering PE testing is higher — prevalence among tested patients is 20–27%, producing false-negative rates of 6–8% in early studies. This is not a flaw in PERC itself but a violation of its precondition: PERC is only safe when pre-test probability is genuinely < 15%. When applied with strict gestalt-based low-PTP criteria, even European studies (Penaloza et al., PERCEPIC) have confirmed safety with 0% false-negative rates.
PROPER Trial — First RCT Evidence
The PROPER trial (Freund et al., JAMA 2018) was the first cluster-randomised controlled trial of PERC vs conventional care in 14 French EDs (n = 1914 low-risk patients). Primary outcome: 1 symptomatic PE (0.1%) at 3 months in the PERC group vs 0 in the control group — meeting the non-inferiority margin. Secondary benefits of the PERC strategy: 10% reduction in CTPA use (13% vs 23%), 36–40 minute reduction in ED length of stay, and 3.3% reduction in hospital admission rates. This trial provides the only RCT-level evidence that PERC-based care does not increase VTE events.
Criterion Interpretation Pitfalls
Heart rate: must be resting rate — do not score based on a single anxious or exertion-related reading; reassess after 5–10 minutes if initially > 100
SpO2: must be room air — if the patient is on supplemental O2 at any flow rate, this criterion cannot be assessed and PERC is invalid
Exogenous oestrogen: includes ALL hormonal therapies — OCP, HRT, testosterone, progesterone-only pills; not limited to oestrogen-containing preparations
Leg swelling: requires visual comparison of calves — if bilateral swelling is present, the criterion is negative (PERC only flags unilateral swelling as a DVT proxy); do not falsely reassure
Prior VTE: must have been objectively documented — patient-reported "blood clot" without imaging confirmation is insufficient and should be treated as positive
Surgery criterion: specifically requires hospitalisation OR endotracheal intubation — minor outpatient procedures in the past 4 weeks do not trigger this criterion
PERC vs D-Dimer — When to Prefer Each
D-dimer has approximately 96% sensitivity for PE but poor specificity — yielding high false-positive rates in older patients, post-surgical patients, pregnant patients, those with active malignancy, and those with systemic inflammation. In these groups, a positive D-dimer almost invariably triggers CTPA regardless of clinical probability. PERC, by contrast, requires no laboratory processing, produces a result in under 60 seconds, shortens ED stay by ~37 minutes, and avoids the D-dimer → CTPA cascade entirely when negative. PERC is most valuable precisely where D-dimer is least useful: younger patients with incidental elevations from non-VTE causes.
Subsegmental PE and the Overtesting Problem
The dramatic rise in CTPA use has exposed a new clinical challenge: incidental detection of subsegmental PE (SSPE), which affects small distal pulmonary arteries and may not require anticoagulation. Studies estimate that 5–10% of CTPA-detected PEs are isolated SSPE with uncertain clinical significance. Kline noted in his JAMA 2018 editorial that CTPA is performed in approximately 1–2% of 120 million annual US ED visits, yet < 5% of scans are positive. PERC is one of the few evidence-based tools specifically designed to interrupt this cycle upstream — before any test is ordered.
Section 4
Next Steps
Decision Tree
01
Establish PTP first: clinical gestalt < 15%, or Wells PE score ≤ 2. If not low → do not apply PERC, proceed directly to D-dimer ± CTPA.
02
Low PTP + ALL 8 PERC criteria absent (PERC score = 0) → PE excluded. No D-dimer. No imaging. Document pre-test probability and PERC assessment explicitly in the clinical record.
03
Low PTP + ANY PERC criterion present (PERC score > 0) → Obtain high-sensitivity D-dimer. Use age-adjusted threshold (age × 10 µg/L) in patients > 50 years.
04
D-dimer negative (below threshold) → PE excluded. Discharge with appropriate safety-netting.
05
D-dimer positive (above threshold) → Proceed to CTPA. Interim LMWH only if CTPA delay > 4 hours and clinical suspicion is high.
Documentation Standard
When applying PERC, explicitly document: (1) pre-test probability estimate and the basis for designating it as low (gestalt or Wells score), (2) each PERC criterion with its assessed value, and (3) the decision reached. In medicolegal settings, undocumented application of PERC is indistinguishable from no assessment. This is particularly important given that PERC leaves a small residual miss rate (~1%) — documentation demonstrates that the clinical threshold was properly evaluated.
Complementary Tools
Wells PE Score
Revised Geneva Score
YEARS Algorithm
Age-Adjusted D-Dimer Calculator
Section 5
Evidence Appraisal
Original Derivation
Clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism.
Kline JA et al. • Journal of Thrombosis and Haemostasis.. 2004;2(8):1247–1255. Derived in 3148 ED patients. Identified 8 objective criteria and established the test-treatment threshold of 1.8%. Proposed PERC as a pre-D-dimer rule-out strategy for low-PTP patients.
Prospective Multicenter Validation
Prospective multicenter evaluation of the pulmonary embolism rule-out criteria.
Kline JA et al. • Journal of Thrombosis and Haemostasis.. 2008;6(5):772–780. n = 8138 patients across 13 EDs (USA and New Zealand). Low suspicion + PERC negative: false-negative rate 1.0% (15/1666). Sensitivity 97.4% (95% CI 95.8–98.5%). Defined low PTP as gestalt < 15%. Confirmed safety below 2% VTE threshold.
Systematic Review and Meta-Analysis
Diagnostic accuracy of pulmonary embolism rule-out criteria: a systematic review and meta-analysis.
Singh B et al. • Annals of Emergency Medicine.. 2012;59:517–520. 12 cohorts, n = 13,885, 6 countries. Pooled sensitivity 97% (95% CI 96–98%), specificity 23%, negative LR 0.17, positive LR 1.24. Confirmed consistent high sensitivity. Significant heterogeneity in specificity (I2 = 97.4%) driven by population PE prevalence differences.
European Validation — PERCEPIC
Pulmonary embolism rule-out criteria (PERC) rule in European patients with low implicit clinical probability (PERCEPIC): a multicentre, prospective, observational study.
Penaloza A et al. • Lancet Haematology.. 2017;4(12):e615–e621. Prospective European multicentre study confirming safety of PERC when combined with strict gestalt-based low-PTP assessment (< 15%). Zero thromboembolic events at 3-month follow-up in PERC-negative, low-PTP patients. Resolved controversy from earlier European studies that had applied PERC without gestalt precondition.
PROPER Trial — First RCT
Effect of the pulmonary embolism rule-out criteria on subsequent thromboembolic events among low-risk emergency department patients: the PROPER randomized clinical trial.
Freund Y et al. • JAMA.. 2018;319(6):559–566. Crossover cluster-RCT, 14 EDs, France, n = 1914 low-risk patients. Primary outcome: 1 PE (0.1%) at 3 months in PERC arm vs 0 in control — non-inferiority met. PERC strategy reduced CTPA use by 10% (13% vs 23%), shortened ED stay by ~37 minutes, and reduced hospital admission by 3.3%. First and only RCT-level evidence supporting PERC.
Section 6
Literature
Dr. Jeffrey Kline
An emergency physician and clinical scientist at Indiana University School of Medicine (subsequently at Wayne State University). Kline developed PERC in response to what he characterised as an epidemic of over-testing in suspected PE — driven by fear of litigation, the proliferation of high-sensitivity D-dimer assays, and the clinical instinct to always do more. His core argument, published in 2004, was mathematically rigorous: when pre-test probability is below ~1.8%, a positive D-dimer is statistically more likely to be a false positive than a true signal of PE, making D-dimer testing actively harmful in that population.
Historical Context
Through the 1990s and 2000s, CTPA became rapidly accessible across US emergency departments and the rate of PE testing expanded dramatically — driven partly by genuine clinical need, partly by medicolegal pressure, and partly by the ease of ordering a sensitive test. The paradox: as more CTPAs were ordered, more incidental and subsegmental PEs were found, many of which were treated with anticoagulation despite uncertain clinical significance, exposing patients to bleeding risk without proven benefit. PERC was the first clinical tool to address this problem not by improving PE detection, but by rationally defining when PE detection is unnecessary.
Score Development and Validation Timeline
01
2004 (J Thromb Haemost): PERC derived in 3148 ED patients. 8 criteria identified via logistic regression. Test-treatment threshold defined at 1.8%.
02
2008 (J Thromb Haemost): Prospective multicenter validation in 8138 patients across 13 sites. False-negative rate 1.0% in low-PTP + PERC-negative group. Safety confirmed in North American ED populations.
03
2011–2012: European studies report higher false-negative rates (6–8%) in populations with PE prevalence > 20%. Controversy about European applicability. Root cause identified as failure to apply gestalt-based low-PTP precondition.
04
2012 (Ann Emerg Med): Singh et al. meta-analysis across 13 cohorts, n = 13,885 — pooled sensitivity 97%. Performance stable across high- and low-prevalence populations when applied correctly.
05
2017 (Lancet Haematol): PERCEPIC study prospectively validates PERC in European EDs using strict < 15% gestalt criterion. Zero missed PEs at 3 months. European controversy resolved.
06
2018 (JAMA): PROPER trial — first RCT. PERC non-inferior to conventional care in 1914 French ED patients. 10% CTPA reduction, 37-minute ED stay reduction. PERC endorsed as standard of care in low-risk ED patients.
