Clinical background
Pulmonary embolism is a diagnosis that is often over-investigated. The D-dimer has low specificity, and a positive result prompts computed tomography pulmonary angiography (CTPA), with its radiation dose, contrast risk and cost. In populations with a low prevalence of pulmonary embolism, the proportion of false-positive D-dimers is high, and the benefit of testing at all is marginal if the pre-test probability is sufficiently low.
The PERC rule was developed to identify patients whose pre-test probability of pulmonary embolism is so low that neither a D-dimer nor imaging is justified. Kline et al. calculated the testing threshold for the D-dimer as 1.8% using the method of Pauker and Kassirer: if the probability of pulmonary embolism lies below this threshold, the risk of a false-positive D-dimer (leading to an unnecessary CTPA) outweighs the benefit of detecting a true pulmonary embolism [1]. PERC is therefore not a tool for finding pulmonary embolism but a tool for safely refraining from further investigation.
Applying the PERC rule
PERC is a block rule with eight binary criteria. The patient is PERC-negative only when all eight criteria are absent. If a single criterion is met, the patient is PERC-positive and the rule cannot be used to exclude pulmonary embolism.
The eight variables are:
- Age ≥50 years: the age limit is absolute; a 49-year-old does not meet the criterion, a 50-year-old does.
- Heart rate ≥100/min: measured on arrival, not after treatment.
- Oxygen saturation <95% on room air: if the patient is already receiving oxygen at the time of measurement, the criterion cannot be assessed.
- Haemoptysis: coughing up blood during the current episode.
- Oestrogen use: refers to exogenous oestrogen, including oral contraceptives and hormone replacement therapy.
- Previous DVT or pulmonary embolism: previous venous thromboembolism at any site.
- Recent surgery or trauma (≤4 weeks, requiring hospital care): the procedure or injury must have been serious enough to require hospital care.
- Unilateral leg swelling: objective asymmetry on clinical examination of the lower legs.
The rule was derived from a cohort of 3,148 patients investigated for suspected pulmonary embolism at ten US emergency departments [1]. Twenty-one variables were collected and analysed by logistic regression with backward elimination, leaving eight variables in the block rule. In the prospective validation within the same study, the rule was tested in a low-risk group of 1,427 patients (in whom a D-dimer was tested initially) and a very-low-risk group of 382 patients (in whom pulmonary embolism was not primarily suspected). The prevalence of pulmonary embolism was 8% and 2% respectively. Sensitivity was 96% and 100%, and specificity 27% and 15%. The prevalence of pulmonary embolism among PERC-negative patients was 1.4% (95% CI 0.5 to 3.0) in the low-risk group and 0% (95% CI 0 to 6.2) in the very-low-risk group [1].
Interpretation in practice
PERC has only two outcomes, and management is binary:
| Outcome | Criterion | Management |
|---|---|---|
| PERC-negative | All eight criteria absent | Pulmonary embolism can be excluded without a D-dimer or imaging. The patient is discharged with advice to return if symptoms worsen. |
| PERC-positive | One or more criteria present | PERC cannot be used. Proceed with the standard algorithm: D-dimer and, if indicated, CTPA. |
It is crucial that PERC-negativity is not a diagnostic answer in itself but a decision not to investigate further. The patient should be informed about signs of deterioration and have a clear route back. The exclusion applies to the current episode; a new presentation of symptoms requires a fresh assessment from the beginning.
Validation and performance
The PROPER trial is the only randomised trial of PERC. It was a cluster-randomised crossover non-inferiority trial at 14 French emergency departments, including 1,916 patients with a low gestalt clinical probability of pulmonary embolism (estimated at under 15% by the treating physician) [2]. In the PERC group, pulmonary embolism was excluded without further testing if all eight criteria were negative; otherwise the standard algorithm with D-dimer and CTPA was followed. The control group always had a D-dimer, with CTPA if positive. The primary endpoint was a thromboembolic event at three months of follow-up. An event occurred in 1 patient (0.1%) in the PERC group versus 0 in the control group, a difference that met the non-inferiority margin of 1.5% [3]. CTPA use fell from 23% to 13% (difference −10 percentage points, 95% CI −13 to −6), the length of stay in the emergency department was shortened by a mean of 36 minutes, and the proportion of hospital admissions fell by 3.3 percentage points [3].
A systematic review and meta-analysis from 2026 included ten studies with a total of 13,672 patients [4]. The pooled prevalence of pulmonary embolism was 7%. Pooled sensitivity was 95% (95% CI 89 to 98), specificity 26% (95% CI 16 to 40), and the negative predictive value 98.2% (95% CI 97.8 to 99.0). The PERC strategy was associated with a significant reduction in CTPA use (RR 0.85, 95% CI 0.80 to 0.91) [4].
The low specificity is expected and is not a sign of poor performance: the rule is designed to be extremely sensitive, and the price is that relatively few patients turn out to be PERC-negative. In the PROPER trial, approximately 37% of patients were PERC-negative, meaning that about a third of low-risk patients could avoid a D-dimer and CTPA [3].
Limitations
PERC presupposes a low pre-test probability. The rule is validated only in patients in whom the treating physician has already judged the clinical probability of pulmonary embolism to be low, either through a structured score (Wells, revised Geneva) or through unstructured clinical judgement (gestalt). Applying PERC to a patient with a moderate or high probability is a sequencing error: the rule cannot lower an already elevated pre-test probability below the testing threshold.
A high prevalence of pulmonary embolism challenges the rule. In European populations with a high prevalence, PERC has been questioned. A Swiss retrospective study of 1,675 patients at six emergency departments found an overall pulmonary embolism prevalence of 21.3%, and among PERC-negative patients with a low clinical probability the prevalence was 6.4% (95% CI 3.7 to 10.8), far above the safe threshold of 2% [5]. A Belgian study of 959 patients with a prevalence of 29.8% found that 5.4% of PERC-negative patients had pulmonary embolism, but when PERC was combined with a low gestalt probability the prevalence fell to 0% (95% CI 0 to 5) [6]. These studies were retrospective and partly included patients with a higher than low probability, which may explain the poorer results. The PROPER trial, which was prospective and strictly limited to a low gestalt probability, by contrast demonstrated safety even in a European setting [3].
The rule does not apply to all patients. PERC is not validated in children, pregnant women, patients on ongoing anticoagulation, patients with severe haemodynamic compromise, or patients in whom an alternative diagnosis is obvious and fully explains the symptoms. Patients already receiving oxygen on arrival cannot be assessed for oxygen saturation on room air, and the criterion is then not applicable.
The oestrogen criterion is gender-neutral in principle but gender-specific in practice. It refers to exogenous oestrogen and therefore mainly captures women. This is deliberate: oestrogen exposure is a genuine risk factor for venous thromboembolism and contributes to the sensitivity of the rule.
References
- Kline JA et al. Clinical criteria to prevent unnecessary diagnostic testing in emergency department patients with suspected pulmonary embolism. J Thromb Haemost 2004. PMID: 15304025
- Freund Y et al. PERC rule to exclude the diagnosis of pulmonary embolism in emergency low-risk patients: study protocol for the PROPER randomized controlled study. Trials 2015. PMID: 26607669
- Freund Y et al. Effect of the Pulmonary Embolism Rule-Out Criteria on Subsequent Thromboembolic Events Among Low-Risk Emergency Department Patients: The PROPER Randomized Clinical Trial. JAMA 2018. PMID: 29450523
- Liu X et al. Diagnostic Validation of PERC and Resource Utilization in Suspected Acute Pulmonary Embolism: A Systematic Review and Meta-Analysis. Clin Appl Thromb Hemost 2026. PMID: 42541393
- Hugli O et al. The pulmonary embolism rule-out criteria (PERC) rule does not safely exclude pulmonary embolism. J Thromb Haemost 2011. PMID: 21091866
- Penaloza A et al. Performance of the Pulmonary Embolism Rule-out Criteria (the PERC rule) combined with low clinical probability in high prevalence population. Thromb Res 2012. PMID: 22424852
