Ruling out PE: Wells, PERC and age-adjusted D-dimer in one workflow

Calculators for this article: Wells criteria for PE, the PERC rule, the age-adjusted D-dimer calculator and sPESI.

Suspected pulmonary embolism is a cognitive-load problem as much as a diagnostic one. The workup is well studied, but it is a chain of conditional steps, each valid only in the population it was tested in. Skip a condition and the rule stops protecting you. This is the sequence, with what each step was shown to do.

Step 1: estimate pretest probability

Everything downstream depends on a pretest estimate. The American College of Physicians' best practice advice starts there: clinicians should use validated clinical prediction rules to estimate pretest probability.^[1]

The Wells score for PE assigns points for clinical signs of DVT (3), PE as the most likely diagnosis (3; the 2000 abstract words this item as no alternative diagnosis), heart rate over 100 (1.5), immobilization or recent surgery (1.5), previous DVT or PE (1.5), hemoptysis (1) and malignancy (1). It can be read in three tiers (under 2 low, 2 to 6 moderate, over 6 high) or two (4 or less "PE unlikely", over 4 "PE likely").^[2] In the derivation study 7.8% of patients with a score of 4 or less had PE.^[2]

In the later emergency department validation, PE was present in 1.3%, 16.2% and 37.5% of patients with low, moderate and high pretest probability.^[3]

Step 2: if pretest probability is low, consider PERC

The pulmonary embolism rule-out criteria (PERC) are eight findings: age under 50, pulse under 100, SaO2 of 95% or more, no hemoptysis, no estrogen use, no surgery or trauma requiring hospitalization within 4 weeks, no prior venous thromboembolism, and no unilateral leg swelling.^[4]

The condition people drop is the entry criterion. PERC was built for patients whose gestalt pretest probability is already low, under 15%.^[4] In the 2008 multicenter evaluation, low suspicion plus PERC negative applied to 20% of patients, and the false-negative rate was 16 of 1,666 (1.0%, 95% CI 0.6% to 1.6%).^[4]

The PROPER cluster-randomized trial tested this in practice among low-risk emergency patients. Three-month thromboembolic events were 0.1% in the PERC group versus none with usual care, within the noninferiority margin, while CT pulmonary angiography fell from 23% to 13%.^[5] PERC does not apply to a patient you already think is moderate or high risk. In that patient it is simply the wrong tool.

Step 3: D-dimer for "PE unlikely", adjusted for age

If pretest probability is low but PERC is positive, or pretest probability is intermediate, the next step is a high-sensitivity D-dimer.^[1] In the Christopher study, "PE unlikely" by the dichotomized Wells score plus a normal D-dimer occurred in 32.0% of patients, and the 3-month rate of nonfatal VTE in the untreated patients was 0.5% (95% CI 0.2% to 1.1%).^[6]

A fixed 500 µg/L cut-off loses specificity with age, because D-dimer rises with age. The ADJUST-PE management study used a cut-off of age × 10 µg/L in patients 50 or older. Among patients whose D-dimer fell between 500 and their age-adjusted cut-off, the 3-month failure rate was 1 of 331 (0.3%, 95% CI 0.1% to 1.7%), and in patients 75 or older the share in whom PE could be excluded rose from 6.4% to 29.7%.^[7] The ACP advice endorses age-adjusted thresholds for patients older than 50.^[1]

Two cautions. The age-adjusted cut-off was validated with assays whose conventional cut-off is 500 µg/L, so check the units your lab reports before multiplying. And it belongs to the non-high-probability group only.

Probability-adapted D-dimer

Two strategies go a step further and raise the D-dimer threshold when clinical probability is lowest. YEARS uses three items (clinical signs of DVT, hemoptysis, PE as the most likely diagnosis): with none of them, PE is excluded below 1,000 ng/mL; with one or more, below 500 ng/mL. The 3-month VTE rate was 0.61%, and CT angiography was not indicated in 48% of patients versus 34% under the standard approach.^[8] PEGeD used low clinical pretest probability with D-dimer under 1,000, or moderate with D-dimer under 500. None of the 1,325 patients ruled out this way had VTE during follow-up (95% CI 0.00% to 0.29%), and chest imaging was used in 34.3% versus 51.9% with a fixed cut-off.^[9] The 2019 ESC guideline states that D-dimer cut-offs adjusted for age or clinical probability can be used as an alternative to the fixed cut-off.^[10]

Step 4: high probability means imaging, not a D-dimer

If pretest probability is high, or the Wells score says "PE likely", the D-dimer cannot rule PE out at that prevalence. The ACP advice is direct: obtain CT pulmonary angiography in patients with high pretest probability and do not order a D-dimer for them.^[1]

The workflow on one line

Where these rules fail

• Wrong population. PERC in a moderate-risk patient, or a D-dimer in a high-risk one, is the most common misuse.
• Pregnancy, hospitalized patients and anticoagulated patients were not the populations in which most of these rules were derived. Check the trial population before applying a rule outside it.
• Unit errors. FEU versus DDU reporting can change the meaning of "500".

Where decision support helps

The reasoning above is not complicated. Holding all four conditions in mind at 3 a.m., in the right order, with the right population for each, is where it slips. Pogosh reads the note, recognizes suspected PE, and offers Wells, PERC and the age-adjusted D-dimer as calculators to run, along with the imaging and labs that fit and the escalation criteria for massive PE. It never runs a calculator for you, and it does not decide whether this patient needs a CT. You do.