PERC vs Wells score: how they work together for PE

What PERC and the Wells score each do

The PERC rule and the Wells score look similar on paper but do different jobs. The Wells score, published by Wells and colleagues in Thrombosis and Haemostasis in 2000, adds points for seven clinical features to estimate the probability of pulmonary embolism in anyone suspected of it, from 0 to 12.5. Its three-level version gives low (under 2), moderate (2 to 6) and high (over 6) probability, and its two-level version gives PE unlikely (4 or less) or likely (over 4). The PERC rule, derived by Kline and colleagues in 2004 and validated in 2008, is not a points score but a block of 8 criteria that must all be absent. It applies only to patients the clinician already judges to be at low risk, a gestalt estimate under 15% in the 2008 study, and its purpose is narrow: to show that such a patient needs no D-dimer. Wells sorts everyone; PERC clears only the lowest-risk group.

Where the criteria overlap

Five of PERC's eight criteria have a close match in the Wells score, but the thresholds and wording differ. Both count heart rate: Wells gives 1.5 points for a rate over 100, while PERC is positive at 100 or more. Both count hemoptysis and a previous DVT or PE. Wells gives 1.5 points for immobilization or surgery in the previous 4 weeks, while PERC asks about surgery or trauma requiring hospitalization within 4 weeks. Wells gives 3 points for clinical symptoms of DVT, and PERC asks about unilateral leg swelling. PERC adds three items that Wells does not use: age 50 or over, oxygen saturation under 95% and estrogen or hormone use. Wells adds two that PERC does not use: malignancy, and the 3-point judgment that no alternative diagnosis is more likely than PE. That last item is why Wells needs clinical judgment in its score, while PERC places the judgment before the rule, in the gestalt estimate.

The order to use them in

Pretest probability comes first, and PERC is applied only when that probability is low. The American College of Physicians set out the sequence in its 2015 best practice advice (Raja and colleagues, Annals of Internal Medicine). Clinicians should first use a validated clinical prediction rule, such as the Wells score, to estimate pretest probability. In patients with a low pretest probability who meet all the PERC criteria, they should obtain neither a D-dimer nor imaging. A high-sensitivity D-dimer should be the first test in patients with an intermediate pretest probability, and in low-probability patients who do not meet all the PERC criteria. In patients with a high pretest probability, the advice is to go straight to CT pulmonary angiography and not to obtain a D-dimer. One detail matters at the first step: PERC was validated with a clinician's gestalt estimate under 15%, not with a Wells cut-off, so a low Wells score supports but does not replace that judgment.

How PERC held up in other settings

PERC performed well where PE was uncommon among the people tested, and less well where it was common. Singh and colleagues pooled 12 cohorts with 13,885 emergency patients from 6 countries (2012), with a pooled sensitivity of 97% and specificity of 23%. Hugli and colleagues then applied PERC retrospectively to 1,675 patients from six European emergency departments, where 21.3% had PE (2011). Only 13.2% were PERC negative, and PE was found in 5.4% of them, and in 6.4% of those who also had a low clinical probability on the revised Geneva score. The authors concluded that PERC could not safely rule out PE in populations with a relatively high prevalence. The PROPER trial later tested PERC prospectively in 1,916 low-risk patients in 14 French emergency departments and found it noninferior to usual care, with 1 missed PE (0.1%) at 3 months. The prevalence of PE in the patients being tested is the key difference.

Worked example: two patients on both

Two patients show how the two tools combine. A 34-year-old woman judged at low risk has pleuritic chest pain after a coughing illness, a pulse of 92 and saturation of 98%, no other PERC findings, and takes a combined oral contraceptive. On Wells she scores 0: low probability and PE unlikely. On PERC she is positive, because estrogen use is one of the eight criteria, so under the ACP sequence she needs a D-dimer. Without the contraceptive she would be PERC negative and need no test at all. A 45-year-old man has a pulse of 108 two weeks after knee surgery that needed a hospital stay, and PE is not the most likely diagnosis. Wells gives 1.5 for heart rate and 1.5 for surgery, a total of 3: moderate on the three-level scale and PE unlikely on the two-level one. PERC is positive twice over and cannot clear him, so a D-dimer is next.

Check each patient in the PERC rule calculator, and score the probability step in the Wells score for PE calculator.