Wells vs revised Geneva score for pulmonary embolism

The main difference: clinical judgment

The Wells score and the revised Geneva score differ most in one item: Wells gives 3 points when PE is judged more likely than any alternative diagnosis, while the revised Geneva score has no judgment item at all. Wells and colleagues published their rule in Thrombosis and Haemostasis in 2000, with seven items and a range of 0 to 12.5. Its two 3-point items are clinical signs of deep vein thrombosis and no alternative diagnosis being as likely as PE, so a large share of the score rests on the assessor. Le Gal and colleagues built the revised Geneva score in the Annals of Internal Medicine in 2006 to be entirely standardized: eight items, all objective history or examination findings, derived in emergency department patients in Europe and validated in a separate cohort. Unlike the original Geneva score, neither needs a blood gas or chest X-ray. Both are used the same way, to decide whether D-dimer or imaging comes next.

How the revised Geneva score is calculated

The revised Geneva score adds points for eight items weighted by their regression coefficients, and the simplified version gives each item 1 point instead. In Le Gal's original, age over 65 scores 1, previous deep vein thrombosis or PE 3, surgery or fracture within a month 2, active cancer 2, unilateral leg pain 3, hemoptysis 2, a heart rate of 75 to 94 beats per minute 3 or 95 or more 5, and pain on deep venous palpation of the leg with unilateral edema 4. Klok and colleagues tested the simplified revised Geneva score in 1,049 patients from two prospective studies in 2008. Its area under the curve was 0.74, against 0.75 for the weighted score, and no patient with a low or intermediate probability, or a PE unlikely result, plus a normal highly sensitive D-dimer had venous thromboembolism in 3 months. The simplified version is quicker to add up at the bedside.

Cut-offs for each score

Each score can be read on a three-level scale or a two-level scale, and the cut-offs differ between them, so always name the scale you used. For the original revised Geneva score, Le Gal's validation cohort found PE in 8% of patients at low probability (0 to 3 points), 28% at intermediate probability (4 to 10) and 74% at high probability (11 or more). The 2019 European Society of Cardiology (ESC) PE guideline also lists two-level cut-offs: PE unlikely at 0 to 5 and likely at 6 or more for the revised Geneva score, and unlikely at 0 to 2 and likely at 3 or more for the simplified version, whose three levels are 0 to 1, 2 to 4 and 5 or more. For Wells, the 2000 paper split the three levels at under 2, 2 to 6 and over 6, and the two levels at 4 or less (PE unlikely) and over 4 (PE likely).

Which score performs better?

Wells and the revised Geneva score perform about equally well, and head-to-head studies found no clear winner. Ceriani and colleagues pooled 29 studies with 31,215 patients in a 2010 meta-analysis. Pooled PE prevalence for the three-level Wells score was 6% at low, 23% at intermediate and 49% at high probability, and for the revised Geneva score 9%, 26% and 76%; the authors concluded that the rules show similar accuracy but are not interchangeable, and that the choice should follow local PE prevalence, the type of patients and the D-dimer assay. Douma and colleagues compared four rules prospectively in 807 Dutch patients in 2011. The share classed as PE unlikely ranged from 62% to 72%, combined with a normal D-dimer each rule excluded PE in 22% to 24% of patients, and every combination had a failure rate of 0.5% to 0.6%. Although 30% of patients had discordant results between rules, none with discordant rules and a normal D-dimer had PE.

Worked example: a patient the scores split on

The scores disagree most when the Wells judgment item decides the result. Take a 40-year-old with sudden pleuritic chest pain and a heart rate of 105, no leg signs, no history of clots, cancer or recent surgery, and no better explanation on assessment. Wells gives 1.5 for a heart rate over 100 and 3 because PE is the most likely diagnosis, a score of 4.5: PE likely, so the next step is imaging. The revised Geneva score gives 5 for a heart rate of 95 or more and nothing else, which is intermediate probability on the three-level scale and PE unlikely on the two-level scale, so a D-dimer would come first. The simplified Geneva score gives 2, also PE unlikely. Now take a 70-year-old with unilateral calf pain, tenderness and edema and a heart rate of 98. Revised Geneva scores 1 + 3 + 5 + 4 = 13, high, and Wells scores at least 3 for the leg signs.

Score your own patient on the Wells score for PE calculator and compare.