Written and maintained by CASRAI Editorial Board
Last updated
“Wells criteria” is ambiguous on its own — there are two distinct clinical decision rules that share the name, one for deep vein thrombosis (DVT) and one for pulmonary embolism (PE), each with its own criteria, point values and validated probability bands. Both were developed by Dr. Philip Wells and colleagues, both are pretest-probability tools rather than diagnostic tests, and both exist for the same operational reason: to sort patients into risk tiers before ordering a D-dimer or imaging study, so that low-risk, D-dimer-negative patients can be safely discharged without a CT pulmonary angiogram (CT-PA) or compression ultrasound. This page covers both scores — their criteria, point values, and 3-tier and 2-tier probability bands — and, because that’s where CASRAI’s own audience sits, focuses on how a hospital builds the score into its clinical decision-support (CDS) tooling and imaging-stewardship metrics rather than on bedside diagnostic technique. For the neurological-status counterpart to this kind of structured bedside score, see CASRAI’s Glasgow Coma Scale guide.
Two different scores, one name
The DVT and PE versions of the Wells score are not interchangeable and are not two versions of the same instrument — they assess different conditions, use different criteria, and produce differently calibrated probability bands. A chart note or CDS rule that just says “Wells score: 3” without specifying which instrument is ambiguous and, in an audit or quality review, unreviewable. Documentation should always name the instrument explicitly: “Wells DVT score” or “Wells PE score” (equivalently, “Wells’ Criteria for Pulmonary Embolism”).
Wells Criteria for DVT
The DVT version assigns one point to each of nine clinical findings, then subtracts two points if an alternative diagnosis is judged at least as likely as DVT:
| Criterion | Points |
|---|---|
| Active cancer (treatment ongoing, within the previous 6 months, or palliative) | +1 |
| Paralysis, paresis, or recent plaster immobilization of the lower extremity | +1 |
| Recently bedridden ≥3 days, or major surgery within the previous 4 weeks | +1 |
| Localized tenderness along the distribution of the deep venous system | +1 |
| Entire leg swollen | +1 |
| Calf swelling >3 cm compared with the asymptomatic leg (measured 10 cm below the tibial tuberosity) | +1 |
| Pitting edema confined to the symptomatic leg | +1 |
| Collateral superficial (non-varicose) veins | +1 |
| Previously documented DVT | +1 |
| Alternative diagnosis at least as likely as DVT | −2 |
DVT pretest-probability bands
Three-tier model (the original, still the version most chart tools display):
- High probability: score ≥3
- Moderate probability: score 1–2
- Low probability: score ≤0
Two-tier (“dichotomized”) model — the version most CDS pathways and D-dimer algorithms actually use, because a binary gate is what a rule engine needs:
- DVT likely: score ≥2 — proceed to compression ultrasound regardless of D-dimer result.
- DVT unlikely: score ≤1 — a negative high-sensitivity D-dimer can safely exclude DVT without imaging; a positive D-dimer still requires ultrasound.
Wells Criteria for PE
The PE version is a separately weighted 7-item rule, with points ranging from 1 to 3 per criterion:
| Criterion | Points |
|---|---|
| Clinical signs and symptoms of DVT (leg swelling, pain with palpation of the deep veins) | +3 |
| PE is the #1 diagnosis, or equally likely to an alternative | +3 |
| Heart rate >100 bpm | +1.5 |
| Immobilization ≥3 days, or surgery in the previous 4 weeks | +1.5 |
| Previous, objectively diagnosed PE or DVT | +1.5 |
| Hemoptysis | +1 |
| Malignancy (treatment within the previous 6 months, or palliative) | +1 |
PE pretest-probability bands
Three-tier model, with the PE prevalence each band was validated against:
- High probability: score >6 (~37.5% PE prevalence)
- Moderate probability: score 2–6 (~16.2% PE prevalence)
- Low probability: score <2 (~1.3% PE prevalence)
Two-tier (simplified/modified) model — the version validated for direct use with D-dimer in the Christopher Study algorithm below:
- PE likely: score >4 (~37.1% PE prevalence) — proceed to CT-PA (or V/Q scan where CT-PA is contraindicated) without waiting on D-dimer.
- PE unlikely: score ≤4 (~12.1% PE prevalence) — order a high-sensitivity D-dimer; a negative result excludes PE without imaging.
How the score gates D-dimer and CT-PA — the decision pathway
The clinical value of either Wells score isn’t the number itself; it’s what the number is allowed to do to the next step in the pathway. The structure is the same for both DVT and PE:
- Likely/high tier: the pretest probability is high enough that a negative D-dimer would not be trusted anyway (D-dimer’s negative predictive value collapses as prevalence rises) — go straight to definitive imaging (compression ultrasound for DVT; CT-PA or V/Q scan for PE).
- Unlikely/low tier: order a D-dimer. A result below the assay’s cutoff, in a genuinely low-pretest-probability patient, has a validated negative predictive value high enough to exclude venous thromboembolism (VTE) without imaging.
- Unlikely/low tier with a positive D-dimer: D-dimer is nonspecific (raised by infection, malignancy, pregnancy, recent surgery, and simply by age), so a positive result in a low-probability patient still requires imaging — the score didn’t rule anything in, it only failed to rule it out non-invasively.
The PE pathway’s landmark validation is the Christopher Study (Christopher Study Investigators, JAMA, 2006), which prospectively tested the simplified 2-tier Wells score paired with a quantitative D-dimer assay and CT-PA, and found the algorithm safely withheld CT-PA from PE-unlikely, D-dimer-negative patients with a low subsequent venous-thromboembolism rate on follow-up. That result is the primary-source basis for treating “Wells unlikely + negative D-dimer” as a validated CT-PA-avoidance pathway rather than a convenience shortcut — it’s the citation an imaging-stewardship committee should point to, not a vendor whitepaper or a specialty society’s paraphrase of it.
A related, narrower instrument worth knowing exists (though it’s a separate tool, not a Wells component): the PERC rule (Kline et al.), an 8-item checklist used before D-dimer in patients who already have a low gestalt or Wells-unlikely probability — if all 8 PERC criteria are negative, the rule supports skipping D-dimer testing entirely, on the reasoning that the false-positive D-dimer rate in that population causes more unnecessary imaging than it prevents. PERC is a further gate on top of Wells-unlikely, not a substitute for scoring Wells in the first place.
The imaging-stewardship and quality-metric angle
For a patient-safety or quality office, the Wells score matters less as a bedside mnemonic and more as the input variable a CDS rule or an order-set hard-stop can act on. A few concrete implementation points:
- CDS alerting on order entry. A CT-PA order placed without a documented Wells score (or with a documented “unlikely” score and no D-dimer result on file) is exactly the kind of order a clinical decision-support rule can intercept — not to block the order, but to force the ordering clinician to document the pretest-probability reasoning, which is also what makes the order defensible on audit.
- Overuse-reduction metrics. CT-PA yield rate (the percentage of CT-PAs that are positive for PE) is a standard imaging-stewardship metric; a low, falling, or persistently sub-benchmark yield rate is the signal that Wells-based triage isn’t actually gating orders in practice, whatever the CDS rule says on paper. ACR Appropriateness Criteria for suspected PE explicitly build pretest-probability categorization into their appropriateness ratings for CT-PA versus other imaging — a hospital’s own order set should mirror that structure, not invent a parallel one.
- Choosing Wisely alignment. The American College of Emergency Physicians’ Choosing Wisely recommendation against CT-PA in patients with a low pretest probability of PE and a negative D-dimer is the professional-society-level version of the same rule this page documents at the instrument level — useful as the citation when a stewardship policy needs to justify itself to medical staff, not just to compliance.
- Documentation completeness, not just score accuracy. An audit that finds the Wells score was calculated correctly but never charted, or charted after the imaging order rather than before it, has found a real process failure even though the clinical judgment was sound — the score has to precede and gate the order to do the stewardship work it’s validated for.
Known limitations — document these, don’t paper over them
- The “alternative diagnosis” criterion is subjective. Both the DVT criterion (−2 points) and the PE “is #1 diagnosis or equally likely” criterion (+3 points) rely on the examining clinician’s own differential, which means the same patient can score differently depending on who examines them. This is the most frequently cited reproducibility weakness of both instruments and is worth naming explicitly in any CDS documentation, rather than presenting the score as more objective than it is.
- Neither score was derived or validated in pregnancy. Both the DVT and PE Wells rules were developed and validated in general adult populations; professional guidance cautions against relying on either score alone in pregnant patients, where baseline D-dimer is often elevated and a different diagnostic approach is generally used.
- D-dimer’s negative predictive value is age- and assay-dependent. A “negative” D-dimer means different things depending on which assay generated it (quantitative ELISA-derived versus qualitative) and, in many current pathways, age-adjusted cutoffs are applied above a threshold age — a stewardship program auditing CT-PA overuse should confirm which assay and cutoff its own lab actually reports before assuming every “negative D-dimer” in the chart means the same thing.
Where this fits in a broader deteriorating-patient and quality-metric program
Wells scoring is one of several structured, point-based bedside instruments a hospital’s clinical-risk and quality program relies on to convert clinical judgment into a documented, auditable number. CASRAI has separate guides for the other major instruments in this family: Glasgow Coma Scale and NIH Stroke Scale (NIHSS) for neurological status, RASS for sedation depth, and Braden Scale and Morse Fall Scale and Johns Hopkins Fall Risk Assessment Tool for pressure-injury and fall risk. For the prevention side of the DVT question specifically — what happens after a patient screens low-risk, or while awaiting definitive imaging — see CASRAI’s DVT prophylaxis and compression-stocking grading guide. All of these sit under CASRAI’s Patient Safety hub.
Frequently asked questions
Is the Wells score for DVT the same as the Wells score for PE?
No. They are two separate instruments that happen to share a name and an author. The DVT score has nine 1-point criteria plus a −2-point subtraction; the PE score has seven criteria weighted from 1 to 3 points. Chart documentation should always specify which one was used.
Does a low Wells score rule out DVT or PE on its own?
No — a low (“unlikely”) Wells score only identifies patients for whom a negative D-dimer is trustworthy enough to exclude the diagnosis without imaging. The score itself doesn’t rule anything out; it determines whether D-dimer is allowed to do that job, or whether the patient needs to go straight to imaging.
Which version of the score — 3-tier or 2-tier — should a CDS rule use?
The 2-tier (dichotomized) version is what the D-dimer-avoidance pathways were actually validated against (the Christopher Study used the simplified 2-tier PE model), and it’s the natural fit for a binary rule-engine gate. The 3-tier version is still common in chart-facing displays because it communicates more clinical nuance to the reading clinician; a well-built CDS tool can display the 3-tier band while gating the order logic on the 2-tier cutoff.
Why does a positive D-dimer still require imaging even after a low Wells score?
Because D-dimer is sensitive but not specific — it rises with infection, malignancy, pregnancy, recent surgery, and simply with age, not only with clot. A low Wells score plus a negative D-dimer is a validated exclusion pathway; a low Wells score plus a positive D-dimer is not — it just means D-dimer wasn’t able to do the excluding, and imaging is still required.
What CT-PA yield rate indicates the pretest-probability gate isn’t working?
There’s no single universal benchmark, but a persistently low positive-CT-PA rate is the standard signal that Wells-based (or equivalent) triage isn’t actually constraining who gets scanned in practice, whatever the documented policy says — it’s the metric an imaging-stewardship or overuse-reduction committee should be tracking against its own baseline over time, not comparing to a borrowed external number without knowing that institution’s case mix.
