Wells score for DVT calculator
Estimate the pretest probability of deep vein thrombosis (DVT) in a patient with a suspected leg DVT. Pick the 2003 two-level model (DVT likely or unlikely) or the original 1997 three-level model (low, moderate or high).
How the Wells score for DVT works
The Wells score for DVT adds one point for each clinical feature present and subtracts two points when another diagnosis is at least as likely as DVT. Wells and colleagues published the three-level model in the Lancet in 1997, with eight one-point features. In the New England Journal of Medicine in 2003 they added a ninth feature, previously documented DVT, widened the surgery window from 4 to 12 weeks, and split patients into two groups, DVT likely or unlikely, instead of three. This calculator switches the criteria and the cut-offs when you change the model.
Wells DVT score = 1 point for each feature present − 2 if an alternative diagnosis is at least as likely as DVT
Range: −2 to 8 in the 1997 model, −2 to 9 in the 2003 model (previously documented DVT counts only in 2003).
| Feature | 1997 (Lancet) | 2003 (NEJM) |
|---|---|---|
| Active cancer | +1 | +1 |
| Paralysis, paresis or recent plaster cast | +1 | +1 |
| Bedridden or major surgery | +1: bedridden more than 3 days, or surgery within 4 weeks | +1: bedridden 3 days or more, or surgery within 12 weeks under general or regional anaesthesia |
| Localised deep vein tenderness | +1 | +1 |
| Entire leg swollen | +1 | +1 |
| Calf swelling vs other leg | +1 | +1 |
| Pitting oedema | +1 | +1 |
| Collateral superficial veins | +1 | +1 |
| Previously documented DVT | Not in the model | +1 |
| Alternative diagnosis as likely | −2 | −2 |
Wells DVT score categories
The Wells DVT score has two sets of categories, one for each paper. The 1997 three-level model sorts patients into low, moderate or high pretest probability; in that study of 593 outpatients, DVT was found in 3%, 17% and 75% of each group. The 2003 two-level model calls DVT likely at 2 points or more and unlikely below 2. In the 2003 trial of 1,096 outpatients, the overall prevalence of DVT or pulmonary embolism was 15.7%.
| Score | Category | What the paper reported |
|---|---|---|
| 0 or less | Low (1997) | DVT in 3% of patients (10 of 329) |
| 1–2 | Moderate (1997) | DVT in 17% of patients (32 of 193) |
| 3 or more | High (1997) | DVT in 75% of patients (53 of 71) |
| Under 2 | DVT unlikely (2003) | With a negative D-dimer, DVT was ruled out without ultrasound |
| 2 or more | DVT likely (2003) | Went on to ultrasound imaging, whatever the D-dimer result |
When the Wells DVT score is less reliable
The Wells score for DVT was built and tested in outpatients with a suspected DVT of the leg, so it was not derived in hospital inpatients or for arm DVT. It gives a pretest probability, not a diagnosis: in both studies the score was combined with ultrasound imaging, and in 2003 with D-dimer testing. The minus-two item, an alternative diagnosis at least as likely as DVT, is a clinical judgement, so two clinicians can score the same patient differently. The 1997 model has no item for a previous DVT.
Questions
Which Wells DVT model should I use?
The 2003 two-level model is the one Wells and colleagues tested with D-dimer in a randomised trial: a patient scored DVT unlikely with a negative D-dimer did not need ultrasound. The 1997 three-level model gives low, moderate or high pretest probability and was used with serial ultrasound and venography. Use the model that matches the diagnostic pathway you follow.
What does "DVT unlikely" mean?
A score under 2 in the 2003 model. In that trial, ultrasound was skipped when such a patient had a negative D-dimer. Among patients in the D-dimer group in whom DVT had been ruled out, 2 had a confirmed venous thromboembolic event in 3 months of follow-up (0.4%).
Why can the Wells DVT score be negative?
An alternative diagnosis at least as likely as DVT subtracts 2 points. A patient with that and no other feature scores −2, which is low probability in the 1997 model and DVT unlikely in the 2003 model.
What changed between the 1997 and 2003 versions?
The 2003 version added previously documented DVT as a one-point item, changed bedridden from more than 3 days to 3 days or more, extended the surgery window from 4 to 12 weeks for surgery under general or regional anaesthesia, and replaced the three categories with two, split at 2 points.
Is my data sent anywhere?
No. The calculation runs in your browser, and the boxes you tick are not sent or stored.
Sources
- Wells PS, Anderson DR, Bormanis J, et al. Value of assessment of pretest probability of deep-vein thrombosis in clinical management. Lancet. 1997;350(9094):1795–1798. doi:10.1016/S0140-6736(97)08140-3
- Wells PS, Anderson DR, Rodger M, et al. Evaluation of D-dimer in the diagnosis of suspected deep-vein thrombosis. N Engl J Med. 2003;349(13):1227–1235. doi:10.1056/NEJMoa023153
- Min SK, Kim YH, Joh JH, et al. Diagnosis and treatment of lower extremity deep vein thrombosis: Korean practice guidelines. Vasc Specialist Int. 2016;32(3):77–104. doi:10.5758/vsi.2016.32.3.77