Medical AIWells criteria pulmonary embolism calculator
Seven weighted clinical criteria estimate the pretest probability of pulmonary embolism. The result can be read three-tier (low, moderate, high) or two-tier (PE unlikely at 4 or less, PE likely above 4), which guides whether D-dimer or direct imaging is the next step.
How to measure each input
- Clinical signs of DVT
- Look for unilateral leg swelling and pain on deep-vein palpation. This item is scored on examination, not on a previous DVT (that is a separate criterion).
- PE most likely diagnosis
- This is the clinician's gestalt: is PE at least as likely as any alternative explanation for the presentation? It is the most subjective and most heavily weighted item.
- Heart rate
- Use a resting heart rate above 100 beats/min recorded at any point during the encounter.
- Immobilisation or recent surgery
- Immobilisation means bed rest (other than to use the bathroom) for 3 or more days; surgery means a procedure under general or regional anaesthesia within the prior 4 weeks.
- Previous DVT or PE
- Counts only when the prior event was objectively confirmed, not merely suspected or treated empirically.
Interpretation
| Band | Meaning |
|---|---|
| < 2, low probability | Low pretest probability (around 2 to 6% prevalence). A negative D-dimer reliably excludes PE without imaging. In the two-tier model this also falls in the PE unlikely group. |
| 2 to 6, moderate probability | Moderate pretest probability (around 20% prevalence). Use D-dimer to triage: a negative result can defer imaging, a positive result prompts CT pulmonary angiography. Scores of 2 to 4 are still PE unlikely on the two-tier model, scores of 5 to 6 are PE likely. |
| > 6, high probability | High pretest probability (around 50 to 65% prevalence). Proceed directly to CT pulmonary angiography; D-dimer should not be used to rule out PE here. This is PE likely on the two-tier model. |
Pitfalls, exclusions and caveats
- The score guides next steps, it does not diagnose or exclude PE. Pair it with D-dimer (age-adjusted where validated) or imaging according to the probability band.
- Do not use D-dimer to exclude PE when the probability is high: the false-negative rate is unacceptable, so image directly.
- The PERC rule applies only when gestalt probability is already low, not as a substitute for the Wells score in moderate or high probability patients.
- The subjective 'PE most likely' item drives much of the variability between clinicians; document the alternative diagnoses considered.
- Wells was derived in adults; it is not validated in pregnancy (where the YEARS or pregnancy-adapted algorithms are preferred) or in children.
- A simplified Wells (each item scored as 1 point, threshold of 1 for unlikely) exists and uses different cut-offs, so do not mix scoring schemes.
Sum of: clinical signs/symptoms of DVT (3) + PE is the most likely diagnosis or equally likely (3) + heart rate > 100 (1.5) + immobilisation >= 3 days or surgery in the prior 4 weeks (1.5) + previous DVT or PE (1.5) + haemoptysis (1) + active malignancy (1). Three-tier: < 2 low, 2 to 6 moderate, > 6 high. Two-tier: <= 4 PE unlikely, > 4 PE likely.The Wells criteria for PE were derived by Wells et al. (2000). This implementation is an educational tool and is not affiliated with the original authors or any guideline body.
Frequently asked
What is the difference between the two-tier and three-tier Wells models?
The three-tier model splits probability into low (< 2), moderate (2 to 6), and high (> 6). The two-tier model collapses these into PE unlikely (4 or less) and PE likely (above 4), which pairs cleanly with a D-dimer rule-out strategy. This calculator shows both.
Can I use D-dimer at any Wells score?
Use D-dimer to exclude PE only at low or moderate probability (or in the PE unlikely group). At high probability the false-negative rate is too high, so go straight to CT pulmonary angiography.
How does Wells relate to the PERC rule?
PERC is applied only after gestalt clinical probability is already judged low, to decide whether even a D-dimer can be skipped. It does not replace the Wells score in moderate or high probability patients.
A score is a start. The product is asked.
Calculators give a number. When the patient in front of you needs the reasoning behind it, with the sources, the product is what does the looking-up.
Medical AI returns evidence-grounded answers backed by real citations. It is a reference tool, and these terms describe how it should and should not be used.
01 Educational purpose
Medical AI is an information and reference tool intended for educational use only. The answers it returns are not medical advice, diagnosis, or treatment. Always consult a qualified doctor or healthcare professional with any question concerning a medical condition.
02 Professional use
Medical AI is designed for use by practicing clinicians. It is not intended for direct patient use and is not a substitute for professional clinical judgment. Apply your own training and current guidelines to every decision an answer informs.
03 Privacy and data
We do not collect, store, or process personally identifiable patient information (PHI or PII). Do not enter names, dates of birth, medical record numbers, or any other patient identifiers into the composer.
04 Information currency
Medical knowledge evolves rapidly. Citations carry their publication date, so consult the primary source and the most recent clinical guideline before acting on anything material.