Wells Score for Pulmonary Embolism
What is the Wells Score?
The Wells score is a validated clinical prediction rule consisting of 7 variables that stratifies patients with suspected PE into risk categories to guide diagnostic testing decisions. 1 The score can be used in either a three-level classification (low, intermediate, high probability) or a two-level classification (PE-unlikely vs PE-likely). 1
How to Calculate the Wells Score
The Wells score assigns points based on the following clinical variables:
- Clinical signs and symptoms of DVT: 3 points 2
- PE is the most likely diagnosis (or equally likely): 3 points 2
- Heart rate >100 bpm: 1.5 points 2
- Immobilization ≥3 days or surgery in previous 4 weeks: 1.5 points 2
- Previous PE or DVT: 1.5 points 2
- Hemoptysis: 1 point 2
- Malignancy (active or treated within 6 months): 1 point 2
Risk Stratification and PE Probability
Three-Level Classification:
- Low probability (Wells score 0-1): ~3-4% prevalence of PE 1, 2
- Intermediate probability (Wells score 2-6): ~13-30% prevalence of PE 1, 2
- High probability (Wells score >6): ~36-67% prevalence of PE 1, 2
Two-Level Classification (Dichotomized):
- PE-unlikely (Wells score ≤4): ~12% prevalence of PE 1
- PE-likely (Wells score >4): ~30% prevalence of PE 1
Clinical Application in Diagnostic Algorithms
For patients with low Wells score (<2) or PE-unlikely (≤4), combine with D-dimer testing to safely exclude PE without imaging. 1 The negative predictive value of this combination is 99.5%, allowing approximately 30% of patients to avoid CT pulmonary angiography. 2
Recommended Diagnostic Approach:
- Wells score ≤4 + negative D-dimer: PE can be safely ruled out; no further imaging needed 1, 2
- Wells score ≤4 + positive D-dimer: Proceed to CT pulmonary angiography 1
- Wells score >4: Proceed directly to CT pulmonary angiography (D-dimer optional) 1
For patients >50 years old, consider using age-adjusted D-dimer cutoff (age × 10 μg/L) to increase diagnostic utility and reduce unnecessary imaging. 2
Performance Characteristics
The Wells score demonstrates good discriminative ability with an area under the ROC curve of 0.74-0.85 across multiple validation studies. 3, 4 When combined with D-dimer testing in low-risk patients, the failure rate (missed PE) is approximately 1.5%, which is considered clinically acceptable. 5
The Wells score performs better than the Revised Geneva Score in most comparative studies, with superior overall accuracy (AUC 0.85 vs 0.76, p=0.005). 3 The Wells score shows higher specificity (67.5% vs 47.0%, p=0.002) though slightly lower sensitivity compared to Geneva. 6
Important Caveats and Limitations
Subjective Elements:
The Wells score contains subjective components, particularly "PE is the most likely diagnosis," which requires clinical judgment. 1 Interrater reliability is moderate for some elements (DVT symptoms κ=0.54, immobilization κ=0.41) but very good for others (malignancy κ=0.87, tachycardia κ=0.94). 1
Special Populations:
In patients with known DVT who are being evaluated for coexisting PE, the Wells score may not perform reliably and should be interpreted with caution. 7 In this specific subset, tachycardia and hemoptysis were the only predictive variables. 7
Gestalt vs Wells Score:
While physician gestalt (unstructured clinical judgment) performs comparably to the Wells score in some studies, the Wells score is more efficient, allowing PE exclusion in 45% of patients compared to only 25% with gestalt alone. 5 Both approaches have similar failure rates (~1.3-1.5%) when combined with D-dimer testing. 5
Comparison with Alternative Scoring Systems
The Wells score and Revised Geneva Score show similar diagnostic performance in direct comparisons, with no significant differences in most studies. 1 However, the Wells score is more widely validated and adopted in clinical practice. 1 The Pisa model showed superior performance in one small study (AUC 0.94) but has not been as extensively validated. 1