What is the recommended scale to estimate the probability of pulmonary embolism (PE)?

Recommended Scales for Estimating Pulmonary Embolism Probability

The Wells score is the recommended clinical prediction rule for estimating the probability of pulmonary embolism (PE) due to its superior accuracy and extensive validation compared to other available scales. 1, 2

Comparison of Available PE Probability Scales

Wells Score

• Most widely used and extensively validated clinical prediction rule 1
• Demonstrates superior discriminative ability with area under the curve (AUC) of 0.85 compared to other scales 2
• Can be used in both three-category (low, moderate, high) or two-category (PE likely or unlikely) schemes 1
• Categorizes patients into:
  • Low risk: ~10% PE prevalence
  • Moderate risk: ~30% PE prevalence
  • High risk: ~65-93% PE prevalence 1, 2

Revised Geneva Score

• Fully standardized clinical decision rule based entirely on objective variables 1, 3
• Less discriminative ability than Wells score (AUC of 0.76) 2
• Has been simplified (Simplified Geneva Score) for easier clinical use 4
• Categorizes patients into:
  • Low risk: ~4-9% PE prevalence
  • Moderate risk: ~22-25% PE prevalence
  • High risk: ~56% PE prevalence 2

PERC (Pulmonary Embolism Rule-Out Criteria)

• Not a probability scale but a rule-out tool for very low-risk patients 1
• Designed to identify patients with <1.8% risk of PE in whom further testing can be safely avoided 1
• Should only be applied to patients already deemed low-risk by clinical assessment 1
• 97% sensitivity and 22% specificity for ruling out PE 1

Pisa Model

• More complex model with higher discriminative ability in some studies (AUC of 0.94) 1
• Relies on advanced chest radiograph interpretation skills beyond average physician capability 1
• Less practical for routine clinical use due to complexity

Clinical Application Algorithm

1. Initial Risk Assessment:

   • Use Wells score as the primary clinical prediction rule 1, 2
   • Alternative: Clinician gestalt assessment (shown to be comparable to structured scores but less standardized) 1, 5

2. For Low Probability Patients (Wells score <2):

   • Apply PERC criteria 1
   • If PERC negative (meets all 8 criteria): No further testing needed 1
   • If PERC positive: Proceed to D-dimer testing 1

3. For Intermediate Probability Patients (Wells score 2-6):

   • Proceed directly to D-dimer testing 1
   • Consider age-adjusted D-dimer cutoff for patients >50 years (age × 10 ng/mL) 1

4. For High Probability Patients (Wells score >6):

   • Proceed directly to imaging (CTPA or V/Q scan) without D-dimer testing 6

Important Considerations and Pitfalls

• The Wells score includes one subjective criterion ("alternative diagnosis less likely than PE"), which may affect interrater reliability 1
• Clinical experience appears to influence gestalt assessment accuracy, with more experienced clinicians sometimes being more cautious 1
• Only about half of clinicians familiar with prediction rules use them in >50% of applicable cases 1
• Special populations (e.g., pregnant women) may require specialized scores like the Pregnancy Adapted Geneva (PAG) Score 7
• D-dimer testing should not be used in high-probability patients or as a screening tool in patients without clinical suspicion of PE 1, 6

Conclusion

While multiple validated scales exist for estimating PE probability, the Wells score offers the best balance of accuracy, clinical utility, and extensive validation. The choice between the Wells score and clinical gestalt may depend on clinician experience, but structured approaches like the Wells score provide better standardization and are recommended for routine clinical practice.