Does using the Padua Prediction Score improve outcomes in preventing venous thromboembolism (VTE) in hospitalized patients?

Padua Prediction Score vs IMPROVE Score for VTE Prevention

The Padua Prediction Score and IMPROVE score are both acceptable risk assessment tools for VTE prophylaxis in hospitalized medical patients, but neither has been proven to improve hard clinical outcomes (mortality, VTE events) when implemented in real-world practice, despite increasing prophylaxis rates. 1

Guideline-Recommended Risk Assessment Models

Both the Padua Prediction Score (≥4 points = high risk) and IMPROVE VTE score (≥2 points = increased risk) are endorsed by major guidelines for stratifying VTE risk in hospitalized medical patients 1, 2. The American Society of Hematology 2018 guidelines present both models as valid options, with the Padua score identifying 39.7% of patients as high-risk (11% VTE incidence without prophylaxis vs 2.2% with prophylaxis) and the IMPROVE score showing a 5.7% VTE rate in high-risk patients (score ≥4) 1.

Critical Evidence Gap: Implementation Does Not Improve Outcomes

Despite guideline endorsement, the most recent and highest-quality implementation studies reveal that mandatory use of the Padua score does not reduce VTE events or mortality in real-world practice. A 2024 retrospective cohort of 18,890 Israeli patients found that Padua score compliance increased prophylaxis use (OR 1.66,95% CI 1.49-1.84) but showed no reduction in mortality (OR 1.13,95% CI 0.97-1.31) or VTE events (OR 1.22,95% CI 0.79-1.8) 3. Similarly, a 2022 Israeli study of 5,117 patients found no significant difference in VTE rates between high-risk (0.27%) and low-risk (0.28%) groups (p=0.768), and prophylactic treatment in high-risk patients did not improve outcomes 4.

Comparative Performance: Padua vs IMPROVE

When directly compared in a 2024 Chinese cohort of 42,257 medical inpatients, both scores showed similar modest predictive ability 5:

• Padua score: AUC 0.735 (95% CI 0.717-0.753), sensitivity 49.4%, specificity 89.6%
• IMPROVE score: AUC 0.711 (95% CI 0.693-0.729), sensitivity 32.5%, specificity 99.0%
• No statistical difference between models (DeLong test p=0.059) 5

The IMPROVE score offers more granular risk stratification (low/intermediate/high vs binary low/high) and more objective assessment criteria, which may facilitate implementation 5.

Poor Performance in Unselected Populations

A critical 2024 Veterans Affairs study of 1,252,460 consecutive hospitalizations demonstrated that both Caprini and Padua scores have poor predictive ability in unselected patient populations 6:

• Padua AUC for 90-day VTE: 0.59 (95% CI 0.58-0.59)
• Performance remained poor in both surgical (AUC 0.56) and non-surgical patients (AUC 0.59)
• This contrasts sharply with the original validation studies conducted in selected high-risk cohorts 6

Practical Clinical Approach

Given the evidence, use a universal prophylaxis strategy rather than relying solely on risk scores for decision-making 1, 2:

1. For all acutely ill hospitalized medical patients: Administer pharmacological prophylaxis with LMWH (preferred), UFH (5000 units SC three times daily), or fondaparinux unless bleeding contraindications exist 1, 2

2. Bleeding risk assessment is more critical than VTE risk scoring: Use IMPROVE bleeding score (≥7 = high bleeding risk) to identify patients requiring mechanical prophylaxis instead 1

3. Mechanical prophylaxis: For patients with active bleeding, platelet count <50,000/mcL, or high bleeding risk, use intermittent pneumatic compression devices (preferred over graduated compression stockings) 1, 2

4. Duration: Continue prophylaxis throughout hospitalization but do not extend beyond discharge 1, 2

Common Pitfalls to Avoid

• Do not withhold prophylaxis based solely on a low Padua or IMPROVE score in acutely ill medical patients, as real-world implementation studies show these scores miss clinically significant VTE events 3, 4
• Do not assume risk scores validated in selected cohorts will perform similarly in general hospital populations 6
• The original Padua validation study excluded many patients and had a 100:1 screening-to-enrollment ratio, limiting generalizability 1
• Important VTE risk factors like elevated CRP, lower Barthel Index, and elevated heart rate are not included in most RAMs 1

Special Populations

For patients with cirrhosis, both Padua (>3 or >4) and IMPROVE (>4) scores can predict VTE risk, though evidence is limited 1. In COVID-19 patients, universal prophylaxis is recommended over individualized risk assessment given high VTE rates 1.