Validating the Improve Venous Thromboembolism (Vte) Risk Score: Retrospective Analysis of Electronic Data from a Large Health System

Background:

Current VTE prophylaxis guidelines strongly promote risk assessment on the individual level. The IMPROVE VTE risk assessment model (RAM) is a composite score formulated to predict individual VTE risk during hospitalization. It is developed from a derivation cohort and has preliminary validation. This score sums 7 risk factors into risk categories: 0‐1—Low Risk, 2‐3—Moderate risk, >=4—high risk. In this study we aimed to validate the IMROVE VTE RAM using billing and EMR data from 2 tertiary medical centers.

Methods:

We identified medical discharges (December 2009 — April 2013) that met the IMROVE protocol (principal discharge diagnosis of a medical illness (CHF, COPD exacerbation, infection, rheumatologic condition, cancer), length of stay >3, age >18, INR=< 1.5, no surgery 90 days prior, no full anticoagulation, no VTE 90 days prior, no obstetrical or psychiatric secondary diagnosis). Cases were defined as a hospital acquired VTE (based on ICD9s) and confirmed by diagnostic study; matched controls were also identified. Risk factors for VTE were measured (age > 60, prior VTE, ICU admission, paralysis, immobility, hypercoagulability, cancer history), and the IMPROVE RAM was assessed

Results:

19,217 patients met inclusion and exclusion criteria. The VTE rate was 0.7%; with 135 cases and 405 controls. 3 risk factors were statistically association with the outcome: Age> 60 OR=1.76 95% CI (1.07‐2.90), prior cancer 3.20, (2.10‐4.86), prior VTE 3.23, (1.75, 5.96). The incidence rates and the 95% confidence intervals (CI) in the three risk groups were: Low risk: 0.33% (0.22‐0.45); Moderate: 0.91% (0.69‐1.13); High risk 1.51% (1.04‐1.99).. The C‐statistic (area under the ROC curve) was 0.702, suggesting a fair degree of discrimination.

Conclusions:

This study, closely resembling the IMPROVE derivation cohort, supports that the 3 category RAM can reliably differentiate low risk patients from those at higher risk for a hospital acquired VTE. With this, it may be reasonable to withhold prophylaxis from low risk patients. Wide application of this RAM will require validation in a prospective study.