A recent study validated the IMPROVE score for predicting bleeding risk in inpatients and guiding venous thromboembolism prophylaxis.
The study included prospective data on 1,668 patients admitted for a medical illness to the Water Reed Army Medical Center from September 2009 through March 2014. Researchers calculated the IMPROVE (International Medical Prevention Registry on Venous Thromboembolism) bleeding risk score for each patient using admission data. Results were published online by CHEST on July 23.
Bleeding events occurred during 2.7% of the admissions: major bleeds in 1.9% and non-major but clinically relevant bleeds in 0.8%. Overall, 20.7% of the studied patients had an IMPROVE score of 7 or higher, and Kaplan-Meier curves showed these patients to have a higher incidence of major and clinically important bleeding (P=0.02 and 0.06, respectively) than other patients. After adjustment for administration of chemical prophylaxis, a score of 7 or higher was significantly associated with major bleeds (odds ratio [OR], 2.6; 95% CI, 1.1 to 5.9; P=0.03), and there was a trend toward significance with clinically important bleeding (OR, 1.9; 95% CI, 0.9 to 3.7; P=0.07).
This first external validation of the IMPROVE score showed that it predicts major bleeding in medical inpatients and may help with assessing relative risks of bleeding and venous thromboembolism before prescribing prophylaxis, the authors concluded. This study's patient population differed from the score's derivation cohort, with more patients having cancer or requiring ICU admission, which strengthens the generalizability of the findings, they said. The lack of significant association between the score and clinically relevant bleeding events may be due to the study's limited ability to identify these events.
The IMPROVE bleeding score should be used in combination with assessment of venous thromboembolism risk to make prophylaxis decisions. One challenge is that a third of patients in this study would be classified as high risk for both bleeding and a clot, the authors noted. Future research should look for additional predictors of bleeding and identify thresholds for administering or withholding prophylaxis, they said.