Validation of the IMPROVE bleeding risk assessment model in surgical patients: Results from the DissolVE-2 Study.

IMPROVE Bleeding Risk Score (BRS) is known to be validated and widely accepted in medical patients. However, its relevance in surgical patients has so far not been explored. External validation of the IMPROVE BRS on bleeding in surgical patients can hopefully improve clinical practice (for surgical patients). Data from 6986 surgical patients were collected from the DissolVE-2 cohort. The Kaplan-Meier method was used to assess the incidences of major bleeding and any bleeding among surgical patients within 14 days of admission. A cut-off value of BRS ≥7 indicated a higher risk of bleeding. Risk factors associated with major and any bleeding were analysed by the Cox regression method. Model discrimination was evaluated by area under the receiver operator characteristic curves (AUC). Calibration curves and Hosmer-Lemeshow χ2 statistics were used to measure the difference between predicted and observed bleeding risks. A total of 6399 surgical patients were included in the final validation cohort. The cumulative incidence rate of any bleeding was 3.9 % (95 % confidence interval [CI], 3.4–4.5), of which the incidence rate of major bleeding was 1.2 % (95 % CI, 0.9–1.6). Among patients with a BRS of ≥7, 16.3 % reported any bleeding, and 26.3 % reported major bleeding. The IMPROVE BRS had a better discriminative power (AUC = 0.69) and excellent goodness of fit (Hosmer-Lemeshow test, P = 0.208) for the prediction of major bleeding events as compared with any bleeding (AUC = 0.55; Hosmer-Lemeshow test, P = 0.004). The calibration plot suggested a more accurate prediction for major bleeding events. Moreover, the IMPROVE BRS had a higher AUC value of 0.83 and better goodness of fit (P = 0.2616) for major bleeding in patients undergoing abdominal surgery than other surgery types. The IMPROVE BRS is a simple and practical technique that can help in predicting the risk of major bleeding in surgical patients, improving functional and safety outcomes of hospitalized patients with surgery. [ABSTRACT FROM AUTHOR]