Comparative performance of the ACUITY and CRUSADE bleeding risk scores for in-hospital bleeding prediction among medically managed patients with acute myocardial infarction: A retrospective cohort study
Chia Siang Kow, Kaeshaelya Thiruchelvam, Jia Le Lim, Olivia Hui Min Sim, Li Ann Yeoh, Sasreeta Paniselvam, Abdullah Faiz ZaihanObjectives
To compare Acute Catheterization and Urgent Intervention Triage strategY (ACUITY) and Can Rapid risk stratification of Unstable angina patients Suppress ADverse outcomes with Early implementation of the ACC/AHA guidelines (CRUSADE) for predicting clinically significant in-hospital bleeding among patients with acute myocardial infarction managed in a non-percutaneous coronary intervention secondary-care setting.
Design
Retrospective cohort study.
Setting
A Malaysian secondary public hospital without on-site percutaneous coronary intervention facilities.
Participants
Adults with ST-elevation myocardial infarction or non-ST-elevation myocardial infarction admitted during 2024.
Main outcome measures
Clinically significant in-hospital bleeding; inter-score agreement, discrimination, calibration, threshold-based sensitivity and specificity, category-based net reclassification improvement (NRI) and exploratory simultaneous logistic regression.
Results
Among 842 patients, 143 (17.0%) experienced bleeding. Agreement between ACUITY and CRUSADE risk categories was fair (κ = 0.270; 95% confidence interval (CI) 0.228–0.313). ACUITY and CRUSADE showed modest discrimination (C-statistics 0.632 and 0.615, respectively; DeLong p = 0.76). Hosmer–Lemeshow p-values were 0.070 for ACUITY and 0.050 for CRUSADE. At cut-offs of ACUITY ≥ 14 and CRUSADE ≥ 40, sensitivity was 76.2% versus 56.6%, and specificity was 46.2% versus 63.4%, respectively; separate paired McNemar tests were significant for both comparisons (both p < 0.001). Global NRI was 0.024 (95% CI −0.061 to 0.108; p = 0.586), indicating no statistically significant overall reclassification improvement. In exploratory simultaneous logistic regression, ACUITY remained associated with bleeding (p = 0.012), whereas CRUSADE did not (p = 0.264).
Conclusions
Both scores showed modest discrimination. At ACUITY ≥ 14 and CRUSADE ≥ 40, ACUITY was more sensitive and CRUSADE more specific, without evidence of a clear overall advantage. Neither score should be used alone to guide management decisions.