Mortality risk stratification in contemporary myocardial infarction–related cardiogenic shock: development of AMICS-ADMIT prediction model and benchmarking of established scores

Abstract

Background Mortality risk stratification in acute myocardial infarction complicated by cardiogenic shock (AMI-CS) remains challenging with heterogeneous temporary mechanical circulatory support (tMCS) use. We developed and temporally validated the admission-based AMICS-ADMIT model and bedside score and benchmarked them against established scores. Methods This retrospective single-centre study, reported according to TRIPOD+AI, included 479 consecutive patients with AMI-CS undergoing percutaneous coronary intervention (PCI) between January 2017 and July 2025. The primary endpoint was in-hospital mortality. AMICS-ADMIT was developed using multivariable logistic regression in the 2017–2022 cohort, internally evaluated by bootstrap resampling, and temporally validated in the 2023–2025 cohort. Missing data were multiply imputed. Performance was assessed using discrimination, calibration, and Brier scores. Benchmarking against CardShock, IABP-SHOCK II, and GRACE was restricted to temporal validation; tMCS-stratified performance was exploratory. Results In-hospital mortality was 42.6% (204/479). The final model included age, left ventricular ejection fraction, prior cardiopulmonary resuscitation, estimated glomerular filtration rate, vasoactive–inotropic score, and ST-segment elevation status. The continuous model achieved AUCs of 0.809 (95% CI 0.764–0.855) in development and 0.834 (95% CI 0.769–0.899) in temporal validation, with Brier scores of 0.176 and 0.167 and favourable calibration. The bedside score achieved AUCs of 0.777 (95% CI 0.728–0.826) and 0.775 (95% CI 0.699–0.851), respectively, with stepwise mortality separation across predefined risk categories. Discrimination did not differ significantly from CardShock, IABP-SHOCK II, or GRACE. Conclusions AMICS-ADMIT showed stable temporal performance and clinically interpretable early risk stratification in contemporary PCI-treated AMI-CS. External multicentre validation is required before clinical implementation.