Admission hemoglobin–shock index phenotypes and 365-day mortality in acute myocardial infarction: a dual-cohort study

Abstract Background Anemia and hemodynamic instability are associated with adverse outcomes in acute myocardial infarction (AMI), but their combined impact is unclear. We evaluated whether an admission hemoglobin–shock index (Hb–SI) phenotype predicts long-term mortality and adds prognostic value. Methods A total of 9222 patients with AMI were analyzed from a tertiary cohort ( n = 4167) and the MIMIC-IV database ( n = 5055). Hemoglobin (Hb) and shock index (SI) were categorized using prespecified thresholds (Hb < 11 g/dL; SI ≥ 0.9) to define four phenotypes. The primary outcome was 365-day mortality. Kaplan–Meier curves and Cox proportional hazards models were used to assess phenotype-associated mortality risk. Results During 365-day follow-up, 196 deaths occurred in the clinical cohort and 502 in the MIMIC-IV cohort. The double-constraint phenotype had the highest absolute mortality in both cohorts. After expanded adjustment, high-drive remained associated with higher mortality in the clinical cohort (HR, 1.82; 95% CI 1.18–2.83), with a borderline association in MIMIC-IV (HR, 1.36; 95% CI 1.00–1.86). Double-constraint was attenuated and no longer significant in the clinical cohort (HR, 1.75; 95% CI 0.71–4.31), but remained associated with higher mortality in MIMIC-IV (HR, 1.51; 95% CI 1.17–1.95). Low reserve showed discordant associations, with higher mortality in the clinical cohort but lower mortality in MIMIC-IV. Continuous modeling showed significant Hb × SI interaction in both cohorts. Addition of the Hb–SI phenotype to the GRACE model produced a statistically significant but numerically small increase in discrimination (ΔAUC = 0.006; P = 0.0187). Conclusion In AMI, the prognostic significance of Hb depends on the hemodynamic context. Hb–SI phenotypes may provide a simple physiological framework for early risk characterization, but low reserve showed cohort heterogeneity and the incremental value beyond GRACE was modest. Prospective validation is needed.