Development and external validation of an admission-based model for 180-day mortality in diabetic acute myocardial infarction.

BACKGROUND

Patients with diabetes mellitus and acute myocardial infarction (DM-AMI) are at high risk, but risk heterogeneity within this subgroup remains insufficiently characterized. Whether a parsimonious admission-based model provides prognostic information beyond GRACE remains uncertain. We aimed to assess admission risk heterogeneity in DM-AMI and develop and externally validate a model for 180-day mortality using routine variables.

METHODS

This retrospective dual-cohort study included AMI patients from a Beijing tertiary hospital (development cohort) and an ICU-based MIMIC-IV cohort (external validation). A Cox model was built using admission variables. Model performance was evaluated using discrimination, calibration, and decision curve analysis with internal bootstrap validation and comparison with GRACE.

RESULTS

Among 4,167 patients, 1,514 had DM-AMI with 90 deaths at 180 days. The model included eight variables (age, heart rate, SBP, glucose, BUN, hemoglobin, RDW, WBC). It showed good discrimination in the development cohort (C-index 0.848; AUC 0.857) and maintained performance externally (AUC 0.744). GRACE comparison showed similar discrimination, while the combined model improved performance (ΔAUC = 0.014, P = 0.031). Calibration in MIMIC-IV indicated risk overestimation (slope 0.527; O/E 0.652).

CONCLUSION

DM-AMI patients exhibit marked admission risk heterogeneity. The model provides complementary prognostic information beyond GRACE but requires recalibration for absolute risk use in external settings.