Hospital length of stay (LOS) after myocardial infarction (MI) is a closely tracked performance metric for cardiology services and has generally declined over time. However, increasing patient complexity may be altering that trend. In an analysis published in The American Journal of Cardiology, a rising comorbidity burden emerged as the strongest factor associated with longer LOS, while a large language model (LLM) demonstrated strong performance in automated chart abstraction.

The study evaluated 6,129 MI admissions at a tertiary cardiac hospital between May 2018 and August 2024. An LLM was used to derive Charlson Comorbidity Index (CCI) scores from clinical case notes, with manual validation in 10% of admissions demonstrating strong agreement (R²=0.81). Time-to-discharge was assessed using Fine and Gray competing-risk models with in-hospital death treated as a competing outcome.

Mean CCI increased by 0.09 points annually (95% CI, 0.06-0.12), driven by higher rates of diabetes, hypertension, renal failure, lymphomas, and metastatic solid tumors. Mean LOS rose concurrently by 6.5 hours per year (95% CI, 3.9-9.1).

CCI was the strongest predictor of LOS (P < 2×10⁻16). Calendar year was not significant after full adjustment (P=0.17), suggesting rising patient complexity may account for the recent increase in MI hospitalization duration.