Trauma surgical intensive care unit (TSICU) patients frequently present with metabolic dysregulation. However, the combined prognostic value of the triglyceride-glucose–body mass index (TyG-BMI) and the stress hyperglycemia ratio (SHR) remains unclear. A retrospective cohort study was conducted using the MIMIC-IV database, with external validation in a TSICU cohort from Tongji Hospital. Adult patients admitted to the TSICU were classified into four groups based on joint TyG-BMI and SHR categories. The endpoints were 28-day and 180-day all-cause mortality. Cox regression, restricted cubic splines, Kaplan–Meier survival curves, and 33 machine learning models were employed to evaluate associations and predictive performance. A total of 872 patients were included in the training cohort and 525 in the validation cohort. Overall, 19.0% and 25.8% of patients died within 28 and 180 days, respectively. Joint SHR–TyG-BMI stratification revealed distinct prognostic profiles. Patients with high SHR and low TyG-BMI had the highest mortality risk, while those with low SHR and high TyG-BMI had the most favorable outcomes. TyG-BMI demonstrated a U-shaped association with short- and mid-term mortality, whereas SHR alone showed limited prognostic value. Machine learning analyses confirmed SHR and TyG-BMI as top predictors, outperforming traditional severity scores. Subgroup analyses across age, diabetes, hypertension, and cardiovascular disease consistently supported their prognostic utility. TyG-BMI and SHR are complementary predictors of short- and mid-term mortality in TSICU patients. High SHR combined with low TyG-BMI identifies patients at greatest risk, whereas preserved metabolic reserve mitigates the adverse effects of stress hyperglycemia. Integrating both acute and chronic metabolic indices may enhance risk stratification and guide individualized metabolic management in surgical intensive care.
Jiang et al. (Sat,) studied this question.