Stress hyperglycemia ratio-integrated index and 28-day mortality in sepsis: a dual-cohort study

Background Stress hyperglycemia is common in sepsis, but admission glucose is influenced by antecedent glycemia. The stress hyperglycemia ratio-integrated index (SHRI) combines hemoglobin A1c (HbA1c)-normalized glycemia with serum potassium. We examined the association between early SHRI and 28-day mortality in adults with sepsis. Methods This retrospective dual-cohort study included 4,495 adults with Sepsis-3-defined sepsis from the Medical Information Mart for Intensive Care (MIMIC-IV) database and 320 patients from an independent hospital validation cohort. SHRI was calculated using HbA1c measured from 90 days before to 24 hours after intensive care unit (ICU) admission and glucose and potassium measured within 6 hours before or after ICU admission; the same exposure definition was applied in both cohorts. The primary outcome was 28-day all-cause mortality. Multivariable Cox proportional hazards models adjusted for age, sex, previously diagnosed diabetes, and day-1 Sequential Organ Failure Assessment (SOFA) score, and restricted cubic splines characterized dose–response associations. Results Within 28 days, 903 patients (20.1%) died in MIMIC-IV and 64 (20.0%) died in the validation cohort. In MIMIC-IV, the adjusted hazard ratio per 0.1 increase in SHRI was 1.07 (95% confidence interval [CI], 1.04–1.11). The corresponding hazard ratio in the validation cohort was 2.00 (95% CI, 1.36–2.94). Restricted cubic spline analyses showed an overall association in both cohorts, with a nonlinear pattern in MIMIC-IV and an approximately linear pattern in the validation cohort. Conclusion Higher early SHRI was associated with higher 28-day mortality in both sepsis cohorts. SHRI may complement established severity measures as an integrated marker of relative hyperglycemia and potassium status.