Stress hyperglycemia ratio and short-term mortality in critically ill septic patients: stratified analysis by diabetes status

BACKGROUND: Sepsis is a dysregulated immune response to infection, often leading to metabolic abnormalities. The stress hyperglycemia ratio (SHR), a reliable predictive indicator, has been widely used in the prediction of adverse outcomes in cardiovascular and cerebrovascular diseases. However, studies on the relationship between the SHR and adverse outcomes in critically ill septic patients are limited. Therefore, this study aimed to investigate the relationship between the SHR and short-term mortality in critically ill septic patients. METHODS: A retrospective analysis of septic patients from the MIMIC-IV database was conducted. The Boruta algorithm was used to identify key features that predict short-term mortality in septic patients. Time-varying Cox proportional hazards model, subgroup analysis, restricted cubic splines (RCSs), and Kaplan‒Meier (K‒M) survival analysis were used to assess the relationships between the SHR and 28-day and 90-day all-cause mortality, along with a subgroup analysis based on diabetic status. RESULTS: This study included 15,876 patients with sepsis, 60.6% of whom were male. Boruta feature selection identified the stress hyperglycemia ratio (SHR) as an important clinical variable. Multivariable Cox proportional hazards models with time‑varying coefficients showed a positive association between SHR and both 28‑day and 90‑day mortality (HR = 1.683, 95% CI 1.489–1.903, P < 0.0001; HR = 1.533, 95% CI 1.397–1.681, P < 0.0001). Restricted cubic spline (RCS) analyses further supported a linear relationship between SHR and 28‑day and 90‑day mortality. Subgroup analyses suggested that the association between SHR and short‑term mortality may be stronger in non‑diabetic than in diabetic patients. Sensitivity analyses were consistent, indicating robustness of the findings. CONCLUSION: The SHR is closely associated with short-term mortality in septic patients, suggesting that it could serve as a potential predictive factor for early risk assessment and clinical intervention.