Glycemic variability improves risk prediction and reveals a mortality threshold in orthopedic trauma patients with sepsis

BACKGROUND: Sepsis is a life-threatening complication in patients with orthopedic trauma, associated with significant mortality. Dysregulated glucose metabolism is a hallmark of critical illness, but the specific role of glycemic variability (GV), beyond hyperglycemia, in this high-risk population remains inadequately explored.Kindly check and confirm the author and their respective affiliations 9 and 10 are correctly identified.Some changes have been made. Please refer to the eProofing system. METHODS: This retrospective cohort study utilized 1946 patients from the Beth Israel Deaconess Medical Center between 2008 and 2022. We included adult intensive care unit (ICU) patients with a primary diagnosis of orthopedic trauma and concurrent sepsis. GV was quantified as the coefficient of variation of all blood glucose measurements during the ICU stay. The primary outcome was all-cause mortality at in-hospital, 30-day, 90-day, and 365-day. Restricted cubic spline (RCS) analysis was used to determine the detailed relationship between GV and mortality and the optimal GV threshold for dichotomization. Propensity score matching and multivariable logistic regression were employed to control for confounders. RESULTS: The study cohort experienced substantial mortality, with rates of 16.0% (in-hospital), 19.8% (30-day), 26.4% (90-day), and 33.9% (365-day). RCS analysis revealed a significant nonlinear, J-shaped relationship between GV and all mortality end points (p < 0.001), characterized by an initial modest increase in risk followed by a sharp escalation beyond a specific threshold. The optimal GV thresholds for predicting in-hospital, 30-day, 90-day, and 365-day mortality were remarkably consistent (0.218, 0.217, 0.205, and 0.210, respectively), leading to a unified average dichotomization threshold of 0.213. After propensity score matching created a well-balanced cohort, high GV remained independently associated with significantly increased risks of mortality across all time points in the fully adjusted model (adjusted odds ratio (OR) for in-hospital mortality: 1.969, 95% confidence interval (CI): 1.379-2.811, p < 0.001; 30-day mortality OR: 1.744, 95% CI: 1.245-2.442, p = 0.001; 90-day mortality OR: 1.735, 95% CI: 1.268-2.375, p < 0.001; 365-day mortality OR: 1.815, 95% CI: 1.352-2.438, p < 0.001). Furthermore, adding GV significantly improved the predictive performance of the Oxford Acute Severity of Illness Score (OASIS) (area under the receiver operating characteristic curve (AUC) increased from 0.684 to 0.699 for in-hospital mortality, p = 0.008) and Charlson Comorbidity Index scores (AUC increased from 0.681 to 0.701, p = 0.001). CONCLUSIONS: In critically ill septic patients with orthopedic trauma, elevated GV is a potent, independent predictor of short- and long-term mortality, exhibiting a J-shaped relationship with mortality risk. GV values above 21.3% are associated with significantly higher mortality risk, and this threshold may serve as a useful prognostic marker for risk stratification. Furthermore, GV provides incremental prognostic value beyond established risk scores, highlighting its potential as a key metric for risk stratification.