Answers are grounded in this paper and cite their sources.
Background: Early glycemic variability (GV) may be associated with mortality in sepsis. Methods: We analyzed 11,907 adult ICU patients meeting Sepsis-3 criteria in MIMIC-IV who had at least six glucose measurements during the first 24 h. Four metrics (SD, CV, MAGE, and GLI) were evaluated. Separate multivariable logistic regression models were fitted in the common sample with all four metrics available (n = 9364; 1499 deaths). Missing anthropometric data were addressed using 20 multiple imputations. Results: The cohort included 2104 deaths (17.7%) under the prespecified calendar-day definition of 28-day mortality. Adjusted ORs per 1-SD increase (95% CIs) were 1.23 (1.14–1.32) for SD, 1.26 (1.19–1.34) for CV, 1.14 (1.07–1.22) for MAGE, and 1.26 (1.19–1.34) for GLI. Pooled spline tests indicated nonlinearity for CV (p < 0.001) and MAGE (p = 0.027), but not SD (p = 0.206) or GLI (p = 0.172). Conclusions: Higher early GV was associated with 28-day mortality, but the observational design, patient selection, and outcome-definition uncertainty limit causal interpretation.
No takes yet. Share an insight, caveat, or question.
Miao et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: