Early Glucose Variability Is Associated with Mortality in Critically Ill Children: A Retrospective Pediatric...

Background: Critical illness is frequently accompanied by dysregulated glucose homeostasis, resulting in increased glucose variability (GV). Beyond absolute hyperglycemia or hypoglycemia, early glycemic instability may reflect disease severity and metabolic stress in critically ill children. Objective: To evaluate the prognostic value of early GV indices during the first 72 h of pediatric intensive care unit (PICU) admission and to examine their association with organ dysfunction and adverse outcomes. Methods: This retrospective observational study included children admitted to the PICU between October 2022 and July 2024 with a length of stay greater than 72 h. Clinical and laboratory data were extracted from electronic health records. GV indices were calculated from glucose measurements obtained during the first 72 h after admission and included mean absolute glucose change (MAG), glycemic lability index (GLI), standard deviation (SD), coefficient of variation (CV), and average consecutive absolute change percentage (ACACP). Associations with severity of illness and outcome were assessed using correlation analysis, exploratory multivariable logistic regression, and receiver operating characteristic (ROC) analysis. Results: A total of 248 patients were included (mean age 7.1 ± 5.9 years; 143 [57.7%] male; mortality 7.3%), with a mean of 15.9 ± 4.0 glucose measurements per patient during the first 72 h. GV indices were strongly intercorrelated (all p < 0.001) and were associated with PELOD-2 and/or lactate levels (p < 0.05). Non-survivors had higher GV values than survivors. In exploratory multivariable logistic regression, SD (OR 4.82, 95% CI 2.2-10.4, p < 0.001) and PELOD-2 score (OR 1.68, 95% CI 1.3-2.2, p = 0.004) were associated with mortality. In ROC analysis, SD and PELOD-2 showed similarly strong discrimination for PICU mortality, with no significant difference by DeLong testing. GLI (AUROC 0.70, p = 0.004), ACACP (AUROC 0.68, p = 0.005), and CV (AUROC 0.66, p = 0.027) showed fair discrimination, whereas MAG and lactate were not significant predictors. Conclusions: In this single-center retrospective PICU cohort, early glucose variability during the first 72 h of admission was associated with illness severity and PICU mortality. Among the evaluated indices, SD showed the strongest association with mortality and remained associated with outcome together with PELOD-2 in an exploratory model. These findings should be interpreted as associative and hypothesis-generating; prospective multicenter studies with standardized glucose monitoring and formal incremental prediction analyses are required before GV can be incorporated into routine prognostic assessment.