Association of continuous glucose monitoring with in-hospital adverse outcomes among critically ill patients: a real-world propensity score-matched study.

BACKGROUND: Critically ill patients often experience substantial glycemic fluctuations, which are associated with adverse clinical outcomes. Conventional point-of-care glucose monitoring provides only intermittent measurements and may delay timely therapeutic adjustment. Continuous glucose monitoring (CGM) enables real-time assessment of glucose trends and may improve glycemic management in intensive care settings. METHODS: This single-center, retrospective, real-world propensity score-matched cohort study included 592 critically ill adults admitted to the intensive care unit. Patients receiving CGM-guided glucose management were matched 1:1 with patients receiving standard point-of-care glucose monitoring. The primary outcome was in-hospital mortality. Secondary outcomes included ICU length of stay, hypoglycemia, severe hypoglycemia, mechanical ventilation duration, and nosocomial infection. Outcomes were analyzed using regression models appropriate to outcome type, with adjustment for prespecified covariates. RESULTS: After propensity score matching, 296 patients were included in the CGM group and 296 in the control group. Compared with standard point-of-care monitoring, CGM use was associated with lower in-hospital mortality (9.5% vs 15.5%; adjusted odds ratio [aOR] = 0.61, 95% CI 0.38-0.98; P = 0.027) and fewer severe hypoglycemic events (4.7% vs 10.8%; aOR = 0.46, 95% CI 0.25-0.85; P = 0.009). The CGM group also had a shorter median ICU length of stay (7 days [IQR 5-10] vs 9 days [IQR 6-13]; P = 0.021) and a lower incidence of nosocomial infection (12.2% vs 17.9%; aOR = 0.63, 95% CI 0.40-0.99; P = 0.048). CONCLUSION: In this retrospective propensity score-matched cohort study, CGM-guided glucose management was associated with improved glycemic safety and more favorable in-hospital outcomes among critically ill adults who remained in the ICU for at least 72 hours. Because of the observational design, survivor-selection bias, residual confounding, and differences in measurement density, these findings should not be interpreted as evidence of causal benefit. Prospective multicenter studies are needed.