Determinants of Continuous Glucose Monitoring Accuracy in Critically Ill Patients Requiring Mechanical Ventilation and Vasopressor Support

Background: Continuous glucose monitoring (CGM) accuracy in critically ill patients may be influenced by altered physiology and concurrent therapies. We assessed independent associations between sensor wear time, arterial blood glucose (ABG), noradrenaline-equivalent (NE) dose, erythrocyte volume fraction, pH, lactate, body-mass index and dialysis to CGM performance (Dexcom G6 ® ) in this setting. Methods: Secondary analysis of a prospective observational study including 40 critically ill adults receiving intravenous insulin, mechanical ventilation and vasopressor therapy. Paired ABG and CGM values ( n = 2946) were analyzed. The primary outcome was percentage absolute relative difference (ARD). Secondary outcomes were the proportions of CGM readings meeting ISO 15197:2013 and CLSI POCT12-A3 criteria. Multivariable generalized linear mixed models with patient-level random intercepts were fitted, applying B-splines for continuous covariates. Results: Median (IQR) ARD was 10.8% (5.2–18.8) (mean ARD 12.7 [95% CI 10.7–15.3] %); 64.5% and 56.0% of readings met ISO and CLSI criteria, respectively. Sensor wear time ( P < 0.001) and pH ( P ≤ 0.02) were independently associated with ARD and both categorical accuracy metrics. Predicted ARD decreased, and the probability of meeting ISO/CLSI criteria increased, during the first 50 h after sensor insertion and then stabilized. The pH–accuracy relationships were approximately linear; in linear-term sensitivity analyses, each 0.1-unit increase in pH was associated with a 1.5 percentage-point higher ARD ( P = 0.009) and 32% lower odds of meeting CLSI criteria ( P = 0.001). NE-dose ( P = 0.006) and ABG ( P < 0.001) were independently associated with ISO-defined accuracy. Spline analyses suggested reduced accuracy at higher NE-doses (>0.6 µg/kg/min) and lower ABG (<5 mmol/L [<90 mg/dL]), although observations were limited in these ranges. Conclusions: In critically ill patients receiving mechanical ventilation and vasopressor therapy, CGM accuracy improves substantially with increasing sensor wear time and is independently associated with arterial pH. Early postinsertion sensor instability and acid–base disturbances appear to be important determinants of CGM performance in critical illness.