Mean amplitude of glycemic excursion and mortality in critically ill patients: A retrospective analysis using the MIMIC-IV database

BackgroundGlycemic variability (GV) is increasingly recognised as a critical determinant of outcomes in critically ill patients. However, standardised criteria for assessing GV remain undefined.ObjectiveThis study aimed to evaluate the relationship between the Mean Amplitude of Glycemic Excursion (MAGE) and mortality in intensive care unit (ICU) patients, and to determine optimal MAGE thresholds for distinct patient populations.MethodsA retrospective cohort of 13 852 critically ill adults with ICU stays exceeding 24 h was analysed. Patients were stratified into MAGE quartiles, and various GV metrics were compared for their predictive performance on mortality. Multivariable-adjusted models were employed to examine associations between MAGE and mortality outcomes.ResultsPatients in higher MAGE quartiles exhibited significantly elevated mortality risks, with the highest quartile associated with ICU mortality (HR 3.59 [95% CI: 2.99-4.31]), in-hospital mortality (HR: 3.43 [95% CI: 2.92-4.02]) and 28-day mortality (HR 2.04 [95% CI: 1.47-2.82]). The relationship between MAGE and mortality was notably stronger in non-diabetic patients (HR: 3.36 [95% CI: 2.90-3.89]) compared to diabetic patients (HR: 1.59 [95% CI: 1.33-1.91]). Restricted cubic spline analyses identified optimal MAGE thresholds of 44.28 mg/dL for the overall population, 58.97 mg/dL for diabetic patients and 17.11 and 37.72 mg/dL for non-diabetic patients. MAGE demonstrated effective predictive performance for all-cause mortality (AUC: 0.6286 [95% CI: 0.6171-0.6400]) compared to other GV metrics. Incorporating MAGE into prognostic models alongside SAPS II and SOFA scores improved performance for all-cause mortality, with net reclassification improvement (NRI) of 0.238 and integrated discrimination improvement (IDI) of 0.008.ConclusionMAGE exhibits effective predictive value for mortality in ICU patients, with distinct thresholds for diabetic and non-diabetic populations. These findings underscore the importance of tailored GV management strategies in critical care settings and support the adoption of MAGE as a standardised metric for GV assessment in ICU settings.