The Association of Continuous Glucose Monitoring Metrics with Hospital-Related Clinical Outcomes in Nonintensive Care Units.

Abstract

Aims: We investigated the association between continuous glucose monitoring (CGM) metrics and clinical outcomes in the nonintensive care unit (non-ICU) setting. Methods: In this observational cohort study, patients on non-ICU floors wore blinded Dexcom G6 Pro CGM. CGM metrics and occurrence of CGM-detected severe hypoglycemia were measured. Clinical data, including infection, diabetic ketoacidosis, renal replacement therapy, thrombosis, and 30-day post-discharge readmissions and emergency department (ED) visits were identified from the medical record and participant phone interview. Multivariate regression assessed predictors of CGM-detected severe hypoglycemia and the associations between CGM metrics and clinical outcomes. Regression models using CGM data or reference glucose data were compared with receiver operating characteristic (ROC) curves. Results: A total of 326 hospitalized adults were enrolled with median % time in range 70-180 mg/dL 44.5% (17.1, 70.2%), % time above range >180 mg/dL 54.8% (28.8, 82.3%), and % time below range 0.6% (0, 0.2%). Predictors of severe hypoglycemia included type 1 diabetes, female gender, lower admission hemoglobin, lower A1c, and longer hospital stay. Regression analyses demonstrated an association of 30-day ED visits with increased %TAR (P = 0.01). ROC curves showed models using CGM data or reference data predicted clinical outcomes similarly. Conclusions: CGM can be useful in identifying patients at risk of inpatient hypoglycemia and 30-day ED visits.