Glucose variability indices predict the episodes of nocturnal hypoglycemia in elderly type 2 diabetic patients treated with insulin

Abstract

AimO determine the applicability of glucose variability (GV) indices derived from continuous glucose monitoring (CGM) data for prediction of nocturnal hypoglycemia (NH) in elderly patients with type 2 diabetes treated with insulin. MethodsWe observed 83 insulin-treated in-patients, 65–80 years of age. Blinded CGM data for 176 nights were analyzed. Daytime (06:00–22:59) mean glucose, Standard Deviation (SD), 2-h Continuous Overlapping Net Glycemic Action (CONGA2) and Mean Absolute Glucose (MAG), pre-midnight (23:00–23:59) mean glucose, SD and MAG, 24-h Mean Amplitude of Glucose Excursions (MAGE), were estimated. Pre-midnight glucose trends were estimated as the absolute difference between glucose values at 23:00 and 23:59 (deltaG). Episode of interstitial glucose ≤70mg/dL observed from 0:00 to 5:59 was considered as NH. ResultsNH was present in 68 out of 176 24-h recordings (39%). Lower daytime mean glucose and CONGA2, and higher MAG values were found in patients with NH as compared to those without (p=0.0002, p=0.0001 and p=0.02, respectively). Pre-midnight mean glucose was lower, while pre-midnight deltaG was higher in patients with NH (p<0.0001 and p=0.02). Antecedent daytime hypoglycemia increased the risk of NH (p<0.0001). In logistic regression analysis, the combination of daytime MAG and pre-midnight mean glucose was the most reliable predictor of subsequent NH (accuracy 75.6%, p=0.0004). ConclusionThe analysis of CGM-derived GV parameters could improve prediction of NH in elderly patients with type 2 diabetes treated with insulin.