Impact of real-time continuous glucose monitoring on glucose variability in critically ill patients

Introduction

Glucose variability has been found to be associated with mortality in critically ill patients, independent of the mean glucose concentration [1]. The aim of this analysis was to assess the impact of real-time continuous glucose monitoring (CGM) on glucose variability in critically ill patients receiving intensive insulin therapy (IIT).

Methods

This is the post hoc analysis of a prospective, randomized, controlled trial [2]. Data of 124 patients admitted to the ICU either receiving IIT according to a real-time CGM system (Guardian^®; Medtronic, Northridge, CA, USA) (n = 63) or according to an algorithm (n = 61) with selective arterial blood glucose measurements (simultaneously blinded CGM) for 72 hours were analysed. Insulin infusion rates were guided according to the same algorithm in both groups. Mean glucose and standard deviation, as a marker of glucose variability, were calculated for the first 24 hours (Glu_MEAN1, Glu_SD1) and for the whole study period (Glu_SD). Statistical comparison of parameters between study groups and between ICU survivors (n = 94) and nonsurvivors (n = 30) was performed using Student's t test.

Results

The variability of sensor glucose during the entire study period was comparable between the real-time CGM group and controls (21.51 ± 1.10 vs 23.44 ± 1.26 mg/dl; P = 0.243). Although the mean sensor glucose during the first 24 hours in the ICU showed a trend towards lower values in the intervention group (real-time CGM vs control: Glu_MEAN1: 104.40 ± 2.00 vs 109.23 ± 1.96 mg/dl; P = 0.087), variability of sensor glucose during the first 24 hours did not differ significantly (real-time CGM vs control: Glu_SD1: 20.85 vs 18.46 mg/dl; P = 0.201). Glu_SD tended to be lower in ICU survivors (ICU survivors vs nonsurvivors: 21.61 ± 8.60 vs 25.10 ± 10.55 mg/dl; P = 0.071) whereas Glu_SD1 (19.06 ± 9.32 vs 21.41 ± 13.21 mg/dl; P = 0.282) and Glu_MEAN1 (106.39 ± 14.83 vs 108.00 ± 18.44 mg/dl; P = 0.627) were not different between ICU survivors and nonsurvivors.

Conclusions

IIT guided by real-time CGM did not result in reduced glucose variability compared with IIT according to an algorithm with selective arterial blood glucose measurements. Although mean sensor glucose during the first 24 hours tended to be lower in the real-time CGM group, glucose variability was not different between the two groups. Glu_SD tended to be lower in ICU survivors irrespective of glucose monitoring.