Defining Glycemic Variability in Very Low-Birthweight Infants: Data from a Continuous Glucose Monitoring System.

Background: Glucose variability (GV) is a matter of interest for researches in recent years. It is connected with oxidative stress, which is crucial in the development of multiple complication of prematurity. However, glycemic variability in preterm infants was poorly investigated. This study aims to investigate glycemic variability obtained from a continuous glucose monitoring (CGM) system in a cohort of very low-birthweight (VLBW) infants.
Methods: A prospective, single-center, open cohort study enrolled 74 VLBW infants with a mean birthweight of 1066 g and median gestational age of 28 weeks. A CGM system (Guardian Real-Time CGM ^® , Medtronic, Northridge, CA) was used to measure interstitial glucose concentration. The glycemic variability was calculated using EasyGV.
Results: Most glycemic variability indices in VLBW infants showed log-normal distribution and for these, geometric mean ÷/ × geometric standard deviation (GSD) was calculated: M-value 2.28 (÷/ × 1.82), mean amplitude of glycemic excursions (MAGE) 1.89 (÷/ × 1.34), average daily risk ratio (ADRR) 2.22 (÷/ × 2.56), lability index 0.46 (÷/ × 1.71), J-index 0.46 (÷/ × 1.71), low blood glucose index 2.05 (÷/ × 1.66), high blood glucose index 1.11 (÷/ × 2.44), continuous overlapping net glycemic action (CONGA) 5.54 (÷/ × 1.16), mean of daily differences (MODD) 1.23 (÷/ × 1.38), and coefficient of variation 1.15 (÷/ × 1.31). Only SD of glucose concentration showed a normal distribution: arithmetic mean 1.24 (+/-0.37). ADRR, J-index, MODD, CONGA, and MAGE are moderately to strongly correlated with SD.
Conclusions: In our cohort of VLBW infants, almost all glycemic variability indices showed skewed positive distribution. The natural central tendency measure for the log-normally distributed data is the geometric mean and for statistical variation is the GSD.