Continuous Glucose Monitoring in the Management of Neonates With Persistent Hypoglycemia and Congenital Hyperinsulinism

Abstract

Background Persistent hypoglycaemia is common in the newborn and is associated with poor neurodevelopmental outcome. Adequate monitoring is critical in prevention, but is dependent on frequent, often hourly blood sampling. Continuous glucose monitoring (CGM) is increasingly being used in children with type 1 diabetes mellitus, but use in neonatology remains limited. We aimed to introduce real-time CGM to provide insights into patterns of dysglycaemia and to support the management of persistent neonatal hypoglycaemia. Methods This is a single centre retrospective study of real-time CGM use over a 4-year period in babies with persistent hypoglycaemia. Results CGMs were inserted in 14 babies: 8 term and 6 preterm infants, 9 with evidence of congenital hyperinsulinism (CHI). A total of 224 days of data were collected demonstrating marked fluctuations in glucose levels in babies with CHI, with a higher sensor glucose SD (1.52±0.79 mmol/l vs 0.77±0.22mmol/l) in infants with CHI compared to preterm infants. A total of 1254 paired glucose values (CGM and blood) were compared and gave a mean absolute relative difference (MARD) of 11%. Conclusion CGM highlighted the challenges of preventing hypoglycaemia in these babies when using intermittent blood glucose levels alone, and the potential application of CGM as an adjunct to clinical care.

The hypoglycaemia error grid: A UK-wide consensus on CGM accuracy assessment in hyperinsulinism

“…-not possible to compute, NA, not applicable. errors (12) by graphical methods that categorised errors into discrete grids (13). However, existing error grids do not accurately represent the risk of CGM inaccuracy for patients with CHI as they are designed to test accuracy in patients with diabetes.…”

“…The first report for the Dexcom system (G5) came in 2019 with 1155 paired values giving a MARD of 17.5%, a mean difference of -1.01 mmol/L (-8.09mg/dL) and no error grid analysis ( 12 ). Two more recent studies from mixed populations including some patients with CHI (60-64%) have shown MARD of 11.0% ( 13 ) and 13.1% ( 14 ) with one study plotting paired values on a CEG. These analyses all include multiple values of glucose >4mmol/L which are of great importance to a patient living with diabetes who may have to adjust insulin doses at higher glucose ranges but of limited practical interest to most patients with CHI who use CGM exclusively for hypoglycaemia detection.…”

“…Hypoglycaemia sensitivity is arguably the most useful measure of a device’s accuracy for patients with CHI as it focuses on the detection of hypoglycaemia ( 15 ). Reported results for hypoglycaemia sensitivity in CHI range from 43-73% ( 12 , 13 ).…”

Continuous Glucose Monitoring in Preterm Infants: The Role of Nutritional Management in Minimizing Glycemic Variability

“…CGM is used in children with type 1 diabetes mellitus, but less so in neonatology. A single-center retrospective study reported the use of real-time CGM over a 4-year period in neonates with hypoglycemia [ 43 ]. Although CGM has been shown to reduce morbidity in type 1 diabetes pregnancy LGA offsprings, a recent large multicenter observational cohort study on CGM with 300 pregnant women with type 1 diabetes found that neonates were more likely to be hypoglycemic [ 44 ].…”

DataSheet_1.pdf

“…There is growing interest in the application of CGM to improve glycaemic control in neonates with CHI (12,13). Win et al reported that CGM showed rapid fluctuations in glucose levels in fourteen neonates with CHI alongside persistent hypoglycaemia, reflecting the high risk of undetected hypoglycaemic episodes when managed on intermittent fingerprick glucose tests (14). Rayannavar et al's observational study demonstrated a high false positive rate for hypoglycaemia readings for children with CHI over a two-week period; the authors determined that CGM should be used as an adjunct to glucose monitoring rather than a sole monitoring device due to its suboptimal accuracy (15).…”