641 - Accuracy of Continuous Glucose Monitoring in Very Preterm Infants

Publication Number: 641.3609

Tabitha M. Tate, University of Minnesota Masonic Children's Hospital, Saint Paul, MN, United States; Scott Lunos, University of Minnesota, Minneapolis, MN, United States; Megan Paulsen, Children's Hospitals and Clinics of Minnesota, Eden Prairie, MN, United States; Brandon M. Nathan, University of Minnesota, Minneapolis, MN, United States; Sara E. Ramel, University of Minnesota Masonic Children's Hospital, North Oaks, MN, United States

Background: Preterm infants are at high risk for glucose instability. Although continuous glucose monitoring (CGM) is widely used in older children and adults, it is not approved for use in preterm infants. Data on accuracy and safety of CGM use in this population is limited.

Objective: Correlate CGM glucose readings with serum blood glucose (SBG) levels in very preterm infants during the first two weeks of life and describe glycemic characteristics during this period.

Design/Methods: Prospective observational cohort study of infants with gestational age at birth < 31 weeks, admitted to the University of Minnesota Masonic Children’s Hospital NICU (N=20). Following parental consent, a Dexcom G6 sensor was placed on the infant's thigh between 0-96 hours of life and utilized for 10 days. The device was set to unblinded mode, with a low alarm set at 60 mg/dL. Definitions of glycemia: Hypoglycemia ( < 60 mg/dL), euglycemia (60-179 mg/dL), hyperglycemia (>180 mg/dL). SBG levels were drawn per unit protocol and at the clinical discretion of the medical team. CGM levels were matched with the closest SBG within 1 hour (N=253 pairs). Descriptive statistics were used to summarize patient demographics, CGM and SBG values. Independent t-test and Pearson correlation were used to assess relationships between CGM and SBG values.

Results: Patient demographics are summarized in Table 1. CGM Summary Measures are outlined in Table 2. Hyperglycemia occurred in 23% of SBG and 26% of CGM measurements. There were no adverse events. CGM and SBG values were strongly correlated (r=0.95, p< 0.0001), though CGM values were higher than the SBG values (Figure 1); estimated mean difference 50.5 (95% CI: 42.1-58.8). Hypoglycemic SBG values did not correlate with CGM values. Euglycemic SBG values had the highest correlation with CGM values (r=0.90, p< 0.0001) with modestly weaker correlation for hyperglycemic values (>180 mg/dL, r=0.69, p< 0.0001). Sensitivity of CGM to detect hypoglycemia was 20.0% (95% CI 4.33%-48.09%), euglycemia was 62.57% (95% CI 55.04% to 69.68%), and hyperglycemia was 98.31% (95% CI 90.91-99.96). Specificity of CGM to detect hypoglycemia was 100% (95% CI 98.46%-100.0%), euglycemia was 82.43% (95% CI 71.83% to 90.30%), and hyperglycemia was 64.10% (95% CI 56.94%-70.83%).

Conclusion(s): In premature infants, CGM was safe and showed strong correlation with SBG values. CGM is sensitive in detection of hyperglycemia and specific in detection of hypoglycemia and euglycemia. The high percentage of time spent in a hyperglycemic range merits ongoing study for associations with other potential morbidities and neurodevelopmental outcomes.

Table 1: Patient Demographics


Table 2: CGM Summary Measures


Figure 1: CGM Correlation with Serum Glucose Values


Table 1: Patient Demographics


Table 2: CGM Summary Measures


Figure 1: CGM Correlation with Serum Glucose Values