Comparison of an Intermittently Scanned (Flash) Continuous Glucose Monitoring System to Standard Self-Monitoring of Capillary Blood Glucose in Gestational Diabetes Mellitus

Abstract

Objective Gestational diabetes mellitus (GDM) requires close surveillance of blood glucose to prevent perinatal morbidity. Self-monitoring of capillary blood glucose (BGM) comes with considerable psychosocial burden. Intermittently scanned continuous glucose monitor (isCGM) devices are discreet and could considerably impact the lifestyle of the patient. They are designed to replace BGM testing in nonpregnant patients. Data on this technology in pregnancy are scant. The aim of this study was to assess concordance of BGM with isCGM in GDM. Study Design Institutional review board approved prospective single-arm study evaluating agreement of isCGM (Freestyle Libre 14-day system) compared with BGM when determining glucose levels fasting and 2-hour postprandial for 14 days. This was documented as percentage of results within Zone A (clinically accurate measurements with no effect on clinical action) or Zone B (values that deviate from reference by >20% but would lead to benign/no treatment) of the Parkes Error Grid (developed for nonpregnant patients with diabetes). Per International Organization for Standardization criteria, agreement was defined as >95% within Zone A or B. Analytical agreement was evaluated using mean and median absolute relative difference (ARD), mean and median absolute difference (AD). Results There were 1,604 pairs of BGM/isCGM observations for 41 patients. Mean glucose values were 102.0 (standard deviation [SD] = 20.5) and 89.4 (SD = 20.1) mg/dL for BGM and isCGM, respectively. Mean and median AD were 15.9 and 13.0 mg/dL, respectively. Mean and median ARD were 15.9 and 12.5%, respectively. Zones A and B contained 76.9 and 22.9% of values, respectively, in the Parkes Error Grid, for a total of 99.8%. Conclusion BGM and isCGM demonstrate clinical agreement. However, glucose values with isCGM trended lower, with greater mean and median ARD than prior studies. Given the strict glycemic control required during pregnancy, physicians should be aware of these differences and their possible clinical implications. Key Points