SAT-649 Comparison of the Accuracy and Concordance of 3 CGM Devices vs SMBG During Aerobic Exercise

Introduction: Real-time continuous glucose monitoring (rt-CGM) and flash glucose monitoring (FGM) devices have become important tools for managing type 1 diabetes. These devices are approved for management decisions in steady-state conditions, however there is a decline in accuracy during aerobic exercise with respect to MARD and lag time.(1) It is possible that newer technologies may be superior to previous devices. Question: With the newest rtCGM, FGM, and long-term CGM devices, do we continue to see an increase in MARD during continuous aerobic exercise? Is there a difference between glucose readings of the 3 devices when worn simultaneously during exercise? Design: A single subject with T1DM, experienced in glucose management during exercise, wore 3 devices simultaneously - the DEXCOM G6 (San Diego, CA; rt-CGM1, worn on the abdomen), the Eversense (Germantown, DM; long-term CGM or rt-CGM2, implanted in the left arm), and the Abbott Freestyle Libre 14-day (Chicago, IL; FGM, worn on the right arm). The rt-CGM2 was calibrated using a blood glucose meter (Ascensia Contour Next) which was also used for comparator SMBG. Glucose was recorded 10 minutes before and after exercise and every 10 minutes during a 60 minute run at moderate intensity. 6 exercise sessions were averaged for data analysis. Subject wore an insulin pump and reduced the basal rate by 50% 90 minutes prior to exercise and resumed the basal immediately post-exercise. Carbohydrates were not used within 3 hours prior to exercise but could be consumed during exercise if needed to avoid hypoglycemia. Results: Glucose value during 60 minutes of exercise dropped from mean of 167 to 114 mg/dL with SMBG, 174 to 115 mg/dL with rt-CGM, 175 to 115 with rt-CGM2, and 150 to 106 mg/dL with FGM. Average measured glucose was 140.0, 145.8, 145.6, and 129.3 mg/dL for SMBG, rt-CGM1, rt-CGM2, and FGM respectively. P-value <0.05 for FGM. MARD (calculated compared to SMBG) for 10 minutes pre-exercise, during exercise, and post-exercise for rt-CGM1 was 5.1%, 11.7%, and 8.6% respectively. For rt-CGM2 MARD was 7.7%, 11.4%, and 10.0% respectively. For FGM, MARD was 12.7%, 5.3%, and 21.3% respectively. Overall MARD was 9.8% for rt-CGM1, 10% for rt-CGM2, and 8.0% for FGM. Conclusions: Blood glucose values dropped with aerobic exercise with observed lag between CGM and SMBG. Rt-CGM1 and Rt-CGM2 showed increased MARD vs SMBG during exercise. Interestingly, lower MARD was seen for FGM during aerobic exercise likely due to bias towards lower glucose levels at baseline as reported by FGM. There was no significant difference seen during exercise between rt-CGM1 and rt-CGM2 despite the differing location of the sensors (transdermal vs subcutaneous) and method of glucose analysis (glucose oxidase vs fluorescence). References: (1) Zaharieva et al. Diabetes Technol Ther 2019; 21: 313-321.