Impact of physical exercise on sensor performance of the FreeStyle Libre intermittently viewed continuous glucose monitoring system in people with Type 1 diabetes: a randomized crossover trial.

Abstract

To evaluate the sensor performance of the FreeStyle Libre intermittently viewed continuous glucose monitoring system using reference blood glucose levels during moderate-intensity exercise while on either full or reduced basal insulin dose in people with Type 1 diabetes. Ten participants with Type 1 diabetes [four women, mean ± sd age 31.4 ± 9.0 years, BMI 25.5±3.8 kg/m2 , HbA1c 55±7 mmol/mol (7.2±0.6%)] exercised on a cycle ergometer for 55 min at a moderate intensity for 5 consecutive days at the clinical research facility, while receiving either their usual or a 75% basal insulin dose. After a 4-week washout period, participants performed the second exercise period having switched to the alternative basal insulin dose. During exercise, reference capillary blood glucose values were analysed using the fully enzymatic-amperometric method and compared with the interstitial glucose values obtained. Intermittently viewed continuous glucose monitoring accuracy was analysed according to median (interquartile range) absolute relative difference, and Clarke error grid and Bland-Altman analysis for overall glucose levels during exercise, stratified by glycaemic range and basal insulin dosing scheme (P<0.05). A total of 845 glucose values were available during exercise to evaluate intermittently viewed continuous glucose monitoring sensor performance. The median (interquartile range) absolute relative difference between the reference values and those obtained by the sensor across the glycaemic range overall was 22 (13.9-29.7)%, and was 36.3 (24.2-45.2)% during hypoglycaemia, 22.8 (14.6-30.6)% during euglycaemia and 15.4 (9-21)% during hyperglycaemia. Usual basal insulin dose was associated with a worse sensor performance during exercise compared with the reduced (75%) basal insulin dose [median (interquartile range) absolute relative difference: 23.7 (17.2-30.7)% vs 20.5 (12-28.1)%; P<0.001). The intermittently viewed continuous glucose monitoring sensor showed diminished accuracy during exercise. Absolute glucose readings derived from the sensor should be used cautiously and need confirmation by additional finger-prick blood glucose measurements.