Continuous glucose monitors overestimate glycemia, with the magnitude of bias varying by postprandial test and individual – A randomized crossover trial.
Description
Background: Continuous glucose monitors are increasingly used to characterize postprandial glycemic responses, yet no study as directly tested how different test foods/beverages alter the accuracy of continuous glucose monitors. Objective: To assess glucose responses to different test foods/drinks when using continuous glucose monitoring (CGM) compared to capillary sampling (criterion). Methods: Fifteen healthy women (n=9) and men (n=6) completed 7 laboratory visits in a randomized crossover design with 48h washout between consecutive visits. During each visit, participants consumed an oral carbohydrate challenge comprising either 50 g glucose (CONTROL) or an equivalent carbohydrate in the form of whole fruits (WHOLE), or blended fruits (BLEND), a commercial fruit smoothie (PRODUCT), the commercial smoothie ingested over 30 ± 4 min (SLOW), the commercial smoothie ingested with 5 g inulin (FIBRE), and the commercial product providing 30 g carbohydrate (DOSE). Glucose concentrations were recorded from both CGM and criterion methods every 15 min for 120 min and expressed as the incremental area under the curve (iAUC). The glycemic index (GI) was calculated relative to CONTROL where appropriate. Exploratory analyses were performed to examine inter-individual heterogeneity of the difference between measurement methods. Results: The GI for PRODUCT was higher with CGM (84±54) versus criterion (60±35; P=0.05) and was also higher for FIBRE with CGM (80±41) versus criterion (56±25; P=0.05). In contrast, with either WHOLE or BLEND, no evidence was observed for differences in GI using CGM (62±38 and 65±17, respectively) versus criterion (57±38 and 55±29; P=0.26 and 0.18, respectively). Furthermore, the difference between CGM versus criterion fasting glucose concentrations varied by participant (p=0.001). Conclusions: Continuous glucose monitoring overestimates the glycemic responses to fruit smoothies in numerous contexts, and in some cases mischaracterized the glycemic index compared to criterion measures. In addition, there is inter-individual heterogeneity of the accuracy of continuous glucose monitors to estimate fasting glucose concentrations. Caution should be applied when inferring absolute or relative glycemic responses to foods using continuous glucose monitors, and capillary sampling should be used when accurate quantification of glycemic response is the priority. Clinical Trial Registry: https://clinicaltrials.gov/study/NCT0633318