1. Daniel A. Baur, Fernanda S. Vargas, Jordan A. Garvey, Christopher W. Bach, and Michael J Ormsbee, FACSM Institute of Sports Sciences and Medicine, Department of Nutrition, Food and Exercise Sciences, Florida State University ABSTRACT INTRODUCTION In a cycling race simulation with ingestion of carbohydrate supplements, CGM and YSI blood glucose measurements correlated. However, CGM glucose measurements substantially differed from YSI, and differences were augmented with large glucose excursions. Interestingly, correlation strength between YSI and CGM increased with large glucose excursions. Due to these factors, it can be concluded that CGM may be of use by athletes to monitor general glucose trends. However, CGM currently lacks the precision necessary for laboratory assessments. METHODS RESULTS CONCLUSION Continuous glucose monitoring (CGM) is commonly utilized by diabetic individuals. Most devices provide near constant (every 5 min) blood glucose feedback over the course of a given day. While obviously of value to this population, the wealth of data provided by CGM is also potentially of great value to sport nutrition scientists for the evaluation of metabolism. Specifically, CGM may provide a relatively non-invasive means of measuring the blood glucose responses to various nutritional strategies during sporting events. However, to confirm the utility of CGM for use in laboratory and field-based studies, further study is required to determine the accuracy of CGM relative to more traditional blood glucose analyzers. Moreover, accuracy must be assessed in the face of various stimuli common to studies evaluating the influence of nutrition on human performance. The aim was the present study was to evaluate the accuracy of a popular CGM device (Dexcom® G4 Platinum) relative to a YSI 2300 blood glucose analyzer during exercise of varying intensities and following meal challenges differing in glycemic indices. Mean absolute relative difference between CGM and YSI was 16.7% ± 15.2 (mean difference = -0.4 ± 20.9) and 15.0% ± 9.6 (-0.2 ± 15.1) for HGI and LGI, respectively. There were positive large (r=0.530, p<0.001) and small correlations (r=0.254, p=0.044) for HGI and LGI, respectively. However, there were wide limits of agreement (95% confidence interval) of ±41.0mg/dL and ±29.6mg/dL for HGI and LGI, respectively (Fig. 2). Limits of agreement were highest during steady state exercise (±46.0 mg/dL) versus rest (31.9 ±mg/dL and late-exercise sprints (±29.2 mg/dL) YSI differed from CGM at 15min pre- exercise (15.9±13.8mg/dL; p=0.023) and at 15min during-exercise (-42.9 ±19.1mg/dL; p=0.001) with HGI, but at no other time points in either condition (Fig. 3). Purpose: The purpose of this study was to examine the accuracy of a CGM (Dexcom® G4 Platinum) for assessing the glycemic effects of two carbohydrate supplements ingested before/during exercise in trained athletes. Methods: Trained male cyclists (n=7; 176.9±5.6cm; 74.4±9.3kg; 60.1±3.3ml/kg/min) ingested a high- (HGI; Gatorade®) and low-glycemic index supplement (LGI; UCAN®) 30min before (60g/600mL water) and every 15min (15g/250mL) during ~3hr of cycling. The CGM was attached 36hr prior to exercise and calibrated every 12hr per manufacturer’s instructions. Accuracy was assessed via comparison with capillary blood glucose measured with a YSI 2300 STAT (YSI, Inc., Yellow Springs, OH). Accuracy between conditions was assessed by Pearson correlation coefficient. Following ANOVA, individual matched pairs were analyzed via paired t-test where appropriate. Bland-Altman analysis determined limits of agreement between conditions. Results: Mean absolute relative difference between CGM and YSI was 16.7% ± 15.2 and 15.0% ± 9.6 for HGI and LGI, respectively. There were positive large (r=0.530, p<0.001) and small correlations (r=0.254, p=0.044) for HGI and LGI, respectively. However, there were wide limits of agreement (95% confidence interval) of ±41.0mg/dL and ±29.6mg/dL for HGI and LGI, respectively. YSI differed from CGM at 15min pre-exercise (15.9±13.8mg/dL; p=0.023) and at 15min during-exercise (-42.9 ±19.1mg/dL; p=0.001) with HGI, but at no other time points in either condition. Conclusions: CGM may be useful for athletes to monitor glycemic trends. However, accuracy is attenuated with large glucose excursions, and correlations may be weaker with narrower glucose ranges. Supported by a product grant from Dexcom, Inc., San Diego, CA CGM accuracy was assessed as part of a study evaluating the efficacy of a novel low-glycemic index (LGI) starch supplement (UCAN®, The UCAN Co., Woodbridge, CT) relative to a traditional high-glycemic index (HGI) supplement (Gatorade®, PepsiCo, Commerce, NY). In a crossover design, trained cyclists (n=7; 176.9±5.6cm; 74.4±9.3kg; 60.1±3.3ml/kg/min) received a supplement 30min before (60g/600mL water) and every 15min (15g/250mL) during ~3hr of cycling (Fig. 1). CGM monitoring (Dexcom® G4 Platinum) began ~36 hr prior to each experimental trial and continued throughout the trials. Comparisons with YSI 2300 were made pre-exercise were made at rest and every 15 min following feeding. During exercise, comparisons were made every 15 min during the first hour of steady state exercise, and following sprint 5 and 10 of the performance test. ACCURACY OF A CONTINUOUS GLUCOSE MONITORING DEVICE FOR THE ASSESSMENT OF CARBOHYDRATE GLYCEMIC EFFECTS DURING EXERCISE RESULTS Fig 2. Limits of agreement between Dexcom and YSI. A, HGI. B, LGI. Fig 3. Glucose responses to HGI and LGI measured via Dexcom and YSI. A, responses to HGI. B, responses to LGI. * denotes p ≤ 0.05 A B AB Fig 1. Exercise and feeding protocol.