Nadine Taleb1,2, Ali Emami1,3, Corinne Suppere1, Virginie Messier1, Laurent Legault4, Jean-Louis Chiasson5,6, Rémi Rabasa-Lhoret1,6,7, Ahmad Haidar8,9. 1. 1 Institut de Recherches Cliniques de Montréal , Montréal, Canada . 2. 2 Division of Sciences Biomédicales, Faculty of Medicine, Université de Montréal , Montréal, Canada . 3. 3 Division of Experimental Medicine, Department of Medicine, McGill University , Montréal, Canada . 4. 4 Montreal Children's Hospital, McGill University Health Centre , Montréal, Canada . 5. 5 Centre de Recherche du Centre Hospitalier de l'université de Montréal (CRCHUM) , Montréal, Canada . 6. 6 Montreal Diabetes Research Center , Montréal, Canada . 7. 7 Nutrition Department, Faculty of Medicine, Université de Montréal , Montréal, Canada . 8. 8 Department of Biomedical Engineering, Faculty of Medicine, McGill University, Montréal, Canada . 9. 9 Division of Endocrinology, Faculty of Medicine, McGill University, Montréal, Canada .
Abstract
BACKGROUND: Despite technological advances, the accuracy of continuous glucose monitoring (CGM) systems may not always be satisfactory with rapidly changing glucose levels, as is notable during exercise. We compare the performance of two current and widely used CGM systems, Dexcom G4 Platinum (Dexcom) and Medtronic Paradigm Veo Enlite system (Enlite), during both rest and exercise in adults with type 1 diabetes (T1D). RESEARCH DESIGN AND METHODS: Paired sensor and plasma glucose (PG) values (total of 431 data pairs for Dexcom and 425 for Enlite) were collected from 17 adults (37.3 ± 13.6 years) with T1D. To evaluate and compare the accuracy of sensor readings, criteria involving sensor bias (sensor minus PG levels), absolute relative difference (ARD), and percentage of readings meeting International Organization for Standardization (ISO) criteria were considered. RESULTS: Both Dexcom and Enlite performed equally well during the rest period, with respective mean/median biases of -0.12/-0.02 mmol/L versus -0.18/-0.40 (P = 0.78, P = 0.66) mmol/L and ARDs of 13.77/13.34% versus 12.38/11.95% (P = 0.53, P = 0.70). During exercise, sensor bias means/medians were -0.40/-0.21 mmol versus -0.26/-0.24 mmol/L (P = 0.67, P = 0.62) and ARDs were 22.53/15.13% versus 20.44/14.11% (P = 0.58, P = 0.68) for Dexcom and Enlite, respectively. Both sensors demonstrated significantly lower performance during exercise; median ARD comparison at rest versus exercise for both Dexcom and Enlite showed a P = 0.02. More data pairs met the ISO criteria for Dexcom and Enlite at rest, 73.6% and 76.9% compared with exercise 48.2% and 53.9%. CONCLUSION: Dexcom and Enlite demonstrated comparable overall performances during rest and physical activity. However, a lower accuracy was observed during exercise for both sensors, necessitating a fine-tuning of their performance with physical activity.
BACKGROUND: Despite technological advances, the accuracy of continuous glucose monitoring (CGM) systems may not always be satisfactory with rapidly changing glucose levels, as is notable during exercise. We compare the performance of two current and widely used CGM systems, Dexcom G4 Platinum (Dexcom) and Medtronic Paradigm Veo Enlite system (Enlite), during both rest and exercise in adults with type 1 diabetes (T1D). RESEARCH DESIGN AND METHODS: Paired sensor and plasma glucose (PG) values (total of 431 data pairs for Dexcom and 425 for Enlite) were collected from 17 adults (37.3 ± 13.6 years) with T1D. To evaluate and compare the accuracy of sensor readings, criteria involving sensor bias (sensor minus PG levels), absolute relative difference (ARD), and percentage of readings meeting International Organization for Standardization (ISO) criteria were considered. RESULTS: Both Dexcom and Enlite performed equally well during the rest period, with respective mean/median biases of -0.12/-0.02 mmol/L versus -0.18/-0.40 (P = 0.78, P = 0.66) mmol/L and ARDs of 13.77/13.34% versus 12.38/11.95% (P = 0.53, P = 0.70). During exercise, sensor bias means/medians were -0.40/-0.21 mmol versus -0.26/-0.24 mmol/L (P = 0.67, P = 0.62) and ARDs were 22.53/15.13% versus 20.44/14.11% (P = 0.58, P = 0.68) for Dexcom and Enlite, respectively. Both sensors demonstrated significantly lower performance during exercise; median ARD comparison at rest versus exercise for both Dexcom and Enlite showed a P = 0.02. More data pairs met the ISO criteria for Dexcom and Enlite at rest, 73.6% and 76.9% compared with exercise 48.2% and 53.9%. CONCLUSION:Dexcom and Enlite demonstrated comparable overall performances during rest and physical activity. However, a lower accuracy was observed during exercise for both sensors, necessitating a fine-tuning of their performance with physical activity.
Authors: John R Petrie; Anne L Peters; Richard M Bergenstal; Reinhard W Holl; G Alexander Fleming; Lutz Heinemann Journal: Diabetologia Date: 2017-10-25 Impact factor: 10.122