Comparisons between accuracy of CGM systems in a pediatric setting can be outdated before they are published. Comment on Nagl et al

Abstract

To the Editor, A recent publication in Pediatric Diabetes from Nagl et al has reported on the comparative accuracy of three continuous glucose monitoring (CGM) systems in children with type 1 diabetes (T1D) aged 9–14 years, during a week diabetes summer camp in 2019, organized by the Austrian Diabetes Union. Of the three systems compared, the FreeStyle Libre (Abbott, Alameda CA) and the Dexcom G6 (Dexcom, San Diego CA) are factory-calibrated systems, whereas the Medtronic Enlite system (Medtronic, Minneapolis MN) required twice-daily calibration using capillary blood glucose measurements. Accuracy of the separate CGM systems were assessed by comparison with fingerprick capillary glucose readings obtained from children at the camp using their own glucometers. The authors concluded that the overall MARDs of the three CGM systems were 13.3% for the FreeStyle Libre system, 10.3% for the Dexcom G6 system and 8.5% for the Medtronic Enlite system. Parkes Error Grid (PEG) analysis showed comparable results for each system, with 98.9% of values for FreeStyle Libre in the clinically relevant zones A + B, with 98.8% and 98.5% of readings for the Dexcom G6 and Medtronic Enlite systems respectively in PEG zones A + B. Interestingly, MARD for the Freestyle Libre was consistent during the day (13.3%) and night-time (13.5%), whereas both the Dexcom G6 and Medtronic Enlite had decreased MARD performance at night compared to daytime (16.0% vs. 10.1% for Dexcom G6; 12.6% vs. 8.1% for Medtronic Enlite). Both factory-calibrated systems tended to underestimate glucose levels in the summer camp setting, which has the effect of alerting users to low glucose in a timely manner, when actual blood glucose may still be in the low euglycemic range. The performance of CGM systems in children is important, since they have benefits for glycemic control, in pediatric and in adult subjects, as proven in randomized clinical trials and in real-world settings. However, we want to emphasize that comparative assessments of accuracy between different systems may be of limited value, given that the evolution of these technologies can outpace comparative analyses. For example, the paper by Nagl et al. focuses on the performance of the three systems as available up to summer 2019. However, at the point of publication in November 2020, the FreeStyle Libre system had already been updated with a new algorithm that makes the accuracy reported by Nagl et al outdated. At the time of the study, Nagl et al assessed the overall MARD of the FreeStyle Libre sensors in their group of children at 13.3%, which is in line with the published MARD of 13.9% for pediatric subjects at that time, which also used paired fingerstick capillary blood glucose measurements. However, the overall MARD for FreeStyle Libre sensors with the most recent generation algorithm in pediatric subjects aged 6–17 years old, whether accessed via the FreeStyle LibreLink smartphone app or the FreeStyle Libre reader, is 9.7% with 100% of sensor readings in consensus error grid (CEG) zones A + B. Similarly, Nagl et al. report that 56.9% of FreeStyle Libre readings overall were within ±15% (or 15 mg/dl) of capillary blood glucose reference values in their pediatric subjects, whereas the most recent generation algorithm has 84.5% of readings within ±15% in children aged 6–17 years old. Performance in hypoglycemia <70 mg/dl is also available for the FreeStyle Libre system with the most recent algorithm, such that 96.1% of readings are within ±15% (or 15 mg/dl) of reference values <70 mg/dL in pediatric subjects. The same algorithm has shown enhanced accuracy for adult subjects. At the time of writing, these established changes and improvements in accuracy and reliability encompass FreeStyle Libre sensors and readers that are available in Austria and all other European, Middle Eastern, and African countries where FreeStyle Libre is currently available with the exception of Turkey, as well as for users of the FreeStyle LibreLink smartphone app that is accessible for all countries in Europe, the Middle East, and Africa where FreeStyle Libre is currently available with the exception of Russia. CGM technology has developed to the point where two of the devices used in the study by Nagl et al. (FreeStyle Libre and Dexcom G6) are authorized for non-adjunctive use in making insulin dosing decisions, and are part of the class of glucose sensing devices designated by the US FDA as integrated CGM (iCGM), a classification intended to further speed innovation and promote the interoperability of CGM with other devices and software. The accuracy of CGM is certainly important in the daily life of children with diabetes, especially as new devices become available for selection. However, beyond accuracy, there are other comparative aspects of CGM systems that are worthy of discussion in this context. These are centred on their utility and alignment with living with diabetes and providing improvements in glucose control and quality of life for children with diabetes. These aspects should also be taken into account during clinical decision-making involving choice of CGM systems for children. Received: 1 March 2021 Accepted: 5 May 2021