Regarding Singh et al,“Effects,Safety,and Treatment Experience of Advanced Hybrid Closed-Loop Systems in Clinical Practice Among Adults Living With Type 1 Diabetes”

We read with interest a recent article by Singh and colleagues related to advanced hybrid closed loop (AHCL) insulin delivery systems in adults with type 1 diabetes ¹ and welcome additional evidence of their effectiveness in a variety of populations (here, Swedish adults) and study designs.

There appear to be more studies emerging on the class effect of AHCL systems, combining outcomes from various manufacturers. The discussion of class effect is usually based on continuous glucose monitoring (CGM) metrics and HbA1c outcomes. Both of these outcome metrics have benefits and challenges in analysis, ² but ultimately provide similar assessment of glycemic changes.

While the class effect of AHCL is a reasonable outcome when combining results from multiple systems, there are methodological challenges with between-system comparisons. Singh and colleagues correctly evaluated and contextualized the aggregated impact of Control-IQ technology (Tandem Diabetes Care, Inc.) with Dexcom G6 (Dexcom, Inc.) and 780G (Medtronic, Inc.). They also reported that CGM-based outcomes were similar between systems, but noted that people using the Control-IQ system had a larger reduction in HbA1c (−0.3%), even when adjusting for baseline characteristics (P = .008). In addition, patient experience metrics were significantly different between the two systems.

This discrepancy between sensor and HbA1c outcomes has been noted in other studies as well, ³ with seemingly different outcomes between systems as measured by impact on HbA1c versus time-in-range. Why might changes in CGM-based metrics agree or disagree with changes in HbA1c values? Part of the answer is that not all CGM systems are created equal—clinically relevant variations with respect to chemistry and sensor construction exist, and complicate efforts to compare sensor outcomes and establish equivalencies across systems.

The two AHCL systems used in the Singh study rely on different algorithms and, importantly, different sensors (Medtronic Guardian 4 used in Medtronic 780G; Dexcom G6 used in t:slim X2 with Control-IQ technology) with different performance attributes. The Guardian 4 sensor mean absolute relative difference (MARD) is 10.6% and 10.8% when worn on the arm and abdomen, ⁴ whereas the Dexcom G6 MARD is 9.0% when worn on the arm. ⁵ Inherent differences in CGM accuracy and performance even beyond MARD may account for purported differences in outcomes ⁶ that are not supported by secondary HbA1c data. Therefore, CGM-derived metrics should not be compared between CGM systems as the instruments of measure are fundamentally different. This can lead to misleading conclusions on superiority of one system versus another, ⁷ when in fact the comparison is methodologically unsound.

Continuous glucose monitoring–based metrics can be a meaningful outcome measure for clinical trials involving AHCL systems, ² but between-system comparisons in real-world datasets must be viewed with caution. The paper by Singh and colleagues allows us to call attention to the value of system-independent outcomes, such as HbA1c, when comparing the efficacy of different AHCL systems.