Intersystem Accuracy in Continuous Glucose Monitoring: When Does This Matter? Free

Since its introduction in 1999, continuous glucose monitoring (CGM) has become a cornerstone in diabetes management. CGMs are now regarded as the standard of care for all insulin-requiring patients with diabetes. Patients have a broad range of options for CGM use. However, the choice of device can be influenced by factors such as cost, insurance coverage, duration of use, compatibility with other technology (such as automated insulin delivery [AID] systems, smart insulin pens, or smartwatches), and individual patient preferences. While the accuracy of CGMs has improved over time, challenges remain. CGM accuracy in real-world settings is often lower in comparison with the values reported on the labels for CGMs that have received FDA approval. With multiple companies producing CGMs, variability in accuracy across systems has emerged, potentially impacting glycemic metrics. The question to ask, therefore, is whether these differences in accuracy impact treatment decisions and safety. As reported in this issue of Diabetes Care, Freckmann et al. (1) compared the performance of three widely used CGM systems, the FreeStyle Libre 3 (FL3), Dexcom G7 (DG7), and Medtronic Simplera (MSP), and found differences in key glycemic metrics. While the degree of these differences has been objectively measured, their impact on both treatment decisions and safety may not always be clear.

Intersystem differences have been highlighted in several previous studies involving older CGMs. With regard to currently available devices, Hanson et al. (2) conducted a head-to-head analysis comparing the point accuracy of the FL3 and DG7 systems. In this study, the number and percentage of sensor glucose readings within the intervals ±15, ±20, and ±40 mg/dL of glucose reference for glucose values <70 mg/dL and within ±15%, ±20%, and ±40% of glucose reference for glucose values ≥70 mg/dL were calculated. The authors found a lower mean absolute relative difference for FL3 compared with DG7 (8.9% and 13.6%, respectively, P < 0.0001). These results alone, however, do not address how treatment decisions or safety would be impacted.

More recently, Eichenlaub et al. (3) compared the accuracy of FL3, DG7, and MSP systems. Differences between paired CGM and comparator measurements (venous and capillary) were assessed, as representative of the percentage of CGM measurements within ±20 mg/dL (for comparator values <100 mg/dL) or ±20% (for comparator values ≥100 mg/dL). The results for all CGM systems varied depending on the comparator method. However, higher accuracy was generally seen with FL3 and DG7 than with MSP across different comparators. Notably, lower accuracy was seen for all three CGM systems compared with findings of previous studies. A subsequent study was conducted with examination of how these discrepancies in accuracy translate to effects on glycemic metrics and their potential clinical implications.

In this comparative analysis, Freckmann et al. (1) found significant differences in glycemic metrics of the FL3, DG7, and MSP and found significant differences in glycemic metrics, likely similar to previously observed variations in accuracy. The study included 23 patients with type 1 diabetes (mean HbA_1c 6.6%), most of whom (65%) were using AID systems. The participants wore all three CGMs in parallel on comparable locations in the upper arm for 14 days, with at least 70% wear time. While for the majority of the study patients were observed in real-world settings, the analysis also incorporated three glucose manipulation sessions designed to assess CGM accuracy.

The findings revealed that FL3 and DG7 were more concordant with each other than with MSP; with the latter there was a numerically lower glucose management indicator (GMI) and more time with glucose in target range (TIR) and time below range (TBR) and less time above range (1). Of note were significant discordances in TBR, with the largest observed difference being 12.9% between FL3 and MSP in a single patient. These discrepancies can be important because inaccurate values may lead to overlooked or incorrect therapeutic adjustments, especially for individuals using AID systems that rely on CGM data to automatically adjust insulin doses. Additionally, with MSP earlier detection of true hypoglycemia, but delayed detection of hyperglycemia, was seen. The authors emphasize the need for improvements in CGM accuracy through the standardization of study procedures in CGM development to enhance reliability and intersystem consistency, as they suggest that these differences can directly impact both patient and provider management.

The authors acknowledge several limitations of this study, including the small sample size and short duration (1). The patient population may have limited generalizability, as race and ethnicity data were not provided. The study focused exclusively on individuals with well-managed type 1 diabetes, average age 53 years. As CGMs are increasingly used in patients with type 2 diabetes who do not require insulin, these findings may not be fully reflective of the broader population. Given the increasing understanding of the discordance between HbA_1c and GMI measures derived from CGM data (4), particularly in populations such as individuals with chronic kidney disease or older adults, in future studies investigators should consider including these groups for more comprehensive results. This is important, as there is increasing reliance on GMI in certain groups where HbA_1c may not be reliable, such as in pregnant individuals and those with advanced renal or liver disease (5). Additionally, the HEDIS guidelines now incorporate GMI with HbA_1c for the HbA1c Control for Patients with Diabetes measure (6).

It is worth noting that this study demonstrates that MSP blood glucose values are consistently lower than comparator data, which results in an artificially higher amount of TIR and TBR. This can distort the perceived effectiveness of MSP compared with FL3 and DG7. When one sensor consistently reports lower values, direct comparisons of CGM glycemic indices across different systems can become unreliable. Based on the data from Freckmann et al. (1), we need to be skeptical when comparing population-based data from MSP versus the other two sensors; ideally, no sensor (when used with an AID system) should be compared with another. Differences in glycemic outcomes are more complicated than simply differences in the algorithms. Often in clinical trials (with or without AID systems) sensors are changed for a variety of reasons, but this also can impact the integrity of the data and thus should be minimized or at the very least reported. It is evident that the differences in sensor readings can lead to greater variability in GMI and discordance with HbA_1c values, making it more challenging to assess the true glycemic control of individuals using CGM devices.

While reducing intersystem discrepancies is important, it does not always translate into changes in clinical management, as the treatment approach may remain the same despite differences in blood glucose data between systems. This suggests that achieving consistent accuracy across CGMs may not be as critical for everyday clinical care as one might assume. However, accuracy can become critical in specific situations, such as preconception and during pregnancy, where stricter blood glucose control is essential to prevent pregnancy-related complications and ensure the health of both the mother and baby (7) (Fig. 1). Accuracy is also crucial for patients with hypoglycemia unawareness, as accurate measurements can help prevent dangerous episodes of low blood glucose. In fact, in this report from Freckmann et al. (1), the overall proportion of TBR (both <70 mg/dL and <54 mg/dL) was 2.9%, 4.9%, and 5.8% for FL3, DG7, and MSP, respectively.

With AID systems, different CGMs are now compatible with the same insulin pump, and this trend is growing. The CGM landscape continues to evolve with the availability of over the counter CGMs and the development of noninvasive sensors (8). Therefore, ensuring intersystem validity across different CGM systems remains ideal but may not necessarily impact all clinical decision-making or lead to changes in treatment decisions.