Response to Yu et al. Article “Healthcare Resource Waste Associated with Patient Nonadherence and Early Discontinuation of Traditional Continuous Glucose Monitoring in Real-World Settings: A Multicountry Analysis”

The report by Yu et al. ¹ describing their analysis of healthcare resource waste associated with nonadherence/early discontinuance of continuous glucose monitoring (CGM) raises issues that warrant discussion. However, aspects of the report are both concerning and puzzling.

All of the studies involve CGM devices that predate current CGM systems. ^{2 –8} New CGM technologies are being introduced ∼18 months, offering greater accuracy, reliability, convenience (e.g., factory calibration and no confirmatory fingerstick testing), and new safety features (e.g., data sharing) compared with earlier generation systems. Persistent use of older CGM systems was severely limited by their poor accuracy and reliability; CGM persistence is impacted by patients' perceptions of the accuracy and reliability of their devices. ^{9 –11} These ongoing improvements will likely improve CGM continuation and adherence. The generalizability of cost–benefit analyses involving rapidly evolving technologies such as CGM limits our ability to draw meaningful conclusions and may significantly overestimate the discontinuation and adherence rates associated with use of current CGM systems.

In addition, the majority of the included studies are limited to Medtronic sensor-augmented insulin pump systems (Medtronic, Inc., Northbridge, CA). ^{2 –4,6}

Differences in accuracy, reliability, and ease of use among the various CGM systems must also be considered. For example, most current CGM systems (real-time CGM [rtCGM]) feature alarms/alerts that automatically warn users of immediate and impending glycemic events. A relatively new intermittently scanned system, FreeStyle Libre (Abbott Diabetes Care, Alameda, CA), does not offer this feature. Other features, such as real-time data sharing of CGM data, can also impact users' perceptions of the value of CGM in their own daily self-management. Required twice-daily calibration and confirmatory fingerstick testing place an added burden on users and may affect persistence. Systems with lesser accuracy can lead to excessive false alarms and subsequent discontinuation due to alarm fatigue. A survey by Chamberlain et al. revealed significant differences in user preferences and frequency of use between two earlier generation CGM systems. ¹²

Adherence and discontinuation rates among sensor-augmented insulin pump users cannot be generalized to the much larger population that use standalone CGM devices. The most recent data from the T1D Exchange registry show that among the 31% of current CGM users, 77% are using a Dexcom CGM device (Dexcom, Inc., San Diego, CA). ¹³ Thus, the preponderance of data from studies using one type of CGM system misrepresents the reported discontinuation and adherence rates of the CGM category as a whole.

We agree that the generalizability of results from randomized controlled trials is somewhat limited; however, they do elucidate important information about modifiable and unmodifiable factors (e.g., training and barriers) that provide valuable guidance to clinicians. Also, they illuminate the important differences between CGM systems.

The authors' rationale for excluding the DIAMOND, ¹⁴ HypoDE, ¹⁵ and GOLD ¹⁶ studies remains somewhat puzzling. According to the authors, these studies “required high levels of adherence for inclusion.” My own review found that none of these studies reported this requirement. It is unfortunate that these trials were excluded. The T1D DIAMOND study showed that 93% of patients in the CGM group were using their device ≥6 days per week at the end of the 6-month study. ¹⁴ CGM persistence increased to 97% at 12 months. ¹⁷ Similar persistence was seen in the T2D DIAMOND study. ¹⁸ These findings clearly support the argument that newer CGM technologies positively impact persistence.

Another puzzling concern is that the authors used Dexcom G5 “wholesale acquisition price” as the basis for their cost calculations. Why was pricing for one current system used to assess costs associated with persistence/discontinuation rates of outdated systems? More importantly, the wholesale acquisition price is not a realistic price reference; significant discounts and similar contractual arrangements often reduce this price by up to 50%. Also, it does not represent prices outside of the United States. A more appropriate approach would have been to examine the actual costs payers incur.

Unfortunately, the authors' report seems to take us in the wrong direction—nowhere in the analysis is the value of CGM use ever mentioned. Recent studies such as the recent DIAMOND trial have shown significant improvements in HbA1c among patients using rtCGM with baseline HbA1c values <8.5% and treated with multiple daily insulin injection therapy, ^14,18 which is much less expensive than insulin pump therapy. ¹⁹ Even greater HbA1c reductions were seen in patients with baseline HbA1c ≥8.5%. ²⁰ The recent HypoDE trial showed that CGM use resulted in significant reductions in hypoglycemic events (including nocturnal events) among T1D patients with problematic hypoglycemia. ¹⁵ One must also consider the indirect costs of hypoglycemia; all levels of hypoglycemia (nonsevere, significant, and severe) confer significant indirect costs on employers as well as individuals with diabetes. ^21,22

Continuing improvements in CGM technology will likely increase persistence moving forward. Improvements in accuracy have enabled some CGM systems to replace fingersticks for diabetes management decisions. For example, the past two generations of Dexcom devices did not require confirmatory fingerstick blood glucose testing, and the most recent sensor (G6) does not require any fingerstick testing. Other systems have sufficient precision and stability to eliminate the need for calibration. Eliminating the burden of daily fingersticks should increase CGM persistence, and it is likely accuracy and reliability will continue to improve over time.

However, other barriers such as cost and access to CGM must be addressed. In five of the seven studies included in the cost analysis report, lack of reimbursement and/or cost of supplies were reported to be strong contributors to CGM discontinuation. ^{2,5 –8} Conversely, a recent study reported a persistence rate of 78% among low-income T1D adolescents who had insurance coverage: two of the eight patients who discontinued use stopped using CGM due to lapses in insurance coverage. ²³

Focusing on resource waste associated with discontinuance/poor adherence using antiquated CGM systems without discussing the clinical benefits and associated cost savings demonstrated with current CGM systems could potentially inhibit future reimbursement policies. Instead, our focus should be placed on understanding why individuals discontinue CGM use, and then address those barriers.