Impact of Continuous Glucose Monitoring Sourcing on Real-World Adherence and Health Care Costs: A Comparative Analysis by Insurance Type

Abstract

This real-world analysis compared adherence rates, total health care costs, and health care resource utilization of individuals receiving continuous glucose monitoring (CGM) supplies through durable medical equipment (DME) providers versus through pharmacies, according to insurance type. Patients in the DME cohort had higher adherence rates than those in the pharmacy cohort. Total costs were lower for Medicare/Medicare Advantage patients getting supplies from a DME than for patients getting supplies from a pharmacy. Among those with commercial insurance, health care utilization was lower in the DME cohort than in the pharmacy cohort. Sourcing CGM supplies from DME providers versus pharmacies may yield benefits on adherence, health care costs, and resource utilization.

Numerous randomized trials and real-world retrospective studies have demonstrated significant clinical benefits of continuous glucose monitoring (CGM) in individuals with diabetes who are treated with insulin and noninsulin therapies (1–3). Recent studies have also reported strong associations among the use of CGM, lower rates of diabetes-related health care resource utilization (4–6), and reductions in health care costs (7–9).

Unlike traditional blood glucose monitoring, which provides a snapshot of an individual's current glucose level, CGM sensors transmit a continuous stream of glucose data to the user's smartphone or handheld reader, which displays a comprehensive profile of each user's glycemic status, including the current glucose value, recent trends, and, importantly, trend arrows that indicate the direction and velocity of changing glucose levels. Current CGM devices also provide programmable alarms alerting users about current and impending adverse glycemic events.

Users can transfer CGM data via cloud-based transmission to their health care provider (HCP) for retrospective analysis using the ambulatory glucose profile (AGP) download software. The AGP report aggregates a large stream of glucose data into actionable information that permits patients and HCPs to quickly identify glucose patterns and make appropriate adjustments. Retrospective analysis creates opportunities for more collaborative patient-clinician interactions, which can be done in person or remotely via telehealth visits. Studies have demonstrated that the use of telemedicine and remote monitoring technologies can significantly improve glycemic management in patients (10–15). In essence, CGM has transformed diabetes management from a reactive to a proactive approach to care.

However, optimizing the benefits of this technology is dependent on patients’ adherence to their prescribed treatment regimens and persistent CGM use (9,16–22). Although adoption of CGM continues to increase (23,24), recent data show that only 50.5% of individuals with diabetes are achieving the American Diabetes Association (ADA)-recommended A1C goal of <7% (25,26).

Although utilizing an interprofessional diabetes care team can play a key role in supporting adherence and improving outcomes (27–29), many physicians’ offices are unable to provide team-based care because of time constraints, inadequate staffing, and lack of experience using CGM and other diabetes technologies, and this inability often leads to therapeutic inertia and poor outcomes (30). One option for improving patient adherence in primary care practices is to use durable medical equipment (DME) providers, which offer comprehensive diabetes management and monitoring services, for the acquisition of CGM supplies.

As recently reported by Allaire et al. (31), CGM adherence was significantly higher, with lower health care costs, for patients who obtained their CGM devices and supplies through a DME supplier than for those who received them through a pharmacy. In this matched-cohort analysis of 2,356 individuals with type 1 or type 2 diabetes in a large, commercial claims database, investigators showed that the total medical costs were significantly lower in the DME cohort compared with the pharmacy cohort ($6,967 vs. $10,635, P <0.001). The study also found that patients who had discontinued CGM were more likely to resume CGM if they received their devices and supplies through a DME provider.

However, because the researchers used direct matching based on insurance type, sex, age, and Charlson Comorbidity Index (CCI) score to avoid bias, the patient population leaned heavily toward commercial insurance. Therefore, differences between DME and pharmacy patients within specific insurance groups, such as Medicare/Medicare Advantage, were not examined.

We hypothesized that adherence, utilization of emergency care, inpatient admissions, and total health care costs between DME and pharmacy patients may differ depending on insurance coverage because of the inherent differences in the covered populations and the benefit structures of each insurance type. The current analysis examined these outcomes within different types of insurance coverage.

Research Design and Methods

Study Design

This retrospective, matched-cohort analysis utilized administrative claims data to investigate the potential differences in adherence rates, health care utilization, and total health care costs among individuals with diabetes receiving CGM supplies through DME providers compared with those obtaining them through a pharmacy by different insurance types, including commercial insurance and Medicare/Medicare Advantage.

Data Source

Administrative claims data from 1 January 2010 through 31 October 2024 were obtained from the Mariner Commercial Claims Database, which represents 75.7 billion claims across all payer types and includes data from 161 million unique patients across the United States. Data are de-identified and comply with the patient confidentiality requirements of the Health Insurance Portability and Accountability Act. Institutional review board approval was not required because individual patient data were not identifiable.

Study Population

Eligible patients were ≥18 years of age and had an initial CGM claim in the first quarter of 2021, the exact date of which served as the index date. Patients with diagnosis codes for renal failure or cancer were excluded. Patients were required to have continuous enrollment for 6 months prior to and 12 months after their index date without evidence of CGM claims before the index date. Patients with a diagnosis of diabetes were identified using International Classification of Diseases, 9th Revision, codes (249.00–250.99, 790.2, 790.21, 790.22, 790.29, 791.5, and 791.6) and International Classification of Diseases, 10th Revision, codes (E08.0–E13.9).

Two subgroups of patients were identified based on their insurance type: commercial insurance or Medicare/Medicare Advantage insurance. Within each insurance subgroup, two patient cohorts were identified by direct matching on the following matching variables: CCI (calculated using all existing claims for each patient over a 2-year period from the index date), age, sex, and diabetes type.

The first cohort, referred to as the pharmacy cohort, consisted of patients who received their CGM device and subsequent supplies over the following 12 months through their insurance pharmacy benefit. These patients were identified using billing codes for the CGM devices and supplies. The second cohort, referred to as the DME cohort, consisted of patients who received their CGM device and supplies from a DME provider over the same 12-month period. Patients in both cohorts were identified using the prespecified CGM and supply codes (Supplementary Table S1). Patients from the DME and pharmacy cohorts were matched directly on CCI score, age range, sex, and diabetes type.

Outcomes

The three outcome measures were adherence, medical costs, and health care resource utilization. Adherence was assessed at months 3, 6, 9, and 12 after each patient’s index date, coinciding with the prescribed 3-month ordering interval for CGM supplies. Patients were deemed adherent if they made ≥90% of all scheduled reorders, which served as a proxy for adherence. Patients without evidence of a reorder during the study period were classified as nonadherent. Total medical costs, comparing DME and pharmacy costs in Medicare/Medicare Advantage patients, were assessed throughout the 12-month follow-up period. Costs included any medical or pharmacy claims reimbursed during this period after each patient’s index date. Health care utilization was assessed by a composite representing the number of emergency room (ER) visits and the number of days spent as an inpatient. Each ER visit was considered to be 1 day for this analysis.

Statistical Analysis

Adherence algorithm

The adherence algorithm used the medication possession ratio (MPR) model, commonly defining adherence as >80%. For this study, the threshold was raised to ≥90%, requiring at least two fills for inclusion in the MPR calculation during the 3-month costing period. Adherence was calculated as the total units supplied (numerator) divided by the number of calendar days between the first and last claim within the coded time frame (denominator).

To ensure comparability, adherence criteria were applied consistently across both the pharmacy and DME cohorts, with supplies in the DME cohort analyzed similarly to prescriptions. Patients in both cohorts were required to have at least two relevant claims for adherence calculations. Differences in adherence and reinitiation rates between the cohorts were examined using z tests, with statistical significance set at P <0.05. This methodology provided a rigorous and standardized approach for evaluating adherence patterns and identifying statistically significant variations.

Cost analysis

Total medical costs included all allowable costs across pharmacy and medical benefits for patients with at least one medical claim over the 12-month period from the index date. Pharmacy costs included all allowable costs for patients with at least one pharmacy claim over the 12-month period from the index date. Outliers were removed at the 98th percentile or greater for commercial and Medicare insurance and at the 97th percentile or greater for Medicare/Medicare Advantage insurance. Differences in mean costs between the DME and pharmacy cohorts were examined by t tests, with statistical significance set at P <0.05.

Results

Study Cohorts

Records for 165,758,790 individuals in the Mariner Database were screened. Of these, 1,379,844 patients had diabetes and used CGM. After applying inclusion/exclusion criteria, 12,138 patients (≥18 years of age) with diabetes, an index CGM claim, no CGM claims in the 3 months prior to their index claim, and continuous enrollment for 12 months after the index date were identified as individuals new to CGM use during the index period. Patents were assigned to one of two insurance subgroups: commercial or Medicare/Medicare Advantage. Within each insurance subgroup, direct matching generated two patient cohorts: 11,284 patients with commercial insurance (pharmacy cohort n = 5,642, DME cohort n = 5,642) and 854 patients with Medicare/Medicare Advantage (pharmacy cohort n = 427, DME cohort n = 427). The mean age of both cohorts was 53 ± 16.3 years. Patients’ baseline characteristics are presented in Table 1.

Table 1.

Patient Characteristics

	Commercial (n = 11,284)	Medicare/Medicare Advantage (n = 854)
Age, years	54.74 ± 15.71	68.65 ± 11.37
Sex  Male  Female	5,244 (46.5) 6,040 (53.5)	346 (40.5) 508 (59.5)
Diabetes type  Type 1 diabetes only  Type 2 diabetes only  Type 1 or type 2 diabetes*  Other/unspecified†	604 (5.4) 5,590 (49.5) 5,056 (44.8) 34 (0.3)	<11‡ 528 (61.8) 316 (37.0) <11‡
CCI score	1.57 ± 1.41	1.73 ± 1.42

Data are mean ± SD or n (%).

*Ambiguity due to inconsistencies in coding.

†Other or unspecified may include diabetes of indeterminant etiology or rarer conditions, such as gestational diabetes, monogenic diabetes, or secondary diabetes.

‡The summary patient counts with results <11 are masked with a “−1” to prevent identification.

Adherence

The percentages of patients who were adherent within each quarter of the 12-month follow-up period by insurance type are presented in Figure 1. At each time period, adherence rates in the DME commercial insurance subgroup were significantly greater than the rates in the pharmacy commercial subgroup (range 59–65% vs. 48–57%). Similarly, adherence rates in the DME Medicare/Medicare Advantage subgroup were significantly greater than in the pharmacy Medicare/Medicare Advantage subgroup (range 78–83% vs. 64–68%). Adherence rates within the DME commercial insurance subgroup were also significantly higher than in the pharmacy commercial insurance subgroup at 6, 9, and 12 months, but with no statistically significant difference at 3 months.

Figure 1.

Adherence rate by diabetes cohort by insurance type.

Health Care Costs

Total health care costs for adherent patients are provided in Table 2. For most comparisons, the mean total costs across time periods within the Medicare/Medicare Advantage group were significantly lower for DME patients relative to pharmacy patients. Total cost-of-care differences were not analyzed for the commercial population because of the inherent variability and unpredictability of plan designs, which hinder accurate cost assessments (32,33). In contrast, Medicare and Medicare Advantage rates are more standardized compared with commercial insurance plans, allowing for more accurate analysis (32,33).

Table 2.

Differences in Health Care Costs for Medicare/Medicare Advantage Patients

Time Point, months	DME Costs, $	Pharmacy Costs, $	t test, P
3	2,608.02	3,333.75	−3.01 (600.84), 0.002
6	5,151.00	6,955.00	−2.89 (585.49), 0.004
9	7,720.63	10,466.41	−3.99 (551.87), <0.001
12	11,154.48	15,029.00	−4.16 (559.73), <0.001

Health Care Utilization

As shown in Figure 2, the mean number of ER visits (analyzed as 1 day each) and inpatient days for patients with commercial insurance was significantly lower across all time periods for DME patients compared with pharmacy patients. For the Medicare/Medicare Advantage patients, the same pattern emerged, with the DME cohort having lower mean ER/inpatient visits, but the differences between the cohorts were not statistically significant (Figure 3).

Figure 2.

Differences in health care utilization ER visits and inpatients days among commercially insured patients.

Figure 3.

Differences in health care utilization ER visits and inpatients days among Medicare/Medicare Advantage patients.

Discussion

Findings from this national claims database analysis demonstrated that patients who obtained their CGM supplies through DME providers had significantly better adherence rates than those who received their CGM supplies from a pharmacy across all of the quarterly time periods within the Medicare/Medicare Advantage subgroup. Within the commercial insurance subgroup, adherence rates were initially the same between the DME and pharmacy cohorts at 3 months but became significantly higher in the DME cohort at the subsequent assessments at 6, 9, and 12 months, with adherence in the DME cohort ranging between 45 and 70% higher than in the pharmacy cohort.

The higher adherence rates observed in the DME cohort were associated with significantly lower total health care costs compared with the pharmacy cohort within the Medicare/Medicare Advantage subgroup across all quarters. This reduction in health care costs is likely attributable to improved clinical outcomes driven by greater adherence to CGM. Additionally, health care utilization, including ER visits and inpatient days, was significantly lower for commercially insured patients in the DME cohort across all time periods compared with the pharmacy cohort.

Medicare/Medicare Advantage patients demonstrated consistently higher adherence compared with commercially insured patients in both the DME and pharmacy cohorts, with the DME Medicare/Medicare Advantage subgroup achieving adherence rates near 80% across all quarterly time periods. This finding may be attributed to Medicare coverage requirements for CGM devices, which mandate adherence to diabetes treatment regimens, consistent device use, and routine in-person or telehealth visits with an HCP (34,35). Another explanation is that, by virtue of their business model, DME suppliers establish direct and trusted relationships with patients, enabling better support for proper device use and adherence to prescribed therapies. In contrast, the pharmacy model tends to prioritize medication dispensing, which may limit engagement and continuity of care and reduce its effectiveness in achieving comparable adherence rates.

Although the DME suppliers’ business model emphasizes ongoing engagement and support, some payer groups, including some Medicare Advantage plans, commercial insurers, and state Medicaid programs, have begun shifting CGM coverage to the pharmacy benefit (either as an expansion of access or a substitution for the DME supplier). Howe and Chavis (36) state that the primary objectives of utilizing the pharmacy channel are to streamline processes, alleviate administrative burdens on clinicians associated with prior authorizations (PAs) required under the DME medical policy, and reduce delays patients may encounter in accessing prescribed CGM devices. Additionally, unlike in the DME channel, CGM devices distributed through pharmacies can generate rebates for payers, creating an additional revenue stream from CGM manufacturers. However, this benefit may not necessarily translate to reduced cost-sharing for consumers. Moreover, some payer groups may restrict CGM formulary options to optimize rebate opportunities.

Under the traditional DME distribution model, patient access to CGM requires treating HCPs to obtain PAs for coverage from DME suppliers. PA requests involve creating and submitting detailed documentation that describes the medical necessity for CGM and supporting evidence, including blood glucose records, patient history of acute complications (e.g., frequent and/or severe hypoglycemia events), and other clinical information. In a 2024 survey of 1,004 practicing physicians, 89% of respondents described these burdens as high or extremely high, and 93% of physicians reported that PAs can delay access to necessary care (37). According to the report, the lack of consistency and transparency in PA requirements is also problematic, leading to erroneous denials for coverage.

With the transition to pharmacy coverage, physicians complete a simplified, one-page form documenting the medical need for CGM, often allowing patients to receive their supplies quickly. Although some large pharmacy chains now offer chronic care management services to their diabetes customers, we could find no outcomes data from these programs in the literature. As recently reported by the American Association for Homecare (38), DME suppliers have a CGM patient retention rate of ∼85%, which is 20–30% higher than under the pharmacy benefit. Moreover, patients who switched from the pharmacy channel to the DME channel expressed greater satisfaction with the service under the DME channel.

The report cited several reasons for the high retention rate. First, DME suppliers are more specialized than pharmacies in counseling and support, providing guidance, training, and product support for the CGM systems they provide. Second, many DME suppliers offer educational resources, onboarding services, and follow-up for their CGM patients utilizing virtual visits and other telehealth technologies to monitor glycemic control (via downloaded CGM data) and overall health status. One large supplier (CCS Medical, Dallas, TX) has developed a predictive model with digital interventions that utilize machine learning to identify patients at risk for nonadherence with CGM use and then address the specific clinical, psychological, administrative, and socioeconomic variables associated with that increased risk (39). Importantly, DME suppliers are proactive in ensuring that the provision of CGM supplies is constant, which supports the ADA recommendation that people with diabetes should have uninterrupted access to their supplies to minimize gaps in CGM use (40).

As discussed above, a common complaint of many HCPs is the time and cost of onerous documentation requirements set by payers for DME coverage. Software programs exist that can reduce the referral burden; however, not all HCPs are aware of or have access to these platforms. Given the clinical and health economic benefits of CGM adherence through the DME channel, payers should consider simplifying their documentation requirements to mirror those used in the pharmacy channel.

The older Medicare population can uniquely benefit from CGM use and the support services provided by DME suppliers. These patients often have multiple comorbidities and cognitive dysfunction, which can affect memory, ability to learn, attention span, and executive function (41). They are also at higher risk of impaired hypoglycemia unawareness and more frequent and/or severe hypoglycemia (42–44). With comprehensive early training and ongoing CGM support provided by DME suppliers, episodes of hypoglycemia can be reduced, which could explain the finding of significant differences in early health care utilization favoring the DME cohort over the pharmacy cohort (45).

The results from this analysis align with a previous study demonstrating significantly higher CGM adherence and substantially lower total health care costs among patients who obtained their CGM devices and supplies through a DME supplier compared with those who received them through a pharmacy (31). In that study, nonadherent patients were more likely to resume CGM if they received their devices and supplies through a DME provider. Additionally, the total cost of care (TCOC) was 53% higher among the pharmacy cohort relative to the DME cohort when strict matching methodologies were applied. Although we did not assess TCOC among patients covered by commercial insurers, the lower health care utilization rates observed in the commercial population provide additional evidence to support the expectation of reduced TCOC within this segment.

This is the first study to assess whether there are differences in CGM adherence, total health care costs, and health care utilization between different distribution channels (DME vs. pharmacy) within various insurance types (commercial and Medicare/Medicare Advantage). By including these multiple outcomes, the study provides a better understanding of how CGM sourcing and insurance type influence patient outcomes. This assessment is particularly important because CGM coverage and benefit structures differ significantly between commercial and Medicare/Medicare Advantage plans, which could affect these outcomes.

Strengths and Limitations

A major strength of the analysis was the use of direct matching between cohorts to demonstrate differences in adherence, medical costs, and health care utilization between the DME and pharmacy cohorts. Another strength was the use of the Mariner Commercial Claims Database, which allowed us to track changes in these outcomes in each cohort over time, using standardized records to detect differences and compare changes. Additionally, the use of the MPR, a validated measure of adherence (46), avoided the inaccuracies that are inherent in relying on patient-reported data.

Although the results of this study are compelling, they should be considered alongside a few caveats. While well-suited for evaluating health care resource utilization and costs, retrospective administrative claims data lack clinical detail, such as reasons for selecting a therapy, the brand or type of device and sensors, and the specific clinical response. In our first study (31), we included Medicaid patients, but given the volume limitations and the increasing complexity of coverage dynamics in this population, we found that the variability introduced too much noise to draw clear, reliable conclusions. To ensure the integrity of our findings, we made the decision to exclude Medicaid from this analysis. Additionally, we did not include insulin pump use as a variable or screening criterion, as our focus was not on differences in CGM uptake based on pump use. Although the study leveraged widely recognized definitions to ensure consistency in assessing adherence, it is important to note that adherence was measured using prescription refill data and the MPR model. However, this method may not accurately reflect actual patient behavior. A patient might refill a prescription without consistently using the CGM device, leading to potential discrepancies between adherence rates and actual usage. There may also be unmeasured confounding variables that could influence CGM adherence and costs despite controlling for factors such as age, sex, and comorbidities. Factors such as socioeconomic status, patient preferences, and provider-level influences, which have been shown to affect CGM utilization and adherence (47,48), were not explored in this analysis. Because the same Current Procedural Terminology codes are used for the leading CGM companies, we were unable to differentiate between the various CGM systems and make comparisons relative to outcomes. One exception is the Senseonics Eversense system, which differs from standard CGM billing in that the HCP bills for both the sensor and the procedure at once, unlike most traditional CGM systems for which sensors and related services are billed separately. However, because of the low number of Eversense users in the United States, we do not believe this influenced our findings.

Conclusion

This study demonstrates the impact and benefits of sourcing CGM supplies from DME providers versus pharmacies relative to patient adherence, health care costs, and utilization. The results show that patients who obtain CGM supplies through DME providers had higher adherence rates and lower overall health care costs compared with those using pharmacy benefits within commercial and Medicare/Medicare Advantage insurance plans. These findings highlight the importance of distribution channels in accessing CGM supplies. Future research is needed to explore the long-term effects of CGM sourcing and additional factors influencing CGM adherence and health care costs to further validate these findings.

This article contains supplementary material online at https://doi.org/10.2337/figshare.28477451.

Funding Statement

CCS Medical provided funding for this study.