Continuous Glucose Monitoring in Individuals With Type 2 Diabetes: A Quality Improvement Program Free

Kootenai Clinic Family Medicine–Hayden is part of the Kootenai Health system and includes a 330-bed community-owned hospital and a wide range of physician clinics, including a family physician residency program. The Kootenai Health system has six primary care clinics, one residency clinic, 37 providers, six mid-level providers (nurse practitioners [NPs] and physician assistants [PAs]), and support from registered nurses (RNs), licensed practical nurses (LPNs), and medical assistants (MAs). The Kootenai Clinic Family Medicine–Hayden is staffed with two physicians, one RN care manager, one RN supervisor, one LPN, and two MAs. The QI program was a collaboration among the physicians, RN supervisor, and RN care manager.

This QI program used a prospective design to assess change in A1C after 3 months of CGM use in people treated with nonintensive insulin, intensive insulin, and noninsulin therapies. Secondary analyses included established CGM metrics generated by CGM download software (38). Metrics of interest included average glucose, glucose management indicator (GMI; an estimated A1C derived from CGM data), time in range (TIR; 70–180 mg/dL), time below range (TBR; <70 mg/dL and <54 mg/dL), and time above range (TAR; >180 mg/dL and >250 mg/dL). Satisfaction surveys were administered at the 3-month visit. Patients were included in the analysis if they met at least one of the following criteria: recent diagnosis of type 2 diabetes, treatment with injected basal insulin or glucagon-like peptide 1 (GLP-1) receptor agonist, history of frequent or increasing hypoglycemic events, elevated A1C ≥7.5%, or not achieving their glycemic targets.

Two study devices were used in the analysis: the FreeStyle Libre 2 system and the FreeStyle Libre 3 CGM systems (Abbott Diabetes Care, Alameda, CA). These systems use the same algorithm to assess glucose levels every 60 seconds. Both require a 1-hour warmup period, have sensors that last for 14 days, and provide real-time alarms to warn users if glucose levels are dropping below or above glucose thresholds. Both CGM systems also have an urgent low glucose alarm at 55 mg/dL. Both have the ability to share glucose values with up to 20 caregivers through the LibreLinkUp app. However, there are some differences between the two systems (Table 1).

The QI program followed a straightforward, standardized approach to CGM initiation and follow-up (Figure 1). For each participant, the RN applied an initial CGM sensor and paired it with the handheld reader (required for Medicaid and Medicare patients) or a smartphone (Android or iPhone) that was compatible with the FreeStyle Libre 2 or FreeStyle Libre 3 app. The smartphone and reader provided immediate information to patients and could be set up to automatically send glucose data to a clinician portal for retrospective analysis. Patients received instructions for placing and removing sensors, setting up and adjusting alarms, connecting to the clinician’s practice computer, and interpreting glucose data. The RN placed an order for sensors through the website of the durable medical equipment supplier for Medicare and Medicaid patients or through a pharmacy for patients with commercial insurance. Patients who did not have their smartphone with them at the initial visit received additional instructions on installing the FreeStyle Libre 2 or Libre 3 app and connecting it to their clinician's practice computer for monitoring. Patients were scheduled for a 2-week follow-up visit with the provider.

At follow-up visits, CGM data were accessed via the LibreView web-based platform and discussed. Changes in medication and/or lifestyle modifications (i.e., diet and exercise) were made as needed. Patients were then monitored during the next month for review of their glucose data and additional therapy adjustments, if needed. Depending on their A1C and GMI, patients were required to follow up with the clinician in 1, 2, and 3 months. If patients were struggling to achieve the A1C or GMI measurement at the 2-week appointment, they followed up monthly. If patients were on track at the 2-week follow-up appointment, they would follow up at the 3-month appointment. All patients were seen at a 3-month interval. At the 3-month follow-up visit, patient CGM data were downloaded and reviewed, therapy was adjusted as needed, A1C was measured for comparison with the pre-intervention A1C value, and a patient satisfaction questionnaire was completed.

Downloading data from CGM sensors and other diabetes devices facilitates more meaningful clinician-patient collaboration and provides patients with a more comprehensive understanding of their diabetes and the impact of food, exercise, medications, and other factors on their glucose management. The LibreView web-based platform features a series of reports that analyze various aspects of the CGM data. For simplicity, clinical staff used the ambulatory glucose profile (AGP) report, which presents a summary of a patient’s glucose status based on established metrics for clinical interpretation of CGM data (38). The report presents these metrics in a standardized format that includes statistical information (average glucose, GMI, and glycemic variability) (Figure 2A), graphical and statistical information regarding the percentage of time the patient has spent within, below, and above the target range (TIR, TBR, and TAR) (Figure 2B). These data are presented in a color-coded bar graph that patients can easily understand, which helps keep patients engaged in their diabetes self-management and prompts more meaningful discussions. The AGP graph (Figure 2C) combines all of the daily glucose profiles (Figure 2D) into a single 24-hour average profile that allows for easy identification of priority time periods on which to focus. Studies have shown that having CGM active for >70% of the time over a 14-day period correlates strongly with 3 months of mean glucose, time in ranges, and hyperglycemia metrics (39,40).

The primary outcome for this evaluation was change in A1C from baseline to the 3-month follow-up visit. Secondary outcomes were changes in CGM metrics, including TIR, TBR, TAR, average glucose, glycemic variability (defined as percentage coefficient of variation), and GMI. Patient-reported treatment satisfaction was also captured using an investigator-developed questionnaire.

Ninety-eight patients were included in this analysis. Patient demographic characteristics are presented in Table 2. All patients used their sensor 100% of the time during the observation period. The analysis demonstrated an overall mean reduction in A1C levels over a 3-month period from 7.9 ± 1.7 to 6.7 ± 0.9%. Mean glucose decreased from 185 to 150 mg/dL. These reductions are reflected in the mean GMI (6.9 ± 0.7%) observed at 3 months. Increases in A1C levels were observed in eight patients; however, four of these patients had an A1C <7.0% and remained at that level during the 3-month period. Among the 70 patients with A1C levels ≥7.0% at baseline, 37 (53%) lowered their A1C to <7.0%. Importantly, 78 (71.4%) patients achieved the ADA glycemic goal of <7.0% (7), and all patients met the National Committee for Quality Assurance Healthcare Effectiveness Data and Information Set metric for “in control” of A1C and GMI <8.0% (41). The mean TIR (70–180 mg/dL) observed at 3 months was 77%, TBR (<70 mg/dL) was 1.34%, and TAR (>180 mg/dL) was 22.6%. At the 3-month follow-up visit, the average number of hypoglycemic events (glucose <70 mg/dL) per day was notably 0.16. Patients reported a high level of satisfaction with their treatment using CGM (Table 3).

A common barrier to CGM use in primary care is confusion about billing and requirements for reimbursement. The following information is included as a guide for clinicians who are interested in integrating CGM use into their clinics.

There are currently two Current Procedural Terminology (CPT) codes that cover CGM initiation: 95249 and 95250. These codes cover placement, hook-up, calibration, patient training, and printout of recorded data. CPT code 95249 is used when the device is owned by the patient (“personal CGM”), and CPT code 95250 is used when the device is owned by the clinic (“professional CGM”). CPT code 95251 is used and reported to insurers when clinicians perform analysis, interpretation, and reporting on a minimum of 72 hours of CGM data. These activities may be conducted with data from patient- or clinic-owned CGM systems. This service is distinct from evaluation and management services and does not include an assessment of or indicate a plan of care for the patient.

Evaluation/management (E/M) CPT codes (99201–99205, 99211–99215, 99241–99245) are reported with the CGM codes if documentation supports the medical necessity of a significant and separately identifiable E/M service performed on the same date as the CGM services. Clinicians must bill the E/M code with the modifier “–25” and submit the 95251 billing on the same day for the same patient if the E/M was a significant and distinct identifiable service that was “above and beyond” the services associated with CPT 95251. Table 4 presents a summary of the CPT codes and associated requirements (42,43).

There are 89 different fee schedule localities in the country, and payments vary significantly from one to another. The geographically adjusted payment rate for any code paid under the physician fee schedule can be accessed through Medicare’s Physician Fee Schedule Lookup Tool (https://www.cms.gov/medicare/physician-fee-schedule/search/overview).

When billing for a CGM data interpretation visit (CPT code 95251), clinicians use a simple statement to document that at least 72 hours of CGM data were available for interpretation and that the findings and recommendations for therapy adjustments were shared with the patient. Clinicians may also want to include a brief description of the recommendations made to the patient. This description may include a list of the recommended medication therapy changes and lifestyle modifications and a screenshot of the AGP report, which can be entered into the electronic medical record. Clinics are advised to obtain copies of the CGM coverage policies from each of their patients’ current insurers to avoid any issues.

This prospective, QI program evaluation demonstrated that the use of CGM improved glycemic control in our patients on intensive insulin, nonintensive insulin, or noninsulin therapy and enhanced their ability to make real-time decisions to better manage their diabetes and improve their glycemic control over a 3-month period. Glycemic improvements were evidenced by significant decreases in A1C and GMI from baseline and increased TIR, which exceeded the goal of 70%.

The addition of CGM into our standard care for type 2 diabetes management also increased patients’ satisfaction with their treatment plans. This finding is particularly important given the relationship between treatment satisfaction, adherence to therapy, and subsequent outcomes. Studies have shown that patients’ satisfaction with their treatment regimen is a strong predictor of medication adherence and positive clinical outcomes (44–46). We hypothesize that the continued use of a CGM beyond 3 months will likely demonstrate a reduction in comorbidities associated with suboptimal diabetes management and enhance the continuity of care for our patients.

The glycemic improvements observed were mostly the result of having comprehensive, yet easily interpretable, data to guide therapeutic decision-making. Moreover, reviewing the data with patients facilitated productive clinician-patient collaborations and discussions. Through these interactions, patients developed a better understanding of their diabetes and the impact of medications, dietary habits, physical activity, and other behaviors/conditions on their glycemic control.

The success of the QI project was quantified by the use of validated measures for primary clinical outcomes and treatment end points for A1C, average glucose, glycemic variability, and TIR (38). Other strengths identified in the project were the ease of use of the CGM system, patient satisfaction, and real-time awareness of glucose levels and the impacts of diet, medication, and lifestyle.

However, there are notable limitations to our findings. Participants were not randomized, and the causality of outcomes could not be established. Additionally, it was not possible to verify the hypoglycemia data, and this work relied heavily on patient surveys and self-reported accounts of self-management goals, and medication adherence. Patients’ TIR before starting CGM was not available. Blinded CGM data before initiation of the unblinded CGM system would have allowed a more comprehensive assessment of improvement in glycemic management. Nevertheless, significant improvements in A1C and other glycemic metrics were observed despite these limitations.

The use of CGM in the population of people with type 2 diabetes has been shown to improve patients’ ability to use real-time glycemic information to better manage diabetes and glucose levels. This QI program was designed to improve A1C outcomes, increase patient engagement in diabetes management, and standardize care for individuals on insulin or noninsulin therapy in a family practice setting. Moving forward, the QI program is being expanded to the other family practice clinics in the Kootenai Health system, as well as specialty practices such cardiac rehabilitation and the cardiology clinic.

The authors thank Christopher G. Parkin, MS, of CGParkin Communications, Inc., for editorial support.

Funding for editorial assistance was provided by Abbott Diabetes Care. Kootenai Health received no funding for the development or implementation of the QI program.

No potential conflicts of interest relevant to this article were reported.

Both authors designed the project protocol, collected and analyzed the data, wrote and revised the manuscript, and approved the final version for submission. Both authors are the guarantors of this work and, as such, had full access to all the data reported and take responsibility for the integrity of the data and the accuracy of the data analysis.