Continuous glucose monitors (CGM), approved by the US Food and Drug Administration in 1999 [1], use a subcutaneous sensor to measure interstitial fluid glucose, reducing monitoring burden and revealing fluctuations that might otherwise be missed. This technology reduced the need for painful fingerstick testing with glucometers that require several steps and supplies [2]. CGM use improves glycemic control [3-10], reduces hypoglycemia and variability, enhances the quality of life [5], and lowers health care use and costs [6-10]. Since diabetes affects 10.5% of the global adult population [11], remains the seventh leading cause of death in the United States [12], and is associated with substantial disease-related distress [13], harnessing the potential of CGM and other emerging technologies may be key to improving outcomes.

Previously, broad CGM use was limited in patients with type 2 diabetes mellitus (T2DM) by a variety of factors. Access to specialized endocrinology care [14] is limited, and despite increased prescribing [15], primary care staff and providers need additional training to be comfortable with the technology [16]. Additionally, CGM use was previously emphasized for patients with insulin-dependent T2DM, and insurance coverage was limited. Policy changes have expanded CGM coverage and accessibility [17,18], making broader CGM use, especially in primary care, more realistic. Many patients with T2DM and prediabetes are managed in primary care settings, and CGM can serve as a patient education tool [19].

In order to maximize CGM use for T2DM in primary care, more work is needed. To date, little is known about the adult patient experience using CGM as part of primary care T2DM management. Without this knowledge, barriers to optimal use cannot be adequately addressed. Recent qualitative work in the United States on outpatient CGM use focused on youth-onset T2DM [20-22], Hispanic adults [23-25], federally qualified health centers [25], or CGM use in combination with other interventions [26]. More work is needed to better understand the common barriers to CGM use in primary care and the issues that primary care providers and staff must help patients overcome. For instance, CGM cost, wearability, discomfort, or fear related to sensor insertion, lack of support, need for guidance on data interpretation, educational needs, and other physical or social factors may all influence CGM use [27,28].

We sought to identify disease- and technology-specific factors influencing CGM use in primary care. Therefore, to guide this work, we chose the Health Belief Model (HBM) [29,30] and the Technology Acceptance Model (TAM) [31]. HBM explains health behaviors through constructs, such as perceived susceptibility, perceived severity, perceived benefits, perceived barriers, self-efficacy, and cues to action. The TAM complements HBM by addressing technology adoption through external variables, perceived usefulness and ease of use, attitude toward using, behavioral intention to use, and actual system use. Combining TAM and HBM enables a holistic evaluation of factors influencing CGM use.

This study aims to understand the perspective of adults with T2DM using CGM in primary care.

This is a qualitative study using directed content analysis examining the experiences of adults with T2DM using CGM in primary care in the United States, using in-person sessions (interview or focus group) with a survey; diabetes-related health data were extracted from the electronic health record (EHR) for each respective participant. We sought to complete 3 focus groups in addition to individual interviews. Study design was guided by the TAM [31] and HBM [30] with an interpretivism paradigm [32]. The interview guide in Multimedia Appendix 1 was internally developed and included questions adjusted to meet fifth-grade reading level. It was internally reviewed but not pilot tested. The interview guide was submitted and approved by the University of Cincinnati Institutional Review Board.

The COREQ (Consolidated Criteria for Reporting Qualitative Research) checklist [33] is available in Checklist 1.

We recruited participants from 3 urban primary care clinics affiliated with a large academic health care system in Southwest Ohio, United States: (1) an internal medicine resident clinic, approximately 7000 patients, 30% have diabetes (80% have hemoglobin A_1c [HbA_1c] <8%, ~18% use CGM); (2) an internal medicine attending clinic, approximately 3500 patients, 24 % have diabetes (81% have HbA_1c <8%); and (3) a combined attending and resident internal medicine-pediatrics clinic, approximately 6000 patients, 11% have diabetes (84% have HbA_1c<8%). These clinic populations are approximately 70% publicly insured (Medicaid, Managed Medicaid, Medicare, or Managed Medicare) and 25% privately insured. The participating clinics were purposively selected due to high public insurance rates, a known marker in the United States for lower socioeconomic status. All clinics are supported by an in-house pharmacy team and nurse practitioners. Providers in all 3 clinics routinely care for patients with T2DM. All sessions took place in-person in an on-site location used for diabetes education.

Participants were recruited by MIK via convenience sampling from the 3 purposively selected clinics. We recruited patients using flyers posted in clinics, announcements to primary care providers and pharmacists, and identifying eligible participants on clinic schedules. MIK contacted potential participants via patient portal messages and phone calls. Eligibility criteria included a diagnosis of T2DM, CGM prescription, 18 years and older, nonpregnant, and the ability to participate in a 1-hour in-person session in English. Participants were assigned to a session type based on their availability.

The research team included an internal medicine primary care physician with clinical informatics training (MIK), a psychology doctorate student (AMC), an internal medicine resident (AS), an undergraduate student (JBW), a senior endocrinology attending (MF), an endocrinology fellow (AZ), and a senior physician scientist in biomedical informatics (EAM). Some team members have first-hand experience monitoring glucose via CGM and glucometers. This work is informed by the belief that chronic disease leads to a significant burden for patients and that technology can help reduce that burden. However, often technical, societal, structural, and individual factors prevent technology from optimizing disease management. This study was developed from MIK’s experience working with patients in primary care and recognition that, while access to CGM was improving, the clinical impact described in the literature seemed to be lagging. MIK served as a supervising physician in the Internal Medicine clinic, where most patients were recruited but did not provide direct care to the patients recruited for this study. The team was supported by the University of Cincinnati Evaluation Services Center (ESC), which provides knowledge, expertise, and experience to serve as an evaluation and applied research partner. ESC provided interviewer training to MIK and supported the development of the protocol, codebook, and overall evaluation. Initial interviews included virtual support from the ESC to ensure quality and provide training for the primary interviewer (MIK).

We conducted 1-hour in-person semistructured sessions (interviews and a focus group). Interviews were conducted by MIK, while the focus group was facilitated by the ESC (EM and AMC). No repeat interviews were conducted. Sessions were audio-recorded and transcribed via conferencing software (Microsoft Teams or Zoom). Transcripts were manually reviewed and edited by study staff. Identifying information was removed. As most sessions included only 1 study interviewer, field notes were limited. No nonparticipants were present during sessions. Transcripts were not returned to patients. Recruitment was terminated when data saturation [34] was achieved, as determined by no new themes, insights, or perspectives emerging from subsequent interviews (MIK and EAM).

Participants completed a paper survey after the session that included questions around perception of CGM using a Likert scale [22], usage habits, and social determinants of health, which was later transcribed into REDCap (Research Electronic Data Capture; Vanderbilt University) by MIK. EHR data were manually extracted via chart review to a REDCap form by MIK. Fields included demographics (age, ethnicity, race, sex, insurance type, and smoking), medical history (length of diabetes diagnosis, complications, and comorbidities), medications, appointment information (number of primary care physician, emergency department, and endocrinology visits in the last year), laboratory data (estimated glomerular filtration rate, cholesterol, HbA_1c at first CGM prescription, and most recent HbA_1c), and completion status of health care maintenance tasks (eye exam, HbA_1c, urine protein, and foot exam). Study data were managed using REDCap [35,36].

Transcripts were coded using MAXQDA [37] (VERBI Software GmbH) and analyzed using a qualitative content analysis [38,39]. A priori codes were based on TAM and HBM, with inductive codes added during review of the first 3 transcripts (MIK and AMC; Multimedia Appendix 2). A total of 2 team members independently coded each transcript. Coding discrepancies were resolved by discussion and consensus. Theme generation was guided by directed content analysis [40] starting with the a priori codes. Inductive codes were evaluated (JBW) and combined with larger themes as appropriate. Data were examined for patterns within and across codes to identify potential themes and subthemes. Iterations of themes and subthemes were validated by team review (MIK, ACZ, AS, JBW, and EAM). Member checking was not performed. An audit trail documented coding and theme development. To strengthen rigor, we used methodological triangulation by comparing survey data on CGM perceptions with quantitative themes. Survey and EHR data were aggregated by session type. Categorical data are reported as counts, percentages; continuous data as mean, standard deviation, range (minimum - maximum).

This study was approved by the University of Cincinnati Institutional Review Board (study number 2023‐0593). The protocol was deemed to be no greater than minimal risk. Written informed consent was obtained from all participants, including disclosure of the study goals. Participants could opt out at any time. Non-essential identifying information has been removed for publication. Participants were informed that direct quotes from the session may be used. Techniques to reduce power imbalances and promote equitable participation were used [38,41], such as using first names, removing badges, and wearing casual business clothing. Participants were compensated with a US $50 cash card.

A total of 46 candidates initiated or received correspondence from the principal investigator; 16 participants (12 interviews and one 4-person focus group) were enrolled between December 2023 and June 2024 (Table 1), when data saturation was reached. Reasons for nonparticipation included inability to be reached, scheduling conflicts, and missed appointments.

The mean age was 56.9 (SD 10.5; range 40‐75) years, 69% (11/16) identified as Black, 100% (16/16) identified as non-Hispanic, and 69% (11/16) as female. Most (15/16, 94%) use public insurance. The mean diabetes duration was 12.3 (SD 5.4; range 3‐22) years, with initial HbA_1c of 9.2% (SD 3.1; range 5.2‐16.9) and most recent HbA_1c of 7.9% (SD 1.9; range 5.4‐10.4). Only 2 participants had an endocrinology visit in the past year. Ten participants used insulin. Mean CGM use was 1.9 (SD 1.4; range 0.1‐5.2) years. Devices included FreeStyle Libre 14 Day, FreeStyle Libre 2, FreeStyle Libre 3, Dexcom G6, and Dexcom G7. Most (12/16, 75%) used a receiver rather than a phone for glucose viewing.

Survey results showed all participants strongly agreed that CGM was useful and would recommend it to a friend. Additionally, 93% (15/16) strongly agreed that CGM was helpful, and 87% (14/16) reported (agreed or strongly agreed) a positive experience using CGM.

Informed by TAM and HBM, 6 themes emerged (Table 2): disease susceptibility, disease severity, influential drivers, perceived ease of use, perceived usefulness, and attitude toward using CGM. Triangulation with survey data confirmed that despite challenges, participants valued CGM and would recommend it.

Disease susceptibility, a component of the HBM, refers to a person’s perception of the risk of acquiring a disease. Subthemes include context of diagnosis, emotional response, and contribution of lifestyle factors (Table 3).

Participants described varied contexts of diagnosis, including specific events, family history, and progression from prediabetes. Reactions to diagnosis ranged from denial and surprise to inevitability, especially among those with a family history. Additionally, participants recognized the impact of lifestyle factors such as obesity and limited exercise on diabetes development.

Disease severity, a component of HBM, captures a person’s feelings on the seriousness and consequences of diabetes. Subthemes include the unstable disease control, medication burden, and fear of complications (Table 3).

Participants noted the unstable nature of disease control, with difficulties such as “sugar addiction” and the potential for life events to derail progress. Despite the focus not being on medications, participants discussed side effects, drug shortages, dose uncertainty, and distrust leading to self-discontinuation. Fear of complications, such as blindness, neuropathy, and death due to poor diabetes control, was prevalent.

Influential drivers, combining cues to action from HBM and external variables from TAM, look at drivers that improve diabetes management and the acceptance of CGM. Subthemes include insurance coverage, support systems, motivation, information sources, and comorbidities (Table 4).

Participants highlighted the critical role of insurance coverage, noting they couldn’t use CGM without it and that device choice was often dictated by coverage. Support systems, including family, community, and health care team members, were also significant. Motivation for disease control stemmed from the desire to live for loved ones, manage symptoms, and optimize HbA_1c. Information about CGM came from advertisements, health care teams, family and friends, and YouTube (Google; 3/16, 19%). Participants reported comorbidities in mental health, cardiovascular, musculoskeletal, neurological, vision, and obesity, with mental health and vision preventing optimal CGM use.

Perceived ease of use, from the TAM, examines how easy or difficult individuals find CGM to use. Subthemes include model differences, initiation, sensor, receiver, alarms, and skin reactions (Table 5).

Participants who used multiple models (19%, 3/16) reported significant differences and strong preferences. Initiation challenges included logistics, troubleshooting, fear of needles, and phone compatibility issues. Sensor barriers involved a lack of application help, optimal location, and water resistance. Receiver issues included connectivity problems, frequent misplacement, and a preference for receivers over phones. Alarms caused significant annoyance, leading some to discontinue use. Various skin reactions required troubleshooting.

Perceived usefulness, from TAM, describes how individuals believe CGM helps manage diabetes. Subthemes include access to continuous data, trust in the data, and data sharing (Table 6).

Participants valued continuous data for constant feedback, learning, responding to extreme values, optimizing medications, and behavior change. They noted improved glucose values and better understanding of their body’s reactions, which kept them motivated. Trust in CGM data was higher compared to traditional glucometer-based monitoring. While few (13%, 2/16) mentioned CGM-specific metrics, most discussed HbA_1c and current glucose values. Data sharing with family and providers was also highly valued.

Attitude toward using CGM, derived from the TAM, captures an individual’s overall views on CGM for diabetes management. Subthemes include empowering disease management, recommendations to others, improvements over glucometer-based monitoring, and further opportunities (Table 7).

Participants felt empowered by CGM, as it allowed easier glucose tracking and understanding of food impact. They adjusted their diets by reducing sweetened beverages, candy, simple carbs, and certain fruits, while increasing fresh produce, grains, diet soda, water, fiber, and lean proteins. All participants highly recommended CGM to others, noting reduced stress in diabetes management. They preferred CGM over traditional glucometer-based monitoring due to no pain, fewer supplies, and less stress. Participants had limited suggestions for future CGM improvements, but desired longer-lasting sensors and more concrete advice around diet and self-motivation.

In this study, we gained insight into the experience of adults with T2DM using CGM in primary care, guided by the HBM and TAM. Participants found that CGM was helpful, empowered them to manage their disease, was an improvement over glucometer-based monitoring, and they would recommend it. Disease susceptibility and severity appeared to motivate diabetes management and influence attitudes toward monitoring blood glucose. Influential drivers include insurance coverage, support systems, motivation, information sources, and comorbidities. The perceived ease of using the technology is related to CGM model differences, initiation process, sensors, receivers, alarms, and skin reactions. Perceived usefulness centers on continuous data, trust in readings, and data sharing. Attitude toward using CGM was generally positive. Our results confirm barriers noted in other populations and uniquely demonstrate the role comorbidities play in interfering with optimal use, strong trust in CGM data versus glucometer-based monitoring, and heavy reliance on receivers rather than phones among adults with T2DM using CGM in primary care.

This study examines the experiences of adults with T2DM using CGM in primary care, most of whom had public insurance (Medicare or Medicaid). Including participants with varied disease severity, insurance type, medication regimens, and CGM use durations provides insights into factors relevant to the complexity of primary care clinics.

Recruitment primarily occurred through primary care physician outreach, but many patients were unreachable or missed scheduled sessions, reflecting the complexity of their lives in which diabetes management is only 1 concern. Convenience sampling underrepresented patients less engaged in care or those who discontinued CGM. Thus, findings are not generalizable for less engaged patients, warranting further study. Purposive sampling was preferred but infeasible due to inaccurate contact information and resource constraints. While it was our intention to complete 3 focus groups, participants opted for the convenience of scheduling interviews.

Although participants received diabetes care in primary care, some may have initially obtained CGM from endocrinology. However, only 2 had endocrinology visits in the past year, suggesting minimal ongoing influence.

Our findings highlight that while adults with T2DM highly value CGM for diabetes management, opportunities to optimize the experience, support, and use of CGM in primary care are numerous. While affirming barriers well-described in other populations in the literature, this study uniquely demonstrates the impact of comorbidities, the trust in CGM data compared to glucometer monitoring, and the reliance on receivers to use CGM technology in an adult T2DM population in primary care.

Successful use of CGM in primary care will require efforts around (1) assessing a patient’s openness and ability to access CGM, (2) developing initiation processes, (3) supporting ongoing troubleshooting, including alarm optimization, and (4) improving the use of data from CGM. The results of this study provide valuable insight for enhancing CGM use for adults with T2DM in primary care by supporting future intervention trials.