At the individual level, the CGM intervention consists of 3 components. First, following the study’s 3 respective health care provider visits (see Health Care Provider Level section), intervention participants will be invited to participate in a post–provider visit phone call with the FQHC’s bilingual or bicultural Hispanic certified diabetes care and education specialist (CDCES) or registered nurse (Figure 3). During the postvisit phone call, the CDCES will review, explain, and answer questions regarding the provider’s T1D plan of care to ensure participant understanding and thereby promote related adherence.

Second, the CGM intervention includes 4 weekly, in-person, CDCES educational visits delivered by the study’s CDCES (Figure 3). The CDCES received 4 hours of training in preparation for delivering the educational visits. The CDCES visits will be held at the participant’s FQHC site of care. The visits are guided by the IMB skills model, adapted for CGM uptake [64]. The IMB-guided sessions provide culturally sensitive, interactive CGM sessions for each participant and their family member (referred to as a dyad) to promote information, motivation, and behavioral skills acquisition for effective uptake and behavioral maintenance of CGM (Figure 2). At week 2 of the CDCES educational sessions, the intervention participants will begin CGM, specifically the FreeStyle Libre 2 system (Table 2).

Each session will integrate confianza (trust), respeto (respect), motivation (health beliefs, motivating factors, and barriers), and skill development. The 4 weekly sessions further integrate the following components of discussion, “hands-on” learning, motivational interviewing, goal-setting, problem-solving, and bidirectional inquiry with an emphasis on a personalized approach and safety (Table 2). Educational handouts, adapted from the Share Plus Intervention, will be provided to the dyad to reinforce session content [65-68]. The educational materials use pictorials and low-literacy language. All educational materials have been reviewed by members of the target population with related modifications to ensure that the handouts are easily understandable and culturally appropriate. The teach-back method will be used at every session. This method provides a way of checking the understanding by asking participants to state in their own words what they need to know or do know.

Third, the CGM intervention includes a total of 7 virtual support group sessions (Figure 3). The virtual support group sessions will be held by a Hispanic bilingual, bicultural peer—an adult with T1D with >2 years of CGM use. The goal of the support group sessions is to reinforce information and skills gained during the 4-week CGM educational sessions and introduce new content concerning, for example, diet and physical activity (see Family or Social Network Level section). The virtual support group sessions will be held using the HIPAA (Health Insurance Portability and Accountability Act)-compliant Zoom (Zoom Video Communications) meeting platform.

At the family or social network level, the core Hispanic values of familismo and collectivismo will be leveraged to promote CGM uptake. Two components of the individual-level intervention will integrate (1) family members (or friends) and (2) a peer educator. A family member will coattend in-person CGM educational sessions (see Individual Level section) with each participant with T1D to provide support in acquiring CGM knowledge and skills and their application in real-world settings.

Previous studies reveal that education alone in Hispanic adults with diabetes is insufficient, and peer support may be warranted for successful behavior change [24,51,53,69-72]. Led by a Hispanic bilingual, bicultural peer with T1D, the virtual support group sessions will provide unique support [73]. Research suggests that the peer format will foster the Hispanic cultural value of collectivismo to promote participant self-disclosure and valuing of opinions, particularly as they pertain to CGM [58]. Moreover, the common bond created through the shared lived experience of T1D and initiating or using CGM, a bond not frequently shared between patients and their providers, will likewise promote support in CGM adoption. The Hispanic bilingual, bicultural peer received 10 hours of Diabetes Community Care Coordinator training from the Association of Diabetes Care and Education Specialists in preparation for delivering the virtual support group sessions [73].

Intervention providers (doctors of medicine [MDs], doctors of osteopathic medicine [DO], and advanced practice registered nurses [APRNs]) received intensive clinical support for T1D care in preparation for the study. We developed and deployed 2 Project ECHO programs to strengthen intervention provider knowledge and skills with a 4-week program in T1D insulin therapy and an 8-week program of Hispanic culturally sensitive CGM education and care (Table 3). Health care providers will see intervention participants at baseline, 3 months, and 6 months or more frequently as indicated (Figure 3).

Four virtual telehealth visits will be delivered via the HIPAA-compliant Zoom meeting platform by a bilingual (English or Spanish) Hispanic CDCES (Figure 4). Session content focuses on increasing participant knowledge and skills related to SMBG. SMBG topics include T1D and blood glucose levels; blood glucose targets, measurement, and monitoring; and causes and management of hypoglycemia and hyperglycemia. American Diabetes Association and Centers for Disease Control educational materials, available in Spanish or English, will be used to guide each session. In addition, control participants will have 5 phone calls with the Hispanic CDCES (Figure 4). The goal of these phone calls is to reinforce SMBG testing knowledge and skills.

Control providers (MDs, DOs, and APRNs) participated in an in-depth, 4-week ECHO program on T1D insulin therapy (Table 3). Control site providers will manage insulin therapy for participants informed by respective HbA_1c levels and SMBG data. Control site providers will see participants at baseline, 3 months, and 6 months or more frequently if clinically warranted (Figure 4).

Demographics, health-related factors, and SDOH will be measured (Table 4). At baseline, demographic (eg, nativity, race, ethnicity, and language preference) and health-related (ie, duration of T1D and specific diabetes chronic complications) data will be collected. SDOH will be measured at baseline and weekly. At baseline, SDOH will be measured with PRAPARE (Protocol for Responding to & Assessing Patients’ Assets, Risks & Experiences). The nationally standardized PRAPARE, developed by the National Association of Community Health Centers, is routinely used at the study FQHC to support social needs screening in its primary care settings. PRAPARE is a 21-item, unsummed measure available in English and Spanish. It will be administered at baseline with data capture of selected items (items 3-8, 11, 14-16, 19; Textbox 1) [76]. Upon identifying a social risk, the CDCES, as indicated, will refer participants, for example, to 211 Connecticut. Accessible by phone or online, 211 Connecticut provides referrals to a wide range of health and human service resources, including transportation, basic needs, housing, and food, among several others. Additional SDOH data (eg, participant transportation to study visits) will be collected weekly for each of the CDCES educational sessions (Textbox 1).

In this study, feasibility will be assessed in terms of (1) recruitment and enrollment, (2) retention, (3) adherence, (4) study procedures and implementation, and (5) acceptability (Textbox 2).

During screening visits, demographic characteristics will be entered into our REDCap recruitment and enrollment database. The recruitment and enrollment database will track recruitment sources, eligibility screening, and enrollment outcomes. Qualitative data on the rationale for study refusal will also be captured in the database. The database will be used to inform modifications to our recruitment strategies to optimize our yield of enrolled participants. In addition, the recruitment and enrollment database will be linked to a retention database. The retention database will track retention yields, retention by differential rates by participant characteristics, and the proportion of completed data by groups (intervention or control).

The REDCap database will be used to track adherence to the intervention and control schedules (Figures 3 and 4). For intervention participants, data on the number of days the CGM was worn and the percentage of time the CGM was active will be collected from the electronic health record (EHR), specifically an electronic copy of the CGM report after each provider visit at 3 and 6 months.

Using the REDCap database, participant intervention and control doses will also be tracked. The number of provider visits completed (3 planned study visits and additional visits as indicated) will be tracked over the 6-month study period. For all participants, the completed number and duration of CDCES and peer-led intervention or control components will be recorded.

Addressing the fidelity of study procedures, data collectors received 6 hours of training in using REDCap in preparation for the RCT. Following training, direct observation revealed that data collectors had achieved high competency in related skills acquisition. During the RCT, fidelity of data collection will be assessed by self-report. Prior to completing a REDCap module for a participant, the database requires that related data entry be endorsed as verified, completed, or incomplete. In addition, in REDCap, open-ended data will be entered on any problems encountered with data collection at baseline, 3 months, and 6 months for each study participant.

With respect to intervention fidelity, specifically the 4 weekly CGM educational sessions, we will use self-report and in vivo observation assessments. A fidelity checklist was developed and integrated into the FQHC’s EHR for each of the 4 sessions, respectively. Using the EHR, the CDCES will have a dedicated template for each of the 4 sessions, corresponding with the goals and content for each session, as displayed in Table 2. The templates have structured domains for entry of quantitative data (achievement of specific session goals and delivery of planned content or not) and corresponding qualitative data, such as participant motivation and goals for CGM, among others. The CDCES will self-report achievement of specific session goals and delivery of planned content or not by using a parallel checklist embedded in each of the session templates. Further, the CDCES will be randomly observed for adherence and competence in delivering the intervention protocol using a fidelity measure designed for this study.

Fidelity of the virtual support group sessions will be assessed by self-report using a web-based data fidelity system. For each of the 7 support group sessions, respectively, the web-based system will capture quantitative and qualitative data. The peer educator, for each session, will be prompted to endorse yes or no regarding the achievement of session goals and delivery of planned content in addition to entering parallel open-ended data.

Acceptability will be assessed with focus group interviews (FGIs) to be held with intervention participants, providers, and CDCES, respectively. FGI interview guides will pose questions on aspects of study processes or procedures (eg, recruitment, retention, data collection, intervention content, frequency of contact, duration of contact, and interventionist delivery).

Physiological outcomes are HbA_1c, time in range, time below range, time above range, and frequency of hypoglycemic- or hyperglycemic-related ED visits or hospitalizations (Textbox 1). HbA_1c will be measured with the DCA 2000 point-of-care analyzer with a capillary blood sample, the standard measure of HbA_1c determination at the study FQHC. CGM glycemic metrics will be captured from FreeStyle Libre View reports (intervention participants only). The frequency of hypoglycemic- or hyperglycemic-related ED visits or hospitalizations over the past 3 months will be based on participant self-report (Textbox 1).

Psychosocial outcomes are depression, quality of life (QOL), social support, and interpersonal processes of care. All psychosocial measures are available in English and Spanish. Depression will be assessed with the 2-item Patient Health Questionnaire (PHQ) and 9-item PHQ. The PHQ-2 is a screening tool for depression. The PHQ-2, using the first 2 items of the PHQ-9, inquires about depressed mood and anhedonia over the past 2 weeks with item scores ranging from 0=not at all to 3=nearly every day. A total PHQ-2 score of >3 suggests major depressive disorder and warrants administration of the PHQ-9 [77]. PHQ-9 item scores likewise range from 0 to 3 (maximum score of 27). A threshold score of >10 indicates mild major depression, >15 or higher indicates moderate major depression, and >20 indicates severe major depression [78].

QOL will be measured by the Medical Outcomes Survey Short Form-36. This well-validated and reliable measure has 8 sections to capture distinct domains of QOL. Each section yields a scored scale, which is a weighted sum of the items in each section. Pooled scores are calculated to generate a total score, ranging from 0 to 100, with lower and higher scores indicating less or greater health-related QOL, respectively (Textbox 1) [77-79].

Social support will be assessed with the Multidimensional Scale of Perceived Social Support (MSPSS). The 12-item MSPSS assesses the perceived adequacy of support from family, friends, and significant others with corresponding subscales. The MSPSS is presented in a 7-point Likert-type format. Response options range from 1=very strongly disagree to 7=very strongly agree, with higher scores revealing greater levels of perceived social support. MSPSS subscale and total scale scores are computed with a specified formula. The MSPSS has demonstrated validity and reliability in Hispanic populations [80,81].

The Interpersonal Processes of Care Survey Short Form measures doctor communication, decision-making, and interpersonal style over the past 12 months. The 18-item Short Form was modified for this study to assess provider (MD, DO, and APRN) communication over the past 3 months. The Interpersonal Processes of Care Survey Short Form has 7 subscales presented in a 5-point Likert-type format; response choices range from 1=never to 7=always. The 3 communication subscales measure lack of clarity (2 items), elicit concerns and responses (3 items), and explain results (2 items). The decision-making subscale assesses patient-centered decision-making (1 item), while 3 interpersonal style subscales measure emotional support and compassion (3 items), discrimination due to race or ethnicity (2 items), and disrespectful office staff (3 items). For each subscale, item responses are averaged with a range of 1-5, with higher scores indicating greater levels [82].

Three aspects of study feasibility will be evaluated using a combination of univariate and bivariate statistical methods. Recruitment sources along with rates and levels of participant retention will be quantified relative to sites and time. Qualitative data on the rationale for study refusal, with the use of data coding, will be quantified. Participant characteristics (eg, age, gender, country of origin, and time since T1D diagnosis) will be summarized using means and SDs or frequencies and percentages, as appropriate to the level of measurement. Associations between participant characteristics and dropout frequency will be investigated via cross-tabulations. Retention will be assessed by participant adherence to the study schedule (Figures 3 and 4). Rates and patterns of adherence to intervention and control procedures and attendance throughout the study duration will be described. Acceptability of the intervention among FQHC providers, CDCES, and intervention participants will be appraised through content analysis of posttrial FGIs [83], using Graneheim and Lundman [84] qualitative content analysis approach to increase trustworthiness. This coding approach involves reading the FGIs several times and identifying and condensing meaning units that are words, sentences, or paragraphs containing aspects related to each other through their content and context. Next, the condensed text will be abstracted with codes, categories, and themes to best represent the perspectives of FQHC providers, CDCES, and intervention participants, respectively.

Completeness and validity of data collection procedures will be assessed by documenting the frequency of missing data and by checking for inconsistent or out-of-range values for all study variables. In-text data on barriers to data collection will be analyzed qualitatively using content analysis [83]. Successful implementation of study procedures will be evaluated via debriefing of study staff and documentation of any departures from the study protocol.

To quantify intervention signals, we will begin by contrasting HbA_1c levels between the intervention and control participants. To do this, we will apply linear mixed modeling (LMM) using study arm assignment (intervention vs control) and time (3 and 6 months) as “main effects,” the corresponding baseline value of HbA_1c as a covariate, and FQHC site variable as well as the “participant” identifier as random effects. This approach will support statistical testing of intervention effects on the mean value of HbA_1c while controlling for differences in baseline levels of the variable, accounting for clustering effects with study sites, and adjusting for correlations within participants over time. After using this strategy to assess the main effect of the intervention on the mean value of HbA_1c, we will investigate the possibility of “intervention-by-time” and “intervention-by-site” interactions and the possibility that any imbalances in baseline participant characteristics (sociodemographic variables and health history factors) between study groups might confound valid estimation of intervention effects. Because the RCT is primarily concerned with feasibility assessment and will involve small samples with limited statistical power, the focus of these analyses will be “effect estimation” rather than formal statistical testing. In the second stage of analysis, we will apply the same modeling strategy to contrast the other physiological outcomes and the psychosocial measures between the intervention and control participants.

Throughout our analyses using LMM, we will attempt to respond to evidence of nonnormality in any of the outcome measures by applying normalizing transformations. If no transformation achieves approximate normality (as is likely to be the case for the time in range, below range, and above range variables as well as the counts of hypo- and hyperglycemic clinical events), we will, instead, apply a generalized linear mixed model with a link function appropriate to the distribution of the outcome measure. Mixed models of repeated measures are robust to the effects of missing observations that occur “at random.” Depending on the extent of missing data (while acknowledging restrictions imposed by the study’s small sample size), sensitivity analyses based on multiple imputations may be conducted to evaluate the potential impact of nonrandom missing data on estimates of the intervention effects.

For members of the intervention arm of the RCT, participant “dose” of the intervention will be described with a combination of means or SDs and frequencies or percentages. Additional analyses using LMM methods will be conducted to evaluate whether measures of intervention dose are associated with differential mean levels of the physiological and psychosocial outcome variables at the 3- and 6-month time points. Finally, we use results of LMM analyses for the following: (1) to calculate standardized effect sizes (Cohen d and eta-squared) for the intervention versus control condition on each of the physiological and psychosocial outcome measures, (2) to determine variance estimates for these outcome variables, (3) to quantify clustering effects (in the form of intraclass correlation coefficients) introduced by the study sites, and (4) to document temporal correlations of outcome measures within participants across the 3 study time points (baseline, 3 months, and 6 months). These statistics will be helpful, but not fully definitive (due to the small sample size), in guiding design decisions and statistical power analyses in preparation for a future, full-scale RCT of the CGM intervention.