Engaging People in Medically Underserved Areas in the Community-Based Healthy Eating and Active Living to Reverse Diabetes (HEAL Diabetes) Program

Abstract

Background/Objectives: Recruiting and retaining low-income participants in community-based diabetes interventions remains a persistent challenge, particularly in medically underserved areas. This study describes engagement strategies and lessons learned recruiting for a 12-month pilot of a community-based, medically tailored nutrition program for diabetes remission and weight loss. Methods: A descriptive, exploratory mixed-methods study was performed to assess the effectiveness of recruitment and engagement strategies in the HEAL Diabetes program and identify areas for improvement. Recruitment and enrollment data were tracked utilizing recruitment logs and field notes. Descriptive statistics were used to analyze recruitment activity and retention rates, while qualitative analysis of fieldnotes identified key barriers and facilitators. Results: Among 83 eligible participants, 63 (75.9%) completed the in-person screening and 35 (55.6% enrollment rate) enrolled. Retention was high, with 30 completing the study. Participants were largely African American (97.1%), female (70.6%), average age of 59.8 years, with a household income below USD 49,000 (74.3%). Recruitment cycles achieved 87.5% of the target before budget constraints halted enrollment. Recruitment was hindered by limited clinical integration, social barriers and life demands, while facilitators to recruiting included trust, flexibility, and tangible support for participation. Conclusions: Conventional recruitment methods, including registry-based approaches, were insufficient for engaging underserved populations. Participant-centric strategies, emphasizing trust, practical support, and structural and cultural relevance, can help enhance enrollment and retention. Effective engagement in community-based diabetes interventions requires multifaceted approaches that address clinical, social, and structural barriers to participation.

1. Introduction

Among low-income communities, racial and ethnic minorities, and other systematically disadvantaged groups, structural barriers such as food insecurity, limited healthcare access, and transportation issues exacerbate the burden of type 2 diabetes (T2D) [1,2,3,4]. Behavioral and psychosocial interventions have demonstrated some success in improving health outcomes among individuals with diabetes [5,6,7]. However, engaging and retaining medically, socially, or financially disadvantaged community members in health interventions can be challenging [8,9,10]. In addition, the same structural barriers (e.g., limited access to preventive care counseling and the lack of safe spaces for exercise) make sustaining health behavior changes difficult [11,12,13,14]. Programs aimed at reaching those disproportionately affected by diabetes are often constrained by low recruitment, poor engagement, and high dropout rates; this is especially relevant for pragmatic trials that seek to reflect everyday settings [15,16,17,18]. Further, these issues are not limited to study design but also reflect the real constraints faced by participants, particularly those managing the daily demands of life in under-resourced and underserved settings.

Low-income and medically underserved participants often must navigate competing priorities (e.g., caregiving responsibilities), financial and transportation constraints, and employment demands that restrict their ability to participate in traditional, clinic-based interventions and programs [19,20]. A range of factors, including social determinants of health (SDOH), mistrust in healthcare, and failing to center participants’ lived realities, contribute to poor engagement and retention in research [20]. Issues, like inadequate or unaffordable transportation to intervention sites, have long been identified as significant obstacles to sustained engagement in research studies and structured health programs [21]. Studies on oncology trials have identified unanticipated participant out-of-pocket costs for clinic visits, additional tests, frequent travel, and loss of work as a major structural barrier for those already experiencing economic disadvantage [22]. In addition, mistrust in the healthcare system also plays a critical role in research participation, particularly among historically marginalized communities [23,24]. Beyond these structural and historical barriers, rigid trial design and the lack of patient-centered intervention approaches pose practical difficulties for many low-income individuals, limiting their ability to engage in health interventions [25]. Additionally, controlled trials and behavioral interventions often require long-term commitment of 6 months to a year or frequent site visits, which often conflict with participants’ schedules and obligations [25]. These challenges underscore the urgent need for more flexible, community-centered research practices that align with participants’ realities and reduce the burden of participation.

Previous studies provide preliminary evidence that participant-centric approaches (e.g., locating interventions directly within communities) can address systemic barriers to participation and help to facilitate improved recruitment and retention among economically disadvantaged participants [25,26,27,28,29]. Further, studies show that participant-centric and multifaceted engagement strategies can mitigate some of these challenges and improve recruitment and retention rates in health interventions [27,30,31,32]. Thus, utilizing participant-centric strategies including locating interventions directly within communities—leveraging trust, providing accessible service locations, and culturally tailored support are likely the key to reaching and sustaining engagement among those most at risk [33,34,35,36]. These studies indicate that lifestyle modification programs and interventions implemented at the neighborhood level that are designed with participants’ lived experiences in mind may better reach the populations most in need.

Participant-centric recruitment and retention strategies that are both culturally and structurally responsive show promise in improving engagement in community-based diabetes interventions, particularly among low-income and underserved populations. The Healthy Eating and Active Living (HEAL) Diabetes Program was implemented as a community-based, medically tailored nutrition intervention to help low-income patients with early diabetes reverse or manage diabetes through sustainable behavior changes (internal program protocol, 2023). This study was conducted as part of the larger randomized controlled pilot study assessing the feasibility of the HEAL Diabetes Program. The current study was conducted to understand and describe the effectiveness of engagement strategies employed and identify areas for improvement. We aim to examine the recruitment and retention strategies employed in HEAL Diabetes, explore participants’ engagement experiences, and identify areas for improvement. Specifically, we compare electronic registry-based recruitment approaches with more personalized, participant-centric strategies to determine which methods are most effective for reaching and retaining individuals who are typically underrepresented in diabetes research and programs.

2. Methods

We employed a descriptive, exploratory mixed methods design to explore recruitment strategies and outcomes for the HEAL Diabetes intervention. Recruitment took place from August 2023 through August 2024, and data collection continues through August 2025. Our goal was to recruit 20 participants per cycle across three recruitment cycles, aiming for a total of 60 participants. The purpose of this study is to report the lessons learned from recruitment, outline the strategies employed, challenges encountered, and practical recommendations for future community-based diabetes interventions.

2.1. Setting

Recruitment, enrollment, data collection, and most intervention procedures for the HEAL Diabetes program took place in three health coach-staffed neighborhood health hubs (NHH) operated by the University of Tennessee Health Science Center (UTHSC) Tennessee Population Health Consortium (TN-PHC) in Memphis, TN. These hubs deploy certified health coaches to serve a majority of low-income and African American populations by offering neighborhood-based essential primary and preventive care services and increasing access to critical chronic disease prevention and self-management resources.

2.2. Identification of Eligible Participants

Program inclusion criteria were adults 18 years or older, a self-reported T2D duration of 0–6 years, HbA1c value ≥ 6.5 percent at the screening visit, BMI ≥ 25 kg/m², and access to a cell phone or smartphone with texting and voicemail capabilities, and the ability and willingness to participate in the program over 12 months. Exclusion criteria were current use of insulin or more than two hypoglycemic medications (either oral or injectable), a recent routine HbA1c greater than or equal to 12%, weight loss of >5 kg within the last six months, inability to understand consent procedures, inability to read or speak English, pregnancy or considering pregnancy, diagnosis or exhibited unstable psychiatric condition and any form of cognitive impairment. In addition, participants were excluded if they were participating in another controlled research trial.

2.3. Data Collection and Analysis

The program coordinator maintained a record of individuals who met basic prescreening criteria established for participation in the community-based diabetes program. In addition to recruitment logs, fieldnotes were recorded throughout each recruitment phase and during program implementation to document barriers, facilitators, and contextual factors influencing the process. These fieldnotes were analyzed using thematic analysis. The research team used a structured coding process in Microsoft Excel Version 16.97.2 (25052611) to identify and organize codes, which were iteratively refined and grouped into overarching themes related to participant engagement. Quantitative data on the number of individuals prescreened, deemed ineligible, declined, and enrolled were also tracked. Descriptive statistics, including counts and proportions, were used to summarize recruitment activity and retention patterns over the course of the study. Analyses were conducted using REDCap (Research Electronic Data Capture), hosted at the University of Tennessee Health Science Center (UTHSC) Center for Biomedical Informatics.

3. Results

This section details key outcomes from the implementation of a participant-centric recruitment strategy within the HEAL Diabetes program. Findings are organized into four domains: (1) recruitment and retention approaches and their effectiveness, (2) demographic profiles of enrolled participants, (3) recruitment and retention metrics, and (4) qualitative insights into participant-reported barriers and facilitators to program engagement.

3.1. Recruitment and Retention Strategies

Given the challenges of engaging low-income populations, we initially implemented a multifaceted recruitment strategy across two distinct phases. In Phase 1, recruitment efforts combined registry-based outreach with community-based strategies to identify eligible participants. In the first phase of recruitment, the Tennessee Population Data Network (TN-POPnet), a statewide chronic disease registry [37], was employed to identify potential subjects. An authorized data scientist extracted a list of adults with documented diabetes diagnoses from a chronic disease registry, focusing on 10 medically underserved ZIP codes. However, the year of diabetes diagnosis and the type of diabetes were unknown. Study staff then conducted cold-call outreach to individuals on this list to assess interest and invite them to participate. However, in Phase 2, registry-based recruitment was suspended to focus efforts on a more participant-centric approach that prioritized trust-building and community engagement.

We partnered with the NHH to recruit eligible patients through physician and health coach referrals, flyer dissemination, and community engagement. Direct outreach to communities leveraged collaborations with local partners, including social service organizations, libraries, and schools. Grassroots outreach efforts included setting up tables at community events, distributing flyers at barbershops, local eateries, and libraries, and hosting informational sessions during meetings. We recruited participants at local health fairs, community events, and clinical sites frequented by our target population. This approach enhanced both participant engagement and sample relevance, aligning more closely with the study’s emphasis on patient-centered insights and real-world experiences.

Our retention strategy was focused on building trusting relationships with participants, which included assigning participant-selected health coaches for consistent follow-ups. In addition, the program was delivered within three university-operated NHHs where potential participants were established clients and were familiar with the site and program personnel. Various study personnel implemented regular check-ins to address barriers and reinforce commitment to the program. We also provided flexibility, offering virtual and in-person sessions to accommodate participants’ schedules, as well as letting them guide future meeting dates. Finally, we welcomed family involvement and peer accountability partners throughout the program.

3.2. Participant Demographics

Table 1. Participant demographics. Demographic characteristics of participants enrolled in the HEAL Diabetes program (N = 35) * Note: gender includes all individuals approached for participation (N = 126), as it was collected during initial outreach.

Variable	Category	Number	Percentage (%)
Race	Black or African American	34	97.0
	Other	1	3.0
Gender	Male *	37	29.4
	Female *	89	70.6
	Total *	126	--
Age (Years)	Under 30	0	0
	31–40	3	8.6
	41–50	3	8.6
	51–60	13	37.1
	61– above	16	45.7
Education	Less than high school	1	2.9
	High school diploma/GED	6	17.1
	Some college/Associate’s	18	51.4
	Bachelor’s degree or higher	9	25.7
Marital Status	Single	13	37.1
	Married/Partnered	8	22.8
	Divorced/Separated	8	22.8
	Widowed	6	17.1
Income (Annual)	<USD 25,000	20	57.1
	USD 25,000–USD 49,999	6	17.1
	USD 50,000–USD 74,999	5	14.3
	≥USD 75,000	4	11.4

displays the demographic characteristics of the participants recruited and enrolled. Of the total 126 participants recruited for the program, 70.6% were women, while men accounted for 29.4% of those recruited. Women also accounted for most participants (74.3%) who were screened and elected to enroll in the study, compared to 25.7% of men. The median age of participants enrolled in the study was 59.8 years, but ages ranged from 33 to 81 years. Most participants were African American (97.1%) and the remaining 2.9% were Caucasian American; 37.1% were single/never married, 22.9% were married, and 17.1% were divorced, and another 17.1% were widowed. Participants had diverse educational backgrounds, ranging from the equivalent of a high school diploma to postgraduate degrees; in fact, 28.6% of participants had some college education, 22.9% had an associate’s degree, 17.1% had a bachelor’s degree, while 11.4% had a General Educational Development (GED). Participants’ household incomes varied, but we successfully recruited a relatively high percentage of low to medium-income participants. At least 74.3% of participants enrolled in the program had an income less than USD 49,000; 37.1% had an annual household income of USD 10,000–USD 24,999, while 20% had an annual income of USD 10,000 or less.

3.3. Enrollment and Retention Outcomes Across Recruitment Phases

3.3.1. Participant Recruitment and Enrollment Rates

We encountered high refusal rates and limited interest in the program in the first phase of recruiting, which included cold-calling participants from a disease registry. We approached a total of 642 individuals. However, of the 519 individuals contacted through TN-POPnet, only three enrolled, reflecting a 0.5% enrollment rate in this preliminary phase of recruitment. In addition, individuals recruited from TN-POPnet made up only 2.4% of participants who were contacted and proceeded to the prescreening process for the program. These outcomes highlighted the limitations of registry-based cold-calling and underscored the need for more tailored recruitment strategies. Given the low engagement through registry-based outreach, we shifted to a second phase of recruitment that employed participant-centric approaches, intended to improve the reach among our population of interest. Of the 123 individuals recruited from participant-centric direct outreach approaches (e.g., marketing, community events, health coach-referral), all underwent the telephone prescreening process.

Overall, a total of 126 people were reached out to and prescreened, including 3 from the registry. Of these, 83 (65.9%) were deemed eligible for in-person screening and 43 (34.1%) were ineligible, with the major reasons including: (1) year of diabetes diagnosis, (2) use of insulin, and (3) not being overweight or obese. Among the 83 eligible participants, 63 (75.9%) attended and completed the in-person screening. Among those who completed the in-person screening, 56 (88.9%) were eligible to enroll in the program. Ultimately, 35 of the 56 participants who completed the in-person screening and were eligible were enrolled in the program, resulting in a 62.5% enrollment rate among those screened. However, of the broader pool of 83 individuals who were initially deemed eligible after the telephone prescreening, only 35 completed all procedures required for enrollment, resulting in an overall engagement rate of 42.2%. We successfully recruited 87.5% of our target sample (35 of 40 participants) over two recruitment cycles within a ten-month period. The first cycle yielded an 85% (17 of 20 participants) recruitment rate, and the second yielded a 90% (18 of 20 participants) recruitment rate. A third recruitment cycle was suspended due to budget constraints. Figure 1 presents a CONSORT flow diagram of participant recruitment for the HEAL Diabetes program.

The figure describes the recruitment process with the number of individuals who were potentially eligible, the number who were excluded, and the number who agreed to participate. (PDF) Assessment of the Capacity to Consent to Treatment in Patients to Acute Medical Wards. Available from: https://www.researchgate.net/publication/26785289_Assessment_of_the_Capacity_to_Consent_to_Treatment_in_Patients_to_Acute_Medical_Wards#fullTextFileContent [accessed on 17 March 2025].

Figure 2 illustrates the comparative effectiveness of participant-centric, community-engaged recruitment strategies versus traditional clinic-centric recruitment (TN-POPnet). Recruitment outcomes varied by strategy. Of the total 35 participants enrolled, 32 (91%) were recruited through community-engaged, participant-centric methods: 15 via health coach referrals, 10 through direct community outreach, and 7 through targeted marketing. In contrast, only three participants (9%) were enrolled via the traditional clinic-centric method (TN-POPnet). Community-based approaches, including health coach referrals, direct outreach, and targeted marketing, led to substantially higher participant enrollment despite the study’s small sample size (N = 35). This suggests that relational, trust-building recruitment methods rooted in community engagement may be more successful in reaching underrepresented populations.

3.3.2. Participant Retention and Follow-Up

Overall Recruitment and Retention

Table 2. Recruitment and retention metrics for the HEAL Diabetes program.

Stage	N	Retention (%)
Total individuals prescreened	126	–
Eligible for in-person screening	83	65.9
Completed screening	63	75.9
Eligible after screening	56	88.9
Consented and enrolled	35	62.5
Retained at 1-month follow-up	31	88.6
Retained at 3-month follow-up	30	96.8

summarizes the overall recruitment and retention outcomes among all participants approached for the HEAL Diabetes Program. Among the 126 individuals who were approached for participation, 65.9% were eligible for in-person screening, 75.9% completed the screening process, 88.9% were deemed eligible after screening, and 62.5% consented and enrolled in the program. Among enrolled participants, 88.6% completed the 1-month follow-up, and 96.8% were retained through the 3-month follow-up.

Table 3. Themes derived from qualitative content analysis of fieldnotes and recruitment log notes.

Clusters	Themes	Examples
Barriers	Poor intervention integration in clinical practice	Limited provider engagement Unanticipated difficulty identifying individuals with diabetes duration of 0–6 years
	Social determinants of health	Inadequate transportation Disability restricted mobility and participation in study activities
	Daily life demands	Time constraints (study duration) Competing responsibilities (caregiving, employment)
Facilitators	Trust building	Community-centric outreach and direct engagement at community events and key neighborhood sites Staff addressed benefits and concerns directly Good reputation within the community due to quality health service delivery
	Adaptability of intervention	Flexible scheduling On-site and virtual options Culturally-tailored delivery
	Tangible support for participation	Transportation assistance Nutrition support

presents the recruitment and retention metrics for the HEAL Diabetes Program. The table outlines the number of participants at each stage of the recruitment process, including the total number approached, those eligible for in-person screening, those who completed screening, those eligible after screening, those who consented and enrolled, and the retention rates at 1-month and 3-month follow-ups. The Retention Rate (%) column represents the percentage of participants who moved from one stage to the next, indicating the retention or drop-off at each point in the recruitment process.

Retention Outcomes by Cohort

Beyond overall metrics, cohort-specific retention patterns provide additional insight into participant engagement. A total of 30 of the 35 enrolled participants completed the study. In the first cycle of recruiting (N = 17), retention rates for the 1, 3, and 6-month post-baseline assessment visits were 87.5%, 100%, and 92.9%, respectively. In the second cycle of recruiting (N = 18), retention rates for the 1 and 3-month post-baseline assessment visits were 88.9% and 100%, respectively. Among those who did not complete the study, four had withdrawn due to time constraints, transportation barriers, or refusal to complete the program; one participant was lost to follow-up by the 6-month check-in but had completed all intervention activities, though their data submission was incomplete. These findings reflect strong short-term retention among enrolled participants, despite earlier attrition during the eligibility and screening phases.

3.3.3. Barriers and Facilitators to Engagement

Table 3 shows the qualitative findings related to barriers and facilitators to participant engagement.

Thematic analysis of the program coordinator’s fieldnotes and recruitment log notes revealed critical themes highlighting both barriers and facilitators to engagement. Key themes related to barriers included poor intervention integration in clinical practice (e.g., limited provider engagement and difficulties in identifying recently diagnosed individuals), daily-life demands (e.g., time constraints and caregiving responsibilities), and SDOH (e.g., inadequate transportation, disability). Many TN-POPnet-identified individuals expressed hesitancy to participate without direct endorsement by their provider, despite reassurance that provider and institutional approval had been obtained. Social and health-related factors, including multiple chronic conditions and mobility limitations, further hindered participation. Potential participants were also hindered by time constraints because of employment, operation hours at the NHH, and caregiving responsibilities. Key themes related to facilitators included trust-building, intervention adaptability, and tangible support for participation. Potential participants reported that trust was strengthened by community-focused outreach, peer referrals, and direct engagement from NHH personnel. Adaptability was another key facilitator; cultural tailoring and providing flexible participation options helped accommodate busy schedules. Finally, tangible support like food vouchers, grocery deliveries, and transportation assistance alleviated financial and logistical burdens.

4. Discussion

The current study underscores the effectiveness of a participant-centric, community-engaged recruitment approach in facilitating both enrollment and retention of low-income individuals in medically underserved areas. The HEAL Diabetes program successfully enrolled 35 of its targeted 40 participants, retaining 30 (85.7%) through program completion. Most participants were recruited through health coach referrals, direct outreach, and community-based marketing, highlighting the value of personalized and contextually grounded strategies. In contrast, attempts to recruit via the electronic medical record–based chronic disease registry (TN-POPnet) were largely ineffective. Of the 519 individuals contacted through TN-POPnet, only 3 enrolled, a 0.5% enrollment rate. These findings illustrate the limitations of traditional, clinic-centric recruitment methods for smaller-scale, community-based interventions.

While we anticipated higher recruiting from the electronic medical record-based chronic disease registry (TN-POP-net), we found it ineffective in recruiting patients for smaller community-based clinical trials. A major limitation of approaching patients by cold-calling from a registry is high refusal rates [38] and limited engagement, likely due to the lack of provider endorsement and limited contextual information, such as diabetes onset. Community-based surveillance systems like the TN-POPnet can provide a potential solution for early identification and recruitment, but to be effectively used, they must be more deeply integrated into clinical workflows and paired with trusted, relationship-based outreach strategies [37,39]. This research is consistent with prior research highlighting the importance of trusted provider referral for community-based health intervention recruitment [40,41]. A previous study conducted among patients with diabetes showed that direct referral was used for fewer patients but resulted in a higher enrollment rate compared to other methods [41]. However, securing referrals remains an ongoing challenge despite prior studies indicating that the use of electronic health records and direct referrals are effective strategies. A previous study conducted in patients with cancer showed that physicians who spent the most time in patient care had the lowest probability of discussing clinical trials with their patients [42]. Despite securing high-level approval from two specific health systems, we lacked direct engagement with individual healthcare providers. While institutional support was obtained, this did not automatically translate into provider-level buy-in or referrals to the program, which are critical for identifying and enrolling eligible participants when using a chronic disease registry [39]. Many frontline clinicians, including primary care physicians and endocrinologists, were unaware of the program. This gap in engagement limited recruitment from the clinical setting.

Furthermore, our research demonstrated substantial practical difficulties in identifying individuals recently diagnosed with diabetes in medically underserved areas, leading to major practical recruitment challenges requiring intensive screening of large numbers to achieve sufficient recruitment. Many of the individuals encountered during recruitment had been living with diabetes for more than six years, making them ineligible for the program despite their interest in participating. This issue likely arose due to the high prevalence of longstanding diabetes among those in our targeted community. This finding aligns with prior research showing that early-stage diabetes is often underdiagnosed or poorly tracked in low-income and medically underserved populations [43]. Inadequate early identification of incident diabetes may make diabetes remission programs particularly difficult to implement without intensive community-based surveillance.

In alignment with existing research, time commitment and frequent site visits remain a challenge to engagement [20,22,25,44]. Many potential participants faced competing priorities, including work and other personal obligations, making it difficult for them to commit to a year-long program. For individuals managing strict work schedules or involved in other activities, setting aside time for regular participation was particularly challenging. Additionally, participants expressed concerns about sustaining engagement given the need for consistent attendance and active involvement in program activities. In addition, many participants were the primary caregivers for grandchildren or other family members, making it difficult to attend program activities. Bierer and colleagues similarly found that underrepresented women (e.g., racial minorities, economically disadvantaged) often cited caregiving responsibility as a barrier to participating and sustaining engagement in clinical research [25].

In addition, we identified multiple SDOH barriers to both recruitment and sustaining participation in the program. Structural barriers like transportation and health status have consistently been cited as significant obstacles to sustained engagement in long-term health interventions [21,45]. Despite efforts to provide transportation assistance, participants were still constrained by transportation barriers, including meeting eligibility requirements (e.g., being 65 years old or older or Medicaid-eligible), rules requiring 48-h advanced scheduling, and failure to distinguish NHH as healthcare clinics, thus restricting access to free medical transportation. Consistent with prior research, health-related challenges, including having multiple chronic conditions and mobility limitations due to disabilities, hinder participation in research and access to health services [46,47].

Existing studies have shown the importance of fostering trust to improve engagement in research and interventions [32,33,48,49,50]. One of the most impactful strategies was actively building trust through direct engagement. Recognizing that many potential participants had skepticism toward healthcare interventions, the program prioritized community-driven outreach. Our findings contrast with a study on diabetes trial recruitment strategies, which reported that media advertisements and community screenings were labor-intensive and time-consuming yet yielded minimal recruitment success [40]. In fact, we found that direct engagement and offering community screenings at key community sites and events were critical for increasing program visibility and establishing trust in research personnel. NHH personnel and the program coordinator played a vital role in bridging this trust gap by providing clear, culturally relevant information about the program’s benefits and addressing concerns directly. In addition, previous positive experiences with other programs provided through the NHH increased participant interest and engagement. Those existing relationships between NHH personnel and previous clients also led to unexpected peer-to-peer and word-of-mouth referrals. Further, receiving direct referrals from NHH health coaches was the most critical tool for identifying and engaging potentially eligible individuals.

Echoing prior findings, intervention adaptability was another important facilitator [18,25,32,34,50]. Providing flexibility for participants (e.g., offering virtual sessions, encouraging whole-family attendance, and allowing participants to direct the scheduling) enabled more individuals to participate despite demanding schedules. By reducing some practical burdens, these adaptations made the program more accessible to those who otherwise would not have been able to engage. Further, in alignment with existing research, recruitment strategies and the program were culturally sensitive (e.g., diverse study personnel, adjusting nutrition education based on participants’ requests) [29,50]. Finally, consistent with prior studies, providing participants direct assistance and tangible support helped to mitigate some financial and other social barriers that deterred participation [21]. Prior studies have shown that providing direct transportation services (i.e., ride-sharing or insurance-provided transportation) has high rates of patient satisfaction and visit completion [21,45]. The HEAL Diabetes program was free-of-charge, provided food vouchers or three months of grocery deliveries, and ride assistance, including access to rideshare services like Uber and Lyft when medical transportation was not available. These provisions not only facilitated participation and encouraged engagement but also reinforced the program’s commitment to addressing the broader social determinants of health that impact diabetes management. Together, these strategies underscore the value of designing diabetes interventions that are both culturally responsive and structurally supportive, particularly for populations facing compounded barriers to program engagement.

Limitations

This study has a few important limitations. Our intentional use of multiple participant-centered strategies to enhance recruitment, enrollment, and retention among financially disadvantaged individuals makes it difficult to determine which specific approaches were most effective. While attrition was low, we were unable to follow up with the few participants lost to follow-up (LTFU), making it difficult to assess how study duration or other contextual factors may have contributed to disengagement. Additionally, although we recorded fieldnotes and documented informal conversations throughout the recruitment process, we did not formally evaluate participants’ experiences with our participant-centered program design and recruitment efforts. Future studies should incorporate direct feedback from participants to better understand which strategies best promote engagement in pragmatic trials.

Finally, the sample size was small for the initial pilot feasibility study, a limitation that limits the broader generalizability of our findings. However, pivoting to community-engaged recruitment strategies allowed us to reach participants in more contextually relevant settings. Despite the small sample size, the transferability of our findings lies in the depth of contextual insight gained through community-based engagement. The practical lessons drawn from our recruitment and engagement process may be especially relevant for similar populations and settings seeking to improve participation among structurally marginalized groups.

5. Conclusions

This study shows that engaging underserved and underrepresented populations in behavior change interventions requires more than the conventional recruitment methods. Registry-based recruiting is insufficient on its own, especially without partnership with healthcare providers. Whereas participant-centric strategies, emphasizing trust-building, practical support, and cultural and structural tailoring, were key to achieving improved enrollment and retention among a predominantly African American, lower-income cohort. Further, this study found that social determinants of health and competing daily demands are continued barriers to intervention engagement. In contrast, facilitators like intervention flexibility, regular outreach, and support services (e.g., transportation, appointment reminders, nutrition support) enable sustained participation. These findings underscore the importance of embedding recruitment and engagement strategies within participants’ lived realities. While participant-centric approaches can enhance the reach of diabetes interventions among marginalized populations, engaging low-income patients in community-based diabetes programs requires a multifaceted approach that addresses participant engagement barriers at the clinical, community, and individual levels.