Texas Strength Through Resilience in Diabetes Education (TX STRIDE): Protocol for a Non-Randomized Cluster-Controlled Trial among African American Adults with Type 2 Diabetes

Abstract

Objective:

African American (AA) adults have the highest prevalence of type 2 diabetes mellitus (T2DM) and face more severe complications compared with Hispanic and non-Hispanic White populations. Lifestyle interventions that address the distress inherent in T2DM and unique stressors faced by AAs that worsen diabetes-related health outcomes are critically needed. We describe the protocol of a non-randomized cluster-controlled trial that uses a resilience-based diabetes self-management education and support (RB-DSMES) intervention to help participants manage the psychosocial and behavioral demands of T2DM.

Participants:

A total of 284 AA adults were recruited through 27 churches in Austin, TX, and the surrounding areas using church announcements and on-site glucose screenings.

Methods:

Churches were allocated to a comparison (standard diabetes self-management education and support; DSMES) or intervention (DSMES plus resilience resource integration; RB-DSMES) arm. Following baseline data collection, both arms received 8 educational classes, 8 support groups, and 2 booster sessions.

Outcomes:

T2DM physical (primary outcome HbA1c) and mental (primary outcome depressive symptoms) health outcomes in intervention vs. comparison arms will be compared at 6-, 12-, and 24-months post-study entry. We also will examine potential mediators of treatment effects at 24 months via changes in resilience resources at 6 months and T2DM self-management and hypothalamic-pituitary-adrenal (HPA) axis function at 12 months.

Discussion:

If successful, this project will provide crucial guidance for addressing the T2DM burden among AAs by establishing the real-world efficacy of the RB-DSMES program and identifying behavioral and biological mechanisms through which the program impacts T2DM health outcomes.

1. Introduction

Diabetes is the 7^th leading cause of disability and premature death in the United States [1], accounting for the highest healthcare spending among all disease categories in adults [2–3]. African Americans (AAs) bear a disproportionate burden of the type 2 diabetes mellitus (T2DM) epidemic, with double the prevalence [4] and a 30% higher incidence [5] compared with non-Hispanic Whites. These disparities reflect the biological and behavioral challenges of managing T2DM and the compounded impact of systemic inequities, cultural stressors, and social determinants of health [6]. Acute and chronic stresses worsen diabetes outcomes by reducing adherence to self-management behaviors [7–9], increasing insulin resistance through neuroendocrine pathways [10], and exacerbating physical and mental health complications [11–12]. The persistent burden of T2DM among AAs has devastating consequences, with disproportionately high rates of end-stage renal disease [13], retinopathy [14], and amputations [15–16]. There is a pressing need for scalable interventions to mitigate these adverse T2DM outcomes in AA adults.

While traditional diabetes interventions have focused primarily on education and behavioral strategies [17–19], there is a growing need to address biopsychosocial contributions to T2DM [20]. One such approach is through the study of resilience—an underexplored but valued construct, particularly in AA communities [21]. Resilience refers to the process of successfully adapting to challenges by maintaining or regaining mental and physical well-being through mental, emotional, and behavioral flexibility, as well as effective coping strategies and support from social resources [22]. Resilience-based approaches offer a promising avenue to mitigate the psychosocial and biological mechanisms driving adverse T2DM outcomes. The Texas Strength Through Resilience in Diabetes Education (TX STRIDE) clinical trial is the first to test the effects of a resilience intervention on T2DM health outcomes through behavioral and biological pathways in AA adults with diabetes (Exhibit 1). By testing the integration of resilience-based components within a standard T2DM self-management curriculum, this trial seeks to bridge the gap between culturally tailored interventions and sustainable health improvements in AA communities.

Exhibit 1.

Conceptual framework of resilience-based diabetes self-management education and support (RB-DSMES)

Note: DHEA: dehydroepiandrosterone; HbA1c: hemoglobin A1c; HPA: hypothalamic-pituitary-adrenal; T2DM: type 2 diabetes mellitus

2. Study design

2.1. Overview of study design and procedures

The primary goal of our ongoing trial was to test the real-world efficacy^a of a resilience-based diabetes self-management education and support (RB-DSMES) intervention using a cluster-controlled trial design [23–24]. We planned to recruit 32 predominantly AA churches (i.e., clusters) in eight cohorts over 4 years, with each cohort including four churches (two intervention, two comparison) randomly assigned to receive standard DSMES or RB-DSMES (standard DSMES + resilience resources). Each church group was anticipated to include approximately 8 to 10 participants and receive a structured curriculum of eight weekly class sessions, eight biweekly support group sessions, and two bimonthly booster sessions, all held in-person at church locations.

The COVID-19 pandemic required adaptations to the study design. In January 2020, baseline data collection for Cohort 1 occurred at four churches, and in-person small-group educational classes were underway. However, in March 2020, all research activities on the principal investigator’s (PI’s) campus were paused due to the pandemic. Six months later, the TX STRIDE trial resumed operations remotely, restarting Cohort 1 and shifting all educational classes and support group sessions to a video conferencing platform (i.e., Zoom). Remote procedures were used for data collection, with research staff assisting via Zoom or telephone. As COVID-19 restrictions eased, onsite data collection at church locations resumed in July 2022, though educational class sessions remained remote for the rest of the trial.

Exhibit 2 shows a sample timeline for one cohort with data collection occurring at baseline and 6-, 12-, and 24-months post-study entry. All study procedures were approved by The University of Texas at Austin Institutional Review Board (Number 2019080008), ensuring compliance with ethical guidelines and human subject protection. Additionally, the study is overseen by a two-member Data Safety and Monitoring Board that meets annually to provide an external, independent assessment of project safety and quality.

Exhibit 2.

Sample timeline for one cohort. Data collected at each time point includes demographics (baseline), medical history and medications (baseline, 6-, 12-, 24-months), resilience resources (baseline, 6-, 12-months), dietary recall (baseline, 6-, 12-months), actigraphy for physical activity (baseline, 6-, 12-months), hair sample for HPA axis function (baseline, 6-, 12-months), and T2DM health outcomes (baseline, 6-, 12-, 24-months).

2.2. Research aims

Aim 1: To compare T2DM physical health outcomes (primary outcome: HbA1c) and mental health outcomes (primary outcome: depressive symptoms) between the RB-DSMES arm vs. the DSMES arm at 6-, 12-, and 24-months.

Aim 2: To test potential mediators of RB-DSMES (vs. DSMES) on T2DM physical and mental health outcomes at 24 months. Potential mediators are resilience resources (measured at 6 months) and self-management behaviors and hypothalamic-pituitary-adrenal (HPA) axis function (measured at 12 months).

2.3. Conceptual framework

Study aims are grounded in our conceptual model of how developing resilience resources influences T2DM outcomes (Exhibit 1). The model also advances knowledge of the behavioral and biological mechanisms by which resilience resources impact T2DM outcomes. Findings from our pilot study suggest that building resilience resources may offset the devastating consequences of T2DM among AAs [25]. Resilience is valued in AA communities as a skill developed through experiences of discrimination and struggle [26]. Despite its potential as a health promotion tool, few interventions have leveraged AAs’ perspective of resilience in diabetes management programs [27–28]. In this study, we conceptualize resilience as a dynamic set of behavioral and psychosocial resources that help individuals manage the chronic stressors of diabetes.

In addition to the potential direct effect of resilience resources on T2DM health outcomes, resilience resources may also impact T2DM health outcomes indirectly by promoting adherence to self-management behaviors and improving HPA axis function. Modifiable self-management behaviors can control and delay disease progression, but racial disparities in these behaviors exacerbate poor T2DM health outcomes. AAs have lower adherence to diabetes medication guidelines [29], consume more high-glycemic index and high-sugar foods and beverages [30], and eat less fiber, fruits, and vegetables compared to NHWs [31], all of which are associated with poor T2DM health outcomes [32–34]. This study examines whether (and potentially which) resilience resources impact self-management behaviors and subsequent T2DM health outcomes among AAs.

Variation in HPA axis function, an indicator of biological stress, is also associated with T2DM health outcomes [35–36], but this relationship is not well-studied longitudinally [36–37]. Although resilience resources are associated with indicators of HPA axis function (i.e., cortisol, DHEA) [38], it is not known whether building resilience resources over time improves long-term HPA axis function measures that can influence T2DM health outcomes. The present study will investigate that hypothesis.

2.4. Participant eligibility

The inclusion criteria were derived using insights from our pilot RB-DSMES intervention [25], which emphasized the importance of including a wide range of ages and stages of disease progression in studies focused on T2DM management. The primary criteria for participant inclusion were: (1) be AA; (2) be diagnosed with T2DM; (3) be 18 years of age or older; and (4) not currently participating in another T2DM management program. Individuals were excluded if they (1) were pregnant or lactating; (2) had medical conditions requiring special diets that conflict with standard T2DM dietary recommendations (e.g., congestive heart failure requiring severe sodium and fluid restrictions, chronic kidney disease requiring low-potassium and low-phosphorus diets); (3) had physical activity restrictions severe enough to preclude walking three times per week (e.g., uncontrolled chronic obstructive pulmonary disease, unstable angina); (4) were undergoing intensive medical treatments that could interfere with study participation (e.g., dialysis, chemotherapy); or (5) were using glucocorticoid-containing medications (e.g., prednisone, methylprednisolone). The study initially planned to exclude individuals with a bald or shaved head due to difficulty obtaining hair samples for cortisol assessment. However, this requirement was removed due to challenges in recruiting male participants for the study and the large number of potential male participants who were bald.

3. Recruitment, enrollment, allocation, and follow-up

3.1. Recruitment

Church-level.

We recruited predominantly AA churches that expressed support for the project, following methods used in prior work [25]. Our research team had previously established trusted relationships with local church leaders who recognized the potential impact of church-based research on improving T2DM management in their congregations.

Participant-level.

Participants were recruited primarily through church announcements and assistance from church leaders. On-site glucose screenings allowed us to identify and recruit eligible individuals with poor diabetes control who might not otherwise volunteer for a T2DM intervention. Participants without a healthcare provider were referred to a study physician and endocrinologist responsible for medical oversight of the study, who facilitated participant medical management of T2DM at a local Federally Qualified Health Center.

3.2. Enrollment

Potential participants who met eligibility criteria and expressed interest in the program were scheduled for baseline testing at an onsite church location. Enrollment officially occurred once participants provided informed consent and completed baseline testing.

3.3. Intervention Allocation and Follow-up

Initially, four churches representing Cohort 1 were randomly assigned to either the intervention condition (RB-DSMES) or the comparison condition (DSMES). However, modifications to the initial allocation protocol were necessary due to the COVID-19 pandemic and associated organizational constraints (e.g., church scheduling conflicts or lack of readiness to host the intervention), participant recruitment challenges, and program instructor-related challenges (e.g., health issues, scheduling conflicts). After conversations with organizational leaders and team members, we agreed to conduct a non-probabilistic intervention allocation with subsequent churches to allow for greater flexibility in participant recruitment. Through this modified approach, we ultimately included 27 churches in the study (n = 284; 89% of the targeted sample).

Participants in both conditions were informed that the study aimed to test the efficacy of our T2DM program. Upon completing the baseline onsite testing, each participant received a Fitbit Inspire health and fitness tracker (Google LLC, Mountain View, CA) for motivation in monitoring daily step targets, a set of resistance activity bands, and a diabetes curriculum notebook (either intervention RB-DSMES notebook or comparison condition DSMES notebook). At present, all cohorts have completed data collection through 12 months (Exhibit 3). Follow-up data collection at 24 months is ongoing.

Exhibit 3.

Diagram showing the flow of clusters (churches) and study participants through each stage of the TX STRIDE trial.

4. Measurement

Data are collected at four time points over 24 months at onsite church locations. Data categories and timing of collection are displayed in Exhibit 2. Demographic characteristics, including age, sex, year of T2DM diagnosis, previous heart disease or stroke, marital status, education level, employment status, current household annual income, and any prior diabetes instruction, were collected at baseline.

4.1. Primary outcomes

The primary physical health outcome, hemoglobin A1c (HbA1c), is determined through finger-prick capillary blood collection and measured on a DCA Vantage Analyzer (Siemens, Tarrytown, NY). This method meets the National Glycohemoglobin Standardization Program certification criteria for a total coefficient of variation <3% in the clinically relevant range [39]. The primary mental health outcome, depressive symptomatology, is measured with the self-report Patient Health Questionnaire-9 (PHQ-9; α = .86) [40], which measures the frequency and severity of depressive symptoms.

4.2. Secondary outcomes

Secondary T2DM physical health outcomes include fasting blood glucose, lipids, body mass index, waist circumference, and blood pressure. Fasting plasma concentrations of glucose, total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides are determined through finger-prick capillary blood collection and analyzed enzymatically with Alere reagent on a Cholestech LDX analyzer (Alere, Waltham, MA). Body mass index is calculated as weight in kilograms divided by height in square meters. Body weight is measured using a Tanita Professional Digital Scale (Model BWB-800, Arlington Heights, IL). Participants wear street clothes and no shoes; coats and belts are removed. Height is determined using a portable stadiometer (Seca 214, Hanover, MD). Waist circumference is measured to the nearest 0.1 cm using a non-stretchable standard tape measure attached to a spring balance, placed 0.1 cm above the iliac crest on a horizontal plane. Blood pressure (systolic and diastolic) is measured with an Omron HEM-907XL Automatic Inflation Blood Pressure Monitor (Omron Healthcare, Lake Forest, IL) following American Heart Association guidelines [41]. After a 5-minute rest period, three consecutive readings are taken 1 minute apart, and the average is used for analyses.

Secondary T2DM mental health outcomes include diabetes distress and general perceived stress. Diabetes distress is measured with the 4-item Diabetes Distress Scale (α = .76) [42–43], assessing the emotional burden due to having T2DM. General perceived stress is evaluated by the 10-item Perceived Stress Scale (α = .82) [44–45], measuring the degree to which situations in one’s life are appraised as stressful.

4.3. Resilience resources

The conceptual definitions and measurements of resilience resources are summarized in Exhibit 4.

Exhibit 4.

Conceptual Definitions and Measurement of Resilience Resources

Resilience Resource	Conceptual Definition	Scale	# Items	Reliability
Adaptation to Stress	Personal qualities that enable an individual to persevere and adapt positively in the face of adversity.	Connor-Davidson Resilience Scale [69–70]	10	.85
		Brief Resilience Scale [71]	6	.80 – .91
Finding Positive Meaning	The extent to which individuals find meaning in the context of T2DM.	Positive Meaning Scale [72]	5	.73
Adaptive Coping Strategies	Cognitive and behavioral strategies that enable an individual to handle adversity.	Coping Orientations to Problems Experienced Scale (Brief COPE) [73–74]	14	.76
Spiritual Coping	Practicing coping behaviors grounded in connection with the Creator or the universe.	Spiritual-Centered Coping subscale, Africultural Coping Systems Inventory [75]	8	.71
Coping with Discrimination	How minority individuals respond to discrimination.	Coping with Discrimination Scale [76]	22	.72 – .90
Self-Efficacy	How self-confident individuals are in managing their T2DM despite life stressors.	Modified Generalized Self-Efficacy Scale [77]	10	.76 – .90
Social Support	Perceived adequacy of support for T2DM management from family, friends, and significant others.	Multi-Dimensional Scale of Perceived Social Support [78]	12	.88
Emotional Regulation	Awareness and acceptance of emotions. Strategies used to regulate emotions.	Brief Version of the Difficulties in Emotion Regulation Scale [79]	16	.92

4.4. Mediator variables

This study examines T2DM self-management behavior and HPA axis function as potential mediators. T2DM self-management behaviors included dietary intake, physical activity, and diabetes self-care adherence. Dietary intake was measured by trained interviewers who collected two 24-hour dietary recalls using the multiple pass technique by telephone (ensuring data for one weekend and one weekday) within 10 days of each data collection session (baseline, 6-, 12-months). All recall data were analyzed using the Nutrition Data System for Research (NDS-R; 2012, University of Minnesota). The NDS-R calculates key dietary variables, including energy intake, macronutrients, total and added sugars, fiber, glycemic index, daily servings of food and beverages, and a Healthy Eating Index—a measure of diet quality based on alignment with national recommendations [46].

Physical activity was measured using a wrist-worn actigraph (wGT3X-BT, Actigraph LLC, Pensacola, FL), which participants wore for seven consecutive days during the same week as their in-person data collection session and the dietary recalls (baseline, 6-, 12-months). The actigraph records data on minutes and percentage of time spent in light physical activity, moderate to vigorous physical activity, and sedentary behavior.

Diabetes self-care adherence was measured by the Self-Care Inventory-Revised (α = .87) (SCI-R). This instrument includes three items to assess glucose self-monitoring and three to assess compliance with prescription medications [47].

HPA axis function was assessed through measurements of hair cortisol and DHEA. Hair samples (3 cm) were collected using thinning shears as close to the scalp as possible near the posterior vertex [48]. To address potential hesitation from participants about providing hair samples, a professional hair stylist collected the samples, which helped reduce discomfort or concern [49]. Hair samples were processed using standardized washing and extraction procedures to minimize the influence of external contaminants, including cosmetic hair products, consistent with previously published protocols [49]. Steroid levels were determined using a commercial high-sensitivity enzyme-linked immunoassay kit (Salimetrics LLC, State College, PA) following the manufacturer’s protocol. Participants with a bald or shaved head did not provide hair samples.

4.5. Measurement adjustments during COVID-19 restrictions

Due to COVID-19 restrictions, data collection temporarily occurred in a remote format, requiring modifications to specific measurement procedures for nine churches at baseline and four churches at 6 months. Participants were mailed a data collection package containing a survey packet, an at-home A1CNow^® kit (PTS Diagnostics, Whitestown, IN), an actigraph, hair collection materials, and step-by-step instructions. Each A1CNow^® kit included supplies for four A1c tests, ensuring sufficient materials for training and obtaining two HbA1c values to assess the reliability of the device. To maintain data integrity, ensure proper test administration, and answer participant questions, research staff scheduled individualized virtual data collection sessions where they provided step-by-step guidance on using the A1CNow^® device [50]. When A1CNow^® was used, values were adjusted using an established correction factor to account for known systematic differences relative to DCA-based HbA1c measurements and to allow for comparability. We previously established the feasibility and bias of the A1CNow^® in the current population and applied this correction factor (y = 0.665 + 1.003x) to each HbA1c value obtained with the A1CNow^® [50].

Hair samples were also collected at home with the assistance of a family member following detailed instructions. Height and body weight were self-reported from a recent doctor’s visit or using home equipment. Data collection of blood pressure, lipids and cholesterol, and waist circumference were omitted during this period. Once completed, the data collection packet was returned to the investigators using prepaid packaging.

5. Comparison arm

The scope and sequence of the DSMES curriculum, aligned with national standards [51], is summarized in the left column of Exhibit 5.

Exhibit 5.

Overview of comparison (standard DSMES) and intervention arms (RB-DSMES)

STANDARD DSMES	RB-DSMES (Standard DSMES + resilience resource integration)
Week 1. What is Diabetes? Why you have T2DM and what you know; participant goals & wishes The role of sugar and the role of insulin Signs & symptoms of diabetes Confirming a diagnosis of diabetes Long term complications of diabetes	Week 1 integrates the following resilience resources into DSMES: Introduce resilience resources; emotions associated with T2DM Achieving resilience; positive outcomes in the face of adversity Discrimination, spiritual coping, emotional regulation, and historical significance of resilience in AAs as a means of adapting to adversity Definitions: diabetes distress, perceived stress Positive meaning: what good has come from your T2DM diagnosis?
Week 2. Getting Your Glucometer to Work for You How to use your glucometer; challenges to testing blood glucose Glycemic targets – blood glucose and A1C Desired values when testing Troubleshooting values that are too high or too low (self-efficacy)	Week 2 integrates the following resilience resources into DSMES: Awareness of maladaptive coping strategies (e.g., denial, fatalism) Applying adaptive coping strategies (action, planning, reframing, acceptance, support) with emphasis on spiritual coping (e.g., prayer) to manage blood glucose
Week 3. The Importance of Carbohydrates Understanding the link between blood glucose and food intake What foods have carbohydrates & what foods are carb-free The technique of carb counting; reading food labels Glycemic index and load Troubleshoot values that are too high or too low	Week 3 integrates the following resilience resources into DSMES: Cultivating social support, with emphasis on testing blood glucose and preparing healthier culturally preferred foods, e.g., at church socials Continue applying adaptive coping strategies and spiritual coping strategies in making healthy food choices Focus on cultural food preferences and diabetes self-management
Week 4. Learning How to Build Your Plate The carb section of the healthy plate The food groups; number of servings per day & serving Cultivating social support from family and friends for food choices Relationship between knowledge and self-efficacy in making healthier food choices; how to enjoy dining out while making healthy choices Choosing healthier options at the store; nutritious recipes on a budget	Week 4 integrates the following resilience resources into DSMES: Seeking social support in making healthy food choices Adaptation to stressful situations; a positive mindset; the responsibility model, how to change an unhelpful thought, and meaningful connections Culturally preferred snacks and tips for healthy meal planning Link food choices/preferences to AA culture and historical events Mindful eating – what is it anyway? What about dessert?
Week 5. A Step Towards Success The benefits of physical activity (aerobic, balance, flexibility, strength) Strategies for adding steps to your day (baseline, benchmark, build) T2DM and physical activity: monitoring your blood glucose The importance of carrying your diabetes identification card How to create a supportive external environment (e.g., fitness partner(s), social support, community resources).	Week 5 integrates the following resilience resources into DSMES: Explain the link between physical activity and emotional regulation Using physical activity to foster mindfulness (emotional regulation, positive meaning, adaptive coping, adaptation to stress) Accountability provided by peer mentors (social support) for performing regular physical activity Being resilient – realistic expectations; reducing sedentary time
Week 6. Understanding Medication The role of medications (oral and injectable) Understand each medication you take and why Organs involved and their role in helping regulate your blood glucose Important other powerful free medications (e.g., activity, reduce stress) Ask the Endocrinologist – questions and answer about medications	Week 6 integrates the following resilience resources into DSMES: Address the cultural belief that God will take care of their diabetes versus personal responsibility and spiritual coping Coping with discrimination in healthcare settings (e.g., interactions with your doctor, lack of access to medicine, feelings of intimidation) Internal (positive meaning, self-confidence), external (social support, health coaches and providers), and existential (faith) resilience resources
Week 7. Emotional Health: A Key Factor in Preventing Complications Preventing short and long-term diabetes complications Macrovascular and microvascular diabetes complications Diabetes and mental health; diabetes and dental health Setting S.M.A.R.T. goals	Week 7 integrates the following resilience resources into DSMES: Emotional regulation – why a key component of diabetes care Adaptation to stress and overworked stress response systems Expressing emotions in a healthy way; empathy, compassion, self-efficacy Coping effectively (awareness, acceptance, action, assessment)
Week 8. Diabetes Care Schedule and Staying Connected Daily and yearly diabetes self-management responsibilities Special topics (e.g., driving and traveling with T2DM, alcohol and T2DM) Taking care of yourself when sick and when to call the doctor	Week 8 integrates the following resilience resources into DSMES: Finding positive meaning in the adversity of being sick (e.g., reflection, renewal, reemphasis on self-care); tiny habits repeatedly done become automatic; self-efficacy despite adversity
Support Groups 1 through 8 Assessment and knowledge gaps Setting short and long-term goals; community resources Group problem solving of personal barriers to self-management Grocery store tours, recipe sharing, cooking demonstrations	Support Groups 1 through 8 integrates the following into DSMES: Informal group discussions; problem solving framed in AA cultural preferences Historical cultural factors that influence health/health behaviors Continual application of resilience resources amid ongoing life stressors
Booster Sessions 1 and 2 Check-in: current health status, challenges to diabetes self-management Diabetes information review and long-terms plans for staying on track	Booster Sessions 1 and 2 integrates the following into DSMES: Continual application of resilience resources for ongoing life stressors Sharing stories of resilience and improved T2DM self-management

6. Intervention arm

The RB-DSMES builds on foundational components of standard DSMES by integrating resilience resources within existing DSMES topics rather than adding separate content (Exhibit 5, right column). Resilience resources (e.g., adaptation to stress, finding positive meaning, adaptive coping, coping with discrimination, spiritual coping, emotional regulation) were embedded into standard DSMES domains such as nutrition, physical activity, medication adherence, and problem-solving through applied examples and facilitated group discussion. RB-DSMES and DSMES sessions followed the same schedule and class length (90 minutes). To support coverage of all curriculum material, participants were encouraged to review notebook materials before and after sessions. Following each class session, a follow-up text message and email was sent to participants reminding them of important class material discussed and referring them to their notebooks for review.

7. Treatment fidelity

Participant attendance was logged to track intervention dosage. To maintain program fidelity, instructors received feedback and additional training and coaching when necessary. As COVID-19 restrictions eased, classes restarted remotely via a video conferencing platform (i.e., Zoom) allowing the PI to attend a majority of the RB-DSMES intervention group and DSMES comparison group classes. This ensured the fidelity of the curriculum and validated the integration of resilience resources into RB-DSMES sessions. To limit contamination/treatment leakage, the RB-DSMES instructors were different than the DSMES instructors. The research assistant assigned to each church prepared a reflection summary after each class to document material covered by instructors in both the standard DSMES comparison group and the RB-DSMES intervention group. Summaries were emailed to instructors and the PI to ensure treatment fidelity. Follow-up check-in meetings with instructors occurred throughout the intervention.

8. Data management and analysis

Data will be merged, edited, and cleaned using SPSS software version 30 (IBM Corp, Armonk, NY). Quality control checks (i.e., checks for missing data, logical inconsistencies, and completeness) will be performed throughout the project.

8.1. Analyses of primary outcomes

Aim 1 examines differences in T2DM physical and mental health outcomes between RB-DSMES and DSMES participants at 6-, 12-, and 24-months. We will use multilevel mixed-effects longitudinal models to estimate treatment effects over these measurement time points. These models include both fixed and random effects and are generalizations of repeated-measures ANOVA that include every participant in the analysis, regardless of missing data, and explicitly account for the hierarchical structure of the data, including the clustering of measurements within participants and participants within churches. In addition to participant-level covariates (e.g., cohort and baseline outcome), models will also include selected church-level covariates derived from aggregated individual-level characteristics, consistent with prespecified imbalance diagnostics, when standardized mean differences exceeded 0.25 (e.g., church-level mean age and sex by intervention condition) to account for between-church differences.

We will fit a separate model for each physical and mental health outcome. Each model will include fixed effects for intervention (RB-DSMES vs. DSMES) interacted with measurement occasion (baseline, 6-, 12-, 24-months) and cohort and baseline covariates along with random effects for participants and churches. While these adjustments improve estimation under a non-randomized cluster design, they cannot eliminate bias from unmeasured confounding at either the participant (i.e., individual) or church (i.e., cluster) level. To test for an overall treatment effect for each outcome, we will compare this model’s fit against the fit of a model without intervention effects using a likelihood ratio test for multilevel models. We will use the model to compare model-based estimated mean outcome values for RB-DSMES and DSMES participants at each measurement occasion to estimate the intervention effects over time. There is some evidence that men achieve lower HbA1c more often than women in response to T2DM treatment [52]. Thus, we will include a treatment × sex interaction term to estimate differential treatment effects by sex.

The planned sample size was informed by simulation-based power analyses conducted during trial planning, using effect size estimates from pilot quasi-experimental studies [25, 53]. Assuming approximately 10 participants per church across 32 churches, simulations accounted for clustering at the church and participant levels and reflected the planned multilevel longitudinal analytic approach. Under these assumptions, the study was estimated to have greater than 90% power to detect intervention effects for the primary outcomes of HbA1c and depressive symptoms at a two-sided α of 0.05. Due to COVID-19-related disruptions, the final sample included 27 churches rather than the 32 originally planned. However, the achieved sample size approached the target specified in the original power calculations and participant retention has been greater than originally predicted.

8.2. Mediation analysis

Aim 2 examines whether the effects of RB-DSMES (vs. DSMES) on T2DM physical and mental health outcomes at 12- and 24-months are mediated by changes in resilience resources at 6- and 12-months. To test this aim, we will first estimate the intervention’s effect on each component of resilience resources at 6- and 12-months, using the multilevel longitudinal model described above. Next, we will use multilevel longitudinal models to estimate the effects of resilience resources at 6- and 12-months on T2DM outcomes at 12- and-24 months, respectively, alongside direct (i.e., unmediated) effects of the intervention. We will combine these fitted models via simulation methods [54] to estimate the extent to which the intervention’s effect on all components of resilience resources simultaneously explains its effect on T2DM health outcomes [55]. The longitudinal design allows us to control for baseline and intermediate values of both resilience resources and T2DM outcomes, which will minimize unobserved confounding.

To examine whether changes in resilience resources driven by RB-DSMES (vs. DSMES) affect health outcomes indirectly, via self-management behaviors and HPA axis function, we will first use a multilevel regression model to estimate the effects of resilience resources at 6 months on self-management behaviors and HPA axis function at 12 months. We will control for all baseline measurements and group assignment. Next, we will estimate the effects of resilience resources at 6 months, as well as self-management behaviors and HPA axis function at 12 months, on T2DM health outcomes at 24 months, again controlling for baseline levels and group. Finally, we will estimate natural direct and indirect effects to assess the extent to which changes in resilience resources contribute to T2DM health outcomes. Hair washing frequency and bleach use will be included as covariates, per recommendations [56]. Confidence intervals will be generated using bootstrapping methods to improve the robustness of our statistical estimates.

Studies examining whether resilience resources predict T2DM health outcomes via self-management behaviors and HPA axis function tend to control for—rather than investigate—the impact of sex on these associations [57]. To address this lack of evidence, we will include an interaction between sex and the predictors to determine whether sex affects Aim 2 causal pathways. We expect to find differences between intervention groups in the effects of resilience resources on T2DM health outcomes and that these differences will be due to direct effects as well as indirect effects via self-management behaviors and HPA axis function.

8.3. Randomization diagnostics and potential solutions

We will conduct diagnostic checks to evaluate imbalance on observed covariates and study outcomes between the intervention and comparison conditions using conventional statistical comparisons at the individual level and standardized differences at the cluster (church) level [58–59]. Standardized mean differences at the church level between ±0.05 and ±0.25 standard deviations will be considered adequate [60]. If church-level differences (> |0.26|) are found, the corresponding church-level covariates will be included directly in the multilevel models to account for imbalance and mitigate potential confounding and selection bias [61–62]. Adjustment for observed characteristics can improve comparability between intervention groups but cannot eliminate unmeasured confounding or establish equivalence.

9. Discussion

Although T2DM is a psychologically and behaviorally demanding chronic condition [63], research suggests that enhancing resilience resources can help adults with T2DM improve their health [64–66]. Building on our previous work [25, 67], this project aims to address the T2DM burden among AA adults by evaluating the real-world efficacy of RB-DSMES and identifying the behavioral and biological mechanisms through which the program may improve T2DM health outcomes. This goal aligns with the NIH National Center for Complementary and Integrative Health’s strategic plan to advance research on the role of resilience in health promotion and disease management [68].

The study is significant because it will (1) test a culturally tailored, theory-derived RB-DSMES intervention with strong potential to enhance T2DM health outcomes, (2) improve understanding of the impact of RB-DSMES on resilience resources and the association between resilience resources and T2DM outcomes, and (3) advance knowledge of the behavioral and biological mechanisms through which resilience resources may impact T2DM outcomes.

We will collaborate with local churches, the Black Nurses Association, and various church organizations to explore potential implementation strategies (e.g., placing the RB-DSMES curriculum and associated materials on the website of each participating church) to expand the program’s reach and ensure its sustainability beyond the funding period. Once the trial is completed, we plan to seek funding to gather information and explore the context for future implementation by incorporating a complementary intervention research component (to approximate the current trial to a Type 1 Hybrid implementation-effectiveness trial design). This component will use mixed-method approaches to conduct a “process evaluation” aimed at identifying implementation barriers and facilitators, as well as potential adaptations needed to improve program reach, adoption, and long-term maintenance.

Highlights:

• Real-world efficacy trial of resilience-based DSMES in African Americans with T2DM

• Intervention examines the effect of resilience on self-management and HPA axis function

• COVID-19 modifications to intervention delivery and data collection are described

• Practical challenges to intervention allocation and analysis solutions are shared

Data availability

No data analyses were used in this article.