Study Engagement in the ROUTE Randomized Controlled Trial: A Diabetes Educator–Led Behavioral Intervention to Promote Continuous Glucose Monitor Use Among Youth With Type 1 Diabetes

Abstract

OBJECTIVE

The purpose of this study was to examine participant engagement in a behavioral intervention designed to optimize the use of continuous glucose monitoring among historically minoritized youth with type 1 diabetes.

RESEARCH DESIGN AND METHODS

We conducted a secondary mixed-methods analysis of quantitative and qualitative data from 60 youth-caregiver dyads who participated in the Research on Optimizing the Use of Technology with Education (ROUTE-T1D) randomized controlled trial. Engagement was measured by intervention session attendance, parent coach engagement, and qualitative interviews with participants.

RESULTS

A diverse population of youth was recruited. Session attendance was high, at 75%. Only greater caregiver education was significantly associated with greater session attendance. There was no correlation between demographic and medical factors and contact with parent coaches. Qualitative results suggest participants felt ease of engagement with the intervention.

CONCLUSIONS

High engagement suggests the intervention was accessible to the diverse study population and demonstrates the feasibility of including diverse populations in diabetes technology research, thereby improving translation to clinical care.

Introduction

Type 1 diabetes (T1D), an autoimmune-mediated condition of insulin deficiency, is one of the most common chronic medical conditions in American youth. It affects 1 in 400 youth under 20 years of age (1) who must be treated with daily subcutaneous insulin (by injection or pump) for life. The incidence of T1D is increasing, especially among non-Hispanic Black (NHB) youth (2,3). Continuous glucose monitoring (CGM), now considered standard of care for T1D management (4) has paved the way for improved glycemia and psychosocial functioning for youth with T1D (5–7). However, disparities exist, with NHB youth having higher hemoglobin A_1c (HbA_1c) levels and lower rates of CGM use compared with their non-Hispanic White peers (8–12). Reasons for these inequities include implicit bias, mistrust of the medical community, and systemic barriers created by adverse social determinants of health such as community factors, social deprivation, and structural racism (13–17). These factors additionally create barriers to research engagement that lead to the underrepresentation of historically minoritized populations in diabetes technology and behavioral health research (18–23). When clinical care decisions are made based on research that excludes the most vulnerable populations, inequities persist and may worsen (24). Addressing these gaps in research requires a deeper, more holistic understanding of the structural and systemic factors that contribute to disparities in youth T1D intervention engagement.

The objective of the current study is to explore intervention engagement in the Research on Optimizing the Use of Technology with Education (ROUTE-T1D) randomized controlled trial using both quantitative and qualitative methods. The ROUTE-T1D study (described below) included primarily youth from historically minoritized backgrounds and offers a unique opportunity to examine how these factors impact study engagement in a behavioral intervention for patients with T1D using diabetes technology and to consider strategies to promote their participation in future studies.

Research Design and Methods

Study Design

We conducted a secondary mixed methods analysis of data from the ROUTE-T1D study, to examine intervention engagement in the single-site randomized controlled trial that evaluated a behavioral intervention to optimize CGM use in youth with T1D (25). Measures of engagement include session attendance, parent coach engagement, and qualitative interviews with participants regarding satisfaction with engagement. All participants in the ROUTE-T1D study were included in this secondary analysis.

Brief Explanation of the ROUTE-T1D Study Protocol and Engagement Strategies

The ROUTE-T1D randomized controlled trial was a diabetes educator–led pilot behavioral intervention for youth and their primary caregiver that was designed to sustain and optimize CGM use, the details of which are published in the study protocol paper (25). Briefly, the study was conducted 13 September 2022 to 11 December 2024 and had a goal of including >50% youth from historically minoritized backgrounds. Participants were recruited from a large urban outpatient diabetes practice. Families were randomized to either the immediate intervention group or the delayed (6 months) waitlist control group. Families in the immediate intervention group immediately participated in 3 monthly appointments led by a diabetes educator who was a part of the diabetes clinical team. Each session provided dyads with strategies including family communication, problem-solving CGM-related challenges, and optimizing CGM use. In addition to the intervention sessions, caregiver participants were matched with a “parent coach.” Parent coaches were referred by members of the diabetes clinical care team and were lay persons and parents of youth with T1D who had been using CGM for ≥1 year. Parent coaches received a virtual training session delivered by a licensed psychologist with background in T1D psychology. Training addressed active listening skills, cultural humility, mental health resources, and delineating parent coach versus medical team responsibilities. Coaches met with the licensed psychologist for ongoing supervision. Parent coaches were expected to make at least three attempts at contact per month for 3 months via phone call, text, and email. Brief questionnaires were completed by parent coaches to track contacts (25). Six months after randomization, families in the delayed intervention group received the three intervention sessions and access to the parent coach. Participants included 60 youth and their primary caregiver. Participants were eligible if they were between the ages of 10 and 15 years old, were diagnosed with T1D for ≥6 months, spoke English or Spanish, and were either new to using a CGM or had <75% CGM use over the last 3 months. Youth were ineligible for the study if they had another major chronic illness, a major psychiatric disorder, or a major developmental disorder that inhibited their ability to complete study activities.

The research team used several strategies to facilitate intervention engagement based on prior research that examined methods to reduce barriers to research participation among historically minoritized populations (18). Intervention sessions were held as telehealth appointments, and scheduling was flexible at times convenient for individual families, such as after school or work. Additionally, families were asked for their preferred method of communication to schedule appointments and provide reminders (e.g., text, email, phone call). Language interpretation and translated materials were available to promote participation by Spanish-speaking participants. Consistent study staff were used to build trust with families. Families were financially compensated at each data collection point and could earn up to $200 for participating in the study. Additionally, before questionnaires were sent at each follow-up period, retention items with the ROUTE-T1D logo (e.g., water bottle, pen, phone holder) were sent to the caregiver and youth participants; no incentives were provided for intervention session attendance. For the parent coach assignment, every attempt was made to match participants with coaches based on similar family experiences. This included matching children with similar interests or hobbies, families speaking the same language, or families that experienced similar challenges with CGM.

Data Collection

Medical and demographic variables were collected at baseline via caregiver report (except for youth sex, which was self-reported by youth), electronic medical record, or Dexcom Clarity. Demographic variables included child age, sex, race or ethnicity, language, caregiver education level, caregiver employment status, caregiver race or ethnicity, caregiver sex, caregiver language, family structure (single-caregiver or multiple-caregiver household), family income, and insurance type. For analyses, race or ethnicity was categorized in the following manner: non-Hispanic Black, Hispanic/Latino/a/x (any race), non- Hispanic White, or another racially minoritized identity. Medical variables included diabetes duration, CGM use duration, mode of insulin delivery (injection vs. pump), HbA_1c, and CGM data. CGM data were collected from Dexcom Clarity with a data collection 90-day time frame. CGM data included time in range (TIR) (i.e., the percentage of time blood glucose levels were within 70–180 mg/dL) and wear time (i.e., percent of the total CGM readings in a 90-day period versus the maximum possible readings). Since all 60 participant dyads completed the intervention, data from both the immediate and delayed group are included in these analyses.

Qualitative Interviews

An optional interview to explore participants’ experience with the study was offered to all participants after they finished the intervention sessions. Transcribed interviews were analyzed by four study team members using rapid qualitative analysis (26). Team members coded two transcribed interviews separately and then met to reconcile summaries and confirm consistency in breadth and depth of information to be captured. The remaining interviews were then divided among team members to code independently. Interview questions were grouped into domains, and a template was used to summarize data by domain. Key quotes were extracted. Data summaries were then transferred into a matrix for synthesis. For the purpose of this secondary analysis, we focused on aspects of the qualitative results that reflected parents’ and children’s perceptions of ease of engagement. Open-ended questions explored barriers or facilitators to scheduling, attending telemedicine sessions, and connecting with the parent coach.

Intervention Engagement

Intervention engagement was tracked after each intervention session and dichotomized into families who attended all three sessions versus families who attended two or fewer. Additionally, parent coach engagement was defined as the number of times caregivers and parent coaches connected via a phone call, text, or email.

Data Analytic Plan

χ² tests and independent samples t tests were used to analyze the associations with intervention session engagement and medical/demographic variables. Independent samples t tests and frequentist one-way ANOVA were used to analyze parent coach engagement associations with medical/demographic variables. A Pearson correlation test was used to analyze the associations between session attendance and parent coach connections. All analyses were performed using IBM SPSS statistics software, version 28.

Ethical Considerations

Caregiver consent and youth assent were obtained during the ROUTE-T1D study enrollment session, which was conducted by a research assistant experienced in pediatric research via Health Insurance Portability and Accountability Act–secure videoconferencing with youth and their primary caregiver; see Straton et al. (25) for details of consent procedure. All procedures were approved by the Institutional Review Board at Children’s National Hospital (Washington, DC) under protocol number Pro00016314. Data were collected between 13 September 2022 and 11 December 2024.

Data and Resource Availability

The data sets generated during and/or analyzed in the current study are available from the corresponding author upon reasonable request.

Results

Sixty youth aged 10–15 years with <75% CGM use and their primary caregivers were enrolled. Ninety-two percent of youth participants were from historically minoritized backgrounds (53% Black/African American, 27% Hispanic/Latino/a/x, 12% another racially minoritized identity [8.3% mixed race and 3.3% Southwest Asian/North African], and 8% White). Among caregivers, 85% were from historically minoritized backgrounds (49% Black/African American, 25% Hispanic/Latino/a/x, 10% another racially minoritized identity [3.3% Asian/Asian American, 3.3% Southwest Asian/North African, and 3.3% mixed race], and 15% White). Mean HbA_1c was 10.6% ± 2.0%, mean CGM wear time was 58% ± 20%, and mean CGM TIR was 30% ± 16%.

Session Engagement by Medical and Demographic Correlates

Overall session attendance was high, at 75% (135 of 180 sessions). Sixty percent of participants (n = 36) attended all three sessions, 15% (n = 9) attended two sessions, 15% (n = 9) attended one session, and 10% (n = 6) did not attend any sessions (Table 1). The primary reason for missed sessions was scheduling conflicts; however, two participants expressed lack of interest as their reason for not attending sessions.

Table 1.

Session attendance in ROUTE-T1D

Overall session attendance	135 (75)
Three sessions attended	36 (60)
Two sessions attended	9 (15)
One session attended	9 (15)
No sessions attended	6 (10)

Data are n (%).

For purposes of analyzing medical and demographic correlates in a small sample, session attendance was dichotomized by those who attended all three sessions (60%) and those who attended two or fewer sessions (40%). Only caregiver education was significantly associated with session attendance, with 79% of caregivers who have an education of up to high school attending two or fewer sessions and 61% of caregivers who have a college degree or higher attending all three sessions (P = 0.002) (Table 2).

Table 2.

Demographic and medical correlates of session engagement in ROUTE-T1D

Characteristic	Overall	Three sessions attended	Two or fewer sessions attended	P
n	60	36	24
Child age (years)	12.9 ± 1.7	13.3 ± 1.7	12.5 ± 1.6	0.08
Child race or ethnicity				0.425
Black, non-Hispanic	32 (53.3)	20 (55.6)	12 (50)
Hispanic/Latinx/a/o	16 (26.7)	7 (19.4)	9 (37.5)
Other historically marginalized identity	7 (11.7)	5 (13.9)	2 (8.3)
White, non-Hispanic	5 (8.3)	4 (11.1)	1 (4.2)
Caregiver race or ethnicity				0.113
Black, non-Hispanic	29 (49.2)	18 (51.4)	11 (45.8)
Hispanic/Latinx/a/o	15 (25.4)	6 (17.1)	9 (37.5)
Other historically marginalized identity	6 (10.0)	3 (8.6)	3 (12.5)
White, non-Hispanic	9 (15.3)	8 (22.9)	1 (4.2)
Child sex				0.593
Female	25 (41.7)	14 (38.9)	11 (45.8)
Male	35 (58.3)	22 (61.1)	13 (54.1)
Caregiver sex				0.462
Female	51 (85.0)	32 (88.9)	19 (79.2)
Male	9 (15.0)	4 (11.1)	5 (20.8)
Caregiver education level				0.002
Up to high school	33 (55.0)	14 (38.9)	19 (79.2)
College degree or more	27 (45.0)	22 (61.1)	5 (20.8)
Caregiver language				0.104
English	53 (88.3)	34 (94.4)	19 (79.2)
Spanish	7 (11.7)	2 (5.6)	5 (20.8)
Family income1				0.053
<$50,000/year	25 (41.7)	12 (37.5)	13 (65)
>$50,000/year	27 (58.3)	20 (62.5)	7 (35)
Insurance type				0.907
Public	43 (71.7)	26 (72.2)	17 (70.8)
Private	17 (28.3)	10 (27.8)	7 (29.2)
Family structure				0.593
Two-caregiver household	35 (58.3)	20 (55.6)	15 (62.5)
Single-caregiver household	25 (41.7)	16 (44.4)	9 (37.5)
Caregiver employment status2				0.617
Not employed full time	18 (30)	10 (28.6)	8 (34.8)
Employed full time	40 (66.7)	25 (71.4)	15 (65.2)
Medical variables
Diabetes duration (years)	5.2 ± 3.4	5.2 ± 3.3	5.2 ± 3.6	0.984
HbA1c (%)	10.5 ± 2.1	10.7 ± 2.3	10.3 ± 1.7	0.454
CGM wear time (%)	45.5 ± 28.3	48.4 ± 29.7	37.4 ± 23.2	0.275
CGM TIR (%)	31.5 ± 16.6	31.5 ± 18.1	30.6 ± 14.6	0.860
Mode of insulin delivery				1.000
Pump	25 (41.7)	15 (41.7)	10 (41.7)
Injection	35 (58.3)	21 (58.3)	14 (58.3)
Duration of CGM use (years)	2.1 ± 1.8	2.4 ± 1.8	1.7 ± 1.8	0.109

Values are n (%) or mean ± SD unless otherwise specified.

¹n = 8 participants declined.

²n = 2 participants declined.

Parent Coach Engagement by Demographic Correlates

Overall, 80% of coach-caregiver dyads made at least one connection. On average, caregivers and coaches connected two times over the 3-month intervention period, with diabetes management in school being the most frequently discussed topic (28%). No significant associations were found between the families’ medical or demographic variables and engagement with the parent coach intervention (P > 0.05). There was a significant correlation between session attendance and parent coach connection, with greater session attendance associated with greater number of parent coach connections (r = 0.379, P = 0.003).

Qualitative Interviews

Of the 60 youth-caregiver dyads, 25 (n = 14 from immediate group, n = 11 from delayed group) completed a semistructured qualitative interview after study completion. Aspects of the interviews related to study engagement included reflections on ease of scheduling and attendance, frequency of sessions, and connecting with parent coach (Table 3). When asked about study pros and cons, caregivers cited factors related to how engagement was facilitated by the team through flexible scheduling options. For example, while some caregivers cited challenges in scheduling sessions because of work and family demands, others found scheduling “convenient,” “accommodating,” and “flexible.” One caregiver appreciated the virtual sessions because “we don’t always have a car, so getting back and forth to [the children’s hospital] was sometimes hard, but the [virtual] meetings were convenient.” Another participant appreciated late afternoon sessions, reporting, “that would be something that I would recommend continuing if possible, like having the after-work hours because [child] is involved in track, so that sometimes presented a challenge, like with the timing, and sometimes I’m driving, so I really appreciate it, that there were extended hours beyond like 9:00–5:00.” Overall, 17 caregivers (68%) reported no scheduling difficulties; 3 caregivers expressed a desire for more flexible timing. Suggestions to improve engagement included offering night and weekend sessions.

Table 3.

Factors related to engagement in qualitative interviews

Theme	Exemplar quotes
Flexible scheduling options enabled attendance	“We don’t always have a car, so getting back and forth to [the children’s hospital] was sometimes hard, but the [virtual] meetings were convenient.”	“That would be something that I would recommend continuing if possible, like having the after-work hours because [child] is involved in track, so that sometimes presented a challenge, like with the timing, and sometimes I’m driving, so I really appreciate it, that there were extended hours beyond like 9:00–5:00.”
Frequency of intervention sessions mattered	“I actually thought they were pretty well spaced apart. It gave us enough time in between to, kind of, implement the things that we had learned prior and make adjustments.”	“I think I would have liked there to just be more calls just because it did make me feel—it helped me feel supported. Like you know, I wasn’t the only one that was worried, you know, about her diabetes.”
Connection with parent coach was challenging	“It was okay. I mean, I wasn’t able to respond to probably as much as I liked to but for the most part, letting them basically them reaching out and letting me know that they’re here for support was the biggest thing.”	“I think [challenges connecting with parent coach] was a combination of missing each other and my forgetfulness. Weekly reminders would have been helpful.”

When asked about appropriateness of intervention frequency, caregivers overall felt comfortable with intervention frequency, although some wanted more frequent or closer together sessions. Nine caregivers found the monthly session spacing appropriate, with one stating that monthly spacing gave them “enough time in between to, kind of, implement the things that we had learned prior and make adjustments.” Five caregivers expressed a desire for more frequent sessions, with one stating, “I think I would have liked there to just be more calls just because it did make me feel—it helped me feel supported.”

Despite the use of multiple communication strategies (email, text, and phone calls), experiences with the parent coach intervention component were more mixed. Some connected with a parent coach several times, while others said they did not connect or had limited contact with a parent coach. Of those who did connect with a parent coach, most felt supported by a peer with similar experience: “wasn’t able to respond to probably as much as I liked to but for the most part, letting them basically them reaching out and letting me know that they’re here for support was the biggest thing.” Another caregiver noted the ease of engaging with “somebody who understood.” Lack of time and attempted but unsuccessful communication were cited as reasons for lack of engagement with a parent coach. “I think it was a combination of missing each other and my forgetfulness. Weekly reminders would have been helpful.” Because of the young age of the youth participants, youth qualitative interviews focused on satisfaction with the intervention rather than factors related to engagement.

Conclusions

In ROUTE-T1D, a diverse study population composed primarily of youth from historically minoritized backgrounds offers a unique opportunity to examine study engagement among those underrepresented in diabetes technology and behavioral research and to consider strategies to promote their participation in future studies. With the use of targeted recruitment strategies to enable participation, we were successful in recruiting a diverse population with high levels of engagement with intervention sessions and parent coaches, suggesting feasibility of engagement in this historically underrepresented cohort.

Although our sample size is small, the lack of significant correlations between demographic or medical factors other than caregiver education with session attendance underscores the feasibility of engaging historically minoritized participants from diverse backgrounds for studies of behavioral interventions and diabetes technology. Among other studies examining engagement in youth with T1D in behavior change interventions, those with lower representation of historically minoritized youth (27–29) found correlations between engagement and demographic and medical factors, while those with more diverse study populations (20,30,31), similar to ours, found no significant association between engagement and demographic or medical factors. Through engaging diverse participants, research findings become more applicable and translatable to clinical practice.

It is notable, since those with higher HbA_1c and lower CGM TIR are historically underrepresented in or excluded from diabetes technology research (32–34), that neither HbA_1c (M_HbA1c = 10.6% ± 2% [92 mmol/mol] at baseline) nor TIR (M_{time-in-range} = 30% ± 16% at baseline) were correlated with session attendance. This finding emphasizes the feasibility of including those with above-target HbA_1c in diabetes technology research. When those with less optimal glycemia are included in research, evidence-based clinical decisions can be informed with consideration of all populations, thereby reducing inequities.

The current finding that caregiver education greater than high school was associated with greater session attendance suggests barriers existed for caregivers with less than high school education. Surprisingly, employment status, which is often associated with education level, was not significantly associated with session attendance. Targeted strategies for caregivers with less than high school education in future interventions may involve reviewing recruitment and intervention materials to ensure best practices for accessibility are used, incorporating strategies to verify caregiver understanding, or considering community-based partners to aid in retention. More work is needed to understand barriers experienced by caregivers with less than high school education so these can be addressed adequately.

Qualitative interviews supported the ease of flexible scheduling options and overall satisfaction with the frequency of the intervention sessions, while offering suggestions for improving accessibility. Future studies may include options for night and weekend intervention sessions. There was a high level of connection between participants and parent coaches, with no correlation between engagement and medical or demographic variables, again suggesting this component of the intervention is accessible and feasible to a diverse population. However, ease of engagement with the parent coach was more mixed, suggesting this is an area for improvement in future studies. Caregivers who engaged with the parent coach found it meaningful, and parent coach contacts were correlated with greater session attendance, suggesting efforts should be made to facilitate parent coach engagement in future studies. Future research should continue to prioritize inclusive recruitment strategies, cultural humility in intervention designs, and long-term follow-up to better understand and mitigate barriers to sustained engagement. Inclusivity in diabetes technology research will lead to informed strategies to address disparities in diabetes care and outcomes.

Health care professionals including physicians, nurse practitioners, nurses, diabetes educators, and dietitians play a crucial role in recognizing and addressing behavior change and technology use in youth with T1D. Use of diabetes educators in behavioral interventions such as ROUTE-T1D can facilitate translation to clinical practice (35). Clinically applicable behavioral interventions such as ROUTE-T1D must be developed with attention to the entire patient population. Proposed solutions to addressing disparities in diabetes care and outcomes include improving health care worker awareness of existing disparities, addressing social determinants of health, using community-based approaches, and engaging in advocacy to dismantle racist policies (17,36–39). Importantly, the inclusion of historically minoritized populations in diabetes technology and psychology research is a strategy to address disparities.

Although there are important findings from this work, study limitations must also be considered. The small sample size limited statistical power in detecting relationships among demographic and medical correlates of study engagement. It is possible that associations exist that were not detected because of the small sample size, underscoring the need for further work with a larger sample. Further, while the intervention was developed to be inclusive of youth and parents from diverse backgrounds, the study team did not engage community members in either intervention development or strategies to support study engagement and retention. Including community members during development and execution of intervention trials may support trial engagement (40). Larger trials are needed to understand how systemic racism impacts exclusion of historically minoritized people in diabetes technology research.

Conclusion

Understanding whether intervention engagement varies by medical and demographic factors is essential to designing effective, equitable, and clinically applicable interventions. ROUTE-T1D was successful in recruiting a diverse sample of historically minoritized youth, who are underrepresented in diabetes technology research, and offers a unique opportunity to explore correlates of study engagement and opportunities to increase research participation in this population. High engagement suggests the diabetes educator–led intervention was accessible to a diverse study population and demonstrates the feasibility of including diverse populations in diabetes technology research. Health care professionals, who are uniquely positioned to deliver such interventions, must consider the relevance of diabetes technology and behavioral research to their various patient populations when delivering evidence-based care. When diverse populations are included in diabetes technology and behavioral health research, clinicians are better equipped to make evidence-based decisions for their patients and improve care and outcomes for all youth living with diabetes.

This article contains supplementary material online at https://doi.org/10.2337/figshare.33030845.

Funding Statement

Research reported in this publication was supported by the National Institutes of Diabetes and Digestive and Kidney Diseases (R01DK131026), https://clinicaltrials.gov/study/NCT05564481, study registration date 4 October 2022. R.S. is the grant recipient.