Integrating Coaching into Digital Behavioral Interventions: A Scalable Training Model and Fidelity Evaluation

Abstract

Background:

Digital interventions offer scalable solutions for program accessibility but face challenges with engagement and adherence. Hybrid delivery models integrating human support through coaching may enhance engagement outcomes, which can improve intervention or clinical outcomes.

Purpose:

This study describes a structured, intervention-agnostic coaching model and associated training procedures designed to reinforce digital intervention content without requiring specialist expertise. It evaluates the feasibility, fidelity, and acceptability of the coaching model, including participant satisfaction.

Methods:

Parents of toddlers born very preterm (n = 113) received digital parent training or health education paired with weekly coaching calls. Eleven coaches from diverse educational backgrounds completed a standardized training protocol. Feasibility was assessed via observations and coach self-report of adherence and competence. Acceptability was evaluated through parent satisfaction surveys. Descriptive analyses summarized fidelity, engagement, and program completion.

Results:

Coaches demonstrated high adherence (M = 94.9%) and competence (M = 88.9%) across 266 observed calls, with minimal variance. Parent call engagement was rated high (M = 2.70; 3-point scale). Most parents (75%) attended ≥ 6 of 11 coaching calls, and 64% completed all six digital program modules. Satisfaction was strong, with 93.6% of parents reporting being satisfied or very satisfied, and over 70% felt connected to their coach. Calls were brief (M = 12.9 minutes), supporting feasibility for busy families.

Conclusions:

An intervention-agnostic coaching model delivered by non-specialists was feasible and acceptable for parent participants. Development of a standardized coaching manual and testing across additional digital interventions are planned to enhance scalability and replication.

Psychosocial and behavioral interventions, defined as interventions emphasizing psychological or social rather than biological factors, have consistently demonstrated efficacy in preventing and treating a broad range of concerns including childhood behavioral problems, mental health concerns, and substance use disorders [1]. However, these interventions face persistent challenges to scalability and accessibility because they are traditionally delivered in person and they are often limited by a provider’s time-bound schedules [2,3]. In-person interventions are vulnerable to systemic and individual barriers including cost, time demands, stigma, and a shortage of trained professionals [4]. Unfortunately, these barriers contribute to the mental health treatment gap, where nearly 50% of youth and 54% of adults with mental health conditions in the United States go untreated [5,6].

To overcome these limitations, self-administered digital interventions, delivered via websites, smartphones, or text messaging [7], have emerged as effective, scalable alternatives to traditional, in-person approaches. These interventions provide the same core content and strategies as traditional in-person interventions while offering individuals the flexibility to engage with material at their own time, pace, and location [8,9]. Unlike in-person formats, digital interventions are not limited to providers’ schedules and can expand reach and improve access despite significant shortages in behavioral health providers [10].

Self-administered digital interventions yield comparable outcomes to traditional, in-person approaches across key behavioral outcomes [11,12]. Engelbrektsson et al. (2023) found that internet-delivered parent training was non-inferior to group-based in-person formats in a 12-month randomized controlled trial. Similarly, Breitenstein et al. (2016) reported that an internet-based parenting program produced equivalent improvements in parenting practices and child outcomes as an in-person group intervention.

Despite the promising findings of efficacy, adherence (e.g., completion of all intervention components) [13] to self-administered digital interventions tends to be lower than face-to-face, guided interventions [14–16]. Lower rates of adherence may be due to reduced accountability and lack of tailoring or personalization. Self-administered digital interventions require a certain level of motivation and discipline from the user to interact with the intervention independently. In addition, adult learners are motivated by tailored content that adds relevance and value, a task difficult to replicate in self-administered programs [17]. This lack of support and accountability for engagement and tailoring of content in self-administered digital interventions may negatively impact outcomes and completion rates.

In response to these limitations, hybrid delivery models [referred to as hybrid delivery] were developed to preserve the increased reach and accessibility of digital interventions while incorporating the improved engagement and treatment outcomes typically seen with in-person interventions [18]. Hybrid delivery pair self-administered digital interventions with in-person or virtual human support that is tailored to a user’s unique strengths, weaknesses, and circumstances [19]. Pineda et al.‘s (2023) taxonomy of digital mental health interventions further classifies hybrid delivery (Type 3) as being primarily therapeutic or technical based on the type of human support provided. Examples of hybrid delivery include individualized feedback, assistance with troubleshooting, and discussion blogs [19,20].

Another common method of hybrid delivery is the use of coaching delivered through modalities such as scheduled calls or messages. Learning new skills and changing behavioral patterns are difficult and coaches seek to improve user motivation and engagement by supporting goal setting, engaging in problem-solving, reinforcing learning across written and verbal modalities, and providing emotional validation and support [19,21]. These techniques are rooted in Self-Determination Theory which emphasizes competence, autonomy and relatedness as key to sustained behavior change [22]. While some coaches may have backgrounds in mental health, coaching itself is distinct from therapy. Coaching interactions are typically brief and centered on reinforcing existing intervention content, enabling hybrid delivery to maintain the scalability of digital delivery while enhancing personalization, accountability, and engagement [19].

Digitally delivered interventions with a coaching component are associated with superior outcomes and reduced rates of attrition compared to self-administered models [19]. Werntz et al. (2023) conducted a systemic meta-review to explore the efficacy of digitally delivered mental health interventions with and without coaching (“human support”). Of the 38 meta-analyses analyzed, 48% of the studies found that coach-supported digital interventions were significantly more effective than self-guided digital interventions. In contrast, only 9% found that self-administered interventions outperformed coach-supported interventions.

Although hybrid delivery offers many advantages relative to self-administered models of digital interventions and traditional, in-person interventions, there are also several key limitations. One such limitation is that hybrid delivery often relies on highly trained specialists such as therapists and psychiatric nurses, which may undermine scalability given financial burden and provider shortages [19]. Although several studies have explored whether non-specialist coaches demonstrate the same efficacy as specialist coaches, findings have been inconsistent [23,24]. Further research is needed to determine whether coaching can be effectively delivered by lay individuals, which could expand its accessibility and reach.

An additional limitation of hybrid delivery with coaching is the lack of standardized, operational definition of coaching, along with inconsistency in adhering to the key components of “coaching” [25]. Applied clinical practice and research studies lack clear operational definitions of coaching, relying instead on vague distinctions such as “not therapy.” In clinical practice, coaching is often defined as appropriate for sub-clinical concerns, while therapy is an intervention for diagnosed mental health conditions. However, distinctions based solely on symptom severity do not clarify how coaching and therapy differ. For example, cognitive-behavioral coaching use many of the same core strategies as cognitive-behavioral therapy, including identification and restructuring of thought patterns [26]. Other models, such as positive psychology, differentiate coaching from therapy primarily through how goals and outcomes are framed, emphasizing the promotion of well-being and strengths rather than treatment of mental health problems [27]. In their meta-analysis, Ward and colleagues [25] highlighted the absence of fidelity monitoring and adherence to coaching’s core components across many interventions. Further, in their narrative review Meyer et al. [28], identify a lack of reporting of clear training protocols, fidelity, and coaching specific outcomes. To address these gaps, the purpose of this paper is to evaluate the feasibility and acceptability of a novel, intervention-agnostic coaching model designed to support hybrid delivery of digitally delivered psychosocial and behavioral interventions by non-specialist providers. Our specific aims are: (1) to describe the coaching model and associated training procedures; (2) to assess the feasibility of the model and training using coach self-report and fidelity observations; and (3) to evaluate the acceptability of coaching, including participant satisfaction with the coaching experience from the perspective of participants.

Methods

Study Design

This article reports the coaching protocol and fidelity results from a randomized trial evaluating the independent and combined effects of delivering digital parent interventions paired with weekly coaching calls [29], hereafter referred as the primary study. Using a descriptive cross-sectional design, we will evaluate the feasibility and acceptability of coaching as a hybrid method for delivering digital interventions.

Sample size and eligibility criteria

In the primary study, parents (N = 224) of a child aged 20–30 months corrected age who were born very premature (gestational age < 32 weeks) were randomized to one of four conditions (digital intervention, digital intervention plus coaching, control digital intervention, or control digital intervention plus coaching). The sample in this study includes parents (n = 113) who were randomized to a coaching condition in the RCT [29].

Eligibility criteria for the primary study were (a) the individual is the parent/legal guardian of a child born very premature (VPT) and 20–30 months corrected age at enrollment, (b) parent can speak and read English, and (c) the parent has access to an electronic device (e.g., smartphone, tablet) to receive phone calls and the intervention. The exclusion criterion was if the child scored 3 SDs below the mean on the Vineland Adaptive Behavior Scale (3^rd ed) [30] communication or socialization index as the interventions would not be appropriate for children with severe-to-profound language and social skill deficits. The primary study focused on the parent population of former VPT infants because these children are twice as likely to demonstrate behavior problems compared with their term-born peers due to neurodevelopmental vulnerabilities [31–33]. For further information on the primary study, see Greene et al. [29].

Participant recruitment

Parents were recruited through Neonatal Intensive Care Units (NICU) follow up clinics at two large midwestern medical centers. Parents were recruited during their in-person appointment at the follow-up clinic to provide information about the study and offer more information. If a parent did not attend or attended but was not reached by the team, research assistants (RAs) called the parent and sent emails to provide information about the study. In addition, potentially eligible families outside of the NICU follow up clinic were identified via medical records and contacted via phone and email.

Procedures

The primary study was approved by the IRB at the Ohio State University. During a phone call, interested parents were screened for eligibility and, if eligible and interested, participated in the informed consent procedures. Upon completion of consent, parents completed baseline measures, followed by randomization to one of the four conditions. In this study, we only include participants in the two coaching conditions.

Coach Protocol

Overview of Coach intervention

The coach intervention builds on prior research focused on the development and pilot testing of the coaching protocol [34]. In this study, coaching is conceptualized as an agnostic framework that leverages principles of communication, problem-solving and behavior change to reinforce existing components of an intervention. Our conceptualization of an intervention-agnostic coaching model means the coaching protocol could be paired with any digital intervention because the coaching methods are not dependent on specific intervention content. However, coaches must have a basic understanding of the intervention to effectively reinforce key concepts and support participant engagement. In addition, coaching does not introduce new intervention concepts but rather uses evidence-based practices to individualize and reinforce components of established interventions.

Each coaching session aimed to (1) establish and maintain rapport with parents, (2) promote accountability for program use, (3) reinforce key intervention content, (4) support and problem-solve around content application, and (5) develop an action plan for the upcoming week. Problem-solving involved identification of problems encountered when applying program content and/or anticipation of difficulties applying new content, generation of solutions and finally selection of solutions. For example, a coach listened to a parent describe that despite their plan to practice praise with their child each morning before preschool, they were not able to practice due to the busyness of the morning time. The coach asked the parent if other times of the day were less busy and more predictable. The parent identified after dinner and before bed as a typically calm time of day and agreed to attempt practice during this window of time. To support these goals, the coaching model incorporated the five essential components of effective coaching outlined by Rush and Shelden (2020): joint planning, observation, action or practice, reflection, and feedback [35]. These components were embedded across a structured 10-week coaching protocol, used consistently for both digital interventions, each of which included six modules for parents to complete. We designed the ten-week coaching schedule to allow parents to complete one module per week while accommodating potential disruptions such as travel, illness, or work. Coaching consisted of an introductory and weekly calls during the 10-week period, followed by a “booster” call one month later. The introductory call focused on building rapport, reviewing the purpose and schedule of the coaching calls, and initiating joint planning by identifying parent priorities and goals and outlining next steps. During the weekly calls, coaches engaged in observation by monitoring parents’ use of the digital intervention via the ezDashboard [36] and through parent self-report. Action or practice was supported through discussions of skill development and encouragement to try out strategies from the digital intervention at home. Reflection was facilitated through conversations that helped parents identify challenges in applying the intervention content or completing the digital intervention modules. Feedback was provided to reinforce effective strategy use and encourage continued engagement. The booster call served as a final check-in to address any remaining content and formally conclude the intervention. Calls were scheduled for 30 minutes for the introductory and booster sessions, and 15 minutes for weekly calls.

Coaches

Coaches were recruited via university and professional job boards and selected through a formal interview. While mental health or coaching backgrounds were not required, evidence of clear verbal communication skills and comfort engaging with parents were prioritized. Two investigators also served as coaches to inform training and provide coaching backup. Coaches were hired on a rolling basis, with eleven individuals from diverse backgrounds serving in this role over the course of the project.

Coach training

Coaches were assigned to one of the two interventions with no crossover to avoid any digital intervention contamination. Coaches worked parttime and completed a standardized 3–4-week training combining asynchronous modules and individualized instruction. Training began with two 1-hour narrated presentations. The first presentation focused on prematurity and child development, providing essential context for working with the population (e.g., parents of children born VPT). The second training presentation focused on communication and interpersonal skills to support the goals of the coaching sessions (e.g., building rapport, fostering intervention completion, facilitating the application of learned strategies, and collaboratively setting weekly action plans). Coaches were trained to avoid therapeutic or expert roles, focusing instead on supportive, goal-oriented conversations that emphasized problem-solving, positive encouragement and reinforcement, and intervention completion. They were also instructed to highlight parent strengths and promote the application of learned strategies across different situations.

To simulate the participant experience, coaches spent 5–6 hours independently navigating the assigned intervention, followed by a one-on-one review with a supervisor to review the intervention content, address questions, and assess readiness for next steps in the coaching training. Coaches were then trained in study operational procedures (e.g., initiating and rescheduling calls, completing post-call documentation, and procedures if a participant missed calls).

Training concluded with practice calls: at least two introductory calls, four to five weekly calls, and two booster calls to demonstrate competence across the full coaching intervention cycle (~ 3–4 hours of practice calls). Coaches received verbal feedback after practice calls, confirmed readiness before working with participants, and completed additional practice if needed. Training was led by the research team, with clinical experts covering content and research staff addressing protocol adherence. Condition-specific supervisors provided post-training supervision to review recent calls, reinforce fidelity to the coaching model, address challenges, and offer feedback.

Variables and Measures

Feasibility: Coaching Fidelity

This study assessed feasibility of the coaching model via coaching fidelity using four metrics: adherence, quality of delivery (competence), coaching dose, and parent engagement (responsiveness). Fidelity was evaluated through independent observation of 33% of the completed coaching calls and coach self-reports submitted after each call.

Adherence.

A fidelity checklist was developed for this study, adapted from the group-based fidelity tool by Breitenstein et al [37] to assess adherence to the coaching protocol and delivery quality (competence). Coaches were rated on whether the task was completed (yes) or not completed (no). Adherence items varied by call type (e.g., introductory (7 items), weekly (6-items), or booster (8-items).

Competence.

Competence was assessed using five items evaluating the coach’s use of (a) active listening, (b) problem solving skills to address barriers to practice or intervention completion, (c) timing and flow of the call, (d) continuity across sessions by referencing prior goals or discussion, and (e) overall quality and affective tone of the interaction. Observers rated these items on a three-point scale: to a great extent, somewhat, or not at all, with detailed coding guidelines provided for consistency.

Observers also provided an overall rating of the tone of the coaching call, defined as the emotional quality of the interaction and the rapport between coach and parent. This dyadic rating considered coach and parent behaviors and was scored as negative, neutral, or positive.

Coaching and program dose.

Coach call attendance (yes/no) and length of call (in minutes) were calculated from the coach self-report. Digital program module completion was determined through automated tracking via the ezDashboard on individual parent usage activity.

Parent engagement during coaching calls.

Parent engagement during coaching calls was assessed through coach ratings and independent observation; however, only the independent ratings are used in this study to reduce potential bias from prior interactions or knowledge of the participant. Engagement was rated using a single global item assessing the extent to which the parent appeared interested and actively involved in the discussion. Observers coded engagement as no/minimal, moderate, or high, based on observable behaviors such as verbal participation, responsiveness, and initiative. The coding manual included clear examples to guide the assessment of parent engagement during coaching sessions. For instance, minimal engagement was characterized by behaviors such as offering only brief or distracted responses. In contrast, high engagement was demonstrated when parents actively participated by asking questions, elaborating on their experiences, or initiating new discussion topics.

Acceptability

Parent satisfaction with coaching.

Parent satisfaction with the coaching experience was assessed at post-intervention using a brief survey. The survey included items evaluating overall satisfaction, perceived helpfulness of the coaching, and the extent to which coaching supported their use of the digital program. Parents completed the satisfaction survey independently after completing the final (booster) call.

Fidelity Coding Procedures

Coder training followed a structured, three-phase process: (1) orientation to the coaching model, (2) instruction on fidelity coding procedures, and (3) practice coding until achieving at least 80% agreement. Three coders were trained to reliably assess adherence to coaching protocols and core competencies. To monitor coder reliability, 10% of all observations were double coded by two trained coders. If coder agreement fell below 80%, coders reviewed the session together to review disagreements and determine a consensus score. Less than 8% of the double-coded fidelity observations fell below 80% agreement. The percent rater agreement across all double coded observations was 88%.

To ensure representative fidelity monitoring, we planned for a random sample of at least 25% of all recorded calls for coding. This sample included calls from a minimum of 25% of coaches and participants, and covered all call types (introductory, weekly, and booster). In total, 33% of completed calls were included for fidelity observation.

Analytic approach

Descriptive statistics were used to summarize the sociodemographic characteristics of parent participants and coaches, as well as key feasibility and acceptability outcomes. Feasibility metrics, including coach adherence, competence, and parent engagement were assessed using both independent fidelity observations and coach self-reports, with call-specific averages computed for each metric. Program engagement was evaluated by calculating the proportion of coaching calls and program modules parents completed. Parent satisfaction with coaching was summarized using survey responses, with results presented as frequencies and percentages across satisfaction domains.

Results

Sample description

The majority of parent participants were mothers (96.5%) and either married or living as married (62.8%; see Table 1). The average age of parents was 33.5 years, with a range of 22 to 54 years. The target children in the study had a mean age of 25.6 months (SD = 2.79). Most parents identified as either Black (32.7%) or White (62.8%).

Table 1.

Sociodemographic Characteristics of Families (n = 113)

Sociodemographic characteristic		n	%
Child Sex	Female	49	43.4
	Male	64	56.6
Relationship with child	Mother (biologic or adoptive)	109	96.5
	Father	4	3.5
Parental marital status	Single	34	30.1
	Married and/or Living as Married	71	62.8
	Divorced/Separated	7	6.2
	Do not wish to answer	1	0.9
Ethnicity	Hispanic/Latino	9	8.0
Race	Asian	1	0.9
	Black	37	32.7
	White	71	62.8
	More than 1 race	3	2.7
	Do not wish to answer	1	0.9
Annual household income	< $30,000	20	17.7
	$30,000 - $79,999	32	28.3
	> $80,000	53	46.9
	Do not wish to answera	8	7.1
		M	SD
Parent Age (years)		33.5	6.1
Child Age (months)		25.6	2.8

^an=1 system missing; n=7 don’t wish to answer

The coaches (n = 11) reported varying levels of experience working with families, ranging from limited (45.5%) to extensive (36.4%; see Table 2). Most held a college degree (54.5%) and were currently enrolled in school (72.7%). Among those enrolled in school, two were studying psychology, two public health, one social work, and three nursing. Three coaches were in undergraduate programs, while the rest were pursuing master’s degrees. The average age of coaches was 30.9 years, with a range of 22 to 53 years. On average, coaches had been in their role for approximately one year (353 days) and worked with 10 parents throughout the project (SD = 7).

Table 2.

Coach demographic characteristics (n = 11)

Sociodemographic characteristic		n	%
Education (highest level)	High School	3	27.3
	College	6	54.5
	Graduate School	2	18.2
Currently in school	Yes	8	72.7
Experience working with families	None	0	0
	A little	5	45.5
	Some	2	18.2
	Quite a bit	4	36.4
Coach is a parent	Yes	3	27.3
		M	SD
Coach age (years)		30.9	11.2

Feasibility

Feasibility of coaching was evaluated using observational fidelity ratings for adherence and competence, coaching dose, and parent engagement. On average, coaches demonstrated high adherence to protocol components (94.9%) and strong competence in delivery (88.9%) across all sessions. Parent engagement was rated high with an average rating of 2.87 (SD = .36; 3-point scale) based on observation and 2.70 (SD = .49) by coach self-report (r = .41). See Table 3 and 4 for call-specific adherence, competence, and parent engagement ratings based on fidelity observation and coach self-report.

Table 3. Coach Feasibility, Dose, and Parent Engagement via observations (n = 266).

	Introductory call (n = 31) M (SD)	Weekly Call (n=210) M (SD)	Booster Call (n = 25) M (SD)	All Calls (N =266) M (SD)
Coach Adherence	.95 (.08)	.95 (.09)	.94 (.09)	.94 (.09)
Coach Competence	1.00 (.00)	.88 (.17)	.91 (.18)	.89 (.17)
Length of Call (mins)	21.5 (7.7)	11.4 (6.1)	14.3 (5.4)	12.9 (7.0)
Parent Engagement *	2.77 (.43)	2.90 (.32)	2.76 (.52)	2.87 (.36)

Note:

^*Parent engagement (1 = no/minimal; 2 = moderate; 3 = high)

Table 4. Coaching Dose and Parent Engagement via coach report (n = 827).

	Introductory call (n = 107) M (SD)	Weekly Call (n=654) M (SD)	Booster Call (n = 66) M (SD)	All Calls (N =827) M (SD)
Length of Call (mins)	22.8 (7.0)	12.0 (5.3)	15.4 (6.3)	13.7 (6.7)
Parent Engagement *	2.77 (.43)	2.70 (.49)	2.73 (.54)	2.71 (.49)

Note:

^*Parent engagement (1 = no/minimal; 2 = moderate; 3 = high)

Coaching and program dose

Across all parents (n = 113), 75% participated in 6 or more of the 11 coaching calls with an average participation rate of 7.4 of the 11 coaching calls (SD = 3.13). The average length of completed calls was 12.9 minutes (SD = 7.0) based on observational coding and 13.7 minutes (SD = 6.7) for coach self-report (r = .95; Table 3 and 4). In addition to completing coaching calls, parents completed an average of 4.6 of 6 modules of the web-based intervention, with 64% completing all six modules.

Acceptability

Parent satisfaction with coaching

Overall, parental feedback reflected high satisfaction with the coaching experience. Of the 113 parents in the coaching condition, 83% completed the post-intervention survey. Among those respondents, nearly 90% of parents felt very comfortable with their coach, and over 70% felt very connected to their coach. Most parents found it easy to attend coaching calls (71%). Coaches played an important role in supporting program use, with 60% of parents reporting substantial help in completing the web-based program and 58% reported the coach helped them apply program content to a great extent. Overall satisfaction was high, with 94% of parents reporting being satisfied or very satisfied with the coaching they received (See Table 5).

Table 5. Parent (n =94) Satisfaction with Coaching.

	Not at all n (%)		Somewhat n (%)	Very n (%)
Comfort with coach	1 (1.1)		9 (9.6)	84 (89.4)
Feeling connected with coach	3 (3.2)		23 (24.7)	67 (72.0)
Ease of attending calls	5 (5.3)		22 (23.4)	67 (71.3)
	Not at all	Very Little	Somewhat	To a great extent
Supported use of digital program	2 (2.2)	8 (8.6)	27 (29.0)	56 (60.2)
Coach helped with applying program content	3 (3.2)	5 (5.4)	31 (33.3)	54 (58.1)
	Very dissatisfied	Dissatisfied	Satisfied	Very Satisfied
Overall satisfaction with coaching	1 (1.1)	5 (5.3)	25 (26.6)	63 (67.0)

Discussion

This study evaluated the feasibility and acceptability of a structured, treatment-agnostic coaching model designed to support hybrid delivery of digital interventions. The three primary aims were: (1) to describe the coaching model and associated training procedures; (2) to assess feasibility using coach self-report and fidelity observations; and (3) to evaluate acceptability from the perspective of participants. The coaching model was designed to reinforce intervention content without requiring specialized behavioral health expertise, thereby addressing scalability challenges common to hybrid delivery. Overall, findings indicate that the model is both feasible and acceptable, with high fidelity in coach delivery and strong parent satisfaction.

Coaching Model and Fidelity

Our results demonstrate that a brief, structured training protocol can produce high fidelity in coaching delivery. Coaches achieved high adherence (95%) and competence (89%), with minimal variance across sessions. This consistency is critical because it reduces a potential “coach effect,” ensuring that intervention outcomes can be attributed to the program rather than individual coach differences. The uniformity achieved with relatively modest training and supervision suggests that the model is scalable and does not require highly specialized expertise.

The diversity of coach background, including age (range = 22–53 years), education (nursing, public health, social work), and experience working with parents underscores the scalability of the model. Notably, 27% of coaches had only a high school diploma and 46% had limited prior experience with parents, yet fidelity remained high. These findings suggest that high-quality coaching can be delivered by nonspecialists when training emphasizes core interpersonal and communication skills, protocol adherence, and sufficient familiarity with intervention content, rather than extensive clinical expertise. This approach is particularly relevant for addressing workforce shortages and expanding access to psychosocial and behavioral interventions and aligning with evidence demonstrating that non-specialist support can enhance engagement in digital interventions [19,23]. By reducing reliance on highly trained specialists, this model addresses scalability challenges identified in prior work [2,3]. Aligning with our findings, recent task-sharing of mental health services literature suggests that pre-training in interpersonal skills and supervision may be key predictors of competence in nonspecialist delivery models [38]. These findings reinforces the importance of relationship-building and communication skills as foundational training targets. While coaching is often considered therapy-adjacent or applied to subclinical populations [23], our findings underscore the distinct value of coaching as a scalable implementation strategy that enables non-specialist providers to effectively reinforce digital interventions.

Acceptability and Parent Satisfaction

High levels of parent satisfaction and program completion are particularly encouraging given the demands faced by parents of young children. Coaching was perceived as helpful for both completing the digital program and applying its content with their child. These findings are noteworthy given the demands on this population of parents of young children born very preterm, many of whom juggle early intervention services and frequent medical appointments. The brevity of coaching calls (average 12.9 minutes) likely contributed to feasibility and acceptability, making participation manageable for busy families. These findings are consistent with prior studies showing that structured, goal-oriented support improves adherence and outcomes in digital interventions [16,21]. Engagement and completion metrics of the digital program and coaching calls further underscore the model’s effectiveness. Parents completed an average of 4.6 of 6 modules, with 64% completing all modules, and attended an average of 7.4 coaching calls. These rates compare favorably to prior research on digital interventions, where attrition is often high [13,14]. The combination of structured coaching and digital delivery appears to enhance program completion without imposing significant time burdens.

Future Directions and Conclusions

This study demonstrated that a structured, treatment-agnostic coaching model can be delivered with high fidelity and strong parent acceptability by non-specialist coaches. Strengths include rigorous fidelity monitoring, a fidelity sampling plan that ensured representation across call types, and a standardized coach protocol. However, limitations such as reliance on descriptive analyses and a single population (parents of children born very preterm) may restrict generalizability. Additionally, while engagement was high, we did not examine whether variations in adherence or competence influenced intervention outcome, a question for future research.

Several lessons emerged that can inform future implementation. While the training protocol was effective, future work could explore whether even more streamlined training maintains fidelity while reducing resource demands. Call structure may also warrant refinement; for example, the booster call, scheduled for 30 minutes, was often shorter (~15 minutes) and may be unnecessary for parents who completed the program within the initial 10 weeks. Tailoring booster calls based on completion status could improve efficiency.

We are developing a standardized manual for coaching procedures to support replication and scalability, consistent with Meyer et al.’s recommendation for replicable coaching models [28]. This manual will be tested with other digital interventions to evaluate generalizability across different content areas and populations. Future research should also examine whether matching coaches to parents based on shared characteristics enhances engagement, though current findings suggest strong rapport can be achieved without such matching. Ultimately, expanding this model to diverse interventions and populations will be critical for addressing persistent gaps in access to psychosocial and behavioral interventions.