DDBT’s Discover phase includes understanding individuals’ prior experiences, needs, and preferences that may facilitate or hinder an intervention’s usability (23). For this study, the exploratory Discover phase included quantitative baseline questionnaires to understand Enhearten adaptation needs in the context of participants’ baseline reports of self-stigma, opioid use, MOUD use, MOUD perceived behavioral control, and mental health symptoms. The Discover phase extended through the first month of Enhearten usage for surveillance of onboarding and feedback on initial experiences.

The DDBT Design/Build phase (23) rapidly adapts the intervention using participant feedback. For our study, Design/Build included iterative activities around the 1-month period to collect interim user feedback (qualitative and quantitative). Participants completed a brief structured questionnaire to understand perceived ease of use, usefulness, and intervention acceptability. They completed a semi-structured interview about which digital intervention components were valued and which needed adaptation. Interviews also included a user design component in which researchers and developers screen shared the Enhearten app to model simplified examples of user-generated ideas in real-time. The total session lasted a maximum of 1 hour for each participant. We rapidly incorporated salient input to inform additional adaptation and deploy modifications prior to the Test phase.

The final DDBT phase, Test (23), pilots a fully functional application to evaluate usability and usefulness after adaptations, and measure implementation outcomes. For our study, this provided preliminary evaluation of Enhearten associations with self-stigma and other outcomes and measured acceptability of the adapted intervention. Participants also engaged in qualitative data collection via member checking by a trained research staff member, aimed at understanding the perceived impact and appropriateness of Enhearten adaptations in response to Design/Build phase feedback (28). Member checking increased methodological rigor in the absence of double coding to ensure adaptations were responsive.

Interview questions focused on perceived benefits of Enhearten use, prioritized (and de-prioritized) features, identified barriers and facilitators to app use, and proposed adaptations ( Supplementary Table S2 ). Design/Build questions were adaptive such that feedback on a particular feature (e.g., redundancy of EMA questions, responsiveness of EMI, additional community-building features) prompted additional follow-up questions to gain more detailed understanding and explore optimizations.

During the Test phase, questions began broadly to understand participant experiences in using the app and then focused on ways adaptations were responsive or not responsive to their prior feedback.

Demographics: At baseline, we asked participants for their race/ethnicity, age category, perinatal status (pregnant or within 12-months postpartum), education level, employment, and state of residence ( Table 1 ).

Opioid use and medications for opioid use disorder (MOUD): At baseline, 1-month, and 2-month follow-ups, we assessed frequency of past 30-day opioid use for non-medical purposes and of past 30-day use of MOUD, each with one question on a scale of 0 to 30 days. For those endorsing any past 30-day MOUD use, we asked about types of MOUD used in the past 30 days.

Stigma and self-stigma: At baseline, 1-month, and 2-month, perceived stigma and self-stigma were assessed using the Brief Opioid Stigma Scale (29), measured on a 5-point scale from “strongly disagree” (1) to “strongly agree” (5). Higher scores indicate stronger endorsement of perceived stigma and self-stigma. This measure was selected due to its focus on three key stigma subscales about individuals who use opioids: 1) Aware: stereotype awareness related to perceived public attitudes (e.g., “Most people believe that a person who is addicted to opioids cannot be trusted”), 2) Agree: stereotype agreement related to internalized stigmatizing beliefs (e.g., “I believe that a person who is addicted to opioids cannot be trusted”), and 3) Harm: self-esteem decrement related to the impact of negative stereotypes on self-worth (e.g., “I currently respect myself less because I cannot be trusted due to my addiction to opioids”). Subscales 2 and 3 assessed components of self-stigma, affording the opportunity to understand associations of Enhearten with self-stigma (to demonstrate convergent validity) – versus externalized perceptions of public stigma (to demonstrate discriminant validity).

MOUD perceived behavioral control: At baseline, 1-month, and 2-month, we assessed perceived behavioral control to use MOUD using 6 items adapted from Banks and colleagues (30), measured on a 7-point scale from “not at all” (1) to “very” (7), with higher scores indicating higher perceived self-efficacy to use MOUD.

Mental health outcomes: At baseline, 1-month, and 2-month, we assessed self-reported symptoms of depression and anxiety using the 5-item Mental Health Index of the Short-Form Health Survey (31). Items were measured on a 6-point scale from “all of the time” (1) to “none of the time” (6), and all items were coded or reverse-coded with higher scores indicating better mental health over the past month.

Technology acceptance outcomes: At 1-month and 2-month, we used the perceived ease of use (PEU) index and the perceived usefulness (PU) index of the Technology Acceptance Model (TAM) Questionnaire (32) to assess Enhearten technology acceptance, and the Acceptability of Intervention Measure (AIM) to understand overall endorsement of the intervention (33). PEU, PU, and AIM each included 4 items measured on a 7-point scale from “totally disagree” (1) to “totally agree” (7). They each represent different components of technology acceptance such as “I find Enhearten easy to use” (PEU) and “using Enhearten helps me with my recovery” (PU), as well as intervention acceptability “Enhearten meets my approval” (AIM).

Most participants identified as White (85%), were 31–50 years old (65%), and postpartum (60%). Participants were from geographically diverse areas, with none reporting current use of opioids for non-medical purposes. About half (45%) were currently taking MOUD, with a larger proportion taking a buprenorphine formulation (40%) compared to methadone (5%) ( Table 1 ). Overall, stigma was moderate at baseline (M=2.70; SD=0.47). In particular, the Aware (public stereotype) subscale was high (M=3.96; SD=0.66), the Agree (internalized beliefs) subscale was moderate (M=2.60; SD=0.84), and the Harm (self-esteem decrement) subscale was low (M=1.54; SD=0.74) ( Supplementary Table S1 ).

Several key insights emerged during app onboarding and training, and through early-stage spontaneous feedback from participants (e.g. verbal comments during onboarding or user-generated questions via follow-up communications). These insights were instrumental in guiding and prioritizing the focus of Design/Build activities.

Insight #1: EMA is a novel component, with advantages and drawbacks. Participants were intrigued by the EMAs, and this contributed to their interest and engagement in the study. PPI were excited to learn that EMIs would be provided directly in response to their completed EMAs, noting the benefit and novelty of this personalization. At the same time, the approach was not intuitive, leading to many questions about expectations (e.g., response time following EMA prompt) and settings (e.g., how to tailor available time windows for EMA prompts).

Insight #2: PPIs engaged in constructive feedback upon first interactions with the technology. Participants found it engaging to be part of the prospective process to shape the Enhearten tool. Several comments foreshadowed themes that more formally emerged in the Design/Build phase. For instance, PPIs were eager to connect with other PPIs in private group chats and cited the importance of using this tool to reinforce their recovery progress. One participant expressed desire for a journal feature that interfaces with the app calendar, to organize a space to reflect on their recovery journey over time.

Semi-structured interviews at 1-month follow-up yielded participants’ perspectives on initial use of Enhearten. These insights were organized into five key themes ( Table 2 ):

These salient themes were drivers for multiple subsequent modifications proposed to participants at 1-month follow-up and prioritized for completion.

Participants found updates to the EMA/EMI feature to be highly effective in adding variety and relevance to their experiences using the app. One participant stated “I really like the new questions for the check-ins. They are better suited to do daily because now I can answer differently every day. I can think about what is changing and it gives me space to celebrate the good things that are happening which is important when pregnant or recently having a baby”.

In general, there was an appreciative sentiment for the study’s collaborative design and the value of participants feeling engaged through the process; for example, one participant indicated “applying the feedback from participants to the app is very noticeable–I can tell you’re listening. You can scroll through videos and find what you like, interact with other women, and the questions asked in the app are better. The updates are very reflective of the feedback I’ve shared”.

Technology Acceptance and Intervention Acceptability results indicated TAM indexes (comprising PEU and PU) and AIM index were high at 1-month and increased by 2-month following the iterative feedback and adaptations ( Table 3 ). Given missing data for 4 participants at the 1-month follow-up, paired analyses for these outcomes were reduced (n=16). The TAM scale aggregating PEU and PU of Enhearten was rated highly (on a scale from 1 to 7) at the 1-month follow-up (M=6.12, SD=0.68) and increased further by 2-month (M=6.61, SD=0.60), t(15)=3.211, p=0.006. Similarly, the AIM scale measuring Enhearten acceptability was rated highly (on a scale from 1 to 7) at 1-month (M=6.25, SD=0.97) and increased further by 2-month (M=6.77, SD=0.54), t(15)=2.534, p=0.023.

Member checking interviews at 2-month follow-up validated the changes made in response to participants’ interview feedback at 1-month follow-up.

Participants greatly appreciated the expanded HIPAA-compliant discussion features created in response to interim 1-month feedback, noting, “I like opening up and knowing it is anonymous. I feel alone sometimes, and it’s nice to share what you’re feeling and get others’ feedback without them knowing who I am.” This was a salient view among several participants, with one recalling a particular instance of using Enhearten to offer support and mentorship, sharing, “I remember a time when someone on the app mentioned being in the hospital and how they were treated and I wanted to share my experiences with that person. I remember the time a doctor did not give me a pain medication because I have a history of opioid addiction and wanted to share that experience with others.” Participants were also hopeful that the community within the Enhearten app would continue to grow, stating, “If there were more moms on there, there would be a better chance of having others going through the same type of things as me”.

Qualitative responses on self-stigma and technology acceptance while using Enhearten were reflected in the quantitative data ( Table 3 ).

The Brief Opioid Stigma Scale results from baseline (M=2.70, SD=0.47) to 2-month follow-up (M=2.27, SD=0.61) indicated perceived stigma and self-stigma significantly decreased following the Enhearten intervention, t(19)=-2.902, p=0.009. Post-hoc subscale analyses suggest this change was driven most notably by the decrease in internalized self-stigma on the stereotype Agreement subscale from baseline (M=2.60, SD=0.84) to 2-month (M=1.91, SD=0.78), t(19)=-3.600, p=0.002.

The MOUD Perceived Behavioral Control scale results from baseline (M = 5.21, SD = 1.45) to 2-month (M = 5.91, SD = 1.49) indicated perceived self-efficacy to use MOUD significantly increased following Enhearten exposure, t(19)=2.686, p=0.015.

The Mental Health Index results from baseline (M = 5.20, SD = 0.94) to 2-month (M = 5.36, SD = 1.04) did not indicate a significant change in mental health outcomes after the intervention, t(19)=1.021, p=0.320.