Wiidookaage’win: Beta-test of a Facebook group intervention for Native women to support opioid use recovery

Abstract

Introduction:

The ongoing opioid misuse epidemic has had a marked impact on American Indian/Alaska Native (AI/AN) communities. Culture- and gender-specific barriers to medically assisted recovery from opioid use disorder (OUD) have been identified, exacerbating its impact for AI/AN women. Wiidookaage’win is a community-based participatory research study that aims to develop a culturally tailored, moderated, private Facebook group intervention to support Minnesotan AI/AN women in medically assisted recovery from OUD. The current study assessed the preliminary feasibility and acceptability of the intervention in a beta-test to inform refinements before conducting a pilot randomized controlled trial (RCT).

Methods:

The intervention was beta-tested for 30 days. Moderators were trained prior to delivering the intervention. Study assessments were conducted at baseline and post-intervention. The post-intervention assessments included substance use (self-report and urine drug screen), treatment acceptability, mental health, and spirituality outcomes. We examined intervention engagement patterns using Facebook metrics and qualitatively explored common topics that emerged in participant posts and comments.

Results:

Ten AI/AN women taking medication for OUD (MOUD) were accrued (age range 25–62 years). Participants had been in opioid recovery a mean of 15.2 months (SD = 16.1; range = 3–60). The study participation rate (accrued/eligible) was 91 %. Nine participants completed the post-intervention survey assessment and eight completed a UDS. Acceptability was high based on the mean treatment satisfaction score (M = 4.8, SD = 0.2 out of a possible 5.0), Facebook group engagement, and positive qualitative feedback. All participants retained at post-intervention continued their MOUD treatment, and none had returned to opioid use.

Conclusions:

The beta-test indicated that the Facebook platform and study procedures generally worked as intended and that the intervention was largely acceptable to study participants. The results of this study phase provided valuable insights to inform refinements prior to conducting a pilot RCT to further assess the feasibility, acceptability, and potential efficacy of the intervention.

1. Introduction

Opioid misuse is common and impactful worldwide, with the highest prevalence in the United States (Degenhardt et al., 2019; Strang et al., 2020). Opioid misuse has a marked adverse impact in American Indian/Alaska Native (AI/AN) communities (Rieckmann et al., 2012; Tipps et al., 2018), with overall drug overdose mortality rates in the U.S. highest for individuals of this racial/ethnic group (Spencer et al., 2022). Data from 2021 in the state of Minnesota indicated that AI/AN individuals were ten times more likely to experience death by drug overdose than white individuals (Minnesota Department of Health, 2021a). These inequities are likely linked to a variety of social determinants of health including intergenerational trauma, systemic racism, and structural barriers to accessing healthcare services (Minnesota Department of Health, 2021b; Venner et al., 2018).

Medications for opioid use disorder (MOUD), including methadone, buprenorphine, and naltrexone are evidence-based yet underutilized (Mauro et al., 2022; Substance Abuse and Mental Health Services Administration, 2022). Estimates indicate that the majority of substance use treatment facilities serving AI/AN populations do not offer MOUD, and the majority of AI/AN individuals admitted to specialty OUD treatment do not receive MOUD (Krawczyk et al., 2021). Barriers to MOUD access and use in AI/AN communities may include structural and organizational barriers such as transportation or trained provider shortages, as well as community and individual factors such as stigma or lack of integration of traditional culture into treatment (Venner et al., 2018). Broadly, treatment retention remains a challenge even for those who access MOUD (Morgan et al., 2018; Timko et al., 2016). Gender-specific challenges to MOUD access and retention, including longer wait times to enter treatment, family conflict, and childcare (Amaro et al., 2023), may exacerbate these barriers for AI/AN women. However, prior opioid research has not focused specifically on this population.

Novel behavioral strategies (Sofuoglu et al., 2019) including those that leverage social support systems (Brigham et al., 2014; Reif et al., 2014) as adjuncts to MOUD treatment are needed, especially for AI/AN women. Alongside the aforementioned MOUD barriers for AI/AN people, prior studies indicate that women enter MOUD treatment with higher rates of post-traumatic stress disorder, depressive symptoms, and perceived social isolation than men (Becker et al., 2017; Di Paola et al., 2022; McHugh et al., 2013), suggesting a need for gender-specific psychosocial support. To develop and implement these supports for AI/AN women, it is important that researchers use a culturally centered, community engaged, and holistic approach. One recent systematic review noted distrust and concern surrounding MOUD treatment in rural AI/AN communities and a need for culturally centered and holistic approaches to OUD support for these groups (Mpofu et al., 2021). Another review highlighted the need for culturally tailored, opioid-specific treatment interventions for Indigenous people and for more culturally-tailored substance use programs in urban Indigenous communities (Richer & Roddy, 2023). Thus, research studies that utilize culturally grounded methods to develop behavioral interventions are needed to address opioid recovery inequities in these communities.

1.1. The Wiidookaage’win study

The Wiidookaage’win study uses the community-based participatory research (CBPR) approach in alignment with previous substance use intervention research conducted in partnership with AI/AN populations (Richer & Roddy, 2023). The CBPR approach ensures that AI/AN storytelling and wisdom are incorporated into the study methodology and intervention—from conceptualization to implementation—to maximize cultural fit, acceptability, and accessibility. In 2021, the lead investigator presented the results of a CBPR project to develop a culturally tailored Facebook smoking cessation intervention (i.e., CAN Quit) for Alaska Native adults (Patten et al., 2023; Sinicrope et al., 2022) to the Minnesota Indian Women’s Resource Center (MIWRC). In response, the MIWRC expressed a need to address opioid overdose inequities and develop a similar Facebook intervention for AI/AN women. The MIWRC co-conceptualized the study and continues to assist with study activities.

Our study team additionally partnered with a study-specific Community Advisory Committee (CAC) and the Healthy Nations Advisory Board (HNAB) of Tribal health leaders in Minnesota to develop the present study. The CAC is comprised of AI/AN women with lived experience, health care providers, Native Elders, and others serving AI/AN people in recovery from OUD (Coyhis & White, 2002). Each CAC member was paid an honorarium of $150 per study meeting attended. The study name, Wiidookaage’win, which means “the place for help, the time of helping” in Ojibwe, was suggested by a Native Elder woman on the CAC and accepted unanimously by the CAC.

Wiidookaage’win aimed to develop and test a culturally tailored, private Facebook-delivered group intervention using evidence-based treatment principles to support AI/AN women in recovery from OUD and currently taking MOUD. Social media has been suggested as a potential avenue to support substance use recovery (Ashford et al., 2019; D’Agostino et al., 2017), and Facebook was selected following its utilization in the CAN Quit study and MIWRC’s positive feedback. Of note, Facebook is the second most widely used social media platform (behind YouTube) by U.S. adults (Gottfried, 2024). We opted to include AI/AN women in recovery using MOUD for multiple reasons: 1) the Facebook intervention is intended as an adjunct rather than a substitute for evidence-based MOUD; 2) we wanted to ensure that participants had access to treatment that would support them to engage fully with the group; 3) we desired to minimize potential adverse risks (e.g., overdose); and 4) the CAC and HNAB felt that providing the intervention as a treatment adjunct for opioid recovery would reinforce AI/AN ways of knowing and cultural values related to community resilience (Coyhis & White, 2002; Mohatt et al., 1988; Teufel-Shone et al., 2018; Walters & Simoni, 2002). To our knowledge, this is the first study aimed at developing a culturally tailored social media intervention for AI/AN women in recovery from OUD.

In a previous study phase, we co-developed study methods and intervention content with the CAC and conducted qualitative interviews with AI/AN women in recovery and interested parties (e.g., healthcare providers, clinicians, and social workers) to guide our intervention content refinement (Young et al., 2024). Herein, we report the results of the beta-test of the Wiidookaage’win Facebook group intervention, which aimed to explore the preliminary feasibility and acceptability of the intervention and inform refinements prior to conducting a pilot randomized controlled trial (RCT).

2. Methods

The study is supported by the National Drug Abuse Treatment Clinical Trials Network (CTN-0123) supported by the National Institute on Drug Abuse (NIDA-UG1DA040316) and was approved by the Mayo Clinic Institutional Review Board (IRB 22–000477). The study is registered with the Clinical Trials Registry (NCT05340855). Participant enrollment and data collection occurred between February 2023 and May 2023. We used Borek et al.’s checklist to improve reporting of group-based behavior-change interventions to guide our reporting of the intervention (Borek et al., 2015).

2.1. Study setting

In Minnesota, AI/AN people comprise 2.8% of the state population (America Counts Staff, 2021). This population includes four Dakota communities and seven Anishinaabe reservations. Study activities and enrollment were focused in the urban metropolitan area of Minneapolis, Minnesota, but participants did not have to reside in Minneapolis to be eligible for the study. Of note, urban AI/AN communities are often intertribal, thus their identities may differ from those residing in reservation-based communities (James et al., 2018; Yuan et al., 2014) and could include members of any of the 574 federally recognized tribes across the United States.

2.2. Participants

2.2.1. Sample size

Our accrual target was 10 AI/AN women, consistent with sample size recommendations for Stage 1 behavioral therapies addictions treatment development (Rounsaville et al., 2001), beta-testing of novel digital health interventions (Baker et al., 2014; Nielsen, 2012), and the total number of group members included in previous studies of Facebook group interventions (Ramo et al., 2015; Sinicrope et al., 2022).

2.2.2. Recruitment

Recruitment was facilitated by addiction treatment and community clinic staff and clinicians within Minneapolis and St. Paul, MN that serve AI/AN people in recovery from OUD across the state. Research team members facilitated multiple in-person informational meetings with these staff and clinicians to distribute recruitment materials, including a study business card, one-page information sheet co-developed with the CAC, and a flyer adapted from the information sheet. Flyers were also posted in local cultural centers and coffeehouses. Additionally, participants from the earlier study phase and screened individuals for the beta-test were encouraged to tell others they know about the study opportunity. Recruitment efforts spanned 10 weeks before we reached our accrual target.

2.2.3. Eligibility

Individuals needed to: self-report AI/AN race (i.e., “Do you identify as American Indian or Alaska Native?”); identify as a woman; reside in Minnesota; be age ≥18; meet criteria for OUD based on the DSM-5 Checklist (American Psychiatric Association, 2013); currently use MOUD; be comfortable speaking and reading in English; have consistent access to broadband Internet on a mobile phone, computer, or tablet; have or be willing to create a Facebook account; be willing to participate in the Facebook intervention for one month; self-report no current suicidality; self-report at least 30 days of opioid abstinence (aside from MOUD and prescribed opioids); screen negative for opioids (aside from disclosed prescriptions) on a Urine Drug Screen (UDS); and be willing to travel to a community clinic in Minneapolis, MN for UDS collection.

2.3. Screening, consent, and enrollment procedures

2.3.1. Screening

Individuals interested in study participation contacted the study team and completed screening questions over the phone, which included all eligibility criteria except for the UDS and suicidality. If the potential participant self-reported medical opioid use besides their MOUD, they were asked to present their prescription at the in-person screening visit.

Individuals provided written informed consent either remotely via DocuSign or during the in-person screening, which took place at the Indian Health Board in Minneapolis, MN. First, participants were administered the last three items of the Concise Health Risk Tracking-Self Report (CHRT-SR) Suicidal Behavior Evaluation (Trivedi et al., 2011), which assessed thoughts of suicide over the past 24 h; individuals deemed ineligible were referred to a study clinician for further evaluation. Potential participants also completed a UDS using a temperature-controlled urine test cup (CLIAwaived^™ Inc., 2017c) in a private bathroom without the study staff present. All individuals who completed the UDS received a $50 cash card. Individuals were only excluded if they tested positive for opioids without confirming their medical use. Ineligible individuals were offered OUD recovery resources.

2.3.2. Enrollment

After completing baseline assessments, eligible participants were emailed an invite link to the Wiidookaage’win Facebook group. Accrual was defined as accepting the invitation. Participants who did not already have a Facebook account were given instructions on how to register. Accrued participants were provided with a smudge kit and a journal to be used throughout the Facebook intervention. Smudging is a cultural space clearing ritual, in which individuals burn herbs and rub the smoke over their bodies to infuse themselves with positive energy or clear negative energies (Indian Health Service). The smudge kit for this study included sage, cedar, and sweetgrass inside an abalone shell. Additionally, participants were emailed an online link to a community resources landing page developed with the CAC and community partners. Resources included local substance use treatment programs, domestic violence and houseless shelters, family services, mental/behavioral health clinics, and local cultural activities. Participants were informed that the Facebook group would begin when all 10 participants were accrued.

2.4. Facebook intervention

The Facebook intervention prototype was guided by recommendations for social media health communications (Centers for Disease Control and Prevention (CDC), 2016). Facebook group content drew from behavioral interventions that have been implemented for substance use recovery and stress management, including Mindfulness-based Relapse Prevention (MBRP) (Bowen et al., 2021), the Stress Management and Resiliency Training (SMART) program (Sood, 2019), and Motivational Interviewing (MI) (Miller & Rollnick, 2013). The study used a CBPR approach to adapt these interventions and co-develop culturally relevant content (including artwork and messaging), as described previously (Young et al., 2024). Broadly, content was developed and iteratively adapted based on CAC and qualitative interview feedback from an earlier phase of the study to integrate and center AI/AN art, nature, and cultural values and practices. As one example, a CAC member with the Native American Community Clinic consulted the clinic’s Spiritual Care providers to develop an introductory post to smudging and the medicinal function of each herb.

Although we developed content for a three-month Facebook intervention (Young et al., 2024), the beta-test lasted 30 days for a preliminary assessment of feasibility and acceptability to facilitate any needed refinements prior to the subsequent RCT. The group was private, hidden, and moderated by study staff to facilitate confidentiality. It was communicated to participants that they were expected to “actively participate” in the group. Participation in the intervention was entirely virtual. There were no live sessions, and participants could log into Facebook to interact with intervention content at any time.

2.4.1. Facebook moderators

The primary intervention moderators were two women on the Mayo Clinic study staff—one American Indian and one biracial—with another woman on the study team acting as a standby moderator. All moderators had educational backgrounds in public health or psychology. In addition, a Native Elder woman who was recommended by the CAC and had prior experience moderating Facebook groups for AI/AN women with OUD was available in a consultant role.

2.4.2. Moderator exchanges

The success of a social media group for health behavior change is dependent upon the skills and confidence of moderators to promote engagement and be proactive to monitor and manage inappropriate posts or misinformation shared among members (Sinicrope et al., 2022; Young, 2013). Moderators were trained in a series of five discussions (approximately 23 h total) which included hands-on practice and didactic instruction. These were facilitated by members of the research team and other experts in OUD treatment, peer recovery support for AI/AN women, AI/AN culture, MBRP, SMART, communication skills in health behavior change, MI, and online community management. We opted to call these moderator exchanges to emphasize the bi-directional sharing of knowledge and questions between moderators and facilitators. The aim of these exchanges was to equip moderators with the skills and confidence necessary to promote positive health outcomes, engagement, and discussion among AI/AN women recovering from OUD in an online social media group intervention. See Supplemental Table 1 for more information on the content of exchanges.

2.4.3. Moderator guidelines and content library

With our CAC and study team, we co-created a Wiidookaage’win group guide, including moderator guidelines and a content library. Moderator guidelines addressed general moderator principles, approaches to welcoming participants and establishing group rules, and strategies for handling challenging conversations, irrelevant posts or comments, or misinformation. Further, based on insights and feedback from CBPR processes (Young et al., 2024) and moderator exchanges, the study team made final refinements to the intervention content and formed a tentative posting schedule to organize the content library.

Content library posts (i.e., original additions to the Facebook feed) typically included 1–2 short paragraphs of text introducing MBRP or SMART exercises, AI/AN cultural practices (e.g., smudging and giving thanks), AI/AN values connected to local Minnesota tribes (e.g., Ojibwe, Dakota), personal stories of OUD recovery from AI/AN women, poetry by AI/AN women, or discussions of resilience in the context of historical trauma. These posts were typically accompanied by a form of media (e. g., videos, music, photographs, artwork) intended to be culturally relevant and/or elicit positive affect, following our qualitative feedback (Young et al., 2024). We also developed moderator prompts (e.g., “What do you already do to practice self-care?”, “Which Ojibwe value relates most to your recovery journey?”) based on MI techniques to promote assets-based language and elicit health behavior change talk (Trivedi et al., 2011) related to the post’s topic. Additionally, the content library included guidelines for sharing our resources link in the initial group description and/or when participants requested support. During the beta-test, the frequency, order, and language of posts was adapted based on participant engagement and needs.

2.4.4. Group facilitation

Moderators posted one selection from the intervention content library every one to two days. They copied and pasted from the library with minimal formatting changes for Facebook to ensure the posts were delivered as designed. To encourage participant posts and comments, the moderators often posted one or two prompt questions as a reply to the post, tagged (@mentioned) specific participants, and used MI techniques (e.g., open-ended questions, reflections) when replying to participants. For example, if a participant shared a challenging time in their life, moderators might reply with a reflection to emphasize the participant’s experience (e.g., “That was a very difficult time for you.” instead of “I’m sorry to hear that.”). Moderators checked the group three to four times daily to interact with the participants, encourage participation, and, if necessary, address inappropriate posts or comments and misinformation as needed. Additionally, the primary moderators held one Zoom meeting with an expert in online community management at Mayo Clinic mid-way through the beta-test to review moderator posts and comments and suggest improvements for group facilitation.

2.5. Assessments

Study assessments were completed in person at baseline and post-intervention. Participants received cash cards as compensation for their time. In addition to the $50 provided for their screening UDS, they received $25 for completing the baseline assessments, $50 for the post-intervention assessment, and $50 for completing a post-intervention UDS.

2.5.1. Study recruitment and retention

Recruitment feasibility measures included the number of potential participants screened, the number eligible, and the number enrolled out of those eligible. Study retention was defined as the proportion of participants completing the post-intervention assessments.

2.5.2. Baseline sociodemographic characteristics

Sociodemographic measures selected from the PhenX Toolkit assessed age, ethnicity, race, education, employment status, and marital status (Hamilton et al., 2011). Salience of AI/AN race was assessed with the question: “By choosing any number between zero and ten, how important is being Native to your overall identity?” (Resnicow et al., 2009) This was scored using an 11-point Likert scale (0 = not at all important to 10 = very important).

2.5.3. Substance use

The Timeline Follow-Back (TLFB) interview (Sobell & Sobell, 1992) was used to assess self-reported opioid use in the past 30 days at phone screening and post-intervention. Study staff completed a UDS with a 12-Panel Instant Drug Test Card (IDTC) (CLIAwaived^™ Inc., 2017b), including panels for oxycodone, morphine, buprenorphine, and methadone metabolite, and a separate IDTC for fentanyl (CLIAwaived^™ Inc., 2017a), following the manufacturer’s recommended procedures at screening and post-intervention.

To assess other substance use at baseline and post-intervention, we utilized an item from the PhenX Toolkit (Indian Health Service) that assesses substance use across seven major categories including stimulants (e.g., methamphetamine, Adderall), cocaine/crack, marijuana, sedatives/tranquilizers, club drugs, hallucinogens, and inhalants/solvents. For each substance endorsed, we used the TLFB interview to assess the days of use over the past 30 days. The 12-Panel IDTC confirmed endorsement of each substance except for inhalants/solvents.

2.5.4. OUD treatment use

Participants self-reported MOUD use at screening. At post-intervention, MOUD retention was assessed using the TLFB to determine MOUD use over the past 30 days and current MOUD use. The TLFB was also used at baseline and post-intervention to assess whether additional OUD treatments (e.g., inpatient or outpatient services, individual or group support services, Narcotics Anonymous) were used within the past 30 days.

2.5.5. Spiritual quality of life

Moderator exchanges addressed the importance of spirituality in AI/AN culture and daily life. Thus, at baseline and post-intervention, we included five items from the World Health Organization Quality of Life — Spirituality, Religious, and Personal Beliefs scale to characterize spirituality in our current sample. The WHOQOL SRPB was validated in a cross-cultural study (WHOQOL SRPB Group, 2006). The five items included in the current study were found to have the strongest factor loading in each facet of the WHOQOL SRPB (connection to a spiritual being, awe and wonder, spiritual strength, peace, and faith) in a previous study (Skevington et al., 2013). Participants responded to each using a five-point Likert scale (0 = not at all to 4 = an extreme amount). Items were summed into a total score that can range from 0 to 20.

2.5.6. Emotional health

Given that OUD and mental health disorders are often comorbid (Santo et al., 2022), the Kessler Psychological Distress Scale (K6) (Kessler et al., 2002) was used to characterize emotional health and monitor psychiatric symptoms at baseline and post-intervention. The K6 was validated among AI/AN adults specifically (Mitchell & Beals, 2011). Each item is scored on a five-point Likert scale (0 = none of the time to 4 = all of the time). Total scores can range from 0 to 24, with a recommended cutoff score of 13 to indicate the presence of a psychiatric disorder (Kessler et al., 2003). If a participant scored ≥13, a licensed clinical study staff member would follow up with the participant for further assessment and initiate a referral for additional mental health assessment and care (Kessler et al., 2003).

2.5.7. Intervention uptake and engagement

At posttest, we used objective, standard engagement metrics (Pagoto et al., 2016) of Facebook views, reactions (e.g., likes, loves), posts (i.e., original additions to the Facebook feed), comments (i.e., replies to a post), and votes on polls to calculate the total number of times participants engaged with the intervention. Participants could select multiple options in polls. In addition to calculating a total engagement count, we also examined engagement metrics individually, as different metrics may indicate different levels of engagement (e.g., creating a comment may require greater engagement than “liking” a post).

2.5.8. Treatment satisfaction

At post-intervention, we assessed satisfaction with the Facebook intervention using 13 items from the Usefulness, Satisfaction, and Ease of Use (USE) questionnaire (Lund, 2001). The measure’s total score is calculated by taking the mean of all 13 items, with possible scores ranging from 1 (low) to 5 (high).

After the USE items, participants were asked how important it would be for future similar interventions to address spirituality on a scale of 0–10 and if they would like to be a peer moderator for a future iteration of the intervention (yes or no). We asked four open-ended questions assessing the participant’s experiences with the intervention, any technical difficulties encountered, recommendations for improvement, and the potential for the intervention to encourage individuals to seek OUD treatment for the first time.

2.5.9. Adverse events

Adverse events were assessed with open-ended questions inquiring about Emergency Department visits, overdoses, or hospitalizations. The consent form encouraged participants to self-report adverse events at any point in the study.

2.6. Statistical methods

2.6.1. Quantitative analyses

We used descriptive statistics to summarize the number of potential participants screened, the number eligible, the reasons for ineligibility, the number enrolled out of those eligible (i.e., participation rate), baseline characteristics, intervention satisfaction, mental health, spiritual quality of life, intervention engagement, and adverse events.

Descriptive statistics were used to summarize return to opioid misuse and MOUD retention. Return to opioid misuse was primarily operationalized as a positive opioid UDS without confirmation of medical use or self-report of any opioid misuse in the past 30 days. Secondarily, we also summarized return to opioid misuse as at least seven consecutive days of self-reported non-medical opioid use in the past 30 days (Williams et al., 2018). We summarized missing UDS outcomes in two different ways in order to explore a range of possible outcomes missing data may reflect: 1) defining a missed UDS as “not positive” (so as not to stigmatize a marginalized population when there are numerous possible reasons for a missed follow-up) (King et al., 2020) and 2) defining a missed UDS as “positive” (to examine the most conservative outcome reflected in the data). MOUD retention was operationalized as the proportion of days a participant reported MOUD use over the past 30 days and current MOUD use at follow-up. Other OUD treatment engagement was operationalized as engagement with other OUD treatment supports over the past 30 days.

2.6.2. Qualitative analyses

Participant-generated Facebook posts, comments, and answers to the post-intervention assessment open-ended questions were entered into Microsoft Excel spreadsheets to compare side-by-side. Content analysis (Krippendorff, 2018) supplemented with QSR NVivo software version 10 (Doncaster, Victoria, Australia) was used for a study team member to generate response themes and frequency of topics discussed during the intervention.

2.7. CAC feedback

After the beta-test was complete, the study team held a Zoom meeting to review quantitative and qualitative findings with the CAC. Qualitative themes and topics were reviewed with the CAC for interpretation. The CAC shared their thoughts on the potential sustainability based on their experiences and feedback from participants.

3. Results

3.1. Participants

A total of 18 individuals were screened by phone, and 17 were potentially eligible, of which 15 completed written informed consent. Of the 15, 14 completed an in-person screening visit. Three of the 14 women were no longer eligible due to a positive UDS for fentanyl (n = 2) or morphine (n = 1). All 11 eligible women completed the baseline survey, of which 10 were accrued, for a study participation rate of 91 %. The remaining eligible individual was lost to follow-up after her in-person visit. Nine out of 10 participants completed post-intervention surveys by phone or in-person, indicating 90 % study retention. A total of two participants could not be reached to complete the post-intervention UDS.

The participant baseline characteristics are described in Table 1. Participants ranged in age from 25 to 62 years, 80 % were single, and 30 % were employed. Mean reported abstinence from opioid misuse was 15.2 months (SD = 16.1; median = 9; range = 3–60), 70 % were receiving additional treatments for OUD (e.g., inpatient or outpatient services, individual or group support services, Narcotics Anonymous), and 30 % reported other current substance use. All but one of the 10 participants had an existing Facebook account at study enrollment. The mean duration between participants accepting the Facebook invitation and the initiation of the Facebook group was 11.1 days (SD = 7.2; range = 0–20).

Table 1.

Baseline characteristics of Wiidookaage’win American Indian women participants (N = 10).

Characteristic	# (%)
Age in years
Mean ± SD	42.2 ± 12.82
Median	35
Range	25–62
Additional race(s)
No	8 (80)
Yes	2 (20)
Hispanic or Latino ethnicity	0 (0)
Marital status
Single, never married	8 (80)
Married/partnered	2 (20)
Occupational status
Unemployed/keeping house/disabled	7 (70)
Working now	3 (30)
Education level
Some college or higher	4 (40)
High school diploma/GED	3 (30)
Some high school	3 (30)
# children in household
Median	1
Range	0–4
Importance of Native identity scorea
Mean ± SD	9.9 ± 0.3
Median	10
Range	9–10
WHOQOL SRPB scale scoreb
Mean ± SD	15.9 ± 2.34
Median	15
Range	18–25
K6 psychological distress scorec
Mean ± SD	5.9 ± 3.56
Median	5
Range	1–12
Months in recovery from opioid misuse
Mean ± SD	15.2 ± 16.14
Median	9
Range	3–60
Prior method(s) of opioid use
Injections	6 (60)
Pills/lozenge	6 (60)
Snorted	3 (30)
Current OUD treatment (in addition to MOUD)	7 (70)
Current other substance use	3 (30)d

^aSingle item. Possible score range 0 to 10. Higher score indicates greater importance of Native race to identity.

^bWorld Health Organization’s Quality of Life Spirituality, Religion, and Personal Beliefs scale. Five items. Possible total score range 0 to 20. Higher score indicates greater perceived impact of spirituality on quality of life.

^cMeasures psychological distress including depressive and anxiety symptoms over the past month. Six Items. Possible total score range 0–24; score of ≥13 indicates presence of a psychiatric disorder (no participants met this cutoff score).

^dAll 3 reported marijuana use.

3.2. Intervention uptake

Forty-seven posts were published to the group, of which 29 (61.7 %) were generated by moderators following the content library. Participants generated the remaining 18 (38.3 %) posts; one participant generated 15 of these 18 posts. The 18 posts were primarily text (94.4 %) and one (5.6 %) was a text/image post. The text posts most commonly consisted of morning greetings and prayers or questions about the study procedures. The text/image post was a phone screenshot of a white wallpaper with the participant reporting that she smudged and took her dose of MOUD. Two moderator-posted videos were auto-removed for violating Facebook’s Community Standards for “Spam.” One included a Tlingit woman talking about how reconnection to AI/AN culture facilitated her recovery from OUD (Centers for Disease Control and Prevention (CDC), 2020); the other included women participating in an Algonquian water song along with spoken word about how AI/AN women are keepers of the water (Within Us Productions LLC, 2018). These posts were intended to foster hope that recovery and reconnection to AI/AN culture are possible by featuring other AI/AN women with similar stories (Young et al., 2024). After these videos were auto-removed, 27 (60 %) moderator and 18 (40 %) participant posts remained for analyses.

3.2.1. Participant engagement

Overall, participants engaged (viewed, reacted, posted, commented, or voted) a median of 33 times (range = 12–98). Participants viewed a median of 27 of the 45 posts (60 %, range = 11–43), commented a median 5 times (range = 0–39), and reacted a median 4 times (range = 0–14). Across three poll posts, participants voted a median of 4 times (range = 1–12) and frequently chose more than one option. Nine (90 %) participants commented or reacted at least once. Three (30 %) participants created at least one post. Four (40 %) voted in polls at least once.

3.2.2. Engagement patterns

Day 29 of the beta-test, which was a Wednesday, was the most active day of the intervention, with 11 individuals (including moderators) engaging in some way. Days 5 and 18 fell on a weekend and were the least active days, with only 2 individuals engaging (including moderators). Engagement trended higher within the last 10 days of the beta-test.

3.2.3. Post engagement

Moderator and participant posts elicited similar engagement (see Table 2). Posts that had higher median engagement totals featured moderator-generated questions about participants’ experiences managing their recovery from OUD, smudging, and giving thanks to other beings. Posts that had lower median engagement totals included moderator-posted MBRP and SMART exercises (e.g., urge surfing, grounding, meditation videos) or featured videos of OUD recovery and resilience stories from AI/AN women. Though moderators often tagged participants to these posts or replied with follow-up prompting questions, participants did not respond.

Table 2.

Participant engagement metrics on Facebook intervention posts and comments.

Engagement metric	Moderator-generated posts and comments	Participant-generated posts and comments
	median, range	median, range
Viewsa	5, 2–9	5, 3–8
Reactionsb	1, 0–4	1, 0–4
Commentsc	2, 0–13	3, 0–11
Votes on pollsd	4, 1–12	N/A
# of times engaged	9, 3–21	8, 4–18

^aCan only be seen on posts.

^bCan react on posts, comments, and replies to other comments.

^cIncludes comments under posts and replies to other comments.

^dPolls can only be made as posts.

3.2.4. Qualitative analysis of participant-generated content

When analyzing the content of comments and posts, participants most frequently discussed how they maintain their recovery, including Native crafts such as beading and practices such as smudging, praying, and giving thanks. Participants also frequently discussed topics related to interconnectedness (e.g., “community,” “family,” “encouragement”) and change (e.g., “learn,” _“clean,” _“better”). See Fig. 1 for other frequently mentioned words and topics.

Fig. 1.

Most frequent words used in participant-generated Facebook posts and comments.

3.3. Treatment satisfaction

The mean total USE score was 4.84 ± 0.18 (range = 4.38–5.0) out of 5.0, indicating high intervention acceptability. All nine participants who completed the post-intervention assessment expressed interest in being a peer moderator for a future iteration of the intervention. Qualitative analysis of post-intervention, open-ended participant feedback revealed the following topics and themes:

3.3.1. Technical problems

Although Facebook auto-removed two moderator posts, few technical issues were reported by participants. One participant reported issues with their own account and could not engage with the intervention until Day 23. Another participant was not sure whether others were not commenting or if their account was having technical problems. Another participant was not sure how to prioritize the intervention posts so they all appeared on her Facebook feed.

3.3.2. Post recommendations

Participants recommended additional videos such as TED Talks, personal stories, and Facebook Live sessions with participants. Participants appreciated seeing different thought-provoking questions and forms of media each day and “felt acknowledged” by seeing recovery stories from other participants. One participant noted that it is important for the moderators to post every day.

3.3.3. Potential for intervention to encourage initiation of OUD treatment

Participants thought that the group could inspire those who are nervous to pursue treatment because seeing others’ personal stories would give them hope and make it seem more plausible that they can start their recovery. One participant noted that involving a bigger group of AI/AN women in recovery and emphasizing the confidentiality of the intervention would be even more helpful to encourage such participants. This feedback was in line with our research team’s preliminary discussions surrounding the future potential for the Facebook group in supporting initiation of OUD treatment.

3.3.4. Overall recommendations

Participants wished that the group lasted longer than 30 days and that it had more participants. They noted that they would have more feedback on recommendations if they had more time to utilize the group. They again highlighted the importance of emphasizing the confidentiality of the group as well as keeping it accessible to only AI/AN women.

3.4. Substance use

All nine participants who completed the post-intervention survey assessment maintained their MOUD use. None of the retained participants returned to opioid use, as evidenced by their self-report (and negative UDS results among those who completed it). When a missing UDS was defined as positive, 20 % of all participants were assumed as returning to opioid misuse, and when a missing UDS was defined as negative, 0 % of participants were assumed as returning to opioid misuse. No participants reported using other substances during the past 30 days, which was confirmed by UDS for those who completed it. Eight out of 9 (89 %) retained participants reported using other OUD treatment supports during the intervention, most commonly group or individual recovery services.

3.5. Spirituality

The mean WHOQOL-SRPB total score was 16.89 (SD = 2.60; median = 18; range = 13–20) out of 20 at post-intervention, compared to 15.9 (SD = 2.34; median = 15; range = 18–25) at baseline. Participants rated the importance for future iterations of the group to address spirituality a mean of 9.25 (SD = 1.39; range 6–10) out of 10.

3.6. Mental health

The mean K6 total score was 4.13 (SD = 3.33; median = 3; range 0–9) out of 30 at post-intervention, compared to 5.9 (SD = 3.56; median = 5; range 1–12) at baseline. Thus 0 % of participants scored above the clinical cut-off of 13 points.

3.7. Adverse events

No adverse events were reported that resulted from intervention participation.

3.8. CAC feedback

CAC members shared that the findings are relevant and impactful for the AI/AN community in the metropolitan area of Minneapolis, Minnesota. They thought it was “very impressive” that every participant expressed interest in being a peer moderator and that this spoke to the acceptability of the intervention. They were intrigued by the lack of participation of individuals younger than age 25 and speculated that age and the types of social media commonly used by different age groups may have impacted interest in the study.

CAC members believed that the intervention could be an effective tool for AI/AN individuals who desire continued social support as an adjunct to their clinic or treatment program. Additionally, they reinforced the themes and common topics identified in our analysis, especially in regard to interconnectedness from group members sharing their recovery stories. They thought the group could help to improve women’s self-esteem by the realization that others in the group care about their recovery. They also responded positively to the high Facebook group engagement observed for posts on Native traditions and noted that reconnection to culture is an important component of recovery for AI/AN women.

3.9. Moderator lessons learned

Intervention moderators gained valuable experience and reflected on lessons learned to improve the Facebook intervention in future iterations. See Table 3 for moderator lessons learned.

Table 3.

Lessons learned by Wiidookaage’win moderators.

Question (Sinicrope et al., 2022)	Moderator Feedback	Description
What were the most challenging parts of moderating the group?	Knowing how to respond to certain posts/comments	• Understanding the type of support participants may be looking for when they share vulnerable experiences (especially since we are not peer moderators and may not share similar experiences)
		• Responding to seemingly incomplete posts by participants
	Knowing when to encourage participation	• Difficult to know if someone might benefit from a followup question, prompt, or tag, or if they simply did not feel like responding to a certain post
		• Challenging not to respond immediately; sometimes unsure whether to wait and see if another participant responds first
		• Challenging to know whether, how, and when to check in on how the participant may be processing and engaging with the posts if they do not engage with a reaction, comment, or post
	Ensuring consistent moderator access to group	• One of the moderators was briefly blocked/suspended by Facebook after posting one of two videos that were removed by Facebook because they were classified as “Spam”
What advice would you give to someone who is about to moderate a group like this?	Make it a team effort	• Involve a consultant experienced in community management and moderating online groups
		• Consistent communication between moderators
		• Ensure that multiple moderators are available for responding (e.g., weekend rotation, when one moderator is blocked by Facebook, due to schedules, etc.)
		• Meet with community partners to understand community and global cultural events
	Be genuinely interested in fostering trust and connection	• Helpful to have connection with participants and/or a Native background; as a Native woman, CS was able to understand and connect with cultural values and references (e.g., humor, spiritual beliefs, language)
		• Having multiple Native women moderators may help participants to feel even more comfortable and connected
		• Be transparent about private/closed nature of the group, who is in the group, and confidentiality
		• Keep study team access to the Facebook group limited to moderators to build and maintain trust
		• Learn and keep track of participant interests expressed in previous comments/posts to help foster connections
		• If a participant shares something vulnerable or difficult, it’s important to directly reach out and ask them about the support they need rather than saying something more passive such as “let us know if you need anything” or “thanks for sharing”
	Be aware and respond to events happening in the community	• It’s important to be aware of upsetting community or global events and to consider when and how to acknowledge these within the group
		• Bring attention to local events that bring AI/AN people together
	Make the virtual platform as personable as possible	• Create a personable introduction (e.g., video or Facebook Live event) to help with comfort level and building connection
		• May be helpful to have moderators conduct in-person screening visits
		• It’s okay to return to older/ previous posts to try to engage or re-engage participants (e.g., tagging participants who are just starting to be more comfortable or engaged on previous posts to ask about their reactions)
What did you think you did best, and what do you think you need to work on?	What worked best	• Motivational interviewing techniques such as open-ended questions and reflections worked well to improve engagement
		• Variety of content and media
		• Being active when members where active and replying in a timely manner; never left a participant or post abandoned for an extended period
		• We private messaged individual participants to check in if they were less engaged in the group; this helped to re-engage folks
		• Posted according to tentative schedule while also being flexible and adjusting based on group engagement
		• Moderators consistently communicated with one another about the group
	What needs to improve	• Balancing wanting to make it an active community while also respecting if participants want to be less overtly active
		• Foresee the possibility of technical difficulties and encourage participants to contact us if they cannot access the group
		• Centering the intervention posts around personal experiences and knowledgesharing instead of completing tasks and watching videos

4. Discussion

In collaboration with community partners, we developed the Wiidookaage’win study, a CBPR project aimed at developing a novel, culturally tailored group Facebook intervention to support opioid recovery in AI/AN women. Though the current beta-test included only 10 participants, it provided valuable lessons learned to inform refinements prior to conducting a pilot RCT. The beta-test indicated that the intervention was generally feasible, acceptable to participants, and worked as intended. Additionally, all of the retained participants abstained from opioid misuse and maintained MOUD use post-intervention, a promising trend that will be explored further in the upcoming pilot RCT.

Recruitment and retention processes were shown to be feasible, with a 91 % study participation rate and 80 % of participants completing all procedures. The collaboration and support of the project by the Indian Health Board in Minneapolis, MN was key in this regard, as their staff allowed our team to conduct in-person procedures in a location familiar and accessible to the local AI/AN community. We also learned that it was helpful to invite participants to accept the invitation into the Facebook group while in-person (using an iPad borrowed from Mayo Clinic if necessary) to improve the likelihood of accrual. Consistent with national survey data demonstrating that Facebook is commonly used by U.S. adults (Gottfried, 2024), 90 % of accrued participants already had a Facebook account, indicating that this is likely a practical platform for the intervention. However, none of our participants were under 25 years of age. Alternative social media platforms (e.g., Instagram, TikTok, Snapchat) may be of greater interest to younger adults (Gottfried, 2024).

Ninety percent of participants commented or reacted at least once during the intervention, which is an encouraging level of engagement based on previous social media group intervention studies (Cheung et al., 2015; Pechmann et al., 2017; Ramo et al., 2015; Sinicrope et al., 2022). This finding also aligns with previous research documenting that AI/AN individuals recovering from substance use will utilize Facebook for social support (Patten et al., 2023; Sinicrope et al., 2022). Importantly, participant engagement was trending higher during the final 10 days of the beta-test, which aligns with previous literature theorizing that as a community develops, an increased proportion of the activity will be generated by community members (Iriberri & Leroy, 2009; Young, 2013). Additionally, participants infrequently engaged on weekends, which will be important to consider in future iterations of the intervention. Overall, these findings will inform improvements to the moderator posting schedule for the upcoming pilot RCT in that we will aim to foster trust and community early on in the intervention and take into account when participants will most likely view posts.

Preparation leading into the beta-test was essential to its success and will also be critical to future testing and dissemination efforts. The moderator exchanges facilitated the integration of MI techniques, AI/AN ways of knowing, and a positive and authentic approach, all of which contributed to the successful moderation of the group. Indeed, moderators frequently elicited change talk from participants, and participants typically elaborated on their initial comments when moderators replied. Of note, no conflicts occurred between participants or moderators, and none of the participant posts were deemed misinformation or inappropriate, each a positive indicator that moderators were proactive to foster a group culture of hope and respect. The intervention also benefitted from having at least one moderator be an American Indian woman, as participants appreciated sharing AI/AN values with one another. The beta-test provided valuable lessons for the moderators, which will be considered in the upcoming RCT to enhance group facilitation, co-moderation, and participant engagement further.

In general, participants appeared to have a positive experience with the intervention based on their high USE scores and the content of posts, comments, and open-ended feedback. Initial feedback from the CAC highlighted the importance of exploring what participants already do to support their own recovery, and content of this type was a positive source of engagement during the beta-test. Selecting, training, and supporting AI/AN women with lived experience as peer moderators in future iterations of the group could be a way towards sustaining the intervention, centering participant recovery stories, and facilitating community implementation of the intervention.

Conversely, posts that elicited lower engagement were task-oriented (e.g., MBRP or SMART exercises) or asked participants to watch videos of moderator-selected recovery stories. Interestingly, however, qualitative interview participants from the previous study phase were highly receptive to three task-oriented posts and all three videos they viewed (Young et al., 2024), and beta-test participants recommended more videos. Live moderator-led video sessions with participants to discuss video posts or practice new MBRP or SMART skills with a debrief on their experience, or “yes/no” polls asking whether participants tried a skill, may be helpful to promote and assess engagement with this type of content. In sum, our findings indicate that future iterations of the group may benefit from examining which metrics best capture engagement on different types of posts and from using a patient-driven, strengths-based (Ezell et al., 2023), and conversational approach to build community and encourage personal exchanges of how reconnection to AI/AN culture supports recovery (Gone & Looking, 2011; Legha & Novins, 2012; Venner et al., 2018). It will be important for future studies to continue to explore facilitators and barriers to group engagement and cohesion.

Though the beta-test provided valuable lessons and was generally encouraging, limitations should be acknowledged. First, though word-of-mouth was encouraged during recruitment, it was not at the forefront of our recruitment methods. More efforts to promote word-of-mouth may have expedited our recruitment efforts. Recent reviews and studies demonstrated that respondent driven sampling (Heckathorn, 1997) is a viable method to reach hidden populations who inject substances and may hesitate to join research studies, especially if enrolled participants receive remuneration for each referred participant (Abadie et al., 2022; Raifman et al., 2022). Respondent driven sampling could also strengthen the application of CBPR among AI/AN people, reinforcing community strengths and interconnectedness, as well as improving the representativeness of AI/AN samples.

Limitations should also be acknowledged regarding the intervention. First, Facebook’s “Spam” detection and subsequent auto-removal of two moderator-posted videos prevented us from including them in our analyses and occasionally led to the temporary suspension of a moderator’s account. Meta’s online Transparency Center describes “Spam” content as “creat[ing] a negative user experience, detract[ing] from people’s ability to engage authentically in online communities and…threaten[ing] the security, stability and usability of our services” (Meta, 2023). Posts with external links (e.g., YouTube) may have been detected as “Spam” due to their low engagement from participants, so uploading videos directly to Facebook instead of linking external websites may ameliorate this concern. Second, the restricted sample size and duration of the intervention may have influenced group uptake and engagement. Indeed, participants indicated in their feedback that they wished the group had included more participants and lasted longer. Additionally, we did not collect any information from participants about whether they connected with one another using Facebook Messenger or beyond the group, which may be important to understand in future iterations of this work.

Furthermore, we recruited participants who were already using MOUD, and many were in recovery for approximately one year or longer. This limits our understanding of the potential applicability of the intervention to more heterogenous groups who may or may not be exploring treatment options. Finally, though the beta-test was meant to inform refinements of the intervention and study processes for future work rather than lead to generalizable findings, it is important to acknowledge that our small sample size limits claims of generalizability and warrants further exploration of the feasibility of broad implementation of the intervention in community contexts. Respondent driven sampling and the modification of Wiidookaage’win content to encourage the exploration of treatment and MOUD may broaden its reach to those newly in recovery.

In conclusion, the beta-test of the Wiidookaage’win Facebook group intervention appeared to work as intended, indicated preliminary feasibility and acceptability of the intervention, and provided valuable lessons for final refinement efforts prior to conducting a pilot RCT. With a larger target accrual of 60 participants, the pilot RCT will aim to provide stronger evidence for the feasibility, acceptability, and potential efficacy of the intervention to support opioid recovery among AI/AN women. It is our hope that this participatory approach to development and evaluation will ultimately result in an easily accessible and culturally tailored intervention that could be disseminated into community settings.

Abbreviations:

AI/AN

American Indian/Alaska Native

CAC

community advisory committee

CBPR

community-based participatory research

CHRT-SR

Concise Health Risk Tracking-Self Report

HNAB

Healthy Nations Advisory Board

IDTC

Instant Drug Test Card

MBRP

Mindfulness-Based Relapse Prevention

MI

Motivational Interviewing

MIWRC

Minnesota Indian Women’s Resource Center

MOUD

medication for opioid use disorder

OUD

opioid use disorder

RCT

randomized controlled trial

SMART

Stress Management and Resiliency Training

TLFB

Timeline Follow-Back

UDS

urine drug screen

USE

Usefulness, Satisfaction, and Ease of Use

WHOQOL SPRB

World Health Organization Quality of Life — Spirituality, Religious, and Personal Beliefs scale

Data availability statement

The datasets generated and analyzed during the current study are not publicly available to protect the privacy of participants, and individuals did not consent to have their data shared publicly. However, data are available from the corresponding author on reasonable request and after review and approval from the Community Advisory Committee.