Process Adaptations to Community-Engaged Research for Victimization Prevention of Trans Women: Failure as a Blueprint towards Non-Exploitative Implementation Science

Abstract

Effective violence prevention interventions are largely inaccessible to trans women and trans femmes, despite clear evidence that disproportionate exposure to experiences of victimization is a social determinant of health disparity. Community-engaged implementation science paradigms hold promise for guiding research psychologists in the delivery of evidence-based programming to address drivers of health disparities impacting trans women and trans femmes. Unfortunately, guidance on how to engage in a process of real-time self-reflection to note where implementation is failing in its goals to establish reciprocal and sustainable (i.e., non-exploitative) community partnerships are lacking. We describe our application of a modified failure modes and effects analysis to guide data-informed adaptations to our community-engaged implementation research project, tailoring and delivering an evidence-based intervention (EBI) to prevent victimization of trans women and trans femmes. By mapping our failure modes, we offer a blueprint for other research psychologists invested in advancing non-exploitative research in partnership with community.

Violence is a public health crisis among trans women and trans femme communities (Valentine et al., 2017), including women and nonbinary people who were not assigned female at birth and identify with a transfeminine experience of gender.¹ As reported by the Human Rights Campaign (HRC, 2021), at least 57 trans people were murdered in the U.S. in 2021, the majority of whom were young Black and/or Latina/Latinx trans women and trans femmes, making it the deadliest year on record for trans communities. This number is likely an underestimate, as tendencies to misgender trans victims in official reports are well-documented (HRC, 2021). Victimization and violence are social determinants of physical and mental health disparities among trans women and trans femmes, contributing to elevated rates of depression, anxiety, suicidality, substance use disorder, and HIV (James et al., 2016; Reisner et al., 2014). As such, interventions to interrupt the violent victimization of trans women and trans femmes may constitute a pathway to mitigate interconnected health disparities impacting these communities.

Empowerment Self Defense (ESD) programs are a critical component of multi-pronged violence prevention (Orchowski, 2020). ESD approaches are group-based interventions that empower participants with physical, verbal, and psychological tools to understand their bodies as sources of strength while explicitly holding perpetrators responsible for violence (National Women’s Martial Arts Federation, 2018). ESD programs evidence decreased rates of victimization in randomized control trials (Gidycz & Dardis, 2014), and are associated with reductions in shame, self-objectification, and stigmatization (Rosenblum et al., 2014). These psychosocial factors, which may result from exposure to anti-trans violence, are known determinants of HIV risk for trans women and trans femmes (Operario et al., 2014). Recent research suggests that self-protective behaviors are commonly employed among Black trans women and trans femmes coping with the aftereffects of violence exposure (Sherman et al., 2021). Moreover, social and psychological stressors specific to the identity-based stigmatization of trans individuals are theorized to increase risk for health disparity concomitant to interpersonal violence along multiple unique pathways (Hendricks & Testa, 2012; Meyer, 2003). As such, ESD approaches hold promise not only to reduce rates of violence against multiply marginalized trans women and trans femmes, but also interrupt downstream impacts on physical and mental health disparities.

Unfortunately, trans women and trans femmes have been largely left out of the violence prevention research literature, and standard ESD programs do not incorporate content specific to gender or racial minority stress. Nearly all research on ESD has focused on educated, upper middle class, White, cisgender women in college settings (Gidycz & Dardis, 2014). The omission of trans communities of color² in this research is alarming, given evidence for racial and ethnic disparities in exposure to violence, financial stability, and health outcomes among trans people (HRC, 2021; James et al., 2016). As evidence-based interventions (EBIs) that bias implementation towards higher resourced populations may exacerbate health disparities (McNulty et al. 2019), efforts to tailor and evaluate ESD interventions for diverse trans women and trans femme communities are urgently needed. To begin to address this need, our research team conducted a community-engaged implementation study to tailor and deliver an ESD program for trans women and trans femmes.

This study was guided by the principles of implementation science. Implementation science refers to “the scientific study of the use of strategies to adopt and integrate evidence-based health intervention into clinical and community settings to improve patient outcomes and benefit population health” (National Institutes of Health, 2017). Broadly, implementation science methods focus on increasing the feasibility and acceptability of health interventions in community settings (Bauer et al. 2015; McNulty et al. 2019). Implementation scientists recognize healthcare disparities as a special case of implementation failure (Woodward et al., 2021). Given the benefits of collaboration with communities of interest for acceptability and feasibility of health equity intervention implementation (Blachman-Demner et al., 2017; Schlechter et al., 2021), community-engaged implementation practices focus increasingly on specifying determinants, processes, and indictors of health equity (Nápoles & Stewart, 2019; Shelton et al., 2020; Woodward et al., 2021). These frameworks borrow from and integrate innovations of community-engaged research, including community-based participatory research (CBPR)—a diverse set of research practices guided by reciprocity, trust, and empowerment (Ortiz et al., 2020) and methods that center community strengths and priorities, invest in long-term commitment, and use knowledge to eliminate health disparities (Israel at al., 2013).

Despite documented benefits in health policy associated with community-engaged research practices (for a review see Ortiz et al., 2020), even well-intentioned researchers may struggle to implement these practices effectively, given structural power imbalances between research teams and community partners (e.g., in terms of access to resources, decision-making, convenience, reputation, etc.; Folayan et al., 2020). Community-engaged collaborative research can also result in exploitation (i.e., the erasure of voices that contribute meaningfully to research; Folayan et al., 2020), thereby thwarting health equity goals. Unfortunately, there is limited guidance on specific processes for researchers interested in engaging in real-time self-reflection to note where research and intervention implementation fails in its non-exploitative goals to engage community members and establish reciprocal and sustainable partnerships.

Purpose of the Current Study

To fill this gap, we analyzed the community-engaged implementation of a violence prevention EBI to trans women and trans femmes to identify barriers to non-exploitative implementation and present actions taken to respond to barriers. Specifically, we describe the retroactive use of a modified failure modes and effects analysis (FMEA; Institute for Healthcare Improvement [IHI], 2017) to model how community-engaged implementation researchers may identify and respond to failure in actuating health equity principles. Our current analysis serves as a blueprint for psychologists invested in self-analysis as one strategy for moving towards non-exploitation and health equity in community-engaged implementation science.

Method

Study Setting

The current analysis leverages data gathered from a larger community-engaged implementation science project to tailor an EBI for violence prevention to meet the needs of trans women and trans femme communities in a large mid-Atlantic metropolis. The parent project employed a two-phase design. Phase 1 entailed formative qualitative work to engage research partners on a community board and interview community members and stakeholders serving the community to inform EBI adaptation and facilitation team training. Phase 2 involved sequential delivery of the EBI to three unique participant cohorts in a single arm trial preregistered with Clinicaltrials.gov (ID: CT04934189).

Figure 1 provides a visual summary of the overall design of the parent project and maps data sources leveraged in the current analysis (i.e., ethnographic field notes from interactions between the research team, EBI practitioners, stakeholders at community organizations, and community members; electronic correspondence; qualitative interviews with community members, stakeholders, and EBI practitioners) on to each stage of the research process. We examined data from these sources using failure modes and effects analysis (FMEA; IHI, 2017). FMEA refers to a commonly used step-by-step methodological approach to process evaluation, designed to promote quality improvement in research and/or program implementation. Key components of FMEA may include identification of failure modes (what failures occurred), indicators (data that suggest failure occurred), causes (mechanisms supporting the occurrence of failure modes), effects (consequences of failure if unaddressed), and research team responses (actions taken to reduce occurrence or mitigate harms of failure). FMEA aligns with implementation science principles, as failure modes inherently represent barriers to feasible and acceptable implementation. All study procedures were approved by the affiliated large public university’s Institutional Review Board and research participants gave informed consent for participation. Data sources gathered from research partners were considered program evaluation activities. Research materials can be provided upon request. To protect confidentiality, data are not available to individuals outside the research team.

Figure 1. Data Sources and Procedures Across Phases 1 and 2 of Study Implementation.

Note. Data sources, their functional role in the study, and data types collected across Phases 1 and 2 of the study. CAB, community advisory board; EBI, evidence-based intervention; ESD, empowerment self-defense.

Researchers’ Backgrounds, Experiences, and Biases

Our analysis centralizes an examination of subjective descriptions and observations of the experiences of community partners, stakeholders, and EBI practitioners to derive meanings in an iterative, inductive process (Levitt et al., 2018; Wertz, 2010). Given this epistemological framework, researcher biases inevitably influence meaning-making and data analysis (Hill, 2012). To minimize the impact of these biases, the research team directly articulated assumptions and biases throughout the research process. All team members believe that structural imbalances in power and resources between researchers and members of historically marginalized communities reinforce risk of exploitation. Team members discussed biases associated with their personal backgrounds, demographic characteristics, and experiences at length to hold one another accountable for minimizing their impact on data interpretation and analysis. The research team included one clinical psychologist, two graduate students, and one bachelor level research assistant. All study team members are White, sexual and/or gender minorities. All researchers involved in this study are committed to social justice, epistemic equity, anti-racism, and the improvement of clinical services to LGBTQ individuals.

Data Sources & Procedures

Community Advisory Board (CAB)

We assembled a CAB of engaged research partners to collaboratively develop the tailored ESD violence prevention curriculum. CAB members were recruited via identification and outreach to community leaders and key stakeholders at organizations known for serving trans women. The CAB included community members (i.e., trans women and trans femmes) and stakeholders (i.e., social service providers with experience providing direct care to the community) who provided formative input about the specific violence prevention needs of the community and potential barriers to acceptability and feasibility of the EBI.

Community members included four trans women and trans femmes: a ballroom housemother (e.g., a guide for those belonging to a subculture of queer and trans people of color that integrates performance, dance, lip-syncing, and modeling) and previous anti-violence educator; an individual working with an organization that advocates for people affected by HIV/AIDS and housing insecurity; an individual working to create space and resources for Black trans artists and culture; and the founder of an organization offering resources for trans and nonbinary communities of color. Stakeholder members were four individuals with ties to organizations that serve trans communities; a peer adherence educator at a community health center for people of color (fourth author), a program director at a community health center providing care to LGBTQ+ communities regardless of their ability to pay; a manager in mental health services at an organization that provides health, wellness, and community programs for the LGBTQ+ community (fifth author), and an EBI practitioner-in-training at a local ESD organization. CAB members were compensated $80 for 4 hours of labor entailing the review and provision of feedback on the research and intervention via board meetings, one-on-one meetings, and electronic correspondence.

Evaluative data from CAB members leveraged in the FMEA were obtained via direct observations, field notes taken during board and one-on-one meetings, electronic correspondence between CAB members and the research team, and written feedback from CAB members on study-related documents (i.e., written revisions or suggestions). All CAB members were invited to contribute to the current manuscript as co-authors. CAB members who agreed to join the authorship team were invited to contribute to manuscript conceptualization, writing, and review.

Community Partnership Development and Maintenance

Development and maintenance of relationships with trans-femme led and allied community organizations to support study implementation required a specific research team role dedicated to coordinating and initiating targeted multi-platform (email, phone, social-media messaging) outreach to prospective community-partners across the two-phase study (i.e., a community-engagement manager, first author). Whenever possible, Zoom and in-person meetings were arranged to support the development of rapport, trust, and human connection. Because we assumed that tailoring and feasible/acceptable delivery of EBI required community-based expertise and ongoing investment in community partnerships, we strived for the total time and effort spent by the research team to be greater than the time and effort asked of community partners. Data sources related to community partnership development and maintenance included electronic correspondence and field notes from community partner meetings and community partner events or organization meetings. The research team collated and collaboratively discussed data to assess the extent of engagement, partnership development, community partner needs, partnership reciprocity, and study/research partnership acceptability to community organizations.

Community Member and Stakeholder Interviews

Trans women and trans femmes and stakeholders serving trans communities (N = 17) were interviewed to gather information about the violence prevention needs of the community in Phase 1 of the study. Stakeholders included trans-specific violence prevention (N = 3), legal support (N = 1), mental health care (N = 1), and housing access (N = 2) experts. Twelve (70.6%) individuals in the interview sample were trans women, one was a trans man (5.9%), and four were nonbinary people (23.5%). Five (29.4%) interviewees were Hispanic/Latinx. Six (35.3%) interviewees were Black, eight (47.1%) were White, one (5.9%) was American Indian, and two (11.8%) were Asian. Interviewees’ ages ranged between 18 and 56 (M = 34.5, SD = 10.54).

Interviews lasted between 56 and 122 minutes (M = 78 minutes, SD = 18 minutes), and were conducted by all members of the research team. Interviewees were compensated $40 for their time. Interviews were semi-structured and prompted for contexts of violence and discrimination, as well as recommendations for adapting the EBI and study implementation process to meet the needs of trans women and trans femmes. Interviews were audio-recorded and analyzed using a rapid-content analysis strategy (Beebe, 2001; Neal et al., 2015) to inform community-derived, data-informed adaptations to the intervention manual and training of the lead EBI practitioner. Rapid content analysis is an intensive, team-based approach for qualitative data analysis that quickly generates preliminary findings to inform intervention development (Beebe, 2001). For reliability, multiple coders discussed and resolved all disagreements through consensus. The current analysis utilized themes related to the study implementation process (i.e., practical, structural, and psychological barriers to delivery, uptake, and sustainability of EBI in community). Themes characterizing violence and discrimination faced by the community will be described in a forthcoming analysis.

Facilitation Team Interviews

We partnered with a local chapter of a global organization providing Empowerment Self-Defense (ESD) programs to deliver the adapted EBI, a 5-week, 20-hour group violence prevention intervention, three times in serial to three unique participant cohorts. The EBI partner organization, a for-profit entity led by White, cisgender women, provided an ESD-trained facilitation team including the lead EBI practitioner, an EBI practitioner-in-training (a White, transmasculine, nonbinary person), and two suited instructors (a Black cisgender man and a White cisgender man) who act as aggressors in role-played scenarios for physical resistance skill practice. Selection of the ESD facilitation team was based on their history of effective EBI dissemination and their existing organizational network to support future dissemination.

In Phase 1 of the parent study, the research team and lead EBI practitioner met weekly and communicated regularly through electronic correspondence to discuss study implementation processes, share community and stakeholder insights and recommendations for program adaptation (e.g., training trans women and trans femmes as part of facilitation team, tailoring roleplay scenarios and didactic material to address fetishization of trans women and trans femmes and community interactions with police), and support the lead EBI practitioner’s competency and adherence to these cultural adaptations. A full description of cultural adaptations to the EBI and the acceptability and feasibility of the tailored ESD program will be published in a separate analysis. In Phase 2, all members of the facilitation team attended weekly meetings with the senior author to share observations and problem-solve challenges to EBI implementation. Research team members (co-lead and third authors) interviewed facilitation team members following each series of the EBI. Demographic information was not collected from the facilitation team. Interviews lasted between 72 and 125 minutes (M = 92 minutes, SD = 23 minutes). Interview questions asked about general intervention impressions, specific session content and series elements, group cohesion, facilitation roles, and power dynamics (inter- and intra-facilitation team).

In the current analysis, we utilized data sources derived through interactions with the facilitation team, including field notes from meetings with the lead EBI practitioner and the facilitation team and electronic correspondence between the lead EBI practitioner, facilitation team members, and research team members. Electronic correspondence and field notes were shared and discussed collaboratively by the research team to characterize acceptability and feasibility of training processes and program implementation. Facilitation team exit interviews were rapid-coded (Beebe, 2001; Neal et al., 2015) to direct content analysis for recommendations on how to improve training and study processes. The current analysis leveraged themes derived from consensus-based research team conversations related to training experiences, power dynamics, and inter-facilitation team relationship processes.

Analytic Plan

The goal of the current analysis was to leverage thematic and observational data gathered over the course of our community-engaged implementation study to identify barriers to non-exploitative implementation. The co-lead authors conducted a modified FMEA (IHI, 2017) to retroactively apply stakeholder and community feedback to steps in study implementation and analyze barriers to non-exploitative implementation. Specifically, the research team leveraged formative Phase 1 research data (i.e., CAB member feedback; community and stakeholder interviews) and Phase 2 process implementation data gathered concurrent with intervention delivery (i.e., facilitator exit-interviews; researcher field notes) to conduct the FMEA. Data analysis was guided by minority stress and implementation science models (e.g., Meyer, 2003; Woodward et al., 2021). Co-lead authors conducted a detailed review of all available data sources (i.e., indicators) and relied on consensus-based decision making to identify associated failure modes, hypothesized causes and effects, and to report research team responses to failure. Description of research team responses to failure were limited to those that could be feasibly taken within the budgetary and time-limitations of the parent study. To support interpretation of the FMEA, we mapped our findings onto key study implementation process steps, presented chronologically in Table 1. Once the co-lead authors reached agreement on failure modes, indicators, causes, effects, and responses, community partners were presented the opportunity to review and confirm agreement. Due to their inapplicability to our analytic goals, we removed the FMEA’s quantitative components (e.g., the calculated risk profile number). The FMEA was not preregistered with Clinicaltrials.gov.

Table 1.

Modified Failure Modes and Effects Analysis of Community-Engagement Implementation Science Approach to Adapt Evidence-Based Violence Prevention Intervention for Diverse Trans Women and Trans Femmes

Process Steps	Failure Modes	Failure Indicators	Failure Causes	Failure Effects	Research Team Responses to Failure
PHASE 1: Collection and Integration of Community Feedback into Existing Intervention
Assemble and engage CAB	1: Shallow collaboration between research team and CAB	25% of CAB members did not engage in meetings or provide feedback Minimal critical feedback on study process/adapted EBI (meeting field notes, electronic correspondence)	Limited trust that investment in research will sustainably benefit community Motivation to engage in ways that do not risk access to researcher resources CAB member capacity given competing priorities to serve community	CAB is unable to guide research team in centering community needs	Credit CAB as contributors (e.g., EBI manual, publication co-authorship) Emphasize expertise of CAB members Integrate individual CAB members as consultants
Partner with community organizations	2: Low acceptability of research program by trans and Black-led community organizations	Limited/critical responses to outreach (community partnership meeting field notes and electronic correspondence)	Lack of credit for community expertise No pathway to community ownership of EBI Exploitation of community by White/cis researchers	Low feasibility of recruitment	Take responsibility Reallocate funding towards community ownership (e.g., community-facilitator training pipeline, EBI delivery in community organization)
Interview community members and stakeholders	3: Voices of trans women and trans femmes of color are not centered in content analyses informing EBI adaptation	67% White interview sample	Limited trust that investment in research will benefit community Exploitation of community by White researchers Competing priorities for survival faced by community, limiting capacity	Low acceptability of adapted intervention to highest priority community members	Ongoing investment in reciprocal community partnerships, prioritizing investment in organizations led by trans communities of color. Investing resources to recruit people of color through community connections
Train the EBI facilitation team	4: Limited acceptability of community-facilitator training by EBI practitioner	Lead EBI practitioner repeatedly selecting White community members over people of color to train in EBI (EBI practitioner meeting field notes) Microaggressions by EBI practitioner, relationship ruptures between lead EBI practitioner and community-facilitators (EBI session and training meeting field notes, facilitation team meeting field notes, exit interview themes)	Implicit biases Poor incentive structures (i.e., for-profit set up of EBI organization) Structural racism	Harms enacted against community Limited community-facilitator acceptability of EBI training Limited feasibility of community-facilitator EBI training	Clinician/PI facilitate ad hoc consultation with EBI practitioners and community-facilitators De-center partnership with EBI organization
PHASE 2: Intervention Delivery; Iterative Adaptation
Plan for sustainable dissemination	5: Adapted intervention is inaccessible to trans women and trans femmes of color after trial ends	Critical feedback from community organizations on need for community ownership of and sustained access to EBI, (community partnership meeting field notes) Recommendations for EBI delivery by trans women or trans femmes of color (community member and stakeholder interview themes)	Planned dissemination through lead EBI practitioner relying on for-profit training and payment mechanisms Limited resources held by community organizations	Limited ongoing benefit to community Exploitation (i.e., researchers benefits from community contributions) Exacerbation of community mistrust in clinical research	Invest in community-facilitator training pipeline Partner to secure funding that flows directly to community organizations Share research/clinical expertise to address unmet community needs

Notes. CAB, community advisory board; PI, principal investigator; EBI, evidenced-based intervention

Adapted from Institute for Healthcare Improvement [IHI]. (2017). QI Essentials Toolkit: Failure Modes and Effects Analysis (FMEA). Cambridge, MA. Retrieved from http://www.ihi.org:80/resources/Pages/Tools/Quality-Improvement-Essentials-Toolkit.aspx

Results

Failure Mode 1: Shallow Collaboration Between Research Team and CAB

Although community advisory board (CAB) members were generally agreeable to the study implementation plan and intervention adaptations, 25% did not attend CAB meetings or provide feedback via electronic communication. CAB members who did engage offered minimal critical feedback, evidenced by field notes from CAB meetings and electronic communication. Because the function of the CAB is to inform adaptations, their agreeableness may indicate a failure to create a space for meaningful engagement and collaboration. We hypothesize that the lack of critical feedback and active collaboration reflects a poor foundation of trust that community investment of time and energy into our research would be of significant, tangible, and sustainable benefit. We hypothesize that efforts to encourage open discussions of procedures and concerns were insufficient given: 1) CAB member awareness of community needs for researcher-held resources (e.g., funds, program-delivery infrastructure), 2) constraints on CAB member’s capacity and prioritization of community services, and 3) CAB member assumption of researcher expertise over their own. If unaddressed, the lack of critical feedback and active collaboration would result in a dilution of CAB contributions to the EBI and its implementation.

To build trust requisite to meaningful collaboration with CAB members, we: 1) credited CAB members in intervention manuals and publications; 2) explicitly and repeatedly affirmed our openness to their feedback as experts; and 3) facilitated one-on-one consultation conversations with individual CAB members. Field notes from these interactions suggest that CAB members were more willing to share their opinions openly when approached one-on-one, as we were able to target discussions to focus on the specific areas of individual CAB member expertise, granting deeper credibility to our stated values.

Failure Mode 2: Low Acceptability of Research Program to Trans and Black-Led Community-Based Organizations

Indicators of a second failure mode emerged during Phase 1 community partnership building outreach efforts, which were met with low-responsiveness and critical feedback represented in researcher field notes from community partnership meetings and electronic correspondence related to outreach efforts. Through our outreach efforts, we met with the founder and chief executive officer (CEO) of a Black-led nonprofit organization that provides access to funding, networks, and training to artists, creators, and entrepreneurs of color (sixth author). Although the CEO declined to partner with our team, they generously provided direct critical feedback explaining their decision. They highlighted how our partnership with the EBI practitioner had clear benefits for our research team (e.g., funding research team positions) and the practitioner (e.g., expanding the EBI practitioner’s consumer market), but minimal comparative benefit to community partners and participants. The research team was “called in” to acknowledge that our reliance on community members to further researcher goals without ensuring lasting benefits to the community (i.e., community access to and ownership of the intervention) was exploitative. If unaddressed, low acceptability of the research program to community organizations and limited willingness of these organizations to refer those they serve would be significant barriers to recruitment and feasibility of the EBI implementation study.

To improve acceptability among community partners, we took responsibility for the failure in our interaction with the CEO and in our study implementation plan. During our interaction, we responded non-defensively to criticism, validated the harms that would result from our initial EBI implementation plan, held significantly more space in our conversation for feedback than our response, and communicated a commitment to make changes to our approach to address their feedback. We do not have data to suggest how our response was interpreted by the CEO. However, their organization was willing to continue providing feedback as we adapted our study implementation plan and collaborate as co-authors in the current analysis, suggesting movement towards a reciprocal relationship.

To improve acceptability among community partners, we reallocated resources to support more equitable exchange and sustainable pathways to community ownership, including redirecting space rental fees in our budget to embed delivery of the intervention in a community-based organization founded and led by Black trans women. To further reallocate resources controlled by researchers and local EBI practitioners, and follow recommendations by community members and stakeholders in Phase 1 interviews, we created a pipeline to facilitator training, integrating community members into the ESD facilitation team. Three Black trans women and one White trans woman joined the facilitation team and were trained in intervention delivery roles. Two community-facilitators, recruited via community connections, began training with the facilitation team in Phase 1 and two were invited from the first series of the EBI during Phase 2. We leveraged a learning collaborative approach to training, where the EBI facilitation team provided training in the adapted EBI to community-facilitators and community-facilitators provided expertise and training in the needs of trans women and trans femmes.

Failure Mode 3: Voices of Trans Women and Trans Femmes of Color Not Centered in Content Analyses Informing EBI Adaptation

Although we attempted to engage community partners who specifically cater to the needs of trans women and trans femmes of color in our recruitment efforts, Phase 1 interview data were derived from a primarily (67%) White sample, suggesting that we did not have a sufficient foundation of trust to support this recruitment plan. Moreover, trans women and trans femmes of color have many competing priorities for survival, limiting their availability to engage in study procedures. Failure to improve representation as we conducted interviews in Phase 1 resulted in a major threat to the validity of research findings and their impact. Themes derived from a primarily White sample do not represent communities at highest risk for violent victimization. By extension, adaptations to the intervention program curriculum informed by these themes may not be responsive to community needs.

To improve representation, we deepened investments in reciprocal partnerships with organizations led by trans communities of color and reallocated funding to support transferring ownership of intervention delivery (e.g., community-facilitator training pipeline). Seven core partnerships made execution of this project possible and improved our study process. Supplemental Table 1 provides a summary of partnerships, their functional role in our process, and our efforts at reciprocal contributions to each partner. These actions resulted in increased community trust and deeper community investment in research activities, indicated by our ability to recruit samples of majority people of color in Phase 2.

Failure Mode 4: Limited Acceptability of Community-Facilitator Training by EBI Practitioner

Low acceptability of the community-facilitator training by the lead EBI practitioner comprised a fourth failure mode. Indicators of this failure mode emerged from behavioral and interpersonal observation of the lead EBI practitioner documented in researcher field notes and electronic communications. First, we observed the lead EBI practitioner repeatedly identify White, rather than trans women and trans femmes of color, as ideal candidates for community-facilitator training, citing characteristics including “natural” coordination/strength in EBI skills, interpersonal savvy/leadership skills, professionalism, session attendance, and punctuality. Second, we observed the lead EBI practitioner enacting microaggressions against community-facilitators, community partners, and trial participants (e.g., appropriation of language, expressed discomfort with participants’ dress/attire, delegation of physical labor to community members, correcting the grammar of community members), amounting to colonization of community space. Further, in training with the community-facilitators, the EBI practitioner struggled to adapt from a directive, hierarchical communication style to a collaborative approach (i.e., one that invites the expertise of community-facilitators in the process and content of training).

We hypothesized that the EBI practitioner’s low acceptability of community-facilitator team training resulted from implicit racial and cisheteronormative biases conflating definitions of professionalism and Whiteness (Gray, 2019). Further, the for-profit set up of the EBI organization may have incentivized resistance to transferring ownership of the EBI to the community. Within a profit-driven model, transfer of EBI ownership to the community may be perceived as a threat to financial security. Microaggressive behaviors may have emerged as a means to manage this threat. Implicit bias and for-profit organization models reinforce and originate from structural racism and White supremacy. Slavery and race-based inequity are embedded in Western capitalist systems (Kendi, 2019). For-profit organizations recreate this dynamic in which the privileged (e.g., White, cisgender people) capitalize on the needs of communities facing oppression (e.g., trans people of color). If we allowed the EBI practitioner to select candidates, access to benefits of the community-facilitator training and intervention ownership would be blocked to trans women and trans femmes of color. Further, the EBI practitioner’s microaggressive behavior actively harmed trans women and trans femmes of color. Relationship ruptures between community-facilitators and the lead practitioner reduced feasibility by threatening our ability to retain community-facilitator trainees, and the EBI practitioner’s approach to facilitator training limited acceptability to community-facilitators.

In response to the EBI practitioner’s selection of only White training candidates, we presented the numerous strengths and skillsets held by several trans women and trans femmes of color and leveraged our power as the research team to select the community-facilitators. Researcher selection of community-facilitator trainees improved representation, but similarly reinforced gatekeeping practices in EBI development. We also immediately and repeatedly communicated the inappropriateness of racist and cisheteronormative attitudes and microaggressive behaviors to the lead EBI practitioner. In response to feedback, the EBI practitioner reasserted a commitment to empowering trans women and trans femme communities of color and made efforts to apologize and repair harms. The senior author (clinical psychologist) held group and individual consultation meetings with the lead EBI practitioner, EBI practitioner-in-training, and each community-facilitator to name microaggressions when they occurred, acknowledge harms, and provide as-needed coaching through barriers to clear interpersonal communication (e.g., misunderstandings, confusion, defensiveness, fear, mistrust).

Given the acuity of this failure mode, the research team critically evaluated the relative harms of continuing to involve the lead EBI practitioner versus terminating the partnership immediately and risking community access to the potentially lifesaving EBI. Ultimately, we took action to prioritize community needs (e.g., equitable access and ownership of the intervention) over the needs of the for-profit EBI organization and the desires of researchers (i.e., to avoid conflict in or dissolution of the relationship with EBI practitioner). We communicated to the EBI partner our intention to terminate the partnership at the end of the pilot trial should indicators of the failure mode continue. In response to this feedback, the lead EBI practitioner removed herself from the facilitation team in the final trial series, citing a desire to minimize harm.

We hypothesize that efforts to support EBI practitioner’s cultural competence and decrease behaviors associated with implicit biases failed because of the deeply entrenched nature of implicit bias. Feedback concerning these behaviors may not have motivated change because the benefits of proprietary intervention ownership (i.e., capitalizing on White, cisgender privilege) outweighed the benefits of community partnership to the EBI practitioner. Field notes derived from weekly debriefing meetings following each session of the final program series indicate that the EBI practitioner-in-training and community-facilitators thrived when given the autonomy to deliver the EBI collaboratively, demonstrating skill and resilience.

Failure Mode 5: Adapted Intervention Inaccessible to Highest Need Trans Women and Trans Femmes When Research Program Ends

Critical feedback from prospective partners emphasizing the need for community ownership of the adapted EBI (Failure Mode 2) elucidated how our original dissemination plan (based on the local EBI organization’s connection to a national network of ESD facilitators) failed to ensure access for trans women and trans femmes with limited resources at highest risk for violent victimization. Many community organizations lack the funding to hire the lead EBI practitioner to deliver the intervention and/or train community members in its delivery, as represented in field notes from interactions with community partners and electronic correspondence. Failure to develop a plan for dissemination accessible to those at most risk for violence would exacerbate existing inequities. To support sustained accessible dissemination, we developed the community-facilitator training pipeline as a pathway for transferring the knowledge, skills, and means of ESD delivery from EBI practitioners to community members, who can in turn, independently deliver the intervention and train fellow community members. We are also partnering in grant writing with community partners (i.e., Brooklyn Ghost Project) focused on full integration of the adapted EBI into existing community organizations (e.g., multisite rollout) and organizational support to deliver the EBI (e.g., structural resources and clinical expertise). Although insufficient to circumvent the long history and ongoing practice of institutional exploitation, these actions are necessary to begin transferring intervention ownership and power to trans women and trans femme communities.

Discussion

The current analysis leverages identification and analysis of failure as a paradigm for adapting community-engaged implementation science research on EBIs for vulnerable communities towards non-exploitation. We applied a modified failure modes and effects analysis (FMEA; IHI, 2017) to systematically evaluate impediments to feasible and acceptable implementation of an evidence-based violence prevention program, Empowerment Self-Defense (ESD), tailored to trans women and trans femmes. Findings highlight key challenges in the negotiation of expertise, power, and resources between EBI practitioners and those our health equity approach to implementation science aims to center–trans women and trans femme community members of color. Centering our partnership with EBI practitioners limited community acceptability and intervention feasibility. Centering community needs decreased EBI practitioner acceptability of the collaboration and revealed structural barriers to flexibly respond to community needs during implementation.

Although adapting to emergent failure modes in our research process required cultural humility, critical self-reflection, flexibility, time, and labor, doing so was entirely feasible under study budgetary and time constraints. Investing in non-exploitative, reciprocal relationships with community partners was necessary for building the trust requisite to acceptable, feasible, and sustainable implementation of an urgently needed EBI. As such, results of the current analysis suggest that psychologists engaged in community-engaged implementation science with health equity goals adopt self-auditing processes like FMEA, to identify, conceptualize, and intervene in exploitative dynamics in the research process as an essential and standard operating procedure.

Considerations for Applying Failure Modes and Effects Analyses

Community-engaged implementation science study process steps may look similar to those analyzed in the current study and share common failure modes. However, failure causes, effects, and indicators are specific to the study design, setting, intervention, and community characteristics, and likely require unique actions to adapt the implementation process towards greater acceptability and feasibility. We offer several broad considerations for researchers hoping to apply this approach to their work: 1) the community of interest and available data indicators; 2) where in the implementation process self-analysis occurs; 3) EBI and practitioner characteristics; and 4) local characteristics of research processes.

Although the current study aimed to center the needs, voices, and expertise of trans women and trans femme communities, we argue that self-auditing and adaptation of implementation processes that maintain failure modes are necessary for research with any community that faces marginalization. In the current study, failure modes and recommended responses were often identified and communicated most clearly by community members and organizations. These findings align with theoretical models that recognize the expertise of individuals in communities that are socially and structurally oppressed as a function of the need to survive these conditions (Collins, 1990; Hooks, 1984). Whenever possible, researchers should utilize data from community organizations and members, integrating EBI practitioners and researcher perspectives to support community-derived data indicators.

Researchers may also consider when in study implementation they initiate self-analysis. We adapted our study process towards non-exploitation after being “called in” by community partners four months into study implementation, suggesting that major and necessary shifts in the research process are feasible in ongoing studies. Community-engaged implementation scientists (e.g., Boothroyd et al., 2017) recommend integrating community voices and expertise in study conceptualization and design (i.e., focus groups or listening circles with trans community members of color). However, it is important to acknowledge that, regardless of when community voices are integrated in the research process, ongoing institutional and structural-level exploitation may limit individual researchers’ ability to mitigate failure. We recommend shifts in funding and institutional review board mechanisms to incentivize paths to reciprocal community partnerships.

In our efforts to adapt and deliver a violence prevention intervention for trans women and trans femmes, our community partners helped us identify the absence of a pipeline for transferring knowledge, skills, and material resources from EBI practitioners to community members as a major barrier to acceptable and sustainable implementation. The absence of pipelines for exchanging knowledge and resources from predominately cis and White-led researchers and EBI practitioners to marginalized communities is a challenge common across psychological science (Alegría et al., 2016; Becker, 2019). As reflected in our analysis, efforts to adapt EBIs that do not build capacity for community ownership are exploitative and limit intervention feasibility and acceptability. When selecting interventions to adapt and deliver, researchers should be prepared to mitigate structural inequities through dissemination plans that include mechanisms for training and compensating community members as practitioners.

In our failure modes and effects analysis, structural barriers to non-exploitative community-engaged implementation science were often identified as causal mechanisms. Our efforts to adapt our implementation process to target these causal mechanisms yielded mixed outcomes. In some cases, structural mechanisms (e.g., for-profit incentive structure of EBI organization) were insurmountable such that adaptations directed at the individual-level (e.g., clinical consultation, direct feedback) were insufficient to address individual attitudes (e.g., implicit biases) reinforcing and reinforced by structural social inequities. We found that structural shifts in implementation (i.e., community-facilitator training pipeline, de-centering EBI organization partnership) were more effective. In keeping with numerous recommendations for the conduct of non-exploitative science to address health inequity (e.g., Buchanan et al., 2021; Rosenthal, 2016), our findings underscore the need to identify and address structural-level inequities that underlie individual-level drivers of study implementation failure.

Limitations

There are several limitations to the current implementation process study. As researchers, we are complicit in a system that drives exploitation of the communities we aim to engage and serve (e.g., researcher benefits like grant-funded positions and publications non-contingent on sustained community access to the EBI). Identifying failures and acknowledging the exploitative nature of one’s own research practices may spur avoidance and defensiveness, limiting scientific objectivity. Research team members conducted interviews, generated field-notes and codes summarizing themes, and conducted the FMEA. As such, the validity of our analyses is limited by barriers to fully understanding structures of oppression and marginalization conferred by our privileged social positions. Non-representation of trans women and trans femme people of color in the research team limited the clarity of our perspectives. Our reliance on community partners to spontaneously provide failure indicators and bring exploitative or non-reciprocal methods to our attention constitutes a serious limitation of the current study. Sustained implementation by and for the community is required to fully address limitations.

Conclusion and Future Directions

In describing our self-analysis of failures in the conduct of community-engaged implementation research, we hoped to provide a blueprint for researchers interested in moving towards non-exploitative implementation science. Our findings demonstrate the feasibility of ongoing self-analysis and adaptation of research processes towards more non-exploitative methods, even under budgetary/time constraints. Preliminary findings suggest that adapting implementation processes towards non-exploitation support greater acceptability and feasibility of the intervention itself, although true non-exploitation requires individual researchers’ continuous efforts and significant structural changes towards non-exploitation.

Our research team acted as a bridge connecting community needs and expertise to scientific evidence, funding, trained facilitators, and clinical consultation. Our findings suggest that these implementation efforts would be more immediately impactful if research teams prioritize investment in mechanisms through which communities develop and deliver interventions by and for themselves. Within this model, research teams shift from the role of power-broker to culturally-humble clinical consultants available to serve community by directing interventions and resources to the mitigation of structural inequities embedded in implementation science. Endeavors to empower communities as owners of these implementation processes require multilevel interventions that address the epistemic exclusion of community voices in research funding, protocol development, and program dissemination (Buchanan & Wiklund, 2021; Roberts et al., 2020). We hope this blueprint for self-analysis of failures and a commitment to engage in data-driven adaptation of research processes towards non-exploitation helps researchers to be better bridges connecting resources to community. However, we perpetuate the ongoing colonization of health equity research (McFarling, 2021) if we center researchers’ role as gatekeepers of resources, rather than restructuring research and public service systems to support and remove barriers for communities who are already doing the work to address health inequity.

Public Significance Statement:

Community-engaged implementation research paradigms aim to expand the reach and impact of evidence-based programming to populations experiencing health disparities. However, few models exist for evaluating implementation of this research for health equity goals. In this article, we analyze modes of failure as a blueprint for researchers hoping to move their study processes towards non-exploitation.