Using community engagement to implement evidence-based practices for opioid use disorder: a data-driven paradigm & systems science approach

Abstract

Community-driven responses are essential to ensure the adoption, reach and sustainability of evidence-based practices (EBPs) to prevent new cases of opioid use disorder (OUD) and reduce fatal and non-fatal overdoses. Most organizational approaches for selecting and implementing EBPs remain top-down and individually oriented without community engagement (CE). Moreover, few CE approaches have leveraged systems science to integrate community resources, values and priorities. This paper provides a novel CE paradigm that utilizes a data-driven and systems science approach; describes the composition, functions, and roles of researchers in CE; discusses unique ethical considerations that are particularly salient to CE research; and provides a description of how systems science and data-driven approaches to CE may be employed to select a range of EBPs that collectively address community needs. Finally, we conclude with scientific recommendations for the use of CE in research. Greater investment in CE research is needed to ensure contextual, equitable, and sustainable access to EBPs, such as medications for OUD (MOUD) in communities heavily impacted by the opioid epidemic. A data-driven approach to CE research guided by systems science has the potential to ensure adequate saturation and sustainability of EBPs that could significantly reduce opioid overdose and health inequities across the US.

INTRODUCTION

Over the past two decades, opioid overdose deaths have increased more than four-fold in the United States (US) and have been the leading cause of unintentional deaths since 2012 (Centers for Disease Control, 2020). Increasingly, communities of color bear a disproportionate burden of opioid use disorder (OUD) and overdose (Substance Abuse and Mental Health Services Administration, 2020). Accumulating research has identified the key role that social determinants, including poverty, unemployment, homelessness, incarceration, stigma, racialized drug laws and policing practices play in driving the opioid epidemic in highly-burdened communities (Altekruse et al., 2020; Dasgupta et al., 2018; Galea et al., 2003; Galea et al., 2004; Kunins, 2020). Researchers have also identified multiple mechanisms through which community-level social determinants contribute to elevated opioid overdose rates, including inadequate access to medications for OUD (MOUD) and healthcare in general, higher rates of housing instability, criminal justice involvement, chronic stress due to limited economic opportunities, social capital and/or social cohesion needed to mobilize a coordinated community response to the opioid epidemic (Altekruse et al., 2020; Collins et al., 2020; Park and Wu, 2020).

Evidence-based practices (EBPs) for OUD include overdose education and naloxone distribution (OEND) (Walley et al., 2013), safer opioid prescribing (Lin et al., 2017), MOUD (Connery, 2015), motivational interviewing (Smedslund et al., 2011), peer navigation (Kelly et al., 2014), and stigma reduction (Livingston et al., 2012) to promote initiation, engagement, and retention in treatment. The availability and reach of EBPs remains suboptimal (Laudet and Humphreys, 2013; Mericle and Grella, 2016; Winhusen et al., 2020). For instance, fewer than 20% of the estimated two million people with OUD received MOUD in 2018 (Substance Abuse and Mental Health Services Administration, 2019; Wu et al., 2016). Moreover, strained health systems are often forced to address parallel epidemics, such as COVID-19, making it even more urgent to focus on innovative, community-driven approaches to stem the opioid epidemic.

Traditionally, approaches to improving accessibility, availability, and sustainability of EBPs for substance use disorders (SUD) have focused on the individual. Community-driven responses, however, are essential for ensuring the implementation, reach and sustainability of culturally-relevant EBPs. Community engagement (CE) approaches have had a lasting impact in many areas of health promotion (Cyril et al., 2015), yet the application of CE for OUD remains limited. CE and other participatory methods, including community-based participatory research (CBPR; (Israel et al., 2017)), (youth) participatory action research (Y/PAR; (Baum et al., 2006; Kohfeldt et al., 2011)), human-centered design (HCD; (Holeman and Kane, 2019)) and citizen science (Den Broeder et al., 2018), re-design traditional “researcher-subject” hierarchies to create iterative cycles of learning grounded in partnership and capacity exchange that can be highly responsive to community-identified priorities (Sprague et al., 2019). This collaborative co-design process has the potential to lower stigmatization and improve social and recovery capital. Numerous scholars have suggested that this critical process optimizes and synergizes community resources for collective response to the opioid epidemic (Holton et al., 2018; Langley-Turnbaugh and Neikirk, 2018; Palombi et al., 2019; Wells et al., 2018).

The size and heterogeneity of communities, availability of resources, shared values and competing priorities influence engagement strategies. Of particular concern is when CE is shaped and constrained by systemic and structural inequities that contribute to inequities in OUD and overdose. This paper introduces a novel CE paradigm that promotes the use of a data-driven and systems science approach to ensure the reach and adoption of community-specific, sustainable EBPs to address the opioid epidemic in heterogenous communities in the US. In this paper, we first define CE and describe the composition, functions and roles of researchers in CE. Next, we discuss unique, ethical considerations of CE research and consider the advantages of using systems science and data-driven approaches. This novel CE paradigm is then demonstrated by the HEALing Communities Study, which aims to reduce the rate of opioid overdose deaths in select US communities by 40% over three years (Chandler et al., 2020). Finally, we conclude with scientific recommendations for the use of data-driven CE in research, which can be applied to ensure the adoption and sustainability of EBPs for OUD treatment and recovery.

WHAT IS COMMUNITY ENGAGEMENT?

CE is the process of working collaboratively with and through groups of people affiliated by geographic proximity, special interest, or similar experiences (Centers for Disease Control, 2011). Within the context of clinical, translational and implementation research, CE refers to initiatives to create and sustain collaborative capacity to deliver EBPs (Centers for Disease Control, 2011). Communities work collaboratively to address problems affecting their well-being by developing academic partnerships and community coalitions that promote the implementation of EBPs by mobilizing resources, influencing systems, and serving as catalysts to change policies, asset access, and programs (Shore, 2006; Sprague Martinez et al., 2020; Wallerstein, 2002). This requires researchers to ally with the community and the community to become part of the research team, thereby creating a mutual understanding for sustained interventions. CE lies on a continuum of level of participation (Centers for Disease Control, 2011; Chung and Lounsbury, 2006), which can ebb and flow over the course of a partnership, ranging from compliant participation (e.g., directed outreach and consultation) to empowering co-investigation and co-creation (e.g., shared leadership) (IAP2, 2011). Although CE may be achieved during a time-limited project, effective CE is often derived from prior experiences in long-term partnerships that can shape approaches to structural problems spanning social, economic, political, and environmental aims (Centers for Disease Control, 2011).

CE for social and behavioral changes can be characterized by different forms of community organizing, including direct service, mutual help, education, advocacy and direct action (Bobo et al., 2001). The CE process is hypothesized to support increasingly greater empowerment of participating community stakeholders as community organizing moves from ‘direct service’ through ‘direct action’ (Bobo et al., 2010). Relatedly, where CE embraces a ‘power with’ approach (Guinier et al., 2009), it has the potential to foster highly emancipatory processes (Labonte and Robertson, 1996; Starhawk, 1987).

Principles of CE research

CE research engages community members and researchers in equitable partnerships at all phases of the research process, making it an attractive model for research with marginalized populations, including people who use drugs (Cyril et al., 2015). CE research must recognize moral standing, interdependence and power of all partners; exhibit empathy and mutual concern by balancing respective interests; awareness of the structural context of others; and grasp the interactions among individual agencies, institutional structures and community settings within which agencies operate, in order to advocate for benefits to traditionally underserved communities (Jennings, 2018; Jennings and Dawson, 2015).

By employing participatory methods, CE involves multiple co-learning cycles that emphasize listening, critical dialogue, action-based reflection, cultural humility, community-based knowledge, and exchange of knowledge. Effective CE elicits a community’s strengths, autonomy, and capacity to solve community problems, as opposed to focusing on deficits (Trickett et al., 2011). In order to develop collaborative, equitable partnerships in all phases of research, CE must be informed by the principles of CBPR, which promotes a bottom-up approach (Wallerstein and Duran, 2006). Collaborative partnerships help to ensure that research is relevant within the context of a community’s resource constraints (e.g., staffing, time, finances). By bringing local partners into the discussion, participatory methods can also help researchers grapple with the ethical and practical challenges of balancing scientific rigor with community preferences and priorities (DiStefano et al., 2013; Rhodes et al., 2010).

GAPS & OPPORTUNITIES IN COMMUNITY ENGAGEMENT TO ADDRESS THE OPIOID EPIDEMIC

CE is shaped by the social-ecological and historical context in which it occurs. Designs that rely on engagement of stakeholders occupying positions of power (e.g., providers, administrators, political leaders) may inadvertently limit participation as Black, Latinx, American Indian, and some Asian/Pacific Islander communities are not adequately represented in the medical, dental and nursing workforce (Sullivan, 2004). The need to establish specialized programs for treatment and recovery may be the consequence of long-standing inequities in insurance availability and healthcare access. “Selecting” neighborhoods based on drug-related arrests or high mortality rates may unduly locate the problem exclusively in communities of color. The influence of structural racism and inequities on CE cannot be readily estimated or eliminated through measurement, research design or statistical controls. CE strategies that fail to incorporate community-specific priorities, values and competing interests may inadvertently perpetuate structural and systemic inequities in their selection and implementation of EBPs, services and policies. This is demonstrated by the low uptake of buprenorphine and other forms of MOUD among Black and Latinx populations, compared to non-Latinx White populations (Lagisetty et al., 2019). In light of structural and systemic inequities, transformative CE models and processes are needed to enhance EBP uptake among communities of color and underserved populations. Given the major role of social determinants in shaping the opioid overdose epidemic at a local level, it is critical to engage diverse community stakeholders to contribute their understanding of how stigma, discrimination and inequities prevent access to prevention, treatment and recovery services (Centers for Disease Control, 2011).

A meta-analysis of diverse CE interventions across health conditions found evidence of small to medium effect sizes for health behavior outcomes, health consequences, self-efficacy and perceived social support, but insufficient evidence for any particular CE model (De Weger et al., 2018; O'Mara-Eves et al., 2015). CE’s mixed findings underscore the need for further research to identify optimal CE strategies for heterogeneous communities and priority populations. Few CE models have been developed for the prevention of SUD, and treatment research utilizing CE remains minimal. Individual-level interventions for OUD have typically involved research with healthcare professionals to improve prescribing and service delivery, without utilizing CE models and other participatory processes to address racial disparities in healthcare access and delivery (Williams and Rucker, 2000). Models such as Communities That Care (Fagan et al., 2011) demonstrate effective engagement of families and youth, but the variations in the quality of EBP implementation associated with intervention warrant further study. Recent studies describe ways in which CE has been used for specific OUD treatment, including a “hackathon” event to identify technology-based opioid related interventions (Srinivasan et al., 2019), mapping pathways to recovery within a community (Adams, 2020), and the development of a community partnership to identify potential activities to address OUD that resulted in a community forum (Palombi et al., 2017).

Current gaps in research include the need for better operationalization and measurement of CE processes (e.g., how to implement CE) and methods to effectively approach and engage stigmatized or underserved communities to enact meaningful change (Cyril et al., 2015). Furthermore, limited research exists on the magnitude of CE needed to affect OUD and other health outcomes (Khodyakov et al., 2011). The failure to adequately engage underserved communities or address social determinants may have catastrophic consequences, exacerbating inequities and increasing mistrust of researchers as interventions disproportionately benefit communities with greater privilege.

COMMUNITY ENGAGEMENT GOVERNANCE AND ETHICS

Successful implementation of EBPs across a community requires multiple layers of engagement both within and across complex organizations and stakeholders (Trochim, W. M. et al., 2006). Through continual cycles of reflection, clarification and affirmation about the purpose and quality of a given partnership, stakeholders have the opportunity to negotiate participation roles, expectations and goals that can be, and often are, tested and adjusted as tensions and conflicts surface. CE requires consideration of the complex relationships between leadership and staff within each organization (Klein and Knight, 2005), as well as competition and power dynamics that might exist between coalition members (Trochim, W. M. et al., 2006). In community-facing projects, effective CE has been strongly associated with leaders who value and are skilled in facilitating open communication and collaborative problem-solving (Shea et al., 2017), and who cultivate legitimacy, fairness, competency and accountability (Lavallee et al., 2014).

Coalitions can be used for transformational CE that is community-driven, -owned, and - managed (Bowen et al., 2010). The life course of such coalitions include three developmental stages: Formation: initial clarification of purpose, with a focus on building rapport and defining functions, roles, membership composition, and recruitment; Operation: establishing operating principles and procedures, with a focus on choosing a leadership and decision-making structure; and Maintenance: evaluation and sustainability (Newman et al., 2011). Coalitions employed for planning, executing, adapting, and implementing health promotion research (Wallerstein et al., 2015; Wallerstein et al., 2011) are cited as a common and effective implementation strategy to support the adoption of EBPs and advocate for policy changes (Powell et al., 2015). Coalitions are comprised of a broad range of stakeholders with shared goals, who engage in policy learning through the selective interpretation and use of data to exercise power and influence over social and policy change. Stakeholders may include government officials, scientists, academics, people with lived experiences and community members. Advocacy coalitions, such as the Harm Reduction Coalition, Coalition to Stop Opioid Overdose, Advocates for Opioid Recovery and Fed Up are examples of coalitions that collaborate to outline policy priorities and strategies to challenge stigma and discrimination related to SUD, remove barriers or facilitate equitable access to harm reduction and recovery support services, promote EBPs, and ensure adequate Medicaid or other insurance coverage to enhance health outcomes. In another example, coalitions have been responsible for developing and expanding community-based overdose prevention programs that provide training, capacity building and evidence generation to support the broad distribution of naloxone to laypersons to prevent opioid overdose (Centers for Disease Control, 2012; Wheeler et al., 2015).

The roles of researchers in CE

Compared to traditional research, CE involves community members earlier and more integrally in the research process. This often encompasses practice-based research, which incorporates local expertise, history of relationships, and shared decision-making for the identification of relevant research questions, and by maximizing involvement throughout data collection, analysis, interpretation, and dissemination of knowledge generated for social change (Paez-Victor, 2002; Westfall et al., 2009). Involving communities throughout all phases of research ensures greater accountability, redefines expertise and ownership, facilitates democratization of knowledge sharing and policy making, and is directly responsive to the needs of the community (Reynolds and Sariola, 2018).

Central to effective CE is the process of building trust between academic and community researchers to encourage participation among underrepresented groups and enhance the relevance and uptake of EBPs (Holzer et al., 2014). This also demonstrates respect for community values, priorities and interests, redresses past harms, minimizes long-standing mistrust and risk of exploitation, resolves existing differences in power, privilege and positionality, allows for all community voices and experiences to be represented in the production of scientific knowledge and ensures that research is relevant and impactful for academic and community researchers alike (Reynolds and Sariola, 2018).

CE research may be initiated by researchers or communities (Mansergh et al., 1996). Irrespective of who initiates, academic and community researchers must be willing to relinquish power at times and rely on other partners’ expertise (Head, 2007). In researcher-initiated CE, power is shared when academic researchers consider how engagement activities intersect with existing power structures, social relationships and broader structural forces; how CE activities (re)shape relationships and everyday social life; and how CE objectives are perceived and achieved (Reynolds and Sariola, 2018). Researcher-initiated CE tends to be more defined, specific, and time-limited. Community-initiated CE tends to be longer term and wider in scope; and is most successful when researchers provide technical assistance and evaluation to help shape programmatic efforts (Mansergh et al., 1996). Researchers must also recognize that CE is resource-intensive, and greater training is needed on engagement and research methods for academic and community researchers, respectively (Luger et al., 2020).

Ethical considerations for CE research

Ethical frameworks founded on the concepts of autonomy, beneficence, non-maleficence and justice (Gillon, 1994), do not adequately address unique ethical dilemmas that arise in CE research, and Institutional Review Boards do not use a standardized approach for reviewing CE research (Silverstein et al., 2008). Consequently, standards for ethical CE research require further contextualization and may be guided by the 4R Framework (Respond, Record, Reflect, and Revise) (Chou and Frazier, 2019).

Given the centrality of the community in CE research (vis-à-vis the individual in biomedical research), there is need for ethical frameworks that are not exclusively grounded in personal autonomy (Baylis et al., 2008; Callahan, 2003). When CE research is driven by community partnerships, researchers must navigate the ethics of cultivating a relationship based on trust. The lack of diverse representation in the scientific and research workforce is concerning as scientific inquiry and ethics are self-regulated by research professionals and scientific bureaucracy (Brandt, 1978; Gutrel, 2014). Furthermore, the protection of rights and welfare of human subjects are often reflective of social values (e.g., structural racism, stigma), causing ethical reviewers to inadvertently overlook unique considerations of CE research with people with SUE) (Souleymanov et al., 2016).

Participating stakeholders represent collective community perspectives and various power differentials in terms of age, gender, race, and/or lived experiences may influence decisions. Moreover, decisions on how often to meet and compensation may differ for professionals versus community members. Navigating “community consent” is difficult when stakeholders disagree on the importance of a research issue or strategy (Centers for Disease Control, 2011). Furthermore, the common elements of informed consent have different ramifications in CE. If the leadership of a community organization decides to join a study, their employees may not have reasonable alternatives to participation. Maintaining confidentiality is also challenging when coalitions are formed in small communities. Publishing findings that risk further stigmatization of underrepresented communities also highlights ethical risks associated with dissemination.

Researchers often prioritize empirically driven science over traditional, community knowledge and practices that lack the advantage of a strong evidence base. Explicit considerations of culture and traditional healing in intervention delivery, implementation models, and quantitative measurement of cultural constructs is necessary to understand unique risks and protective factors for diverse populations (Burlew et al., 2013). Addressing these ethical concerns is imperative for equipoise in CE research involving dissemination and implementation of EBPs that are known to be effective.

A DATA-DRIVEN APPROACH TO COMMUNITY ENGAGEMENT

A data-driven approach to CE is essential for ensuring the selection, adaptation, implementation and sustainability of EBPs. The ‘evidence’ driving programmatic, structural, and policy interventions and their evaluation relies on observable, quantitative metrics and qualitative descriptions of experiences, viewpoints, and other phenomena. These datapoints are crucial for decision-making, ensuring cultural relevance and evaluation of EBPs and policies. Descriptions and perceived experiences can improve or inhibit the success of interventions (e.g., impact of stigma and discrimination as a function of race/ethnicity, class, sex/gender, criminal justice history, or dealing with concurrent interpersonal violence) and can be used to identify barriers and gaps in the service delivery system and strengthen intervention and prevention efforts. Ongoing collection of these metrics is needed to inform coalition action plans and refine implementation strategies related to the deployment of EBPs.

Systems science as a data-driven approach for CE

The purpose of systems science models is to improve decision making and identify effective points for penetration of EBPs and systems change by assisting stakeholders in making their mental models explicit and by testing hypotheses about system behavior by recognizing unintended consequences of policies (Carey et al., 2015). Systems science includes quantitative and qualitative modeling and analytical approaches that focus on understanding the behavior of complex systems by studying system components and their dynamic interactions at multiple levels (Carey et al., 2015). Systems science is a powerful tool for CE and advances equity by incorporating voices from all stakeholders (including those frequently marginalized) and permitting identification of the community-specific drivers of inequities through the use of local data. When used for CE, coalitions are commonly involved in developing and validating systems science models for the purpose of decision-making (Koh et al., 2018), by synthesizing coalition discussions, developing common agendas, selecting EBPs, and identifying barriers and facilitators of EBP implementation.

Agent-based modeling (ABM) and system dynamics modeling (SDM)

Among systems science methodologies, ABM and SDM have been extensively used in public health research (Koh et al., 2018), but have limited application within the context of the opioid epidemic (Jalali et al., 2020a, b). Both approaches are used to formalize mental models to cope with the complexity underlying public health challenges and uncover novel ways to intervene in the system to solve real-world problems. ABM is used to model systems outcomes by simulating behaviors and interactions of individual agents (e.g., patients, providers, policymakers, organizations) within a responsive environment (Rose et al., 2015). It is particularly apt at modeling interactions among and between agents and the environment.

SDM is depicted as informal conceptual maps or formal simulation models and is useful for revealing and understanding endogenous sources (i.e., non-linear balancing and reinforcing feedback structures) that underlie emergent systems behavior (Richardson, 2011). For example, the causal loop diagram in Figure 1 shows two competing feedback loops. Taken together, this simplified model suggests that the effective application of both Naloxone and MOUD is critical to reducing opioid overdose and deaths. Naloxone use plays a more critical role in reducing deaths in the short term, but expansion of MOUD is needed to reduce overdose deaths in the long run. Moreover, the benefit of OUD mitigation efforts is weakened without stigma reduction efforts, thus creating an entrenched system where overdose deaths cannot be decreased. As research shows, this feedback perspective and the delay between cause and effect is typically ignored in prevailing mental models (Sterman, 2010). Thus, embedding systems thinking in CE research helps avoid the common pitfall of either-or thinking in coalition deliberations. Through stakeholder-driven deliberation and iterative modeling activities, formal visual system pictures and diagrams are developed to help make informal and often flawed mental models explicit. In this process, stakeholder engagement is fostered, which lays the foundation for systems change. For example, the model in Figure 1 can help stakeholder groups recognize components that are related through two important feedback loops around OEND and MOUD that can readily solve some problems regarding OUD and associated stigma. Studying these loops may help change flawed mental models around how Narcan distribution is leading to more drug use. To realize that Narcan can save lives in the short term, we need to expand MOUD referral and retention to diminish OUD prevalence across communities.

Figure 1 –

Example of a Causal Loop Diagram for the opioid epidemic: In the reinforcing loop (R1 - Linkage to MOUD to reduce Overdose), efforts to address OUD leads to more patients receiving MOUD, which means fewer overdoses, less opioid-related stigma, and greater support for OUD programs. On the other hand, the balancing loop (B1 - Stigma Leveling Off Efforts to Address OUD) shows that increased harm reduction efforts such as distribution of Narcan to address OUD may also lead to fewer overdose deaths initially. However, Narcan by itself does not prevent use of drugs, which means more individuals remaining affected by OUD and at risk for further overdose. This, in turn, perpetuates stigma and counteracts the stigma-reducing benefit of the first loop. Because the system is dynamic, the timing and speed of progression in each loop is important. For instance, when Naloxone is widely available and used, overdose deaths can be efficiently averted. However, this may lead to more individuals overdosing again, and in the absence of an equally or more efficient MOUD referral and retention effort, stigma may end up increasing and remaining high over time, further eroding support for efforts to address OUD.

Building consensus

ABM and SDM can be developed through Group Model Building (GMB), a participatory process that involves coalitions in the modeling process (Rose et al., 2015). GMB is conducted by researchers familiar with modeling for the purpose of aiding stakeholders in understanding a complex problem, designing policy interventions and creating buy-in to implement model-based policy recommendations (Vennix, 1999). SDM modeling is particularly appropriate for fostering CE; and SDM modelers have incrementally developed a set of standard scripts to enhance the transparency and replicability of the GMB method (Hovmand et al., 2011). GMB techniques bring a disciplined strategy to the challenges of implementing complex CE social interventions, helping to cope with dynamic complexity, uncertainty and challenges in consensus building among coalition members (Hovmand and Ford, 2009).

Within the GMB discourse, Community-Based System Dynamics (CBSD) is an explicit attempt to involve community members in the SDM process (Hovmand, 2014). During CBSD, community participants express their mental models by contributing to drawing causal diagrams and feedback loop structures. In this process, prevailing mental models, assumptions and policies are challenged by drawing upon insights that are revealed during the modeling process (Hovmand, 2014). CBSD fosters CE by having coalitions examine the underlying system from endogenous or feedback perspectives to formulate insights. Consequently, this facilitates stakeholder empowerment to advocate for systems change based on these insights.

Both qualitative causal maps and simulation models assist in facilitating CE. Simple and self-explanatory concept models during GMB workshops assist in the development of a shared understanding among participants that dynamic behavior is a consequence of the system structure (Richardson, 2013), whereas formal simulation models provide real-time computational tools for co-learning, strategy adaptations and collective action through systems transformation.

Synthesizing and displaying data to drive community decision making

Data are key to effective engagement of researchers and community coalitions. Data provide a common, concrete focus when bringing people together with disparate expertise, positions, knowledge, and politics. The process of selecting metrics and phenomena for which data will be collected and reported is a valuable and insightful process of within CE (Luger et al., 2020). Effective use of data to enhance CE requires thoughtful processing and transformation of raw/underlying data into a format that can be easily consumed by coalitions. For example, visualizations (e.g., graphs, summary tables) can help rapidly monitor or assess the impact of program implementation or policy changes (Dowding et al., 2015; Joshi et al., 2017; Nash, 2020). Although researchers are particularly adept at data analysis, the selection of metrics and dissemination of findings allows for more effective CE (e.g., increasing clarity of presentation of findings, strengthening interpretation and reducing the likelihood of misinterpretation). Community-tailored dashboards that offer data visualizations via the Internet can provide coalitions with convenient access to the latest available information and results (Wu et al., 2020); web browser features that include interactivity to zoom, filter, or overlay along temporal, spatial, or by categorical dimensions can strengthen the impact and utility of visualizations for each stakeholder (Jacko, 2012; Shneiderman, 1996).

Beyond monitoring and evaluation, data analyses and visualizations can provide insights, such as health inequities in a community, when metrics are separated by race/ethnicity, gender and other sociodemographic factors. The implementation of practices and policies can be refined or changed for optimization when data are deployed as part of a rapid-cycle quality improvement process (American Diabetes Association, 2004; Berwick, 1998; Langley et al., 2009). Data can also be used to improve prediction and scenario planning; quantitative and qualitative data can inform systems modeling (e.g., ABM and simulations modeling, described above) to support coalitions’ decisions around what, how, and/or the degree services and policies in a community should change.

Key considerations for a data-driven approach

The quality and success of a data-driven approach is fundamentally shaped by the availability, timeliness and quality of the data collected. Although researchers emphasize validity and reliability, there may be a tradeoff with community members needing to act on data that are more real-time or more “on hand.” Official death certificates are often used by researchers, yet it is not unusual for those data to take more than a year to finalize. However, locally sourced alternatives with less lag (and lower validity and/or reliability) might include records from a community hospital or coroner/medical examiner’s office. Another example is social media data from venues such as Reddit, Facebook or Instagram that cater to people with OUD and their family members. These data have tremendous reach and range, yet the external and internal validity of such findings may be undetermined or limited. Researchers and community members need to undergo the process of grappling and understanding why and when there may be discrepancies between official statistics and locally sourced and/or more real-time metrics. Data-driven CE is exemplified by community members and researchers working together to combine and synergize analyses and visualizations across existing data sources.

Evaluating adoption, implementation, maintenance, reach, and effectiveness of CE research

Randomized control trials (RCT) are widely viewed as the “gold-standard” for studying intervention effectiveness. However, many of the assumptions underlying RCT limit its utility for studying CE. Even if random assignment yields experimental groups that are similar on all contextual variables, randomization does not remove the influence of those variables on outcomes. Contextual differences, from the distance someone needs to travel for MOUD to variation in the local drug supply, are necessarily at the heart of the CE. Heterogeneity of intervention effects associated with community diversity may be obscured by RCT analysis. This issue cannot be resolved by subgroup analysis alone because factors that influence responses to an intervention may not be known in advance. In order to identify individual and contextual influences associated with heterogeneous responses to an intervention, it is useful to get away from the notion of the monolithic “intervention effect”. Rather, what is required are methods to identify and examine associations between processes of engagement and changes in response over time, as well as contextual variables that shape those associations (Feldstein and Glasgow, 2008).

Dynamic study designs for CE research include Multiphase Optimization Strategy (MOST), step-wedge cluster randomized trials and comprehensive dynamic trials. MOST provides a framework for implementing sustainable interventions that are scalable and able to maintain fidelity given resource constraints (e.g., time, money, staffing) (Collins et al., 2007). The three phases of MOST include preparation (identify and operationalize optimization criterion—i.e., reduce the number of opioid overdose deaths by 40% over three-years); optimization (EBP selection); and evaluation (through RCT). If evaluation findings indicate that the optimization criterion was not achieved, the cycle repeats itself, starting with the preparation phase. Step-wedge designs (Hemming et al., 2015) are particularly well-suited to evaluate service-delivery EBPs and EBPs that do not rely on individual-level participation. Step-wedge trials involve random and sequential crossover of clusters (i.e., communities) from control to intervention until all clusters are exposed and are, thus, well-equipped to account for community heterogeneity. Moreover, step-wedge designs are pragmatic for CE research as policies are typically rolled out over a period of time, and outcomes are measured from routinely collected data. Comprehensive dynamic trial (CDT) designs (Rapkin and Trickett, 2005) capture on-going management and adaptation of interventions necessary to maximize and sustain impact. In CDT, community decision-makers seek to evolve an intervention over time, to achieve high effectiveness with minimal time and resources. Interventions are not treated as fixed conditions but as sets of variables used in analysis to indicate exposures to core intervention elements (Rapkin, 2019).

CE approaches to implementing sustainable evidence-based practices

Sustainability, defined as the maintenance of EBP components (Scheirer, 2005), must occur within the political, social and financial contexts that EBPs are implemented (Scheirer and Dearing, 2011). Sustainability is not a single activity, but rather a process that must begin at the early stages of implementation, planned on an ongoing basis and considered integral to the CE process. Sustainability plans are an ethical responsibility of CE governing bodies, service delivery agencies, community organizations, and researchers (Akerlund, 2000; Mancini and Marek, 2004; Substance Abuse and Mental Health Services Administration, 2008). Sustainability activities may include grant writing to allocate future funding, hiring staff to maintain CE operations, support infrastructure needed to continue operations and activities of coalitions, and training activities to support monitoring and evaluation. Moreover, it is beneficial for coalitions to collaborate with researchers to identify sustainability strategies prior to the onset of implementation by developing process and outcome indicators that are specific, measurable, achievable, realistic, and time-bound (SMART) in order to identify appropriate data-driven strategies (Centers for Disease Control, 2015). On-going access to data can inform sustainability, identify successes and failures, and promote engagement of stakeholders for the continued deployment of EBPs.

Sustained change requires navigating multiple layers of engagement within and across complex organizations and stakeholders (Trochim, William M et al., 2006). This can be achieved through change management, the science, research and tools that consider the human side of transformational change (Hiatt, 2006). Applying change management principles to CE can help build momentum for sustained changes by addressing expected resistance upfront and aligning stakeholders as change agents to support adaptation of new organizational practices, processes, and policies.

CE research in practice: A systems science approach to addressing the opioid epidemic

The HEALing Communities Study (HCS) utilizes CE research to support the strategic selection, implementation and adoption of EBPs to reduce opioid overdose deaths by 40% over three years in highly-burdened, heterogenous communities across New York, Massachusetts, Kentucky and Ohio (Chandler et al., 2020). Through the use of coalitions, HCS researchers actively engage with multi-sectorial community members to address local barriers to the implementation of EBPs. HCS is among the first to use systems science and data-driven CE approaches to implement community-specific interventions to address the opioid epidemic in the US (Sprague Martinez et al., 2020). Systems science methods (i.e., GMB, ABM, SDM) are used to synthesize coalition discussions, develop common agendas, and help stakeholders identify and address challenges and enabling factors that may affect the implementation of EBPs to effectively reduce OUD, overdose, and death in local communities. The modeling results facilitate community-engaged action planning. Prioritization, implementation, adaptation and sustainability of EBPs overtime are informed by integrating the perspectives of diverse stakeholders represented within the model structure and feedback loops that underlie the opioid challenge in each intervention county. Furthermore, the data-driven and systems science approach is used in creating demand for EBP and stigma reduction campaigns. Stigma reduction campaigns are tailored to the specific needs and creativity of communities, and vary based on available media channels, readiness and technical capacity of community coalitions (Lefebvre et al., 2020). Community-tailored dashboards, co-created by researchers and coalitions, are used to identify the key drivers of local opioid crises, inform selection of EBPs, and monitor their impact on OUD, opioid overdose deaths, and other related metrics (Wu et al., 2020). This data-driven approach allows for community-informed decision making and ensures that the goals of researchers and community members align. Moreover, this approach can be used to inform, monitor, and evaluate EBPs that redress structural barriers and stigma that affects access to and quality of health and social services for marginalized and stigmatized populations.

RECOMMENDATIONS FOR THE EXPANSION OF CE RESEARCH

The paper’s authors are drawn from a multi-disciplinary team of researchers from addiction medicine, economics, public health and systems science. This paper is intended for researchers and community-based stakeholders who plan to be involved with CE research for the treatment and prevention of OUD and overdose. It is our hope that this paper inspires researchers and community members to increase the use of data-driven CE to address the opioid epidemic.

To promote the use of CE research, we recommend that CE should be guided by the principles of CBPR, which take a bottom-up approach to addressing the opioid epidemic in communities that are underserved and/or underrepresented. Participatory methods involve co-learning, co-listening, critical dialogue, action-based reflection, cultural humility, community-based knowledge, and an exchange of knowledge. Communities and researchers should build on each other’s collective strengths, resources and assets, and balance knowledge-generation to benefit all community partners, while integrating social justice in the deliberation and implementation of EBPs.

Data-driven approaches should be used to inform coalition strategies in an unbiased manner. Systems science offers a helpful set of tools for CE across local agencies, complex organizations, and hierarchical systems by understanding a system’s structure and inter-relationships. To accomplish this, we suggest building relationships across systems, understanding stakeholder roles and interdependencies, and identifying champions to ensure endorsement, collaboration and sustainability of EBPs tailored to the local context and changing circumstances.

The composition of coalitions should include all constituencies and stakeholders, including service providers, government officials, law enforcement, people with lived experiences, and families who have lost relatives due to OUD. Ethical issues must be addressed up front, and researchers must be cognizant of differences in power and relationships among community members. Finally, plans for dissemination of findings and sustainability of EBPs must be considered to continue implementing change after the research period ends.

CONCLUSION

CE research approaches to combat the opioid epidemic are rare, and systems science has yet to be fully used to address the opioid crisis in the US. Greater investment in CE research is needed to ensure contextual, equitable, and sustainable access to EBPs in communities heavily impacted by the opioid epidemic in order to produce a collective vision that results in community action. Quantitative and qualitative data-driven approaches to CE that are grounded in systems science have the potential to ensure adequate saturation and sustainability of contextually and culturally relevant EBPs to significantly reduce opioid overdose deaths. This approach provides a path to redress historical racial/ethnic and socioeconomic inequities in access to OUD treatment and recovery services by ensuring marginalized voices are represented through diverse community coalition membership, and by redefining expertise and power.

Highlights:

• A novel approach to community-engagement to address the opioid epidemic

• This approach addresses community heterogeneity and health inequities

• Contextual, equitable, and sustainable implementation to evidence-based practices is discussed

• A case study of systems science to address the opioid epidemic is included