Medicaid reimbursement for community violence intervention and prevention (CVI): a multi-state policy implementation case study

Abstract

Background

Violence has fluctuated in the United States in recent years. Additionally, policing practices have been challenged, especially in neighborhoods of color. Community Violence Intervention (CVI) programs have emerged as an effective policy to address violence through neighborhood-centered resources, trauma-informed care, and credible messengers, without full reliance on law enforcement officials. However, inconsistent funding challenges the feasibility and sustainability of these programs. In 2021 several states introduced policies to allocate Medicaid reimbursement for CVI services offering a promising solution to a more sustainable stream of funding.

Methods

This study uses rigorous qualitative analysis to evaluate the implementation of Medicaid reimbursement policies in California, Illinois, and Connecticut, applying the Exploration-Preparation-Implementation-Sustainment (EPIS) model. An analysis of secondary documentation and semi-structured interviews with key stakeholders from the first three states to implement the policy. Stakeholders were recruited from a variety of policy, medical, and non-profit sectors to provide their perspectives and expertise on implementation.

Results

Interviews with stakeholders from policy, medical, and non-profit sectors and a deep analysis of secondary documentation identifies key successes and barriers to effective implementation of Medicaid reimbursement policies across the United States. Acknowledging the barriers of implementation highlights where policy planning and development fails to be properly implemented on the ground. Findings emphasize the need for state-specific policy adaptation, collaboration amongst policymakers and practitioners, and sufficient training for on-the-ground CVI staff members.

Conclusions

Implementation of a Medicaid reimbursement policy for CVI programs could improve the efficacy and sustainability of such programs. However, states need to be aware of the challenges that may arise during the planning and implementation phases. The findings from this study reveal that policy makers, service providers and medical professionals need to be involved and collaborative throughout the planning and implementation process of the policy. States that are planning to implement these policies should assess whether they are ready to implement the policy to ensure that it is successful in the long term.

Background

There have been noteworthy fluctuations in violence across major cities in the United States since 2019, with murders up by 20 to 30 percent in some areas (Grawert & Kim, 2022). The collateral consequences of violence are undeniable, including poorer health, social, professional, educational, and economic outcomes for the injured (Abba-Aji et al., 2024; Kelly et al., 2010; Kilpatrick & Acierno, 2003; Lee et al., 2014; Petrie & Zatzick, 2010; Rich, 2009; Rich & Grey, 2009), and at the community level, poor macro-level health and mental health (Baranyi et al., 2021, Eberly et al., 2022; Finegood et al., 2020) and fewer economic development opportunities (Bowes, 2007). There is a particular negative impact on Black and Latino males (among whom gun violence is one of the leading causes of injury and death) (NNHVIP, 2019; Wintemute, 2015).

Justice system responses, such as proactive law enforcement and punitive sentencing policy, have typically been the primary strategies used to address violence in the United States. However, these have shown to have varying degrees of success (Braga et al., 2019; Cohen & Ludwig, 2003; Sherman & Rogan, 1995; McGarrell et al., 2001). Likewise, the collateral consequences (e.g., jail overcrowding, mass incarceration and supervision, racial and ethnic disparities, poor community-police relations, see Boppre & Harmon, 2017, Weisburd et al., 2019) and mixed empirical findings on how these interventions impact crime (see Stemen, 2017), have invited criminologists and policymakers to consider the potential utility of community-based strategies to reduce violence and heal individuals and communities (Braga, 2022).

Community-based violence intervention (CVI) programs leverage the expertise of local community members and resources to prevent and intervene in gun violence without police intervention (Bureau of Justice Assistance, 2023). These programs can be based throughout the community using violence interrupters (street-based CVI) or within hospital settings (hospital-based violence intervention programs) to prevent and mitigate the long-term effects of community violence. Funding allocations to these programs often confront several challenges. In many cases, funding is not sufficient to carry out all the goals of the programs in the long term, challenging the sustainability and fidelity of the program (Buggs, 2022). Additionally, funding allocations may change due to shifts in political leaders and ideologies on how and when law enforcement should be used. These challenges have pushed stakeholders to seek funding allocations that would allow for a more sustainable approach to CVI implementation.

In 2021 and 2022, several U.S. states introduced and implemented state-wide policies which would enable Medicaid reimbursement to CVI providers and their services. These policies offer the potential for more sustainable funding sources for CVI programs, providers, and clients. Yet, given that CVI programs have only recently emerged in national policy as a strategy for reducing community violence, there needs to be a stronger understanding of how these policies are being implemented on the ground.

This study examines the implementation of this new and promising strategy, Medicaid reimbursement for CVI services (hereafter, “the policy”), via a review of secondary documentation and semi-structured interviews with key stakeholders from three states, California, Illinois, and Connecticut (the first states to implement the policy. Stakeholders were recruited from a variety of policy, medical, and non-profit sectors to provide their perspectives and expertise on implementation. Our findings reveal that the challenges and achievements associated with the policy and its implementation vary by context, and policy recommendations are provided to consider these factors in the future design and implementation of this policy in other contexts.

Evidence supporting CVI programs

The goal of CVI is to reduce violence by leveraging the knowledge and resources that are already present in communities. To do this, these initiatives rely on credible messengers, or trusted community leaders with knowledge of the neighborhood to identify areas and individuals who are most likely to experience or perpetuate violence. Given that violence disproportionately impacts low-income neighborhoods of color, CVI programs place emphasis on empowering community leaders to work together using a coordinated community response (CCR) to promote racial and socioeconomic equity (Ranjan & Dmello, 2022).

CVI techniques have long been utilized in urban areas across the United States. In the 1930s the Chicago Area Project sought to minimize crime by addressing family problems, steering juvenile justice policy away from individual pathologies and towards addressing community-wide challenges using trusted community members (Schlossman et al., 1983). These techniques were adopted in the 1950s and 1960s to reinforce the importance of community expertise and address issues of gang violence and youth violence during times of political and cultural upheaval (Spergel, 1965). These strategies were pushed aside when the Johnson Administration called for increased spending on community policing and corrections, despite warnings of the consequences of over policing in Black, economically disenfranchised neighborhoods (The National Advisory Commission on Civil Disorders, 1967).

In 2020 and 2021 the rise in community violence and pushback on law enforcement shifted attention back to community-centered strategies (Dawson et al., 2023). In 2021, the Biden Administration announced support for community violence intervention and prevention (CVI) programs as a comprehensive strategy to respond to gun violence. The American Rescue Plan Act (ARPA) explicitly enabled city, county, and state governments to use funds to address unprecedented spikes in community violence (U.S. Congress, 2021). These policies led to a swell of federal funding to CVI programs and other community-driven solutions to intervene in cycles of violence following COVID-19 (The White House, 2021; U.S. Congress, 2021).

There are two leading forms of CVI models that operate without police involvement: street outreach/violence interruption models and hospital-based violence intervention programs (HVIPs). Street outreach/violence interruption models, coined as community violence intervention at the roots (CVI-R), utilize trusted community leaders to mediate conflicts within the community that may lead to violence without relying on law enforcement to step in (Pugliese et al., 2022; Slutkin et al., 2015). Additionally, CVI-R programs assist in connecting individuals to services to prevent future conflict. These programs have shown promise in reducing gun violence perpetration and victimization (Buggs et al., 2022; Delgado et al., 2017; Webster et al., 2018), and in improving outcomes associated with the social determinants of health (SDOH), including increased employment outcomes (Corburn & Fukutome-Lopez, 2019; Moreno Leon et al., 2020) and increased drug treatment enrollment among program participants (Moreno Leon et al., 2020).

The HVIP model similarly relies on violence intervention specialists with community knowledge and lived experience (e.g., past violent injury). This model, however, takes place within the hospital setting, wherein trained community members are dispatched to an individual who has suffered a violent altercation before they have been discharged from the hospital. HVIPs have demonstrated positive outcomes among participants in several key areas, including reduced reinjury, weapon carrying, and rearrest for violent crime, and increased post-injury employment and service-seeking behaviors (Aboutanos et al., 2011; Cooper et al., 2006; Snider et al., 2020; Zatzick et al., 2014; Zun et al., 2006). Quasi-experimental and qualitative research studies have therefore been crucial to identifying these individual-level outcomes that occur within the neighborhood. Findings have shown that HVIP participants tend to experience fewer repeat victimizations and hospitalizations, less medical debt and greater usage of crime victims’ compensation programs, fewer symptoms of PTSD, and better health/mental health overall (Chong et al., 2015; Evans & Vega, 2018; Gorman et al., 2022; Holler et al., 2022; Juillard et al., 2016). The growth of promising research on CVIs and pushback on traditional justice system responses have leveraged policymakers to consider funding allocation towards CVI programs to address issues of gun violence.

“Funded to fail:” the status of financial support for community violence intervention and prevention (CVI) and the promise of a new policy

“Many programs are funded to fail. They are funded just barely enough to keep their lights on, and barely enough to have part-time implementation for things like violence interruption and violence prevention and intervention work. So that really was [at] the core of, ‘How do we find sustainable long-term funding…to really put a base on these organizational structures?’” - SID 4 (Connecticut)

CVI service delivery is resource intensive; it often requires staff to go into communities or hospitals around the clock, maintain frequent and sustained contact with participants as they recover, and address the complex immediate and long-term needs of program participants. Programs typically depend on financial support from a constellation of sources, including (but not limited to) philanthropy, private foundations and organizations, and federal and state programs and legislation. While these funding streams provide essential support for CVI, they do not inherently offer long-term sustainability. Grants can ensnare programs into a boom-and-bust cycle that demands overloaded employees to continuously monitor new requests-for-proposals (RFPs), complete what can be lengthy and labor-intensive applications, and report deliverables to remain in compliance with funders. Funding from grantors and philanthropic entities alike can be uncertain due to shifting political climates, priorities, and economic factors. The American Rescue Plan Act of 2021 (ARPA) funding–now an integral support for many CVI initiatives–must be allocated by the end of 2024 and spent by 2026 which is not ample time when considering the numerous tasks that CVI staff take on. Dubbed the “ARPA cliff” (see Brachman & Haskins, 2023), programs are being forced to consider other funding supports as yet another funding stream dries up.

When resources become low, staff layoffs and reductions in service delivery are likely to follow. Violence Prevention Professionals (VPP) and Violence Intervention Specialists (VIS) are examples of frontline workers who have essential roles yet arguably stand to lose the most from unsustainable funding. In many cases VPP/VIS carry out the most dangerous aspects of the work and therefore require considerable time, training, and mental support to do their job effectively. In addition, these workers have survived a violent injury themselves, are from the community they are serving and therefore would benefit from increased organizational support (Ranjan et al., 2023). Sustainable funding could enable CVI programs to retain VPP/VIS, increase their pay, and support their health and wellness while delivering much needed services.

In response to resource constraints more states are considering policies that can provide sufficient and sustained funding for CVI. Much recent attention has been directed towards the use of Medicaid to reimburse CVI providers for violence intervention and prevention services (i.e., “the policy”). Medicaid is a federal program that is designed to bring healthcare to low-income individuals and administered at the state level. It provides coverage for transportation to medical care, inpatient/outpatient hospital visits, home health services, and many other essential benefits (see https://www.medicaid.gov/). Under the policy, VPP/VIS could become Medicaid-certified providers and bill for the time spent delivering case management or crisis intervention services to participants, for example.

This proposed shift in CVI funding follows years of momentum and growing public support. After Medicaid expansion under the Affordable Care Act (ACA) of 2010, a larger portion of the low-income population in most U.S. states was eligible for healthcare coverage under Medicaid, resulting in the program becoming the single largest payer of costs associated with gunshot wounds (GSWs) (Coupet et al., 2018). Since then, the Federal government has expressed its support for the policy via public-facing press releases and webinars (Health Alliance for Violence Intervention [HAVI], n. d.).¹ A recent publication compiled by 135 organizations and individuals (e.g., National Institute for Criminal Justice Reform, Urban Peace Institute, Community-based Public Safety Collective) identified the policy as a key component to building capacity within the CVI landscape (CVI Action Plan, 2024). The HAVI has promoted the policy as an avenue for advancing equity and addressing the compounding harms of a high rate of gun violence and low rate of insurance for young Black Americans (Fischer et al., 2021).

There are some areas where the policy has been successfully implemented at the city-level. Healing Hurt People (HHP) is a hospital- and community-linked violence intervention program based in Philadelphia that provides proof-of-concept for the policy at the city level. According to their website, HHP provides “an integrated care model of trauma focused healing services…to survivors of violent injury…or witnesses to such violence between the ages of 8 and 35” (see https://drexel.edu/cnsj/healing-hurt-people/overview/). In 2018, HHP became the first CVI program to bill Medicaid for violence intervention services via Community Behavioral Health (a division of the Department of Behavioral Health and Intellectual disAbility Services and a Medicaid-managed care hub). In addition to supplemental funding from the Pennsylvania Commission on Crime and Delinquency, the Robert Wood Johnson Foundation, and the Drexel University Dornsife School of Public Health, HHP’s frontline workers bill Medicaid for the services that they provide to participants.

Previous reports and studies highlight the complexities of implementing Medicaid policies and programs statewide (Crable et al., 2022a; Kenney et al., 2016; Weissert & Goggin, 2002). Implementation of statewide policies often requires years of advocacy, planning, and compromise, and sometimes the final product and the original vision/purpose no longer align. Other times implementation must move quickly due to short windows of opportunity that may shift over election cycles and legislative sessions. The current push for CVI programs across the United States has opened one such window for the policy. At the time of conducting this research, the policy was passed in only three states at the outset of the study in 2023 (CT, CA, IL) and has now passed in seven (MD, OR, CO, and NY) and is anticipated in several more. Despite knowing the potential intricacies, there is no published research (to our knowledge) specific to the policy’s implementation. This is a critical gap given that CVI programs have become a nationally recognized policy (The White House, 2021) yet these programs have known resource and sustainability deficits. As will be explored in detail here, policymakers understand that the passing of the policy and the on-the-ground implementation of the policy are two entirely different achievements. Both processes can be complicated and therefore require further scrutiny. Detailed guidance that is rooted in an implementation science framework is vital for success in states that are considering or have already passed the policy.

The current study

Data and method

The goal of our study is to provide a thick description of the policy and its implementation across the United States, providing a level of detail that allows readers to envision how the policy would unfold, or is unfolding, within their own state, and to use relevant information to inform initial and ongoing implementation efforts. Connecticut, Illinois, and California were selected as the study areas because they were the first to pass the policy and therefore have the most developed implementation. We used a qualitative case study methodology, which allows for a multi-modal research strategy to investigate a phenomenon (here, a statewide policy shift) in depth and within its natural setting (Yin, 2009; Priya, 2021). This methodology is useful for exploring many facets of policy implementation, including how these may differ across context. Case study designs have proven useful in other qualitative research studies on large-scale changes in policy or practice (see, e.g., Delcher et al., 2023; Rengifo et al., 2017).

We approached the current study from the emic perspective, which is a research orientation in which the “insider's” or “informant’s” views of reality are elicited, and emphasis is placed on “native or respondent categories and meanings” (Morey & Luthans, 1984, p.29). In line with a confirmatory paradigm (Guba & Lincoln, 1994) and an inductive approach to qualitative analysis, we aimed to elicit information and themes on how the stakeholders as expert informants have experienced the policy and its implementation in their state. After this elicitation was completed and data collection reached saturation, we situated our discussion within the EPIS (Exploration-Preparation-Implementation-Sustainment) framework to offer a clear set of policy recommendations for practical application (see Discussion).

Data collection

Data collection was completed in two stages. The first stage (formulation via documentation review and pre-sample meetings) was preparatory in nature; it positioned the research team to approach data collection in a well-informed and strategic manner. The second stage included the bulk of data collection (e.g., semi-structured interviews with a purposive snowball sample). More details on each stage are provided below.

Stage 1

Stage 1 (formulation) had three components: 1a) the identification of a sampling frame, 1b) fact-finding about the policy through a review of documentation, and 1c) pre-sample meetings. In stage 1a we determined the sampling frame by consulting with the HAVI (Health Alliance for Violence Intervention), a national advocacy organization that works closely with states to implement the policy. In stage 1b we conducted a review of secondary documentation (e.g., laws, legislation, policy documents, published and unpublished research and evaluations, expert conference presentations, committee reports, news features, publicly available data) to determine the planned components and scope of the policy.

Stage 1c included a one-hour, unstructured pre-sample meeting with one high-level stakeholder in each state to elicit information that would help determine the content of a semi-structured interview guide for the full sample (see stage 2). These individuals were either named by the national advocacy organization or were cited in recent reports and news stories as having played a pivotal role in the policy’s passage or implementation. During these meetings, we introduced the purpose of our study and asked the respondents to share any information they deemed relevant. All three stakeholders chose to recount the history of the policy in terms of its enabling legislation and described their respective role during this process. The stakeholders also focused on identifying key collaborators and enumerating the most impactful barriers to and facilitators of the policy’s passage and implementation. At the end of each meeting, we requested additional sources of secondary documentation and names of potential interviewees from their state to form a purposive snowball interview sample. With the information gleaned from these meetings (and in line with the EPIS framework, see Discussion), we constructed a semi-structured interview guide² to be used during stage two of the data collection process.

Stage 2

Stage 2 included one-hour, semi-structured interviews with policy stakeholders, which were recorded and transcribed using a secure virtual platform. We followed Montclair State University’s (MSU) guidelines for informed consent documentation and the processes set forth for conducting minimal-risk research.³ The semi-structured interview guide is provided in Appendix A.

We first interviewed those who were likely to have a high-level understanding of the policy and its implementation (e.g., collaborative committee members, advocacy organization representatives), and then those who were considered “frontline” implementers (e.g., Medicaid administrators, HVIP and community organization staff). The first three interviews were more structured than those that came after as the research team adjusted its approach to permit the full sharing of applicable knowledge given time constraints. We shifted to less structured interviews where stakeholders had more space to speak freely, eliciting additional information or guiding the conversation towards the interview guide questions as needed to fill gaps in information and understanding. This less structured approach allowed the emic perspective to emerge and fostered a much richer and more nuanced set of results that better achieved the goals of our study (Priya, 2021; Swain & King, 2022).

Upon completing each interview, we elicited from the stakeholder additional policy documentation and the names of other potential interviewees. Specifically, we requested the names of individuals from organizations or entities that were mentioned during the interview, and from groups of stakeholders who were not yet well-represented in the sample. This process continued until we reached a level of saturation in which few new insights were obtained during ongoing data collection (Guest et al., 2006). This sampling method yielded a pool of 50 individuals by the end of the project (17 of whom were interviewed). Eleven individuals invited for an interview opted out due to scheduling difficulties (n = 2) or they did not respond (n = 9) to the invitation.

Analysis

Recorded interviews and transcripts were first analyzed for emergent categories within states. Categories were identified by hand by one study author and reviewed and revised by another who was present during the interviews until both agreed that the categories were accurate and complete. Information from the documentation review was solicited both before and after the interview portion of data collection, to both prepare for the interviews as well as fill in gaps in knowledge and context afterwards. Reliability in the quantitative sense was not our research goal; we aim to present transferable results via thick description and dependability.

Dependability refers to the consistency of the findings across research participants and over time and relies on the evaluation of findings from research participants (Anney, 2014). We followed a member-checking process to ensure dependable results that are both valid (i.e., trustworthy) and confirmable (i.e., traceable to the source) (Birt et al., 2016). We used Qualtrics to facilitate this process, which has not been used for member checking in prior qualitative research to our knowledge. In the Qualtrics survey we listed each of the categories from the stakeholders in a given state and asked them to 1) affirm or refute the accuracy of the categories, and 2) to provide additional detail or context as needed. We then grouped the categories across states to produce overarching themes (presented below). Data collection, analysis, and member checking continued until saturation. In this way, our analysis began from the first interview and informed the continuation of the project (e.g., the interview questions changed based on the findings and were unique to different stakeholders depending on their roles, location, and knowledge base).

Results

Sample

We drew information from a vast array of data (via ongoing documentation review and semi-structured interviews) to assess implementation in the first three states to pass the policy: Connecticut, California, and Illinois. We conducted in-depth interviews with 17 individuals who have had varying degrees of direct experience in the policy’s design, passage, implementation, and usage. This quantity is in line with similar qualitative research that seeks to achieve saturation (Delcher et al., 2023; Guest, 2006). Interviewees recruited for the study include frontline medical practitioners, non-profit program directors and staff, elected government officials, and policy directors, consultants, and analysts. These individuals represented hospitals and HVIPs, governmental bodies, advocacy organizations, Medicaid administrations, CVI and community mental health programs. The interviewees’ range of positions, experiences, and views provided an unprecedented holistic approach to understanding the challenges and achievements of policy implementation.

An analysis of secondary documentation was completed to better understand the implementation context and characteristics of the policy in each state and to help triangulate the findings and fill gaps in knowledge. This review included an analysis of gray materials such as state laws, bill history, media publications, webinars, conference presentations, reports, news features, and publicly available data. The review suggests that these policies were all introduced, reviewed, and implemented between 2021 and 2022, with California’s policy being the most recent (implemented in June 2022). Table 1 below highlights by state the guidelines on who qualifies for the Medicaid-covered benefit, the requirements for Violence Prevention Professionals (VPPs)/Violence Intervention Specialists (VIS), and an overview of how costs will be billed to state Medicaid centers. The between-state variation in Table 1 highlights the contextual differences that may influence implementation.

Table 1 Policy information by state

The results that follow begin with an overview of the case profiles for each state, setting the stage for their implementation contexts. These are followed by a summary of the themes that emerged from the semi-structured interviews and review of documentation. Across the three states, 22 categories emerged (seven in Illinois, six in California, and nine in Connecticut). For space considerations and given the sheer volume of results, detailed descriptions of each category (organized by “case” or state) are provided in the supplemental materials. Here, we provide a higher-level view of the policy and its implementation and prioritize the most salient findings by presenting overarching themes.

Case profiles

Case profile #1: Connecticut

Connecticut (CT) has a population of 3,606,000, where much of the population is concentrated in mid-size cities like Bridgeport, New Haven, and Hartford (all with populations between 100,000 and 200,000 people) (American Community Survey, 2022). The majority of Connecticut’s population identify as white (more than 80%) and 13% identify as Black or African American (U.S. Census Bureau, 2023). Notably, Connecticut has the second largest socioeconomic divide in the United States, not including D.C. and Puerto Rico, with a GINI score of 0.5, closely following New York which has a GINI index of 0.52 (Statistica Research Department, 2023). Evidence of this divide continues to grow. A recent report showed an increase in homelessness in the state for the second year in a row (Monk, 2023). These shifts in poverty and homelessness appear to interact with fluctuations in gun violence which also increased during and following the COVID-19 pandemic (Smith, 2023). Most rising trends of gun-related homicides are focused in the largest Connecticut cities pushing policymakers to consider alternative methods of violence prevention in mid-sized Connecticut cities.

The Medicaid reimbursement policy was introduced into the Connecticut state senate in January 2021 (AB 1929). The policy took effect in June 2022 (Haigh, 2021).

Case profile #2: Illinois

Illinois (IL) has a population of 12,582,000. Chicago makes up 2.5 million of this population. Other larger cities in Illinois are Aurora, Joliet, and Naperville all with populations between 100,000 and 200,000 residents. Issues of gun violence are largely focused on Chicago (Mason et al., 2023). While there have been several community and police-related policies aiming to address issues of gun violence in Chicago since the 1990s, gun violence has still not been stabilized. In 2021 over 750 people died from a gun-related incident in Chicago (Howe & Boyle, 2022) the highest number of gun violence deaths in the last 25 years (Wolf et al., 2021).

CVI programming has long been prominent in Illinois to address fluctuating trends of gun violence in Chicago and other urban areas of Illinois. The first CVI-based programs were developed and implemented in Chicago in 2000, which became popularly known as the Cure Violence initiative. Since this initial intervention many CVI-related programs have been implemented throughout Chicago and other parts of Illinois.

The Medicaid reimbursement policy or the Community Health Worker Certification and Reimbursement Act was introduced in January 2021 and passed in April 2021. The bill focuses on addressing barriers faced by community providers, especially those who are serving neighborhoods of color.

Illinois stakeholders come from diverse professional backgrounds including medical centers, academic institutions, and small local non-profit organizations. Additionally, stakeholders responded from varying levels in their organizations ranging from Director positions to staff members at local HVIPs in Chicago. This diversity in stakeholders allows for a more comprehensive understanding of the multilevel implementation of the policy. While the majority of the interviewees come from Chicago, stakeholders pointed out a need for more policy interventions in other areas of Illinois.

Case profile #3: California

California is the largest of the three states, with a population of 39,965,000. Unlike other states, no single race constitutes a majority of California’s population (39% Latino, 35% white, 15% Asian American). Most of the population is concentrated in major cities like Los Angeles, San Francisco and Sacramento. In the 1990s, California’s gun violence rates stood out among the rest of the states, showing an average that was three times higher than the national average of shootings among young people (Office of Gun Violence Prevention, 2023). Like the rest of the United States, California experienced increases in gun violence between March 2020 and January 2021. Before the COVID-19 pandemic, gun violence rates in California were at a record low (Office of Gun Violence Prevention, 2023). Despite fluctuations in gun violence over the last three decades, California has managed to keep gun violence somewhat contained since the 2010s.

The policy passed in California in August 2022 and became effective immediately. The focus of the policy is to ensure that community violence prevention services are provided by qualified professionals and covered under Medi-Cal, or California’s state Medicaid program. Unlike other states, billing through Medi-Cal occurs at the county level through Medicaid administrators known as managed care programs or MCPs, while Illinois and Connecticut administer benefits at the state level.

Themes

Six themes (see Table 2) were generated from the 22 state-level categories (see Appendix B for a list of categories by theme and state, and supplementary materials for thorough descriptions of each category). In the paragraphs that follow we describe each theme in detail, using case examples from individual states and illustrative quotes as helpful. The purpose of this section is to provide thick description and transferable knowledge about the policy and its implementation as generated through an inductive coding process. Policy recommendations stemming from these themes are situated within the EPIS theoretical framework and described in the Discussion section to follow.

Table 2 Summary of themes

Theme 1 (“Communities/CBOs”): policymakers, community members, and CBOs in the CVI space have complex dynamics and do not share equal access to designing and implementing the policy, which has implications for gaining access to the policy

This theme includes categories relating to the relationships between community members and/or community-based organizations and the policy or its implementation.⁴ Interviewees from all three states described complex dynamics between community members, community-based organizations (CBOs), and those designing and implementing the policy. Responses characterized the perceived barriers between communities and the ability to access the policy, but also highlighted efforts intended to reduce such barriers.

The ability of CBOs to access the policy was a primary concern. In many states the provision of violence intervention and prevention services falls into one of two categories: 1) hospital-based (and community-linked) (“HVIP”) or 2) community-based (“CBO”), and these different types of entities do not often have equal access to the policy. For example, some interviewees note that the process of billing Medicaid and receiving reimbursement requires training and infrastructure that some programs do not have access to. Because of their hospital-based setting, HVIPs possess the infrastructure and Medicaid billing experience that many non-HVIP CBOs lack at the time of policy passage. The associated “red tape” of being a Medicaid-certified provider (e.g., billing, compliance, auditing) is a key barrier that many CBOs simply cannot overcome without additional resources and support.

“I think that's one of our issues…[for] a lot of the CBOs there's sort of two barriers…one is like, do they know how to bill Medicaid…The second is, do they have access to a clinician? Because the way that it's written…in California, like Connecticut, is you need a licensed clinician to sign off on the initial referral. And in California you need [a licensed] clinician…to develop the Care Plan…a lot of CBOs don't have access to a physician or have a licensed clinical social worker or licensed marriage and family therapist on staff, so they're going to have to figure that out as well.” - SID 9, California

Even CBOs intent on accessing the policy are frustrated by a perceived lack of training and technical assistance (TTA). In Illinois the frustration extends beyond the Medicaid certification process because their policy's structure requires a “team-based” approach in which agencies providing the frontline services must build relationships and infrastructure with agencies tasked with providing the other requisite team members (e.g., Qualified Mental Health Professionals). In and beyond Illinois, there is an expressed need for TTA regarding the provider certification and enrollment process, setting up Managed Care Organization (MCO) contracts, billing, and compliance.

Findings additionally show that the process of applying and receiving “quick funding” (e.g., grants) may shift CBOs’ focus away from the lengthy process of becoming a Medicaid-certified provider delaying their uptake of the policy. CBOs are typically more familiar with grant applications, compliance, and reporting than with Medicaid processes. Thus, despite the many aforementioned limitations of grant funding, findings suggest that they still provide CBOs a familiar way in which to receive funding.

Beyond the differential access HVIPs and CBOs have to the policy, low-income communities and/or communities of color (i.e., those most affected by community violence, see Centers for Disease Control & Prevention, 2020) may exhibit a high level of mistrust towards government institutions stemming from concerns of institutional racism in which many governments, hospitals, and universities have been complicit. With the policy firmly rooted in a government-sponsored healthcare insurance system (i.e., Medicaid), gaining the trust of communities and the organizations that serve them can be challenging, and any mistrust may manifest as disinterest in or non-support for the policy.

There are efforts to address the challenges experienced by communities and CBOs to improve community support and access to the policy. In California, stakeholders are developing tools including an implementation guide and training webinars (see supplemental materials for links) that include the voices of local and national advocates and policymakers to provide accessible information on the policy. Separately, stakeholders in California aim to establish a community of practice or “intermediary hub,” described as a group of local CBOs who have a history of doing violence prevention work, that will provide the necessary resources to access the policy. An effort in Illinois involves the development of a centralized, multi-organizational training, billing, and certification entity to streamline and speed up these respective processes.

It should be noted that, at the time of writing, very few entities (HVIPs and CBOs alike) have successfully billed under the policy in any of the three states. While some view this as a failure of implementation, others suggest a slow uptake may be desirable during the early stages, as fewer providers billing under the policy lightens the administrative burden on the state Medicaid administrator as potential insufficiencies in the process can be addressed.

Theme 2 (“Design”): the specific design, structure, and components of the policy will impact its goodness-of-fit in context

This theme includes categories relating to the specific design, structure, and components of the policy.⁵ Interviewees reflected upon the policy’s goodness-of-fit in their context and suggested ways in which it might be improved upon. They emphasized a particular need to consider who should have access to the policy, who qualifies for services under it, and how the reimbursement process should be designed around the use of a relatively new type of provider: the violence intervention specialist/violence prevention professional (VIS/VPP).

“I think we may have said ‘community-based’ one too many times in the legislation because they excluded hospitals from even being allowed as providers. So the only hospitals right now that can provide the service are those that have a community mental health or a behavioral health center designation…That means like 80% of the hospitals who already do this work don't qualify. That is getting changed…but we're still working through that, as the Legislature ends in a couple of days.” - SID 3, Illinois

During the early implementation stakeholders limited access to the policy to pilot the model and reduce confusion prior to expansion. Stakeholders in Connecticut and Illinois did this by limiting the types of organizations that could access the Medicaid-covered benefit (e.g., only HVIPs and CBOs that are hospital-linked, or only members of a Violence Prevention Community Support Team [VP-CST]). While this practice is often used in policy implementation, many doing CVI work in both states expressed disappointment, feeling effectively “shut out” during these initial stages of limited policy implementation.

The policy outlined the qualifications required for becoming a certified VPP/VIS which differed between states (see Table 1). These requirements became an additional point of concern. These qualifications often required considerable training, the employment of trained medical personnel, or demonstrated experience working in VPP and VIS positions. To ensure that these qualifications were being met, states attempted to facilitate training for CVI staff by providing multiple pathways in which individuals could receive certification. Illinois, for example, enlisted a well-established HVIP to tap a statewide coalition for training examples, which yielded two certification pathways in addition to the Medicaid administrator’s extant community health worker (CHW) certification. Under this model VPPs may become certified under any one of these three pathways, and the state Medicaid administrator retains the authority to recognize other certifications in the future as they become available. California also allows multiple certification pathways, and Connecticut relies on the training model designed and delivered by the HAVI (see Table 1).

Another key factor of the policy is the billing and reimbursement structure. There was room for discretion in this regard as none of the three states had a system in place for reimbursing VPP/VIS for violence intervention and prevention-related services. Connecticut opted for a time-based reimbursement model for the sake of simplicity while implementing this new service provider. Under this system, 15-min units of service are used as a way to reduce uncertainty in how various services by VPP/VIS might be defined (i.e., which procedure codes should be used). While not explicitly mentioned in the interviews, review of secondary documents affirmed that California utilized a similar reimbursement strategy as Connecticut.

Illinois opted for a more complicated system of reimbursement that is rooted in a community mental health model. Illinois’ VP-CST model requires three key roles: 1) a team lead that meets the qualifications of a Qualified Mental Health Professional (QMHP), 2) a Peer Support Worker (PSW) with lived experience, and 3) at least one other staff member that meets the qualifications of a Mental Health Professional (MHP) who directly supervises the PSW. For this reason, Illinois’ reimbursement rates appear to be about three times higher than those in Connecticut or California. Some argue that a fee-for-service model in general incentivizes billable productivity over the quality of services and may narrow the scope of billable services (e.g., face-to-face contacts only). Other potential models are noted in the discussion/recommendations section.

Reimbursement rates (i.e., the amount of money billed for each unit of service) are another important consideration. Rates vary from state to state and reflect the norms and practices of each state Medicaid administrator. Stakeholders and service providers are often not in agreement on how these rates are set, and regularly express dissatisfaction with them. Across Connecticut and California it was determined that the reimbursement rates were too low (see Table 1) considering the cost of startup as a Medicaid provider and the complicated, stressful, and sometimes dangerous nature of violence intervention and prevention work. Interviewees in tune with Medicaid administration suggested, however, that starting out with a low reimbursement rate for a new provider is somewhat expected, and that there is room for negotiation and advocacy down the line such that rates will be raised incrementally and become a feasible incentive for providers. Furthermore, the money organizations receive from reimbursements is better viewed as a budgetary supplement and not a complete funding stream.

In addition to VPP/VIS qualifications and reimbursement, states must determine who will qualify for services under their policy. These qualifications vary across states. To receive services from a VP-CST in Illinois one must have a) been a victim of violence or b) have “chronic exposure” to violence, and demonstrate medical necessity according to the Illinois Medicaid Comprehensive Assessment of Needs and Strengths (IM + CANS) instrument. To receive services under the policy in California one must be a victim of violence or “at risk” of violence victimization. Policies do not tend to explicitly define these categories, which is considered a double-edged sword: specificity in language may facilitate the development of billing codes and policies later on; vagueness, however, may leave the primary decision-making to the state Medicaid administrator as opposed to legislators, who may have gaps in knowledge about the needs and capacity of communities and providers.

One thing is certain about the policy’s design, structure, and components: they will (and should) change over time, particularly as states move from initial implementation into sustainment. With a new category of provider and oftentimes new service categories or structures, Medicaid administrators anticipate a need to continuously monitor and adapt. Small adjustments can typically be made ad hoc, though larger changes (e.g., significant rate increases, expansion of covered services and providers) will likely require higher level approvals such as state statute amendments and/or Federal approval. Stakeholders are correct to assume that these changes may be time consuming and/or difficult to achieve. Some stakeholders associated with Medicaid administration offer that the meaningful inclusion of frontline providers, advocates, and other community stakeholders in the policy design process may prevent the need for larger adjustments after implementation. There is further opportunity to engage these groups when states inevitably have to design amendments to the policy.

Theme 3 (“Collaboration”): designing, passing, and implementing the policy requires collaborative, multi-stakeholder efforts and advocacy

This theme includes categories relating to collaborative, multi-stakeholder efforts to advocate for and/or implement the policy.⁶ Interviewees from Illinois and Connecticut summarized their efforts to recruit and include feedback from a broad group of stakeholders. These efforts were rooted in a desire for stakeholder inclusivity, which at times contributed to challenging dynamics during the policy design and implementation processes.

“...We have some of these structures, like our Commission on Gun Violence Intervention and Prevention. There is, I would say,…good conversation, and there is very good collaboration…meaning that all the organizations that I've mentioned, both [violence] interruption groups and…’pure’ HVIP groups, and others, are all kind of at multiple of the same tables...We have yet, as I said, to get really over that next hurdle of really getting solid reimbursement done by community organizations, and I will be interested to see kind of where it goes. But the communication and the collaboration is there.”

-SID 4, Connecticut

Connecticut, the smallest of the three sites and the first to implement the policy, underwent a lengthy collaborative process leading up to its passage. During this time, two key pilot HVIPs tracked and shared their data with a small group of stakeholders. From this relatively informal effort grew the CT HVIP Collaborative, a broader convening of stakeholders prior to bill passage that had an initial focus on philanthropy, though later shifted to pursuing a Medicaid-covered benefit. Once it was considered politically viable, an advisory committee was formed to advocate for the policy (under HB 5677, and with support from key “legislative champions”⁷). Once the bill was passed, the advisory committee disbanded and has since developed into the Commission on Gun Violence Intervention and Prevention: a formal working group of state and national stakeholders. This Commission continues to play an active role in crafting the policy and has subcommittees that are focused on sustainability and long-term implementation.

Stakeholder inclusivity was at the forefront of these collaborative efforts, resulting in participation from clergy, community-based organizations, municipalities/law enforcement, philanthropic organizations, advocacy groups, and youth groups. Having a diverse set of stakeholders fostered overall buy-in for the policy, though also brought to the table many unique perspectives, and existing relationship dynamics between stakeholders who have long coexisted in the CVIPI space. Some, for example, may have long standing relationships with specific legislators and may have competed with others in the group for grant funding and in serving clientele from the same community. Tensions between stakeholders can contribute to a power struggle that delays implementation or reduces buy-in. Like Connecticut, Illinois also convened a semi-formal collaborative work group of local, state, and national stakeholders, including street outreach workers and social service providers, to brainstorm various design elements of the policy prior to its legislative passage. Only a select few, however, were directly involved in the legislative and advocacy efforts, leaving the others (and particularly the hospital-associated stakeholders) to feel as though the bill language did not include their perspectives, and became too focused on the work of community-based organizations. Because of this, some stakeholders were uncertain of the merit of the policy from its infancy.

Theme 4 (“Context”): state context (e.g., political, social, resource) has important implications for the policy and its implementation

This theme includes categories relating to the contexts under which the policy was implemented.⁸ The case profiles presented in the beginning of the results section include statistics from secondary administrative data and a review of literature, however the information detailed below reflects what interviewees from Connecticut and California shared. It includes descriptions of the contextual elements that stakeholders were perhaps most concerned with during the implementation process.

“...Frankly, we have the blessing of being a very small state geographically…You drive 90 [minutes] in any direction from this place in Wallingford, and you're in another state. So we do have that advantage of going for us here...we know it's not the same around the country.” - SID 1, Connecticut

Connecticut interviewees noted the advantages of implementing the policy in a relatively small state, highlighting that Connecticut is the perfect pilot state because of its low number of violent victimizations annually. This translates to a low potential cost of the policy, meaning that these costs could be absorbed rather than going through the administrative burden of adding a new line to the state budget. Connecticut’s small size also meant that many stakeholders involved in the implementation were accustomed to working together on issues of violence. As noted under Theme 3: Collaboration, however, stakeholder familiarity can be a double-edged sword in the presence of “turf considerations,” or the idea that some stakeholders may have a long history of collaborating or receiving support from specific people or entities.

California’s state-level characteristics and unique Medicaid administration structure also affected the policy’s implementation. California is large geographically and, unlike other states, their Medicaid program (“Medi-Cal'') is implemented by county-level managed care organizations (MCOs). Thus, some believe that lessons in statewide policy implementation from other states would apply to their context. Implementation may be unique even between counties, as some MCOs have a monopoly in one county, and MCOs in other counties share the work.

Theme 5 (“Momentum”): the speed with which the policy is designed and passed/adopted can have downstream effects on implementation

This theme includes categories relating to the speed with which the policy was passed and/or implemented.⁹ Interviewees from Connecticut and Illinois noted the importance of balancing the need to capitalize on momentum while also understanding the potential downstream effects of passing legislation without carefully considering all the elements and stakeholders whose voices should be included.

“I think it's fair to say that we went fast on this. We wanted to get out of the gate…we had real, significant support from the Governor and the Governor's office on doing this, they just kind of “got it.” We didn't build a large coalition of healthcare providers that were focused on this. We got it into a violence intervention statute, and so there wasn't a lot of awareness among hospitals that [the legislation] was out there, or community mental health providers. It just kind of landed.” - SID 3, Illinois

A series of events and conditions catalyzed support for the policy in Connecticut and Illinois: 1) the onset of the COVID-19 pandemic, 2) momentary increases in violent crime in many large cities, 3) civil unrest in the aftermath of George Floyd’s murder, and eventually 4) the impending “ARPA cliff” and related concerns about sudden resource shifts. These conditions led to calls for more sustainable and equitable approaches to CVIPI and a desire to treat violence as a public health issue (further explored in Theme 6: Framing). Under these conditions stakeholders from many unique disciplines learned to collaborate to address a broad need–a skill that translated well to working on the policy.

Building on this initial push, regular meetings were held before the legislative passage of the policy. In Connecticut, this included monthly meetings between stakeholders to better understand frontline violence intervention work and structure the policy to fit the state context. These meetings helped to maintain momentum and kept stakeholders responsive to one another, as much of the work on the policy happened in the interim. Public hearings, while potentially labor intensive for the organizers, were also cited as an important venue for building momentum that is rooted in testimony from diverse stakeholders on a range of topics (e.g., the impacts of gun violence and injury, the work of pilot HVIPs).

Some suggested that there may even have been too much momentum, in Illinois and Connecticut in particular, as the policy was passed before many of the logistics of implementation were determined. The speed with which the policy is passed was considered a double-edged sword. In Illinois, the need to produce policy language quickly resulted in only select members of the collaborative working group (see Theme 3: Collaboration) drafting the legislation without first building a larger coalition, primarily because this was perceived as something that would take additional time and perhaps miss the window of opportunity. To capitalize on momentum, the policy was drafted in conjunction with the state Medicaid administrator and included in a larger community violence intervention statute (HB158). In retrospect, the “race to get something done” may have hampered important elements of the legislation. For example, the legislation initially included an explicit call for an implementation planning process, but this was removed as stakeholders feared it would delay the policy. In retrospect and given evidence of a slow uptake after legislative passage, more specific policy language may have been beneficial to consider and include.

Theme 6 (“Framing”): the way the policy and/or the issues of violence and CVI are framed publicly may impact stakeholders’ ability to secure support

This theme includes categories relating to the framing of the policy and/or the issues of violence and violence prevention/intervention.¹⁰ Interviewees in California and Connecticut shared the ways in which stakeholders in their states worked to change and provide consistent messaging around the issue of violence and regarding the policy specifically. Framing became very important for designing and securing support for legislation, both in small working groups and broader audiences, including state legislatures and the general public.

Across the states it was apparent that framing violence as a public health issue was the most appropriate and reflective of the intentions behind this policy shift. During early conversations, stakeholders in California noted that this framing calls upon policymakers to find ways to bring healing resources to victims, to redirect the economic benefit of safety net systems to the communities that they are intended to serve, and to believe in the wisdom of communities to direct their own solutions to the issue of violence. The nation’s first HVIP, Youth ALIVE!, provided a practitioner's perspective during early meetings to break down how funding VPPs aligns with the broader Medi-Cal goals and fits within a public health approach to CVI. Even within a public health lens, though, there are concerns about the medicalization of the violence prevention profession. These concerns relate to the harms that the medical system has brought upon poor communities and communities of color historically.

“You know, many of us have been working for many years, who talk about how the things that drive health outcomes very rarely have much to do with traditional medical model interventions…but we don't seem to know how to do anything other than diagnose and code.” - SID 16, California

In Connecticut, stakeholders also sought to frame gun violence as a public health issue–one which should be addressed through targeting the social determinants of health. This was the impetus for stakeholders to work closely with hospitals and HVIPs, and positioned the efforts toward passing the policy in line with hospital and hospital network priorities. Beyond a public health framing, both the Executive and Legislative branches perceived that they had “something to gain” from the policy given the current social and political context (e.g., constituent support for policies that might reduce gun violence without perceived impingements on Second Amendment rights).

Connecticut policymakers considered it important to develop and maintain consistent statewide messaging before, during, and after implementation. For example, they suggested defining violence intervention within policy language, including which types of organizations fall under this definition. To facilitate consistency, the state's established trade association for hospitals and other healthcare providers took up the task of communicating what was working in one location of the state to another, and also crafted a comprehensive “wish list” for the policy along the way. The central location of this entity made it so that stakeholders could physically meet in a neutral space to share ideas. The participation of key hospital healthcare systems made it possible for HVIP advocates to promote the model in regions of the state where these systems operate other hospitals. National entities (e.g., the HAVI, Giffords) helped to foster an understanding of the national landscape in the policy’s development (e.g., what is happening in other states, what is planned, what is working or not working). These efforts, in combination, contributed to consistent statewide messaging about violence, CVI, and the policy.

Discussion and policy recommendations

Our findings are timely considering oncoming resource shifts in the CVI space (e.g., the so-called “ARPA-cliff”), the dawn of a new Presidential administration in 2024, and momentary reductions in violent crime across major U.S. cities (Lopez & Boxerman, 2024). Now more than ever it is important to consider all viable options for sustainable CVI programming; here, we have focused on the statewide implementation of Medicaid reimbursement for violence intervention and prevention services (i.e., “the policy”). From a review of documentation and interviews with high-level stakeholders across Connecticut, California, and Illinois (i.e., the first three states to implement the policy), six themes emerged and suggest a high degree of concern among stakeholders regarding: 1) the relationship dynamics of community members and CBOs to the policy and its implementation, and 2) the specific components and structure of the policy.

The exploration-preparation-implementation-sustainment (EPIS) model

To frame the discussion, we next provide an overview of the study’s animating theoretical framework: the Exploration-Preparation-Implementation-Sustainment (EPIS) model (Moullin et al., 2019, see Fig. 1). Our semi-structured question guide and policy recommendations are rooted in the EPIS model, which has traditionally been used to frame studies of organizational-level implementation of evidence-based practices. Recent research, however, has extended its utility to the study of statewide policy implementation (Becan et al., 2018; Crable et al., 2022). The EPIS model emphasizes four key components: 1) the phases of implementation, 2) contextual levels, 3) innovation factors, and 4) bridging factors (see Fig. 1). We describe these components below, adapting them to the policy of interest: the statewide adoption of Medicaid reimbursement for CVI services.

Fig. 1

EPIS model (adapted from Crable et al., 2022)

Key component #1: phases of implementation

The EPIS model recognizes that implementation happens in distinct phases, each of which brings its own challenges and considerations. The four phases are Exploration, Preparation, Implementation, and Sustainment. We provide more detail on each of these stages in the policy recommendations, which are organized accordingly.

Key component #2: inner and outer contexts

The EPIS model affirms that within each of the four phases of implementation there are factors that make up the inner and outer policy contexts. The outer context refers to the environment beyond the policy’s direct use and implementation and may include the political, service, and policy environment in the state, as well as rates of violence and public attitudes towards gun violence and CVI (for example). Moullin et al. (2019) reiterate that the outer context also can also reflect the “inter-organizational relationships between entities…that influence and make the outer context dynamic,” which in this study may include governments, hospital networks, training and technical assistance providers, and advocacy organizations (p.3).

The inner context includes the entities with direct involvement in the use of the policy. This may include community- and/or hospital-based violence intervention and prevention programs (including, e.g., their organizational structures, resources, internal policies, staffing practices) and their staff who provide and bill for services. This also may include state Medicaid administrators and managed care organizations (MCOs) which often design, implement, and monitor the structures and processes associated with billing under the policy. The inner and outer contextual levels interact in complex ways to highlight the multilayered nature of statewide policy implementation.

Key component #3: innovation factors

The third key component of EPIS relates to the policy (or innovation) itself, with an emphasis on the policy’s characteristics, developers, and fit within context. This component acknowledges that some adaptations may be necessary in order to improve the policy-context fit while also maintaining its core components. Context includes all of the factors that are listed in Fig. 1 under Inner Context and Outer Context.

Key component #4: bridging factors

The fourth component of EPIS includes “bridging factors;” i.e., the factors that facilitate the interconnectedness and reciprocal influence between the outer and inner context entities, and across the stages of implementation. These factors are essential for the Sustainment phase.

Policy recommendations

Next, we present a series of policy recommendations that are organized by the four phases of the EPIS model (Fig. 2). Recommendations are intended to assist those who are considering or actively implementing the policy in their state and stem directly from the study results. There is an emphasis on recommendations within the Preparation and Implementation phases as a natural consequence of the interviewees’ concerns and recent experiences. Throughout, policymakers should consider their unique state context and the possibility that not all recommendations will apply. Summaries of and additional considerations for the recommendations at each stage appear below Fig. 2.

Fig. 2

Policy recommendations by EPIS stage

Exploration phase policy recommendations

In phase 1 (Exploration), stakeholders identify the challenges that CVI programs experience, recognizing that the existing resources for CVI are inadequate to meet beneficiary needs and sustain programs long-term. Upon identifying this need, stakeholders then 1) decide whether to adopt the policy and 2) consider which adaptations may be needed to fit the policy and its implementation to their context. We put forth three recommendations for stakeholders to consider during this phase:

• 1. Determine the need for and feasibility of the policy. This recommendation stems primarily from Theme 4 (Context), in which interviewees described the nuances of their current CVI funding landscape and infrastructure, Medicaid administration, and the social and political climate around addressing violence. We suggest that stakeholders specifically consider the following: Is there a need for CVI funding beyond what is offered in the current landscape? What are the healthcare and Medicaid structures in your state, and how might this impact the policy’s implementation? Has your state enabled similar policies in the past, and were there any lessons learned? Would the policy require an amendment to the state budget? What would this process entail?

• 2. Identify potential stakeholders and consider how to involve them. This recommendation stems primarily from Themes 1 (Communities/CBOs) and 3 (Collaboration), in which the interviewees described the ebb and flow and oftentimes fragile nature of collaborative working relationships in the CVI space. We suggest that stakeholders specifically consider the following: Who is or has been active in the CVI space (e.g., HVIPs, CBOs, advocacy groups), and what are their strengths and potential contributions? Should efforts regarding the policy include some or all of them? Do potential stakeholders have the following: a) financial or political “firepower,” or b) previous experience with Medicaid policy? Is there a sense of trust and alliance between stakeholders? How can this be developed or maintained? Can a “neutral” leader guiding the policy conversation, host meetings, and communicate between stakeholders?

• 3. Strategically frame the issue of violence and the policy itself. This recommendation stems primarily from Theme 6 (Framing), in which the interviewees noted that the way stakeholders talk about violence and CVI will impact the components of the policy and who it may be supported by. We suggest that stakeholders specifically consider the following: Through which systems has your state addressed the issue of violence (e.g., medical, public health, criminal justice)? Is there a need to change this approach/framing? Does the framing align with current evidence-based practices in CVI (e.g., social determinants of health), the structure of the policy, the groups of stakeholders involved, and the public-facing messaging? Which current events could help spur policy momentum (e.g., crime trends, politics, resource shifts)? Do stakeholders understand what CVI entails theoretically and in practice? Which policy outcomes will be important to track given its framing? We provide a list of potential outcomes in the Conclusion under recommendations for future research.

Preparation phase policy recommendations

Stakeholders move into phase 2 (Preparation) upon deciding to adopt the policy. This decision may come via legislation, administrative rule, or both (in succession). In this phase, stakeholders identify potential barriers to and facilitators of the policy’s implementation and develop an implementation plan. Implementation supports are also identified during this phase, which may include training and technical assistance, consultation from other states or national organizations, and supplementary resources to facilitate the use of the policy over the next two phases (Implementation and Sustainment). The decisions made during Preparation can have strong and lasting impacts on, therefore we put forth the following recommendations:

• 1. Determine whether to enable the policy via administrative rule or legislation (or both). This recommendation stems primarily from Theme 5 (Momentum) in which the interviewees discussed their states’ pathways and timeframes for enabling the policy. We suggest that stakeholders specifically consider the following: Which approach will result in a timelier passing of the policy? More sustainable? Which entities will have more influence on the policy, its implementation, and future amendments under either approach? What are the relative benefits and drawbacks of pursuing either approach (or both)? Can you mandate that an implementation plan be created under either approach?

• 2. Prepare to design the policy. This recommendation stems primarily from Theme 2 (Design), in which the interviewees described the painstaking attention to detail required in designing a policy that is both effective and ‘works’ for everyone involved. We suggest that stakeholders specifically consider the following: Have you solicited guidance and feedback from national experts in CVI and Medicaid policy (e.g., HAVI, Center for Health Care Strategies)? Has an advisory committee been formed to spearhead these efforts? Are there upcoming legislative initiatives that could house the policy, or should it stand alone? Given its implications for billing and other aspects of implementation, how specific should the policy language be (e.g., what constitutes CVI)? What are the relative strengths and drawbacks of a fee-for-service model versus alternative payment structures (see Health Care Payment Learning & Action Network, 2017)? Under which program should the Medicaid-covered benefit be housed (e.g., health, mental health, behavioral health)? What are the downstream implications of this structure (e.g., billing a new code under an extant provider type vs. developing a new provider type altogether)?

• 3. Strategically design the Medicaid-covered benefit under the policy. Like recommendation #2, this recommendation also stems primarily from Theme 2 (Design). Rooted in feedback that we received from interviewees from California, we suggest that stakeholders explicitly consider these five aspects of the benefit, and include them (as appropriate) in the policy’s language:

  • a. Scope. I.e., which codes can be billed, in which setting, at which time, and under whose supervision? Does the definition of medical necessity reflect individual and structural drivers of poor health outcomes (i.e., SDOH)?

  • b. Credentialing. I.e., who can provide the services, and how will they be credentialed?

  • c. Paneling. I.e., How do providers sign up to get paid? At which level will there be National Provider Identifiers (NPIs)—organizational or individual? Under managed care systems, is the benefit scaled in the managed care contract? Are there tools and TTA in place for those navigating this process for the first time (e.g., common MCO contracts)?

  • d. Payer. I.e., who is reimbursing for services? Are health plans positioned for this work (as primarily agents of compliance)?

  • e. Rates. I.e., how much are providers reimbursed for services? Have rates been developed in consultation with stakeholders who are familiar with the work? Do the rates make the policy a viable funding source for CVI providers (e.g., ~ $100,000/year in reimbursements is about the minimum for scaling a full-time worker, about $55 per 30 min)? Can you index automatic adjustments to rates in response to inflation or other factors, so they do not need to go through a lengthier approval process?

• 4. Secure support for the policy (specific to a legislative process). This recommendation stems primarily from Themes 3 (Collaboration) and 5 (Momentum), in which interviewees discussed the need to identify stakeholders who will take ownership over the policy and its legislative passage and enlist others to do so as well, creating a groundswell of nonpartisan support. We suggest that stakeholders specifically consider the following: Have you assessed potential legislative support for the policy and used this information to be strategic about outreach? Do you have at least one legislative “champion” or a coalition of champions to advocate for the policy? Is there the capacity to host public hearings for stakeholder input and to secure buy-in? Have you used personal stories and data to illustrate the potential benefit of the policy? What opportunities exist to secure support from both ends of the political spectrum? Can the policy be a bipartisan effort?

Implementation phase policy recommendations

In phase 3 (Implementation), the use of the policy is initiated, and this process is guided by the supports established during the Preparation phase. Stakeholders are to monitor the implementation process and adjust their strategies as needed. During this phase, we recommend that stakeholders consider the following:

• 1. Spread the word about the policy. This recommendation stems primarily from Theme 1 (Communities/CBOS), in which the interviewees suggested the importance of developing and disseminating materials to help stakeholders become aware of and knowledgeable about the policy. We suggest that stakeholders specifically consider the following: Has an All Plan Letter been disseminated? Can you host webinars and information sessions open to all interested parties? What funding is needed and available for this?

• 2. Create a robust system of training and technical assistance. This recommendation also stems primarily from Theme 1 (Communities/CBOs), in which the interviewees expressed the high need for ongoing TTA, particularly for CVI providers who have never billed Medicaid for services. We suggest that stakeholders specifically consider the following: Who is positioned to participate in a robust TTA system (e.g., state administrator of the Medicaid-covered benefit, HVIPs, CBO partners with established community relationships, hospitals, universities, and training institutes)? Do providers have the knowledge and capacity to avoid compliance issues (which may delay or stop reimbursements)? Which providers have the least experience with/capacity to bill Medicaid, and are TTA efforts focused on them? Is there proper funding for TTA?

• 3. Provide implementation oversight. This recommendation stems primarily from Theme 3 (Collaboration), in which the interviewees expressed the importance of encouraging knowledgeable stakeholders to ‘lean in’ and use their expertise to troubleshoot the complicated policy implementation process. We suggest that stakeholders specifically consider the following: Can the administrator of the Medicaid-covered benefit maintain regular contact with providers during implementation to ensure the structure of the policy aligns with their work? Is there a Commission, group, or community of practice that can track and oversee implementation, being responsive to issues as they arise?

• 4. Adjust the benefit/policy as needed. This recommendation stems primarily from Theme 2 (Design) in which the interviewees discussed how the specific components of the policy could work in theory versus how they worked in practice, and the need for flexibility post-implementation. This is particularly the case in states in which the policy was passed rapidly and without specific attention to certain policy details. We suggest that stakeholders specifically consider the following: Has the policy been implemented as designed? If not, are there adjustments needed to the policy itself, or is there a need for bolstered TTA? Are there elements of the policy that do not fit the policy context in practice? What is the best avenue for amending the policy (i.e., legislative, administrative, or both)? Is there funding available for ongoing research and evaluation regarding the policy?

Sustainment phase policy recommendations

In phase 4 (Sustainment), the policy continues to be used (with adaptation as needed), and the processes and support remain ongoing. Through Sustainment, stakeholders achieve their goal of addressing the need for more sustainable CVI programming. The achievement of other supplemental goals (e.g., cost-savings, violence reduction) is also assessed. At the time of data collection, no state taking part in this study had reached the Sustainment phase, therefore the recommendations provided below are based on what interviewees believe will be important during this phase.

• 1. Develop/fund intermediaries for ongoing support. This recommendation stems primarily from Themes 1 (Communities/CBOs) and 3 (Collaboration), in which the interviewees discussed the importance of tapping on a diverse network of CVI providers and stakeholders to support one another. We suggest that stakeholders specifically consider the following: Are there financial resources to bring together providers into a community of practice to define best practices and identify other areas of support (e.g., access to an electronic medical record [EMR], capacity building)? Would an intermediary or community of practice be more feasible and sustainable than bringing in an outside TTA provider?

• 2. Monitor ongoing implementation. This recommendation stems primarily from Theme 2 (Design), in which the interviewees noted that the policy and/or its implementation are expected to need adjustment and therefore require monitoring. We suggest that stakeholders specifically consider the following: How many providers were able to bill for service under the policy in its first year? Which barriers are providers experiencing, and how can these be removed? Do reimbursement rates appear sufficient according to the providers? Are there opportunities for expansion into reimbursement under other service models not currently covered under the policy (e.g., violence interruption)? What are the available funding mechanisms for monitoring implementation?

• 3. Continue collaborative CVI efforts. This recommendation stems primarily from Theme 2 (Design), in which interviewees expressed that the policy is not likely cost-covering, rather it contributes to a larger constellation of funding sources for CVI. We suggest that stakeholders specifically consider the following: Are other CVI investments in place to reduce violence as the policy alone will be insufficient (e.g., grants, philanthropy)?

• 4. Track outcome data and fidelity. This recommendation stems primarily from Theme 3 (Collaboration), in which the interviewees noted the value of collecting and sharing data to inform ongoing implementation. We suggest that stakeholders specifically consider the following: Are there universities or other entities available to develop data tracking systems? Are you seeing the anticipated changes in outcomes under the policy? If not, what can be adjusted? Are stakeholders being made aware of outcomes and the level of fidelity under the policy?

Conclusions

Recent fluctuations in violence across the United States have underscored the inadequacies of traditional justice system responses, such as punitive criminal justice policies that promote racial disparities and mass incarceration. This has prompted a shift toward community-based violence intervention and prevention (CVI) programs, which leverage local expertise to prevent and reduce the harms associated with violence. While these initiatives are often under-resourced, a recent policy innovation–Medicaid reimbursement for violence intervention and prevention services (i.e., the policy)–offers one promising avenue for sustainability. The policy continues to gain popularity; however, little is known about how to implement it effectively.

To address this knowledge gap, we conducted a multistate case study of the policy and its implementation. Our qualitative analysis of secondary documentation and semi-structured interviews with stakeholders from the first three states to adopt the policy was rooted in the Exploration-Preparation-Implementation-Sustainment (EPIS) framework, and the results revealed that the implementation process has many challenges and complexities. Several common themes emerged relating to: 1) the relationships between community members and/or CBOs and the policy/its implementation, 2) the specific design, structure, and components of the policy, 3) collaborative, multi-stakeholder efforts to advocate for and/or implement the policy, 4) the context (e.g., political, social, resource) in which the policy is being implemented, 5) the speed with which the policy was passed and/or implemented, and 6) the framing of the policy and/or the issues of violence and violence prevention/intervention. Our analysis suggested that, across all these domains, the decisions made early on in the policymaking process can have a great impact on whether it is considered a successful pursuit.

A few key findings bear reiteration. First, interviewees articulated a clear difference between passing the policy legislatively and implementing the policy administratively. These are unique processes, with unique stakeholders and priorities. Second, the policy should be viewed as one piece in a large and diverse funding puzzle, as Medicaid reimbursement rates (in the states included in this study) are not cost covering. To this end, a final point for reiteration is that states can (and should) be creative in designing the policy, including exploring nontraditional billing models. CVI work can be traumatizing, exhausting, and perhaps inappropriate for fee-for-service models that place a high premium on traditional views of productivity and success. A “bundle” approach, for example (in which some portion of a provider’s budget is covered by Medicaid in exchange for providing data which evidences the work they are doing), can help reform the relationship between CVI providers and Medicaid.

As the policy continues to roll out in additional states, we present two sets of recommendations for future research that are informed by our analysis and relate primarily to potential outcomes and methodologies. In terms of outcomes, future research should cast a wide net, with the understanding that there are many ways to measure the policy’s impacts, and that this can and should be done at multiple levels. For beneficiaries, the policy may impact the amount of services they receive, overall well-being and mental health symptomatology, skill acquisition via CBT, criminal justice involvement, and violence perpetration and victimization. Generally, any outcomes under the SDOH umbrella could be appropriate to assess at the individual service recipient level.

Future research should also consider drawing from these findings to observe the policy’s impacts on specific CVI programs and its staff members (e.g. VPP/VIS workers). Potential outcomes include pay and sustainability in frontline positions, levels of burnout and turnover, and the number and types of services delivered and billed for. Evaluators should prioritize speaking with VPP/VIS and CVI leadership to gain a better understanding of their daily work with survivors of violence and other potential important outcomes.

At the state level, future research should assess the level of flexibility in accessing the benefit, the number of organizations and providers billing, and the cost savings to Medicaid (as the greatest payer of treatment for GSW, see Coupet et al., 2018). With the policy continuing to grow in popularity, future research should also consider implementation case studies in other states that were not included here to more deeply assess the diffusion of these policies across the United States (Starke, 2013). The findings from our study could be used specifically to construct a readiness tool for states considering the policy. The level of detail and context provided here permits states to consider for themselves what ‘readiness’ may entail.

From a methodological perspective, future policy evaluations should follow the principles set forth in models of coordinated community response (CCR) and community-based participatory research (CBPR), such as engaging a diverse set of stakeholders in developing and participating in research and providing continuous feedback and communication. These principles operate at multiple levels (e.g., individual, relational, community, and societal) and in multiple forms based on the goals or needs of the specific evaluation (Ranjan & Dmello, 2022). This approach falls in line with the theoretical underpinnings of CVI, which is rooted in the notion that communities have the wisdom and preparedness to help their own to break cycles of violence and recover from its harms.

There are several important limitations to the current study: the member checking process was limited in that only about half of the sample participated, we received limited feedback upon distributing the draft manuscript to the full sample for feedback, and there was greater representation in the interviews from Connecticut and Illinois than from California (therefore the results and policy recommendations may reflect more closely the experiences and contexts in these two states). Despite these limitations we have provided here a robust and detailed exploration of the policy and its implementation across multiple diverse contexts, with practical implications and recommendations for policymakers to consider.

Appendices

Appendix 1

Semi-structured interview guide

“THE POLICY”: Statewide implementation of Medicaid reimbursement for violence intervention and/or prevention services.

• 1. Tell us about your professional role.

• 2. Tell us about your role in relation to “the policy”(see above).

  • a Did you have direct input in the crafting of the policy?

  • b How has your professional role been impacted by the policy and/or its implementation?

• 3. [Innovation factors] Tell us about the characteristics of the policy in [their state]. [Prompt for specific policy components (e.g., reimbursement for VPPs) and characteristics (e.g., county vs. state-level Medicaid agencies, agency policy vs. legal statute). Prompt for additional written documentation on the policy (remember to collect at end of interview)]

  • a Who all had a “say” in crafting the design and components of the policy? [Note that this will be explored in more detail below]

• 4. Describe the stakeholders who have influence with the “inner context” of the policy and its implementation (i.e., those who receive, provide, or process reimbursable services under the policy, or those who designed the specifics of the policy). [Prompt using examples (e.g., patients or community members, funders/contractors, managed care organizations, state Medicaid agency’s organizational characteristics and service environment [e.g., benefits, provider contracts] or direct service providers like clinicians or practitioners)]

• 5. Describe the stakeholders or entities that comprise the “outer context” of the policy and its implementation (i.e., those at the Federal and state levels who may have input in or influence over the policy’s planning and implementation, but are not directly involved in receiving, providing, or processing reimbursable services). [Prompt using examples (e.g., governments, professional societies, advocacy groups, inter-organizational networks)]

• 6. In your opinion, how does the policy “fit” the implementation contexts in your state? [If needed, reiterate their answers from the two previous questions]

• 7. How would you describe the relationships between the outer and inner context entities? [Prompt for bridging factors like community-academic partnerships, purveyors or intermediaries, enhanced financial reimbursement arrangements (see Crable et al., 2022).]

  • a Have any specific activities been undertaken between entities (e.g., meetings, conferences, advisory committees, commissions, MOUs) to solidify or improve their relationships or degree of collaboration? [Prompt for timelines about how long entities have collaborated, if ever]

• 8. When did it first become apparent that the policy was needed in [their state]? [Prompt for the surrounding circumstances like budget or sustainability issues]

• 9. What signaled that it might be an appropriate time to consider a policy change such as Medicaid reimbursement? [Prompt for violent crime trends, social/political momentum at the state and Federal levels, data being collected/shared by hospitals or community groups]

  • a Were multiple stakeholders in agreement that the timing was appropriate (i.e., was there collective momentum/motivation)? Please explain.

• 10. Which individual, community, and provider needs is the policy intended to address?

  • a How were these needs identified?

• 11. Was there a process for determining that implementing the policy would be a worthwhile undertaking?

  • a If so, what was the process? Who was involved, and what was considered?

• 12. Which specific policy components or adaptations were considered to reflect [their state’s] unique context? [Prompt for adaptations at the individual, organizational, and systemic levels]

• 13. Were other policies, initiatives, or approaches under consideration besides Medicaid reimbursement?

• 14. Was a detailed implementation plan created?

  • a If so, by whom? What did it entail? How was it communicated/distributed/monitored?

• 15. What were some of the barriers to implementation?

  • a How were these determined? Were these identified by policymakers prior to implementation? How did they plan to surmount them? [Prompt for common issues like developing the infrastructure, determining reimbursement rates, professionalization of VPPs]

• 16. What were some of the facilitators of implementation?

  • a How were these determined? Were these identified by policymakers prior to implementation? How were they utilized? Were they helpful (in retrospect)?

• 17. Describe the “implementation climate” leading up to the policy’s implementation in [their state]. [Prompt for whether there seemed to be general support for or resistance against the policy, among whom, and how it affected the lead-up to implementation]

• 18. Describe the implementation process for the policy. [Prompt for timelines, stakeholder involvement, process/political hurdles, etc.]

  • a Has this process been slower, faster, or about as expected?

• 19. Was the initial implementation process monitored?

  • a If so, by whom?

  • b Were any necessary adjustments (to the implementation process or the policy itself) identified at this stage?

• 20. Which lessons were learned during the implementation process (if any)? [Prompt for whether they would advise other states to pursue different strategies]

• 21. Is there anything you would change about the policy in its current state? Please describe.

• 22. Describe the current status of the policy’s implementation in [their state]. [Prompt for expected vs. actual progress]

• 23. What are the planned ‘next steps’ (if any) for the policy’s implementation and sustainment?

• 24. Which adjustments do you feel are needed in order to:

  • a Improve the policy’s implementation; and

  • b Sustain the policy as designed?

• 25. What is the long-term vision for the policy and its impacts?

• 26. How is/will the policy’s “success” be determined? [Prompt for cost/benefit analyses, individual/organization/systemic measures, implementation fidelity and sustainability measures]

• 27. Is there anything else you would like to share about the policy or its implementation that we didn’t cover today?

Appendix 2

Table 3 State categories grouped by theme