How Are Cities Designing These Programs?

Cities and health systems building HVIPs have made distinct choices across six areas.

Design Question 1: Where Does the Program Live Institutionally?

Does the HVIP live in the hospital’s clinical infrastructure, a public health department, an academic medical center structure, or a community-based nonprofit that contracts with the hospital? Each arrangement shapes staffing authority, funding pathways, and the program’s ability to survive leadership transitions.

Hospital trauma department model: The program is embedded in the clinical team, with specialists reporting through hospital violence prevention program leadership. This is the most common structure among mature programs. The University of Chicago’s Violence Recovery Program has operated this way since 2018, with its team integrated into the Level 1 trauma center’s operations and reporting through program director Dwayne Johnson, who was named director in 2024 after six years as a founding specialist. Grady Hospital’s IVYY Project follows the same structure, with Dr. Randi Smith as medical director providing clinical authority alongside the community-based specialist team. The advantage of the trauma department model is integration: specialists are present when trauma patients arrive because they work in the same building with the same notification system. The risk is budget vulnerability — when hospital systems tighten clinical budgets, non-billable program staff are often the first to be reduced.

Public health department co-ownership: Chicago’s model is notable for splitting ownership between the hospital (which hosts the program operationally) and the public health department (which provides a significant share of the funding and the public health framing). Dr. Simbo Ige’s department administers the $3 million city investment in HVIP programming, targeting the 15 neighborhoods that account for roughly half of Chicago’s gun violence. This dual-authority structure provides fiscal redundancy — even if hospital leadership changes, the public health department’s investment gives the program an independent institutional patron. Mobile’s HALO program uses a similar tri-party structure: city, Mobile County Health Department, and USA Health University Hospital each contribute to program governance and resources.

Academic medical center model: Yale, UC Davis, VCU, and Prisma Health operate their programs through academic medical center structures that combine research and clinical service. The academic framing creates three advantages: access to research funding, the ability to evaluate the program rigorously (which builds the evidence base), and a training mission that extends impact beyond the home institution. VCU’s model in Richmond became the template for Virginia’s statewide expansion precisely because it developed both a program and an exportable training infrastructure. The risk of the academic model is institutional pace — academic medical centers move slowly, and the credible messenger workforce that makes HVIPs effective is not always comfortable in highly bureaucratic academic environments.

Community-based nonprofit partnership: Some programs separate clinical hosting (done by the hospital) from community-based staffing (done through a nonprofit partner). The hospital provides space, clinical integration, and sometimes a portion of funding; the nonprofit recruits, trains, and manages the credible messengers who do bedside and follow-up work. This structure can access community talent that hospital HR systems would not recruit, and it builds community organizational capacity alongside clinical capacity. The risk is coordination overhead and potential misalignment between hospital clinical culture and community-based organizational culture.

Dr. Smith at Grady describes the IVYY Project’s structure as “bedside-clinic-community” — not just a care model but an institutional structure: the hospital provides the bedside and clinic phases; the community partnership network provides the community phase.¹

Design Question 2: Who Does the Work — What Does the Workforce Look Like?

The spectrum runs from pure credible messenger staffing to pure clinical staffing.

Credible messenger-centered model: Atlanta’s IVYY Project employs what the program explicitly calls “credible messengers” — trained experts who live in the community and often have lived experience with violence themselves.¹ Nkosi Cave, a violence intervention specialist at the IVYY Project, describes the dynamic: “When we meet people at the bedside, we kind of see a reflection of ourselves and our communities in our patients.”¹

Clinical staff with credible messenger support: The Prisma Health model in Columbia, South Carolina deploys “trauma, internal medicine, psychiatry, and injury prevention specialists, and university researchers” alongside the community-based component.²

Licensed social work with community specialist integration: The University of Chicago model involves licensed social workers in clinical roles — Christine Goggins, lead violence recovery specialist and licensed social worker, co-led the national training design — alongside community-based specialists.⁵ Director Dwayne Johnson has described the goal: “Our goal is to have our model duplicated throughout the nation.”⁵

The Boston University researchers found that “sustained engagement” — more than four contacts in the first eight weeks — was what produced durable outcomes. Patients who received only initial contact showed outcomes “roughly equal between treatment and control strategies.”³

Named staffing scale: The University of Chicago maintains approximately 20 violence recovery specialists working around the clock. Louisville’s Pivot to Peace uses a partnership with Goodwill and Volunteers of America to provide its community-based staff. Minneapolis’s Next Step has named staff including director Kentral Galloway. None of these programs are two-person operations. Scaling to 24-hour coverage at a high-volume trauma center requires workforce investment that most programs have not yet made.

Design Question 3: How Is the Program Activated When a Patient Arrives?

HVIP notification systems vary substantially, and the notification design determines what fraction of eligible patients the program actually reaches.

Manual notification through nursing or physician: The most common model in newer and smaller programs. A nurse or physician who identifies a patient with a violence-related injury notifies the HVIP team. This model depends entirely on clinical staff remembering to make the notification, which introduces a significant gap. Programs in this model often serve only a fraction of their eligible patient population not because patients decline, but because the notification never happens.

Automatic notification through electronic health record (EHR) integration: When the hospital’s EHR flags violence-related injury codes automatically and routes a notification to the HVIP team, the program reaches a much higher fraction of eligible patients. This requires technical integration that smaller hospitals may not be able to implement, and it may generate false positives (patients who don’t meet program eligibility), but it removes the human error element from the notification chain.

Dedicated trauma pager system: The Hope and Shield Network at University of Cincinnati Medical Center provides physicians with “trauma pagers so they can be notified when an injured patient arrives and can immediately engage with patients and families during inpatient stays and at outpatient follow-up.”⁴ This bridges the gap between manual notification and full EHR integration — it is low-tech but reliable if the pager protocols are followed.

QR code notification: Pittsburgh’s HVIP uses QR codes that nurses can scan at the bedside to instantly notify the intervention team — a low-cost technical solution that removes the need for nurses to remember a phone number or initiate a multi-step notification process.⁵

Durham’s dual-mechanism lesson applied to HVIPs: Durham’s HEART program in the mobile crisis space achieved the highest documented capture rate by running two simultaneous notification channels rather than relying on either alone. The same principle applies to HVIPs: programs that rely on a single notification channel — whether manual physician notification or EHR automation — will miss patients that a dual-channel system would catch. Programs seeking to serve the broadest eligible population benefit from building redundancy into the notification architecture rather than assuming any single mechanism will fire reliably in a high-volume trauma environment.³

Durham’s dual-mechanism lesson from the mobile crisis space: In the mobile crisis team field, Durham’s HEART program achieved the highest documented capture rate through a dual mechanism (automated CAD-based dispatch and embedded clinicians) that did not rely on any single notification pathway. The same principle applies to HVIPs: programs that rely on a single notification channel will miss patients that a dual-channel system would catch. Programs seeking to serve the broadest population of eligible patients should build redundancy into the notification architecture.

The Digital Violence Responder — upstream notification before the hospital: Sinai Hospital in Baltimore has piloted a “first-in-the-nation” approach that extends hospital-based intervention upstream of the emergency department entirely. The Digital Violence Responder is a hospital staff member trained in de-escalation who “monitors social media for credible threats and alerts a call center that sends trained mediators to defuse the conflicts” before violence occurs. Reported by WYPR Baltimore, the model “builds on community-based violence interrupter efforts, where nonprofits use trusted messengers — and now digital tools — to intervene early.” This design does not replace bedside HVIP — it adds a pre-injury intervention tier connected to the hospital’s violence prevention infrastructure. It is the only documented hospital-based model that attempts to act on intelligence about impending violence rather than responding to injury after it has happened.⁶

Design Question 4: What Is the Service Model — What Does Engagement Actually Include?

The service model question is where programs diverge most widely. The Austin Travis County HVIP documented that the most common patient needs at intake are mental health resources, legal assistance, employment, housing, and education.⁵ Programs that address all five areas require different staffing, different community partnerships, and different funding than programs that address only immediate safety planning.

Bedside-only programs conduct the initial crisis intervention and safety assessment, connect patients to existing services through referral, and end the formal HVIP engagement at discharge. This is the minimum viable model. It reaches patients at the intervention window but does not sustain the relationship that the Boston University study found necessary for durable outcomes.

Sustained case management programs maintain active contact for months after discharge. Dr. Amy Makley of Cincinnati describes what “sustained” means in practice: “instead of helping the victim for a day or two treating their physical wounds, the support will continue for months after leaving the hospital.”⁵ Minneapolis’s Next Step provides “ongoing trauma counseling, as well as legal services, work placement, or school enrollment assistance” long after the patient has left the hospital.⁵

Multidisciplinary service programs embed psychiatric care, substance use treatment, and legal services alongside the case management function. The Prisma Health model in Columbia, South Carolina deploys “trauma, internal medicine, psychiatry, and injury prevention specialists, and university researchers” as an integrated team.⁷ Yale’s program in New Haven adds victim compensation navigation and unconditional cash stipends — $1,000 for family members of homicide victims, two $500 payments for violence survivors — recognizing that immediate financial barriers can prevent engagement regardless of service quality.⁸

What programs provide most commonly: Dr. Randi Smith’s program at Grady describes the core service set as help navigating victim compensation funds, connections to trauma-informed therapy, safety planning for discharge, and ongoing case management. Some programs offer emergency relocation assistance or job training slots.⁶

The most variable dimension across documented programs is the depth and breadth of services they provide. Programs range from primarily therapeutic (trauma counseling, PTSD treatment, mental health services) to primarily practical (housing navigation, employment placement, legal services) to comprehensive (all of the above, delivered through sustained case management).

The bedside-clinic-community architecture: The IVYY Project at Grady has formalized the most clearly articulated service model in the field through a peer-reviewed paper in Trauma Surgery & Acute Care Open. Its three-phase model distinguishes:

Bedside: Violence intervention specialists at the hospital during the acute treatment phase, providing “immediate care to victims of violence at the time of injury,” ensuring medical treatment and psychological support, and creating “a seamless transition to ongoing wraparound services.”
Clinic: A “one-stop shop” combining ongoing medical care beyond the acute phase with social services: “physicians, advanced practice providers, wound care specialists, mental health experts, a social worker,” serving as “a critical bridge between immediate bedside care and long-term community resources.”
Community: Connections to organizations for ongoing wraparound support addressing mental health, education, employment, financial stability, legal aid, housing, transportation, and food security.⁹

This model is the most architecturally explicit in the field. Programs that have the capacity to implement all three phases are likely to produce outcomes closer to the 2-3% reinjury rates in the best-documented programs. Programs that only implement Phase One (bedside) without sustained case management are likely to produce outcomes closer to what the Boston University study found for “any engagement” — “roughly equal between treatment and control.”

Service components with documented outcomes association:

Victim compensation navigation: accessing funds that most patients are unaware of and unable to navigate independently
Trauma counseling: addressing PTSD and trauma-related symptoms that increase violence risk
Employment and education: the 89% employment/diploma/GED completion rate among program graduates is associated with four-times-greater success in the intervention overall
Legal services: the University of Chicago embedded two full-time civil lawyers through Recovery Legal Care in 2022, providing bedside civil legal help — one of the more innovative service expansions in the field¹⁰
Housing navigation: addressing one of the most concrete barriers to the patient’s ability to leave a dangerous environment
Cash assistance: Yale’s $1,000 stipend pilot for family members of homicide victims, and $500 payments for survivors, reflects growing recognition that immediate cash removes barriers (transportation to safety, rent, food) that case management cannot¹¹

What most programs cannot yet provide: Long-term residential support for patients who need to relocate for safety and lack independent housing capacity. Emergency relocation assistance exists as a service in some programs but is resource-constrained. The gap between knowing a patient needs to move and being able to fund and execute that move is real and large.

Design Question 5: How Does the Program Connect to What Comes Before and After?

Upstream: the hospital as the only activation point: Unlike mobile crisis programs that can be reached through multiple dispatch channels, HVIPs activate only through hospital admission. This means their effectiveness depends on the quality of emergency medical services getting the patient to a participating hospital, and on the notification system working within that hospital once the patient arrives. Hospitals that serve high-violence neighborhoods but have not built HVIP programs represent entire gaps in the system — patients arrive, get surgery, and leave without any intervention.

Downstream: the handoff to community organizations: The University of Chicago’s 60+ agency network, Grady’s warm handoff model, and Louisville’s partnership with Goodwill and Volunteers of America all represent different approaches to the downstream connection.⁸

The Trauma Recovery Center connection: Hospitals that can refer patients to a TRC for longer-term therapeutic work create a two-institution continuum that neither the HVIP nor the TRC could produce alone. Buffalo’s design, with the TRC explicitly built in partnership with a pre-existing HVIP, is the clearest documented model of this architecture. Sacramento’s UC Davis system connects the bedside violence intervention specialist to a Trauma Recovery Center for “long-term therapy and social services, and help with food, housing and employment,” according to reporting by CBS News on the program’s launch.¹²

The street CVI connection: The UChicago VRP has developed relationships with street outreach workers through the Metropolitan Peace Academy, creating mutual trust between hospital-based and street-based specialists. UChicago Medicine’s program materials describe the Metropolitan Peace Academy relationship as enabling warm handoffs from hospital specialists to street outreach workers — sustaining engagement with patients who return to neighborhoods where both systems are active.⁸

Design Question 6: How Do Programs Sustain Funding?

The funding question is where most programs face their most acute vulnerability. HVIPs are not billable in the way that surgery and clinical procedures are. The work that distinguishes them from standard medical care (case management, community relationship-building, transportation to appointments, peer support) is largely not reimbursable under standard insurance frameworks.

Chicago’s $3 million investment flows through the public health department. Virginia’s $8.5 million in 2025 funds 12 statewide programs. Connecticut was the first state to sponsor Medicaid reimbursement for community services to do violence intervention outreach, creating a reimbursable pathway that most states have not replicated.¹³ DOJ awarded Dr. Randi Smith’s IVYY Project $2 million under the Community Violence Intervention and Prevention Initiative.¹⁴

Angela Kimball, of the advocacy organization Inseparable, has articulated a structural challenge that applies directly to HVIPs: police and fire departments are funded for the capacity to be ready — officers are paid whether or not they take a call in a given hour. Violence recovery specialists funded through encounter-based mechanisms are paid only for active patient contact. The supervision, training, community relationship maintenance, and availability during slow periods are not covered by encounter-based billing.¹⁰

UChicago combines hospital institutional funding, city investment, federal grants, and philanthropic support. Grady combines DOJ grants, hospital institutional support, and academic medical center research funding. Louisville uses city Office of Violence Prevention budget, federal CVI grants, and nonprofit partner infrastructure.¹⁰

Design Question 7: Who Governs and How Are Programs Held Accountable?

Governance is not a design question that most program profiles address explicitly, but it shapes everything from staff authority to outcome measurement to whether the program survives leadership transitions.

Physician-led governance: In programs led by trauma surgeons (Grady, VCU, Prisma Health, Cincinnati) the physician provides both medical authority (which enables clinical integration) and institutional credibility (which protects the program from the “not a hospital’s job” resistance). Physician champions can defend HVIP budget lines in a language that hospital administrators understand. The risk is succession: when the founding physician transitions out of the role, programs that have not built broader institutional endorsement can rapidly lose resource support.

Public health department governance: Programs where the city or county health department holds administrative authority (Chicago, Mobile) benefit from a public health frame that survives individual hospital leadership transitions. The public health department’s funding continues even when hospital leadership changes. Louisville’s Office of Violence Prevention represents a hybrid: it is a city agency that coordinates both the hospital-based and street-based components of Pivot to Peace, providing unified reporting, accountability, and political ownership through the mayor’s office.

Outcome accountability: The cities with the most legible HVIP outcomes are the ones with formal reporting structures. Louisville’s Office of Violence Prevention publishes annual reports. Virginia’s Hospital and Healthcare Association tracks and reports re-injury rates across all 12 statewide programs. New York City’s citywide HVIP system reports aggregate patient numbers. These reporting structures do two things: they build the public record that supports continued investment, and they create internal pressure to improve outcomes rather than simply maintain operations.

Programs without formal reporting structures may be operating with good outcomes or poor ones; neither their communities nor their funders can tell the difference. Building reporting infrastructure is not bureaucratic overhead — it is what gives programs the political capital to survive difficult budget cycles.

Design Question 8: How Do Programs Handle Co-Occurring Mental Health and Substance Use?

Many patients who arrive at trauma centers with gunshot wounds have co-occurring mental health conditions, substance use disorders, or both. How programs respond to these co-occurring presentations shapes both their reach and their outcomes.

Integrated screening and service: Programs with embedded clinical staff (Prisma Health’s psychiatry component, VCU’s clinical team) can screen for and directly address mental health and substance use conditions in ways that purely peer-staffed programs cannot. Yale’s program provides psychiatric care and substance use treatment alongside the violence intervention services. This integration is resource-intensive but addresses conditions that, if untreated, become risk factors for future violence involvement.

Referral-based co-occurring services: Most programs refer patients to external mental health and substance use services rather than providing them directly. The effectiveness of this approach depends entirely on the accessibility of those services — whether they exist, accept the patient’s insurance, have capacity, and can be accessed without transportation barriers the patient cannot overcome independently. In cities with strong behavioral health infrastructure, referral-based co-occurring services can work. In cities where mental health outpatient capacity is constrained, the referral becomes another document in the patient’s folder that goes nowhere.

The PTSD dimension specifically: Dr. Amy Makley at Cincinnati documented that violence survivors “experience high rates of mental health disease and post-traumatic stress disorder that extend beyond the initial injuries.” Untreated PTSD produces hypervigilance, reactivity, and risk-taking behavior that increase the likelihood of future violence exposure. Programs that do not address PTSD are intervening in the social and material conditions of violence without addressing one of the psychological conditions most directly linked to reinjury risk. This is one of the clearest gaps between what the evidence suggests is needed and what most programs are funded to provide.¹⁵

Design Question 9: How Do Programs Build and Maintain Credibility With the Patient Population?

HVIP credibility is not institutional — it is personal and relational. Programs that succeed in sustained engagement succeed because individual workers build enough trust with individual patients to stay in relationship through the disruptions, missed contacts, and moments of backsliding that characterize the recovery trajectory.

The credible messenger selection question: What qualifies someone to be a credible messenger for a specific patient population? Programs answer this differently. Some prioritize neighborhood or network proximity: workers who grew up in the same streets or knew the same people. Some prioritize personal history with violence: workers who were themselves shot, incarcerated, or lost close family members. Some prioritize time since involvement: workers who have enough distance from their own history to engage professionally without being retraumatized.

The selection process matters enormously. A credible messenger who lacks adequate distance from their own trauma can be destabilized by repeated patient contact. A credible messenger without authentic community connection can fail to build the trust that makes the model work. Programs that have built strong credible messenger pipelines (UChicago’s VRP is the most documented example) have done so over years, not through a single hiring cycle.

The institutional trust problem: Trauma centers in high-violence urban neighborhoods are often institutions that the surrounding community has complicated relationships with. Hospitals have histories of research misconduct in communities of color, of treating patients dismissively, and of discharging people without adequate support. An HVIP that deploys credible messengers alongside clinical staff navigates this institutional history every time a specialist arrives at a patient’s bedside. The credible messenger’s personal credibility creates an access point; it does not erase the institutional history that the hospital carries. Programs that acknowledge this explicitly in their training and culture are more effective than programs that treat it as irrelevant.¹⁶

Design Question 10: What Does Launch and Scale-Up Look Like?

The programs with the strongest outcomes are mostly programs that have been operating for years and have iterated through early design challenges. The launch and scale-up trajectory matters because many cities and hospitals contemplating HVIPs in 2025 and 2026 are starting, not sustaining.

Training and replication from documented programs: The University of Chicago VRP has developed the most systematic replication model in the field. Its four-day training workshop, facilitated by lead specialist Christine Goggins, brought OSF HealthCare Saint Anthony Medical Center in Rockford, Illinois through a structured program launch process. The $99,000 Illinois Department of Public Health grant to UChicago for training and technical assistance to other Illinois hospitals formalized this function. The VCU model in Richmond is similarly the template for Virginia’s statewide expansion.¹⁷

The practical implication: hospitals launching new HVIPs do not need to build the model from scratch. They can access training from programs with documented track records, adapt the model to their institutional context, and measure outcomes against the benchmarks that mature programs have established.

The first-year challenge: Programs in their first year face challenges that mature ones do not: building credibility with hospital clinical staff unfamiliar with the HVIP workforce model, establishing notification systems that actually work, recruiting credible messengers in a limited talent pool, and building community organization connections that take years to develop. The programs currently reporting first-year data, some with impressive initial numbers, should be read as promising starts, not established outcomes. The Cincinnati Hope and Shield Network’s “zero reinjuries in 50 patients” in six months is encouraging but is too small and too new to be treated as equivalent to VCU’s five-year 3.6% recidivism rate.

The training infrastructure gap in most states: Outside of Illinois and Virginia, there is no systematic state-level infrastructure to support hospital HVIP launches. A hospital in Mississippi, Kansas, or Arizona that wants to launch an HVIP must either find a national training partner (the Health Alliance for Violence Intervention, HAVI, provides some technical assistance), seek out a program like UChicago or VCU directly, or develop the model independently. The absence of state-level infrastructure means that program quality at launch varies substantially — and programs launched with weak design are likely to produce weaker outcomes than programs launched with strong training and technical support.

What the Design Evidence Points Toward

No single design is optimal for every context. The University of Chicago’s 20-specialist, around-the-clock model is the right design for a high-volume South Side Level 1 Trauma Center. It would be over-resourced for a rural trauma center seeing 40 violent injuries per year, and under-resourced for a New York City public hospital system treating 3,500+ patients annually.

What the evidence consistently supports is a set of conditions that appear across the most effective programs regardless of scale:

Sustained engagement rather than one-time bedside contact — the dosage effect documented in the Boston study
Credible messenger workforce with authentic community connection — not a clinical staff substitute but a distinct workforce model with distinct qualifications
Multiple notification pathways to reach the broadest fraction of eligible patients
Downstream connections to community organizations, Trauma Recovery Centers, and street CVI programs — the hospital phase is the beginning, not the whole intervention
Diverse funding streams — no single source is a sustainable plan
Formal outcome measurement — what gets measured gets funded and improved

Programs that achieve all six tend to be the ones with documented outcomes at the lower end of the reinjury range (2-4%). Programs that achieve some but not others tend to be the ones with solid but not exceptional outcomes. Programs that achieve none of them tend to be the ones that exist on paper but serve a small fraction of eligible patients.

The implementation question for a city or hospital system is not “should we do this?” — the evidence answers that. It is “what design will our context and resources actually support, and how do we build toward the full model over time?”

The Design Choices That Most Programs Get Wrong in Year One

For cities and hospitals currently planning or in early implementation, the documented failure patterns are worth naming explicitly.

Notification system underinvestment: Programs that rely on busy nurses to remember to make a call to the HVIP team, without a dedicated pager, QR code, or EHR trigger, systematically miss eligible patients. The first design fix is almost always notification architecture, not workforce or service design. A program with a 40% notification rate is serving 40% of its eligible population regardless of how good its specialists are.

Confusing pilot scale with program scale: A two-specialist program covering a Level 1 Trauma Center that treats 400 gunshot patients per year is running a demonstration, not a program. It will serve some patients well and miss most. Decision-makers who fund pilots and then evaluate whether “the program works” based on pilot outcomes are evaluating the wrong thing. The program hasn’t happened yet; the pilot has. The Louisville model (85% capture rate from 628 patients across two years) is what a functioning program looks like.

Hiring for credentials rather than credibility: The most common workforce error is treating HVIP specialist hiring as a standard social work or case management hire. The credential that makes the model work, lived experience of the same violence, networks, and neighborhoods as the patient population, is not on a resume. Programs that hire based on clinical credentials and hope credibility follows have typically produced lower acceptance rates and lower sustained engagement than programs that hire for community connection and build clinical infrastructure around that workforce.

Treating the bedside moment as the intervention: The Boston University Annals evidence is clear: any engagement (bedside contact without sustained follow-up) produces outcomes “roughly equal between treatment and control.” The bedside is the door, not the intervention. Programs that measure success by bedside contacts rather than 90-day engagement rates are measuring the door, not what lies behind it.