How Are Cities Designing These Programs?

SAMHSA’s 2025 National Guidelines describe a set of design decisions each jurisdiction must address when building crisis stabilization centers. This card documents what cities have chosen at each decision point and what the sources report about outcomes.

Decision 1: What Acuity Levels Will the Center Serve?

The first and most consequential design decision is who the center is built to serve. Crisis stabilization centers range from low-acuity peer-support walk-in spaces to high-acuity secure 23-hour observation units. The Arizona ASU/Connections study and the SAMHSA 2025 guidelines both describe a tiered model as the design standard.

The spectrum:

At the low-acuity end is the “living room” model: a peer-staffed walk-in space where people experiencing distress can access support, conversation, and resource connection without clinical assessment or medical intervention. Cleveland’s center built this model as its intake framework, with “a living room model intake area to feel less clinical,” staffed by peer support specialists. Albuquerque’s center has a dedicated peer-based living room as one of its three zones, specifically designed for “low-acuity walk-in or referral services in a welcoming environment accessed by a separate public entrance and lobby.”

At the high-acuity end is the secure 23-hour observation unit: a clinically supervised environment with psychiatric evaluation, medication management, medical monitoring, and security to prevent elopement for people who may be dangerous to themselves or others. The Phoenix Crisis Response Center operated by Connections Health Solutions includes a “secure 23-hour observation unit staffed by an interdisciplinary team of 24/7 onsite medical providers (psychiatrists, nurse practitioners, or physician assistants), registered nurses, case managers, behavioral health technicians, and peers.”

In between are the middle tiers: 3-to-5-day crisis stabilization stays for people who stabilize over days rather than hours, and extended residential care up to 14 days for people transitioning from acute crisis to community re-engagement.

What cities have found: The Albuquerque three-zone model reflects the SAMHSA guideline that different acuity levels require different clinical environments. A person in severe psychosis requiring medication adjustment and a person needing someone to talk to after a difficult day are not served by the same clinical environment. Separating the zones allows the peer-focused living room to function without the security and clinical intensity that the higher-acuity unit requires, which keeps the lower-acuity zone accessible and non-threatening for the people it is designed for.

SAMHSA’s 2025 National Guidelines identify tiered design — separate staffing, physical environments, and clinical governance for each acuity level — as an implementation standard.

Decision 2: Voluntary vs. Involuntary Admissions

The second decision is whether the center accepts people who do not want to be there.

Most crisis stabilization centers are designed primarily or exclusively for voluntary admissions. SAMHSA’s 2025 National Guidelines identify voluntary, recovery-oriented engagement as the design standard for crisis stabilization centers, distinguishing them from coercive alternatives.

The involuntary dimension: Some centers operate with a hybrid model, accepting both voluntary patients and people brought under legal hold authority. Arizona’s Tucson Crisis Response Center provides “23-hour observation and inpatient services only to individuals who arrive voluntarily and focuses on populations with substance use disorder needs,” while the Maricopa facility includes secure capacity for involuntary patients. Bellingham’s Anne Deacon Center For Hope provides “a 12-hour involuntary hold while they sober up or get medications to steady them” as one service within a broader continuum.

The design tension: The Arizona ASU/Connections study documents that Tucson’s center maintains a voluntary-only model while the Maricopa facility maintains separate secure capacity for involuntary patients — a design that avoids mixing populations in the same space.

What this means for first responders: The voluntary-primary model creates the most important operational question for police departments: what do they do when they arrive with someone who is refusing to go? If the center cannot accept involuntary admissions, and the person meets the legal standard for an involuntary hold, the officer needs either a second destination (an ED with psychiatric hold authority) or the authority to complete the hold at the crisis center. Dr. Balfour’s research identifies the protocol that well-designed centers use: evaluate first, determine whether voluntary engagement is possible, and if not, arrange transfer to a facility with hold authority. The critical point is that the center does not turn officers away before assessment.

Decision 3: Institutional Home and Operator

Who builds the center, who funds it, and who runs it are three questions that often have different answers. The institutional home shapes the center’s political durability, its access to Medicaid billing infrastructure, its relationship with the health system, and its operational culture.

County or city government-built, nonprofit-operated: The most common model for larger investments. Fulton County built and funded the center; Grady Health System operates it. Franklin County (Columbus) funded construction through ADAMH and multiple government sources; Recovery Innovations operates services. This model separates capital and political risk (government bears the construction and policy risk) from operational risk (the contractor bears workforce, clinical, and daily operational risk).

Hospital system-integrated: Salt Lake City’s Huntsman Mental Health Institute is operated by the University of Utah Huntsman Mental Health Institute, combining state and philanthropic funding with university health system resources. This model provides access to existing clinical infrastructure, training pipelines, and credentialing systems, but creates different accountability dynamics than a community-governed facility.

State-led, locally executed: Georgia’s DHBDD model involves state operational funding, state-coordinated planning, and a state-contracted operator (Grady Health System in Fulton County), with local government providing construction capital. Virginia’s “Right Help, Right Now” program created a similar structure: state funds flow to localities that develop and build facilities, with state-contracted or locally-selected operators running services.

Nonprofit primary: Connecticut’s REST Center; Southcentral Foundation in Alaska; Community Medical Services in Waukesha County, Wisconsin. Nonprofits operating under government contracts provide flexibility, specialized expertise, and Medicaid billing infrastructure, but face the contract vulnerability risk that caused CAHOOTS’ collapse in the mobile crisis space and the Arkansas CSU’s failure: when the government partner cuts funding, the nonprofit operator cannot sustain operations independently.

What cities have found: The Northwest Arkansas CSU collapse began when the nonprofit contractor could not sustain operations independently after state funding cuts. Arizona’s statewide system has operated through multiple budget cycles using a Medicaid-financed model under AHCCCS governance.

The voluntary-primary model creates an operational question for police departments: what happens when an officer arrives with someone who refuses to go voluntarily? Arizona’s design resolves this by maintaining both a voluntary walk-in facility (Tucson CRC) and a secure observation unit with involuntary hold capacity (Phoenix CRC’s secure unit). Dr. Balfour’s research identifies the two-step protocol that well-designed centers use: assess first, determine whether voluntary engagement is possible, and if not, arrange transfer to a facility with hold authority rather than turning the officer away. The critical principle from Balfour’s research: the center does not turn away before assessment.

The co-occurrence routing problem. SAMHSA’s 2025 National Guidelines specifically identify co-occurring presentations — someone in psychiatric crisis who is also acutely intoxicated — as a population that crisis stabilization centers must address with explicit protocols, because routing such individuals to sobering centers misses their psychiatric need, while routing them to psychiatric-only crisis centers may exceed those facilities’ capacity for medical intoxication monitoring. The Bellingham model — which provides a short-term involuntary hold while patients sober — is one documented design response. Centers designed for voluntary psychiatric presentations that routinely encounter intoxicated patients without a protocol for that presentation create both clinical and safety gaps.

Decision 4: Law Enforcement Drop-Off Design

Dr. Balfour’s research identifies two specific design failures: officers waiting in queue for transfers (which eliminates the time advantage over ERs) and officers being turned away (which eliminates the reliability that drives police use).

The Memphis standard: Memphis built a dedicated first-responder drop-off area with key card access for law enforcement, specifically designed so that officers can complete a transfer in 15 to 30 minutes rather than the several hours an ER transfer typically requires. The operational significance is not incidental: an officer who spends two hours at an emergency room for a psychiatric transfer has lost two hours of patrol coverage. An officer who completes a transfer in 20 minutes has not.

The Albuquerque integration: Albuquerque’s center includes “dedicated space for both law enforcement and the Albuquerque Community Safety department,” reflecting the city’s hybrid emergency response model in which ACS civilian responders work alongside traditional first responders. The physical co-location of ACS and law enforcement spaces reflects an operational reality: many crisis center transfers involve a handoff from a police officer who responded first to an ACS worker who provides follow-up care.

The “no wrong door” operational requirement: Dr. Balfour’s research establishes that when officers are turned away, they stop bringing people. The operational pattern: centers with unpredictable admission practices cause officers to default to emergency rooms, which are slower but reliable. The guarantee of acceptance (for assessment at minimum, even if the outcome is a transfer to a higher-acuity facility) is the single most important operational feature for building law enforcement trust.

Hawaii implemented the operational benefit clearly: before the center opened, “a police officer could end up waiting hours with a patient.” After: they “can be back to the beat in like five or 10 minutes.” The time recovery per officer across a full year of transfers is a meaningful contribution to police capacity that police departments can quantify in their own terms.

The King County guaranteed-acceptance policy is worth examining in detail because it solves a documented problem in the most direct way possible. Police departments that have experienced refusals at crisis centers — even a small number of them — revert to emergency room transfers as the default. A written policy guaranteeing acceptance for assessment, regardless of presentation complexity, removes the uncertainty that causes officer reversion. The policy does not mean the center accepts everyone for long-term treatment; it means the center assesses everyone before making a disposition decision.

Memphis’s key card drop-off design addresses the time dimension of the same problem. A police officer who must enter through a public entrance, navigate intake procedures, and wait for a clinical assessment is losing patrol time in a way that is functionally similar to an ER transfer. The dedicated law enforcement entrance with key card access, documented in Memphis’s design, is a physical solution to an operational problem. The 15-to-30-minute transfer window that results is short enough that officers can complete a transfer mid-shift without disrupting patrol coverage in the way that ER transfers do.

Decision 5: Mobile Crisis Team Integration

The center’s relationship with mobile crisis teams is as important as its relationship with law enforcement. Mobile crisis teams need a reliable destination for the cases they cannot resolve in the field. Crisis centers need a steady source of appropriate referrals. The connection between them determines whether both work better than either would alone.

King County’s guaranteed-acceptance policy: King County, Washington has established that anyone brought by a mobile crisis team will be accepted at the crisis stabilization center. This policy removes the uncertainty that causes mobile teams to default to emergency rooms: if the team knows the center will accept the person, they can make the transport decision confidently rather than spending time on a phone call trying to determine whether a transfer is appropriate.

Columbus’s integrated model: The Franklin County Crisis Care Center is designed to serve as a hub for Columbus’s mobile crisis system, not just as a passive destination. Columbus Police Chief Elaine Bryant and Fire Chief Jeffrey Happ described the operational benefit at the facility’s opening: “It helps the officers because we have somewhere to take people when they are in crisis. The last thing we want is enforcement. We want to get people the help they need.”

Indiana’s design: St. Joseph County’s first behavioral health crisis center serves as “a headquarters for the county’s mobile crisis team, which will operate out of the facility.” Co-locating the mobile crisis team with the stabilization center physically connects the two components and creates natural coordination in dispatch, warm handoffs, and follow-up.

What cities have found: King County’s guaranteed-acceptance policy and St. Joseph County’s co-location model represent two documented approaches to mobile team and center coordination.

The Washington State HCA 2024 report identifies coordination gaps between 988, mobile crisis, and facility-based components as a primary operational challenge documented nationally across multiple state systems. In jurisdictions where the three tiers were built by different agencies at different times without shared protocols, the gaps consistently show at the handoff points: a 988 counselor who cannot confirm in real time whether the crisis center has capacity before dispatching a mobile team; a mobile team that transports a person to a center without advance notification; a center that receives a walk-in without knowing the person’s prior crisis history from the 988 contact. Each of these gaps is documented not as a design flaw in the programs themselves but as a coordination failure between programs that were built without interoperability in mind.

Franklin County Columbus addressed this at the infrastructure level by building on the county’s existing electronic bed board, which had tracked psychiatric capacity across facilities since 2009, giving dispatchers real-time confirmation of center availability before transport. Dayton’s three-tier model built shared operational protocols between the crisis hotline, mobile teams, and the center as a design prerequisite before opening, ensuring each component had documented clarity on what the others could handle, when to escalate, and what a completed handoff looked like from each side.

Decision 6: Discharge Planning and Follow-Up

SAMHSA’s 2025 guidelines identify discharge planning — medication supply, scheduled follow-up, case manager assignment — as a required design element, not optional.

What strong discharge looks like: Nebraska’s Crisis Stabilization and Resource Center provides the clearest model: after stabilization, “patients work with a case manager after they’re released” specifically to support reintegration. Columbus’s facility design includes the explicit goal of linking people to “community mental health and substance use treatment providers, as well as resources related to non-medical needs, including housing, food and transportation” at discharge.

The medication gap: Franklin County’s model includes a specific provision that addresses a common post-discharge failure: the center supplies “each guest with all prescribed medications to carry them safely through to their next scheduled community provider appointment.” This addresses a documented gap: people who leave a stabilization episode with a prescription but without the medication, money to fill it, or transportation to a pharmacy frequently do not maintain medication continuity, and the crisis recurs. In the National Alliance on Mental Illness’s research on incarcerated people with mental illness, over 50% of those taking psychiatric medication at admission did not continue receiving it during incarceration, a pattern that reflects the same continuity failure in a different setting. Crisis centers that give people medication to bridge the gap to their next appointment are addressing this specific failure mode directly.

What centers cannot control: The best discharge planning in a crisis stabilization center is limited by the availability of outpatient behavioral health care into which people are being discharged. In communities with outpatient waitlists of weeks to months, a thorough discharge plan that schedules a first appointment is better than no plan, but it does not guarantee continuity. Crisis centers in regions with thin outpatient capacity face a structural ceiling on their effectiveness that no amount of internal program design can overcome. The National Association of State Mental Health Program Directors has documented significant variation in outpatient capacity nationally; centers in states with the most acute outpatient shortages face the highest risk of post-discharge recurrence.

Peer follow-up as a bridge: Programs including Nebraska’s Crisis Stabilization and Resource Center have incorporated peer specialist follow-up calls in the 24 to 48 hours after discharge as a specific intervention to bridge the gap between discharge and the first outpatient appointment. The peer who was part of the intake process at the center can make a follow-up call using the relational trust established during the crisis episode, check on medication continuity, confirm the appointment, and help problem-solve barriers. This function does not require a licensed clinician and extends the center’s therapeutic reach beyond the walls of the facility.

Decision 7: Population Specialization vs. General Admissions

The choice between serving a general adult population and designing for specific subpopulations has significant implications for clinical model, staffing, and physical design.

General adult model: Memphis, Albuquerque, Columbus, and Atlanta all operate general adult models serving adults 18 and older with any behavioral health crisis presentation. It produces the broadest reach and serves the largest volume but requires a clinical model capable of handling the full spectrum of adult psychiatric and substance use presentations.

Youth specialization: Facilities designed specifically for youth require separate physical environments (adolescents should not share space with adults in acute psychiatric crisis), clinical staff trained in child and adolescent psychiatry, family-inclusive service models, and connections to school-based follow-up. Johnson County, Kansas’s 10-bed youth center and Alaska’s Juneau facility represent the specialized end of this spectrum. States where youth-specific capacity does not exist route adolescents to adult facilities (creating clinical mismatch) or to pediatric emergency departments (creating boarding without treatment). The Johnson County mental health director’s observation that 65 to 70% of youth in the juvenile justice system have a diagnosable mental health condition frames the alternative to youth-specific crisis care starkly: the juvenile detention facility becomes the default psychiatric unit.

Homeless-specific models: Portland’s Behavioral Health Resource Center represents the most explicit specialization for people experiencing homelessness with co-occurring mental illness and substance use. The design reflects the clinical reality that this population has often failed in general congregate settings and requires an environment specifically built around trust-building and basic needs provision alongside clinical care. Access to “kitchens, bathrooms and showers, along with private spaces” is not a service amenity; it is the therapeutic precondition for engagement with people who have been excluded from other settings.

Pregnancy and substance use specialization: Baltimore’s planned center with 30 beds for teens, 20 for adults, and 40 stabilization beds specifically addresses the intersection of opioid use disorder and pregnancy, a population with specific medical protocols, legal complexity around mandatory reporting, and heightened stakes where treatment engagement during the crisis window is most critical.

Design logic: The rationale for specialized facilities is that populations with distinct clinical needs — youth, people experiencing homelessness with chronic illness, pregnant women with opioid use disorder — are poorly served by facilities not designed for those needs. Whether specialization produces measurably better outcomes has not been studied comparatively across facility types; the argument is design-logic rather than independently evaluated evidence. The tradeoff is access: a specialized facility that is not geographically accessible to its target population is less useful than a general facility that is available. For most jurisdictions, the sequence has been: build general adult capacity first, then develop specialized capacity for populations that are poorly served by the general model.

Decision 8: 988 Integration and Crisis Line Coordination

Crisis stabilization centers exist in a three-tier system: someone to call (988 or a local crisis line), someone to respond (mobile crisis teams), and somewhere to go (the center). The center’s integration with the front end of the system determines how many appropriate referrals it receives and how effectively the three tiers work as a continuum.

The 988 connection: The 988 Suicide and Crisis Lifeline, which handled nearly 5 million contacts in its first year of operation, can initiate mobile crisis dispatch for callers who need in-person response. When mobile teams cannot resolve calls in the field and transport to a crisis center, 988 dispatch needs to know the center’s availability, acceptance policies, and geographic coverage. In jurisdictions where 988 and local crisis dispatch are not connected, mobile teams make transfer decisions without knowing whether the center has capacity.

Dayton’s three-tier integration: Montgomery County built an explicit design framework with three linked components: someone to call (the Montgomery County Crisis Now Hotline), someone to respond (Mobile Crisis Response Teams), and someone to go to (the crisis center). Each tier has clear entry, exit, and handoff protocols. The crisis line screens and dispatches. The mobile team responds and determines whether field stabilization is possible or transport to the center is indicated. The center accepts the handoff and connects to ongoing care.

What integration requires operationally: The center must communicate real-time capacity to dispatch; 988 counselors and mobile crisis dispatchers need to know whether the center is available before a team is sent with a patient. Some centers use electronic bed boards, shared systems that show available capacity, allowing dispatchers to confirm acceptance before transport. The Franklin County Crisis Care Center’s relationship to the existing Columbus behavioral health system was partly designed around the county’s existing electronic bed board infrastructure, which had been tracking psychiatric capacity across facilities since 2009.

The gap in most jurisdictions: The Washington State HCA 2024 report identifies coordination gaps between 988, mobile crisis, and facility-based components as a primary operational challenge.

Decision 9: Capital Investment and Physical Design

The Psychiatric Times documents the design philosophy that therapeutic environments reduce defensive responses in people in psychiatric distress.

The therapeutic environment principle: The Psychiatric Times documents the design philosophy: “homelike and hospitality-oriented dormitory-style rooms” that “give some privacy and reinforce dignity” with “a calming, living room setting with artwork, natural light, and sensory rooms.” This is not mere aesthetics. Bright fluorescent lighting, exam tables, and institutional furniture signal to a person in psychiatric distress that they are in a clinical-coercive environment, which activates defensive responses that make therapeutic engagement harder. Natural light, personal space, and peer-oriented common areas signal safety.

Safety-designed interiors: The Fulton County center’s design specifically incorporated “soft edges and weighted furniture” to reduce self-harm risk without creating an institutional atmosphere. Weighted furniture is a clinical intervention, not a decorating choice: it reduces agitation in people with sensory dysregulation while providing comfort for those who are simply distressed.

Scale, cost, and configuration: Large facilities serving high volumes require multi-zone configurations, dedicated law enforcement entrances, multiple acuity levels, and operational infrastructure that small centers do not need. Capital cost ranges accordingly: Salt Lake City’s $64 million facility, Memphis’s $34 million center, and Juneau’s $18 million center reflect different scale and scope. Clark County, Nevada acquired and converted an existing building for $10 million, preserving most therapeutic design principles at substantially lower cost. Communities without capital for new construction have developed conversion models that begin with simpler configurations and expand as program volume and funding grow. Columbus’s Franklin County facility is explicitly phased: initial observation units opened in September 2025, with urgent care and family resource center components following in 2026, and long-term support capacity by 2027.

Rural implications: Physical design requirements do not change in rural settings, but the economics do. A rural county of 200,000 people cannot sustain the fixed costs of a 55,000-square-foot facility that operates 24/7 with 120 staff. Benton County, Washington’s “first comprehensive behavioral health and recovery center” was built at a scale appropriate for a rural county’s population and budget, demonstrating that the design principles transfer across scale, even if the facility footprint does not.

Decision 10: Staffing Model and Peer Integration

Memphis, Salt Lake City, and Johnson County all document a staffing model combining licensed clinical staff alongside peer support specialists.

The multi-disciplinary standard: Memphis’s staffing roster illustrates the full model: “120 psychiatrists, nurse practitioners, registered nurses, peer support specialists, counselors and social workers.” Salt Lake City’s Huntsman facility operates with 200 mental health professionals. Johnson County’s youth center was staffed with “a clinician, case manager, nurse, as well as several behavioral health specialists.” The Fulton County center hired 110 employees including nurses and security staff before opening.

The peer support function: Albuquerque’s three-zone model documents peer support as the primary service at the low-acuity tier, with peers serving as the first point of contact.

The staffing math: Salt Lake City’s Huntsman facility reports 200 mental health professionals required for its scale of operation. The Northwest Arkansas CSU collapsed when staff resignation left four employees — a level documented as insufficient to safely operate the facility.

The retention problem: The Northwest Arkansas collapse began with staff resignations before formal closure — the Becker’s Behavioral Health report documents that underfunding produced departures that accelerated the timeline.

SAMHSA’s 2025 National Guidelines identify 24/7 staffing as a design requirement that must account for all shifts, weekends, leave, and turnover in workforce planning. Memphis staffs 120 professionals; Salt Lake City staffs 200. These figures reflect the arithmetic of continuous coverage at volume, not aspirational ideals.

The Northwest Arkansas collapse accelerated after staff resignations left four employees — a level the facility itself documented as insufficient to safely operate. That sequence illustrates the staffing floor problem: a program with fewer staff than the minimum needed to operate safely is not a program operating at reduced capacity. It is a program that cannot operate.

The Northwest Arkansas CSU collapse began with staff departures before formal closure — the Becker’s Behavioral Health report documents that funding cuts produced turnover that accelerated the timeline. SAMHSA’s 2025 National Guidelines identify workforce planning, including turnover costs, as a prerequisite for program design. SAMHSA’s 2025 National Guidelines identify workforce planning as a prerequisite for program design, not a detail to address after construction.

SAMHSA’s 2025 National Guidelines identify peer support specialist compensation and supervision as staffing design elements that require explicit planning.

Bottom Line

The SAMHSA 2025 National Guidelines describe these design decisions as prerequisites for implementation, not post-opening refinements. The documented collapses — Arkansas and Montana — share the pattern of funding structures decided without adequate operational revenue planning.