Community Assistance Response and Engagement (CARE) Pilot Program Statistical Report

The City of Boulder’s Community Assistance Response and Engagement (CARE) program was launched to provide a non-police response to low-acuity behavioral health, human service and medical calls that did not present with safety concerns. CARE paired a licensed behavioral health clinician with a dedicated paramedic to respond to selected 911 calls where no violence, weapons or criminal activity were involved. The goal was to resolve crises safely in the community, reduce reliance on law enforcement and connect residents to care.

During its first full year of operation (April 2024 through March 2025), CARE completed 246 responses, serving at least 179 distinct individuals. Most people had only one contact with the team and the majority of incidents were resolved without the need for police or hospital transport. Over half of all CARE calls ended with the person remaining safely in the community, while nearly three-quarters of responses resulted in referrals to additional services such as case management, walk-in crisis care or medical treatment.

Strict dispatch protocols, designed to protect both responders and the public, limited the types of calls CARE could handle. This conservative approach led to low call volume and under-utilization of paramedics dedicated to the program. Staff interviews highlighted the value of CARE’s compassionate, health-focused approach—community members often engaged more readily with teams that did not include police—but also pointed to a need for more flexible dispatch criteria and better alignment across partner agencies.

CARE’s clinical case management services emerged as a major strength. Case managers provided short-term, individualized support to help clients access benefits, treatment and housing. This ongoing engagement reduced the likelihood of repeated crises and strengthened collaboration among local service providers.

Overall, CARE proved that non-police crisis response can be safe, effective and well-received by the community. However, the number of calls appropriate for this model was relatively small. The findings suggest that Boulder should retain a non-police option as part of its broader behavioral health continuum, but dedicated paramedic staffing may not be the most efficient structure under current demand. Going forward, refining dispatch protocols, improving interagency coordination, exploring community paramedicine models with a broader scope of practice and expanding case management capacity will help Boulder build a more integrated and sustainable crisis response system.

Communities across the United States are exploring alternatives to traditional law enforcement responses for individuals experiencing behavioral health crises. This report describes an effort by the City of Boulder to send a multidisciplinary team consisting of a behavioral health clinician and a paramedic in response to 911 and police non-emergency line calls for certain types of behavioral health crises. This effort was called CARE: Community Assistance Response and Engagement.

The City of Boulder has a long history of providing alternatives to police-only responses to behavioral health crises. The Crisis Intervention Response Team (CIRT) was launched in February 2021. CIRT is a co-response team of licensed behavioral health clinicians from the city’s Housing and Human Services Department (HHS) who respond with Boulder Police Department (BPD) officers. Both clinical experience of clinicians in the field and input from stakeholders suggested that some incidents likely did not require police involvement to resolve. Additionally, some persons in crisis do not respond well to the presence of a police officer, regardless of how well-intentioned and well-trained the officer may be.

Consistent with best practice in the field and peer programs in Eugene, Oregon and Denver, Colorado, the city wanted pairs of professionals to respond to these calls. There was also anecdotal evidence that some behavioral health crises involved a medical component or that medical problems were co-occurring with behavioral health crises. Boulder Fire-Rescue (BFR) was well-positioned to hire and supervise paramedics who could fill this role and hoped that having dedicated paramedics for health crises could divert lower-acuity calls from the 911 system.

Discussion of these issues between the Boulder City Council, HHS, BFR, BPD, and Boulder Police and Fire Communications (BPFC, Boulder’s 911 call center) started in mid-2022, about a year after CIRT launched, and the CARE pilot program began in December 2023. There was initially one CARE unit consisting of a dedicated paramedic and behavioral health clinicians shared with CIRT. In March of 2024, a second CARE unit was introduced to enhance capability.

This report was prepared by Bauman Consulting Group under contract with the City of Boulder to provide analysis and evaluation services. This report follows a prior report on CIRT’s crisis intervention responses during 2022 and 2023 and focuses on the CARE program. Additional analyses of work done by the team’s clinical case managers, who serve community members referred by CIRT or CARE, are also included in this report.

The initial indicators used by 911 dispatchers for a CARE call included mental health crises, anxiety, panic attacks, depression, delusions, paranoia, hallucinations and suicidal ideation. CARE could be dispatched to these incidents when there was no criminal activity, no threat of violence, and no weapons involved.

The call-taking process initially lacked a structured questioning framework, relying on the call taker’s discretion and a process map adapted from Denver’s STAR program. This process resulted in some dispatcher uncertainty over what types of calls were appropriate for CARE. This uncertainty tended to result in under-utilization of CARE out of an abundance of caution and a tendency to default to more established response structures.

In late March of 2024, BPFC began using structured questioning protocols from the International Academies of Emergency Dispatch (IAED). These protocols are a systematic, guided triage process for emergency medical dispatch based on caller-provided information. The goal was to accurately determine the event type, resources required and response mode. These protocols had previously been in use by BPFC for medical calls and were applied to calls involving behavioral health as well. The protocol was revised again in May 2024, expanding the scope of work for CARE to allow for slightly more uncertainty regarding the breathing status of the person and unknown problems. The final dispatch process map, using IAED protocols, is attached as Appendix II.

Throughout the duration of the CARE pilot program, all departments involved — BPFC, HHS, BFR, and BPD — worked collaboratively to produce a non-police response to crisis calls that met community need without compromising safety. Interviews with staff suggested that this was largely successful, with few situations escalating beyond the training and capacity of CARE teams. As described in more detail below, however, this conservative approach ultimately resulted in low utilization of the CARE team. While not zero, the number of calls for which CARE was appropriate was low.

To ensure that changes in dispatch protocols do not impact this report’s findings, much of the remainder of this report focuses on the one-year period from April 1, 2024, through March 31, 2025, which also aligns with the time period when the CARE team was fully staffed and generally available from 9:00 am to 7:00pm, seven days a week.

CARE was a collaboration between HHS, BPFC, BFR and BPD. At the level of field staff, two paramedics were dedicated solely to CARE and behavioral health clinicians were shared with CIRT. CIRT was an already well-established program with strong ties to BPD both at the administrative level and among field staff.

We conducted interviews with behavioral health clinicians and line-level supervisors, along with management and executive staff at HHS. We also interviewed BPFC and BFR executive staff. Different program visions emerged from these interviews, with HHS staff having envisioned a flexible, non-police response that could maximize engagement and follow-up, while BFR’s approach emphasized rapid response and intervention through protocol-driven deployments. The result was inconsistent expectations around roles, goals and success metrics which is natural when bringing departments with different cultures together as CARE attempted to do.

Clinicians expressed frustration with the strict dispatch protocols. CIRT clinicians are able to assign themselves to incidents where police have been dispatched. CARE, on the other hand, was limited to only those incidents where BPFC’s protocols determined that CARE was the appropriate resource. This was true even when there were incidents involving persons or places to which CARE had previously responded. Clinicians understood that the protocols were in place to ensure the safety of both CARE team members and the community. The clinicians agreed with the dispatch protocols in principle, but in practice, clinicians felt the protocols were overly rigid.

Another issue that caused frustration was that some calls that appeared appropriate for CARE at the dispatch level ended up needing police involvement. This could occur in situations where a third party was reporting that someone was in mental health crisis, but they didn’t know their location. CARE staff did not have access to law enforcement tools such as databases to find someone’s current address, or the ability to ping someone’s phone. BPD also got involved in situations where the subject of the call was intoxicated and needed transport to detox, or when someone involved in the call requested a police report. These situations highlighted challenges with determining the right response during the initial call to dispatch.

CARE paramedics were assigned exclusively to the CARE program. Due to limited call volume, dispatch protocols and a narrowly defined scope of the program, they were not consistently utilized to the full extent of their training — particularly compared to what is possible in a broader community-paramedic model. Dispatch protocols also required that a paramedic perform an assessment to determine whether an individual met defined patient-criteria.

While behavioral health clinicians could flex between CARE and CIRT, paramedics did not have the same cross-assignment options, resulting in periods within their 10-hour shifts without active calls. However, paramedics contributed through follow-ups and supportive outreach when not on CARE calls which is closer alignment to a community-paramedic model. The observed under-utilization reflects the design and scope of the CARE program rather than any limitation of the paramedic role.

Despite these operational and administrative challenges, the staff that were interviewed for this report found value in non-police responses to health crises. Clients would sometimes engage more willingly with teams that did not include law enforcement, and staff reported positive outcomes from case management as well. While some clinicians had safety concerns before the program began, all clinicians we interviewed reported that there were few safety concerns in the field and attributed this to the careful dispatch protocols.

Crisis Intervention and Response

The CARE team consisted of licensed behavioral health clinicians, case-managers and two paramedics. The CARE program shared behavioral health clinicians and case management staff with CIRT; the paramedics were dedicated CARE resources. Responses were requested by the City of Boulder’s Police and Fire Communications Center (BPFC, 911 dispatch) after a community member called 911 or the police department non-emergency line to report a concern. Unlike CIRT, CARE could not assign itself to incidents in the computer-aided dispatch system — responses were dispatched through the 911 call center.

Clinical Case Management

Clinical case managers worked alongside behavioral health clinicians to support individuals following a crisis, addressing the underlying service gaps that often contribute to repeated emergencies. During the CARE analysis period, three clinical case managers provided short-term, intensive assistance to connect people with benefits, treatment, and community resources. Many clients struggled with barriers such as insurance issues, long waitlists, or difficulties accessing medications. Because these needs often couldn’t be resolved during an acute crisis, the addition of clinical case management gave the team a way to sustain engagement and reduce the likelihood of future crises. The model was intentionally low-barrier and person-centered, emphasizing trust and individualized support.

Each case manager typically carried a caseload of 10–15 people at a time, allowing for consistent contact and relationship-building. They coordinated closely with treatment providers, helping align fragmented services and ensure clients received cohesive care. This collaboration extended to community outreach, with case managers strengthening relationships among local providers and advocating for clients who might otherwise be turned away. Clinicians report that case management improved both client outcomes and their own sense of effectiveness, giving them confidence that care could extend beyond the moment of crisis.

There were 246 documented CARE responses from April 1, 2024, through March 31, 2025. The number of responses per day ranged from zero to three, with an average of 0.67 responses per day. More than half (52.3%) of calendar days from April 1, 2024 to March 31, 2025 had zero CARE responses. When there were any CARE responses, the most common number of calls per day was one (32.3% of calendar days), with some days having two responses (11.0% of days) or three responses (4.4% of days).

The average number of responses varied by day of week, with the highest average on Monday (0.96) and the lowest on Saturday (0.48). Table 1 shows the average number of CARE responses for each day of the week.

Table 1: Average count of CARE responses by day of week, April 1, 2024 - March 31, 2025

Since CARE had its own unique call type set up in the dispatch system, demand was measured by the number of CARE calls generated throughout the day, whether the team was in service or not. The left pane of Figure 1 shows a time of day and day of week heat map for this proxy measure of demand. Percentage of the total number of calls is symbolized, with periods of higher demand shown in orange and periods of lower demand shown in blue. More than two-thirds (68.1%) of demand occurred between 9:00 a.m. and 7:00 p.m. Demand was higher earlier in the week, particularly on Mondays, Demand was low between midnight and 7:00 a.m., with only 11.9% of demand during those hours.

The right pane of Figure 1 shows a heat map of CARE responses. CARE responses generally follow the staffing pattern for CARE, with 99.2% of responses beginning during the hours of 9:00 a.m. to 7:00 p.m. Like demand, responses occurred more frequently earlier in the week, particularly on Mondays.

Figure 1: Demand and CARE Responses time of day and day of week heat maps

The average number of hours spent on calls for service per week was 4.9, with an overall monthly average of 21.2 hours. There was no clear seasonal pattern to utilization, as shown in Figure 2.

Figure 2: Total monthly hours on calls for service, CARE

In interviews, multiple CARE staff discussed how some of the low utilization may have been due to lack of availability of a CARE team on some days. Availability was impacted due to illness, vacation or training, which, given the low staffing of the CARE team, could take the team offline for an entire day. Anecdotally, this may explain some of the zero-response days — however it is unlikely to explain the low utilization on days when the team responded to at least one call. On days with at least one response in the community, the average time in the field responding to calls was 1.5 hours, with 40% of days with at least one call having an hour or less spent on responding to requests for assistance. Time in the field is not the entirety of work conducted by CARE staff; both clinicians and paramedics have substantial reporting requirements on top of time spent in the field with community members. Even so, an average of 1.5 hours in the field suggests the team was underutilized.

CARE teams interacted with more than 179 distinct individuals in the 246 responses to requests for service. The exact number of distinct individuals[1] is not known because 34 individuals were anonymous, with no identifying information recorded. Of the 179 individuals that CARE was able to identify, most (87.7%) had one contact with CARE.Thirteen individuals (7.3%) had two contacts, and eight persons (4.5%) had three or more contacts. The maximum number of contacts for any individual was six.

All demographic categories have a relatively large number of people with unknown characteristics. This is largely due to the difficulty of asking for demographic information from people who are experiencing a behavioral health crisis. CARE staff ask community members for identifying information primarily to enable continuity of the person’s care. CARE does not, however, require positive identification when responding to a call for assistance from a community member. The result is often better care for the person experiencing a crisis, but this comes at the expense of data that is less than complete.

As shown in Figure 3, the gender of persons the CARE team responded to was split equally between female (40.2%) and male (39.0%), with 17.1% of unknown or unreported gender. The remaining percentage (3.7%) were transgender.

[1] CARE uses a combination of name and date of birth to avoid creating duplicate records of clients. When duplicate records are found, they are consolidated into a single record for the person.

Figure 3: Gender

Age was not known (or was not recorded) in one out of five contacts (22.0%). Small percentages of people were under the age of 18 (3.7%) or 89 years old or older (1.2%). Young adults aged 18-24 (8.5%) were slightly less common than people aged 25-64. The most common age category among people whose age was known was 55-64, at 14.2% of people.

Figure 4: Age

The majority of CARE contacts were White (58.5%). There were some contacts with Black or African American community members (3.7%), and a small number of contacts with persons of other races (2.8%). Similar to age, a substantial portion of contacts were of unknown or not recorded race (35.0%).

Figure 5: Race

The concern can be thought of as the reason for the response — why were CARE staff involved? CARE teams could record up to three concerns for each response (primary, secondary and tertiary concerns) because people often have multiple concurrent issues. The primary concern is based on the staff’s clinical impression of the predominant issue addressed during the contact. Secondary and/or tertiary concerns are contributing factors to the primary concern. For example, suicidal thoughts (also referred to as suicidal ideation) may be the primary concern for the call, with a secondary concern of trauma. Not all responses had secondary or tertiary concerns.

Figure 6 shows all primary concerns. The most common primary concern in 2024-2025 was suicidal ideation (17.1% of all CARE responses), followed by substance use/intoxication (15.4% of all CARE responses). Situational reactions are third (14.2%). Situational reactions are circumstances where the person’s behavior is driven by a reaction to a specific situational stressor rather than a mental illness or substance intoxication. Examples include a relationship breakup, failing a test, or being fired from a job. Situational reactions are often paired with underlying issues that make persons more vulnerable to behavioral or emotional dysregulation such that external events precipitate a crisis reaction. Medical issues were the fourth most common primary concern, at 12.6% of all responses (31 responses).

Figure 6: Primary Concern, All CARE Responses

Focusing on only the primary concern can tend to undercount some concerns that often co-occur with other concerns. Suicidal ideation, for example, is a primary concern in 17.1% of responses — but also appears as a secondary or tertiary concern in an additional 10.1% of responses, for a total of 27.2% of responses where suicidal ideation is recorded as any concern. Figure 7 shows the percent of all responses where each type of concern is listed as a primary, secondary or tertiary concern.

Figure 7: Any Concern, All CARE Responses

The four most common concerns remain suicidal ideation, substance use, situational reactions and medical issues. Trauma, however, is now the fifth most common concern; when trauma is a concern, it is more often a secondary or tertiary concern to suicidal ideation.

Diversions are noted by CARE when their intervention changed the outcome of the call. Diversion was not applicable in most CARE calls (93.1%). Diversion from higher levels of care or emergency mental health hold occurred in 6.5% of incidents. The low rate of diversion is likely related to dispatch protocols that were specifically designed to dispatch teams other than CARE when the incident was likely to require a higher level of care or police involvement.

The preferred outcome in mental or behavioral health crisis response is typically for the community member to remain in the community, which means the community member was not transported to a higher level of care. Remaining in the community has several advantages over hospital stays, including providing care in a less disruptive manner, improved recovery rates, easier access to support networks, lower health care costs and reduction of stigma.

Overall, more than half (52.0%) of CARE responses resulted in the person remaining in the community. More than one in five (23.2%) resulted in voluntary transport. CARE was unable to contact the person in 17.1% of responses. BPD arrested three people (1.2%) in responses with an initial CARE response requested. In two of these responses, BPD cancelled the CARE response because the party was wanted on a warrant. On the third, an intoxicated community member was charged with assault on an American Medical Response (AMR) paramedic and also had an outstanding warrant.

Referrals to services were common. Referrals were made in 72.8% of responses. Community members could be referred for immediate treatment, or ongoing or future treatment. Community members could also be referred to both immediate and ongoing services. Table 2 shows these types of referrals, the definition of each and examples. The type of treatment referred to during CARE responses appears in Figure 8.

Table 2: Referral Types

Figure 8: Referrals by type of treatment

CARE referred community members to 22 different services. Some services were more commonly referred than others. Community members can be referred to multiple services, depending on their needs. The total number of referrals reported below therefore exceeds the number of incidents.

The most common referral was follow-up by CARE (37.4%), followed by the team’s case managers (27.6%). A referral back to CARE could mean that the clinician was going to follow up with that person by phone or in person, or it could mean that the team provided their contact information to someone on the call with an encouragement for them to reach out as needed. No referrals were made in just over a quarter of responses (27.2%). Table 3 shows the percent of CARE responses that resulted in a referral to each service.

Table 3: Percent of referrals by service, 2024-2025

Case management consists of resource navigation, where persons are directed toward resources, and more intense active case management. Resource navigation is for persons who are high-functioning and able to access services independently, but who need referrals for low acuity concerns. Resource navigation clients may be able to meet their needs using existing services, but they could use some redirection or problem-solving during a period of crisis. Resource navigation clients may also not want more active case management but will accept resource navigation.

The most common needs addressed by resource navigation were mental health services (57.8%), benefits (36.4%) and housing (31.6%). Figure 9 shows the percent of all resource navigation sessions that addressed each need. More than one need can be addressed in a session. Just over 100 persons (103) received resource navigation services during the study period.

Figure 9: Needs Addressed by Case Management: Resource Navigation Clients

Active case management clients receive more intensive coordination of services, typically over a longer time period. Case managers enroll clients into active case management when clients need more support to connect with service providers in the community and address their goals. Case managers meet with clients as often as needed. Thirty-nine people received these services over 460 sessions; some of these people also received resource navigation. The needs addressed for active case management clients were different from those of resource navigation clients. More than half (52%) of active case management clients received services related to needs not specifically listed (i.e., other needs), suggesting a broad array of needs addressed. Healthcare services (43.7%), benefits (38.3%) and mental health services (37.2%) were the next most common.

Figure 10: Needs Addressed by Case Management: Active Clients

The Community Assistance Response and Engagement (CARE) pilot program represented one way Boulder can provide a non-police response to behavioral health crises. Launched in late 2023 as a partnership between Housing and Human Services (HHS), Boulder Fire-Rescue (BFR), Boulder Police Department (BPD) and Police and Fire Communications (BPFC), CARE paired licensed behavioral health clinicians with dedicated paramedics to provide a medical and clinical response outside of traditional law enforcement channels.

After refining dispatch protocols, over its first year of full implementation (April 2024 – March 2025), CARE completed 246 responses, serving at least 179 distinct individuals. CARE’s work focused on crisis intervention, clinical case management and referrals to community services. The majority of clients engaged only once, though a small number had multiple contacts. CARE’s most common concerns included suicidal ideation, substance use, situational stressors and medical issues, often in overlapping combinations.

Program outcomes suggest that CARE responses frequently enabled individuals to remain safely in the community, with over half of all responses ending without hospital or justice system involvement. Diversion rates were modest, largely reflecting the dispatch protocols that limited CARE’s deployment to lower-acuity incidents. Nevertheless, the program demonstrated that a carefully structured non-police response can resolve crises effectively and connect individuals to needed services. Referrals to additional supports were common (73% of responses), particularly to CARE, CARE case management and walk-in or emergency services.

Staff interviews underscored both the strengths and challenges of the pilot. Clinicians, paramedics and supervisors recognized that CARE offered an important alternative to police response and reduced barriers for community members who might otherwise avoid engaging with services. At the same time, differences in organizational culture, rigid dispatch protocols and low call volume contributed to inconsistent expectations across departments, complicating role clarity and leaving paramedics underutilized. These challenges highlight the importance of aligning program design, dispatch protocols and staffing models to the realities of community need.

Overall, the CARE pilot program showed promise as a complement to Boulder’s broader behavioral health response system. The pilot demonstrated that some crises can be resolved without police presence, but that identification of those incidents at the time of dispatch is difficult, given the inherent uncertainty often present during a behavioral health crisis. Low utilization of the paramedic’s skillset within CARE reflects how the program’s dispatch protocols and operational constraints limited opportunities to apply their clinical skillset. This observation relates specifically to the current CARE deployment model and should not be interpreted as a statement about the value or effectiveness of paramedics in other models. Future work should explore how a broader community paramedicine deployment model could leverage the full skillset of paramedics more effectively.

Our work in Boulder and best practices in behavioral health crisis response suggests that having a continuum of responses that involves interprofessional teams of police, behavioral health clinicians, case managers, paramedics and other healthcare professionals is likely to result in the best outcomes for members of the community. Often, the populations involved are difficult to serve, and flexible responses that allow team members to lean on the training and expertise of other team members as needed is crucial to success.

From our analysis of quantitative data and interviews with staff in Boulder, the key finding is that the number of 911 calls that can safely and practically be responded to without a police officer is too small to have resources solely dedicated to a non-police response modality. At the same time, the number of such calls is not zero — a non-police response should exist as one response modality available to the community. One option for this non-police response is a more flexible use of existing crisis team clinicians and case managers, which the City of Boulder has indicated it will pursue. As Boulder considers the future of crisis response, the CARE program’s early lessons provide a foundation for building a more responsive, collaborative and sustainable approach to behavioral health crisis intervention.

AMR: American Medical Response

BFR: Boulder Fire-Rescue

BPD: Boulder Police Department

BPFC: Boulder Police and Fire Communications

CARE: Community Assistance Response and Engagement

CIRT: Crisis Intervention Response Team

EDGE: Mental Health Partners Early Diversion Get Engaged

HHS: Housing and Human Services

IAED: International Academies of Emergency Dispatch

Figure 1: Left pane (Demand)

Figure 1: Right pane (Responses)

Figure 2: Total Monthly Hours on Calls for Service, CARE

Figure 3: Gender

Figure 4: Age

Figure 5: Race

Figure 6: Primary Concern, All CARE Responses

Figure 7: Any Concern, All CARE Responses

Figure 8: Referrals by Type of Treatment

Figure 9: Needs Addressed by Case Management: Resource Navigation Clients

Figure 10: Needs Addressed by Case Management: Active Clients