Expanding Beyond Policing: Rethinking the Meaning of Public Safety

BY BETSY GARDNER • June 14, 2022

We need to start redefining 911. We need to redefine public safety. We don’t need to figure out how to ‘make it work,’ instead we need to redefine it and ask, ‘how do we do it differently?’ Because it’s time.

Bridgette Dean, Director of the Sacramento Department of Community Response


According to the Washington Post’s Fatal Force database 1,045 people were shot and killed by police in the past year (as of May 15, 2022). Community members, advocates, and many public officials are exploring new models of public safety that remove armed officers from the scene of non-violent incidents or mental-health crises as one way of reducing fatal police encounters. One year ago we wrote about the equity-centered and mental health focused work in Denver, New York City, and San Francisco. Each city was developing their own model of an alternative or co-response to 911 calls that prioritized the health and safety of the community. By bringing in crisis first responders with specialized training to certain emergency calls, cities could strengthen trust and reduce community-police conflict, reserve police efforts for calls that are better fits for their training and reduce use of force – plus provide those in mental health distress with suitable care.

While many folks long advocated for new models of public safety, to specifically address racial disparities in arrests and deaths, there was much broader public attention to this issue in the summer of 2020. Since then, more and more cities have explored alternative responses (which send out non-law enforcement teams) or co-responses (which pair mental health/crisis professionals with a law enforcement officer). In our meetings with city leaders from across the country, we have seen interest in this work increase exponentially, with officials trading ideas for funding, models, and messaging and requesting lessons from cities already diving into this work.

Based on the Harvard’s Government Performance Lab’s (GPL) alternative to 911 Emergency Response Initiative in Durham, NC; Harris County, TX; Long Beach, CA; Philadelphia, PA; and Phoenix, AZ this article will discuss different models in these and other cities, present both best practices and potential challenges, and provide advice for local governments.

Why Multiple Models:

CAHOOTS (Crisis Assistance Helping Out on the Streets) in Eugene, Oregon. CAHOOTs is a promising model with a strong track record of successfully responding to “social service type calls,” but depending on the needs of the jurisdiction, may not be the best fit in every case. It’s important for a wider variety of models to be developed and tested to give policymakers more options as they consider their communities’ needs and varying local government structures.

The GPL team understands these nuances, as they’ve spent the past year working with the five different jurisdictions on alternatives to 911. According to Ana Billingsley and Gabriela Solis, assistant director and project leader respectively, the team purposefully chose five jurisdictions that varied in size, approach to alternative response, and resourcing to support. In doing so, they better understand best practices and make recommendations that are tailored to a variety of different places, governments, and regulations.

For example, Harris County, Texas had a strong tradition of law enforcement diversion work at the sheriff level already, so the county’s alternative to 911 work is being spearheaded by the county public health department in close collaboration with the sheriff, while Philadelphia operates a city-run co-response model (detailed below). It’s important for the GPL team to present a variety of actionable, successful options so that all cities can identify a model that can work for them.

In Sacramento, Director of the Department of Community Response (DCR) Bridgette Dean explained that the city is currently using its 311 line for alternative response calls. Initially, when Dean started as director of the newly founded DCR in 2020 they were using 911, but due to an evolving relationship with the police department, her DCR team moved over to 311 by April 2021.

There are drawbacks to moving these calls to 311. Dispatchers aren’t traditionally trained for emergencies, so this switch requires additional training and support; plus, many of the calls are based on location rather than an individual. The Sacramento Police does still direct pertinent calls to DCR, and although Dean’s team doesn’t have access to the police notes from these calls, there is inherently more flexibility and space for her to build this work through 311. And the community has adapted to using 311 for their alt-911 calls (for issues like mental health and homelessness) thanks to a strong campaign to rebrand 311 as the outlet for these services.

For Dean, moving out of the 911 system wasn’t the original plan, but she saw it as a step toward a true alternative response, where the DCR operated as a separate unit. Ultimately Dean wants the DCR to become the third point on the emergency response triad (with fire and police) and be able to directly support the community at times of need, independently of law enforcement.

In Philadelphia, the model looks quite different; city leaders are dedicated to several mental and behavioral health responses, including training police in crisis intervention, running an overdose response team through the fire department and EMS, and piloting the co-response Crisis Intervention Response Team (CIRT) through 911. Julia Hinckley, the Deputy Chief of Staff and Policy Director for Mayor Jim Kenney, explained that the city is dedicated to working with departments on multiple levels in order to best utilize all the resources and sort through the different “decision trees of 911 calls.”

The CIRT pilot is a co-response model, which means that a police officer with crisis intervention training and a master’s level behavioral health clinician from a local nonprofit are paired together to respond to 911 calls coded as behavioral health, in an unmarked police vehicles. While there is also a current pilot with the Department of Behavioral Health and Intellectual Disability Services that is an alternative, non-police behavioral health response, CIRT represents a concrete step toward a vision of public safety that branches off the traditional police response while still remaining under the broader policing umbrella.

For Hinckley and her colleague, Deputy Policy Director Mariele McGlazer, coordinating these “multiple doors” and bringing together multiple departments for co-response works because “everyone is philosophically on the same page” when it comes to reducing unnecessary police presence at non-violent calls for service. McGlazer reported that in focus groups many officers reported that they hadn’t felt equipped or well trained to respond to mental/behavioral crisis calls and felt that their involvement in these calls could be both inappropriate and could divert resources from efforts to address serious crime. However, she also stressed that “the devil is in the details” when it comes to the cultural shift that comes with partnering response teams across departments and disciplines, and the city is still working to bring the projects from pilot to scale.

Groundwork for Racial Equity:

An important aspect underlying and grounding this work, for all cities, is racial equity. As discussed in our most recent piece, Black people are more than three times as likely to be killed during an encounter with police than white people, an extreme and disturbing disparity. Many cities have ramped up their non-police or co-response programs since 2020 and the high-profile murders of Black Americans by police officers as a way to address and prevent these deadly encounters. However, weaving racial equity into alternative or co-response programs relies on having pre-existing and transparent racial equity work at the city or county level.

Long Beach, California is another jurisdiction in the GPL cohort. City leaders had already done racial reconciliation work at the local level and were able to implement their alternative to 911 program with the support and trust of the racial reconciliation committee. In Phoenix, the city is prioritizing hiring non-police responders from the neighborhoods they will be working in, making sure that the folks working in those areas are familiar with and representative of those communities. Solis pointed out that the hiring in Phoenix is particularly thoughtful, as local leaders are looking beyond specific credentials and academic markers to create a more inclusive recruitment and hiring pool. Rather than exclusively targeting clinicians, who are 75.6 percent female and 76.3 percent white according to 2021 data from the Bureau of Labor Statistics, Phoenix is looking to peer support specialists and bachelor’s level behavioral health workers where there is more diversity.

In Philadelphia, city leaders have already been working on closing racial disparities in arrests and legal system involvement. With this background, officials are applying the racial equity lens to this work by monitoring outcomes to the co-response model by race; using anonymized data, the city is breaking down CIRT data by race to identify racial inequities. Hinckley and McGlazer know that there are already high rates of racial disparity in police response and arrests and view the CIRT model as a way to reduce arrests — both overall and by race. This is supported by the other crisis intervention and behavioral health responses (what Hinckley referred to as the “decision tree” of possible 911 responses) that also prioritize racial equity.

The Role of Data:

While data has a crucial role in this work — tracking outcomes and improvements by race, deciding how many calls can be shifted from police to an alternative response, measuring and defining success — it is challenging to access, clean, and understand data across multiple departments in cities and counties. The GPL team has spent a significant portion of their time assisting cohort cities with data work and has identified several important considerations for local jurisdictions looking to implement an alternative or co-response program.

All cities should start their planning with a review of their emergency calls to figure out which ones are both safe and suitable for a response other than traditional police. This information is imperative for planning a team as it helps cities make data informed decisions about eligible calls for alternative teams, location of teams, team schedules, team training and the number of teams needed to meet the unique local demand.

When starting their review of calls, cities should be prepared to find the best way to leverage the 911 data available, knowing that no data set is perfect. While each city can separate the calls along different criteria (i.e., some cities will only have an alt response for mental health crisis calls while others include calls about trespassing, homelessness, and neighbor disputes), the general exclusionary criteria is violence and/or the presence of a weapon. However, Solis explained that this isn’t easy to identify in 911 call data, as these criteria often are not tracked in call notes and are difficult to analyze. Instead of relying solely on information collected at the time 911 call was made, governments should supplement 911 call data with call disposition data to understand how the call ended on scene. In Long Beach, CA, the planning workgroup honed in on non-violent eligible calls by digging deeper into which calls resulted in non-criminal justice related dispositions, such as “assisted citizen” or “left a note”, These dispositions, the planning work group felt, were an indicator of appropriateness for alternative response.

Of course, data isn’t only required in the beginning. In reviewing existing practices, Dean pulled data on the police department’s co-response model that partnered a clinician with a police officer; unfortunately, the clinicians reported that they didn’t have much control during responses. Pulling more data and research, Dean found that after an officer and clinician responded to an emergency call, the same callers made more than nine follow up calls; yet data shows that a clinician-based or solely social worker response significantly cut redundant calls, with 71 percent of those callers never needing to follow up for further emergency assistance.

As discussed in an earlier section, the Sacramento DCR has run into data sharing issues now that it’s operating through 311 and not 911. Dean is strongly advocating for better open data and a single data system to share notes between fire, police, and DCR. Initial learnings from the GPL shows that the most successful cities are ones that were able to include data experts like the chief data officer from the beginning, since they are in the perfect position to track, iterate, and evaluate data from multiple sources and have experience effectively communicating data to internal and external stakeholders.

Defining Success:

For any alternative or co-response program, there is a significant amount of pressure to deliver results, but measuring success is possible on several different metrics — although not all are equally measurable.  According to Solis, the primary measure of success for alternative emergency response interventions is whether teams are matched to the right calls and respond effectively. Tracking how often teams have to call for police backup can be one indicator of whether the right calls are being diverted to the alternative response team—for example in Portland after one year in operation, the Portland Street Response teams called for police back up in less than 3 percent of calls. The GPL team is also tracking the potential for cost savings from reducing emergency room visits (often the final destination for mental health care when police respond), but they ultimately want to focus on using funding most effectively rather than reducing funds.

In Philadelphia, one measure of success for city leaders is a reduction in the most restrictive level of psychiatric care, an involuntary commitment (known as a Section 302). Hinckley reported that they’ll be collecting data on how many interventions/interactions result in any kind of restrictive outcome, which she defined as both a 302 and/or an arrest. Their goal is to increase connections to community services by decreasing restrictive outcomes, something that they will also be breaking down by race to ensure that these changes are applied equitably.

For Dean, the Sacramento community is responsible for determining if the alt-response is a success. “We are client centered which means they determine the need,” said Dean. “A key piece of alt-911 response models is saying ‘what do you need, what do you want to do, where do you want to go’.” One achievement she noted is the reduction in calls about homelessness to the SPD. Thanks in part to the federal Homeless Management Information System (HMIS), folks across the city have a shared understanding of the data on Sacramento residents experiencing homelessness and can all see call responses and service delivery. This is an important indicator of broad community support for the alternative response, because as Dean explained “homelessness is a microcosm of all of the alt-911 needs in one group.”

Luckily for cities just getting started with their planning, there are now multiple teams to look to for how to think about defining success. Aloka Narayanan, the GPL Innovation Fellow on the alternative response team, worked to gather and organize key performance metrics for the alt-911 city cohort. She reviewed open data available in cities like Denver and Portland that have established alternative response teams and categorized the data metrics into categories, listed here as a resource for governments planning alternative 911 emergency response teams.

Knowing the variety and total number of impact metrics means that cities can focus across these categories and bring in relevant stakeholders early in the alternative response process, plus they can also judge the correct model of alt- or co-response based on local funding and capacity. Since many jurisdictions are also focused on decreasing disproportionality in policing and the legal system more broadly, metrics can also help keep a focus on equity in implementation.


This is some of the most important work we can be doing as a local government, to build trust and improve our response to people in what can be their worst, most challenging moments.

Mariele McGlazer, Deputy Policy Director for Philadelphia


Data-Smart is committed to continuing to share resources and best practices around non- and co-police responses, and any cities interested in implementing a similar model are encouraged to join the GPL’s Alternative 911 Emergency Response Community of Practice.

About the Author

Betsy Gardner

Betsy Gardner is the editor of Data-Smart City Solutions and the producer of the Data-Smart City Pod. Prior to joining the Ash Center, Betsy worked in a variety of roles in higher education, focusing on deconstructing racial and gender inequality through research, writing, and facilitation. She also researched government spending and transparency at the Lincoln Institute of Land Policy. Betsy holds a master’s degree in Urban and Regional Policy from Northeastern University, a bachelor’s degree in Art History from Boston University, and a graduate certificate in Digital Storytelling from the Harvard Extension School.