The Mental Health Crisis Calls That Most Often Result in Force

Cross-referencing the mental health crisis literature with use-of-force data to identify where escalation is most preventable, and what coordinated response looks like.

Few categories of police work are discussed more, and measured worse, than encounters with people in mental health crisis. The encounters are common, they are disproportionately represented among the most serious outcomes, and the data describing them is fragmented across federal collections, non-governmental databases, and academic studies that rarely use the same definitions. This article assembles what can be said with reasonable confidence, flags what cannot, and draws out the training implications.

How often crisis is present when force is used

The most-cited national figure comes from the Washington Post's police shooting database, which has tracked fatal shootings by on-duty officers since 2015. Across that record, roughly one in five to one in four people fatally shot by police showed signs of mental illness at the time of the encounter.¹ The Post's running tally has identified more than 1,400 people with mental illness among those fatally shot since the database launched.¹ For a single year of reference, an analysis of 2015 data found that about 23 percent of people killed in police encounters displayed signs of mental illness.¹

A separate and narrower figure is also useful: roughly 12 percent of fatal police shootings occurred while officers were specifically responding to a behavioral health call.¹ The gap between that 12 percent and the 20-to-25 percent "signs of mental illness" figure is instructive. It tells us that in most fatal encounters involving a person in crisis, the call was not dispatched as a mental health call. Officers arrived for a reported disturbance, a welfare check, a trespass, or a suspicious-person report, and the crisis became apparent only on scene, or never became formally classified at all.

The most striking single statistic in this area comes from the Treatment Advocacy Center's 2015 report "Overlooked in the Undercounted," which estimated that people with untreated severe mental illness are roughly 16 times more likely to be killed during an encounter with law enforcement than other people approached or stopped by officers.²

Two honest caveats apply to all of these numbers. First, none of them is drawn from a complete national count, because no such count exists for crisis encounters specifically. Second, and more important, these figures describe the worst outcomes, the fatal ones. They say very little about the vastly larger number of crisis encounters that end without force, and nothing at all about how often skilled response prevented force. The denominator problem is severe: we can count the tragedies far better than we can count the successes.

Why crisis calls escalate

The crisis intervention literature converges on a consistent set of escalation drivers, and they map cleanly onto how a call unfolds in time.

Misclassification at dispatch. Because most crisis encounters are not dispatched as mental health calls, officers frequently arrive primed for a different problem, with a tactical posture matched to a disturbance or a crime rather than to a person in distress. The framing of the first thirty seconds is set before the officer ever sees the subject.

Pace and proximity. Crisis behavior, including a person who is non-responsive to commands, agitated, or holding an object, often triggers a rapid closing of distance and a rapid escalation of verbal intensity. The literature on crisis de-escalation consistently identifies time and distance as the two resources most often surrendered early and most valuable when preserved.

Command-and-control communication. A person in psychiatric crisis may be unable to comply with rapid, repeated verbal commands, not unwilling. When non-compliance is read as defiance rather than incapacity, the encounter escalates on a misunderstanding. This is the single mechanism that crisis-specific training most directly targets.

Absence of a coordinated alternative. When officers have no co-responder, no mental health clinician, and no ready transport-to-treatment option, the realistic menu of outcomes narrows to arrest or force. Coordinated response models exist precisely to widen that menu.

What the CIT evidence actually shows

The Crisis Intervention Team (CIT) model, developed in Memphis in 1988 and disseminated nationally with support from CIT International, is the most widely adopted framework for police response to crisis. Honest reporting on its evidence base requires presenting both the encouraging and the disappointing findings.

On the encouraging side, the research is fairly consistent at the officer level. CIT-trained officers show greater knowledge of mental illness, improved attitudes, reduced stigma, and greater self-efficacy in crisis encounters.³ In scenario-based studies, CIT-trained officers were significantly more likely to report verbal engagement or negotiation as their highest level of force, and they consistently rated non-physical actions as more effective, and physical force as less effective, than non-CIT officers did.⁴ One frequently cited finding is that CIT-trained officers used force in only about 15 percent of encounters rated as high violence risk, and that when they did use force they tended toward lower-lethality methods.⁴ CIT-trained officers are also more likely to refer or transport a person to treatment, and less likely to arrest, than officers without the training.³

On the disappointing side, the picture for objective system-level outcomes is weaker. Multiple reviews note that there is limited peer-reviewed evidence that CIT reduces objective measures such as arrests, officer injury, citizen injury, or use of force at the agency level.³ Because CIT programs vary enormously in their components and in the supporting infrastructure around them, study results range from no measurable effect in some jurisdictions to a moderate effect in others.³

The reconciliation of those two findings is the practically important point. CIT reliably changes how officers think, what they know, and what they intend to do. It does not reliably change system outcomes on its own, because system outcomes also depend on dispatch classification, the availability of clinicians and crisis facilities, and the realistic alternatives to arrest. Training is necessary. It is not sufficient. A program that improves officer skill but is dropped into a system with no treatment destination will show muted results, and that is a finding about the system, not a verdict on the training.

Coordinated response: what staging and co-response add

The crisis literature increasingly distinguishes between training officers to respond better and changing who responds and with what support. Co-responder models pair officers with mental health clinicians; some jurisdictions route appropriate calls to clinician-led teams without a police presence at all. The evaluation evidence on these models is still maturing and varies by jurisdiction, so this article does not assert a single national effectiveness figure.

What can be said is structural. When a clinician is available, the encounter has a person whose entire role is assessment and connection to care rather than scene control. When a transport-to-treatment destination exists, the binary of arrest-or-force is broken. And when crisis calls are correctly identified at dispatch, the responding officer arrives with the right mental model. None of these are training interventions in the narrow sense, but training is what prepares an officer to use them: to recognize a crisis quickly, to hold time and distance while a co-responder arrives, and to hand off effectively.

What this means for your agency's training

The data and the literature, read together and read honestly, support a focused set of conclusions.

Crisis is present in a substantial minority of the most serious encounters, but it usually arrives mislabeled. Training therefore cannot be confined to calls dispatched as mental health calls. Officers need to recognize crisis indicators inside calls that came in as something else, because that is where most crisis encounters actually live.

The officer-level evidence for crisis training is solid: it improves recognition, attitudes, and the intention to use verbal tactics first. Agencies should adopt it for that reason, while being honest that training alone will not move agency-wide injury or arrest statistics if the surrounding system offers officers no alternative to arrest or force.

The escalation drivers are specific and rehearsable: misreading the call, surrendering time and distance, and interpreting incapacity as defiance. These are not abstractions. They are concrete decision points that a scenario can reproduce and an officer can practice.

CodeBlu's scenario-based approach is well matched to the recognition-and-decision problem at the center of crisis response. The honest claim is bounded: scenario rehearsal builds the officer-level skills that the CIT literature shows training can reliably build. It does not substitute for the dispatch reform, clinician availability, and treatment-destination infrastructure that the same literature shows are also required. Agencies should pursue both, and should measure their own crisis-encounter outcomes rather than relying on national figures that were never designed to evaluate a single department.

Recommended CodeBlu scenarios this article supports

1. Crisis recognition inside a misclassified call: a scenario dispatched as a disturbance or welfare check that is, in fact, a person in psychiatric crisis.
2. Holding time and distance: rehearsing the deliberate choice to slow the encounter and preserve space while additional resources arrive.
3. Incapacity versus defiance: practicing the read on a subject who cannot, rather than will not, comply with verbal commands.
4. Co-responder handoff: coordinating with and transferring scene communication to an arriving mental health clinician.
5. Subject in crisis holding an object: a decision-point scenario focused on the most common high-risk crisis configuration without defaulting to closing distance.

---

Footnotes

1. Washington Post, "Police shootings database, 2015 to present." https://www.washingtonpost.com/graphics/investigations/police-shootings-database/ (accessed May 2026). Figures on share of people fatally shot showing signs of mental illness, the cumulative count of more than 1,400, and the 12 percent behavioral-health-call share are drawn from analyses of this database. ↩ ↩² ↩³ ↩⁴

2. Treatment Advocacy Center, "Overlooked in the Undercounted: The Role of Mental Illness in Fatal Law Enforcement Encounters," 2015. https://www.tac.org/wp-content/uploads/2023/11/Overlooked-in-the-Undercounted.pdf ↩

3. R. Rogers et al., "Effectiveness of Police Crisis Intervention Training Programs," Journal of the American Academy of Psychiatry and the Law, 2019. https://jaapl.org/content/early/2019/09/24/JAAPL.003863-19 ↩ ↩² ↩³ ↩⁴

4. M. Compton et al., research on use-of-force preferences and perceived effectiveness among CIT and non-CIT officers. https://pmc.ncbi.nlm.nih.gov/articles/PMC3122295/ ↩ ↩²