For EMS directors, fire chiefs, and public health administrators watching this shift from outside a police academy, the headline is simple: training works. A randomized controlled trial out of Louisville found that officers who went through PERF’s ICAT program used force 28% less often, with fewer injuries on both sides of the call. But training has a ceiling. The agencies getting the most out of these federal dollars are pairing it with something no curriculum can fully teach: real-time information about the person on the other end of the call.
What Counts as De-Escalation Training for Police in 2026
De-escalation training today means scenario-based programs that put officers through simulated crisis calls, not a lecture hall and a slide deck. The most established model is ICAT (Integrating Communications, Assessment, and Tactics), developed by PERF and built around slowing down high-risk, low-threat encounters long enough to talk someone down instead of taking action.
The National Policing Institute and PERF are now piloting a hybrid successor called ACCT (Advanced Crisis Communication and Tactics), which folds in elements of Crisis Intervention Team (CIT) training. It’s being tested with departments in Las Vegas, Nevada; Marquette, Wisconsin; and Daphne, Alabama, ahead of a wider release. Several agencies have also moved training into virtual reality, where officers run the same crisis scenario multiple times and get immediate feedback on tone, distance, and timing rather than a single pass-fail run.
The Federal Push Behind the Curriculum
This isn’t a grassroots trend. It’s federal policy. The Law Enforcement De-escalation Training Act of 2022 (Public Law 117-325) directed the Department of Justice to fund exactly this kind of training. BJA’s FY25 De-escalation and Crisis Response Training Program awarded up to $700,000 per agency to departments serving people with behavioral health conditions, intellectual and developmental disabilities, and traumatic brain injuries. A companion program, the FY25 Virtual Reality De-escalation Site-Based Initiative, specifically funds VR training paired with university research partners to evaluate results.
Both application windows closed in June 2026. If your agency didn’t apply this round, the money didn’t disappear; it just moved into implementation. Awarded departments are three years into building out training, and BJA typically reopens similar solicitations annually, so this is the year to watch what gets funded and to start building a case for the next cycle.
What the Research Actually Shows
The strongest evidence so far comes from Louisville. Researchers ran a randomized controlled trial after the Louisville Metro Police Department rolled out ICAT training in 2019, comparing trained officers against a control group. The results, published in Criminology & Public Policy, showed 28% fewer use-of-force incidents, 26% fewer citizen injuries, and 36% fewer officer injuries among trained officers.
A RAND review of the broader evidence base found the research is still thinner than it should be for a tactic this widely adopted, but what exists points the same direction: training changes behavior on the street, not just in the classroom. Early studies comparing VR training to live-action scenarios found the two perform comparably on skill acquisition, with VR showing a slightly bigger reduction in officers’ bias toward people with mental illness.
The Gap Training Alone Doesn’t Close
Here’s what the research doesn’t say: training doesn’t replace the need for behavioral health expertise on the call. ICAT and its successors teach officers to slow down, read a scene, and buy time. What they can’t teach is a clinical read on whether someone is in psychosis, coming down from a substance, or in a mental health crisis tied to a diagnosis the responding officer has no way of knowing about walking in.
That’s the piece of a training curriculum, however well designed, that can’t close on its own. An officer who’s been through 24 hours of ICAT is better equipped to de-escalate a scene. An officer who also has access to a co-responder’s clinical judgment, or to a case history showing this isn’t the person’s first crisis call, is working with a different set of tools entirely.
Where Co-Responder Data Strengthens the Outcome
This is where co-responder programs earn their place alongside training. A review of co-responder team evaluations from the International Association of Chiefs of Police found consistent, if still preliminary, evidence that pairing officers with behavioral health clinicians improves de-escalation outcomes, connects people to services instead of jail, and reduces arrests. No randomized trials exist yet for co-responder models specifically. Still, the pattern across quasi-experimental studies is the same as the one Louisville found for training: outcomes improve when the person facing the crisis receives more than one kind of expertise on scene.
Shared data is what makes that expertise usable in the moment. An officer trained in ICAT still has to make a read in seconds. A co-responder model that surfaces call history, prior diagnoses on file with consent, or a flag that this address has come up before turns that read into something closer to informed judgment. Training teaches the officer how to slow down the moment. Data tells them what they’re slowing down for.
What This Means for Agencies and Grant Writers
If you’re building a case for the next BJA cycle, or reporting back on this year’s award, the strongest applications and outcome reports won’t treat training and co-response as separate line items. Departments that can show a training investment backed by a functioning co-responder or data-sharing infrastructure have a stronger story to tell a funder than training numbers alone. That’s true whether you’re a police department applying directly or a behavioral health partner writing the co-responder half of the proposal.
The federal money is flowing toward exactly the combination that Julota’s mission points to: agencies, clinicians, and community partners working from the same information rather than around each other. That’s not a slogan here; it’s what shows up in the outcome data when training and co-response are built to reinforce each other rather than sit on separate budget lines.
