Customized De-escalation Training Shows Early Gains for Long-Term Care Staff

Healthcare workers in mental health and long-term care settings face some of the highest risks of workplace violence of any profession, and the physical and psychological toll on staff can be severe. Many hospitals and care organizations now mandate behavioral safety skills training, yet the evidence on which training methods actually work remains frustratingly inconsistent. A team at the Centre for Addiction and Mental Health (CAMH) in Toronto has now published a detailed account of how they built, delivered, and evaluated a customized version of their de-escalation program for a long-term care facility in Ontario, offering one of the clearest real-world case studies of what tailored safety training looks like in practice, and where it falls short.

The program, called TIDES (Trauma-Informed De-escalation Education for Safety and Protection), was developed at CAMH in 2018 and has been mandatory for the hospital’s own staff for seven years. It spans the full arc of risk management, from prevention and early identification of escalating behavior through collaborative de-escalation, team-based responses, and, when necessary, physical skills to facilitate restraint. The curriculum is grounded in trauma-informed and patient-centred care principles and combines didactic teaching, simulation, and practice-based learning, with evaluation woven throughout using trainee, observer-rated, leadership, and organizational data sources. Trainers are allied healthcare professionals with both clinical and safety training experience who undergo onboarding and follow-up mentoring.

What makes the new report notable is not the training itself but the customization process behind it. The authors argue that a one-size-fits-all approach rarely fits any organization well, because hospitals and care facilities differ in their objectives, their identified safety issues, and their resources. Instead, they adapted two established instructional design frameworks: the ADDIE model (Analysis, Design, Development, Implementation, Evaluation), introduced in the 1970s and among the most widely used methodologies for instructional design, and the needs-assessment-based program planning model described by Kaufman and colleagues. Both frameworks begin by identifying and prioritizing needs, then use that information to drive design, development, implementation, and evaluation.

The site was an Ontario long-term care facility caring for roughly 160 residents annually across five units, employing 330 resident-facing staff. About half the residents are older adults and half have chronic mental illness. The immediate impetus was escalating behavior by a single resident, but the broader context was the COVID-19 pandemic, which had disrupted supports and led to increased police calls for residents exhibiting escalated behavior. Facility leadership contacted the TIDES team in response to this mounting crisis, and the urgency of the situation shaped everything that followed.

The needs assessment combined staff focus groups drawn from different shifts and departments with detailed facility reports on serious incidents of escalated behavior, known as Code Whites, that had led to 911 calls. The TIDES team summarized themes from both sources and discussed them with leadership and staff, producing five objectives that were simultaneously priorities for the facility and feasible targets for a one-day training: reducing responsive behaviors, increasing resident cooperation, decreasing staff and resident injuries, reducing the frequency of Code Whites, and minimizing the need for police intervention. The trainers also proposed additional topics, including perceptions of safety, self-efficacy, and confidence, based on their experience with other organizations.

Evaluation was guided by Kirkpatrick’s four-level framework, which prioritizes organizational results as the central driver of what outcomes to assess. Staff surveys were administered at baseline, immediately after training, and at one-month follow-up, covering injuries, confidence in dealing with escalating residents and police-involved incidents, and training effectiveness. The plan also included observer-rated assessments of physical safety skills and a director survey, plus a longitudinal analysis of monthly administrative data covering the twelve months before and after training. All tools followed standard ethical practices, with informed consent, anonymous collection, and processing by a trained, blinded analyst.

The preliminary results, drawn from 118 trainees who had completed the training at the time of reporting, are cautiously encouraging. For three of the five facility-identified objectives, staff average effectiveness scores rose from 3.8 at baseline to between 4.2 and 4.5 on a five-point scale. Self-efficacy mean scores climbed from 53.4 to 72.3, the proportion of staff feeling somewhat to very safe around aggressive residents rose from 68 to 86 percent, and the share rating the training as effective or very effective increased from 65 to 83 percent. At baseline, 22 percent of staff reported feeling somewhat unsafe or not safe at all at work in the past month, while 58 percent reported feeling somewhat to completely safe, a figure that will be re-examined at follow-up.

Organizational signals were also positive, though more anecdotal. Managers reported that a shared language for understanding trauma, triggers, and behavioral escalation helped integrate the training with existing long-term care practices. Anecdotal reports indicated that calls to police had stopped since training began, and facility leadership requested a three-day Train-the-Trainer curriculum to build in-house capacity, with the parent organization now discussing expansion to two other regional sites and possibly nationally. Open-ended trainee feedback highlighted three improvement themes: the culture of safety when working with older adults, team response to Code White situations, and self-protection skills at the bedside, with requests for more scenarios, clearer team leadership roles, and even a physical bed for practice.

But the report is equally candid about what was lost to pragmatism. The observer-rated skills assessments were dropped as impracticable given the facility’s limited time and space, and leadership changes at the facility derailed the director survey, meaning a true pre-training organizational baseline is now impossible. More troubling, the one-month follow-up staff survey has so far returned a non-response rate of nearly 90 percent, leaving no way to document even self-reported skill retention. This matters because the team’s previous work found that while self-reported confidence remained high a month after training, observer-rated skill retention dropped by 25 to 33 percent depending on the skill. Self-reported confidence, in other words, is not an adequate proxy for objectively measured ability.

The authors frame their contribution honestly: a theory-grounded, replicable customization process that demonstrates short-term staff-level outcomes aligned with organizational priorities, but no evidence yet on organizational-level impact, sustainability, or resident perspectives. They suggest that future evaluations could be reported in formats amenable to meta-analysis, with means, standard deviations, and raw frequencies readily available, and point to realist reviews that identify configurations of context, mechanism, and outcome rather than single factors. The key ingredient in their experience was early and continual engagement of the facility at multiple levels, from administration to frontline staff, which staff reportedly appreciated deeply. Conceptual frameworks, they conclude, remain valuable road maps even when pragmatic constraints force deviations from them, because they expose exactly which gaps remain and guide more tailored, actionable decisions about what to measure next.

Subject of Research: Development and evaluation of a customized workplace violence prevention training program for long-term care staff

Article Title: Developing and evaluating a customized evidence-based safety training program for healthcare staff: a quality improvement approach

Article References: Lin, E., Boyd, H., Bratsalis, E., Hodgson, K., Islam, F., Khan, M. I., Crestois, N., & Sockalingam, S. (2026). Developing and evaluating a customized evidence-based safety training program for healthcare staff: a quality improvement approach. Frontiers in Psychiatry, 17, 1901445. https://doi.org/10.3389/fpsyt.2026.1901445

Image Credits: AI Generated

DOI: 10.3389/fpsyt.2026.1901445

Keywords: workplace violence, safety training, long-term care, de-escalation, trauma-informed care, quality improvement, Kirkpatrick framework, ADDIE model, healthcare staff, mental health care, program evaluation, CAMH