The Ballroom Session That Changed Nothing

Separate the two diagnoses first. The 92% quiz score proves staff could recognize correct answers immediately after a lecture - that is knowledge, the second level of evaluation. Satisfaction is reaction, the weakest level. Neither predicts behavior.

Signs of a design and transfer problem:

• One massed, off-site session with no practice, role-play, or feedback builds no skill under stress.
• No pre-work, no spaced reinforcement, no job aids, no post-session observation.
• Supervisors were never trained and hold no expectation that the skill is used, so nothing on the unit supports or rewards it.

Evidence still needed before you conclude 'training is the fix':

• Have a unit manager walk through one recent incident step by step. Does the staff member not know what to do, or know and not be able to do it here?
• Check staffing at peak escalation times. If no second person is available, de-escalation is not executable and the barrier is scheduling, not skill.
• Confirm whether the two later incidents involved staff who attended, and what they actually did.

Report a diagnosis, not a verdict: likely a design failure compounded by an environment that does not permit or reward the behavior.

Agree with the policy; disagree that it substitutes for practice.

Argument to make: A signature proves exposure to a document, not capability. Under stress, people default to behavior they have rehearsed, not to language they read once. In a patient-facing setting, 'we issued a policy' is weak evidence if an incident is later reviewed by a regulator, a payer, or a plaintiff.

Concrete counter-proposal:

• Keep the policy, but rewrite it as observable behaviors, not aspirations.
• Reinvest the remaining budget in short, scenario-based practice on the two highest-risk units, and train those unit supervisors first so they can coach and observe.
• Run it as a 90-day pilot with pre/post measures and a decision point, which answers the cost concern with data instead of argument.

Cost framing: put the remaining training budget next to the cost of one incident - injury, restraint, turnover of the involved staff member, and claim cost. The comparison usually ends the debate in one slide.

Build four levels, but weight the last two.

• Reaction: brief post-session survey, used only to fix logistics.
• Learning: scenario-based skills check scored against a behavioral checklist, not a multiple-choice quiz; require a pass to complete.
• Behavior: supervisor observation of at least 10% of staff per month using the same checklist, plus a spot-check on whether job aids are actually in use on the unit.
• Results: incidents per 1,000 patient days, restraint use, injury claims, and turnover on the pilot units, tracked against a pre-training baseline and a comparison unit.

Guardrails:

• Establish the baseline before launch or you will never be able to show effect.
• Watch for reporting drift - a safety push can raise incident counts without a real change in events. Pair the rate with near-miss reporting volume.
• Set the review at 90 and 180 days with a named owner, and publish a one-page result to the COO. Evaluation that does not reach the sponsor does not survive the next budget cycle.