Boundaries and escalation
The things a CHW must never do, how to recognize a red flag, and how to route crisis and out-of-scope situations to a licensed supervisor quickly and calmly.
Learning objectives
By the end of this module, the CHW can:
- Recite the do-not-do list without hesitation.
- Recognize the red flags that require escalation to a licensed supervisor.
- Respond safely to a crisis disclosure and route it correctly.
- Explain why boundary skills fade and why boosters focus on them.
Facilitator outline
The trust clients place in a CHW is exactly why staying in scope protects them. Set a serious but non-anxious tone.
Walk each item below and give one concrete example of what it looks like in a session, and what to do instead.
Walk the escalation card: which situations trigger escalation, what the CHW does immediately, what goes to the licensed supervisor and in what timeframe, and crisis routing to 988 or 911.
Run the role-play below. Practice the immediate response and the handoff, calmly and without leaving the person alone.
Boundary and "do-not-do" skills fade faster than anything else. That is why the 6 to 8 week booster focuses here. Close with a teach-back on the red flags.
The do-not-do list
These are the actions a CHW never takes, no matter how confident an AI answer sounds or how much a client asks.
- Do not diagnose. Naming a condition is a clinician's job. Describe, support, and refer.
- Do not treat AI output as medical advice. An AI answer about symptoms, conditions, or medications is never a clinical recommendation, no matter how detailed.
- Do not give medication guidance. Any question about starting, stopping, dosing, or combining medication goes to a licensed professional.
- Do not give legal or immigration advice. Route to a qualified legal resource; do not interpret law from an AI answer.
- Do not discourage professional care. The tool and the CHW support reaching care; they never replace it.
- Do not let AI stand in for an interpreter on consent, diagnosis, treatment, or insurance. The client has the right to a qualified human interpreter.
When a red flag appears
The routing is the same every time. You do not leave the client, you loop in a licensed supervisor within the agreed window, and any threat to life goes to a crisis line immediately.
Structured lay-provider training programs test boundaries by scoring the harmful actions a provider must not do, using role-play (the EQUIP and ENACT approach; Kohrt and colleagues). We use the same method in the role-play kit.
Among trained frontline health workers followed for 18 months, the skills that decayed most were the "do-not-do" steps, and refresher training was recommended (Gobezayehu and colleagues, 2014). Inoculation-style skills also fade within about two months without a booster (Maertens and colleagues). This is why the booster schedule targets boundaries and spot-the-error practice.
Role-play: crisis disclosure mid-session
Pair up. One CHW plays a client who, during a hands-on activity, quietly says something that signals risk; the other responds. The client says:
The responder's task is to stay calm, take it seriously, not leave the person alone, and follow the escalation card: respond warmly, ask directly and simply, and move to crisis routing (988, with interpreters, or 911 if there is immediate danger) while looping in the licensed supervisor. Good facilitation does not minimize, does not try to counsel or treat, and does not hand the person a chatbot. Debrief against the card and name what the CHW did that kept the person safe.
Materials
- Escalation rule cardPrintable
- If you are in crisisGuide
- What to ask AI, and what to ask peopleGuide
- Your right to an interpreterGuide
- Can each CHW recite the do-not-do list and the red flags from memory?
- Did the crisis role-play end with correct routing and a supervisor loop-in, not counseling?
- Did I make clear that these are the skills we deliberately refresh at the booster?