Training hub Kit
Role-play kit
Scripted client scenarios for practice, plus an observation checklist a trainer uses to decide when a CHW is ready to deliver on their own.
Run each scenario in pairs, then rotate. One person plays the client from the script; the other facilitates. A trainer or peer observes with the checklist at the bottom of this page. Structured role-play with an itemized checklist is the standard way lay health workers are trained and assessed (the ENACT and EQUIP approach; Kohrt and colleagues), including scoring the harmful actions a facilitator must not take.
How to run a scenario
- Read the setup aloud. The client player uses the opening lines, then improvises in character.
- The facilitator responds in real time. Keep it to 4 to 6 minutes.
- The observer scores the checklist and notes any harmful behavior.
- Debrief with the questions provided, then switch roles.
Client brings AI misinformation
Setup
A client arrives convinced of a health claim they got from a chatbot or social media. The claim is wrong and could keep them from care.
Client opening lines
"I read that antidepressants permanently change your brain and that therapy is just talking, so there's no real point. The AI explained the whole thing. So I don't think I need to go."
What good facilitation looks like
- Stays warm and curious; does not argue or shame the client for what they read.
- Separates the claim from the client: "Let's look at where that comes from together."
- Models the checking habit: leave the answer, check who says so elsewhere, or bring it to a clinician.
- Redirects to a trustworthy next step, such as asking a provider directly at a visit.
Harmful behaviors to avoid
- Debating clinical facts as if the CHW were the medical authority.
- Offering a counter-diagnosis or telling the client what medication to take.
- Dismissing the client ("that's just nonsense") in a way that ends the conversation.
Debrief questions
- Did the facilitator address the claim without becoming the medical expert?
- What checking step did the client leave with?
Client overtrusts the chatbot
Setup
A client treats every AI answer as final and is about to act on specific details the tool produced, without verifying them.
Client opening lines
"The chatbot gave me the clinic, the phone number, and said I qualify for free care. I already have my whole plan. I don't need to call anyone to check."
What good facilitation looks like
- Coaches the habits from Module 2: guess first, treat the answer as a draft, expect uncertainty.
- Reframes the AI's plan as a strong first draft to confirm together.
- Names one concrete verification step, such as calling the clinic to confirm coverage and new-patient status.
- Lowers trust to a useful level without pushing the client to abandon the tool.
Harmful behaviors to avoid
- Confirming the invented details as true because they sound plausible.
- Swinging the client to the opposite extreme so they distrust all help and stall.
Debrief questions
- Which habits did the facilitator actually coach?
- Was the client's trust calibrated, or just replaced with fear?
Client fears data sharing
Setup
A client is afraid that anything typed into the chatbot could reach immigration authorities and is ready to opt out entirely.
Client opening lines
"What if immigration can see this? I've heard they pull everything. I don't want my name in there. Maybe I just shouldn't use it."
What good facilitation looks like
- Treats the concern as reasonable and does not talk the client out of caution.
- States the actual protections using the transparency guide's wording, no stronger.
- Gives two or three concrete rules the client controls: ask general questions, never enter status or ID numbers, cover identifiers on any photo.
- Leaves the client both safer and still able to use the tool.
Harmful behaviors to avoid
- Promising total safety or overstating what LINC guarantees.
- Dismissing the fear so the client disengages from care altogether.
- Encouraging the client to enter status or ID numbers "because it's fine."
Debrief questions
- Was the reassurance honest about both protections and limits?
- Did the client leave with rules they control?
Crisis disclosure mid-session
Setup
During a hands-on activity, a client quietly discloses thoughts of self-harm. This scenario practices the most important handoff in the training.
Client opening lines
"Some days I don't see the point of any of this. Last week I thought about not being here anymore."
What good facilitation looks like
- Stays calm, takes it seriously, and does not leave the person alone.
- Responds warmly and asks directly and simply about safety.
- Follows the escalation card: crisis routing to 988 (with interpreters) or 911 for immediate danger, and loops in the licensed supervisor.
- Stays with the person through the handoff.
Harmful behaviors to avoid
- Minimizing ("you'll be fine") or changing the subject.
- Trying to counsel or treat the person instead of routing to crisis support.
- Handing the person a chatbot for a crisis. Crisis is always human.
- Leaving the person alone to make the call.
Debrief questions
- Did the facilitator route correctly and involve the supervisor?
- What specifically kept the person safe in the moment?
Medication question beyond scope
Setup
A client asks the CHW to help decide about medication, and wants to use an AI answer to settle it. This is out of scope.
Client opening lines
"The AI says I can stop my medication now that I feel better, and it explained the dose to taper. Can you just help me do what it says?"
What good facilitation looks like
- Names the boundary plainly and kindly: medication decisions are for a licensed professional.
- Explains that an AI answer about medication is never a clinical recommendation.
- Helps the client draft the question to bring to their prescriber, and helps them get an appointment or interpreter.
- Escalates per the card if there is any sign of immediate risk from stopping.
Harmful behaviors to avoid
- Advising the client to start, stop, taper, or combine any medication.
- Reading an AI dosing answer aloud as if it were medical guidance.
- Reassuring the client that stopping is fine.
Debrief questions
- Did the facilitator hold the boundary while still helping the client move forward?
- What did the client leave equipped to do (bring the question to a prescriber)?
Clinic refuses an interpreter
Setup
Front-desk staff push the client to use a family member or a phone app instead of a qualified interpreter for an intake appointment.
Client opening lines (playing clinic staff)
"We're short today. Can't your daughter translate? Or just use the app on your phone, that's usually fine for the forms."
What good facilitation looks like
- Coaches the client to ask clearly for a free, qualified interpreter and to decline the shortcut politely but firmly.
- Gives the client an actual sentence: "I have the right to a free interpreter for this, and I'd like one for the consent form."
- Names when AI or an app is not enough: consent, diagnosis, treatment, insurance.
- Stays calm and keeps the client in control of the request.
Harmful behaviors to avoid
- Telling the client to just use the app to save everyone time.
- Allowing a child or family member to interpret a consent or diagnosis conversation.
- Backing down as soon as staff resist.
Debrief questions
- Did the client leave with a concrete script and the confidence to use it?
- Did the facilitator name the situations that require a human interpreter?
Observation checklist
The observer rates each item as needs work, adequate, or good. Items marked in red are safety items. Use this during observed practice to decide readiness for solo delivery.
| Facilitation item | Needs work | Adequate | Good |
|---|---|---|---|
| Builds rapport; warm and respectful with the client | |||
| Uses plain language; avoids jargon and technical AI terms | |||
| Uses teach-back to confirm the client understood | |||
| Opens and checks the AI output together with the client | |||
| Coaches "guess first" and treats AI answers as drafts | |||
| Names the language limits and the interpreter right when relevant | |||
| Gives calibrated reassurance; neither overpromises nor frightens | |||
| Protects privacy: no status or ID numbers entered; photos covered | |||
| Stays in scope: no diagnosis, no medication, legal, or medical advice | |||
| Recognizes red flags and escalates appropriately | |||
| Responds calmly and safely to distress or crisis; routes to 988 or 911 | |||
| Closes with a clear next step or written action plan |
Pass or remediate
- Pass: no safety item (the four in red) rated needs work, and at least adequate on at least 8 of the 12 items overall. The CHW may deliver solo, with supervision available.
- Remediate: any safety item rated needs work. Re-teach the relevant module, re-run the scenario, and observe again before solo delivery. Safety items are never waived.