Vet After Hours Call Triage: The 4-Question Decision Script
At 10:45 PM, a cat owner calls. Their cat jumped off the counter and is holding up one front leg. It happened 20 minutes ago. The cat is eating, is alert, and keeps trying to walk. The owner is panicking and wants to know if they need to drive 40 minutes to the emergency vet tonight.
The right answer — ER now vs. first morning appointment vs. watch-at-home with a call if anything changes — depends on four pieces of information. Getting those four pieces quickly, in the right order, is what vet after hours call triage looks like when it works.
This post is a practical 4-question decision script that any practice can use: printed next to the phone for an on-call tech, encoded into an AI answering service, or handed to a staff member taking calls after hours. The goal is consistent, defensible routing — not a veterinary diagnosis — so every after-hours caller gets a clear next step.
This isn’t veterinary medical advice. The script routes calls to appropriate care. When in doubt, always direct the owner to an emergency vet.
Why Vet After Hours Call Triage Needs a Written Script
Without a script, after-hours triage is improvised. The person taking the call decides based on their experience, their energy level, and how the caller describes the concern. The routing varies. The advice varies. And the outcomes vary in ways that expose your practice to inconsistent client experiences and, at worst, real liability.
A written script fixes that. It doesn’t replace clinical judgment; it captures the right information in the right order so judgment can be applied consistently. A 911 dispatcher reads from a card on a cardiac call for the same reason. Not a competence problem. A speed problem.
The 4-question vet after hours call triage script is designed for non-veterinarians: a trained technician, a practice manager, or an AI answering service. It identifies the three outcomes a triage call can reach and asks only what’s needed to get there.
The 3 Triage Outcomes Every After-Hours Call Reaches
Before the questions, the three buckets:
Emergency ER referral. The concern can’t wait until morning. The owner needs to be in a vehicle heading to an emergency facility now, with the name, address, and phone number in hand.
First-available appointment. The concern should be seen by a veterinarian, but the window is hours, not minutes. Book the earliest available slot, capture the concern in detail for the chart, and tell the owner what to watch for in the meantime.
Home-care watch with call-back criteria. Minor concerns where a specific set of observable changes would move the situation into one of the first two buckets. The owner gets a clear list of “if you see X, go to ER; if you see Y, call us first thing.” No medical advice — just observation criteria.
Every call falls into one of these three. The script gets you there in under three minutes.
Two of those buckets are more forgiving than they feel at 11 PM. The AVMA’s July 2025 look at 2024 visit data found nearly one in two emergency patients were seen within 30 minutes, and only 8.4% of owners waited more than two hours. On the non-urgent side, almost one in five owners got a same-day appointment. Useful context for whoever is running the script: sending a stable animal to first-available isn’t parking it for a week, and sending a critical one to the ER isn’t feeding it into a queue.
The 4-Question Vet After Hours Call Triage Script
Ask these four questions, in this order, without detours. Total time: 60–180 seconds depending on the caller.
Question 1: What is the animal, and what are you seeing right now?
“Tell me the species, and describe exactly what you’re seeing — I want the most concerning thing first.”
This is the triage opener. You want the most alarming piece of information upfront. “My dog is having trouble breathing” routes differently than “my dog has been scratching at his ear for two days.” By asking for the most concerning thing first, you surface the ER-level signals before the caller buries them in narrative.
Species matters because the same symptom reads differently across dogs, cats, and exotics. A cat breathing with its mouth open is a veterinary emergency in almost every circumstance. A dog panting on a hot night is not.
Question 2: Is the animal standing, walking, and breathing normally?
“Is the animal up and moving — and is their breathing normal, even if labored or faster than usual?”
This is the stability question. Normal ambulation and normal (or near-normal) breathing are the two strongest indicators that an animal is not in immediate systemic crisis. If either is abnormal, that’s an automatic ER signal regardless of what the presenting complaint is.
The question is phrased as two yes/no checks. The caller answers quickly. A “no” on either part flags the call for escalation.
Question 3: What happened, and how long ago?
“Walk me through what happened — and how long ago did it start?”
Mechanism and timing are the two levers that most change the ER/morning/home split:
- Mechanism: Trauma (hit by car, fall from height, animal bite) skews toward ER regardless of apparent stability, because internal injuries aren’t visible. Ingestion (plant, food, medication, household chemical) triggers the ASPCA Animal Poison Control Center referral immediately for anything with known toxicity. Gradual onset symptoms (has been acting off for two days) generally have more time flexibility than acute-onset ones.
- Timing: A dog who ate a grape 20 minutes ago is different from a dog who ate grapes three hours ago and has been vomiting since. Time elapsed since onset is often the difference between decontamination being useful and it not.
Ingestion calls are worth over-indexing on, because the odds are better than most people assume that the thing on the floor is genuinely toxic. ASPCA Poison Control fielded more than 376,000 exposures in 2025, and the breakdown of what pets actually got into is a useful mental index for whoever is on the phone at midnight:
| Category | Share of exposures (ASPCA Top 10 Toxins of 2025) |
|---|---|
| Over-the-counter medications and supplements | 16.9% |
| Human food and drink | 16.4% |
| Chocolate | 13.6% |
| Veterinary products (chewable meds, mostly) | 9.1% |
| Plants and fungi | 7.5% |
| Rodenticides | 6.5% |
| Household products | 6.4% |
Over-the-counter human medication has held the top spot for eight consecutive years. So when an owner says “he got into my purse,” treat that as a toxin call until proven otherwise — not a wait-and-see. And note the veterinary-products line: flavored chewables that pets self-dose from a chewed-open bottle are a real category, not an edge case.
Note: This script captures information to route the call — it doesn’t provide veterinary medical advice. For any toxin ingestion concern, direct the owner to the ASPCA Animal Poison Control Center or an emergency facility.
Question 4: What’s your best contact number if we get cut off?
“Before I give you next steps — what’s the best number to reach you on, and what’s your full name?”
Get the contact before the next-step conversation. If the call drops, if the owner has to run for their keys, if anything disrupts the next 60 seconds — the lead is retained. This is a lesson from any call intake discipline: callback number before dispatch.
Name plus number. That’s all you need at this point.
How the 4 Questions Map to a Routing Decision
| Signal from Q1–Q3 | Routing outcome |
|---|---|
| Respiratory distress, inability to stand, suspected toxin ingestion | Emergency ER referral — now |
| Trauma with unknown internal status | Emergency ER referral — now |
| Alert and mobile, non-critical presenting complaint, onset within 12 hours | First-available appointment |
| Alert and mobile, gradual-onset concern, owner has been watching it for days | First-available appointment or home-care watch |
| Minor behavioral note, no physical symptoms, owner just wants it on the chart | Home-care watch, call-back criteria provided |
The routing table doesn’t cover every case. It covers most cases, and it’s transparent enough that whoever is running the script — human or AI — can explain why they made the call they made.
Map the Phone Buckets to the Triage Categories Your ER Already Uses
Your three phone outcomes aren’t arbitrary. They’re a compressed version of the colour-coded systems emergency departments run at the door, most of which descend from the Manchester triage system used in human medicine: red (immediate), orange (very urgent), yellow (urgent), green (standard), blue (non-urgent).
That lineage matters for two reasons. First, it gives the ER you’re referring to a shared vocabulary — “this sounds orange to me” travels better than “she seemed pretty worried.” Second, the veterinary versions have now been measured, so you know roughly how well a structured script performs versus gut feel.
A 2025 study in the Journal of Veterinary Emergency and Critical Care tested a five-point veterinary triage system (VetTriS) on 164 dogs and cats presenting to an emergency department. Three numbers from it are worth pinning to your card:
| Measure | Result (VetTriS, J Vet Emerg Crit Care 2025) |
|---|---|
| Target waiting time met | 85.4% of patients |
| Overtriage (rated more urgent than warranted) | 9.8% |
| Undertriage (rated less urgent than warranted) | 0.6% |
Interobserver agreement came in at a kappa of 0.69 — substantial, meaning two different people running the same structured system landed on the same category most of the time. That’s the whole argument for a written script in one statistic.
Look at the gap in the VetTriS results between 9.8% overtriage and 0.6% undertriage. A well-built triage system doesn’t aim for accuracy in both directions equally. It’s deliberately tilted so that when it’s wrong, it’s wrong toward more urgent. Your phone script should carry the same bias. An owner who drives to the ER and gets told “he’s fine, go home” has lost an evening. An owner told to wait until morning with a blocked cat has lost the cat.
Here’s how the three phone outcomes line up:
| Phone script outcome | ER triage equivalent |
|---|---|
| Emergency ER referral | Red / orange — seen immediately or within ~15 minutes |
| First-available appointment | Yellow — needs a veterinarian, window measured in hours |
| Home-care watch with call-back criteria | Green / blue — standard or non-urgent, with defined escalation triggers |
Phone triage is harder than door triage in one specific way: you can’t see the animal, take a pulse, or check mucous membranes. You’re working from an owner’s description of something they’re frightened by. So a phone script has to lean on the observations an untrained person can actually make reliably — is it standing, is it breathing, what happened, how long ago — which is exactly what the four questions ask.
Veterinary Triage Red Flags That Skip the Script Entirely
Some answers end the script. If any of these show up — at Question 1, at Question 2, anywhere — stop asking questions, give the ER referral, and get the callback number on the way out the door. Don’t finish the sequence for tidiness.
- Open-mouth breathing in a cat, or any breathing that’s noisy, labored, or has the animal sitting with elbows out. Respiratory distress has the shortest clock of anything on this list.
- Unproductive retching in a large, deep-chested dog — the classic bloat presentation. A Great Dane heaving with nothing coming up is a surgical emergency, not a wait-till-morning.
- A male cat straining in the litter box. Owners describe this as constipation. Urethral obstruction is what it usually is, and it becomes fatal within a day or so.
- Collapse, seizure, or an inability to stand. Doesn’t matter how quickly the animal recovered afterward.
- Bleeding that hasn’t stopped after five minutes of direct pressure, or a wound where a body cavity is visibly open.
- A dystocia call — an animal actively in labor with strong contractions and nothing delivered for 30+ minutes.
That list belongs on the same card as the four questions, because it’s what a trained tech carries in their head and a new hire doesn’t. Encoded into an AI answering service, it becomes a hard escalation trigger: the phrase fires, the script exits, the referral goes out.
One more thing to flag on your card, not in the caller’s ear: an animal is not the only thing an owner is weighing. A March 2026 ASPCA report found six in ten American pet owners aren’t confident they could afford a pet medical emergency — the organization estimates that puts over 100 million pets at risk. Some of the pushback you get on an ER referral isn’t disbelief. It’s cost. Give the referral anyway, clearly and without hedging, and let the owner make the decision with correct information.
Build the ER Referral List Before You Need It
A triage script that ends in “you should go to the ER” and nothing else is half a script. Whoever runs it needs a specific name, address, and phone number to hand over, and 11 PM is the wrong time to be searching for one.
Build the card once, keep it next to the phone (and in your AI receptionist’s knowledge base):
- Two facilities, not one. Nearest and next-nearest. The nearest one is sometimes on diversion, closed for staffing, or full.
- Drive time, not distance. “Twenty-five minutes on the 40” is a usable sentence. “Eighteen miles” isn’t.
- Actual current hours. Overnight coverage at emergency practices changes with staffing. Re-verify quarterly.
- Capability level. The Veterinary Emergency and Critical Care Society certifies facilities at three levels based on hours, equipment, and staffing — Level I is the highest and requires a full-time board-certified criticalist. Certification runs two years, so a facility’s status can lapse. Knowing which of your local options can do emergency surgery overnight versus stabilize-and-transfer changes where you send a bloat call.
- Exotics. If your practice sees birds, reptiles, or small mammals, the nearest ER that will actually take them is often a different facility from the dog-and-cat one.
Then send the information ahead. The 2025 AAHA referral guidelines put the weight on making sure the receiving hospital has what it needs before the patient arrives — records transferred, communication consistent, expectations managed. After hours that’s a smaller version of the same job: a text or portal note to the ER with species, presenting complaint, time of onset, and the owner’s name saves the receiving team from starting the history from zero at the front desk.
Teletriage vs. Telemedicine: The Line Your Phone Script Must Not Cross
This is the question practice owners actually ask before they’ll let anyone — a tech, an outsourced service, an AI — answer the after-hours line: what are we allowed to say?
The AVMA draws the boundary in its definitions, and the boundary is the whole reason this script is written the way it is. Teletriage is “the safe, appropriate, and timely assessment and management (immediate referral to a veterinarian or not) of animal patients via electronic consultation with their owners.” The AVMA is explicit about what teletriage isn’t: “A diagnosis is not rendered.” Telemedicine, by contrast, “may only be conducted within an existing veterinarian-client-patient relationship.”
So: sorting urgency is teletriage. Naming the problem is telemedicine. The four questions stay firmly on the first side of that line — they gather observations and produce a routing decision, not an assessment of what’s wrong with the animal.
Don’t assume that only applies to somebody else’s phone line. The AVMA House of Delegates adopted a revised Model Veterinary Practice Act in July 2025, and its teletriage definition was drawn wide deliberately — AVMA staff clarified it covers not just pet poison hotlines but “other providers who may be assisting animal owners in determining whether or not to see a veterinarian and with what urgency.” That’s a description of your after-hours line. Whoever or whatever runs it is doing teletriage, and teletriage rules are the ones that apply.
Two more things worth knowing before you write your own script:
- A VCPR can’t be created over the phone. The AVMA Model Veterinary Practice Act states that a veterinarian-client-patient relationship “cannot be established solely by telephonic or other electronic means,” and the AVMA’s VCPR guidance notes the FDA has taken the same position. A new caller who’s never been to your practice doesn’t become a client because you triaged their dog at midnight.
- There’s a narrow emergency carve-out. The same AVMA guidance says an exception “may be made for advice given in an emergency situation until a patient can be seen by a veterinarian.” Read that as: get them moving toward care, not as licence to manage the case by phone.
State practice acts define the VCPR differently, and the AVMA advises practitioners to know how their own state words it. If your on-call rota crosses a state line, that’s a conversation to have with your state board rather than a detail to figure out at 2 AM.
This isn’t legal advice. Veterinary practice acts vary by state and change. Have your own counsel or state board review any after-hours triage protocol before you put it into service.
The practical upshot for your script: every line in it should be a question, a routing decision, or an observation criterion. Nothing in it should be a conclusion about what’s wrong with the animal. That constraint is also what makes the script safe to hand to a non-veterinarian — or to encode into software.
Write the Call Down: What the Record Has to Show by Morning
A triage call that routes correctly and never reaches the chart is half-finished. The owner calls back at 9 AM, someone else picks up, and the practice starts from zero. Or the ER calls for history and nobody there knows a call happened at all.
The AAVSB — the association state veterinary boards belong to — published model medical-recordkeeping regulations in August 2025, recommended by its Regulatory Policy Committee that April. Model regulations aren’t law anywhere until a board adopts them. But they’re the clearest current statement of what regulators think a record should contain, and four required elements land directly on an overnight phone call:
| AAVSB model requirement | What it means for a 2 AM triage call |
|---|---|
| §2(c)(20) — “Summary of medically relevant communications with the Client” | The call is a record event on its own. Log it. |
| §2(c)(21) — “Any services or treatment recommendations declined by the Client" | "Told them ER now, they said they’d wait for morning” goes in writing. |
| §2(c)(4) — date and time of entry | Timestamp the call, not just the morning follow-up. |
| §2(c)(6) — identity of whoever created the record, including “if the creation or update of a record involves the use of artificial intelligence” | If software took the call and wrote the note, the record should say so. |
That last row is the one most practices haven’t thought about yet. The AAVSB’s own commentary explains why: if an individual or software other than the licensee helps write the record, “that should be clearly noted as this information may be helpful in a disciplinary case.” Whatever answers your phone overnight, its notes should be attributable to it.
Give the declined-referral line its own weight. An owner who turns down an ER referral is the highest-risk call your practice takes that night, and the note you write is the only record that conversation ever leaves. Write what you recommended, in the words you used, and what they said back.
So the minimum after-hours triage note is short:
- Time of the call
- Owner name and callback number — Question 4, doing double duty
- Species and presenting complaint — Question 1
- What happened and how long ago — Question 3
- Standing, walking, breathing — Question 2
- The routing decision you gave
- What the owner said they’d do
- Whether you sent a heads-up to the receiving ER
Eight lines, and six of them are answers the script already collected. That’s not a coincidence — it’s why the questions are ordered the way they are. The intake and the record are the same document.
Running the Script at Scale: AI vs. Human Coverage
A printed script works. An AI answering service running the same script works every time, identically, at 2 AM and at 10 PM, on call #3 and call #47 of the week.
The veterinary emergency answering service post walks through the 2 AM experience in detail — what good sounds like versus what voicemail sounds like. The practical point here: an AI that runs this 4-question script doesn’t drift on tired nights, doesn’t skip Question 2 because the caller is crying, and doesn’t forget to get the callback number before giving next steps.
The script is what makes AI coverage trustworthy for triage. A general-purpose answering service that just takes a message is not running triage — it’s delaying it. The AI receptionist vs. answering service comparison covers the structural difference between those two approaches, and the AI receptionist for veterinary clinics overview covers what the rest of the call load looks like once triage is handled.
For the practice owner weighing coverage options, the can AI handle emergency calls analysis is worth reading — it goes deep on how AI escalation works when the stakes are high.
What This Script Does Not Cover
Three things outside the scope of triage that belong in your standard operating procedure:
- Wildlife and stray animals. Different jurisdiction, different liability. Have a wildlife rehab referral number ready and a clear policy on what your practice will and won’t advise.
- Behavioral emergencies. An animal that is a bite or aggression risk to the owner requires a different script — one focused on safety before anything else. Don’t triage a bite risk the same way you’d triage a vomiting cat.
- Owner distress without a clear animal emergency. Sometimes the call is about a person’s anxiety more than an animal in crisis. Acknowledge the concern, run the four questions, and route from the answers — not from the emotional level of the call.
Frequently Asked
Q: Should a non-veterinarian be running this script? A: Yes — that’s the point. The script is designed to be run by trained technicians, practice managers, or AI systems. It routes calls to appropriate care; it doesn’t make veterinary judgments. The veterinary judgment happens at the emergency facility or at the clinic appointment. The script gets the owner to the right door.
Q: What if the owner refuses to answer the questions and just wants a “yes or no, should I go to the ER”? A: Acknowledge the urgency, then anchor to Question 2: “I hear you — let me ask you one quick thing so I can give you the right answer. Is the animal standing and breathing normally right now?” That one question tells you almost everything you need for an emergency-vs-not split. If the owner still won’t engage, default to an ER referral — better to send a non-emergency to the ER than to miss a real one.
Q: How do we handle toxin ingestion calls specifically? A: Question 3 (mechanism) will surface it. Any suspected ingestion of a known toxin — medications, certain plants, household chemicals, xylitol, grapes, chocolate in quantity — routes to the ASPCA Animal Poison Control Center number and/or an emergency facility immediately. Don’t hold these calls for home-care advice. This isn’t veterinary medical advice — direct the owner to ASPCA Poison Control or emergency care for any ingestion concern.
Q: Can this script be used by an AI answering service after hours? A: Yes. The script is structured as four linear questions with branching logic at Q1 (species/symptom), Q2 (stability), and Q3 (mechanism/timing). That structure maps directly to how an AI voice system handles a call — fixed questions, conditional routing, escalation triggers. The resulting intake is the same every time.
Q: What do we say when the owner tells us they can’t afford the emergency vet? A: Give the referral anyway, then stop talking about money. You’re not qualified to price the visit, and guessing at a number the ER will charge helps nobody. What you can do is offer the facts: the name and number of both nearby facilities, and the suggestion that they call ahead and ask about payment options directly. Cost is a real constraint for most owners — a March 2026 ASPCA survey found six in ten owners aren’t confident they could cover a pet medical emergency — but softening a referral to make it easier to hear is how a genuine emergency becomes a morning appointment that arrives too late.
Q: How long should an after-hours triage call actually take? A: 60 to 180 seconds for the four questions, plus a minute for next steps. Anything past five minutes usually means one of two things: the caller is distressed and needs the questions re-anchored (go back to Question 2), or the call is stability-normal and non-urgent and should have been routed to a first-available appointment two minutes ago. Log the outliers — a pattern of long calls on one symptom type usually means the script needs a branch it doesn’t have.
Q: Can whoever answers the phone give the owner advice, or only route the call? A: Route, plus observation criteria. The AVMA’s definition of teletriage covers assessment and referral and states plainly that “a diagnosis is not rendered” — that’s the line. Telling an owner “watch for open-mouth breathing and go straight to the ER if you see it” is a routing instruction. Telling them “sounds like an allergic reaction, give him a Benadryl” isn’t, and it isn’t something a non-veterinarian should be saying regardless. Practice acts vary by state; this isn’t legal advice, and your protocol should be reviewed by counsel or your state board.
Q: How do we know if our triage script is sending too many people to the ER? A: You probably want it erring that direction. The 2025 VetTriS study of a five-point veterinary triage system found overtriage in 9.8% of patients and undertriage in just 0.6% — the asymmetry is the design, not a flaw. Audit the calls anyway: pull a month of after-hours ER referrals and check outcomes with the receiving facility. If a specific symptom keeps producing referrals that turn out to be non-urgent, that symptom needs its own branch in the script, not a looser rule across the board.
Q: What do we actually need to write down after an after-hours triage call? A: Time of the call, owner name and callback number, species and presenting complaint, what happened and how long ago, the two stability answers, the routing decision you gave, and what the owner said they’d do. The AAVSB’s 2025 model recordkeeping regulations treat a “summary of medically relevant communications with the Client” and “any services or treatment recommendations declined” as record elements — so a declined ER referral is the one line you can’t skip. Your state board sets the binding rule; check how yours words it.
Q: If an AI answers the call, does the record need to say so? A: The AAVSB’s 2025 model regulations say the record should identify whoever created or updated it, and that “if the creation or update of a record involves the use of artificial intelligence, such information shall be included in the record.” The accompanying commentary is direct about the reason — it matters in a disciplinary case. Model regulations don’t bind you until your state adopts them, so verify with your board. Building the disclosure in from day one costs nothing.
Q: What’s the difference between a triage script and a phone script for booking? A: A booking script assumes the call is a routine appointment request and focuses on gathering scheduling information. A triage script assumes the call might be an emergency and focuses on ruling that out first. Both have their place — after hours, the triage script runs first, and if the call is non-urgent, it flows into booking. The veterinary answering service post covers how routine after-hours bookings work once triage confirms the situation isn’t urgent.
Get the Vet Triage Checklist
The 4-question script and routing table above condense into a one-page printable your on-call team can use today — or the exact input format an AI answering service needs to run triage correctly.