Daniel's been thinking about triage again, but this time he's pointed at the version where the patient can't tell you what's wrong. He asks how emergency pediatric specialists do what they do, given that a big chunk of their patients are non-verbal and another big chunk are barely verbal, and the stakes are urgent care stakes. He wants to know how they assess, how they sort, and what the rest of us can learn from a discipline that has to work without the patient's own account.
This one lands close to home for me. I spent years in pediatrics, and the triage part was always the part that kept me up at night. Not the complex cases, not the surgeries, not the rare diagnoses. The fifteen seconds at the door when you have to decide if a kid who can't talk is sick or really sick.
So before we get into how they actually do it, we need to understand why this is a different problem, not just a harder version of the same problem. What makes it structurally different?
Right. Adult triage assumes a cooperative narrator. The patient says, chest pain, came on after lunch, feels like pressure, radiates to the jaw. You get a story. Pediatric triage, especially for the under-fives, you get a parent who slept four hours and a child who is either crying, limp, or somewhere in between. The primary data source is gone. And the secondary data source, the vital signs, is lying to you.
Lying how? That's a strong claim.
A child's cardiovascular system compensates. When a kid starts to get sick, really sick, the body clamps down peripheral blood vessels, shunts blood to the core, and pushes the heart rate up. Blood pressure stays normal. It looks fine on paper. A kid can be in early septic shock with a blood pressure that would make an adult nurse yawn. Then the compensation fails, and the kid crashes. Fast. One minute they're sitting up, the next minute they're gray and unresponsive. The adult vital-sign thresholds that trigger escalation are systematically late for children. By the time a child's blood pressure drops, you're not in early trouble anymore. You're in the deep end.
So the instrument panel is giving you false reassurance until it suddenly gives you a catastrophe.
And that changes the entire risk calculus of wait and see. In adults, you can sometimes watch and reassess. In kids, watching is the dangerous default. Because the crash doesn't come with a warning light. It comes after the warning light has been unplugged.
That's a chilling way to put it. The warning light has been unplugged. So the clinician is flying without the instruments they were trained to trust.
And that's the part that's hard to internalize if you've only ever worked with adults. In adult medicine, the vital signs are your safety net. In pediatrics, they're a lagging indicator. The safety net is something else entirely, and we'll get to what that something else is.
Daniel's prompt has three threads, really. The communication challenge of pediatrics, the heightened stakes of urgent care, and the question of what other fields can steal from pediatric emergency specialists. But to answer any of that, we have to sit with the structural problem first. This is not adult triage with smaller patients and adjusted thresholds. It's a different discipline.
And there's a third structural problem Daniel didn't name but the research keeps surfacing. Most emergency departments aren't actually pediatric-ready. So the specialists he's asking about are often working inside systems that were built around adult patients and never fully retrofitted. We'll get to that, because it changes what the expertise means.
Let's start with the assessment problem. What it actually looks like when you have seconds to sort a child who can't tell you what's wrong.
Picture a fourteen-month-old. Fever of a hundred and two. Irritable, not feeding well, pulling at one ear. The differential spans teething at one end and early sepsis at the other. Otitis media in the middle. A urinary tract infection. Maybe meningitis. You cannot ask, does it hurt when I press here. You cannot ask, is the pain sharp or dull. You cannot ask, on a scale of one to ten. You get a parent who says something is off, but the parent can't articulate what, because the parent is exhausted and scared and not trained to read their own child's physiology.
So what does the triage nurse actually do? Walk me through the first thirty seconds.
She reads the child from the doorway. Before she touches the kid, before she puts a cuff on, before she gets a temperature. There's a framework called the Pediatric Assessment Triangle. Three axes. Appearance, work of breathing, circulation to skin. And it's designed to be used in seconds, from across the room, with no equipment.
Walk me through the axes. Start with appearance, because that seems like the one that's hardest to define.
Appearance is the big one. It's tone, interactivity, consolability, gaze, cry. Is the child alert and looking around, or is the child limp and staring through you? Is the child crying vigorously, or is the cry weak and high-pitched? Does the child make eye contact with the parent, or is the child too exhausted to engage? A child who is too quiet, too still, too flat, that's a red flag even if every number on the monitor looks fine.
So the child's behavior is the vital sign.
The child's behavior is the vital sign that matters most. The protocol is built around the recognition that a kid's affect is the earliest and most reliable indicator of serious illness. Before the heart rate climbs, before the blood pressure drops, before the fever spikes, the child looks wrong. The child who is just tired, who doesn't reach for the parent, who doesn't track a stranger walking in, that child gets escalated even with normal vitals. Because the normal vitals are a lagging indicator.
And the other two axes?
Work of breathing. You're looking at the child's positioning. A kid in respiratory distress will sit forward, tripod, use the neck muscles. You're looking for retractions, the skin pulling in between the ribs or above the collarbone. You're listening for audible sounds, stridor, grunting, wheezing. A grunting infant is a specific alarm bell. Grunting is the body's way of generating positive end-expiratory pressure, trying to keep the alveoli open. It means the child is working hard to breathe, and that's a pre-arrest sign in a baby.
I want to pause on that, because it's a good example of the difference between knowing the mechanism and just memorizing the sign. A grunt sounds like a small thing. But if you understand that the child is literally trying to create pressure in their own airways to keep them from collapsing, the grunt stops being a sound and becomes a signal.
That's the level of understanding the discipline demands. It's not enough to know that grunting is bad. You have to know why it's bad, because that's what tells you how bad it is and how fast you need to move. A grunting infant is not a wait-and-see situation. That child is working at the edge of their physiological reserve.
And circulation to skin.
Pallor, mottling, cyanosis. A child who is compensating for shock will shunt blood away from the skin. The skin goes pale, then mottled, then blue at the extremities. You see a kid with cold, mottled legs and a fast heart rate, you don't need a blood pressure. You know the child is in trouble. The skin is the window into the circulation, and it's free, and it's instant.
So the Pediatric Assessment Triangle is a structured observational framework that works before you have any data. That's the thing that jumps out at me. It's not a checklist of questions. It's a discipline of looking.
And that's the part that generalizes. But let me stay clinical for a minute, because there's a protocol layer on top of the PAT. The Emergency Severity Index, the ESI, is the dominant triage sorting system in American emergency departments. It has pediatric adaptations. ESI assigns a level from one to five, one being immediate, five being non-urgent. For adults, it leans heavily on vital signs and the patient's complaint. For kids, the pediatric versions weight the PAT heavily and build in different vital-sign thresholds for different ages.
And the accuracy problem.
ESI pediatric triage accuracy is an active area of study. Under-triage and over-triage both happen. Under-triage is the nightmare. A child who is compensating gets sorted as a four or a five, sits in the waiting room, and crashes. Over-triage is the quieter problem. A child with a viral cough gets sorted as a two, takes a bed, and the waiting room backs up. Crowding is itself a safety hazard. So the tradeoff is asymmetric. Under-triage kills. Over-triage degrades the whole system.
You'd rather have a crowded waiting room of well children than one dead child sent home.
That's the rational error. But the crowding isn't free. It slows everything down, it burns staff, it makes the next under-triage more likely. The discipline is living inside that tension. And the tension is not theoretical. It plays out every single shift in every busy emergency department.
Let's talk about the parent-reported history. You mentioned the parent as a data source. How does that work in practice? Because the parent is also a mess, right? They're scared, they're sleep-deprived, they're not objective.
The parent is a proxy historian. And the discipline is treating the parent's account as structured data, not anecdote. You ask specific questions. When did the fever start. Has the child been drinking. How many wet diapers today. Is the child acting like himself. The parent may not be able to say what's wrong, but the parent can tell you what's changed. The parent has the baseline. The clinician doesn't.
So the parent is the calibration standard.
The parent is the calibration standard. And the skill is borrowing the caregiver's baseline without being captured by it. Because the parent is also terrified, and the parent's anxiety doesn't map cleanly onto the child's acuity. A panicked parent can bring in a kid with a cold. A calm parent can bring in a kid in early shock. You have to read the parent's report for the signal, and the child's body for the truth.
That's a hard skill. Reading two data streams at once, one of which is emotional and one of which is physiological, and trying to extract signal from both.
And the two data streams can contradict each other. The parent says, he's just not himself, and the child is sitting there looking around, interactive, normal color. Or the parent says, I think he's fine, but the child is gray and limp. You have to hold both and decide which one to trust. Most of the time, the child's body wins.
There's a concrete example that keeps coming up in the literature. The child who looks just tired. The PAT appearance axis flags lethargy or poor interactivity, which triggers escalation even though the vital signs are technically within normal limits. The protocol is built around the recognition that the child's behavior is the vital sign that matters most. The tired child is the dangerous child.
And that's the hardest thing to teach. A new nurse sees a kid who is quiet and thinks, good, the kid is calm. An experienced pediatric nurse sees a kid who is quiet and thinks, why isn't this child looking around. The quiet child is the one who scares me. A screaming toddler is almost always fine. A limp, silent toddler is the one you carry to the back.
I want to dig into that teaching problem, because it's not obvious. Why is it hard to teach someone to be scared of a quiet child?
Because the quiet child doesn't look scary. The quiet child looks peaceful. A screaming toddler is visibly distressed, and the instinct is to respond to the distress. The quiet child doesn't trigger the same alarm. It takes experience to learn that the absence of distress is sometimes the distress. That's a counterintuitive lesson, and it goes against the grain of normal human intuition.
So the training has to override the instinct.
The training has to override the instinct, and the instinct is deep. It's not just clinical inexperience. It's a human response. We respond to visible suffering. We don't respond to visible absence. The child who is too quiet is showing an absence of the normal response, and that's harder to see than a positive signal.
So that's the clinical mechanism. But there's a bigger question underneath it. What happens when the system around that clinician isn't built for pediatric patients?
This is where it gets uncomfortable. The two thousand twenty-one National Pediatric Readiness Project assessment found that while most emergency departments have some pediatric equipment, far fewer have the full set of guidelines-compliant supplies. And pediatric emergency care coordinators are not universal. The coordinator role is the person whose job is to make sure the department is actually ready for kids, the equipment, the training, the protocols, the drills. And a lot of departments don't have one.
The specialists Daniel is asking about are often working in systems that were designed around adults.
Right. The typical emergency department in this country is built for chest pain and car accidents. The equipment is adult-sized. The protocols are adult-weighted. The staff training is adult-focused. Pediatrics is a minority of the volume, maybe ten to twenty percent of visits, but it's the population where the assessment is hardest and the margin for error is smallest. And the system treats it as an add-on.
The knock-on effect is the one that bothers me. When a hospital isn't pediatric-ready, the triage nurse's judgment carries more weight, because the system isn't catching what she misses. That's the opposite of how high-reliability systems are supposed to work. You want the system to be robust to individual error, not dependent on individual excellence.
Pediatric triage expertise is a single point of failure in many institutions. If the one nurse who really knows kids is off shift, or leaves, or retires, the safety margin collapses. And nobody notices until a child crashes in the waiting room.
That's the nightmare scenario. The system looks fine until the moment it doesn't. And the moment it doesn't is a child in a waiting room chair.
The waiting room is the worst place for a pediatric crash. It's public. It's crowded. The family is there. Other families are there. The staff is not prepared. It's the failure mode that everyone in the field is trying to prevent, and it happens anyway.
What can other disciplines learn from this? Daniel asked that directly, and I want to give it real weight. Because the temptation is to say, well, my field doesn't deal with life and death, so this doesn't apply. But I think that's a cop-out.
The transferable insight is not, use the Pediatric Assessment Triangle in your standup meeting. The transferable insight is, when your primary data source is unreliable, build your process around observation and structured proxies rather than around self-report. That's the deep lesson.
Break that apart.
Four pieces. First, structured observational frameworks that work before you have data. The PAT is a discipline of looking, and it's teachable. You can train someone to read appearance, work of breathing, circulation to skin in an afternoon. The framework doesn't replace expertise, but it gives the non-expert a place to start. Second, the explicit acknowledgment that behavior is data even when words aren't. A user who goes quiet in a support ticket, a customer who stops logging in, an employee who stops asking questions, those are signals. The absence of self-report is itself information.
Third?
The asymmetric cost of errors. Under-triage and over-triage are not equally bad. The protocol should reflect that. In most domains, the default is to treat false positives and false negatives as symmetric, because that's what the math assumes. But in pediatric triage, the cost of missing a sick kid is death, and the cost of over-sorting a well kid is crowding. The protocol is built around the asymmetry. Any discipline that makes high-stakes decisions about people should ask itself, which error is worse, and does the process actually reflect that?
Fourth, the proxy historian.
The parent as structured data source. When the person can't tell you what's wrong, you find the person who knows them best, and you ask that person structured questions. Not, what do you think is wrong, but, when did this start, what has changed, what does normal look like. The proxy's account is data. It's not anecdote. It's the baseline you don't have.
The limits of the analogy matter too. Most domains don't have the stakes or the time pressure of a pediatric emergency department. Nobody dies if your sprint retro misreads the team's mood. The lesson is less, install a triage protocol in your startup, and more, when self-report is unreliable, observation and structured proxies beat asking harder.
The open question underneath all of this. If pediatric triage is this hard, and this dependent on individual expertise, and most emergency departments aren't fully ready, what does that say about how we staff and train for it? The research doesn't answer that. It's a live tension. The field knows the readiness gap exists. It knows the accuracy problem exists. It hasn't figured out whether the answer is more systematic readiness or just hoping the exceptional nurse is on shift.
The exceptional nurse. That's the thing I keep circling.
I worked in a pediatric emergency department, night shift, registration clerk. Not clinical. I checked families in. I watched the waiting room.
The waiting room.
The triage nurse was a woman named Dolores. She could look at a kid from across the room and move them up the queue before the parents had finished signing the clipboard. The protocol didn't tell her to do it. She just knew. And she was right more often than the protocol. The hardest part of my job wasn't the triage decisions. It was managing the parents. They were often more panicked than the child. Sometimes I had to calm the parent down before the clinical staff could even assess the kid. Dolores was triaging the parent as much as the patient.
She kept a notebook.
A personal notebook. Not an official log. She wrote down patterns she noticed. Kids who won't make eye contact with the parent are sicker than kids who won't make eye contact with me. I don't know if that's true. She believed it. And she was right more often than the protocol.
The protocol was written down afterward by people trying to teach what Dolores already knew.
That's what it looked like to me. I sat at that desk for a year and a half. I saw a lot of kids come through. The ones Dolores pulled to the front were the ones who needed it. I never saw her wrong in a way that mattered.
The tacit knowledge that never makes it into the formal protocol. That's the thing I can't stop thinking about. The PAT is an attempt to codify what Dolores did by instinct. But the codification is incomplete. The notebook had the patterns that didn't fit the framework. Eye contact with the parent versus eye contact with the stranger. That's a specific, granular observation. It's not in any protocol. And it probably saved lives.
The field knows this. The protocols are necessary but insufficient. The tacit knowledge does the real work in hard cases. And the tacit knowledge walks out the door when Dolores retires.
There's no easy way to capture it. You can't write a protocol for every pattern. The patterns are too many, too contextual, too dependent on the specific nurse's experience. The notebook is the closest thing to a capture mechanism, and it's personal, not institutional.
The institutions that do this well have some way of transferring that tacit knowledge. Mentorship, shadowing, paired shifts. But that's slow, and it doesn't scale. The alternative is to accept that the tacit knowledge is a scarce resource and build systems that don't depend on it as much. But that's a different kind of investment, and most places haven't made it.
The single most common wrong belief about pediatric triage is that it's adult triage with smaller patients and adjusted vital-sign thresholds.
It's not. The primary data source is gone, the physiology of compensation means the standard thresholds are late, and the entire discipline is built around reading behavior and borrowing the parent's baseline. Change the vital-sign chart and you haven't changed the problem.
The open question that stays with me is what happens when the expertise isn't in the room. If pediatric triage depends this heavily on individual judgment, and most emergency departments aren't fully pediatric-ready, then the safety margin is thinner than anyone wants to admit. The research on ESI pediatric accuracy is ongoing, and the readiness gap is a known problem. The question is whether the field moves toward more systematic readiness or continues to rely on exceptional individuals who may or may not be on shift when your kid needs them.
That question doesn't have a clean answer. The exceptional individuals are real, and they save lives. But they're not a system. They're a patch. And a patch fails when it's not there.
Which brings us back to Daniel's original question. What can the rest of us learn from a discipline that has to work without the patient's own account? I think the answer is, more than we're comfortable with. The lesson about structured observation, about behavior as data, about asymmetric error costs, about proxy historians, those apply anywhere the primary data source is unreliable. Which is a lot more places than we like to admit.
The quiet child in the waiting room is not just a pediatric problem. It's a general problem. The signal is there. You just have to know how to look.
Thank you to our producer, Hilbert Flumingtop, for keeping the show running.
This has been My Weird Prompts. Find us at my weird prompts dot com, and if you have a question you want us to dig into, email us at show at my weird prompts dot com.
We'll be back soon.
