Emergency triage
A Socratic walk-through of emergency triage — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why can treating patients in the order they arrive cost more lives than treating them out of order?
First come, first served is the fairest rule we have. It cannot be gamed by status, it needs no judgement, and everybody understands it. Yet the moment a bus overturns, every emergency system in the world abandons it — and claims that abandoning it saves lives.
That is a strong claim. Nobody is treated faster because the order changed; the same surgeons work at the same speed. So where could the extra survivors come from?
Reasoning it through
REASONING #Begin with the case where the claim is false, because it tells us what the argument depends on. Suppose the department is quiet: four patients, six staff, nothing waiting. Everyone is seen almost immediately whatever the order, and reordering buys nothing. So triage is not a property of medicine. It is a property of scarcity — it only does work when demand exceeds capacity, and its value should grow as that gap grows.
Now let the gap open, and ask what varies between the people waiting. Two things, independently.
The first is how fast their situation decays. An obstructed airway or an arterial bleed is measured in minutes. A closed fracture of the forearm is essentially unchanged after four hours, and its owner will do as well treated tonight as now. Delay is not a uniform cost applied to everyone; it varies across the room by orders of magnitude.
The second is how much of the bottleneck each person consumes. If the scarce thing is surgeon-minutes, or a ventilator, or the one CT scanner, then a case needing six hours of theatre is not comparable to one needing ten, however sick each is.
Put those together and the quantity to maximise stops being "who is sickest". It becomes closer to how much treating this person now, rather than later, changes their outcome — per unit of the scarce resource it costs. First come, first served is optimal only where everyone's decay rate is identical; the more those rates differ, the more it throws away.
Two uncomfortable consequences follow from the same arithmetic, and they are one consequence, not two. The walking wounded go last — numerous, loud, and costing almost nothing by waiting. And the very worst injured may also go last, marked expectant: not because they matter less, but because their expected gain is small and their resource cost enormous, so treating them consumes capacity that would have saved several others.
A second mechanism is usually missed. The sort must be performed at speed, by whoever is present, under noise and fear — which is why field systems reduce clinical judgement to a handful of crude tests, can the patient walk, are they breathing, can they follow a command, and why deliberately imperfect algorithms outperform better judgement in practice. The categories are a communication channel as much as an optimiser: a coloured tag tells everyone downstream what to do without a conversation.
Here is the test. If triage works by exploiting differences in decay rate under scarcity, its benefit should scale with the demand-to-capacity ratio and with the spread of those rates, and vanish when either is small. The refuting observation would be triage improving outcomes in an uncrowded department, or failing to improve them in a genuinely overwhelmed one.
The analogy
THE ANALOGY #Think of a fire crew arriving with one hose to find three buildings alight. They do not aim it at whichever alarm was called in first. The shed already fully involved will burn out whatever they do; the brick house barely smoking will still be standing in an hour. The water goes on the timber house just catching, where the same minute of hose changes the outcome most.
buildings make no claim on being saved and nobody grieves an unranked shed, whereas the moral weight of ranking people appears nowhere in the arithmetic that produced the ranking — and a fire chief sees all three buildings at once, while a triage officer decides about each patient without knowing who is still in the wreckage.
Clarifying the model
THE MODEL #The misreading to dislodge is that triage means "treat the sickest first". It does not, and a system that did would be worse than first come, first served in a mass-casualty event, because the sickest are exactly where effort is most likely to be spent for the least return. Triage sorts by marginal benefit per unit of scarce capacity, which puts some of the sickest near the end of the queue and some of the least sick there too.
A refinement that connects the steps: the decay rates are not fixed, so the sort is not either. Categories are reassigned as people deteriorate or stabilise, and re-triage on arrival at hospital routinely moves patients between groups. A single sort would be an optimisation of a snapshot that has already gone stale.
Now the honest limits. You cannot randomise a disaster, so the evidence for triage protocols is largely simulation, retrospective review and expert consensus rather than trial. Studies of triage accuracy consistently find substantial over- and under-triage, but the published rates vary so widely between systems, injury mixes and definitions that quoting a figure would be pretending to a precision the literature does not have — so I will not. The "golden hour" is likewise better treated as a doctrine than a measurement; its direct evidential support is thin and contested. And whether the expectant category saves net lives in practice, as opposed to in models, remains genuinely argued — as does whether the larger gains come from field sorting at all, or from decompressing the hospital behind it.
A picture of it
THE PICTURE #How to readThis is a kanban board repurposed — the columns are not stages a patient passes through but the four groups a triage officer sorts into, and the order left to right is the order of treatment, not of severity. Read each heading as the reason for its position: the first is treated first because a few minutes of cheap intervention changes the outcome, the second because nothing is lost by waiting. The fourth column is what makes triage hard, and its heading is the whole argument — it sits at the end for the same reason the third does, low benefit for the capacity consumed, reached from the opposite extreme. Cards move between columns as patients change, which is why re-triage exists.
What became clearer
WHAT CLEARED #Order matters only when capacity is short, and it matters because the cost of waiting is wildly unequal across the people waiting. First come, first served silently assumes a minute means the same to everyone in the room; triage replaces that assumption with an estimate of what a minute is worth to each, weighted by what treating them costs of the one thing there is not enough of. The result pushes both the least injured and, painfully, some of the most injured toward the back — not a cruelty added to the system, but the same arithmetic reaching two ends of one scale.
Where to go next
ONWARD #- How crisis standards of care are declared and withdrawn, and who decides that scarcity is severe enough to change the rule.
- Two neighbours mark the boundaries: Single-queue service shows how pooling servers lowers waits while holding the order fixed, and Pricing the sacred asks how tragic trade-offs are made bearable; this piece asks only why reordering, by itself, saves anyone.
Key terms
TERMS #| Term | What it means |
|---|---|
| Triage | sorting patients by the benefit treatment now would produce, relative to the scarce capacity it consumes. |
| Expectant category | those whose injuries are judged unsurvivable given available resources, offered comfort care while others are treated. |
| Over-triage | assigning a higher priority than the injuries warrant, consuming capacity needed elsewhere. |
| Under-triage | assigning a lower priority than warranted, the error that directly costs the individual. |
Every term the collection defines is gathered in the glossary.