The Emergency Severity Index gets misread constantly, and almost always in the same way: people assume all five levels measure how sick someone is. They don’t. The top two levels ask about danger. The bottom three ask a completely different question — how much of the department is this going to take?

Once that split is clear, triage stops feeling like a judgement call about vibes and starts being a short decision tree you can defend.

The two questions, in order

Levels 1 and 2 are acuity decisions. You’re asking whether this patient is dying now, or could be shortly.

Levels 3, 4 and 5 are resource decisions. Once you’ve established the patient isn’t in that danger category, the level is set by how many resources they’ll need — not by how uncomfortable they are.

That’s the whole architecture. The rest is detail.

Level 1 — needs a life-saving intervention now

The question is simple: does this patient require an immediate life-saving intervention?

Intubation, emergent airway management, defibrillation, chest compressions, immediate significant medication for haemodynamic instability. If the answer is yes, they’re a 1 and the room is already moving.

An unresponsive patient, an obstructed airway, a peri-arrest rhythm, an apnoeic patient — these are 1s.

Level 2 — high risk, and it can’t wait

This is the level that separates good triage from average triage, and the one most often under-called. Level 2 is for the patient who isn’t dying in front of you but plausibly could be soon. Three doors lead here:

A high-risk situation — the clinical picture carries a serious threat. Chest pain that sounds cardiac, a possible stroke inside the treatment window, suspected sepsis, ectopic pregnancy, testicular torsion, an immunocompromised patient with a fever.
New confusion, lethargy, or disorientation — a change in mental status is a level 2 finding on its own.
Severe pain or distress, where the presentation genuinely warrants immediate attention.

Level 2 is the one that gets missed

The classic misses are the well-appearing patient with a time-critical diagnosis: the stroke still inside the window, the atypical MI, the early sepsis whose vitals haven’t declared yet. Under-triage of a level 2 to a level 3 is the error with the most serious consequences in the whole system. If you’re genuinely torn between 2 and 3, the safer call is 2.

Levels 3 to 5 — counting resources

Only once 1 and 2 are excluded does the question change to: how many different resources will this patient need to reach a disposition?

Level	Resources
3	Two or more
4	One
5	None

A resource is a distinct type of intervention — labs, imaging, IV fluids or IV/IM medication, a specialist consult, a simple procedure. Things that are part of any assessment don’t count: the history and exam themselves, point-of-care testing, oral medications, a prescription, a tetanus shot, crutches or a sling.

The distinction that trips people up is oral versus IV. Oral medication isn’t a resource; IV medication is. Same drug, different route, different level.

Some worked examples:

Ankle injury needing an X-ray — one resource, level 4.
Abdominal pain needing labs, imaging and IV fluids — two or more, level 3.
Simple prescription refill or suture removal — no resources, level 5.

You're estimating, not prescribing

You aren’t ordering the workup — you’re predicting what an experienced clinician would order for this presentation. The point is sorting speed, not accuracy about any individual chart. Don’t stall in triage trying to get the count exactly right.

The danger-zone vitals check

There’s one more step people skip. Before you settle on a level 3, look at the vital signs. If the heart rate, respiratory rate, or oxygen saturation falls outside the accepted range for the patient’s age, consider upgrading to level 2.

This exists precisely to catch the patient whose story sounds routine but whose numbers don’t. It’s the safety net on the most dangerous under-triage in the tool, and it’s the step most often left out under pressure.

What good triage documentation records

Triage is a decision that gets reviewed, so record what drove it: the presenting complaint in the patient’s words, a full set of vitals, pain score, relevant history and allergies, the level assigned — and, when a patient is upgraded, why.

“Level 2 — chest pain with radiation to jaw, diaphoretic, HR 112” is a defensible entry. “Level 2 — chest pain” is not.

The bottom line

Ask the danger question first and the resource question second. Never run them together, and never let the resource count talk you out of a level 2 you’ve already justified. The tool is fast because the order is fixed.

Keep going

For deterioration after the patient is through the door, early warning scores covers the track-and-trigger side. For the time-critical presentation most likely to be under-triaged, blood pressure in acute stroke and TNK vs tPA.

Sources

References

ENA — Emergency Nurses Association — triage standards and ESI education
AHRQ — Emergency Severity Index Implementation Handbook — the algorithm, decision points, and worked cases
StatPearls — Triage — triage systems overview