Most in-hospital cardiac arrests aren’t sudden. Look back through the chart and the warning signs are usually there for hours — a respiratory rate creeping up, a blood pressure drifting down, a patient who got quietly harder to rouse. The physiology declares itself well before the crisis. The problem has never been that the signs are absent; it’s that they arrive one set of observations at a time, on different shifts, to different people, and nobody assembles them.

Early warning scores exist to do that assembling automatically.

What the score is made of

The Modified Early Warning Score turns five routine bedside measurements into points, and adds them up:

Respiratory rate
Heart rate
Systolic blood pressure
Temperature
Level of consciousness (typically AVPU — Alert, Voice, Pain, Unresponsive)

Each is scored against a normal band, with points rising as a value moves further from it in either direction. Nothing here requires equipment you don’t already have at the bedside, which is the whole point — the score is built from observations that are being taken anyway.

Respiratory rate is the one that gets faked, and the one that matters most

It’s the most sensitive early marker of deterioration and the most likely to be estimated rather than counted. A tachypnoeic patient is compensating for something — sepsis, acidosis, a pulmonary embolus, pain — and a rate charted as “18” because nobody counted throws away the single most predictive number on the page. Count it. For a full minute if anything looks off.

What the number means

Higher scores mean more monitoring, faster. Typical escalation looks something like: a low score continues routine observations; a middle score increases their frequency and prompts a nursing review; a high score triggers an urgent provider review or a rapid response call.

The specific thresholds are your facility’s, and this genuinely varies — many hospitals have moved to NEWS2, which adds oxygen saturation and supplemental oxygen use, and treats a confused patient as a red flag in its own right. Know your own track-and-trigger policy and the numbers attached to it.

Two structural points matter more than any threshold:

A single high subscore counts. An aggregate in the reassuring range can hide one profoundly abnormal value — most policies escalate on any single extreme parameter regardless of the total, and if yours doesn’t say so explicitly, treat it that way anyway.

The trend beats the number. A score of 3 that was 0 four hours ago is a different patient from a score of 3 that’s been 3 all day. Deterioration is a direction, and the score is most useful read as a series.

The score does not have a veto over you

An early warning score is a safety net, not a permission slip. If a patient looks wrong and the number is reassuring, escalate anyway — scores are calibrated for populations and will miss individuals, particularly the young, the pregnant, and anyone on medication that blunts a heart-rate response. Beta blockers can hold a heart rate flat through a shock state. “But their MEWS was 2” has never been an adequate account of why nobody called.

When the score climbs

The number is the prompt; the work is what follows.

Reassess properly, and in person. Full set of vitals with a counted respiratory rate, a look at the patient, and a check that the readings are real — a cuff on the wrong arm or a cold probe manufactures a crisis.
Reach for the obvious reversibles. Airway, oxygen, position, glucose, pain, a bladder scan. Some climbing scores resolve at this step.
Escalate on the policy, and say the number. “Mr —’s MEWS has gone from 1 to 5 since 0800, mostly respiratory rate and blood pressure” is a sentence that gets a response. Vague concern often doesn’t.
Increase the observation frequency and hold it there until the picture is clearly improving — not until the next routine round.
Document the score, the escalation, and the response. Who you called, when, and what was decided. A deteriorating patient generates a record that gets read closely afterwards.

Why this belongs to nursing

Early warning scores are one of the few interventions where the mechanism is almost entirely nursing labour. Somebody has to count the respiratory rate, notice the trend, do the arithmetic, and make the call. Rapid response teams are only as good as the trigger that summons them, and the trigger is you.

That’s also why the failure modes are nursing ones: estimated respiratory rates, scores calculated but not acted on, and escalation delayed because the patient “didn’t look that bad yet.”

The bottom line

Deterioration announces itself in ordinary vital signs, hours ahead. An early warning score is just a structured way of hearing it. Count the respirations honestly, read the trend rather than the snapshot, and escalate on the number or on your judgement — whichever fires first.

Try it

Add up the five parameters with the MEWS calculator.

Keep going

The other proactive score on a med-surg floor: the Braden Scale. For documenting a changing patient so the trend is visible to whoever comes next, neuro checks: documenting them defensibly covers the principle.

Sources

References

RCP — National Early Warning Score (NEWS2) — the current standardised successor to MEWS
StatPearls — Early Warning Scores — scoring systems and evidence
AHRQ — Rapid Response Systems — escalation and response team evidence