How Early Warning Scores Like NEWS2 Flag Deterioration

A nurse walks past a bed and thinks the patient looks "off," yet the chart shows numbers that each seem almost normal. An aggregate early warning score exists to turn that unease into something countable. Tools like the National Early Warning Score 2 (NEWS2) take a handful of routine vital signs, award points for how far each strays from a healthy range, add them up, and produce a single total that rises as the body drifts toward trouble. Cross a total of 5, and the chart calls for an urgent bedside review. The score is a prompt to look harder and escalate sooner, not a diagnosis in itself.

Key points#

• NEWS2 scores six vital signs; more extreme readings earn more points, and the points are summed.
• A total of 5 or more, or any single sign scoring its maximum of 3, triggers urgent review.
• "Discrimination" (often an AUROC) measures how well a score separates people who deteriorate from those who do not; the trigger threshold trades sensitivity against false alarms.
• Compared head to head, NEWS2 screens more sensitively for general deterioration, while qSOFA more specifically confirms the sickest patients with suspected infection.

Why one number replaced a wall of charts#

Before a shared score existed, hospitals across the NHS each kept their own local charts, and a patient moving between wards could be judged by different rules in different rooms. The original National Early Warning Score arrived in 2012 after the Royal College of Physicians brought a multidisciplinary group together to standardize the approach. As Bryan Williams describes in Clinical Medicine (2022), the score was not invented from intuition; it was built by linking routinely recorded physiological measurements to hard outcomes such as death, cardiac arrest, and unplanned intensive care admission in large clinical databases, then tuned using feedback from the wards that used it.

Six parameters carry the signal: respiratory rate, oxygen saturation, systolic blood pressure, pulse rate, level of consciousness or new confusion, and temperature. Each earns more points the further it drifts from normal, and a two-point uplift is added for anyone who needs supplemental oxygen to hold their target saturation, because needing help to breathe is itself a warning.

The 2017 revision to NEWS2 patched two weak spots the first version exposed to criticism. A second oxygen saturation scale was added for people with hypercapnic respiratory failure, usually from COPD, whose safe target is 88 to 92% rather than the higher band, so the chart no longer punishes an appropriate reading. And new-onset confusion was written into the consciousness check, turning the AVPU scale into ACVPU, because sudden delirium is an early and easily overlooked sign that something is wrong.

Reading the numbers: discrimination and thresholds#

Two ideas decide whether any warning score earns its place at the bedside.

The first is discrimination, the ability to tell apart patients who will deteriorate from those who will not. It is usually summarized as the area under the receiver operating characteristic curve (AUROC), where 0.5 is no better than a coin flip and 1.0 would be flawless. Real clinical scores land in between, and small differences in AUROC translate into real patients caught or missed.

The second is the trigger threshold, the line where a continuous number becomes an instruction to act. Set the line low and you catch more deteriorating patients (higher sensitivity) but sound more false alarms (lower specificity). Set it high and you get the reverse. There is no threshold that avoids both problems; the choice reflects what a system is willing to trade.

NEWS2 does not rely on a single line. Low totals prompt routine monitoring. Any one parameter hitting its maximum of 3 points prompts a closer look even when the total is otherwise modest, because a single severely abnormal sign can matter as much as several mild ones. An aggregate of 5 or more is the key trigger for urgent review. This banding matters, because a total of 5 built from one dangerously abnormal vital sign is a different clinical picture from a 5 spread across several borderline ones, and both deserve eyes at the bedside.

NEWS2 and qSOFA are built for different jobs#

qSOFA comes from a separate lineage. The Sepsis-3 task force (Singer and colleagues, JAMA 2016) proposed the quick Sequential Organ Failure Assessment as a fast bedside prompt for patients with suspected infection outside the intensive care unit. It awards one point each for a respiratory rate of 22 or higher, altered mental status, and a systolic blood pressure of 100 or lower, with 2 or more flagging elevated risk. In the derivation cohorts, a qSOFA of 2 or more carried a several-fold rise in hospital mortality. It was designed to rule in the sickest, not to screen everyone.

That design goal shapes what happens when the two tools meet the same patients. Across emergency department and prehospital cohorts, qSOFA tends to be highly specific but insensitive: when it fires, serious illness is likely, yet it stays silent for many patients who go on to deteriorate. NEWS2, with more parameters and finer gradations, tends to catch more of those patients at some cost to specificity. In a multicenter prehospital study of suspected infection (Frontiers in Medicine, 2023), NEWS2 discriminated in-hospital mortality slightly better than qSOFA, with AUROCs of roughly 0.76 versus 0.73, and emergency department cohorts tend to show the same pattern. The practical reading is that NEWS2 works better as a general screen for undifferentiated deterioration, while qSOFA serves as a quick rule-in once infection is already suspected. It is also why UK guidance keeps NEWS2 as the standard track-and-trigger tool for general screening.

Where these scores mislead#

No aggregate number replaces clinical judgment, and NEWS2 has known blind spots the Royal College names directly. Applied carelessly, its oxygen scale can nudge clinicians toward over-oxygenating patients with hypercapnic respiratory failure, which is why the second scale exists. It can also read poorly in spinal cord injury, especially tetraplegia or high paraplegia, where autonomic disruption distorts heart rate, blood pressure, and temperature so the vital signs no longer track the underlying illness.

A score also never explains why a patient is failing; it only says that they are. Its real value is that it buys time and attention, and that value holds only if the response system, the people and pathways summoned when the total reaches 5, is actually staffed and ready to act. A high number that no one answers helps no one.