Did the Emergency Four-Hour Target Shorten Waiting, or Change How Hospitals Count It?

Geoffrey Chen

When Numbers Start Making Decisions · Season Two, “When the Measure Becomes the Target” · Article 1

1. The patient can stay still while the score improves

Four hours after a patient arrives at an emergency department, have they left it? The answer seems straightforward: look at whether the patient has physically departed the emergency area. In a performance system built from timestamps, however, “departure” can acquire more than one administrative meaning.

A 2019 study reanalysed 32,184 emergency presentations in New South Wales under different definitions of the four-hour measure. Substituting a “ready to depart” time for physical departure raised apparent performance by six percentage points. Applying one local interpretation of “transferred” raised the result for admitted patients by sixteen percentage points. Simulations also found that reported performance could improve without changing total length of stay if the timing of an admission decision changed or more emergency short-stay capacity was used. Hession and colleagues: “Gaming National Emergency Access Target performance using emergency department short stay units and clinical initiatives”

That finding does not prove that hospitals routinely falsified records. Nor does it show that short-stay units lack clinical value. It establishes a narrower and more important point: once a time measure starts affecting reputation, management pressure and resource allocation, its definitions become part of the field on which organisations act. The number may still be described as “the proportion completed within four hours”, while the reality being measured quietly shifts from “the patient reached an appropriate destination sooner” to “the database received a qualifying code before the deadline”.

Season Two examines this change. Season One asked how thresholds divide a continuous reality into two institutional positions. This season asks what happens after a threshold becomes a target. How do people and organisations reorganise work around it? When a score rises, how can we distinguish a genuine improvement in the world from an improvement in how the world is recorded?

2. Four hours is not the time until a doctor first sees you

Australian governments agreed to the National Emergency Access Target, or NEAT, in 2011. The original path was intended to reach a national target of 90 per cent by 2015: within four hours of arrival, emergency patients would be admitted to hospital, referred or transferred elsewhere, or discharged. The target was not 100 per cent because some patients should remain longer for clinical reasons. National Health Performance Authority: “Time patients spent in emergency departments in 2012 and 2013”

Two features of that definition are easily obscured in ordinary discussion.

First, it is not a measure of the time before treatment begins. Someone may be triaged and treated quickly but wait ten hours in emergency because no inpatient bed is available. Another patient may first see a clinician later but go home inside four hours. The NEAT measure covers total time from presentation to physical departure. AIHW metadata defines its numerator as completed presentations with a length of stay no greater than 240 minutes and its denominator as all completed presentations. AIHW METEOR: “National Emergency Access Target”

Second, the measure tests patient flow through the whole hospital, not merely the speed of emergency doctors. When a patient needs admission but no ward bed is available, the emergency department cannot produce departure alone. Imaging, pathology, specialist consultation, ward acceptance, bed management and community services can all affect the clock. A measure displayed at the emergency door can therefore demand coordinated change throughout the institution.

That is one of its strengths. Crowding and prolonged emergency stays are not merely uncomfortable; they are associated with safety risks. The target converted a distributed problem, for which no single department held the whole picture, into a visible shared constraint. Patient flow ceased to be background misfortune and became an outcome that hospital leadership had to manage.

3. Why four hours and 90 per cent?

The human body does not encounter a new natural danger at minute 241, and 90 per cent is not the only medically valid proportion. The figures are institutional structures designed to make a continuous risk manageable.

The Australian target drew on the United Kingdom’s four-hour rule and on evidence linking emergency crowding, access block and poor outcomes. An Australian retrospective study covering 59 hospitals and data from 2010 to 2014 found that improved four-hour performance was associated with lower risk-adjusted in-hospital mortality among admitted emergency patients. The relationship remained observable to a compliance level of about 83 per cent, but was not similarly established above that point. Sullivan and colleagues: “The National Emergency Access Target and the 4-hour rule—time to review the target”

This was an observational study. It cannot show that the four-hour target alone caused the mortality reduction. The improvement may have resulted from whole-of-hospital reforms prompted by the target, better bed coordination, earlier senior decisions or other changes occurring at the same time. It does show that the clock was not simply chosen for an attractive dashboard. Moving patients safely out of crowded emergency departments has a legitimate clinical purpose.

The precise proportion remains a policy choice. A target set too low may fail to produce reform. One set too high may force justified clinical exceptions to serve the clock. Later analysis has suggested that a figure around 80 per cent may be more defensible than treating 90 per cent as an unquestionable fact of nature. A threshold must be clear to coordinate action, but it must also remain revisable so that a historical decision does not masquerade as a permanent medical truth.

4. How a measure enters everyday hospital action

The target differs from an ordinary statistic because it entered daily operations. Managers can see which patients are approaching the limit. Teams plan admission or discharge earlier. Specialists are asked to respond faster. Bed managers experience pressure to release capacity. A number that once described an annual average becomes a countdown that influences the next action.

That change can produce three kinds of result.

The first is genuine improvement. A hospital reduces repeated handovers, requests tests earlier, brings senior decision-makers forward and coordinates beds or home support. Time falls without compromising care. The measure has revealed a bottleneck and forced structural cooperation.

The second is local optimisation. Attention concentrates on patients nearing four hours. Those who have waited two hours and those already beyond eight hours may receive less management attention. Work that matters for quality but does not count towards the target can be displaced. Total flow may improve, but the improvement is uneven.

The third is definitional improvement. A patient is administratively transferred to a short-stay unit, marked ready to depart, or assigned another code that stops the emergency clock while their actual location and care pathway change little. The report improves more than the patient’s experience.

All three can coexist. Evidence of gaming does not erase genuine gains. Evidence that some patients left sooner does not prove that every higher score represents clinical improvement. A mature evaluation must separate the mechanisms rather than use one finding to dismiss the others.

5. What a single target makes invisible

The four-hour proportion asks only whether the episode ended within 240 minutes. It does not directly answer whether treatment began within the clinically appropriate triage time, pain was managed, diagnosis was accurate, discharge was premature, the destination was suitable, the patient returned, or staff were placed under unsustainable strain.

Two hundred and thirty-nine minutes and 241 minutes may feel almost identical to patients, yet appear on opposite sides of the performance table. A patient moved to an unsuitable unit at three hours and fifty-nine minutes counts as success. A complex patient kept safely in emergency for four hours and ten minutes counts as failure.

Time still matters. It is simply one dimension of the outcome. Once strong consequences attach to a measure, the measured organisation adapts. A hospital is not a stationary object and the indicator is not a thermometer standing outside it. Together they form a new operating structure. The useful lesson is more demanding than repeating Goodhart’s law: when measurement enters a structure, it changes attention, resources, roles and feasible actions. We must therefore inspect changes in the number together with changes in the system that produced it.

6. How can we know whether patients benefited?

A more reliable emergency performance system should read at least five groups of evidence together:

  1. The full time distribution: not only the four-hour proportion, but the median, 90th percentile, admitted-patient stays and extreme waits.
  2. Clinical outcomes: deaths, deterioration, returns, unplanned readmissions and unsafe discharge.
  3. The patient’s location: whether departure was physical, and whether a short-stay unit delivered suitable care instead of moving an administrative boundary.
  4. Distributional differences: whether triage groups, ages, disabled people, mental-health patients and those needing admission benefited similarly.
  5. Work processes: the definitions of timestamps, amended records, bunching near the limit and staff reports about target pressure.

National reporting continues to show the four-hour proportion, but now places it in a broader distribution. AIHW reports that in 2024–25, 53 per cent of completed emergency presentations ended within four hours. Ninety per cent ended within 11 hours and 16 minutes, while the 90th-percentile stay for patients subsequently admitted approached 19 hours. AIHW: “Emergency department care”

The latter figures reveal the long tail hidden by a single proportion. If two systems have the same four-hour result but one reduces the longest waits from twenty hours to twelve, that is a material improvement. If a hospital merely moves many cases from just over four hours to three hours and fifty-nine minutes while the long tail worsens, it has not achieved the same thing.

Conclusion: keep the clock, but do not let it stand in for the patient

The four-hour measure can expose crowding, coordinate hospital-wide action and sustain attention to patient flow. It cannot bear the entire judgement of hospital quality. It has encouraged real reform and created room to optimise definitions. Both statements can be true.

My judgement is not that the measure should be abandoned. Its dominance as a single compliance rate should be reduced. Physical departure needs a consistent definition, and the result must be checked against long waits, clinical outcomes, returns, short-stay-unit use and outcomes for different groups. The target proportion itself should be periodically reviewed rather than inherited forever.

What deserves reward is not the closure of a database record at minute 239. It is a patient reaching appropriate care at an appropriate time without safety, dignity or somebody else’s wait being sacrificed. A shared clock can help a hospital coordinate, but it cannot answer where this particular patient ought to go. That commitment still belongs to people and institutions capable of understanding the clinical situation, mobilising resources and accepting responsibility for the result.

Principal sources

Discover more from Geoffrey Chen

Subscribe to get the latest posts sent to your email.