Article Text

Download PDFPDF

An early warning? Universal risk scoring in emergency medicine
  1. Damian Roland,
  2. Timothy J Coats
  1. Leicester University, Leicester, UK
  1. Correspondence to Dr Damian Roland, Emergency Medicine Academic Group, Room 003A, Robert Kilpatrick Clinical Sciences Building, Leicester Royal Infirmary, Leicester LE1 5WW, UK; dr98{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

The concept of triage is well established in emergency medicine. Originating in the Napoleonic wars it has been used internationally to determine the clinical need of patients presenting to emergency departments (ED). Triage systems have typically relied on the use of experienced staff or decision support systems to judge the time a patient can afford to wait before treatment commences. There are various triage models in place, but all follow this same principle and can be applied to both adults and children.

Despite being well established in emergency medicine the concept of ‘triage’ has only recently been adopted on hospital wards. It had become increasingly recognised that patients who deteriorated in hospital, especially those who subsequently died or were admitted to intensive care, demonstrated measurable physiological changes hours before recognition by medical and nursing staff.1 This led to the development of ‘early warning scores’ (a form of ward triage system), which have proliferated from their initial inception nearly 15 years ago.2 The NHS modernisation agency popularised the term ‘track and trigger systems’ (TTS) to encompass the variety of different approaches employed. They defined TTS as periodic observations of selected basic physiological signs (tracking) with predetermined calling or response criteria (trigger). The response involves more frequent observation, an alert to senior nursing/medical staff, or a review by a specialist critical care outreach team. An excellent systematic review of the subject demonstrated that a wide variety of TTS were in use, but ‘with little evidence of reliability, validity and utility’.3 The authors recommended the need for larger prospective studies with timed recordings of all important outcomes, including mortality, cardiac arrest and admission to intensive care.

The development of inhospital TTS from ED triage has now come full circle with groups implementing TTS in ED as an adjunct to triage4 prompted by the finding that patients who were admitted to intensive care directly from an ED did better than those admitted from general wards.5 However, we cannot be certain a TTS derived from inpatients will also be useful in the ED. As TTS are predictive models they are unlikely to work well if applied to a patient group with different characteristics to the derivation dataset. Current TTS are derived originally from hospital inpatients who have received treatment and therefore will have different physiology to patients in the untreated ED arrival state. We should not expect that a TTS derived from inpatients will perform as well when applied to patients in the ED. This effect was seen in a study comparing various TTS with the Manchester triage system, which found that the TTS failed to identify a greater number of critically ill patients.6

In July 2007, the National Institute for Health and Clinical Excellence in the UK released a short clinical guideline entitled ‘Acutely ill patients in hospital: recognition of and response to acute illness in adults in hospital’. One of the key priorities of the document was the recommendation that a physiological TTS should be used to monitor all adult patients in acute hospital settings. The confidential enquiry into maternal and childhood death published in 2008 recommended a similar approach in children's hospitals. However, the concept an NHS-wide TTS used at all stages of the acute care pathway is intellectually flawed.7 Different parts of the NHS have patients with very different pre-test probabilities of illness. If the same system is used throughout the NHS, areas with a low prevalence of serious acute disease (such as primary care) will have high false-positive rates (and low false-negative rates) compared with areas with high prevalence (such as the ED). There are, of course, advantages to having a universal method of severity scoring—the huge success of the Glasgow coma scale has demonstrated the importance of having a consistently measured way of assessing a patient. However a ‘NHS early warning score’ is unlikely to perform well in all patient groups, and may even give a large amount of false reassurance or false cause for concern, depending on the incidence of acute severe disease in the patients.

Triage on arrival is well established but as the demand for urgent care increases there is a need for systems to recognise the ‘sick patient’, identify patient deterioration, and predict the requirement for a higher level of care within the hospital. The best means to enable this beyond initial triage have yet to be defined. There is undoubtedly a need for an ED TTS, but simply using an inpatient-derived model, although an attractively simple concept, is potentially flawed. A TTS is unlikely to perform well across different NHS settings so any score proposed for ED use must be validated in ED patients.



  • Competing interests None.

  • Provenance and peer review Not commissioned; not externally peer reviewed.