Improving NEWS2 predictive accuracy

  • Research type

    Research Study

  • Full title

    Improving the ability of the National Early Warning Score (NEWS2) system to predict critical outcomes through additional patient data or amendments to the scoring process

  • IRAS ID

    344472

  • Contact name

    Chris Plummer

  • Contact email

    chris.plummer@nhs.net

  • Sponsor organisation

    The Newcastle upon Tyne Hospitals NHS Foundation Trust Newcastle Joint Research Office

  • Clinicaltrials.gov Identifier

    11002, R&D

  • Duration of Study in the UK

    1 years, 0 months, 0 days

  • Research summary

    In 2012, the Royal College of Physicians developed a National Early Warning Score (NEWS) tool to help healthcare professionals monitor patients and identify any deterioration in their status early enough to allow effective intervention to improve outcomes. This system was refined in 2017 as NEWS2, and has been widely adopted throughout the NHS. Previous studies have found that NEWS2 performs well in predicting patients who are likely to die in the next 24 hours, but that its ability to predict other important clinical outcomes over longer time periods is limited. Significantly, the system was not developed with data from older patients, and therefore is less good at predicting outcomes in this growing population. This is
    particularly important as older adults are especially vulnerable to sudden physiological changes. In an ageing population, a tool that better predicts outcomes early enough for effective interventions, and can learn from continually accumulating data, would change clinical practice and significantly improve patient safety. Healthcare professionals would have the opportunity to prioritise evidence-based treatment to those most likely to benefit.

    In this study, we will use approximately 7 years of historical, anonymised patient data from the Newcastle upon Tyne Hospitals NHS Foundation Trust to explore ways in which a new alerting system could improve on NEWS2. We will use a large number of routinely collected patient variables which are linked with patient outcomes, and use these linked data to determine which variables, and with which algorithm, best predicts outcomes. We will then test the models developed on another set of patient data to determine their positive and negative predictive accuracy. After this study, we intend to test the best models on data from other NHS Trusts, to determine how well they perform in different patient populations and different healthcare environments.

  • REC name

    Wales REC 4

  • REC reference

    25/WA/0060

  • Date of REC Opinion

    1 Apr 2025

  • REC opinion

    Further Information Favourable Opinion