Elsevier

Resuscitation

Volume 82, Issue 8, August 2011, Pages 1013-1018
Resuscitation

Clinical paper
Centile-based early warning scores derived from statistical distributions of vital signs

https://doi.org/10.1016/j.resuscitation.2011.03.006Get rights and content

Abstract

Aim of study

To develop an early warning score (EWS) system based on the statistical properties of the vital signs in at-risk hospitalised patients.

Materials and methods

A large dataset comprising 64,622 h of vital-sign data, acquired from 863 acutely ill in-hospital patients using bedside monitors, was used to investigate the statistical properties of the four main vital signs. Normalised histograms and cumulative distribution functions were plotted for each of the four variables. A centile-based alerting system was modelled using the aggregated database.

Results

The means and standard deviations of our population's vital signs are very similar to those published in previous studies. When compared with EWS systems based on a future outcome, the cut-off values in our system are most different for respiratory rate and systolic blood pressure. With four-hourly observations in a 12-h shift, about 1 in 8 at-risk patients would trigger our alerting system during the shift.

Conclusions

A centile-based EWS system will identify patients with abnormal vital signs regardless of their eventual outcome and might therefore be more likely to generate an alert when presented with patients with redeemable morbidity or avoidable mortality. We are about to start a stepped-wedge clinical trial gradually introducing an electronic version of our EWS system on the trauma wards in a teaching hospital.

Introduction

In 1997, Morgan and colleagues described the first early warning score (EWS) system, designed to alert clinicians to deteriorating patients using aggregate weighted scoring of vital signs.1 Many variations of this scheme have since been published.2 Despite evidence that physiological instability precedes critical clinical deterioration,3, 4, 5, 6 EWS systems have not been shown to improve patient outcomes.2, 7, 8

Most EWS systems use weights and cut-off values for vital signs that are derived from expert opinion.2, 9, 10 We argue that clinical conjecture can be improved upon by statistical determination of the ranges of normality in the at-risk population, for each vital sign observation. In this paper we describe the development of a centile-based system derived from such information.

Section snippets

Data collection

A large dataset comprising 64,622 h of vital-sign data, acquired from 863 acutely ill in-hospital patients using bedside monitors, was used to investigate the statistical properties of the four main vital signs: heart rate (HR), respiratory rate (RR), peripheral arterial oxygen saturation (SpO2), and systolic blood pressure (SysBP). The data came from three clinical studies in the United Kingdom (UK) and United States (US) between 2004 and 2008.6, 11, 12, 13

The methodologies of these studies

Results

The patient population characteristics in each study are shown in Table 1.

Discussion

We have presented the ranges of values for the four main vital signs in at-risk hospitalised populations. We suggest that clinicians would want to be alerted when their patient strays outside the limits of normality defined by these ranges. Conversely, it seems reasonable that patients whose vital signs remain within the limits of normality should not score in an alerting system. Our EWS uses the 10th and 90th centiles to define the no-score limits. A centile-based system provides a statistical

Conclusion

Our centile-based statistical approach is in line with the original objectives of early warning scores, namely the identification of abnormal physiology at the time of the test.1, 22 Observations are treated as being abnormal if they lie at the extremes of the distributions of vital signs acquired from representative sets of at-risk hospitalised patients. EWS systems based on such an approach have the potential to identify patients with abnormal vital signs and are more likely to trigger when

Conflict of interest statement

No conflicts of interest to declare.

Acknowledgements

The work described in this paper was funded by the NIHR Biomedical Research Centre Programme. Dr David Clifton is supported by the Wellcome Trust and EPSRC under grant number WT 088877/Z/09/Z.

References (22)

  • D.R. Goldhill et al.

    A physiologically-based early warning score for ward patients: the association between score and outcome

    Anaesthesia

    (2005)
  • Cited by (102)

    • Development and validation of early warning score system: A systematic literature review

      2020, Journal of Biomedical Informatics
      Citation Excerpt :

      Decision trees were employed in three studies [30,32,33], while more advanced tree models using ensemble methods, such as bagging, boosting, and random forest, were utilized in two studies [37,40]. Other statistical learning techniques, including Naïve Bayes classification and Kernel-base density, were also applied for modeling in this review [26,42,44]. In total, we reviewed two studies that used unweighted activation criteria [17,25], 13 studies that utilized aggregated weighted scores [6,18–22,26,27,30,33,39,43,44], and 14 studies that applied complex computerized scores [23,24,28,29,31,32,34–38,40–42].

    View all citing articles on Scopus

    A Spanish translated version of the abstract of this article appears as Appendix in the final online version at doi:10.1016/j.resuscitation.2011.03.006.

    View full text