Decision-tree early warning score (DTEWS) validates the design of the National Early Warning Score (NEWS)

James S. Briggs,Gary B. Smith,Stuart W. Jarvis,Paul E. Schmidt,Peter I. Featherstone,Paul Meredith,David R. Prytherch,Tessy Badriyah

doi:10.1016/j.resuscitation.2013.12.011

Abstract

Aim of study: To compare the performance of a human-generated, trial and error-optimised early warning score (EWS), i.e., National Early Warning Score (NEWS), with one generated entirely algorithmically using Decision Tree (DT) analysis. Materials and methodsWe used DT analysis to construct a decision-tree EWS (DTEWS) from a database of 198,755 vital signs observation sets collected from 35,585 consecutive, completed acute medical admissions. We evaluated the ability of DTEWS to discriminate patients at risk of cardiac arrest, unanticipated intensive care unit admission or death, each within 24h of a given vital signs observation. We compared the performance of DTEWS and NEWS using the area under the receiver-operating characteristic (AUROC) curve. ResultsThe structures of DTEWS and NEWS were very similar. The AUROC (95% CI) for DTEWS for cardiac arrest, unanticipated ICU admission, death, and any of the outcomes, all within 24h, were 0.708 (0.669–0.747), 0.862 (0.852–0.872), 0.899 (0.892–0.907), and 0.877 (0.870–0.883), respectively. Values for NEWS were 0.722 (0.685–0.759) [cardiac arrest], 0.857 (0.847–0.868) [unanticipated ICU admission}, 0.894 (0.887–0.902) [death], and 0.873 (0.866–0.879) [any outcome]. ConclusionsThe decision-tree technique independently validates the composition and weightings of NEWS. The DT approach quickly provided an almost identical EWS to NEWS, although one that admittedly would benefit from fine-tuning using clinical knowledge. We believe that DT analysis could be used to quickly develop candidate models for disease-specific EWSs, which may be required in future.

Highlights

In 2010, our group developed a novel early warning scoring system – ViEWS (VitalPAC Early Warning Score) – for use in the early recognition and response to patient deterioration.[1]
The area under the receiver-operating characteristic (AUROC) for decision-tree Early Warning Scores (EWSs) (DTEWS) for cardiac arrest, unanticipated ICU admission, death, and any of the outcomes, all within 24 h, were 0.708 (0.669–0.747), 0.862 (0.852–0.872), 0.899 (0.892–0.907), and 0.877 (0.870–0.883), respectively
The Decision Tree (DT) approach quickly provided an almost identical EWS to National Early Warning Score (NEWS), one that admittedly would benefit from fine-tuning using clinical knowledge

Summary

Introduction

In 2010, our group developed a novel early warning scoring system – ViEWS (VitalPAC Early Warning Score) – for use in the early recognition and response to patient deterioration.[1] ViEWS was constructed using an iterative, pragmatic, ‘trial and error’ approach, with the cut-offs for its scoring bands being deliberately adjusted to maximise its ability to predict in-hospital death within 24 hours of a vital signs observation. We have since shown that the ability of NEWS to discriminate patients at risk of cardiac arrest, unanticipated intensive care unit admission or death within 24 hours of a NEWS value is superior to that of 33 other Early Warning Scores (EWSs).[3] We wished to compare the structure and discriminative performance of the human-generated, trial and error-optimised EWS, NEWS, with an EWS generated entirely algorithmically using Decision Tree (DT) analysis, against the combined outcome of cardiac arrest, unanticipated intensive care unit admission or death within 24 hours of a given vital signs observation

Results

Discussion

Conclusion