The performance of prognostic scores in predicting outcome for critically ill patients in the emergency department (ED)

Abstract

Background: The Prince of Wales Emergency Department Score (PEDS) was developed to predict likelihood of mortality and evaluate the intensive-care needs of undifferentiated patients presenting to the resuscitation room. PEDS is specific for critically ill ED patients and could aid in triage for intensive care unit admission. Objectives: To validate and refine PEDS, and compare its performance with six previously published scores for risk-stratifying critically ill ED patients: MEDS (Mortality in the Emergency Department Sepsis), MEES (Mainz Emergency Evaluation Score), MEWS (Modified Early Warning Score), Simple Clinical Score (SCS), WPSS (Worthing Physiological Scoring System), and REMS (Rapid Emergency Medicine Score). Design: Prospective observational study with ethical approval from The Chinese University of Hong Kong. Setting: The ED of the Prince of Wales Hospital in Hong Kong. Participants: There were 239 consecutive adult patients in triage category 1 and 2, presenting to the resuscitation room from 9:00 a.m.–5:00 p.m., Monday–Friday, September 2008–February 2009. Interventions: Physiological, point-of-care, laboratory, radiological, and demographic data to calculate all scores were collected. Main outcome measures: Composite primary outcome of 7-day mortality or intensive care admission. Results: Area under receiver operating characteristic curve for PEDS validation cohort: 0.784 (95% confidence interval [CI] 0.726–0.834); refined PEDS: 0.795 (95% CI 0.734–0.844); MEDS: 0.585 (95% CI 0.520–0.648), p < 0.001; MEES: 0.731 (95% CI 0.670–0.786), p = 0.202; MEWS: 0.744 (95% CI 0.684–0.798), p = 0.328; SCS: 0.721 (95% CI 0.660–0.777), p = 0.133, WPSS: 0.749 (95% CI 0.689–0.802), p = 0.328, REMS: 0.721 (95% CI 0.660–0.770), p = 0.176 (p-values for significance-of-difference from refined PEDS). Conclusion: The performance of all scores deteriorated when prospectively validated. Refined PEDS performed best, but was not significantly better than the other scores, except MEDS. Refined PEDS appears promising as a helpful adjunct to subjective clinical opinion.