188 Original article Predictive validity comparison of two five-level triage acuity scales Andrew Worster a , Christopher M. Fernandes a , Kevin Eva b and Suneel Upadhye a Introduction Each of the two most commonly used five- level triage tools in North America, the Emergency Severity Index and the Canadian Triage and Acuity Scale have been used as a measure of emergency department resource utilization in addition to acuity. In both cases, it is believed that patients triaged as having a higher level of acuity require a greater number of emergency department resources. We compared the ability of each tool to predict the emergency department resources for each emergency department visit and associated hospital admission and in-hospital mortality rates. Methods This is an observational, cohort study of a population-based random sample of patients triaged at two emergency departments over a 4-month period. Correlational analyses were performed to examine the relationship between the triage assessment and: (i) resource utilization, (ii) hospital admission, and (iii) in-hospital mortality. Results From 486 patients, analyses revealed the greatest correlation was between Emergency Severity Index and diagnostic resources [ – 0.54 (95% confidence intervals: – 0.58, – 0.50)] and the poorest correlation was between Canadian Triage and Acuity Scale and mortality [ – 0.16 (95% confidence intervals: – 0.20, – 0.12)]. No statistically significant differences (P < 0.005) were observed between each tool ’s ability to predict any of the outcomes measured. Conclusion No statistically significant difference was observed in the ability of Emergency Severity Index v. 3 and Canadian Triage and Acuity Scale to predict emergency department resource utilization or immediate patient outcomes. This ability is, at best, only moderate indicating that other, more accurate tools than measures of triage acuity are required for this purpose. European Journal of Emergency Medicine 14:188–192  c 2007 Lippincott Williams & Wilkins. European Journal of Emergency Medicine 2007, 14:188–192 Keywords: Canadian Triage Acuity Scale, emergency department, Emergency Severity Index, predictive validity, triage a Division of Emergency Medicine and b Department of Clinical Epidemiology and Biostatics, McMaster University, Hamilton, Ontario Canada Correspondence to Dr Andrew Worster, MD, MSc, McMaster University, Emergency Department, 1200 Main Street West, Hamilton, ON, Canada L8N 3ZS Tel: + 1 905 521 2100; fax: + 1 905 521 2608; e-mail: aworster@rogers.com Received 15 April 2006 Accepted 27 September 2006 Introduction Triage is an initial assessment and sorting process applied to patients upon presentation to an emergency depart- ment (ED). Multiple triage tools, typically three to five- level categorical measurement scales using an ordinal ranking of illness or injury acuity, have been created to assist in the triage process. Like all health measurement scales, triage scales must meet at least two important criteria to accurately perform as intended. The first criterion is validity; the ability of the scale to measure what it purports to measure. Second is reliability, the scale’s ability to yield consistent results over repeated observations under the same conditions. A scale that is reliable but not valid has no value [1]. By contrast, one cannot assess the validity of a scale until it is demon- strated to be reliable. The two most commonly used, five- level triage tools in North America, the Emergency Severity Index (ESI) and the Canadian Triage and Acuity Scale (CTAS) have both been shown to have a high interrater reliability [2–4]. ESI was developed specifically as a measure of ED resource utilization in addition to acuity and studies demonstrating its predictive validity as a measure of ED resource utilization have been reported [5–7]. CTAS, also a five-level triage acuity measure with the same interrater reliability as ESI, was not developed specifically as a measure of ED resource utilization but has been assessed for this purpose [8,9]. The premise in both cases is that patients triaged as having a higher level of acuity require a greater number of ED resources during their ED visit. This study sought to prospectively compare the predictive validity of these triage tools as measures of ED resource utilization and immediate patient outcomes. Methods Objectives The primary objective was to measure and compare the ability of each of two five-level triage tools to predict the 0969-9546  c 2007 Lippincott Williams & Wilkins