How reliable is emergency department triage?

Study objective: To measure interrater and intrarater agreement for an emer gency department triage system. Methods: A 2-phase experimental study was conducted using previously descri bed in-person scripted encounters with emergency nurses who perform patient triage and attending emergency physicians at a tertiary referral center. S tandardized patient scenarios were presented twice over 6 weeks. Participan ts rated severity for each patient using a 5-tier triage system (nurses onl y) and estimated the probability of hospital admission, the most appropriat e time frame to physician evaluation (5 choices, from "lmmediate" to "More than 24 hours"), the need for a monitored ED bed, and the need for diagnost ic services. Interrater agreement was measured by a coefficient of agreemen t for multiple raters and multiple categories. Results: Of the 37 participants (fewer than 90% of those eligible), 19(51%) completed both phases (12 nurses, 7 physicians). Four (33%) of the nurses assigned the same severity ratings for the 5 cases in phase 2 as they did i n phase 1. Intrarater agreement among the 12 nurses rating triage severity was .757. Interrater agreement of nurses and physicians was substantial reg arding need for ED monitoring, and moderate to substantial for other triage assessments. Conclusion: There was general agreement in interrater assessment of triage classification. Continued work is necessary to more fully delineate areas o f variation.