Forecasting emergency department arrivals: a tutorial for emergency department directors.

This article is a tutorial for emergency department (ED) medical directors needing to anticipate ED arrivals in support of strategic, tactical, and operational planning and activities. The authors demonstrate our regression-based forecasting models based on data obtained from a large teaching hospital's ED. The versatility of the regression analysis is shown to readily accommodate a variety of forecasting situations. Trend regression analysis using annual ED arrival data shows the long-term growth. The monthly and daily variation in ED arrivals is captured using zero/one variables while Fourier regression effectively describes the wavelike patterns observed in hourly ED arrivals. In our study hospital, these forecasting methods uncovered: long-term growth in demand of about 1,000 additional arrivals per year; February was generally the slowest month of the year while July was the busiest month of the year; there were about 20 fewer arrivals on Fridays (the slowest day) than Sundays (the busiest); and arrivals typically peaked at about 10 per hour in the afternoons from 1 p.m. to 6 p.m., approximately. Because similar data are routinely collected by most hospitals and regression analysis software is widely available, the forecasting models described here can serve as an important tool to support a wide range of ED resource planning activities.

The emergency severity index triage algorithm version 2 is reliable and valid.

OBJECTIVES: Initial studies have shown improved reliability and validity of a new triage tool, the Emergency Severity Index (ESI), over conventional three-level scales at two university medical centers. After pilot implementation and validation, the ESI was revised to include pediatric and updated vital signs criteria. The goal of this study was to assess ESI version (v.) 2 reliability and validity at seven emergency departments (EDs) in three states. METHODS: In part 1, interrater reliability was assessed using weighted kappa analysis of written training cases and postimplementation by a random sampling of actual patient triages. In part 2, validity was analyzed using a prospective cohort with stratified random sampling at each site. The ESI was compared with outcomes including resource consumption, inpatient admission, ED length of stay, and 60-day all-cause mortality. RESULTS: Weighted kappa analysis of interrater reliability ranged from 0.70 to 0.80 for the written scenarios (n = 3289) and 0.69 to 0.87 for patient triages (n = 386). Outcomes for the validity cohort (n = 1042) included hospitalization rates by ESI triage level: level 1, 83%; 2, 67%; 3, 42%; 4, 8%; level 5, 4%. Sixty-day all-cause mortality by triage level was as follows: level 1, 25%; 2, 4%; 3, 2%; 4, 1%; and 5, 0%. CONCLUSIONS: ESI v. 2 triage produced reliable, valid stratification of patients across seven sites. ESI triage should be evaluated as an ED casemix identification system for uniform data collection in the United States and compared with other major ED triage methods.