Abstract
STUDY OBJECTIVE: Descriptions of emergency department (ED) census often do not differentiate between patients and encounters, and there is no guidance about which unit of analysis is most appropriate. We explore differences between patient- and encounter-level accounting of ED utilization. METHODS: Data extracted from hospital databases were used to identify all registered patients at 3 different but geographically proximate EDs: urban academic, urban community, and suburban community. Data were available from 2000 to 2009 for the academic ED and for all 3 EDs from 2003 to 2007. For each ED, we calculated number of encounters, proportion of encounters for "annual new patients" (ie, not seen previously that year), and number of "new patients" (ie, not seen previously during study period). We also determined the annual number of encounters per patient for each ED. RESULTS: At the academic ED, there were 890,397 encounters involving 256,805 patients. Annual encounters (≈89,000) and patients (≈49,000) remained relatively stable over time. Patients were new in 36.1% (95% confidence interval [CI] 35.8% to 36.4%) of year 2 encounters, 25.3% (95% CI 25.1% to 25.6%) of year 5 encounters, and 22.4% (95% CI 22.1% to 22.7%) of year 10 encounters. For community EDs, 50.9% (95% CI 50.4% to 51.5%) and 53.7% (95% CI 53.1% to 54.2%) were new in year 2, and by the fifth year, 35.0% (95% CI 34.5% to 35.5%) and 36.2% (95% CI 35.7% to 36.7%) were new. In the academic ED, 56% of patients had a single encounter during 5 years and less than 6% had more than 8 encounters during that period. In community EDs, 62% of patients had a single encounter during 5 years and less than 3% had more than 8 encounters overall. CONCLUSION: EDs provide care to a relatively static population, with truly new patients composing only a minority of encounters. Although multiple encounters per patient are common, highly frequent use occurs for only a minority of ED patients, and then only for a discrete period. Encounters and patients are not equivalent units of analysis, and policymakers and researchers should determine which is most appropriate for their decisionmaking.