International audienceWe use discrete event simulation to model and analyze a real-life emergency department (ED). Our approach relies on the appropriate integration of most real-life ED features to the simulation model in order to derive useful practical results. Data is supplied from the ED of the urban French hospital Saint Camille. Our purpose is to optimize the human resource staffing levels. We want to minimize the patient average length of stay (LOS), by integrating the staffing budget constraint and a constraint securing that the most severe incidents will see a doctor within a specified time limit. The second constraint allows to avoid the perverse effect of only considering the LOS metric that would delay the treatment of the most urgent patients. We use simulation-based optimization, in which we perform a sensitivity analysis expressing LOS as a function of the staffing budget and also the average door-to-doctor time for urgent patients (DTDT). We show that the budget has a diminishing marginal effect on the problem solution. Due to the correlation between LOS and DTDT, we also observe that the DTDT constraint may significantly affect the feasibility of the problem or the value of the optimal solution

BackgroundPredictors of unscheduled return visits (URV), best time-frame to evaluate URV rate and clinical relationship between both visits have not yet been determined for the elderly following an ED visit.MethodsWe conducted a prospective-observational study including 11,521 patients aged ≥75-years and discharged from ED (5,368 patients (53.5%)) or hospitalized after ED visit (6,153 patients). Logistic Regression and time-to-failure analyses including Cox proportional model were performed.ResultsMean time to URV was 17 days; 72-hour, 30-day and 90-day URV rates were 1.8%, 6.1% and 10% respectively. Multivariate analysis indicates that care-pathway and final disposition decisions were significantly associated with a 30-day URV. Thus, we evaluated predictors of 30-day URV rates among non-admitted and hospitalized patient groups. By using the Cox model we found that, for non-admitted patients, triage acuity and diagnostic category and, for hospitalized patients, that visit time (day, night) and diagnostic categories were significant predictors (p<0.001). For URV, we found that 25% were due to closely related-clinical conditions. Time lapses between both visits constituted the strongest predictor of closely related-clinical conditions.ConclusionOur study shows that a decision of non-admission in emergency departments is linked with an accrued risk of URV, and that some diagnostic categories are also related for non-admitted and hospitalized subjects alike. Our study also demonstrates that the best time frame to evaluate the URV rate after an ED visit is 30 days, because this is the time period during which most URVs and cases with close clinical relationships between two visits are concentrated. Our results suggest that URV can be used as an indicator or quality.