Study setting and population
This study was conducted in the ED of a university hospital, in Nîmes, France, with an average annual census of 75,000 visits per year. The ED welcomes all medical and surgical emergencies for adults and children, except ophthalmological and gynecological emergencies. Our hospital consists of an 870-bed (medicine, surgery, and obstetrics), university-affiliated tertiary care hospital with 76,000 inpatient admissions and 317,000 ambulatory visits per year. We included all ED patient visits in our study.
Study design
We conducted a “before-after” study of segmentation of the ED including all patient visits to the ED from January 1, 2011 to September 30, 2011 and from January 1, 2012 to September 30, 2012. The ED segmentation consisted of the development of a new patient care geographical layout on a pre-existing site and changing the organization of patient flow. It took place on May 10, 2012. All patient visits to the ED between January 1, 2011 and the implementation of the organizational change were considered as belonging to the “before” period, the others to the “after” period. Winter was defined as the period from January 1 to May 10. Summer was defined as the period from May 10 to September 30. Therefore, the “before” periods included winter 2011, summer 2011, and winter 2012. The “after” period was defined as summer 2012 (Fig. 1). This study was approved by the Institutional Review Board of Nîmes University Hospital and was declared to the National Commission for Data Processing and Civil Liberties.
Emergency department segmentation
Entities “before” segmentation included the triage unit (one physician, one nurse) using the Canadian Triage and Acuity Scale (CTAS), pediatric emergencies (one physician, one resident, one nurse), and a medico-surgical unit for adults (three physicians, four residents, three nurses) where most of the patients were admitted. The medico-surgical unit included a resuscitation room for life-threatening emergencies (CTAS 1 and 2). Before segmentation, there was no allocation of patients to a specific physician or nurse. Patients who would have been admitted to a ward if a bed was available, waited most of the time in the ED hallways (Fig. 2–a). The segmentation was defined as a new organization of the caregivers into seven sectors corresponding to seven architectural entities: the medico-surgical units I and II (one physician, one resident, one nurse), the traumatic emergencies (one physician, one resident, one nurse), the pediatric emergencies (one physician, one resident, one nurse), the inpatient waiting area (one physician, one nurse), the patients’ triage sector (one nurse in the morning, two in the afternoon) and the resuscitation room (one resident, one nurse). The physician responsible for traumatic emergencies also supported the triage nurse if needed, while one of the physicians from the medico-surgical units supported the resuscitation room if needed. The referral to each sector was made by the triage nurse (Fig. 2–b). Segmentation therefore did not require an increase in physician staffing, but required an increase in nurse staffing. Patients who would have been admitted to a ward if a bed was available waited in the inpatient waiting area.
Measurements
The data were collected through the hospital’s electronic medical record system (InterSystems, Cambridge, United States). The patients were sorted according to CTAS. Inpatient LOS was calculated for hospitalized patients. In our hospital as in many countries, mortality rate is generally higher during the winter and because “before” and “after” periods occurred in different seasons, we analyzed mortality data according to the same time periods the preceding year. Thus, we choose three “before” periods, to adjust for differences in mortality rates across the various seasons.
Outcomes
Our primary outcome was the inpatient all-cause mortality rate of all patients admitted from the ED. Secondary outcomes were the 24-hour and 30-day inpatient mortality, delays to FMC, ED LOS, inpatient LOS, and triage delays.
Data analysis
Qualitative variables were compared using a Chi-squared or Fisher exact test. Quantitative variables were compared using Student t-test or analysis of variance. In cases of non-parametric distributions, the Wilcoxon-Mann-Whitney test was utilized. The results were presented as means (SD) or medians (IQR) where appropriate. For qualitative variables, numbers and percentages were presented. CTAS 1 times were not analyzed because patients’ management was immediate. All tests were two-tailed and all statistical analyses were carried out in R (www.r-project.org). P values below .05 were considered statistically significant.