Abstract
Objectives
Emergency departments (EDs) have been confronted with growing demand for several decades. However, the strain on EDs is generally not evenly distributed across the year, as some days are at increased risk of overcrowding compared to others. Identifying the days during which teams are unable to meet actual service demands could be beneficial as a means to avoid, in the most drastic cases, redirecting patients to other care structures (ambulance diversion). The objective of this study was to identify the factors explaining ED overcrowding in a general hospital.
Methods
We conducted a retrospective study at a single emergency department between 1 January 2017 and 31 December 2021. The days with the highest deviation from baseline regarding length of stay were defined as being overcrowded. The factors associated with overcrowding were evaluated using a logistic regression model adjusted for the day’s characteristics.
Results
The study period comprised 183 overcrowded days and 1643 uncrowded days. The factors associated with the risk of crowding in multivariable analysis were the ED crowding status of the previous day, the number of patients at the beginning of the day, the proportion of patients age 75 or greater, the number of radiological exams per patient, and the number of radiological exams of the skull and brain region. The mitigating factors were the proportion of patients with less urgent needs during triage, an increased proportion of patients needing trauma care, the proportion of patients < 7 years old, weekend days, and the epidemic period following the month of March 2020.
Conclusion
Our study has highlighted case-mix factors and chronological factors related to the risk of overcrowding. These factors were communicated to hospital management, which allowed the hospital to reassess its operations.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12873-025-01460-0.
Keywords: Emergency medical services, Overcrowding, Emergency room visits, Emergency services, Hospital/trends, Health services accessibility
Introduction
For a decade, as in numerous countries, the French health care system has experienced a major crisis related to the capacity of its hospital emergency departments (EDs) to respond to the demand for unscheduled care [1]. EDs in France are seeing a growing rise in demand for services, while emergency staffing remains limited by hospitals’ training and recruitment capacities. Improvements in access to care in unscheduled-care facilities has become a national priority.
A study conducted in a teaching hospital in the United States showed a higher margin of error in triage as ED occupancy rates increased: a patient would be classified as being more urgent if the treating department was experiencing high occupancy at the time of the patient’s arrival [2]. One cohort study of 2913 patients with severe sepsis or septic shock showed that the proportion of patients treated with antibiotics within 3 h decreased on days of high ED occupancy, although the study did not demonstrate a significant effect on mortality [3, 4].
Numerous studies have investigated the causes of overcrowding in emergency departments. Some have used qualitative methods, which provides important information, but which does not allow a precise estimation of the extent of the phenomena studied [5]. Other studies have been in the form of inquiries to management in specific hospitals, sometimes based on questionnaires using 5-point scales [6], rather than a direct analysis based on local data [7]. In cases where detailed ED data were available, the studies have often covered lengths of time inferior or equal to 1 year [8, 9], and the quantitative sections have examined patient triage without considering patient diagnoses on a finer scale [10, 11]. However, forecasting the circumstances indicative of overcrowding in an ED can facilitate organizational responses both beforehand and downstream, to avoid, in the worst cases, situations such as ambulance diversion, although these situations are rarely observed in France.
The objective of this study was therefore to identify the factors explaining ED overcrowding in hospitals, in particular those factors associated with the timing of a patient’s visit and with patient characteristics.
Methods
Design and study population
We conducted a retrospective study of the flow of patients through a single emergency department of Troyes hospital, located in the east of France. The hospital under study is the largest hospital in the Aube administrative department, which has a population of 310,000 inhabitants and a medical density of 298 physicians for 100,000 inhabitants (among the lowest in France) [7]. The hospital has 442 medical beds, 127 surgical beds, and 63 beds dedicated to gynecology and obstetrics. With a volume of more than 45,000 annual admissions, the ED of Troyes hospital is known for a high stream of activity according to national statistics [12].
The study population consisted of all patients admitted to the ED between 1 January 2017 and 31 December 2021. Within these 5 years, a total of 292,622 admissions were observed, or the equivalent of 58,524.40 visits per year. A 5-year study period was chosen so as to best represent the activity of the ED.
Patients arriving in the ED are initially assessed by a triage nurse. This nurse directs them to the next step according to their condition: the short circuit (patients whose medical condition is judged least urgent) or the long circuit (more complex situations requiring additional assessments such as biological exams, imaging, etc., and/or requiring a specialist’s opinion). Patients judged as having a precarious medical condition are directed to critical care services in the ED, also called déchocage (or trauma center), for immediate intervention.
Main outcome
ED overcrowding is a condition in which the emergency personnel are unable to cope with the number of patient visits, which results in a decrease in ED performance, and lower efficiency in the orientation and treatment of patients. As the ED experiences an increase in its workload (due to the number of patients present in the ED or to the medical complexity of these patients), its staff has to manage multiple requests simultaneously, which results in a decrease in the time that can be allocated for the management each patient, and therefore an increase in the average length of stay of patients.
The principal outcome of our study was a dichotomous classification of each day as overcrowded or uncrowded. Days with the highest average length of stay (LOS), which is consistent with a decrease in ED performance, can generally be considered as overcrowded.
The concept of crowding was therefore defined by using an indicator based on the average LOS for each day. Nonetheless, an examination of LOS is not sufficient to determine the degree of crowding on a single day because variations of this indicator over the long term can arise, with the risk of a duration of several weeks being considered as overcrowded if the average LOS at baseline increases over this period. Indeed, in a normal hospital, overcrowded days are by definition the exception rather than the norm. Thus, across longer stretches of time, overcrowding is better conceptualized as a deviation from the mean of the week or the month in progress as opposed to exceeding a fixed value.
To account for long-term LOS variations in the ED, a mean value of LOS 
and a local standard deviation of LOS 
were calculated for each period of the study, using a generalized additive model (GAM). This model estimated the local average LOS, T, as a variable explained by time, expressed in weeks from the start of the study, based on the observed data. The model thus estimated the long-term trend in LOS.
Then, a standardized difference from the mean LOS of the surrounding days 
was calculated, by a transformation yielded by subtracting the local mean 
estimated by the GAM model from the LOS of the day 
, then dividing this result by the local standard deviation 
:
 |
with 
and 
corresponding to the local means and standard deviations estimated by the respective GAM models.
The indicator of crowding 
calculated for each day can thus be interpreted as the deviation from the mean LOS of a single day compared to the usual LOS observed across the surrounding days. The days of overcrowding were then defined as those on which the deviation from the local mean 
exceeded the 90th percentile of the distribution (10% of the highest values for the period).
Variables measured
The variables examined were: the number of patients present in the ED at midnight, the number of arrivals of that day, the level of crowding observed the previous day, the day of the week (weekend or not), the season, the period before or after March 2020, the average age of the patients (i.e., the proportion of patients aged ≥ 75, between ages 15 and 30, or < 7), the proportion of non-urgent patients (score 1 on the PS classification [13]), the proportion of women, the average number of radiological exams per patient (as it is a commonly practiced routine exam, the electrocardiogram was not counted among these tests), the average number of radiological tests of the skull and brain region per patient, the proportion of patients admitted for trauma or wounds, and the proportion of patients admitted for psychiatric reasons. The diagnoses were coded by the International Classification of Diseases, 10th edition. To avoid a correction of the diagnoses based on small group size, the diagnoses were grouped using a clustering algorithm based on the identification of diagnostic groups arising in a concomitant manner [14]. Only the groups most relevant to explaining the functions of an emergency department were retained for analysis (groups chosen by expert opinion). Other variables were measured to assess the effect of crowding on the emergency department: time before being cared for (time elapsed before any medical prescription, lab test, diagnosis, radiologial exam, surgery or intervention is made or medical history is recorded), percentage of patients in an access block situation (defined as patients hospitalised after 8 h or more in the emergency department), overall time to hospitalisation for hospitalised patients (in minutes), and the percentage of patients who left without being seen.
We conducted a bivariate analysis of the characteristics associated with the deviation from the mean LOS ≥ 10th percentile, according to the variables potentially associated with this variable, chosen by expert opinion. We performed χ2 tests for the categorical variables and Student’s t tests for continuous variables.
Using a multivariable logistic regression analysis, we modeled the probability of a crowded day (deviation from the local mean ≥ 90th percentile). Variables included in the multivariable model were selected by expert opinion.
To account for the possibility of functional changes due to the presence of an epidemic period, a post hoc sensitivity analysis was conducted, taking into account only the period from 1 January 2017 to 29 February 2020, which was not affected by lockdowns due to the Coronarovirus disease 19 (COVID-19).
A p value of < 0. 05 was considered statistically significant. The analyses were performed using the R software, version 4.2.3 (The R Foundation for Statistical Computing, Vienna, Austria).
Ethics and approvals
This study was conducted in conformity with the national legislation relative to epidemiological studies. As the study was purely observational and retrospective, based on pre-existing and anonymous data, it did not need approval by an ethics committee, according to law L1121-1 of the French public health code (n°2012 − 300, 5 march 2012). The study was conducted in accordance with all relevant legislation concerning medical information and has been registered with the national register of observational studies (dossier 23218515). Patients were informed of the study via the register of ongoing studies in the hospital, and they had the option to refuse the use of their personal data. The data used in this study were extracted from the database used by the ED of the hospital called RESURGENCES.
Results
There were 183 days of overcrowding and 1643 uncrowded days (Fig. 1; Table 1). This represents therefore approximately 10% of total days with overcrowding, as conforms with the adopted definition.
Fig. 1.
Changes in daily average length of stay and crowding status during the study period
Table 1.
Characteristics of overcrowded and uncrowded days in the ED during the study period
| Overcrowded day (average LOS ≥ 90th percentile) | Uncrowded day (average LOS < 90th percentile) | P-value | |
|---|---|---|---|
| n | 183 | 1643 | |
| Average patient age (SD) | 42.0 (3.3) | 40.3 (3.4) | < 0.001 |
| Average percent of patients aged ≥ 75 (SD) | 16.8 (3.8) | 14.9 (3.4) | < 0.001 |
| Average percent of patients with age 15–30 (SD) | 22.1 (3.8) | 22.5 (3.8) | 0.12 |
| Average percent of patients < 7 years old (SD) | 12.2 (3.7) | 13.4 (4.3) | < 0.001 |
| Average percent of women (SD) | 50.9 (3.7) | 50.8 (4.2) | 0.81 |
| Daily median time to first medical action (min): median (Q1, Q3) | 109 (92, 129) | 81 (67, 98) | < 0.001 |
| Daily percentage of patients that left without being seen: median (Q1, Q3) | 5.7 (3.7, 8.3) | 3.5 (2.2, 5.6) | < 0.001 |
| Daily median waiting time before hospitalization (hours): median (Q1, Q3) | 7.6 (6.6, 9.0) | 6.5 (5.5, 7.7) | < 0.001 |
| Average daily percentage of access block patients (SD) | 11.0 (3.6) | 8.5 (3.8) | < 0.001 |
| Number of patients present at the beginning of the day [midnight] (SD) | 41.3 (10.1) | 31.0 (9.6) | < 0.001 |
| Average percent of patients with less urgent needs [PS1 classification] (SD) | 65.4 (5.5) | 69.5 (5.0) | < 0.001 |
| Average percent of patients in the mental health cluster (SD) | 8.8 (2.6) | 9.4 (2.7) | 0.01 |
| Average percent of patients in the infectious diseases cluster (SD) | 9.78 (3.47) | 10.70 (3.84) | 0.002 |
| Average percent of patients admitted for wounds and trauma (SD) | 19.4 (4.4) | 21.3 (4.8) | < 0.001 |
| Prior day overcrowding: n days (%) | 48 (26.2) | 135 (8.2) | < 0.001 |
| Average number of daily arrivals (SD) | 149.4 (22.2) | 153.5 (23.4) | 0.024 |
| Average number of long-circuit arrivals (SD) | 49.77 (8.11) | 51.62 (7.80) | 0.002 |
| Season: n days (%) | 0.07 | ||
| Fall | 55 (30.1) | 400 (24.3) | |
| Summer | 37 (20.2) | 423 (25.7) | |
| Winter | 53 (29.0) | 398 (24.2) | |
| Spring | 38 (20.8) | 422 (25.7) | |
| Day of the week: n days (%) | < 0.001 | ||
| Weekend | 25 (13.7) | 496 (30.2) | |
| Weekday | 158 (86.3) | 1147 (69.8) | |
| COVID-19 epidemic period (March 2020–end of study): n (%) | 68 (37.2) | 603 (36.7) | 0.97 |
| Average percentage of patients with lab tests (SD) | 43.2 (5.8) | 42.7 (5.6) | 0.24 |
| Average number of radiological exams per patient above the median: n (%) | 135 (73.8) | 776 (47.2) | < 0.001 |
| Average number of radiological exams of the skull and brain region per patient above the median: n (%) | 127 (69.4) | 780 (47.5) | < 0.001 |
Unit of measurement for the table is a 24-hour day. Days are classified as overcrowded or uncrowded based on the average length of stay. SD = Standard Deviation. PS = Patient state
The overcrowded days were not distributed evenly between the different days of the week. Only 25 overcrowded days occurred on the weekend (13.7% of overcrowded days), suggesting a protective role (p < 0.001). Overcrowded days were often preceded by another day of overcrowding, as was the case for 48 days (26.2%, p < 0.001).
Regarding the motive for a visit to the ED, patients within the mental health cluster were less represented on overcrowded days (on average 8.8% of visits, Standard Deviation SD 2.6%) in comparison to the uncrowded days (on average 9.4% of visits, SD 2.7%; p = 0.01). Additionally, the proportion of patients admitted for trauma was slightly lower on overcrowded days: 19.4% (SD 4.4%) versus 21.3% (SD 4.8%) respectively (p < 0.001).
However, the average number of radiological exams per patient was higher than the median value on 135 (73.8%) of the days of increased crowding (p < 0.0001), as opposed to 47.2% on other days. We observed a similar effect for radiological tests of the skull and brain region.
In multivariable analysis (Fig. 2; Table 2), the protective factors against crowding for a day in the ED included: weekend days (odds ratio (OR) 0.416, 95% CI [0.244, 0.686]), lockdown periods during the COVID-19 pandemic (OR 0.047, 95% CI [0.024, 0.090]), the percentage of patients with non-urgent needs (OR 0.899, 95% CI [0.861, 0.937]), the percentage of patients < 7 years old (OR 0.878, 95% CI [0.817, 0.942] p < 0.001), and the percentage of patients admitted for wounds or trauma (OR 0.888, 95% CI [0.842, 0.935]). The number of patients at the beginning of the day (at midnight) increased the risk of crowding: OR for 10 additional patients 3.769, 95% CI [2.990, 4.808] p < 0.001). The percentage of patients age ≥ 75 slightly increased the risk of crowding (OR 1.079, 95% CI [1.010, 1.153] p = 0.02), and the presence of overcrowding the previous day also increased the risk (OR 1.934, 95% CI [1.187, 3.109] p = 0.01).
Fig. 2.
Forest plot of the multivariable logistic regression analysis modeling the probability of an overcrowded day in the ED
Table 2.
Multivariable logistic regression analysis modeling the probability of an overcrowded day in the ED
| OR (95% CI) | P-Value | |
|---|---|---|
| Average patient age | ||
| Percent of patients aged ≥ 75 | 1.079 (1.010, 1.153) | 0.02 |
| Percent of patients with age 15–30 | 0.962 (0.906, 1.020) | 0.19 |
| Percent of patients < 7 years old | 0.878 (0.817, 0.942) | < 0.001 |
| Percent of women | 0.984 (0.939, 1.032) | 0.51 |
| Number of patients present at the beginning of the day [midnight] OR for 10 additional patients | 3.769 (2.990, 4.808) | < 0.001 |
| Percent of patients with less urgent needs (PS1 classification) | 0.899 (0.861, 0.937) | < 0.001 |
| Percent of patients in the mental health cluster | 0.959 (0.887, 1.036) | 0.29 |
| Percent of patients in the infectious diseases cluster | 0.942 (0.872, 1.017) | 0.13 |
| Percent of patients admitted for wounds and trauma | 0.888 (0.842, 0.935) | < 0.001 |
| Prior day overcrowding | 1.934 (1.187, 3.109) | 0.01 |
| Number of daily arrivals (OR for 10 additional patients) | 0.944 (0.814, 1.095) | 0.45 |
| Number of long-circuit arrivals (OR for 10 additional patients) | 0.732 (0.525, 1.007) | 0.06 |
| Season | 0.36 | |
| Fall | 1 (Réf.) | |
| Summer | 0.615 (0.354, 1.059) | |
| Winter | 0.900 (0.520, 1.553) | |
| Spring | 0.827 (0.481, 1.415) | |
| Weekend (Ref.: Monday to Friday) | 0.416 (0.244, 0.686) | < 0.001 |
| COVID-19 epidemic period (March 2020 until end of study) | 0.047 (0.024, 0.090) | < 0.001 |
| Average number of radiological exams per patient above the median | 2.307 (1.494, 3.602) | < 0.001 |
| Average number of radiological exams of the skull and brain region per patient above the median | 1.528 (1.027, 2.292) | 0.04 |
The risk of overcrowding grew with the number of radiological exams per patient above the median OR 2.307 [1.494, 3.602]. Furthermore, a number of radiological exams of the skull and brain region per patient above the median also increased the risk of crowding, with an OR of 1.528 (95% CI [1.027, 2.292]. The sensitivity analysis including only the first period of the study showed similar results for the majority of the variables (Tables S1 and S2).
Discussion
Our study highlights the factors associated with emergency department crowding for a 24-hour period. The most notable predictors included the average number of radiological exams per patient (risk factor), weekend days (protective factor), and the crowding status of the previous day.
Over the past decade, hospital operations and the demand for care have shown a tendency to increase, mainly because of population aging, associated with an increase in the prevalence of chronic diseases in developed countries [15].
Our research team observed a significant difference between the average age of the patients admitted on crowded days (42.0, SD 3.3) versus an average age of 40.3 (SD 3.4) for patients admitted on uncrowded days (p < 0.001). Studies by Kawano et al. as well as Knapman et al. [9, 11], whose methodologies were similar to our own, had also underscored the important role of the age of arriving patients for the LOS of all patients in the ED. A literature review showed that ED overcrowding was correlated with the degree of urgency of the patients [16]. Patients presenting to the ED are often older and with chronic pathologies for which treatment is complex [4]. Automatization to establish a profile of risk for these patients using algorithms based on their medical histories is a promising avenue of research but one which raises ethical questions [17]. Conversely, patients < 7 years old, when they represent a larger proportion of the case mix on one day, were associated with a lower probability of crowding. Wounds and traumas, as well as patients with less urgent needs who were more rapidly treated, were also associated with a lesser probability of crowding.
The number of arrivals per day was comparable between crowded and uncrowded days (a statistically significant difference in our bivariate analysis showed a greater number of patient arrivals on uncrowded days, but the effect size was not clinically significant, nor was the effect statistically significant in multivariable analysis). However, the number of patients at the start of the day (midnight) was strongly predictive of overcrowding in both the bivariate and multivariable analyses.
The average number of radiological exams per patient was greater on crowded days (p < 0.001). This result does not necessarily translate to a problem with patient flow within radiology services because the execution of these exams could suggest a case mix marked by greater urgency or specific pathologies (such as stroke).
The function of triage is of key importance in the organization of patient treatment in the ED because it represents the first link in the care chain. The Centre Hospitalier Universitaire of Liège (Belgium) evaluated the advance nursing triage of patients presenting to their ED for thoracic pain [18]. Performing an electrocardiogram and cardiac enzymes assay at the moment of triage reduced the delay of care for these patients, as well as their length of stay in the ED. Within the framework of cooperation between healthcare professionals, specifically trained triage nurses (Infirmiers d’Accueil et d’Orientation) are authorized to proscribe radiographic exams in certain cases at the moment of triage upon patient intake to the ED. The Centre Hospitalier Universitaire de Tenon (Paris), which has studied this practice, has found a significant decrease in the waiting time for medical care, and in the delay for ED treatment more broadly, for those patients admitted to the ED with a suspicion of ankle sprain [19]. Likewise, one way of reducing crowding in the ED at Troyes hospital would therefore be an advance patient triage, with certain tasks (choice of additional exams) delegated to specifically trained triage nurses [18].
Expert opinions recommend the presence of a physician designated for intake and organization of care, or an attending emergency physician who knows ED triage and treatment protocols (a médecin d’accueil et d’organization or MAO). This provider can rapidly interpret electrocardiograms, take decisions for a patient’s initial orientation, and prioritize patient care [20]. The presence of an attending emergency physician is officially advised in structures with more than 50,000 intakes per year [21]. The fact that a day of overcrowding increases the probability that the following day will also be overcrowded is consistent with the existence of periods of overcrowding in hospitals that last more than one day. Thus, ED management can adapt hospital operations, such as by limiting scheduled hospitalizations, to make treatment beds available.
ED crowding was more observable during the week than on the weekend. Another study had shown that the percentage of ED admissions and patients in “boarding” zones was greater on Mondays and Tuesdays [22]. Scheduled care is more common in the beginning of the week in all hospital departments, which creates the risk of limiting the number of treatment beds available to hospitalize the patients who need it. The availability of beds for unscheduled sick patients is a fundamental question. One of the main causes of ED crowding is the wait for a bed in the units, called “boarding” [23]. Furthermore, general practitioners (GPs) respond more easily at the beginning of the week to their patients in emergency care, after a period of absence. These data were observed in a study conducted in the ED of Bordeaux teaching hospital, which studied the messages by GPs whose patients were being treated by the ED. About three-quarters of messages were addressed during the week, as opposed to one quarter during the weekend [24]. On the other hand, the season was only minimally associated with the phenomenon of overcrowding, which could be due to the way in which the phenomenon was defined, as a deviation from the local trend in our study.
The period after March 2020 was marked by the COVID-19 epidemic [25] and was associated with a low risk of crowding in adjusted analysis. This period, marked by a lockdown of the population (restricted movements outside the home), was known in the international literature for a lower number of admissions, but a lesser portion of patients with reduced urgency [26]. Because of a number of concurrent variables lending to a higher likelihood of crowding in the second period, and because the number of crowded days were by conception relatively constant between periods (by using the deviation from the mean of the period), the probabilities of overcrowding were naturally adjusted downward by the coefficient representing the period of the epidemic. Nevertheless, sensitivity analysis conducted only for the pre-epidemic period shows global results coherent with the principal analysis. Only the percentage of patients age ≥ 75 and the average number of radiological exams of the skull and brain region per patient above the median were no longer statistically significant in the sensitivity analysis, even though the OR values stayed relatively close to those of the principal analysis, probably because of the smaller sample size in the sensitivity analysis.
Our study presents certain limitations. First, it focused on a single center, and therefore the vast majority of patients were inhabitants of a single administrative department. Primary care data, particularly regarding primary care density and utilization, which may influence emergency department crowding, was not available. Patients aged ≥ 75 represented a smaller proportion of the case mix of the emergency department than in other French hospitals (17%) [12]. Our study took into account the diagnoses of the admitted patients, which represents an improvement over other articles on this subject. Nonetheless, only certain diagnoses could be considered because of a limitation regarding the number of variables than could be included in the model. Also, it was not possible to adjust the analyses on human resources because the human resources database had not been made available for the study. A protocol to count the number of biological tests and their impact could also be devised for further studies.
The use of a definition of overcrowding based on a deviation from the mean value of the period poses a risk in evaluating difficult or calm periods across several weeks, but that was the easiest way to avoid having the majority of overcrowded days cluster in the second part of the study, at the moment when the value of the baseline length of stay was higher than in the first period. The fact of finding significant predictors both in the principal analysis and the sensitivity analysis (capturing the period before 2020) is reassuring as regards the robustness of the results using this approach. A serious limitation also lies in the definition of the indicator itself, which does not take into account contextual or organizational factors, both of which play an important role. However, in our view, this does not diminish the relevance of such an indicator in the emergency department, where it may serve as a valuable tool for service management.
Finally, there may have been unobserved confounding factors, such as the socioeconomic status of patients, and the proportion of patients who left without being seen, which were not specifically addressed within our study. Other elements, in particular the use of comorbidity scores, could also enrich factors associated with overcrowding [27].
Conclusion
This study has made it possible to identify factors associated with ED overcrowding, including days of the weekend, the execution of radiological exams, and patient profiles (the proportions on a single day of patients with less urgent needs, admissions for trauma, or children < 7 years old being associated with the least crowding).
When the ED is overcrowded, the wait time increases for all phases of emergency services care. To reduce the length of stay and the wait times, one possible action plan is to initiate the advance ordering of additional exams.
Moreover, we could envision reinforcing the medical and paramedical teams for known periods of crowding, such as days outside of the weekend, in particular at the beginning of the week. Further research could test other ED strain components, for example overcrowding as perceived by the emergency department staff (days with and without crowding according to ED paramedical staff and physicians).
Finally, a global vision of the intake process by administrative personnel would allow the hospital to better organize downstream emergency services, creating a greater number of available beds for hospitalization. The present study has made it possible to communicate these elements to coordinating physicians, and other hospital centers could optimize their care processes by implementing a similar analysis. Implementation of this kind of analysis and its influence on average length of stay has not yet been evaluated but could be the object of a future study.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
None.
Author contributions
Conceptualization: JC, AW, SS; Data curation: AW, JC; Formal analysis: JC; Investigation: JC, MD,; Methodology: AW, SS, JC; Project administration: SS; Resources: SS; Supervision: SS; Writing – original draft: AW, AS, JC, MD, SS, ABP; Writing – review & editing: AW, AS, JC, MD, ABP, DL, SS;
Funding
None.
Data availability
All the data collected in the course of this survey are presented in the article and its supplementary material. Data are available from the corresponding author on written request.
Declarations
Ethics approval and consent to participate
According to the French legislation (articles L.1121-1 paragraph 1 and R1121-2, Public health code), the approval of the ethics committee was not needed to use anonymous data for retrospective observational study. Furthermore, same articles stated that ethic approval was not needed to access to hospital databases by medical investigators for a research purpose.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Hellmann R, Feral-Pierssens AL, Michault A, Casalino E, Ricard-Hibon A, Adnet F, et al. The analysis of the geographical distribution of emergency departments’ frequent users: a tool to prioritize public health policies? BMC Public Health. 2021;21(1):1689. 10.1186/s12889-021-11682-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chen W, Linthicum B, Argon NT, Bohrmann T, Lopiano K, Mehrotra A, et al. The effects of emergency department crowding on triage and hospital admission decisions. Am J Emerg Med. 2020;38(4):774–9. [DOI] [PubMed] [Google Scholar]
- 3.Gaieski DF, Agarwal AK, Mikkelsen ME, Drumheller B, Cham Sante S, Shofer FS, et al. The impact of ED crowding on early interventions and mortality in patients with severe sepsis. Am J Emerg Med. 2017;35(7):953–60. 10.1016/j.ajem.2017.01.061. [DOI] [PubMed] [Google Scholar]
- 4.Morley C, Unwin M, Peterson GM, Stankovich J, Kinsman L. Emergency department crowding: A systematic review of causes, consequences and solutions. Pone. 2018;13(8):0203316. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Estey A, Ness K, Saunders LD, Alibhai A, Bear RA. Understanding the causes of overcrowding in emergency departments in the capital health region in alberta: a focus group study. CJEM. 2003;5(2):87–94. 10.1017/s1481803500008216. [DOI] [PubMed] [Google Scholar]
- 6.Derlet RW, Richards JR. Emergency department overcrowding in Florida, new York, and Texas. South Med J. 2002;95(8):846–9. [PubMed] [Google Scholar]
- 7.Bond K, Ospina MB, Blitz S, Afilalo M, Campbell SG, Bullard M, et al. Frequency, determinants and impact of overcrowding in emergency departments in canada: a National survey. Healthc Q. 2007;10(4):32–40. 10.12927/hcq.2007.19312. [DOI] [PubMed] [Google Scholar]
- 8.Van Der Linden MC, Khursheed M, Hooda K, Pines JM, Van Der Linden N. Two emergency departments, 6000km apart: differences in patient flow and staff perceptions about crowding. Int Emerg Nurs. 2017;35:30–6. 10.1016/j.ienj.2017.06.002. [DOI] [PubMed] [Google Scholar]
- 9.Knapman M, Bonner A. Overcrowding in medium-volume emergency departments: effects of aged patients in emergency departments on wait times for non-emergent triage-level patients. Int J Nurs Pract. 2010;16(3):310–7. 10.1111/j.1440-172X.2010.01846.x. [DOI] [PubMed] [Google Scholar]
- 10.Aboagye-Sarfo P, Mai Q, Sanfilippo FM, Preen DB, Stewart LM, Fatovich DM. Growth in Western Australian emergency department demand during 2007–2013 is due to people with urgent and complex care needs. Emerg Med Australas. 2015;27(3):202–9. 10.1111/1742-6723.12396. [DOI] [PubMed] [Google Scholar]
- 11.Kawano T, Nishiyama K, Anan H, Tujimura Y. Direct relationship between aging and overcrowding in the ED, and a calculation formula for demand projection: a cross-sectional study. Emerg Med J. 2014;31(1):19–23. 10.1136/emermed-2012-202050. [DOI] [PubMed] [Google Scholar]
- 12.Naouri D, El Khoury C, Vincent-Cassy C, Vuagnat A, Schmidt J, Yordanov Y. The French emergency national survey: a description of emergency departments and patients in France. Plos One [Internet]. 2018;13(6). Available from: https://journals.plos.org/plosone/article10.1371/journal.pone.0198474. [DOI] [PMC free article] [PubMed]
- 13.Afilal M, Yalaoui F, Dugardin F, Amodeo L, Laplanche D, Blua P. Forecasting the emergency department patients flow. J Med Syst. 2016;40(7):175. 10.1007/s10916-016-0527-0. [DOI] [PubMed] [Google Scholar]
- 14.Wartelle A, Mourad-Chehade F, Yalaoui F, Chrusciel J, Laplanche D, Sanchez S. Clustering of a health dataset using diagnosis co-occurrences. Appl Sci. 2021;11(5):2373. 10.3390/app11052373.
- 15.Ansah JP, Chiu CT. Projecting the chronic disease burden among the adult population in the united States using a multi-state population model. Front Public Health. 2022;10:1082183. 10.3389/fpubh.2022.1082183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Savioli G, Ceresa IF, Gri N, Piccini GB, Longhitano Y, Zanza C, et al. Emergency department overcrowding: Understanding the factors to find corresponding solutions. J Pers Med. 2022;12(2):279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Haimovich AD, Shah MN, Southerland LT, Hwang U, Patterson BW. Automating risk stratification for geriatric syndromes in the emergency department. J Am Geriatr Soc. 2024;72(1):258–67. 10.1111/jgs.18594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Jobé J, Ghuysen A, D’Orio V. Advanced nurse triage for emergency department. Rev Med Liege. 2018;73(5–6):229–36. [PubMed] [Google Scholar]
- 19.Valentian M, Mewasing BI, Burgraff E, Mintandjian A, Lefranc V, Verbbrugghe R, et al. Intérêt d’une Demande anticipée de radiographies par l’infirmière organisatrice de l’accueil, Dans les traumatismes de Cheville. Ann Fr Med Urgence. 2016;7(2):77–84. [Google Scholar]
- 20.Choquet CM-AC. Pascal Gillet. Recommandations formalisées d’experts - Triage 2013 [Internet]. [cited 2022 Nov 17]. Available from: https://www.sfmu.org/upload/referentielsSFMU/rfe_triage2013.pdf
- 21.Gilles Vuides Stéphane, Bourgeois, Frédéric L. Panorama. 2017, Organisation des services d’urgence. [Internet]. 2014 [cited 2023 May 1]. Available from: https://ies-sud.fr/wp-content/uploads/2019/03/Panorama_2017_CTRU_WEB.pdf
- 22.Asaro PV, Lewis LM, Boxerman SB. Emergency department overcrowding: analysis of the factors of renege rate. AEM. 2008;14(2):157–62. [DOI] [PubMed] [Google Scholar]
- 23.Oberlin M, Andrès E, Behr M, Kepka S, Le borgne P, Bilbault P. Emergency overcrowding and hospital organization: causes and solutions. Rev Med Interne. 2020;41(10):693–9. [DOI] [PubMed] [Google Scholar]
- 24.Lemonnier N. Devenir des patients adressés par un médecin avec courrier médical, aux urgences adultes du CHU Pellegrin, à Bordeaux [Internet] [Thèse d’exercice]. [Bordeaux, France]: Université de Bordeaux. Faculté de Médecine; 2016. Available from: https://dumas.ccsd.cnrs.fr/dumas-01420423/document
- 25.Guan WJ, Ni ZY, Hu Y, Liang WH, Ou CQ, He JX, et al. Clinical characteristics of coronavirus disease 2019 in China. N Engl J Med. 2020;382(18):1708–20. 10.1056/NEJMoa2002032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ghaderi H, Stowell JR, Akhter M, Norquist C, Pugsley P, Subbian V. Impact of COVID-19 Pandemic on emergency department visits: a regional case study of informatics challenges and opportunities. AMIA Annu Symp Proc. 2022;496–505. [PMC free article] [PubMed]
- 27.Kuang A, Xu C, Southern DA, Sandhu N, Quan H. Validated administrative data based ICD-10 algorithms for chronic conditions: A systematic review. J Epidemiol Popul Health. 2024;72(4):202744. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Citations
- Wartelle A, Mourad-Chehade F, Yalaoui F, Chrusciel J, Laplanche D, Sanchez S. Clustering of a health dataset using diagnosis co-occurrences. Appl Sci. 2021;11(5):2373. 10.3390/app11052373.
Supplementary Materials
Data Availability Statement
All the data collected in the course of this survey are presented in the article and its supplementary material. Data are available from the corresponding author on written request.