Abstract
Background
Prolonged emergency department length of stay (EDLOS) is a worldwide issue associated with increased mortality, decreased patient satisfaction and poor quality of care. The factors influencing EDLOS have not been comprehensively studied in the context of Swedish EDs. This study’s objective is to determine the input-, throughput- and output factors associated with EDLOS, at two urban EDs in Sweden.
Methods
Data was collected from two hospitals. All patient visits during the two-year study period were included. Patients who left without being seen by a physician were excluded. The explanatory factors included patient characteristics, medical data, and hospital bed occupancy data. Multi-variable linear regression analysis was used to test the associations between the factors and EDLOS.
Results
The top contributors to prolonged EDLOS were diagnostic imaging, which added between 64 and 149 min of EDLOS, diagnostic testing at central laboratory (53–99 min), followed by intra-ED zone transfer (46–94 min). Arriving during crowding or being admitted during high hospital bed occupancy had a significant but relatively small absolute effect on the outcome.
Conclusions
Throughput factors had far greater impact on EDLOS than both input- and output factors. Adapting strategies to the structural and procedural characteristics of each setting may enhance the effectiveness of improvement efforts.
Clinical trial number
Not applicable.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12873-025-01283-z.
Keywords: Emergency department, Length of stay, Explanatory factors, Patient flow
Background
Emergency Department (ED) Length of Stay (LOS) is the amount of time patients spend in the ED from arrival to discharge or admission [1]. Prolonged EDLOS is associated with increased mortality [2], decreased patient satisfaction [3] and poor quality of care [4]. This complex problem has been longstanding [5], worsening [6] and worldwide [7]. Different health jurisdictions have implemented national time-targets [8] and other clinical interventions [9], but improvements in wait times have been variable and unsustained [10].
The reason for prolonged EDLOS is complex and influenced by multiple factors [11], some of which are from processes in the ED, and some, such as boarding, are hospital and system issues [12]. A convenient way of organizing factors associated with EDLOS, is to apply Asplin’s input-throughput-output model [13]. This model was originally designed to identify components of the health care system that contribute to ED crowding, but the same components have been used to study patient flow and EDLOS [14].
Until 2017, Swedish EDs followed the international trend of increasing ED visits and longer LOS. However, since 2017, according to the Swedish Board of Health and Welfare [15], ED visits have decreased, but EDLOS has continued to increase [16]. A recent study based on data from Swedish ED visits in 2015–2019, suggests that patients admitted to in-hospital care may be more vulnerable to the effects of crowding. It also concluded that these effects are more prominent in counties with high levels of hospital bed occupancy [17]. Another Swedish study revealed that admitted patients spend more time in the ED, than patients discharged home, suggesting an output-related bottleneck to ED patient flow [18]. However, there are no Swedish studies investigating the impact of investigations, ex. laboratory testing and diagnostic imaging [19] or hospital occupancy [20]. A more comprehensive study of the factors associated with EDLOS is needed.
This study’s objective is to determine the input-, throughput- and output factors associated with EDLOS, at two urban EDs in Sweden.
Methods
This study was designed as a cross-sectional observational study using data from electronic patient records. Since only routinely collected data was used, informed consent was waived. The study was approved by the Swedish Ethical Review Authority (DNR-2020-01614).
Study setting
The study was conducted in two EDs, located in two separate administrative regions in central Sweden. Örebro University Hospital (OUH) has a catchment area of approximately 200,000 people and Mälarsjukhuset in Eskilstuna (MSE) has 120,000 people [21], but both EDs receive 50–65 000 visits annually [16]. Both hospitals were organized in a way that has been described in the literature as a “Hybrid model” [22], which refers to a dual authority, shared between the ED and other departments. Both OUH and MSE have 24-hour general surgery, anesthesiology, obstetrics, and pediatric coverage. OUH is a level 1 trauma center with an average of 402 beds during the study period, while MSE had 240 beds. Both EDs are organized with low proportions of physicians specialized in emergency medicine (EM), a newly established specialty in Sweden. A pre-triage nurse assesses patients and assigns them to the most appropriate medical specialty, based on the chief complaint. Within the ED patients are managed in separate zones, based on the specialty assignment made by the pre-triage nurse. Certain patient groups are redirected entirely from the ED and managed elsewhere. For example, psychiatric patients were not assessed in either of the studied EDs. For details on the case-mix and specific patient flow strategies for each hospital, see Appendix A. Both EDs used the Rapid Emergency Triage and Treatment System (RETTS) [23] as their main triage assessment tool.
Study material
Two cohorts consisting of data extracted from consecutive electronic medical records of patients who visited the two EDs from January 1, 2018 to December 31, 2019 were included. The ED visit was defined as one patient’s continuous stay in the ED. The records of all ED patients, irrespective of age, were included. Observations of patients who left the ED without being seen (LWBS) were excluded.
Data collection
The data sources included the ED patient registry, electronic health care records including laboratory and radiology records, and the Swedish population registry [24]. Demographics and ED process data, such as arrival time, were collected from the respective ED registries, but medical interventions, such as diagnostic imaging and laboratory tests, were retrieved separately from the respective hospital department data holdings and merged into the main dataset through time-stamps and personal ID numbers [24].
Dependent variable
The primary dependent variable was EDLOS measured in minutes, defined as the time between patient arrival and patient departure, either by hospital admission or discharge home.
Independent variables
An outline of the independent variables collected and used in the regression analysis is provided below in the input-throughput-output model (Fig. 1). A detailed description of the independent variables is presented in Appendix B.
Fig. 1.
Patient flow in relation to Asplin’s model of emergency department crowding [13] and collected variables for the study
Input factors
Patient demographics included sex (male/female) and age. Using the National Library of Medicine method of age grouping, age was categorized into children (0–17), adults (18–64), older adults (65–79) and elderly (80+) [25]. Other input factors collected were referral, mode of arrival and arrival time. Referral meant that upon arrival, the patient presented a written referral from a general practitioner or equivalent. Mode of arrival was either ambulance or walk-in. Time of arrival was categorized both by day of week and hour of day. After univariate regression analysis day of week was categorized into Weekend (Saturday and Sunday), Monday and Other day. Hour of day was categorized to correspond with personnel shift hours, which meant 8 AM − 4:59 PM, 5 PM − 9:59 PM and 10 PM − 7:59 AM.
Throughput factors
ED crowding was measured using a model previously used in studies of ED crowding, conducted in settings similar to those used in the current study [17, 26]. The crowding exposure was defined as the mean ED census during the shift that the exposed patient arrived, divided by the expected census for the same shift. The expected census at each ED was estimated using a linear regression model with year, weekday/weekend, and arrival hour as predictors. The crowding exposure was categorized into three categories: low (0–75% of observations), moderate (76–95% of observations) and high (highest 5% of observations). Triage level was categorized into the five levels used by the RETTS acuity assessment [23] performed by the triage nurse. Time-to-physician was defined as time elapsed from arrival to first recorded contact with a physician.
Laboratory work was categorized to reflect different levels of analysis complexity. The specific analyses were chosen based on clinical experience as well as on the results of previous research [19]. The categories were: no labs; labs excluding troponins and D-dimer, meaning any sample of blood, spinal fluid or tissue collected and sent to the hospital’s central laboratory, but excluding observations where troponins or D-dimer was analyzed; labs including troponins, not D-dimer, were blood samples sent to central laboratory for troponin-I levels; labs including D-dimer, not troponins, were blood samples sent to central laboratory for D-dimer levels; labs including D-dimer and troponins; and point-of-care testing only, were samples of blood or urine that were analyzed only in the ED, without being sent to the central laboratory. Only recorded analyses performed during the ED stay were included.
Medical imaging was divided into: X-ray (plain imaging excluding interventional radiology, fluoroscopy, and angiography), CT (computed tomography) and ultrasound (sonographic non-interventional imaging, not bedside). Other examinations included magnetic resonance imaging, interventional radiology and other, to the ED context, rare types of examinations. Only examinations done within the ED stay were included.
Intra-ED zone transfer occurs when a patient is reassigned to a different specialty after being assessed by the index specialty assigned at pre-triage — e.g., when a patient presenting with abdominal pain is initially assigned to the surgical zone but transferred to the medical zone due to abnormal ECG findings. This transfer is not a physical transfer of the actual patient, but rather an administrative transfer and an indication that another medical specialty is medically responsible, while the patient remains in the ED.
Output factors
Patient disposition was categorized into discharged or admitted during low/moderate/high hospital bed occupancy. Hospital bed occupancy was categorized as low (< 85%), moderate (85–95%) or high (> 95%). The Swedish National Board of Health and Welfare’s recommendation of 85% bed occupancy [27], and two other studies were used to determine the cut-off points. The two studies found that hospital bed occupancy above 95% was associated with increased rates of 30-day unplanned readmissions [28] and decreased rates of in-hospital admissions from the ED, suggesting that shortage of hospital beds could influence the ED physician’s decision to admit or discharge [29].
Data cleaning and imputation
Data was assessed for quality, duplication and missing data. Duplicates and observations with poor data quality, such as negative duration for EDLOS or corrupted personal ID number, were removed. We considered EDLOS exceeding 48 h to be highly unlikely, thus indicative of data entry errors. Therefore, observations with EDLOS above 2880 min were considered outliers and removed from the data set. The cut-off was chosen based on clinical plausibility and supported by the empirical distribution of the data, where only a handful of observations exceeded this threshold. Because an intra-departmental transfer for the same patient visit generated a new encounter in the administrative database, these observations were merged into a single encounter. Multivariate imputation by chained equations was performed to estimate values for any variable with more than 5% missing values.
Statistical analysis
Descriptive statistics were used for Table 1. A p-value less than 0.01 was considered to indicate statistical significance. For the primary outcome, a multivariable linear regression analysis was performed, using stepwise backward elimination technique to exclude nonsignificant covariates. The main analyses were performed for each hospital separately. An assessment of potential multi-collinearity was done using a Variance Inflation Factor (VIF) test (see Appendix C). Data was analyzed using Stata version 18 [30].
Table 1.
Demographics: ED patient visits from 1 january, 2018 through 31 december, 2019, by study site
| Örebro university hospital (OUH) | Mälarsjukhuset Eskilstuna (MSE) | |||
|---|---|---|---|---|
| Patient visits, n | 123,730 | 98,317 | ||
| Age, mean (median) | 41.9 (41.1) | 45.5 (46.5) | ||
| 0–17, n (%) | 32,588 (26.3) | 20,201 (20.5) | ||
| 18–64, n (%) | 56,337 (45.5) | 47,966 (48.8) | ||
| 65–79, n (%) | 21,550 (17.4) | 18,607 (18.9) | ||
| 80+, n (%) | 13,255 (10.7) | 11,553 (11.7) | ||
| Female sex, n (%) | 60,874 (49.1) | 49,857 (50.7) | ||
| Arrival day, n (%) | ||||
| Monday | 18,760 (15.2) | 15,568 (15.8) | ||
| Tuesday-Friday | 69,694 (56.3) | 56,000 (57.0) | ||
| Weekend | 35,276 (28.5) | 26,759 (27.2) | ||
| Arrival time of day, n (%) | ||||
| 8:00 AM-4:59 PM | 65,368 (52.8) | 56,230 (57.1) | ||
| 5:00 PM-9:59 PM | 34,332 (27.7) | 24,937 (25.3) | ||
| 10:00 PM-7:59 AM | 24,030 (19.4) | 17,232 (17.5) | ||
| Arrival by ambulance | 24,792 (20.0) | 19,203 (19.5) | ||
| Referral a, n (%) | 9,991 (8.1) | 1,982 (2.0) | ||
| Top reasons for visit, n (%) | Abdominal pain 18,920 (16.3) | Abdominal pain 12,550 (14.6) | ||
| Dyspnea 8,575 (7.4) | Injury of wrist and hand 6,539 (7.6) | |||
| Chest pain 8,474 (7.3) | Dyspnea 6,503 (7.6) | |||
| Fever 7,118 (6.2) | Chest pain 5,669 (6.6) | |||
| Injury of head 5,703 (4.9) | Injury of head 4,115 (4.8) | |||
| Injury of wrist and hand 5,415 (4.7) | Injury of ankle and foot 3,608 (4.2) | |||
| Unspecified reason 4,314 (3.7) | Pain in limb 3,108 (3.6) | |||
| Infectious disease 3,042 (2.6) | Fever 3,004 (3.5) | |||
| Pain in limb 2,823 (2.4) | Dizziness and giddiness 2,331 (2.7) | |||
| Dizziness and giddiness 2,790 (2.4) | Injury of lower leg 2,134 (2.5) | |||
| Crowding, observed/expected census b | Observed mean (SD) | Expected mean (SD) | Observed mean (SD) | Expected mean (SD) |
| Low 0–75% c | 29.2 (11.4) | 30.5 (10.0) | 24.9 (9.9) | 26.4 (8.6) |
| Moderate 76–94% c | 37.5 (14.9) | 27.0 (10.8) | 32.9 (12.9) | 23.5 (9.2) |
| High 95–100% c | 29.3 (15.4) | 16.0 (9.1) | 26.1 (13.9) | 14.2 (8.3) |
| Triage level, n (%) | ||||
| Red | 5,202 (4.2) | 2,418 (2.5) | ||
| Orange | 21,787 (17.6) | 18,496 (18.8) | ||
| Yellow | 60,477 (48.9) | 46,377 (47.1) | ||
| Green | 23,659 (19.1) | 19,269 (19.6) | ||
| Blue | 12,603 (10.2) | 11,767 (12.0) | ||
| Median time-to-physician (IQR) | 62 min (31–121) | 59 min (26–119) | ||
| Mean time-to-physician (SD) | 89.2 min (82.3) | 88.3 min (90.2) | ||
| Laboratory work, n (%) | ||||
| No laboratory work | 53,997 (43.6) | 41,011 (41.7) | ||
| Labs, excl. Troponins and D-dimer | 41,346 (33.4) | 41,190 (41.9) | ||
| Labs, incl. Troponins not D-dimer | 20,831 (16.8) | 7,688 (7.8) | ||
| Labs, incl. D-Dimer not Troponins | 657 (0.5) | 1,265 (1.3) | ||
| Labs incl. Troponins & D-Dimer | 1,176 (1.0) | 1,128 (1.2) | ||
| POC d analysis only | 5,723 (4.6) | 6,045 (6.2) | ||
| Diagnostic imaging, n (%) | ||||
| No diagnostic imaging | 92,192 (74.5) | 60,328 (61.4) | ||
| X-ray | 11,413 (9.2) | 20,766 (21.1) | ||
| Ultrasound | 2,223 (1.8) | 2,691 (2.7) | ||
| Computed tomography | 17,654 (14.3) | 14,542 (14.8) | ||
| Other examination e | 248 (0.2) | 0 (0) | ||
| Intra-ED zone transfer, n (%) | 4,574 (3.7) | 3,508 (3.6) | ||
| Patient disposition, n (%) | ||||
| Discharged home | 90,479 (73.1) | 73,384 (74.6) | ||
| Admitted to ward during: | ||||
| Low bed occupancy (< 85%) | 7,440 (6.0) | 2,487 (2.8) | ||
| Moderate bed occupancy (85–95%) | 19,614 (15.9) | 12,038 (12.2) | ||
| High bed occupancy (> 95%) | 6,197 (5.0) | 10,490 (10.7) | ||
| Hospital bed occupancy during study period, mean (SD) | 89.4% (0.08) | 93.3% (0.07) | ||
| Patient outcomes | ||||
| Median EDLOS (IQR) | 190 min (118–287) | 203 min (125–309) | ||
| Mean EDLOS (SD) | 215.2 min (132.5) | 231.5 min (147.5) | ||
| EDLOS > 4 h, n (%) | 43,886 (35.5) | 39,084 (39.8) | ||
| LWBSf, n (%) | 1,593 (1.3) | 3,168 (3.12) | ||
| Deceased in ED, n (%) | 118 (0.1) | 92 (0.1) | ||
| 24-hour mortality, n (%) | 348 (0.3) | 293 (0.3) | ||
a = Patient presented a referral from a general practitioner (or equivalent) upon arrival
b = Observed mean census by shift– divided by expected mean census of the same shift, approximated by linear regression
c = Crowding exposure categories based on the distribution of the quotient derived by the observed/expected census. Low = bottom 75% of observations, Moderate = 76–94% of observations, High = top 95% of observations
d = Point-of-care
e = Advanced diagnostic imaging excl. CT– e.g. Magnetic resonance imaging (MRI) or fluoroscopy
f = Left the ED without being seen by a physician
Results
After excluding duplicates and LWBS, 222,047 patient visits were included: 123,730 at OUH and 98,317 at MSE (Fig. 2). There were missing values for time-to-physician and triage level, 11.9% and 5.6% respectively. Missing values were estimated by imputation.
Fig. 2.
- Overview of data flow
The median EDLOS for OUH and MSE were 190 and 203 min respectively. At OUH, the patients’ mean age was lower (41.1 years compared to 46.5 at MSE) and the proportion of pediatric patients (age < 18) was greater (26.3% compared to 20.5% at MSE). Abdominal pain was the most common reason for visit at both study sites. At MSE the mean hospital bed occupancy was higher (93.3% compared to 89.4%) and a higher proportion of patients were admitted during high bed occupancy (10.7 at MSE compared to 5.0% at OUH) (Table 1).
The linear regression analyses (Table 2) revealed that at OUH, all variables had a significant effect on the dependent variable EDLOS in minutes. At MSE, there were nonsignificant associations for arrival day of week and for being admitted during low hospital occupancy compared to being discharged home. The overall model fit was statistically significant (P < 0.01) with R2-estimates of 0.60 and 0.55 at OUH and MSE respectively.
Table 2.
Linear regression analysis. Estimated coefficients and 95% confidence intervals (CIs) for associations with EDLOS
| Independent variable | Coefficient (95% CI) | |
|---|---|---|
| Örebro University Hospital | Mälarsjukhuset Eskilstuna | |
| Age 0–17 | 11.8 (10.6–13.0) * | -8.3 (-10.1– -6.6) * |
| 18–64 (reference) | 0 | 0 |
| 65–79 | 7.6 (6.3–9.0) * | 10.3 (8.6–12.1) * |
| 80+ | 14.7 (13.1–16.4) * | 11.3 (9.2–13.5) * |
| Sex, female | 3.0 (2.1–3.9) * | 2.7 (1.4–3.9) * |
| Arrival day | ||
| Monday | 2.0 (0.7–3.3) * | 0.5 (-1.3–2.2) |
| Tuesday-Friday (ref.) | 0 | 0 |
| Weekend | -5.2 (-6.3– -4.2) * | 0.8 (-0.6–2.3) |
| Time of day | ||
| 8:00 AM-4:59 PM (ref.) | 0 | 0 |
| 5:00 PM-9:59 PM | -12.6 (-13.7– -11.5) * | -6.8 (-8.3– -5.3) * |
| 10:00 PM-7:59 AM | -13.8 (-15.1– -12.6) * | -10.2 (-11.9– -8.5) * |
| Arrival by ambulance | 19.3 (17.9–20.7) * | 19.8 (18.0–21.6) * |
| Referral | 2.9 (1.1–4.6) * | 5.7 (1.3–10.1) |
| Crowding | ||
| Low 0–75% (ref.) | 0 | 0 |
| Moderate 76–94% | 5.4 (4.2–6.6) * | 9.6 (8.0–11.1) * |
| High 95–100% | 9.8 (7.6–11.9) * | 11.5 (8.8–14.3) * |
| Triage level | ||
| Red (ref.) | 0 | 0 |
| Orange | 52.1 (49.6–54.7) * | 64.2 (60.0–68.4) * |
| Yellow | 38.6 (36.1–41.2) * | 65.8 (61.5–70.0) * |
| Green | 30.0 (27.2–32.7) * | 57.7 (53.3–62.2) * |
| Blue | 19.0 (16.0–22.1) * | 52.6 (47.8–57.3) * |
| Time-to-Physician a | 0.9 (0.9, 0.9) * | 0.9 (0.9, 0.9) * |
| Laboratory analysis | ||
| No labs done (ref.) | 0 | 0 |
| Labs, excluding Troponins and D-dimer | 61.1 (59.9–62.4) * | 59.6 (58.0–61.3) * |
| Labs, incl. Troponins not D-dimer | 52.8 (51.3–54.4) * | 61.8 (59.2–64.5) * |
| Labs, incl. D-Dimer not Troponins | 83.3 (76.9–89.7) * | 68.3 (62.7–74.0) * |
| Labs incl. Troponins & D-Dimer | 98.3 (93.5–103.1) * | 98.7 (92.8–104.6) * |
| POC b analysis only | 38.8 (36.5–41.1) * | 26.0 (23.3–28.8) * |
| Diagnostic imaging | ||
| No radiology (ref.) | 0 | 0 |
| X-ray | 67.7 (66.0–69.4) * | 63.8 (62.1–65.5) * |
| Ultrasound | 148.5 (145.0–152.0) * | 121.2 (117.4–125.1) * |
| Computed tomography | 131.7 (130.3–133.1) * | 125.3 (123.5–127.1) * |
| Other examinationc | 143.1 (132.8–153.4) * | -c |
| Intra-ED zone transfer | 93.5 (90.9–96.1) * | 46.4 (43.0–49.8) * |
| Disposition | ||
| Discharged (ref.) | 0 | 0 |
| Admitted, < 85% hospital bed occupancy | 3.5 (1.5–5.6) * | 0.7 (-3.4–4.7) |
| Admitted, 85–95% bed occupancy | 8.0 (6.5–9.4) * | 4.5 (2.4–6.7) * |
| Admitted, > 95% bed occupancy | 7.4 (5.2–9.7) * | 16.5 (14.3–18.8) * |
| Intercept | 28.2 (25.3–31.0) * | 15.4 (10.9–20.0) * |
* = Statistically significant association with the dependent variable, p-value < 0.01
a = Continuous variable, unit = 1 min increments
b= Point-of-care
c= No “Other examinations” (i.e. MRI, fluoroscopy angiography etc.) were carried out during ED stay at MSE, the variable was omitted from the model
Arrival by ambulance was the largest contributor to increased EDLOS among the input factors at both study sites, and was associated with an increase of 19 min. Patients aged 65 and above were more likely to have an increased EDLOS at both study sites, compared to patients aged 18 to 64. Patient age < 18 was associated with an increased EDLOS (12 min) at OUH, but a decreased EDLOS (-8 min) at MSE. Arrival time during the evening- or night shift decreased the estimated EDLOS at both study sites.
For the throughput factors, diagnostic imaging was the greatest contributor to the estimated EDLOS at both study sites with non-bedside ultrasound adding 148 min at OUH and CT adding 125 min at MSE. Laboratory analysis including troponins and D-dimer increased EDLOS with 98 and 99 min at OUH and MSE respectively. Patients whose laboratory work included only POC-analysis had an estimated increase in EDLOS of 39 and 26 min at OUH and MSE respectively, compared to patients who had no laboratory work performed. Intra-ED zone transfer added 94 min to the estimated EDLOS at OUH and 46 min at MSE. Moderate (76–95% of all observations) ED crowding increased the estimated EDLOS with 5 and 10 min at OUH and MSE respectively.
For the output factors, being admitted to in-hospital care increased the estimated EDLOS at OUH by 4, 8 and 7 min, during low (< 85%), moderate (85–95%) and high (> 95%) hospital bed occupancy respectively. At MSE there was no significant effect of being admitted during low hospital bed occupancy compared to being discharged from the ED. At moderate and high bed occupancies, the estimated increases in EDLOS were 5 and 17 min respectively.
Discussion
In this study of factors associated with EDLOS at two hospitals in Sweden, we found that the factors with the greatest impact were throughput related. Diagnostic imaging and laboratory testing surpassed the impact of input and output factors by a large margin. Intra-ED zone transfer was associated with significant and substantial increases in EDLOS estimated at 94 and 46 min at OUH and MSE respectively. This effect is, to our knowledge, the first recorded estimation of the impact of a Hybrid model of organizing emergency departments [22] and both the magnitude of the effect, as well as the difference between study sites, highlights the importance of ED organization regarding authority and patient responsibility. The effect of hospital bed occupancy on EDLOS was more prominent in the hospital with fewer hospital beds and higher mean bed occupancy. While there were similarities between the study sites, our study also revealed important differences in the effects of several factors, which highlights the uniqueness of each ED.
At both study sites, older age was associated with an increased EDLOS, compared to patients aged 18 to 64. Similar results have been seen in previous studies and can most likely be attributed to higher patient complexity [18, 31]. Female sex was also associated with a small but significant increase in EDLOS, which is not as easily explained by complexity and not as frequently reported. This association has, however, been described previously [18], but the underlying cause remains to be investigated. Young age (< 18 years) was associated with increased EDLOS at OUH, but with decreased EDLOS at MSE. These contrasting findings may reflect the organizational differences between the study sites. Specifically, MSE lacked a designated area tailored for children, which may have contributed to a tendency to expedite the disposition of pediatric patients [32]. Alternatively, the finding is reflective of a cultural difference of the study sites. OUH, seeing more pediatric patients, might be more prone to time-consuming practices of alleviating pain and distress associated with common procedures [33].
Laboratory analysis and diagnostic imaging were associated with significant increases in EDLOS, ranging between one and two hours of added time across both study sites. These results are in line with previous research and therefore not surprising [20]. Laboratory testing, depending on complexity, was associated with additional EDLOS between 55 and 93 min, which can be compared to an overall effect of 72 (95% CI: 66–78) minutes for blood testing, estimated in a large nationwide study in United States [34]. Advanced imaging (ADI), including CT and MRI, was estimated to increase EDLOS by 121–143 min, which can be compared to the results of another study using the United States National Hospital Ambulatory Medical Care Survey, who found ADI to be associated with an increased EDLOS ranging between 47 and 123 min, depending on patient symptoms. For abdominal pain, the most common complaint in our study, the 95% CI was 97–141 min [35]. The amount of additional time associated with diagnostic testing and imaging, suggests that alternative methods, such as point-of-care (POC) ultrasound [36] and POC-testing [37] should be given careful consideration in clinical practice. Our results should also be interpreted in the light of several studies claiming that a substantial amount of diagnostic imaging is deemed inappropriate [38–40].
There was a moderate overall effect of ED crowding. A high level of crowding was associated with an increased EDLOS of 10 and 11 min, at OUH and MSE respectively. However, this effect was observed only in the top 5th percentile of observations. For the majority of the ED population at OUH, crowding was associated with 5 min of additional EDLOS or less. These results contrast with previous research [18, 41], where crowding has been identified as a main contributor to prolonged EDLOS globally. In another study conducted in Sweden, crowding was associated with increases in EDLOS ranging between 45 and 117 min, depending on chief complaint [42]. One possible explanation for this difference, is that none of the abovementioned studies has taken the possible effect of diagnostic testing and imaging into account. It is also plausible to assume that ED crowding has greater effect on staff workload [43] and patient safety [44], than it has on the length of stay for individual patients.
At both of the study sites, the output factors were substantially smaller than most throughput factors, which contradicts the results from a large overview in North America [45] as well as the conclusions of an international systematic review [7]. This can partly be explained by the relatively moderate rates of bed occupancy during the study period. It is, however, important to note that the data for this study were collected before the COVID-19 pandemic. Since then, the total number of hospital beds in Sweden has decreased [46], and the number of patients admitted to wards already at maximum capacity has increased [47]. It is also noteworthy that the median EDLOS for both EDs in this study was approximately three hours. This was somewhat below the national average for Swedish EDs and significantly below the median EDLOS in more densely populated regions such as Stockholm and Gothenburg [15].
Strengths and limitations
A strength of this study was the use of robust data from multiple sources and a large sample size. There were missing data that required imputation; however, the absence of significant changes following imputation implied that the data were robust. Although this study benefits from robust data and a large sample size, it has several limitations. The generalizability of our findings may be limited by the specific settings in which the study was conducted. Also, the use of EDLOS in its continuous form as the dependent variable challenges the assumption of normality for linear regression with ordinary least squares (OLS). However, the standardized residuals of the predicted values were found to be normally distributed (see Appendix C) and even though a generalized linear model (GLM) might have strengthened the statistical integrity, a previous study using similar data has shown very similar results when comparing OLS and GLM [35]. The intuitive interpretation of estimated coefficients offered by OLS linear regression was ultimately decisive for the choice of statistical analysis.
Additionally, our study used the bed occupancy rate as a proxy for access block. Ideally, a direct measurement of the time patients spent waiting for a hospital bed would have been preferable. Unfortunately, the absence of data on bed-request time constrains our ability to provide more precise measurements in this regard. Another limitation is that a lot of time has passed between data collection and publication, which means that some of the explanatory factors may not still be applicable. One such factor is the hospital bed occupancy rate, which is likely to have increased substantially since the time of the data collection [46]. In addition, the choice of proxy measure for crowding can be seen as a limitation. Using a simpler and widely acknowledged measure like the ED bed occupancy rate [48], might have increased the interpretability of the results. However, it was decided, based on our own knowledge of the study sites and previous research at similar settings, to use a measure with better potential for capturing the effect of unexpected surges in patient flow [17, 26].
Clinical implications
The study results can be used by ED management and health care policy makers, aiding in the prioritization of interventions aimed at reducing EDLOS. At OUH, the difference between being admitted during low and high hospital occupancy was less than 4 min, meaning that the effect of access block on EDLOS was very weak. Therefore, an attempt to increase the number of hospital beds might prove to be an expensive and inefficient intervention, if reduced EDLOS is the aim. Additionally, the large effect of complex laboratory testing and advanced diagnostic imaging supports the development of clinical pathways. Either with the aim to delay or refrain from a test– or to increase the use of alternative diagnostic methods, such as POC-sonography and -testing.
The large difference between the effects of intra-ED zone transfer observed at the two study sites suggests that the procedure of reassigning patients to other specialties has potential to be made more time efficient. Moreover, we suggest that each ED use its operational data to identify its unique bottlenecks to ED flow.
Conclusions
In conclusion, this study highlights the substantial impact of throughput factors - particularly diagnostic imaging and intra-ED zone transfers - on emergency department length of stay (EDLOS). The findings underscore the importance of context-sensitive interventions that address operational inefficiencies contributing to prolonged EDLOS. Adapting strategies to the structural and procedural characteristics of each setting may enhance the effectiveness of improvement efforts.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgements
We would like to show our gratitude to Mr. Arne Karlsson of OUH for assisting with data collection and to Dr. Ruzan Udumyan at Örebro university for advice regarding the statistical analysis and missing data handling.
Abbreviations
- ADI
Advanced diagnostic imaging
- CI
Confidence interval
- ECG
Electrocardiogram
- ED
Emergency department
- EM
Emergency medicine
- EDLOS
Emergency department length of stay
- GLM
Generalized linear model
- IQR
Interquartile range
- LOS
Length of stay
- LWBS
Left without being seen (by a physician)
- MSE
Mälarsjukhuset Eskilstuna (Hospital)
- OLS
Ordinary least squares
- OUH
Örebro university hospital
- POC
Point-of-care
- RETTS
Rapid Emergency Triage and Treatment System
- SD
Standard deviation
- VIF
Variance inflation factor
Author contributions
Conceptualization: JA, IC and LK. Methodology and formal analysis: JA and IC. Writing– original draft: JA. Writing– Review and Editing: LK, IC, LN and AG.
Funding
Open access funding provided by Örebro University. Preparation of study, collection of data, and writing of manuscript carried out by authors under terms of employment the individual had with their respective affiliations. The authors received no external funding.
Data availability
Unfortunately, the data set is not permitted to be available for open access. Data used for this study is sensitive patient data from Örebro University Hospital and Mälarsjukhuset Eskilstuna. These data are not publicly available due to restrictions governed by the Swedish Ethical Review Authority, and Swedish laws in relation to research in, and distribution of health care data. However, Örebro University researchers are mandated to archive research data for 10 years if an audit is required. To gain access to the data, please contact the Swedish Ethical Review Authority and cite registration number: 2020-01614.
Declarations
Ethics approval and consent to participate
This study was conducted in accordance with the ethical standards laid down in the Swedish Act concerning the Ethical Review of Research Involving Humans [49] and adhered to the principles of the Declaration of Helsinki as revised by the World Medical Association [50]. As this project was a retrospective observational study, there were no risks of physical harm or pain to participants. The likelihood of discomfort due to potential breaches of privacy was assessed as very low. The research protocol was reviewed and approved by the Swedish Ethical Review Authority (DNR-2020-01614), which also granted a waiver of informed consent due to the impracticality of obtaining consent from a large and diverse sample. The nature of our study necessitated the analysis of a significantly large sample size, rendering it impractical and unfeasible to obtain informed consent from each individual participant. The sheer scale and diversity of the sample made it logistically impossible to engage with each participant on an individual basis. Waiver of the need for informed consent, was applied for at the Swedish Ethical Review Authority, who approved the request (DNR-2020-01614).
Consent for publication
All authors have given consent for publication.
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.Hruska K, Castren M, Banerjee J, Behringer W, Bjornsen LP, Cameron P et al. Template for uniform reporting of emergency department measures, consensus according to the Utstein method. Eur J Emerg Med. 2019;26(6):417–22. 10.1097/MEJ.0000000000000582. PMID: 31464714. [DOI] [PubMed]
- 2.Berg LM, Ehrenberg A, Florin J, Ostergren J, Discacciati A, Goransson KE. Associations between crowding and Ten-Day mortality among patients allocated lower triage acuity levels without need of acute hospital care on departure from the emergency department. Ann Emerg Med. 2019;74(3):345–56. [DOI] [PubMed] [Google Scholar]
- 3.Walsh M, Knott JC. Satisfaction with the emergency department environment decreases with length of stay. Emerg Med J. 2010;27(11):821–8. [DOI] [PubMed] [Google Scholar]
- 4.Ackroyd-Stolarz S, Guernsey JR, MacKinnon NJ, Kovacs G. The association between a prolonged stay in the emergency department and adverse events in older patients admitted to hospital: a retrospective cohort study. BMJ Qual Saf. 2011;20(7):564–9. [DOI] [PubMed] [Google Scholar]
- 5.Derlet RW, Richards JR. Overcrowding in the nation’s emergency departments: complex causes and disturbing effects. Ann Emerg Med. 2000;35(1):63–8. [DOI] [PubMed] [Google Scholar]
- 6.Pitts SR, Pines JM, Handrigan MT, Kellermann AL. National trends in emergency department occupancy, 2001 to 2008: effect of inpatient admissions versus emergency department practice intensity. Ann Emerg Med. 2012;60(6):679–86. e3. [DOI] [PubMed] [Google Scholar]
- 7.Morley C, Unwin M, Peterson GM, Stankovich J, Kinsman L. Emergency department crowding: A systematic review of causes, consequences and solutions. PLoS ONE. 2018;13(8):e0203316. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.New Zealand Ministry of Health. Targeting emergencies - Shorter stays in emergency departments. 2011.
- 9.Yarmohammadian MH, Rezaei F, Haghshenas A, Tavakoli N. Overcrowding in emergency departments: A review of strategies to decrease future challenges. J Res Med Sci. 2017;22:23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.De Freitas L, Goodacre S, O’Hara R, Thokala P, Hariharan S. Interventions to improve patient flow in emergency departments: an umbrella review. Emerg Med J. 2018;35(10):626–37. [DOI] [PubMed] [Google Scholar]
- 11.Paling S, Lambert J, Clouting J, González-Esquerré J, Auterson T. Waiting times in emergency departments: exploring the factors associated with longer patient waits for emergency care in England using routinely collected daily data. Emerg Med J. 2020;37(12):781–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.McKenna P, Heslin SM, Viccellio P, Mallon WK, Hernandez C, Morley EJ. Emergency department and hospital crowding: causes, consequences, and cures. Clin Exp Emerg Med. 2019;6(3):189–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Asplin BR, Magid DJ, Rhodes KV, Solberg LI, Lurie N, Camargo CA. Jr. A conceptual model of emergency department crowding. Ann Emerg Med. 2003;42(2):173–80. [DOI] [PubMed] [Google Scholar]
- 14.Kwok ESH, Geymonat C, Peters K, Bickerton K, Mackenzie T, Lamothe R, et al. A novel emergency department surge protocol: implementation of a targeted response plan. J Clin Outcomes Manage. 2015;22(11):495–503. [Google Scholar]
- 15.Socialstyrelsen. Statistikdatabas för akutmottagningar, väntetider och besök 2023 [Available from: https://sdb.socialstyrelsen.se/if_avt_manad/
- 16.Socialstyrelsen. Statistik om akutmottagningar, väntetider och besök 2021 2022. Contract No.: Art.nr: 2022-10-8141.
- 17.Af Ugglas B, Lindmarker P, Ekelund U, Djärv T, Holzmann MJ. Emergency department crowding and mortality in 14 Swedish emergency departments, a cohort study leveraging the Swedish emergency registry (SVAR). PLoS ONE. 2021;16(3):e0247881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Perdahl T, Axelsson S, Svensson P, Djarv T. Patient and organizational characteristics predict a long length of stay in the emergency department - a Swedish cohort study. Eur J Emerg Med. 2017;24(4):284–9. [DOI] [PubMed] [Google Scholar]
- 19.Cheng I, Zwarenstein M, Kiss A, Castren M, Brommels M, Schull M. Factors associated with failure of emergency wait-time targets for high acuity discharges and intensive care unit admissions. CJEM. 2018;20(1):112–24. [DOI] [PubMed] [Google Scholar]
- 20.Payne K, Risi D, O’Hare A, Binks S, Curtis K. Factors that contribute to patient length of stay in the emergency department: A time in motion observational study. Australas Emerg Care. 2023;26(4):321–5. 10.1016/j.auec.2023.04.002. Epub 2023 May 2. PMID: 37142544. [DOI] [PubMed]
- 21.Myndigheten för vård- och omsorgsanalys. En akut bild av Sverige: Kartläggning av akutsjukvårdens organisation och arbetsfördelning. 2018. Report No.: 2018:13.
- 22.Moellekaer A, Duvald I, Obel B, Madsen B, Eskildsen J, Kirkegaard H. The organization of Danish emergency departments. Eur J Emerg Med. 2019;26(4):295–300. [DOI] [PubMed] [Google Scholar]
- 23.Widgren B. Rätt med RETTS beslutstöd - Utmaningar och Möjligheter Inom Akutsjukvård. 2 ed. nd ed. Göteborg: B4PRESS förlag; 2019. [Google Scholar]
- 24.Swedish Tax Agency. Population registration in Sweden. Report No.: SKV 717B; 2007. [Google Scholar]
- 25.Geifman N, Cohen R, Rubin E. Redefining meaningful age groups in the context of disease. Age (Dordr). 2013;35(6):2357–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Af Ugglas B, Djärv T, Ljungman PLS, Holzmann MJ. Emergency department crowding associated with increased 30-day mortality: a cohort study in Stockholm region, sweden, 2012 to 2016. J Am Coll Emerg Physicians Open. 2020;1(6):1312–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Swedish National board of Health and Welfare. Riktvärden för beläggningsgrad och vårdplatser. 2023. Report No.: 2023-5-8573.
- 28.Friebel R, Fisher R, Deeny SR, Gardner T, Molloy A, Steventon A. The implications of high bed occupancy rates on readmission rates in england: A longitudinal study. Health Policy. 2019;123(8):765–72. [DOI] [PubMed] [Google Scholar]
- 29.Blom MC, Jonsson F, Landin-Olsson M, Ivarsson K. The probability of patients being admitted from the emergency department is negatively correlated to in-hospital bed occupancy - a registry study. Int J Emerg Med. 2014;7(1):8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.StataCorp. Stata statistical software: release 18. College station. TX: StataCorp LLC; 2023. [Google Scholar]
- 31.Casalino E, Wargon M, Peroziello A, Choquet C, Leroy C, Beaune S, et al. Predictive factors for longer length of stay in an emergency department: a prospective multicentre study evaluating the impact of age, patient’s clinical acuity and complexity, and care pathways. Emerg Med J. 2014;31(5):361–8. [DOI] [PubMed] [Google Scholar]
- 32.Gripko M, Joseph A, MohammadiGorji S. Effects of the physical environment on children and families in hospital-based emergency departments: A systematic literature review. J Environ Psychol. 2023;86. [DOI] [PMC free article] [PubMed]
- 33.Trottier ED, Doré-Bergeron MJ, Chauvin-Kimoff L, Baerg K, Ali S. Managing pain and distress in children undergoing brief diagnostic and therapeutic procedures. Paediatr Child Health. 2019;24(8):509–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Kocher KE, Meurer WJ, Desmond JS, Nallamothu BK. Effect of testing and treatment on emergency department length of stay using a National database. J Emerg Nurs. 2013;39(3):221–5. 10.1016/j.jen.2012.02.015. Epub May 18. [DOI] [PubMed] [Google Scholar]
- 35.Kanzaria HK, Probst MA, Ponce NA, Hsia RY. The association between advanced diagnostic imaging and ED length of stay. Am J Emerg Med. 2014;32(10):1253–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Whitson MR, Mayo PH. Ultrasonography in the emergency department. Crit Care. 2016;20(1):227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Hausfater P, Hajage D, Bulsei J, Canavaggio P, Lafourcade A, Paquet AL, et al. Impact of Point-of-care testing on length of stay of patients in the emergency department: A Cluster-randomized controlled study. Acad Emerg Med. 2020;27(10):974–83. [DOI] [PubMed] [Google Scholar]
- 38.Alimohammadi H, Zareh Shahamati S, Karkhaneh Yousefi A, Safarpour Lima B. Potentially inappropriate brain CT-scan requesting in the emergency department: A retrospective study in patients with neurologic complaints. Acta Biomed. 2021;92(5):e2021302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Ståhlbrandt H, Björnfot I, Cederlund T, Almén A. CT and MRI imaging in sweden: retrospective appropriateness analysis of large referral samples. Insights Imaging. 2023;14(1):134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Novoa Ferro M, Santos Armentia E, Silva Priegue N, Jurado Basildo C, Sepúlveda Villegas CA. Del Campo estepar S. Brain CT requests from emergency department: reality. Radiologia (Engl Ed). 2022;64(5):422–32. [DOI] [PubMed] [Google Scholar]
- 41.Wiler JL, Handel DA, Ginde AA, Aronsky D, Genes NG, Hackman JL, et al. Predictors of patient length of stay in 9 emergency departments. Eur Heart J Acute Cardiovasc Care. 2012;1(2):99–108. 10.1177/2048872612447049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Wickman L, Svensson P, Djärv T. Effect of crowding on length of stay for common chief complaints in the emergency department: A STROBE cohort study. Medicine. 2017;96(44):1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Wretborn J, Khoshnood A, Wieloch M, Ekelund U. Skane emergency department assessment of patient load (SEAL)-A model to estimate crowding based on workload in Swedish emergency departments. PLoS ONE. 2015;10(6):e0130020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Rasouli HR, Esfahani AA, Nobakht M, Eskandari M, Mahmoodi S, Goodarzi H, et al. Outcomes of crowding in emergency departments; a systematic review. Arch Acad Emerg Med. 2019;7(1):e52. [PMC free article] [PubMed] [Google Scholar]
- 45.Joseph JW, White BA. Emergency department operations: an overview. Emerg Med Clin North Am. 2020;38(3):549–62. [DOI] [PubMed] [Google Scholar]
- 46.Organisation for Economic Cooperation and Development [OECD]. OECD Health Statistics 2023 - Healthcare resources, hospital beds by sector [Available from: https://stats.oecd.org/Index.aspx?ThemeTreeId=9
- 47.Swedish Association of Local Authorities and Regions [SKR]. Väntetidsstatistik - Överbeläggningar och utlokaliserade patienter [Available from: https://www.vantetider.se/vantetiderivarden/vantetidsstatistik/specialiseradvard/overbelaggningar.54399.html
- 48.Badr S, Nyce A, Awan T, Cortes D, Mowdawalla C, Rachoin JS. Measures of emergency department crowding, a systematic review. How to make sense of a long list. Open Access Emerg Med. 2022;14:5–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.World Medical Association. World medical association declaration of helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191–4. 10.1001/jama.2013.281053. [DOI] [PubMed] [Google Scholar]
- 50.Svensk författningssamling (Swedish Code of Statutes). The act concerning the ethical review of research involving humans (SFS 2003:460). Stockholm: Government Offices of Sweden; 2003. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Unfortunately, the data set is not permitted to be available for open access. Data used for this study is sensitive patient data from Örebro University Hospital and Mälarsjukhuset Eskilstuna. These data are not publicly available due to restrictions governed by the Swedish Ethical Review Authority, and Swedish laws in relation to research in, and distribution of health care data. However, Örebro University researchers are mandated to archive research data for 10 years if an audit is required. To gain access to the data, please contact the Swedish Ethical Review Authority and cite registration number: 2020-01614.