Characteristics of emergency department visits in Jordanian public hospitals

Abstract

Background

Emergency Department (ED) provides the initial point of care for many patients. Healthcare management systems globally and locally are now racing to utilize information systems to gain insights into the characteristics of patients. Learning about system’s characteristics can better enable.

Objectives

To evaluate (ED) visits in Jordanian Public hospitals and characterize the clinical demographic and epidemiological features of ED visits. Study the hospitalization of ED visits, define the patients’ profile associated with visit-related characteristics, and related features in Jordanian Public hospitals.

Patients and methods

Retrospective cohort study of patients with at least 1 ED visit to Jordanian Public Hospitals between January 1, 2018, and November 31, 2023, derived from electronic healthcare record system (aka Hakeem) for all Jordanian public hospitals. Using ministry of health National Hospital ED data. Descriptive statics and inferential statistics including univariate and multivariate regression by means of the odds ratio and 95% confidence interval for the factors associated with hospitalization ED visits.

Results

Between 2018 and 2023, Jordan’s Emergency Departments recorded 9.16 million visits, with 6.37% resulting in hospitalization. Common complaints included pain and fever, with hospitalization linked to severity and patient demographics. Key factors influencing outcomes were triage class, age, shift, and regional differences, providing crucial insights for optimizing ED management and resource allocation in Jordan.

Introduction

The role of Emergency Department (ED) has changed over time from providing rooms for the resuscitation of severely ill or injured patients to becoming a complex healthcare service setting, using sophisticated technology to provide healthcare that was once given in hospital wards [1]. EDs now serve as a critical access point in healthcare systems, especially for hospital admissions [2, 3]. Patients can seek care at any ED, regardless of their ability to pay, and this care is available 24 hours a day [4]. It has been reported that the ED utilization is higher than other healthcare settings. In the United States, for example, 11% of the outpatient services are provided in the ED and almost half of the hospital admissions go through the ED. Thus, also cost of ED services (EDs) has been estimated to represent five to 6%– approximately $200 billion– of United States (US) health care spending [5]. Several studies have reported dramatic increase in ED visits [4–6]. This increased emergency department utilization accounted for about 25% of visits by less than 10% of the population [7]. Studies have indicated that most frequent visitors are patients who consult primary health care physicians, have insurance, and use inpatient and outpatient health care services [1, 8, 9]. In the Middle East context, the reported literature on emergency department visitors and their characteristic, along with the determinant factors, are scarce [2, 10].

ED and Hospital Admissions

The literature highlights a significant increase in ED utilization, which now surpasses other healthcare settings in demand [2, 11]. ED overcrowding is a significant global issue, driven primarily by the boarding of patients who have been admitted but are awaiting inpatient beds. Boarding leads to extended ED stays and exacerbates crowding, a problem that has been well-documented in numerous studies. For example, research indicates that ED crowding is associated with delays in care, which can lead to adverse patient outcomes such as increased mortality [12, 13]. Historically, acute hospital inpatients were often admitted directly by primary care or specialty physicians, resulting in shorter ED stays and higher rates of brief inpatient admissions. However, the current trend involves detailed evaluations in the ED before admission, contributing to longer ED stays and increased crowding [2].

International Perspectives on ED and Admission

Globally, the problem of ED utilization varies in severity depending on the healthcare system. For instance, in France, legal requirements prevent EDs from denying care based on symptoms or vital signs, leading to substantial congestion. French healthcare leaders argue that solving this issue require an improved system-wide solution that incorporate inpatient capacity constraints [2, 11]. The same applies to India, where limited inpatient bed availability often results in patients being kept in the ED until they are treated and discharged, further straining ED resource. In Hong Kong, with 2 million ED visits annually, the hospital admission rate ranges from 25% to 30% [2]. The literature on hospitalization from EDs underlines the essential role of EDs in modern healthcare systems, particularly in facilitating hospital admissions. However, the increasing utilization of EDs and the boarding of admitted patients have led to significant pressure on ED resources. Addressing such challenges involves a multidisciplinary approach that includes effective bed management, the adoption of predictive analytics, and tailored operational strategies [14]. These measures are crucial for improving patient outcomes and ensuring the efficient operation of healthcare systems. In Jordan, ED hospitalization outcome have not been yet studied, or even mentioned in related clinical studies. However, in the region, other countries have touched on the subject. The authors of [1] estimated the frequency of ED visits, and described the patients’ profile along with visit-related characteristics, a Saudi-hospital. Furthermore, [15] have investigated several operational parameters regarding the visits to one Saudi hospital relying solely on descriptive statistics. Setting specific empirical evidence can enlighten healthcare policymakers, practitioners, hospital administrators, and ED heads on implementing improvement initiative leading to higher utilization of ED services, and insuring safety for patients. This study aims primarily to characterize and evaluate the ED visits in Jordanian public hospitals as well as estimate the magnitude hospitalization outcome from ED, and identify any trends in ED visits.

Methods

Study design and setting

This is a retrospective cohort study based on the data extracted from the EDs in all public hospitals in the Hashemite Kingdome of Jordan. With the use of the Electronic Health Systems (EHS, AKA-Hakeem), combined data from the ED component of the 2018-2023 of all public hospitals. Data were extracted from the national Electronic Health System (EHS, “Hakeem”) for ED visits between January 1, 2018 and November 30, 2023, encompassing all eligible public hospitals. Hakeem collects information about the visits to hospitals in Jordan. The collection has been conducted annually since the establishment of the electronic health system 2015, however it’s not until 2018 when proper implementation of the system took place. The Hakeem system has been operational since 2015, but comprehensive adoption across public hospitals occurred by 2018, ensuring reliable electronic data capture from 2018 onward.

Population

The study aims to identify all ED visits recorded in the 2018–2023 date. All ED visits documented from 2018 through 2023 were initially considered for inclusion. The study excludes private hospitals, army-related (military) hospitals, and hospitals with fewer than 1,000 ED visits per year to focus on consistently reporting centers.

Data collection

Data Source

Data for this study were extracted from the Hakeem electronic medical record system, which captures clinical encounters across all public hospitals in Jordan. The initial dataset comprised approximately 9.16 million emergency department (ED) visit records between January 2018 and November 2023.

Variables Collected

For each ED visit, demographic data (age, sex, and nationality), clinical characteristics (free-text chief complaint and medical department), and operational variables were recorded. Operational data included the triage acuity level—assigned according to the Canadian Triage and Acuity Scale (CTAS)—the date and time of arrival, the time of departure, the overall length of stay, and visit outcome [16]. ED outcomes were classified into five mutually exclusive categories: discharged home, hospitalized, against medical advice, deceased, or unknown. Additional temporal variables were derived, including shift timing, which was categorized as morning, evening, or night in accordance with the Ministry of Health’s standard 8-hour shift schedule. The day of the week (Sunday through Saturday) was also recorded to allow for analysis of weekly service trends. Utilization history was captured through the total number of ED visits per patient during the study period, in addition to timestamps for individual arrivals and departures, which facilitated the identification of frequent users and calculation of visit duration. An input-based crowding metric was developed by computing the number of ED arrivals per hour to serve as a proxy for real-time patient load at arrival. To enable time-based analysis, all visit times were categorized into three predefined 8-hour shifts in line with the standard operational schedule of the Jordanian Ministry of Health.

Data Cleaning and Preparation

The raw data were initially exported to Microsoft Excel and subsequently cleaned and processed for statistical analysis. The cleaning process involved the removal of duplicate records, cases with missing essential fields, and visits from hospitals that did not meet the study’s inclusion criteria. A flow diagram (Fig. 1) outlines the data cleaning and inclusion/exclusion process.

Fig. 1.

Data Cleaning and Processing

Ethics Approval

This study was approved by the Institutional Review Boards of the Hashemite University and the Jordanian Ministry of Health. All data were fully de-identified prior to analysis. Informed consent was therefore not required. The study adhered to the principles of the Declaration of Helsinki.

Qualitative data processing

Free-text data often introduce discrepancies in datasets, especially when most other variables are numeric or categorical and require minimal cleaning. In our study, one crucial field—the triage chief complaint—was collected as free text. Unlike predefined numeric or categorical entries, free-text fields allow a wide variety of responses, leading to numerous inconsistencies in the raw data [17, 18]. Consequently, careful pre-processing of this free-text field was essential before analysis. To address this challenge, we implemented a structured multi-step pre-processing approach. First, automated frequency analysis was applied to identify the most common chief complaint entries. Then, clinical experts manually reviewed the free-text responses to correct spelling variations, consolidate synonymous terms, and standardize the terminology. Multiple experts participated in the manual coding process to achieve consensus and ensure high inter-rater reliability—a method commonly employed to enhance data validity in similar studies [19]. This process yielded a standardized set of chief complaints. Approximately 85% of all chief complaint entries (encompassing over 640 distinct terms) were successfully standardized through this approach. The remaining 15% of entries (more than 40,000 infrequent or ambiguous cases) could not be confidently standardized and were categorized as miscellaneous. The standardization was guided by the World Health Organization’s ICD-11 classification principles [12], implemented as a four-tier hierarchy: Normalization of similar terms: Semantically similar terms were unified (e.g., all variants of “pain” were mapped to a standard term “PAIN”). Primary symptom identification: If multiple symptoms were listed in one complaint, the primary symptom was identified and recorded. Secondary complaint capture: Any additional symptom or qualifier in the complaint was captured as a secondary element. System classification: Each complaint was then categorized by physiological system (e.g., a complaint of “leg pain” was classified under the Orthopedic system). This method improved both the clinical coherence and the analytical utility of the chief complaint data, aligning with established emergency department coding practices [20, 21]. Table 1 provides examples of the standardization and classification process for several chief complaints.

Table 1.

Example of the classification process

Chief_complaint_1	STANDARAD	CLASS 1	CLASS 2	SYSTEM
Vomiting	VOMITTING	VOMITTING	VOMITTING	GASTROINTESTINAL
Check up	CHECK UP	CHECK UP	UNSPECFIED	GENERAL
Cough	COUGH	COUGH	UNSPECFIED	RESPIRATORY
Follow up	FOLLOW UP	FOLLOW UP	UNSPECFIED	GENERAL
Gastric pain	GASTRIC PAIN	GASTRIC	PAIN	GASTROINTESTINAL
Headache	HEADACHE	HEADACHE	UNSPECFIED	NEUROLOGIC

In addition to standardizing the chief complaint field, we coded several other variables to facilitate analysis. Nationality was used as a proxy for migration status: patients were categorized as Jordanian or non-Jordanian, with Syrian nationals representing the largest refugee group in the sample. Each patient’s total number of ED visits during the study period was calculated and classified into three utilization frequency categories—occasional (1–2 visits), frequent (3–9 visits), and very frequent ( visits)—following published definitions [19]. For regression analyses, ED visit frequency was further dichotomized into visits versus visits (i.e. more than two vs. two or fewer visits). Triage acuity for each visit was recorded using the Canadian Triage and Acuity Scale (CTAS) and later grouped into urgent (CTAS 1–3) and non-urgent (CTAS 4–5) categories, consistent with standard practice [19]. The outcome of each ED visit was recorded as a binary variable indicating hospital admission versus discharge [19]. Finally, hospitals in the study were grouped into three broad geographic regions (North, Middle, South) to examine regional variations in ED utilization patterns.

Data analysis

Data analysis was conducted using both descriptive and inferential statistical methods. Initially, the demographic and clinical characteristics of the ED visit cohort were summarized using frequencies and percentages for categorical variables and appropriate summary statistics (means or medians) for continuous variables. Comparisons of categorical variables between patient groups (e.g., admitted vs. non-admitted) were performed using Chi-square tests, or Fisher’s exact tests if expected frequencies were low. Continuous variables were analyzed using parametric tests (e.g., t-tests, ANOVA) or non-parametric equivalents (e.g., Mann–Whitney U test), depending on data distribution. Univariate logistic regression analyses were conducted to explore the crude associations between individual predictors (patient age, sex, nationality, triage acuity level, chief complaint category, shift timing, hospital region, and ED visit frequency) and the likelihood of hospitalization after an ED visit. Variables demonstrating a statistically significant association in univariate analysis or considered clinically relevant were then included in a multivariate logistic regression model. This model aimed to identify independent predictors of hospital admission. Results were reported as unadjusted odds ratios (ORs) and adjusted odds ratios (AORs), with corresponding 95% confidence intervals (CIs). Model performance was evaluated using the concordance statistic (Area Under the Receiver Operating Characteristic Curve, AUC) to measure discriminatory ability, and the Hosmer–Lemeshow test assessed model calibration. All statistical analyses were performed using Stata software (version 15). To ensure consistency with international reporting standards, age groups and other categorical variables were aligned with classifications recommended by the World Health Organization [22], facilitating comparisons with similar international studies.

Results

ED visit volume and patient characteristics (2018–2023)

During the 2018–2023 period, Jordan’s emergency departments saw notable fluctuations in utilization rates. Figure 2 illustrates the annual volume of ED visits (per 100,000 population), stratified by year and age group.

Fig. 2.

Annual volume of ED visits (per 100,000 population), stratified by year and age group

There was a steady rise in visit rates from 2018 to 2019, followed by a sharp decline in 2020, coinciding with the COVID-19 lockdowns. A marked rebound occurred in 2021 after restrictions eased, but visit rates in 2022 and 2023 declined again slightly, suggesting a possible new baseline post-pandemic. Across all years, older adults had higher ED utilization rates than younger age groups (Fig.1). The cohort of ED visits was roughly balanced in terms of patient sex, with males accounting for 53.1% of visits and females 46.9%. The age distribution of patients was skewed toward younger groups– for example, children under 14 comprised the largest age group of visits– but older patients constituted a disproportionately larger share of those who were hospitalized (as detailed below). The majority of encounters (87% of all visits) took place during daytime hours (morning and evening shifts), with only about 13% of visits occurring overnight. There was minimal variation by day of week in ED attendance; Tuesdays had a slightly higher volume of visits than other days, but the difference was not pronounced. The vast majority of ED users ( 95%) were Jordanian nationals, reflecting the national population, with a small fraction of patients of other nationalities (primarily Syrian) using ED services. Patients presented to the ED with a wide range of complaints, but a few chief complaints accounted for a large proportion of visits. Table 2 summarizes the 20 most common presenting complaints.

Table 2.

The 20 most common Complaints presented at ED

	Complaint	Frequency	Percentage	Cum
1	PAIN	2,018,530	26.2	26.2
2	FEVER	966,829	12.55	38.74
3	Flu Like Symptoms	673,654	8.74	47.49
4	COUGH	542,256	7.04	54.52
5	EMERGENCY ROOM	494,548	6.42	60.94
6	STOMACHACHE	290,400	3.77	64.71
7	ABDOMINAL PAIN	240,139	3.12	67.83
8	UNKNOWN	239,049	3.1	70.93
9	FOLLOW UP	217,746	2.83	73.75
10	VOMITTING	145,217	1.88	75.64
11	HEADACHE	132,330	1.72	77.36
12	HYPERTENSION	117,170	1.52	78.88
13	LABOR PAIN	115,437	1.5	80.38
14	SHORTNESS OF BREATH	111,091	1.44	81.82
15	HEARTBURN	108,365	1.41	83.22
16	GESTATIONAL WEEKS	108,184	1.4	84.63
17	GENERAL WEACKNESS	104,973	1.36	85.99
18	CHEST PAIN	90,226	1.17	87.16
19	URINARY	67,844	0.88	88.04
20	DIZZINESS	57,298	0.74	88.78

The single most frequent presentation was unspecified pain, which accounted for 26.2% of all ED visits. This was followed by fever (12.6%), influenza-like illness (8.7%), and cough (7.0%). Cumulatively, these top four complaints made up over half of all ED cases. Other common presentations included functional dyspepsia, abdominal pain, and symptoms requiring evaluation for conditions like hypertension or complications of pregnancy. When categorizing visits by broad medical specialty systems, general/minor ailments (approximately 26.9% of all visits) and ENT-related issues (24.8%) were the leading categories, together comprising about half of all ED visits. No other single specialty category (e.g., gastrointestinal, musculoskeletal, cardiovascular, etc.) exceeded 15% of visits individually.

Disposition outcomes and bivariate associations

Of the 7.54 million ED visits analyzed, the majority (around 93%) resulted in patients being treated and sent home, while a smaller proportion (approximately 7%) ended in hospital admission. Only a very small fraction of cases (<0.1%) resulted in death in the ED or patients leaving against medical advice. We observed significant differences in patient and visit characteristics between visits that ended in hospitalization and those that did not (Table 3).

Table 3.

ED Visit Characteristics

Characteristics	Deceased	Against Medical Advice	Home	Hospitalized	Unknown	P-Value
Triage - Class
Unknown	0 (0.00)	182 (0.82)	103.931 (1.38)	2 (0.00)	33.493 (3.63)	0.000
(01) Resuscitation	41 (5.37)	366 (1.66)	401.513 (5.32)	51.127 (8.85)	38.948 (4.22)
(02) Emergency	209 (27.36)	1.663 (7.53)	568.751 (7.54)	65.166 (11.28)	94.203 (10.21)
(03) Urgent	254 (33.25)	7.076 (32.05)	1.736.996 (23.03)	125.309 (21.70)	180.618 (19.57)
(04) Less Urgent	160 (20.94)	9.127 (41.34)	2.636.922 (34.96)	161.752 (28.01)	286.653 (31.05)
(05) Non Urgent	100 (13.09)	3.663 (16.59)	2.093.755 (27.76)	174.182 (30.16)	289.166 (31.33)
Age- group
<14yrs	216 (28.27)	7.645 (34.63)	2.666..817 (35.36)	126.832 (21.96)	295.057 (31.96)	0.000
14-24yrs	85 (11.13)	2.784 (12.61)	1.204.310 (15.97)	114.063 (19.75)	146.948 (4.22)
25-44yrs -	100 (13.09)	2.628 (11.90)	1.217.326 (16.14)	130.232 (22.55)	151.009 (16.36)
45-54yrs	89 (11.65)	2.255 (10.21)	922.222 (12.23)	82.213 (14.24)	113.626 (12.31)
55-65yrs	102 (13.35)	2.546 (11.53)	762.084 (10.10)	55.568 (9.62)	100.691 (10.91)
>65yrs	172 (22.51)	4.219 (19.11)	768.909 (10.20)	68.630 (11.88)	116.491 (12.62)
Shift
Morning	304 (39.79)	8.563 (38.79)	2.862.185 (37.95)	192.053 (33.25)	327.860 (35.52)	0.000
Evening	366 (47.91)	11.174 (50.61)	3.749.217 (49.71)	281.306 (48.71)	493.816 (53.50)
Night	94 (12.30)	2.340 (10.60)	930.266 (12.34)	104.179 (18.04)	101.416 (10.99)
System
All	0 (0.00)	68 (0.55)	907.966 (14.35)	129.873 (29.26)	75.711 (11.86)	0.000
Cardiology	13 (5.35)	563 (4.53)	102.818 (1.62)	5.274 (1.19)	13.875 (2.17)
Dentist	0 (0.00)	0 (0.00)	3.598 (0.06)	47 (0.01)	400 (0.06)
Dermatology	1 (0.41)	87 (0.70)	12.483 (0.20)	635 (0.14)	755 (0.12)
Ent	61 (25.10)	3.222 (25.90)	1.593.976 (25.19)	51.896 (11.69)	190.202 (29.79)
Gastrointestinal	42 (17.28)	3.292 (26.46)	719.310 (11.37)	61.861 (13.94)	71.343 (11.17)
General	9 (3.70)	515 (4.14)	1.771.339 (27.99)	75.067 (16.91)	150.359 (23.55)
Genitourinary	37 (15.23)	565 (4.54)	60.135 (0.95)	1.805 (0.41)	27.068 (4.24)
Internal	8 (3.29)	344 (2.77)	53.973 (0.85)	2.077 (0.47)	4.521 (0.71)
Laboratory	0 (0.00)	2 (0.02)	31.123 (0.49)	17.015 (3.83)	1.179 (0.18)
Mental Health	0.(0.00)	0.(0.00)	1.064 (0.02)	472 (0.17)	15 (0.00)
Neurologic	11 (4.53)	264 (2.12)	138.380 (2.19)	7.539 (1.70)	13.081 (2.05)
Obstetrics & Gynecology	7 (2.88)	310 (2.49)	100.188 (1.58)	15.965 (3.60)	13.688 (2.14)
Ophthalmology	1 (0.41)	178 (1.43)	12.926 (0.20)	319 (0.07)	1.680 (0.26)
Orthopedic	15 (6.17)	1.118 (8.99)	162.201 (2.56)	14.904 (3.36)	16.672 (2.61)
Pediatric	0 (0.00)	1 (0.01)	3.414 (0.5)	150 (0.03)	1.479 (0.23)
Pharmacy	0 (0.00)	0 (0.00)	1.017 (0.02)	14 (0.00)	7 (0.00)
>Respiratory	34 (13.99)	1.822 (14.65)	605.528 (9.57)	55.989 (12.62)	52.066 (8.15)
Substance	0 (0.00)	0 (0.00)	375 (0.01)	310 (0.07)	3 (0.00)
Unspecified	4 (1.65)	90 (0.72)	45.553 (0.72)	2.342 (0.53)	4.428 (0.69)
Day of the Week
Sunday	63 (8.25)	3.367 (15.25)	1.120..290 (14.85)	83.245 (14.41)	139.632 (15.13)	0.000
Monday	85 (11.13)	3.383 (15.32)	1.107.346 (14.68)	90.796 (15.72)	139.872 (15.15)
Tuesday	100 (13.09)	3.644 (16.51)	1.097.513 (14.55)	80.151 (13.88)	136.180 (14.75)
Wednesday	224 (29.32)	3.665 (16.60)	1.089.783 (14.45)	84.411 (14.62)	128.305 (13.90)
Thursday	171 (22.38)	3.038 (13.76)	1.069.580 (14.18)	83.073 (14.38)	125.445 (13.59)
>Friday	94 (12.30)	2.460 (11.14)	998.964 (13.25)	74.723 (12.94)	118.609 (12.85)
Saturday	27 (3.53)	2.520 (11.41)	1.058.192 (14.03)	81.139 (14.05)	135.049 (14.63)
Gender
Female	336 (43.98)	9.456 (42.83)	3.505.108 (46.48)	306.321 (53.02)	429.972 (46.58)	0.000
Male	428 (56.02)	12.621 (11.41)	4.036.560 (53.52)	271.307 (46.98)	493.120 (53.42)
Nationality
Egypt	5 (0.65)	187 (0.85)	35.364 (0.47)	3.002 (0.52)	5.236 (0.57)	0.000
Jordan	739 (96.73)	21.133 (95.72)	7.156.644 (94.89)	532.599 (92.22)	879.825 (95.31)
Occupied Palestinian	10 (1.31)	414 (1.88)	138.177 (1.83)	8.711 (1.51)	21.841 (2.37)
Others	5 (0.65)	163 (0.74)	27.006 (0.36)	2.947 (0.51)	3.653 (0.40)
Syrian Arab	5 (0.65)	180 (0.82)	184.477 (2.45)	30.279 (5.24)	12.537 (1.36)
Republic
Region
>Middle	764 (100.00)	22.077 (100.00)	4.347.683 (57.65)	224.241 (38.83)	790.190 (85.60)	0.000
>North	0 (0.00)	0 (0.00)	2.827.371 (37.49)	323.901 (56.08)	130.807 (14.17)
South	0 (0.00)	0 (0.00)	366.614 (4.86)	29.396 (5.09)	2.095 (0.23)
Duration
1 hr or less	165 (21.60)	2783 (12.61)	3936494 (52.20)	344348 (59.62)	384762 (41.68)
1.0–2.0 hrs.	50 (6.54)	1230 (5.57)	1161149 (15.40)	65306 (11.31)	120919 (13.10)
2.0–3.5 hrs.	53 (6.94)	1437 (6.51)	779857 (10.34)	49678 (8.60)	84543 (9.16)
3.5–6.0 hrs.	62 (8.12)	1916 (8.68)	560500 (7.43)	34661 (6.00)	64657 (7.00)
6.0–10.0 hrs.	49 (6.41)	2754 (12.47)	375830 (4.98)	22379 (3.87)	76198 (8.25)
>10 hrs	385 (50.39)	11957 (54.16)	727838 (9.65)	61166 (10.59)	192013 (20.80)
Visits
1-3 visits	446 (58.38)	14201 (64.32)	3226727 (42.79)	256510 (44.41)	447935 (48.53)
4-10 visits	286 (37.43)	7125 (32.27)	3364200 (44.61)	225151 (38.98)	396022 (42.90)
>10 vsists	32 (4.19)	751 (3.40)	950741 (12.61)	95877 (16.60)	79135 (8.57)

For instance, triage acuity was strongly associated with disposition (X² 0.001). High-acuity cases labeled as Resuscitation were rare among total visits (only 5% of all ED cases) but constituted a larger share of admissions (nearly 9% of hospitalized cases), implying that the most critical presentations often led to inpatient care. Conversely, Non-Urgent visits (which made up about 28% of total cases) had a relatively low likelihood of hospital admission, as many of these lower-acuity patients were discharged home. Patient age also showed a clear pattern in outcomes (X² 0.001). Younger patients (<14 years) represented the largest group in the ED overall, and the vast majority of them were discharged without needing admission. In contrast, older patients were more likely to require hospitalization: for example, adults over 65 years, while comprising only about 10% of total visits, accounted for a higher proportion of ED admissions compared to younger groups. Timing of visit had some impact as well. ED visits during the night shift (11pm–7am) were somewhat more likely to end in hospitalization than visits during daytime shifts– this is reflected in night shift visits forming 18% of hospitalized cases but only 12% of overall visits. Visits on different days of the week did not show major disparities in admission rates; although Tuesday had the highest visit volume, the distribution of hospitalizations across weekdays was fairly even Table 3. There was no significant difference in admission rates between male and female patients on a bivariate level (p=0.11), as both sexes had similar proportions of visits resulting in hospitalization. Nationality and region were statistically associated with disposition (0.001), but these differences were driven by the large sample size– nearly all patients were Jordanian from the Central region, with slightly higher admission rates observed among patients from the Northern region and among Syrian nationals (who formed a small subset of ED users).

Multivariable analysis: predictors of ED hospital admission

We fitted a multivariable logistic regression model (with penalization to account for the large dataset and numerous predictors) to identify independent predictors of ED visits resulting in hospital admission. Table 4 presents the adjusted odds ratios (AORs) and 95% confidence intervals for the key factors.

Table 4.

Adjusted Odds Ratios (AORs) and 95% confidence intervals for the key factors

Characteristics	Home	Hos	Univariate			Multivariate
			OR	95% CI		AOR	95% CI
Triage - Class	0	1
(00) Unknown	137406 (1.62)	2 (0.00)	Ref			Ref
(01) Resuscitation	440868 (5.19)m	51127 (8.85)n	1	1	.001	3585.565	896.615	14338.685
(02) Emergency	664826 (7.83)	65166 (11.28)	.845	.835	.856	2752.667	688.376	11007.324
(03) Urgent	1924945 (22.68)	125309 (21.70)	.561	.555	.567	2121.808	530.632	8484.355
(04) Less Urgent	2932862 (34.55)	161752 (28.01)	.476	.471	.481	1641.673	410.559	6564.44
(05) Non Urgent	2386684 (28.12)	174182 (30.16)	.629	.623	.636	1631.989	408.135	6525.742
Age- group	0	1
<14yrs	2969735 (34.99)	126832 (21.96)	Ref			Ref
14-24yrs	1353401 (15.95)	114063 (19.75)	1.973	1.957	1.99	1.436	1.409	1.464
25-44yrs -	1371063 (16.15)	130232 (22.55)	2.224	2.206	2.242	1.533	1.494	1.573
45-54yrs	1038192 (12.23)	82213 (14.24)	1.854	1.837	1.871	1.097	1.06	1.136
55-65yrs	865423 (10.20)	55568 (9.62)	1.503	1.488	1.519	.77	.736	.805
>65yrs	889791 (10.48)	68630 (11.88)	1.806	1.789	1.823	.87	.819	.925
Shift
Morning	3,198,913 (37.69)	192053 (33.25)	Ref			Ref
Evening	4,254,575 (50.13)	281,306 (48.71)	1.101	1.095	1.108	.766	.759	.773
Night	1,034,117 (1218)	104,179 (18.04)	1.678	1.665	1.691	.499	.49	.509
System:
CARDIOLOGY	117269 (1.68)	5274 (1.19)	Ref			Ref
DENTIST	3998 (0.06)	47 (0.01)	.261	.974	1.153	.396	.295	.531
DERMAT- OLOGY	13326 (0.19)	635 (0.14)	1.06	.627	.665	1.618	1.476	1.774
ENT	1787461 (25.61)	51896 (11.69)	.646	1.683	1.783	.688	.661	.716
GASTROI- NTESTINAL	793987 (11.38)	61861 (13.94)	1.732	1.525	1.613	1.873	1.8	1.948
General and minor complaints	2905967 (41.64)	204940 (46.18)	1.568	.433	.483	1.685	1.622	1.752
Genitourinary	87805 (1.26)	1805 (0.41)	.457	.745	.826	.553	.52	.589
INTERNAL	58846 (0.84)	2077 (0.47)	.785	11.329	12.108	1.438	1.355	1.526
LAORATORY	32304 (0.46)	17015 (3.83)	11.712	13.871	16.856	6.941	6.649	7.246
MENTAL HEALTH	1079 (0.02)	742 (0.17)	15.291	1.066	1.145	15.022	11.455	19.699
NEUROLOGIC	151736 (2.17)	7539 (1.70)	1.105	3.01	3.21	1.347	1.286	1.41
OBSTETRIC & GYNECOLOGY	114193 (1.64)	15965 (3.60)	3.109	.428	.538	3.015	2.886	3.149
Unspecfied			.48	1.783	1.901	.742	.655	.839
ORTHOPEDIC	180006 (2.58)	14904 (3.36)	1.841	.578	.804	2.635	2.527	2.747
PEDIATRIC	4894 (0.07)	150 (0.03)	.682	.179	.515	1.183	.965	1.45
PHARMACY	1024 (0.01)	14 (0.00)	.304	1.834	1.943	.393	.231	.667
RESPIRATORY	659450 (9.45)	55989 (12.62)	1.888	15.653	21.244	1.204	1.156	1.254
SUBSTANCE	378 (0.01)	310 (0.07)	18.235	.989	1.093	16.573	14.134	19.433
UNSPECFIED	50076 (0.72)	2342 (0.53)	1.04	.044	.046	1.056	.984	1.132
Day
Sunday	1263353 (14.88)	83245 (14.41)	Ref			Ref
Monday	1250686 (14.74)	90796 (15.72)	1.102	1.091	1.113	.999	.984	1.014
Tuesday	1237438 (14.58)	80151 (13.88)	.983	.973	.993	1.012	.996	1.027
Wednesday	1221977 (14.40)	84411 (14.62)	1.048	1.038	1.059	1.095	1.078	1.111
Thursday	1198235 (14.12)	83073 (14.38)	1.052	1.042	1.063	.978	.963	.993
Friday	1120128 (13.20)	74723 (12.94)	1.012	1.002	1.023	.95	.935	.965
Saturday	1195788 (14.09)	81139 (14.05)	1.03	1.02	1.04	.996	.981	1.012
Gender
Female	3944882 (46.42)	306,231 (53.02)	Ref			Ref
Male	. 4,542,732 (53.52)	271,307 (46.98)	.769	0.76	0.77	628	.622	.633
Nationality
Egypt	40792 (0.48)	3002 (0.52)	Ref			Ref
Jordan	8058344 (94.94)	532599 (92.22)	.898	.865	.932	.867	.822	.915
Occupied Palestinian	160443 (1.89)	8711 (1.51)	.738	.707	.77	.916	.863	.973
Others	30836 (0.36)	2947 (0.51)	1.299	1.232	1.369	1.044	.966	1.128
Syrian Arab Republic	197199 (2.32)	30279 (5.24)	2.086	2.007	2.169	1.741	1.646	1.842
Region
Middle	5160718 (60.80)	224241 (38.83)	Ref			Ref
North	2958178 (34.85)	323901 (56.08)	2.52	2.506	2.534	.814	.806	.823
South	368709 (4.34)	29396 (5.09)	1.835	1.812	1.858	.914	.899	.93
Visits
1-3 visits	3689311 (43.47)	256510 (44.41)	Ref			Ref
4-10 visits	3767635 (44.39)	225151 (38.98)	.86	.854	.865	1.154	1.144	1.165
>10	1030668 (12.14)	95877 (16.60)	1.338	1.328	1.348	1.057	1.044	1.07
Duration
1 hr or less
1.0–2.0 hrs.
2.0–3.5 hrs.
3.5–6.0 hrs.
6.0–10.0 hrs.
>10 hrs

Triage acuity emerged as the strongest independent predictor of hospital admission. Using the Urgent category as the reference (due to its highest raw admission proportion), patients triaged as Non-Urgent had significantly lower odds of admission (AOR 0.14, 0.001). Surprisingly, even patients triaged as Resuscitation exhibited lower odds of admission than those categorized as Urgent (AOR 0.60, 0.01). Patients in the Emergency and Less Urgent categories also had lower adjusted odds (AORs 0.7–0.8, 0.01). Age was a strong predictor of admission. Compared to children under 14, older adults—especially those aged >65—had significantly higher adjusted odds of hospitalization (AOR >1.5, 0.001), after adjusting for all other covariates. A modest but statistically significant sex difference was also observed: female patients had slightly higher adjusted odds of admission than males (AOR 1.1, ). This corresponds with the descriptive pattern that females made up 53% of hospitalized cases, despite constituting 47% of total ED visits. Among operational variables, ED crowding (CROWDNESS) was inversely associated with hospitalization ( 0.001). For example, each additional 50 patients present at the time of arrival was associated with a decrease in the likelihood of admission (AOR per 50-patient increase < 1.0). After adjusting for triage acuity, age, sex, and crowding, time-of-visit variables (shift and weekday) were not statistically significant (p >0.05). Although admission rates were descriptively higher during night shifts, these differences were explained by patient-level factors. Nationality (Jordanian vs. non-Jordanian) was not significant in the adjusted model. The elevated admission rates for Syrian patients in unadjusted data were attenuated when controlling for confounders like age and acuity. Region of hospital showed a small but significant effect: visits to Northern hospitals had higher odds of admission than those in the Central region (AOR 1.2, 0.01). Model diagnostics revealed no evidence of collinearity or overfitting. The penalized regression approach yielded stable estimates despite the dataset’s size. Key findings are summarized in Tables 3 and 4.

Discussion

This nationwide study of ED visits (2018–2023) provides critical insights into emergency care utilization and hospital admission determinants in Jordan. ED attendance rose substantially over the study period, peaking before a sharp decline in 2020 during COVID-19 lockdowns, followed by a rebound in 2021– a pattern mirroring global trends [4]. Despite partial recovery, ED volumes in 2021–2023 remained below projected pre-pandemic levels [4], suggesting a lasting shift in healthcare-seeking behavior. Notably, only 6.4% of ED visits resulted in hospitalization, which is roughly half the rate reported in high-income settings (e.g. 13% in the U.S.) [23]. This discrepancy indicates that Jordan’s EDs manage a large load of lower-acuity cases on the front line, underscoring their role as a safety net for non-critical care. High ED reliance for non-admitted care has also been observed globally and raises concerns about primary care access [8]. Consistent with expectations, advanced triage acuity was the strongest predictor of admission: patients triaged Urgent had the highest admission odds, reflecting the intended prioritization of severely ill patients, as documented in international studies [24]. By contrast, those labeled Non-Urgent rarely required admission, reinforcing the effectiveness of triage in filtering low-risk cases. Intermediate triage categories (e.g. patients with ENT or general complaints) constituted a substantial portion of visits yet had moderate admission rates, suggesting that many moderate-acuity cases are managed and discharged from the ED. This points to potential inefficiencies or gaps in redirecting non-emergent cases to outpatient settings, a challenge also noted in other healthcare systems [8].

Patient demographics significantly influenced admission outcomes. Increasing age was strongly associated with higher hospitalization likelihood, which aligns with well-established patterns of greater comorbidity and complexity among older ED patients [25, 26]. Our findings accord with prior observations that elderly patients drive a disproportionate share of admissions [8], reaffirming the need for geriatric-focused care strategies. A modest gender difference was observed: female patients had slightly higher odds of hospitalization than males. Females accounted for a majority of ED visits and admissions in our data (53% of admissions vs 47% of visits), a finding that resonates with reports from other contexts where women utilize ED services more frequently [13, 27]. The reasons may include differences in health-seeking behavior, disease profiles (e.g. obstetric or gynecologic emergencies), or clinical decision thresholds, though the disparity in our study was small. Prior literature on sex differences in ED outcomes is mixed– some studies note higher acuity or mortality in males [28, 29]– but our results underscore that even subtle gender-related patterns in ED utilization warrant attention to ensure equitable care.

Operational and temporal factors also played a notable role. ED crowding was inversely associated with admission probability: during periods of high patient volume, the odds of hospitalization decreased. This suggests that when resources are strained, clinicians may adopt more conservative admission practices, possibly discharging patients who might otherwise be admitted. Such an effect is plausible given capacity pressures, and it is consistent with evidence that ED crowding can alter clinical decision-making and compromise optimal care [13]. Extensive research has linked overcrowding to delayed treatments and worse outcomes [13]; our findings add that it may also raise the threshold for admission, highlighting a systemic risk during peak demand. We also observed lower adjusted admission odds during evening and night shifts. Although unadjusted data showed fewer admissions at night (perhaps reflecting fewer available inpatient beds or on-call specialists after hours), this temporal difference disappeared after controlling for patient acuity. This implies the diurnal variation is largely due to case-mix– patients presenting overnight tend to be less severely ill– rather than a deliberate bias in admission decisions. It aligns with typical ED patterns where off-peak hours see proportionally more low-acuity visits. Nevertheless, the need for consistent decision standards across shifts is important, as is ensuring adequate nighttime staffing and access to inpatient services for those who do need admission [24].

Population background characteristics provided additional context to our results. In unadjusted analyses, nationality appeared influential: Syrian refugees had higher crude admission rates compared to Jordanians. This elevated hospitalization risk echoes regional reports that refugees often present with greater health needs due to barriers in accessing routine care[30]. Indeed, Jordan’s public hospitals have absorbed substantial healthcare demand from displaced Syrians [31]. However, in our adjusted models nationality was not a significant predictor, suggesting that the higher admission rates among Syrian patients were explained by factors like their age and acuity on presentation rather than differential treatment. In other words, Syrian refugees in our cohort tended to be sicker or older, which drove their higher likelihood of admission. Still, the very fact that many refugees reach the ED in worse condition underscores disparities in healthcare access. This finding aligns with broader evidence that refugee populations face challenges in primary care continuity, potentially leading to more frequent ED use and hospitalizations [30]. It reinforces the importance of strengthening community and preventive services for refugee communities to manage chronic conditions before they escalate to emergencies. We also found that patients with repeated ED visits (“frequent flyers”) had increased odds of hospitalization. This is in line with the profile of frequent ED users reported internationally– they often suffer from poorly controlled chronic illnesses or psychosocial issues that result in higher acuity presentations and greater need for inpatient care [32, 33]. Far from being “inappropriate” users, these patients typically have legitimate health needs and consume intensive resources, as noted in prior studies [8, 34]. Our data thus concur with the global literature depicting frequent ED attenders as a high-risk group, not simply a source of avoidable visits. Finally, a modest regional variation was observed: visits to hospitals in the Northern region had slightly higher adjusted admission odds than those in the Central region. While the difference was small, it may reflect underlying disparities in hospital resources or admission practices across regions. Such variation has been scarcely reported in the literature and may be specific to Jordan’s healthcare system (e.g. differing hospital capacities or referral patterns), meriting further local investigation.

Overall, our findings paint an integrated picture in which clinical acuity (triage level) and system factors (crowding, access to care) predominantly shape ED disposition outcomes. Each result finds correspondence in prior research– from the primacy of triage urgency [24] and the vulnerability of older adults [8] to the challenges of ED crowding [13] and refugee health inequities [30]– suggesting that the patterns observed in Jordan reflect broader phenomena in emergency medicine. Where our results diverged from expectations (e.g. the Resuscitation triage anomaly or a slight female predominance in admissions), they highlight important questions for further study and local quality improvement.

Implications

These findings carry significant implications for emergency care practice and health policy in Jordan and similar health systems. Optimizing triage processes should be a priority. The counterintuitive outcome for Resuscitation-level patients signals a need to audit and standardize triage criteria. Ensuring that triage categories accurately stratify patients by true acuity is essential for safety– any inconsistency in labeling critical cases could lead to under-triage [35]. Training programs and quality checks for triage staff are recommended to improve fidelity. Additionally, the large proportion of intermediate-acuity presentations suggests an opportunity to refine care pathways for non-urgent cases. Strengthening referral networks and outpatient clinics could help divert patients with primary care needs away from EDs. For example, expanded after-hours urgent care centers or enhanced triage screening to fast-track low-acuity patients to next-day clinics might reduce ED load. Our data show that the ED is often used to fill gaps in primary care; policymakers should recognize the ED’s role as a healthcare “barometer” and invest not only in emergency services but also in preventive and primary care to alleviate unnecessary ED utilization [30].

Managing ED crowding is another key implication. The inverse relationship between crowding and admissions implies that high census periods potentially compromise patient care decisions. Hospital administrators and health planners should implement strategies to mitigate crowding and maintain quality. These could include improving inpatient bed turnover to reduce boarding, implementing surge capacity protocols, and using real-time crowding metrics to trigger auxiliary support. In parallel, enhancing hospital capacity and staffing is crucial. The substantial absolute number of ED-related hospitalizations (over half a million during 2018–2023) demands adequate inpatient beds, especially for critical and acute cases. Health system planning should use these data to forecast needed inpatient capacity and allocate resources accordingly. For instance, knowing that ED admissions have grown in volume, hospitals may need to expand observation units or acute care wards to accommodate surges.

Our findings regarding frequent ED users call for targeted interventions. Patients with repeated ED visits likely have complex needs that are not met through standard care continuity. Implementing case management programs for high utilizers– for example, assigning care coordinators or “ED care plans” to patients who frequent the ED– could connect these individuals with appropriate outpatient services (such as chronic disease clinics, mental health services, or social support) and prevent avoidable returns [36]. Similar programs in other countries have shown success in improving outcomes for frequent ED visitors by addressing underlying issues (poorly controlled chronic illness, lack of social support, etc.). We recommend that hospital networks identify their top frequent attenders and develop multidisciplinary care plans, potentially including scheduled follow-up visits, social work consultations, or home care outreach. By proactively managing these patients outside the ED, the healthcare system can improve continuity of care [30].

Strengths and limitations

Strengths

This study benefits from a large-scale, multi-year dataset encompassing all ED visits to government (Ministry of Health) hospitals nationwide over nearly six years. With over 300,000 visits analyzed, the sample size provides a comprehensive and representative overview of emergency care demand in the Jordanian public sector. This broad scope enhances the generalizability of findings to similar contexts and lends robust statistical power for detecting even modest associations. The use of an integrated electronic health record system (Hakeem) across all included hospitals ensured standardized data capture. This consistency in recording key variables– from patient demographics and triage levels to timestamps for crowding assessment– strengthens the reliability of our measures. We were able to examine a wide array of factors (demographic, clinical, temporal, and operational) simultaneously, which is a testament to the richness of the dataset. Furthermore, rigorous analytical methods were employed. We utilized multivariable logistic regression with adjustments for potential confounders, and the model diagnostics indicated no significant issues (e.g. multicollinearity or overfitting) [24]. By applying penalized regression techniques suited for large datasets, we obtained stable estimates despite the enormous sample size. The consistency of our findings with known ED patterns (as discussed above) also attests to the validity of the results [30]. In sum, the study’s extensive coverage and methodological rigor provide confidence in the observed associations.

Limitations

Several limitations must be acknowledged. First, as a retrospective analysis of administrative data, our study is subject to the inherent constraints of such data. We lacked detailed clinical information such as comorbid conditions, precise diagnoses, or severity scores (beyond the triage acuity) for each patient. Important factors like the presence of chronic diseases, socioeconomic status, or follow-up outcomes (e.g. 30-day mortality) were not captured, which may confound or mediate the relationships we observed. For example, the reasons behind a “Resuscitation” patient not being admitted could not be discerned without clinical chart review. Second, while triage acuity provides a proxy for illness severity, it may be inconsistently applied; mis-triage could bias the associations with admission. Third, the study’s scope was limited to public sector hospitals. We excluded private hospitals and military hospitals from the analysis (as these either did not report data consistently or cater to different patient populations) [24]. Consequently, our findings pertain specifically to Jordan’s public ED system and may not fully generalize to the private healthcare sector or to other countries with different healthcare structures. Fourth, the period of analysis included the COVID-19 pandemic, an extraordinary disruption to healthcare utilization. While we have interpreted and adjusted for the pandemic’s impact on ED visits, the long-term trends could be influenced by pandemic-related shifts in behavior that are still evolving. Fifth, the outcome of interest was hospital admission from the ED, which we used as an indicator of case severity and resource need. However, this binary outcome does not capture other meaningful endpoints, such as ED revisit rates, inpatient length of stay, or mortality, which would provide a more holistic view of care quality. Finally, causality cannot be established due to the observational design; associations (e.g. between crowding and admission decisions) should not be interpreted as direct cause-and-effect without further prospective study. Despite these limitations, the study offers valuable epidemiological evidence. We mitigated some concerns by focusing on adjusted associations (reducing confounding) and highlighting where results align with established knowledge, but caution is warranted in interpretation.

Future research

Given the prominent role of frequent ED users in admissions, future research could also focus on this subgroup. Longitudinal designs following frequent attenders over time would help determine the underlying causes of their repeat visits and admissions. Do they have uncontrolled chronic conditions, poor outpatient follow-up, or social determinants (like homelessness) driving ED reliance? Answering these questions would support the design of targeted case management interventions. Trials or program evaluations of interventions (for example, assigning health navigators to frequent ED users) could measure impacts on subsequent ED use and hospitalization, thereby building an evidence base for best practices in Jordan’s context. Finally, further research should broaden the scope beyond admission as an outcome. Patient-centered outcomes such as satisfaction, functional recovery, or longer-term health status post-ED visit are critical to assess. Additionally, system-level outcomes like cost-effectiveness of potential interventions (e.g. expanding primary care vs. adding ED beds) would be highly informative for policy. By pursuing these lines of inquiry– from mechanistic studies of triage and decision-making to intervention trials for high-risk groups– researchers can help translate the findings of this study into actionable improvements in emergency care delivery.

Conclusion

In conclusion, this study offers a comprehensive assessment of ED visit characteristics and hospitalization outcomes in Jordanian public hospitals over 2018–2023. We documented a dramatic rise in ED utilization over time, tempered by the transient disruption of the COVID-19 pandemic, reflecting global emergency care trends [4]. Our analysis revealed that a complex interplay of patient factors (triage acuity, age, gender, repeat visits) and operational conditions (crowding levels, time of visit) influence the likelihood of hospital admission from the ED. Triage acuity emerged as the central determinant: patients deemed Urgent had the highest admission odds, whereas Non-Urgent cases rarely required inpatient care, affirming the triage system’s critical role in prioritizing limited resources. We also found that older adults and frequent ED visitors carry a disproportionately high burden of admissions, while female patients and refugee patients showed notable patterns that warrant culturally sensitive attention. High crowding conditions were associated with reduced admissions, underscoring how resource constraints can tangibly impact patient disposition decisions.

Together, these findings highlight both challenges and opportunities for Jordan’s health system. They underscore the ED’s dual role as a gateway to inpatient care and a safety net for broad community health needs. The study provides evidence to inform targeted improvements: enhancing triage accuracy, bolstering surge capacity to combat crowding, strengthening primary care to reduce non-urgent ED visits, and addressing the needs of vulnerable populations (such as refugees and chronic frequent-users). By implementing changes in these areas, health authorities can improve emergency care outcomes and system efficiency. In summary, our research not only quantifies the patterns of ED utilization and admissions in a Middle Eastern context but also contextualizes them against international literature and public health imperatives. We hope these insights will guide policymakers and clinicians in Jordan and similar settings in developing data-driven strategies to optimize emergency care delivery and ensure that the ED continues to fulfill its crucial role in the healthcare continuum [30].

Authors' contributions

Departments recorded 9.16 million visits, with 6.37% resulting in hospitalization. Common complaints included pain and fever, with hospitalization linked to severity and patient demographics. Key factors influencing outcomes were triage class, age, shift, and regional differences, providing crucial insights for optimizing ED management and resource allocation in Jordan. dataset, conducted preliminary statistical analyses, and contributed to the interpretation of textual data using advance text analysis. AA (Ahmad Alrawashdeh): Contributed to the conceptual design of the study, supervised clinical relevance of the emergency department data. All authors have read and approved the final manuscript and agree to be accountable for all aspects of the work.

Funding

No Funding.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

IRB form available upon request.

Consent for publication

NA.

Competing interests

The authors declare no competing interests.