Admission rates and outcomes of elderly patients in emergency department observation units. A Spanish multicentre study

1. Introduction

As the global population ages, older adults account for a growing proportion of emergency department (ED) visits [1, 2]. In many countries, patients aged ≥65 represent up to 30% of ED activity [3]. These patients often present with higher risk and complexity due to functional impairment, cognitive decline, polypharmacy, and multiple chronic conditions [4, 5].

Emergency Department Observation Units (EDOUs) offer short-term clinical observation and management for patients who require further assessment before a disposition decision. These units benefit patients whose condition may evolve or need additional testing or response monitoring. EDOUs typically operate with stays under 48 hours [4] and provide an alternative to inpatient admission when discharge is not immediately safe or feasible.

EDOUs may benefit older patients by avoiding unnecessary hospitalization while offering close observation. Geriatric conditions such as falls, delirium, syncope, heart failure, or respiratory infections often fall within the scope of EDOU care [4, 5, 6, 7, 8, 9, 10, 11]. Evidence suggests that EDOUs can improve patient flow, reduce costs, and maintain safety, particularly when combined with protocolized management [12, 13, 14, 15, 16, 17, 18].

Despite these potential advantages, few studies have examined how ED teams use EDOUs for older patients or how this decision affects outcomes [19, 20, 21]. Most prior research has focused on general adult populations or EDOU performance metrics rather than patient selection.

This study, therefore, aims to identify clinical and functional factors associated with the decision to manage patients aged ≥65 years in an EDOU. We also assess how EDOU use relates to key short-term outcomes, including hospitalization, ED revisits, and mortality. Our findings may support future development of evidence-based criteria for EDOU admission in geriatric emergency care.

2. Methods

3. Results

Table 1 presents the structural and functional characteristics of the participating EDOUs, providing context for their utilization.

The EDEN-20 study included 23,955 ED visits across 48 Spanish hospitals. Of these, 6393 visits (26.7%) involved patients admitted to an EDOU (EDOU group), while 17,562 (73.3%) did not (non-EDOU group) (Fig. 1). The median age of all patients was 78 years (IQR: 13); 10,433 patients (43.6%) were aged 80 or older, and 55% were women.

Fig. 1.Flowchart for patient inclusion in the EDEN-20 cohort. EDOU: Emergency Department Observation Unit; EDEN: Emergency Department and Elder Needs. Note: bold text is used to highlight key cohort definitions and main outcome categories.

The median ED length of stay was 8.3 hours (IQR: 15.22) for the EDOU group and 2.7 hours (IQR: 3.24) for the non-EDOU group. A total of 1078 patients (17%) in the EDOU group had ED stays longer than 24 hours (including time in the EDOU), compared to only 226 patients (1.3%) in the non-EDOU group (OR = 16.0; 95% CI: 14.0–18.0; p < 0.001).

Compared to non-EDOU patients, those in the EDOU group were older (median age 81 vs. 77 years; p < 0.001), with a higher proportion aged ≥80 years, a greater burden of comorbidity, worse functional capacity (as measured by the Barthel Index), more frequent mobility impairment, and higher rates of cognitive decline. Patients in the EDOU group more often arrived by ambulance and were more frequently referred by a physician or from another hospital. Table 2 shows these variables along with their unadjusted ORs and statistical significance.

The same table also presents data on vital signs. The presence of arterial hypotension, arterial hypertension, significant tachypnoea, tachycardia, bradycardia, hypoxemia, and a Glasgow Coma Scale (GCS) score <15 was significantly associated with EDOU admission. The strongest associations were observed for tachypnoea (OR = 4.0; 95% CI: 3.4–4.7) and GCS <15 (OR = 3.8; 95% CI: 3.2–4.6).

Regarding laboratory abnormalities (Table 2), erythrocytosis (OR = 2.5; 95% CI: 2.0–3.0) and anaemia (OR = 2.0; 95% CI: 1.7–2.4) showed the strongest associations with EDOU admission. Of the 31 variables analysed in the EDEN-20 dataset, 27 showed statistically significant differences between the EDOU and non-EDOU groups.

Table 3 displays the 10 most common ED diagnoses in each group. In the EDOU group, the leading diagnoses were: heart failure, pneumonia, urinary tract infection, chest pain, other specified respiratory disorders, atrial fibrillation and flutter, syncope and collapse, acute exacerbation of chronic obstructive pulmonary disease, unspecified lower respiratory infection, and cerebral infarction.

Multivariable logistic regression analysis, which included the 29 variables that differed significantly between groups, identified five independent predictors of EDOU admission: age ≥80 years, tachypnoea, bradycardia, GCS <15, and anaemia (Table 4, Fig. 2).

Fig. 2.Adjusted associations for variables analyzed in the EDEN-20 Study with admission in emergency department-observation units (EDOU). The multivariable model included only significant variables (p < 0.05) on univariate analysis. The odds ratio in bold numbers denotes statistically significant differences (p < 0.05). ED: Emergency Department; EDOU: Emergency Department Observation Unit; CI: Confidence Interval.

Regarding the outcomes, the need for hospital admission was 54% in the EDOU group compared to 15% in the non-EDOU group (aOR = 2.4; 95% CI: 2.0–2.8). The EDOU group also showed higher rates of 30-day rehospitalization (aOR = 1.6; 95% CI: 1.2–2.3). Notably, EDOU care showed a negative, but non-significant, association with 30-day ED revisit among discharged patients (aOR = 0.9; 95% CI: 0.7–1.2).

No significant associations were observed between EDOU use and prolonged hospital stay (aOR = 1.2; 95% CI: 0.9–1.5), 30-day hospital admission after ED discharge (aOR = 1.1; 95% CI: 0.9–1.4), in-hospital mortality (aOR = 1.0; 95% CI: 0.7–1.5), or all-cause 30-day mortality (aOR = 1.0; 95% CI: 0.8–1.4) (Table 5).

4. Discussion

This exploratory study reveals previously unreported sociodemographic, functional, and clinical characteristics of a representative sample of EDOUs in Spain. It also identifies factors associated with admission to EDOUs—an area scarcely addressed in the literature—and describes selected short-term outcomes following observation unit care.

ED visits by older adults are associated with higher rates of hospital admission, prolonged inpatient stays, unplanned readmissions, functional decline, and mortality [21, 25]. For many of these patients, EDOUs may offer a safe and efficient alternative to conventional admission, helping to avoid unnecessary hospitalization [18].

In our study, patients admitted to EDOUs were older, with greater comorbidity, functional dependence, and cognitive impairment than those who were not admitted to observation units. This suggests a more complex clinical profile, consistent with previous findings [15, 19].

Except for arterial hypertension, all other abnormal physical examination findings were significantly associated with the decision to admit to the EDOU (Table 4). These signs likely influenced clinicians during the triage and initial evaluation phases. Caterino et al. [26] found that among older patients in EDOUs, age >65 was not predictive of hospital admission and that hypertension was the only significantly associated vital sign. However, to our knowledge, no previous studies have examined the role of physical signs in predicting admission to an observation unit.

Regarding laboratory values, most abnormalities were associated with admission to an EDOU, except for leukopenia, thrombocytopenia, hypoglycaemia, and elevated lactate. Anaemia and erythrocytosis showed the strongest associations, possibly reflecting transfusion needs and exacerbated chronic obstructive pulmonary disease (COPD), respectively.

The analysis of emergency diagnoses (Table 3) showed that the five most common reasons for EDOU admission were heart failure, pneumonia, urinary tract infection, chest pain, and other specified respiratory disorders. Ross et al. [15] listed the 20 most common diagnoses among patients aged ≥65 admitted to EDOUs in the United States, including conditions such as dehydration, syncope, COPD, cellulitis, abdominal pain, atrial fibrillation, anaemia requiring transfusion, and social problems. All 10 most frequent diagnoses in our study are included within Ross’s top 20. Surprisingly, diagnoses like back pain, abdominal/pelvic pain, or trauma-related injuries were not prevalent in our cohort, possibly because these cases are often managed in standard ED consultation areas or referred directly to specialist care.

Our multivariable regression model identified five independent factors associated with EDOU admission: age ≥80 years, tachypnoea, bradycardia, GCS <15, and anaemia. Although not predictive, the model provides an explanatory framework to identify the clinical profile associated with the decision to admit to EDOUs. Understanding these patterns is key to standardizing practice and optimizing care for older adults in the ED.

Observation unit use among adult ED patients is estimated at around 10% overall [6], and approximately 2% for geriatric patients [15]. In our study, 27% of patients aged ≥65 were managed in an EDOU—a proportion rarely reported in the biomedical literature. Furthermore, 54% of hospital admissions occurred after an EDOU stay, compared with just 15% directly from the ED (p < 0.001). These figures highlight the central role of EDOUs in managing geriatric patients in Spain.

Previous studies in the U.S. and U.K. have reported inpatient admission rates from EDOUs among older adults ranging from 16% to 78% [15, 27, 28, 29, 30]. Our results fall within this range. In unadjusted analyses, 30-day hospital readmission after ED discharge (10.6% vs. 5.8%; OR = 2.0; 95% CI: 1.7–2.2) and 30-day all-cause mortality (9.2% vs. 3.0%; OR = 3.2; 95% CI: 2.9–3.7) were significantly higher in the EDOU group. However, these differences disappeared after adjusting for covariates in the regression model, except for a slight increase in 30-day rehospitalization after observation unit stay (11.3% vs. 10.8%; OR = 1.6; 95% CI: 1.2–1.3; p < 0.05).

In an extensive cohort study, Dharmarajan et al. [31] evaluated outcomes after observation stays in older adults and reported that 20% of hospital revisits and 50% of all subsequent hospitalizations occurred after an initial observation stay. The 30-day mortality rate in their study was 1.8%, much lower than ours, likely due to the younger age and lower comorbidity burden in their population.

More recently, Berger et al. [32] analysed ED return rates after EDOU discharge in a general population and found an overall revisit rate of 9.4%. In contrast, our cohort had a revisit rate of 24%, possibly reflecting the older age and higher complexity of our patients.

It is important to note that EDOU admission criteria were not standardized across centres. Variability in local protocols, unit structure, and bed availability likely influenced patient selection [33, 34, 35]. Therefore, the observed association between EDOU care and outcomes may partly reflect differences in patient complexity rather than the effect of EDOU management itself.

5. Limitations

Our findings should be interpreted in light of several limitations. First, the 48 participating EDOUs were not randomly selected; they volunteered to participate. Nevertheless, they represent 12 of Spain’s 17 autonomous communities, including university, high-complexity, and regional hospitals, minimizing potential selection bias.

Second, this analysis is based on a secondary use of a multipurpose registry. Therefore, some associations may be influenced by variables not captured in the original design.

Third, we did not model potential interactions between clinical variables (e.g., bradycardia and beta-blocker use, GCS, and baseline cognitive impairment). Similarly, due to data limitations, we could not apply frailty indices (e.g., Clinical Frailty Scale) or comorbidity scores (e.g., Charlson Index). Instead, we used the Barthel Index as a validated proxy for baseline functional status.

Fourth, we could not confirm whether any individual appeared more than once in the dataset. Although the inclusion period was short, the possibility of repeat visits cannot be excluded. We analysed outcomes per visit, and the data were extracted from electronic health records without patient identifiers.

Lastly, although we described the structural features of each EDOU (Table 1), we did not analyse whether unit type (e.g., Type I vs. Type II) or organizational model (e.g., open vs. closed) influenced outcomes. Future analyses using multilevel or hierarchical models are warranted.

This was an exploratory multicentre registry analysis, and we did not apply formal corrections for multiple comparisons. Although multivariable models were used to control for confounding, univariate associations should be interpreted cautiously. Future studies should confirm these findings in prospectively designed cohorts.

6. Conclusions

Observation units are increasingly used to manage older adults presenting to emergency departments. In this multicentre study involving 48 Spanish hospitals, we found that admission to an EDOU was associated with indicators of greater clinical complexity, such as advanced age, abnormal vital signs, impaired consciousness, and anaemia.

Short-term outcomes, including hospital admission and 30-day readmission, were more frequent among EDOU patients. However, these associations likely reflect patient complexity rather than adverse effects of EDOU care itself. EDOU use was not associated with increased mortality or prolonged hospital stays.

These findings underscore the need to establish standardized criteria for EDOU admission in geriatric emergency care. Developing evidence-based, protocol-driven strategies may improve patient selection and promote safer, more efficient use of observation resources.

Availability of data and materials

The data presented in this study are available on reasonable request from the corresponding author.

Author contributions

FJMP—designed the research study. FJMP, IBF, JGC, GBP, ÒM, CF, SA, AAA, JJ, PLL, OJTU, PEH, PPE, MCSP, MRG, MFPP, MGT, MBG, MVB, FLL, RMM, VTG, RJGH, JPG, CLA, MDP, MER, MIV, MRR, AMG—performed the research. FJMP, IBF, JGC, GBP, SA, ÒM—analyzed the data. FJMP, IBF, JGC, ÒM, SA—wrote the manuscript. All authors contributed substantially to the study conception, data acquisition, and data verification, critically revised the manuscript drafts, and read and approved the final version.

Ethics approval and consent to participate

The EDEN project received approval from the Clinical Research Ethics Committee of Hospital Clínico San Carlos de Madrid (protocol HCSC/22/005-E). Given the registry’s non-interventional nature, Spanish legislation allows centralized approval by a lead ethics committee, with required notification to local committees. The requirement for informed consent was waived by the same Ethics Committee due to the study’s retrospective, non-interventional design and the use of anonymized data. The study adhered to the principles outlined in the Declaration of Helsinki.

Acknowledgment

We thank all members of the SIESTA network for their commitment and valuable contributions to the development of this study.

SIESTA-Network Researchers:

Hospital Clínico San Carlos, Madrid: Juan González del Castillo, Cesáreo Fernández Alonso, Jorge García Lamberechts, Alejandro Melcon Villalibre, Sara Vargas Lobé, Laura Fernández García, Beatriz Escudero Blázquez, Estrella Serrano Molina, Julia Barrado Cuchillo.

Hospital Universitario Infanta Cristina, Parla: Ángel Iván Diaz Salado, Alicia Fuente Gafor io, Cristina Güemes de la Iglesia, Beatriz Honrado Galán.

Hospital Santa Tecla, Tarragona: Enrique Martín Mojarro, Sílvia Flores Quesada, Lidia Cuevas Jiménez.

Hospital Universitario de Canarias, Tenerife: Guillermo Burillo Putze, Aarati Vaswani-Bulchand, Patricia Eiroa-Hernández.

Hospital Norte Tenerife: Patricia Parra-Esquivel, Montserrat Rodríguez-Cabrera.

Hospital General Universitario Reina Sofía, Murcia: Pascual Piñera Salmerón, Lorena Bernabé Vera, Juan José López Pérez, Paula Lázaro Aragüés, Mariya Ovsepyan, M. Encarnación Sánchez Cánovas.

Hospital Universitario del Henares, Madrid: David Ampuero Martinich, Martín Ruiz Grinspan, Cristóbal Rodríguez Leal, Rocío Martínez Avilés.

Hospital Clínic, Barcelona: Òscar Mir, Sònia Jiménez, Sira Aguiló Mir, Francesc Xavier Alemany González, María Florencia Poblete Palacios, Claudia Lorena Amarilla Molinas, Ivet Gina Osorio Quispe, Sandra Cuerpo Cardeñosa.

Hospital General Universitario de Elche: Matilde González Tejera, Ana Puche Alcaraz, Cristina Chacón García.

Hospital Universitario y Politécnico La Fe, Valencia: Leticia Serrano Lázaro, Javier Millán Soria, Jésica Mansilla Collado, María Bóveda García.

Hospital Universitario Dr Balmis, Alicante: Pere Llorens Soriano, Adriana Gil Rodrigo, Begoña Espinosa Fernández, Mónica Veguillas Benito, Sergio Guzmán Martínez, Gema Jara Torres, María Caballero Martínez.

Hospital Universitario de Bellvitge, L’Hospitalet de Llobregat, Barcelona: Javier Jacob Rodríguez, Ferran Llopis, Elena Fuentes, Lidia Fuentes, Francisco Chamorro, Lara Guillen, Nieves López.

Hospital de Axarquía, Málaga: Coral Suero Méndez, Lucía Zambrano Serrano, Rocío Muñoz Martos, Rocío Lorenzo Álvarez.

Hospital Regional Universitario de Málaga: Manuel Salido Mota, Valle Toro Gallardo, Antonio Real López, Lucía Ocaña Martínez, Esther Muñoz Soler, Mario Lozano Sánchez, Eva María Fragero Blesa.

Hospital Santa Barbara, Soria: Fahd Beddar Chaib, Rodrigo Javier Gil Hernández.

Hospital Valle de los Pedroches, Pozoblanco, Córdoba: Jorge Pedraza García, Paula Pedraza Ramírez.

Hospital Universitario Reina Sofía, Córdoba: F. Javier Montero-Pérez, Carmen Lucena Aguilera, F. de Borja Quero Espinosa, Ángela María Cobos Requena, Esperanza Muñoz Triano, Inmaculada Bajo Fernández, María Calderón Caro, Sierra Bretones Baena.

Hospital Universitario Gregorio Marañón, Madrid: Juan Fernández Herranz, Marta Rincón Francés, Irene Arnaiz Fernández Esther Gargallo Garcia, Juan Antonio Andueza Lillo, Iria Miguens Blanco, Ioana Muñoz Betegón, Dariela Edith Micheloud Giménez.

Hospital Universitario de Burgos: Verónica Castro Jiménez, Lucía González Ferreira, Rocio Hernando González, María Eugenia Rodríguez Palma, Javier Varona Castrillo.

Complejo Asistencial Universitario de León: Albert Carbó Jordá, Enrique González Revuelta, Héctor Lago Gancedo, Miguel Moreno Martín, Alberto Alvárez Madrigal, Marta Iglesias Vela, Mónica Santos Orús.

Hospital Universitario Morales Meseguer, Murcia: Rafael Antonio Pérez-Costa, María Rodríguez Romero, Esperanza Marín Arranz, Ana Barnes Parra.

Hospital Francesc de Borja de Gandía, Valencia: Natalia Morera Sendra, Beatriz González García, Fátima Peiró Monzó.

Hospital Universitario Severo Ochoa, Leganés. Madrid: María José Hernández Martínez, Ana Benito Blanco, Vanesa Abad Cuñado, Julia Martínez-Ibarreta Zorita, Irene Cabrera Rodrigo.

Hospital Clínico Universitario Virgen Arrixaca, Murcia: Miguel Parra Morata Eva Quero Motto, Nuria Tomas Garcia, Laura Bernal Martínez.

Hospital Universitario Lorenzo Guirao, Cieza, Murcia: Alberto Artieda Larrañaga, José Joaquín Giménez Belló.

Hospital Universitario Dr. Josep Trueta, Girona: María Adroher Muñoz, Ester Soy Ferrer, Eduard Anton Poch Ferrer.

Hospital de Mendaro, Guipuzkoa: Jeong-Uh Hong Cho.

Hospital Universitario Miguel Servet, Zaragoza: Patricia Trenc Español, Fernando López López, Jorge Navarro Calzada, Belén Gros Bañeres, Cristina Martín Durán, María Teresa Escolar Martínez-Berganza, Iciar González Salvatierra.

Hospital Comarcal El Escorial, Madrid: Sara Gayoso Martín.

Hospital Do Salnes, Villagarcia de Arosa, Pontevedra: María Goretti Sánchez Sindín.

Hospital de Barbanza, Ribeira, A Coruña: Martina Silva Penas, Azucena Prieto Zapico.

Hospital del Mar, Barcelona: Bárbara Gómez y Gómez, Isabel Cirera Lorenzo, Patricia Gallardo Vizcaíno, Margarita Puiggali Ballard.

Hospital Santa Creu y Sant Pau, Barcelona: Aitor Alquezar Arbé, Carlos Romero Carrete, Sergio Pérez Baena, Laura Lozano Polo, Roser Arenos Sambro, José María Guardiola Tey, Carme Beltrán Vilagrasa.

Hospital de Vic, Barcelona: Lluís Llauger.

Hospital Valle del Nalón, Langreo, Asturias: Ángel Peláez González, Celia Rodríguez Valles.

Hospital Altagracia, Manzanares, Cuidad Real: Laura Márquez Quero.

Hospital Nuestra Señora del Prado de Talavera de la Reina, Toledo: Laura Molina, Mónica Cañete, Ricardo Juárez González.

Hospital Universitario Vinalopó, Elche, Alicante: María Martínez Juan María José Blanco Hoffman, Pedro Ruiz Asensio.

Hospital de Móstoles, Madrid: Fátima Fernández Salgado, Eva de las Nieves Rodríguez, Gema Gómez García.

Hospital Virgen del Rocío, Sevilla: Rafaela Ríos Gallardo, Teresa Pablos Pizarro, Mariano Herranz García, Laura Redondo Lora, Ana Gómez Caminero, Claudio Bueno Mariscal, Amparo Fernández-Simón Almela, Esther Pérez García.

Hospital General Universitario Dr. Peset. Valencia. María Amparo Berenguer Diez. María Ángeles de Juan Gómez. María Luisa López Grima, Rigoberto Jesús del Rio Navarro.

Hospital Universitario Son Espases. Palma de Mallorca: Núria Perelló Viola. Bernardino Comas Diaz, Sandra Guiu Martí, Juan Domínguez Casasola.

Clínica Universitaria Navarra, Madrid: Nieves López-Laguna.

Hospital Clínico Universitario de Valencia: José J. Noceda Bermejo, María Teresa Sánchez Moreno, Raquel Benavent Campos, Jacinto García Acosta, Alejandro Cortés Soler.

Hospital Alvaro Cunqueiro, Vigo: María Teresa Maza Vera, Raquel Rodríguez Calveiro, Paz Balado Dacosta, Violeta Delgado Sardina, Emma González Nespereira, Carmen Fernández Domato, Elena Sánchez Fernández-Linares.

Hospital Universitario de Salamanca: Ángel García García, Francisco Javier Diego Robledo, Manuel Ángel Palomero Martín, Jesús Ángel Sánchez Serrano.

Hospital de Zumarraga, Gipuzkoa: Patxi Ezponda.

Hospital Virxe da Xunqueira, La Coruña: Andrea Martínez Lorenzo.

Hospital Universitario Río Hortega, Valladolid: Inmaculada García Rupérez, Pablo González Garcinuño, Raquel Hernando Fernández, José Ramón Oliva Ramos, Virginia Carbajosa Rodríguez.

Hospital Central Asturias: Alba Martínez Alonso, Angela López Carrillo, Belén Pérez Fernández, Carmen Pérez Fonseca, Claudia Corugedo Ovies, Claudia Marinero Noval, Eugenia Prieto Piquero.

Hospital Juan Ramón Jiménez, Huelva: Asumpta Ruiz Aranda, María José Marchena, José María Santos Martin.

Funding

This research received no external funding.

Conflict of interest

The authors declare no conflict of interest.