Johanna T. Meßner1
Melanie L. Conrad2,3
Dennis Freuer4
Christine Meisinger4
Fabian B. Fahlbusch2
Florian Weber1*
- 1Interdisciplinary Pediatric Emergency Department, Faculty of Medicine, University of Augsburg, Augsburg, Germany
- 2Neonatology and Pediatric Intensive Care, Faculty of Medicine, University of Augsburg, Augsburg, Germany
- 3Institute of Microbiology, Infectious Diseases and Immunology, Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin, Germany
- 4Epidemiology, Medical Faculty, University of Augsburg, University Hospital Augsburg, Augsburg, Germany
Overcrowding in pediatric emergency departments (PEDs) is an increasing global challenge. While adult emergency medicine has developed several validated measures to assess overcrowding, pediatric-specific methods remain scarce. In recent years, new approaches have emerged, including the first predictive models capable of anticipating crowding before it occurs. This Mini-Review provides a narrative synthesis of recent conceptual and methodological developments in the measurement and prediction of overcrowding in PEDs, building on the literature published since 2021. It conceptually examines unidimensional metrics, multidimensional scores, and emerging predictive models, emphasizing the shift from retrospective assessment to temporally oriented approaches and the need for pediatric-specific validation and multicenter evaluation.
1 Introduction
Overcrowding in emergency departments (EDs) is a growing global challenge that affects healthcare quality, patient safety, and staff well-being (1–6). Pediatric EDs (PEDs) differ substantially from adult settings due to lower admission rates (7–9), higher proportions of non-urgent visits (2, 8, 10, 11), and unique patient flow characteristics (5, 8, 12). These factors limit the direct transferability of adult-derived tools to pediatric contexts.
Reliable measurement of crowding is a prerequisite for effective intervention planning, however, most existing tools such as the National Emergency Department Overcrowding Scale (NEDOCS) or the Emergency Department Work Index (EDWIN) were designed for adult emergency medicine (3, 13). Their use in pediatric settings has only recently been tested, often revealing the need for adaptation. In 2021, the systematic review by Abudan et al. (1) emphasized that no standardized method existed to specifically measure PED overcrowding and called for predictive systems capable of early detection. Since then, two studies have proposed forecasting approaches that anticipate crowding—marking a paradigm shift from reactive detection to proactive management. This Mini-Review maps the development of these approaches from 2021 to 2025 as an updated extension of the synthesis by Abudan et al. (1).
2 Methods
This Mini-Review applies a narrative synthesis approach, informed by a structured literature search to support transparency and conceptual focus. A targeted literature search was conducted in PubMed, the Cochrane Library, and the Pediatric Emergency Care journal, covering publications from 2021 to 2025 using the search terms “overcrowding” and “pediatric emergency department.” The selected time window was chosen to capture recent conceptual and methodological developments following the systematic review by Abudan et al. (1). The search strategy was deliberately focused and time-restricted, with the aim of capturing recent conceptual and methodological advances in the assessment and prediction of PED overcrowding. Only peer-reviewed journal articles were considered; gray literature, conference abstracts, and unpublished reports were not included to ensure methodological transparency and consistency of reported metrics. Eligible study designs comprised observational studies, validation studies, and model development and validation studies, including predictive time-series and forecasting approaches. Articles that referred to overcrowding solely as a descriptive or contextual concept without specifying how crowding was operationalized or measured were excluded, as such studies do not allow meaningful comparison of overcrowding assessment or prediction approaches. Duplicates were removed prior to screening, and titles and abstracts were assessed for eligibility according to predefined inclusion criteria. Titles and abstracts were independently reviewed by two reviewers (JTM and FW) to identify publications of conceptual relevance, with full-text articles subsequently assessed for inclusion; any discrepancies were resolved through discussion and consensus. Given the narrative and concept-oriented scope of this Mini-Review, no formal risk-of-bias assessment was performed. Unidimensional metrics and predictive models were distinguished by their temporal orientation, as predictive models enable anticipatory operational planning through forecasts of future crowding—even when based on a single input variable such as patient census—rather than retrospective point-in-time assessment. Pediatric validation status was classified consistently as “pediatric-specific” for tools originally developed for PEDs, “adapted from adult tool” for modified adult-derived scores, and “evaluated in PEDs” for adult tools formally tested in pediatric settings. To enhance transparency of the literature selection process, a study selection flow diagram is provided (Figure 1). The approaches described across the identified publications were conceptually organized into three categories: (1) unidimensional metrics, (2) multidimensional scores, and (3) predictive models, drawing on a total of 13 publications addressing the measurement or prediction of overcrowding in PEDs. The focused and time-restricted nature of the literature search represents an inherent limitation of this Mini-Review. Relevant studies published outside the selected time frame or indexed in additional databases may not have been captured. In addition, restricting inclusion to peer-reviewed journal articles may have excluded emerging approaches reported only in conference proceedings or gray literature.
Figure 1. Study selection flow diagram for this narrative Mini-Review. Flow diagram illustrating the identification, screening, and inclusion of studies addressing the measurement or prediction of overcrowding in PEDs since 2021. The diagram is provided to enhance transparency of the literature selection process within the narrative synthesis approach applied in this Mini-Review.
3 Synthesis of recent evidence
Overcrowding assessment tools in pediatric emergency care were grouped into three conceptual categories: unidimensional metrics, multidimensional scores, and predictive models (Table 1). While the first two categories measure crowding retrospectively, the third category represents a move toward predictive management, representing a proactive evolution in managing patient flow and staffing resources.
Table 1. Overview of approaches used to assess or predict overcrowding in pediatric emergency departments, conceptually grouped into unidimensional metrics, multidimensional scores, and predictive models.
Unidimensional crowding indicators such as length of stay (LOS), left without being seen (LWBS), and occupancy rates remain the most widely used metrics for PED crowding (5, 14). Their advantage lies in simplicity and data accessibility. However, they capture only one aspect of the complex crowding phenomenon. Metrics like LOS and LWBS were originally developed for adult EDs and require pediatric validation (15). Other studies have proposed additional operational markers such as waiting time, patient-to-nurse ratios, and daily patient volume (11, 16). While these are easy to quantify, they remain retrospective and cannot predict upcoming surges in demand.
Multidimensional scores originally developed for adult emergency departments, such as NEDOCS (17, 18) and EDWIN (19, 20), have been evaluated in pediatric settings. These studies confirmed correlations with established crowding metrics. These tools integrate several operational parameters, including patient volume, waiting time, and resource use, into a composite index. PEDOCS (21) and SOTU-PED (Score Objectif de Tension dans les services d'Urgences pédiatriques; Objective Tension Score in Pediatric Emergency Departments) (22) were the first scoring systems tailored specifically to PEDS. Both aim to provide objective, real-time identification of crowding levels and to account for the unique patient flow patterns in children. The modified NEDOCS (mNEDOCS) replaces adult-specific variables with pediatric-relevant indicators, such as the number of patients in resuscitation rooms (18). While this adaption improves contextual relevance in pediatric settings, validation remains limited by small cohorts and site-specific calibration.
The most recent and conceptually transformative development is the emergence of predictive modelling for PED overcrowding. In 2022, Almeida et al. (23) developed a time-series model to predict daily PED visits and support proactive resource planning. In addition to using a daily patient census as the core variable, the authors incorporated calendar effects (school vs. non-school days) and weather data. The model was based on data from over 600.000 admissions between 2010 and 2017 at a public hospital in Lisbon, Portugal. The objective was to identify temporal patterns and assess the influence of school-calendar and weather-related factors in order to develop a forecasting model for daily patient volumes. The results of the Almeida study revealed clear cyclical patterns in patient volumes. For example, a strong annual cycle was observed, with peaks in January and February, primarily driven by respiratory infections. Another cycle, occurring roughly every four months, correlated with school holiday periods. Using these cyclical patterns, the authors developed a time-series forecasting model with a mean absolute percentage error (MAPE) of 10.7% ± 1.1% in cross-validation, indicating promising predictive performance.
In 2025, Akbasli et al. (24) expanded on this concept with SO-SAFED (Shift Optimization and System for Anticipating and Forecasting Emergency Department Crowding), an artificial intelligence (AI)–based system using the daily patient census to predict overcrowding in a PED before it occurs. By calculating crowding several hours in advance, SO-SAFED allows proactive staff allocation and operational planning. The model was built around a single variable: the daily number of patient visits. Using data from over 350.000 PED visits, the authors developed and tested 20 time-series models (ranging from traditional statistical approaches to advanced deep learning algorithms) to forecast patient volumes. From this study, it became evident that modern AI models provide significantly better predictions than traditional statistical methods. A key advantage of AI-based approaches is automated model maintenance through MLOps (Machine Learning Operations) architectures. This allows the model to be continuously updated with current data, to respond to data drift, and to always select the best performing model for any given situation. During simulation, average forecast accuracy improved from 44% to 60%. These forecasts were then used to optimize physician staffing shifts increasing the total number of employed physicians. Schedules were adjusted in 69 out of 84 shifts, which led to a reduction of the number of patients per physician during peak hours by over 4, contributing to better care, more efficient use of resources and reduced workload. The results showed that with only one input, the daily patient volume, the system could effectively forecast crowding and guide real-time staffing adjustments. This approach can also be seen as unidimensional, but unlike the other unidimensional metrics, it was used to build a predictive model.
Both of these new models (23, 24) represent a paradigm shift from detection to prediction in pediatric emergency care settings. Unlike retrospective scores and metrics, they enable early intervention to potentially prevent the adverse downstream effects of overcrowding.
4 Discussion
Most existing PED overcrowding tools remain reactive, identifying problems only after capacity has been exceeded. Multidimensional scores such as NEDOCS and EDWIN offer structured, quantitative assessment but rely heavily on adult-derived variables (3, 13, 25). Their pediatric adaptations, including mNEDOCS and PEDOCS, demonstrate potential yet require multicenter validation to confirm generalizability. Retrospective unidimensional metrics continue to dominate daily operational monitoring due to simplicity. However, their limited scope prevents a nuanced understanding of complex system dynamics.
The emergence of predictive models marks a conceptual shift in how PED overcrowding can be understood and managed. Importantly, this paradigm shift is defined by temporal orientation rather than by the number of variables included. Predictive models—particularly those by Almeida et al. (23) and Akbasli et al. (24)—apply temporal modeling to historical data to generate forward-looking estimates of crowding. This also explains why predictive approaches may rely on a single primary input variable, such as patient census, while still representing a fundamentally different conceptual framework from retrospective unidimensional metrics. Their integration into clinical workflows could transform capacity management by enabling data-driven staffing and surge preparedness.
Nonetheless, predictive models face several important limitations. Current predictive models for pediatric emergency department overcrowding rely on historical utilization patterns and may therefore be vulnerable to structural disruptions such as pandemics, policy changes, seasonal surges, or shifts in care-seeking behavior. In addition, prospective evidence demonstrating clinical impact on patient outcomes, safety metrics, or quality of care is currently lacking. From a methodological perspective, AI-based time-series models carry inherent risks of overfitting and data leakage, particularly in single-center datasets. Few studies have compared predictive approaches with simpler baseline models in real-world deployment, including algorithm-driven staffing decisions. Ethical considerations surrounding algorithm transparency and data governance also warrant attention. Moreover, many existing neonatal machine-learning models lack explainability, a critical barrier to clinician trust and regulatory acceptance (26–28). Consequently, prospective, multicenter validation and evaluation of real-world clinical impact will be essential to establish robustness, safety, and added value beyond simpler baseline approaches.
In addition to these methodological challenges, several ethical and human factors remain insufficiently explored. Beyond technical validation, future research should address the ethical and operational consequences of predictive overcrowding management, including potential effects on triage fairness, staff workload, and patient experience. Clear governance structures, transparency of algorithmic decision support, and defined human oversight will be essential, as will the development of secure, interoperable data infrastructures to support multicenter implementation. Integrating these aspects will be crucial to ensure that future predictive tools not only enhance operational efficiency but also uphold ethical and human-centered standards of pediatric emergency care.
5 Conclusion
Approaches to assessing overcrowding in PEDs are evolving from retrospective metrics toward predictive, temporally oriented models. While unidimensional and multidimensional scores remain important for monitoring current system strain, predictive approaches introduce a forward-looking perspective that may enable anticipatory operational planning. However, current evidence is limited to single-center studies, underscoring the need for multicenter validation and robust governance frameworks before broader clinical implementation.
Author contributions
JM: Writing – review & editing, Writing – original draft, Investigation, Visualization, Formal analysis, Data curation. MC: Writing – original draft, Writing – review & editing, Project administration. DF: Writing – review & editing, Writing – original draft, Project administration. CM: Project administration, Writing – original draft, Writing – review & editing. FF: Conceptualization, Resources, Writing – review & editing, Writing – original draft, Supervision. FW: Writing – review & editing, Funding acquisition, Writing – original draft, Conceptualization, Resources.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This research was supported by the B. Braun Foundation (Grant ID: BBST-D-25-00100). The funding organization had no role in the design of the study, the collection, analysis, or interpretation of data, or in the writing of the manuscript.
Acknowledgments
Data analysis of the present work was performed by Johanna Messner in partial fulfillment of the requirements for obtaining the degree “Dr. med.” at the University of Augsburg, Medical Faculty.
Conflict of interest
The author(s) declared that the research this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was used in the creation of this manuscript. OpenAI (ChatGPT) was used for language editing. The authors have reviewed and approved all changes made.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1. Abudan A, Merchant RC. Multi-dimensional measurements of crowding for pediatric emergency departments: a systematic review. Glob Pediatr Health. (2021) 8:2333794X21999153. doi: 10.1177/2333794X21999153
2. Sinclair D. Emergency department overcrowding—implications for paediatric emergency medicine. Paediatr Child Health. (2007) 12(6):491–4. doi: 10.1093/pch/12.6.491
3. Sartini M, Carbone A, Demartini A, Giribone L, Oliva M, Spagnolo AM, et al. Overcrowding in emergency department: causes, consequences, and solutions—a. Narrative Review. Healthcare. (2022) 10(9):1625. doi: 10.3390/healthcare10091625
4. Savioli G, Ceresa IF, Gri N, Bavestrello Piccini G, Longhitano Y, Zanza C, et al. Emergency department overcrowding: understanding the factors to find corresponding solutions. J Pers Med. (2022) 12(2):279. doi: 10.3390/jpm12020279
5. Menon NVB, Jayashree M, Nallasamy K, Angurana SK, Bansal A. Bed utilization and overcrowding in a high-volume tertiary level pediatric emergency department. Indian Pediatr. (2021) 58(8):723–5. doi: 10.1007/s13312-021-2278-3
6. Long CM, Mehrhoff C, Abdel-Latief E, Rech M, Laubham M. Factors influencing pediatric emergency department visits for low-acuity conditions. Pediatr Emerg Care. (2021) 37(5):265. doi: 10.1097/PEC.0000000000001553
7. Georgio G, Guttmann A, Doan QH. Emergency department flow measures for adult and pediatric patients in British Columbia and Ontario: a retrospective, repeated cross-sectional study. J Emerg Med. (2017) 53(3):418–26. doi: 10.1016/j.jemermed.2017.05.004
8. Rathlev NK, Holt NM, Harbertson CA, Hettler J, Reznek MA, Tsai SL, et al. 2017 AAAEM benchmarking survey: comparing pediatric and adult academic emergency departments. Pediatr Emerg Care. (2021) 37(12):e1278–84. doi: 10.1097/PEC.0000000000002002
9. Cag Y, Ozcetin M, Ozdemir AA, Elveren H. Evaluation of using pediatric emergency rooms. North Clin Istanb. (2018) 6(2):134–40. doi: 10.14744/nci.2018.79990
10. Ghadeer A, Aldandan HA, Alnajjar JK, Alamer JS, Almusallam MH, Alneamah SA, et al. Factors influencing visits to the pediatric emergency department. Cureus. (2024) 16(1):e51995. doi: 10.7759/cureus.51995
11. Adriani L, Dall’Oglio I, Brusco C, Gawronski O, Piga S, Reale A, et al. Reduction of waiting times and patients leaving without being seen in the tertiary pediatric emergency department: a comparative observational study. Pediatr Emerg Care. (2022) 38(5):219. doi: 10.1097/PEC.0000000000002605
12. Doan Q, Wong H, Meckler G, Johnson D, Stang A, Dixon A, et al. The impact of pediatric emergency department crowding on patient and health care system outcomes: a multicentre cohort study. CMAJ Can Med Assoc J. (2019) 191(23):E627–35. doi: 10.1503/cmaj.181426
13. Improta G, Majolo M, Raiola E, Russo G, Longo G, Triassi M. A case study to investigate the impact of overcrowding indices in emergency departments. BMC Emerg Med. (2022) 22:143. doi: 10.1186/s12873-022-00703-8
14. Sorz-Nechay A, Leeb F, Haas M, Djordjevic E, Yeghiazaryan L, Greber-Platzer S. Relieving the pediatric emergency department by referring low triaged patients using the Manchester triage system. Eur J Pediatr. (2025) 184(7):412. doi: 10.1007/s00431-025-06230-5
15. Hwang U, McCarthy ML, Aronsky D, Asplin B, Crane PW, Craven CK, et al. Measures of crowding in the emergency department: a systematic review. Acad Emerg Med. (2011) 18(5):527–38. doi: 10.1111/j.1553-2712.2011.01054.x
16. Frazier SB, Gay JC, Barkin S, Graham M, Walsh M, Carlson K. Pediatric emergency department to primary care transfer protocol: transforming access for patients’ needs. Healthcare. (2022) 10(3):100643. doi: 10.1016/j.hjdsi.2022.100643
17. Gorski JK, Arnold TS, Usiak H, Showalter CD. Crowding is the strongest predictor of left without being seen risk in a pediatric emergency department. Am J Emerg Med. (2021) 48:73–8. doi: 10.1016/j.ajem.2021.04.005
18. Jadcherla Y, Stoner M, Helwig S, Lo C, Shi J, MacDowell D, et al. Measuring overcrowding in a large academic tertiary care pediatric emergency department. Pediatr Emerg Care. (2024) 40(11):801–5. doi: 10.1097/PEC.0000000000003257
19. Kane A, Tackett S, Ngo T, Burkett GS, Wilson ME, Ryan LM, et al. Impact of an emergency department saturation tool on a pediatric hospital’s capacity management policy. Pediatr Emerg Care. (2023) 39(6):457–61. doi: 10.1097/PEC.0000000000002964
20. Chen SM, Stewart G, Sentman E, Helwig S, Rust L, Hoffman J, et al. Validation of the emergency department work Index in a pediatric freestanding community emergency department. Pediatr Emerg Care. (2024) 40(10):705–10. doi: 10.1097/PEC.0000000000003247
21. Weiss SJ, Ernst AA, Sills MR, Quinn BJ, Johnson A, Nick TG. Development of a novel measure of overcrowding in a pediatric emergency department. Pediatr Emerg Care. (2007) 23(9):641–5. doi: 10.1097/PEC.0b013e31814a69e2
22. Noel G, Jouve E, Fruscione S, Minodier P, Boiron L, Viudes G, et al. Real-Time measurement of crowding in pediatric emergency department: derivation and validation using consensual perception of crowding (SOTU-PED). Pediatr Emerg Care. (2021) 37(12):e1244–50. doi: 10.1097/PEC.0000000000001986
23. Almeida HS, Sousa M, Mascarenhas I, Russo A, Barrento M, Mendes M, et al. The dynamics of patient visits to a public hospital pediatric emergency department: a time-series model. Pediatr Emerg Care. (2022) 38(1):e240–5. doi: 10.1097/PEC.0000000000002235
24. Akbasli IT, Birbilen AZ, Teksam O. Artificial intelligence-driven forecasting and shift optimization for pediatric emergency department crowding. JAMIA Open. (2025) 8(2):ooae138. doi: 10.1093/jamiaopen/ooae138
25. 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: 10.2147/OAEM.S338079
26. Keles E, Bagci U. The past, current, and future of neonatal intensive care units with artificial intelligence: a systematic review. Npj Digit Med. (2023) 6(1):220. doi: 10.1038/s41746-023-00941-5
27. Helm K, Epp A, Gerstlauer M, Kaspar M, Hinske LC, Conrad ML, et al. Oxycardiorespirograms in neonatal monitoring—current gaps and future potential of artificial intelligence: a mini-review. Front Pediatr. (2025) 13:1680074. doi: 10.3389/fped.2025.1680074
Keywords: clinical decision support (CDS) systems, overcrowding assessment, pediatric emergency department (PED), predictive modeling, temporal demand analysis
Citation: Meßner JT, Conrad ML, Freuer D, Meisinger C, Fahlbusch FB and Weber F (2026) Mini-Review: From measurement to prediction—a conceptual paradigm shift in assessing overcrowding in pediatric emergency departments since 2021. Front. Pediatr. 13:1746637. doi: 10.3389/fped.2025.1746637
Received: 14 November 2025; Revised: 17 December 2025;
Accepted: 24 December 2025;
Published: 13 January 2026.
Edited by:
Neven Saleh, Future University in Egypt, EgyptReviewed by:
Ahmed Salaheldin, University of Hertfordshire, United KingdomCopyright: © 2026 Meßner, Conrad, Freuer, Meisinger, Fahlbusch and Weber. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Florian Weber, Rmxvcmlhbi5XZWJlckB1ay1hdWdzYnVyZy5kZQ==