Background: Overcrowding in Emergency departments adversely impacts efficiency, patient outcomes, and resource allocation. Accurate triage systems are essential for prioritizing care and optimizing resources. While traditional methods provide a foundation, they often lack precision in addressing modern healthcare complexities. Artificial intelligence (AI) and machine learning (ML) offer advanced capabilities to enhance triage accuracy, improve patient prioritization, and support clinical decision-making, addressing limitations of conventional approaches and paving the way for adaptive triage solutions. Objective: This systematic review aims to assess the use of artificial intelligence (AI) and machine learning (ML) in determining the outcomes of patients presenting in Emergency department (ED) triage. Methods: A systematic search was conducted on April 21, 2023, using electronic databases including PubMed/Medline, Cochrane Library, Ovid, and Google Scholar, without year restrictions. The main outcome of this review was to assess the use of AI and ML in the ED Triage. Articles that used different models of AI and ML to predict various outcomes of patients in the ED setting were included. Results: A total of 17 studies were included in this systematic review. Fifteen studies assessed the role of machine learning methods in emergency department triage, while two studies evaluated the role of AI and machine learning in prehospital triage. The results of our systematic review favor the use of machine learning methods and artificial intelligence in emergency triage. Machine learning models were found to be superior to conventional emergency severity score methods in determining triage, diagnosis, and early management of patients. Among the machine learning methods, the boosting model was slightly more effective. Conclusion: Our study supports the notion that AI and ML are the future of Emergency departments. They aid in predicting patient outcomes and determining appropriate management strategies more efficiently, thereby enhancing decision making in the ED.
The aim of this paper is to describe the methods and processes used to develop a theory-informed, dedicated serratus anterior plane block (SAPB) education and training program for Emergency Medicine clinicians. This program was informed by Kern's six-step approach to curriculum development. Firstly, expert consensus on education and training was developed using a modified Delphi process. Secondly, contemporary medical literature was incorporated with educational pedagogical theoretical frameworks including Bloom's taxonomy of educational objectives, backwards design, and constructive alignment, scaffolded learning, and segmented learning. Finally, these were aligned with existing national policies around regional anaesthesia in the Emergency Department. The resultant SAPB evidence-based education package is supported by the most current evidence for using the SAPB as a regional anaesthesia procedure for patients with rib fracture pain. It serves as an educational tool for clinicians already proficient in the principles of plane blocks with additional guidance on establishing local workshops for practical education as well as ongoing supervision and governance for this procedure. The development of a theory-informed, dedicated SAPB education and training program for Emergency Medicine clinicians required a balanced combination of clinical science and procedural expertise with educational theory. This provides a framework for future works when developing educational resources to introduce new clinical procedures into routine Emergency Medicine practice.
We examined whether the 116 emergency department transfers from residential aged care homes prevented by an intervention that provided on-site assessment and management following a fall led to a detectable shift in system-level emergency department utilisation. This 17-month before-and-after study was conducted at three metropolitan hospital emergency departments of a single health service in Melbourne, Australia. Monthly post-fall emergency department transfers from 108 residential aged care homes were compared before (1 May 2021 to 30 April 2022) and after (1 May 2022 to 30 September 2022) the intervention using two routinely collected administrative datasets. The primary analysis included patients eligible for the intervention. Sensitivity analyses examined patients ineligible for the intervention and a season-matched before period (1 May 2021 to 30 September 2021). Mean (SD) monthly post-fall emergency department transfers from residential aged care homes among patients eligible for the intervention were 92.2 (14.9) before and 74.2 (7.2) after implementation (mean difference [95% CI]: 18.0 [2.9, 33.1], p = 0.023; season-matched: 28.2 [10.3, 46.1], p = 0.007). Monthly post-fall emergency department transfers from residential aged care homes among patients ineligible for the intervention did not significantly differ before and after the intervention (mean difference [95% CI]: 1.8 [-3.0, 6.7], p = 0.434; season-matched: 1.8 [-5.1, 8.7], p = 0.566). We detected 78% (n ≈ 90) of the verified 116 emergency department transfers prevented. On-site assessment and management following a fall at residential aged care homes may reduce system-level emergency department utilisation. Examining sustainability and cost-effectiveness appears warranted.
Pressure on emergency departments is increasing, perpetuating the phenomenon known as ramping or ambulance ramping, where patients are forced to wait on stretchers in corridors until they are allocated a suitable treatment space. This once temporary response to emergency department crowding has now become the norm for many emergency departments. The ramping environment impacts the patient experience well beyond the undesirability of extended waiting times. Nevertheless, there remains little research regarding how patients experience the ramping environment and what might make their wait more comfortable. Semi-structured interviews were performed with patients who had waited at least an hour on the ramp at two Australian emergency departments. An inductive thematic analysis was undertaken to explore the patient experience while ramped and aspects of the built environment and care that participants perceived could better support their health and waiting experience. Twenty-six patients participated in the study. Four major themes and 13 subthemes were developed from the data. Participants described psychological distress, physical discomfort, an unsuitable built environment, and challenges to maintain physical health. They suggested modifications to the built environment and care received that may ameliorate some of their concerns. Ramping can create additional stresses to the patient experience. Making changes to the physical aspects of the built environment may be challenging in the short term; however, other aspects of the ramp were identified that could be considered to maximise patient-centred care.
We aimed to describe the profile of patients with low back pain who received pathology testing in emergency departments or after inpatient admission, and to describe the ordered tests. A retrospective study of electronic medical records from three emergency departments in Sydney, New South Wales, Australia, from January 2016 to October 2021, was undertaken. We included patients diagnosed with a lumbar spine condition at discharge from the emergency department and extracted their demographic and episode of care characteristics. Pathology tests were ordered in 23.8% of 15,300 episodes of care. Patients who received pathology testing were typically older, were female, had a non-English preferred language, required an interpreter, arrived by ambulance during working hours, had their condition triaged as urgent, had a diagnosis of a serious low back pain pathology, were admitted to inpatient wards, and had an increased length of stay. Full blood count, electrolytes, urea, creatinine (EUC) and liver function tests were the most ordered tests. The characteristics of patients with low back pain receiving pathology testing in emergency departments as well as those of the tests were described. The results suggest that guideline recommendations were partly followed, but an investigation into the appropriateness of pathology testing is needed to confirm this hypothesis and ensure the relevance of low back pain care in emergency departments.
Family, domestic and sexual violence (FDSV) is a global health issue affecting one in four women. Emergency departments (EDs) are often the first point of healthcare contact for people experiencing FDSV. Virtual emergency departments have been introduced in Australia to improve access to care and reduce pressure on physical EDs, but little is known about FDSV presentations in this setting. This study aimed to identify and describe FDSV presentations to the Victorian Virtual Emergency Department (VVED). A retrospective cohort study examined FDSV presentations to VVED between 1 July 2023 and 30 June 2024. Cases were identified using VVED, injury surveillance, and clinician documentation data. Records were filtered by age, sex and injury intent, followed by file review to confirm eligibility. Demographic and presentation characteristics were summarised using descriptive statistics. Twenty-one FDSV presentations were identified among 173,936 VVED presentations (0.012%). All were female, median age 37 years. Most were born in Australia (86%) and lived in socioeconomically disadvantaged areas (62%). Intimate partner violence was the most common form of FDSV (81%), and injuries above the clavicle (head, face and neck injuries) were the most frequent presentation (53%). Most patients were referred via healthcare professionals (76%). One-third were directed to physical EDs, while most were managed virtually with follow-up arranged through general practitioners or other healthcare providers. FDSV presentations were rarely identified in the virtual emergency setting, suggesting under-detection. Virtual EDs offer opportunities to support people experiencing violence, but improved clinician training, documentation and referral pathways are needed.
To synthesise evidence on opportunistic screening for asymptomatic sexually transmitted infections and bloodborne viruses (STI/BBV) in emergency departments (EDs), focusing on detection, public health outcomes and feasibility. An integrative review was conducted using Whittemore and Knafl's framework. PubMed, EBSCO, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), Cochrane, and Embase were searched (2012-2024). Eligible sources included quantitative, qualitative, mixed-methods and policy papers. Studies were appraised with the Mixed Methods Appraisal Tool (MMAT) and synthesised thematically. A total of 20 studies were included. Thematic synthesis identified four key domains: routine or opt-out ED screening increased case detection and was cost-effective; barriers included privacy concerns, workflow constraints and low risk perception; behaviour-based eligibility better aligned screening with actual risk factors, whereas population-based approaches often misclassified exposure; and implementation was facilitated by confidentiality safeguards, streamlined workflows, self-collection and electronic prompts. Routine, opportunistic asymptomatic STI/BBV screening embedded within emergency department workflows reliably increases case detection of Chlamydia trachomatis, Neisseria gonorrhoeae, syphilis, and HIV and is cost-effective. Behaviour-based eligibility better aligns screening with actual transmission risk than identity- or population-based stratification, and implementation strategies that safeguard confidentiality, support self-collection, deliver results discreetly, and integrate electronic prompts into routine workflows consistently improve acceptance and feasibility. For Australian emergency departments, the evidence supports adopting routine, opportunistic, behaviour-based screening as a feasible, high-yield public health intervention. Conclusions are bounded by the predominance of United States studies, limited longitudinal follow-up, and the use of a single reviewer.
This study examined the distribution of emergency departments relative to remoteness and socioeconomic disadvantage and explored service availability for children in non-metropolitan Western Australia. We conducted a spatial analysis using the Australian Government datasets. Child population counts and emergency facility locations were integrated into Quantum Geographic Information System and classified by remoteness using the Modified Monash Model (MMM) and by area-level socioeconomic disadvantage using the Index of Relative Socioeconomic Disadvantage (IRSD). Buffer zones were generated around each facility to quantify the proportion of children residing within specified distance thresholds. Service availability was assessed using facility-to-child and bed-to-child ratios. Emergency department availability declined with increasing remoteness: MMM5 contained the largest share of facilities (42.6%), followed by MMM6 (31.2%) and MMM7 (26.2%). However, ratios appeared more favourable in very remote areas, with facility-to-child ratios improving from 1:848 (MMM5) to 1:741 (MMM7). Bed availability showed a similar pattern, ranging from 1:83 in MMM5 to 1:60 in MMM7. Socioeconomic analyses revealed a concentration of facilities and bed capacity in the most disadvantaged areas, with IRSD 1 hosting 57.4% of all facilities and 35.8% of total bed capacity. Geographic and socioeconomic inequities in emergency care provision persist across rural and remote Western Australia. Apparent improvements in ratios in very remote areas are driven by low population density rather than greater service investment. These findings support a shift to needs-based paediatric emergency service planning, using remoteness and IRSD-stratified proximity and capacity measures to prioritise communities with the greatest access shortfalls.
Vaccine-preventable diseases (VPDs) are re-emerging worldwide, with declining vaccination coverage since the COVID-19 pandemic resulting in outbreaks of previously rare infections in Australasia. Emergency departments are increasingly at the forefront of recognising and managing these diseases while responding to their broader public health implications. This editorial examines the factors contributing to the resurgence of VPDs and declining immunisation rates, and their implications for ACEM training and EM practice. The accompanying articles in this issue provide practical perspectives on managing diphtheria and measles in the emergency department, addressing vaccine hesitancy and preparing for VPD-related examination questions.
The primary aim of this project was to test the feasibility of collating valid and useful ED data to inform the future establishment of a binational (i.e., Australia and Aotearoa New Zealand) clinical registry for emergency care. A secondary aim of this project was to utilise multi-site ED data to assess the crude impact of COVID-19 restrictions on overall ED presentations. This was a retrospective observational study using data routinely collected by EDs. The inclusion criteria were all patients presenting to participating EDs between 2017 and 2020. The primary outcome was data completeness. The secondary outcome was the number of ED presentations. Variables for which there was at least 80% data completeness across all participating sites were eligible for simple demonstrations of multi-jurisdictional analyses. There were data for 11 ED networks and 4,405,762 ED presentations. There were just nine variables which met the 80% completeness cutoff: network and site identification, birthdate, gender, First Nations self-identification, ED arrival date and time, ED arrival mode, triage category, ED triage date and time. Compared to the expected number of overall annual ED presentations, there was a reduction of 16% (95% CI 14% to 17%) in the actual number of overall ED presentations in 2020. Data completeness for routinely collected data collated from multiple sites was satisfactory for just a small number of variables. These findings inform the opportunities and challenges to establishing a binational clinical registry for emergency care.
Emergency departments are seeing an increase in mental health presentations across the globe; however, there are no standard mental health assessment models of care currently being used in emergency departments. This narrative review aims to search the literature and report the mental health models of care used globally. This narrative review conducted a search of 4 medical databases, Medline Complete, CINAHL, Embase and PsychINFO; the research team screened titles and abstracts and reviewed the full text. Risk of bias and quality assessment was conducted by 2 independent researchers. A total of 2461 articles were identified for title and abstract screening; of these, 84 were reviewed in full text and subsequently, 79 articles were excluded, resulting in 5 articles to be included in this narrative review. Two articles assessed the Psychiatric Assessment and Planning Unit (PAPU) model and reported on reduced length of stay; a third article reported a similar model and improved length of stay. Two articles reported models on sub-sample cohorts of mental health: individuals experiencing suicide ideation and individuals experiencing alcohol and substance use issues. The results of this narrative review highlight the lack of literature concerning models of care and mental health assessment in emergency departments. Of the few articles reviewed, a successful model of care and/or assessment requires a dedicated physical space in the emergency department (such as a PAPU), including mental health short stay beds and dedicated mental health staff.
Townsville, North Queensland, experienced record rainfall in the 2024-2025 wet season with an associated surge in the number of patients with melioidosis. A large proportion presented to the Townsville University Hospital (TUH) emergency department (ED), presenting an opportunity to assess the surge from an ED perspective. This was a retrospective case series of all patients with melioidosis that presented to TUH ED between 1 November 2024 and 30 April 2025. Cases for the entire region were sourced from the Statewide Reference Laboratory, and then individually assessed to ascertain whether they first presented to the TUH-ED. There were 57 patients with confirmed melioidosis that presented to the TUH ED in the 2024-2025 wet season, 40 males and 17 females. Risk factors for melioidosis were present in 51 (88%), with diabetes being the most common (47%). ED blood cultures were taken in 54 patients and were positive in 36 (67%). All patients were admitted, with a median length of stay of 11 days; 12 (21%) patients had an intensive care admission, and hospital mortality was eight (14%). Pulmonary involvement was evident in 79% and prostatic involvement in 30% of males. Meropenem was administered in ED to 53% overall, and 80% of those with severe illness. The surge of patients with melioidosis demonstrated a wide spectrum of infection foci and severity of illness. In melioidosis endemic regions following significant rainfall, we recommend to culture widely, administer meropenem for severe infections and consider melioidosis as a cause for unusual presentations.
EDs are not routine immunisation settings, yet they often provide care for patients whose immunisation needs may be unrecognised, unmet or not recently reviewed. This is particularly relevant for people who experience fragmented access to healthcare, vulnerability, language or cultural barriers, or who have questions or concerns about immunisation. Immunisation care in emergency departments should be opportunistic and proportionate rather than framed as universal vaccine delivery. For emergency clinicians, the practical task is to assess where a presentation reveals an immunisation-related need and respond in a way that fits the patient, clinical presentation and local service capacity. This may include administering an indicated vaccine, providing education, addressing concerns, documenting unmet need or supporting follow-up with relevant immunisation services in the community. Consistent with the National Immunisation Strategy 2025-2030, a team-based approach is important to support immunisation care in ED settings.
To analyse temporal performance drift and optimal retraining frequency for an ensemble machine learning model to predict inpatient admission from paediatric emergency department (ED) triage data. This study utilised 409,307 ED presentations from 1 July 2018 to 30 June 2024 at Perth Children's Hospital. An ensemble stacking model (XGBoost, TabNet, multi-layer perceptron and logistic regression base learners with a logistic regression meta-learner) incorporated structured triage features and tuned BioClinicalBERT-derived embeddings from free-text notes. The model ran prospectively through a 5-year rolling-window simulation, testing nine retraining cadences from weekly to triennial and a static model. Training, retraining and validation datasets were temporally separate and prior to the test set. Primary outcomes were discrimination via the area under the receiver operator characteristic (AUROC) and calibration as absolute mean daily bed error (AMDBE). Weekly retraining achieved a mean AUROC of 0.843 (SD 0.016) and AMDBE of 2.57 (SD 1.79) over the 5-year simulation. Fortnightly and monthly cadences were non-inferior (AMDBE 2.61 and 2.73), whereas longer intervals showed progressive calibration degradation (p < 0.001) and stable AUROC. Concept drift was most pronounced in the static model, with a mean AMDBE of 10.6 in 2024 compared to 1.79 for the weekly model. Notably, monthly retraining required only 25% of the weekly computational burden with non-inferior performance. Monthly, or more frequent, model retraining sustains discrimination and calibration for paediatric ED admission prediction. This effectively mitigated concept drift and enabled accurate simulated daily bed-demand forecasting, providing evidence to support the clinical testing of such modelling.
Optimal initial laryngoscope blade selection for paediatric emergency intubation is uncertain. We examined whether Miller or Macintosh blades are associated with differences in first-pass success (FPS) and adverse events in children < 5 years. Prospectively collected data from the Australia and New Zealand ED Airway Registry (ANZEDAR) between March 2010 and March 2024 were analysed for children under 5 years of age. We report demographics, FPS and adverse events by initial blade type. Multivariable models examined factors associated with FPS and hypoxia. Among 201 children, 88 (43.8%) were intubated with a Miller blade and 113 (56.2%) with a Macintosh blade. In unadjusted analyses, children intubated with Miller blades were younger (median 0.4 years, IQR 0.08-1.35 vs. median 1.6 years, IQR 0.75-2.00), p < 0.001), had lower FPS (63.6% vs. 80.5%; OR 0.42, 95% CI 0.22-0.80; p = 0.008) and had a higher incidence of hypoxia (33.0% vs. 17.7%; OR 2.28, 95% CI 1.19-4.46; p = 0.01) compared with children intubated with Macintosh blades. Hypotension rates did not differ. In this cohort of young children intubated in the ED, Macintosh blade use was associated with higher FPS and fewer hypoxic events compared with Miller blade. Age was a significant confounder and should be considered when choosing which laryngoscope blade to use in young children.
To compare emergency department (ED) utilisation and related costs between people with and without dementia in their last 3 years of life (including the year of death and two full years prior to the year of death). This retrospective study used linked data (2013-2015) from 8389 people with dementia and 7813 people without dementia who died in 2015 in Queensland, Australia. ED utilisation data from the Queensland public hospitals were linked to cost data from the National Hospital Costing Data Collection using the patient's unique identifier. Two-sample t-test, Chi-square test and Generalised Estimating Equations regression were used for data analysis. After adjusting for potential confounders, we show that people with dementia were more likely to have ED presentations (OR = 2.001, p < 0.01), short-stay unit admissions (OR = 1.435, p < 0.01) and arrive by ambulance (OR = 2.367, p < 0.01) than those without dementia. The average ED cost per episode for people with dementia is AUD 943.24 compared to AUD 912.82 for people without dementia, with a mean difference of AUD 30.43 (p < 0.01). ED costs for individuals with dementia were 3% higher, amounting to an estimated AUD 5 million for Australia's health sector in 2023. People with dementia have higher ED utilisation and costs than those without dementia. The results highlight the need for improved ED care models and targeted resource allocation to accommodate the complex needs of people with dementia.
Generative artificial intelligence (AI) is reshaping the way clinicians record their clinical notes. AI-scribe systems leverage generative AI capabilities to transcribe clinical encounters into draft clinical notes. In this study, we assessed clinician uptake and estimated modelled documentation time savings for an AI-scribe system in an emergency department (ED). ED physicians and trainees were provided access to an AI-scribe for 5 weeks. Data from the first week were excluded. The transcript of each presentation, the initial AI-generated clinical note and the final EMR clinical notes were used to calculate time to finalise AI-assisted notes. Forty ED consultants and 23 trainees accessed the AI-scribe. Over the study period, nine consultants (22.5%) and 11 registrars (48%) used the system. The AI-scribe was used for 248 ED presentations, including 185 (74.6%) by trainees and 63 (25.4%) by consultants. The system generated 44,489 words. Following clinician review, 18,140 words were added and 2274 were deleted prior to submitting the final clinical notes. For a clinician with an average typing speed, use of the AI-scribe resulted in a time saving of 7.1 h of documentation. This was reduced to 4.9 h for rapid typer clinicians. Overall, the initial AI-generated notes were modified on 1143 occasions. The most frequently revised section was the history of presenting illness (23.3%) followed by the management plan (22.9%). Uptake of AI-scribe was higher among trainees than consultants, and the platform achieved substantial time savings. Future studies are required to quantify real-time productivity gains over longer periods.
Across the globe, there has been a resurgence in vaccine-preventable diseases reflecting declining immunity, vaccine hesitancy and gaps in vaccine coverage. Vaccination was seen as one of the most successful public health interventions of the 20th century-as a consequence, many vaccine-preventable diseases such as measles and diphtheria were relegated to textbooks. With the re-emergence of vaccine-preventable diseases (VPDs), it is an urgent priority to equip emergency physicians with the knowledge to rapidly identify and manage potential cases and strengthen public health measures to curb transmission.
To evaluate camera-based, advanced sensor technology for contactless vital sign monitoring (versus standard monitoring) posttriage in adult emergency department patients. A single-site, prospective, observational study was conducted at Townsville University Hospital Emergency Department between May 2025 and August 2025. Adult (≥ 18 years) walk-in patients were eligible. A convenience sample of consenting participants was recruited. Two sets of vital signs-blood pressure, heart rate, respiratory rate, oxygen saturation, temperature-were obtained. The first set was measured immediately after triage using a dual-camera system comprising a thermographic camera and a second camera with proprietary software. The second set was obtained immediately posttriage using standard monitoring equipment. Reliability and acceptability were the primary and secondary outcomes respectively. Reliability (accuracy and precision) was determined through a Bland-Altman analysis to assess agreement between the two sets of vital signs. Acceptability of technology was assessed using a participant questionnaire. A total of 300 patients were recruited, of whom 288 had complete data on paired vital signs and were included in the Bland-Altman analysis. Minimal bias was observed across most vital signs, indicating limited systematic error. Respiratory rate, oxygen saturation and temperature showed the closest agreement, whereas systolic blood pressure and heart rate exhibited the widest limits of agreement. Acceptability was high, with most participants reporting comfort, ease of use and overall satisfaction. Compared with standard vital sign monitoring, camera-based technology demonstrated good accuracy but variable precision across vital signs. Further algorithm refinement is required before reliable clinical integration.
To describe the epidemiology, healthcare utilisation and outcomes of patients presenting to emergency departments (EDs) with chronic liver disease (CLD) in Queensland, Australia. This statewide data linkage study included adult patients with CLD-related diagnoses across 104 Queensland Health EDs between 1 January 2016 and 31 August 2023. Emergency, inpatient and mortality data were linked. Patients were stratified by cirrhosis status and decompensation. Outcome was 30-day mortality. Poisson regression assessed trends, and Cox regression evaluated mortality. Amongst 15,999,186 ED presentations, 23,578 (0.15%) were related to CLD, involving 11,961 patients. Presentations increased by 2% annually (IRR 1.02, 95% CI 1.02-1.03). Cirrhosis accounted for 18,735 presentations (79.5%). Overall, 20,312 presentations (86.1%) resulted in hospital admission, 918 (4.5%) were admitted to intensive care units (ICU), and 963 (4.1%) resulted in in-hospital death. Amongst patients with cirrhosis, 16,968 (90.6%) resulted in admission, 867 (5.1%) were admitted to ICU and 899 (4.8%) died in hospital. Predictors of 30-day mortality included cirrhosis (adjusted hazard ratio (aHR) 6.92, 95% CI 5.36-8.94), malignancy (aHR 3.21, 95% CI 2.90-3.55), hepatorenal syndrome (aHR 3.15, 95% CI 2.72-3.66), encephalopathy (aHR 2.03, 95% CI 1.78-2.32) and spontaneous bacterial peritonitis (aHR 1.47, 95% CI 1.20-1.80). Presentation to tertiary hospitals was associated with lower mortality (aHR 0.75, 95% CI 0.68-0.82). CLD-related ED presentations are increasing and place substantial demand on hospital services in Queensland. Decompensation events strongly predict mortality and healthcare utilisation. ED-initiated risk stratification and coordinated care models to improve outcomes for patients with cirrhosis require development and evaluation.