Patient flow in emergency departments: a comprehensive umbrella review of solutions and challenges across the health system.

BACKGROUND: Globally, emergency departments (EDs) are overcrowded and unable to meet an ever-increasing demand for care. The aim of this study is to comprehensively review and synthesise literature on potential solutions and challenges throughout the entire health system, focusing on ED patient flow. METHODS: An umbrella review was conducted to comprehensively summarise and synthesise the available evidence from multiple research syntheses. A comprehensive search strategy was employed in four databases alongside government or organisational websites in March 2023. Gray literature and reports were also searched. Quality was assessed using the JBI critical appraisal checklist for systematic reviews and research syntheses. We summarised and classified findings using qualitative synthesis, the Population-Capacity-Process (PCP) model, and the input/throughput/output (I/T/O) model of ED patient flow and synthesised intervention outcomes based on the Quadruple Aim framework. RESULTS: The search strategy yielded 1263 articles, of which 39 were included in the umbrella review. Patient flow interventions were categorised into human factors, management-organisation interventions, and infrastructure and mapped to the relevant component of the patient journey from pre-ED to post-ED interventions. Most interventions had mixed or quadruple nonsignificant outcomes. The majority of interventions for enhancing ED patient flow were primarily related to the 'within-ED' phase of the patient journey. Fewer interventions were identified for the 'post-ED' phase (acute inpatient transfer, subacute inpatient transfer, hospital at home, discharge home, or residential care) and the 'pre-ED' phase. The intervention outcomes were aligned with the aim (QAIM), which aims to improve patient care experience, enhance population health, optimise efficiency, and enhance staff satisfaction. CONCLUSIONS: This study found that there was a wide range of interventions used to address patient flow, but the effectiveness of these interventions varied, and most interventions were focused on the ED. Interventions for the remainder of the patient journey were largely neglected. The metrics reported were mainly focused on efficiency measures rather than addressing all quadrants of the quadruple aim. Further research is needed to investigate and enhance the effectiveness of interventions outside the ED in improving ED patient flow. It is essential to develop interventions that relate to all three phases of patient flow: pre-ED, within-ED, and post-ED.

Towards Accurate Search for Neonatal Heartbeat: Weighted Algorithm for Reliable ECG Analysis of Premature Infants.

Accurate identification of the QRS complex is critical to analyse heart rate variability (HRV), which is linked to various adverse outcomes in premature infants. Reliable and accurate extraction of HRV characteristics at a large scale in the neonatal context remains a challenge. In this paper, we investigate the capabilities of 15 state-of-the-art QRS complex detection implementations using two real-world preterm neonatal datasets. As an attempt to improve the accuracy and reliability, we introduce a weighted ensemble-based method as an alternative. Obtained results indicate the superiority of the proposed method over the state of the art on both datasets with an F1-score of 0.966 (95% CI 0.962-0.97) and 0.893 (95% CI 0.892-0.894). This motivates the deployment of ensemble-based methods for any HRV-based analysis to ensure robust and accurate QRS complex detection.

Innovative Implemented Tools for Outpatient Clinic Scheduling.

Every year there are approximately three million new specialist clinic appointments at local hospital networks in Victoria. CSIRO, in collaboration with Austin Health, have developed two algorithms to assist with waitlist management in their outpatient specialist clinics. This study describes the implementation of these algorithms in software tools developed to support their use and trial in the clinical setting at Austin Health. We discuss the system design and development of both these software tools. We also review the implemented workflow of the tools and discuss how these tools seek to improve current systems.

Modelling Planned vs. Actual Start Time to Control the Efficiency of Surgery.

To control the efficiency of surgery, it is ideal to have actual starting times of surgical procedures coincide with their planned start time. This study analysed over 4 years of data from a large metropolitan hospital and identified factors associated with surgery commencing close to the planned starting time via statistical modelling. A web application comprising novel visualisations to complement the statistical analysis was developed to facilitate translational impact by providing theatre administrators and clinical staff with a tool to assist with continuous quality improvement.

An Approach for Generating Realistic Australian Synthetic Healthcare Data.

Healthcare data is a scarce resource and access is often cumbersome. While medical software development would benefit from real datasets, the privacy of the patients is held at a higher priority. Realistic synthetic healthcare data can fill this gap by providing a dataset for quality control while at the same time preserving the patient's anonymity and privacy. Existing methods focus on American or European patient healthcare data but none is exclusively focused on the Australian population. Australia is a highly diverse country that has a unique healthcare system. To overcome this problem, we used a popular publicly available tool, Synthea, to generate disease progressions based on the Australian population. With this approach, we were able to generate 100,000 patients following Queensland (Australia) demographics.

Elucidating Discrepancy in Explanations of Predictive Models Developed Using EMR.

The lack of transparency and explainability hinders the clinical adoption of Machine learning (ML) algorithms. While explainable artificial intelligence (XAI) methods have been proposed, little research has focused on the agreement between these methods and expert clinical knowledge. This study applies current state-of-the-art explainability methods to clinical decision support algorithms developed for Electronic Medical Records (EMR) data to analyse the concordance between these factors and discusses causes for identified discrepancies from a clinical and technical perspective. Important factors for achieving trustworthy XAI solutions for clinical decision support are also discussed.

A Step Towards Building Health Digital Twins: Patient Phenotype Representation for Health Outcome Prediction.

Precision medicine aims to provide more effective interventions and preventive options to patients by considering their individual risk factors and by employing available evidence. This proof of concept study presents an approach towards generating holistic virtual representations of patients, a.k.a. health digital twins. The developed virtual representations were applied in two health outcome prediction case studies for readmission and in-hospital mortality predictions. The results demonstrated the effectiveness of the virtual representations to facilitate predictive analysis in practicing precision medicine.

Emergency Department Demand and the First Year of the COVID-19 Pandemic.

We present a retrospective analysis of Emergency Department daily patient flow across 84 hospitals in Queensland, Australia over a four-year period from 2017 - 2020, leading up to and including the start of the COVID-19 pandemic. Daily ED demand significantly increased year-on-year over the study period, though significant increases in 2020 were likely attributed to ED fever screening clinics. Compliance against a four-hour ED Length of Stay target had been slightly decreasing since 2017, and the first year of the pandemic showed significant improvements in target compliance compared to previous years for all patients including the cohort admitted from ED. The length of stay for ED patients was also significantly less in 2020 (mean = 3.1 hours) compared to previous years. As an area of topical interest, a special focus on influenza-like illness presentations to ED helps quantify changes in volume of this cohort. This knowledge assists hospitals in planning and responding to variations in hospital demand.

Improving preoperative prediction of surgery duration.

BACKGROUND: Operating rooms (ORs) are one of the costliest units in a hospital, therefore the cumulative consequences of any kind of inefficiency in OR management lead to a significant loss of revenue for the hospital, staff dissatisfaction, and patient care disruption. One of the possible solutions to improving OR efficiency is knowing a reliable estimate of the duration of operations. The literature suggests that the current methods used in hospitals, e.g., a surgeon's estimate for the given surgery or taking the average of only five previous records of the same procedure, have room for improvement. METHODS: We used over 4 years of elective surgery records (n = 52,171) from one of the major metropolitan hospitals in Australia. We developed robust Machine Learning (ML) approaches to provide a more accurate prediction of operation duration, especially in the absence of surgeon's estimation. Individual patient characteristics and historic surgery information attributed to medical records were used to train predictive models. A wide range of algorithms such as Extreme Gradient Boosting (XGBoost) and Random Forest (RF) were tested for predicting operation duration. RESULTS: The results show that the XGBoost model provided statistically significantly less error than other compared ML models. The XGBoost model also reduced the total absolute error by 6854 min (i.e., about 114 h) compared to the current hospital methods. CONCLUSION: The results indicate the potential of using ML methods for reaching a more accurate estimation of operation duration compared to current methods used in the hospital. In addition, using a set of realistic features in the ML models that are available at the point of OR scheduling enabled the potential deployment of the proposed approach.

Development of an Interactive Dashboard to Analyse Physiological Signals in the Neonatal Intensive Care Unit.

Premature babies and those born with a medical condition are cared for within the neonatal intensive care unit (NICU) in hospitals. Monitoring physiological signals and subsequent analysis and interpretation can reveal acute and chronic conditions for these neonates. Several advanced algorithms using physiological signals have been built into existing monitoring systems to allow clinicians to analyse signals in real time and anticipate patient deterioration. However, limited enhancements have been made to interactively visualise and adapt them to neonatal monitoring systems. To bridge this gap, we describe the development of a user-friendly and interactive dashboard for neonatal vital signs analysis written in the Python programming language where the analysis can be performed without prior computing knowledge. To ensure practicality, the dashboard was designed in consultation with a neonatologist to visualise electrocardiogram, heart rate, respiratory rate and oxygen saturation data in a time-series format. The resulting dashboard included interactive visualisations, advanced electrocardiogram analysis and statistical analysis which can be used to extract important information on patients' conditions.Clinical Relevance- This will support the care of preterm infants by allowing clinicians to visualise and interpret physiological data in greater granularity, aiding in patient monitoring and detection of adverse conditions. The detection of adverse conditions could allow timely and potentially life-saving interventions for conditions such as sepsis and brain injury.

Predicting Disengagement to Better Support Outcomes in a Web-Based Weight Loss Program Using Machine Learning Models: Cross-Sectional Study.

BACKGROUND: Engagement is key to interventions that achieve successful behavior change and improvements in health. There is limited literature on the application of predictive machine learning (ML) models to data from commercially available weight loss programs to predict disengagement. Such data could help participants achieve their goals. OBJECTIVE: This study aimed to use explainable ML to predict the risk of member disengagement week by week over 12 weeks on a commercially available web-based weight loss program. METHODS: Data were available from 59,686 adults who participated in the weight loss program between October 2014 and September 2019. Data included year of birth, sex, height, weight, motivation to join the program, use statistics (eg, weight entries, entries into the food diary, views of the menu, and program content), program type, and weight loss. Random forest, extreme gradient boosting, and logistic regression with L1 regularization models were developed and validated using a 10-fold cross-validation approach. In addition, temporal validation was performed on a test cohort of 16,947 members who participated in the program between April 2018 and September 2019, and the remaining data were used for model development. Shapley values were used to identify globally relevant features and explain individual predictions. RESULTS: The average age of the participants was 49.60 (SD 12.54) years, the average starting BMI was 32.43 (SD 6.19), and 81.46% (39,594/48,604) of the participants were female. The class distributions (active and inactive members) changed from 39,369 and 9235 in week 2 to 31,602 and 17,002 in week 12, respectively. With 10-fold-cross-validation, extreme gradient boosting models had the best predictive performance, which ranged from 0.85 (95% CI 0.84-0.85) to 0.93 (95% CI 0.93-0.93) for area under the receiver operating characteristic curve and from 0.57 (95% CI 0.56-0.58) to 0.95 (95% CI 0.95-0.96) for area under the precision-recall curve (across 12 weeks of the program). They also presented a good calibration. Results obtained with temporal validation ranged from 0.51 to 0.95 for area under a precision-recall curve and 0.84 to 0.93 for area under the receiver operating characteristic curve across the 12 weeks. There was a considerable improvement in area under a precision-recall curve of 20% in week 3 of the program. On the basis of the computed Shapley values, the most important features for predicting disengagement in the following week were those related to the total activity on the platform and entering a weight in the previous weeks. CONCLUSIONS: This study showed the potential of applying ML predictive algorithms to help predict and understand participants' disengagement with a web-based weight loss program. Given the association between engagement and health outcomes, these findings can prove valuable in providing better support to individuals to enhance their engagement and potentially achieve greater weight loss.

Mortality and readmission differences associated with after-hours hospital admission: A population-based cohort study in Queensland Australia.

Background and Aims: Policy makers and health system managers are seeking evidence on the risks involved for patients associated with after-hours care. This study of approximately 1 million patients who were admitted to the 25 largest public hospitals in Queensland Australia sought to quantify mortality and readmission differences associated with after-hours hospital admission. Methods: Logistic regression was used to assess whether there were any differences in mortality and readmissions based on the time inpatients were admitted to hospital (after-hours versus within hours). Patient and staffing data, including the variation in physician and nursing staff numbers and seniority were included as explicit predictors within patient outcome models. Results: After adjusting for case-mix confounding, statistically significant higher mortality was observed for patients admitted on weekends via the hospital's emergency department compared to within hours. This finding of elevated mortality risk after-hours held true in sensitivity analyses which explored broader definitions of after-hours care: an "Extended" definition comprising a weekend extending into Friday night and early Monday morning; and a "Twilight" definition comprising weekends and weeknights.There were no significant differences in 30-day readmissions for emergency or elective patients admitted after-hours. Increased mortality risks for elective patients was found to be an evening/weekend effect rather than a day-of-week effect. Workforce metrics that played a role in observed outcome differences within hours/after-hours were more a time of day rather than day of week effect, i.e. staffing impacts differ more between day and night than the weekday versus weekend. Conclusion: Patients admitted after-hours have significantly higher mortality than patients admitted within hours. This study confirms an association between mortality differences and the time patients were admitted to hospital, and identifies characteristics of patients and staffing that affect those outcomes.

Epidemiology and outcomes of early-onset AKI in COVID-19-related ARDS in comparison with non-COVID-19-related ARDS: insights from two prospective global cohort studies.

BACKGROUND: Acute kidney injury (AKI) is a frequent and severe complication of both COVID-19-related acute respiratory distress syndrome (ARDS) and non-COVID-19-related ARDS. The COVID-19 Critical Care Consortium (CCCC) has generated a global data set on the demographics, management and outcomes of critically ill COVID-19 patients. The LUNG-SAFE study was an international prospective cohort study of patients with severe respiratory failure, including ARDS, which pre-dated the pandemic. METHODS: The incidence, demographic profile, management and outcomes of early AKI in patients undergoing invasive mechanical ventilation for COVID-19-related ARDS were described and compared with AKI in a non-COVID-19-related ARDS cohort. RESULTS: Of 18,964 patients in the CCCC data set, 1699 patients with COVID-19-related ARDS required invasive ventilation and had relevant outcome data. Of these, 110 (6.5%) had stage 1, 94 (5.5%) had stage 2, 151 (8.9%) had stage 3 AKI, while 1214 (79.1%) had no AKI within 48 h of initiating invasive mechanical ventilation. Patients developing AKI were older and more likely to have hypertension or chronic cardiac disease. There were geo-economic differences in the incidence of AKI, with lower incidence of stage 3 AKI in European high-income countries and a higher incidence in patients from middle-income countries. Both 28-day and 90-day mortality risk was increased for patients with stage 2 (HR 2.00, p < 0.001) and stage 3 AKI (HR 1.95, p < 0.001). Compared to non-COVID-19 ARDS, the incidence of shock was reduced with lower cardiovascular SOFA score across all patient groups, while hospital mortality was worse in all groups [no AKI (30 vs 50%), Stage 1 (38 vs 58%), Stage 2 (56 vs 74%), and Stage 3 (52 vs 72%), p < 0.001]. The time profile of onset of AKI also differed, with 56% of all AKI occurring in the first 48 h in patients with COVID-19 ARDS compared to 89% in the non-COVID-19 ARDS population. CONCLUSION: AKI is a common and serious complication of COVID-19, with a high mortality rate, which differs by geo-economic location. Important differences exist in the profile of AKI in COVID-19 versus non-COVID-19 ARDS in terms of their haemodynamic profile, time of onset and clinical outcomes.

A discrete event simulation for improving operating theatre efficiency.

BACKGROUND: Increasing demand in healthcare services has posed excessive burden on healthcare professionals and hospitals with finite capacity. Operating theatres are critical resources within hospitals that can become bottlenecks in patient flow during high demand conditions. There are substantial costs associated with running operating theatres that include keeping professional staff ready, maintaining operating theatres and equipment, environmental services and cleaning of operating theatres and recovery rooms, and these costs can increase if theatres are not used efficiently. In addition to cost, operating theatre inefficiency can result in surgery cancelations and delays, and eventually, poor patient outcomes, which can be exacerbated under the increase in demand. METHODS: The allocation of surgeries to operating theatres is explored using a simulation model for patients admitted to inpatient beds and sent for surgery. We proposed a discrete event simulation (DES) to model incoming flow to operating theatres of a major metropolitan hospital. We assessed how changing the configuration of surgery at the target hospital affects Key Performance Indicators relating to theatre efficiency. In particular, the model was used to assess impacts of six different scenarios by defining new/hypothetical theatre case-mix, opening and closing times of theatres, turnaround (changeover) time, and repurposing the theatres. Target performance metrics included theatre utilisation, pre-operative length-of-stay, average reclaimable time, the percentage of total theatre time in a year that could be reclaimed, and patient waiting time. A web-based application was developed that allows testing user-defined scenarios and interactive analysis of the results. RESULTS: Extending the opening hours of operating theatres by 1 hour almost halved the number of deferred electives as well as over-run cases but at the expense of reduced theatre utilisation. A one-hour reduction in opening hours resulted in 10 times more deferred elective cases and a negligible increase in theatre utilisation. Reducing turnaround time by 50% had positive effects on theatre management: increased utilisation and less deferred and over-run elective cases. Pooling emergency theatres did not affect theatre utilisation but resulted in a considerable reduction in average wait time and the proportion of the delayed emergency cases. CONCLUSIONS: The developed DES-based simulation model of operating theatres along with the web-based user interface provided a useful interrogation tool for theatre management and hospital executive teams to assess new operational strategies. The next step is to embed simulation as ongoing practices in theatre planning workflow, allowing operational managers to use the model outputs to increase theatre utilisation, and reduce cancellations and schedule changes. This can support hospitals in providing services as efficiently and effectively as possible.

New bed configurations and discharge timing policies: A hospital-wide simulation.

OBJECTIVE: Optimising patient flow is becoming an increasingly critical issue as patient demand fluctuates in healthcare systems with finite capacity. Simulation provides a powerful tool to fine-tune policies and investigate their impact before any costly intervention. METHODS: A hospital-wide discrete event simulation is developed to model incoming flow from ED and elective units in a busy metropolitan hospital. The impacts of two different policies are investigated using this simulation model: (i) varying inpatient bed configurations and a load sharing strategy among a cluster of wards within a medical department and (ii) early discharge strategies on inpatient bed access. Several clinically relevant bed configurations and early discharge scenarios are defined and their impact on key performance metrics are quantified. RESULTS: Sharing beds between wards reduced the average and total ED length of stay (LOS) by 21% compared to having patients queue for individual wards. The current baseline performance level could be maintained by using fewer beds when the load sharing approach was imposed. Earlier discharge of inpatients resulted in reducing average patient ED LOS by approximately 16% and average patient waiting time by 75%. Specific time-based discharge targets led to greater improvements in flow compared to blanket approaches of discharging all patients 1 or 2 hours earlier. CONCLUSIONS: ED access performance for admitted patients can be improved by modifying downstream capacity or inpatient discharge times. The simulation model was able to quantify the potential impacts of such policies on patient flow and to provide insights for future strategic planning.

Air ambulance retrievals of patients with suspected appendicitis and acute abdominal pain: The patients' journeys, referral pathways and appendectomy outcomes using linked data in Central Queensland, Australia.

INTRODUCTION: Acute appendicitis is the most common cause of acute abdominal pain presentations to the ED and common air ambulance transfer. AIMS: describe how linked data can be used to explore patients' journeys, referral pathways and request-to-activation responsiveness of patients' appendectomy outcomes (minor vs major complexity). METHODS: Data sources were linked: aeromedical, hospital and death. Request-to-activation intervals showed strong right-tailed skewness. Quantile regression examined whether the longest request-to-activation intervals were associated with appendicitis complexity in patients who underwent an appendectomy. RESULTS: There were 684 patients in three referral pathways based on hospital capability levels. In total, 5.6 % patients were discharged from ED. 83.3 % of all rural origins entered via the ED. 3.8 % of appendicitis patients were triaged to tertiary hospitals. Appendectomy patients with major complexity outcomes were less likely to have longer request-to-activation wait times & had longer lengths of stay than patients with minor complexity outcomes. CONCLUSIONS: Linked data highlighted four aspects of a functioning referral system: appendectomy outcomes of major complexity were less likely to have longer request-to-activation intervals compared to minor (sicker patients were identified); few were discharged from EDs (validated transfer); few were triaged to tertiary hospitals (appropriate level for need), and no deaths relating to appendectomy.

Using a Quality Framework to Explore Air Ambulance Patients' Journey Outcomes in Central Queensland, Australia.

INTRODUCTION: In Australia, aeromedical retrieval provides a vital link for rural communities with limited health services to definitive care in urban centers. Yet, there are few studies of aeromedical patient experiences and outcomes, or clear measures of the service quality provided to these patients. STUDY OBJECTIVE: This study explores whether a previously developed quality framework could usefully be applied to existing air ambulance patient journeys (ie, the sequences of care that span multiple settings; prehospital and hospital-based pre-flight, flight transport, after-flight hospital in-patient, and disposition). The study aimed to use linked data from aeromedical, emergency department (ED), and hospital sources, and from death registries, to document and analyze patient journeys. METHODS: A previously developed air ambulance quality framework was used to place patient, prehospital, and in-hospital service outcomes in relevant quality domains identified from the Institutes of Medicine (IOM) and Dr. Donabedian models. To understand the aeromedical patients' journeys, data from all relevant data sources were linked by unique patient identifiers and the outcomes of the resulting analyses were applied to the air ambulance quality framework. RESULTS: Overall, air ambulance referral pathways could be classified into three categories: Intraregional (those retrievals which stayed within the region), Out of Region, and Into Region. Patient journeys and service outcomes varied markedly between referral pathways. Prehospital and in-hospital service variables and patient outcomes showed that the framework could be used to explore air ambulance service quality. CONCLUSION: The air ambulance quality framework can usefully be applied to air ambulance patient experiences and outcomes using linked data analysis. The framework can help guide prehospital and in-hospital performance reporting. With variations between regional referral pathways, this knowledge will aid with planning within the local service. The study successfully linked data from aeromedical, ED, in-hospital, and death sources and explored the aeromedical patients' journeys.

Identifying patients at risk of unplanned re-hospitalisation using statewide electronic health records.

Preventing unplanned hospitalisations, including readmissions and re-presentations to the emergency department, is an important strategy for addressing the growing demand for hospital care. Significant successes have been reported from interventions put in place by hospitals to reduce their incidence. However, there is limited use of data-driven algorithms in hospital services to identify patients for enrolment into these intervention programs. Here we present the results of a study aiming to develop algorithms deployable at scale as part of a state government's initiative to address rehospitalizations and which fills several gaps identified in the state-of-the-art literature. To the best of our knowledge, our study involves the largest-ever sample size for developing risk models. Logistic regression, random forests and gradient boosted techniques were explored as model candidates and validated retrospectively on five years of data from 27 hospitals in Queensland, Australia. The models used a range of predictor variables sourced from state-wide Emergency Department(ED), inpatient, hospital-dispensed medications and hospital-requested pathology databases. The investigation leads to several findings: (i) the advantage of looking at a longer patient data history, (ii) ED and inpatient datasets alone can provide useful information for predicting hospitalisation risk and the addition of medications and pathology test results leads to trivial performance improvements, (iii) predicting readmissions to the hospital was slightly easier than predicting re-presentations to ED after an inpatient stay, which was slightly easier again than predicting re-presentations to ED after an EDstay, (iv) a gradient boosted approach (XGBoost) was systematically the most powerful modelling approach across various tests.