Attitudes towards artificial intelligence in emergency medicine.

OBJECTIVE: To assess Australian and New Zealand emergency clinicians' attitudes towards the use of artificial intelligence (AI) in emergency medicine. METHODS: We undertook a qualitative interview-based study based on grounded theory. Participants were recruited through ED internal mailing lists, the Australasian College for Emergency Medicine Bulletin, and the research teams' personal networks. Interviews were transcribed, coded and themes presented. RESULTS: Twenty-five interviews were conducted between July 2021 and May 2022. Thematic saturation was achieved after 22 interviews. Most participants were from either Western Australia (52%) or Victoria (16%) and were consultants (96%). More participants reported feeling optimistic (10/25) than neutral (6/25), pessimistic (2/25) or mixed (7/25) towards the use of AI in the ED. A minority expressed scepticism regarding the feasibility or value of implementing AI into the ED. Multiple potential risks and ethical issues were discussed by participants including skill loss from overreliance on AI, algorithmic bias, patient privacy and concerns over liability. Participants also discussed perceived inadequacies in existing information technology systems. Participants felt that AI technologies would be used as decision support tools and not replace the roles of emergency clinicians. Participants were not concerned about the impact of AI on their job security. Most (17/25) participants thought that AI would impact emergency medicine within the next 10 years. CONCLUSIONS: Emergency clinicians interviewed were generally optimistic about the use of AI in emergency medicine, so long as it is used as a decision support tool and they maintain the ability to override its recommendations.

Applications of natural language processing at emergency department triage: A narrative review.

INTRODUCTION: Natural language processing (NLP) uses various computational methods to analyse and understand human language, and has been applied to data acquired at Emergency Department (ED) triage to predict various outcomes. The objective of this scoping review is to evaluate how NLP has been applied to data acquired at ED triage, assess if NLP based models outperform humans or current risk stratification techniques when predicting outcomes, and assess if incorporating free-text improve predictive performance of models when compared to predictive models that use only structured data. METHODS: All English language peer-reviewed research that applied an NLP technique to free-text obtained at ED triage was eligible for inclusion. We excluded studies focusing solely on disease surveillance, and studies that used information obtained after triage. We searched the electronic databases MEDLINE, Embase, Cochrane Database of Systematic Reviews, Web of Science, and Scopus for medical subject headings and text keywords related to NLP and triage. Databases were last searched on 01/01/2022. Risk of bias in studies was assessed using the Prediction model Risk of Bias Assessment Tool (PROBAST). Due to the high level of heterogeneity between studies and high risk of bias, a metanalysis was not conducted. Instead, a narrative synthesis is provided. RESULTS: In total, 3730 studies were screened, and 20 studies were included. The population size varied greatly between studies ranging from 1.8 million patients to 598 triage notes. The most common outcomes assessed were prediction of triage score, prediction of admission, and prediction of critical illness. NLP models achieved high accuracy in predicting need for admission, triage score, critical illness, and mapping free-text chief complaints to structured fields. Incorporating both structured data and free-text data improved results when compared to models that used only structured data. However, the majority of studies (80%) were assessed to have a high risk of bias, and only one study reported the deployment of an NLP model into clinical practice. CONCLUSION: Unstructured free-text triage notes have been used by NLP models to predict clinically relevant outcomes. However, the majority of studies have a high risk of bias, most research is retrospective, and there are few examples of implementation into clinical practice. Future work is needed to prospectively assess if applying NLP to data acquired at ED triage improves ED outcomes when compared to usual clinical practice.

Western Australian medical students' attitudes towards artificial intelligence in healthcare.

INTRODUCTION: Surveys conducted internationally have found widespread interest in artificial intelligence (AI) amongst medical students. No similar surveys have been conducted in Western Australia (WA) and it is not known how medical students in WA feel about the use of AI in healthcare or their understanding of AI. We aim to assess WA medical students' attitudes towards AI in general, AI in healthcare, and the inclusion of AI education in the medical curriculum. METHODS: A digital survey instrument was developed based on a review of available literature and consultation with subject matter experts. The survey was piloted with a group of medical students and refined based on their feedback. We then sent this anonymous digital survey to all medical students in WA (approximately 1539 students). Responses were open from the 7th of September 2021 to the 7th of November 2021. Students' categorical responses were qualitatively analysed, and free text comments from the survey were qualitatively analysed using open coding techniques. RESULTS: Overall, 134 students answered one or more questions (8.9% response rate). The majority of students (82.0%) were 20-29 years old, studying medicine as a postgraduate degree (77.6%), and had started clinical rotations (62.7%). Students were interested in AI (82.6%), self-reported having a basic understanding of AI (84.8%), but few agreed that they had an understanding of the basic computational principles of AI (33.3%) or the limitations of AI (46.2%). Most students (87.5%) had not received teaching in AI. The majority of students (58.6%) agreed that AI should be part of medical training and most (72.7%) wanted more teaching focusing on AI in medicine. Medical students appeared optimistic regarding the role of AI in medicine, with most (74.4%) agreeing with the statement that AI will improve medicine in general. The majority (56.6%) of medical students were not concerned about the impact of AI on their job security as a doctor. Students selected radiology (72.6%), pathology (58.2%), and medical administration (44.8%) as the specialties most likely to be impacted by AI, and psychiatry (61.2%), palliative care (48.5%), and obstetrics and gynaecology (41.0%) as the specialties least likely to be impacted by AI. Qualitative analysis of free text comments identified the use of AI as a tool, and that doctors will not be replaced as common themes. CONCLUSION: Medical students in WA appear to be interested in AI. However, they have not received education about AI and do not feel they understand its basic computational principles or limitations. AI appears to be a current deficit in the medical curriculum in WA, and most students surveyed were supportive of its introduction. These results are consistent with previous surveys conducted internationally.

Applications of machine learning to undifferentiated chest pain in the emergency department: A systematic review.

BACKGROUND: Chest pain is amongst the most common reason for presentation to the emergency department (ED). There are many causes of chest pain, and it is important for the emergency physician to quickly and accurately diagnose life threatening causes such as acute myocardial infarction (AMI). Multiple clinical decision tools have been developed to assist clinicians in risk stratifying patients with chest. There is growing recognition that machine learning (ML) will have a significant impact on the practice of medicine in the near future and may assist with diagnosis and risk stratification. This systematic review aims to evaluate how ML has been applied to adults presenting to the ED with undifferentiated chest pain and assess if ML models show improved performance when compared to physicians or current risk stratification techniques. METHODS AND FINDINGS: We conducted a systematic review of journal articles that applied a ML technique to an adult patient presenting to an emergency department with undifferentiated chest pain. Multiple databases were searched from inception through to November 2020. In total, 3361 articles were screened, and 23 articles were included. We did not conduct a metanalysis due to a high level of heterogeneity between studies in both their methods, and reporting. The most common primary outcomes assessed were diagnosis of acute myocardial infarction (AMI) (12 studies), and prognosis of major adverse cardiovascular event (MACE) (6 studies). There were 14 retrospective studies and 5 prospective studies. Four studies reported the development of a machine learning model retrospectively then tested it prospectively. The most common machine learning methods used were artificial neural networks (14 studies), random forest (6 studies), support vector machine (5 studies), and gradient boosting (2 studies). Multiple studies achieved high accuracy in both the diagnosis of AMI in the ED setting, and in predicting mortality and composite outcomes over various timeframes. ML outperformed existing risk stratification scores in all cases, and physicians in three out of four cases. The majority of studies were single centre, retrospective, and without prospective or external validation. There were only 3 studies that were considered low risk of bias and had low applicability concerns. Two studies reported integrating the ML model into clinical practice. CONCLUSIONS: Research on applications of ML for undifferentiated chest pain in the ED has been ongoing for decades. ML has been reported to outperform emergency physicians and current risk stratification tools to diagnose AMI and predict MACE but has rarely been integrated into practice. Many studies assessing the use of ML in undifferentiated chest pain in the ED have a high risk of bias. It is important that future studies make use of recently developed standardised ML reporting guidelines, register their protocols, and share their datasets and code. Future work is required to assess the impact of ML model implementation on clinical decision making, patient orientated outcomes, and patient and physician acceptability. TRIAL REGISTRATION: International Prospective Register of Systematic Reviews registration number: CRD42020184977.

Artificial intelligence and machine learning in emergency medicine.

Interest in artificial intelligence (AI) research has grown rapidly over the past few years, in part thanks to the numerous successes of modern machine learning techniques such as deep learning, the availability of large datasets and improvements in computing power. AI is proving to be increasingly applicable to healthcare and there is a growing list of tasks where algorithms have matched or surpassed physician performance. Despite the successes there remain significant concerns and challenges surrounding algorithm opacity, trust and patient data security. Notwithstanding these challenges, AI technologies will likely become increasingly integrated into emergency medicine in the coming years. This perspective presents an overview of current AI research relevant to emergency medicine.

Impact of the Four-Hour Rule in Western Australian hospitals: Trend analysis of a large record linkage study 2002-2013.

BACKGROUND: In 2009, the Western Australian (WA) Government introduced the Four-Hour Rule (FHR) program. The policy stated that most patients presenting to Emergency Departments (EDs) were to be seen and either admitted, transferred, or discharged within 4 hours. This study utilised de-identified data from five participating hospitals, before and after FHR implementation, to assess the impact of the FHR on several areas of ED functioning. METHODS: A state (WA) population-based intervention study design, using longitudinal data obtained from administrative health databases via record linkage methodology, and interrupted time series analysis technique. FINDINGS: There were 3,214,802 ED presentations, corresponding to 1,203,513 ED patients. After the FHR implementation, access block for patients admitted through ED for all five sites showed a significant reduction of up to 13.2% (Rate Ratio 0.868, 95%CI 0.814, 0.925) per quarter. Rate of ED attendances for most hospitals continued to rise throughout the entire study period and were unaffected by the FHR, except for one hospital. Pattern of change in ED re-attendance rate post-FHR was similar to pre-FHR, but the trend reduced for two hospitals. ED occupancy was reduced by 6.2% per quarter post-FHR for the most 'crowded' ED. ED length of stay and ED efficiency improved in four hospitals and deteriorated in one hospital. Time to being seen by ED clinician and Did-Not-Wait rate improved for some hospitals. Admission rates in post-FHR increased, by up to 1% per quarter, for two hospitals where the pre-FHR trend was decreasing. CONCLUSIONS: The FHR had a consistent effect on 'flow' measures: significantly reducing ED overcrowding and access block and enhancing ED efficiency. Time-based outcome measures mostly improved with the FHR. There is some evidence of increased ED attendance, but no evidence of increased ED re-attendance. Effects on patient disposition status were mixed. Overall, this reflects the value of investing resources into the ED/hospital system to improve efficiency and patient experience. Further research is required to illuminate the exact mechanisms of the effects of FHR on the ED and hospital functioning across Australia.

Which patients should be transported to the emergency department? A perpetual prehospital dilemma.

OBJECTIVE: To examine the ability of paramedics to identify patients who could be managed in the community and to identify predictors that could be used to accurately identify patients who should be transported to EDs. METHODS: Lower acuity patients who were assessed by paramedics in the Perth metropolitan area in 2013 were studied. Paramedics prospectively indicated on the patient care record if they considered that the patient could be treated in the community. The paramedic decisions were compared with actual disposition from the ED (discharge and admission), and the occurrence of subsequent events (ambulance request, ED visit, admission and death) for discharged patients at the scene was investigated. Decision tree analysis was used to identify predictors that were associated with hospital admission. RESULTS: In total, 57 183 patients were transported to the ED, and 10 204 patients were discharged at the scene by paramedics. Paramedics identified 2717 patients who could potentially be treated in the community among those who were transported to the ED. Of these, 1455 patients (53.6%) were admitted to hospital. For patients discharged at the scene, those who were indicated as suitable for community care were more likely to experience subsequent events than those who were not. The decision tree found that two predictors (age and aetiology) were associated with hospital admission. Overall discriminative power of the decision tree was poor; the area under the receiver operating characteristic curve was 0.686. CONCLUSION: Lower acuity patients who could be treated in the community were not accurately identified by paramedics. This process requires further evaluation.

Paramedic Checklists do not Accurately Identify Post-ictal or Hypoglycaemic Patients Suitable for Discharge at the Scene.

OBJECTIVES: The objective of this study was to assess the accuracy and safety of two pre-defined checklists to identify prehospital post-ictal or hypoglycemic patients who could be discharged at the scene. METHODS: A retrospective cohort study of lower acuity, adult patients attended by paramedics in 2013, and who were either post-ictal or hypoglycemic, was conducted. Two self-care pathway assessment checklists (one each for post-ictal and hypoglycemia) designed as clinical decision tools for paramedics to identify patients suitable for discharge at the scene were used. The intention of the checklists was to provide paramedics with justification to not transport a patient if all checklist criteria were met. Actual patient destination (emergency department [ED] or discharge at the scene) and subsequent events (eg, ambulance requests) were compared between patients who did and did not fulfill the checklists. The performance of the checklists against the destination determined by paramedics was also assessed. RESULTS: Totals of 629 post-ictal and 609 hypoglycemic patients were identified. Of these, 91 (14.5%) and 37 (6.1%) patients fulfilled the respective checklist. Among those who fulfilled the checklist, 25 (27.5%) post-ictal and 18 (48.6%) hypoglycemic patients were discharged at the scene, and 21 (23.1%) and seven (18.9%) were admitted to hospital after ED assessment. Amongst post-ictal patients, those fulfilling the checklist had more subsequent ambulance requests (P=.01) and ED attendances with seizure-related conditions (P=.04) within three days than those who did not. Amongst hypoglycemic patients, there were no significant differences in subsequent events between those who did and did not meet the criteria. Paramedics discharged five times more hypoglycemic patients at the scene than the checklist predicted with no significant differences in the rate of subsequent events. Four deaths (0.66%) occurred within seven days in the hypoglycemic cohort, and none of them were attributed directly to hypoglycemia. CONCLUSIONS: The checklists did not accurately identify patients suitable for discharge at the scene within the Emergency Medical Service. Patients who fulfilled the post-ictal checklist made more subsequent health care service requests within three days than those who did not. Both checklists showed similar occurrence of subsequent events to paramedics' decision, but the hypoglycemia checklist identified fewer patients who could be discharged at the scene than paramedics actually discharged. Reliance on these checklists may increase transportations to ED and delay initiation of appropriate treatment at a hospital. Tohira H , Fatovich D , Williams TA , Bremner A , Arendts G , Rogers IR , Celenza A , Mountain D , Cameron P , Sprivulis P , Ahern T , Finn J . Paramedic checklists do not accurately identify post-ictal or hypoglycaemic patients suitable for discharge at the scene. Prehosp Disaster Med. 2016;31(3):282-293.

Is it Appropriate for Patients to be Discharged at the Scene by Paramedics?

BACKGROUND: Outcomes of patients who are discharged at the scene by paramedics are not fully understood. OBJECTIVE: We aimed to describe the risk of re-presentation and/or death in prehospital patients discharged at the scene. METHODS: We conducted a retrospective cohort study using linked ambulance, emergency department (ED), and death data. We compared outcomes in patients who were discharged at the scene by paramedics with those who were transported to ED by paramedics and then discharged from ED between January 1 and December 31, 2013 in metropolitan Perth, Western Australia. Occurrences of subsequent ambulance requests, ED attendance, hospital admission and death were compared between those discharged at the scene and those discharged from ED. RESULTS: There were 47,330 patients during the study period, of whom 19,732 and 27,598 patients were discharged at the scene and from ED, respectively. Compared to those discharged from ED, those discharged at the scene were more likely to subsequently: request an ambulance (6.1% vs. 1.8%, adjusted odds ratio [adj OR] 3.4; 95% confidence interval [CI] 3.0-3.9), attend ED (4.6% vs. 1.4%, adj OR 3.3; 95% CI 2.8-3.8), be admitted to hospital (3.3% vs. 0.8%, adj OR 4.2; 95% CI 3.4-5.1). Those discharged at the scene tended towards an increased likelihood of death (0.2% vs. 0.1%, adj OR 1.8; 95% CI 0.99-3.2) within 24 hours of discharge compared to those discharged from ED. CONCLUSION: Patients attended by paramedics who were discharged at the scene had more subsequent events than those who were transported to and discharged from ED. Further consideration needs to be given to who is suitable to be discharged at the scene by paramedics.

Western Australia coastal shark bites: A risk assessment.

BACKGROUND: As Western Australia's (WA) government enacts shark bite mitigation, the personal risk of shark bite in WA has not been studied. AIMS: Model the risk of large (>3m) white shark bite (Carcharodon carcharias, LWS) in southwest WA. METHOD: An observational study inclusive of 1 January 1974 to 31 December 2013 was conducted. Analysis of prey abundance, location, water temperature, and water activity participation. Shark bite risk was benchmarked against serious or fatal recreational cycling crash risk in WA. RESULTS: Total and fatal shark bites have grown exponentially over 40 years (3 to 29 total, 0 to 7 fatal per 5 years), correlated with the 10 per cent annual growth in WA humpback whale (Megaptera novaeangliae) abundance (rtotal=0.96 95%CI 0.77-0.99, p<0.001; rfatal=0.96 95%CI 0.81-0.99, p<0.001) but not water activity participation rtotal= 0.25, 95%CI -0.45-0.76, p=0.48). LWS were implicated in 10 of 12 fatalities. Metropolitan Perth beach summer/autumn bathing less than 25m from shore in water less than 5m deep (risk lower than 1 in 20 years) is estimated to be at least 50x safer than cycling. Off-shore diving and surf sports off Perth, during winter/spring have a similar risk to cycling. Winter/spring off- shore diving south of Perth has between 3 and 11 times the cycling risk. CONCLUSION: WA's shark bite risk is likely to increase as whale abundance continues to increase off the WA coast. However, the risk to bathers less than 25m from shore in shallow water during the WA summer, is likely to remain very low, and well below the risk of other recreational activities undertaken in WA.

Measuring the incidence of hospital-acquired complications and their effect on length of stay using CHADx.

OBJECTIVES: To use an automated Classification of Hospital Acquired Diagnoses (CHADx) reporting system to report the incidence of hospital-acquired complications in inpatients and investigate the association between hospital-acquired complications and hospital length of stay (LOS) in multiday-stay patients. DESIGN: Retrospective cross-sectional study for calendar years 2010 and 2011. SETTING: South Metropolitan Health Service in Western Australia, which consists of two teaching and three non-teaching hospitals. MAIN OUTCOME MEASURES: Incidence of hospital-acquired complications and mean LOS for multiday-stay patients. RESULTS: Of 436 841 inpatient separations, 29 172 (6.68%) had at least one hospital-acquired complication code assigned in the administrative data, and there were a total of 56 326 complication codes. The three most common complications were postprocedural complications; cardiovascular complications; and labour, delivery and postpartum complications. In the subset of data on multiday-stay patients, crude mean LOS was longer in separations for patients with hospital-acquired complications than in separations for those without such complications (17.4 days v 5.4 days). After adjusting for potential confounders, separations for patients with hospital-acquired complications had almost four times the mean LOS of separations for those without such complications (incident rate ratio, 3.84; 95% CI, 3.73-3.96; P < 0.001). CONCLUSIONS: An automated CHADx reporting system can be used to collect data on patients with hospital-acquired complications. Such data can be used to increase emphasis on patient safety and quality of care and identify potential opportunities to reduce LOS.

Evidence-based paramedic models of care to reduce unnecessary emergency department attendance--feasibility and safety.

BACKGROUND: As demand for Emergency Department (ED) services continues to exceed increases explained by population growth, strategies to reduce ED presentations are being explored. The concept of ambulance paramedics providing an alternative model of care to the current default 'see and transport to ED' has intuitive appeal and has been implemented in several locations around the world. The premise is that for certain non-critically ill patients, the Extended Care Paramedic (ECP) can either 'see and treat' or 'see and refer' to another primary or community care practitioner, rather than transport to hospital. However, there has been little rigorous investigation of which types of patients can be safely identified and managed in the community, or the impact of ECPs on ED attendance. METHODS/DESIGN: St John Ambulance Western Australia paramedics will indicate on the electronic patient care record (e-PCR) of patients attended in the Perth metropolitan area if they consider them to be suitable to be managed in the community. 'Follow-up' will examine these patients using ED data to determine the patient's disposition from the ED. A clinical panel will then develop a protocol to identify those patients who can be safely managed in the community. Paramedics will then assess patients against the derived ECP protocols and identify those deemed suitable to 'see and treat' or 'see and refer'. The ED disposition (and other clinical outcomes) of these 'ECP protocol identified' patients will enable us to assess whether it would have been appropriate to manage these patients in the community. We will also 'track' re-presentations to EDs within seven days of the initial presentation. This is a 'virtual experiment' with no direct involvement of patients or changes in clinical practice. A systems modelling approach will be used to assess the likely impact on ED crowding. DISCUSSION: To date the efficacy, cost-effectiveness and safety of alternative community-based models of emergency care have not been rigorously investigated. This study will inform the development of ECP protocols through the identification of types of patient presentation that can be considered both safe and appropriate for paramedics to manage in the community.

Can we monitor heart attack in the troponin era? Evidence from a population-based cohort study.

BACKGROUND: Troponins (highly sensitive biomarkers of myocardial damage) increase counts of myocardial infarction (MI) in clinical practice, but their impact on trends in admission rates for MI in National statistics is uncertain. METHODS: Cases coded as MI or other cardiac diagnoses in the Hospital Morbidity Data Collection (MI-HMDC) in Western Australia in 1998 and 2003 were classified using revised criteria for MI developed by an International panel convened by the American Heart Association (AHA criteria) using information on symptoms, ECGs and cardiac biomarkers abstracted from samples of medical notes. Age-sex standardized rates of MI-HMDC were compared with rates of MI based on AHA criteria including troponins (MI-AHA) or traditional biomarkers only (MI-AHAck). RESULTS: Between 1998 and 2003, rates of MI-HMDC decreased by 3.5% whereas rates of MI-AHA increased by 17%, a difference largely due to increased false-negative cases in the HMDC associated with marked increased use of troponin tests in cardiac admissions generally, and progressively lower test thresholds. In contrast, rates of MI-AHAck declined by 18%. CONCLUSIONS: Increasing misclassification of MI-AHA by the HMDC may be due to reluctance by clinicians to diagnose MI based on relatively small increases in troponin levels. These influences are likely to continue. Monitoring MI using AHA criteria will require calibration of commercially available troponin tests and agreement on lower diagnostic thresholds for epidemiological studies. Declining rates of MI-AHA ck are consistent with long-standing trends in MI in Western Australia, suggesting that neither MI-HMDC nor MI-AHA reflect the true underlying population trends in MI.

The economic benefits of health information exchange interoperability for Australia.

OBJECTIVE: To estimate costs and benefits for Australia of implementing health information exchange interoperability among health care providers and other health care stakeholders. DESIGN: A cost-benefit model considering four levels of interoperability (Level 1, paper based; Level 2, machine transportable; Level 3, machine readable; and Level 4, machine interpretable) was developed for Government-funded health services, then validated by expert review. RESULTS: Roll-out costs for Level 3 and Level 4 interoperability were projected to be $21.5 billion and $14.2 billion, respectively, and steady-state costs, $1470 million and $933 million per annum, respectively. Level 3 interoperability would achieve steady-state savings of $1820 million, and Level 4 interoperability, $2990 million, comprising transactions of: laboratory $1180 million (39%); other providers, $893 million (30%); imaging centre, $680 million (23%); pharmacy, $213 million (7%) and public health, $27 million (1%). Net steady-state Level 4 benefits are projected to be $2050 million: $1710 million more than Level 3 benefits of $348 million, reflecting reduced interface costs for Level 4 interoperability due to standardisation of the semantic content of Level 4 messages. CONCLUSIONS: Benefits to both providers and society will accrue from the implementation of interoperability. Standards are needed for the semantic content of clinical messages, in addition to message exchange standards, for the full benefits of interoperability to be realised. An Australian Government policy position supporting such standards is recommended.

Electronic injury surveillance in Perth emergency departments: validity of the data.

OBJECTIVE: To assess the validity of data collected by a new injury surveillance system in metropolitan public hospital ED in Western Australia. METHODS: A reference group of four experts used text descriptions recorded in the injury surveillance system to independently assign codes for intent and cause of injury for each case in the sample. These codes were then compared with the intent and cause codes which triage nurses had assigned to these cases. The level of agreement between these codes were assessed by means of descriptive statistics. Systematic coding errors were also identified and analysed. RESULTS: Of the 419 cases with adequate text descriptions, triage nurses agreed with the reference group of experts in 91.9% (intent) and 79.2% (cause) of cases. Falls accounted for 28.6% (n = 120) of cases and falls code agreement was 86.7%. Self-harm accounted for 5.3% (n = 22) of cases and self-harm code agreement was 77.3%. Systematic errors were detected in the coding of agent of injury, the underlying mechanism of injury and falls involving a mode of transport. CONCLUSIONS: The new injury surveillance system can be successfully used in ED and provides a valid mechanism for monitoring injuries. Refinements to reduce systematic coding errors might improve the validity and quality of the data. A larger sample is needed to assess the validity of the self-harm code.

Hospital workload and adverse events.

CONTEXT: Hospitals are under pressure to increase revenue and lower costs, and at the same time, they face dramatic variation in clinical demand. OBJECTIVE: : We sought to determine the relationship between peak hospital workload and rates of adverse events (AEs). METHODS: A random sample of 24,676 adult patients discharged from the medical/surgical services at 4 US hospitals (2 urban and 2 suburban teaching hospitals) from October 2000 to September 2001 were screened using administrative data, leaving 6841 cases to be reviewed for the presence of AEs. Daily workload for each hospital was characterized by volume, throughput (admissions and discharges), intensity (aggregate DRG weight), and staffing (patient-to-nurse ratios). For volume, we calculated an "enhanced" occupancy rate that accounted for same-day bed occupancy by more than 1 patient. We used Poisson regressions to predict the likelihood of an AE, with control for workload and individual patient complexity, and the effects of clustering. RESULTS: One urban teaching hospital had enhanced occupancy rates more than 100% for much of the year. At that hospital, admissions and patients per nurse were significantly related to the likelihood of an AE (P < 0.05); occupancy rate, discharges, and DRG-weighted census were significant at P < 0.10. For example, a 0.1% increase in the patient-to-nurse ratio led to a 28% increase in the AE rate. Results at the other 3 hospitals varied and were mainly non significant. CONCLUSIONS: Hospitals that operate at or over capacity may experience heightened rates of patient safety events and might consider re-engineering the structures of care to respond better during periods of high stress.

ECHO: the Western Australian Emergency Care Hospitalisation and Outcome linked data project.

OBJECTIVE: To describe and assess the quality of the data resources linked for the Western Australian Emergency Care Hospitalisation and Outcome (ECHO) project. METHODS: The ECHO project links electronic records from the WA Emergency Department Information System to the St John Ambulance Service Pre-Hospital Care Database, the WA Hospital Morbidity Data System and the WA Mortality Database. Linkages are created using standard probabilistic matching techniques with extensive clerical review. Commencing with all metropolitan Perth public emergency departments from July 2000, these linkages will be updated annually for at least five successive years. The proportion of actual linkages between emergency department records and ambulance, admission and death records was assessed in comparison to expected linkage rates. RESULTS: Of 578,200 total emergency department records, there were 144,897 emergency presentations recorded as arriving by ambulance, of which 135,332 (93.4%) were linked to an ambulance record pertaining to the same episode. Of the 165,650 presentations recorded as admitted, 162,216 (97.9%) were linked to a hospital morbidity record relating to the same episode. Furthermore, 96.2% of the 2,084 cases recorded as 'dead on arrival' and 98.9% of the 624 cases recorded as 'died in emergency' were linked to a corresponding death record. CONCLUSIONS: Linkage quality consistent with international standards has been achieved, resulting in an information infrastructure capable of supporting an extensive research agenda focusing on the interaction and outcomes of both pre-hospital and within-hospital emergency medical care services.

Interface between residential aged care facilities and a teaching hospital emergency department in Western Australia.

OBJECTIVE: To estimate the appropriateness of emergency department (ED) presentations by people aged>or=65 years living in residential care facilities. DESIGN, SETTING AND PARTICIPANTS: Retrospective cohort study of older residents of residential care facilities who presented to the ED of the Royal Perth Hospital, Western Australia, between January and June 2002. Data were reviewed by an expert clinical panel. MAIN OUTCOME MEASURES: Appropriateness of ED presentation, presenting complaint, involvement of a general practitioner/locum doctor prior to transfer, proportion of patients admitted to hospital from the ED, survival to discharge. RESULTS: 541 residents aged>or=65 years were transferred by ambulance to the ED, comprising 8.3% of all ED presentations of people in this age group. The mean age of the study cohort was 83.7 years (SD, 7.0 years), of which 68% were women. Of the 541 presentations, 326 (60%) resulted in hospital admission, and of these, 276 (85%) survived to hospital discharge. Musculoskeletal disorders accounted for 25% of all presentations, and 22% were falls-related; pneumonia (11% of presentations) was the single largest presenting complaint. ED attendance was deemed "inappropriate" for 71/541 cases (13.1%; 95% CI, 10.5%-16.2%); in only 25% of ED presentations was a GP/locum doctor involved prior to transfer. CONCLUSIONS: The majority of ED presentations by aged care residents were considered to be appropriate, but there was scope for improvement in coordinating care between the hospital ED and residential care institutions.

The association between hospital overcrowding and mortality among patients admitted via Western Australian emergency departments.

OBJECTIVE: To examine the relationship between hospital and emergency department (ED) occupancy, as indicators of hospital overcrowding, and mortality after emergency admission. DESIGN: Retrospective analysis of 62 495 probabilistically linked emergency hospital admissions and death records. SETTING: Three tertiary metropolitan hospitals between July 2000 and June 2003. PARTICIPANTS: All patients 18 years or older whose first ED attendance resulted in hospital admission during the study period. MAIN OUTCOME MEASURES: Deaths on days 2, 7 and 30 were evaluated against an Overcrowding Hazard Scale based on hospital and ED occupancy, after adjusting for age, diagnosis, referral source, urgency and mode of transport to hospital. RESULTS: There was a linear relationship between the Overcrowding Hazard Scale and deaths on Day 7 (r=0.98; 95% CI, 0.79-1.00). An Overcrowding Hazard Scale>2 was associated with an increased Day 2, Day 7 and Day 30 hazard ratio for death of 1.3 (95% CI, 1.1-1.6), 1.3 (95% CI, 1.2-1.5) and 1.2 (95% CI, 1.1-1.3), respectively. Deaths at 30 days associated with an Overcrowding Hazard Scale>2 compared with one of <3 were undifferentiated with respect to age, diagnosis, urgency, transport mode, referral source or hospital length of stay, but had longer ED durations of stay (risk ratio per hour of ED stay, 1.1; 95% CI, 1.1-1.1; P<0.001) and longer physician waiting times (risk ratio per hour of ED wait, 1.2; 95% CI, 1.1-1.3; P=0.01). CONCLUSIONS: Hospital and ED overcrowding is associated with increased mortality. The Overcrowding Hazard Scale may be used to assess the hazard associated with hospital and ED overcrowding. Reducing overcrowding may improve outcomes for patients requiring emergency hospital admission.

Internet-accessible emergency department workload information reduces ambulance diversion.

OBJECTIVE: To determine the effect of pre-emptive ambulance distribution based on the implementation of a real-time, Internet-accessible emergency department (ED) workload schematic and prehospital Australasian Triage Scale (ATS) allocations on ambulance diversion in Western Australia. METHODS: Comparison of July-December 2002 and July-December 2003 metropolitan Perth ED cubicle occupancy, ambulance diversion, ambulance distribution, and ambulance unloading delays at four inner and four outer metropolitan EDs. RESULTS: Ambulance diversion fell from 1,788 hours in 2002 to 1,138 hours in 2003 (p < 0.001) despite an increase in mean weekly ED cubicle occupancy from 31 patients (95% confidence internal [CI] 29-33) in 2002 to 39 patients in 2003 (95% CI 36-43, p < 0.001). Inner metropolitan ED ambulance attendances fell 2.7% from 27,475 in 2002 to 26,743 in 2003, and outer metropolitan correspondingly rose from 5,877 to 6,628 ambulance attendances (p < 0.001). Unloading delays were similar in 2002 (219, 0.66%) and 2003 (223, 0.67%, p = 0.84); however, median duration of unloading delays increased from 38 minutes (interquartile range [IQR] 18-68) in 2002 to 50 minutes (IQR 25-108) in 2003 (p < 0.001). CONCLUSIONS: The implementation of pre-emptive ambulance distribution using Internet-accessible ED information and prehospital ATS allocations was associated with reduced ambulance diversion, probably consequent upon the redistribution of ambulances from inner to outer metropolitan EDs. The rise in ED cubicle occupancy between the study periods suggests that this approach to reducing ambulance diversion should be viewed only as complementary to direct efforts to reduce ambulance diversion by improving hospital inpatient flow and the balance between acute and elective hospital inpatient accommodation.