Research Letter: Dispensing of psychotropic medications in Australia between 2013 and 2022.

OBJECTIVE: The objective of this study was to determine the proportion of Australians dispensed psychotropic medications between 2013 and 2022 according to their age. METHODS: Services Australia provided a de-identified 10% random Pharmaceutical Benefits Scheme sample that allowed us to determine, for each year, the proportion of Australians dispensed at least one script for antipsychotics, antidepressants, anxiolytics, or hypnotics. The classification of medications followed Anatomical Therapeutic Chemical coding. Participants were stratified into 10-year age groups from 0-9 to ⩾90 years, and sex was coded as male/female. We retrieved population numbers from the Australian Bureau of Statistics. RESULTS: The number of records per year ranged from 1,540,520 to 1,746,402, and 54.10% were for females. A greater proportion of older adults, particularly those aged ⩾70 years, were dispensed antipsychotics, antidepressants, anxiolytics and hypnotics than any other age group. The proportion of people who dispensed antipsychotics, anxiolytics and hypnotics declined between 2013 and 2022 but increased for antidepressants, most markedly for adolescents and young adults. Females were more frequently dispensed antidepressants, anxiolytics and hypnotics than males, but males were more frequently dispensed antipsychotics than females, albeit not in later life. CONCLUSION: Older age groups and females are the most frequent recipients of psychotropic medications dispensed in Australia.

The Safety of Methoxyflurane for Emergency Pain Relief in Children and Adolescents: A Retrospective Cohort Study.

OBJECTIVES: The use of methoxyflurane is becoming increasingly popular in the treatment of pain in an emergency setting, in part due to its ease of administration. However, little is known about the risk of serious adverse events in children and adolescents. The aim of this study was to examine the safety of methoxyflurane in a pediatric population. METHODS: The study was a retrospective cohort study of pediatric prehospital events using probabilistic linked health data. All ambulance transfers in Western Australia between 1990 and 2016 involving children and adolescent patients were identified. Patients were categorized based on administered analgesia: methoxyflurane, an opioid analgesic, both methoxyflurane and an opioid analgesic, or no analgesic. Hospital and mortality data were linked to transferred patients to identify deaths, adverse drug reactions, liver and kidney toxicity, and re-admissions to hospital following ambulance transfer. Generalized linear models, adjusting for sociodemographic and ambulance transfer characteristics, were used to compare outcomes between children exposed to methoxyflurane and the other three groups. RESULTS: The study cohort consisted of 37,211 children, including 9,472 patients (25.5%) treated with methoxyflurane alone, 2,764 (7.4%) treated with an opioid analgesic, 1,235 (3.3%) treated with both methoxyflurane and an opioid analgesic, and 23,740 (63.8%) treated with no analgesic. Death in children and adolescents was uncommon, with less than five deaths (<0.1%) observed in the 12 months following treatment with methoxyflurane and no deaths in those treated with both methoxyflurane and an opioid analgesic. Adverse drug reaction was rare (<0.1%) in patients treated with methoxyflurane, as was liver and kidney toxicity with no case observed. At 90-days follow-up, there was no significant difference in hospitalization in patients treated with methoxyflurane and those treated with methoxyflurane and an opioid analgesic (adjusted OR:1.01, 95%CI:0.85-1.21). Compared with methoxyflurane treated patients, patients treated with an opioid analgesic were more likely to be hospitalized (aOR:1.23, 95%CI:1.09-1.39), while patients treated with no analgesic were less likely to be hospitalized (aOR:0.85, 95%CI:0.79-0.92). CONCLUSIONS: In children and adolescents transported by ambulance, the use of methoxyflurane was not associated with an increased risk of hospitalization, death, serious adverse drug reactions or liver and kidney toxicity.

Health in a Virtual Environment (HIVE): A Novel Continuous Remote Monitoring Service for Inpatient Management.

The aim of this study was to describe the implementation of a novel 50-bed continuous remote monitoring service for high-risk acute inpatients treated in non-critical wards, known as Health in a Virtual Environment (HIVE). We report the initial results, presenting the number and type of patients connected to the service, and assess key outcomes from this cohort. This was a prospective, observational study of characteristics and outcomes of patients connected to the HIVE continuous monitoring service at a major tertiary hospital and a smaller public hospital in Western Australia between January 2021 and June 2023. In the first two and a half years following implementation, 7541 patients were connected to HIVE for a total of 331,118 h. Overall, these patients had a median length of stay of 5 days (IQR 2, 10), 11.0% (n = 833) had an intensive care unit admission, 22.4% (n = 1691) had an all-cause emergency readmission within 28 days from hospital discharge, and 2.2% (n = 167) died in hospital. Conclusions: Our initial results show promise, demonstrating that this innovative approach to inpatient care can be successfully implemented to monitor high-risk patients in medical and surgical wards. Future studies will investigate the effectiveness of the program by comparing patients receiving HIVE supported care to comparable patients receiving routine care.

Supply of nirmatrelvir/ritonavir and molnupiravir for patients with COVID-19 in the first eight months since listing on the Australian Pharmaceutical Benefits Scheme: A retrospective observational study.

OBJECTIVES: To compare the supply of molnupiravir and nirmatrelvir/ritonavir in relation to patient characteristics and other co-prescribed medicines and to estimate the number of patients without contraindications to nirmatrelvir/ritonavir who were treated with molnupiravir. STUDY DESIGN, SETTING: Retrospective observational study of patients identified in the Pharmaceutical Benefits Scheme (PBS) 10 % sample dataset who were supplied with either molnupiravir or nirmatrelvir/ritonavir between May and December 2022. We supplemented the PBS dataset with aggregated counts from published literature to determine prevalence of clinical contraindications to nirmatrelvir/ritonavir. MAIN OUTCOME MEASURES: We used multivariable Poisson regression to estimate risk ratios (RR) of receiving nirmatrelvir/ritonavir over molnupiravir. RESULTS: We identified 54,550 patients who received either nirmatrelvir/ritonavir (26.8 %) or molnupiravir (73.2 %). Their average age was 71.6 (SD = 13.4) years and 57.1 % were female. Patients were less likely to receive nirmatrelvir/ritonavir with increasing age (RR = 0.50; 95 % CI: 0.48-0.53; for ages 85 + compared to < 65 years) or who had received medicines contraindicated for use with nirmatrelvir/ritonavir (RR = 0.66; 95 % CI: 0.64-0.68). During the study period, we estimated that between 28.4 % and 45.4 % of patients aged ≥ 65 years had received molnupiravir in the absence of pharmacological and clinical contraindications to nirmatrelvir/ritonavir. CONCLUSION: Many prescriptions were written for molnupiravir where there were no contraindications to nirmatrelvir/ritonavir. The benefits that followed from prompt government action in approving and obtaining nirmatrelvir/ritonavir were therefore likely to be less than they could potentially have been. Governments should consider investing in quality improvement systems to ensure the best outcomes in terms of efficacy and safety.

Cross-Sectional and Longitudinal Associations Between Treatment for Herpes Virus Infection and the Dispensing of Antidementia Medicines: An Analysis of the Australian Pharmaceutical Benefits Scheme Database.

Background: Evidence from previous observational studies suggest that infection by herpes simplex virus (HSV) and varicella zoster virus (VZV) increase the risk of dementia. Objective: To investigate if older adults exposed to HSV treatment have lower risk of dementia than the rest of the population. Methods: We used the 10% Australian Pharmaceutical Benefits Scheme (PBS) database from 2013 to 2022 to ascertain the cross-sectional, time-series and longitudinal association between exposure to HSV treatment and the dispensing of antidementia medicines. Participants were men and women aged 60 years or older. We used Anatomical Therapeutic Chemical (ATC) codes to identify medicines dispensed for the treatment of HSV and dementia. Results: During the year 2022 6,868 (1.2%) of 559,561 of participants aged 60 years or over were dispensed antidementia agent. The odds ratio (OR) of being dispensed an antidementia agent among individuals dispensed treatment for HSV was 0.73 (99% CI = 0.56-0.95). Multilevel logistic regression for the 2013-2022 period for those dispensed HSV treatment was 0.87 (99% CI = 0.75-1.00). Split-time span series from 2013 was associated with hazard ratio of 0.98 (99% CI = 0.89-1.07) for individuals dispensed relative to those not dispensed HSV treatment. All analyses were adjusted for age, sex, and the dispensing of medicines for the treatment of diabetes, hyperlipidemia, hypertension, and ischemic heart disease. Conclusions: The dispensing of antiviral medicines for the treatment of HSV and VZV is consistently, but not conclusively, associated with decreased dispensing of antidementia medicines. This suggests that treatment of HSV and VZV infections may contribute to reduce the risk of dementia.

Predicting multifaceted risks using machine learning in atrial fibrillation: insights from GLORIA-AF study.

Aims: Patients with atrial fibrillation (AF) have a higher risk of ischaemic stroke and death. While anticoagulants are effective at reducing these risks, they increase the risk of bleeding. Current clinical risk scores only perform modestly in predicting adverse outcomes, especially for the outcome of death. We aimed to test the multi-label gradient boosting decision tree (ML-GBDT) model in predicting risks for adverse outcomes in a prospective global AF registry. Methods and results: We studied patients from phase II/III of the Global Registry on Long-Term Oral Anti-Thrombotic Treatment in Patients with Atrial Fibrillation registry between 2011 and 2020. The outcomes were all-cause death, ischaemic stroke, and major bleeding within 1 year following the AF. We trained the ML-GBDT model and compared its discrimination with the clinical scores in predicting patient outcomes. A total of 25 656 patients were included [mean age 70.3 years (SD 10.3); 44.8% female]. Within 1 year after AF, ischaemic stroke occurred in 215 (0.8%), major bleeding in 405 (1.6%), and death in 897 (3.5%) patients. Our model achieved an optimized area under the curve in predicting death (0.785, 95% CI: 0.757-0.813) compared with the Charlson Comorbidity Index (0.747, P = 0.007), ischaemic stroke (0.691, 0.626-0.756) compared with CHA2DS2-VASc (0.613, P = 0.028), and major bleeding (0.698, 0.651-0.745) as opposed to HAS-BLED (0.607, P = 0.002), with improvement in net reclassification index (10.0, 12.5, and 23.6%, respectively). Conclusion: The ML-GBDT model outperformed clinical risk scores in predicting the risks in patients with AF. This approach could be used as a single multifaceted holistic tool to optimize patient risk assessment and mitigate adverse outcomes when managing AF.

Observational evidence linking psychotropic medicines to the dispensing of opioid agents in later life.

BACKGROUND: The use of opioid medicines is common in developed countries, particularly among older adults and those with mental health disorders. It is unclear if the association between mental disorders and opioid medicines is causal, or is due to reverse causality or confounding. METHODS: We used a 10% random sample of the Australian Pharmaceutical Benefits Scheme (years 2012-2022) to examine the cross-sectional, case-control and longitudinal association between the dispensing of antidepressants, anxiolytics, hypnotics, antipsychotics and lithium, and opioid medicines. We used logistic regression, structural equation models (SEM), and Cox regression to analyze the data. Analyses were adjusted for age (years), sex, and number of non-psychotropic medicines dispensed during the year. RESULTS: The 2022 file contained 804 334 individuals aged 50 years or over (53.1% women), of whom 181 690 (22.6%) received an opioid medicine. The adjusted odds ratio of being dispensed opioid medicines was 1.44 (99% CI = 1.42-1.46) for antidepressants, 1.97 (99% CI = 1.92-2.03) for anxiolytics, 1.55 (99% CI = 1.51-1.60) for hypnotics, 1.32 (99% CI = 1.27-1.38) for antipsychotics, and 0.60 (99% CI = 0.53-0.69) for lithium. Similar associations were noticed when we compared participants who were or not dispensed opioid medicines in 2022 for exposure to psychotropic agents between 2012 and 2021. SEM confirmed that this association was not due to reverse causality. The dispensing of antidepressants was associated with increased adjusted hazard (HR) of subsequent dispensing of opioid medicines (HR = 1.29, 99% CI = 1.27-1.30). Similar associations were observed for anxiolytics, hypnotics and antipsychotics, but not lithium. CONCLUSIONS: The dispensing of opioid medicines is higher among older individuals exposed to antidepressants, anxiolytics, hypnotics and antipsychotics than those who are not. These associations are not due to reverse causality or study design. Preventive strategies seeking to minimise the risk of inappropriate use of opioid medicines in later life should consider targeting this high-risk population.

Cross-sectional, case-control and longitudinal associations between exposure to glucagon-like peptide-1 receptor agonists and the dispensing of antidepressants.

AIM: To determine if the dispensing of glucagon-like peptide (GLP)-1 receptor agonists is associated with increased dispensing of antidepressants. MATERIALS AND METHODS: We used cross-sectional, case-control and retrospective cohort study designs to examine the association between dispensed GLP-1 receptor agonists and antidepressants between 2012 and 2022 in the 10% random sample of the Australian Pharmaceutical Benefits Scheme (PBS) data. PBS-listed GLP-1 receptor agonists, exenatide, dulaglutide and semaglutide were the exposures. Outcomes were the odds ratio [ORs; 99% confidence interval (CI)] and hazard ratio (99% CI) of being dispensed any antidepressant. Analyses were adjusted for demographic measures and the dispensing of medicines to manage cardiovascular diseases or anxiety/insomnia. Statistical tests were two-sided at the 1% level of significance. RESULTS: In total, 358 075 of 1 746 391 individuals were dispensed antidepressants, and 8495 of the 24 783 dispensed a GLP-1 receptor agonist were also dispensed an antidepressant in 2022 (OR 1.44; 99% CI 1.38-1.50); 24 103 of the 1 746 391 participants had been dispensed a GLP-1 receptor agonist between 2012 and 2021, and of these 8083 were dispensed antidepressants in 2022 (OR 1.52; 99% CI 1.46-1.59). The 2012 cohort included 1 213 316 individuals who had not been dispensed antidepressants that year. The hazard ratio of being dispensed an antidepressant between 2013 and 2022 following the dispensing of a GLP-1 receptor agonist was 1.19 (99% CI 1.12-1.27). Additional analyses restricting the time of exposure confirmed these associations for all PBS-listed GLP-1 receptor agonists. CONCLUSIONS: Individuals exposed to GLP-1 receptor agonists are at greater risk of being dispensed antidepressants. The possible impact of GLP-1 receptor agonists on the mood of consumers requires ongoing vigilance and further research.

Analysis and evaluation of explainable artificial intelligence on suicide risk assessment.

This study explores the effectiveness of Explainable Artificial Intelligence (XAI) for predicting suicide risk from medical tabular data. Given the common challenge of limited datasets in health-related Machine Learning (ML) applications, we use data augmentation in tandem with ML to enhance the identification of individuals at high risk of suicide. We use SHapley Additive exPlanations (SHAP) for XAI and traditional correlation analysis to rank feature importance, pinpointing primary factors influencing suicide risk and preventive measures. Experimental results show the Random Forest (RF) model is excelling in accuracy, F1 score, and AUC (>97% across metrics). According to SHAP, anger issues, depression, and social isolation emerge as top predictors of suicide risk, while individuals with high incomes, esteemed professions, and higher education present the lowest risk. Our findings underscore the effectiveness of ML and XAI in suicide risk assessment, offering valuable insights for psychiatrists and facilitating informed clinical decisions.

Sex- and Age-Specific Differences in Risk Profiles and Early Outcomes in Adults With Acute Coronary Syndromes.

BACKGROUND: Adults <55 years of age comprise a quarter of all acute coronary syndromes (ACS) hospitalisations. There is a paucity of data characterising this group, particularly sex differences. This study aimed to compare the clinical and risk profile of patients with ACS aged <55 years with older counterparts, and measure short-term outcomes by age and sex. METHOD: The study population comprised patients with ACS enrolled in the AUS-Global Registry of Acute Coronary Events (GRACE), Cooperative National Registry of Acute Coronary Syndrome Care (CONCORDANCE) and SNAPSHOT ACS registries. We compared clinical features and combinations of major modifiable risk factors (hypertension, smoking, dyslipidaemia, and diabetes) by sex and age group (20-54, 55-74, 75-94 years). All-cause mortality and major adverse events were identified in-hospital and at 6-months. RESULTS: There were 16,658 patients included (22.3% aged 20-54 years). Among them, 20-54 year olds had the highest proportion of ST-elevation myocardial infarction compared with sex-matched older age groups. Half of 20-54 year olds were current smokers, compared with a quarter of 55-74 year olds, and had the highest prevalence of no major modifiable risk factors (14.2% women, 12.7% men) and of single risk factors (27.6% women, 29.0% men), driven by smoking. Conversely, this age group had the highest proportion of all four modifiable risk factors (6.6% women, 4.7% men). Mortality at 6 months in 20-54 year olds was similar between men (2.3%) and women (1.7%), although lower than in older age groups. CONCLUSIONS: Younger adults with ACS are more likely to have either no risk factor, a single risk factor, or all four modifiable risk factors, than older patients. Targeted risk factor prevention and management is warranted in this age group.

Development of a list of Australian potentially inappropriate medicines using the Delphi technique.

BACKGROUND: Older people are at high risk of medicines-related harms. otentially inappropriate medicines (PIMs) list has been developed to assist clinicians and researchers to identify medicines with risks that may potentially outweigh their benefits in order to improve medication management and safety. AIM: To develop a list of PIMs for older people specific to Australia. METHODS: The study obtained expert consensus through the utilisation of the Delphi technique in Australia. A total of 33 experts partook in the initial round, while 32 experts engaged in the subsequent round. The primary outcomes encompass medicines assessed as potentially inappropriate, the specific contexts in which their inappropriateness arises and potentially safer alternatives. RESULTS: A total of 16 medicines or medicine classes had one or more medicines deemed as potentially inappropriate in older people. Up to 19 medicines or medicine classes had specific conditions that make them more potentially inappropriate, while alternatives were suggested for 16 medicines or classes. CONCLUSION: An explicit PIMs list for older people living in Australia has been developed containing 19 drugs/drug classes. The PIMs list is intended to be used as a guide for clinicians when assessing medication appropriateness in older people in Australian clinical settings and does not substitute individualised treatment advice from clinicians.

Adherence to pharmacotherapy: sine qua non for reducing cumulative risk of premature coronary disease in familial hypercholesterolemia.

PURPOSE OF REVIEW: Familial hypercholesterolemia (FH) is a dominant and highly penetrant monogenic disorder present from birth that markedly elevates plasma low-density lipoprotein (LDL)-cholesterol concentration and, if untreated, leads to atherosclerotic cardiovascular disease (ASCVD). The risk of ASCVD can be substantially reduced with lipid-lowering treatment (LLT). However, adherence to LLT remains a major challenge in FH patients and an under-recognized issue. We review several barriers to treatment adherence and implementation strategies for improving adherence in patients with FH. RECENT FINDINGS: Barriers that negatively affect patient adherence to treatment include the misunderstanding of perceived and actual risk of FH and the benefits of LLT, inadequate knowledge, lack of standardization of treatment, insufficient monitoring of LDL-cholesterol level, and inequalities in healthcare resources. Education of patients, carers and healthcare providers, guideline-directed treatment goals, regular monitoring, medication regimen simplification and greater access to established and new drugs are crucial enablers for improving adherence to treatment. However, given FH is present from birth, strategies for life-long adherence from childhood or young adulthood is critically important and requires further study. To be effective, strategies should be multifaceted, targeted and patient-centred involving a multidisciplinary-team with support from family, communities and peer groups. SUMMARY: FH confers a significant risk for ASCVD from a young age. Achieving better medication adherence is foundational for improving clinical outcomes and reducing the burden of atherosclerosis over a lifetime. Identification of key barriers and enablers are critical for implementing better adherence to treatment across the life-course of patients with FH.

Prospective Association Between the Dispensing of Antidepressants and of Medications to Treat Osteoporosis in Older Age.

BACKGROUND: Osteoporosis is a common condition associated with fragility fractures, especially in older individuals and women. Antidepressants have emerged as a potential risk factor, but their association with bone fragility remains uncertain because the results of past studies are difficult to generalize. We aimed to investigate the association between antidepressant exposure and subsequent treatment for osteoporosis in a nationally representative sample of Australians. METHODS: Cohort study using a 10% random sample of the Pharmaceutical Benefits Scheme (PBS) data for 2012, that included 566,707 individuals aged older than or equal to 50 years not dispensed osteoporosis medications. The effect of exposure to antidepressants during 2012 (prevalent or incident) or later (up to 2022) was examined using Cox regression models adjusted for age, sex, comorbidities and other psychotropic medications. RESULTS: Over 10 years, 73,360 (12.94%) received osteoporosis medications; 16,216 (22.10%) had been dispensed antidepressants in 2012. The hazard of osteoporosis medication dispensing was higher among those exposed to antidepressants (HR = 1.16, 99% CI = 1.14-1.18; average duration of follow up: 8.0 ± 3.1 years, range: 1-10 years). The hazard of osteoporosis medication diminished with increasing age, and the effect of antidepressants was 37%-76% more pronounced among men in the 50s and 60s. Different classes of antidepressants had a similar risk profile. CONCLUSION: The dispensing of antidepressants in older age is associated with higher hazard of subsequent dispensing of medications for osteoporosis, and this association is more marked for young older adults, particularly men. Clinicians should monitor the bone health of older individuals treated with antidepressants in order to decrease the morbidity associated with fragility fractures.

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.

Dispensing of antineoplastic medications and their impact on the dispensing of anti-dementia drugs for adults aged ≥60 years: A cohort study.

History of cancer has been associated with decreased risk of dementia, but it is unclear if this is due to the use of antineoplastic medications. Participants were 442,795 adults aged ≥60 years, of whom 235,841 (53.26 %) were women. Those dispensed antineoplastic medications during 2012-2013 had lower odds of being dispensed an anti-dementia drug between 2015 and 2021 (age/sex-adjusted OR = 0.60, 95%CI = 0.55-0.66). The dispensing of antineoplastic medications was associated with an adjusted hazard ratio of 0.72 (95%CI = 0.65-0.80) of subsequent dispensing of an anti-dementia drug. Understanding the mechanisms that support this association may contribute to the introduction of novel approaches to dementia prevention.

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.

Contemporary trends in anaphylaxis burden and healthcare utilisation in Western Australia: A linked data study.

Background: Anaphylaxis is a growing public health problem in Australia. To determine the extent of the problem, we linked multiple health datasets to examine temporal trends in anaphylaxis events across the health system in Western Australia (WA). Methods: We identified an anaphylaxis cohort from 1980 to 2020 using linked datasets from ambulance, emergency departments, hospital inpatients and deaths. Age-standardised anaphylaxis event rates were calculated from 2010 to 2020. Dataset-specific rates for anaphylaxis were also examined, to show differences in health care utilisation. Annual percent change in rates (2010-2019) were estimated using age-adjusted Poisson regression models. Results: A total of 19 140 individuals (mean age 31 years; 51% female) experienced 24 239 anaphylaxis events between 2010 and 2020. From 2010 to 2019, the average annual percent increase (95% CI) in rates was 5.3% (4.8-5.8%), from 70.3 to 113.9, with rates reducing to 76.5/100 000 population in 2020. Adolescents and young adults aged 5-14 years and 15-24 years had the greatest increase of 6.9% (5.6-8.1%) and 6.8% (5.6-8.0) respectively, with those over 25 years increasing by approximately 5% per year and children 1-4 years showing the lowest annual increase of 2.6% (1.1-4.2%). The highest absolute rates were seen in under 1 year (269.7/100 000; 2019). There has been an acceleration of trends from 2015 to 2019, underpinned by large increases in 15-24 and 25-34 years. All databases, show similar increasing trends, with ambulance attendance (33.7 per 100 000), emergency presentation (89.8 per 100 000) and hospital admissions (46.2 per 100 000), for anaphylaxis highest in 2019. However, whilst ambulance and emergency presentations have grown by 8.9% (95%CI 7.9-9.8%) and 6.6% per year (95%CI 6.0-7.2%), respectively, hospitalisations appear to be steadying with only a 0.9% (95%CI 0.2-1.6%) yearly rise. Conclusion: Rates of anaphylaxis continue to increase, with WA having higher rates than previous estimates for Australia. Whilst rates are still high in infants, lower trends in children compared to older ages may indicate better prevention of allergy. Results show more people experiencing anaphylaxis now receive care in emergency and ambulance, rather than hospital. Further exploration of the patient care journey through prehospital and inpatient care is required, to understand the changing health demands of people who experience anaphylaxis.

Deep learning model to predict exercise stress test results: Optimizing the diagnostic test selection strategy and reduce wastage in suspected coronary artery disease patients.

BACKGROUND: Cardiac exercise stress testing (EST) offers a non-invasive way in the management of patients with suspected coronary artery disease (CAD). However, up to 30% EST results are either inconclusive or non-diagnostic, which results in significant resource wastage. Our aim was to build machine learning (ML) based models, using patients demographic (age, sex) and pre-test clinical information (reason for performing test, medications, blood pressure, heart rate, and resting electrocardiogram), capable of predicting EST results beforehand including those with inconclusive or non-diagnostic results. METHODS: A total of 30,710 patients (mean age 54.0 years, 69% male) were included in the study with 25% randomly sampled in the test set, and the remaining samples were split into a train and validation set with a ratio of 9:1. We constructed different ML models from pre-test variables and compared their discriminant power using the area under the receiver operating characteristic curve (AUC). RESULTS: A network of Oblivious Decision Trees provided the best discriminant power (AUC=0.83, sensitivity=69%, specificity=0.78%) for predicting inconclusive EST results. A total of 2010 inconclusive ESTs were correctly identified in the testing set. CONCLUSIONS: Our ML model, developed using demographic and pre-test clinical information, can accurately predict EST results and could be used to identify patients with inconclusive or non-diagnostic results beforehand. Our system could thus be used as a personalised decision support tool by clinicians for optimizing the diagnostic test selection strategy for CAD patients and to reduce healthcare expenditure by reducing nondiagnostic or inconclusive ESTs.