Leveraging existing data to improve antimicrobial resistance-related mortality estimates for Australia.

Antimicrobial resistance (AMR) is a global pandemic, however, estimating its burden is a complex process. As a result, many countries rely on global estimates to infer burden within their own setting. With a growing number of recent publications quantifying AMR burden in Australia, and an expansion of surveillance programs, enumerating AMR mortality for Australia is feasible. We aimed to leverage existing published data to assess methodological factors contributing to the considerable variation in AMR-related mortality and provide two reliable estimates of AMR mortality in Australia. This is a necessary step towards generating meaningful measures of AMR burden in Australia.

Graphical Association Analysis for Identifying Variation in Provider Claims for Joint Replacement Surgery.

Identifying potentially fraudulent or wasteful medical insurance claims can be difficult due to the large amounts of data and human effort involved. We applied unsupervised machine learning to construct interpretable models which rank variations in medical provider claiming behaviour in the domain of unilateral joint replacement surgery, using data from the Australian Medicare Benefits Schedule. For each of three surgical procedures reference models of claims for each procedure were constructed and compared analytically to models of individual provider claims. Providers were ranked using a score based on fees for typical claims made in addition to those in the reference model. Evaluation of the results indicated that the top-ranked providers were likely to be unusual in their claiming patterns, with typical claims from outlying providers adding up to 192% to the cost of a procedure. The method is efficient, generalizable to other procedures and, being interpretable, integrates well into existing workflows.

Readmission Risk Based on Debility and Psychosocial Measures: The Western 9 Algorithm.

Many studies have stressed the importance of psychosocial determinants of health such as food insecurity, housing instability, and education level, as important drivers of health outcomes. Western Health developed a set of nine screening questions (the Western 9) based on psychosocial measures to profile patients enrolled in the HealthLinks Chronic Care initiative. An aggregate score was then converted into a Low, Medium and High risk profiles. The aim of the study was to see if the Western 9 questions added additional discriminative power to existing risk of readmission algorithms. Results show that the inclusion of the risk profiles significantly improved model fit and calibration compared to a baseline risk model. Suggestions for further refinement include developing weighted indices for the Western 9 to improve model fit.

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.

Explainable machine learning for real-time deterioration alert prediction to guide pre-emptive treatment.

The Electronic Medical Record (EMR) provides an opportunity to manage patient care efficiently and accurately. This includes clinical decision support tools for the timely identification of adverse events or acute illnesses preceded by deterioration. This paper presents a machine learning-driven tool developed using real-time EMR data for identifying patients at high risk of reaching critical conditions that may demand immediate interventions. This tool provides a pre-emptive solution that can help busy clinicians to prioritize their efforts while evaluating the individual patient risk of deterioration. The tool also provides visualized explanation of the main contributing factors to its decisions, which can guide the choice of intervention. When applied to a test cohort of 18,648 patient records, the tool achieved 100% sensitivity for prediction windows 2-8 h in advance for patients that were identified at 95%, 85% and 70% risk of deterioration.

Predicting Patient Deterioration: A Review of Tools in the Digital Hospital Setting.

BACKGROUND: Early warning tools identify patients at risk of deterioration in hospitals. Electronic medical records in hospitals offer real-time data and the opportunity to automate early warning tools and provide real-time, dynamic risk estimates. OBJECTIVE: This review describes published studies on the development, validation, and implementation of tools for predicting patient deterioration in general wards in hospitals. METHODS: An electronic database search of peer reviewed journal papers from 2008-2020 identified studies reporting the use of tools and algorithms for predicting patient deterioration, defined by unplanned transfer to the intensive care unit, cardiac arrest, or death. Studies conducted solely in intensive care units, emergency departments, or single diagnosis patient groups were excluded. RESULTS: A total of 46 publications were eligible for inclusion. These publications were heterogeneous in design, setting, and outcome measures. Most studies were retrospective studies using cohort data to develop, validate, or statistically evaluate prediction tools. The tools consisted of early warning, screening, or scoring systems based on physiologic data, as well as more complex algorithms developed to better represent real-time data, deal with complexities of longitudinal data, and warn of deterioration risk earlier. Only a few studies detailed the results of the implementation of deterioration warning tools. CONCLUSIONS: Despite relative progress in the development of algorithms to predict patient deterioration, the literature has not shown that the deployment or implementation of such algorithms is reproducibly associated with improvements in patient outcomes. Further work is needed to realize the potential of automated predictions and update dynamic risk estimates as part of an operational early warning system for inpatient deterioration.

After-hours emergency department care: Does time or day of arrival affect survival?

OBJECTIVE: To determine whether after-hours presentation to EDs is associated with differences in 7-day and 30-day mortality. The influence of patient case-mix and workforce staffing differences are also explored. METHODS: We conducted a retrospective observational study of 3.7 million ED episodes across 30 public hospitals in Queensland, Australia during May 2013-September 2015 using routinely collected hospital data linked to hospital staffing data and the death registry. Episodes were categorised as within/after-hours using time of presentation. Staffing was derived from payroll records and explored by defining 11 staffing ratios. RESULTS: Weekend presentation was slightly more associated (7-day mortality odds ratio 1.05, 95% confidence interval [CI] 1.01-1.10) or no more associated (30-day mortality odds ratio 1.01, 95% CI 0.98-1.03) with death than weekday presentation. When weeknights are included in the 'after-hours' period, odds ratios are smaller, so that after-hours presentation is no more associated (7-day mortality odds ratio 1.03, 95% CI 0.99-1.08) or less associated (30-day mortality odds ratio 0.95, 95% CI 0.93-0.97) with death. No significant after-hours patient case-mix differences were observed between weekday and weekend presentations for 7-day mortality. In other combinations of outcome and after-hours definition, some differences (especially measures relating to severity of presenting condition) were found. Staffing ratios were not strongly associated with any within/after-hours differences in ED mortality. CONCLUSIONS: After-hours presentation on the weekend to an ED is associated with higher 7-day mortality even after controlling for case-mix.

The Effects of Telemonitoring on Patient Compliance With Self-Management Recommendations and Outcomes of the Innovative Telemonitoring Enhanced Care Program for Chronic Heart Failure: Randomized Controlled Trial.

BACKGROUND: Telemonitoring enables care providers to remotely support outpatients in self-managing chronic heart failure (CHF), but the objective assessment of patient compliance with self-management recommendations has seldom been studied. OBJECTIVE: This study aimed to evaluate patient compliance with self-management recommendations of an innovative telemonitoring enhanced care program for CHF (ITEC-CHF). METHODS: We conducted a multicenter randomized controlled trial with a 6-month follow-up. The ITEC-CHF program comprised the provision of Bluetooth-enabled scales linked to a call center and nurse care services to assist participants with weight monitoring compliance. Compliance was defined a priori as weighing at least 4 days per week, analyzed objectively from weight recordings on the scales. The intention-to-treat principle was used to perform the analysis. RESULTS: A total of 184 participants (141/184, 76.6% male), with a mean age of 70.1 (SD 12.3) years, were randomized to receive either ITEC-CHF (n=91) or usual care (control; n=93), of which 67 ITEC-CHF and 81 control participants completed the intervention. For the compliance criterion of weighing at least 4 days per week, the proportion of compliant participants in the ITEC-CHF group was not significantly higher than that in the control group (ITEC-CHF: 67/91, 74% vs control: 56/91, 60%; P=.06). However, the proportion of ITEC-CHF participants achieving the stricter compliance standard of at least 6 days a week was significantly higher than that in the control group (ITEC-CHF: 41/91, 45% vs control: 23/93, 25%; P=.005). CONCLUSIONS: ITEC-CHF improved participant compliance with weight monitoring, although the withdrawal rate was high. Telemonitoring is a promising method for supporting both patients and clinicians in the management of CHF. However, further refinements are required to optimize this model of care. TRIAL REGISTRATION: Australian New Zealand Clinical Trial Registry ACTRN12614000916640; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=366691.

Immunisation status of children presenting to the emergency department: Linkage of a longitudinal birth cohort with national immunisation data.

AIM: To describe the relationship between emergency department (ED) diagnosis of infectious disease and immunisation status in children ≤5 years. We also aimed to demonstrate feasibility of proof-of-concept linkage between disparate databases. METHODS: Data from a cohort of 3404 children born in Southeast Queensland/Far North New South Wales between 2006 and 2011 were linked to Australian Childhood Immunisation Registry data and Emergency Department Information System data for presentations between 2006 and 2014. Immunisation status was assigned using the 2009 National Immunisation Program schedule. RESULTS: Of 1490 children (79% of those consented) with data on immunisation status, 87.2 and 84.6% were fully immunised by 12 and 24 months, respectively. Adding partially immunised children increased this to 93.2 and 91.4% at 12 and 24 months, respectively. Nearly two-thirds of all children made at least one ED presentation. Children presenting to ED with an infectious disease did not differ in immunisation status compared to children with other (non-infectious disease type) presentations but were younger, more likely to live with other children and had a longer ED stay and higher admission rate. Respiratory syncytial virus (RSV) was more frequently diagnosed in unimmunised children. CONCLUSIONS: In an existing birth cohort, immunisation rates were lower than the national average. RSV was more prevalent in unimmunised children presenting to ED, but immunisation status was not significantly associated with other infectious disease presentations. Linkage between national immunisation data and Australian ED data is feasible and has the potential to identify previously unrecognised factors related to child immunisation status and health-care utilisation.

Parental-reported allergic disorders and emergency department presentations for allergy in the first five years of life; a longitudinal birth cohort.

BACKGROUND: To measure rates of parental-report of allergic disorders and ED presentations for allergic disorders in children, and to describe factors associated with either. METHODS: An existing cohort of 3404 children born between 2006 and 2011 (Environments for Healthy Living) with prospectively collected pre-natal, perinatal and follow-up data were linked to i) nationwide Medicare and pharmaceutical data and ii) Emergency Department (ED) data from four hospitals in Australia. Parental-reported allergy was assessed in those who returned follow-up questionnaires. ED presentation was defined as any presentation for a suite of allergic disorders, excluding asthma. Univariate analysis and multivariate logistic regression were used to descibe risk factors for both parental-reported allergy and ED presentation for an allergic disorder. RESULTS: The incidence of parental-reported child allergy at 1, 3 and 5 years of age was 7.8, 7.8 and 12.6%, respectively. Independent predictors of parental-report of allergy in multivariate analysis were parental-report of asthma (OR 2.2, 95% CI 1.4-3.4) or eczema (OR 4.3, 95% CI 3.1-6.1) and age > 6 months at introduction of solids (OR 1.3, 95% CI 1.0-1.7). Factors associated with ED presentations for allergy, which occurred in 3.6% of the cohort, were presence of maternal asthma (OR 2.3 95% CI:1.1, 4.9) and child born in spring (OR 1.7, 95% CI 1.1, 2.7). CONCLUSIONS: More than 10% of children up to 5 years have a parental-reported allergic disorder, and 3.6% presented to ED. Parental-report of eczema and/or asthma and late introduction of solids were predictors of parental-report of allergy. Spring birth and maternal asthma were predictors for ED presentation for allergy.

Exploratory multivariate analysis of hospital admissions data in conjunction with workforce data.

The prevalence of electronic health data has brought us a step closer to understanding of the dynamics of hospital admissions. However, little research has investigated hospital admission data in conjunction with information about the environment where the patient was admitted, such as staffing level and hospital type. This paper studied this crucial but often neglected issue by investigating hospital admission records together with workforce data. Exploratory multivariate analysis methods, such as principal component analysis (PCA) and multiple correspondence analysis (MCA), were applied to study important variables associated with admission and workforce data. The exploratory results obtained shed light on the contribution of these variables to the typology of hospital admissions.

Emergency department presentations in infants: Predictors from an Australian birth cohort.

AIM: Infants under 12 months of age are disproportionately represented amongst emergency department (ED) presentations, and infants are more likely to be frequent ED users. This study aimed to describe and identify psychosocial predictors of ED presentation in infants. METHODS: A prospective birth cohort from Queensland and New South Wales (Environments for Healthy Living) was used to understand infant health service use. Baseline and 12-month questionnaire data pertaining to children born between 2006 and 2011 were used to identify predictors of ED presentation, using multiple regression analysis. RESULTS: Of the 2184 children in the cohort with available baseline and 12-month data, 579 (27%) presented at least once to an ED during their first 12 months of life. Statistically significant predictors of ED presentation in the multivariate analysis included the mother having asthma (odds ratio (OR) 1.67, 95% confidence interval (CI) 1.15-2.39) and a higher Kessler-6 score (a measure of psychological distress) of the primary carer at baseline (OR 1.04, 95% CI 1.01-1.08). Maternal education level was not associated with ED presentations of infants. CONCLUSIONS: This study describes maternal and child factors of children who present to the ED in the first year of life. Factors related to an infant's support system were found to be predictors for an ED presentation in the first year of life. This study emphasises the need to review the maternal medical history and psychosocial situation. There may be benefits for health-care practitioners to take the opportunity (such as during routine childhood immunisation) to perform a brief screening tool (such as the Kessler-6) to understand psychological distress experienced by mothers. This may influence the likelihood of a child presenting to an ED within the first 12 months of life.

SnoMAP: Pioneering the Path for Clinical Coding to Improve Patient Care.

The increasing demand for healthcare and the static resources available necessitate data driven improvements in healthcare at large scale. The SnoMAP tool was rapidly developed to provide an automated solution that transforms and maps clinician-entered data to provide data which is fit for both administrative and clinical purposes. Accuracy of data mapping was maintained.

Weekly Hospital Workforce Data: A Data Visualisation Exercise.

Quantifying the health workforce in terms of overall staff numbers and their ratio to patients under their care can strengthen analytical studies designed to inform policy regarding how hospital services are delivered. Information about staffing is traditionally obtained via location-specific audits or self-reported information gleaned from surveys which hold potential biases around time-dependence and recall. In contrast, work presented in this paper describes the derivation of useful workforce metrics from routine hospital financial and clinical information systems that overcome these biases. Staffing data is aggregated, visualised and linked to patient demand to gain insight into spatial and temporal variations in hospital staffing and workload. Overall, hospital staff resourcing varies noticeably across a week, with staff numbers and staff-to-patient ratios dropping to low levels at night and across a weekend. Exploration of staff-to-staff ratios allows further insight into staff dynamics across a week and the variation of supervision level.

Analysing the emergency department patient journey: Discovery of bottlenecks to emergency department patient flow.

OBJECTIVE: Despite significant workflow reform to comply with the federally mandated National Emergency Access Target (NEAT), Australian public hospitals continue to face significant barriers in achieving good ED patient flow. This study was undertaken to identify and analyse the impact of individual waypoints on an ED patient's journey and identify which waypoints act as bottlenecks to a hospital's 4 h ED disposition performance. METHODS: This study involves retrospective analysis and simulation employing 2 years of ED administrative data from a sample of two major and two large metropolitan hospitals in Queensland, Australia. The main outcome measures included waypoint wait times (Treatment Delay and Departure Delay), ED length of stay (EDLOS) and compliance with the NEAT target, measured for all (overall NEAT) and admitted (Admitted NEAT) patients. Variations in outcome measures were analysed as functions of hour of day, day of week, departure status and triage category. Simulations identified the impact of potential ED workflow changes in the context of NEAT performance. RESULTS: Departure Delay accounted for 60 and 20% of EDLOS across large and major metropolitan hospitals, respectively. Higher gains in NEAT compliance are associated with improvements in departure delay rather than treatment delay. Simulation identified that halving Departure Delay improves Admitted NEAT by up to 22 and 4% at large and major metropolitan hospitals, respectively. CONCLUSIONS: The results reinforces the need for a whole-of-hospital effort to address flow bottlenecks, and identify moving a patient from emergency to inpatient care as the critical bottleneck in ED system performance.

Predicting Unpanned Return to Hospital for Chronic Disease Patients.

Preventing unplanned returns, including readmissions and representations to the emergency department is increasingly becoming a performance target for hospitals across the globe. Significant successes have been reported from interventions put in to place by hospitals to reduce their incidence. However, despite several risk stratification algorithms being proposed in recent years, there is limited use of these algorithms in hospital services to identify patients for enrolment into these intervention programs. This study identifies constraints limiting the practical use of such algorithms. We also develop and validate models that focus on clinically relevant patient cohorts and are thus better suited to practical deployment in hospitals, while still offering good predictive ability.

The National Emergency Access Target (NEAT) and the 4-hour rule: time to review the target.

OBJECTIVE: We explored the relationship between the National Emergency Access Target (NEAT) compliance rate, defined as the proportion of patients admitted or discharged from emergency departments (EDs) within 4 hours of presentation, and the risk-adjusted in-hospital mortality of patients admitted to hospital acutely from EDs. DESIGN, SETTING AND PARTICIPANTS: Retrospective observational study of all de-identified episodes of care involving patients who presented acutely to the EDs of 59 Australian hospitals between 1 July 2010 and 30 June 2014. MAIN OUTCOME MEASURE: The relationship between the risk-adjusted mortality of inpatients admitted acutely from EDs (the emergency hospital standardised mortality ratio [eHSMR]: the ratio of the numbers of observed to expected deaths) and NEAT compliance rates for all presenting patients (total NEAT) and admitted patients (admitted NEAT). RESULTS: ED and inpatient data were aggregated for 12.5 million ED episodes of care and 11.6 million inpatient episodes of care. A highly significant (P < 0.001) linear, inverse relationship between eHSMR and each of total and admitted NEAT compliance rates was found; eHSMR declined to a nadir of 73 as total and admitted NEAT compliance rates rose to about 83% and 65% respectively. Sensitivity analyses found no confounding by the inclusion of palliative care and/or short-stay patients. CONCLUSION: As NEAT compliance rates increased, in-hospital mortality of emergency admissions declined, although this direct inverse relationship is lost once total and admitted NEAT compliance rates exceed certain levels. This inverse association between NEAT compliance rates and in-hospital mortality should be considered when formulating targets for access to emergency care.

Development of a novel image-based program to teach narrow-band imaging.

OBJECTIVES: Narrow-band imaging (NBI) is a widely available endoscopic imaging technology; however, uptake of the technique could be improved. Teaching new imaging techniques and assessing trainees' performance can be a challenging exercise during a 1-day workshop. To support NBI training, we developed an online training tool (Medimq) to help experts train novices in NBI bronchoscopy that could assess trainees' performance and provide feedback before the close of the 1-day course. The present study determines whether trainees' capacity to identify relevant pathology increases with the proposed interactive testing method. METHODS: Two groups of 20 and 18 bronchoscopists have attended an NBI course where they did a pretest and post-test before and after the main lecture, and a follow-up test 4 weeks later to measure retention of knowledge. We measured their ability to mark normal and abnormal 'biopsy size' areas on bronchoscopic NBI images for biopsy. These markings were compared with areas marked by experts on the same images. RESULTS: The first group results were used to pilot the test. After modifications, the results of the improved test for group 2 showed trainees improved by 32% (total class average normalized gain) in detecting normal or abnormal areas. On follow-up testing, Group 2 improved by 23%. CONCLUSIONS: The overall class average normalized gain of 32% shows our test can be used to improve trainees' competency in analyzing NBI Images. The testing method (and tool) can be used to measure the follow up 4 weeks later. Better follow-up test results would be expected with more frequent practice by trainees after the course.

Model Selection and Variable Aggregation of Australian Hospital Data.

BACKGROUND: Hospital administrative data commonly consist of hundreds of variables with many consisting of hundreds, if not thousands, of distinct categories, especially for disease groups. Conventional approaches to develop regression models for prediction either fail completely due to multicollinearity or sparsity issues or take too long and consume too many computer resources. METHODS: We demonstrate how regularisation and variable aggregation techniques such as Elastic Net can overcome some of these problems. Parameter estimates from univariate generalised linear models (GLM) and Elastic Net models were used to aggregate disease groups into a more manageable number and predict the probability of mortality for a given patient. RESULTS: When employed for variable aggregation and variable selection, Elastic Net models ran at least four times faster than GLMs, though producing a less discriminative model. When applied to final models for predicting hospital mortality, though, both Elastic Net and GLM models demonstrated similar predictive power and efficiently solved an otherwise complex problem. CONCLUSION: Elastic Net regularisation and variable aggregation provide an efficient mechanism for solving healthcare modelling problems.