Christopher Bain

Short-Term Forecasting of Emergency Inpatient Flow

Hospital managers have to manage resources effectively, while maintaining a high quality of care. For hospitals where admissions from the emergency department to the wards represent a large proportion of admissions, the ability to forecast these admissions and the resultant ward occupancy is especially useful for resource planning purposes. Since emergency admissions often compete with planned elective admissions, modeling emergency demand may result in improved elective planning as well. We compare several models for forecasting daily emergency inpatient admissions and occupancy. The models are applied to three years of daily data. By measuring their mean square error in a cross-validation framework, we find that emergency admissions are largely random, and hence, unpredictable, whereas emergency occupancy can be forecasted using a model combining regression and autoregressive integrated moving average (ARIMA) model, or a seasonal ARIMA model, for up to one week ahead. Faced with variable admissions and occupancy, hospitals must prepare a reserve capacity of beds and staff. Our approach allows estimation of the required reserve capacity.

Impact of High-Sensitivity Troponin I Testing with Sex-Specific Cutoffs on the Diagnosis of Acute Myocardial Infarction.

BACKGROUND High-sensitivity cardiac troponin I (hs-cTnI) assays show sex-dependent differences in the 99th percentile of healthy populations, with concentrations in women approximately 50% lower. The adoption of sex-specific cutoffs seems appropriate, although it is not yet clear what effect these will have on acute myocardial infarction (AMI) diagnosis and management. METHODS We conducted a retrospective pre- and postchangeover analysis of troponin I testing in the 6 months before and after moving from the contemporary Abbott Architect TnI assay (cTnI) to hs-cTnI at 2 tertiary centers in Australia and New Zealand. The cTnI cutoff was 30 ng/L for both sexes, whereas a female-specific cutoff of 16 ng/L was adopted upon changeover to hsTnI. RESULTS Changeover from the cTnI assay to the hs-cTnI assay increased the number of female patients with increased troponin I concentrations at both sites (from 29.7% to 34.9% and from 22.4% to 30.8%; P < 0.001). There was no statistically significant change in the number of men with increased concentrations in the same time period (P = 0.09). The increased percentage of women with increased troponin I was not associated with an increase in the number of women with AMI diagnoses at either center. Angiographic data available from 1 center showed no change in the percentage of angiograms performed in women. CONCLUSIONS Although increasing the proportion of women with increased troponin I, adopting sex-specific cutoffs with the hs-cTnI assay did not lead to an increase in AMI diagnoses in females, or in the number of women undergoing angiography.

Common laboratory tests predict imminent death in ward patients.

OBJECTIVE To estimate the ability of commonly measured laboratory variables to predict an imminent (within the same or next calendar day) death in ward patients. DESIGN Retrospective observational study. SETTING Two university affiliated hospitals. PATIENTS Cohort of 42,701 patients admitted for more than 24 hours and external validation cohort of 13,137 patients admitted for more than 24 hours. INTERVENTION We linked commonly measured laboratory tests with event databases and assessed the ability of each laboratory variable or combination of variables together with patient age to predict imminent death. MEASUREMENTS AND MAIN RESULTS In the inception teaching hospital, we studied 418,897 batches of tests in 42,701 patients (males 55%; average age 65.8 ± 17.6 years), for a total of >2.5 million individual measurements. Among these patients, there were 1596 deaths. Multivariable logistic modelling achieved an AUC-ROC of 0.87 (95% CI: 0.85-0.89) for the prediction of imminent death. Using an additional 105,074 batches from a cohort of 13,137 patients from a second teaching hospital, the multivariate model achieved an AUC-ROC of 0.88 (95% CI: 0.85-0.90). CONCLUSIONS Commonly performed laboratory tests can help predict imminent death in ward patients. Prospective investigations of the clinical utility of such predictions appear justified.

A prospective randomised controlled trial comparing tracheal intubation plus manual in-line stabilisation of the cervical spine using the Macintosh laryngoscope vs the McGrath(®) Series 5 videolaryngoscope.

Cervical spine immobilisation can make direct laryngoscopy difficult, which might lead to airway complications. This randomised control trial compared the time to successful intubation using either the Macintosh laryngoscope or the McGrath® Series 5 videolaryngoscope in 128 patients who had cervical immobilisation applied. Intubation difficulty score, Cormack & Lehane laryngoscopic view, intubation failures, changes in cardiovascular variables and the incidence of any complications were recorded. The mean (SD) successful intubation time with the Macintosh laryngoscope was significantly shorter compared with the McGrath laryngoscope, 50.0 (32.6) s vs 82.7 (80.0) s, respectively (p = 0.0003), despite the McGrath laryngoscopes having a lower intubation difficulty score and a superior glottic view. There were five McGrath laryngoscope intubation failures, three owing to difficulty in passing the tracheal tube and two to equipment malfunction. Equipment malfunction is a major concern as a reliable intubating device is vital when faced with an airway crisis.

Common laboratory tests predict imminent medical emergency team calls, intensive care unit admission or death in emergency department patients

To estimate the ability of commonly measured laboratory variables to predict imminent (within the same or next calendar day) medical emergency team (MET) calls, ICU admission or death.

Engineering a safe landing: engaging medical practitioners in a systems approach to patient safety

Background: Several event studies, including the Australian Safety and Quality in Healthcare Study, emphasize gaps in safety for hospitalized patients. It is now recognized that system‐based factors contribute significantly to risk of adverse events and this has led to a shift in focus of patient safety from the autonomous responsibility of medical clinicians to a systems‐based approach. The aim of this study was to determine medical practitioner awareness of, level of engagement in and barriers to engagement in a systems approach to patient safety and quality.

Text mining electronic hospital records to automatically classify admissions against disease: Measuring the impact of linking data sources

OBJECTIVE Text and data mining play an important role in obtaining insights from Health and Hospital Information Systems. This paper presents a text mining system for detecting admissions marked as positive for several diseases: Lung Cancer, Breast Cancer, Colon Cancer, Secondary Malignant Neoplasm of Respiratory and Digestive Organs, Multiple Myeloma and Malignant Plasma Cell Neoplasms, Pneumonia, and Pulmonary Embolism. We specifically examine the effect of linking multiple data sources on text classification performance. METHODS Support Vector Machine classifiers are built for eight data source combinations, and evaluated using the metrics of Precision, Recall and F-Score. Sub-sampling techniques are used to address unbalanced datasets of medical records. We use radiology reports as an initial data source and add other sources, such as pathology reports and patient and hospital admission data, in order to assess the research question regarding the impact of the value of multiple data sources. Statistical significance is measured using the Wilcoxon signed-rank test. A second set of experiments explores aspects of the system in greater depth, focusing on Lung Cancer. We explore the impact of feature selection; analyse the learning curve; examine the effect of restricting admissions to only those containing reports from all data sources; and examine the impact of reducing the sub-sampling. These experiments provide better understanding of how to best apply text classification in the context of imbalanced data of variable completeness. RESULTS Radiology questions plus patient and hospital admission data contribute valuable information for detecting most of the diseases, significantly improving performance when added to radiology reports alone or to the combination of radiology and pathology reports. CONCLUSION Overall, linking data sources significantly improved classification performance for all the diseases examined. However, there is no single approach that suits all scenarios; the choice of the most effective combination of data sources depends on the specific disease to be classified.

Advancing data management and usage in a major Australian health service: The REASON discovery platform™

The acquisition of data, its aggregation and usage is critical to the ongoing function of hospitals. This statement remains true across the numerous dimensions of that functioning - from clinical care to management and operations to research. At Alfred Health we have constructed a standards based informatics platform called REASON, with support for all of these functions in mind. Whilst analogous to more advanced examples of such platforms in the US, we believe this work is unique in the Australian context. In this paper we will outline work to date, lessons learned, and examples of how the health service has been positively impacted so far by the work. Finally, we will examine some future directions for the platform.

Designing a more efficient, effective and safe Medical Emergency Team (MET) service using data analysis

Introduction Hospitals have seen a rise in Medical Emergency Team (MET) reviews. We hypothesised that the commonest MET calls result in similar treatments. Our aim was to design a pre-emptive management algorithm that allowed direct institution of treatment to patients without having to wait for attendance of the MET team and to model its potential impact on MET call incidence and patient outcomes. Methods Data was extracted for all MET calls from the hospital database. Association rule data mining techniques were used to identify the most common combinations of MET call causes, outcomes and therapies. Results There were 13,656 MET calls during the 34-month study period in 7936 patients. The most common MET call was for hypotension [31%, (2459/7936)]. These MET calls were strongly associated with the immediate administration of intra-venous fluid (70% [1714/2459] v 13% [739/5477] p<0.001), unless the patient was located on a respiratory ward (adjusted OR 0.41 [95%CI 0.25–0.67] p<0.001), had a cardiac cause for admission (adjusted OR 0.61 [95%CI 0.50–0.75] p<0.001) or was under the care of the heart failure team (adjusted OR 0.29 [95%CI 0.19–0.42] p<0.001). Modelling the effect of a pre-emptive management algorithm for immediate fluid administration without MET activation on data from a test period of 24 months following the study period, suggested it would lead to a 68.7% (2541/3697) reduction in MET calls for hypotension and a 19.6% (2541/12938) reduction in total METs without adverse effects on patients. Conclusion Routinely collected data and analytic techniques can be used to develop a pre-emptive management algorithm to administer intravenous fluid therapy to a specific group of hypotensive patients without the need to initiate a MET call. This could both lead to earlier treatment for the patient and less total MET calls.

Spectrum of illness among returned Australian travellers from Bali, Indonesia: a 5-year retrospective observational study: Illness in Aust. travellers from Bali

Bali, Indonesia, presents significant infectious and non‐infectious health risks for Australian travellers. Understanding this spectrum of illnesses has the potential to assist clinicians in evaluating unwell returning travellers and guide provision of pre‐travel advice.