Using patients' experiences of adverse events to improve health service delivery and practice: protocol of a data linkage study of Australian adults age 45 and above

Evidence of patients' experiences is fundamental to creating effective health policy and service responses, yet is missing from our knowledge of adverse events. This protocol describes explorative research redressing this significant deficit; investigating the experiences of a large cohort of recently hospitalised patients aged 45 years and above in hospitals in New South Wales (NSW), Australia.

Data Integration Protocol In Ten-steps (DIPIT) : a new standard for medical researchers

The exponential increase in data, computing power and the availability of readily accessible analytical software has allowed organisations around the world to leverage the benefits of integrating multiple heterogeneous data files for enterprise-level planning and decision making. Benefits from effective data integration to the health and medical research community include more trustworthy research, higher service quality, improved personnel efficiency, reduction of redundant tasks, facilitation of auditing and more timely, relevant and specific information. The costs of poor quality processes elevate the risk of erroneous outcomes, an erosion of confidence in the data and the organisations using these data. To date there are no documented set of standards for best practice integration of heterogeneous data files for research purposes. Therefore, the aim of this paper is to describe a set of clear protocol for data file integration (Data Integration Protocol In Ten-steps; DIPIT) translational to any field of research.

Psychiatric disorders and unmet needs in Australian police cells

To determine the prevalence of current psychiatric disorders and unmet needs in a sample of police cell detainees in Victoria. A cross-sectional descriptive study was conducted, including data linkage with the Victoria Police database and the Victorian Psychiatric Case Register. In Melbourne, Australia, 150 detainees were recruited from two busy metropolitan police stations. Outcome measures included estimated rates of psychiatric disorders, using the Structured Clinical Interview for DSM-IV-TR, and individual needs, using the Camberwell Assessment of Need – Forensic Version. One quarter (n = 32, 25.4%) of detainees had a prior admission to a psychiatric hospital, and three quarters met current criteria for a diagnosable mental disorder. The most common disorders were substance dependence (n = 81, 54%) and mood disorders (n = 60, 40%). A third met diagnostic criteria for both a mental illness and a substance use disorder. The odds of being classified with mood (OR = 10.1), anxiety (OR = 2.2), psychotic (OR = 15.4) and substance use disorders (OR = 26.3) were all significantly higher in the current sample as compared with the general population. Detainees with a mental illness identified significantly more needs and significantly more unmet needs (e.g. psychological distress) than those who did not rate as having a current mental illness. There remains a pressing need to evaluate standardized screening tools for mental illnesses in police cells to provide timely access to assessment and treatment services. The need for functional interagency collaborations are highlighted and discussed.

A systematic approach to chronic heart failure care : a consensus statement

The National Heart Foundation of Australia assembled an expert panel to provide guidance on policy and system changes to improve the quality of care for people with chronic heart failure (CHF). The recommendations have the potential to reduce emergency presentations, hospitalisations and premature death among patients with CHF. Best-practice management of CHF involves evidencebased, multidisciplinary, patient-centred care, which leads to better health outcomes. A CHF care model is required to achieve this. Although CHF management programs exist, ensuring access for everyone remains a challenge. This is particularly so for Aboriginal and Torres Strait Islander peoples, those from non-metropolitan areas and lower socioeconomic backgrounds, and culturally and linguistically diverse populations. Lack of data and inadequate identifi cation of people with CHF prevents effi cient patient monitoring, limiting information to improve or optimise care. This leads to ineff ectiveness in measuring outcomes and evaluating the CHF care provided. Expanding current cardiac registries to include patients with CHF and developing mechanisms to promote data linkage across care transitions are essential. As the prevalence of CHF rises, the demand for multidisciplinary workforce support will increase. Workforce planning should provide access to services outside of large cities, one of the main challenges it is currently facing. To enhance community-based management of CHF, general practitioners should be empowered to lead care. Incentive arrangements should favour provision of care for Aboriginal and Torres Strait Islander peoples, those from lower socioeconomic backgrounds and rural areas, and culturally and linguistically diverse populations. Ongoing research is vital to improving systems of care for people with CHF. Future research activity needs to ensure the translation of valuable knowledge and high quality evidence into practice.

Feasibility of trialling cord blood stem cell treatments for cerebral palsy in Australia

Umbilical cord blood may have therapeutic benefit in children with cerebral palsy (CP), but further studies are required. On first appearance it seems that Australia is well placed for such a trial because we have excellence in CP research backed by extensive CP registers, and both public and private cord blood banks. We aimed to examine the possibilities of conducting a trial of autologous umbilical cord blood cells (UCBCs) as a treatment for children with CP in Australia.

Identification and outcomes of clinical phenotypes in amyotrophic lateral sclerosis/motor neuron disease: Australian National Motor Neuron Disease observational cohort

OBJECTIVE: To capture the clinical patterns, timing of key milestones and survival of patients presenting with amyotrophic lateral sclerosis/motor neuron disease (ALS/MND) within Australia.

METHODS: Data were prospectively collected and were timed to normal clinical assessments. An initial registration clinical report form (CRF) and subsequent ongoing assessment CRFs were submitted with a completion CRF at the time of death.

DESIGN: Prospective observational cohort study.

PARTICIPANTS: 1834 patients with a diagnosis of ALS/MND were registered and followed in ALS/MND clinics between 2005 and 2015.

RESULTS: 5 major clinical phenotypes were determined and included ALS bulbar onset, ALS cervical onset and ALS lumbar onset, flail arm and leg and primary lateral sclerosis (PLS). Of the 1834 registered patients, 1677 (90%) could be allocated a clinical phenotype. ALS bulbar onset had a significantly lower length of survival when compared with all other clinical phenotypes (p<0.004). There were delays in the median time to diagnosis of up to 12 months for the ALS phenotypes, 18 months for the flail limb phenotypes and 19 months for PLS. Riluzole treatment was started in 78-85% of cases. The median delays in initiating riluzole therapy, from symptom onset, varied from 10 to 12 months in the ALS phenotypes and 15-18 months in the flail limb phenotypes. Percutaneous endoscopic gastrostomy was implemented in 8-36% of ALS phenotypes and 2-9% of the flail phenotypes. Non-invasive ventilation was started in 16-22% of ALS phenotypes and 21-29% of flail phenotypes.

CONCLUSIONS: The establishment of a cohort registry for ALS/MND is able to determine clinical phenotypes, survival and monitor time to key milestones in disease progression. It is intended to expand the cohort to a more population-based registry using opt-out methodology and facilitate data linkage to other national registries.

Analysis of protein sequence and interaction data for candidate disease gene prediction

Linkage analysis is a successful procedure to associate diseases with specific genomic regions. These regions are often large, containing hundreds of genes, which make experimental methods employed to identify the disease gene arduous and expensive. We present two methods to prioritize candidates for further experimental study: Common Pathway Scanning (CPS) and Common Module Profiling (CMP). CPS is based on the assumption that common phenotypes are associated with dysfunction in proteins that participate in the same complex or pathway. CPS applies network data derived from protein–protein interaction (PPI) and pathway databases to identify relationships between genes. CMP identifies likely candidates using a domain-dependent sequence similarity approach, based on the hypothesis that disruption of genes of similar function will lead to the same phenotype. Both algorithms use two forms of input data: known disease genes or multiple disease loci. When using known disease genes as input, our combined methods have a sensitivity of 0.52 and a specificity of 0.97 and reduce the candidate list by 13-fold. Using multiple loci, our methods successfully identify disease genes for all benchmark diseases with a sensitivity of 0.84 and a specificity of 0.63. Our combined approach prioritizes good candidates and will accelerate the disease gene discovery process.

Validation of de-identified record linkage to ascertain hospital admissions in a cohort study

Background Cohort studies can provide valuable evidence of cause and effect relationships but are subject to loss of participants over time, limiting the validity of findings. Computerised record linkage offers a passive and ongoing method of obtaining health outcomes from existing routinely collected data sources. However, the quality of record linkage is reliant upon the availability and accuracy of common identifying variables. We sought to develop and validate a method for linking a cohort study to a state-wide hospital admissions dataset with limited availability of unique identifying variables.

Methods A sample of 2000 participants from a cohort study (n = 41 514) was linked to a state-wide hospitalisations dataset in Victoria, Australia using the national health insurance (Medicare) number and demographic data as identifying variables. Availability of the health insurance number was limited in both datasets; therefore linkage was undertaken both with and without use of this number and agreement tested between both algorithms. Sensitivity was calculated for a sub-sample of 101 participants with a hospital admission confirmed by medical record review.

Results Of the 2000 study participants, 85% were found to have a record in the hospitalisations dataset when the national health insurance number and sex were used as linkage variables and 92% when demographic details only were used. When agreement between the two methods was tested the disagreement fraction was 9%, mainly due to "false positive" links when demographic details only were used. A final algorithm that used multiple combinations of identifying variables resulted in a match proportion of 87%. Sensitivity of this final linkage was 95%.

Conclusions High quality record linkage of cohort data with a hospitalisations dataset that has limited identifiers can be achieved using combinations of a national health insurance number and demographic data as identifying variables.

Using clinical trial data and linked administrative health data to reduce the risk of adverse events associated with the uptake of newly released drugs by older Australians: a model process

BACKGROUND: The study was undertaken to evaluate the contribution of a process which uses clinical trial data plus linked de-identified administrative health data to forecast potential risk of adverse events associated with the use of newly released drugs by older Australian patients. METHODS: The study uses publicly available data from the clinical trials of a newly released drug to ascertain which patient age groups, gender, comorbidities and co-medications were excluded in the trials. It then uses linked de-identified hospital morbidity and medications dispensing data to investigate the comorbidities and co-medications of patients who suffer from the target morbidity of the new drug and who are the likely target population for the drug. The clinical trial information and the linked morbidity and medication data are compared to assess which patient groups could potentially be at risk of an adverse event associated with use of the new drug. RESULTS: Applying the model in a retrospective real-world scenario identified that the majority of the sample group of Australian patients aged 65 years and over with the target morbidity of the newly released COX-2-selective NSAID rofecoxib also suffered from a major morbidity excluded in the trials of that drug, indicating a substantial potential risk of adverse events amongst those patients. This risk was borne out in post-release morbidity and mortality associated with use of that drug. CONCLUSIONS: Clinical trial data and linked administrative health data can together support a prospective assessment of patient groups who could be at risk of an adverse event if they are prescribed a newly released drug in the context of their age, gender, comorbidities and/or co-medications. Communication of this independent risk information to prescribers has the potential to reduce adverse events in the period after the release of the new drug, which is when the risk is greatest.Note: The terms 'adverse drug reaction' and 'adverse drug event' have come to be used interchangeably in the current literature. For consistency, the authors have chosen to use the wider term 'adverse drug event' (ADE).

Monitoring and evaluating surgical care: defining perioperative mortality rate and standardising data collection

BACKGROUND: Case volume per 100 000 population and perioperative mortality rate (POMR) are key indicators to monitor and strengthen surgical services. However, comparisons of POMR have been restricted by absence of standardised approaches to when it is measured, the ideal denominator, need for risk adjustment, and whether data are available. We aimed to address these issues and recommend a minimum dataset by analysing four large mixed surgical datasets, two from well-resourced settings with sophisticated electronic patient information systems and two from resource-limited settings where clinicians maintain locally developed databases. METHODS: We obtained data from the New Zealand (NZ) National Minimum Dataset, the Geelong Hospital patient management system in Australia, and purpose-built surgical databases in Pietermaritzburg, South Africa (PMZ) and Port Moresby, Papua New Guinea (PNG). Information was sought on inclusion and exclusion criteria, coding criteria, and completeness of patient identifiers, admission, procedure, discharge and death dates, operation details, urgency of admission, and American Society of Anesthesiologists (ASA) score. Date-related errors were defined as missing dates and impossible discrepancies. For every site, we then calculated the POMR, the effect of admission episodes or procedures as denominator, and the difference between in-hospital POMR and 30-day POMR. To determine the need for risk adjustment, we used univariate and multivariate logistic regression to assess the effect on relative POMR for each site of age, admission urgency, ASA score, and procedure type. FINDINGS: 1 365 773 patient admissions involving 1 514 242 procedures were included, among which 8655 deaths were recorded within 30 days. Database inclusion and exclusion criteria differed substantially. NZ and Geelong records had less than 0·1% date-related errors and greater than 99·9% completeness. PMZ databases had 99·9% or greater completeness of all data except date-related items (94·0%). PNG had 99·9% or greater completeness for date of birth or age and admission date and operative procedure, but 80-83% completeness of patient identifiers and date related items. Coding of procedures was not standardised, and only NZ recorded ASA status and complete post-discharge mortality. In-hospital POMR range was 0·38% in NZ to 3·44% in PMZ, and in NZ it underestimated 30-day POMR by roughly a third. The difference in POMR by procedures instead of admission episodes as denominator ranged from 10% to 70%. Age older than 65 years and emergency admission had large independent effects on POMR, but relatively little effect in multivariate analysis on the relative odds of in-hospital death at each site. INTERPRETATION: Hospitals can collect and provide data for case volume and POMR without sophisticated electronic information systems. POMR should initially be defined by in-hospital mortality because post-discharge deaths are not usually recorded, and with procedures as denominator because details allowing linkage of several operations within one patient's admission are not always present. Although age and admission urgency are independently associated with POMR, and ASA and case mix were not included, risk adjustment might not be essential because the relative odds between sites persisted. Standardisation of inclusion criteria and definitions is needed, as is attention to accuracy and completeness of dates of procedures, discharge and death. A one-page, paper-based form, or alternatively a simple electronic data collection form, containing a minimum dataset commenced in the operating theatre could facilitate this process. FUNDING: None.

Building cloud-based healthcare data mining services

The linkage between healthcare service and cloud computing techniques has drawn much attention lately. Up to the present, most works focus on IT system migration and the management of distributed healthcare data rather than taking advantage of information hidden in the data. In this paper, we propose to explore healthcare data via cloud-based healthcare data mining services. Specifically, we propose a cloud-based healthcare data mining framework for healthcare data mining service development. Under such framework, we further develop a cloud-based healthcare data mining service to predict patients future length of stay in hospital.

A hybrid approach to clustering in big data

Clustering of big data has received much attention recently. In this paper, we present a new clusiVAT algorithm and compare it with four other popular data clustering algorithms. Three of the four comparison methods are based on the well known, classical batch k-means model. Specifically, we use k-means, single pass k-means, online k-means, and clustering using representatives (CURE) for numerical comparisons. clusiVAT is based on sampling the data, imaging the reordered distance matrix to estimate the number of clusters in the data visually, clustering the samples using a relative of single linkage (SL), and then noniteratively extending the labels to the rest of the data-set using the nearest prototype rule. Previous work has established that clusiVAT produces true SL clusters in compact-separated data. We have performed experiments to show that k-means and its modified algorithms suffer from initialization issues that cause many failures. On the other hand, clusiVAT needs no initialization, and almost always finds partitions that accurately match ground truth labels in labeled data. CURE also finds SL type partitions but is much slower than the other four algorithms. In our experiments, clusiVAT proves to be the fastest and most accurate of the five algorithms; e.g., it recovers 97% of the ground truth labels in the real world KDD-99 cup data (4 292 637 samples in 41 dimensions) in 76 s.