Mortality Patterns Related to Hospital Admissions in NSW

Speaker: Blanca Gallego Luxan¹, Oscar Perez Concha²

Affiliation: Centre for Health Informatics, UNSW¹, Centre for Health Informatics, UNSW²

Time: Monday 24/10/2011 from 14:00 to 15:00

Venue: Access Grid UWS. Presented from Campbelltown (26.1.50), accessible from Parramatta (EB.1.32) and Penrith (Y239).

Abstract:

Part1: Effect of hospital AEs in during admission and post-discharge mortality.

By linking data from more than 14 million hospital admissions with death registry data we have been able to analyse the impact of adverse events during hospitalisation on death rates, in far greater detail than in past studies. We found that the presence of at least one adverse event during admission is associated with 3.38% death rate, compared to 0.89% without. At 1 year post-discharge, the percentages of admissions preceding death become 12.98% vs. 6.10%, respectively. Risk was higher for certain patient groups, for example, a 15-fold increased risk of dying during admission for patients diagnosed with angina pectoris and with other primary gonarthrosis. Mortality up to 1 year post-discharge was greatly increased for principal diagnoses of open wound of lower leg and malignant neoplasm of the skin of upper limb among others.

Part2: Weekend Temporal Patterns of patient mortality.

We analysed temporal trends and patterns of mortality over 7 years (2000-2007) of hospital admissions by computing yearly adjusted mortality rates and adjusted odds ratios for each diagnosis-related group (DRG) and admission pathway (emergency department versus other admissions). We found that the overall adjusted mortality rate of weekday admissions presented a negative linear trend (1.22% [1.20-1.23]-year 1, 1.10% [1.08-1.11]-year 7), whereas the counterpart for weekend admissions showed a positive trend for the first four years followed by a negative trend (1.34% [1.30-1.66]-year 1, 1.44% [1.39-1.76]-year 4, 1.23%[1.18-1.47]-year 7). Patients belonging to a set of 10 diagnostic groups (8 of them cancer diagnoses) were consistently at a higher risk of dying if admitted during the weekend.

Biography¹:

Dr Gallego leads the Modelling and Simulation in Health research team at the Centre for Health Informatics, (UNSW). This group is developing new empirical models for the analysis, assessment and prediction of healthcare delivery and for the integration of new sources of information into public health and clinical decision making.

Trained as a physicist, she obtained a Ph.D. in Atmospheric and Oceanic Sciences from the University of California at Los Angeles (UCLA) after which she worked as a postdoctoral fellow with the Integrated Sustainability Analysis group at the University of Sydney, before joining the Centre for Health Informatics in 2006.

Dr Gallego has extensive international research experience in data analysis and computational modelling and has made significant and innovative contributions to the design, analysis and development of models derived from complex empirical data for a wide range of applications such as patient safety, biosurveillance, corporate sustainability reporting, ecological footprint analysis and climate variability.

Biography²:

Dr Perez Concha's research interests are in the field of machine learning, pattern recognition, modelling and prediction of spatio-temporal systems, non linear dynamical systems and chaos theory.

Oscar obtained his MS (2003) in Telecommunication Engineering at the Technical University of Madrid (UPM) and his Ph.D in Computer Science (2008) from the Carlos III University of Madrid (UC3M). Subsequently, he worked as a postdoctoral research associate in the area of tracking and activity recognition in video sequences in the Research Centre for Innovation in IT Services and Applications (iNEXT) at University of Technology, Sydney (UTS).

He joined the Centre for Health Informatics (UNSW) in 2011.

As a part of the Patient Safety Program Grant at the Australian Institute of Health Innovation (AIHI), Oscar is researching time series of hospitalisation data for the analysis, assessment, and prediction of spatio-temporal behaviour in the Health Care System.