Preventable hospitalisations: geographic variation is more a measure of the health of the population than of GP supply

Michael Falster, from the Centre for Big Data Research in Health, UNSW tells us about his research which has won the HSRAANZ Best PhD Student Prize.

“Preventable hospitalisations” are used internationally to measure the performance of primary healthcare, and are a key health performance indicator in the Australian National Healthcare Agreement. While this indicator is used to allocate funding to health services, there is surprisingly little empirical evidence to supports its use and interpretation.

In research conducted at the University of New South Wales and Western Sydney University, we used linked questionnaire and administrative hospital data for 267,091 participants in the Sax Institute’s 45 and Up Study to explore the relative roles of personal and health system factors in driving geographic variation in preventable hospitalisation.

We found that GP supply, measured as full-time workload equivalents, was not a significant predictor of preventable hospitalisation, and explained only a small amount (2.9%) of the geographic variation in hospitalisation rates. Conversely, more than one third (36.9%) of variation was driven by the socio-demographic composition, health and behaviours of the population. These personal characteristics explained a greater amount of the variation for chronic conditions (37.5%) than acute (15.5%) or vaccine-preventable conditions (2.4%).

These results have direct implications for the measurement and interpretation of preventable hospitalisations in Australia, and suggest the most appropriate policy responses are long-term policy strategies to promote healthy living, as well as targeted local interventions to efficiently manage the current burden of high-risk and chronically ill patients. When used for performance comparison purposes, reporting of geographic rates of preventable hospitalisation by individual conditions or potential pathways for intervention could help guide policymakers to best respond to the indicator in a timely manner.

The research was part of the NHMRC-funded APHID (Assessing Preventable Hospitalisation InDicators) Study, with partners the Australian Commission on Safety and Quality in Health Care, the NSW Bureau of Health Information and the NSW Agency for Clinical Innovation. Through these policy partners we continue to advise on the evidence-based use of this health performance indicator in Australia.

The research paper is available at,.9.aspx

Falster MO, Jorm LR, Douglas KA, et al. Sociodemographic and health characteristics, rather than primary care supply, are major drivers of geographic variation in preventable hospitalizations in Australia. Med Care 2015;53:436-445

Michael Falster is a Biostatistician and Research Fellow at the Centre for Big Data Research in Health (CBDRH) at UNSW Australia where he is also completing his PhD . Michael has over 10 years’ experience working in public health, biostatistics and epidemiological research, and is currently project coordinator on the Assessing Preventable Hospitalisation InDicators (APHID) Study, an NHMRC funded partnership grant using linked data to explore contributors to geographic variation in ‘preventable’ hospitalisations.  Michael’s work and interests are characterized by finding innovative statistical methods for quantifying and exploring variation in health and health care, such as: multilevel models for deconstructing geographic variation in health inequalities and outcomes; data visualizations exploring temporal patterns of health events; spatial methods for identifying and analyzing hospital patient catchments; and data algorithms for characterizing longitudinal patterns of healthcare use.  Michael has experience in diverse fields such as health services research, injury, Aboriginal health, cancer epidemiology and perinatal research, and experience analyzing complex linked data sources including survey, hospital, Medicare, mortality, perinatal, cancer notification and emergency department data sets. Having a background in health, policy and statistics, Michael is interested in translating complex statistical methods and findings towards a policy audience.