Broad descriptions of variation are a useful way of finding disparities, but to generate change they need to be carefully analysed and applied
The Supplement accompanying this issue of the MJA addresses variation in health care delivery as an important theme in ensuring appropriate care across the health system. While all would agree that health care should be provided as and where needed, and that only appropriate care should be delivered, difficulties arise when seeking consensus on what variation is “appropriate” and what should be done about it.
Complex systems are inherently variable. Indeed, a multitude of data shows variation in clinical practice but there is less consensus on its implications. For example, while one focus is high utilisation, there is also an issue with underutilisation, raising issues of equity. Aboriginal and Torres Strait Islander people receive fewer interventions for a range of major health conditions than non-Indigenous Australians; the age standardised rate of coronary procedures is 40% lower.1 Overall, the lowest quintile by socio-economic status of area of residence has 78.8 elective hospital admissions involving elective surgery per 1000 population compared with 91.5 for the highest quintile.2
Geographic variation is the focus of the recent Australian Atlas of Health Care Variation,3 summarised in the Supplement by Buchan and colleagues.4 The Atlas compares extreme high and low rates of interventions, using about 350 areas. An earlier report by the Australian Commission on Safety and Quality in Health Care (ACSQHC) and the Australian Institute of Health and Welfare calculated a systematic component of variation and provided a comprehensive picture covering 60 Medicare Locals, which were in place at that time.5
The two reports focus attention on variation, emphasising unwarranted variation, but crucially they do not define “unwarranted”. While presenting a high profile call for action, priority areas for action are not suggested.
It is not sufficient to publish an Atlas of Variation … and expect it to have remarkable impact. An Atlas of Variation needs to be an integral part of a larger transformational change program.6
The limited available evidence indicates that public reporting alone has minimal impact on changing clinical practice.7 Change may also be facilitated by considering local factors.8 Atkinson and colleagues identified 13 factors required for system change, including microsystem capacity, credible evidence, engagement, peer support and integration into routine practice.9
In 2016, the Grattan Institute released an analysis of variation in preventable conditions across Victoria and Queensland.10 The report set the bar higher for addressing variation, noting the need for a time series when targeting action. The Institute found that 15% of areas exceeded their variation benchmark each year, but only 5% exceeded it for 3 successive years. It proposed identifying areas showing sustained variation and developing response strategies in consultation with all stakeholders.
Financial incentives are another approach. In the Supplement, Hall and van Gool report disappointing evidence that financial incentives improve quality, and note possible unintended effects through gaming or patient selection.11 Moreover, complex Australian health care financing arrangements make it difficult to share the economic benefits of reductions in practice variations among the various stakeholders. Caution is therefore necessary in considering financial incentives.
A weakness of variation data is that they are always explicable by specific variables, such as patient factors or legitimate process differences. One way to deal with this is to use variation data to identify where change may appear warranted, and then focus on the development of mutually agreed standards of best practice care. Contributions in the Supplement reflect the important work that the ACSQHC has undertaken in this regard: Chew and colleagues describe the process for developing clinical care standards in Australia;12 important examples are provided by Caplan and colleagues (cognitive impairment)13 and Turnidge and colleagues (antimicrobial stewardship).14 However, as noted above, these need to be incorporated into broader micro and macro change strategies.
Wilcox and McNeil argue that clinical registries can play a role in identifying quality interventions and clinical pathways, and indeed the ACSQHC has published guidelines for clinical registries.15 Registries are potentially a link between high level policy and practice variables of interest to a given clinical community. A strategy is now needed to link clinical registries with national hospital data collections, to better inform understanding of in-patient activity variation and performance.
In conclusion, broad descriptions of variation are a useful way of finding disparities, but will not of themselves generate change. To point the way to useful policy change, they need to be carefully analysed and applied.
Variation needs to be addressed as part of the broader quality and safety agenda, where political commitment has been hard won over the past 20 years. Regular, considered reporting of variation may assist, but is not in itself a magic bullet.