Variation has long been used as an indicator in the analysis and measurement of processes across a variety of industries. Large variation in outcomes, inputs or outputs is often considered as a sign that a process is not working as planned and examination and improvement may be required. In health, large amounts of variation in utilisation and provision may signal a cause for concern, raising questions of quality and equity.
In this post Professor Anne Duggan, Senior Staff Specialist, Gastroenterology, John Hunter Hospital, and Luke Slawomirski, Program Manager Implementation Support, at the Australian Commission on Safety and Quality in Health Care, discuss the recently published Exploring Healthcare Variation in Australia: Analyses resulting from an OECD study and what it might indicate about service provision:
Australian patients may be getting unnecessary interventions or missing out on needed care depending on where they live or which health services they attend.
A discussion paper co-authored by the Australian Commission on Safety and Quality in Health Care and the Australian Institute of Health and Welfare shows regional variation in the rates of common hospital procedures in 2010-11.
For example, people living in northern South Australia were three times more likely to be admitted for knee arthroscopy than residents of inner west Sydney or far north Queensland [Figure 1].
1. Rates are age and sex standardised to the 30 June 2001 Australian population; 2. Data for rates based on a small number of admissions are unshaded. Source: AIHW analysis of National Hospital Morbidity Database.
Figure 1 is taken from the paper and shows variation in knee arthroscopy admission rates in 2010-11 by Medicare Local of patient residence; it also reveals the high proportion of procedures undertaken in the private sector.
Variation in cardiac care was also observed. People living in the Grampians region (Victoria) were over three times more likely to have a coronary bypass than Fremantle residents [Figure 2]. For coronary angiography (an invasive diagnostic procedure), the difference across the country was 7-fold.
Note: The five groups are based on age and sex standardised rates. The range within each group is as follows: Lowest (32–58); 2nd (59–67); 3rd (68–73); 4th (74–82); Highest (83–105). Source: AIHW analysis of National Hospital Morbidity Database.
Figure 2. This map, taken from the paper, shows variation in admission rates for coronary bypass in 2010-11 by Medicare Local of patient residence.
Regional variation in admissions for hysterectomy (without any diagnosis of cancer) was 3-fold. Women living in metropolitan areas were less likely to undergo this procedure than women living in regional areas [Figure 3].
The analysis was based on where patients lived, defined by geographic boundaries of Medicare Locals. However, the approach can be used on any chosen geographic area.
1. Rates are age standardised to the 30 June 2001 Australian population; 2. Peer groups were established based on three criteria: (a) proximity of each Medicare Local to major metropolitan cities; (b) proximity to major hospitals; and (c) socioeconomic status. See Section B for further information. Source: AIHW analysis of National Hospital Morbidity Database.
Figure 3 is taken from the paper, and shows variation in admission rates for hysterectomy (without diagnosis of cancer) in 2010-11 by Medicare Local of patient residence; Medicare Locals are arranged by peer group of similar remoteness/rurality.
Wouldn’t we expect to see some variation?
Some level of variation in how health care is provided and utilised is expected. For instance, it may reflect differences in the healthcare needs of people living in a certain region. The study accounted for these differences by standardising the data for age and gender, and by comparing ‘like with like’ regions in terms of rurality and remoteness. Variation was also observed within these similar regions (see hysterectomy chart).
There may, of course, be other factors. For example, the people of northern SA may play more sport than their Sydney or northern Queensland counterparts, and sustain more knee injuries requiring keyhole surgery.
Nevertheless, judging by the extent and pattern of the observed variation, it is likely that at least some of the variation is unwarranted: that is, driven by factors other than patient need or preference.
Who receives the ‘right’ level of care?
Without better data on outcomes of care it is difficult to know which of the regions is delivering the most appropriate rate. Appropriate care means patients receiving best-evidence care that is suited to their needs and preferences.
Appropriateness is important for two reasons:
- When patients receive unnecessary treatment they are placed under undue risk (no healthcare intervention is risk-free). In addition scarce resources are wasted.
- When patients miss out on care they need, we have a problem with ensuring equitable access to health care.
What is needed is research into the causes of variation and patient preferences and outcomes. Action is also required by clinical groups and government agencies to identify the actual extent of unwarranted variation, and to put in place measures to ensure appropriate care.
More appropriate care can improve not only equity of access, but also value for money and productivity in our healthcare system.
These are very important considerations at a time when health budgets and the notion of universal access to care are both under unprecedented strain.