By Krystyn Villaflores, La Trobe University
According to the Australian Institute of Health and Welfare, expenditure on health totalled $130.3 billion in 2010/11, rising $7.8 billion from the previous year. That averages out to an annual spending of $5796 on the health needs of each Australian. With the strain on the health care system projected to continue rising steadily in the decades to come, one needs to look at how and where the money is being spent, and the ways in which we can improve efficiency.
One way to dissect health spending is to look at the impact made by certain conditions, since approximately two thirds of expenditure can be attributed to disease groupings. Cardiovascular disease alone contributes an estimated cost of $8 billion yearly, greater than that of any other condition. Seeing the enormous costs associated with disease, it is impossible for one to deny the need for research in this area.
This begs the question, in what way can one conduct research that will minimise the stress put on the public and private health sectors attributable to disease? This is a question I decided to tackle for my summer vacation research scholarship, more specifically, by looking at disease prediction models.
One of the main goals of prognostic disease modelling is to quantify individual risk of developing disease, which has an obvious impact on the health care system. This risk is the basis of treatment and intervention protocols being implemented for an individual. The importance of obtaining precise and unbiased risk estimates cannot be understated, as only those whose risk is estimated to be over a particular threshold will be recommended treatment, leading to very apparent negative effects associated with both over- and underestimation. Too-conservative estimates result in individuals forgoing preventative treatment measures, leading to long-term stress on the health care system and poorer individual outcomes. Erroneously high-risk estimates lead to patients being prescribed unnecessary treatments and medications. With the cost of health care skyrocketing and the impact of pharmaceutical benefits set to soar more than any other component of Australian Government health expenditure, these are ill effects we simply cannot afford.
The estimation of this risk becomes more complicated under particular study designs, and this gets to the heart of my research. The gold standard of study designs when it comes to medical and epidemiological research is the cohort study. This involves following a large number of individuals, often tens of thousands, over a period of several years and taking stock of an assortment of measurements. With the advent of new technology the array of things that we can test and measure in this field is ever growing. Unfortunately, this often comes with a high price tag, making the cohort study often inappropriate. Without progressing to statistical geek speak to describe the intricacies of my project, we proposed a method for obtaining valid risk estimates from only a subset of individuals. What we found was that under the conditions that we tested, it is possible to acquire accurate risk estimates, in the process refuting some of the current literature. This is a really exciting result, as it opens the doors for research in this area; using this methodology means a significant reduction in time, effort and costs. With further development in this area, and corroborating evidence surely to follow, in the years to come our pockets and bodies will definitely be thankful!
Krystyn Villaflores was one of the recipients of a 2013/14 AMSI Vacation Research Scholarship.