The challenge
Identifying at-risk patients for the Australian Government's Health Care Homes trial
More than 10,000 patients have been registered in the Australian Government's Health Care Homes trial, which aims to provide improved care for people with chronic and complex health conditions.
Patients enrolled in the trial, which will run until 30 June 2021, receive integrated care that is tailored to their individual needs.
They are looked after by a committed care team, have access to a shared care plan and receive better coordinated care.
Identifying which patients would benefit most from the Health Care Homes program was key to ensuring the trial can achieve better outcomes for patients, carers, and clinical staff.
Our response
A predictive risk algorithm to identify patients at risk of hospitalisation
Partnering with Precedence Health Care and Sonic Health Services, our data scientists at the Australian eHealth Research Centre developed and validated a predictive risk model to help identify patients who would be eligible for the Health Care Homes trial.
Developed using de-identified primary care data, the model considers patient demographics, their socio-economic index of relative advantage/disadvantage, physiological information, medications, chronic conditions, pathology categories according to abnormal levels in test results, and lifestyle factors such as alcohol and tobacco use.
Using this data, the predictive risk model algorithm determines the probability that a patient will make an unplanned hospital visit within 12 months.
The results
Our model has been installed in practices and services around Australia
In November 2018, our predictive risk tool was installed in approximately 120 health services around Australia participating in the Health Care Homes program.
Patients identified by our algorithm were contacted by the participating health service and invited to undertake a further assessment of their eligibility and Health Care Homes tier level.
After the two-year Health Care Homes trial, an independent evaluation will be carried out.
Access the predictive risk model algorithm
With the ability to identify people most at-risk of hospitalisation, our predictive risk model algorithm may in fact help to reduce the instances of hospitalisation, enabling primary healthcare providers to proactively deliver targeted care to their patients.
For this reason, the Department of Health has made the predictive risk model algorithm freely available for use under an Apache 2.0 open source license.