Real-time Clinical Decision Support Systems To Prevent Deterioration In Acute Care
By Robin Blythe, AusHSI Research Fellow – Policy Analysis
As the Australian hospital system becomes increasingly digital, it gains the capability for advanced analytics to improve patient care. Clinical decision support systems, which can provide clinicians with appropriate, actionable advice at patients’ bedsides, sit at the frontier of this movement. In the case of inpatient deterioration, when a patient’s condition begins to rapidly worsen in the hospital, these systems can be critically important to avert major adverse outcomes including admission to intensive care and death.
Yet, when examining these systems in practice, it seems that applying technological advancements to the clinical setting has not delivered better outcomes. Published evidence has highlighted that current decision support systems are too simple, telling healthcare providers what they already know, or are too complex to implement within existing clinical workflows.
My PhD research is focused on understanding what makes clinical decision support models useful for addressing clinical deterioration. My project will consist of a series of related studies. The first study, a scoping review of existing deterioration detection methods in practice and their impact on patient outcomes, is soon to be submitted for publication. The second study is currently being prepared, and will use a modified Delphi panel to qualitatively assess what kind of information clinicians value when treating deteriorating patients. This will be followed by gaining an understanding of how up-to-date patient information, such as vital signs and laboratory results, can be integrated into our decision process. The final step is to combine the learnings from the first three studies into the development and evaluation of a rigorous statistical model. The knowledge of information that clinicians value can be expressed through appropriate ways to summarise it with the aim of creating a decision tool that clinicians find useful.
Developed models will seek to address two questions: why is the patient deteriorating, and what can we do about it? Building a clinical model capable of understanding what is causing deterioration and how it can be addressed is an emerging field in medical informatics. It relies on clinical knowledge, rather than other fields such as machine learning which rely on associations between variables without any consideration of whether those relationships make sense in practice.
This research program is made possible by the Digital Health Collaborative Research Centre (DHCRC). The DHCRC is a national partnership between academics, private, and public sectors to develop and commercialise digital solutions for improved healthcare delivery.