Calibrating Clinical Risk Calculators for New Zealand

The team working on the Clinical Risk Calculators project at the official “kick off meeting” held at Orion Health. From left to right, Dr Kevin Ross (GM, PDH), Dr Robyn Whitakker (WDHB), Delwyn Armstrong (WDHB), Prof. Bernhard Pfahringer (UoA), Dr Tom Robinson (WDHB), Dr Mirza Baig (Project PI), Prof. Rod Jackson (UoA), Anna Spyker (OH), Dr Kelly Atkinson (PDH), May-Lin Tye (OH), Dr Edmond Zhang (OH), Reece Robinson (OH), Dr Farhaan Mirza (AUT) and Dr Ehsan Ullah (ADHB). Every decision is a balancing act between risk and opportunity, but when a patient’s life is at stake, getting the right data to the clinician in real-time and understanding what the data means – so they can make the right call – is critical. As part of the groundbreaking programme Precision Driven Health, we have commenced a new research project that is examining existing clinical risk assessments and calculators and adapting them to our local New Zealand population. An example is the calculations done by healthcare providers to find out how soon a patient might return to hospital upon discharge. Those patients deemed ‘high risk’ are likely to receive more community care treatment than other patients, because the goal is to keep everyone out of hospital for as long as possible. This project involves a number of senior clinicians and academic researchers from district health boards and universities, as well as researchers from Orion Health. It will contribute to a growing body of international research that is examining how machine learning techniques can be applied to health data, and thereby enable the practice of precision medicine. Recent studies have indicated that current risk calculators are out of date and too generic for most groups. This project will test the hypothesis “Are risk calculators more effective when they are validated using the local population and context?” As part of the study, the team will design and develop a predictive model for assessing the clinical risk of hospital readmission. This will test the readmission risk prediction within a short period of time such as 30, 60, and 90 days after the date of discharge. Improving and updating risk calculators, and ensuring that they can be applied in a way that takes into account local population and practice, has the potential to greatly improve health outcomes. This is because a more accurate risk calculator can help prioritise the right patients, reducing delays in diagnosis and treatment that may result in serious – even fatal – systematic effects. There is also the issue of over-treatment, which has a significant impact on the patient, provider and overall healthcare system. Current risk assessments are based on a combination of several seemingly modest factors which may result in a much higher total risk than a single, more impressively raised factor. Therefore, systems to help estimate total risk accurately are required. Improving the accuracy of clinical risk assessments and calculators through this research will, we believe, provide some very tangible improvements in the delivery of healthcare globally. We are excited to have this national research project underway.