Predictive risk adjustment tools use relationships in historic, routinely collected electronic health data to determine the expected future health care resource use of each individual in a population. The tools can be used to estimate future events for people at different levels of risk, providing commissioners with more accurate estimates of likely future costs.
In the US and Europe, risk adjustment models are used widely to help determine health payments, either for fixing ‘capitated’ budgets or for deciding reimbursement rates for individual patients. In the NHS, the most widespread use of these techniques so far has been in the use of ‘case finding’ tools, such as the Patients At Risk of Re-hospitalisation (PARR) and the ’Combined Model’.
The Nuffield Trust has considerable experience in the application of predictive risk modelling techniques
Predictive risk and health care: an overview provides a useful overview of how risk adjustment techniques are currently being used in the NHS, considering the principal applications of risk adjustment. The research summary also looks at emerging developments, including modelling with social care data, predicting the impact of preventive care and making shorter-term predictions of readmissions.
The Nuffield Trust has considerable experience in the application of predictive risk modelling techniques. This includes work on the development of case finding tools such as PARR and the Combined Model; a feasibility study of models that predict the future use of social care by individuals; work on person-based resource allocation; and a range of national evaluations of interventions to reduce hospital use such as the Whole System Demonstrator Project, integrated care pilots, and selected Partnership for Older People Projects (POPPs) and Virtual Wards.
This summary, by Theo Georghiou, Adam Steventon, Professor John Billings, Ian Blunt, Dr Geraint Lewis, and Dr Martin Bardsley, will be of interest to commissioners and policy-makers, along with researchers and others with an interest in this field.