reports

Developing decision trees to classify patients suited for similar interventions by combining clinical judgments with Leeds Risk Stratification Tool

London, UK: London School of Economics and Political Science.

London School of Economics and Political Science, London, UK

The objectives of the project are to provide a better understanding about how risk stratification tool is supposed to support decision making in NHS organisations, and to gain insights about how the Leeds Risk Stratification Tool is developed.

The approaches of the project are literature review, decision tree approach and interviews. The literature review will discuss the tools commissioned by the Department of Health, and focus on the underlying Adjusted Clinical Group (ACG) system employed by the Leeds Risk Stratification Tool. The decision tree approach is complemented by interviews with CCGs and NHS Leeds, as decision trees are developed by combing clinical judgments and outputs from the Leeds Risk Stratification Tool.

The ultimate deliverables of the project are two decision trees, which classify patients suited for similar interventions. The first decision tree identifies patients predicted to have medium resource use because of barriers to appropriate care. The second one identifies patients at high risk of avoidable unplanned hospitalisation who would benefit from enhanced case management.

The rationale of building two decision trees lies in the difference between these two cohorts of patients: while people with medium resource use do not realise they are ill, and they are reluctant to take part in intervention programmes; people with high risk of hospitalisation know their progressive health problems, but they have gaps in care management. The value of the decision tree modeling process is to help NHS Leeds and CCGs explore reasons why patients are having risk: The chief reason why patients have predictive medium resource use is that they have barriers in intervention; the chief reason why patients have high risk of hospitalisation is due to gaps in care. Therefore, the first decision tree helps to reduce the accelerating risk by early intervention targeted at barriers; and the second decision tree reduce unplanned hospitalisation by Intermediate Care and End-of-Life Care.