A new report from the PHG Foundation sets out a common framework for the evaluation and comparison of different models for predicting the risk of different diseases.
Risk prediction models for common diseases are an important element of public health practice, allowing screening and preventative interventions to be targeted to those at greatest risk of developing disease. As new data emerges, including on potentially predictive new biomarkers, the number of available and updated models increases – but there is no systematic approach to assessing their clinical performance.
Quality standards in risk prediction sets out a practical framework to assess risk prediction models, building on earlier PHG Foundation work on risk prediction for coronary heart disease, and based on the recommendations of international experts in the field. The new approach is based on assessments of three key ‘quality domains’ - the risk prediction model in question, the medical context in which it is to be used, and issues relating to clinical implementation of the model.
Comment: Using the best possible risk prediction model is critical for delivery of optimal health care, not only to identify and treat those at genuinely increased risk of developing serious disease, but also to ensure that others do not receive unnecessary treatments. Targeting treatments with the greatest accuracy avoids waste and minimises side-effects. This new approach to assessing and comparing current, updated and novel risk prediction models is therefore a valuable tool.