Prediction model studies are increasingly common in the medical literature, and typically describe the development or validation of a prediction model. These models estimate the individualised probability or risk that a certain condition will occur in the future by combining information from multiple prognostic factors from an individual. Several reviews have demonstrated that there is an abundance of prediction models predicting the same outcomes for the same (or similar) patients or individuals. Unfortunatley, there is often conflicting evidence about their predictive (and comparative) performance. A systematic review of is therefore helpful to identify and appraise studies that developed or validated a prediction model, with meta-analysis needed to summarise their evidence.
In this talk, I will discuss the rationale and necessary steps to undertake a meta-analysis of a model's predictive performance. Subsequently, I will describe how these results can used to infer upon the need for further tailoring, and to inform the synthesis of prediction models.