A clinical prediction model (CPM) is a tool for predicting healthcare outcomes, usually within a specific population and context. A common approach is to develop a new CPM for each population and context, however, this wastes potentially useful historical information. A better approach is to update or incorporate the existing CPMs already developed for use in similar contexts or populations. In addition, CPMs commonly become miscalibrated over time, and need replacing or updating. In this paper we review a range of approaches for re-using and updating CPMs; these fall in three main categories: simple coefficient updating; combining multiple previous CPMs in a meta-model; and dynamic updating of models. We evaluated the performance (discrimination and calibration) of the different strategies using data on mortality following cardiac surgery in the UK: We found that no single strategy performed sufficiently well to be used to the exclusion of the others. In conclusion, useful tools exist for updating existing CPMs to a new population or context, and these should be implemented rather than developing a new CPM from scratch, using a breadth of complementary statistical methods.
展开▼
