Does predicting who will fall tomorrow tell you who is frail today?

A frequently raised question about frailty indices is whether certain variables should be weighted more heavily than others. In this commentary, we outline some of the pros and cons of weighting. Weighting can improve prediction of outcomes, but it also carries risks: underestimating the effects of existing interventions, prioritising factors that may not align with classical frailty phenotypes, and the risk that a model will not predict outcomes in the same way in a new population as it did in the one it was trained on. We offer an alternative perspective that frailty indices need not be seen only as prediction tools. Used as latent measures of health, they can support and complement clinical assessment. The true test of any data-derived frailty measure should be whether clinicians find it helpful in practice and whether its use can reduce the rate of adverse outcomes in a randomised controlled trial.