Editorial

Predictive risk modelling using routine data: underexploited potential to benefit patients Huge amounts of data on individuals are now routinely collected electronically in the public sector in the UK. The NHS, as a single payer, can mandate data collection with a degree of consistency for almost the whole popu-lation of the UK, and creates one of the best sources of routinely collected data on health care in the world. These data are best for patient encounters in inpatient, outpatient and A&E settings, as well as the comprehen-sive coverage of people registered in general practice. As they have unique identifiers, with the right security mechanism in place to protect confidentiality,1 it is poss-ible to link records over time and across services. These can be augmented at a local level by data on social car