Admission to hospital is directly related both to the underlying health of the patient and tohealth policy or coverage within a given country or region. Patients undergoing treatment inregions with relatively liberal admission policies will have more frequent and more“serious” exacerbations, while those in regions with conservative admission policies willhave less frequent and/or fewer “serious” episodes. This bias has serious implications forprevalence estimates in epidemiologic studies, effect estimates in studies examining thelink between exacerbations and disease trajectory, and site selection and treatment out-comes in clinical trials.

Adistinct advantage of this approach is thatrisk prediction can be performed remotelyand can be made arbitrarily complex toimprove its accuracy. In this “populationsurveillance” approach, individuals who areidentified as high-risk can then be contactedfor preventive disease management.

In contrast, in this study,we focused on exacerbation risk prediction for discharged COPDpatients, because their health condition is likely to be lessaccessible.

ACCEPT can combine predicted risk with effectestimates from randomised trials to enable personalisedtreatment.