Development and external validation of competing risk models for major osteoporotic fracture and hip fracture in an ethnically diverse UK population of 15.6 million adults

Fragility fractures are a major cause of mortality, disability and loss of independence in adults worldwide, yet existing fracture-risk prediction algorithms incompletely capture the growing populations of individuals living with learning disability, chronic disease, HIV and cancer survivorship. We developed sex-specific prediction models of 10-year major osteoporotic and hip-fracture risk in 9,465,613 UK adults in primary care and evaluated them in held-out QResearch (n = 3,155,655) and external CPRD (n = 3,011,531) cohorts. The corresponding incident numbers of major osteoporotic fractures during follow-up were 166,442, 55,413 and 55,145. Cause-specific Cox models estimated risk with competing mortality. Model A re-estimated established QFracture predictors; Model B additionally incorporated learning disability, Down syndrome, multiple sclerosis, visual impairment, HIV, selected cancers and cancer-directed endocrine therapies. Discrimination was high across validation cohorts (C-statistics, 0.766–0.900). Models A and B showed better overall calibration than QFracture and higher estimated net benefit over parts of the prespecified threshold range; residual miscalibration remained in some external-validation strata. Most population-level improvement arose from contemporary estimation within the competing risk framework. Added predictors produced clinically informative reprioritization of selected individuals. These models could enable more accurate, inclusive fracture-risk assessment.