Objectives: Previous fracture prediction models have been based on the assumption of a stable risk of subsequent fractures over time. The aim of the present work was to develop a nomogram for prediction of 5-year and 10-year individualised absolute fracture risks for postmenopausal women taking into account the time relation between fractures.
Methods: A population-based prospective study was performed in 23 general practice centres located in the southern part of The Netherlands. At baseline (1992-1994), 4203 postmenopausal women between 50 and 80 years participated and 2372 of them also participated 10 years later. Baseline measurements included lumbar spine bone mineral density (BMD) and clinical risk factor evaluation. The incidence of fractures was ascertained. Bayesian model averaging and Cox's proportional hazards model were used.
Results: After enrolment, 382 (16.1%) women had a clinical fracture. Fracture risk was associated with advancing age (HR 1.09 per SD (5 years); 95% CI 1.01 to 1.17), lumbar spine BMD (HR 1.23 per -1 SD; 95% CI 1.10 to 1.37) and a prior fracture, with HR 3.27 (95% CI 2.50 to 4.30) for a recent prior fracture (≤5 years previously) and HR 1.97 (95% CI 1.43 to 2.71) for a non-recent prior fracture after menopause (>5 years previously). Women with a recent prior fracture had 66% higher risk of an incident fracture than those with a non-recent prior fracture (HR 1.66; 95% CI 1.15 to 2.40).
Conclusions: The nomogram developed can help doctors to inform patients more effectively and thus better manage patient care by providing an individualised fracture risk taking into account the time relationship for fractures.