Prediction of absolute risk of non-spinal fractures using clinical risk factors and heel quantitative ultrasound

Abstract

The relationship between osteoporosis risk factors, bone quantitative ultrasound (QUS) and non-spinal fracture risk was estimated in a cohort of 5,201 postmenopausal women from Spain who were prospectively evaluated during three years. Several clinical risk factors and low heel QUS values were independently associated with non-spinal fracture risk. Low-trauma, non-spinal fractures are a growing source of morbidity and mortality in the elderly. The aim of the present study was to examine the association of heel quantitative ultrasound (QUS) and a series of osteoporosis and fracture risk factors, with incident low energy non-spinal fractures in a population of elderly women, and to incorporate them into fracture prediction models. 5,201 women aged 65 or older were enrolled in a three-year cohort study. Participants completed an osteoporosis and fracture risk factors questionnaire. QUS was measured at the heel with a gel-coupled device. Cox-proportional hazard analyses were performed to evaluate the association with the first incident low-trauma non-spinal fracture. Three hundred and eleven women (6.0%) sustained a total of 363 low-trauma fractures, including 133 forearm/wrist, 54 hip, 50 humerus, 37 leg and 17 pelvic fractures. For every standard deviation decrease in the quantitative ultrasound index, the adjusted hazard ratios (95% CI) for any non-vertebral, hip, forearm/wrist, and humerus fractures were 1.31 (1.15-1.49), 1.40 (1.01-1.95), 1.50 (1.19-1.89) and 1.35 (0.97-1.87), respectively. Similar results were observed with other QUS variables. The best predictive models indicated that age, a history of falls, a previous low-trauma fracture, a family history of fracture, a calcium intake from dairy products of less than 250 mg/day, and lower values of QUS parameters were independently associated with the risk of non-spinal fractures. Both clinical risk factors and QUS are independent predictors of risk of fragility non-spinal fractures. A prediction algorithm using these variables was developed to estimate the absolute risk of non-spinal fractures in elderly women in Spain.