Summary: This study examined the effects of the use of clinical risk factors (CRFs) alone, BMD alone or the combination using the FRAX tool for the detection of women at risk of hip fracture. BMD tests alone selected women at higher risk and a greater number of hip fracture cases were identified compared to the use of CRFs alone. The combined use of CRFs and BMD identified fewer women above a threshold risk than the use of BMD alone, but with a higher hip fracture risk and thus had the more favourable positive predictive value (PPV) and number needed to treat (NNT).
Introduction: Algorithms have recently become available for the calculation of hip fracture probability from CRFs with and without information on femoral neck BMD. The aim of this study was to examine the effects of the use of CRFs alone, BMD alone or their combination using the FRAX tool for the detection of women at risk of hip fracture.
Methods: Data from 10 prospective population based cohorts, in which BMD and CRFs were documented, were used to compute the 10-year probabilities of hip fracture calibrated to the fracture and death hazards of the UK. The effects of the use of BMD tests were examined in simulations where BMD tests were used alone, CRFs alone or their combined use. The base case examined the effects in women at the age of 65 years. The principal outcome measures were the number of women identified above an intervention threshold, the number of hip fracture cases that would be identified, the positive predicted value and the NNT to prevent a hip fracture during a hypothetical treatment with an effectiveness of 35% targeted to those above the threshold fracture risk. We also examined BMD values in women selected for treatment. Sensitivity analysis examined the effect of age and limited use of BMD resources.
Results: BMD tests alone selected women at higher risk of hip fracture than the use of CRFs alone (6.1% versus 5.3%). BMD tests alone also identified a greater number of hip fracture cases (219/1,000) compared to the use of CRFs alone (140/1,000). The combined use of CRFs and BMD identified fewer women above a threshold risk than the use of BMD alone (168/1,000 versus 219/1,000, respectively), but with a higher hip fracture risk (PPV, 8.6% versus 6.1%), and consequently a lower number needed to treat (NNT) (33 versus 47). In sensitivity analyses, the PPV and NNT were always better for the combination than either BMD or CRFs alone across all ages studied (50-70 years).
Conclusions: The use of FRAX in combination with BMD increases the performance characteristics of fracture risk assessment.