Background
Many individuals who sustain a fracture have normal/moderate deficits in bone mineral density (BMD). Recent technologies for assessing fracture risk (trabecular bone score [TBS], advanced hip analysis [AHA] and quantitative ultrasound [QUS])may improve prediction but have not yet been assessed in concert. We aimed to develop a best-fit model of these tools and to assess if this model improves fracture risk prediction over femoral neck BMD alone in those with BMD above diagnostic threshold for osteoporosis.
Methods
Participants (613 men, 395 women, aged 40-92yr) from the Geelong Osteoporosis Study (GOS) were followed from date of assessment to first fracture, death or end of study. BMD was measured at hip and lumbar spine (GE Lunar Prodigy Pro, Madison, WI), TBS was calculated from lumbar spine scans (TBS iNsight V2.2) and AHA was measured from hip scans (GE Lunar, Madison WI). QUS was performed on the left heel using a Lunar Achilles ultrasonometer. Incident fractures were self-reported and verified via radiological report. Cox-proportional hazards and logistic regressions were used to evaluate predictive accuracy, with Harrell’s C and area under the Receiver Operating Curve (AUROC) values respectively.
Results
There were 74 fractures in 63 participants (fracture rate: 1.1/1000 person years; 95%CI 0.8-1.4) over a median follow up time of 8.0 (IQR 6.8-8.7) years for men and 4.7 (IQR 3.9-5.3) years for women. Multivariable Cox-proportional hazards modelling for any fracture, including bone parameters of femoral neck BMD, AHA buckling ratio, TBS and QUS stiffness index adjusted for confounders showed a small improvement over adjusted femoral neck BMD alone (Harrell’s C: 0.696 vs 0.709; AUROC: 0.682 [95%CI:0.614-0.751] vs 0.696 [95%CI:0.629-0.763]).
Conclusion
Novel techniques for measuring bone showed modest improvements over femoral neck BMD as measured by DXA. Further studies are needed to assess if potential sex related differences exist.