Reasons for discrepancies in hip fracture risk estimates using FRAX and Garvan calculators

Abstract

ObjectivesBoth the FRAX and Garvan calculators are used to estimate absolute risk of fracture, but they sometimes produce different estimates. We sought to determine which patient characteristics contribute to these discrepancies. Study designTen-year hip fracture risk was estimated for 122 women, using both FRAX and Garvan with bone mineral density (BMD). Main outcome measuresDifferences in estimates of hip fracture were assessed, both in absolute terms and with respect to a treatment threshold of 3%. ResultsGarvan estimates were higher than FRAX estimates across the range of ages and BMDs studied. A history of falls or of multiple fractures increased risk calculated by Garvan 3–6-fold, but did not account for all differences between calculators. Discrepancies around a 3% treatment threshold occurred in 31/122 (25%). Women aged 70–74 years, and women with osteopenia were most likely to have discordant estimates. Most discordant estimates (29/31) had a Garvan estimate ≥3% and FRAX <3%. Falls, multiple fractures, ethnicity and a history of parental hip fracture contributed to some discordant estimates. ConclusionsHip fracture risk estimates are usually higher with Garvan than FRAX, and these differences could impact on treatment decisions in about a quarter of patients. Falls and multiple fractures have a strong influence on Garvan risk estimates, when present. Clinically important discrepancies tend to occur in patients who are at borderline fracture risk. In patients with hip fracture risks near the treatment threshold with one calculator, use of the other calculator should be considered to help guide treatment decisions.