Performance of U.S.-FRAX by race and ethnicity for short-term hip and major osteoporotic fracture prediction in U.S. women with rheumatoid arthritis.

Women with RA face elevated fracture risk, yet performance of the U.S.-FRAX across racial and ethnic groups in RA remains understudied. In this retrospective cohort study, we examined 14 533 women aged ≥65 yr with RA from the U.S. national Rheumatology Informatics System for Effectiveness registry, linked to Medicare claims (2016-2018), to assess the performance of race- and ethnicity-adapted U.S.-FRAX calculators without BMD for predicting short-term (2-yr) hip and major osteoporotic fracture (MOF) risk across racial and ethnic groups. Most participants were White (86.5%), with smaller proportions of Black or African American (7.8%), Hispanic or Latina (4.7%), and Asian (1.1%) women. Overall fracture incidence was high (15.3 per 1000 person-years (PY) for hip fracture; 55.8 for MOF), with variability by race and ethnicity. Hip fracture incidence (per 1000 PY) was lowest among Asian (6.5) and Black or African American (7.0) women and highest among Hispanic or Latina (16.3) and White (16.2) women. MOF incidence (per 1000 PY) was lowest in Black or African American (26.0) and highest in White (58.7) women. Discrimination of U.S.-FRAX for short-term fractures was modest, with overall cross-validated area under the curves (AUCs) of 0.71 (95% CI, 0.69-0.73) for hip fracture and 0.66 (95% CI, 0.65-0.68) for MOF. No significant differences in AUCs were found between Black or African American, Hispanic or Latina, and White women with RA. Short observation period for fractures precluded formal calibration testing, but our data suggests U.S.-FRAX without BMD may underestimate MOF risk, particularly among Black or African and Hispanic or Latina women with RA. Performance in Asian women was not reported due to low fracture rates. The modest discrimination and insufficient calibration data in all racial and ethnic RA groups studied underscore the need for larger studies in diverse RA populations. Ultimately, RA-specific fracture risk assessment tools may be needed.