We investigated the fracture risk assessment tool (FRAX) Canada calibration and discrimination according to income quintile in 51,327 Canadian women, with and without a competing mortality framework. Our data show that, under a competing mortality framework, FRAX provides robust fracture prediction and calibration regardless of socioeconomic status (SES).
Introduction: FRAX® predicts 10-year fracture risk. Social factors may independently affect fracture risk. We investigated FRAX calibration and discrimination according to SES.
Methods: Women aged ≥50 years with baseline femoral neck bone mineral density (BMD) were identified from the Manitoba Bone Density Program, Canada (n = 51,327), 1996-2011. Mean household income, extracted from 2006 census files, was categorized into quintiles. Ten-year fracture probabilities were calculated using FRAX Canada. Incident non-traumatic fractures were studied in relation to income quintile in adjusted Cox proportional hazards models. We compared observed versus predicted fractures with and without a competing mortality framework.
Results: During mean 6.2 ± 3.7 years of follow up, there were 6,392 deaths, 3,723 women with ≥1 major osteoporotic fracture (MOF), and 1,027 with hip fractures. Lower income was associated with higher risk for death, MOF, and hip fracture in adjusted models (all p < 0.005). More women in income quintile 1 (lowest) versus quintile 5 experienced death (19 vs. 8%), MOF (10 vs. 6%), or hip fracture (3.0 vs. 1.3 %) (all p ≤ 0.001). Adjustment for competing mortality mitigated the effect of SES on FRAX calibration, and good calibration was observed. FRAX provided good fracture discrimination for MOF and hip fracture within each income quintile (all p < 0.001). Area under the curve was slightly lower for income quintiles 1 versus 5 for FRAX with BMD to predict MOF (0.68, 95% CI 0.66-0.70 vs. 0.71, 95% CI 0.69-0.74) and hip fracture (0.79, 95% CI 0.76-0.81 vs. 0.87, 95% CI 0.84-0.89).
Conclusion: Increased fracture risk in individuals of lower income is offset by increased mortality. Under a competing mortality framework, FRAX provides robust fracture prediction and calibration regardless of SES.