Fracture risk in Asian-Canadian women is significantly over-estimated by the Canadian Association of Radiologists-Osteoporosis Canada risk prediction tool: retrospective cohort study.

Abstract

Health risk prediction tools based on largely White populations may lead to treatment inequity when applied to non-White populations where outcome rates differ. We examined the calibration of the Canadian Association of Radiologists-Osteoporosis Canada (CAROC) fracture risk prediction tool in self-identified Asian-Canadian women. Retrospective cohort study of women over age 50 using the Manitoba BMD Registry. At first BMD, the intake questionnaire collected self-identification of ethnicity and fracture risk factors. 10-year fracture risk was estimated using CAROC and categorized into low, medium, or high fracture risk. Linked administrative databases identified incident osteoporotic fractures. Observed fracture rates were compared between White and Asian-Canadians and compared to the original CAROC risk stratification. There were 63,632 and 1703 women who self-identified as White-Canadian or Asian-Canadian, respectively, covering approximately 600,000 patient-years follow-up. There were 6588 incident fractures; a similar percentage of patients were assigned to each risk stratum at baseline by CAROC. A progressive rise in 10-year observed fracture rates occurred for each CAROC stratum in the White-Canadian population but much lower fracture rates than predicted in Asian-Canadian patients (p < 0.001). Fracture incidence rate ratios were 1.9-2.6 foldhigher in White- vs Asian-Canadian patients for all strata (p < 0.001). In the CAROC moderate and high-risk categories, observed fracture rates in Asian-Canadian patients were typically lower than predicted, indicating poor model calibration. In Asian-Canadian women, observed osteoporosis fracture rates are lower than predicted when using the CAROC tool. Over-estimation of fracture risk may influence shared decision-making discussions.