Fracture risk thresholds for majority incident fracture identification: A registry-based cohort analysis.

Abstract

The World Health Organization (WHO) is reviewing its positions on osteoporosis. The current analysis was performed to inform whether FRAX-based risk thresholds can identify the majority of individuals who will sustain fractures during follow-up and thereby help to address limitations inherent in the use of BMD T-scores. The study population consisted of 38,525 women aged 50 years and older (mean age 66.0 years) with baseline fracture risk prediction (FRAX major osteoporotic fracture probability estimated with BMD). Incident fractures identified through linked administrative healthcare data were categorized as any fracture, major osteoporotic fracture, and hip/vertebral fracture. We considered three time horizons: incident fracture within 10 years, within 5 years and within 2 years. During 10 years of follow-up there were 5353 women (13.7%) sustaining any incident fracture; hip/vertebral fractures within 2 years were registered in 1475 (3.8%). It was possible to define age-dependent and overall (age-independent) risk thresholds that identified the majority of incident fractures regardless of the fracture outcome assessed. As a rule, a lower risk threshold was required in younger versus older individuals. In contrast, a BMD T-score of -2.5 or lower failed to identify the majority of incident fractures under most scenarios. Diagnostic odds ratios from osteoporotic T-scores were consistently lower than from corresponding FRAX-based age-independent risk thresholds. FRAX-based risk thresholds can identify the majority of older women who will sustain incident fractures during the next 2-10 years.