Revisiting the Osteoporosis Self-Assessment Tool (OST): evidence from a multi-cohort study in Asia supporting age-dependent thresholds.

Chandran, Manju; Wong, Chiew Meng Johnny; Helgason, Benedikt; Leong, Yong Eu; Chan, Angelique; Malhotra, Rahul; Gupta, Preeti; Jha, Dheeraj et al. · Osteoporos Int · 2026

prospective_cohort · Level II

Where this comes from

Abstract

The Osteoporosis Self-Assessment Tool (OST) is a useful discriminator of densitometric osteoporosis, but its long-used fixed threshold performs poorly, especially in younger postmenopausal women. In this multi-ethnic cohort study, optimal screening thresholds varied strikingly by age, supporting a shift from a single cut-off to age-dependent thresholds. The Osteoporosis Self-Assessment Tool (OST) is used worldwide to identify individuals likely to have osteoporosis on dual-energy X-ray absorptiometry (DXA) and fixed thresholds are widely used. We evaluated OST performance in a multi-ethnic population and examined whether age-dependent thresholds better reflect its role as a screening tool. Women (n = 2787) and men (n = 1465), ≥ 50 and ≥ 60-years respectively, from three cohorts were studied. Osteoporosis was defined as a DXA T-score ≤  - 2.5 at the lumbar spine, total hip, or femoral neck. We evaluated the originally identified (≤ - 1), the historically propagated- (≤ - 4), Youden index-derived-, and screening-oriented-OST thresholds targeting 80% sensitivity. Stability was assessed using fivefold cross-validation supported by a real-world patient cohort. OST showed acceptable discrimination (AUC 0.79 in women; 0.72 in men). The - 1 threshold showed only moderate sensitivity in younger women (0.54, aged 50-59 years) and poor specificity in older women (0.21, aged 70-79 years; 0.03 at ≥ 80 years). The - 4 threshold showed poor sensitivity overall, particularly in younger women (0.02, aged 50-59 years; 0.14, aged 60-69 years). Youden thresholds demonstrated clear age-dependence shifting from - 0.15, (50-59 years) to - 6.20 (≥ 80 years), with stability on cross-validation ((mean - 2.18, SD 0.45)). Screening thresholds targeting 80% sensitivity also showed a steep age-dependent gradient (+ 0.25 to - 5.39). Similar age-dependent patterns were observed in men. OST thresholds show substantial variation in performance across age groups. Age-dependent thresholds provide a more appropriate framework for osteoporosis screening.