Spine–hip bone mineral density discordance in postmenopausal Asian women: fracture burden and incremental risk stratification beyond hip BMD

Abstract

Summary

In 4059 postmenopausal women, spine–hip discordance was common (41.3%) and associated with heterogeneous fracture risk. Hip-only screening missed 680 vertebral fractures. Major hip-lower discordance showed the highest fracture burden. Each 1-SD increase in T-score offset independently increased hip fracture odds by 28%.

Purpose

To evaluate the prevalence and clinical significance of spine–hip T-score discordance in a large cohort of postmenopausal Asian women and to determine whether a standardized discordance metric provides incremental fracture risk information beyond hip bone mineral density (BMD) alone.

Methods

We conducted a retrospective cross-sectional study of 4059 postmenopausal women who underwent dual-energy X-ray absorptiometry (DXA) at a tertiary referral hospital. Participants were classified according to spine–hip diagnostic concordance as concordant, minor discordance (a one-category difference, spine lower or hip lower), or major discordance (a two-category difference, spine lower or hip lower). Three osteoporosis classification strategies (hip-only, spine-only, and combined dual-site DXA) were compared for detection of prevalent fractures. Multivariable logistic regression models were used to evaluate whether the T-score offset (lumbar spine T-score minus hip T-score) provided incremental predictive information beyond hip BMD.

Results

Diagnostic discordance was present in 41.3% of participants, including 39.1% with minor discordance and 2.2% with major discordance. The spine-lower pattern was approximately twice as common as the hip-lower pattern (25.7% vs. 13.4%). Women with major discordance (hip lower) demonstrated the highest prevalence of vertebral fracture (60.0%) and hip fracture (28.6%), exceeding rates observed in the concordant osteoporosis group (53.0% and 16.6%, respectively). Hip-only classification missed 680 prevalent vertebral fractures and showed a sensitivity of 45.1%. For hip fracture detection, hip-only screening demonstrated the highest overall discrimination (AUC 0.701), whereas combined dual-site classification provided the greatest sensitivity (73.0%). Consistent with recent evidence that hip BMD is the dominant densitometric predictor of hip fracture, hip BMD remained the strongest single predictor in our cohort. Importantly, after adjustment for age, body mass index, and hip BMD, each 1-standard deviation increase in T-score offset was independently associated with a 28% higher odds of hip fracture (adjusted OR 1.28; 95% CI 1.09–1.51; P = 0.003), indicating statistically significant incremental predictive information beyond absolute hip BMD alone.

Conclusion

Spine–hip T-score discordance was common in this clinical DXA cohort and was associated with clinically meaningful heterogeneity in fracture burden. Major discordance (hip lower) identified a high-risk phenotype that may be obscured by relatively preserved lumbar spine BMD. Although hip BMD remained the strongest single predictor of hip fracture, the standardized T-score offset may provide complementary information regarding skeletal heterogeneity, although prospective validation is required.