Spine–hip bone mineral density discordance and fracture risk misclassification: clinical value of multi-site DXA in postmenopausal women

Spine–hip bone mineral density (BMD) discordance frequently complicates fracture-risk assessment in clinical practice. Although its prevalence is well recognized, the extent to which discordance contributes to diagnostic misclassification across dual-energy X-ray absorptiometry (DXA) strategies remains insufficiently quantified. This retrospective cross-sectional study included 4,059 postmenopausal women aged ≥ 50 years who underwent DXA of the lumbar spine, total hip, and femoral neck between 2022 and 2024. Diagnostic performance (sensitivity, specificity, and area under the receiver operating characteristic curve [AUC]) was compared across single-site and combined-site strategies. Multivariable logistic regression evaluated independent associations with prevalent fractures, while continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI) quantified the incremental value of multi-site assessment. Lumbar spine–total hip discordance occurred in 41.3% of the cohort, including 2.2% with major discordance. Major discordance was associated with a higher fracture burden, particularly when total hip BMD was lower than lumbar spine BMD, suggesting concealed proximal femoral fragility despite apparently reassuring spine values. Single-site DXA strategies missed up to 55.0% of prevalent fractures. Although selected reduced-site models showed competitive discrimination for specific outcomes, no single skeletal site consistently performed best across all fracture endpoints. Three-site assessment provided the most robust overall performance, achieving sensitivities of 77.8% for vertebral fractures and 85.7% for hip fractures, with superior discrimination (AUC up to 0.789). Higher lumbar spine T-scores were paradoxically associated with increased hip fracture risk, likely reflecting degenerative artifact and masking of true skeletal fragility. Addition of femoral neck measurements yielded substantial reclassification gains for hip fracture risk (NRI up to 0.591; IDI up to 0.049). Reliance on single-site DXA results leads to substantial fracture-risk misclassification. Spine–hip discordance appears to reflect regionally heterogeneous skeletal fragility rather than mere measurement inconsistency. Routine multi-site assessment—particularly with inclusion of the femoral neck—provides the most robust strategy for case detection and fracture-risk stratification, thereby offering a more accurate basis for clinical decision-making in postmenopausal women. An apparently reassuring lumbar spine T-score in an older woman should not exclude osteoporosis elsewhere, particularly at the total hip.