Signed Total Hip-Femoral Neck T-Score Discordance and Prevalent Proximal Femur Fracture in Elderly Hip-DXA Patients

초록

Background and Objectives: Total hip (TH) and femoral neck (FN) T-scores from the same dual-energy X-ray absorptiometry (DXA) examination often diverge. We evaluated whether the signed TH-FN difference is associated with prevalent proximal femur fracture and whether it has individual-level discriminative value, explicitly separating statistical association from discrimination. Materials and Methods: In a single-center retrospective case-control study (complete case N = 3752; 448 fractures), multivariable logistic regression adjusting for the TH T-score and 12 covariates estimated the association of the signed difference, interaction with severity, and standalone and incremental discrimination (AUC, cross-validated ΔAUC). Additional analyses examined standardized and BMD-scale exposures, a low-energy-restricted cohort, a subtype case-case comparison, and adjustment for the ISCD lowest-site T-score including the lumbar spine. Results: The signed difference was associated with fracture (adjusted OR 1.65, 95% CI 1.39-1.97; interaction OR 1.06, 0.75-1.50), persisting with inpatient-restricted controls (OR 1.74), standardized and BMD-scale exposures, and low-energy restriction (OR 1.68). Standalone discrimination was weak (AUC 0.547) and out-of-sample incremental value negligible (ΔAUC +0.005, 95% CI -0.009 to +0.014). Adjusting for the ISCD lowest-site T-score abolished the full-cohort association (OR 1.04, 0.88-1.23)-the adjusted signal is femoral neck information already captured by the lowest-site rule-while a residual association persisted within the TH ≤ -2.5 subgroup (OR 1.34, 1.02-1.76). Fracture proportions are within-sample case-control fractions, not population incidence. Conclusions: The signed TH-FN discordance is a robust, routinely derivable correlate of prevalent proximal femur fracture whose standalone discrimination is nonetheless limited-a concrete demonstration that statistical significance need not confer predictive utility. It is best positioned as a supplementary, hypothesis-generating marker; any clinical role requires prospective validation against FRAX and the ISCD lowest-site rule.

키워드