Abstract / Summary
Conventional assessment of femoral neck fractures — from Garden classification to the evaluation of reduction quality — relies on two-dimensional (2D) radiography, which cannot simultaneously depict rotational malalignment and translational displacement. We used computed tomography (CT)-based three-dimensional (3D) reconstruction to quantify pre- and postoperative spatial displacement, to evaluate the efficacy of surgical correction, and to explore the prognostic value of 3D parameters for osteonecrosis of the femoral head (ONFH). In this retrospective single-centre study, 200 patients with unilateral femoral neck fractures treated by internal fixation were analysed. Three-dimensional parameters — d1 (fovea capitis displacement), d2 (femoral head displacement) and α (spatial angulation) — were measured before and after surgery using CT-based 3D reconstruction. Univariable and multivariable logistic regression and receiver operating characteristic analysis were performed to explore predictors and to evaluate model discrimination, with bootstrap internal validation (1,000 resamples). All three 3D parameters increased across Garden grades (all P < 0.001), although types I and II did not differ significantly from one another. In radiographically satisfactory reductions, postoperative residuals did not differ significantly from low-displacement preoperative baselines for several comparisons (Garden II vs. preoperative type I: P = 0.053, 0.064 and 0.093 for d1, d2 and α, respectively; Garden IV vs. preoperative type I for d2 and α: P = 0.071 and 0.078), suggesting that 2D-satisfactory reduction may conceal residual 3D displacement. ONFH developed in 27 patients (13.5%) during a median follow-up of 25 months. A combined model based on postoperative d2 and α yielded an apparent AUC of 0.674 (95% CI 0.556–0.793, P = 0.004) and an optimism-corrected AUC of 0.664. Discrimination was better in stable fractures (AUC 0.793) than in unstable fractures (AUC 0.631, P = 0.059), where it did not reach statistical significance. A severity gradient across the Garden types is corroborated in three-dimensional space, although Garden types I and II did not differ significantly from one another. Two-dimensionally satisfactory reduction may still leave substantial residual displacement. Postoperative d2 and α showed a prognostic signal that was most evident in stable fractures and warrants confirmation. Although internally stable, the model showed modest overall discrimination and requires external validation before any clinical application. These 3D parameters should therefore be regarded as investigational complements to, rather than replacements for, conventional 2D assessment. Not applicable.