Computed tomography-derived muscle metrics and functional outcomes after proximal femoral nailing in older women with intertrochanteric fractures
TL;DR
CT-derived gluteal muscle quantity showed the most consistent nominal association with function after PFN, but no association survived FDR correction and incremental value beyond routine clinical predictors was limited.
Abstract
Aim: To examine whether computed tomography (CT)-derived cross-sectional area (CSA) and radiodensity (Hounsfield units, HU) of the psoas, gluteus maximus, and gluteus medius-minimus are associated with postoperative function in women aged ≥65 years treated with a proximal femoral nail (PFN) for intertrochanteric fracture.Materials and Methods: Sixty-three women aged ≥65 years undergoing PFN fixation for intertrochanteric fracture were analyzed retrospectively. Muscle measurements were obtained on standardized axial CT slices by manual tracing. Function was assessed with the Functional Ambulation Classification (FAC), Parker Mobility Score, Barthel Index, and modified Harris Hip Score (mHHS). Analyses included Spearman correlations with Benjamini-Hochberg false discovery rate (FDR) correction (24 comparisons in a predefined plan), HC3 robust linear regression adjusted for age and ASA, and area under the receiver-operating-characteristic curve (AUC) at clinically meaningful thresholds.Results: The median age was 74 years (IQR 67.5-82.0); median follow-up nine months. Gluteus maximus CSA correlated with Barthel (rho=0.34, p=0.007) and FAC (rho=0.30, p=0.015); an exploratory gluteal CSA composite showed a slightly stronger correlation with Barthel (rho=0.35, p=0.005). Gluteus maximus CSA discriminated community ambulators (FAC≥4) with AUC=0.71 (95% CI 0.55-0.85), and gluteus medius-minimus CSA discriminated Barthel≥90 with AUC=0.73 (0.60-0.85). No association survived FDR correction. Added to age, ASA, and pre-fracture ambulation, the gluteal CSA composite contributed only 0.5-2.2% incremental R².Conclusions: CT-derived gluteal muscle quantity showed the most consistent nominal association with function after PFN, but no association survived FDR correction and incremental value beyond routine clinical predictors was limited. The proposed cut-offs are hypothesis-generating and require prospective validation.