Reliability and agreement study of three-dimensional measurement for femoral head displacement indicators after femoral neck fractures

The outcomes of femoral neck fractures (FNFs) are influenced by the displacements of femoral head preoperatively and the reduction quality of surgery. CT-based three-dimensional (3D) reconstruction and measurement techniques have been widely employed in numerous studies to evaluate femoral head displacements both preoperatively and postoperatively. This study aims to measure specific indicators of displacement and residual displacement of femoral head after FNFs, and to assess the reliability and agreement of the measurement methods. Pre- and post-operative CT data of FNF patients treated with cannulated screws (2015-2020) were measured and analyzed. Bilateral proximal femurs were reconstructed using Mimics 22.0. Key measurements included: displacement of fovea of femoral head (d1), displacement of center of femoral head (d2), and 3D angle (α). The measurements were carried out by three experienced observers independently. Intraclass Correlation Coefficient (ICC) and Concordance Correlation Coefficient (CCC) were used to evaluate reliability and agreement. The standard error of measurement (SEM), Bland-Altman plots and 95% limits of agreement (LoA) were used to evaluate agreement. This study analyzed 200 CT scans from 100 patients (63.5 ± 4.9 years; 56 F/44 M). All measurement parameters (d1, d2, α) showed significant changes (P < 0.001) after fixation. Inter-observer reliability and agreement were great (ICCs: 0.894-0.982; CCCs: 0.737-0.946) for preoperative indicators and good-to-moderate (ICCs: 0.605-0.846; CCCs: 0.336-0.644) for postoperative indicators. Bland-Altman plots revealed acceptable agreement for all comparisons between two observers pre- and postoperatively with ≥ 92% data points falling within 95% LoAs. The SEMs for most measurement indicators were low both preoperatively and postoperatively (range: 0.857-3.565), except for angles measured in the transverse and sagittal planes. This study confirms that CT-based 3D reconstruction is valid for preoperative femoral neck fracture assessment (ICC > 0.894). While postoperative measurements show moderate agreement and greater measurement error, particularly in the sagittal plane, it should be noted that ICC/CCC values can be influenced by variability among samples. In future research, it is essential to focus on the standardized registration of models across all three anatomical planes during 3D reconstruction to further enhance measurement accuracy and stability.