An evaluation of reliability in measurement of native femoral torsion on CT imaging.

Abstract

This study assesses intra- and interobserver reliability in measuring femoral torsion on repeated computed tomography (CT) scanograms and explores variability across training levels and patient characteristics. Patients from 2001 to 2017 with femur fractures who underwent multiple CT scanograms were included. Measurements were performed by an orthopaedic traumatologist, two orthopaedic residents (PGY-5 and PGY-3), and a musculoskeletal radiologist. Intraobserver reliability was assessed using intraclass correlation coefficients (ICCs), and interobserver reliability using repeated measures ANOVA. Correlations between measurement variability and body mass index (BMI), age, height, gender, and laterality were evaluated using Pearson's coefficient. Twenty-nine patients with unilateral uninjured femurs were analyzed. Interobserver analysis revealed significant differences in femoral torsion (p = 0.006) and length (p = 0.019). Intraobserver reliability showed strong agreement in femoral torsion (ICC 0.620-0.950) and length (ICC 0.992-0.999), but moderate to poor agreement in femoral neck axis (ICC 0.394-0.627) and posterior condylar axis (ICC 0.561-0.665). Femoral length measurements were the most consistent across all reviewers. Higher BMI correlated with increased variability in femoral torsion measurements (r = 0.378; p = 0.048). No significant correlation was found for age (p = 0.110), height (p = 0.363), gender (p = 0.610), or laterality (p = 0.830). These findings highlight the reliability of CT scanograms to assess femoral length and torsion when done by the same physician. Although moderate to poor agreement in femoral neck axis and posterior femoral condyle axis, overall femoral torsion was found to be highly reliable. Reliable measurement of native femoral torsion in patients with a higher BMI may be difficult. To ensure reliable measures of femoral torsion, the same clinician should evaluate all pertinent studies for an individual patient. Level III Retrospective cohort study.