Reliability of EOS compared to conventional radiographs for evaluation of lower extremity deformity in adult patients.

Abstract

The purpose of this study was to compare reliability of lower extremity imaging measurements using EOS and conventional X-ray (CR) of adult patients with mechanical axis malalignment. Ten patients (20 lower limbs) of mean age of 31.6years (range 21-39) with post-traumatic deformities who presented for evaluation of osteotomies and/or ligament and cartilage reconstructions were prospectively enrolled. Two independent observers performed full-length anterior-posterior (AP) measurements 2weeks apart on both CXR and two-dimensional (2D) EOS images. Measurements included weight-bearing axis (WBA), varus/valgus angle (V/V), femoral length (FL), tibial length (TL), femoral mechanical axis (FMA), tibial mechanical axis (TMA), and total limb length (TLL). Reliability was determined with random effects modeling of intraclass correlation coefficients (ICC) set to consistency. Three statistical operations were performed to compare interrater validity in CXR and EOS: students' two-sample t test, paired two-sample t test, and Pearson's correlative r-statistical agreement. There was a statistically significant difference for V/V, FL, and TLL (all p< 0.01) between CXR and EOS. A relatively large proportion of the population consistently had larger V/V measures for EOS compared to CXR. In contrast, the FL and TLL measures were consistently larger for CXR compared to EOS. The differences between CXR and EOS measurements were statistically significant, though the small differences in values were not clinically meaningful. Agreement of all measures remained high (r= 0.84-0.99). Using 2D EOS for lower extremity measurements is reproducible, reliable, and comparable to the gold standard, standing long leg radiographs.