Interobserver and Intraobserver Reliability of the Radiographic Analysis of Femoroacetabular Impingement and Dysplasia Using Computer-assisted Measurements

Background: A comprehensive evaluation of hip radiographs in the young adult with hip pain has become increasingly complex and time consuming. The interobserver reliability of manually performed measurements of femoroacetabular impingement, including the alpha angle, has been questioned. Methods to improve the reliability of a radiographic evaluation may increase the clinical utility of these parameters.

Purpose: To determine the interobserver and intraobserver reliability of a computer-assisted radiographic analysis of the young adult hip in a clinically relevant setting.

Study Design: Cohort study (diagnosis); Level of evidence, 3.

Methods: A reliability study of a comprehensive computer-assisted radiographic evaluation was performed, which included 25 radiographic parameters of proximal femoral morphology, acetabular morphology, hip osteoarthritis, and pelvic tilt/rotation. Anteroposterior pelvis and 45° Dunn lateral radiographs of 70 consecutive patients undergoing hip preservation surgery were included. Each radiograph was analyzed by 4 experienced hip surgeons. The reliability of continuous measurements was analyzed using intraclass correlation coefficients (ICCs), while categorical parameters were analyzed using κ values and percentages of agreement.

Results: The interobserver reliability of the parameters of proximal femoral morphology, acetabular morphology, and osteoarthritis was generally substantial to excellent. Parameters with lesser interobserver reliability included the alpha angle (ICC, 0.43), Tönnis osteoarthritis classification (κ = 0.22), and classification of pelvic tilt (using the coccyx or sacrococcygeal joint) (κ = 0.43 and 0.61, respectively).

Conclusion: A computer-assisted analysis of young adult hip radiographs generally demonstrates substantial to excellent levels of interobserver reliability for most parameters. However, alpha angle measurements demonstrated only moderate interobserver reliability, despite excellent intraobserver reliability. Measurements of the joint space width appear to be significantly more reliable than the use of the Tönnis osteoarthritis classification in this population. The classification of pelvic tilt utilizing the coccyx or sacrococcygeal joint is only moderately reliable.