Correlation of Measurements of the Prearthritic Hip Between Plain Radiography and Computed Tomography

Background Previous studies have compared radiographic and computed tomography (CT) imaging for the evaluation of prearthritic hip pain. However, the intermodality, interrater, and intrarater consistencies of those parameters have not been investigated. Objective To determine whether radiographs with an anteroposterior pelvis view and 45°‐Dunn lateral view reliably correlate with CT in the context of lateral center edge (LCE), Tonnis, alpha, and beta angle measurements for femoroacetabular impingement or hip dysplasia diagnosis. Design Retrospective study. Setting Academic orthopedic institution. Patients Fifty consecutive participants with hip pain in the institutional hip registry with radiographs and CT imaging on file were evaluated between 2013 and 2014. Materials and Methods Radiologic data (50 CTs and 50 radiographs) were evaluated by 3 physicians. LCE, Tonnis, alpha, and beta angles were measured on radiographs and CTs in 2 rounds of readings. In round 1, the center of rotation on CT imaging was standardized by 1 rater. In round 2, individual raters chose CT images using a quadrant method, and reproducibility was assessed. Reliability statistics were operationalized with intraclass correlation coefficients (ICCs). Main Outcome Measurements Intermodality, intrarater, and interrater reliability of CT vs radiographic measurements. Results The intermodality reliability for all raters was excellent (ICC [95% CI]: 0.84 [0.76‐0.90] to 0.97 [0.96‐0.98]). Intrarater reliability for both modalities showed excellent reliability (ICC = 0.75‐0.96). Interrater reliability of CT measures of LCE, Tonnis, and alpha angles demonstrated excellent agreement (ICC ≥ 0.88). Beta angle measures demonstrated good agreement (ICC [95% CI] = 0.68 [0.49‐0.81]). Interrater reliability of radiographic measures showed excellent agreement (ICC = 0.82‐0.94). Conclusion Equivalent angle measurement readings on CT and radiographs were consistent among physicians. CT measurements correlated well with radiographic measurements. This suggests that if a standardized procedure is used to find the center of the femoral head, a positive correlation among LCE, alpha, beta, and Tonnis angles measured on CT can be obtained between multiple readers. Level of Evidence III