Reliability of Radiographic Imaging Characteristics for Osteochondritis Dissecans of the Capitellum

Background: A primary challenge in the treatment of capitellar osteochondritis dissecans (OCD) is accurate imaging assessment. Radiographic classification consensus is not available in the current literature, and correlation of radiographs with lesion stability and resultant best treatment is lacking. Purpose: To determine the inter- and intrarater reliability of the presence or absence and common radiographic characteristics of capitellar OCD lesions. Study Design: Cohort study (Diagnosis); Level of evidence, 3. Methods: Anteroposterior, lateral, and oblique radiographs for 29 cases were reviewed by 7 orthopaedic surgeons. Images were assessed for elbow anthropometry and morphology, OCD presence, lesion characteristics, the presence of progeny bone and progeny features, and radial head abnormalities. Intra- and interrater reliability was assessed using Fleiss and Cohen kappa for nominal variables and intraclass correlation coefficients (ICCs) for continuous variables. Results: Surgeons demonstrated substantial to excellent inter- and intrarater reliability when assessing elbow characteristics: anthropometric (interrater ICC, 0.94-0.99; intrarater ICC, 0.82-0.96) and morphologic (Fleiss, 0.61-0.76; Cohen, 0.68). When the OCD lesion was assessed, fair to moderate interrater agreement was found for classifying the absence or presence of a lesion (Fleiss, 0.28-0.46) and the location of the OCD (Fleiss, 0.24-0.52), poor agreement for assessing the contour of the lesion (Fleiss, 0.00-0.09), and excellent agreement for measuring the size of the lesion (ICC, 0.82-0.94). Poor to fair interrater agreement was found for radial head abnormalities (Fleiss, 0.00-0.27). Progeny bone visualization and fragmentation demonstrated moderate interrater agreement (Fleiss, 0.43-0.47) where displacement of the bone demonstrated poor interrater agreement (Fleiss, 0.11-0.16). Intrarater agreement for OCD lesion characteristics, progeny bone visualization, and progeny bone features was moderate to excellent. Conclusion: Given only the fair to moderate agreement among raters for identifying OCD on radiographs, this imaging modality may not serve as a dependable screening tool in isolation. Additional imaging should be obtained if the clinical presentation suggests capitellar OCD and a definitive diagnosis is not possible with radiographs. However, clinicians can reliability measure the size of radiographically apparent OCD, suggesting that radiographs may serve as an appropriate imaging modality for follow-up care.