Association Between 45° Flexion Anteroposterior Elbow Radiographs and Diagnostic Accuracy of Capitellum Osteochondritis Dissecans

Background: An anteroposterior (AP) radiograph of the elbow in 45° of flexion has been suggested to increase the diagnostic accuracy of capitellum osteochondritis dissecans (OCD).

Purpose: To assess the diagnostic performance, inter- and intraobserver reliability, and confidence level for identifying capitellum OCD using plain radiographs (AP, lateral, and 45° flexion AP).

Study design: Cohort study (Diagnosis); Level of evidence, 3.

Methods: This was a retrospective study of pediatric and adolescent patients with capitellum OCD and a control group. Six independent clinicians who were blinded to the official radiologists' reports reviewed images on 2 separate occasions, 1 week apart. A 5-point Likert scale was used to assess the clinicians' level of confidence. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated for individual and combinations of radiographic views. Inter- and intraobserver reliability was determined using Cohen kappa (κ) coefficients.

Results: A total of 28 elbows (mean age, 12.5 ± 2 years) were included. There were no differences in age (P = .18), sex (P = .62), or laterality (P > .999) between groups. There were marked variations in the diagnostic accuracy between views: sensitivity (AP, 85.1; lateral, 73.2; 45° flexion AP, 91.7), specificity (AP, 89.3; lateral, 91.7; 45° flexion AP, 91.1), PPV (AP, 88.8; lateral, 89.8; 45° flexion AP, 91.1), NPV (AP, 85.7; lateral, 77.4; 45° flexion AP, 91.6), and accuracy (AP, 87.2; lateral, 82.4; 45° flexion AP, 91.4). Standard radiographs (AP and lateral views) failed to diagnose capitellum OCD in 4.8% of cases. The sensitivity of the 3 combined views was 100%. Confidence intervals in the clinicians' diagnostic assessments were similar for each view (AP, 4.0; lateral, 4.0; and 45° flexion AP, 4.1). Interobserver reliability was substantial for AP and lateral views (κ = 0.65 and κ = 0.60, respectively) but highest for the 45° flexion AP radiographs (κ = 0.72). Intraobserver reliability for the 45° flexion AP view was moderate to almost perfect (κ = 0.45 to 0.93).

Conclusion: The 45° flexion AP view can detect capitellum OCD with excellent accuracy, a high level of confidence, and substantial interobserver agreement. When added to standard AP and lateral radiographs, the 45° flexion AP view aids in the identification of capitellum OCD.