Background: We aimed to evaluate the diagnostic utility of pelvic radiographs versus magnetic resonance imaging (MRI) of the sacroiliac joints in children with suspected sacroiliitis.
Methods: This was a retrospective cross-sectional study of children with suspected or confirmed spondyloarthritis who underwent pelvic radiograph and MRI within 6 months of one another. Images were scored independently by five raters. Interrater reliability was calculated using Fleiss's kappa coefficient (κ). Test properties of radiographs for depiction of sacroiliitis were calculated using MRI global sacroiliitis impression as the reference standard.
Results: The interrater agreement for global impression was κ = 0.34 (95% CI 0.19-0.52) for radiographs and κ = 0.72 (95% CI 0.52-0.86) for MRI. Across raters, the sensitivity of radiographs ranged from 25 to 77.8% and specificity ranged from 60.8 to 92.2%. Positive and negative predictive values ranged from 25.9 to 52% and from 82.7 to 93.9%, respectively. The misclassification rate ranged from 6 to 17% for negative radiographs/positive MRI scans and from 48 to 74% for positive radiographs/negative MRI scans. When the reference standard was changed to structural lesions consistent with sacroiliitis on MRI, the misclassification rate was higher for negative radiographs/positive MRI scans (9-23%) and marginally improved for positive radiographs/negative MRI scans (33-52%).
Conclusion: Interrater reliability of MRI was superior to radiographs for global sacroiliitis impression. Misclassification for both negative and positive radiographs was high across raters. Radiographs have limited utility in screening for sacroiliitis in children and result in a significant proportion of both false negative and positive findings versus MRI findings.
Keywords: Juvenile spondyloarthritis; Magnetic resonance imaging; Misclassification; Radiograph; Sacroiliitis.