An assessment of intra- and interobserver agreement of reporting orthopaedic findings on presale radiographs of Thoroughbred yearlings.

The research aims at studying the level of agreement between different observers in identifying orthopaedic findings from radiographs of Thoroughbred yearlings. In this study, four specialized veterinarians individually assessed radiographs with the results showcasing excellent negative agreement with variable outcomes for positive agreements.

Methodology

• The study adopted a retrospective method by analyzing archived radiographs.
• Four veterinary radiology specialists independently examined 167 sets of radiographs on two occasions per set. The radiographs explored orthopaedic findings in various parts of the horses, including the fore feet, fore and hind fetlocks, carpi, tarsi, and stifles.
• There were 27 specific findings that were analysed for agreement using three parameters – Kappa statistic (κ), percentage of positive agreement (Ppos), and percentage of negative agreement.

Findings

• The absence of radiographic findings was consistently agreed upon, indicating an excellent percentage of negative agreement for nearly all orthopaedic findings.
• Agreement on the presence of findings displayed variability. Particular findings such as extra carpal bones, osseous cyst-like lesions of the ulnar carpal bone, sagittal ridge defects of the third metacarpus, fractures of the fore and hind proximal sesamoid bones and others had a consistent agreement with a κ≥0.5 and Ppos≥50%.
• On the other hand, the presence of lucency within the proximal sesamoids showed a consistently lower agreement with an intra- and interagreement κ<0.4 and Ppos<40%.

Conclusion

• Even though the absence of orthopaedic findings was agreed upon, there was inconsistent agreement regarding the presence of specific orthopaedic findings.
• Larger and straightforward to categorize radiographic findings saw a better agreement than smaller or complex ones.
• The findings suggest a need for more accurate definitions and training to improve agreement among observers, especially for orthopaedic findings with poor or fair to good agreement.