An assessment of intra- and interobserver agreement of reporting orthopaedic findings on presale radiographs of Thoroughbred yearlings.
Authors: Jackson M A, Vizard A L, Anderson G A, Mattoon J S, Lavelle R B, Smithenson B T, Lester N V, Clarke A F, Whitton R C
Journal: Equine veterinary journal
Summary
# Editorial Summary Radiographic examination forms a cornerstone of presale veterinary assessments in Thoroughbred yearlings, yet standardisation of how different veterinarians interpret these images remains poorly understood. This retrospective study had four board-certified veterinary radiologists independently review 167 sets of presale radiographs on two separate occasions, assessing 27 orthopaedic findings across the distal limbs and joints to determine both intra-observer consistency (same reader on different occasions) and inter-observer agreement (different readers). Reassuringly, observers showed excellent agreement when identifying the *absence* of abnormalities across most findings, but agreement was considerably more variable when findings were actually present—with larger, clearly defined lesions (such as extra carpal bones, sagittal ridge defects, and certain osteochondrosis lesions) demonstrating robust agreement (kappa ≥0.5), whilst more subtle changes like lucencies within the proximal sesamoids showed poor consistency (kappa <0.4). For equine professionals purchasing or assessing yearlings, these results highlight that a negative radiographic report carries substantial weight, but positive findings—particularly smaller or ambiguous lesions—may warrant second opinions or additional specialist interpretation, as individual observer variation could materially influence purchasing or management decisions based on presale radiographs.
Read the full abstract on PubMed
Practical Takeaways
- •When purchasing yearlings, radiographic reports should be interpreted cautiously—observers are much more reliable at confirming absence of lesions than presence; request a second opinion for critical findings, especially subtle lesions like sesamoid lucencies
- •Larger, clearly defined lesions (carpal bones, sesamoid fractures, ridge defects) are reported consistently between radiologists, making these findings more trustworthy for purchase decisions than subtle findings
- •Be aware that lucency within proximal sesamoids has poor interobserver agreement; if this finding influences your purchasing decision, seek a second radiologist's opinion to improve confidence
Key Findings
- •Excellent negative agreement (absence of findings) observed across all orthopaedic findings except regular vascular channels in proximal sesamoids
- •Eight major findings (extra carpal bones, ulnar carpal cyst-like lesions, third metacarpal sagittal ridge defects, proximal sesamoid fractures, hind sesamoid vascular channels, tarsal osteochondrosis, and medial femoral condyle cyst-like lesions) achieved consistent intra- and interobserver agreement with κ≥0.5 and positive agreement ≥50%
- •Lucency within proximal sesamoids showed consistently poor agreement with κ<0.4 and positive agreement <40% across all observers
- •Agreement on presence of findings was highly variable, while agreement on absence of findings was uniformly excellent, indicating observer bias toward negative reporting