Observer agreement study of cervical-vertebral ratios in horses.
Authors: Scrivani P V, Levine J M, Holmes N L, Furr M, Divers T J, Cohen N D
Journal: Equine veterinary journal
Summary
# Editorial Summary: Observer Agreement in Cervical-Vertebral Ratio Measurement Cervical-vertebral ratios are widely used radiographic indices to assess horses with suspected cervical stenotic myelopathy, yet their clinical utility depends critically on measurement consistency—a factor that hadn't been rigorously quantified until this 2011 observer agreement study. Scrivani and colleagues analysed measurements taken by a board-certified radiologist and imaging resident at two commonly evaluated sites (C3-4 and C6-7) in 75 horses, using Bland-Altman analysis to establish limits of agreement and repeatability thresholds. Intraobserver variability ranged from ±4–8% at C3-4 and ±6–7% at C6-7 for intravertebral ratios, whilst interobserver differences were marginally larger (±5–8% and ±6–8% respectively); intervertebral ratios showed greater scatter, particularly at C6-7 where limits reached ±13% between examiners. These measurement fluctuations—typically 5–10% even with experienced radiologists—carry real diagnostic implications: a borderline ratio could shift from clinically significant to reassuring depending on who performs the measurement and when, potentially leading to both unnecessary investigations and missed diagnoses. When integrating cervical-vertebral ratios into clinical decision-making, all equine professionals should recognise that no single measurement should be interpreted in isolation; contextual clinical findings, serial measurements, and radiologist expertise remain essential to avoid misdiagnosis driven by inherent measurement variability.
Read the full abstract on PubMed
Practical Takeaways
- •When interpreting cervical-vertebral ratios, account for inherent measurement variability of 5-10% to avoid false positive or negative diagnoses
- •Intervertebral ratios show greater variability between examiners than intra-vertebral ratios; consider this when comparing measurements over time or between clinics
- •Serial measurements for monitoring progression of cervical disease should ideally be performed by the same examiner to minimize interobserver variation
Key Findings
- •Cervical-vertebral ratio measurements varied by 5-10% within and between examiners at C3-4 and C6-7 sites
- •Intra-vertebral ratio measurements at C3-4 showed 95% repeatability with differences ≤4% typically and ≤8% always
- •Intervertebral ratio measurements demonstrated lower repeatability (92-96%) with differences up to 11% at C3-4 and 10% at C6-7
- •Measurement variability of 5-10% may lead to misdiagnosis and limits clinical usefulness of cervical-vertebral ratios in practice