Visual assessment of the cervical vertebral maturation stages a study of diagnostic accuracy and repeatability

Abstract Objective: To evaluate the diagnostic accuracy and repeatability of the visual assessment of the cervical vertebral maturation (CVM) stages. Materials and Methods: Ten operators underwent training sessions in visual assessment of CVM staging. Subsequently, they were asked to stage 72 cases equally divided into the six stages. Such assessment was repeated twice in two sessions (T1 and T2) 4 weeks apart. A reference standard for each case was created according to a cephalometric analysis of both the concavities and shapes of the cervical vertebrae. Results: The overall agreement with the reference standard was about 68\% for both sessions and 76.9\% for intrarater repeatability. The overall kappa coefficients with the reference standard were up to 0.86 for both sessions, and 0.88 for intrarater repeatability. Overall, disagreements one stage and twp stage apart were 23.5\% (T1) and 5.1\% (T2), respectively. Sensitivity ranged from 53.3\% for CS5 (T1) to 99.9\% for CS1 (T2), positive predictive values ranged from 52.4\% for CS5 (T2) to 94.3\% for CS6 (T1), and accuracy ranged from 83.6\% for CS4 (T2) to 94.9\% for CS1 (T1). Conclusions: Visual assessment of the CVM stages is accurate and repeatable to a satisfactory level. About one in three cases remain misclassified; disagreement is generally limited to one stage and is mostly seen in stages 4 and 5.