Background

Sanders et al. [1, 2] describe a simplified system for determining digital skeletal age (DSA) and its use in predicting the likelihood a curve will reach surgical magnitude. We assessed the inter-and intra-rater reliability and prognostic implications of this classification system using data from a multicenter trial of outcomes in AIS (BrAIST).

Material and methods

36 subjects were randomly selected. We determined the predicted prognosis by cross-classifying the DSA and Cobb angle using Sanders’ estimates.

Results

Kappa coefficients ranged from 0.76 to 0.88. For example, one rater’s reading corresponded to a 0% risk of the curve reaching surgical magnitude, while the other rater’s reading for the same subject corresponded to a 92% risk.

The high level of agreement in DSA found by Sanders et. al was replicated in this study, and would be considered “substantial” to “almost perfect” using widely applied standards [3] Despite this agreement, different prognoses were predicted for 11% of these subjects.

Conclusions

Clinicians and researchers should consider seeking a second review of the DSA (especially if it appears to be in the DSA 2 to 3 range), and the Cobb angle, prior to using it to make prognostic predictions and treatment decisions.