The validity of automatic methods for estimating skeletal age in young athletes: a comparison of the BAUSport ultrasound system and BoneXpert with the radiographic method of Fels

Abstract

This study examined the validity of two automated methods (BAUSport, BoneXpert software using Fels, Greulich-Pyle, Tanner-Whithouse III protocols) for estimating skeletal age (SA) in young athletes in comparison to a reference standard (Fels). 85 male and female athletes, nine to seventeen years of age, from multiple sports were assessed for SA as part of an annual medical and health screening programme. Intra-class correlations demonstrated high degrees of association between the automatic methods for estimating SA (BAUSport r = .98; BoneXpert r = .96–.99) and the discrepancy between SA and chronological age (SA-CA) (BAUSport r = .93; BoneXpert r = .88–.97), with the reference standard. Concordance analyses for the categorisation of participants as early, on-time and late maturing also demonstrated substantial levels of agreement for both methods (BAUSport Kappa = .71; BoneXpert Fels Kappa = .63) with the reference standard. Bland-Altman plots comparing the automatic methods with the reference standard identified statistically significant fixed biases, ranging in magnitude from small to large. Collectively, these results suggest that BoneXpert and BAUSport can provide comparable estimates of SA and SA-CA in young athletes relative to the Fels method. Biases in the estimation of SA should, however, be considered and the automatic methods should be implemented as part of a comprehensive growth and maturity screening protocol. The non-invasive nature of the BAUSport method affords particular advantages (no radiation exposure, portability) in contexts where the regular estimation of SA is recommended.