The sphericity deviation score: Is it a useful and reproducible parameter to assess the femoral head sphericity in Perthes disease?

The sphericity deviation score (SDS) is a continuous parameter to describe femoral head shape in patients with Perthes disease. The aims of this study were to evaluate the reproducibility of SDS and its association with the 3-group Stulberg classification. Anteroposterior and frog-leg lateral radiographs of 41 healed Perthes hips were assessed. Six observers with varying levels of experience, independently performed SDS measurements on 2 occasions, 2 weeks apart. Associations between SDS and the modified 3-group Stulberg classification were tested using ANOVA or Kruskal–Wallis tests and Spearman’s rank correlation coefficient (rho). Intra- and interobserver agreement was assessed with Bland–Altman 95% limits of agreement (LoA), and reliability was analyzed by intraclass correlation coefficients (ICC). The median SDS values in hips with round femoral head were significantly higher when measured by inexperienced observers than those measured by experienced observers (30 vs 17; p<0.001). The SDS values did not significantly correlate with the Stulberg classification among inexperienced observers. Among experienced observers, a strong correlation was found (Spearman’s rho 0.670; p<0.001); however, only 2 of 3 observers were able to distinguish between round and oval hips. Interobserver agreement for SDS was unacceptable (LoA >40) for inexperienced observers and marginal (LoA 20–40) for experienced observers. Interobserver reliability was poor for inexperienced observers (ICC 0.15; p=0.02) and moderate for experienced observers (ICC 0.69; p<0.001). SDS is a complex parameter that did not consistently differentiate between femoral head shapes when used by inexperienced observers. Even among experienced pediatric orthopedic researchers, SDS showed marginal interobserver agreement. The method should be used only by experienced observers in scientific studies. • The Sphericity Deviation Score (SDS) correlates significantly with the Stulberg classification only when measured by experienced pediatric orthopedic researchers. • SDS is a complex and time-consuming parameter that often failed to distinguish clearly between adjacent Stulberg categories (e.g., round vs. ovoid hips). • Among inexperienced observers, the limits of agreement (LoA) of SDS measurements was unacceptable and the reliability (ICC) was poor. • SDS should be used only by experienced observers in scientific studies. Even among experienced researchers, interobserver LoA was marginal, indicating that further methodological refinements are desirable. retrospective cohort study.