RESUMO
Background and purpose - Skeletal maturity is a crucial parameter when calculating remaining growth in children. We compared 3 different methods, 2 manual and 1 automated, in the radiological assessment of bone age with respect to precision and systematic difference. Material and methods - 66 simultaneous examinations of the left hand and left elbow from children treated for leg-length discrepancies were randomly selected for skeletal age assessment. The radiographs were anonymized and assessed twice with at least 3 weeks' interval according to the Greulich and Pyle (GP) and Sauvegrain (SG) methods by 5 radiologists with different levels of experience. The hand radiographs were also assessed for GP bone age by use of the automated BoneXpert (BX) method for comparison. Results - The inter-observer intraclass correlation coefficient (ICC) was 0.96 for the GP and 0.98 for the SG method. The inter- and intra-observer standard error of the measurement (SEm) was 0.41 and 0.32 years for the GP method and 0.27 and 0.21 years for the SG method with a significant difference (p < 0.001) between the methods and between the experienced and the less experienced radiologists for both methods (p = 0.003 and p < 0.001). In 25% of the assessments the discrepancy between the GP and the SG method was > 1 year. There was no systematic difference comparing either manual method with the automatic BX method. Interpretation - With respect to the precision of skeletal age determination, we recommend using the SG method or preferably the automated BX method based on GP assessments in the calculation of remaining growth.