OBJECTIVE: To compare the performance of a new sonographic (US) method of bone age estimation with other methods currently in use. DESIGN AND PATIENTS: One hundred and fifteen subjects underwent left hand/wrist radiography and US examination of the hip for bone age assessment. For each patient, measurements of skeletal age were available based on Greulich-Pyle and Tanner and Whitehouse, the latter being presented in three subtypes (RUS, carpals, and B20) in addition to the US values. To assess agreement between methods, each method was compared with every other method. Differences between calculated skeletal age and chronological age were assessed, and the sensitivity, specificity, and predictive values of each method computed. RESULTS: Coupled B20/RUS values showed the best agreement, with 95% of observations within 2.45 years of each other, followed by carpals/B20, B20/GP, and GP/RUS. The US method agreed the least (difference of 4.19-5.13 years) with the other methods. The US method provided 85.8% (US vs RUS) to 91.3% (US vs GP) concordant results in recognizing differences between skeletal and chronological age, and showed a 72.5% sensitivity and a 56.8% specificity. CONCLUSION: Although the US method promises to permit a safe and cost-effective assessment of skeletal age, its low accuracy makes it currently unsuitable for clinical use.
OBJECTIVE: To compare the performance of a new sonographic (US) method of bone age estimation with other methods currently in use. DESIGN AND PATIENTS: One hundred and fifteen subjects underwent left hand/wrist radiography and US examination of the hip for bone age assessment. For each patient, measurements of skeletal age were available based on Greulich-Pyle and Tanner and Whitehouse, the latter being presented in three subtypes (RUS, carpals, and B20) in addition to the US values. To assess agreement between methods, each method was compared with every other method. Differences between calculated skeletal age and chronological age were assessed, and the sensitivity, specificity, and predictive values of each method computed. RESULTS: Coupled B20/RUS values showed the best agreement, with 95% of observations within 2.45 years of each other, followed by carpals/B20, B20/GP, and GP/RUS. The US method agreed the least (difference of 4.19-5.13 years) with the other methods. The US method provided 85.8% (US vs RUS) to 91.3% (US vs GP) concordant results in recognizing differences between skeletal and chronological age, and showed a 72.5% sensitivity and a 56.8% specificity. CONCLUSION: Although the US method promises to permit a safe and cost-effective assessment of skeletal age, its low accuracy makes it currently unsuitable for clinical use.