Objective: The purpose of this study was to investigate the impact of knowing chronologic age on the variability of pediatric bone age determination using the method of Greulich and Pyle.
Materials and methods: Radiographs of the left hand of 107 patients were interpreted by four radiologists on two separate occasions, once with and once without knowledge of the patient's chronologic age at time of interpretation. Twenty-five radiographs were randomly selected and reevaluated twice by each radiologist. Interobserver and intraobserver variability were calculated and compared for the two conditions. The distribution of studies with normal and abnormal findings was then compared across knowledge conditions for all observers and by individual observer, using two standard deviations above and below chronologic age as the range of "normal".
Results: When the chronologic age was known, the interobserver reliability coefficient for knowledge of chronologic age was 0.954 and when not known, 0.952. The intraobserver reliability coefficients when chronologic age was known ranged from 0.944 to 0.967, and when not known from 0.938 to 0.980. Observers interpreted 58% (248/428) of the radiographs as having normal findings when chronologic age was known and 48% (205/428) when chronologic age was not known.
Conclusion: Knowing chronologic age before assessing bone age radiographs does not affect the reproducibility of assessment. However, observers are more likely to interpret the radiograph as showing normal findings when chronologic age is known than if the interpretation is performed with the observer unaware of chronologic age.