Methods of reducing observer variation in age estimation from panoramic radiographs

To compare inter- and intra-observer variation in age estimation using subjective and objective methods.

Fifty-six panoramic radiographs of patients aged 13 to 25 years were assessed subjectively twice by four observers for the stage of root development of the lower third molars on a seven-point scale. Two of the observers were calibrated before the first evaluation, and the other two before the second. The same observers also digitized predefined reference points in a subsample of the third molars ten times by means of a computer-based digitizer system. Observer variation in classifying different root stages and in digitizing coordinates and measuring root lengths were studied.

Significant systematic differences existed between two of the observers in the subjective assessment. One non-pre-calibrated observer changed her pattern most. Scattergrams from the digitizer measurements showed a good precision within and between observers with no significant differences in calculating root lengths. When, however, the teeth were analysed separately, some significant differences were found.

The overall observer agreement was best using the objective method on several teeth. With only one tooth there were some significant differences. Subjective evaluation was improved by prior calibration.