Evaluation of intraexaminer and interexaminer agreement on classifying lower third molars according to the systems of Pell and Gregory and of Winter.

Purpose: This study evaluated the intraexaminer and interexaminer reproducibility of 2 systems (Pell and Gregory [P&G] and Winter) for classifying the anatomic position of third molars (M3s) and compared these 2 traditional classification systems with a novel computer-based system (Radio Memory software; Belo Horizonte, Brazil).

Materials And Methods: Using a descriptive observational study design, we enrolled a sample of 4 surgeons who evaluated M3 position on panoramic radiographs. They classified M3 position (P&G) and inclination (Winter) on 2 different occasions separated by 15 days. M3 position and inclination were also classified using the Radio Memory software. Intraexaminer and interexaminer agreement were computed, as was agreement between the examiners and the Radio Memory software.

Results: The radiographic sample comprised 40 M3s. Intraexaminer agreement on classifying the molars according to their inclination was greater than 85% in all cases (mean agreement, 89.37%), whereas reproducibility of classification according to tooth position was observed in only 2 examiners (mean agreement, 66.25%). For the first observations, interexaminer agreement was observed for both classifications, whereas for the second observation, reproducibility among all of the examiners was observed only for the Winter classification. No homogeneity was recorded between the measurements made by the observers and the objective results regarding the position and inclination of the molars determined by the Radio Memory software.

Conclusions: The classification of impacted lower M3s based on their position and following the P&G criteria lacks both intraexaminer and interexaminer reproducibility. However, the levels of intraexaminer and interexaminer agreement are very high when attempting to classify the M3s according to their inclination (Winter classification). These results suggest the advisability of using an objective measurement method to minimize the error introduced by observer interpretation.