Does increased staging complexity influence expert agreement in dental age estimation?

Dental age estimation (DAE) is a well-established procedure in current practice, and it is frequently required in forensic and legal investigations. This study aimed to evaluate the intra- and inter-observer reliability of dental development staging among forensic practitioners using two widely applied staging systems, according to Demirjian et al. 1973 and Moorrees et al. 1963. The study included 250 anonymized panoramic radiographs of individuals between 6 and 14 years (median age, 10 years), comprising 51.2% males and 48.8% females. Eighteen observers with prior training in DAE independently assessed tooth development on panoramic radiographs using the Demirjian and Moorrees methods, with each method evaluated in two separate rounds. Intra-observer reliability was evaluated using weighted and unweighted Cohen's Kappa, while inter-observer agreement was assessed using Gwet's AC1 and AC2 coefficients. Weighted analyses accounted for the ordinal structure of developmental stages, whereas unweighted analyses reflected exact agreement. Intra-observer reliability was consistently high across both systems, with median weighted Cohen's Kappa values ranging from 0.897 to 0.957 and median unweighted Cohen's Kappa values ranging from 0.836 to 0.944. Inter-observer agreement values averaged across rounds assessed with Gwet's AC2 ranged from 0.914 to 0.987, whereas mean Gwet's AC1 ranged from 0.306 to 0.908, indicating that disagreement was primarily due to adjacent-stage differences. Moorrees staging showed slightly higher reliability under weighted conditions, while Demirjian demonstrated higher exact agreement under unweighted conditions. Agreement increased with the number of observers but reached a plateau at approximately 8-10 observers, suggesting an optimal number of observers for forensic consensus. The results indicated that difficulty in distinguishing adjacent stages, unclear stage boundaries, and overlapping anatomical structures were the main sources of uncertainty. The Moorrees system was perceived as more complex, while Demirjian showed difficulty focused on the root development stages. These findings highlight the importance of methodological considerations in dental age estimation and suggest that refinement of staging systems may further improve the reliability of expert assessment.