The third molars are one of few useful predictors for assessing the degree of maturity in adolescence and young adulthood. It has application in age estimation in the age group of 14–23 years, in general, and in juvenile/adult status prediction, in particular. Using a 10-stage grading of third molars, Gunst et al. developed regression formulae on a large sample of Belgians (n=2513) for estimating age. Their research has been recommended as a ‘reference study’ in age estimation guidelines. The present study has ventured to determine if estimating age in Indians using the Belgian formulae produced results comparable to those reported in the Belgian study; in addition, this study attempts to determine if the same formulae predicted juvenile/adult status (age </≥18 years) accurately in Indians. The sample comprised 268 orthopantomograms (OPGs) of subjects aged between 14 and 23 years. The OPGs included a mix of one, two, three and four third molars. In total, 916 teeth were assessed using the same 10-stage grading. Age in each OPG was estimated by applying the relevant Belgian regression formulae (regression formulae are available for one, two, three and four third molars). To determine if the formulae produced age estimates comparable to those in the Belgian study, the percentage of Indian subjects whose actual age fell within the 68% confidence interval (CI) (calculated from the ±1 S.D. value available for each Belgian formula) was ascertained. If ≥68% of Indian subjects’ age fell inside this interval, it indicates that the Belgian formulae are applicable in Indians. To assess the suitability of the Belgian formulae in predicting juvenile/adult status in Indians, the accuracy of the age estimation per se was not considered, rather, the number of correct </≥18 years age predictions only was noted. Overall, ∼74% of Indian subjects’ actual age fell within the 68% CI; with regards to the Belgian formulae being able to correctly predict juvenile/adult status, 78% of all subjects were categorized to the correct age group (</≥18 years). This was marginally better in males (79.4%) than females (77.3%), and in juveniles (80.3%) than adults (76.1%). These suggest that Belgian formulae are valid in age estimation per se of Indians; however, the formulae's accuracy rate in determining whether a subject is </≥18 years old in the Indian context, while similar to other statistical approaches, indicate that it may not be reliable-enough, and must be applied carefully and judiciously in juvenile/adult identification.
Read full abstract