Accuracy of identifying juvenile/adult status from third molar development using prediction probabilities derived from logistic regression analysis.

The use of third molars in predicting juvenile/adult status (</≥ 18 years) has important legal ramifications. Third molar development was assessed using Köhler's grading on 268 orthopantomograms of Indian subjects. Logistic regression analysis was applied to determine allocation accuracy of juvenile/adult status and the level of probability that is "reliable" in predicting juvenile/adult status. Allocation accuracies ranged between 75.8% and 78.2% for the sexes combined, with minimal male-female differences. Adults were categorized more accurately than juveniles, suggesting that Köhler's grading puts Indian juveniles at greater risk of unwarranted punishment. In both sexes, juvenile/adult status was "reliably" predicted when the probability was >80% using individual third molars (excepting the lower right third molar in males); combining upper and lower third molars on the left/right sides, "reliable" predictions were possible when the probability was >80% and >90% for females and males, respectively. Overall, "reliable" juvenile/adult status prediction was achieved in c. 36% of subjects.