OBJECTIVE

Age estimation methods using mandibular third molar (M3) root formation were tested. Diagnostic accuracy of M3 to predict age 18 was tested.

DESIGN

Methods were tested on a target sample of 300 dental panoramic radiographs (age 11-25). Diagnostic accuracy was assessed on separate reference data (n = 1,663, age 9-25). Root stage was the diagnostic test predicting 18 years of age.

METHODS

Root stage of M3 was assessed and age estimated (n = 157) using published methods that use Demirjian or Moorrees root stages. The difference between dental and known ages was assessed. Diagnostic tests and likelihood ratios were calculated for reference data.

MAIN OUTCOME MEASURE

Mean difference (bias), standard deviation and absolute mean difference between dental age and known ages. Likelihood ratio of age 18, given M3 root stage.

RESULTS

Only six of 37 methods estimated age with bias not significant to zero. Mean absolute difference between dental and known age for these methods ranged from 1.45 to 1.97 years. Standard deviation of bias for all methods was around 2 years and 95% confidence interval of estimated age is ± 4 years. The best methods using Demirjian and Moorrees stages are detailed. Likelihood ratio of being at least 18 if M3 was mature was 13.61. If M3 was 'A1/2' (apex half closed) or mature, on the balance of probabilities, estimated age was at least 18.

CONCLUSION

Most methods using M3 root formation estimate age with significant bias. If M3 is mature, age 18 is more than likely attained.