Dental age estimation is of vital importance in the communities of forensic science and clinical dentistry[9, 20, 29–32]. Does a practical and accurate method exist to elucidate the inner relationships between the teeth and chronological age? Many works have been put forward and aroused drastic discussions. From decades of age to the last ten years, there have been huge explorations in the community of dental age estimation. In retrospect, Demirjian, along with his colleagues, performed clinical research between tooth development and chronological age in 1973[8]. They have proven a correlation between tooth development and actual age. Since then, several methods have been upgraded or constructed based on oral radiographs and tested in different populations and regions. Of them, some should be voiced here for their non-invasion and convenience, such as the Willems method[10], Cameriere method[33], and London atlas method[11].
The Willems method was first proposed by Willems et al. based on a population of Belgian-Caucasian teenagers in 2001[10]. The Willems method, in other words, a modified Demirjian dental age assessment method, simplified the former method by leaving out a step of data processing and improved the accuracy of dental age assessment. The initial Willems method remained a sex-specific trait, similar to the Demirjian method mentioned. Later, Willems et al. rethought the old method and constructed a reduced non-gender specific method (Willems II) to better serve the community of dental age estimation[26]. Willems II method provides a common standard for both genders. The Willems II method will display good performance in bioarchaeological studies, while the sex of skeletal remains is unknown. Although some studies on the Willems method have been conducted in other regions on the planet, to the best of our knowledge, the practical effect of the Willems II method has only been reported in only 3 studies and has not been fully investigated thus far [15, 34, 35]. Hence, the better choice of the two Willems methods applied in a population of eastern China is meaningful to explore.
The Willems method has gained popularity to a great extent as an easy-to-check approach to estimating the actual age. The Willems I method’s over/underestimation of age has been reported worldwide. overestimation of chronological age was validated in a variety of counties or regions such as India[36],Turkey[37],Thailand[38] ,Spain[39],Kenya[40]and Poland [41]. elsewhere, there were also reports of underestimation in north China[21], Tunisia[35], and Sri Lanka[42]. In allusion to the differences of the Willems method among several regions, a systemic review, and meta-analysis we conducted previously concerning the Willems I method revealed the Willems method overestimated CA by 0.18 years and 0.06 years for boys and girls, respectively[4]. It also revealed that ethnicity specificity was necessary when adopting the Willems I method. This conclusion was consistent with other similar meta-analyses[22, 23, 43]. In the current work, the underestimation of age was + 0.44 ± 1.17 (p < 0.05) and + 0.09 ± 1.37 (p = 0.091) for the boy and girl groups, respectively. From the whole perspective, compared with the boys’ group, the girls’ group showed a more accurate effect of age assessment as a whole. This phenomenon was also viewed in our previous works and other printed studies[5, 44].
For the Willems II method, only a few works have been reported over the last decades(Table 4). Urzel et al. observed a mean difference between CA and DA of 0.03 years for both genders [15]. As is depicted in the Fig. 3, we did not display the exact positions of DA by the Willems II method for both genders because the original paper only showed a mean difference value of 0.00 years from a whole scale. They concluded that both methods were appropriate when applied to a French population, although the Willems I method was more accurate according to their data. Another similar study from an Indian population revealed that the mean values were 0.06 ± 0.80, -0.11 ± 0.79, and − 0.01 ± 0.80 years for boys, girls, and the total sample, respectively [34]. Among Tunisian sub-adults [35], the Willems II method tended to underestimate age by 0.91, and 0.64 years for boys and girls, respectively. Whereas the Willems I method indicated an underestimation of chronological age by 0.40 years for boys, and by 0.69 years for girls. In their investigation, Willems I was more accurate when compared with Willems II in the Tunisian population. In the current study, the mean values of CA-DA were 0.57 ± 1.20 (p < 0.05) and − 0.09 + 1.34 (p = 0.106) years for boys and girls, respectively. A more accurate estimation was observed in the girls’ group in the two methods. These results revealed that Willems I was more accurate and reliable for boys, while Willems II was a better option for girls. Our results are in accordance with the aforementioned studies.
Despite the accuracy assessment of Willems I and Willems II, the study still has limitations that must be elucidated. The samples we collected ranged from 11 to 16 years old. We did not retrieve lower age groups younger than 11 years old. Larger sample size will help better understand the suitability of the methods.
In conclusion, the present study investigated the applicability of the Willems I and Willems II methods with a sample of eastern Chinese teenagers aged 11 to 16 years old. These outcomes suggest that Willems I is more accurate for boys, while the Willems II method prefers girls; both methods may not be a perfect match for the eastern Chinese Han population. An ethnicity-specific model based on the Willems method or further modifications is encouraged to prosper the science of dental age estimation.