Abstract
Background
Demirjian’s method of age estimation has been reported to overestimate age and Willems’ method to give consistently more accurate results. Not enough, however, is known about the applicability of Chaillet’s standards.
Aim
The present study aimed to compare the accuracy of Demirjian’s, Willems’ and Chaillet’s standards in age estimation of 5 to 15 year-old Indian children.
Design
In this cross-sectional observational study, three methods were compared for accuracy in estimating the age of 1200 Indian children aged 5-15 years.
Results
Demirjian’s method overestimated age by +0.24 ± 0.80 years, +0.11 ± 0.81years and +0.19 ± 0.80 years in boys, girls and the total sample, respectively. With Willems’ method, overestimations of +0.09 ± 0.80 years, +0.08 ± 0.80 years and +0.09 ± 0.80 years were obtained in boys, girls and the total sample, respectively. Chaillet’s method underestimated age by -0.12 ± 0.69 years, -0.45 ± 0.88 years and -0.25 ± 0.83 years in boys, girls and the total sample, respectively. Statistically significant differences were observed between dental and chronological ages with all methods (p < 0.001). Significant sex-based differences were observed only with Demirjian’s and Chaillet’s methods (p < 0.05).
Conclusion
Willems’ method was the most accurate in age estimation, followed by Demirjian’s and Chaillet’s methods. While Demirjian’s method was more accurate than Chaillet’s in females, Chaillet’s method better predicted the age of males.
KEYWORDS : Age estimation, Demirjian’s method, Willems’ method, Chaillet’s method, Indian children
Introduction
First proposed more than four decades ago, the Demirjian method (1) has emerged as the most widely researched and applied technique in dental age (DA) estimation of children and adolescents. This is owing largely to the simplicity of the method, as well as the radiographic and schematic illustrations of tooth development and accompanying description, which the original (1) and subsequent works (2) provided. However, Demirjian’s French-Canadian standards have been reported to consistently overestimate age in most populations, including a Belgian-Caucasian population studied by Willems and co-workers, (3) which highlighted the need for population-specific databases. In order to avoid this overestimation, these latter researchers adapted Demirjian’s maturity scores using a weighted ANOVA and constructed new tables for boys and girls, from which a maturity score could be directly expressed
in years. The Willems method is increasingly recognized as a more accurate predictor of age in recent years. Chaillet and co-workers, (4) adapted Demirjian’s scores for the same Belgian population using polynomial functions and constructed another set of tables which they observed to be more accurate for this population.
The dental literature abounds with reports (4-18) on the applicability of the Demirjian method and its variations to non-Indian and Indian populations. Extensive research (7, 8, 11, 14, 16-20) has been carried out using the Willems method as well. However, previous research on Indian populations is limited by inadequate sample sizes, age structures, grouping and approaches to statistical analysis. (16-19) Tests of the application of the Chaillet standards to non-Indian populations are few, (4, 5, 21-23) and while one study (24) did test these standards in an Indian population, the sample size was relatively small, the age range was 9 to 20 years and the Demirjian 8-teeth method was used for dental staging. Hence, the present study proposed to evaluate and compare the validity of the Demirjian, Willems and Chaillet standards in determination of chronological age (CA) of a reasonably large sample of 1200 Indian children aged 5 to 15 years.
METHODOLOGY:
This study was designed as a cross-sectional observational study. Ethical clearance was obtained from the Ethical Committee, Pacific Dental College and Hospital, Udaipur, India. Parents/ guardians had signed an agreement with the dental institution that dental records and radiographs could be used only for research and educational purposes without the possibility of personal identification.
Sampling method: A convenience sampling method was employed, all radiographs being digital and made during the period from January 2012 to September 2015 of children aged between 5.0 and 15.9 years who had sought treatment at the Department of Paediatric Dentistry, Pacific Dental College and Hospital, Udaipur, Rajasthan, India, and required an orthopantomograph (OPG) as part of the investigation protocol.
Selection criteria: Both parents of all children were of Indian origin and nationality. Only patients with a documented date of birth and date of radiography in the oral health record were included to facilitate verification of the chronological age (in completed decimal years) for each subject. Orthopantomographs with image distortion due to improper position or movement of the patient during exposure, and incomplete image or lack of clarity resulting from an improper exposure technique were excluded. Also, radiographs were excluded from the study if the patient had any history of surgical/medical treatment or systemic illness with the potential to cause significantly delayed or early development, significant numbers of teeth other than third molars missing either congenitally or due to disease and trauma, or malformation of teeth or obvious dental pathology that could affect tooth development.
Final sample: Of the 1303 radiographs collected, 103 did not meet the selection criteria owing to either congenital absence of several teeth (n = 22), lack of image clarity (n = 08) or inadequate information regarding the date of birth (n = 73). Thus, a final sample of 1200 OPGS of 699 male and 501 female Indian children aged 5 to 15 years was selected for the study. The distribution of radiographs by age and sex is presented in Table 1. Radiographs of patients aged 5.0 to 5.9 years were included in age group 5, of those aged 6.0 to 6.9 years in age group 6 and so on. Thus, age group 15 consisted of children aged 15.0 to 15.9 years.
Table 1. Distribution of the study sample by age and gender.
| Chronological age (years) | Females | Males | Total | ||||
|---|---|---|---|---|---|---|---|
| Age group | Age range | N | % | N | % | N | % |
| 5 | 5.0 - 5.9 | 24 | 4.79 | 23 | 3.29 | 47 | 3.92 |
| 6 | 6.0 - 6.9 | 39 | 7.78 | 40 | 5.72 | 79 | 6.58 |
| 7 | 7.0 - 7.9 | 46 | 9.18 | 58 | 8.30 | 104 | 8.67 |
| 8 | 8.0 - 8.9 | 50 | 9.98 | 58 | 8.30 | 108 | 9.00 |
| 9 | 9.0 - 9.9 | 55 | 10.98 | 78 | 11.16 | 133 | 11.08 |
| 10 | 10.0 - 10.9 | 55 | 10.98 | 100 | 14.31 | 155 | 12.92 |
| 11 | 11.0 - 11.9 | 40 | 7.98 | 82 | 11.73 | 122 | 10.17 |
Calculation of chronological age: The dates of birth and of making the OPG were obtained from the hospital records. A function of Microsoft Excel was used to calculate the difference between the recorded date of birth and the date on which the OPG was made, to obtain the chronological age (CA) in decimal years.
Data collection: All digital radiographs meeting the selection criteria were saved as jpeg images and viewed on the same LCD monitor using Windows Photo Viewer and a magnifying glass for improved visualization. The examiners were allowed to use resources of image improvement such as brightness, contrast and zoom. Each OPG was coded with a numerical ID to avoid examiner bias. Age and sex of the subjects were thus unknown to the examiner. Nomenclature for teeth assessed was assigned according to the Fédération Dentaire Internationale (FDI) system. Seven mandibular teeth of the left side (excluding the third molar) were evaluated by the Demirjian’s dental staging method. (1) Once the stage that most accurately described the stage of development of the tooth in question was identified, the corresponding alpha-numeric rating (0 to H) was assigned to that tooth. The alpha-numeric stages 0 to H were converted to the revised self-weighted gender-specific numerical scores of Demirjian and Goldstein. (2) The individual scores for were summed to obtain a total maturity score or dental score (DS), which was converted to a dental age (DA) using the Demirjian, Tanner and Goldstein tables. (1)
In Chaillet’s method, the maturity scores provided by Chaillet et al. (4) were summed to obtain the total maturity score, which was converted to a dental age using the tables constructed by the same authors, while in the Willems method, the maturity scores provided by Willems et al. (3) for each tooth were summed to directly provide the dental age in years.
Reproducibility of measurements: Two well-trained examiners independently evaluated 100 radiographs using Demirjian’s method of dental staging, after a period of mutual calibration without any knowledge of age or sex, in order to allow an analysis of inter-examiner agreement. Ultimately, a single examiner assessed all radiographs. Intra-examiner agreement was assessed by having one examiner re-evaluate the same 100 radiographs after a period of two months without any knowledge of sex or age or of the stages assigned in the first evaluation.
Data analysis: All statistical analyses and data management were performed using Statistical Package of Social Sciences 19.0 (SPSS Inc., Chicago, IL, USA) for Windows and MS-Excel (Microsoft Office 2010). Analyses were made for each sex and age group, and for the total sample. Kolmogorov-Smirnov and Shapiro-Wilk tests were performed to test the normality of the data. As the sample showed a non-normal distribution, non-parametric tests were applied. For all tests, a p value ≤ 0.05 was considered statistically significant.
The accuracy of each method of age estimation was determined by mean difference between estimated dental age and the chronological age (DA–CA) for each sex and age group, and the total sample. A positive result indicated an overestimation, and a negative result indicated an underestimation of age. Box-plot graphs are used to present the mean DA-CA of each sex and age group, and the total sample, with whiskers indicating the range. Absolute accuracy was determined by means of the absolute differences between DA and CA of girls and boys and the total sample for each method. The Wilcoxon signed rank test was applied to assess the significance of DA-CA for both methods for each sex and age group, for the total sample and between methods. Independent t-test was employed for comparisons of DA-CA between sexes. The correlation between DA and CA was analysed using Spearman’s rank correlation coefficient for each sex and for the total study sample. Inter- and intra-examiner agreements are expressed as percentages. Cohen’s kappa coefficient was used to calculate the degree of reliability of these agreements. Regression analyses were performed and gender-specific equations were derived for all three methods.
RESULTS
The mean age (± SD) of the entire sample was 10.75 ± 2.72 years, those of girls and boys being 10.68 ± 2.87 and 10.81 ± 2.60, respectively. Inter- and intra-examiner agreements were 86% and 93% respectively, with Kappa values of 0.81 and 0.90 indicating almost perfect agreement.
In the present study, the mean chronological and Demirjian dental ages for girls were 10.68 ± 2.87 and 10.79 ± 2.86 years, respectively, while for boys the values were 10.81 ± 2.60 and 11.05 ± 2.71 years, respectively. The mean differences between dental and chronological ages for boys, girls and the total sample (+0.24 ± 0.80, +0.11 ± 0.81 and +0.19 ± 0.80 years, respectively) were statistically significant (p < 0.01). Significant differences between mean dental and chronological ages were observed in age groups 6, 8, 9, 12, 14 and 15 for girls and 6, 7, 8, 10, 11, 14 and 15 for boys (p ≤ 0.05). In girls, Demirjian’s method overestimated age by +0.06 to +0.54 years in all age groups, with the exception of groups 8 and 14, for which underestimations of -0.21 and -0.37 years, respectively, were obtained. In boys, overestimations ranged from +0.15 to + 0.59 years in most age groups, with underestimations of -0.13 and -0.02 years for groups 5 and 12, respectively (Table 2).
Table 2. Comparison of chronological age (CA) and Demirjian dental age (DA) by gender and age group.
| Gender | Age group (years) | N | CA | DA | DA-CA | p value* | p value# |
|---|---|---|---|---|---|---|---|
| Mean ± SD (years) | |||||||
| GIRLS | 5 | 24 | 5.46 ± 0.33 | 5.53 ± 0.58 | + 0.07 ± 0.43 | 0.543 | 0.846 |
| 6 | 39 | 6.57 ± 0.32 | 7.11 ± 0.58 | + 0.54 ± 0.38 | < 0.001 | < 0.001 | |
| 7 | 46 | 7.52 ± 0.26 | 7.65 ± 0.67 | + 0.13 ± 0.47 | 0.097 | 0.167 | |
| 8 | 50 | 8.51 ± 0.31 | 8.30 ± 0.73 | - 0.21 ± 0.44 | 0.049 | 0.048 | |
| 9 | 55 | 9.48 ± 0.30 | 9.84 ± 0.77 | + 0.36 ± 0.43 | 0.001 | 0.001 | |
| 10 | 55 | 10.55 ± 0.32 | 10.72 ± 0.85 | + 0.17 ± 0.52 | 0.173 | 0.101 | |
| 11 | 40 | 11.44 ± 0.32 | 11.58 ± 0.84 | + 0.14 ± 0.54 | 0.352 | 0.357 | |
| 12 | 55 | 12.49 ± 0.32 | 12.75 ± 0.87 | + 0.26 ± 0.53 | 0.048 | 0.049 | |
| 13 | 57 | 13.46 ± 0.30 | 13.52 ± 0.83 | + 0.06 ± 0.50 | 0.607 | 0.502 | |
| 14 | 59 | 14.48 ± 0.28 | 14.11 ± 0.62 | - 0.37 ± 0.39 | < 0.001 | < 0.001 | |
| 15 | 21 | 15.48 ± 0.27 | 15.73 ± 0.51 | + 0.25 ± 0.26 | 0.024 | 0.023 | |
| Total | 501 | 10.68 ± 2.87 | 10.79 ± 2.86 | + 0.11 ± 0.81 | 0.003 | 0.002 | |
| BOYS | 5 | 23 | 5.56 ± 0.29 | 5.43 ± 0.98 | - 0.13 ± 0.42 | 0.551 | 0.503 |
| 6 | 40 | 6.52 ± 0.31 | 7.05 ± 0.78 | + 0.53 ± 0.44 | < 0.001 | < 0.001 | |
| 7 | 58 | 7.48 ± 0.29 | 7.72 ± 0.67 | + 0.24 ± 0.46 | 0.015 | 0.004 | |
| 8 | 58 | 8.47 ± 0.29 | 8.89 ± 0.61 | + 0.42 ± 0.97 | < 0.001 | < 0.001 | |
| 9 | 78 | 9.46 ± 0.28 | 9.61 ± 0.82 | + 0.15 ± 0.45 | 0.111 | 0.069 | |
| 10 | 100 | 10.45 ± 0.29 | 10.76 ± 0.84 | + 0.31 ± 0.47 | < 0.001 | < 0.000 | |
| 11 | 82 | 11.51 ± 0.30 | 11.66 ± 0.70 | + 0.15 ± 0.37 | 0.035 | < 0.001 | |
| 12 | 91 | 12.44 ± 0.30 | 12.42 ± 0.74 | - 0.02 ± 0.45 | 0.705 | 0.541 | |
| 13 | 82 | 13.41 ± 0.31 | 13.56 ± 0.78 | + 0.15 ± 0.36 | 0.088 | 0.106 | |
| 14 | 58 | 14.47 ± 0.31 | 15.06 ± 0.58 | + 0.59 ± 0.46 | < 0.001 | < 0.001 | |
| 15 | 29 | 15.24 ± 0.25 | 15.80 ± 0.40 | + 0.56 ± 0.35 | < 0.001 | < 0.001 | |
| Total | 699 | 10.81 ± 2.60 | 11.05 ± 2.71 | + 0.24 ± 0.80 | < 0.001 | < 0.001 | |
| Total sample | 1200 | 10.75 ± 2.72 | 10.94 ± 2.78 | + 0.19 ± 0.80 | < 0.001 | < 0.001 | |
*Paired t test, #Wilcoxon Signed Rank test: p ≤ 0.05 = significant
The mean chronological and Willems’ dental ages for girls were 10.68 ± 2.87 and 10.76 ± 3.04 years, respectively, while for boys the values were 10.81 ± 2.60 and 10.90 ± 2.66 years, respectively. The mean differences between dental and chronological ages for boys, girls and the total sample (+0.09 ± 0.80, +0.08 ± 0.80 and +0.09 ± 0.80 years, respectively) were statistically significant (p < 0.05). Significant differences between mean dental and chronological ages were observed in age groups 6, 11, 13 and 14 for girls and 8, 10, and 15 for boys (p ≤ 0.05). In girls, Willems’ method overestimated age by +0.01 to +0.61 years in most age groups, while underestimating age by -0.02 to -0.25 years, in groups 7, 8, 10 and 13. In boys, overestimations ranged from +0.01 to +0.35 years in all age groups, with the exception of age group 12 for which an underestimation of -0.10 years was obtained (Table 3).
Table 3. Comparison of chronological age (CA) and Willems’ dental age (DA) by gender and age group.
| Gender | Age group (years) | N | Mean age ± SD (years) | Mean DA-CA (years) | p value# | |
|---|---|---|---|---|---|---|
| CA | DA | |||||
| GIRLS | 5 | 24 | 5.46 ± 0.33 | 5.47 ± 0.86 | + 0.01 ± 0.84 | 0.964 |
| 6 | 39 | 6.57 ± 0.32 | 6.81 ± 0.70 | + 0.24 ± 0.64 | 0.033 | |
| 7 | 46 | 7.52 ± 0.26 | 7.39 ± 0.78 | - 0.13 ± 0.75 | 0.191 | |
| 8 | 50 | 8.51 ± 0.31 | 8.46 ± 0.73 | - 0.05 ± 0.63 | 0.765 | |
| 9 | 55 | 9.48 ± 0.30 | 9.52 ± 0.78 | + 0.04 ± 0.79 | 0.811 | |
| 10 | 55 | 10.55 ± 0.32 | 10.53 ± 0.81 | - 0.02 ± 0.85 | 0.694 | |
| 11 | 40 | 11.44 ± 0.32 | 11.73 ± 0.89 | + 0.29 ± 0.84 | 0.032 | |
| 12 | 55 | 12.49 ± 0.32 | 12.53 ± 0.83 | + 0.04 ± 0.88 | 0.744 | |
| 13 | 57 | 13.46 ± 0.30 | 13.21 ± 0.71 | - 0.25 ± 0.70 | 0.018 | |
| 14 | 59 | 14.48 ± 0.28 | 15.09 ± 0.93 | + 0.61 ± 0.93 | < 0.001 | |
| 15 | 21 | 15.48 ± 0.27 | 15.63 ± 0.37 | + 0.15 ± 0.42 | 0.126 | |
| Total | 501 | 10.68 ± 2.87 | 10.76 ± 3.04 | + 0.08 ± 0.80 | 0.049 | |
| BOYS | 5 | 23 | 5.56 ± 0.29 | 5.74 ± 0.73 | + 0.18 ± 0.73 | 0.301 |
| 6 | 40 | 6.52 ± 0.31 | 6.62 ± 0.77 | + 0.10 ± 0.84 | 0.183 | |
| 7 | 58 | 7.48 ± 0.29 | 7.62 ± 0.78 | + 0.14 ± 0.73 | 0.153 | |
| 8 | 58 | 8.47 ± 0.29 | 8.69 ± 0.69 | + 0.22 ± 0.76 | 0.043 | |
| 9 | 78 | 9.46 ± 0.28 | 9.56 ± 0.88 | + 0.10 ± 0.90 | 0.410 | |
| 10 | 100 | 10.45 ± 0.29 | 10.65 ±0.82 | + 0.20 ± 0.80 | 0.015 | |
| 11 | 82 | 11.51 ± 0.30 | 11.53 ± 0.75 | + 0.02 ± 0.75 | 0.705 | |
| 12 | 91 | 12.44 ± 0.30 | 12.34 ± 0.70 | - 0.10 ± 0.67 | 0.145 | |
| 13 | 82 | 13.41 ± 0.31 | 13.42 ± 0.76 | + 0.01 ± 0.81 | 0.891 | |
| 14 | 58 | 14.47 ± 0.31 | 14.54 ±0.68 | + 0.07 ± 0.75 | 0.553 | |
| 15 | 29 | 15.24 ± 0.25 | 15.59 ±0.76 | + 0.35 ± 0.82 | 0.048 | |
| Total | 699 | 10.81 ± 2.60 | 10.90 ± 2.66 | + 0.09 ± 0.80 | 0.002 | |
| Total sample | 1200 | 10.75 ± 2.72 | 10.84 ± 2.83 | + 0.09 ± 0.80 | < 0.001 | |
#Wilcoxon Signed Rank test: p ≤ 0.05 = significant
The mean chronological and Chaillet’s dental ages for girls were 10.68 ± 2.87 and 10.23 ± 2.83 years, respectively, while for boys the values were 10.81 ± 2.60 and 10.69 ± 2.68 years, respectively. The mean differences between dental and chronological ages for boys, girls and the total sample (-0.45 ± 0.88, -0.12 ± 0.69 and -0.25 ± 0.83 years, respectively) were statistically significant (p < 0.001). Significant differences between mean dental and chronological ages were observed in age groups 8, 10, 11 and 12 for girls and 5, 6, 8 and 14 for boys (p < 0.05). In girls, the Chaillet method underestimated age by -0.04 to -0.93 years in all age groups, except group 5 for which an overestimation of +0.20 years was obtained. In boys, underestimations ranged from -0.01 to -0.36 years in all age groups, while overestimating age by +0.06 to +0.29 years in group 5, 8 and 14 (Table 4).
Table 4. Comparison of chronological age (CA) and Chaillet dental age (DA) by gender and age group.
| Gender | Age group (years) | N | CA | DA | DA-CA | p value# |
|---|---|---|---|---|---|---|
| Mean ± SD (years) | ||||||
| GIRLS | 5 | 24 | 5.46 ± 0.33 | 5.66 ± 0.84 | 0.20 ± 0.64 | 0.308 |
| 6 | 39 | 6.57 ± 0.32 | 6.53 ± 0.67 | - 0.04 ± 0.30 | 0.613 | |
| 7 | 46 | 7.52 ± 0.26 | 7.12 ± 0.94 | - 0.40 ± 0.67 | 0.062 | |
| 8 | 50 | 8.51 ± 0.31 | 8.00 ± 1.04 | - 0.51 ± 0.60 | 0.005 | |
| 9 | 55 | 9.48 ± 0.30 | 9.17 ± 0.66 | - 0.31 ± 0.53 | 0.384 | |
| 10 | 55 | 10.55 ± 0.32 | 10.11 ± 0.73 | - 0.44 ± 0.63 | 0.002 | |
| 11 | 40 | 11.44 ± 0.32 | 10.88 ± 0.63 | - 0.56 ± 0.73 | 0.004 | |
| 12 | 55 | 12.49 ± 0.32 | 11.56 ± 0.69 | - 0.93 ± 0.91 | 0.001 | |
| 13 | 57 | 13.46 ± 0.30 | 12.64 ± 0.96 | - 0.82 ± 1.17 | 0.893 | |
| 14 | 59 | 14.48 ± 0.28 | 14.35 ± 0.99 | - 0.13 ± 1.06 | 0.150 | |
| 15 | 21 | 15.48 ± 0.27 | 14.69 ± 0.70 | - 0.79 ± 0.90 | 0.476 | |
| Total | 501 | 10.68 ± 2.87 | 10.23 ± 2.83 | - 0.45 ± 0.34 | 0.005 | |
| BOYS | 5 | 23 | 5.56 ± 0.29 | 5.85 ± 0.62 | 0.29 ± 0.71 | 0.001 |
| 6 | 40 | 6.52 ± 0.31 | 6.48 ± 0.88 | - 0.04 ± 0.59 | 0.097 | |
| 7 | 58 | 7.48 ± 0.29 | 7.47 ± 0.86 | - 0.01 ± 0.25 | 0.614 | |
| 8 | 58 | 8.47 ± 0.29 | 8.53 ± 0.80 | 0.06 ± 0.37 | 0.008 | |
| 9 | 78 | 9.46 ± 0.28 | 9.32 ± 0.95 | - 0.14 ± 0.58 | 0.801 | |
| 10 | 100 | 10.45 ± 0.29 | 10.22 ± 0.89 | - 0.23 ± 0.62 | 0.182 | |
| 11 | 82 | 11.51 ± 0.30 | 11.38 ± 0.79 | - 0.13 ± 0.65 | 0.504 | |
| 12 | 91 | 12.44 ± 0.30 | 12.17 ± 0.84 | - 0.27 ± 0.77 | 0.832 | |
| 13 | 82 | 13.41 ± 0.31 | 13.05 ± 0.91 | - 0.36 ± 0.86 | 0.825 | |
| 14 | 58 | 14.47 ± 0.31 | 14.71 ± 0.92 | 0.24 ± 0.92 | 0.007 | |
| 15 | 29 | 15.24 ± 0.25 | 15.18 ± 0.69 | - 0.06 ± 0.91 | 0.795 | |
| Total | 699 | 10.81 ± 2.60 | 10.69 ± 2.68 | - 0.12 ± 0.20 | < 0.001 | |
| Total sample | 1200 | 10.75 ± 2.72 | 10.50 ± 2.75 | - 0.25 ± 0.32 | < 0.001 | |
#Wilcoxon Signed Rank test: p ≤ 0.05 = significant
The lowest mean DA-CA was obtained with the Willems method, followed by the Demirjian and Chaillet method. Significant sex-based differences were observed in mean DA-CA with the Demirjian and the Chaillet methods (p = 0.005 and p < 0.0001, respectively) but not the Willems method (p > 0.05). With the Demirjian method the mean DA-CA was significantly lower in girls than in boys, whereas, the reverse was true with the Chaillet’s method (Table 5).
Table 5. Intra-method comparison between genders of mean dental and chronological ages (DA-CA).
| Gender | N | Demirjian | Chaillet | Willems | |||
|---|---|---|---|---|---|---|---|
| Mean DA-CA ± SD (years) | p value | Mean DA-CA ± SD (years) | p value | Mean DA-CA ± SD (years) | p value | ||
| Girls | 501 | + 0.11 ± 0.81 | 0.005 | - 0.45 ± 0.34 | <0.001 | 0.08 ± 0.80 | 0.830 |
| Boys | 699 | + 0.24 ± 0.80 | - 0.12 ± 0.20 | 0.09 ± 0.80 | |||
Independent t-test; p≤ 0.05 = significant
When the Spearman correlation coefficient test was performed for girls, boys and the total sample, strong linear correlations were observed between CA and DA for all methods, r values ranging from 0.875 to 0.966 and p values < 0.001 (Table 6). An inter-method comparison of mean DA-CA values revealed statistically significant (p < 0.05) differences in girls, boys and the total sample (Table 7).
Table 6. Correlation between chronological and dental ages by method.
| Method | r / p values | Females | Males | Total sample |
|---|---|---|---|---|
| Demirjian | r value | 0.957 | 0.962 | 0.961 |
| p value | < 0.001 | < 0.001 | < 0.001 | |
| Chaillet | r value | 0.837 | 0.877 | 0.841 |
| p value | < 0.001 | < 0.001 | < 0.001 | |
| Willems | r value | 0.966 | 0.959 | 0.962 |
| p value | < 0.001 | < 0.001 | < 0.001 |
Spearman’s rank correlation coefficient: r = Spearman’s rho, p = significant
Table 7. Inter-method comparison of mean dental and chronological ages (DA-CA) by gender and age group.
| Method with mean DA-CA | Difference in mean DA-CA | 95% CI of DA-CA | Absolute difference | p value# | ||
|---|---|---|---|---|---|---|
| (years ± SD) | Years | |||||
| Females (N = 501) | Demirjian | Willems | -0.03 ± 0.80 | 0.05 to 0.16 | 0.61 | < 0.001 |
| Chaillet | -0.56 ± 0.85 | 0.48 to 0.62 | 0.62 | < 0.001 | ||
| Willems | Chaillet | 0.53 ± 0.84 | -0.53 to -0.40 | 0.65 | < 0.001 | |
| Males (N = 699) | Demirjian | Willems | -0.15 ± 0.80 | 0.09 to 0.16 | 0.56 | < 0.001 |
| Chaillet | -0.36 ± 0.75 | 0.05 to 0.14 | 0.58 | 0.012 | ||
| Willems | Chaillet | 0.21 ± 0.75 | -0.16 to 0.44 | 0.56 | < 0.001 | |
| Total (N = 1200) | Demirjian | Willems | -0.10 ± 0.80 | 0.03 to 0.26 | 0.58 | < 0.001 |
| Chaillet | -0.44 ± 0.83 | -0.20 to 0.54 | 0.60 | < 0.001 | ||
| Willems | Chaillet | -0.34 ± 0.82 | -0.34 to -0.19 | 0.62 | < 0.001 | |
#Wilcoxon Signed Rank test: p ≤ 0.05 = significant
Box-plots in Figures 1-5 present the mean DA-CA of each sex and age group, and the total sample,with whiskers indicating the range.Regression analyses were performed and the following equations were derived:
Figure 1.
Box plot of mean DA-CA of females aged 5-10 years
Figure 2.
Box plot of mean DA-CA of females aged 11-15 years
Figure 3.
Box plot of mean DA-CA of males aged 5-10 years
Figure 4.
Box plot of mean DA-CA of males aged 11-15 years
Figure 5.
Box plot of mean DA-CA by gender for the total sample
For the Demirjian method:
Males: CA = 0.914 + 0.883 × DA
Females: CA = 0.316 + 0.961 × DA
For the Willems method:
Males: CA = 0.622 + 0.934 × DA
Females: CA = 0.894 + 0.909 × DA
For the Chaillet method:
Males: CA = 2.104 + 0.791 × DA
Females: CA = 0.768 + 0.969 × DA
DISCUSSION
Owing to the consistent overestimations of age reported with Demirjian’s French-Canadian standards in most populations, Willems and his co-workers (3) and Chaillet et al., (4) developed new dental maturity standards using weighted ANOVA and polynomial functions, respectively, for the same Belgian population. While the Demirjian and Willems methods have been widely tested, the Chaillet standards have not, especially on Indian populations. Therefore, the present study compared the Demirjian, Willems and Chaillet methods on a sample of orthopantomographs of Indian children aged 5 to 15 years, obtained by a convenience sampling method. This method is preferred by most researchers because it is fast, inexpensive and easy and the subjects are conveniently accessible.
Only the mandibular teeth were evaluated in the present study because, unlike the developing maxillary permanent teeth whose radiographic views are often obstructed by bony structures of the maxilla, the teeth of the mandible are quite clearly visible in an OPG. Further, a very high degree of symmetry is known to exist between the teeth of the left and right sides, only the seven mandibular teeth of the left quadrant were assessed. Third molar germs were excluded from assessment because of the high degree of variability observed in third molar genesis and development.
While assessing dental age, it is important not only to consider the proximity of the estimated age to the actual or chronological age, but also the reproducibility of the age estimation method. In the present study, agreements between and within examiners for Demirjian’s method of dental staging were obtained in percentages and measured by Cohen’s kappa coefficient. This coefficient is a more robust measure than simple percent agreement calculation, taking into account the agreement occurring by chance. Inter- and intra-examiner agreements for Demirjian’s dental staging were observed to be 91% and 90%, respectively, with a kappa coefficient of 0.83. The difference in mean DA-CA was not significant between examiners or between two assessments by the first examiner. Other studies have reported kappa values ranging from 0.68 to 0.92 for inter-examiner agreements (9, 10) and from 0.67 to 0.96 for intra-examiner agreements. (11, 12)
The Demirjian method has been found to consistently overestimate age in various populations, with a majority of studies reporting overestimations of up to +1.23 years in males and +1.20 years in females. (13) The Willems method has, in recent years, been accepted as a more accurate method of age estimation with reports of overestimations of up to +0.55 in males and +0.53 years in females. (14) Reports of underestimations of age with both these methods, although available, are few. In the present study, overestimations of age by +0.24 and +0.09 years for males and by +0.11 and +0.08 years for females, were obtained using the Demirjian and the Willems methods respectively. Studies testing the Chaillet’s multi-ethnic international maturity standards (21) have reported overestimations of +0.28 and +0.37 years in males and +0.09 and +0.21 years in females of Bosnian-Herzegovinian (22) and Spanish Caucasian populations, (23) respectively. Underestimations have been reported of -0.48 and -0.18 years in males and -0.61 and -0.59 years in females of Venezuelan (23) and French (8) populations, respectively. In the present study, underestimations were obtained of -0.12 years in males and -0.45 years in females, with Chaillet’s original standards for Belgian children.
Significant sex-based differences were observed with the Demirjian and Chaillet methods in the present study but not with the Willems method. This sex difference has been attributed to the faster biological and dental maturation in girls, which leads to a higher dental compared to chronological age. (15) However, some other studies (12, 16) have reported a higher dental age compared to chronological age in boys than in girls.
The accuracy or precision of an age estimating method may be affected by the quality of the reference material (sample), reliability of the method and biological variability in dental development. (25, 26) Hence, it is important to accept that no age estimation method can predict the exact age of every individual. In the present study, significant correlations were observed between the chronological and estimated dental ages with all the three methods tested. The Willems method was the most accurate in age estimation, followed by the Demirjian and the Chaillet methods. While the Demirjian method was more accurate than the Chaillet method in females, the Chaillet method better predicted the age of males compared to the Demirjian method. Most studies on non-Indian (7, 27) and Indian (16, 17) populations have found the Willems method to be more accurate than the Demirjian method, while one Indian study (18) reports the reverse.
The mean prediction errors obtained with all three methods for the sample ranged from 1.08 to 3.0 months. Although smaller intervals are desirable, differences between chronological and estimated ages of up to 12 months can be considered to be within normal standards. (28) While the low mean prediction errors suggest that the published standards of the age estimation methods tested could be suitable, gender-specific formulae were derived in the present study, which could increase the accuracy of these methods when applied to Indian populations.
The main limitation of this study would be the use of a convenience sample, which may be subject to sampling bias and may not be representative of the entire population. (29) However, obtaining a random sample was not practical. Further, exposing children to X-radiation for obtaining orthopantomographs when not indicated for diagnosis or treatment may raise ethical concerns. With the use of convenience sampling, we could prevent unnecessary exposure to radiation. By including a large sample of varied age and of both sexes, we have attempted to reduce the sampling bias of a convenience sampling.
CONCLUSIONS
From the results of this study, it was concluded that the Willems and Demirjian methods overestimated age, while the Chaillet method underestimated age. The Willems method most accurately predicted the age (mean prediction error = 1.08 months for both sexes) of the study sample, followed by the Demirjian and Chaillet methods. The Demirjian method was significantly more accurate in girls compared to boys while the reverse was true for the Chaillet method. The Willems method was equally accurate in both sexes. Significant differences between estimated dental age and chronological age were observed with each of the methods. The gender-specific formulae derived in the present study could increase the accuracy of these methods when applied to Indian populations.
Footnotes
The authors declare that they have no conflict of interest.
REFERENCES
- 1.Demirjian A, Goldstein H, Tanner JM. A new system of dental age assessment. Hum Biol. 1973. May;45(2):211–27. [PubMed] [Google Scholar]
- 2.Demirjian A, Goldstein H. New systems for dental maturity based on seven and four teeth. Ann Hum Biol. 1976. Sep;3(5):411–21. 10.1080/03014467600001671 [DOI] [PubMed] [Google Scholar]
- 3.Willems G, Van Olmen A, Spiessens B, Carels C. Dental age estimation in Belgian children: Demirjian’s technique revisited. J Forensic Sci. 2001. Jul;46(4):893–5. 10.1520/JFS15064J [DOI] [PubMed] [Google Scholar]
- 4.Chaillet N, Willems G. Dental maturity in Belgian children using Demirjian’s method and polynomial functions: New standard curves for forensic and clinical use. J Forensic Odontostomatol. 2004;22(2):18–27. [PubMed] [Google Scholar]
- 5.Chaillet N, Nystrom M, Kataja M. Dental maturity curves in Finnish children: Demirjian’s method revisited and polynomial functions for age estimation. J Forensic Sci. 2004. Nov;49(6):1324–31. 10.1520/JFS2004211 [DOI] [PubMed] [Google Scholar]
- 6.Galić I, Nakaš E, Prohić S, Selimović E, Obradović B, Petrovečki M. Dental age estimation among children aged 5–14 years using the Demirjian method in Bosnia-Herzegovina. Acta Stomatol Croat. 2010;44(1):17–25. [Google Scholar]
- 7.Ambarkova V, Galić I, Vodanović M, Biočina-Lukenda D, Brkić H. Dental age estimation using Demirjian and Willems methods: Cross sectional study on children from the Former Yugoslav Republic of Macedonia. Forensic Sci Int. 2014. Jan;234:187.e1–7. 10.1016/j.forsciint.2013.10.024 [DOI] [PubMed] [Google Scholar]
- 8.Urzel V, Bruzek J. Dental age assessment in children: a comparison of four methods in a recent French population. J Forensic Sci. 2013. Sep;58(5):1341–7. 10.1111/1556-4029.12221 [DOI] [PubMed] [Google Scholar]
- 9.Chen JW, Guo J, Zhou J, Liu RK, Chen TT, Zou SJ. Assessment of dental maturity of western Chinese children using Demirjian’s method. Forensic Sci Int. 2010. Apr 15;197(1-3):119.e1–4. 10.1016/j.forsciint.2009.12.009 [DOI] [PubMed] [Google Scholar]
- 10.Leurs IH, Wattel E, Aartman IH, Etty E, Prahl-Andersen B. Dental age in Dutch children. Eur J Orthod. 2005. Jun;27(3):309–14. 10.1093/ejo/cji010 [DOI] [PubMed] [Google Scholar]
- 11.Javadinejad S, Sekhavati H, Ghafari R. A comparison of the accuracy of four age estimation methods based on panoramic radiography of developing teeth. J Dent Res Dent Clin Dent Prospects. 2015. Spring;9(2):72–8. 10.15171/joddd.2015.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Feijóo G, Barbería E, De Nova J, Prieto JL. Permanent teeth development in a Spanish sample. Application to dental age estimation. Forensic Sci Int. 2012. Jan 10;214(1-3):213.e1–6. 10.1016/j.forsciint.2011.08.024 [DOI] [PubMed] [Google Scholar]
- 13.Abu Asab S, Noor SN, Khamis MF. The accuracy of Demirjian method in dental age estimation of Malay children. Singapore Dent J. 2011;32(1):19–27. 10.1016/S0377-5291(12)70012-3 [DOI] [PubMed] [Google Scholar]
- 14.Kumaresan R, Cugati N, Chandrasekaran B, Karthikeyan P. Reliability and validity of five radiographic dental‐age estimation methods in a population of Malaysian children. J Investig Clin Dent. 2016. Feb;7(1):102–9. 10.1111/jicd.12116 [DOI] [PubMed] [Google Scholar]
- 15.Baghdadi ZD. Dental maturity in Saudi children using the Demirjian method: a comparative study and new prediction models. ISRN Dent. 2013;2013:390314. 10.1155/2013/390314 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gupta S, Mehendiratta M, Rehani S, Kumra M, Nagpal R, Gupta R. Age estimation in Indian children and adolescents in the NCR region of Haryana: a comparative study. J Forensic Dent Sci. 2015. Sep-Dec;7(3):253–8. 10.4103/0975-1475.172453 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Patnana AK, Vabbalareddy RS, Vanga NRV. Evaluating the reliability of three different dental age estimation methods in Visakhapatnam children. Int J Clin Pediatr Dent. 2014. Sep-Dec;7(3):186–91. 10.5005/jp-journals-10005-1262 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Mohammed RB, Sanghvi P, Perumalla KK, Srinivasaraju D, Srinivas J, Kalyan US, et al. Accuracy of four dental age estimation methods in southern Indian children. J Clin Diagn Res. 2015. Jan;9(1):HC01–08. Available from http://www.jcdr.net/ReadXMLFile.aspx?id=5495 10.7860/JCDR/2015/10141.5495 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mohammed RB, Krishnamraju PV, Prasanth PS, Sanghvi P, Lata Reddy MA, Jyotsna S. Dental age estimation using Willems method: A digital orthopantomographic study. Contemp Clin Dent. 2014;5(3):371–6. 10.4103/0976-237X.137954 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hegde S, Patodia A, Dixit U. Willems I VS Willems II: A comparative study of accuracy in 5–15 year old Indian children. Forensic Sci Int. 2016. Sep;266:117–22. 10.1016/j.forsciint.2016.05.012 [DOI] [PubMed] [Google Scholar]
- 21.Chaillet N, Nyström M, Demirjian A. Comparison of dental maturity in children of different ethnic origins: international maturity curves for clinicians. J Forensic Sci. 2005. Sep;50(5):1164–74. 10.1520/JFS2005020 [DOI] [PubMed] [Google Scholar]
- 22.Galić I, Vodanović M, Janković S, Mihanović F, Nakaš E, Prohić S. Dental age estimation on Bosnian–Herzegovinian children aged 6–14 years: Evaluation of Chaillet’s international maturity standards. J Forensic Leg Med. 2013;20(1):40–5. 10.1016/j.jflm.2012.04.037 [DOI] [PubMed] [Google Scholar]
- 23.Cruz-Landeira A, Linares-Argote J, Martínez-Rodríguez M, Rodríguez-Calvo MS, Otero XL, Concheiro L. Dental age estimation in Spanish and Venezuelan children. Comparison of Demirjian and Chaillet’s scores. Int J Legal Med. 2010. Mar;124(2):105–12. 10.1007/s00414-009-0380-5 [DOI] [PubMed] [Google Scholar]
- 24.Mohammed RB, Srinivas B, Sanghvi P, Satyanarayana G, Gopalakrishnan M, Pavani BV. Accuracy of Demirjian’s 8 teeth method for age prediction in South Indian children: A comparative study. Contemp Clin Dent. 2015. Jan-Mar;6(1):5–11. 10.4103/0976-237X.149283 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Staaf V, Mörnstad H, Welander U. Age estimation based on tooth development: A test of reliability and validity. Scand J Dent Res. 1991. Aug;99(4):281–6. 10.1111/j.1600-0722.1991.tb01029.x [DOI] [PubMed] [Google Scholar]
- 26.Bagić IČ, Sever N, Brkić H, Kern J. Dental age estimation in children using orthopantomograms. Acta Stomatol Croat. 2008;42(1):11–8. [Google Scholar]
- 27.Maber M, Liversidge H, Hector MP. Accuracy of age estimation of radiographic methods using developing teeth. Forensic Sci Int. 2006. May 15;159 Suppl 1:S68–73. 10.1016/j.forsciint.2006.02.019 [DOI] [PubMed] [Google Scholar]
- 28.Moraes ME, Bastos MS, Santos LR, Castilho JC, Moraes LC, Medici Filho E. Dental age in patients with Down syndrome. Braz Oral Res. 2007. Jul-Sep;21(3):259–64. 10.1590/S1806-83242007000300012 [DOI] [PubMed] [Google Scholar]
- 29.Frey LR, Botan CH, Kreps GL. Investigating communication: An introduction to research methods. 2nd ed. Needham Heights, MA: Allyn & Bacon; 2000. [Google Scholar]