[Downloaded free from http://www.ijdr.in on Wednesday, October 9, 2019, IP: 223.225.44.

218]

Original Research

Accuracy of Demirjian’s and Indian‑Specific Formulae in Age Estimation

Using Eight‑Teeth Method in Kanyakumari Population

Abstract Akhil S,
Background: Most of the scientific formulae for age estimation in forensic odontology were tested Isaac Joseph T1,
among western population and hence cannot be applied to the Indian population consistently. Girish KL1,
Therefore, it was in this context that Dr.  Ashith B. Acharya had carried out a study using the
modified Demirjian’s method in Indian population and found out that the study gave inferior results Pradeesh Sathyan1
for age estimation. So he developed Indian‑specific regression analysis and worked out a formula. Department of Oral Pathology,
Aim: This study was done to validate age using Demirjian’s eight‑teeth method and to compare Malabar Dental College and
Research Centre, Edappal,
the effectiveness of Demirjian’s formula and Indian‑specific formula in Kanyakumari population. Malappuram, Kerala,
Material and Methods: Digital orthopantomographs of 150  patients fulfilling the inclusion and 1
Department of Oral Pathology,
exclusion criteria in the age group of 8–24  years were used in the study. The third quadrant in the Sree Mookambika Institute of
radiograph was assessed visually from mandibular central incisor to the third molar using Demirjian’s Dental Sciences, Kulasekharam,
modified criteria chart. Calculation of the dental age was done using Demirjian’s formula and Ashith Kanyakumari, Tamil Nadu, India
B. Acharya’s Indian‑specific formula. The difference between chronological age and dental age was
calculated, and the mean absolute error (MAE) was obtained. Results: The MAE was 0.20 years for
the whole of Kanyakumari population, and for males it was 0.10  years and for females 0.29  years
with Indian‑specific formula, whereas the MAE was 2.66, 1.86, and 3.51 years, respectively, for the
whole of Kanyakumari population, males, and females using Demirjian’s formula. Conclusion: The
observations from this study suggest that the MAE was less between chronological age and estimated
dental age which was calculated using Indian‑specific formula, compared with the values obtained
using Demirjian’s formula. Thereby we conclude that Indian‑specific formula is more reliable in age
estimation of Kanyakumari population.

Keywords: Age estimation, Demirjian’s method, mean absolute error

Introduction In the present scenario, most of the age

estimation techniques are invasive, requiring
Age estimation forms one of the most
lengthy processing time, use of expensive
important subdisciplines of forensic
instruments, and the services of an expert
sciences and is of paramount importance
pathologist to deduce the age of a person.
in medicolegal issues. It plays an
But the biggest pitfall had been the lack of
important role in several dental spheres.
the usefulness of these methods in vivo. It
In forensic dentistry, there is always a
is in this case that the branch of radiology
need to determine the age of unidentified
comes handy as it gives an insight into the
skeletons or individuals who have no
developmental stages of the teeth, which Address for correspondence:
record/documentation of their chronological
provides a baseline data for age estimation Dr. Girish KL,
age.[1‑3] Various biological age estimation
in children and adolescents.[5] Department of Oral
methods have been developed and they Pathology, Sree Mookambika
were based on skeletal, morphological, There are various radiographic methods Institute of Dental Sciences,
secondary sexual character, and dental available for age estimation. But the most Kulasekharam ‑ 629 161,
Kanyakumari,
findings. The aim of determining age is to commonly used one is the Demirjian’s Tamil Nadu, India.
give the forensic odontologist an accurate method which is based on French‑Canadian E‑mail: [email protected]
age range within a biological profile. population. This is mainly because of
To provide the greatest possible value, the simplicity of the method, as well as
the method used should have the lowest radiographic and schematic illustrations of Access this article online

achievable standard deviation (SD).[4] the tooth development and accompanying Website: www.ijdr.in
description, which the original and the DOI: 10.4103/ijdr.IJDR_768_17
subsequent works provided. This method Quick Response Code:
This is an open access journal, and articles are distributed under
the terms of the Creative Commons Attribution-NonCommercial-
ShareAlike 4.0 License, which allows others to remix, tweak, and
build upon the work non-commercially, as long as appropriate How to cite this article: Akhil S, Joseph TI,
credit is given and the new creations are licensed under the Girish KL, Sathyan P. Accuracy of Demirjian's and
identical terms. Indian-specific formulae in age estimation using
eight-teeth method in Kanyakumari population. Indian
For reprints contact: [email protected] J Dent Res 2019;30:352-7.

352 © 2019 Indian Journal of Dental Research | Published by Wolters Kluwer - Medknow
[Downloaded free from http://www.ijdr.in on Wednesday, October 9, 2019, IP: 223.225.44.218]

Akhil, et al.: Dental age estimation in Kanyakumari population

uses orthopantomographs to estimate the extent of In this study, the third quadrant in the radiograph was
mineralized dental tissues and the shape of the chamber of assessed visually from 31 to 38 using Demirjian’s modified
seven left permanent lower teeth. This method classified criteria chart, which included 10 developmental stages
the development of teeth into eight stages and arrived at an instead of 8. The stages were then entered into a separate
age estimation method.[4,6] scoring proforma following which the sex‑specific maturity
score for each tooth was entered depending on the scoring
The original method used only the seven mandibular teeth
grades developed by Demirjian. Calculation of dental age
on the left side and assigned a gender‑specific maturity
using the mandibular third quadrant teeth was carried out
score to each tooth. The scores were summed up and
separately using Demirjian’s[9] and Ashith B. Acharya’s
compared with the centile charts to arrive at the estimated
Indian‑specific formulae.[11] The value so obtained was
age.[4] The original Demirjian’s method excluded the third
designated as the calculated dental age.
molar because of variability in its development, eruption,
and anatomy. But the third molar provides the only reliable The difference between the chronological age and dental
radiological parameter for age estimation in the age group age was calculated and the mean absolute error  (MAE)
of 16–23  years. Chaillet and Demirjian in 2004 proposed was obtained. MAE is considered as a standard measure
changes to the original method by incorporating the third to estimate the effectiveness of the methods in age
molar and developed a new maturity score based on a estimation studies. The accuracy of age prediction is
French population.[7‑9] usually represented by the MAE, which is calculated as
the difference between the calculated age and the actual
Most of the scientific formulae for age estimation in forensic
age at the time of exposure. The number of values that
odontology were tested among the western population
fell in the error group of <±1  year, within 1.1–2  years,
and hence cannot be applied to the Indian population
and >±2 years was noted. Error of <±1 year is considered
consistently. Hence, it was in this context that Dr.  Ashith
a good result and error rate of >±2  years is considered as
Acharya had carried out a study using the modified
inaccurate.[11,12]
Demirjian’s method in the Indian population and found
out that the study gave inferior results for age estimation. Statistical analysis
So he developed Indian‑specific regression analysis and
The mean and SD were calculated using the descriptive
worked out a formula.[10,11] This study was done to validate
statistics. Data were analyzed by Statistical Package for
age using Demirjian’s eight‑teeth method and to compare
Social Services  (SPSS 20.0) trial version. Normality of
the effectiveness of Demirjian’s formula and Dr.  Ashith
the data was assessed using Kolmogorov–Smirnov test.
Acharya’s Indian‑specific formula in Kanyakumari
Data were found to be normally distributed, and hence,
population.
“paired‑t” test was used to find statistical significance. The
Materials and Methods limit of statistical significance was set as 0.05 (P < 0.05).

The study sample consisted of digital orthopantomographs Results

of patients, which were taken from the archives of
The dental age of each individual was calculated by
Departments of Oral Medicine and Radiology and
evaluating the panoramic radiographs using Demirjian’s
Orthodontics. The sample size was calculated to be 132
eight‑teeth method, Demirjian’s formula, and Indian‑specific
and approximated to 150 with the power of 80%, and data
formula developed by Ashith B. Acharya. Accordingly,
for the sample size calculation were taken from the study
the mean dental age of males was 15.61  years and that of
conducted by Sarkar SS et al. in 2013. The study group
females was 17.47  years; 16.54  years was the calculated
comprised 150 digital panoramic radiographs of patients
mean dental age for the total sample size according to
within the age group of 8–24  years, equally distributed
Indian‑specific formula; whereas using Demirjian’s method,
between the sexes and which fulfilled the inclusion and
the mean dental age of males, females, and that of total
exclusion criteria [Table 1].
sample was 13.85,14.25, and 14.08 years, respectively. The
All the radiographs taken for routine diagnostic purpose interquartile range of chronological age for study sample
and orthodontic treatment were considered. The was 6, whereas it was 5 and 5.63 for dental age calculated
subjects selected for the study were contacted to verify by Demirjian’s method and Indian‑specific formula,
chronological age, and informed consent was obtained. respectively.

Table 1: Exclusion criteria

OPG of patients with caries/periapical pathology
OPG of patients with developmental anomalies
OPG of patients with crowding of teeth where the root structures of the teeth were not clearly discernible
OPG of patients with bilaterally missing mandibular teeth
OPG: Orthopantomograph

Indian Journal of Dental Research | Volume 30 | Issue 3 | May-June 2019  353

[Downloaded free from http://www.ijdr.in on Wednesday, October 9, 2019, IP: 223.225.44.218]

Akhil, et al.: Dental age estimation in Kanyakumari population

The effectiveness of age estimation by the said criteria, The original Demirjian’s method using the seven mandibular
Demirjian’s method, and Indian‑specific formula was teeth had a high accuracy but poor reliability.[5] Subsequently,
compared in terms of MAE between estimated and actual Chaillet and Demirjian modified their method to incorporate
age, and the number of age estimates that were either the third molar and developed a new maturity score based
<±1, 1.1–2, and >±2  years, respectively. The MAE which on a French population.[9] Ashith B. Acharya in his study
was calculated by finding out the difference between used the eight‑teeth method in 547 Indians  (199  males and
chronological age and calculated dental age was 0.20 years 348  females) age 7–25  years discussed the effectiveness
for the whole of Kanyakumari population. The MAE for of original formula and Indian‑specific formula. The study
males was 0.10  years and that for females was 0.29  years showed that the cubic functions by Chaillet and Demirjian
using Indian‑specific formula, and using Demirjian’s provided the best fit for French population and same cubic
formula the MAE was 1.86  years for males, 3.51  years functions misclassified the Indian sample. Regression
for females, and 2.66  years for the whole of Kanyakumari analysis was performed considering the variations for the
population. The number of age estimates that were either Indian sample to derive Indian‑specific formula.[11]
<±1, 1.1–2, and >±2  years were calculated which gave the
error rate and accuracy of the results [Table 2]. In this study, the dental age was evaluated using modified
Demirjian’s eight‑teeth method, Demirjian’s formula, and
All the values calculated were expressed in terms of Indian‑specific formula developed by Ashith B. Acharya
mean  ±  SD. The results obtained were tabulated and using 150 digital orthopantomographs [Figures 1 and 2].
subjected to statistical analysis. Statistical analysis did not The sample size was equally distributed among both males
yield any significant statistical variation  (p  >  0.05) using and females. Accordingly, using Demirjian’s formula, the
Indian‑specific formula in all the samples. When Demirjian’s mean dental age of males, females, and of the total sample
formula results were subjected to statistical analysis, there size was 13.84, 14.24, and 14.04 years, respectively. Using
was a significant difference  (P  <  0.05) between calculated Indian‑specific formula, the mean dental age of males
age and chronological age in all samples [Table 3]. was 15.61  years and that of females was 17.47  years;
16.54  years was the calculated mean dental age for the
Discussion total sample size [Graphs 1 and 2].
Forensic odontologists often face the problem of
The effectiveness of age estimation by Demirjian’s
determining the age of unknown bodies and also living
formula and Indian‑specific formula was compared in
persons. Tooth is considered as a dependable body part for
terms of MAE between estimated and actual age.[11] The
forensic age estimation since environmental factors have
study revealed that the MAE in Demirjian’s formula
little influence on it.[13] Various methods for age estimation
which was calculated by finding out the difference
using teeth were formulated through the years. Among the
between chronological age and calculated dental age
different methods used for age determination in individuals,
was 2.66  years for the whole of study population
the radiological method has got a number of advantages
which was greater than studies conducted by Acharya
over histological, biochemical, and genetic methods.[14]
AB  (1.29  years),[11] Sarkar S et al.  (1.57  years),[15] and
Mohammed RB et al.  (1.61  years).[16] The MAE calculated
Table 2: Distribution of percentage (number) of samples by Indian‑specific formula was found to be 0.20  years
under different error rates for the whole of study population which was much lesser
Method Groups Error rate than studies conducted by Kumar VJ et al.  (1.18  years),[5]
˂±1 1.1‑2 ˃±2 Acharya AB  (0.87  years),[11] Sarkar S et al.  (0.94  years),[15]
Demirjian’s Male 37.3 (28) 24 (18) 38.7 (29) and Mohammed RB et al. (0.53 years).[16] The values in our
formulae Female 14.7 (11) 13.3 (10) 72 (54) study were much closer to the actual age compared with
Whole sample 26 (39) 18.7 (28) 55.3 (83) the studies done by various other authors, which could be
Indian‑specific Male 52 (39) 28 (21) 20 (15) as a result of careful selection of sample and only one‑fifth
formulae Female 52 (39) 30.7 (23) 17.3 (13) of the sample size above the age group of 20  years where
Whole sample 52 (78) 29.3 (44) 18.7 (28) maximum variations are usually seen.

Table 3: Statistical analysis between the two formulae

Method Groups Chronological age Dental age Mean absolute error P‑value
Demirjian’s formulae Male 15.71±3.548 13.85±2.111 1.86 0.000
Female 17.76±3.928 14.25±2.036 3.51 0.000
Kanyakumari population 16.73±3.870 14.08±2.076 2.66 0.000
Indian‑specific formulae Male 15.71±3.548 15.61±3.029 0.10 0.538
Female 17.76±3.928 17.47±3.296 0.29 0.814
Kanyakumari population 16.73±3.870 16.53±3.289 0.20 0.538

354 Indian Journal of Dental Research | Volume 30 | Issue 3 | May-June 2019

[Downloaded free from http://www.ijdr.in on Wednesday, October 9, 2019, IP: 223.225.44.218]

Akhil, et al.: Dental age estimation in Kanyakumari population

Figure  1: Orthopantomograph of an 11‑year‑old male Figure 2: Orthopantomograph of a 17‑year‑old female

Graph 1: Comparison of mean of chronological and dental age in Indian Graph 2: Comparison of mean of chronological and dental age in Demirjian’s
specific formula formula

In this study, the MAE which was found out using Sarkar S et al.  (1.54  years),[15] Mohammed RB
Demirjian’s formula for males was 1.86  years which et al.  (1.55  years),[16] and Kiran CS et al.  (0.83  years).[17]
was lesser than the MAE of the whole population of This was because more number of samples was present
the same study and greater than studies conducted by in the age group above 20  years in females where the
various other authors like Acharya AB  (0.94  years),[11] maximum number of variations is present.
Kiran CS et al.  (0.89  years),[17] whereas studies done The MAE for females in our study using Indian‑specific
by Sarkar S et al.  (1.63  years)[15] and Mohammed RB formula was 0.29  years and it was much less than the
et al.  (1.66  years)[16] showed an MAE equal to our study. studies conducted by various other authors like Kumar  VJ
Using Indian‑specific formulae, the MAE for males et al.  (1.05  years),[5] Acharya AB  (0.99  years),[11] Sarkar S
was 0.10  years which was much lesser than the MAE et al. (0.94 years),[15] Mohammed RB et al. (0.85 years),[16]
of the whole population of the same study and studies Kiran CS et al.  (1.91  years),[17] and Tandon
conducted by various other authors like Kumar VJ et al. (0.87 years).[18]
et al.  (1.2  years),[5] Acharya AB  (0.70  years),[11] Kiran CS
et al. (1.72 years),[17] and Tandon et al. (0.85 years).[18] The The MAE is a quantity used to measure how close
MAE for males in our study was equal to that of studies predictions are to the eventual outcomes. It is the mean
done by Sarkar S et al. (0.10 years)[15] and Mohammed RB error irrespective of positive or negative sign.[17] The MAE
has been advocated by a number of authors as a measure
et al.  (0.21  years).[16] Thus, age estimation of males using
to quantify methods’ accuracy. Errors of <±1  year have
Indian‑specific formula in our study was more accurate
been considered as “good results,” and errors >±2  years
than that of the whole population in the same study and the
have been designated as “inaccurate.”[11] Accordingly, the
studies by various authors.
age estimates that were either <±1, 1.1–2, and >±2  years
Using Demirjian’s formula, the MAE for females was were calculated and this gave the error rate and accuracy
found to be 3.51  years. When compared, it was more than of the results. The MAE was <±1  years in 37.33%
the MAE in males and that of the whole population of the of males, 14.67% of females, and 26% of the whole
same study using Demirjian’s method. The MAE of females population using Demirjian’s formula; whereas using
of our study was much more than the studies conducted Indian‑specific formulae, 52% of males, 52% of females,
by various other authors like Acharya AB  (1.55  years),[11] and 52% of whole population had an MAE <±1  year.

Indian Journal of Dental Research | Volume 30 | Issue 3 | May-June 2019  355

[Downloaded free from http://www.ijdr.in on Wednesday, October 9, 2019, IP: 223.225.44.218]

Akhil, et al.: Dental age estimation in Kanyakumari population

Graph 4: Number of age estimates in different error groups using

Graph 3: Number of age estimates in different error groups using Indian
Demirjian’s formula
specific formula

About 18% of males, 13.3% of females, and 18.7% of the Acknowledgement

whole sample size had MAE in the range of 1.1–2  years, The authors thank the Departments of Oral Medicine and
respectively, using Demirjian’s formula, whereas 28% of Orthodontics, Sree Mookambika Institute of Dental Sciences,
males, 30.7% of females, and 29.3% of whole sample size Kulasekharam.
were found in the same age range using Indian‑specific Financial support and sponsorship
formula [Graphs 3 and 4].
Nil.
The error rate of >±2  years was established in 38.7% of
males, 72% of females, and 55.3% of the whole population Conflicts of interest
using Demirjian’s formula. About 20% of males, 17.3% There are no conflicts of interest.
of females, and 18.7% of the whole population had error
rate of >±2  years using Indian‑specific formula. The References
error rate of <±1 and 1.1–2  years in our study was in 1. Mani  S, Naing  L, John  J, Samsudin  A. Comparison of two
accordance to studies conducted by Ashith B. Acharya and methods of dental age estimation in 7‑15  year‑old Malays. Int J
Kumar VJ et al. The error rate of >±2  years was higher Paediatr Dent 2008;18:380‑8.
in our study than that of Acharya  (7%)[11] and Kumar VJ 2. Chaillet  N, Nyström M, Demirjian  A. Comparison of dental
et al  (10.75%)[8] using both formulae. The higher error maturity in children of different ethnic origins: International
maturity curves for clinicians. J Forensic Sci 2005;50:1164‑74.
rate of >±2  years in our study could be probably due to
3. Willems  G, Van Olmen  A, Spiessens  B, Carels  C. Dental age
the higher sample size in age group of above 20  years, estimation in Belgian children: Demirjian’s technique revisited.
where there is always an underestimation of age by 7.5 and J Forensic Sci 2001;46:893‑5.
7.85  years by Demirjian’s formula and 4.5 and 3.25  years 4. Demirjian A, Goldstein H, Tanner J. A new system of dental age
by Indian‑specific formula, respectively, for males and assessment. Hum Biol 1973;45:211‑21.
females. The results clearly show that Indian‑specific 5. Kumar  JV, Gopal  SK. Reliability of age estimation using
formula had more number of values under error rate of Demirjian’s 8 teeth method and Indian specific formula.
J Forensic Dent Sci 2011;3:19‑22.
<±1 year when compared with Demirjian’s formula making
6. Hegde  RJ, Khare  SS, Saraf  TA, Trivedi  S, Naidu  S. Evaluation
it more reliable. This suggests that the variation is not of the accuracy of Demirjian method for estimation of dental age
significant and that Demirjian’s eight‑teeth method along among 6‑12  years of children in Navi Mumbai: A  radiographic
with Ashith B. Acharya’s Indian‑specific formula is suitable study. J Indian Soc Pedod Prev Dent 2015;33:319‑23.
for estimation of age in Kanyakumari population. 7. Willens  G. A  review of the most commonly used dental age
estimation techniques. J Forensic Odonto Stomatol 2001;19:9‑17.
Conclusion 8. Mincer  HH, Harris  EF, Berryman  HE. The ABFO study of third
molar development and its use as an estimator of chronological
In conclusion, it is proved that age estimation using age. J Forensic Sci 1993;38:379‑90.
Demirjian’s formula resulted in an inferior age prediction 9. Chaillet  N, Demirjian  A. Dental maturity in South France:
of Kanyakumari population, whereas there was very less A  comparison between Demirjian’s method and polynomial
difference between estimated dental age and chronological functions. J Forensic Sci 2004;49:1059‑66.
age calculated using Indian‑specific formula. Age estimation 10. Koshy  S, Tandon  S. Dental age assessment: The applicability
of the assessment in south Indian children. Forensic Sci Int
in males was more accurate than that of the whole population
1998;94:73‑85.
using Indian‑specific formula. Further studies with larger 11. Acharya AB. Age estimation in Indians using Demirjian’s 8‑teeth
sample size are required to validate whether Indian‑specific method. J Forensic Sci 2011;56:124‑7.
formula can be used to find out age in different populations in 12. Solari  AC, Abramovitch  K. The accuracy of age and precision
India since India is a vast land with distinct ethnic diversity. of third molar development as an indicator of chronologic age in

356 Indian Journal of Dental Research | Volume 30 | Issue 3 | May-June 2019

[Downloaded free from http://www.ijdr.in on Wednesday, October 9, 2019, IP: 223.225.44.218]

Akhil, et al.: Dental age estimation in Kanyakumari population

Hispanics. J Forensic Sci 2002;47:531‑5. 16. Mohammed  RB, Srinivas  B, Sanghvi  P, Satyanarayana  G,
13. Mathew  DG, Rajesh  S, Koshi  E, Priya  LE, Nair  AS, Gopalakrishnan  M, Pavani  BV. Accuracy of Demirjian’s 8
Mohan  A. Adult forensic age estimation using mandibular first teeth method for age prediction in South Indian children:
molar radiographs: A  novel technique. J  Forensic Dent Sci A comparative study. Contemp Clin Dent 2015;6:5‑11.
2013;5:56‑9. 17. Kiran  CS, Reddy  RS, Ramesh  T, Madhavi  NS, Ramya  K.
14. Acharya  AB, Sivapathasundharan  B. Forensic odontology. In: Radiographic evaluation of dental age using Demirjian’s
Shafer’s textbook of oral pathology. 7th  edn. India: Elsevier eight‑teeth method and its comparison with Indian formulas in
Private Ltd; 2012. p. 879‑907. South Indian population. J Forensic Dent Sci 2015;7:44‑8.
15. Sarkar  SS, Kailasam  SB, Kumar  PM. Accuracy of estimation 18. Tandon  A, Agarwal  V, Arora  V. Reliability of India‑specific
of dental age in comparison with chronological age in Indian regression formula for age estimation of population in and
population: A  comparative analysis of two formulas. J  Forensic around Bahadurgarh, Haryana (India). J Oral Biol Craniofac Res
Leg Med 2013;20:230‑3. 2015;5:193‑7.

Indian Journal of Dental Research | Volume 30 | Issue 3 | May-June 2019  357