- Email: [email protected]
A cut-off value of third molar maturity index for indicating a minimum age of criminal responsibility: Older or younger than 16 years?
Journal of Forensic and Legal Medicine 65 (2019) 108–112
Contents lists available at ScienceDirect
Journal of Forensic and Legal Medicine journal homepage: www.elsevier.com/locate/yjflm
Research Paper
A cut-off value of third molar maturity index for indicating a minimum age of criminal responsibility: Older or younger than 16 years?
T
Sudheer B. Ballaa,∗, Subramanyeswara S. Chinnib, Ivan Galicc, Aditya Mohan Alwalad, Pramod Machanie, Roberto Camerieref a
Department of Forensic Odontology, Panineeya Mahavidyalaya Institute of Dental Sciences, Hyderabad, Telangana, India Consultant Forensic Odontologist, Bangalore, Karnataka, India c Departments of Research in Biomedicine and Health, University of Split School of Medicine, Spinciceva 2, 21000, Split, Croatia d Department of Oral and Maxillofacial Surgery, MNR Dental College, Sangareddy, Telangana, India e Nightingales Regional Lead, Bangalore, Karnataka, India f Age Estimation Project, Institute of Legal Medicine, University of Macerata, Macerata, Italy b
A R T I C LE I N FO
A B S T R A C T
Keywords: Dental age estimation Minimum age of criminal responsibility Third molar maturity index Orthopantomograms 16 years
Providing appropriate legal mechanisms, that evaluate the progression of development from the age of childhood innocence to maturity and full responsibility, considered one of the difficult areas of criminal justice policy. The minimum age of criminal responsibility (MACR) in children varies among countries and differs widely owing to history and culture. Due to rising and brutality of criminal offenses, particularly by juveniles, few countries have lowered the minimum age of criminal responsibility, and many have considered/considering to do the same. India is one such country in which is under the proposal of lowering the age of criminal responsibility to 16 years. As there is lack of useful age assessment methods, that can indicate whether if the individual in question is younger (< 16 years) or older than MACR (≥16 years), the present study was undertaken to derive a specific cut-off value of the third molar maturity index (I3M) for this purpose. The sample consisted of 1078 orthopantomograms (OPTs) from Andhra Pradesh, India, aged between 11 and 20 years (45.4% males and 54.6% females). The reproducibility of I3M was calculated by intra-class correlation coefficients, which showed an intra-rater and inter-rater agreement of 0.912 and 0.891, respectively. The sample was divided into training dataset (819 OPTs), to test I3M and gender as independent variables and MACR as a dependent variable by logistic regression analysis, and test dataset (259 OPTs). A receiver operating curve (ROC) analysis was performed to evaluate the specific cut-off value of I3M for predicting MACR status. A logistic regression analysis showed that gender was not statistically significant for predicting MACR status while ROC analysis indicated a cut-off value of I3M = 0.293 as best for predicting MACR status. The performance of derived cut-off value was tested in a test data set. The sensitivity of the test was 90.6% and 90%, while specificity was 86% and 87.1% in males and females, respectively. The proportion of correctly classified individuals was 88.0% and 88.7% in males and females, respectively. The estimated Bayes post-test probability in males was 87.2% and while in females it was 88.3%. The results highlight the contribution of the derived cut-off value of I3M for discriminating individuals around MACR and should be evaluated in other populations.
1. Introduction Assessing dental age is not just limited to the prediction of growth in children for diagnosis and treatment planning, but can also be used to resolve issues related to prosecution in the criminal and civil courts.1 At times, forensic experts needed to answer question regards juvenile's attainment of the minimum age of criminal responsibility (MACR) and age of majority. The former is considered one of the most sensitive areas
∗
of criminal justice policy as it is vital to apply appropriate legal mechanisms that reflect the transition from the age of childhood innocence through to maturity and full responsibility under the criminal law.2 In recent times, Juvenile delinquency has attracted much legislative debate and judicial intervention.3 It refers to the antisocial or the criminal activity of the child which violates the law and often considered to be a gateway to adult crime since a large percentage of criminal careers has their roots in childhood. The legal difference between a criminal and a
Corresponding author. E-mail address: [email protected] (S.B. Balla).
https://doi.org/10.1016/j.jflm.2019.05.014 Received 23 March 2018; Accepted 19 May 2019 Available online 22 May 2019 1752-928X/ © 2019 Elsevier Ltd and Faculty of Forensic and Legal Medicine. All rights reserved.
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
12–14 years. Since third molars are the only teeth available growing at the age of 18 years, their development was considered most suitable evidence for juvenile/adult discrimination. Mincer et al.19 were the first to verify the usefulness of Demirjian's stages (DS) and third molars for the discrimination of juvenile/adult status. Later Cameriere et al.20 have demonstrated the better performance of the specific value of third molar maturity index (I3M), I3M < 0.08, in discriminating adults or minors compared to DS system. The applicability of latter in predicting the age of majority was tested successfully in various populations.5,21–30 It is vital to adopt accurate age estimation methods prior considering steps to prevent and treat delinquency. Moreover, recent trends in India posed a task to forensic experts and pointed out the lack of useful age estimation methods that can indicate whether if the individual in question is younger or older than MACR (16 years). Currently, there are no methods available in practice or in literature to indicate MACR among Indian individuals. This study was undertaken to derive a cut-off value of I3M which can be useful for predicting being younger (< 16 years) or older than MACR (≥16 years).
delinquent is a matter of a day on the calendar as if one commits an illegal act a day before he ceases to be a juvenile is considered a delinquent act, otherwise as a criminal offense. Here the unlawful act is the same, but the perpetrator of the act is a day older, and therefore, a criminal.4 1.1. MACR and judicial trend in India Article 40.3 of the Convention on the rights of the child defines MACR as “the age below which a person is completely immune from any criminal liability due to lack of maturity and judgment to understand the consequences of one's actions”.5 In practical terms, a separate justice system for children that has rehabilitation and reintegration should be established. In India, the criminal system is governed and regulated by two major legislations including the Indian Penal Code, 1860 (IPC) and the Criminal Procedure Code, 1970 (CrPC). The IPC has set the minimum age of criminal responsibility as 12 years.6 According to Juvenile Justice (Care and protection of children) Act, 2000, children cannot be put into the same category as adults and hence required special provisions for them. This act has set the age of criminal responsibility at 18 years that concurs definition of child under the UN convention on the rights of the child.6,7 Juvenile offenders are accorded special treatment under this act and are not tried in the same courts as adult offenders. In extreme cases under Juvenile Justice Act, the offenders will be sent to detention for a maximum period of three years in reformatory centers. In India, there has been a sharp rise in crimes committed by juveniles or young adults. According to the statistics of National Crime Records Bureau (NCRB), 27% of the juvenile crimes in 2015 were perpetrated by the individuals between ages 12–16 years, 1.4% by those between 7 and 12 years, and 71.6% by those between 16 and 18 years.8 Owing to the rise and brutality of criminal offenses particularly by juveniles, there was an increasing sense of urgency to create legal avenues for some deterrence to warn off the under-age perpetrators.9
2. Materials and methods A final sample of panoramic radiographs (OPTs) of 1078 living subjects was analyzed retrospectively (Table 1). OPTs used in this study were initialized for different clinical purposes on patients from the Panineeya Institute of Dental Sciences, Hyderabad, Telangana and from private dental clinics, Andhra Pradesh, India. Ethics Committee for research involving human subjects approved the usage of OPTs in this study. Each OPT was separately recorded in Excel file. Record comprised an identification number, gender, date of birth, date of OPT. Real age in years was calculated as the difference between dates of OPT and birth. The inclusion criteria for retrieving the sample for the analysis were: OPT of the good technical quality, age between 11 and 20 years, no evidence of hereditary or systematic diseases which can affect the dental and skeletal development, including hormonal deficiency or hormonal therapy, gastrointestinal diseases. Exclusion criteria are any missing or incomplete data on birth, agenesis of any permanent teeth including all four third molars, extracted second and third molars, rotated or impacted third molars. All OPTs were recorded in JPG format. Only third molars from the left side of the mandible were evaluated according to Cameriere's method. A freeware image-processing program (Adobe® Photoshop® CS4) was used for all adjustment of a grayscale and brightness and contrast improvement. Briefly, the measurements of the projections of open apices, Ai, and height, Li, of the third molars were recorded and I3M was calculated as the proportion of open apices and heights or
1.2. Raising concerns As per NCRB statistics, in 2010, the toll of rape cases by juveniles stood at 858, it became 1149 in 2011, 1316 in 2012, which further increased up to 1388 in 2013.9 Moreover, in last ten years, heinous crimes by juveniles have increased four to five times. One such incident is frightening gang-rape- cum-murder in Delhi in which one of accused is a minor who allegedly committed the most brutal act has shifted the spotlight on juvenile offenders, their age bar, nature of crime and brutality involved.10 Keeping in mind the rise of offenses by minors, there is a need for the change of the current MACR (18 years) and to make special provisions for serious offenses. Following this, there are serious issues raised by law, regarding amendments to the criminal law and MACR. This issue has reached the apex court of India, asking for complete struck off the Act of 2000 and to make changes in various provisions to enhanced punishment to a juvenile in conflict with the law.11 Later the government considered a proposal that under 18 years accused of heinous crimes are tried as adults and passed a bill redefining a juvenile's age in these offenses.12,13
Table 1 Age and Sex distribution of the data (training and test datasets) and the parentheses in training dataset represents the number of participants with closed apices of left second mandibular molar.
1.3. Need of the hour and need for the study In specific legal requirements, usage of low-invasive methods with higher accuracy and precision are required to assign an age to a living child without legal documents.14,15 To further improve the diagnostic accuracy of age estimates, the Study Group on Forensic Age Diagnostics (AGFAD) recommended usage of methods in combination that includes a physical examination with determination of anthropometric measurements, examination of skeletal development that includes handwrist bones and evaluation of dental status by dental exams.16–18 The mineralization of the second molars completes between the age of 109
Age (years)
Malesb training dataset
Females training dataset
Total training dataset
Males test dataset
Females test dataset
Total test dataset
11 12 13 14 15 16 17 18 19 20
26 35 45 57 54 31 41 33 31 20
39 50 47 53 62 42 41 39 51 22
65 (0) 85 (6) 92 (29) 110 (62) 116 (98) 73 (70) 82 (81) 72 (72) 82 (82) 42 (42)
10 11 10 15 17 13 11 10 11 9
10 10 9 17 18 18 15 17 10 18
20 21 19 32 35 31 26 27 21 27
Total
373
819
117
142
259
(0) (2) (15) (24) (43) (30) (41) (33) (31) (20)
(0) (4) (14) (38) (55) (40) (40) (39) (51) (22)
446
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
I3M = [Ai/Li]. I3M is evaluated in a similar way to the ratio Ai to Li, when i = 6 or 7, as reported for the first and second lower molars, as described in Cameriere et al.20 Additionally, if the development of the root of TM completed, then I3M = 0.0 was recorded. Pearson correlation coefficient was performed to test the correlation between I3M and real age. The first author performed the measures of the sample. The analyses were performed using a blind approach, and the observer was not able to detect identification number, gender or age of the subjects. The first author and one other observer additionally evaluated randomly selected 50 OPTs two weeks after finished the first measurements. The intraclass correlation coefficient (ICC) was calculated to assess intra-rater and inter-rater agreement of I3M.31 A sample of OPTs was randomly divided into training dataset (819) and test dataset (259). A logistic regression linear model was tested on the training dataset with dependent variables E = 1 and E = 0 to verify the significance of the independent variables I3M and gender on the binary answer whether a subject is ≥ 16 years (E = 1) or < 16 years (E = 0). The dummy variable gender equals to 1 for boys and 0 for girls. The predictive accuracy of the specific cut-off value of I3M to test whether a subject is ≥ 16 years or < 16 years was established by determining the receiver operating characteristic (ROC) analysis. The area under the curve (AUC) and 95% confidence interval (95% CI) was calculated with a non-parametric method.20,32,33 If the cut-off is closer to the apex of the ROC curve to the upper left corner, it is the greater accuracy of the test.33 The best performance of the discrimination test was established as the maximum value of Youden's index (J). The greatest Youden's index was used to determine a specific cut-off value of I3M.34 The performance of the established cut-off value of I3M was evaluated by the two-by-two contingency table (CT) on the test dataset of OPTs. Precisely, CT in the first row lists the results of those who are ≥16 years and have I3M less than cut-off value (true positives, TP), following those who are < 16 years and have I3M less than cut-off value (false positives, FP). In the second row list those who are ≥16 years and have I3M ≥ than cut-off value (false negatives, FN), and finally, those who are < 16 and have I3M ≥ cut-off value (true negatives, TN). The percentage of accurate classification (Ac), sensitivity (Se) or the percentage of the subjects ≥16 years who have I3M < cutoff value, and the specificity (Sp) or the percentage of individuals younger than 16 who have I3M ≥ cut-off value, were calculated. The positive predictive value (PPV) and the negative predictive value (NPV) or the proportions of positive and negative results that are truly positive and truly negative results, respectively were calculated.35 Also, the positive likelihood ratio (LR+) and negative likelihood ratio (LR-) were also calculated to express how many times more or less likely a test result is to be found in (≥16 years) compared to (< 16 years) participants.35 The post-test probability p of being 16 years of age or older can help to discriminate between those individuals who are 16 and over and under 16. According to Bayes theorem, the post-test probability may be written as:
p=
Se × p0 Se × p0 + (1 − Sp) × (1 − p0 )
Fig. 1. Scatter-plot of the relationship between age (years) and the third molar maturity index.
p < 0.05. 3. Results The final sample consisted of 1078 OPTs. Table 1 demonstrates the distribution of the training and test datasets according to gender and age. The number of subjects with closed apices of second molars in each age cohort was presented in parenthesis of training dataset of same table. Closed apices were first noticed in both boys and girls at 12 years age. None of the apices were open after 16 years of age in both genders. Pearson correlation coefficients for the correlation between I3M and real age were better for females, 0.780 (p < 0.001), than for males, 0.778 (p < 0.001), Fig. 1. The intra-class correlation coefficient for the intrarater agreement was 0.912 while for the inter-rater agreement was 0.891. Obtained results of ICC showed excellent results both on the precision and the repeatability of the variable I3M. The logistic model showed the significance of variable I3M (p < 0.001) while gender was not statistically significant (p = 0.053), Table 2. The full model containing I3M as a predictor was statistically significant, χ2 (1, 819) = 556.3, p < 0.001, indicating that model was able to discriminate between those who are ≥16 years and < 16 years. The whole model explained between 0.493 (Cox & Snell R Square) and 0.662 (Nagelkerke R Square) of the variance in ≥16 years and < 16 statuses. A linear logistic model may be written as: Logit (p) = 2.086–5.102 × I3M
(2)
The value for AUC was 0.936 (95% CI, 0.920 to 0.953). A cut-off value of I3M, to discriminate that individuals are ≥16 years or < 16 years, was established for the highest value of the Youden index.35 The highest Youden index was 0.739 for the cut-off value of I3M = 0.293. The validity and efficiency of a cut-off value of I3M < 0.293 to discriminate individuals between ≥16 years or < 16 years was
(1)
where p is post-test probability and p0 is the probability that the subject in question is ≥ 16 years, given that he or she is aged between 11 and 20 years, which represent the target population. Probability p0 was calculated as the proportion of subjects between 16 and 20 years of age who live in the Andhra Pradesh according to demographic data from the 2011 census (http://www.censusindia.gov.in/2011census/Cseries/C-13.html) and those between 11 and 20 years which was evaluated from data from the same web source. This proportion was considered to be 51.3% for males and 51.9% for females. Statistical analysis was performed by IBM SPSS 20.0 software program (IBM® SPSS® Statistics, Armonk, NY). The significant threshold was set at
Table 2 Parameter estimates of the third molar maturity index (I3M) and gender (g) as explanatory variables and ≥16 years (E = 1) and < 16 years (E = 0) age as dichotomous dependent variable on Logistic Regression.
Constant I3M gender
110
B
Std.Error
Wald statistics
df
p
1.887 −5.176 0.424
0.185 0.340 0.219
103.618 231.190 3.748
1 1 1
< 0.000 < 0.000 0.053
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
errors.37,38 To reduce such errors related to subjectivism in the assessment of dental age and for a greater confidence interval, mensuration techniques can be adapted to extract dimensional information from images.39 In 2008, Cameriere et al.20 introduced such method that requires measurement of apical pulp width and tooth length as visualized in OPTs and ratio be calculated from them, which means that derived index is scale-invariant and can also be applied to non-digital media that lacks information regarding magnification factor.21 Up to the age of 14, most of the permanent teeth finished their maturation so only third molars, which can maturate up to the early twenties, can be used for discriminating at 16 years threshold.38,40 In the present study, the authors analyzed the OPTs of 819 South Indian subjects aged between 11 and 20 years by logistic regression analysis to derive a cut-off value using third molar maturity index, to discriminate individuals who are around MACR (16 years of age). Only I3M statistically significantly contributed to the regression model while gender did not. ROC analysis showed that cut-off value of I3M < 0.293 the best discriminate subjects as 16 years of age and older or under 16. For each threshold, there is a combination of sensitivity and specificity obtained for different cut-off values that can be connected in a graph, resulting in ROC curve. The more the ROC curve moves towards the left upper corner, represents the perfect dichotomous test with 100% sensitivity and 100% specificity, the better test it is.41 A measure of performance for the test is the area under the ROC curve, and it varies between 0.5 (totally uninformative test) to 1 (a test that perfectly discriminates).42 In this paper, the resulting ROC curve (Fig. 2) has an area under the curve (AUC) of 0.936 indicating that this test predicts the likelihood of age being at least 16 years.35 The results of the analysis of the test sample highlighted the applicability of derived cut-off value, and similar performance. The cut-off value of I3M < 0.293 showed better sensitivity, 90.6% in males and 90.0% in females, than specificity, 85.9% in males and 87.1% in females. Females were slightly better classified (88.7%) than males (88.0%). Further, the quantities obtained from 2-by-2 contingency table (Table 3) to test the performance of cut-off value for discriminating subjects around MACR were discussed. What is the probability of an individual at a specific I3M cut-off value being at least 16? This is expressed as the positive predictive value, and many studies have verified the forensic usefulness of Positive and Negative predictive values (PPV & NPV) and Likelihood ratios (LRs) to diagnostically analyze the probability of discriminating individuals.24,43 In the present study, predictive values were used, less false positives indicate a higher PPV. A
Table 3 Contingency table describing discrimination performance of the test (≥16 years of < 16 years) for cut-off values of third molar maturity index (I3M) in females and males. Males
Total Males Age
Females
Total Females
Age
≥16
< 16
Test I3M < 0.293 I3M ≥ 0.293
48TP 5FP
9FN 55TN
Total
53
64
≥16
< 16
57 60
72TP 8FP
8FN 54TN
80 62
117
80
62
142
Abbreviation: TP, true positive; FN, false negative, FP, false positive, TN, true negative. Table 4 The quantities from 2-by-2 contingency tables (95% confidence interval) to test the performance of cut-off value of I3M < 0.293 in males to discriminate subjects as ≥16 years and < 16 years of age. Quantities
Males
Females
Accuracy Sensitivity Specificity PPV NPV LR + LR Bayes PTP
88.0% (95% CI, 82.2%–93.9%) 90.6% (95%CI, 82.7%–98.4%) 85.9% (95%CI, 77.4%–94.5%) 84.2% (95%CI, 74.7%–93.7%) 91.7% (95%CI, 84.7%–98.7%) 6.44 (95%CI, 3.49 to 11.87) 0.11 (95%CI, 0.05 to 0.25) 87.2% (95%CI, 78.1%–96.2%)
88.7% (95%CI, 83.5%–93.6%) 90.0% (95%CI, 83.4%–96.6%) 87.1% (95%CI, 78.8%–95.4%) 90.0% (95%CI, 83.4%–96.6%) 87.1% (95%CI,78.8%–95.4%) 6.97 (95%CI, 3.64 to 13.38) 0.11 (95%CI, 0.06 to 0.22) 88.3% (95%CI, 80.1%–96.4%)
Abbreviation: LR+, positive likelihood ratio; LR+, negative likelihood ratio; Bayes PTP, Bayes post-test probability.
evaluated on the test dataset, Table 3. Table 4 shows the values of the test for males and females. In males, 103 out of 117 subjects, 88.0% (95% CI, 83.9%–91.1%) were correctly classified. The sensitivity, or the proportion of subjects being ≥16 years of age whose test was positive, was 90.6% (95% CI, 82.7%–98.4%). The specificity, the proportion of subjects < 16 years whose test was negative, was 85.9% (95% CI, 77.4%–94.5%). PPV and NPV were 84.2% (95% CI, 74.7%–93.7%) and 91.7% (95% CI, 84.7%–98.7%), respectively. Positive likelihood ratio (LR+) and negative likelihood ratio (LR−) were 6.44 (95% CI, 3.49 to 11.87) and 0.11 (95% CI, 0.05 to 0.25) respectively. The estimated Bayes post-test probability was 87.2% (95% CI, 78.1–96.2%). In females, 126 out of 142 participants were correctly classified, 88.7% (95% CI, 83.5%–93.6%). The sensitivity was 90.0% (95% CI, 83.4%–96.6%) and specificity was 87.1% (95% CI, 78.8%–95.4%). PPV and NPV were 90.0% (95% CI, 83.4%–96.6%) and 87.1% (95% CI, 78.8%–95.4%), respectively. Positive likelihood ratio (LR+) and negative likelihood ratio (LR−) were 6.97 (95% CI, 3.64 to 13.38) and 0.11 (95% CI, 0.06 to 0.22), respectively. The estimated Bayes post-test probability was 88.3% (95% CI, 80.1%–96.4%). 4. Discussion Predicting one's attainment of MACR remains concern both in legal and forensic aspects. Moreover, through odontology, one can provide a solution for the application of specific techniques depending on the fact that tooth mineralization is more controlled by genes and less by environmental factors.36 Adopting reliable and efficient scientific methods to determine the status of an individual, i.e., above or below 16 years of age, within a specific population has become increasingly important to resolve court cases. Although Demirjian's mineralization stages of tooth mineralization were applied in general, appreciation of these mineralization stages is often difficult, and the choice of assigning a stage is quite subjective and commonly associated with prediction
Fig. 2. Receiver operating characteristic curves for ≥16 years or older status of the third molar maturity index. 111
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
sensitive test will have a higher NPV in the observed population. Higher PPV was observed in females, i.e., 90.0% than in males, 84.2% when, while NPV was greater in males, 91.7%, than in females, 87.1%. A likelihood ratio is used to interpret the discriminative score as the strength of evidence and is adapted to convey expert evaluative opinions to the legal system.44 It combines the sensitivity and specificity into a single value that indicates how much the test result will reduce the uncertainty of a given diagnosis.45 For males, the LR + ratio was 6.44 and LR-was 0.11 while for females LR+ was 6.97 and LR-was 0.11. When the cut-off value was applied, the percentage of false negatives in males was 7.7%, in females 5.6% and the proportion of false positives were 4.3% and 5.6% respectively. From the forensic and legal point of view, it is important to minimize the percentage of errors. As it considered being a more serious mistake to charge an individual younger than 16 years with the charges that apply to the subjects who are above MACR and it may lead to unfair treatment of these children in society.46 The error rate is one concept that assesses the performance of the test,41 and it is referred as a weighted average of errors among false negatives and false positives. Using the obtained cut-off value of I3M in the present study, we obtained an error rate of 12% in males and 11.3% in females. Guidelines for the age estimation in criminal proceedings suggest a multidisciplinary approach.47 AGFAD recommended physical examination with basic anthropometric measurements: height, weight, pubertal characteristics, discerning signs of sexual maturity or presence of any developmental disorders, X-ray study of hand-wrist and a dental examination recording status of dentition and radiographic study.48 In the study by Cameriere et al.,49 using X-ray images of hand-wrist and CAD program, evaluated ossification of the carpals which showed good agreement with chronological age and ossification of epiphyses of ulna and radius, for which mineralization lasts until 16 or 17 years.49 It suggests that possible future developments in lines of mineralization of ulna and radius and determining MACR, can be done and be a possible addition for better discrimination of these subjects.
indianexpress.com/article/india/india-others/juvenile-justice-bill-passed-in-lok-sabha/. 14. Gulsahi A, Tirali RE, Cehreli SB, De Luca S, Ferrante L, Cameriere R. The reliability of Cameriere's method in Turkish children: a preliminary report. Forensic Sci Int. 2015;249:319 e1–5. 15. De Luca S, De Giorgio S, Butti AC, Biagi R, Cingolani M, Cameriere R. Age estimation in children by measurement of open apices in tooth roots: study of a Mexican sample. Forensic Sci Int. 2012;221:155 e1–7. 16. Schmeling A, Reisinger W, Geserick G, Olze A. Age estimation of unaccompanied minors. Part I. General considerations. Forensic Sci Int. 2006;159(Suppl 1):S61–S64. 17. Schmeling A, Grundmann C, Fuhrmann A, et al. Criteria for age estimation in living individuals. Int J Leg Med. 2008;122:457–460. 18. Schmeling A, Geserick G, Reisinger W, Olze A. Age estimation. Forensic Sci Int. 2007;165:178–181. 19. Mincer HH, Harris EF, Berryman HE. The A.B.F.O. study of third molar development and its use as an estimator of chronological age. J Forensic Sci. 1993;38:379–390. 20. Cameriere R, Ferrante L, De Angelis D, Scarpino F, Galli F. The comparison between measurement of open apices of third molars and Demirjian stages to test chronological age of over 18 year olds in living subjects. Int J Leg Med. 2008;122:493–497. 21. Franklin D, Karkhanis S, Flavel A, Collini F, DeLuca S, Cameriere R. Accuracy of a cut-off value based on the third molar index: validation in an Australian population. Forensic Sci Int. 2016;266:575 e1–6. 22. Cavrić J, Galić I, Vodanović M, et al. Third molar maturity index (I3M) for assessing age of majority in a black African population in Botswana. Int J Leg Med. 2016;130:1109–1120. 23. Deitos AR, Costa C, Michel-Crosato E, Galic I, Cameriere R, Biazevic MG. Age estimation among Brazilians: younger or older than 18? J Forensic Leg Med. 2015;33:111–115. 24. Galic I, Lauc T, Brkic H, et al. Cameriere's third molar maturity index in assessing age of majority. Forensic Sci Int. 2015;252:191 e1–5. 25. De Luca S, Biagi R, Begnoni G, et al. Accuracy of Cameriere's cut-off value for third molar in assessing 18 years of age. Forensic Sci Int. 2014;235:102 e1–6. 26. Cameriere R, Pacifici A, Viva S, Carbone D, Pacifici L, Polimeni A. Adult or not? Accuracy of Cameriere's cut-off value for third molar in assessing 18 years of age for legal purposes. Minerva Stomatol. 2014;63:283–294. 27. Zelic K, Galic I, Nedeljkovic N, et al. Accuracy of Cameriere's third molar maturity index in assessing legal adulthood on Serbian population. Forensic Sci Int. 2016;259:127–132. 28. Quispe Lizarbe RJ, Solís Adrianzén C, Quezada-Márquez MM, Galić I, Cameriere R. Demirjian's stages and Cameriere's third molar maturity index to estimate legal adult age in Peruvian population. Leg Med. 2017;25:59–65. 29. De Luca S, Aguilar L, Rivera M, et al. Accuracy of cut-off value by measurement of third molar index: study of a Colombian sample. Forensic Sci Int. 2016;261:160 e1–5. 30. Dardouri AA, Cameriere R, De Luca S, Vanin S. Third molar maturity index by measurements of open apices in a Libyan sample of living subjects. Forensic Sci Int. 2016;267:230 e1–e6. 31. Ferrante L, Cameriere R. Statistical methods to assess the reliability of measurements in the procedures for forensic age estimation. Int J Leg Med. 2009;123:277–283. 32. Martin-de las Heras S, Garcia-Fortea P, Ortega A, Zodocovich S, Valenzuela A. Third molar development according to chronological age in populations from Spanish and Magrebian origin. Forensic Sci Int. 2008;174:47–53. 33. Galic I, Mihanovic F, Giuliodori A, Conforti F, Cingolani M, Cameriere R. Accuracy of scoring of the epiphyses at the knee joint (SKJ) for assessing legal adult age of 18 years. Int J Leg Med. 2016;130:1129–1142. 34. Schisterman EF, Perkins NJ, Liu A, Bondell H. Optimal cut-point and its corresponding Youden Index to discriminate individuals using pooled blood samples. Epidemiology. 2005;16:73–81. 35. Fletcher R, Fletcher S. Diagnosis. In: Fletcher R, Fletcher S, eds. Clinical Epidemiology the Essentials. Baltimore: Wolters, Kluwer, Lippincott. Williams & Wilkins; 2005:35–58. 36. Gleiser I, Hunt Jr EE. The permanent mandibular first molar: its calcification, eruption and decay. Am J Phys Anthropol. 1955;13:253–283. 37. Foti B, Lalys L, Adalian P, et al. New forensic approach to age determination in children based on tooth eruption. Forensic Sci Int. 2003;132:49–56. 38. Liversidge HM. Dental age revisted. In: Irish JD, Nelson GC, eds. Technique and Application in Dental Anthropology. Cambridge: Cambridge University Press; 2008:234–252. 39. SWGDE Guidelines for Forensic Image Analysis. Scientific Working Group on Digital Evidence. . Available from: https://www.swgde.org/documents/Current%20Documents/ SWGDE%20Guidelines%20for%20Forensic%20Image%20Analysis. 40. Cameriere R, Ferrante L, Cingolani M. Age estimation in children by measurement of open apices in teeth. Int J Leg Med. 2006;120:49–52. 41. Habbema JDF, Eijkemans R, Krijnen P, Knottnerus JA. Analysis of data on the accuracy of diagnostic tests. In: Knottnerus J, Buntinx F, eds. The Evidence Base of Clinical Diagnosis. London: Wiley-Blackwell; 2009:118–145. 42. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29–36. 43. Liversidge HM, Marsden PH. Estimating age and the likelihood of having attained 18 years of age using mandibular third molars. Br Dent J. 2010;209:E13. 44. Martire KA, Kemp RI, Sayle M, Newell BR. On the interpretation of likelihood ratios in forensic science evidence: presentation formats and the weak evidence effect. Forensic Sci Int. 2014;240:61–68. 45. Florkowski CM. Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests. Clin Biochem Rev. 2008;29(Suppl 1):S83–S87. 46. Thevissen PW, Kvaal SI, Willems G. Ethics in age estimation of unaccompanied minors. J Forensic Odontostomatol. 2012;30(Suppl 1):84–102. 47. Schmeling A, Dettmeyer R, Rudolf E, Vieth V, Geserick G. Forensic age estimation. Deutsches Arzteblatt international. 2016;113:44–50. 48. Schmeling A, Olze A, Reisinger W, Geserick G. Age estimation of living people undergoing criminal proceedings. Lancet. 2001;358:89–90. 49. Cameriere R, Ferrante L, Mirtella D, Cingolani M. Carpals and epiphyses of radius and ulna as age indicators. Int J Leg Med. 2006;120:143–146.
5. Conclusion In conclusion, the choice of third molar maturity index, I3M < 0.293, can be used as a supplementary method to indicate MACR age (≥16 years) for forensic and legal purposes. The specific cutoff value for MACR should be verified in other population. References 1. Thevissen PW, Fieuws S, Willems G. Third molar development: evaluation of nine tooth development registration techniques for age estimations. J Forensic Sci. 2013;58:393–397. 2. Urbas G. The age of criminal responsibility (No. 181). Criminology Aio. 2000; 2000:1–6 Canberra: Trends and Issues in Crime and Criminal Justice http://wwwaicgovau/media_ library/publications/tandi_pdf/tandi181pdf. 3. Mahalaxmi T. Doli incapax: Re- accessing the age of criminal responsibility. RHIMRJ. 2015;2:1–4. 4. Burrows AH. The problem of juvenile delinquency. J Educ Sociol. 1946;19:382–390. 5. Balla SB, Galic I, P K, Vanin S, De Luca S, Cameriere R. Validation of third molar maturity index (I3M) for discrimination of juvenile/adult status in South Indian population. J Forensic Leg Med. 2017;49:2–7. 6. Bhatia S. The Minimum Age of Criminal Responsibility in India: is it to be blamed for the increasing youth crime? Rostrum’s Law Review. 2014;2. 7. Bajpai A. Child Rights in India: Law, Policy, and Practice. New Delhi; New York: Oxford University Press; 2003. 8. Bureau NCRed. Crime in India 2015 Statistics. 2015; 2015. Available from: ncrb.gov.in/ StatPublications/CII/CII2015/FILES/CrimeInIndia2015.pdf. 9. Staff FP. Trying Minors as Adults for Rape: Will Amending Juvenile Justice Act Deter Crime? Firstpost. Firstpost; 2014. 10. 2012 Delhi gang rape case: a chronology. Available at: http://www.huffingtonpost.in/ 2017/05/05/2012-delhi-gang-rape-case-a-chronology_a_22070838/. 11. Mukesh Y, Pooja R. Age of criminal responsibility of juvenile in India vis-a-vis global scenario: a critical review. J Indian Acad Forensic Med. 2013;35:262–271. 12. India to change age of criminal responsibility for minors accused of rape and murder. Available at: http://www.telegraph.co.uk/news/worldnews/asia/india/10499798/Indiato-change-age-of-criminal-responsibility-for-minors-accused-of-rape-and-murder.html. 13. 16 to 18: LS passes bill redefining a juvenile's age in serious crimes. Available at: http://
112
Contents lists available at ScienceDirect
Journal of Forensic and Legal Medicine journal homepage: www.elsevier.com/locate/yjflm
Research Paper
A cut-off value of third molar maturity index for indicating a minimum age of criminal responsibility: Older or younger than 16 years?
T
Sudheer B. Ballaa,∗, Subramanyeswara S. Chinnib, Ivan Galicc, Aditya Mohan Alwalad, Pramod Machanie, Roberto Camerieref a
Department of Forensic Odontology, Panineeya Mahavidyalaya Institute of Dental Sciences, Hyderabad, Telangana, India Consultant Forensic Odontologist, Bangalore, Karnataka, India c Departments of Research in Biomedicine and Health, University of Split School of Medicine, Spinciceva 2, 21000, Split, Croatia d Department of Oral and Maxillofacial Surgery, MNR Dental College, Sangareddy, Telangana, India e Nightingales Regional Lead, Bangalore, Karnataka, India f Age Estimation Project, Institute of Legal Medicine, University of Macerata, Macerata, Italy b
A R T I C LE I N FO
A B S T R A C T
Keywords: Dental age estimation Minimum age of criminal responsibility Third molar maturity index Orthopantomograms 16 years
Providing appropriate legal mechanisms, that evaluate the progression of development from the age of childhood innocence to maturity and full responsibility, considered one of the difficult areas of criminal justice policy. The minimum age of criminal responsibility (MACR) in children varies among countries and differs widely owing to history and culture. Due to rising and brutality of criminal offenses, particularly by juveniles, few countries have lowered the minimum age of criminal responsibility, and many have considered/considering to do the same. India is one such country in which is under the proposal of lowering the age of criminal responsibility to 16 years. As there is lack of useful age assessment methods, that can indicate whether if the individual in question is younger (< 16 years) or older than MACR (≥16 years), the present study was undertaken to derive a specific cut-off value of the third molar maturity index (I3M) for this purpose. The sample consisted of 1078 orthopantomograms (OPTs) from Andhra Pradesh, India, aged between 11 and 20 years (45.4% males and 54.6% females). The reproducibility of I3M was calculated by intra-class correlation coefficients, which showed an intra-rater and inter-rater agreement of 0.912 and 0.891, respectively. The sample was divided into training dataset (819 OPTs), to test I3M and gender as independent variables and MACR as a dependent variable by logistic regression analysis, and test dataset (259 OPTs). A receiver operating curve (ROC) analysis was performed to evaluate the specific cut-off value of I3M for predicting MACR status. A logistic regression analysis showed that gender was not statistically significant for predicting MACR status while ROC analysis indicated a cut-off value of I3M = 0.293 as best for predicting MACR status. The performance of derived cut-off value was tested in a test data set. The sensitivity of the test was 90.6% and 90%, while specificity was 86% and 87.1% in males and females, respectively. The proportion of correctly classified individuals was 88.0% and 88.7% in males and females, respectively. The estimated Bayes post-test probability in males was 87.2% and while in females it was 88.3%. The results highlight the contribution of the derived cut-off value of I3M for discriminating individuals around MACR and should be evaluated in other populations.
1. Introduction Assessing dental age is not just limited to the prediction of growth in children for diagnosis and treatment planning, but can also be used to resolve issues related to prosecution in the criminal and civil courts.1 At times, forensic experts needed to answer question regards juvenile's attainment of the minimum age of criminal responsibility (MACR) and age of majority. The former is considered one of the most sensitive areas
∗
of criminal justice policy as it is vital to apply appropriate legal mechanisms that reflect the transition from the age of childhood innocence through to maturity and full responsibility under the criminal law.2 In recent times, Juvenile delinquency has attracted much legislative debate and judicial intervention.3 It refers to the antisocial or the criminal activity of the child which violates the law and often considered to be a gateway to adult crime since a large percentage of criminal careers has their roots in childhood. The legal difference between a criminal and a
Corresponding author. E-mail address: [email protected] (S.B. Balla).
https://doi.org/10.1016/j.jflm.2019.05.014 Received 23 March 2018; Accepted 19 May 2019 Available online 22 May 2019 1752-928X/ © 2019 Elsevier Ltd and Faculty of Forensic and Legal Medicine. All rights reserved.
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
12–14 years. Since third molars are the only teeth available growing at the age of 18 years, their development was considered most suitable evidence for juvenile/adult discrimination. Mincer et al.19 were the first to verify the usefulness of Demirjian's stages (DS) and third molars for the discrimination of juvenile/adult status. Later Cameriere et al.20 have demonstrated the better performance of the specific value of third molar maturity index (I3M), I3M < 0.08, in discriminating adults or minors compared to DS system. The applicability of latter in predicting the age of majority was tested successfully in various populations.5,21–30 It is vital to adopt accurate age estimation methods prior considering steps to prevent and treat delinquency. Moreover, recent trends in India posed a task to forensic experts and pointed out the lack of useful age estimation methods that can indicate whether if the individual in question is younger or older than MACR (16 years). Currently, there are no methods available in practice or in literature to indicate MACR among Indian individuals. This study was undertaken to derive a cut-off value of I3M which can be useful for predicting being younger (< 16 years) or older than MACR (≥16 years).
delinquent is a matter of a day on the calendar as if one commits an illegal act a day before he ceases to be a juvenile is considered a delinquent act, otherwise as a criminal offense. Here the unlawful act is the same, but the perpetrator of the act is a day older, and therefore, a criminal.4 1.1. MACR and judicial trend in India Article 40.3 of the Convention on the rights of the child defines MACR as “the age below which a person is completely immune from any criminal liability due to lack of maturity and judgment to understand the consequences of one's actions”.5 In practical terms, a separate justice system for children that has rehabilitation and reintegration should be established. In India, the criminal system is governed and regulated by two major legislations including the Indian Penal Code, 1860 (IPC) and the Criminal Procedure Code, 1970 (CrPC). The IPC has set the minimum age of criminal responsibility as 12 years.6 According to Juvenile Justice (Care and protection of children) Act, 2000, children cannot be put into the same category as adults and hence required special provisions for them. This act has set the age of criminal responsibility at 18 years that concurs definition of child under the UN convention on the rights of the child.6,7 Juvenile offenders are accorded special treatment under this act and are not tried in the same courts as adult offenders. In extreme cases under Juvenile Justice Act, the offenders will be sent to detention for a maximum period of three years in reformatory centers. In India, there has been a sharp rise in crimes committed by juveniles or young adults. According to the statistics of National Crime Records Bureau (NCRB), 27% of the juvenile crimes in 2015 were perpetrated by the individuals between ages 12–16 years, 1.4% by those between 7 and 12 years, and 71.6% by those between 16 and 18 years.8 Owing to the rise and brutality of criminal offenses particularly by juveniles, there was an increasing sense of urgency to create legal avenues for some deterrence to warn off the under-age perpetrators.9
2. Materials and methods A final sample of panoramic radiographs (OPTs) of 1078 living subjects was analyzed retrospectively (Table 1). OPTs used in this study were initialized for different clinical purposes on patients from the Panineeya Institute of Dental Sciences, Hyderabad, Telangana and from private dental clinics, Andhra Pradesh, India. Ethics Committee for research involving human subjects approved the usage of OPTs in this study. Each OPT was separately recorded in Excel file. Record comprised an identification number, gender, date of birth, date of OPT. Real age in years was calculated as the difference between dates of OPT and birth. The inclusion criteria for retrieving the sample for the analysis were: OPT of the good technical quality, age between 11 and 20 years, no evidence of hereditary or systematic diseases which can affect the dental and skeletal development, including hormonal deficiency or hormonal therapy, gastrointestinal diseases. Exclusion criteria are any missing or incomplete data on birth, agenesis of any permanent teeth including all four third molars, extracted second and third molars, rotated or impacted third molars. All OPTs were recorded in JPG format. Only third molars from the left side of the mandible were evaluated according to Cameriere's method. A freeware image-processing program (Adobe® Photoshop® CS4) was used for all adjustment of a grayscale and brightness and contrast improvement. Briefly, the measurements of the projections of open apices, Ai, and height, Li, of the third molars were recorded and I3M was calculated as the proportion of open apices and heights or
1.2. Raising concerns As per NCRB statistics, in 2010, the toll of rape cases by juveniles stood at 858, it became 1149 in 2011, 1316 in 2012, which further increased up to 1388 in 2013.9 Moreover, in last ten years, heinous crimes by juveniles have increased four to five times. One such incident is frightening gang-rape- cum-murder in Delhi in which one of accused is a minor who allegedly committed the most brutal act has shifted the spotlight on juvenile offenders, their age bar, nature of crime and brutality involved.10 Keeping in mind the rise of offenses by minors, there is a need for the change of the current MACR (18 years) and to make special provisions for serious offenses. Following this, there are serious issues raised by law, regarding amendments to the criminal law and MACR. This issue has reached the apex court of India, asking for complete struck off the Act of 2000 and to make changes in various provisions to enhanced punishment to a juvenile in conflict with the law.11 Later the government considered a proposal that under 18 years accused of heinous crimes are tried as adults and passed a bill redefining a juvenile's age in these offenses.12,13
Table 1 Age and Sex distribution of the data (training and test datasets) and the parentheses in training dataset represents the number of participants with closed apices of left second mandibular molar.
1.3. Need of the hour and need for the study In specific legal requirements, usage of low-invasive methods with higher accuracy and precision are required to assign an age to a living child without legal documents.14,15 To further improve the diagnostic accuracy of age estimates, the Study Group on Forensic Age Diagnostics (AGFAD) recommended usage of methods in combination that includes a physical examination with determination of anthropometric measurements, examination of skeletal development that includes handwrist bones and evaluation of dental status by dental exams.16–18 The mineralization of the second molars completes between the age of 109
Age (years)
Malesb training dataset
Females training dataset
Total training dataset
Males test dataset
Females test dataset
Total test dataset
11 12 13 14 15 16 17 18 19 20
26 35 45 57 54 31 41 33 31 20
39 50 47 53 62 42 41 39 51 22
65 (0) 85 (6) 92 (29) 110 (62) 116 (98) 73 (70) 82 (81) 72 (72) 82 (82) 42 (42)
10 11 10 15 17 13 11 10 11 9
10 10 9 17 18 18 15 17 10 18
20 21 19 32 35 31 26 27 21 27
Total
373
819
117
142
259
(0) (2) (15) (24) (43) (30) (41) (33) (31) (20)
(0) (4) (14) (38) (55) (40) (40) (39) (51) (22)
446
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
I3M = [Ai/Li]. I3M is evaluated in a similar way to the ratio Ai to Li, when i = 6 or 7, as reported for the first and second lower molars, as described in Cameriere et al.20 Additionally, if the development of the root of TM completed, then I3M = 0.0 was recorded. Pearson correlation coefficient was performed to test the correlation between I3M and real age. The first author performed the measures of the sample. The analyses were performed using a blind approach, and the observer was not able to detect identification number, gender or age of the subjects. The first author and one other observer additionally evaluated randomly selected 50 OPTs two weeks after finished the first measurements. The intraclass correlation coefficient (ICC) was calculated to assess intra-rater and inter-rater agreement of I3M.31 A sample of OPTs was randomly divided into training dataset (819) and test dataset (259). A logistic regression linear model was tested on the training dataset with dependent variables E = 1 and E = 0 to verify the significance of the independent variables I3M and gender on the binary answer whether a subject is ≥ 16 years (E = 1) or < 16 years (E = 0). The dummy variable gender equals to 1 for boys and 0 for girls. The predictive accuracy of the specific cut-off value of I3M to test whether a subject is ≥ 16 years or < 16 years was established by determining the receiver operating characteristic (ROC) analysis. The area under the curve (AUC) and 95% confidence interval (95% CI) was calculated with a non-parametric method.20,32,33 If the cut-off is closer to the apex of the ROC curve to the upper left corner, it is the greater accuracy of the test.33 The best performance of the discrimination test was established as the maximum value of Youden's index (J). The greatest Youden's index was used to determine a specific cut-off value of I3M.34 The performance of the established cut-off value of I3M was evaluated by the two-by-two contingency table (CT) on the test dataset of OPTs. Precisely, CT in the first row lists the results of those who are ≥16 years and have I3M less than cut-off value (true positives, TP), following those who are < 16 years and have I3M less than cut-off value (false positives, FP). In the second row list those who are ≥16 years and have I3M ≥ than cut-off value (false negatives, FN), and finally, those who are < 16 and have I3M ≥ cut-off value (true negatives, TN). The percentage of accurate classification (Ac), sensitivity (Se) or the percentage of the subjects ≥16 years who have I3M < cutoff value, and the specificity (Sp) or the percentage of individuals younger than 16 who have I3M ≥ cut-off value, were calculated. The positive predictive value (PPV) and the negative predictive value (NPV) or the proportions of positive and negative results that are truly positive and truly negative results, respectively were calculated.35 Also, the positive likelihood ratio (LR+) and negative likelihood ratio (LR-) were also calculated to express how many times more or less likely a test result is to be found in (≥16 years) compared to (< 16 years) participants.35 The post-test probability p of being 16 years of age or older can help to discriminate between those individuals who are 16 and over and under 16. According to Bayes theorem, the post-test probability may be written as:
p=
Se × p0 Se × p0 + (1 − Sp) × (1 − p0 )
Fig. 1. Scatter-plot of the relationship between age (years) and the third molar maturity index.
p < 0.05. 3. Results The final sample consisted of 1078 OPTs. Table 1 demonstrates the distribution of the training and test datasets according to gender and age. The number of subjects with closed apices of second molars in each age cohort was presented in parenthesis of training dataset of same table. Closed apices were first noticed in both boys and girls at 12 years age. None of the apices were open after 16 years of age in both genders. Pearson correlation coefficients for the correlation between I3M and real age were better for females, 0.780 (p < 0.001), than for males, 0.778 (p < 0.001), Fig. 1. The intra-class correlation coefficient for the intrarater agreement was 0.912 while for the inter-rater agreement was 0.891. Obtained results of ICC showed excellent results both on the precision and the repeatability of the variable I3M. The logistic model showed the significance of variable I3M (p < 0.001) while gender was not statistically significant (p = 0.053), Table 2. The full model containing I3M as a predictor was statistically significant, χ2 (1, 819) = 556.3, p < 0.001, indicating that model was able to discriminate between those who are ≥16 years and < 16 years. The whole model explained between 0.493 (Cox & Snell R Square) and 0.662 (Nagelkerke R Square) of the variance in ≥16 years and < 16 statuses. A linear logistic model may be written as: Logit (p) = 2.086–5.102 × I3M
(2)
The value for AUC was 0.936 (95% CI, 0.920 to 0.953). A cut-off value of I3M, to discriminate that individuals are ≥16 years or < 16 years, was established for the highest value of the Youden index.35 The highest Youden index was 0.739 for the cut-off value of I3M = 0.293. The validity and efficiency of a cut-off value of I3M < 0.293 to discriminate individuals between ≥16 years or < 16 years was
(1)
where p is post-test probability and p0 is the probability that the subject in question is ≥ 16 years, given that he or she is aged between 11 and 20 years, which represent the target population. Probability p0 was calculated as the proportion of subjects between 16 and 20 years of age who live in the Andhra Pradesh according to demographic data from the 2011 census (http://www.censusindia.gov.in/2011census/Cseries/C-13.html) and those between 11 and 20 years which was evaluated from data from the same web source. This proportion was considered to be 51.3% for males and 51.9% for females. Statistical analysis was performed by IBM SPSS 20.0 software program (IBM® SPSS® Statistics, Armonk, NY). The significant threshold was set at
Table 2 Parameter estimates of the third molar maturity index (I3M) and gender (g) as explanatory variables and ≥16 years (E = 1) and < 16 years (E = 0) age as dichotomous dependent variable on Logistic Regression.
Constant I3M gender
110
B
Std.Error
Wald statistics
df
p
1.887 −5.176 0.424
0.185 0.340 0.219
103.618 231.190 3.748
1 1 1
< 0.000 < 0.000 0.053
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
errors.37,38 To reduce such errors related to subjectivism in the assessment of dental age and for a greater confidence interval, mensuration techniques can be adapted to extract dimensional information from images.39 In 2008, Cameriere et al.20 introduced such method that requires measurement of apical pulp width and tooth length as visualized in OPTs and ratio be calculated from them, which means that derived index is scale-invariant and can also be applied to non-digital media that lacks information regarding magnification factor.21 Up to the age of 14, most of the permanent teeth finished their maturation so only third molars, which can maturate up to the early twenties, can be used for discriminating at 16 years threshold.38,40 In the present study, the authors analyzed the OPTs of 819 South Indian subjects aged between 11 and 20 years by logistic regression analysis to derive a cut-off value using third molar maturity index, to discriminate individuals who are around MACR (16 years of age). Only I3M statistically significantly contributed to the regression model while gender did not. ROC analysis showed that cut-off value of I3M < 0.293 the best discriminate subjects as 16 years of age and older or under 16. For each threshold, there is a combination of sensitivity and specificity obtained for different cut-off values that can be connected in a graph, resulting in ROC curve. The more the ROC curve moves towards the left upper corner, represents the perfect dichotomous test with 100% sensitivity and 100% specificity, the better test it is.41 A measure of performance for the test is the area under the ROC curve, and it varies between 0.5 (totally uninformative test) to 1 (a test that perfectly discriminates).42 In this paper, the resulting ROC curve (Fig. 2) has an area under the curve (AUC) of 0.936 indicating that this test predicts the likelihood of age being at least 16 years.35 The results of the analysis of the test sample highlighted the applicability of derived cut-off value, and similar performance. The cut-off value of I3M < 0.293 showed better sensitivity, 90.6% in males and 90.0% in females, than specificity, 85.9% in males and 87.1% in females. Females were slightly better classified (88.7%) than males (88.0%). Further, the quantities obtained from 2-by-2 contingency table (Table 3) to test the performance of cut-off value for discriminating subjects around MACR were discussed. What is the probability of an individual at a specific I3M cut-off value being at least 16? This is expressed as the positive predictive value, and many studies have verified the forensic usefulness of Positive and Negative predictive values (PPV & NPV) and Likelihood ratios (LRs) to diagnostically analyze the probability of discriminating individuals.24,43 In the present study, predictive values were used, less false positives indicate a higher PPV. A
Table 3 Contingency table describing discrimination performance of the test (≥16 years of < 16 years) for cut-off values of third molar maturity index (I3M) in females and males. Males
Total Males Age
Females
Total Females
Age
≥16
< 16
Test I3M < 0.293 I3M ≥ 0.293
48TP 5FP
9FN 55TN
Total
53
64
≥16
< 16
57 60
72TP 8FP
8FN 54TN
80 62
117
80
62
142
Abbreviation: TP, true positive; FN, false negative, FP, false positive, TN, true negative. Table 4 The quantities from 2-by-2 contingency tables (95% confidence interval) to test the performance of cut-off value of I3M < 0.293 in males to discriminate subjects as ≥16 years and < 16 years of age. Quantities
Males
Females
Accuracy Sensitivity Specificity PPV NPV LR + LR Bayes PTP
88.0% (95% CI, 82.2%–93.9%) 90.6% (95%CI, 82.7%–98.4%) 85.9% (95%CI, 77.4%–94.5%) 84.2% (95%CI, 74.7%–93.7%) 91.7% (95%CI, 84.7%–98.7%) 6.44 (95%CI, 3.49 to 11.87) 0.11 (95%CI, 0.05 to 0.25) 87.2% (95%CI, 78.1%–96.2%)
88.7% (95%CI, 83.5%–93.6%) 90.0% (95%CI, 83.4%–96.6%) 87.1% (95%CI, 78.8%–95.4%) 90.0% (95%CI, 83.4%–96.6%) 87.1% (95%CI,78.8%–95.4%) 6.97 (95%CI, 3.64 to 13.38) 0.11 (95%CI, 0.06 to 0.22) 88.3% (95%CI, 80.1%–96.4%)
Abbreviation: LR+, positive likelihood ratio; LR+, negative likelihood ratio; Bayes PTP, Bayes post-test probability.
evaluated on the test dataset, Table 3. Table 4 shows the values of the test for males and females. In males, 103 out of 117 subjects, 88.0% (95% CI, 83.9%–91.1%) were correctly classified. The sensitivity, or the proportion of subjects being ≥16 years of age whose test was positive, was 90.6% (95% CI, 82.7%–98.4%). The specificity, the proportion of subjects < 16 years whose test was negative, was 85.9% (95% CI, 77.4%–94.5%). PPV and NPV were 84.2% (95% CI, 74.7%–93.7%) and 91.7% (95% CI, 84.7%–98.7%), respectively. Positive likelihood ratio (LR+) and negative likelihood ratio (LR−) were 6.44 (95% CI, 3.49 to 11.87) and 0.11 (95% CI, 0.05 to 0.25) respectively. The estimated Bayes post-test probability was 87.2% (95% CI, 78.1–96.2%). In females, 126 out of 142 participants were correctly classified, 88.7% (95% CI, 83.5%–93.6%). The sensitivity was 90.0% (95% CI, 83.4%–96.6%) and specificity was 87.1% (95% CI, 78.8%–95.4%). PPV and NPV were 90.0% (95% CI, 83.4%–96.6%) and 87.1% (95% CI, 78.8%–95.4%), respectively. Positive likelihood ratio (LR+) and negative likelihood ratio (LR−) were 6.97 (95% CI, 3.64 to 13.38) and 0.11 (95% CI, 0.06 to 0.22), respectively. The estimated Bayes post-test probability was 88.3% (95% CI, 80.1%–96.4%). 4. Discussion Predicting one's attainment of MACR remains concern both in legal and forensic aspects. Moreover, through odontology, one can provide a solution for the application of specific techniques depending on the fact that tooth mineralization is more controlled by genes and less by environmental factors.36 Adopting reliable and efficient scientific methods to determine the status of an individual, i.e., above or below 16 years of age, within a specific population has become increasingly important to resolve court cases. Although Demirjian's mineralization stages of tooth mineralization were applied in general, appreciation of these mineralization stages is often difficult, and the choice of assigning a stage is quite subjective and commonly associated with prediction
Fig. 2. Receiver operating characteristic curves for ≥16 years or older status of the third molar maturity index. 111
Journal of Forensic and Legal Medicine 65 (2019) 108–112
S.B. Balla, et al.
sensitive test will have a higher NPV in the observed population. Higher PPV was observed in females, i.e., 90.0% than in males, 84.2% when, while NPV was greater in males, 91.7%, than in females, 87.1%. A likelihood ratio is used to interpret the discriminative score as the strength of evidence and is adapted to convey expert evaluative opinions to the legal system.44 It combines the sensitivity and specificity into a single value that indicates how much the test result will reduce the uncertainty of a given diagnosis.45 For males, the LR + ratio was 6.44 and LR-was 0.11 while for females LR+ was 6.97 and LR-was 0.11. When the cut-off value was applied, the percentage of false negatives in males was 7.7%, in females 5.6% and the proportion of false positives were 4.3% and 5.6% respectively. From the forensic and legal point of view, it is important to minimize the percentage of errors. As it considered being a more serious mistake to charge an individual younger than 16 years with the charges that apply to the subjects who are above MACR and it may lead to unfair treatment of these children in society.46 The error rate is one concept that assesses the performance of the test,41 and it is referred as a weighted average of errors among false negatives and false positives. Using the obtained cut-off value of I3M in the present study, we obtained an error rate of 12% in males and 11.3% in females. Guidelines for the age estimation in criminal proceedings suggest a multidisciplinary approach.47 AGFAD recommended physical examination with basic anthropometric measurements: height, weight, pubertal characteristics, discerning signs of sexual maturity or presence of any developmental disorders, X-ray study of hand-wrist and a dental examination recording status of dentition and radiographic study.48 In the study by Cameriere et al.,49 using X-ray images of hand-wrist and CAD program, evaluated ossification of the carpals which showed good agreement with chronological age and ossification of epiphyses of ulna and radius, for which mineralization lasts until 16 or 17 years.49 It suggests that possible future developments in lines of mineralization of ulna and radius and determining MACR, can be done and be a possible addition for better discrimination of these subjects.
indianexpress.com/article/india/india-others/juvenile-justice-bill-passed-in-lok-sabha/. 14. Gulsahi A, Tirali RE, Cehreli SB, De Luca S, Ferrante L, Cameriere R. The reliability of Cameriere's method in Turkish children: a preliminary report. Forensic Sci Int. 2015;249:319 e1–5. 15. De Luca S, De Giorgio S, Butti AC, Biagi R, Cingolani M, Cameriere R. Age estimation in children by measurement of open apices in tooth roots: study of a Mexican sample. Forensic Sci Int. 2012;221:155 e1–7. 16. Schmeling A, Reisinger W, Geserick G, Olze A. Age estimation of unaccompanied minors. Part I. General considerations. Forensic Sci Int. 2006;159(Suppl 1):S61–S64. 17. Schmeling A, Grundmann C, Fuhrmann A, et al. Criteria for age estimation in living individuals. Int J Leg Med. 2008;122:457–460. 18. Schmeling A, Geserick G, Reisinger W, Olze A. Age estimation. Forensic Sci Int. 2007;165:178–181. 19. Mincer HH, Harris EF, Berryman HE. The A.B.F.O. study of third molar development and its use as an estimator of chronological age. J Forensic Sci. 1993;38:379–390. 20. Cameriere R, Ferrante L, De Angelis D, Scarpino F, Galli F. The comparison between measurement of open apices of third molars and Demirjian stages to test chronological age of over 18 year olds in living subjects. Int J Leg Med. 2008;122:493–497. 21. Franklin D, Karkhanis S, Flavel A, Collini F, DeLuca S, Cameriere R. Accuracy of a cut-off value based on the third molar index: validation in an Australian population. Forensic Sci Int. 2016;266:575 e1–6. 22. Cavrić J, Galić I, Vodanović M, et al. Third molar maturity index (I3M) for assessing age of majority in a black African population in Botswana. Int J Leg Med. 2016;130:1109–1120. 23. Deitos AR, Costa C, Michel-Crosato E, Galic I, Cameriere R, Biazevic MG. Age estimation among Brazilians: younger or older than 18? J Forensic Leg Med. 2015;33:111–115. 24. Galic I, Lauc T, Brkic H, et al. Cameriere's third molar maturity index in assessing age of majority. Forensic Sci Int. 2015;252:191 e1–5. 25. De Luca S, Biagi R, Begnoni G, et al. Accuracy of Cameriere's cut-off value for third molar in assessing 18 years of age. Forensic Sci Int. 2014;235:102 e1–6. 26. Cameriere R, Pacifici A, Viva S, Carbone D, Pacifici L, Polimeni A. Adult or not? Accuracy of Cameriere's cut-off value for third molar in assessing 18 years of age for legal purposes. Minerva Stomatol. 2014;63:283–294. 27. Zelic K, Galic I, Nedeljkovic N, et al. Accuracy of Cameriere's third molar maturity index in assessing legal adulthood on Serbian population. Forensic Sci Int. 2016;259:127–132. 28. Quispe Lizarbe RJ, Solís Adrianzén C, Quezada-Márquez MM, Galić I, Cameriere R. Demirjian's stages and Cameriere's third molar maturity index to estimate legal adult age in Peruvian population. Leg Med. 2017;25:59–65. 29. De Luca S, Aguilar L, Rivera M, et al. Accuracy of cut-off value by measurement of third molar index: study of a Colombian sample. Forensic Sci Int. 2016;261:160 e1–5. 30. Dardouri AA, Cameriere R, De Luca S, Vanin S. Third molar maturity index by measurements of open apices in a Libyan sample of living subjects. Forensic Sci Int. 2016;267:230 e1–e6. 31. Ferrante L, Cameriere R. Statistical methods to assess the reliability of measurements in the procedures for forensic age estimation. Int J Leg Med. 2009;123:277–283. 32. Martin-de las Heras S, Garcia-Fortea P, Ortega A, Zodocovich S, Valenzuela A. Third molar development according to chronological age in populations from Spanish and Magrebian origin. Forensic Sci Int. 2008;174:47–53. 33. Galic I, Mihanovic F, Giuliodori A, Conforti F, Cingolani M, Cameriere R. Accuracy of scoring of the epiphyses at the knee joint (SKJ) for assessing legal adult age of 18 years. Int J Leg Med. 2016;130:1129–1142. 34. Schisterman EF, Perkins NJ, Liu A, Bondell H. Optimal cut-point and its corresponding Youden Index to discriminate individuals using pooled blood samples. Epidemiology. 2005;16:73–81. 35. Fletcher R, Fletcher S. Diagnosis. In: Fletcher R, Fletcher S, eds. Clinical Epidemiology the Essentials. Baltimore: Wolters, Kluwer, Lippincott. Williams & Wilkins; 2005:35–58. 36. Gleiser I, Hunt Jr EE. The permanent mandibular first molar: its calcification, eruption and decay. Am J Phys Anthropol. 1955;13:253–283. 37. Foti B, Lalys L, Adalian P, et al. New forensic approach to age determination in children based on tooth eruption. Forensic Sci Int. 2003;132:49–56. 38. Liversidge HM. Dental age revisted. In: Irish JD, Nelson GC, eds. Technique and Application in Dental Anthropology. Cambridge: Cambridge University Press; 2008:234–252. 39. SWGDE Guidelines for Forensic Image Analysis. Scientific Working Group on Digital Evidence. . Available from: https://www.swgde.org/documents/Current%20Documents/ SWGDE%20Guidelines%20for%20Forensic%20Image%20Analysis. 40. Cameriere R, Ferrante L, Cingolani M. Age estimation in children by measurement of open apices in teeth. Int J Leg Med. 2006;120:49–52. 41. Habbema JDF, Eijkemans R, Krijnen P, Knottnerus JA. Analysis of data on the accuracy of diagnostic tests. In: Knottnerus J, Buntinx F, eds. The Evidence Base of Clinical Diagnosis. London: Wiley-Blackwell; 2009:118–145. 42. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29–36. 43. Liversidge HM, Marsden PH. Estimating age and the likelihood of having attained 18 years of age using mandibular third molars. Br Dent J. 2010;209:E13. 44. Martire KA, Kemp RI, Sayle M, Newell BR. On the interpretation of likelihood ratios in forensic science evidence: presentation formats and the weak evidence effect. Forensic Sci Int. 2014;240:61–68. 45. Florkowski CM. Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests. Clin Biochem Rev. 2008;29(Suppl 1):S83–S87. 46. Thevissen PW, Kvaal SI, Willems G. Ethics in age estimation of unaccompanied minors. J Forensic Odontostomatol. 2012;30(Suppl 1):84–102. 47. Schmeling A, Dettmeyer R, Rudolf E, Vieth V, Geserick G. Forensic age estimation. Deutsches Arzteblatt international. 2016;113:44–50. 48. Schmeling A, Olze A, Reisinger W, Geserick G. Age estimation of living people undergoing criminal proceedings. Lancet. 2001;358:89–90. 49. Cameriere R, Ferrante L, Mirtella D, Cingolani M. Carpals and epiphyses of radius and ulna as age indicators. Int J Leg Med. 2006;120:143–146.
5. Conclusion In conclusion, the choice of third molar maturity index, I3M < 0.293, can be used as a supplementary method to indicate MACR age (≥16 years) for forensic and legal purposes. The specific cutoff value for MACR should be verified in other population. References 1. Thevissen PW, Fieuws S, Willems G. Third molar development: evaluation of nine tooth development registration techniques for age estimations. J Forensic Sci. 2013;58:393–397. 2. Urbas G. The age of criminal responsibility (No. 181). Criminology Aio. 2000; 2000:1–6 Canberra: Trends and Issues in Crime and Criminal Justice http://wwwaicgovau/media_ library/publications/tandi_pdf/tandi181pdf. 3. Mahalaxmi T. Doli incapax: Re- accessing the age of criminal responsibility. RHIMRJ. 2015;2:1–4. 4. Burrows AH. The problem of juvenile delinquency. J Educ Sociol. 1946;19:382–390. 5. Balla SB, Galic I, P K, Vanin S, De Luca S, Cameriere R. Validation of third molar maturity index (I3M) for discrimination of juvenile/adult status in South Indian population. J Forensic Leg Med. 2017;49:2–7. 6. Bhatia S. The Minimum Age of Criminal Responsibility in India: is it to be blamed for the increasing youth crime? Rostrum’s Law Review. 2014;2. 7. Bajpai A. Child Rights in India: Law, Policy, and Practice. New Delhi; New York: Oxford University Press; 2003. 8. Bureau NCRed. Crime in India 2015 Statistics. 2015; 2015. Available from: ncrb.gov.in/ StatPublications/CII/CII2015/FILES/CrimeInIndia2015.pdf. 9. Staff FP. Trying Minors as Adults for Rape: Will Amending Juvenile Justice Act Deter Crime? Firstpost. Firstpost; 2014. 10. 2012 Delhi gang rape case: a chronology. Available at: http://www.huffingtonpost.in/ 2017/05/05/2012-delhi-gang-rape-case-a-chronology_a_22070838/. 11. Mukesh Y, Pooja R. Age of criminal responsibility of juvenile in India vis-a-vis global scenario: a critical review. J Indian Acad Forensic Med. 2013;35:262–271. 12. India to change age of criminal responsibility for minors accused of rape and murder. Available at: http://www.telegraph.co.uk/news/worldnews/asia/india/10499798/Indiato-change-age-of-criminal-responsibility-for-minors-accused-of-rape-and-murder.html. 13. 16 to 18: LS passes bill redefining a juvenile's age in serious crimes. Available at: http://
112