Accuracy of the cutoff value of the third molar maturity index: an Egyptian study

Abstract

Background

This study aimed to test the sensitivity and specificity of the third molar maturity index (I_3M) cutoff value to discriminate between individuals above and below 18 years of age in an Egyptian sample.

Material and methods

Digital images of 247 orthopantographs (97 boys and 150 girls) were evaluated. The cutoff value of I_3M obtained from the results of Cameriere et al. (Int J Legal Med 122:493–497, 2008) was evaluated in both girls and boys.

Results

Sensitivity (the proportion of individuals being 18 years of age or older) was 95% and 73% in boys and girls, respectively. Meanwhile, specificity (the proportion of individuals younger than 18) was 100% in boys and 97% in girls. The proportion of correctly classified boys was 97% while it was 59% in girls.

Conclusion

It can be concluded that the third molar maturity index can discriminate between individuals who are 18 years and over and those under 18 with higher accuracy in boys.

Background

The age of 18 is the age of legal responsibility. This is important especially with the increased illegal migration. Many illegal African immigrants pass to Europe through Egypt. Illegal immigrants tend to lack reliable documents since they either lose them or even destroy them. The Egypt’s state statistics agency (CAPMAS) report said that many children in the age range 10–17 were married and 93.4% of them were girls, a violation of Egyptian laws which the government aims to combat (ahramonline accessed 11 June 2019). So, providing suitable legal procedures reflecting the transition from childhood to maturity and full criminal responsibility is an important issue in criminal justice (De Luca et al. 2016). That is why forensic age estimation of living individuals is a very important tool to overcome deliberate falsification of age in medicolegal and criminal cases (Cameriere et al. 2015; Gupta et al. 2015).

Teeth development as a mean to evaluate age in children has the advantage of being in a uniform manner from infancy to late adolescence. It is under strong genetic control and shows less environmental influences compared to skeletal maturation (Franklin et al. 2015).

The third molar development assessment is usually applied for those aged 13–23 years (Galic et al. 2015). For estimating the age of 18, Cameriere et al. (2008) published a method that depends on the assessment of the relation between age and the third molar maturity index (I_3M). They found that a cutoff value of 0.08 in Italian population showed 70% sensitivity and 98% specificity. It has been tested in many populations and showed high sensitivity and specificity (Deitos et al. 2015; Galic et al. 2015; De Luca et al. 2016). No studies until now have assessed such method on the Egyptian population. So, we aimed to test the sensitivity and specificity of this cutoff value to discriminate between Egyptian individuals above and below 18 years of age.

Material and methods

Sample

A sample of 259 orthopantographs (OPGs) from Egyptian subjects aged 13–24 was analyzed. This cross-sectional study collected OPGs at random from the digital archive of the Mansoura University, Faculty of Dentistry, during years 2015 and 2017, as well as from private dental clinics, following the ethical standards laid down by the Declaration of Helsinki (2013) and was approved by Mansoura University IRB (code: R.18.11.345). Exclusion criteria were dental pathology affecting the lower left jaw and unknown date of birth.

Following the initial sampling, 12 OPGs were excluded because they did not meet the inclusion criteria, leaving 247 OPGs (97 boys and 150 girls). The digital images were evaluated using an open-source image computer-aided drafting program used to process and analyze digital images (Digimizer Image Analysis program).

Assessment of the third molar maturity index (I_3M)

The left lower third molar was examined using the I_3M index described in the study by Cameriere et al. (2008) where if the apices of the third molar are completely closed, then I_3M = 0. Otherwise, the distance between the inner sides of the open apex, i.e., apical distance (AD) of mandibular left third molars, was measured and divided by the tooth length (L). If the tooth had two roots (ADa and ADb), the sum of these distances was divided by the tooth length (L) (Fig. 1). Measurements were carried out by two different observers.

Fig. 1

Measurements of a tooth with one root (a) and two roots (b) where AD apical distance, L length

Twenty randomly selected OPGs were assessed to check for inter-observer reliability and were re-examined after a month interval for intra-observer reliability.

Statistical analysis

A Microsoft Excel file was used in this study to register the identification number, gender, chronological age, and I_3M value. The cutoff value of 0.08 for I_3M for both genders, obtained from the results of Cameriere et al. (2008), was evaluated. An individual is considered to be 18 years of age or older if I_3M is lower than 0.08. The sensitivity of the test (i.e., the proportion of subjects > 18 years of age who have I_3M < 0.08) together with specificity (i.e., the proportion of individuals younger than 18 who have I_3M > 0.08) was evaluated.

According to Bayes’ theorem, post-test probability was written as

$$ \mathrm{PPV}=\frac{\mathrm{Se}\ {p}_0}{\mathrm{Se}\ {p}_0+\left(1-\mathrm{Sp}\right)\left(1-{p}_0\right)} $$

where positive predictive value (PPV) is the positive predictive value and p₀ is the prevalence of Egyptian subjects aged between 18 and 24 years of age among all subjects aged between 13 and 24 years old (the studied sample) who live in Egypt based on the demographic data retrieved from Egypt Demographics Profile (2018). The estimated prevalence for girls is 59% and 60% for boys.

Intra-class correlation coefficient (ICC) was used to test the intra-observer and inter-observer agreement between the two observers. To evaluate intra-observer and inter-observer agreement between both observers, Cohen’s kappa was used.

Results

Age distribution of the sample is shown in Table 1. Individuals < 18 years were 100 (42 boys and 58 girls) while those ≥ 18 years were 147 (55 boys and 92 girls). The mean chronological age of boys and girls were 18.85 ± 3.61 years and 19.11 ± 4.16 years, respectively. The minimum age at which complete maturation (union of apices) of the third molar (I_3M = 0) was noticed to be 20.07 years for girls and 19.44 years for boys.

Table 1 Sample of panoramic radiographs from Egyptian individuals according to sex and age categories

The validity of I_3M presenting the number of individuals who have I_3M ≥ 0.08 and I_3M < 0.08 and those who are younger than 18 and those not younger than 18 is shown in Tables 2 and 3.

Table 2 Contingency table of the cutoff value’s performance for girls (n, 150)

Table 3 Contingency table of the cutoff value’s performance for boys (n, 97)

The close association between adult age and positivity of the test (i.e., I_3M < 0.08) in girls is shown in Table 2. It was found that 123 out of 150 (82% (95% CI 76–88)) were correctly classified. The sensitivity of the test was 73% (95% CI 64–82) and its specificity was 97% (95% CI 92–100). The PPV was 59%.

Table 3 displays the close association between adult age and the positivity of the test in the boys. From 97 individuals, 94 were accurately classified (97% (95% CI 93–100)). The sensitivity of the test was 95% (95% CI 89–100) and the specificity was 100%. The PPV was 60%.

The ICC for intra-rater and inter-rater agreement of I_3M was 98.7% (95% CI 98.4–99.2) and 97.8% (95% CI 96–98.2), respectively. Cohen’s kappa for intra-rater and inter-rater agreement was 0.93 and 0.86.

Discussion

It is important to discriminate minors from adult especially in the absence of documents to avoid criminal violations (Deitos et al. 2015). From this point of view, methods used for age estimation applied in forensic medicine must be accurate, precise, and well-validated before application (Rivera et al. 2017).

The estimation of dental age in Egyptians was reported in previous studies (El-Bakary et al. 2010; Zaher et al. 2011; El Morsi et al. 2015; Wahdan et al. 2017). But, to the best of our knowledge, this is the first Egyptian study that validates the use of the third molar maturity index cutoff value of 0.08 to discriminate those aged 18 years or older from minors (less than 18 years).

In his original study, Cameriere et al. (2008) evaluated the I_3M in OPGs of 906 Caucasian Italian individuals aged between 14 and 23 years (53.6% females and 46.4% males). The cutoff value of 0.08 showed 70% sensitivity and 98% specificity with 83% correct classification. This cutoff value was tested in many populations around the world and showed high sensitivity and specificity although percentages were different.

Results of this Egyptian sample showed relatively faster development (earlier complete maturation of the third molar) in boys. Generally, many studies showed male precedence of the third molar development, e.g., Hispanics (Solari and Abramovitch 2002), Japanese (Olze et al. 2004), Serbian (Zelic et al. 2016), and Saudi (AlQahtani et al. 2017). Nevertheless, females were reported to show faster development of third molars in one study (Liversidge 2008).

When the cutoff value of “I_3M = 0.08” was evaluated in the present study (247 cases), the high sensitivity (Se) of the test in boys (95%) but relatively lower sensitivity in girls (73%) was documented. Sensitivity values were reported to be over 95% in Turkish males (128 boys from 293 total sample) (Gulsahi et al. 2016), Colombians (288 children), (De Luca et al. 2016), and Serbians (598 cases) (Zelic et al. 2016). Meanwhile, it was 88% in Botswana (1294 individuals) (Cavrić et al. 2016) and 90% in Australians (143 cases) (Franklin et al. 2016). Unfortunately, sensitivity was only 52% in the Saudi population (300 cases) (AlQahtani et al. 2017).

Meanwhile, a high specificity (Sp) of the cutoff value (100% in boys and 97% in girls) was found in the present sample. On the other hand, specificity was 100% in the Turkish study (Gulsahi et al. 2016) and in Saudi males (AlQahtani et al. 2017), and over 85% in Colombians (De Luca et al. 2016), Serbians (Zelic et al. 2016), Botswana (Cavrić et al. 2016), and Australian population (Franklin et al. 2016).

It seems that values of sensitivity and specificity of the present study are more in boys ensuring previous published data in Serbians (Zelic et al. 2016). On the other hand, girls showed more values in a Colombian study (De Luca et al. 2016).

Different sensitivity and specificity in these populations may be either due to actual differences between them or differences in the age distribution and sample size, as well as dissimilarity among observers. However, the examination of the third molars may provide reasonable accuracy for the likelihood that a person is at least 18 years old (Acharya et al. 2014).

In the present Egyptian sample, the proportion of correctly classified persons was 97% in boys and 82% in girls, even better than 83% reported by Cameriere et al. (2008). More or less similar proportions of correctly classified individuals were reported in Turkish (97.6% for boys and 92.7% in girls) (Gulsahi et al. 2016), Serbians (95% in males and 91% in females) (Zelic et al. 2016), and Chileans (83%) (Cameriere et al. 2018).

Meanwhile, the positive predictive value (PPV) in the present Egyptian sample was 60% in boys and 59% in girls. Although the PPV is highly useful, it is sensitive to the prevalence of the positive test group (older age category) and can rarely be generalized beyond the study sample (Gulsahi et al. 2016). In a legal and forensic context, it is ethically safer to underestimate age than to overestimate age due to constitutional and judicial implications when involving a possible adult (Cavrić et al. 2016).

Technically, unacceptable errors can be detected if an adult was wrongly diagnosed as a minor (less than 18 years). They will get the full benefit of safeguarding through being children in care. However, if the reverse occurs and a minor was diagnosed as over 18 years, this is an ethically unacceptable error as it leads to violation of minors’ rights (Garamendi et al. 2005).

In the context of estimating a person’s age, none of the techniques can be used with high reliability when the person has completed his growth. However, from the legal point of view, it is often enough to indicate whether a person is an adult or a minor (Schmeling et al. 2008).

Conclusion

In conclusion, the third molar maturity index can discriminate between individuals who are 18 years and over and those under 18 with higher accuracy in boys than girls. As the present findings are in correlation with findings for other populations, the I_3m cutoff value is justifiable to be used in medicolegal practice.