Mandibular Measurements for Sex and Age Estimation in Brazilian Sampling.

OBJECTIVE: The present study aimed to evaluate the possibility of sex and age estimation in a Brazilian sample, using the following five mandibular measurements: coronoid height, gonial angle, bigonial distance, ramus height, maximum length.
MATERIALS AND METHODS: In total, 103 mandibles were measured, 53 female and 50 male, grouped according to the age in 5 different groups. Two different observers performed all these measurements. For statistical analysis, one-way ANOVA was used with a significance level of 5% to verify if the mandibular measurements were related to age. The Student t-test was used for comparisons between the sex samples. Eighty three samples were used to develop a logistic regression model. The intra-observer and inter-observer differences were evaluated using the Kappa coefficient.
RESULTS: The sex differences were statistically significant in all the variables, and all of them were larger in men, apart from the variable related to the gonial angle (GA), which was larger in women. For the logistic regression formula, the variables used were: bigonial distance and mandibular ramus, with 90% accuracy. For age, the measurements did not show a pattern. The inter and intra-observer values were greater than 0.85.
CONCLUSION: In conclusion, it is possible to estimate sex using mandibular measurements, but it is not recommended to use these measurements for age estimation.

Introduction

The identification of unknown human remains in cases of putrefaction, skeletonization or mutilation is common in forensic practice (, ), and in these cases, secondary methods, such as the establishment of an anthropological profile, are useful and important (), through the sex, age, ancestry, and stature estimation, among other parameters of forensic interest. (, )

Sex is highly important for forensic purposes, mainly because it is necessary to estimate other anthropological parameters. (, ) There are different methodologies for estimation, which involve analyses of teeth and bones (–). Several authors have studied the mandible for sex and age estimation, mainly because it is robust and protected by the facial musculature, making it easier to be found intact. ()

Additionally the adult mandible can be used to estimate sex with a high degree of accuracy (, ), the average values ​​of mandibular measurements are larger for men than for women. (, , ) When studying the relationship between the gonial angle and sex, some authors observed that the mean value is larger in women than it is in men. (, ) Results obtained in several studies showed that the mandibular ramus and the coronoid height have a high discriminatory power for each sex, and can be used for sexual differentiation. (, , )

Usually methods of age estimation involve bone and dental analysis. (–) Bone measurements have been very useful in these cases, because they are fast and simple to apply. The mandibular ramus is generally intact when identification is required, and has therefore been the subject of much research to estimate age. () The gonial angle changes during life, at birth being rather obtuse, then decreasing as one grows up, and increasing again in old age. () During growth, the mandibular ramus increases more in height than the mandibular body in length, making the angle more acute. (, )

There is a scientific consensus that genetic influences skeletal features, and because of that the miscegenation presented in Brazil makes difficult to estimate age and sex based on international measurements and methodologies, since they were often for European populations. (, , ) It is very important that each population have their own measurements and formula to estimate age and sex, (, , ) validating or creating methods for Brazilian population. Thinking in this importance, the present study aimed to evaluate sex and age estimation in a Brazilian sample, using mandibular measurements.

Materials and methods

The study was approved by the Ethics Committee on Research under number CAAE: 50415215.0.0000.5419, fulfilling all the requirements of the Brazilian Resolution 466/12. ()

In total, 103 mandibles were measured, 53 female and 50 male, from the Bone Museum of the Department of Social Dentistry (UNICAMP, Piracicaba Dental School). This museum consists of a contemporary sample. The ancestry is difficult to estimate since the population of Brazil is heterogeneous and mixed. Only mandibles in good conditions were included: mandibles without fractures and with previous records about sex and age. Mandibular bones with presence of trauma, malformation, bone lesions or deformities that did not provide a complete mandible were excluded, as well as the ones which had no prior record of sex and age.

The following five measurements were used: coronoid height, gonial angle, bigonial distance, ramus height and maximum length. All these measurements were performed by two observers, without access of the data from the skeletons (age and sex), and without talking to each other. The measurements were repeated, for both observers, a month after the first measurement, in 30% of the sample.

The measurements were performed in some specific landmarks of the mandible, based on the book “Human Osteology” (). The measurement of the coronoid height process was performed using the Coronion (Cr) - the point at the tip of the coronoid process of the mandible, and the Gonion (Go) – point along the rounded posteroinferior corner of the mandible, between the ramus and the body, as can be seen in Figure 1-A. The gonial angle (GA) is formed by the intersection between the mandibular body and the mandibular ramus, as highlighted in Figure 1-B. The measurement of bigonial distance was taken between the two Gonion (Go) points on both sides, as observed in Figure 1-C. The measurement of the mandible ramus height was performed using the Condylion superior (Co) - the most superior point on the mandibular condyle, and the Gonion (Go), as can be seen in Figure 1-D. And the measurement of the mandible maximum length was performed using the Condylion (Co), and the Pogonion (Pg) – most anterior midline point of the chin of the mandible, represented in Figure 1-E.

Figure 1

Didactic scheme for the measurements performed: (A) coronoid height measurement (Cr-Go); (B) determination of the gonial angle (GA); (C) measurement of the bigonial distance (Go-Go); (D) measurement of ramus height (Co-Go); and (E) measurement of the mandible maximum length (Co-Pg).

All distances, with the exception of the gonial angle, were measured with the aid of a 0.1mm resolution digital caliper and +/- 0.1mm accuracy of the Western™ brand (Etilux™, São Paulo, Brazil) showing the measurement performed in millimeters. For the determination of the gonial angle, a 180o protractor was used in such a way that the protractor ruler touched the base of the mandible, and the degree marker was tangent to the posterior edge of the mandibular ramus, thus finding the value of the angle.

The mandibles were grouped according to age: G1: 0-20 years; G2: 21-40 years; G3: 41-60 years; G4: 61-80 years; G5: 80-100 years, regardless of sex. The sample size of the groups was: 4, 24, 29, 35, 11 samples, respectively.

For statistical analysis, the one-way ANOVA was performed, with a 5% significance level to verify if the mandibular measurements were related to age. The Student t-test was applied for comparisons between the sex samples. These tests were performed using Graphpad Prism Software version 7.01 for Windows (GraphPad™, La Jolla, California, USA). Both tests were run on unpaired samples. A logistic regression model was developed using 83 samples which were chosen randomly, and then, 20 measurements were used for testing this model. A stepwise automatic selection was used for choosing the best measurements for sex estimation. The intra-observer and inter-observer differences were evaluated for each measurement performed on the mandibles by the intra-class correlation coefficient (ICC), using the Kappa coefficient.

Results

The results for sex estimation are described in Table 1. All the measurements were statistically significant, since they had a value of p≤0.01 and all of them were larger in men, except for the gonial angle (GA), which was larger in women.

Table 1

Range in millimeters, of the values found in the measurements to estimate the sex with the respective value of p.

Measurements	Range (mm)		Value of p
	Female	Male
Co-Go	45.88 – 62.4	52.05 – 69.3	<0.0001
Cr-Go	42.4 – 70.75	51 – 69.75	<0.0001
GA	118 - 140	104 – 139.8	0.0064
Go-Go	81.65 – 94.9	87.85 – 111.2	<0.0001
Co-Pg	101.2 – 125.8	112.6 – 129.7	<0.0001

All the variables were analyzed to develop the logistic regression formula, however, the following two variables were better sex predictors: bigonial distance and mandibular ramus. Because of that, those variables were chosen to compose the following logistic regression formula (Figure 2). If the value for sex from the model is higher than 0.5, it is classified as male and if is lower than 0.5, it is classified as female. The accuracy of the proposed model was 90%.

Figure 2

Formula for sex estimation is developed using the following two measurements: bigonial distance and mandibular ramus. Values higher than 0.5 were classified as male, and if lower than 0.5 they were classified as female.

The mandibles grouped according to age with the mean and respective standard deviation of each of the measurements can be better analyzed in Table 2.

Table 2

Mean and standard deviation of the values of the mandibular measurements for the age estimation.

Measurements	Mean (Standard deviation)
	G1	G2	G3	G4	G5
Co-Go	54.37 (6.5)	58.84 (4.57)	53.9 (5.55)	54.48 (6.96)	52.78 (5.17)
Cr-Go	56.77 (8.38)	59.69 (6.96)	53.93 (3.48)	56.09 (8.43)	55.57 (4.51)
GA	118.5 (2,39)	122.4 (8.07)	129.6 (7.75)	126.2 (8.66)	131.1 (4.11)
Go-Go	83.68 (5.07)	94.11 (6.45)	91.7 (6.40)	94.53 (8.16)	89.72 (6.53)
Co-Pg	109 (8.17)	119.1 (4,93)	117.7 (7.32)	116.9 (6.60)	116 (7.27)

Groups G2 and G3 did have a statistically significant difference between them in three variables: Co-Co; Cr-Go; GA, with value of p≤0.05. Furthermore, it was observed that comparing the groups in the variable Co-Go, the G2 also had a statistically significant difference (p≤0.05) with G3, G4, G5. However, the groups G4 and G5 demonstrated a statistical relevance only regarding this variable. The measurements of bigonial distance (Go-Go) and maximum length of the mandible (Co-Pg), had no groups that were statistically relevant. There was an increase in the gonial angle (GA) in all the groups, except for G4, where there was a small decrease in value. The other measurements did not have an increase or decrease pattern in values from G1 to G5.

The agreement analysis of inter-observers had values greater than 0.89, and the intra-observer agreement analysis had values ranging from 0.85 to 0.99 (Table 3).

Table 3

Inter-observer and intra-observer correlation index.

Measurement	Inter-observer	Intra-observer 1	Intra-observer 2
Co-Go	0.93	0.89	0.94
Cr-Go	0.92	0.97	0.99
GA	0.89	0.97	0.97
Go-Go	0.92	0.97	0.99
Co-Pg	0.93	0.85	0.97

Discussion

The methodology used is useful for sex estimation with high accuracy, especially the measurements of Go-Go and Co-Go. It was possible to create a logistic regression model using those measurements that presented 90% accuracy for sex estimation in this sample. However, this methodology was not ideal for prediction the age; only Co-Go was effective for this purpose. The study had high values in the inter and intra-observer analysis, showing good accuracy in the measurements made for the same person or for more than one observer. This indicates that the methodology is reproductive, thus being useful in forensic cases.

There is a scientific consensus that males have measurements larger than females. This study shows that in mandibles measurements it is no different, therefore, agreeing with other cited authors (–, , –). Gillet et al. (), using a multislice computed tomography, found larger values for males than females. They concluded that the mandible is not the most dimorphic structure among other cephalic structures, but a useful tool when only this structure is available and not the entire skull. Another Brazilian study () was able to create discriminant functions with mandibular measurements, and they concluded that the mandible variables are important for sex estimation. However, GA is a very discrepant measurement, sometimes being larger in males (, , ), other times in females (, , , ), which was the case in our study. Belaldavar et al. () used only the gonial angle for sexual dimorphism, and their values were higher for females, although the sex prediction with this measurement was low and not useful for forensic purposes. The gonial angle ​​may be related to the size of the mandible, the smaller the ramus, more obtuse is the angle (), thus explaining why we found higher values for females than males.

It was observed that bigonial distance and mandibular ramus were the most dimorphic variables, being able to develop one logistic regression model with 90% accuracy. Vodanovic et al. () using 18 mandible variables, found that the length of the mandibular body; mandibular angle; and a minimum ramus breadth provided the best separation of the sexes, thus disagreeing with our study. However, the accuracy in their study was 92.06% using 9 variables, value close to presented in this study (90%), using only 2 variables. It is worth mentioning that Vodanovic et al. () observed that with less variables such as three or one, this accuracy is lower, 88.24% and 69.23% to 81.36% respectively.

In our study it was not possible to use the same size of samples in each age group, however, this was also the case in other studies (, , ) and in those, it was considered effective for age estimation. Gonial angle showed to increase with age (, ), but another study found that this measurement decreases with age (), which seems to be more logical, since the angle is inversely proportional with the measurement of the mandibular ramus which increases with age (, ). In this study, the bigonial distance did not have values with an increase or decrease pattern, agreeing with another study (). However, because of the increase of the entire mandible with age, this measurement was supposed to grow as observed in other studies (, ). Regarding the mandible maximum length, it decreased with older age. Even though no pattern was found, Rajkumari et al. () were able to create a multiple linear regression to predict the age using mandible measurements, some of them similar to ours. This would be explained precisely by the fact that if the gonial angle decreases with age, the mandibular ramus will be closer to the mandibular mentum, thus making the mandible maximum length smaller.

When the intra and inter-observer evaluations were performed, all values were high, showing that the study could be reproduced by anyone. In contrast, Gamba et al. (), despite using measurements that were the same or similar to those of the present study, found very contrasting values for intra and inter-observer evaluations. Their research, however, performed the measurements in computed tomography and not in bone pieces directly, with which can be inferred that it is better to perform measurements on bone structures rather than on images, even if those images are shown to be accurate.

One limitation found in this study was the sample: once it needed to be identified, having the correct sex and age and not just the estimation of those variables, we did not have many samples available to use. Because of that, it was not possible to have the same number of samples in age groups. This probably brings some complications to our results for age estimation. However, with a regular number of sample (, ) or with different size of samples in each group (, , ), it was considered effective for age estimation, with high degree of accuracy ().

Conclusion

It is difficult to estimate age and sex, which would be based on international measurements and methodologies in the Brazilian mixed population, since these methods are often based on European populations. It is important to develop specific formulas or to validate methodologies in Brazilian population, because it is normal in forensic practice to use some modified international methodologies, especially if they are metric ones. This research showed to be important, since the developed formula for sex estimation in Brazilian presented 90% accuracy, hence it is possible to estimate sex using mandibular measurements. However, the age estimation does not follow a pattern between the groups, therefore it is not recommended to use only these measurements for this purpose.