TY - JOUR
T1 - Discriminant analysis of mandibular measurements for the estimation of sex in a modern Brazilian sample
AU - Lopez-Capp, Thais Torralbo
AU - Rynn, Christopher
AU - Wilkinson, Caroline
AU - de Paiva, Luiz Airton Saavedra
AU - Michel-Crosato, Edgard
AU - Biazevic, Maria Gabriela Haye
N1 - This investigation was funded by the State of São Paulo’s Research Fund (FAPESP, process numbers 2014/13340-7, 2014/23727-6 and 2011/18577-7).
PY - 2018/5
Y1 - 2018/5
N2 - The present study aimed to evaluate the accuracy of mandibular measurements for sex determination in a Brazilian population. The sample was composed of 100 mandibles, of which 53 were female and 47 were male, and the average age was 57.03 years. The mandible measurement protocol was composed of 15 measurements, of which six were bilateral and nine were unique. Mandibles were directly measured using a digital caliper and a protractor. The descriptive analysis of the present study revealed higher mean values for male mandibles compared to those for female mandibles with the exception of the left mandibular angle. Among the 21 measures analyzed in this group, 15 were statistically significant (p < 0.05). Univariate discriminant analyses produced a mean percentage of correct predictions that varied between 49 and 79%. The association of variables increased the percentage of correct prediction of sex to vary from 76 to 86%. The receiver operator characteristic (ROC) curve analysis indicated that the best variable for estimating sex was bigonial breadth (BGB; area under the ROC curve (AUC) = 0.764) followed by the right maximum ramus height (MRHr; AUC = 0.763). A reference table for estimating sex in a Brazilian population using mandible measurements was developed based on the ROC curve analysis. Mandibular measures provide a simple and reliable method for sex discrimination in Brazilian adults due to the sexual dimorphism revealed by analysis of the metric variables and the satisfactory results demonstrated by discriminant formulas, ROC curve analysis, and the reference table.
AB - The present study aimed to evaluate the accuracy of mandibular measurements for sex determination in a Brazilian population. The sample was composed of 100 mandibles, of which 53 were female and 47 were male, and the average age was 57.03 years. The mandible measurement protocol was composed of 15 measurements, of which six were bilateral and nine were unique. Mandibles were directly measured using a digital caliper and a protractor. The descriptive analysis of the present study revealed higher mean values for male mandibles compared to those for female mandibles with the exception of the left mandibular angle. Among the 21 measures analyzed in this group, 15 were statistically significant (p < 0.05). Univariate discriminant analyses produced a mean percentage of correct predictions that varied between 49 and 79%. The association of variables increased the percentage of correct prediction of sex to vary from 76 to 86%. The receiver operator characteristic (ROC) curve analysis indicated that the best variable for estimating sex was bigonial breadth (BGB; area under the ROC curve (AUC) = 0.764) followed by the right maximum ramus height (MRHr; AUC = 0.763). A reference table for estimating sex in a Brazilian population using mandible measurements was developed based on the ROC curve analysis. Mandibular measures provide a simple and reliable method for sex discrimination in Brazilian adults due to the sexual dimorphism revealed by analysis of the metric variables and the satisfactory results demonstrated by discriminant formulas, ROC curve analysis, and the reference table.
KW - Forensic anthropology
KW - Forensic dentistry
KW - Human identification
KW - Mandible
KW - Sex determination by skeleton
UR - http://www.scopus.com/inward/record.url?scp=85029907645&partnerID=8YFLogxK
U2 - 10.1007/s00414-017-1681-8
DO - 10.1007/s00414-017-1681-8
M3 - Article
C2 - 28951960
SN - 0937-9827
VL - 132
SP - 843
EP - 851
JO - International Journal of Legal Medicine
JF - International Journal of Legal Medicine
IS - 3
ER -
