Age estimation by pulp/tooth area ratio in canines: study of a Portuguese sample to test Cameriere's method.

Age estimation in adults is an important problem in both anthropological and forensic fields, and apposition of secondary dentine is often used as an indicator of age. In recent papers, Cameriere et al. studied the pulp/tooth area ratio of canines for this purpose. The present study examines the application of the pulp/tooth area ratio by peri-apical X-ray images as an age indicator in a Portuguese identified sample. The statistical model was then compared with results from an Italian identified sample, to establish whether a common regression model for both samples could be developed. The Portuguese sample consisted of 126 canines of male and 132 of female from subjects 20 to 84 years old, from the osteological collection of the Museum of Anthropology at Coimbra University. The Italian sample consisted of 114 canines of male and 86 of female from subjects 20 to 79 years old, analyzed in Cameriere et al. (2007), and came from the Frassetto osteological collection of Sassari (Sardinia), now housed in the Museum of Anthropology, Department of Experimental and Evolutionistic Biology, University of Bologna. Statistical analysis was performed in order to obtain multiple regression formulas for dental age calculation, with chronological age as dependent variable, and gender and pulp/tooth area ratio on upper (RA(u)) and lower canines (RA(l)) as independent variables. ANCOVA analysis showed that gender was not significant but that variables RA(u) and RA(l) were. The regression model for the Portuguese sample yielded the following equations: Age=101.3-556.68 RA(u) (upper canines) and Age=92.37-492.05 RA(l) (lower canines). Both models explained about 97% of total variance, and mean prediction errors were ME=2.37 years and 2.55 years, respectively. Comparisons between the equation referring to the Portuguese sample and the equivalent linear equations proposed by Cameriere et al. for the Italian sample did not reveal significant differences between the linear models, suggesting that a common regression model could be applied for both samples. The common regression model, describing age as a linear function of RA(u) and RA(l), yielded the following linear regression formulas: Age=100.598-544.433 RA(u); Age=91.362-480.901 RA(l), and explained 86% and 93% of total variance, respectively. Mean prediction errors were ME=2.68 years and 2.73 years, respectively.