Symmetry of dental caries: nearest-neighbor analysis with Monte Carlo simulation

Objective: To test the symmetry of caries lesions (CL) at site level using nearest neighbor analysis with Monte Carlo (MC) method. This new approach allows taking into account the unequal probability of sites to have the disease and to use all information available contrary to Hujoel et al (1994) method that only includes discordant pairs.

Methods: Data from the Third National Health and Examination Survey (1989-94) was used. The analysis was performed separately for each age group: 3-5, 12-15, 16-20, 21-25, 26-30 and 30-35 yrs-old and also stratified by the level of dental caries (bellow and above the median). A specific program was developed to perform the statistics with MC simulation: Ji = ‡”k ‡”t/2 aij di dj where i = homologous of j, K = number of individuals, t = number of teeth, aij = adjacent teeth of i, di= disease in i, j= disease in j. The null and alternative hypotheses were respectively that the CL were distributed at random and that CL were symmetric.

Results: The analyses suggested that CL independent of the age range were symmetric (p < 0.001). When the unequal probability of sites to have the disease was not considered the p-values increased, although non-significant finding (p= 0.08) was only observed for

3-5 years old group. The symmetry was also maintained independent of the stratification by level of dental caries, except for individuals with low levels of dental caries among 3-5 and 30-53 yrs-old. By contrast, when the same data was analyzed with Hujoel et al �es method, the conclusions were that the Cl were asymmetrically distributed, which may be explained by the exclusion of concordant pairs.

Conclusions: This cross-sectional study shows that CL is symmetrically distributed. However, this pattern needs confirmed in longitudinal studies using space-time interactions analysis. Grant-FAPESP 02/09544-9.