Method:
The study was independently reviewed and approved by the Research Ethics Committee of the Piracicaba Dental School, University of Campinas. This cross sectional study was carried out in public schools (22) and Family Health Units (34). A probabilistic sample of 11,179 individuals was composed using a sampling error of 5%, DMF = 5.16 with SD = 4.54, response rate of 20% and a confidence level of 95%. Two calibrated examiners performed clinical exams (K>0.85 for all indices). The dependent variable studied was the number of decayed teeth. The independent variables were classified into individual (clinical, demographic, behavioural, socioeconomic, self-perceived oral health, access to dental services, quality of life) and contextual (Social Vulnerability Index of São Paulo State, Index of Social Exclusion of Piracicaba city, family education level of parents, monthly income, households with wastepipe and garbage collection). A multilevel regression model was estimated by PROC GLIMMIX ie "Generalized Linear Mixed Models". The model fit was assessed by -2 Res Log Likelihood with a significance level of 5%.
Result: In the first model the intra-class coefficient of correlation was r = 0.012 (i.e. the variation in the number of decayed teeth between the neighbourhoods represented approximately 1% of the total variation. For model 2, the variables prosthesis need (p <0.0001), residents in the same house having caries (p = 0.0118), number of people in the family (p = 0.0342) and risk of caries (p <0, 0001) were associated with number of decayed teeth. In model 3 the variable number of decayed teeth increases with the worsening social exclusion index (p = 0.0070).
Conclusion: Both individual and contextual variables influence the number of decayed teeth in adolescents of Piracicaba, SP, Brazil.