Validação diagnóstica de questionário para pesquisa epidemiológica da doença periodontal

Abstract

The self-reported measures have been used for epidemiological evaluation of chronic diseases in large populations, offering advantages in cost and logistics. Some questionnaires have been tested in different populations to assess periodontal disease, but results are still scarce. The focus of this study was to analyze the performance of a set of measures on self-reported periodontal disease to estimate the prevalence of periodontitis in a Brazilian sample. The study sample consisted of 284 individuals, aged 18-60 years, from multiethnic and socioeconomic heterogeneous group, in Belo Horizonte - Brazil. Each participant answered a questionnaire with 22 questions covering sociodemographic variables and self-perception of periodontal condition. On clinical examination probing depth (PD), clinical attachment level (CAL), bleeding on probing (BOP), mobility, visible plaque index and bleeding index were recorded. Periodontitis was defined as: a) mild/ moderate- the presence of at least two interproximal sites with NIC _ 4 mm or PS_ 5 mm in different teeth; b)severe- the presence of at least two interproximal sites with NIC _ 6mm and one site with with PS _ 5mm in different teeth(CDC, 2007). The prevalence of periodontitis determined by clinical examination was 28.87%. The final predictive logistic regression model with demographic variables and self-reported included three measures, age, family income and number of teeth, associated with periodontitis. The model for health and disease showed sensitivity (SS) = 35,44%, specificity (ES) = 90,82% and accuracy (AROC) = 77,42%, for the independent variable mild/moderate disease, SS = 18,37%, ES = 97,96% and AROC = 73,17%, for severe disease SS = 24,24%, ES = 97,96% and AROC= 82,95%. It was observed that the severity of disease improves the identification of affected subjects in the sample and the accuracy of the model is best represented in this case. Measures of selfreported periodontal disease showed moderate predictive value for the occurrence of disease and may be used in combination with socio-demographic characteristics for predicting periodontitis.