Medidas de auto-relato para periodontite: Avaliação de acurácia, validação discriminante e associação com o senso de coerência

Abstract

Clinical examination, the gold standard for assessment of periodontal disease (PD), has a cost and difficult logistic, making epidemiological research in periodontics difficult and costly. An alternative is the use of self-report measures into a questionnaire that can determine the PD in the population. The aim of this study was evaluate the accuracy of individual self-report measures and multivariate models, test the internal and discriminant validity of these measures and their association with the sense of coherence (SOC). The final sample was composed of 284 subjects, aged 18 to 60 years, both genders, heterogeneous racial and socio economic group. In the clinical assessment was registered probing depth, clinical attachment level, bleeding on probing, visible plaque index and mobility. Different cutoffs for periodontitis definition, gingivitis and continuous measures for periodontal parameters were used to determine oral health outcomes. The self-report measures were answered through a questionnaire (Cyrino et al., 2011) and SOC through the short version of 13 items (Bonanato et al., 2009). Were determined the sensitivity (SS), specificity (ES) and area under ROC curve (AROC) for each measure. Multivariate predictive models for periodontitis were created using logistic regression analysis. The set of self-report measures was subjected to factor analysis which examined the internal reliability and canonical discriminant analysis validated the dimensions found.Were also logistic and liner regression models to assess the association between oral health outcomes and biological, socio-economic variables and SOC. The individual accuracy values varied considerably. Regarding SS, the values ranged from 7.40 (smoke) to 88.89 (flossing); ES, ranged from 13.89 (flossing) to 97.22 (tooth mobility). In the final logistic model, the significant variables were dental migration, bone loss and oral health health (SS=85.19, SP=66.67; AROC=79.78). The self report measures were separated into 3 domains: Oral health (frequency of flossing fl = 0.895), perception of symptoms and previous treatments (surgery fl = 0.618), risk factors (diabetes fl = 0.806). The perception oral health was associated with SOC (p=0.010), family income (p=0.001), smoke (p=0.004) and flossing (p=0.017). The perception of gum disease (p=0.001) and perception of periodontal disease (p=0.015) also were associated of SOC. Self-report measures have a useful SS and ES values, and it is possible create a new latent variable that represents a number of other variables that are correlated and thus summarize the questionnaire.