Test-retest reliability of self-perceived oral health in an adult population in Rio de Janeiro, Brazil

This study evaluated the test-retest reliability of a single item, namely self-perceived oral health, included in a questionnaire applied in a longitudinal study (the "Pro-Saude" or Pro-Health Study, 2001). The questionnaire was applied twice in a sample of 101 employees at a State university in Rio de Janeiro, Brazil. Self-perceived oral health was based on a single item, with five choices, ranging from "very good" to "very bad". Agreement was estimated using weighted kappa statistics and stratified according to gender, age, income, and schooling. Overall kappa coefficient was 0.80. Higher values were found in women (k = 0.84), young adults (k = 0.85), subjects with medium-level schooling (k = 0.86), and those earning more than six times the minimum wage (k = 0.91). Lower kappa estimates were obtained for individuals over 40 years of age (k = 0.67 for 40-49 years and k = 0.69 for >= 50). The reliability of the single item "self-perceived oral health" varied from "considerable" to "almost perfect" in all population strata, suggesting that this item can be used in future analyses in the scope of the "Pro-Saude" Study.