Impact of oral health literacy on self-reported missing data in epidemiological research.

Abstract

To investigate whether oral health literacy (OHL) impacts missing data obtained through self-reporting in oral health epidemiological research. A cross-sectional study was conducted with parents (n=344) of 4- to 5-year-old children randomly selected from public schools within the city of Curitiba, Brazil. Parental OHL was measured using the Brazilian version of the Rapid Estimate of Adult Literacy in Dentistry (BREALD-30). Parents answered a set of questionnaires comprising 88 items concerning sociodemographic and economic data, children's access to dental services, oral hygiene behaviour, diet and mealtime behaviour. The total number of unanswered items (TUI) and the number of unanswered items in each type of question (open-ended, dichotomous, multiple choice with up to 4 options and with 5-9 options) for each participant was compared across different levels of OHL (chi-squared, Mann-Whitney, Kruskal-Wallis and Spearman's correlation test). Multiple Poisson regression was used to estimate rate ratios (RR) of TUI between OHL scores and their respective 95% confidence interval (95% CI). Approximately one-third of studied parents (37%) exhibited low OHL (BREALD-30≤21). The prevalence of missing data in at least one item was 85.5%. Low OHL was associated with failing to respond open-ended items (P=0.003) and multiple-choice items with up to 4 (P=0.003) and between 5 and 9 options (P=0.030). There was a negative correlation between OHL scores and TUI (r=-0.195; P<0.001), as well as with the number of unanswered items in all types of questions (P<0.01), except dichotomous questions. Parents with lower OHL were more likely to show higher values of TUI (RR 0.95: 0.93-0.98), when adjusted by income and education. Participants with lower OHL were significantly more likely to fail to complete research questionnaires. The impact of OHL on missing data was greater with more complex types of items.