[Analysis of prevalence of self-reported hearing loss and associated factors: primary versus proxy informant]

The objective was to evaluate differences between prevalence rates for self-reported hearing loss and associated factors, obtained from responses by primary and proxy informants in a Population-Based Study on Human Communication Disorders (DCH-POP in Portuguese). This was a study on epidemiological methods using data from a cross-sectional household survey with a sample of 1,253 individuals from Southern Brazil. To verify differences between prevalence rates comparing primary informants and proxy informants, we used the chi-square or Fisher's exact test for categorical variables and Mann-Whitney for continuous variables. The log-binomial model was adjusted for hearing loss as the dependent variable, considering three datasets: the entire sample, only primary informants, and only proxy informants, estimating association by prevalence ratios. In the final models, only the independent variables age and dizziness were associated with hearing loss, independently of the dataset that was used. Proxy informants generally underestimated the prevalence rates for the target outcomes, when compared to primary informants.