[Reliability of reported data from adolescent and their mothers in a health survey]

To estimate the reliability of data obtained from questionnaires applied to adolescents and their mothers in a health survey.

Data from population-based survey conducted with a one-stage random cluster area sampling of residents in the urban area of Salvador, Brazil, were assessed. The study sample comprised 82 pairs of adolescent workers aged 10 to 14 years and their mothers. After interviewing the adolescents, in a second visit, their mothers were invited to answer a similar questionnaire applied by a blinded interviewer. Respiratory and depressive symptoms, sleep pattern changes and distress in the neighborhood and at school were the variables analyzed. Reliability was measured using proportion of agreements and kappa index of agreement statistics.

Overall agreement and kappa index were low. Mothers have provided more positive answers than adolescents, resulting in a negative bias. Reliability was higher among boys for all answers when compared to girls, except for sleep patterns changes. For depressive symptoms, mother's education level was negatively associated with agreement for both genders.

Information from adolescents differs from that obtained from mothers. It is worth noticing that poor agreement does not impair validity, which needs to be examined in further studies.