[Quality analysis of self-reported information on sleep duration in schoolchildren in the Study of Cardiovascular Risk Factors in Adolescents (ERICA)].

The study aimed to assess sleep behavior in adolescents 12 to 17 years of age participating in ERICA (Study of Cardiovascular Risk Factors in Adolescents), based on sociodemographic characteristics, school shift, and type of school, and to describe the correction of inconsistent sleep times in a nationally representative study in Brazil. Data were collected in 2013 and 2014. Four questions were asked on habitual time for sleeping and waking on weekdays and weekends, with 24 possible answers, one for each hour of the day. Analysis of inconsistencies considered the distribution of frequencies of answers as to sleeping and waking times, in addition to compatibility with the school shift. Sleep duration during the week and on weekends was obtained by the difference between sleeping and waking times, and differences of ≤ 4 or ≥ 14 hours were excluded. Mean total sleep duration in seven days was calculated by the formula (weekday sleep duration x 5 + weekend sleep duration x 2)/7. The following groups were created: original data (answers that did not require correction), corrected data (inconsistent, but amenable to correction), and excluded data (inconsistent, and for which there was no criterion for correction). Correction recovered inconsistent information for 5,988 adolescents, 8% of the 74,589 participants. A total of 7,937 (10.6%) answers were excluded. Adolescents whose information was corrected or excluded were younger, predominantly males, from public schools, and from the North of Brazil. Correction minimized losses and lent greater consistency to the data treatment. The study contributes to the improvement of data collection tools in observational studies, lending transparency to the way of dealing with inherent limitations in the data collection method.