Adolescent food intake: Validation and calibration of a food frequency questionnaire in a complex sampling study

• Autor(a) principal: Barroso, Rosemary da Rocha Fonseca
• Data de Publicação: 2021
• Outros Autores: Aquino, Rosana, Amorim, Leila Denise Alves Ferreira
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Research, Society and Development
• DOI: 10.33448/rsd-v10i6.16075
• Texto Completo: https://rsdjournal.org/index.php/rsd/article/view/16075
• topic: Adolescentes; Cuestionario de frecuencia alimentaria (CFA); Validación; Calibración; Muestreo complejo.; Adolescentes; Questionário de frequência alimentar (QFA); Validação; Calibração; Amostragem complexa.; Adolescents; Food frequency questionnaire (FFQ); Validation; Calibration; Complex sampling.

Studies on dietary patterns depend on the validity of methods, such as the food frequency questionnaire (FFQ), to estimate long-term food intake, which can be overestimated, and the method needs to be validated. It is questioned whether validation studies nested in studies with complex sample design (CSD) should ignore data dependence. This study aims to validate the FFQ applied to school adolescents and estimate the calibration factors in a study with complex sampling. The validity of the FFQ is tested against dietary records (DR) applied to 85 adolescents in a study with CSD, with estimation of calibration factors (λ) for energy, nutrients, and dietary groups (DG). Intraclass correlation coefficient was calculated; mean differences were compared by modeling, and agreement analyzes were performed. The means between the methods differed for energy and nutrients (p <0.05), but four DG were similar. Limits of agreement were not acceptable for nutrients. Correlation coefficients of fiber, zinc, and fatty acids, adjusted for energy, ranged from 0.,16 to 0,.22 and for DG ranged from 0.07 to 0.36. The attenuation factors corroborated the differences (λ <0,40). The FFQ presented relative validity, especially for DG, and λ, which brought the FFQ estimates to FR, being relevant when nutrients are used in epidemiological analyzes. The adjustment of the analyzes by the CSD confirmed that the structure of data dependence must be considered, even in validation studies, and it should be avoided to treat such data as if they came from a simple random sample.