Abstract
The aim of this study was to analyze the validity of self-reported anthropometric measurements (weight and height) for classifying the nutritional status of Brazilian adults and elderly people using data from the 2019 National Health Survey (PNS). The PNS sample is made up of permanent private households from all of Brazil's federative units and this is a cross-sectional study in which 6,571 records were identified with measured and reported data, with no missing data for one variable being identified when in the presence of another. Validation was carried out with 6,381 data after removing atypical data. The variables used for stratification were: gender, age, race/color, schooling, and income, and the weighted Kappa Coefficient and the Intraclass Correlation Coefficient (ICC) were used to analyze agreement between the nutritional status categories. Accuracy was analyzed based on sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). For construct validity, a Poisson regression was performed for each outcome (measured and self-reported), with the independent variables "gender", "color/race", "schooling", and "family income". All the analyses showed positive results for validation. There was greater reproducibility among adults (18 to 59 years old) compared to the elderly and among men compared to women. This validation indicates a concrete possibility of carrying out an association of observational studies using reported nutritional status as the outcome variable, as an efficient strategy which could minimize the operational difficulties often encountered.