[Validity of an instrument for screening households at risk of food insecurity in different strata of the Brazilian population].

The absence of risk screening tools for food insecurity compromises the ability to assess, monitor, and provide immediate assistance to those in hunger, especially during emergencies such as the COVID-19 crisis. Hence, this study sought to test the validity of an instrument for Screening Households at Risk of Food Insecurity (TRIA) in different strata of the Brazilian population TRIA uses questions 2 and 4 of the Brazilian Food Insecurity Scale (EBIA), originally validated using data from the Brazilian National Survey of Demography and Health of Children and Women (PNDS 2006). In this study, using data from the Brazilian National Household Sample Survey (PNAD 2013), its reproducibility was tested by repeating the original combinatorial procedures, examining whether the parameters of sensitivity, specificity, accuracy, and positive predictive values (PPV) and negative values (NPV) would result in the same arrangement of questions. Moreover, convergent validity was analyzed by comparing the strength of association between food insecurity and dietary variables using two binomial regression models (TRIA x EBIA). Results indicated that the combination of questions 2 and 4 performed best among the population strata studied, and presented optimal convergent validity. PPV and NPV adjusted for food insecurity prevalence in states ranged from 42.8% (Santa Catarina) to 87.6% (Amazonas) and 95.8% (Amazonas) to 99.5% (Santa Catarina), respectively. In conclusion, besides being reproducible, TRIA presented excellent validity parameters, especially among vulnerable groups. It can thus be used in care practice and as an instrument of food and nutritional surveillance in Brazil.