Abstract
BACKGROUND: Ultra-processed foods are recognized as indicators of an unhealthy diet in epidemiological studies. In addition to ultra-processed foods, the NOVA food classification system identifies three other groups with less processing. Unprocessed foods that, together with minimally processed foods (MPF), make NOVA group 1, NOVA group 2 is processed culinary ingredients, and NOVA group 3 is processed foods. OBJECTIVE: To assess the relative validity of the semi-quantitative food frequency questionnaire (SFFQ) to estimate the energy intake for each group NOVA classification system by comparing it with two 24 h-dietary-recall (24DRs) Mexican adults. METHODS: We analyzed dietary information from 226 adults included <60 and ≥60 years with complete SFFQ and two 24DRs from the National Health and Nutrition Survey 2012. We reported mean differences, Spearman correlation coefficients, intra-class correlations coefficients, Bland-Altman plots, and weighted kappa between measures. RESULTS: The percentage energy intake from unprocessed and minimally processed foods group, Spearman correlation coefficients was 0.54 in adults <60 years and 0.42 in adults ≥60 years, while ultra-processed foods group was 0.67 and 0.48, respectively. The intra-class correlation coefficients in the unprocessed and minimally processed foods group was 0.51 in adults <60 years and 0.46 in adults ≥60 years, and for the ultra-processed foods group were 0.71 and 0.50, respectively. Bland-Altman plots indicated reasonably consistent agreement for unprocessed and minimally processed foods group and ultra-processed foods group in adults <60 years and adults in the ≥60 age group. Weighted kappa was 0.45 in the ultra-processed foods group to adults <60 years and was 0.36-≥60 years. CONCLUSION: The SFFQ had acceptable validity to rank the percentage of energy intake from unprocessed and minimally processed foods group and ultra-processed foods group in Mexican adults, both in adults under 60 years and who were 60 years old or older.