Assessing nutritional composition and ingredients of packaged foods in Brazil: an in-store census method for creating a comprehensive food label database.

Background: The consumption of ultra-processed packaged foods has surged worldwide with important health implications. It is pertinent to study the composition of packaged foods through information provided on labels. However, there is limited methodological discussion in the field. This study aimed at discussing methodological evolution and challenges of in-store census methods for assessing the composition of packaged foods, and characterizing a Brazilian food label database. Methods: The first Brazilian food label database reported in the scientific literature, based on data of in-store census method, was created in 2010 by the Nutrition in Foodservice Research Centre (NUPPRE at the Federal University of Santa Catarina). The in-store census method involves collecting primary data directly from the labels of packaged foods available for sale through retail food outlets. In 2020, the in-store census was carried out in partnership with the FoodSwitch Program. The NUPPRE/FoodSwitch Brazil 2020 database was developed in four steps: pre-data collection, data collection, data tabulation, and database construction and processing. The database was characterized by calculating the prevalence of foods per food group and foods that declared mandatory nutrition and health information on food labels according to Brazilian regulation. Results: The nutritional profile and ingredients of packaged foods was obtained from four food label censuses (2010, 2011, 2013 and 2020), supporting the Brazilian government on food labeling regulations and public policies. The experience prompted reflections about the methodological aspects of food label studies, and enabled improvements to the research process, such as a more accurate data collection, the inclusion of all packaged foods and beverages available for sale in the supermarket and the inclusion of more variables to the analysis. It is noteworthy the relevance of building nationwide food labeling databases. However, there are important challenges regarding the costs and efforts needed to maintain and update the data, especially in continental countries such as Brazil. The NUPPRE/FoodSwitch Brazil 2020 database consists of 7,828 packaged foods, 94% of the sampled brands sold nationwide. Most foods presented the mandatory information according to Brazilian regulation. Conclusion: This study proposed a series of methodological procedures to be carefully considered, designed, and executed during planning, data collection, data tabulation, and database processing. Greater rigor and detail are needed in the methods section of scientific articles, to aid replication.