https://doi.org/10.37527/2023.73.S1
1Graduate Program in Public Health, University of Campinas, São Paulo, Brazil, 2Center for Epidemiological Studies in Nutrition and Health (NUPENS), University of São Paulo, São Paulo, Brazil, 3Center for Food Studies and Research (NEPA), University of Campinas, São Paulo, Brazil.
Introduction. Food advertising on labels is used by food industries as a strategy for promoting and selling products. Commercial databases (commonly used by the retail sector) can provide a significant volume of information on nutritional composition and advertising present in food and beverages sold in different countries. However, the reliability of this information needs to be evaluated before using in monitoring food labeling. Objectives. To analyze the reliability of data from a commercial database on advertising on food and beverage labels sold in Brazil between 2018-2021. Methods. The advertising data found on the labels were collected and organized by the company Mintel. To evaluate the reliability, trained researchers collected the advertising information using images of the front face of the same products present in the Mintel database, identified by barcodes. The agreement between the two sources of data was evaluated by adjusted Kappa (> 0.70 was satisfactory). Results. A total of 6.914 food and beverages were used in the concordance analysis. There was high agreement between the data (Kappa > 0.81) for 13 types of advertising: licensed characters, commemorative dates, sports events, Physical activity, famous athletes, price promotions, portion size, school elements, recommendations from Dietary Guidelines, family relationship, convenience, fun and adventure, and unconventional effects of the product. Six categories obtained substantial agreement (Kappa between 0.61-0.80): recipes and culinary tips, product superiority, brand or company appeal, taste, color, odor and texture, preparation convenience, and unconventional food format. A single category of advertising, the presence of QR code or website on the packaging, obtained poor agreement (Kappa = 0.09). Conclusions. The secondary advertising data showed mostly high agreement with the data collected by trained researchers. This demonstrates that these data, generally used by the retail sector, can also be used for monitoring food labeling when primary collection is not possible, which is often costly and time-consuming depending on the sample to be analyzed.
Keywords: secondary data analysis, food and beverage labels, advertising.