Using commercial dataset for extracting information on nutrition composition for policy evaluation purposes

Authors

Beatriz Silva Nunes, Graduate Program in Collective Health, School of Medical Sciences, https://ror.org/04wffgt70University of Campinas, Campinas, Brazil.
Camila Aparecida Borges, Center for Epidemiological Studies in Nutrition and Health, University of São Paulo (USP), São Paulo, Brazil.
Mariana Fagundes Grilo, Center for Food Studies and Research, https://ror.org/04wffgt70University of Campinas (UNICAMP), Campinas, Brazil.
Ana Clara Duran, Graduate Program in Collective Health, School of Medical Sciences, https://ror.org/04wffgt70University of Campinas, Campinas, Brazil.

Document Type

Journal Article

Publication Date

12-23-2025

Journal

Public health nutrition

Volume

28

Issue

1

DOI

10.1017/S1368980025101456

Keywords

Composition nutritional; Evaluating; Food labelling; Monitoring; Nutrition policy; Secondary data analysis

Abstract

OBJECTIVE: This study assessed the suitability of nutritional composition data from a commercial dataset for policy evaluation in Brazil. DESIGN: We compared the proportions of packaged foods and beverages, classified according to the Nova food classification and the nutritional composition of matched products using data from a commercial database of food labels (Mintel-Global New Products Database (GNPD)) and the Brazilian Food Labels Database (BFLD), collected in 2017 as a 'gold standard.' We evaluated the agreement between the two datasets using paired t tests, Wilcoxon-Mann-Whitney test and the Intraclass Correlation Coefficient (ICC) for energy, carbohydrates, total sugars, proteins, total fats, saturated fats, trans-fats, sodium and fiber. SETTING: Brazil. PARTICIPANTS: Totally, 11 434 packaged foods and beverages collected in 2017 provided by BFLD and 67 042 packaged foods and beverages launched from 2001 to 2017 provided by Mintel-GNPD. RESULTS: The proportions of ultra-processed foods (UPF) were similar in both datasets. Paired products exhibited an excellent correlation (ICC > 0·80), with no statistically significant difference in the mean values (P ≥ 0·05) of most nutrients analysed. Discrepancies in fibre and fat content were noted in some UPF subcategories, including sweet biscuits, ice cream, candies, dairy beverages, sauces and condiments. CONCLUSION: The Mintel-GNPD dataset closely aligns with the BFLD in UPF distribution and shows a similar nutritional composition to a sample of matched foods available for purchase in stores, indicating its potential contribution to monitoring and evaluating food labelling policies in Brazil and in studies of food and beverages composition in food retail through the verification of policy compliance.

APA Citation

Nunes, Beatriz Silva; Borges, Camila Aparecida; Grilo, Mariana Fagundes; and Duran, Ana Clara, "Using commercial dataset for extracting information on nutrition composition for policy evaluation purposes" (2025). GW Authored Works. Paper 8370.
https://hsrc.himmelfarb.gwu.edu/gwhpubs/8370

Department

Public Health Student Works