This is a cross-sectional study in which we analyzed the use of NHC on the front-of-package of foods and beverages available in supermarkets in Mexico City during the period of January to March 2017. The selection of outlets was carried out randomly and included the biggest supermarket chains in Mexico (such as the Walmart group, La Comer, Soriana, and Chedraui) and other types of establishments. The selection of establishments was made according to the urban Basic Geostatistics Areas (AGEB by its acronym in Spanish; it is a geographical area delimited by streets, avenues, walkways, or any other feature whose land use is mainly living, industrial, or commercial usage and its population is greater than 2,500 inhabitants). The selection of AGEBs was determined according to the level of marginalization defined by the National Institute of Statistics and Geography (low, middle, and high) and population density (>20,000 inhabitants). The selection of establishments in each AGEB was carried out randomly and proportionate-to-size.
Data were collected from 136 establishments of different types: supermarkets (n=52), price club (n=8, a type of supermarket where membership is required, and which generally offers products contained in multi-packages for consumers and other establishments), wineries (n=32), convenience stores (n=20), mini supermarkets (n=17) and other types of establishments (n=7). The selected establishments included supermarket chains with more than 70% of the market share in Mexico .
To collect data, nutrition undergraduate students were trained by researchers from the National Institute of Public Health (INSP). Photographs were taken of each side of the package with a smartphone. When a product had a singular shape (cylinder, sphere or bag), the fieldworkers captured all relevant information.
It was ensured that the name of the product, the front of the package (including HNC), the type of package, GDA labelling, bar code, list of ingredients, nutrient facts table, and price were captured. The fieldworkers walked through all the aisles of the establishments to capture all products available (except those that were repeated in different stores). For each visit to a particular establishment, a checklist was generated with the registered products; if a product had already been registered in another establishment, the product was excluded. Bar codes were used to identify duplicate products. Each field worker collected information on the same category of products from all included establishments. Before taking photographs, we consulted the legal representatives of each supermarket chain and /or the manager in charge for authorization.
Relevant information on food and beverage packages (as detailed below) was captured in RedCaP by 8 research assistants who were previously trained and exported in electronic spreadsheets.
Health and nutrition claims
All claims that appeared on front-of-package were registered and classified according to the protocols and taxonomy of INFORMAS (Table 1) [13,23]. The classification included three categories and their subcategories (see Additional file 1): 1) nutrition claims (health-related ingredient claim, nutrient content claim and nutrient comparative), 2) health claims (general health claim, nutrient and other function claim and reduction of disease risk claim) and 3) Other claims (i.e. organic, gluten free). The format of each claim was also recorded as verbal, numerical or symbolic. Products containing combinations of numerical and verbal format were registered as numerical.
Food and beverages were grouped according to their degree of processing according to the NOVA classification system as unprocessed or minimally processed foods (such as fresh fruits or vegetables, whole grain cereals, plain milk, and seeds with no added ingredients), processed culinary ingredients (such as salt, sugar or oil), processed foods (such as canned fruits and vegetables, salted seeds or meat with salt for preservation), and ultra-processed foods (such as carbonated beverages, ready to eat foods like pizza or hamburgers, pastries and instant sauces) . The classification was made using the information available in the nutrition information panel and list of ingredients such as sugars, sodium, fat and added sweeteners, and other ingredients such as emulsifiers, preservatives, binders, humectants, stabilizers, brighteners, colorants, and flavorings.
Nutritional quality of the food products was calculated using the nutrient profile criteria of the Mexican FOPL regulation. Energy information was reported in calories, saturated fats, trans fats and free sugars (such as sugar, sucrose, fructose, corn syrup, honey) in grams per 100 g/mL, sodium in milligrams per 100g/mL, and use of non-sugar sweeteners reported in the list of ingredients. For products that require preparation for consumption, the reconstituted content was considered. This nutrient profile is applicable to products containing added sugar, sodium or fat (n=14,191), so unprocessed or minimally processed foods and processed culinary ingredients were automatically classified without excess of nutrients of concern (healthier). To establish the cut-off points, the calories per gram were calculated for free sugars (4 kcal), saturated fats (9 kcal) and trans fats (9 kcal). The following criteria were applied: Excess of: a) calories: ≥ 275 kcal per 100g for foods, ≥ 70 kcal per 100 mL for beverages or ≥ 8 kcal per 100 mL from free sugars for beverages; b) free sugars: ≥ 10% of total energy from free sugars; c) saturated fat: ≥10% of total energy from saturated fat; d) trans fat: ≥ 1% of total energy from trans fat; e) sodium: ≥1 mg of sodium per 1 kcal or ≥ 300 mg per 100g, ≥ 45 mg per 100 mL for non-caloric beverages; and, use of non-sugar sweeteners: reported in list of ingredients. For products without disaggregated content of free sugars on the package, we calculated free sugars according to the algorithms proposed by the PAHO nutrient profile model .
Scenarios for the use of health and nutrition claims
Two scenarios were applied to show the differences in the use of HNC before and after the implementation of the regulations.
The first scenario (current scenario) analyzes the use of HNC on packaged food products in Mexico in 2017. The second scenario (regulatory scenario) was established following the specifications of the third and final stage of the new FOPL regulation in Mexico. The regulation requires that products with warning labels (excessive in calories, free sugars, saturated fats, trans fats and sodium) or warning legends (contains non-sugar sweeteners or added caffeine) must not: a) use health claims or b) use nutrition claims , and c) display nutrition claims in the front of package.
For this scenario and according to the Mexican FOPL regulation, the following types of claims are considered to be included in the regulation: nutrient content claim, nutrient comparative claim, nutrient and other function claim and reduction of disease risk claim. Consequently, these types of claims were covered in the analyses for the second scenario (Table 5 and Table 6). For health-related ingredient claims and general health claims (as per the INFORMAS taxonomy), the Mexican regulations do not apply.
We used the nutrient profile criteria of the Mexican FOPL regulation to determine which products are still permitted to display HNC.
The analyses were performed using the statistical package STATA version 14. To verify the objectivity in the classification of claims, we performed a reliability test between two raters (Table 2). A random sample of products that contained claims (n=436) was taken and claims were classified according to the content and format by the two raters. The proportions of claims classified by category and subcategories were compared. The consistency in the reliability tests was determined using the Kappa Coefficient, values above 0.8 indicate very good consistency. The variables included in the analysis correspond to the categorical type, so they were presented as frequencies and percentages. The use of HNC was presented by food group, claims type (Table 4) and nutrient profile (Figure 1 and Table 5). Chi-square tests were used to determine differences in the proportion of products with HNC between the current scenario and the regulatory scenario (Table 5 and Table 6). Two logistic regression models were fitted to determine the Odds Ratio (OR) of using HNC in the current scenario and the regulatory scenario. Both models were adjusted for the components of the Mexican FOPL regulation and we report the results for each threshold of the nutrient profile separately. The analysis that included the regulatory scenario only considered products with claims (n = 8,746). For all tests, the value P<0.05 was considered as statistically significant.