Modifications in the Consumption of Energy, Sugar, and Saturated Fat among the Mexican Adult Population: Simulation of the Effect When Replacing Processed Foods that Comply with a Front of Package Labeling System

A Mexican Committee of Nutrition Experts (MCNE) from the National Institute of Public Health (INSP), free from conflict of interest, established food content standards to place the front-of-package (FOP) logo on foods that meet these nutrition criteria. The objectives were to simulate the effect on nutrient intake in the Mexican adult population (20–59 years old) after replacing commonly consumed processed foods with those that meet the FOP nutrition-labeling criteria. Twenty-four hour dietary recalls were collected from the 2012 Mexican National Health and Nutrition Survey (n = 2164 adults). A food database from the INSP was used. Weighted medians and 25–75 inter-quartile ranges (IQR) of energy and nutrient intake were calculated for all subjects by sociodemographic characteristics before and after replacing foods. Significant decreases were observed in energy (−5.4%), saturated fatty acids (−18.9%), trans-fatty acids (−20%), total sugar (−36.8%) and sodium (−10.7%) intake and a significant increase in fiber intake (+15.5%) after replacing foods, using the MCNE nutrition criteria. Replacing commonly consumed processed foods in the diet with foods that meet the FOP nutrition-labeling criteria set by the MCNE can lead to improvements in energy and nutrient intake in the Mexican adult population.


Introduction
Excessive body weight is one of the main public health problems in Mexico [1,2]. According to the 2016 National Survey of Health and Nutrition (ENSANUT, by its acronym in Spanish), the combined prevalence of overweight and obesity affects 72.5% of the Mexican population, representing 49.4 million people [3]. Obesity is recognized as a risk factor for non-communicable chronic diseases, such as diabetes mellitus and cardiovascular disease, which are the main causes of mortality in the country [4]. This growing obesity epidemic can be attributed to physical inactivity and significant changes in diet, such as the rise in consumption of processed foods, which has led to an increased intake of total fat, saturated fat, sugar and sodium, and a decrease in fiber intake among the population [2,5]. Although individuals have some responsibility for the quality of their diet, the environment can strongly influence decision-making for food selection and consumption [6,7].
Food labeling has been recognized as a tool with the potential to influence food choices and dietary habits of consumers [8,9]. Considering this, the Mexican government launched a clear

ENSANUT 2012
The ENSANUT is a complex design survey, in which data from demographic characteristics, health, nutrition and access to health services were collected from a nationally representative sample. The general datasets analyzed during the current study are available from the Mexican National Health and Nutrition Survey repository [20]. The detailed datasets analyzed during the current study are also available from the corresponding author on reasonable request. This survey had a complex probabilistic design with state representation by urban (population ≥ 2500 inhabitants) and rural (population < 2500 inhabitants) localities. The sampling frame was integrated with information from the Census of Population and Housing 2005, disaggregated by Geo-statistic Basic Areas and the list of newly emerging localities in the 2010 Census. Data collection of the ENSANUT 2012 was conducted between October 2011 and May 2012. Information on 50,528 households was obtained in each of the 32 states in the country, with a response rate of 87%. A detailed description of the sampling procedures and survey methodology has already been published elsewhere [21].

Dietary Information
Individual food consumption was obtained by the previously validated 24HR method [22][23][24]. The dietetic information was collected by standardized staff on a sub sample of 10,886 subjects (about 11% of the surveyed population of ENSANUT 2012); only those between the ages of 20 and 59 years were included in this analysis (n = 2281). This method consists of recording, through an interview, all food consumed by the individual the preceding day to estimate the average consumption of the population. An (iterative) multiple step or five-step method was used in order to capture more accurately the interviewee food intake and avoid underestimation. This method is an adapted version of the 24HR of the National Cancer Institute of the United States [22]. This method consists of five steps: (1) quick list of food consumed; (2) foods often forgotten; (3) time and occasion of consumption; (4) review and detail of ingested foods; and (5) final scan to help individuals completely and accurately remember all the foods eaten during the previous 24 h. Adults aged 20 to 59 years who had one completed 24HR were included in the analysis. Pregnant and/or lactating women (n = 98) and subjects with aberrant data such as a BMI under 10 kg/m 2 or greater than 58 kg/m 2 (n = 2) were excluded from the analysis. Aberrant dietary data were analyzed on a case-by-case basis and corrected when possible. Cases of individual consumption below −3 standard deviations (SD) and 3 SD above the average energy consumption and those with aberrant or missing data were excluded from analysis (n = 117, 0.78%) [25,26]. The final test sample consisted of 2164 individuals.

INSP Food Composition Database
An extensive and diverse food database assembled by researchers from the INSP (unpublished) [27] was used. The nutrition content of food was obtained from various sources such as the United States Department of Agriculture (USDA) [28], the Mexican Equivalent Food System [29], the nutrition tables of the National Institute of Medical Sciences and Nutrition Salvador Zubiran [30], standardized recipes and food labels. This database was used to determine the average intake of energy and macronutrients per capita per day, as well as to identify foods that meet FOP nutrition-labeling criteria from the MCNE and that are candidates to replace commonly consumed processed foods that do not meet these criteria. To calculate the added sugar amount in food, the USDA database of added sugar content in foods was used [31]. Finally, the nutrition content of new foods was obtained directly from product labels.

FOP Nutrition-Labeling Criteria
Processed foods reported in the 24HR were evaluated with MCNE nutrition criteria and COFEPRIS criteria. The description of each set of criteria is presented below.

MCNE Nutrition Criteria
Foods were divided in 19 categories; cutoffs were set for the energy, saturated fat, trans fat, added sugar, sodium and fiber content per 100 g, 100 mL, or in % of total fat or energy of food. Cut-off points are different for each category (see Appendix A, Table A1).

COFEPRIS Nutrition Criteria
Foods were divided into 26 categories, and limits were set for energy, saturated fat, total sugar and sodium content per 100 g or 100 mL, per serving or in % of total fat or energy per product. Cut-off points are different for each category. These criteria excluded sugar-based products, like chocolate products, jam, jelly, syrup, honey and soft drinks (see Appendix B, Table A2). Such food products are not able to carry the COFEPRIS logo because they are not considered healthy and are not supposed to comply with the cut-off points for nutrients of concern.
Both nutrition criteria groups were considered to classify foods into two groups: those that do not comply with the criteria and those that do comply and are candidates to replace the first ones. To measure the food replacement three scenarios were calculated and compared for both MCNE and COFEPRIS nutrition criteria: Scenario 1: Represents the actual consumption. The median nutritional intake of the Mexican adult population based on data from a single 24HR per person from ENSANUT 2012. Scenario 2: Simulates the food replacement. The median nutrient intake when replacing processed foods reported in a 24HR from ENSANUT 2012 that do not comply with the MCNE nutrition criteria, with similar foods that do comply with the criteria. When it was not possible to find a replacement, foods were not replaced. This allowed the maximum potential change in intake, adhering as much as possible to the consumption of foods reported. Scenario 3: Simulates the food replacement with similar foods that did comply with criteria applying a correction factor with energy. Since foods that comply with criteria had less energy, the replacement could lead to a decrease in energy intake. For such a reason, the median intake of nutrients from Scenario 2 was corrected by the difference in energy density between the original intake and replacement, applying a correction factor. Therefore, when a food (for example, cereal bars, All Bran or Special K, 446.86 kcal/100 g) was replaced by a food with a lower energy density (in this case the Alpen light cereal bars, 290.5 kcal/100 g), a multiplication factor was applied (here 446.86/290.5 = 1.54) so that the total amount of energy consumed was the same as the amount of energy supplied by the food that had been replaced. This procedure was done for each product with the exception of sugar-sweetened beverages, bakery products and dairy, since correction would have resulted in unrealistic and very high amounts of consumption. Subsequently, the last two steps were repeated, but only replacing certain food groups or categories from the diet to see if significant differences were found in the intake of energy and nutrients after replacement using the MCNE nutrition criteria.

Statistical Analysis
Due to the skewed distribution, the energy and nutrient intake of the population were expressed in medians and weighted interquartile ranges (p25-p75). To evaluate statistically significant differences between the scenarios, energy and nutrient intake were log-transformed, and means were compared using linear regression models. A p value less than 0.05 was established to consider statistically significant differences. Statistical analyses were performed using Stata version 12.1, and the SVY module was used to adjust the sampling design of ENSANUT 2012 (College Station, TX, USA) [32].

Ethical Considerations
All participants from ENSANUT 2012 signed a pre-interview informed consent. This study protocol was reviewed and approved by the Ethics, Research and Biosafety Committee from the National Institute of Public Health.

Results
A total of 2164 adults aged 20-59 years were included in the analysis. The characteristics of the study sample are presented in Table 1. From the 695 foods and beverages that were identified in the 24HR, 354 items were classified as processed, excluding alcoholic beverages and dietary supplements. From the processed foods, 75.7% (n = 268) were classified as not meeting the MCNE nutrition criteria, and from these, only 44% (n = 118) could be replaced by a food that met the criteria; only those were used to calculate the nutrient intakes of Scenarios 2 and 3. From the total processed foods, only 45.6% (n = 167) were classified as not meeting the COFEPRIS nutrition criteria, and from these, 32.3% (n = 54) could be replaced by a food that met these nutrition criteria. Median intakes of energy and nutrients before and after the replacement of food for both nutrition criteria are presented in Table 2. At the national level, energy, saturated fat, trans fat, total sugar, sodium and fiber intake showed significant changes (p < 0.05) when commonly consumed foods were replaced via simulation with those that met the MCNE criteria. When data were corrected for energy intake, the difference in nutrition intake was still evident, but was less for energy, saturated fat, trans fat, total sugar and sodium intake, and greater for fiber intake. When replacing foods using COFEPRIS criteria, only significant decreases were observed for trans fat and sodium intake (p < 0.05). When the simulated nutrition intake of both criteria was compared, a lower intake of energy, saturated fat and sugar, as well as a higher intake of fiber, was observed for the MCNE criteria, compared to the COFEPRIS criteria. No significant difference in trans fat and sodium intake between the two criteria was observed.   Figure 1 shows the percentage change in the median intake of energy, saturated fat, trans fat, sodium, total sugars and fiber, if the Mexican population were to replace all unhealthy processed foods in their diet with those that meet MCNE criteria (with and without adjustment of energy using a multiplication factor). Similar to the description of Scenario 3 when corrected by energy, the median energy intake was significantly reduced by 5.4%. In addition, significant reductions in the median intake of saturated fat (−18.9%), trans fat (−20%), sugar (−36.8%) and sodium (−10.7%) were observed, and median fiber intake increased by (15.5%). The greatest reduction was observed in the total sugar intake, both nationally and by demographic characteristics. After testing the replacement of some food categories separately, only the replacement of sugar-sweetened beverages, using the MCNE criteria, produced a significant reduction in the sugar intake of the Mexican adult population (−28.3%). No significant differences in the rest of the nutrient and energy intakes were observed (data not shown).

Discussion
The FOP nutrition-labeling regulation is a strategy that may facilitate the adoption of healthy eating, by promoting better consumer decisions regarding the processed foods they consume, and thus helping to improve the nutrition and health status of the population [7,8]. This is the first study in Mexico that evaluates, through simulation, the potential impact of the replacement of commonly consumed foods in the diet, with products that meet FOP nutrition-labeling criteria, on the nutrition intake of the population. Some of the assumptions if consumers were to comply with the FOP nutrition-labeling criteria will be a significant decrease in the intake of dangerous nutrients such as, saturated fat, trans fat and sugar; furthermore, the largest and significant decrease will be that of sugar.
Based on the 24HR method, this analysis showed that if the Mexican adult population aged 20 to 59 years replaced processed foods commonly consumed in the diet with those that meet the MCEN nutrition criteria without changing other aspects of the diet, we would observe a significant decrease in the intake of energy, saturated fat, trans fat and total sugar. The excessive consumption of these critical nutrients for public health has been associated with a high prevalence of obesity and chronic non-communicable diseases such as diabetes mellitus, hypertension and dyslipidemias [33]. In contrast, we would observe a significant increase in fiber intake, of which adequate intake is associated with chronic disease reduction [34]. The subgroups that benefited the most from the replacement of food were men, adults 20-39 years old, those from urban locality, and the inhabitants of the northern region, possibly because of a greater consumption of processed foods than the rest. In contrast, those that least benefited from the replacement of food were those classified as underweight and normal weight, and those with low socioeconomic status; subgroups that may have had a lower consumption of processed foods compared to the rest.
Importantly, by using the MCNE criteria, the largest decrease was observed in sugar intake. This is because most of the replacement was made in sugar-sweetened beverages and the Mexican population, as has been observed and reported in previous studies, has a high consumption of these products [35]. When replacing only sugar-sweetened beverages for those with a very low sugar content, using the MCNE criteria, a significant decrease in the total sugar intake by the population was observed. This shows that by only replacing this group of beverages, a significant reduction in sugar intake could be observed, as well as a positive impact on the population health, since its excessive consumption has been associated with obesity and diabetes [35][36][37][38].
The results of this study, using the MCNE nutrition criteria, are consistent with those observed in similar studies, which performed a simulation of the replacement of processed foods by those that meet the criteria of the program "Choices International" (criteria on which the MCNE was based to establish their own). In 2009, Roodenburg et al., tested the potential impact of the replacement of processed foods in the diet with those that met the criteria of the "Choices International" program by modeling the nutrition intake of the Dutch adult population, observing substantial reductions in the intake of energy, saturated fat, trans fat, sugar and sodium, as well as an increase in fiber intake [39]. However, in this study, a greater reduction was observed in the intake of trans fat, compared with other nutrients, possibly because of a higher content in foods available on the Dutch market or food consumption patterns in the country. Later in 2011, a simulation of the replacement of processed foods in Greece, Spain, United States, Israel, China and South Africa was conducted. In this evaluation, three typical menus of each country based on population nutrient intake were assessed using the Choices program criteria. Three menus were then developed for each country replacing processed foods that did not meet the Choices criteria with those that did. This methodology showed that replacement may have a positive impact on reducing excess intake of nutrients with upper limits and increasing fiber intake among the adult population in the countries studied [40].
When simulation of replacing foods in the diet with those that meet the COFEPRIS nutrition criteria was performed, no significant decreases in the intake of energy, saturated fat and sugar were found, and no significant increase in fiber intake was found; only decreases in the intake of trans fats and sodium were observed. In addition, when decreases in the intake of energy, saturated fat, sugar and increase in fiber intake between MCNE and COFEPRIS criteria were compared, significant differences were found, finding a more favorable effect when using MCNE criteria. COFEPRIS nutrition criteria are based on those established by the food industry itself (which are not in accordance with the WHO dietary intake recommendations), being more permissible and less stringent than those of MCNE. Because of this, a greater number of products can be classified with a suitable nutrition profile, as observed in this analysis. With the MCNE criteria 25% of processed foods were classified as having an adequate nutrition content, while up to 56% of processed foods were classified in this category when COFEPRIS criteria were utilized. The latter causes consumers to perceive these products with a better nutrition quality and make unhealthy decisions regarding food choices. This would cause, as demonstrated in this study, negative positive impact on population nutrient intake and does not contribute to preventing and reversing obesity and non-communicable diseases in the country. Therefore, the COFEPRIS nutrition criteria are not considered an effective strategy for regulating FOP nutrition-labeling in Mexico.
Another important finding in this study was that a high proportion of processed foods consumed by Mexican adults (approx. 75%) does not meet the MCNE nutrition criteria, and less than half of these (44%) could be replaced with a healthier product that meets these criteria. For this reason, it is important not only to establish a more severe FOP nutrition-labeling regulation, but also to encourage the reformulation of processed foods by the industry. Some studies have found a successful reduction in nutrients of concern like sugar, saturated fat and sodium with stringent FOP labeling targets; also, studies showed that a logo perceived as credible and recognized helps in the choice of healthier foods [41][42][43]. At the same time, the population must be oriented towards a greater consumption of natural, traditional or low-level processing foods.

Strengths of the Study
Other studies have proved that modelling serves the purpose of accurately estimating the intake of nutrients when processed foods are replaced with healthier options [44][45][46]. These types of studies are often used to predict dietary changes before they are implemented in populations. Therefore, this information could be translated into nutrition policies and could potentially help policy makers to make better policies that will likely improve the health of the population and decrease diet-related non-communicable diseases [19].

Limitations of the Study
A limitation of the study is that we are not certain that people will replace foods with others that do comply with the criteria. This limitation is likely to happen in real life as people base their food choices and intakes not only on the FOP labeling but also on their activity and the whole environment [47].
Another limitation is that the dietary intake of the Mexican adult population in ENSANUT 2012 might be underestimated due to implications of the 24HR method [22], since it is difficult for a person to remember all foods and exact amounts consumed or the ingredients used in more complex food preparations. In addition, this questionnaire does not provide the variety or the exact amount of the usual food intake. This would result in a bias measurement of actual consumption. Furthermore, the database of the nutrient content of food may be inadequate because it uses different secondary sources that can provide, for some foods, a different nutrition content from the specific food consumed by the individual.

Conclusions
Based on simulations using data from the ENSANUT 2012, improvements can be observed in the energy and nutrient intake if Mexican adults replaced unhealthy processed foods commonly consumed in their diet with those that meet the MCNE nutrition criteria for FOP labeling. These results demonstrate the potential impact that could occur if the actual COFEPRIS nutrition criteria were adjusted towards the WHO recommendations, with emphasis on the category of sugar-sweetened beverages. This strategy, along with the establishment of clear food labeling, could be effective to regulate the FOP nutrition-labeling in Mexico, encouraging healthy decision-making regarding the purchase and consumption of foods, and thus improving the nutrient intake of the population.

Acknowledgments:
We would like to thank Katherine Endres for her support in the language editing of this manuscript.

Food Group Definition Nutritional Criteria
Breakfast cereal products All kinds of breakfast cereal products.
Saturated fat ≤13% energy Trans fat ≤0.1 g/100 g Sodium ≤500 mg/100 g Added sugar ≤20 g/100 g Fiber ≥5 g/100 g Milk and dairy products All kinds of milk and dairy products.

Appendix C Stratification for Analysis
The replacement effect was stratified nationally by sex (male or female), age group, BMI category, country region, locality and socioeconomic status only for substitution with MCNE nutrition criteria. Regarding the age, the sample was divided into two groups: 20 to 39 years old (young adult) and 40 to 59 years (middle adult). The WHO criteria were used to categorize the BMI into four groups: malnourished (<18.5 kg/m 2 ), normal BMI (18.5 to 24.9 kg/m 2 ), overweight (25.0-29.9 kg/m 2 ) and obese (≥30.0 kg/m 2 ) [48]. The locality was classified as rural if the population was <2500 inhabitants or urban if it had ≥2500 inhabitants. The country was divided into three regions: North, Central and Mexico City, and South. These three regions have common socio-economic and geographic characteristics and are grouped as follows: (A) North; (B) Centre; and (C) South. Finally, a socio-economic index was constructed using the Main Component Analysis with variables of the characteristics of living place, goods and services. Six variables (building materials of the floor and ceiling and possession of refrigerator, stove, television and computer) were selected. The first component that accounted for 41.1% of the total variability was selected as an index. Finally, socioeconomic status (SES) was classified into three categories using as cut-off points the percentiles 33 and 67 of the index (low, medium, high).