Perceived diet and lifestyle changes during connement for Covid-19 pandemic in Mexico: ESCAN-COVID19Mx Survey

Background: Connement measures that were put in place in Mexico to reduce the transmission of the SARS-CoV-2 virus can affect individual’s lifestyle and well-being; Therefore, the aim of the study was to assess perceived changes in diet quality, emotional eating, physical activity and lifestyle, in a group of Mexican adults before and during the COVID-19 connement. Methods: 8289 adults answered an online questionnaire between April and May 2020. Data about diet quality, emotional eating, physical activity and sociodemographic characteristics, weight and height were collected. Before and after connement differences by sociodemographic characteristics were assessed with Wilcoxon, Anova and linear regression analyses. Results: Most participants were women (80%) between 18 and 38 years old (70%), with a low degree of marginalization (82.8%) and a high educational level (84.2%); 53.1% had a normal weight and 31.4% were overweight. Half (46.8%) of the participants perceived a change in the quality of their diet. Lifestyle changes were present among some of the participants, 6.1% stopped smoking, 12.1% stopped consuming alcohol, 53.3% sleep later, 9% became more sedentary, and increased their screen (43%), sitting and lying down time (81.6%). Conclusion: Mexicans staying at home during the COVID-19 connement, that completed the ESCAN-COVID19Mx Survey, perceived positive changes in the quality of their diet, and smoking and alcohol consumption, but negative changes in the level of physical activity and quality of sleep.


Background
Around the world, strict con nement measures were put in place to reduce the transmission of the SARS-CoV-2 virus (Ebrahim et al., 2020). In Mexico, con nement measures included: social distancing, social isolation mainly for the vulnerable population, cancellation of mass events and gatherings, and school closings. Hygiene measures such as social distancing practices, avoiding touching the face, frequent handwashing, and the use proper sneezing and coughing etiquette were also promoted (Diario O cial de la Federación, 2020). The effect that these measures could have on the wellbeing of the population remain unknown. Social distancing has proved to have a positive impact on the transmission of other contagious diseases. For example, it has shown to reduce the transmission of in uenza cases within a range of 1-50%, with the closure of schools and work at home being the two strategies with the best results (Rashid et al., 2015). However, these measures could also have a negative impact on the food security of the population, especially among the most vulnerable groups (Dunn et al., 2020;Power et al., 2020).
Con nement caused changes in human behaviours, services, economy, and food systems (Huff et al., 2015;Hobbs, 2020). Initial surveys from COVID-19 in the US showed that during the COVID-19 pandemic, participants were worried about their health, and had made some lifestyle changes. For example, participants stopped going to the gym (50%) and stocked up on food (74%) (Nelson et al., 2020). During the COVID-19 pandemic, panic buying and hoarding of food and essential products have been registered, causing shortages of some products (Hobbs, 2020;Nelson et al., 2020;Romano Gutiérrez, 2020). Stockpiled foods included bottled water, milk, rice, frozen food dough and canned foods (Hobbs, 2020;Romano Gutiérrez, 2020;Siche, 2020).
People living in social isolation may be more likely to initiate harmful behaviours such as smoking, excessive alcohol consumption and overeating as a mechanism of psychological relief (Valtorta et al., 2016;Leigh-Hunt et al., 2017). Despite the fact participants from these studies were adults with little social interaction due to causes other than the epidemic, it is possible that the social distancing and anxiety caused by the COVID-19 state of emergency (Losada-Baltar et al., 2020), have similar effects or are more accentuated on the behaviours of the population.
Results from a telephone survey conducted by the Mexican Institute of Public Health (ENSARS-CoV-2 ) in which data from 1000 Mexicans was collected between May 11th and 30th 2020, showed participants had less diversity in their diet, reduced their level of physical activity, and increased sedentary behaviours and sleeping hours (Shamah-Levy et al., 2020). Therefore, it becomes relevant to document the effects of the pandemic on the population's eating habits and lifestyle, since an individual's nutritional status and poor diet quality may impact on comorbidities and subsequent complications.
Thus, the aim of this study is to describe diet and lifestyle changes during con nement in COVID-19 pandemic in Mexican adults (≥ 18 years) according to sociodemographic characteristics. Our results could help guide actions and strategies on nutrition and health after the contingency.

Materials And Methods
The aim of this study was to describe diet and lifestyle changes during con nement in COVID-19 pandemic in Mexican adults (≥ 18 years) according to sociodemographic characteristics.

Study design and sample
For this cross-sectional study, we structured the ESCAN-COVID19Mx survey. The survey was intended to be answered by: a) Mexican adults (aged ≥ 18 years); b) of any sex; c) residing in Mexico between March 17 and May 30, 2020); d) that con rmed staying at home during the con nement period; e) with access to an electronic device and to the internet; and f) read the informed consent included in the online instrument and agreed to participate. The ESCAN-COVID19Mx was designed to be self-reported through an electronic platform (Google forms ©), where all responses were compulsory. The study was approved by the Ethics and Research Ethics committees of Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán reg GAS-3376-20-20-1 (INCMNSZ) and the Instituto Nacional de Salud Pública CI 1666 (INSP).
The questionnaire was distributed using a snowball sampling technique through social media (WhatsApp, Telegram, Facebook, Twitter) and email. Sample size was estimated at 2401 participants, hypothesising that a proportion of 50% of the population con ned at home would present negative changes in their diet quality, with a margin of error of 2%.

ESCAN-COVID19Mx Survey
The ESCAN-COVID19Mx survey has ve components divided as follows: I) Sociodemographic characteristics; II) Diet quality III); Emotional Eating (over and under eating from negative emotions; IV) Physical activity and sedentary behaviour; and V) Lifestyle factors (Quality of life). All the scales used as reference were validated in the Mexican population so that they were self-reported in the ESCAN-COVID19Mx survey and could be answered in less than 15 minutes.
Sociodemographic characteristics: We collected information about the participants´ sex, age, place of residence, education, marital status, number of children in the household, weight and height. AMAI system 2018 (Asociación Mexicana de Agencias de Inteligencia de Mercado y Opinión Revisión Nivel Socioeconómico, 2018) was used to assess participants' socioeconomic classi cation, and level of marginalisation was obtained through the participant's postal code.
Diet quality. The Mini-Survey to Evaluate the Quality of Food Intake (Mini-ECCA) was included to assess diet quality during and before the COVID-19 con nement. The Mini-ECCA has been validated in healthy Mexican adults (Bernal-Orozco et al., 2018). Participants who perceived changes in their diet during the pandemic were asked to answer the Mini-ECCA twice; once in reference to their diet prior to con nement and the second time in reference to their diet during con nement. Diet Quality Index (DQI). The Mini-ECCA was used to obtain an index with a value of 0-12 according to their consumption of water, vegetables, sh, fruit, legumes, sweets, sugar-sweetened beverages, meats, fats, cereal, junk food and eating away from home. Higher values represent a better diet quality.
Emotional eating. We used the Spanish version of the Adult Eating Behaviour Questionnaire (AEBQ-Esp) validated in Mexican adults from the original AEBQ (Hunot et al., 2016) to assess emotional eating. We did not assess perceived emotional eating prior to con nement, as appetitive traits show continuity and stability through time (Ashcroft et al., 2008). Two of the seven subscales from the AEBQ-Esp were used: (1) emotional overeating and (2) emotional undereating. Each subscale included 5 items each: "Eat more" or "Eat less" when I am angry, worried, upset, anxious and/or irritated. Each question was answered using a 5-point Likert Scale: "Strongly disagree", "disagree", "neither agree, nor disagree", "agree" and "strongly agree" (Hunot-Alexander et al., 2020).
Physical activity and sedentary behaviour. This component was constituted by four questions about the participant's physical activity and sedentary behaviour during and before the contingency period. The questions about physical activity inquire on the days per week that the participant performed 30 minutes or more of intense physical activity. To measure sedentary behaviour, the participants were asked about the hours per day they remain sitting or lying down (Beeken et al., 2012(Beeken et al., , 2017. The Physical Activity and Sedentary Behaviour Index (PASBI) was constructed with the following two scales: a) Vigorous physical activity: participants were assigned a score (0-12) based on the times per week they completed 30 minutes or more of intense physical activity. These scores were distributed as follows: "Did not do any vigorous physical activity" 0 points, "1-2 times per week" 3 points, "3-4 times per week" 6 points, "5-6 times per week" 9 points, "Every day or several times per day" 12 points. b) Sedentary behaviour: Based on the time per day the participant was sitting down or lying down the participants were. The scores for sedentary behaviour were given as follows: "0 to 2 hours" 12 points, "3 to 6 hours" 8 points, "7 to 10 hours" 4 points, "11 hours or more" 0 points. The PASBI is the result of adding both items, producing an ordinal scale of 18 levels with scores ranging from 0 to 24.
Lifestyle factors (Quality of life). For these questions, appearance and content validity was calculated among the research group and three rounds of evaluation were performed per question by consensus.
Two rounds of the pilot testing were conducted in 10 and 15 individuals each, with the last approved version in consensus included in the survey. The section was made up of the following aspects: food expenses, health information sources, compliance with con nement, changes in body weight, screen time, smoking, alcohol drinking and sleeping habits. A Lifestyle quality Index (LSQI) was calculated which included the following six aspects: Smoking, alcohol consumption and sleeping habits, LSQI and the two components of the PASBI. For the LSQI and the PASBI components, their respective 12-point scores were used. To obtain the smoking, alcohol and sleeping habits scales, a score of 0 to 2 was assigned to the responses issued in each question, where 0 was considered the least recommended behaviour, and a maximum value of 2 to the recommended behaviour. The total score for each component was added and weighted so that each provided the same contribution to the index (12 points). The score of the LSQI obtained was from 4 to 72 on the scale that evaluated lifestyle retrospectively; and from 3 to 72 on the scale that evaluated the lifestyle quality during the contingency. A higher score was considered a better lifestyle quality.

Statistical Analyses
Sociodemographic characteristics of the participants were described using indicators of central tendency and dispersion for quantitative variables while frequencies and percentages were used for qualitative variables. The DQI, PASBI and LSQI were described with medians and interquartile ranges (IQR 25-75). To investigate whether there were differences in the abovementioned indices before and during con nement, by sociodemographic characteristics and weight category, Wilcoxon's signed ranks test and chi 2 were used as appropriate. Means for each emotional eating subscale were calculated and linear regressions for both emotional over and under eating with BMI were calculated. Statistical signi cance was de ned as p < 0.05. All data analyses and statistical tests were performed using R (v.3.6.1).

Sample characteristics
Respondents (n = 8289) fully completed the ESCAN-Covid19MX survey between April 13th and May 16th. The majority of participants were women (80%), participants where within an age range of 18 to 38 years (70%), living in areas with very low levels of marginalization (82.8%) and higher education level (84.2%).

Page 7/23
Based on self-reported weight and height, half of the respondents (53.1%) had a BMI between 18 and 24.9 kg/m 2 , a third had a BMI between 25 and 29.9 kg/m 2 (31.4%), and 13.3% had a BMI ≥ 30 kg/m2. Figure 1 shows the response rate by place of residence.
As shown in Table 1, there were signi cant differences in age, education, place of residence and weight status of participants by gender. A higher proportion of women had low and normal weight, whereas a higher proportion of men had overweight or obesity (p < 0.005). Nearly half of the participants (46.8%) perceived a change in the quality of their diet. There was an increase in the consumption of homemade foods (28.4%). The median of the Diet quality Index (DQI) improved by 3 points (-1 to 2, p < 0.001) for the full sample. The DQI was higher during con nement in all groups regardless of education level, marginalization level or place of residence (p < 0.001). Moreover, there was a positive association between BMI and the DQI (p < 0.001). Table 2 shows the changes in the DQI by age, nutritional status, schooling, degree of marginalization and geographic region of residence.  Figure 2 shows the change rate in the full sample in the DQI before and during con nement, suggesting a shift to better diet quality.

Emotional Eating
Our results showed that 70% of the respondents perceived no changes in their emotional eating during con nement. Results from the AEBQ-Esp Scale showed that the emotional overeating subscale mean was 2.63 (± 0.88) whereas the mean of emotional undereating subscale was 2.51 (± 0.97). Linear regressions between emotional overeating and emotional undereating with BMI showed no signi cant relationships between either subscales and BMI (ß=-0.937, p = 0.854 and ß=-1.19, p = 0.806 respectively).

Physical Activity and Sedentary Behaviour Index (PASBI)
Participants who did not engage in any intense physical activity prior to con nement reported engaging on more intense physical activity during con nement, as the rate of people engaging on intense physical activity increased from 31.2-39.9% (p < 0.001). However, those who reported performing intense physical activity 3-4 times/week prior to con nement decreased from 25 to 18.5% (p < 0.001). Moreover, Table 3 shows there was a statistically signi cant reduction in the PASBI score during con nement among all age groups, education levels, marginalization, and place of residence. The sources used by the participants to keep themselves informed about the COVID-19 pandemic where social media (78.8%), leaders and health professionals (73%) and health authorities, government, and health institutions (78.8%). Regarding money spent on food, 38.9% of the participants perceived having a higher food expenditure than usual during con nement, but 47.8% reported having a lower family income.
Nearly half of the households (44%) included minors.
Observing physical activity results per week in Table 3, the one performed most frequently among the population was 3 to 4 times per week before and during con nement, but even this category of weekly exercise frequency decreased by 6.3%, being more accentuated in men (6.5%) than in women. In women, weekly exercise was more affected in the category of 5 to 6 times and several times a day it was decreased (-3.2 and − 0.3%, respectively).
Regarding lifestyle factors, majority of participants reported that during con nement they would leave their homes only if it was necessary (85.5%), 43% increased their time in front of screens, 81.6% spent more hours sitting or lying down, and 42.3% consumed food in front of screens once a day. The lifestyle quality index (LSQI) for the general population prior to con nement was 42 points ] and 41 points during the con nement (IQR 34-47). As a possible consequence of the changes in the quality of diet and physical activity, the LSQI was also affected.

Discussion
The ESCAN-COVID19Mx aimed to evaluate the perceived changes in the quality of diet, lifestyle, emotional eating and physical activity of Mexican adults during and prior the COVID-19 pandemic. A total of 8289 Mexicans who reported staying at home during the con nement period, completed the online questionnaire.
Results from our study showed an improvement in the quality of the diet among those who were in an age range of 18 to 59 years, had a normal weight or overweight, higher education level, lived in Mexico City, the north and centre of the country, and who lived in areas of low marginalization.
These results were unexpected since, when registering the hoarding of frozen and canned foods at the beginning of the pandemic (Hobbs, 2020;Romano Gutiérrez, 2020;Siche, 2020), we anticipated a possible increase in the consumption of ultra-processed foods resulting in a lower diet quality of the diet.
However, a third of the respondents had greater access to homemade foods and greater consumption of fresh foods during the stage of con nement.
In parallel with our study, results from a survey by National Institute of Public Health in Mexico (ENSARS-CoV-2 ) showed that during the COVID-19 con nement, their participants had a higher consumption of cereals, fresh fruits and vegetables, as well as meat, sh and shell sh, eggs, dairy products, legumes, oils, condiments, coffee and tea. Also, that they increased their consumption of sweetened beverages, sweets and snacks. Unlike our study, their report describes a diet diversity index with an average score of 13.5 (0-18) (Shamah-Levy et al., 2020). In France, results of the Nutri-Net cohort showed a reduction in the consumption of fresh foods and an increase in foods high in sugar at the beginning of the COVID-19 pandemic (Deschasaux-Tanguy et al., 2020). On the contrary, the COVIDiet study in Spain, showed that during con nement, participants showed greater adherence to the Mediterranean diet, which is characterized by the consumption of olive oil, fruits and vegetables and evidence suggests is optimal to prevent non-communicable diseases and preserve good health (Rodríguez-Pérez et al., 2020).
Recent studies in other countries have also investigated emotional eating behaviours during the COVID-19 pandemic reaching con icting results (Deschasaux-Tanguy et al., 2020; Rodríguez-Pérez et al., 2020; Romeo-Arroyo, Mora and Vázquez-Araújo, 2020; Scarmozzino and Visioli, 2020;Sidor and Rzymski, 2020). However, to our knowledge, no study has assessed eating behaviour based on emotional over and under eating, that is, the way negative emotions impact on food consumption and their possible impact on body weight (Hunot et al., 2016;Hunot-Alexander et al., 2020). Using the AEBQ-Esp we measured the participants' emotional overeating and emotional undereating during the COVID-19 con nement. During the validation of the AEBQ tool in Mexicans prior to con nement the mean of eating more through emotions was 2.54 (± 0.87) and eating less through emotions was 2.79 (± 0.88) (Hunot-Alexander et al., 2020). Results from this study showed no associations with either emotional over or undereating and BMI. These results differ with other publications that show that emotional overeating is positively associated with BMI in adult populations (Hunot et al., 2016;Mallan et al., 2017;Hunot-Alexander et al., 2020). Results from our study suggest that con nement has negatively impacted on the physical activity of the population. Our results are similar to ndings from studies in other countries (Deschasaux-Tanguy et al., 2020;Lesser and Nienhuis, 2020) in which they have reported a reduction in physical activity and an increase in sedentary time during the COVID-19 con nement. This reduction in physical activity was expected due to the closure of gyms and recreation centres, as people do less active commuting. It could be possible that reductions in physical activity, could have been due to an excess of activities whilst at home, such as looking after children, home schooling, plus individual home o ce work. This could partly explain the results obtained in this study and why women reduced their exercise frequency. Reduction in physical activity during con nement is alarming as prior to the pandemic, less than half (42%) of the adults in Mexico, referred doing physical-sports practice in their free time (INEGI, 2020). This reduction could have a long-term impact on lifestyle and could lead to the appearance of type 2 diabetes and metabolic syndrome (Rezende et al., 2014). In this study, we found that sedentary lifestyle increased from 31.3 to 40%, while in the ENSARS-CoV-2 survey, sedentary lifestyle rose from 39.5 to 45% (Shamah-Levy et al., 2020). Again, and as seen with diet quality, the physical activity and sedentary lifestyle indices were different regardless of the degree of marginalization and the region of the country (p < 0.001).
The LSQI constructed with information on smoking, alcohol consumption, diet quality, physical activity and sedentary behaviour showed a deterioration in lifestyle factors in both young adults and middle age adults during the COVID-19 con nement, but not in the elderly. However, it is striking that the change was not signi cant in participants with BMI < 18 kg/m 2 (p = 0.298), ≥ 30 kg/m 2 (p = 0.609), with a basic education level (p = 0.081), medium, low to very low degree of marginalization (p = 0.217, 0.855 and 0.825, respectively) and those who lived in the southern region of Mexico (p = 0.096). Studies prior to the COVID-19 pandemic showed that people in social isolation were likely to initiate or increase behaviours such as smoking or alcohol consumption (Leigh-Hunt et al., 2017). Nevertheless, to our knowledge, none of the studies in the context of the COVID-19 pandemic have shown a signi cant increase in tobacco and alcohol consumption. A study in Spain reported a decrease in the consumption of alcoholic beverages by 57.3% (Rodríguez-Pérez et al., 2020); while in Italy a study showed a signi cant reduction in smoking (Di Renzo et al., 2020). In our study, some of the participants reported having changes in their smoking and alcohol habits, both positive and negative; however, for the majority these behaviours remained constant as smoking increased by 3%, while alcohol consumption rose to 10.2%. This situation in in contrast with the ndings of the ENSARS-CoV-2 where cigarette consumption increased to 21% and alcohol to 12% (Shamah-Levy et al., 2020).
It is possible that the reduction of tobacco and alcohol consumption may have been for health concerns or that participants solely engaged in these behaviours socially. More research is needed to understand the factors that in uenced these behaviour changes and whether these changes were permanent.
Sleeping habits were affected in more than a quarter of the participants. Our results were similar to those in the ENSARS-CoV-2 (Shamah-Levy et al., 2020). It is worth noting that in our study 45% of the participants had a BMI above 25 kg/m 2 , when it is known that the prevalence of overweight and obesity in Mexico is around 75% (Shamah-Levy, Campos-Nonato and Cuevas-Nasu, 2019). This could be related to the socioeconomic level of the population in the study, where marginalization was low and had a higher level of education.
One possible limitation of the ESCAN-COVID19Mx is that the instrument did not collect information on the type of economic activities, household income or food security of the participants. Also, we did not inquire on the reasons for making the behaviour changes and we did not reapply the survey at the end of con nement. Therefore, it is not possible to understand the causality of these changes or whether these were maintained. The recruitment strategy through social networks and snowball methodology meant that there may be groups of Mexicans who were not correctly represented in this survey. Due to the crosssectional nature of our study, causal inferences should not be made between sociodemographic characteristics and changes in the quality of diet, physical activity, and lifestyle. Likewise, the participants self-reported their weight and height; therefore, it is possible that their BMI could be under or over reported.
The main strength of our study is that this is the rst study with a sample size greater than the estimated and obtained responses from all over the country about the changes in the quality of diet, physical activity and lifestyle in Mexicans who stayed home during the COVID-19 con nement. Results from this study show the importance of improving diet quality in people who work from home. A possible reason for this could be that most of the people who participated in this survey continued working from home during lockdown, however in this study we did not collect information about the participant's current working status to con rm this.
The ESCAN-COVID19Mx was carried out before the new frontal warning labelling came into place on the 1st of October 2020 (White and Barquera, 2020); thus, this survey could serve as reference point to evaluate changes in food consumption for this new food regulation.
The COVID-19 pandemic has emphasised the importance of preventing chronic degenerative diseases as evidence suggest that people living with these conditions are at higher risk of developing a more severe condition (Zheng et al., 2020). Therefore, it is important to encourage healthy lifestyle behaviours among the population during and after this time of crisis to prevent the risk of complications due to infectious and chronic degenerative diseases. One of the main challenges that are being faced during the COVID-19 con nement is that there is a lack of resources and programs that help the general population improve their health under these conditions.

Conclusions
The results from this study showed that during the COVID-19 con nement, Mexicans that were staying at home and who had a low degree of marginalization and a higher educational level, increased the quality of their diet by consuming more fresh foods prepared at home, yet they increased the consumption of sugar-sweetened beverages as a general tendency, reduced their physical activity and possibly the quality of their lifestyle decreased.

Figure 2
Change in the Diet quality index (DQI) prior and during the con nement

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. STROBEnutchecklistESCANCOVID19MxNutrition.docx