Prevalence and Risk Factors of Food Insecurity among Mexican University Students’ Households

Household food insecurity (FI) remains a major public health challenge worldwide. Data about perceived FI and its risk factors in Mexican university students are lacking. We aimed to assess FI’s prevalence and factors affecting it among university students’ households in Mexico. This cross-sectional analysis involved 7671 university students’ households using the 2018 Mexican National of Household Income and Expenditure Survey data. Variables analyzed included sociodemographic characteristics, and the 12-item validated Mexican Scale for Food Security (EMSA). Multivariable logistic regression modelling was performed to identify FI risk factors. The overall household FI prevalence was 30.8%. According to FI severity, prevalence rates were 16.3% for mild-FI, 8.8% for moderate-FI, and 5.7% for severe-FI. Low socioeconomic status (OR = 2.72; 95%CI: 2.09–3.54), low education level of household’s head (OR = 2.36; 95%CI: 1.90–2.94), self-ascription to an indigenous group (OR = 1.59; 95%CI: 1.41–1.79), attending public university (OR = 1.27; 95%CI: 1.13–1.43), female-headed household (OR = 1.26; 95%CI: 1.13–1.40), having worked recently (OR = 1.19; 95%CI: 1.07–1.33), and being in second year of studies (OR = 1.17; 95%CI: 1.03–1.33), were significantly related to FI. Our results confirm that FI is highly prevalent among Mexican university students’ households and that sociodemographic factors are essential in addressing this concern. Findings highlight the need for preventive programs and policies to alleviate FI.


Introduction
Food insecurity (FI) is a multidimensional concept that exists whenever "the availability of nutritionally adequate and safe foods or the ability to acquire acceptable foods in socially acceptable ways is limited or uncertain" [1]. FI remains a significant public health challenge worldwide. According to the latest estimates, in 2019, 25.9% of the world population and 31.7% of the Latin American and Caribbean population suffered moderate or severe FI [2]. Furthermore, 22.7% of the Mexican population suffered moderate or severe FI in the 2018-2019 period [3]. However, these rates are expected to rise due to the coronavirus disease (COVID-19) pandemic [2]. Some risk factors associated with FI A cross-sectional analysis was conducted using data from the 2018 Mexican National of Household Income and Expenditure Survey (ENIGH, for its acronym in Spanish). ENIGH is an open access database that can be downloaded from the National Institute of Statistics and Geography (INEGI, for its acronym in Spanish) website [21]. ENIGH is conducted biennially in collaboration with INEGI and the National Council for the Evaluation of the Social Development Policy (CONEVAL, for its acronym in Spanish).
Details of the survey design and methodology have been published elsewhere [22,23]. Briefly, ENIGH employed probabilistic, two-stage, and cluster sampling methods. The sampling unit was dwelling; the observation units were households and people, and the population under study were households and people who lived at home. All information was recorded using standardized questionnaires applied through face-to-face interviews with the household's head, between 21 August 2018 and 28 November 2018. The information included in the ENIGH database is about the amount, obtention, and distribution of households' income and expenses, and the occupational and sociodemo-graphic characteristics of the household members. The database contains information on 74,647 households from 32 states of Mexico. The study complies with the tenets of the Declaration of Helsinki. Considering the anonymous nature of the publicly available data and that no additional contact with participants was needed, this study was not reviewed by a Research Ethics Board.
We included households with complete data of the Mexican Scale for Food Security (EMSA, for its acronym in Spanish) [24] and with at least one university student living at home at the time of survey application. It is possible that more than one university student lived in the household, but only one university student per household was included in the present analysis. The selection of the university student for this analysis was not intentional. There is a database with the characteristics of the participants and another database with the characteristics of the households. When merging both databases, the characteristics of a university student (irrespective of whom this was) were joined with the characteristics of their home.

Sociodemographic Characteristics
The following characteristics, available in the database for each university student, were selected: age, sex, marital status (single: single, separated, divorced or widower; being in a relationship: de facto or married), type of university attended (private or public), years of study (first, second or third to fifth years), receiving a scholarship (yes or no), being an indigenous speaker (yes or no), self-ascription to an indigenous group (yes, if a person considers himself/herself as indigenous, according to his/her culture), and working during the last month (yes or no).
In addition, university student's household characteristics were included: the presence of children under 18 years of age in the household (yes or no), mother's presence in the household (yes or no), father's presence in the household (yes or no), household's head sex, household's head educational level (graduate or postgraduate; elementary, middle, or high school; no educational instruction, kindergarten, or incomplete elementary school), and population size where the household is established (≥100,000 inhabitants, urban area, large city; 2500 to 99,999 inhabitants, urban area, town; <2500 inhabitants, rural area). Socioeconomic status (high, middle-high, middle-low, or low) was also considered. The categorization was constructed by ENIGH personnel through a statistical analysis of 24 indicators within the household (socioeconomic, physical, and equipment features). The household was classified as nuclear (only one primary family group), one-person (home made up of a single person who is the household's head), extended (household's head and his/her primary family group plus other family groups or other relatives), composite (nuclear or extended household plus people unrelated to the household's head), co-resident (household consisting of two or more people who are unrelated to the household's head). For analytical purposes, however, the household type variable was recategorized as nuclear household and others.

Household Food Security Status
The degree of household food security was measured using the validated Mexican Scale EMSA [24]. This instrument has been used in previous studies in the Mexican population [4,25] and by CONEVAL, to measure FI and perceived food access [26]. The scale includes 12 questions with yes/no response options. Each item inquires if, in the last three months and due to lack of money or resources, an adult in the household ever: (1) had a low varied diet; (2) skipped breakfast, lunch, or dinner; (3) ate less than you should; (4) ran out of food; (5) felt hungry but did not eat; and (6) ate only once or stopped eating one day. Question 5 and 6 were only answered by those who answered "yes" in the first four questions. When children under 18 years of age live in the household, it additionally asks if a minor at home: (7) had a low varied diet; (8) ate less than he/she should; (9) received less amount of food at meals; (10) felt hungry but did not eat; (11) went to bed hungry, and (12) ate only once or stopped eating on a given day [24].
Based on the total number of affirmative answers on the EMSA, a score was obtained, ranging from 0 to 6 for households without children or 0 to 12 for households with children. According to this score, households were classified into four different categories: household food security (0 points), mild household FI (1-2 or 1-3 points for households without or with children, respectively), moderate household FI (3-4 or 4-7 points for households without or with children, respectively) and severe household FI (5-6 or 8-12 points for households without or with children, respectively) [24].

Statistical Analysis
Descriptive data are presented as mean and standard deviation for quantitative variables or number and percentage for qualitative variables. Differences in quantitative variables were examined using the ANOVA test, and different distribution in qualitative variables were examined using the Chi-square test. We calculated FI and food security prevalence with a 95% confidence interval (CI) in all Mexican households and each of the 32 states of Mexico.
We used univariable and multivariable logistic regression analyses to assess risk factors, including all sociodemographic characteristics, related to household FI (compared to food-secure households) and are presented as odds ratio (OR) and 95%CI. For multivariable analysis, all factors showing a p-value < 0.1 in the univariable analysis were included.
We excluded, in the multivariable model, the following variables: "children under 18 years of age in the household", because it has collinearity with the dependent variable; "mother's presence in the household" and "father's presence in the household", because they were highly correlated with "sex of the head of household" variable; and "indigenous speaker", since it has collinearity with the variable "self-ascription to an indigenous group". A choropleth map of household FI status in Mexico was generated in Microsoft Excel ® from Microsoft 365. Statistical analysis was performed by STATA ® software, version 15.0 (StataCorp, College Station, TX, USA). A p-value < 0.05 was established as statistically significant.

Participants Sociodemographic Characteristics
A total of 7671 university students and households were included in the study. Overall, the mean students' age was 22.2 ± 5.3 years, 50.9% were female, 89.8% were not in a partner relationship, and 41.5% had worked last month. In addition, most of them were in a public or a government university (71.9%), did not receive a grant (84%), and more than a half were either in the first or second year of their studies (58.1%). Only 1.7% was an indigenous speaker, and 22.7% self-reported as belonging to an indigenous group (Table 1, total column). In terms of the composition of student household: most were middle-low or middle-high socioeconomic status (77.1%), lived in a nuclear-type family (66.7%), where no children under 18 years of age was present (74.4%), and generally the mother (78.2%) and/or the father (59.7%) were present. Most of the household's head had elementary, middle, or high school level (71.5%) and were males (70%). Only 19.4% had a bachelor's or postgraduate degree. Half of the households were located in large cities (≥100,000 in inhabitants) (53.9%) ( Table 1, total column).  .3) 1 According to the Mexican Scale for Food Security (EMSA, Spanish acronym). 2 The association between qualitative variables was analyzed with the Chi 2 test. Comparison of age between food insecurity categories was evaluated with the onefactor ANOVA test. A p-value < 0.05 was established as statistically significant. 3 Data is presented as mean ± standard deviation. 4 Data is presented as frequency (percentage). 5 A person who considers himself/herself as indigenous, according to his/her culture. 6 Others: One-person household, extended household, composite household, co-resident household.

Association between Sociodemographic Characteristics and Household Food Insecurity
Univariable logistic regression analysis (   Regarding university students' households, we found an inverse association between socioeconomic status and FI; the higher the socioeconomic status, the lower the risk of FI  (Table 2). In contrast, paternal presence in the household (OR: 0.89; 95%CI: 0.81, 0.98; p = 0.021) was a protective factor for the risk of household FI. Our multivariable regression model, which included variables identified with a p-value < 0.1 in the univariate analysis (Table 2), found that all the aforementioned factors (except being in the first year of studies, household type, and population size) remained as independent predictors of risk of household FI (all p < 0.05).

Discussion
A third of the Mexican university students' households in our study live with some level of FI. The household characteristics were positively associated with FI, especially having a middle or lower socioeconomic status, a household head with a low educational level (high school or lower), and a female-headed household. Furthermore, FI was more likely in students with a self-ascription to an indigenous group, those attending a public university, those having worked last month, and those in the second year of studies. However, FI was less likely in older students. To our knowledge, this is the first study determining FI prevalence and factors affecting it among a representative sample of Mexican university students' households.
In our study, FI prevalence was high (30.8%), similar to the mean prevalence reported in previous reviews about FI prevalence in university students in other countries (27-42%), ranging from 10% to 84% [12,13,27]. The most frequent questionnaires used to measure FI in studies on university students were the US Department of Agriculture Food Security Survey Module and the 10-item Adult Food Security Module. However, FI prevalence observed Nutrients 2021, 13, 3426 9 of 14 in this study was lower in comparison with the national prevalence (55%), determined with the Latin American and Caribbean Food Security Scale (ELCSA Spanish acronym) [3], extended version of the EMSA. The difference in prevalence rates may be due to the low frequency in this study of some characteristics (in less than 30% of participants) that have been positively associated with FI in studies carried out in the Mexican population [4,[28][29][30] and in university students from other countries [13,15,16,18]. These characteristics (percentage in our study) are: self-ascription to a minority group (22.7%) [13][14][15][16]18,28,29,31], presence of children under 18 years of age at home (25.6%) [4,13,29], woman household's head (30%) [4,29], households located in rural areas (<2500 inhabitants) (21%) [4,29], and household's head with low or without education (13.1%) [4,29,30,32].
Most university students in our study lived in nuclear (66.7%) and extended (28.9%) household type. Only 2.7% had a one-person household and 0.8% a co-resident household. This living situation of university students differs from other countries. Only 35.3% of college students from Australia, lived with their parents [19], 84.4% of freshmen from US [33] and 31.4% of university students from Nigeria [34] lived on campus. This is relevant because traditional and extended households are more likely to be food secure than unitary households [4]. Besides, students living with their parents or other relatives are less likely to have FI, than students who lived alone [31]. However, there are other households and individual characteristics that could increase the FI risk in this population group.
Having low socioeconomic status was positively associated with FI. Low income is one of the main risk factors for FI. In other contexts, lower income was positively associated with FI in UK adults [5], Australian adults [35], and in US college students [36]. In the Mexican population, any income level in comparison to very low income was positively associated with food security and negatively with FI [4]. In addition, FI was more frequent in those with the lowest family income [29]. Low-income households may not be able to buy enough food to meet their needs. In 2018, 48.8% of the Mexican population (61.1 million people) were living in poverty with an insufficient income to buy the basic food basket, goods, and services; and 16.8% (21 million people) were living in extreme poverty with an insufficient income to buy the basic food basket [37,38].
Educational level of a household's head lower than graduate or postgraduate was positively associated with FI, similar to other studies conducted on college students [32,33] and the general Mexican population [4,29,30]. Moreover, in the general population from Australia, having less than university-level education was positively associated with FI [35]. Education broadens the possibility of having a better job and a better income, allowing to acquire enough food to cover the nutritional needs of the members of the household [39]. In Mexico, the average quarterly income increased with education level. Those with postgraduate studies had the highest income, and those with an incomplete primary school had the lowest income [23,40]. Additionally, a low education level may have an impact on how household resources are managed.
Having a female-headed household was positively associated with FI, similar to studies conducted on Mexican population [4,29]. In a systematic review and meta-analysis, it was evidenced that female-headed households were 75% (95% CI: 49%, 96%) more likely to be food insecure than male-headed households [41]. This type of household is more vulnerable to deprivation situations, compared to men-headed households. Reasons behind are that: (1) they are single-parent (mother and children) or extended households, either due to separation, divorce, or widowhood [26,39]; (2) there is a greater number of members under 15 and over 65 years, that represents a greater potential economic dependence [26]. Another consideration is that (3) because of gender roles, women typically assume responsibility for domestic work and caring for other people. Therefore, women tend mainly to seek part-time jobs or reduced hours as a strategy to combine both activities. However, this implies inserting themselves into more precarious jobs, with lower remuneration and less stability [26]; (4) there is income inequality between men and women. Particularly in Mexico, although Mexican laws promote labor equality between men and women, women receive less income from their work regardless of age, education and marital status [23,26,40,42,43]. Finally, (5) feminized occupations tend to have relatively lower wages [26].
We observed that attending a public university was positively associated with FI. This association was not observed in a study on university students from the US [32]. In Mexico (2018-2019), there were more private schools of higher education (n = 3252) than public schools (n = 2283); however, more university students attended public schools (2,773,338 students) than private schools (1,170,206 students). Public schools obtain financing from federal, state and autonomous funds, while private schools obtain financing from private sources and subsidies [45]. Therefore, public universities are economically more accessible than private universities to students of any socioeconomic level or degree of vulnerability.
Employed students were more likely to be food insecure [36], while being a nonemployed student was negatively associated with FI than students working full-time [32]. It has been evidenced that financially independent students were more likely to report FI [13,46]. Otherwise, receiving financial support from family was negatively associated with being food insecure [16]. Furthermore, students may work because they need an economic income. Nevertheless, students, to combine work and study, may work part-time and therefore, due to their student status and the work they perform, their salaries are low. In US university students, having one or more part-time jobs was positively associated with FI, with respect to not working [17,18]. Moreover, many younger students may not have the experience and skills to manage financial resources or save. A study showed that spending on other items instead of buying food was positively associated with FI in students [15]. We do not know if this happens in Mexico, but it could be that if students work, it is to meet their personal expenses (cellphone . . . ).
Finally, in the present study, being in second year of studies was positively associated with FI compared with being in third to fifth years. Other studies also have identified academic years as predictors of university students' FI [15,18,47]. In a longitudinal study, FI was positively associated with the end of the first semester and the end of the second semester compared with the start of the first semester [47]. In the US, students in the sophomore and junior years were more likely to experience FI than freshmen [17] or graduate students [15]. Regarding our results, students with more years of study may have more knowledge or preparation than students in the first years of university and can access a better-paying job. The former can also explain the negative association observed, in this and other studies [14,35], between age and FI.
The available energy food supply in Mexico exceeds the requirements to cover the demand. The problem is that not all people have access to it [52,53]. Over the course of several years, various food policies have been implemented in Mexico [54]. In 2018, 35% of Mexican households benefited from at least one food aid program. These supports were received mainly by those living in rural locations and the southern states of Mexico [3]. In the 2012-2016 period, it was identified that food aid social programs had a positive effect on households located only in rural localities. In these households, the proportion of households with food security increased, and those with moderate and severe FI decreased [55].
On the other hand, in 2018 [56] and 2021 [57,58], the government offered financial support scholarships for university students who meet specific requirements. Additionally, public and private universities in Mexico, and non-governmental organizations (public and private), also offer financial scholarships and discounts on tuition fees for students who meet some requirements [59]. However, the main objective of this support is to promote school retention and completion of studies. To our knowledge, there are no food-type supports for university students in Mexico. Therefore, particularly for university students, it is suggested to implement financial counseling at the individual level, grant food pantries at the institutional level, and make changes at the policy level to increase financial aid [13] and create food aid programs for university students. At the national level, in addition to maintaining government programs to alleviate hunger and improving people's nutritional status, the following are required: improvements in education, well-paying jobs, social security [38,52], availability of healthy food at affordable prices [38], and sustainable diet and production system [60].
The main strengths of this study are the large sample size, which includes participants from all states of Mexico, participants from all socioeconomic levels, and participants who live in different types of localities. It used a validated instrument to evaluate Mexicans' household food security status (EMSA) [24]. Likewise, the participants in this study come from a randomly selected sample. Therefore, the FI prevalence observed probably reflects the reality of Mexican university students' households in 2018. Moreover, these findings indicate an advance in understanding the relationship between FI and sociodemographic variables in this population group. However, the present study has some limitations. FI prevalence in university students and the general population from Mexico may have increased due to the COVID-19 pandemic. Poverty in the Mexican population increased in 2020 (55.7 million people), compared to 2018 (51.9 million people) [61]. In addition, in other contexts, due to the pandemic FI prevalence increased in US households [6], and changes in the FI prevalence are suggested in Australian adults [35], and university students [31,46]. Therefore, a current analysis of the FI prevalence in university students is required. Additionally, the cross-sectional nature of the study design limits the potential to discern causative relationships. Consequently, the results should be interpreted with caution. As the sample of this study consisted of Mexican university students' households, the results cannot be generalized to the general population.

Conclusions
In conclusion, a high prevalence of FI was detected among Mexican university student households. One out of three Mexican households suffers from mild, moderate, or severe FI. Further characterization of sociodemographic factors underlying FI may help to identify intervention opportunities to mitigate the consequences of FI. The population of university students is affected by FI, but their study is underrated in Mexico. This analysis encourages future research focused on the consequences of FI, such as unhealthy dietary patterns, bad food choices, low academic performance, and health disorders in Mexican university students.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/nu13103426/s1, Table S1: Prevalence of food insecurity in Mexican university students' households, according to state, Table S2: Perception of access to food in Mexican university students' households. Institutional Review Board Statement: Ethical review and approval were waived for this study, due to the anonymous nature of the publicly available data and that no additional contact with participants was needed.
Informed Consent Statement: Patient consent was waived due to the anonymous nature of the publicly available data.