Dietary trends and socioeconomic disparities among young adults during the COVID-19 pandemic

Jennifer Parker (  jpt7@psu.edu ) Pennsylvania State University Allentown Campus: Penn State Lehigh Valley https://orcid.org/00000003-3773-3526 Simranpreet Kaur Pennsylvania State University Allentown Campus: Penn State Lehigh Valley John Marlo Medalla Pennsylvania State University Allentown Campus: Penn State Lehigh Valley Anairobi Imbert-Sanchez Pennsylvania State University Allentown Campus: Penn State Lehigh Valley Jeanette Alexandra Bautista Pennsylvania State University Allentown Campus: Penn State Lehigh Valley

regions such as in India and Sub-Saharan Africa [7], the circumstances were also dire in the wealthiest countries, including in the United States. The National Bureau of Economic Research declared the COVID-19 pandemic to have caused the worst American recession since the Great Depression with unemployment rates reaching the highest levels in recorded history [8]. Growing numbers of people in the United States and in other advanced industrialized nations, were suffering joblessness, nancial strains, and lack of access to nutrient rich food [9].
Lifestyle behaviors during lockdown conditions were found to reinforce these disturbing trends [10].
Studies found that more time spent at home promoted hypercaloric intake with larger meal sizes and increased frequency of snacking [11,12]. One study found this to be especially true for females whose energy intake was about 20% greater during the pandemic [11]. Other studies found that poor habits, including eating more comfort foods and surplus calories to feel better, were associated with depressed moods and anxious feelings [4,[13][14][15]. Concerning issues with inactivity and weight control were also tied to the pandemic [12-14, 16, 17].
But not all depictions of the pandemic's impact on diet were negative. There were rosier accounts of families making the best of times hunkered down together, cooking nutritious recipes, and gathering around the dinner table. One study found that with people staying at home and a decline in "eating out," there has been more consumption of home cooked meals, less fried foods (which tend to be a product of restaurant and fast foods), and with it, positive dietary change and prevention of obesity [18]. Private industry studies reported more cooking, use of recipes, and con dence in healthy meal preparation among their consumers [19,20]. Narrative pieces in Gastronomica stressed a movement toward "creative living" where authenticity emerged to reconceptualize home cooking as a means for gaining a sense of control and agency during a time when everything else seemed out of control [21,22]. In these depictions, the COVID-19 pandemic seemed to have reclaimed the kitchen, food, and cooking, as central to dietary health and wellness.
Unfortunately, in this growing literature on dietary practices during the COVID-19 pandemic, little attention has been paid to group level differences by socioeconomic status. This lack of attention is surprising given the ever-widening inequalities brought on by the pandemic and increasing evidence of class-based disparities in COVID-19 health outcomes [23,24]. Given the evidence on pandemic induced inequalities and the crucial link between diet and health, this study aimed to investigate how dietary practices may or may not be re ected in this growing socioeconomic divide.
The few studies that have examined socioeconomic differences in dietary patterns during COVID-19 provide evidence for nutritional inequities based on social class. In one study, high socioeconomic advantage was associated with lower odds of being affected by poor appetite or overeating while in lockdown [25]. Another study focused on children and adolescents in Brazil and found that lower-class families in home con nement and those from the Northeast region consumed a less healthy diet, with less fruits, juices, vegetables, and beans, than their more privileged counterparts [26]. The present research contributes to this socioeconomic lens with a focus on dietary practices among young adults, a largely neglected demographic in studies on COVID-19's impact on diet. The only known study that has focused on a similar age group found a host of negative dietary trends during the pandemic including overeating, inactivity, and weight gain. But this study focused speci cally on college students and neglected to differentiate groups by socioeconomic status [27].
Young adults are an important demographic to consider since they have experienced some of the most severe pandemic-related disruptions during a formative stage of development, a stage that is known to have profound implications on career paths, life-long economic security, and building a foundation for future bodily health [28]. During the COVID-19 pandemic, young adults were uprooted from college environments, experienced disproportionate amounts of job losses in hospitality and retail sectors [29,30] and had social lives abruptly curtailed during a time of life when peer groups are especially important [31]. These changes are even more critical considering that before the pandemic, young adults were already experiencing alarming rates of depression, stress, and suicide [32,33] setting them up to be exponentially vulnerable to COVID-19 related health problems.
Young adults' dietary practices during the pandemic could either help or hinder the vast challenges they face at a time when a healthy diet is essential to both physical and mental well-being. This study speci cally asks, how has the COVID-19 pandemic impacted the diets of young adults in the United States and how do the effects vary by socioeconomic group? Have dietary practices among young adults re ected and exacerbated the American class divide? Or is diet an aspect of life that young people experienced on an even playing eld during the pandemic? Dietary related practices on a broad range of measures were investigated. The intention was to contribute knowledge that could be useful for developing socioeconomically sensitive policies aimed at mitigating harmful long-term outcomes of the COVID-19 pandemic while preventing nutritional inequities during future public health crises.

Design
This study employed a cross-sectional design and targeted young adults between the ages of 18 and 28 living in the United States. An anonymous questionnaire [34] was developed speci cally for this study and was distributed through the Qualtrics platform online via private social media sites (Facebook and Instagram). The questionnaire was available for a period of 20 days from November 10 to November 30, 2020. This method was effective for the purposes of this study because it allowed for widespread dissemination at a time when face-to-face distribution was restricted due to the pandemic and when internet usage was up due to home con nement measures [35]. This study received approval from the Institutional Review Board of the university where this study took place.

Variables and measures
The questionnaire comprised six content areas: a) composition of the diet, b) sources of food, c) dietary habits, d) weight and physical activity, e) food insecurity, and f) nancial impact on diet.
a. Composition of the diet was measured by "constant sum" questions to assess how the overall distribution of the diet across ve food categories changed from before the Covid-19 pandemic to the rst few months of the pandemic. Food groups were based on USDA (U.S. Department of Agriculture) and PCRM (Physicians Committee for Responsible Medicine) guidelines for a healthy diet and included meats, dairy, grains, fruits/vegetables, and junk food.
b. Sources of food were also measured by "constant sum" questions to assess changes in where food came from across ve sources, including restaurants, fast food establishments, pre-prepared food at home, home cooked food by scratch, and junk food.
c. Dietary behaviors were measured based on an eight-item index scale (HEI, Healthy Eating Index) developed and adapted from previous questionnaires conducted during the COVID-19 pandemic [4,15,36,37]. Each item was based on a Likert scale (from 1 to 5) and analyzed individually, and then, compositely (as a continuous index variable). The intent was to evaluate changes in eating habits in respect to how those habits were perceived to become healthier or unhealthier from before the pandemic to the rst few months of the pandemic. Items ranged from overall perception of healthy eating to "sense of hunger and satiety" and "amount of meals prepared at home." Possible scores on the composite index ranged from 8 (much less healthy) to 40 (much healthier).
d. Weight change was measured on a 5-item Likert scale from "signi cantly decreased" to "signi cantly increased" to assess degree of weight gain or loss from before the pandemic to the rst few months of the pandemic. Physical activity during the rst few months of the pandemic was also measured and was based on four levels of activity, from "sedentary" to "very active".
e. Food insecurity was evaluated based on items adopted from the USDA Measurement of Household Food Security Survey with a focus on evaluating both extent of worry about having enough food, and degree to which one actually had enough to eat[38].
f. Financial circumstances were assessed using three separate measures intended to measure changes in overall nancial position and its relation to diet, including spending on groceries, and perceptions of how nancial conditions impacted spending on diet.
g. Socioeconomic status was measured by household income with two categories, lower household income (<$70,000) and upper household income (≤$70,000). This boundary was based on median household income in the United States (2019) ($68,400) [39]. Students living on their own were requested to report the income of the household where they were a declared dependent. Independent adults were asked to exclude the income of non-related household members.
Data analysis SPSS software (version 27) (SPSS Inc., Chicago, IL, USA) was used to conduct descriptive analysis and comparisons by income group. To examine changes in dietary composition and sources of food over time (continuous measures), paired sample t-tests were conducted. Chi-squared tests were computed to assess relationships between categorical variables including dietary behaviors, physical activity levels and weight change, food insecurity, and nancial circumstances. A t-test was employed to analyze household income-based differences in the HEI (Healthy Eating Index), a continuous variable. Analysis of Variance (ANOVA) tests were computed to measure associations between the HEI and physical activity levels. The alpha coe cient (Cronbach's Alpha) for the eight items comprising the HEI was .787, suggesting a relatively high internal consistency [40]. Skewness of continuous variables ranged from .054 to 1.824 and kurtosis ranged from − .858 to 6.458, indicating robust distributions within acceptable ranges for parametric testing [41,42].

Participants
A total of 361 people attempted the survey. After screening participants who did not meet the inclusionary criteria and/or did not complete at least 75% of the survey, 254 participants were included. Lower-income participants comprised 62% of the sample and upper-income participants comprised 38% of the sample.

Dietary consumption and sources of food
Composition of the diet changed from before the pandemic to the rst few months of the pandemic. Consumption of meats signi cantly declined from 29% (SD = 18.12) of the diet before the pandemic to 27% (SD = 16.56) during lockdown (p = .039). Grains also declined from 14% (SD = 11.79) to 13% (SD = 11.73), (p = .028). Meanwhile the proportion of junk food increased from 16% (SD = 13.06) to over 19% (SD = 14.61), (p = .008). The decline in meat consumption was signi cant only among upper-income participants (p = .044) while the decline in grains was signi cant only among lower-income participants only (p = .014). Meanwhile, the increase in junk food consumption was signi cant only among lowerincome participants (p = .049) ( Table 1).
Where participants got their food also changed. Food from restaurants declined from 18% (SD = 15.74) before the pandemic to 13% (SD = 15.91) in the rst few months of the pandemic (p < .001). Junk food sources increased from under 10% (10.53) to over 12% (11.41), (p < .001). The changes, though, were different between income groups. Upper-income participants, on average, got less food from restaurants (p = .005) and fastfood establishment (p = .013) and more from home cooked meals that were made from scratch (p = .04). Meanwhile, lower-income participants also got less food from restaurants (p = .008) but made up for the gap differently with a signi cant increase in junk food sources (p = .002) ( Table 1).  Second, when the HEI was broken down by item, several concerning trends were revealed (Table 2). In regard to overall healthiness of eating behaviors, well over half (55%) reported that their habits had become less healthy or much less healthy during the pandemic. About a quarter reported no change and nearly 1 in 5 (18 percent) reported that their habits had gotten healthier or much healthier. There were signi cant differences by income group (p = .041). Those reporting less healthy and much less healthy eating habits were disproportionately lower-income and those reporting healthier or much healthier eating habits were disproportionately upper-income. Nearly two-thirds (61%) of lower-income participants versus only 15% of upper-income participants reported worse eating habits during the pandemic. Meanwhile, upper-income participants reported having healthier or much healthier habits at more than twice the rate (27%) of their lower-income counterparts (13%).
More than half (53%) reported less healthy or much less healthy habits regarding sense of hunger and satiety while far fewer (14%) reported healthier or much healthier habits in this regard. There were signi cant differences by income group (p = .020). Lower-income participants were more represented among those who reported less or much less healthy habits (59%) versus 43% among upper-income participants. And upper-income participants were more represented among those who reported healthier changes (18%) versus only 12% among lower income participants who reported the same.
A large percentage of respondents (40%) reported healthier habits in number of meals prepared at home during the pandemic while far fewer (27%) reported unhealthier home meal preparation habits. There were signi cant differences by income group (p = .012). A far higher rate of upper-income (52%) than lower-income (32%) participants reported healthier or much healthier practices while more lower income (31%) than upper-income (22%) reported unhealthier or much unhealthier practices.
More than 40% reported worse habits regarding their indulgence in restaurant and/or fast foods during the COVID-19 pandemic. Another 40% reported better habits. There were signi cant differences by income group (p = .021) as lower-income participants were disproportionately represented among those who report unhealthier or much unhealthier habits (47% versus 35%) and upper-income participants were disproportionately represented among those who reported healthier or much healthier habits (51% compared to 34% of lower-income participants).
The overwhelming majority (67%) reported worse habits during COVID-19 when it came to indulging in unhealthy foods during moments of boredom and distress. Only 13% reported being better in this regard.
The trend was signi cantly worse for lower-income participants (p = .014). Nearly three-quarters (73%) of lower-income participants reported a worse time with unhealthy food indulgence during times of stress or boredom versus 57% of upper-income participants. More upper-income (16%) versus lower-income (11%) participants reported having healthier or much healthier eating habits in situations of boredom and distress.
There were other notable changes in reported eating behaviors where no signi cant differences by income level were found. For one, the majority (60%) reported unhealthier or much unhealthier snack consumption while only about 1 in 10 (11%) reported better habits. About a third (34%) reported healthier or much healthier intake of immunity-boosting foods, while 40 percent reported no change and about a quarter (26%) reported worse habits in this regard. Finally, more than a third (37%) reported less healthy or much less healthy habits involving the number of meals eaten per day. Another third (37%) reported no change and about a quarter (27%) reported healthier or much healthier practices.

Weight change and physical activity
About 42% of participants, overall, reported gaining weight during the pandemic, with six percent reporting signi cant increase in weight. Another 40% reported no change in their weight and 17 percent reported losing weight. Lower-income participants reported a signi cantly greater proportion of this weight gain (p = .024). Half (50%) of lower-income participants reported gaining weight with nearly eight percent reporting a signi cant increase. Less than a third (29%) of upper-income participants reported gaining weight and only three percent reporting a signi cant increase in weight (Table 3).
Most participants (66%) reported being sedentary or only lightly active compared to about a third (34%) who reported being active or very active. When asked about how much their activity level changed from before the pandemic to the rst few months of the pandemic, about a third (33%) reported being less active and slightly more than a third reported being more active (39%). Differences in activity levels and change in physical activity between income groups were not found to be statistically signi cant. Still, the differences we found, especially at the two extremes of the activity ladder, are noteworthy. Considerably more lower income (36%) than upper income (26%) participants reported being sedentary during lockdown while more upper income (22%) than lower income (10%) participants reported being highly active. The higher rate of reported sedentary lifestyle among lower income participants during the lockdown appears to be related to a higher rate of both decreased activity during the COVID-19 and already existing sedentary lifestyle prior to the pandemic (Table 3).

Food insecurity
During the pandemic, a small percentage of participants reported serious issues with food insecurity with about seven percent not having enough to eat and ve percent worrying about where their next meal was coming from "regularly" or "most of the time." Food insecurity was disproportionately concentrated among lower-income participants with eight percent reporting not having enough to eat on a regular or most of the time basis (versus ve percent of their upper-income counterparts) (p = .001) and six percent worrying about where their next meal was coming from on a regular or most of the time basis versus only two percent of upper-income participants (p = .001) who had the same worry (Table 5).

3.5-Financial circumstances
Nearly half (43%) of the participants reported that their nancial circumstances took a turn for the worse because of COVID-19. About the same amount (46%) reported no change and 11% reported better nancial circumstances. Meanwhile most (58%) reported increased spending on groceries during the pandemic and a quarter (28%) reported no change. There were no signi cant differences between income groups. When asked how their nancial circumstances impacted their diet during COVID-19, most (60%) reported that there had been no change while about a third (31%) reported that their nancial situation had made their diet worse or much worse during the pandemic. Fewer than 10 percent reported that their nancial situation led to an improvement in their diet. Lower-income participants were disproportionately represented among those who reported that their nancial circumstances made their diet worse or much worse during the COVID-19 pandemic (39% versus only 18% of upper-income participants) (p = .013) (Table 6).

Discussion
The results of this study show a clear trend toward more unhealthy consumption behaviors and dietary habits of young adults during the COVID-19 pandemic. The overall consumption of junk food signi cantly increased, over a third gained weight, and about two-thirds reported less healthy eating habits overall, unhealthier snack consumption, and more emotional eating. These results echo ndings from other studies both in the United States and in other wealthy nations that have found associations between the pandemic and dietary deterioration [4,[11][12][13][14][15][16][17]27].
The previous neglect of a focus on the young adult demographic in this growing literature, aside from rare exceptions [27], was a primary motivation for the current study. As we have argued, the abrupt and unprecedented degree of change experienced at a formative stage of life makes this group important to understand on its own. Another aim was to investigate socioeconomic differences, given the widening inequalities induced by the pandemic. Group level differences by household income followed a clear pattern. Lower-income participants were disproportionately represented among those who reported unhealthy consumption and behaviors while the opposite was true for upper-income participants. While both groups reported more negative, than positive, changes, when there were positive changes, such as less reliance on fast foods and consuming more home-cooked food from scratch, upper-income participants were disproportionately represented among those who reported them.
The differences we found may be explained, in part, by the de ning characteristics of the American class divide during COVID-19. A uent Americans have been more likely to remain employed, work from home, and not suffer income decline while lower-income Americans have disproportionately suffered job loss, nancial strain, and housing precariousness. Among lower-income Americans still working, they have been more likely to work outside the home than their upper-income counterparts, and often in two or more job while facing risk of exposure to the virus [43,44]. The economic impact has been especially harsh for young workers who are disproportionately employed in entry level service sectors where job losses were particularly severe [29]. These disparate class-based experiences may help explain dietary inequities.
Cooking at home, for instance, requires planning, time, investment in raw products, and a modicum of stability [45], requirements that the a uent may have enhanced capacity to meet during the pandemic and the lower-income may have decreased capacity to meet. This study's nding that home-based cooking from scratch signi cantly increased, and reliance on fast food signi cantly decreased only among upper-income participants, indicates that the pandemic's positive in uences on healthy eating[18, 21,22] may be a silver lining enjoyed primarily by the more privileged. On the other hand, a reliance on pre-prepared, processed, convenience and/or fast foods may better correlate with the circumstances of those with less stable, downward spiraling lives, as they are cheap, readily available, and calorie dense. In a recent study involving focus groups, lower-income women reported that barriers to healthy eating included cost, convenience, and preparation time. This team of authors also noted that comfort foods may be used by as a coping mechanism for the multiple stressors and anxieties in their lives [46]. Geography may play a role, too. Lower-income households are already more likely to reside in food deserts and mirages where access to foods high in nutrition is limited [47,48] In this way, the results of this study support existing research on how socioeconomic status impacts dietary practices [49,50] while contributing evidence on how the pandemic may exacerbate nutritional inequities.
Lower-income participants' disproportionate representation in several other unhealthy trends is additional cause for concern. While alarming numbers reported gaining weight and sedentary lifestyle during the pandemic, the proportions were much more severe among those from lower-income households. When more than one in two lower-income young adults report gaining weight during the pandemic (versus one in four among upper-income participants), it suggests a class-based vulnerability to a host of metabolic diseases associated with unhealthy weight gain, such as type 2 diabetes, dyslipidemia and hypertension, all of which have been shown to downgrade immune responses, and make one more vulnerable to infections and less responsive to antivirals and vaccinations [4,[51][52][53]. The Center for Disease Control and Prevention (CDC) has identi ed obesity and increased BMI as a risk factor for COVID-19 related illnesses, regardless of age [54]. Even more disturbing is that lower-income Americans already have higher rates of obesity than their more a uent counterparts, a disparity that has long been recognized [55,56].
Studies have shown that the increased risk to obesity during the pandemic is due to a sustained positive energy balance with reduced energy exposure [10]. Since junk foods are generally energy dense, and nutritionally lacking, it likely poses a threat to those reporting an increase in its consumption, especially in conjunction with a sedentary or low active lifestyle. The positive relationship between activity levels and healthy eating behaviors, irrespective of household income level, is an important nding, as it suggests that all socioeconomic groups bene t from the positive in uence of an active lifestyle on healthy eating behaviors. The fact that more lower income than upper income participants reported being sedentary during the lockdown suggests a need for promoting and ensuring opportunities for active lifestyle among the low income who may not have the same access to pandemic-safe recreational facilities, including parks, sports arenas, and home-based equipment.
The prevalence of food insecurity found in this study and lower-income participants' disproportionate representation in it, corresponds with what is already known about the pandemic's in uence on food accessibility. Food insecurity was heightened during the pandemic, becoming a catastrophic problem associated with panic, hoarding, and other alarming behaviors, while a icting the most economically vulnerable [57]. The fact that half of lower-income participants in this study struggled with getting enough to eat at least once or twice during the rst few months of the pandemic and that more than 10 percent did not have enough to eat on a regular basis, is deeply concerning. On the other hand, it should not be missed that nearly a quarter of upper-income participants reported struggling with getting enough to eat at least once or twice during the pandemic, and for more than 5%, it was a regular occurrence. It is clear, that even though income background is signi cantly correlated with food insecurity, it does not make anyone immune from the perils of hunger and malnutrition [47].
It is important to emphasize that both lower-and upper-income participants in this study reported worsened nancial circumstances during the pandemic. This re ects the ways in which economic losses have impacted a broad swath of the American population. On the other hand, it is also important to note that signi cantly more lower-income participants reported that their nancial situation negatively impacted their diet. This may indicate one of the ways in which pandemic related economic losses impacted the daily lives of lower-income groups more severely. In a TIAA survey conducted during the pandemic it was found that more than half of individuals who make less than $50,000 per year have never created an emergency fund[58]. This nding follows what studies have always shown, that the wealthier you are, the more equipped you are to absorb emergencies, and the poorer you are, the less you have to fall back on, and the more vulnerable you are to a downward spiral where diet is often the rst casuality [59].
Most generally, the widening socioeconomic disparities in society at large that have been caused by the COVID-19 pandemic appear to be re ected in the dietary inequities found in this study. The pandemic has impacted dietary composition, eating practices and weight change in primarily negative ways, but the lower-income are disproportionately represented in these unhealthy trends while the opposite is true for upper-income participants. It is important for public health policies aimed at improving the well-being of the U.S. population to be cognizant of these income-based inequities and how they re ect "differential access to the resources required to access high-quality diets and physical activity" [55]. Solutions should be geared toward reshaping scal, social and physical environments, rather than relying solely on behavioral interventions [55,60].

Conclusions And Limitations
This research contributes to a growing body of research focused on understanding the pandemic's impact on diet. With a focus on young adults, it lls a gap by addressing an age demographic that has been largely neglected in this literature. With an emphasis on differences by household income, this study provides a lens into how pandemic related socioeconomic disparities might be playing out in the diets and overall well-being of young Americans. Very little work has been attentive to socioeconomic disparities in dietary behaviors during the pandemic, despite its importance to overall health and the wellknown links between health status and vulnerability to the COVID-19 virus. Given the way the pandemic has both exacerbated and exploited already existing socioeconomic disparities, it is imperative that we have an understanding of how diet and other food related concerns might be interwoven into these disparities. The health-related problems that arise out of these disparities can be remedied, both now, and in preparation for the next public health crisis.
This study had limitations. Since the data was collected online, it was able to capture substantial representation of socioeconomic groups but less balanced representation in regard to gender. The majority of the participants were female which is the same outcome for other studies that have examined dietary behaviors during the COVID-19 pandemic that have relied on social media sites such as Facebook. This is possibly because the majority of Facebook users are female. Moreover, women were disproportionately impacted by joblessness during the pandemic and could have had more time to respond to the survey than their male counterparts [14]. It should also be noted that while the study sample was relatively proportionately balanced in regard to most race and ethnic groups for this age demographic, that it was under-represented by African Americans, a group that has been disproportionately harmed by the pandemic. It is possible that this under-representation could have impacted the results even though the focus of this study was not on race/ethnic differences. Additionally, the study relied on self-reported data so recall bias may be a factor in the results. And since the data for both periods, before and in the rst few months of the pandemic, were reported retrospectively, recall bias could possibly affect the internal reliability of the study [61]. But since the gap in time was very small and the shift from pre-pandemic to when the pandemic began unprecedented with a very high contrast, we believe that recall bias was minimal.