Disordered eating during COVID-19 pandemic is associated with nutritional status, negative mood changes, and body image in university students

Abstract

Our objective was to evaluate disordered eating and associated factors in university students in the early period of the COVID-19 pandemic in Brazil. This is an analytical cross-sectional observational study, and data collection took place between May and June 2020, using an online semi-structured questionnaire. In total, 936 students were evaluated. Present signs/symptoms of the flu syndrome (OR = 1.605), negative mood changes (OR = 1.628), weight gain (OR = 1.739), obesity (OR = 3.089), follow fitness/health profile on social media (OR = 2.050), having inadequate body perception (OR = 2.416), and body dissatisfaction (OR = 2.612) increased the chances of presenting a higher score on the disordered eating scale.

Introduction

Coronavirus is a family of viruses that cause respiratory infections, and the new agent of coronavirus (nCov-2019) was discovered in December 2019 after reported cases in China. On 26 February 2020, the first positive case for COVID-19 in Brazil was confirmed (World Health Organization [WHO], 2020) and, on 11 March 2020, the WHO declared that it was a pandemic (OPAS, 2020). Since the appearance of the virus, various levels of preventive and restrictive measures have been implemented, on a global scale, with the aim of slowing down or preventing the spread of the disease. Among these measures is social distancing, which has validated effectiveness to control the spread of the virus, but which has also resulted in several negative consequences, such as changes in the economy and health, increased negative feelings (Phillipou et al., 2020), in addition to changes in eating behavior (Cecchetto et al., 2021), body shape and in the mental health of individuals (Di Renzo, Gualtieri, Pivari, et al., 2020; Haddad et al., 2020).

These restrictive measures impacted the entire population, including university students, who, as a result of the absence of academic activities caused by the COVID-19 pandemic, presented high levels of stress, and a consequent increase in dysfunctional eating behaviors (Flaudias et al., 2020). Such issues are already originally relevant in this group, since the beginning of university studies is a major stressor and is associated with an increase in the incidence of depression, anxiety, alcohol abuse, and risky eating behavior (Unikel et al., 2016). These changes in students’ lives favor the development of dysfunctional eating behavior and risk for eating disorders (Carvalho et al., 2013; Kessler and Poll, 2018; Moraes et al, 2016). The presence of disordered eating in university students was evidenced in a study carried out by Alvarenga et al. (2013). Their results revealed practices compatible with this clinical condition, such as high frequency of purgative practices (vomiting), ingestion of liquids instead of solid foods, omission of meals, restrictive diets, or use of a compensatory method in this population.

Disordered eating includes dysfunctional practices similar to those that occur in classic eating disorders (anorexia, bulimia, and binge), but that do not meet the diagnostic criteria of the American Psychiatric Association. This concept includes unhealthy behaviors for weight control, such as fasting or very restrictive diets, consumption of liquid diets in order to lose weight and use of laxatives and/or diuretics (Díaz de León-Vázquez and Unikel Santoncini, 2019). Studies that evaluated disordered eating in university students point to a high frequency of this risk behavior (Alvarenga et al., 2012; Díaz de León-Vázquez et al., 2017) and emphasize that the average age for the onset of eating disorders is between 18 and 21 years old (Hudson et al., 2007), being more prevalent in females (Silva et al., 2011).

Dissatisfaction with body image is closely related to disordered eating practices, and even among students with body weight within the eutrophy threshold, body distortion is present, given the high prevalence of body dissatisfaction in university students (Alvarenga, Philippi, et al., 2010; Carvalho et al., 2013; Leal et al., 2020). In this context, the role of the media as a mediator in the negative relationship between body dissatisfaction and disordered eating is already well documented (Alvarenga, Dunker, et al., 2010; Walker et al., 2015). According to Sidani et al. (2016), there is a greater concern about food among individuals who continually use social media, such as young adults. Constant access to media and social media is an important risk factor for exacerbating body dissatisfaction in women (Alvarenga, Philippi, et al., 2010). Souza et al. (2016) demonstrated in a literature review that, for Australian university students, the use of Facebook^® was correlated with greater body dissatisfaction, and the longer the time of exposure to the media, the greater the levels of dissatisfaction. The same was observed among American university students who had Instagram^® as the most accessed media.

In the current scenario, the COVID-19 pandemic has acted as a strong stressor for different social groups, contributing to changes in the eating behavior of the young adult population. Although some works have already been published showing such alterations (Catucci et al., 2021; da Mota Santana et al., 2021; Flaudias et al., 2020), the presence of disordered eating in university students in this period has not yet been evaluated. Therefore, this study aimed to analyze the disordered eating of university students and the factors associated with it in the initial period of the COVID-19 pandemic in southeastern Brazil. The hypothesis of this work is that the presence of signs and symptoms of flu syndrome, negative changes in mood, weight gain, and obesity may be directly associated with a more disordered eating in this period. Still, it is believed that contact with fitness profiles on social media, which negatively impact body perception and satisfaction, may also present this association.

Material and methods

Design and study population

This is a descriptive, observational, cross-sectional, and analytical epidemiological study. The target population of this research was made up of students from the onsite undergraduate courses of a federal university in southeastern Brazil (Espírito Santo), over 18 years of age, of both sexes. All students regularly enrolled on all campuses at that university were invited to participate. The survey was conducted via an online form and recruitment took place through institutional email (mediated by the university), and through dissemination through academic centers and social media.

The inclusion criteria were: being enrolled in any onsite undergraduate course at the university; being 18 years of age or older; and having agreed to the informed consent form. Pregnant and lactating women were excluded from the study due to changes in eating behavior and nutritional status resulting from these specific conditions. Nutrition students were excluded because they have been shown to have different eating behaviors, which could generate a bias in the sample (Alvarenga et al, 2012; Fiates and Salles, 2001; Fredenberg et al., 1996). The decision to exclude this group in our study was also due to the feasibility of comparing our study with those published by the author of the scale (Alvarenga et al., 2012, 2020).

To calculate the sample size, the OpenEpi^® software (online version) was used. The sample size was considered for a proportion or for a descriptive study, according to the formula: n = [EDFF × Np(1 − p)]/[(d²/Z²_{1 − α/2} × (N − 1) + p × (1 − p)], where n is the population size; EDFF, the drawing effect (equals 1); d, 5% significance level; and p, the expected frequency of the result in the population. According to data provided by the university, there were 17,796 students enrolled in onsite courses in the semester of data collection. Therefore, considering this number of students, the prevalence of disordered eating in university students (6.5%; Alvarenga et al., 2020), 95% confidence interval and maximum error of 5%, a minimum necessary sample of 93 participants was calculated for the study.

Data collection

The informed consent form and the self-reported data collection form were sent to students by email through a questionnaire registered on the Google Forms^® platform. The form contained the following fields: sociodemographic and academic data; variables related to the pandemic; life habits; anthropometrics; and eating behavior. Participants were informed about the inclusion and exclusion criteria and those who agreed to participate ticked the option “I accept” in the informed consent form.

Sociodemographic and academic variables

Age: students were asked about their age in years at the time of collection. This variable was treated as a numeric variable.

Sex: students were asked about their biological sex by checking one of the options “Female” or “Male.”

Course at the university and course time: students were asked about the course and period they were enrolled in. The courses were later categorized as being in the health field or not.

Variables related to the pandemic

The characteristics referring to the pandemic corresponded to the period of the month prior to the survey.

Signs/symptoms of the flu syndrome: students were asked whether they presented signs/symptoms of the flu syndrome.

Negative mood changes: students reported whether they perceived negative mood changes in the last month, presenting one or more of these feelings: anguish, fear, anxiety, anger, sadness, loneliness, boredom. Therefore, participants who marked one or more of these options were classified as “yes,” and if the students did not notice any changes, they marked the option “no.”

Life habits and anthropometrics variables

Practice of physical activity: students were asked if they were practicing physical activity, with the answer options “no” or “yes.”

Weight gain: students reported whether they had noticed weight gain in the last month.

Body weight, height, and Body Mass Index (BMI): students reported their current weight (in kg) and height (in m). These data were used to calculate the Body Mass Index (BMI), which was further classified according to the WHO criteria for adults as “underweight,” “normal range,” “overweight,” or “obese” (WHO, 2000).

Behavioral variables

Profiles on social media: students were asked whether they followed fitness/health profiles on social media.

Body perception and satisfaction: to assess body perception and satisfaction, the analysis instrument by Kakeshita et al. (2009) was used, which consists of silhouette scales developed to assess Brazilian children and adults of both sexes, validated for digital use by Freire and Fisberg (2017). Students observed the digital scale and answered two questions at each stage: “point to the figure that best represents your body today,” which was called actual BMI; “point to the figure that best represents the body you would like to have,” which was called the desired BMI. Real BMI was defined by self-reported weight and height information. The perception and satisfaction/dissatisfaction with body image was evaluated according to the discrepancies between the selected figures (Freire and Fisberg, 2017; Lira et al., 2017).

Disordered eating: the Disordered Eating Attitude Scale-short (DEAS-s) was used to assess disordered eating (study outcome variable; Alvarenga et al., 2020). DEAS-s measures disordered eating attitudes, defined as abnormal beliefs, thoughts, feelings, behaviors, and relationships in relation to eating. The scale consists of 17 questions, which are scored according to a template prepared by the authors of the instrument, with a range of points ranging from 17 to 80 points (Alvarenga et al., 2020). For this variable to be transformed into a binary categorical variable, the score of each question was added, and from the obtained median (23 points), the participants were distributed in the categories: “Below the median,” when the value was below 23 points, or “Equal to or above the median,” when the score was equal to or greater than this value. Higher scores on this scale represent more dysfunctional eating attitudes (Mazzolani et al., 2021).

Data analysis

Data were analyzed using IBM SPSS Statistics for Windows software, version 23.0 (IBM Corp, Armonk, NY, USA). The normality of the variables was assessed using the Shapiro–Wilk test. To describe the study variables, medians with interquartile ranges and absolute and relative frequencies were used. After distributing the variables between the categories determined according to the score on the scale that evaluates eating disordered (“Below the median,” and “Equal to or above the median”), the differences were analyzed: between the medians, the Mann–Whitney U test was used, and to analyze the differences in proportions, the chi-square (χ²) or Fisher’s exact test was used. To quantify the participation of independent variables in the outcome of interest (score equal to or above the median on the disordered eating scale), a multivariate analysis was performed, including, in the binary logistic regression model, the independent variables that presented a significance level of up to 20% in the bivariate tests. The hypothesis of absence of multicollinearity and the adjustment of the model according to the Hosmer–Lemeshow test were also considered. The significance level for all tests was 5%.

Results

The characterization of the sample can be seen in Table 1. A total of 936 students were evaluated, with a median age of 22 years, and most of them female (n = 667; 71.3%). Among the participants, most attended the first half of the course (n = 565; 60.4%) and were not from the field of health (n = 665; 71%).

Table 1.

Sociodemographic, academic, lifestyle, nutritional, and behavioral data of undergraduate students at a university in southeastern Brazil.

Variables	N	%
Age (median ± interquartile range)	22 ± 5	–
Sex
Male	269	28.7
Female	667	71.3
Course time
First half	565	60.4
Second half	371	39.6
Graduation field
Health	271	29
No health	665	71
Presence of signs/symptoms of the flu syndrome (last month)
No	708	75.6
Yes	228	24.4
Negative mood changes (last month)
No	172	18.4
Yes	764	81.6
Physical activity
No	415	44.3
Yes	521	55.7
Weight gain (last month)
No	502	53.6
Yes	434	46.4
Nutritional status
Normal range	514	54.9
Underweight	89	9.5
Overweight	210	22.4
Obese	123	13.1
Follow fitness/health profile on social media
No	619	66.1
Yes	317	33.9
Body perception
Adequate	140	15
Inadequate	796	85
Body dissatisfaction
No	95	10.1
Yes	841	89.9
Disordered eating (median ± interquartile range)	23 ± 12	–

N = 936.

Of the total number of participants, the majority reporting not having signs/symptoms of the flu syndrome (n = 708; 75.6%). Regarding negative mood changes during this period, this change was reported by 81.6% of participants (n = 764). The practice of physical activity were reported by more than half of the participants (n = 492; 52.6% and n = 521; 55.7%, respectively). Of the total, 434 students (46.4%) reported weight gain in the last month, although the majority had a normal weight (n = 514; 54.9%).

Regarding behavioral variables, the majority did not follow a fitness/health profile on social media (n = 619; 66.1%). Most students had inadequate body perception (n = 796; 85%), with a high percentage of body dissatisfaction (n = 841; 89.9%). Regarding the assessment of disordered eating, the median of the total score was 23 ± 12 points (minimum of 17 and maximum of 80 points).

Table 2 shows the sociodemographic, academic, lifestyle, nutritional, and behavioral data, distributed according to the median of the disordered eating score. The objective of this analysis was to verify which variables were associated with the outcome, and variables that presented a significance level greater than 20% were included in the multiple analysis. Being older (p = 0.036), having signs/symptoms of the flu syndrome (p = 0.006), negative mood changes (p < 0.001), weight gain (p < 0.001), and obesity (p < 0.001) were associated with the category equal to or above the median. Also, following a fitness/health profile on social media, presenting inadequate body perception and body dissatisfaction (p < 0.001 for all variables) were also associated with this category.

Table 2.

Sociodemographic, academic, lifestyle, nutritional, and behavioral data distributed according to the median of the total disordered eating score of undergraduate students at a university in southeastern Brazil.

Variables	Disordered eating score
	Below the median	Equal or above the median	p Value	Total
	N (%)	N (%)		N (%)
Age (median ± IQR) a	22 ± 4	22 ± 5	0.036*	–
Sex b			0.193
Male	138 (30.8)	131 (26.8)		269 (28.7)
Female	310 (69.2)	357 (73.2)		667 (71.3)
Course time c			0.894
First half	269 (60)	296 (60.7)		565 (60.4)
Second half	179 (40)	192 (39.3)		371 (39.6)
Graduation field c			0.220
Health	121 (27)	150 (30.7)		271 (29)
No health	327 (73)	338 (69.3)		665 (71)
Presence of signs/symptoms of the flu syndrome (last month) c			0.006*
No	357 (79.7)	351 (71.9)		708 (75.6)
Yes	91 (20.3)	137 (28.1)		228 (24.4)
Negative mood changes (last month) c			<0.001*
No	104 (23.2)	68 (13.9)		172 (18.4)
Yes	344 (76.8)	420 (86.1)		764 (81.6)
Physical activity c			0.429
No	205 (45.8)	210 (43)		415 (44.4)
Yes	243 (54.2)	278 (57)		521 (55.6)
Weight gain (last month) c			<0.001*
No	289 (64.5)	213 (43.6)		502 (53.6)
Yes	159 (35.5)	275 (56.4)		434 (46.4)
Nutritional status b			<0.001*
Normal range	274 (61.2)	240 (49.2)		514 (54.9)
Underweight	68 (15.2)	21 (4.3)		89 (9.5)
Overweight	75 (16.7)	135 (27.6)		210 (22.4)
Obese	31 (6.9)	92 (18.9)		123 (13.2)
Follow fitness/health profile on social media c			<0.001*
No	327 (73)	292 (59.8)		619 (66.1)
Yes	121 (27)	196 (40.2)		317 (33.9)
Body perception c			<0.001*
Adequate	99 (22.1)	41 (8.4)		140 (14.9)
Inadequate	349 (77.9)	447 (91.6)		796 (85.1)
Body dissatisfaction c			<0.001*
No	71 (15.8)	24 (4.9)		95 (10.1)
Yes	377 (84.2)	464 (95.1)		841 (89.9)

N = 936.

IQR: interquartile range.

^aMann–Whitney U test.

^bChi-square test.

^cFisher’s exact test.

^*Statistically significant values (p < 0.05).

Table 3 shows the results obtained with the multiple analysis by binary logistic regression. The hypothesis is that the variables that were associated with the outcome in the bivariate analyzes (sociodemographic—age and sex; related to the pandemic—presence of signs/symptoms of flu-like syndrome and negative changes in mood; weight gain; nutritional status; and behavioral variables—following fitness/health profile on social media and body perception/satisfaction) could impact on a lower or greater chance of a more disordered eating.

Table 3.

Multiple analysis according to the highest disordered eating score of undergraduate students at a university in southeastern Brazil.

Variables	Crude				Adjusted
	p Value	OR	CI95%		p Value	OR	CI95%
			Lower	Upper			Lower	Upper
Age	0.799	1.002	0.986	1.018	0.097	0.985	0.967	1.003
Sex
Male		1				1
Female	0.181	1.213	0.914	1.611	0.737	1.058	0.761	1.470
Presence of signs/symptoms of the flu syndrome (last month)
No		1				1
Yes	0.006*	1.531	1.131	2.074	0.006*	1.605	1.144	2.252
Negative mood changes (last month)
No		1				1
Yes	<0.001*	1.867	1.333	2.616	0.013*	1.628	1.110	2.387
Weight gain (last month)
No		1				1
Yes	<0.001*	2.347	1.803	3.054	<0.001*	1.739	1.298	2.329
Nutritional status
Normal range		1				1
Underweight	<0.001*	0.353	0.210	0.592	<0.001*	0.367	0.212	0.635
Overweight	<0.001*	2.055	1.476	2.862	0.002*	1.772	1.224	2.564
Obese	<0.001*	3.388	2.177	5.273	<0.001*	3.089	1.899	5.024
Follow fitness/health profile on social media
No		1				1
Yes	<0.001*	1.814	1.376	2.391	<0.001*	2.050	1.505	2.793
Body perception
Adequate		1				1
Inadequate	<0.001*	3.093	2.094	4.568	<0.001*	2.416	1.576	3.705
Body dissatisfaction
No		1				1
Yes	<0.001*	3.641	2.248	5.898	<0.001*	2.612	1.558	4.378

Binary logistic regression crude and adjusted. Hosmer e Lemeshow test: Sig. 0.451. R² Nagelkerke: 0.228.

OR: odds ratio; CI: confidence interval.

^*Statistically significant values (p < 0.05).

The results show that having signs/symptoms of the flu syndrome increased the chances of a score equal to or above the median on the disordered eating scale by 60.5% (OR = 1.605; CI = 1.144–2.252; p = 0.006). The presence of negative mood changes increased these chances by 62.8% (OR = 1.628; CI = 1.110–2.387; p = 0.013). In addition, weight gain also significantly increased the chances of a higher score on the DEAS-s (OR = 1.739; CI = 1.298–2.329; p < 0.001), with these chances being three times higher in those with obesity (OR = 3.089; CI = 1.899–5.024; p < 0.001), compared to those with normal BMI. In contrast, being underweight reduced the chances of a score equal to or above the median by 63% (OR = 0.367; CI = 0.212–0.635; p < 0.001).

Following a fitness/health profile on social media doubled the chances of participants scoring equal to or above the median on the DEAS-s (OR = 2.050; CI = 1.505–2.793; p < 0.001), an association also observed with inadequate body perception (OR = 2.416; CI = 1.576–3.705; p < 0.001) and body dissatisfaction (OR = 2.612; CI = 1.558–4.378; p < 0.001).

Discussion

It has been shown that signs/symptoms of the flu syndrome, negative mood changes, weight gain, obesity, following a fitness/health profile on social media, in addition to inappropriate body perception and body dissatisfaction, are associated with greater chances of a higher score on the disordered eating scale in undergraduate university students during the COVID-19 pandemic. Such results are relevant considering the strong impact on physical and mental health resulting from the presence of these disorders in the lives of this population (Ammar et al., 2020; Cecchetto et al., 2021).

Recent studies have shown changes in eating habits, lifestyle habits, and exacerbation of symptoms of dysfunctional eating behaviors both in the general population and in individuals with pre-existing disorders during this period (Di Renzo, Gualtieri, Pivari, et al., 2020; Phillipou et al., 2020), which makes it essential to study disordered eating in this context. There are several instruments developed to assess disordered eating, among which the DEAS-s stands out, which is a self-administered scale developed and validated for Brazilian university students, by Alvarenga et al. (2020).

Eating behavior is influenced by several factors (Unikel et al., 2016), and, in the present study, an association was observed between the presence of signs/symptoms of the flu syndrome and higher scores on the DEAS-s. This flu syndrome, in the context of the pandemic, can contribute to the presence of negative feelings (such as stress, fear, and anxiety), which culminate in an increase in dysfunctional eating practices (Cooper et al., 2020). Likewise, the presence of negative mood changes was significantly associated with higher scores for disordered eating in our study. This relationship must be carefully investigated, as it collaborates with the increase in the practice of dysfunctional eating behaviors (Saade et al., 2019).

In this context, the fear of contracting the disease was a condition present in society (Cooper et al., 2020), and in these situations there may be an increase in emotional eating. Emotion-guided eating is used to deal with situations where stress levels are high, and it is common to turn to food in response to negative emotions (Rodgers et al., 2012). This panorama is present even during the COVID-19 pandemic (Ammar et al., 2020; Brooks et al., 2020; Cooper et al., 2020; Di Renzo, Gualtieri, Cinelli, et al., 2020; Rodríguez-Pérez et al., 2020), contributing to changes in eating behaviors. Zachary et al. (2020) point out that, in the presence of stress and boredom, students increased their search for food. Flaudias et al. (2020) found an association between anxiety, stress, and greater practice of restrictive diets. Another important point to be considered is loneliness, a basic premise of this period of social distancing at the beginning of the pandemic, which is a risk factor for eating disorders (Levine, 2012).

Changes in eating behaviors can also be related to changes in body composition. In our study, almost half of the population reported having perceived weight gain during the first month of the social distancing measures, and this gain was positively associated with a higher score on the DEAS-s. Weight gain in this period has been portrayed as a consequence of measures of social restriction in the world population (Di Renzo, Gualtieri, Pivari, et al., 2020; Keel et al., 2020; Rodríguez-Pérez et al., 2020), which caused a reduction in physical activity and a greater search for food (Di Renzo, Gualtieri, Cinelli, et al., 2020; Sánchez-Sánchez et al., 2020).

Furthermore, being concerned about nutritional status is associated with greater concern about food (Keel et al., 2020). It is known that students who are overweight are more likely to develop disordered eating (Leal et al., 2020). Thus, a warning signal is lit, as, in the long term, these unhealthy weight loss practices, paradoxically, contribute to increased weight gain, favoring obesity (Haynes et al., 2018). Being overweight in the pre-pandemic period contributed to greater psychological distress (stress, depression, and emotional eating) and dysfunctional eating behaviors in young adults during the pandemic (Puhl et al., 2020).

Our data also showed that having obesity tripled the chances of a higher score on the disordered eating scale. Silveira, cited by Dunker et al. (2021), states that “[. . .] Obesity and disordered eating behaviors result from cultural contexts that motivate an unhealthy relationship with food, eating habits and physical activity” (p. 269). This relationship seems to be accentuated by the media, as social media reinforce cultural ideals of beauty (Stice, 2002). We identified a significant association between following a fitness/health profile on social media and higher scores on the DEAS-s. The use of social media is considered high among university students (Volpato and Motta, 2021), and the COVID-19 pandemic seems to have contributed to this growth, given that face-to-face meetings have been replaced by virtual ones (Pedral et al., 2020). This fact should be viewed with caution, especially among university students, given that the greater the time spent with social media, the greater are the concerns about weight and body shape (Rodgers et al., 2020). In addition, among those who spent more time watching the news regarding the pandemic, an increase in restrictive diet practices and binge-eating behaviors was observed (Flaudias et al., 2020).

Furthermore, it is important to highlight the relationship between adequate body perception and body satisfaction. A inadequate body perception refers to a discrepancy between the actual body image (represented here by the silhouette corresponding to the reported BMI) and the perceived one (according to the silhouette chosen as representative of the current body shape). In a recent publication, it was shown that those who perceived themselves as overweight were less likely to be satisfied with their body image (Ramos et al., 2022), which can impact eating behavior. Considering the relationship between overweight and disordered eating, it is known that young people with obesity who overestimate their body shape are more likely to engage in dysfunctional eating behaviors (Nunes et al., 2001) because they are more dissatisfied with their body image (Leal et al., 2020). This association was also evidenced in our study, which showed that having inadequate body perception and being dissatisfied with one’s own image were associated with greater chances of having more disordered eating.

The ideal of beauty determined by society, when not achieved, culminates in body dissatisfaction, with the consequent development of dysfunctional eating behaviors as a way to fit into the imposed beauty standards (Cruz-Sáez et al., 2020; Leal et al., 2020; Stice, 2002). Body dissatisfaction is influenced by several factors, including nutritional status (Mazzolani et al., 2021) and negative feelings (Czepczor-Bernat et al., 2021). In our study, even normal weight students had a high prevalence of body dissatisfaction, a relationship also found by other authors (Costa et al., 2019; da Mota Santana et al., 2021). It is important to highlight, therefore, that having normal BMI is not a protective factor against the absence of unhealthy practices for weight control and body dissatisfaction (Nunes et al., 2001), which demonstrates the need to act preventively in relation to dysfunctional eating behaviors, even when the nutritional status is within the parameters considered adequate.

The results of this study allowed us to assess the impact of the COVID-19 pandemic on the diet of university students, a public primarily affected in relation to body image issues and disordered eating, as discussed extensively above. Through this information, it is possible to develop measures for the remission, even if partial, of this situation, and to carry out preventive work in order to prepare students and health professionals for a possible similar scenario in the future.

Conclusion

This study demonstrated that having signs/symptoms of the flu syndrome, negative mood changes, weight gain, obesity, following a fitness/health profile on social media, in addition to inadequate body perception and body dissatisfaction, were associated with greater chances of higher scores on the disordered eating scale, in university students during the COVID-19 pandemic. As it is a cross-sectional study, it is not possible to determine causality between the variables. Data collection based on self-reported information also represents another limitation of this study. However, these results can contribute to the establishment of public measures aimed at mitigating the negative effect of disordered eating associated with the COVID-19 pandemic, given that the verified effects can last for long periods in these individuals. This study can also contribute to preventive and/or early measures in the face of changes in eating behavior in possible similar pandemic situations in the future.

Limitations

The data from this work are important to understanding the reality of this population and contribute to the formation of public policies aimed at these students, in the current and future context. However, the study has some limitations, and among these we highlight the transversality of the data, which does not allow inferring causality. In addition, data collection was carried out in an online format, using self-reported information. This was due to the current context of the pandemic, and several works cited in this study followed the same methodology, with no evident compromise in the reliability of the data. Longitudinal studies, with data collected in person, can be carried out in the future in order to provide more accurate data on the late effects of the pandemic.