Loneliness is associated with eating disorders among a national sample of U.S. college students during the COVID-19 pandemic

Abstract

Objective:

To identify the association between loneliness and eating disorder symptomatology among a national sample of U.S. college students during COVID-19.

Participants:

Cross-sectional data from the 2020–2021 Healthy Minds Study (N = 96,645) were analyzed.

Methods:

Loneliness was measured using the UCLA 3-item Loneliness Scale and eating disorder symptomology was measured using the SCOFF questionnaire. Multiple modified Poisson regression analyses were conducted, adjusting for confounding variables.

Results:

Greater loneliness was associated with both a positive eating disorder screen (risk ratio [RR] 1.09, 95% confidence interval [CI] 1.09–1.10) and greater number of eating disorder symptoms (RR 1.07, 95% CI 1.06–1.08). Gender modified this relationship, and men who endorsed greater loneliness had higher risk of eating disorder symptomatology compared to women.

Conclusions:

Loneliness during the COVID-19 pandemic was associated with a greater risk of eating disorder symptomatology among college students. Findings underscore the need for social support and eating disorders programming on college campuses.

Introduction

Social support, or the emotional and material resources afforded to one through interpersonal relationships, has been identified as an important factor in protecting against the development of psychopathology.¹ Greater social support may reduce the risk of psychopathology through buffering the impact of stress, increasing adjustment, or some third variable such as personality traits.¹ Conversely, loneliness and isolation may contribute to and exacerbate mental health concerns through mechanisms such as disrupted interoceptive processing, altered reactivity within attentional and affective networks (i.e. attributing more attention to negative social stimuli), and changes in cognitive functioning.² Indeed, Quadt et al³ liken loneliness to a “social allostatic load” that produces wear and tear on the body through repeated activations. While research has typically focused on examining the impact of loneliness in older adulthood, emerging adults are still at high risk for loneliness⁴ and concurrently there are many risk factors present during this developmental period associated with the development of psychopathology.⁵

Socio-emotional difficulties are theorized to play an important role in eating disorder development, and maintenance,⁶ with research indicating that people with eating disorders are highly sensitive to social rejection,⁷ are hypervigilant to social threats (i.e. angry faces),^{7, 8} engage in unfavorable social comparisons,⁹ and exhibit difficulties with labeling and regulating emotions.¹⁰ Eating disorder behaviors may increase in response to events that arouse negative emotions, especially as individuals seek to regulate such emotions.^{11, 12} For example, suppression and rumination of unwanted emotions such as anger and loneliness increase symptoms among people with binge eating disorder.¹³ Paradoxically, eating disorder behaviors are often secretive and isolating, which may increase negative affect and risk of social isolation.¹⁴ However, it is important to note that some research indicates emotion regulation difficulties are not a key factor involved in the relationship between eating disorders and loneliness.¹⁵

Additionally, prior research suggests that weight-related teasing and bullying, negative self-perceptions, and biased perceptions of social relationships and situations may contribute to the relationship between loneliness and eating disorders.^{16, 17} Taken together, difficulties with social relationships, especially feelings of inadequacy (related to social comparison) coupled with maladaptive or ineffective emotion regulation strategies may increase feelings of loneliness which individuals seek to regulate via eating disorder behaviors. Furthermore, eating disorder behaviors themselves may isolate affected persons from social networks.¹⁸ Indeed, loneliness has been posited as both a risk and maintenance factor for eating disorders.¹⁴

While concerns about college students’ mental health was mounting prior to the COVID-19 pandemic, the COVID-19 pandemic had a notable impact on their mental health. College students have reported difficulty coping with disruptions caused by the pandemic, and are reporting increases in stress, anxiety, depression, loneliness, and social isolation.^{19, 20} The increase in loneliness and social isolation are particularly concerning given that emerging adults are already a developmentally vulnerable population for eating disorders²¹ and such additional risk factors may compound risk. Indeed, several studies have noted increases in eating disorder symptoms during the course of the COVID-19 pandemic.^{22, 23} Certain situational factors arising during the pandemic such as stress, anxiety, reduced social support, food insecurity, body and weight-focused messages from media, restricted access to healthcare may also increase eating disorder risk during the pandemic.²⁴ Notably, a small sample of people receiving eating disorder treatment reported both positive and negative effects of social isolation on eating disorder symptoms during COVID-19. Participants attributed having less privacy to reduced binge eating, but conversely, participants also reported that increased social isolation and eating due to loneliness worsened symptoms.²⁵

While several empirical studies have noted the link between loneliness and isolation during COVID-19 and depression and anxiety,^{26, 27} fewer have looked at the relationship between loneliness during COVID-19 and eating disorder symptomatology.²⁵ As such, the aim of the current study was to examine the association between loneliness and eating disorder symptomatology. Given previous research indicating a relationship between loneliness and psychopathology generally, as well as with eating disorders specifically, we hypothesized that loneliness would be positively associated with eating disorder symptomatology. As a secondary exploratory aim, we assessed whether gender modified the relationship between loneliness and eating disorder symptomatology. Prior research has documented gender differences for eating disorders^28–30 and loneliness,⁴ underscoring the need to understand whether this identifier is a modifying factor for loneliness and eating disorders. Given this exploratory aim, there was no a priori hypothesis.

Methods

Data are from the 2020–2021 Healthy Minds Study (HMS). HMS is an annual, cross-sectional study of college and university students across the U.S. HMS surveys participants on a range of biological, psychological, and social health factors. Colleges and universities voluntarily elect to participate. At institutions with ≥4000 students, 4000 students were randomly recruited to participate in the survey; at institutions with <4000 students, all students were recruited. To be eligible to participate, students must have been at least 18-year old. Students were invited to participate via email and the survey was administered online via Qualtrics. Students were incentivized to participate using a drawing for Amazon gift cards. Informed consent was obtained from all participating students. HMS was approved by all participating institutions. Given the data are anonymous and publicly available, this analysis was exempt from further ethics approval.

A total of 140 colleges and universities participated in the 2020–2021 HMS. The response rate for the fall 2020 study was 14%, while the response rate for the spring 2021 study was 15%. Overall, a total of 137,916 participants completed the survey. This analysis was restricted to participants who were between the ages of 18–26 years (N = 110,915), representing over 80% of the sample, given that this time period aligns closely with the traditional college years and emerging adulthood,³¹ and is a critical period for the onset of eating disorders.²¹ Additionally, participants were only included in this study if they completed the eating disorder and loneliness items, resulting in a final analytic sample of 96,645 participants.

Measures

Dependent variable

The eating disorder screen was measured using the five-item SCOFF questionnaire.³² A positive eating disorder screen was measured by a score of ≥2 “yes” responses, which has been shown to have a sensitivity of 86% and a specificity of 83%.³³

Independent variables

Loneliness was measured using the UCLA 3-item Loneliness Scale.³⁴ A sum score is created by adding the total of the three items resulting in a range of 3–9, with higher scores representing greater loneliness.³⁴ Interval reliability of the three items was good with α = 0.85.

Demographic and confounding variables

Demographic variables included self-reported age, gender, race/ethnicity, and sexual orientation. Confounding variables included residence, relationship status, current financial stress, sense of belonging to the campus community, any alcohol use in the past 2 weeks, depression score,³⁵ and anxiety score.³⁶ The demographic and confounding variables were controlled for in our analyses given that they may contribute to both eating disorder symptoms and loneliness.^{4, 28, 30, 37}

Statistical analysis

Descriptive statistics were performed to characterize the sample. Multiple modified Poisson regression analyses³⁸ were conducted and risk ratios (RR) and 95% confidence intervals (CI) were used to determine the association between the loneliness and eating disorder screen. An additional regression model was conducted to test for effect modification by gender. All regression analyses adjusted for the demographic and confounding variables. Preconstructed nonresponse sample weighting was applied to all analysis. Sample weights were constructed based on gender, race/ethnicity, academic level, and grade point average. Participants with underrepresented demographic characteristics are assigned greater sample weights.³⁹ All analyses were conducted in 2022 using Stata 17.

Results

Overall, the sample of 96,645 participants was demographically diverse (Table 1). Slightly more than a quarter (28.2%) of the sample screened positive for an eating disorder, while the mean score on the loneliness scale was 5.8 (SD 2.0).

Table 1.

Characteristics of participants from the 2020–2021 Healthy Minds Study (N = 96,645).

	Mean (SD) / %
Age	20.7 (2.2)
Gender
Cisgender women	56.8
Cisgender men	39.0
TGNB	4.2
Race/ethnicity
White or Caucasian, non-Hispanic, non-Arab	60.3
Black or African American, non-Hispanic	10.7
Hispanic/Latino/a	7.9
Asian or Asian American	8.9
Other race/ethnicitya	2.4
More than 1 race/ethnicity	9.8
Sexual orientation
Heterosexual	74.9
Gay or lesbian	6.0
Bisexual	12.0
Queer, questioning, or other	7.1
Residence
On-campus housing, residence hall	24.3
On-campus housing, apartment	5.6
Fraternity or sorority	1.1
On- or off-campus co-operative housing	1.2
Off-campus, non-university housing	30.2
Off-campus, with my parents (or relatives)	35.7
Other	1.7
Relationship status
Single	56.9
In a relationship	37.6
Married, in a domestic partnership, or engaged	4.6
Divorced, separated, or widowed	0.1
Other	0.7
Financial stress, current
Never stressful	7.3
Rarely stressful	19.7
Sometimes stressful	34.7
Often stressful	23.6
Always stressful	14.7
Sense of belonging to the campus community
Overall agree	65.3
Overall disagree	34.7
Any alcohol use, past 2 weeks	47.6
Depression score, PHQ-9	9.5 (6.8)
Anxiety score, GAD-7	8.1 (6.1)
Positive eating disorder screen, ≥ 2 on SCOFF	28.2
SCOFF score
0	48.2
1	23.6
2	15.5
3	8.7
4	3.3
5	0.7
Loneliness score, range 3–9	5.8 (2.0)

Note: Preconstructed nonresponse sample weighting was applied to all analyses.

PHQ: Patient Health Questionnaire; GAD: Generalized Anxiety Disorder; TGNB: Transgender/gender non-binary.

^aOther race/ethnicity included participants who identified as Middle Eastern/Arab, Native Hawaiian, Pacific Islander, American Indian, or Alaskan Native.

Results from Poisson regression analyses showed significant associations between loneliness and eating disorder symtomatology (Table 2). Higher scores on the loneliness scale were associated with both a positive eating disorder screen and greater number of eating disorder symptoms, while adjusting for demographic and confounding variables. Gender modified the association between loneliness and eating disorder symptomatology, whereby men with higher loneliness scores had significantly higher risk of both a positive eating disorder screen (RR 1.05, 95% CI 1.03–1.07 for interaction) and greater number of eating disorder symptoms (RR 1.03, 95% CI 1.02–1.05 for interaction) compared to women (p < .001 for interactions).

Table 2.

Associations between loneliness and eating disorder symptomatology among participants in the 2020–2021 Healthy Minds Study (N = 96,645).

	Positive eating disorder screena		SCOFF scoreb
	RRc (95% CI)	p	RRc (95% CI)	p
Loneliness scored	1.09 (1.09–1.10)	<.001	1.07 (1.06–1.08)	<.001

Note: Preconstructed nonresponse sample weighting was applied to all analyses.

Boldface indicates statistical significance (p < .05).

RR: Risk ratio; CI: Confidence interval.

^aPositive eating disorder screen indicated by a score of ≥2 on the SCOFF questionnaire.

^bRange: 0–5 with higher scores indicating greater eating disorder symptoms.

^cAdjusted for age, gender, race/ethnicity, sexual orientation, residence, relationship status, financial stress, sense of belonging to the campus community, any alcohol use, past 2 wk, depression score, and anxiety score.

^dRange: 3–9 with higher scores indicating greater loneliness.

Discussion

The findings from the study confirm the study hypothesis that loneliness is associated with a greater risk of a positive eating disorder screen, as well as a greater number of eating disorder symptoms. These findings are particularly salient as they are independent of key confounding variables, including sense of belong to the campus community, relationship status, residence, and depression and anxiety symptomatology. Findings also underscored that loneliness among men significantly increased the risk of eating disorders symptoms compared to women. The relationship between loneliness and eating disorders are contextualized by the COVID-19 pandemic, where significant changes to the delivery of higher education was occurring.

The association between loneliness and eating disorders among college students underscores the impact of loneliness on mental health² and expands prior research on the impacts of loneliness that has typically focused on older populations.⁴ Given the concerns regarding the mental health of college students and the ever-evolving social and psychological impacts of the COVID-19 pandemic,^{19, 20} the findings from the study provide important implications for health care and college health professionals. Specifically, health professionals should understand the adverse impacts of loneliness on the mental health of young people, which may, in turn, be associated with maladaptive coping strategies (e.g. eating disorder behaviors). Thus, the need for effective screening and intervention strategies (e.g., perspective-taking, being present)⁴⁰ are needed to mitigate both of these issues. Educators in higher education should also be aware of the relationship between loneliness and mental health, as they may be in an optimal position to facilitate social connectedness within their classrooms. Additionally, given that the COVID-19 pandemic has restricted access to treatment,²⁴ there is a need for awareness campaigns to inform college students of the effects of loneliness and resources available to reduce such feelings. These implications may be particularly important for male college students.

Strengths of this study in the use of a large, national sample of college and university students representing campuses across geographic regions of the U.S. However, this study is not without limitations. First, all items are based on self-report, which increases the risk of reporting bias. Second, the SCOFF questionnaire is a screening tool that does not provide a depth and breadth of eating disorder symptomology as other commonly used measures (e.g. Eating Disorder Examination Questionnaire). However, the SCOFF may be a more practical eating disorder measure to include in clinical settings and large-scale survey research given the concise 5-question form. Additionally, we analyzed the SCOFF in two forms, resulting in similar findings, adding further support for the findings. Third, the data are cross-sectional which precludes causal conclusions. Furthermore, there are likely multiple mechanisms that explain the relationship between loneliness and eating disorder symptoms that were not present in the HMS survey. Future research should aim to identify these specific mechanisms. Fourth, there is the potential for nonresponse bias due to low response rates. However, the overall sample was over 96,000 students and we applied preconstructed nonresponse sample weighting to all our analyses to reduce the threat of bias. Additionally, surveys such as HMS that have a large sampling frame often obtain reliable results.⁴¹ Lastly, the results should be considered within the context of the relatively small effect sizes.

Conclusion

Among a large, national sample of U.S. college students, loneliness is associated with eating disorder symptomatology, including both a positive eating disorder screen and a greater number of eating disorder symptoms. These findings expand prior research on the impacts of loneliness on poor mental health and are contextualized by the ongoing COVID-19 pandemic. These findings are particularly relevant to health care and college health professionals and underscore the need for multiple awareness, assessment, and intervention strategies to promote the health and well-being of college students.

Abbreviations

CI

confidence interval

HMS

Healthy Minds Study

RR

risk ratio

Availability of data and material

The Healthy Minds Study is available to researchers. Please visit http://healthymindsnetwork.org for more information.