Anxiety Connects Social Media Use to Food and Alcohol Disturbance and Disordered Eating When Social Support is Low: A Moderated Mediation Analysis

Abstract

Food and alcohol disturbance (FAD) is an understudied form of disordered eating, in which the consumption of calories is restricted in preparation for drinking alcohol. Guided by previous literature, the present study examined the direct, indirect, and interactive relationships between social media use, anxiety, social support, FAD, and disordered eating among young adults. Recruited from a large southwestern public university, the sample included 679 undergraduate students who completed an online survey in spring of 2022 and who indicated that they consume alcohol and are 18 to 29 years old. Two moderated mediation analyses assessed the indirect effects of anxiety on the relationships between social media use with FAD and disordered eating, and the conditional contribution of social support. Results indicated that social media use was related to disordered eating both directly and indirectly through anxiety, but it was only related to FAD through anxiety. Furthermore, indirect effects connecting social media use to FAD and disordered eating were conditional upon social support. Our findings suggest FAD and disordered eating may be coping mechanisms for anxiety stemming from social media exposure, though these associations appear to be attenuated when social support is high. As such, these findings may be relevant for shaping future intervention and prevention efforts for emerging adults experiencing FAD and disordered eating.

1. Introduction

Alcohol use and disordered eating are common among young adults, particularly those in college.^1,2,3,4 A growing body of research examines alcohol use and disordered eating together, in which calories are restricted in preparation for prolonged drinking episodes.⁵ This form of behavior is known as “food and alcohol disturbance” (FAD), and studies indicate that over 50% of U.S. college students have engaged in FAD behavior.⁶ FAD often involves harmful actions like fasting, excessive physical activity, and misusing laxatives.⁵ Reasons that people engage in FAD behavior include limiting cumulative caloric intake, enhancing alcohol’s effects, and saving money when going out.^7,8 Previously labeled “drunkorexia,” the term FAD was adopted by researchers due to the negative connotation associated with drunkorexia.⁹ However, inconsistencies in terminology, definition, and scale selection have limited scientific understanding of FAD. Despite this, adverse consequences such as unwanted/unplanned sexual activity, memory loss, and increased levels of violence/aggression are linked to FAD, highlighting the need for more research to explore the correlates of this health risk behavior.⁵

FAD is not currently recognized as a mental health disorder in the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition,¹⁰ and FAD’s classification as an alcohol use disorder (AUD) or eating disorder is currently debated. Literature suggests FAD shares more similarities with eating disorders than AUD,¹¹ but relative to AUD and other disordered eating behaviors, FAD is highly understudied. Research suggests many correlates of disordered eating also correlate with FAD.^11,12 Given the overlapping features between disordered eating and FAD, it is plausible to conceptualize FAD within Fairburn’s transdiagnostic model of eating disorders.¹³ This framework describes factors such as body satisfaction, mood regulation, and interpersonal difficulties as contributing to deliberate food restriction. One variable that can strongly relate to multiple dimensions of this model that has not been fully explored in the FAD literature is social media use (SMU).

1.1. Social Media Use: Associations with FAD and Disordered Eating

Research indicates a heightened vulnerability among young adults for developing substance use and eating disorders.^14,15 Young adults in college face a particular risk, partly because alcohol use and disordered eating are perceived as normative.^3,16–19 Notably, alcohol use and disordered eating are not confined to physical spaces but extend into the digital realm, particularly social media, where these behaviors are frequently depicted and endorsed.^20,21 Given the prevalence of SMU among young adults, the exposure to such norms on social media may contribute to increases in disordered eating behavior like FAD in this population.^22–27

In recent years, the proliferation of social media platforms has revolutionized the way people interact.²⁸ Although SMU offers a convenient way for people to connect with others, some research indicates that SMU negatively impacts mental health and maladaptive health behaviors.^29,31 According to social comparison theory, self-evaluation stems from comparisons to others.³² In the context of Fairburn’s transdiagnostic model, SMU may contribute to the overvaluation of shape and weight, fostering unrealistic beauty standards and promoting social comparison. In support of this, past studies indicate a positive relationship between SMU and FAD.³³

Research also demonstrates a strong relationship between the use of image-based social media platforms (e.g., Snapchat) and anxiety among college students.^34–36 Anxiety plays a significant role in the etiology and maintenance of eating disorders and can be linked to various components of Fairburn’s model. Anxiety-related concerns about body image and fear of weight gain can contribute to the overvaluation of shape and weight, reinforcing disordered eating behaviors.³⁷ Moreover, individuals with elevated anxiety levels may engage in dietary restraint as a maladaptive coping mechanism to alleviate feelings of distress or uncertainty. Further, there is evidence that SMU, anxiety, and forms of disordered eating may be linked.^38–42 That is, individuals may engage in disordered eating behaviors as coping mechanisms for anxiety.³⁷

While the potential adverse effects of SMU are apparent, it is important to note that SMU may also have beneficial impacts. For instance, SMU allows people to connect with loved ones more easily^43,44 and foster feelings of social support.⁴⁵ Social support refers to the interpersonal resources provided by others that can improve an individual’s well-being,⁴⁶ and aligns with Fairburn’s emphasis on interpersonal difficulties. Evidence suggests social support may provide emotional regulation and reduced isolation, which may buffer against maladaptive health outcomes.^47,48 Previous research indicates that social support is negatively correlated with anxiety, disordered eating, and FAD.^48–51 Furthermore, social support may buffer the relationship between anxiety and disordered eating.^52,53 However, although the literature has examined the interplay between psychosocial factors such as SMU, anxiety, and social support with disordered eating in young adults, an examination of these constructs with FAD is lacking.^5,54

1.2. Current Study

The aim of this study was to examine the direct, indirect, and interactive relationships between SMU, anxiety, and social support with FAD and disordered eating among college students. To do this, we used a moderated mediation framework to examine how anxiety accounts for the relationships between SMU with FAD and disordered eating, and how each direct and indirect pathway may be moderated by social support (see Figure 1).

Figure 1.

Conceptual model illustrating specified pathways for two separate moderated mediation analyses

The hypotheses for this study are provided below.

1. SMU would positively and directly relate to both FAD and disordered eating.

2. SMU would indirectly relate to FAD and disordered eating through anxiety.

3. The associations between SMU, anxiety, FAD, and disordered eating would be buffered by social support.

2. Method

2.2. Participants and Procedure

This cross-sectional survey was administered to undergraduate students enrolled in a psychology course at a southwestern public Hispanic Serving Institution during the spring semester of 2022. Prospective participants were invited to complete a 45-minute online survey on health behaviors administered through the departmental human subjects pool, which is comprised of both psychology and non-psychology majors. Interested participants accessed the Qualtrics-hosted study through a link, which displayed an online informed consent form. To counteract ordering and fatigue effects, questionnaire blocks that were not strategically positioned for purposes of survey logic (e.g., only showing alcohol items to participants reporting any drinking) were randomized within Qualtrics. Participants clicked “next” to indicate consent and begin the survey. Upon completion of the survey, participants were incentivized with course credit. All research and recruitment procedures were approved by the Institutional Review Board at the authors’ university. A total of 984 people clicked on the link to begin, with 858 participants completing the survey. For this analysis, participants who reported no alcohol use were excluded. Given the emphasis on young adulthood based on prior work, we restricted the sample to individuals between the ages of 18 and 29 to promote external validity. After accounting for these inclusion criteria, the analytic sample for the current study was 679 (82.5% female) ranging from 18 to 29 years old (M = 20.95, SD = 2.18). Participants self-identified as 41.4% White, 36.4% Hispanic, 10.6% Black, 1.9% Asian, and 8.2% multi-racial or as a member of a racial/ethnic group not listed.

2.3. Measures

2.2.1. Statistical Controls

Statistical analyses adjusted for the following: a) sex assigned at birth, coded as 0 = male and 1 = female; b) body mass index (BMI), calculated using self-reported height and weight (weight (lbs) / [height (in)]² x 703); and c) alcohol consumption, which was measured with a single item.⁵⁵ Participants were provided a definition of a standard drink and were asked how many drinks they consume in a typical day while drinking.⁵⁶ Responses were open-ended. These statistical controls were selected based on previous research that controlled for sex and BMI that may confound disordered eating, and typical alcohol consumption that may confound FAD behaviors.^2,5

2.2.2. Food and Alcohol Disturbance

The Drunkorexia Behaviors and Motives Scale (DMBS) contains a 12-item subscale assessing FAD behaviors (i.e., the ways they might restrict calories in anticipation of consuming alcohol later).⁵⁷ Participants indicate their frequency of each behavior (e.g., “By eating less all day”) using a 5-point Likert-type scale ranging from never to very often. Responses are summed, and higher scores represent higher levels of FAD behaviors. The DMBS behaviors subscale has been administered to college populations and found to have excellent reliability (α = .98).⁵⁸ Similarly, the DMBS behaviors subscale had excellent internal consistency in this sample (α = .96).

2.2.3. Disordered Eating

The Eating Disorder Diagnostic Scale (EDDS) consists of 22 items to assess disordered eating relevant to anorexia nervosa, bulimia nervosa, and binge-eating disorder.⁵⁹ The first 18 items of the EDDS scale are questions that comprise a symptom composite measure. Four of the questions ask participants to respond using a 7-point Likert-type scale ranging from not at all to extremely (e.g., “Over the past 3 months, has your weight influenced how you think about (judge) yourself as a person?”), eight questions ask participants to respond with either yes or no (e.g., “During episodes of overeating and loss of control, did you eat until you felt uncomfortably full?”), and six questions ask participants to respond with a number ranging from 0 to 14 (e.g., “How many times per week on average over the past 3 months have you made yourself vomit to prevent weight gain or counteract the effects of eating?”). To adjust for the different response formats, the scores for these questions were standardized and then summed to arrive at the symptom composite value, with higher scores suggesting higher levels of disordered eating symptoms.⁵⁹ Past studies have used the EDDS to measure disordered eating among college students and have found the EDDS to have good reliability (α = .89).⁶⁰ The EDDS symptom composite measure had excellent internal consistency in this sample (α = .91).

2.2.4. Anxiety

The Depression, Anxiety, and Stress Scale (DASS-21) includes a 7-item anxiety subscale.⁶¹ For each item (e.g., “I felt I was close to panic”), participants indicate how much the statements applied to them over the past week, using a 4-point Likert-type scale ranging from did not apply to me at all to applied to me very much or most of the time. Responses are summed, such that higher scores represent higher levels of anxiety. Researchers have used the DASS-21 to measure anxiety in college student populations and discovered the DASS-21 anxiety subscale to have good reliability (α = .85).⁶² The internal consistency of the DASS-21 anxiety subscale was good (α = .86) in this sample.

2.2.5. Social Support

The Multidimensional Scale of Perceived Social Support (MSPSS) is a 12-item measure of perceived social support from family, friends, and significant others.⁶³ For the MSPSS, participants responded to a 7-point Likert-type scale ranging from very strongly disagree to very strongly agree to indicate their level of agreement with the items about social support (e.g., “There is a special person who is around when I am in need”). The average score of all 12 items was computed for a total score of social support. Previous research validating the MSPSS in a sample of college students found the measure to have good internal consistency (α = .87).⁶⁴ Likewise, the MSPSS scale had excellent internal consistency (α = .92) in this study.

2.2.6. Social Media Use

A 10-point Likert-type scale was created for the purpose of this study assessing how often participants used any social media, with response options ranging from never to all the time. Higher scores on this measure reflect higher levels of SMU. This item closely resembles the single-item evaluations of SMU frequency employed in previous studies.⁶⁵ For descriptive purposes, participants also reported the specific social media platforms (e.g., Instagram, Snapchat, Facebook) they typically engage with.

2.3. Analytic Approach

All statistical analyses were conducted using IBM SPSS version 29. Hypothesis testing was conducted via moderated mediation analyses using the PROCESS macro for SPSS (Model 59).⁶⁶ We examined two models, one for each outcome (i.e., FAD and disordered eating). Both models examined pathways between SMU and the outcome through anxiety, and the moderating effects of social support on the indirect effect model. Confidence intervals (CIs) were estimated with 5,000 bootstrap samples, which allows for more precise and stable estimates of statistics and their CIs in the presence of non-normal distributions.⁶⁷ All continuous covariates forming interaction terms were centered. Interactions were considered significant if the p-value associated with them was less than 0.05 and interpreted by plotting and analyzing data points at +1 and −1 standard deviations from the mean. Both models were adjusted for participants’ BMI and sex, and the FAD model was adjusted for quantity of alcohol consumption.

4. Results

4.1. Preliminary Results

Missing data were assessed using Little’s MCAR test, and the data were found to be missing completely at random (χ² = 16.515, df = 18, p = .557). Skewness and kurtosis were within normal limits for the outcomes (FAD: skewness = 1.38, kurtosis = 0.97; disordered eating: skewness = 0.41, kurtosis = −0.79). Outliers were identified for FAD (n = 41) and disordered eating (n = 1); however, the outliers did not affect the interpretation of the results and all data were included in this analysis. Table 1 shows the descriptive statistics and bivariate correlations between the study’s variables. The results suggested that, at the zero-order level, SMU was positively associated with anxiety, social support, and disordered eating, though unrelated to FAD. Anxiety was negatively associated with social support, while positively related to disordered eating and FAD. Finally, social support was negatively correlated with disordered eating but was unrelated to FAD.

Table 1.

Descriptive Statistics and Zero-Order Correlations Among Study’s Variables

	M	SD	1	2	3	4	5	6	7	8
Outcomes
1. FAD	19.52	10.63	—
2.Disordered Eating	22.91	15.15	.45***	—
Covariates
3. Social Media Use	8.15	1.64	−.01	.19***	—
4. Anxiety	12.65	10.79	.22***	.40***	.11**	—
5. Social Support	5.34	1.38	−.06	−.17***	.09*	−.16***	—
Statistical Controls
6. Sex (female n/%)	560	82.5	.12**	.19***	.07	.14***	.04	—
7. BMI	25.56	8.06	.08*	.20***	.02	−.01	−.05	−.05	—
8. Alcohol	2.61	2.43	.18***	.04	.07	−.01	.12**	−.01	−.03	—

Note. FAD=Food and Alcohol Disturbance; BMI=Body Mass Index;

^***p<.001,

^**p<.01,

^*p<.05

4.2. FAD Model

The model examining FAD was significant and accounted for 11% of the variance in FAD (R = 0.33, F(8, 617) = 9.19, p < .001). SMU significantly and positively related to anxiety and did not directly relate to FAD after adjusting for alcohol use, sex, and BMI (see Table 2). The interaction between SMU and social support was non-significant, as was the interaction between anxiety and social support. In terms of the conditional effects of anxiety on FAD at different levels of social support, anxiety had a significant positive effect on FAD at all levels of social support. There were significant indirect effects of SMU on FAD via anxiety at lower and average levels, but not at higher levels of social support.

Table 2.

FAD Moderated Mediation Model

Outcome: FAD
Covariates	B	SE	t	p
Social Media Use	−0.29	0.25	−1.17	.243
Anxiety	0.41	0.08	5.25	<.001
Social Support	−0.29	0.31	−0.93	.351
Social Media Use * Social Support	−0.03	0.16	−0.21	.837
Anxiety * Social Support	−0.10	0.05	−1.87	.063
Statistical Controls
Sex (female)	3.01	1.10	2.73	.007
BMI	0.13	0.05	2.38	.018
Alcohol	0.81	0.17	4.87	<.001
Outcome: Anxiety
Covariates	B	SE	t	p
Social Media Use	0.36	0.13	2.77	.006
Social Support	−0.65	0.16	−4.14	<.001
Social Media Use * Social Support	−0.03	0.08	−0.32	.752
Statistical Controls
BMI	−0.12	0.03	−0.44	.662
Sex	1.87	0.56	3.31	.001
Alcohol	0.01	0.09	.10	.919
Conditional Direct Effects of Anxiety on FAD
	B	SE	LLCI	ULCI
Lower Social Support	−0.25	0.31	−0.86	0.37
Average Social Support	−0.30	0.26	−0.81	0.20
Higher Social Support	−0.34	0.36	−1.05	0.37
Conditional Indirect Effects of Social Media Use on FAD via Anxiety
	B	SE	LLCI	ULCI
Lower Social Support	0.21	0.11	0.01	0.43
Average Social Support	0.14	0.06	0.03	0.27
Higher Social Support	0.08	0.07	−0.02	0.26

Note: N = 679. BMI=Body Mass Index. LLCI = lower limit confidence interval, ULCI = upper limit confidence interval.

4.3. Disordered Eating Model

The model examining disordered eating was also significant and accounted for 26% of the variance in disordered eating (R = 0.51, F(7, 664) = 33.27, p < .001). SMU significantly and positively related to perceived anxiety and disordered eating after adjusting for sex and BMI (see Table 3). The interactions between SMU and social support and between anxiety and social support were non-significant. Similar to the FAD model, there were indirect effects of SMU on disordered eating via anxiety at lower and average levels of social support, but not at higher levels of social support.

Table 3.

Disordered Eating Moderated Mediation Model

Outcome: Disordered Eating
Covariates	B	SE	t	p
Social Media Use	0.86	0.24	3.65	<.001
Anxiety	0.73	0.07	10.01	<.001
Social Support	−1.15	0.28	−4.14	<.001
Social Media Use * Social Support	−0.08	0.15	−0.55	.582
Anxiety * Social Support	0.03	0.05	0.62	.535
Statistical Controls
Sex (female)	4.37	1.02	4.28	<.001
BMI	0.29	0.05	6.38	<.001
Outcome: Anxiety
Covariates	B	SE	t	p
Social Media Use	0.36	0.13	2.85	.005
Social Support	−0.65	0.15	−4.41	<.001
Social Media Use * Social Support	−0.03	0.08	−0.38	.70
Statistical Controls
BMI	−0.01	0.03	−0.11	.91
Sex (female)	2.09	0.54	3.88	<.001
Conditional Direct Effects of Anxiety on Disordered Eating
	B	SE	LLCI	ULCI
Lower Social Support	0.97	0.29	0.39	1.54
Average Social Support	0.84	0.24	0.36	1.31
Higher Social Support	0.74	0.34	0.07	1.39
Conditional Indirect Effects of Social Media Use on Disordered Eating via Anxiety
	B	SE	LLCI	ULCI
Lower Social Support	0.27	0.11	0.03	0.49
Average Social Support	0.26	0.09	0.08	0.45
Higher Social Support	0.24	0.14	−0.03	0.54

Note: N = 679. BMI=Body Mass Index. LLCI = lower limit confidence interval, ULCI = upper limit confidence interval.

4. Discussion

This study examined SMU, anxiety, social support, FAD, and disordered eating among young adults and contributes to the expanding literature on FAD behavior among young adults in an increasingly digital and interconnected world. Three hypotheses guided this study, and, while hypotheses were only partially supported, our findings illuminated the complex nature of these relationships, improving scientific understanding of how these factors collectively impact the well-being of college students.

First, we hypothesized that SMU would be positively and directly related to both FAD and disordered eating. Social media platforms contain curated images and content, which can elicit social comparison, body image concerns, and anxiety and may lead to maladaptive coping mechanisms, such as disordered eating behaviors.^38–42 Our findings on the relationship between SMU and disordered eating complements prior research, reinforcing the notion that SMU may play a role in the development of disordered eating behaviors. However, the lack of a direct relationship between SMU and FAD conflicts with prior research.³³ While our study shared similarities with research conducted by Foster et al³³ by focusing on a predominantly female college sample, distinctions in the FAD scale and SMU measurement might underlie these inconsistencies. First, Foster et al³³ used the Compensatory Eating and Behaviors in Response to Alcohol Consumption Scale to measure FAD⁶⁸ and our study employed the DMBS, which may elicit different aspects of FAD behavior. Second, Foster et al³³ included an empirical measure of SMU (i.e., screen time on social media applications evidenced by participant’s smartphones), which may provide a more objective evaluation of time spent on social media, reducing susceptibility to memory recall bias inherent in our study-generated single-item scale for SMU.

Our second hypothesis was that the relationship between SMU with FAD and disordered eating would operate through anxiety, based on past research supporting this link and the link between both FAD and disordered eating with anxiety.^37,54 This hypothesis was supported as the indirect effects of SMU to each outcome variable were significant. This suggests that SMU, although not directly related to FAD, relates to increased anxiety among individuals, which in turn relates to increased FAD behaviors. These findings align with the coping mechanism theoretical framework,⁶⁹ which suggests that individuals may resort to disordered eating patterns as a way to cope with heightened anxiety, aiming to regain a sense of control amid the challenges they face. This framework is particularly relevant for engaging in FAD as it also applies to alcohol use. That is, some individuals drink alcohol to cope with negative affect such as anxiety, and endorsing coping motives is a risk factor for problematic drinking behaviors.⁷⁰

Our final hypothesis was that social support would buffer the relationships between SMU, anxiety, FAD, and disordered eating. This hypothesis was partially supported. In both the FAD and disordered eating models, results indicated that the significant indirect pathway connecting SMU to FAD and disordered eating through anxiety was not significant when social support was higher. This suggests that the presence of strong social support networks provides alternative, healthier ways to address anxiety, reducing the reliance on FAD and disordered eating as coping strategies. Findings from past literature examining whether social support may buffer the negative influence of stressors on disordered eating are mixed. Some studies have found social support to be a significant moderator between poor mental health and disordered eating,^52,52,71 while others have not.^72,73 Our findings suggest that social support’s role in this context may be better understood within a moderated mediation framework. Specifically, the results of our study indicate that social support may not serve as a direct buffer between stressors and disordered eating behaviors, but rather as a contributor to the underlying mechanism that connects SMU to maladaptive eating behaviors.

4.1. Limitations and Future Directions

There are several limitations for the current study that warrant consideration. First, the cross-sectional design does now allow for causal inferences. Second, individuals might underreport sensitive behaviors like disordered eating, leading to potential measurement inaccuracies.^74,75 The reliance on self-reporting may also introduce recall and memory bias, affecting the precision of the data collected.^76,77 Next, it is worth noting that our sample was predominantly composed of females. While existing research currently suggests no gender differences in FAD,⁷⁸ research also indicates that females rely on social support networks more than males.⁷⁹ Therefore, this demographic skew may potentially constrain the generalizability of our findings. Also, the analytic sample consisted solely of college students and thus may not generalize to non-college young adult populations. In addition, previous research indicates that FAD engagement may vary as a function of race.⁸⁰ Unfortunately, we were not adequately powered to conduct meaningful subgroup analyses. Future research with larger and more diverse samples is needed to explore potential variations in FAD and related behaviors across different racial or ethnic groups effectively. Finally, we used the original EDDS to measure disordered eating. The EDDS has strong psychometric properties, can be computed to yield a total score of disordered eating, and provides valuable information pertaining to subclinical levels of disordered eating.⁸¹ However, the updated EDDS-5 more closely maps onto criteria for eating disorders established by the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition⁹ that would allow for a more nuanced and clinically relevant assessment of disordered eating.

4.2. Conclusion

This study implies that FAD and disordered eating may emerge as maladaptive responses to anxiety associated with SMU among young adult college students. As such, preventive measures could benefit from emphasizing more adaptive stress-coping responses, and providing comprehensive mental health support within academic institutions that considers lesser-known disordered eating behaviors such as FAD. Furthermore, the identification of the varied indirect associations between SMU, anxiety, and both FAD and disordered eating based on level of social support offers a unique opportunity for intervention. Recognizing the protective function of social support underscores the need to cultivate and strengthen peer and community support networks. Targeted intervention strategies may therefore focus on fostering an environment that encourages young adult college students to reach out for support, thereby mitigating the potential negative impact of SMU and anxiety on their health and well-being. Therefore, this study is a step toward improved understanding of FAD and disordered eating behaviors within college young adult populations, emphasizing not only individual coping strategies but also the vital role of interpersonal support networks in promoting students’ mental and emotional health.

Data Availability Statement

Data are available by request from the last author.