1 INTRODUCTION
The coronavirus (COVID‐19) pandemic (WHO, 2020) has resulted in high death and unemployment rates and disruptions in daily living, which may contribute to increased mental health concerns, including elevated anxiety (Wang et al., 2020). Given established links between anxiety and eating disorder (ED) pathology (Swithbourne & Touyz, Swinbourne & Touyz, 2007) and compulsive exercise (Meyer, Taranis, Goodwin, & Haycraft, 2011), COVID‐19 may exacerbate risk for ED pathology and compulsive exercise (Cooper et al., 2020; Fernández‐Aranda et al., 2020; Rodgers et al., 2020; Touyz, Lacey, & Hay, 2020). Additionally, ED and compulsive exercise behaviors can function to reduce anxiety (Engel et al., 2016; Goodwin, Haycraft, & Meyer, 2012), and therefore may assist in managing COVID‐19 anxiety. Currently, research examining the impact of COVID‐19 on ED pathology and compulsive exercise is limited. Preliminary work evidences COVID‐19 related increases in compulsive exercise and ED behaviors (Baenas et al., 2020; Branley‐Bell & Talbot, 2020; Papandreou, Arija, Aretouli, Tsilidis, & Bulló, 2020; Phillipou et al., 2020; Schlegl, Maier, Meule, & Voderholzer, 2020; Termorshuizen et al., 2020); however, more work is necessary to understand factors impacting risk.
COVID‐19 anxiety may be generally associated with negative psychological outcomes; however, individual factors likely impact the extent to which it increases risk for ED pathology and compulsive exercise. Intolerance of uncertainty, the tendency to experience significant distress in response to uncertain situations (Birrell, Meares, Wilkinson, & Freeston, 2011), may possibly exacerbate risk for ED behaviors. Intolerance of uncertainty is a transdiagnostic factor (Mahoney & McEvoy, 2012) that is hypothesized to promote increased compulsive or “safety” behaviors (Boswell, Thompson‐Hollands, Farchione, & Barlow, 2013; Holaway, Heimberg, & Coles, 2006) in response to perceived uncertainty. Intolerance of uncertainty also contributes to understanding the etiology and maintenance of EDs (for reviews, see Brown et al., 2017; Kesby, Maguire, Brownlow, & Grisham, 2017) and is associated with increased compulsive exercise (Scharmer, Reilly, Gorrell, & Anderson, 2020), such that ED and compulsive exercise behaviors may function to alleviate distress related to uncertainty (Brown et al., 2017). Thus, intolerance of uncertainty may impact ED and compulsive exercise risk related to COVID‐19 distress.
Examining the impacts of COVID‐19 specific intolerance of uncertainty may also inform our understanding of ED and compulsive exercise risk. Although similar to trait intolerance of uncertainty, COVID‐19 intolerance of uncertainty refers to reactions to pandemic‐related uncertainty and may differ from trait intolerance of uncertainty within an individual. For example, an individual with low trait intolerance of uncertainty may experience more distress in reaction to COVID‐19 uncertainty. Recent work suggests that reactions to uncertainty in specific domains are more strongly associated with maladaptive behaviors and psychological distress (e.g., Mahoney & McEvoy, 2012). Accordingly, individual differences in COVID‐19 intolerance of uncertainty may uniquely impact risk for psychological distress and associated safety behaviors (i.e., ED and compulsive exercise behaviors). COVID‐19 represents an important source of stress (Gruber et al., 2020) and uncertainty about the pandemic is pervasive. Thus, examining risk specifically related to COVID‐19 intolerance of uncertainty may further assist in understanding the impact of COVID‐19 on ED pathology and compulsive exercise.
The current study explored ED and compulsive exercise risk during the COVID‐19 by examining how COVID‐19 anxiety impacts intolerance of uncertainty, ED pathology, and compulsive exercise. We predicted that higher COVID‐19 anxiety, trait intolerance of uncertainty, and COVID‐19 intolerance of uncertainty would be associated with more severe compulsive exercise and ED pathology. We also predicted that greater trait and COVID‐19 intolerance of uncertainty would increase risk for ED pathology and compulsive exercise related to COVID‐19 anxiety.
2 METHOD
2.1 Participants and procedures
Undergraduate participants (N = 295; 65.1% female) were recruited for a study examining uncertainty and coping during COVID‐19 through the research pool at a large university in the northeastern United States. A majority of participants (Mage = 19.7, SD = 2.0) resided in New York state (95.0%) and reported social distancing (92.5%) during data collection (March–April 2020). Additional participant information appears in supplementary materials. All participants provided informed consent, completed online questionnaires, and received course credit for participation. The study was approved by the university’s institutional review board.
2.2 Measures
2.2.1 Eating Disorder Examination‐Questionnaire
The Eating Disorder Examination‐Questionnaire (EDE‐Q) is a widely validated (Berg, Peterson, Frazier, & Crow, 2012; Fairburn & Beglin, 1994), 36‐item self‐report questionnaire that assesses cognitive and behavioral ED symptoms. The global EDE‐Q score was the primary measure of ED pathology and demonstrated high internal consistency (Cronbach’s α = .95).
2.2.2 Fear of illness and virus evaluation
The fear of illness and virus evaluation (FIVE) is a 35‐item measure that assesses fear and anxiety associated with infectious disease (Ehrenreich‐May, 2020). The current study modified the original version of the FIVE to specifically reference the “coronavirus public health emergency.” This scale demonstrated high internal consistency (Cronbach’s α = .94).
2.2.3 State‐Trait Anxiety Inventory—trait subscale
The State‐Trait Anxiety Inventory—trait subscale (STAI‐T) is a widely used measure of trait anxiety that has strong psychometric properties (Barnes, Harp, & Jung, 2002; Spielberger, 1983). In our sample, internal consistency for the STAI‐T was good (Cronbach’s α = .91).
2.2.4 Compulsive exercise test
The compulsive exercise test (CET) is a 24‐item self‐report questionnaire that assesses cognitive and behavioral features of compulsive exercise (Taranis, Touyz, & Meyer, 2011). The CET total score was a general measure of compulsive exercise in the current study and demonstrated internal consistency (Cronbach’s α = .87).
2.2.5 Intolerance of uncertainty scale‐short form
The intolerance of uncertainty scale‐short form (IUS)‐12 is a 12‐item, measurement of general intolerance of uncertainty (Carleton et al., 2007). In our sample, internal consistency was good (Cronbach’s α = .93). A total intolerance of uncertainty score was calculated by summing the Prospective and Inhibitory intolerance of uncertainty subscales, representing the primary measure of trait intolerance of uncertainty.
2.2.6 Intolerance of COVID‐19 uncertainty scale
Similar to past research exploring specific intolerance of uncertainty (Mahoney & McEvoy, 2012), we altered IUS‐12 items to specifically assess intolerance of uncertainty related to the coronavirus pandemic. Internal consistency for the revised measure was good (Cronbach’s α = .94).
2.2.7 The Godin leisure‐time exercise questionnaire
The Godin leisure‐time exercise questionnaire (GLETQ) is a three‐item self‐report questionnaire that assesses the frequency and intensity of physical activity over a seven‐day period, resulting in an overall GLTEQ score (Godin, 2011; Godin & Shephard, 1985). Participants reported frequency and intensity of exercise over the past week and in a typical week prior to COVID‐19.
2.3 Analytic plan
Bivariate correlations examined associations between COVID‐19 anxiety, trait anxiety, trait intolerance of uncertainty, COVID‐19 intolerance of uncertainty, ED pathology, and compulsive exercise. Additionally, multiple linear regression analyses, controlling for gender and STAI‐T, using Hayes’ PROCESS (Hayes, 2013) examined interactions between COVID‐19 anxiety and trait and COVID‐19 intolerance of uncertainty in models predicting ED pathology and compulsive exercise. Additional information about study analyses is available in the supporting materials.
3 RESULTS
Participants’ reported mean scores on the EDE‐Q, CET, and IUS‐12 were within on standard deviation of those from similar samples at the same institution prior to COVID‐19 (e.g., Scharmer et al., 2020) (Table 1). Participants reported moderate physical activity on the GLETQ (M = 33.96, SD = 33.70). Results from dependent samples t‐tests indicated reductions in exercise associated with COVID‐19, t(243) = 6.43, p < .001, d = .41.
Descriptive statistics and bivariate correlations for study variables
| Variable | M | SD | 2 | 3 | 4 | 5 | 6 |
|---|---|---|---|---|---|---|---|
| 1. EDE‐Q | 1.67 | 1.31 | .49* | .46* | .36* | .35* | .30* |
| 2. CET | 12.04 | 3.29 | — | .13 | .12 | .21* | .14 |
| 3. STAI‐T | 46.36 | 10.31 | — | .34* | .47* | .41* | |
| 4. FIVE | 47.51 | 15.46 | — | .47* | .60* | ||
| 5. IUS‐12 | 31.02 | 10.61 | — | .67* | |||
| 6. COVID‐19 IUS‐12 | 31.14 | 11.93 | — |
- Note: N = 295. Modified significance level p < .001.
- Abbreviations: CET, Compulsive Exercise Test; COVID‐19 IUS‐12, COVID‐19‐specific adapted forms of IUS‐12; EDE‐Q, Eating Disorder Examination‐Questionnaire; FIVE, Fear of Illness and Virus Evaluation; IUS‐12, Intolerance of Uncertainty Scale‐12; STAI‐T, State–Trait Anxiety Inventory—Trait form.
Bivariate correlations appear in Table 1. EDE‐Q scores were positively associated with COVID‐19 anxiety, trait intolerance of uncertainty, and COVID‐19 intolerance of uncertainty. CET scores were positively associated with trait intolerance of uncertainty but not COVID‐19 anxiety and intolerance of uncertainty.
Multiple regression analyses yielded significant models for CET and EDE‐Q scores (Table 2). Interaction terms in both models were significant; however, contrary to hypotheses, associations between COVID‐19 anxiety and CET/EDE‐Q scores were stronger for individuals with lower trait and COVID‐19 intolerance of uncertainty.
Model fit statistics, regression coefficients, and dominance weights for models predicting EDE‐Q and CET scores
| DV | Model | Predictors | R2 | ΔR2 | F | b | SE (b) | 95% CI | t | p | Dominance R2 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CET | 1 | .064 | 3.438 | .005 | ||||||||
| Gender | −.590 | 0.417 | −1.412, .232 | −1.414 | .159 | .001 | ||||||
| STAI‐T | .016 | 0.022 | −.028, .060 | .722 | .471 | .019 | ||||||
| FIVE | .013 | 0.014 | −.015, .041 | .932 | .352 | .002 | ||||||
| IUS‐12 | .059 | 0.023 | .014, .104 | 2.607 | .010 | .040 | ||||||
| FIVE*IUS‐12 | .013 | 4.184 | −.002 | 0.001 | −.004, .00 | −2.045 | .042 | |||||
| CET | 2 | .051 | 2.759 | .019 | ||||||||
| Gender | −.500 | 0.424 | −1.335, .335 | −1.179 | .239 | |||||||
| STAI‐T | .032 | 0.022 | −.012, .075 | 1.425 | .155 | |||||||
| FIVE | .020 | 0.015 | −.009, .049 | 1.369 | .172 | |||||||
| COVID IU | .027 | 0.023 | −.017, .072 | 1.206 | .229 | .019 | ||||||
| FIVE*COVID IU | .020 | 5.507 | −.002 | 0.001 | −.004, .000 | −2.347 | .020 | |||||
| EDE‐Q | 3 | .319 | 26.693 | .000 | ||||||||
| Gender | .198 | 0.059 | .083, .314 | 3.372 | .001 | .060 | ||||||
| STAI‐T | .018 | 0.003 | .012, .024 | 5.593 | .000 | .119 | ||||||
| FIVE | .007 | 0.002 | .003, .012 | 3.096 | .002 | .052 | ||||||
| IUS‐12 | .004 | 0.004 | −.003, .011 | 1.120 | .264 | .039 | ||||||
| FIVE*IUS‐12 | .027 | 9.446 | −.001 | 0.000 | −.001, .000 | −3.074 | .002 | |||||
| EDE‐Q | 4 | .316 | 27.774 | .000 | ||||||||
| Gender | .213 | 0.060 | .094, .332 | 3.526 | .001 | |||||||
| STAI‐T | .012 | 0.003 | .013, .025 | 6.322 | .000 | |||||||
| FIVE | .008 | 0.003 | .003, .014 | 3.074 | .002 | |||||||
| COVID IU | .001 | 0.004 | −.006, .008 | .312 | .756 | .024 | ||||||
| FIVE*COVID IU | .027 | 10.851 | −.001 | 0.000 | −.001, .000 | −3.294 | .001 | |||||
- Note: N = 295. Bold values indicate p < .05.
- Abbreviations: CET, Compulsive Exercise Test; CI, confidence interval; COVID IU, COVID‐19‐specific adapted forms of IUS‐12; EDE‐Q, Eating Disorder Examination‐Questionnaire; FIVE, Fear of Illness and Virus Evaluation; IUS‐12, Intolerance of Uncertainty Scale‐12; STAI‐T, State–Trait Anxiety Inventory—Trait form.
3.1 Post hoc analyses
Given that results contrasted hypotheses, we aimed to better understand determinants of risk for ED pathology and compulsive exercise by further examining important predictors in each model through dominance analyses (Budescu, 1993). Results indicated that trait intolerance of uncertainty was the strongest predictor of compulsive exercise, and trait anxiety most strongly predicted ED pathology (Table 2).
4 DISCUSSION
This study explored risk for ED pathology and compulsive exercise associated with COVID‐19 anxiety. Findings indicated that higher levels of COVID‐19 anxiety and COVID‐19 intolerance of uncertainty were associated with more severe ED pathology, but not compulsive exercise. However, trait intolerance of uncertainty was associated with both ED pathology and compulsive exercise. Results also suggested that risk for compulsive exercise and ED pathology associated with COVID‐19 anxiety is higher for individuals with lower levels of trait and COVID‐19 intolerance of uncertainty.
Our findings are consistent with existing work documenting associations between trait intolerance of uncertainty, ED pathology, and compulsive exercise (Brown et al., 2017; Kesby et al., 2017; Scharmer et al., 2020). Results indicating that higher levels of COVID‐19 anxiety and intolerance of uncertainty were associated with more severe ED pathology are also consistent with recent work evidencing COVID‐19 increases in ED behaviors (e.g., Branley‐Bell & Talbot, 2020; Phillipou et al., 2020) and publications predicting increases in ED risk associated with COVID‐19 anxiety (e.g., Cooper et al., 2020; Rodgers et al., 2020). However, findings that neither COVID‐19 anxiety nor COVID‐19 intolerance of uncertainty were associated with compulsive exercise are surprising given established links between compulsive exercise and anxiety (Meyer et al., 2011) and intolerance of uncertainty (Scharmer et al., 2020). This finding suggests that COVID‐19 anxiety and intolerance of uncertainty do not impact risk for compulsive exercise. Although these results contrast our hypotheses, they align with participants’ reports that compulsive exercise did not increase in response to COVID‐19 and recent work indicating COVID‐19 related reductions in exercise in nonclinical ED samples (e.g., Phillipou et al., 2020). Given evidence of COVID‐19 related increases in exercise among individuals with current or past ED diagnoses (e.g., Phillipou et al., 2020; Schlegl et al., 2020), future work examining the impacts of COVID‐19 anxiety and intolerance of uncertainty on compulsive exercise in clinical samples is warranted.
We predicted that trait and COVID‐19 intolerance of uncertainty would exacerbate ED and compulsive exercise risk associated with COVID‐19 anxiety. However, results suggest that COVID‐19 anxiety increases risk for ED pathology and compulsive exercise more strongly for individuals with lower trait and COVID‐19 intolerance of uncertainty. These findings may reflect a hierarchical structure for risk, in which individual factors (e.g., intolerance of uncertainty, anxiety) are more important than situational risk factors (e.g., COVID‐19 anxiety). Trait‐level predictors “dominated” other predictors in post‐hoc analyses, suggesting that trait anxiety and intolerance of uncertainty are more important than COVID‐19 factors in determining risk for compulsive exercise and ED pathology. Thus, individuals with trait risk factors may experience elevated ED pathology and compulsive exercise regardless of COVID‐19 anxiety. In contrast, for those with lower trait risk, COVID‐19 anxiety increases ED and compulsive exercise risk.
This study specifically examined COVID‐19 distress; however, findings may have implications for anxiety and stress more generally. Although COVID‐19 represents a unique stressor (Gruber et al., 2020), the acute effects of COVID‐19 may resemble reactions to other major events that cause life disruptions. Thus, our finding that COVID‐19 anxiety increases risk for ED pathology and compulsive exercise among individuals with lower trait risk may suggest that other acute stressors will have similar impacts. In contrast, for individuals with higher trait‐level risk, situational determinants may be less important. Future work replicating our results in different situations using longitudinal designs is warranted.
Study limitations should be acknowledged when interpreting results. The cross‐sectional nature of this study prevents our ability to draw causal conclusions; future use of longitudinal designs will further improve our understanding of risk for compulsive exercise pathology and CE associated with COVID‐19 and other events that require lifestyle modifications. This study also examined a nonclinical sample. Future work in clinical samples will further elucidate the impacts of COVID‐19 anxiety on ED pathology and CE. Additionally, future work examining the impacts of other characteristics (e.g., race/ethnicity) and life circumstances (e.g., employment) on ED and CE risk associated with COVID‐19 is warranted. Our assessment of exercise behaviors may also limit conclusions from this study. Future work that includes objective, in addition to, retrospective self‐report, assessments of exercise behaviors is warranted. Finally, COVID‐19 intolerance of uncertainty represents a novel construct. The operationalization of this construct in the current study resembles definitions of similar constructs in other recent work (e.g., Mahoney & McEvoy, 2012); however, the importance and validity of this variable remains unknown. Future work further examining COVID‐19 intolerance of uncertainty will further improve our understanding of distress associated with the pandemic.
The current study adds to a growing body of literature on the psychological impacts of COVID‐19 and other major events associated with life disruptions. Findings suggest that individual factors impact links between COVID‐19 anxiety and risk for ED pathology and CE, such that COVID‐19 anxiety increases risk among those with lower trait‐level risk. These results suggest that, for individuals with trait‐level risk factors, existing prevention and intervention efforts targeting intolerance of uncertainty and anxiety may be most effective. In contrast, interventions aimed at managing COVID‐19 distress may be most appropriate for individuals with lower trait‐level risk factors. Future work continuing to examine the longitudinal effects of COVID‐19 on ED pathology and CE is warranted given the unique and evolving circumstances.
ACKNOWLEDGMENTS
Sasha Gorrell receives funding support from NIMH (T32 MH018261).
CONFLICT OF INTEREST
The authors have no conflicts of interest to declare. This research did not receive any specific grant from funding agencies in the public, commercial, or not‐for‐profit sectors.
The data that support the findings of this study are available from the corresponding author upon reasonable request.
REFERENCES
