Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 May 1.
Published in final edited form as: Int J Eat Disord. 2023 Jan 31;56(5):1001–1010. doi: 10.1002/eat.23901

Divergent Trajectories of Positive Affect Following Maladaptive and Non-maladaptive Exercise Among Individuals with Binge-spectrum Eating Disorders

Elizabeth W Lampe 1,2,*, Sasha Gorrell 3, Kathryn Smith 4, Adam M Payne-Reichert 2, Adrienne S Juarascio 1,2, Stephanie M Manasse 2
PMCID: PMC10286557  NIHMSID: NIHMS1866939  PMID: 36719039

Abstract

Objective.

Individuals with eating disorders (EDs) often engage in maladaptive exercise (e.g., feeling driven, or to “compensate” for eating) which maintains eating pathology. Maladaptive exercise has been theorized to help individuals with EDs regulate emotions by enhancing positive affect (PA) and reducing negative affect (NA) associated with binge episodes and poor body image. However, no research has considered the presence of non-maladaptive exercise or evaluated this theory in binge-spectrum EDs.

Method.

This study evaluated affective trajectories pre- and post-exercise and examined exercise type (maladaptive or non-maladaptive) as a moderator. We recruited adults with binge-spectrum EDs (N=107; n=84 endorsed exercise) to complete a 7–14-day ecological momentary assessment protocol assessing NA (anxiety, guilt, sadness), PA (excitement, cheerfulness), and exercise.

Results.

There was a significant quadratic trajectory of PA pre-exercise, suggesting that the upward trajectory of PA leveled out as it moved closer to an exercise episode. Exercise type significantly moderated the cubic trajectory of PA post-exercise, such that the upward trajectory of PA accelerated in the hours following non-maladaptive exercise episodes while the downward trajectory of PA accelerated in the hours following maladaptive exercise episodes. No trajectories of NA demonstrated significant effects.

Conclusions.

Pre-exercise trajectories of PA may reflect positive expectancies around exercise. Post-exercise trajectories of PA suggest that non-maladaptive exercise promotes increased PA. Future research should evaluate when, and for whom, exercise serves to regulate affect and examine other momentary risk-factors of exercise in EDs.

Keywords: Negative Affect, Positive Affect, Exercise, Binge-eating Disorder, Bulimia Nervosa

Introduction

Binge-spectrum eating disorders (EDs; characterized by a recurrent sense of loss of control over one’s eating, such as bulimia nervosa [BN] or binge-eating disorder [BED]) are serious mental illnesses that affect millions of individuals worldwide (American Psychiatric Association, 2013), and are associated with increased mortality and high treatment costs (Van Hoeken & Hoek, 2020). Binge-spectrum EDs are characterized by binge-eating (i.e., the subjective experience of loss-of-control over eating), which may be followed by engagement in inappropriate weight-loss behaviors (e.g., self-induced vomiting, maladaptive exercise).

Maladaptive exercise is a common, yet complex, phenomenon in binge-spectrum ED samples (American Psychiatric Association, 2013; Claudino et al., 2019). Approximately 40–60% of adults who reported binge eating in an ED treatment setting also reported maladaptive exercise (Monell, Levallius, Mantilla, & Birgegård, 2018), however, we do not yet understand the maintaining mechanisms underlying maladaptive exercise. Such understanding is necessary to inform treatment approaches. For those with EDs, engagement in maladaptive exercise has been described in the literature as either ‘compensatory’ (i.e., as a means of controlling one’s weight, shape or amount of fat, or burning off calories; Davis, Guller, & Smith, 2016; C. Fairburn, Cooper, & O’Connor, 2014) and/or ‘driven’ (i.e., feeling compelled to exercise due to fears of weight gain, or to avoid other negative consequences associated with not exercising such as increases in negative affect; Hausenblas & Downs, 2002; Meyer, Taranis, Goodwin, & Haycraft, 2011). However, even among individuals with binge-spectrum EDs who exercise, there is considerable variability in the degree to which individuals engage in maladaptive exercise. For example, not all exercise episodes are necessarily maladaptive for a given individual (Chubbs-Payne et al., 2020; Lampe et al., 2021). However, it is currently unknown to what extent exercise that is non-maladaptive (i.e., not driven or compensatory) is overlapping with exercise that is adaptive (i.e., exercise promoting social or health benefits) in this population. Due to this lack of theoretical and empirical conceptualization of adaptive exercise engagement in the context of EDs, we will focus on non-maladaptive exercise engagement (i.e., exercise characterized by the absence of maladaptive characteristics). There is evidence that non-maladaptive exercise is a distinct construct from maladaptive exercise in this population. For example, a growing body of research has demonstrated that while increased maladaptive exercise is associated with higher ED pathology, increased non-maladaptive exercise is associated with lower ED pathology (E. W. Lampe, Forman, Juarascio, & Manasse, 2021; Mathisen, 2018; Mathisen, Sundgot-Borgen, Bulik, & Bratland-Sanda, 2021). Thus, it is important to understand functions of both maladaptive and non-maladaptive exercise so that treatments can better target maladaptive exercise and promote non-maladaptive exercise.

Several theoretical conceptualizations of maladaptive exercise have emerged (Davies, 2015; J. Mond & Gorrell, 2021), of which two models have gained wider support within the literature. The first model describes exercise behavior similarly to a behavioral addiction, whereby one might engage in exercise to initially experience a positive affective state, but then over time, a tolerance develops (i.e., increased exercise intensity is needed in order to experience an effect), along with withdrawal symptoms (e.g., anxiety, guilt; Hausenblas & Downs, 2002). The second model emphasizes compulsively engaging in exercise as a means to reduce or avoid negative affective states or feared negative consequences (e.g., weight gain; Meyer et al., 2011). Both models describe behavior that is maintained through negative reinforcement (e.g., mitigating negative affect; NA) and, to a lesser extent, positive reinforcement (e.g., increases in positive affect; PA). For both models, there is empirical evidence for the validity of affect-regulation in cross-sectional research on exercise in ED samples, where it is clear that exercise motivated by negative affect regulation is associated with higher compulsivity and ED symptoms (Bratland-Sanda et al., 2011; De Young & Anderson, 2010; Taranis & Meyer, 2011). There is additional evidence that exercise is associated with both lower NA and higher PA among university samples with higher obligation to exercise and eating pathology (LePage, Price, O’Neil, & Crowther, 2012; Thome & Espelage, 2004). However, little is known about the temporal relationship between affect and exercise within the broader binge-type ED spectrum (e.g., binge-eating disorder).

Further, theoretical discourse has not adequately accounted for the complexity of exercise behavior in the context of EDs. Based on current conceptualizations of other compensatory behaviors (e.g., purging, laxative use) in binge-spectrum EDs, NA is thought to increase leading up to, and decrease following, engagement in a given maladaptive behavior (Hilbert & Tuschen-Caffier, 2007; Schaefer, Smith, et al., 2020; Smyth et al., 2007). However, exercise may diverge from this pattern evidenced in other purging-type behavior cycles, in particular because engagement in exercise can be either maladaptive or non-maladaptive in nature. As such, it is possible that the way in which exercise serves to regulate affect may differ depending on the type of exercise (i.e., maladaptive or non-maladaptive). Understanding whether exercise serves to regulate affect for either—or both— maladaptive and non-maladaptive exercise episodes among individuals with binge eating pathology would allow us to better adapt current treatment approaches to discourage maladaptive exercise engagement and encourage non-maladaptive exercise engagement.

A momentary understanding of the affect-regulatory function of exercise in EDs is indicated and can be achieved with ecological momentary assessment (EMA). EMA involves the repeated assessment of participants’ affective states and behaviors (e.g., exercise) in a naturalistic setting. EMA is uniquely positioned to test prospective, momentary relationships as this approach allows for the evaluation of moment-to-moment changes in affective experiences before and after engagement in exercise, including the examination of both within- and between person differences. EMA also confers the advantage of examining trajectories of change in a particular variable, which allow for understanding of the rate of change in affect leading up to and following an exercise episode.

Past research using EMA to examine affect and exercise in non-ED populations has found increases in PA pre- and post-exercise, whereas changes in NA are inconsistent (Liao, Shonkoff, & Dunton, 2015). Though EMA has been employed in research on affect within EDs (Schaefer, Engel, & Wonderlich, 2020), only a few studies have used EMA to examine predictive relationships between exercise and affect in the context of anorexia (Engel et al., 2013; Karr et al., 2017). These studies found that NA decreased significantly following exercise (Engel et al., 2013), and that exercise enhanced PA. Improvements in PA following exercise were theorized to facilitate further exercise engagement via positive reinforcement (Karr et al., 2017). However, the existing published work has focused entirely on samples of individuals diagnosed with anorexia nervosa. Further understanding of non-maladaptive exercise is important within binge-spectrum EDs because exercise may have positive effects on body image, appetite and emotion regulation, and cognitive functioning, all of which have been linked to binge eating symptoms (Vancampfort et al., 2014). Further, exercise interventions have shown some utility for binge eating (though underlying mechanisms such as affect are not well-understood; Blanchet et al., 2018). Finally, it is important to examine non-linear trajectories of affect around exercise episodes as non-linear trajectories of affect have been identified around other ED behaviors (e.g., binge eating, restriction; Haynos et al., 2017; Schaefer, Smith, et al., 2020).

Thus, the current study aimed to evaluate trajectories of both NA (anxiety, guilt, and sadness) and PA (cheerfulness, excitement) around engagement in exercise among a sample of treatment-seeking individuals with binge-spectrum EDs. We hypothesized that 1) NA will decrease following exercise engagement, and 2) PA will increase following exercise engagement. A second aim of the current study was to evaluate the moderating role of exercise type (maladaptive or non-maladaptive) on affective trajectories. In line with conceptual models of exercise within EDs (Hausenblas & Downs, 2002; Meyer et al., 2011), we hypothesized that 1) NA will increase leading up to maladaptive exercise engagement but not non-maladaptive exercise engagement, and 2) PA will increase following non-maladaptive, but not maladaptive, exercise engagement.

Methods

Participants

We recruited adults seeking treatment for clinically-significant binge eating from the community (N=107) who were enrolled to participate in a remote treatment study in the Center for Weight Eating and Lifestyle Sciences (WELL Center) at Drexel University. Participants were excluded from the current analyses if they did not endorse exercise engagement during the study period (n=23), resulting in a final sample of 84 individuals. Participants were included in the study if they were ≥ 18 years old, experienced an average of at least one objective or subjective binge-eating episode per week over the previous 12 weeks, had a smartphone and were willing to complete surveys on it for the course of the study, had at least 7 days before their first treatment session, were located in the United States, and were willing and able to participate in remote intervention and assessments. Participants were excluded from the study if they were unable to fluently speak, write, and read English, had BMI < 18.5, were planning to begin, or were currently participating in another weight loss treatment or psychotherapy for binge eating and/or weight loss in the next 16 months, or had a mental handicap or were currently experiencing other severe psychopathology that would limit their ability to engage in the treatment program (e.g., severe depression, substance dependence, active psychotic disorder). While maladaptive exercise in common in BN (Udo & Grilo, 2018), recent studies have found that individuals with BED may also be engaging in maladaptive exercise (Monell et al., 2018; Wons, Michael, Lin, & Juarascio, 2021), thus we included both individuals with BED and BN in the current study. Although the DSM-5 requires individuals to have “objectively large” binge episodes in order to have a diagnosis of BN or BED, the experience of loss-of-control (i.e., the feeling that one is driven or compelled to keep eating and feel they cannot stop) is the hallmark of binge episodes regardless of size and is the feature most associated with impaired quality of life and elevated psychopathology (rather than binge size or frequency; Latner, Hildebrandt, Rosewall, Chisholm, & Hayashi, 2007; J. M. Mond, Latner, Hay, Owen, & Rodgers, 2010). Fifty-eight participants in the current analysis were also included in other published studies examining the associations between exercise and body dissatisfaction (Srivastava, Lampe, Wons, Juarascio, & Manasse, 2022) and fear of weight gain (Lampe, Wons, Taylor, Juarascio, & Manasse, 2022) and examining reasons for exercise engagement (Lampe et al., 2021).

Procedures

Recruitment and screening.

Participants were recruited from the community using radio and social media advertising for ED treatment. Once eligibility was determined for treatment, participants were offered the opportunity to participate in a one to two-week EMA study and were consented if interested. Participants were then contacted by study personnel to set up the EMA application on their smartphone. Participants also received psychoeducation on several relevant constructs to the EMA surveys such as maladaptive exercise. In defining maladaptive exercise, driven exercise was described as, “Feeling driven or compelled to exercise. For example, exercising even when you are sick, or injured, or have other obligations, or feeling upset if you are unable to exercise,” and compensatory exercise was described as, “Exercising to compensate (or make up for) calories that you have consumed throughout the day.” Participants then completed the EMA study for 7–14 days, depending on the timing of their first treatment session.

Ecological Momentary Assessment Surveys.

The current study included both signal-contingent (in response to a notification) and event-contingent (following engagement in an ED behavior) EMA surveys sent to participants’ smartphones. Six EMA signals were semi-randomly spread out over the day such that participants received ~3 in the morning and ~3 in the afternoon/evening. Participants were also instructed to complete an EMA survey after engaging in an ED behavior such as binge eating, purging, or engaging in another compensatory behavior (excluding exercise, as that was asked about separately within the surveys). Participants received up to $12.00 for each day they completed EMA surveys, with $2.00 being deducted per missed survey. Participants also received $50 for completing ≥ 85% of all surveys. All EMA prompts were scheduled between participant’s self-reported average wake- and bed-times and participants had 20 minutes to complete each signal-contingent survey before the prompt expired.

Measures

Eating Pathology.

The Eating Disorder Examination 17.0 (EDE) (C. G. Fairburn, Cooper, & O’Connor, 1993) was used to assess eating pathology over the previous three months. The EDE is a well validated, semi-structured diagnostic interview and was administered by a trained rater.

Exercise.

Exercise was assessed at each survey using the question “Have you exercised since the last survey?” If participants responded in the affirmative, they then responded to two items assessing maladaptive exercise: 1)“To what extent did you feel driven to exercise?” and 2) “To what extent did you exercise to compensate for eating?” These EMA items correspond to the two criteria for maladaptive exercise engagement assessed in the EDE interview: either 1) exercising in a ‘driven’ or ‘compulsive’ way, and 2) exercising as a means of controlling one’s weight, shape or amount of fat, or to burn off calories. Each of the EMA questions assessing maladaptive exercise was answered on a 5-point Likert scale from one (Not at all) to five (Extremely/Completely). Consistent with prior EMA literature, a maladaptive exercise episode was defined as endorsement of ≥ 3/5 either compensatory or driven exercise (Lampe et al., 2021; Lampe et al., 2022; Srivastava et al., 2022). This criterion for defining maladaptive exercise was chosen based on clinical significance cut-offs used in extant literature (Karr et al., 2017; Lampe et al., 2021; Lampe et al., 2022; E. W. Lampe et al., 2021; Srivastava et al., 2022). Because existing definitions of non-maladaptive exercise in the context of binge-spectrum EDs are lacking, we considered any exercise as non-maladaptive which was not categorized as maladaptive (i.e., non-maladaptive episodes were endorsed by participants as < 3/5 on both compensatory and driven exercise items), following CDC guidelines for healthy exercise (2020).

Affect.

Following questions about exercise engagement, both NA and PA were measured at each EMA survey by asking participants, “Right now, to what extent do you feel…?” Selected NA items were chosen from the PANAS (Watson, Clark, & Tellegen, 1988) and included: sad, guilty, and anxious. PA items (also chosen from the PANAS) included: excited and cheerful. Both NA and PA items were chosen due to established relationships with ED behaviors (Berg et al., 2017; Brownstone et al., 2013; Deaver, Miltenberger, Smyth, Meidinger, & Crosby, 2003; Haedt-Matt & Keel, 2015). Participants responded to each prompt using 5-point Likert scales which ranged from one (Not at all) to five (Extremely). NA and PA at each survey were obtained by summing participants’ responses to each of the corresponding items, in accordance with previous EMA studies of affect in EDs (Engel et al., 2013; Schaefer, Smith, et al., 2020; Smyth et al., 2007). Cronbach’s alpha for NA was 0.80 and for PA was 0.79 within this sample, indicating high internal consistency.

Statistical Analyses

Pre- and post-exercise trajectories of PA and NA were modeled separately using piecewise linear, quadratic, and cubic functions centered on the survey at which the exercise episode was reported. We used a mixed-effects polynomial regression (Hedeker & Gibbons, 2006) approach to 1) model discrete temporal patterns for pre-exercise and post-exercise, 2) accommodate nonlinear trajectories, 3) use data collected at differing time intervals (as not all participants had the same number of observations following exercise engagement), 4) accommodate data with missing observations, 5) include exercise episodes that occurred at varying time points, and 6) model nonlinear relationships across time both at the individual and sample levels.

Multilevel models included momentary observations of affect and exercise (Level 1) which were nested within subjects (Level 2). Consistent with previous approaches (Schaefer, Smith, et al., 2020), estimated effects included linear (time prior to the exercise, time following the exercise), quadratic ([time prior to the exercise]2, [time following the exercise]2), and cubic ([time prior to the exercise]3, [time following the exercise]3) effects. Quadratic and cubic effects were estimated to capture any potentially non-linear changes in affect pre- and post-exercise. Within trajectory analyses, linear effects indicate whether the initial slope of the regression line (i.e., changes in affect immediately surrounding an exercise episode) is positive, negative, or remains flat. Quadratic effects indicate whether the regression line deflects downward or upward from the linear component as it moves farther from the exercise episode. This component represents any acceleration or deceleration in the rate of change in affect farther out from the exercise episode. Finally, cubic effects capture further deflections from the quadratic slope of the regression line happening farthest from the exercise episode. All models used a random intercept for subject and a common intercept for pre- and post-exercise trajectories.

Each model examined whether the pre-exercise linear, quadratic, and cubic effects differed significantly from zero and whether the post-exercise linear, cubic, and quadratic effects of the trajectory significantly differed from the pre-exercise trajectory. We also examined a model that included an interaction term for exercise type and time prior to- or following the exercise episode. Because antecedent and consequent affect ratings are highly associated, only the first exercise episode was used when more than one episode was reported in a single day. All analyses used a first-order autoregressive covariance structure (AR1) to account for serial correlations. Maximum-likelihood estimation methods were used to estimate parameters. Models were evaluated which included covariates for binge eating and compensatory behaviors at the previous survey, however results did not meaningfully differ and models are presented without covariates for parsimony. All analyses were conducted using SPSS Version 28.0.

Results

Sample Descriptives

The current sample (80.4% female) included 84 treatment-seeking adults (Mage=41.36, SD=13.52) with trans-diagnostic binge eating (i.e., BN, BED, or Other Specified Feeding or Eating Disorder [OSFED]-BN/BED) and mean BMI=34.83 (SD=9.06). Participants primarily identified as Caucasian (75.2%) with others identifying as: African American (10.5%), Asian (2.9%), American Indian/Alaska Native (1.0%), more than one race (6.7%), or unknown or prefer not to say (3.8%); 10.4% identified as Hispanic or Latinx. At baseline, participants endorsed an average of 25.26 (SD=21.42) binge episodes and 6.99 (SD=12.45) compensatory behaviors over the past month (including maladaptive exercise), per the EDE interview. Thirty-eight participants were diagnosed with bulimia nervosa-spectrum EDs (BN; 45.2%), and 46 participants were diagnosed with BED-spectrum EDs (54.8%). Specifically, 12 participants were diagnosed with OSFED-BED of low frequency (14.3%), 26 with full-threshold BED (31.0%), 12 with OSFED-BN of low frequency (14.3%), and 34 with full-threshold BN (40.5%). Low frequency diagnoses were assigned when participants experienced < 12 LOC eating episodes (i.e., any combination of subjective or objective binge episodes) in the past three months and/or (in the case of subthreshold BN) < 12 compensatory behavior episodes.

EMA Recordings and Missing Data

Average participant compliance (i.e., percent signal-contingent ratings completed for each prompt), was 87.28% (SD=14.11), which is similar to previous EMA work within ED samples (Engel et al., 2013; Schaefer, Smith, et al., 2020). Participants completed the protocol for an average of 12.16 days (SD=2.47). A total of 2,503 antecedent affect ratings (M=18.62, SD=13.95 per person) and 1,743 consequent affect ratings (M=8.72, SD=5.08 per person) were reported relative to within-day exercise episodes.

Exercise Engagement

Baseline Exercise Engagement.

Thirty-six participants (42.9%) endorsed maladaptive exercise during the past month (28 days) on the EDE interview. Per the EDE, participants endorsed an average of 10.21 (SD=19.91) maladaptive exercise episodes over the past month (Mcomp=4.55, SDcomp=9.64, Mdriven=5.67, SDdriven=11.03). Of the 36 participants who endorsed maladaptive exercise on the EDE, 34 (94.44%) also reported at least one episode of EMA-measured maladaptive exercise during the EMA period. EDE- and EMA-assessed maladaptive exercise engagement were significantly, positively, and moderately correlated (r=0.49, p<.001).

EMA-measured Exercise Engagement.

Of the 107 participants enrolled in the parent study, 84 (78.50%) reported exercise engagement during the EMA period and were included in the current analyses. Twenty-eight (33.33%) participants reported only maladaptive exercise episodes, 22 (26.19%) reported only non-maladaptive, and 34 (40.48%) reported both maladaptive and non-maladaptive exercise episodes during the EMA period. Among those who reported both non-maladaptive and maladaptive exercise engagement, 50.54% (SD=28.35%) of their exercise episodes were maladaptive on average. Of the 62 (73.81%) total participants that endorsed at least one episode of maladaptive exercise (i.e., those who endorsed either maladaptive exercise only, or both maladaptive and non-maladaptive exercise engagement); these participants endorsed an average of 4.82 (SD=7.27) episodes of maladaptive exercise across the EMA period. See Table 1 for further description of exercise-episode types and participant exercise profiles.

Table 1.

Number (N) and Percent (%) of Exercise Episode Types and Participant Exercise Profiles

Exercise Episode Types N %
Driven only 184 27.8
Compensatory only 15 2.3
Driven and compensatory 206 31.2
Non-maladaptive 256 38.7
Participant Exercise Profiles
Maladaptive exercisers 62 57.9
Non-maladaptive exercisers 22 20.6
Open in a new tab

Note. Maladaptive exercisers were those that engaged in any level of maladaptive exercise, while non-maladaptive exercisers were those that engaged in solely non-maladaptive exercise across the recording period.

Trajectories of Negative Affect

There was no significant effect of NA pre- or post-exercise (Table 2). Whether exercise was maladaptive (i.e., driven and/or compensatory) versus non-maladaptive did not significantly moderate the pre- or post-exercise trajectories of NA (Table 2 and Figures 1 & 2).

Table 2.

Trajectories of Negative Affect around Exercise Episodes.

Moderator Time Period Parameter B S.E. p 95% CI
Negative Affect -- Pre-exercise Linear 0.016 0.023 0.481 −0.029–0.062
Quadratic 0.002 0.002 0.417 −0.002–0.006
Cubic −0.0002 0.0002 0.508 −0.001–0.0003
Post-exercise Linear −0.026 0.098 0.794 −0.218–0.167
Quadratic −0.007 0.023 0.748 −0.052–0.037
Cubic 0.0003 0.002 0.856 −0.003–0.004
Type Pre-exercise Linear −0.030 0.052 0.568 −0.131–0.072
Quadratic 0.001 0.005 0.847 −0.008–0.010
Cubic 0.0003 0.001 0.604 −0.001–0.002
Post-exercise Linear −0.190 0.446 0.670 −1.065–0.685
Quadratic 0.042 0.140 0.762 −0.233–0.318
Cubic −0.001 0.010 0.889 −0.022–0.019
Cubic 0.001 0.001 0.530 −0.001–0.002
Open in a new tab
*

(p<.05)

Note. Exercise type was coded as 0=non-maladaptive exercise and 1=maladaptive exercise.

Figure 1.

Figure 1.

Open in a new tab

Trajectories of overall NA pre- and post-exercise.

Figure 2.

Figure 2.

Open in a new tab

Trajectories of overall PA pre- and post-exercise (*p<.05).

Trajectories of Positive Affect

There was a significant negative quadratic trajectory of PA pre-exercise, suggesting that the upward trajectory of PA decelerated as it moved closer to an exercise episode (linear estimate=.016, quadratic= −.007 [p=.005]; Table 3 and Figure 2). We did not find any significant trajectory of PA post-exercise. Exercise type significantly moderated the cubic trajectory of PA post-exercise (linear estimate=−.330, quadratic=.251, cubic=−.024 [p=.045]; Table 3 and Figure 2) such that the upward trajectory of PA accelerated in the hours following non-maladaptive exercise episodes while the downward trajectory of PA accelerated in the hours following maladaptive exercise episodes.

Table 3.

Trajectories of Positive Affect around Exercise Episodes.

Moderator Time Period Parameter B S.E. p 95% CI
Positive Affect -- Pre-exercise Linear 0.016 0.026 0.548 −0.035–0.067
Quadratic −0.007 0.002 0.005* −0.012– −0.002
Cubic 0.0002 0.0003 0.359 −0.0003–0.001
Post-exercise Linear 0.031 0.120 0.779 −0.185–0.246
Quadratic −0.023 0.025 0.367 −0.073–0.027
Cubic 0.002 0.002 0.245 −0.002–0.007
Type Pre-exercise Linear −0.022 0.058 0.707 −0.136–0.093
Quadratic −0.008 0.005 0.134 −0.018–0.002
Cubic 0.001 0.001 0.247 −0.001–0.002
Post-exercise Linear −0.330 0.508 0.516 −1.327–0.667
Quadratic 0.251 0.160 0.117 −0.063–0.565
Cubic −0.024 0.012 0.045* −0.047– −0.001
Open in a new tab
*

(p<.05)

Note. Exercise type was coded as 0=non-maladaptive exercise and 1=maladaptive exercise.

Discussion

The current study was the first to use EMA to examine trajectories of affect before and after exercise engagement among individuals with transdiagnostic binge eating. We examined trajectories of both negative (anxiety, guilt, and sadness) and positive (excitement, cheerfulness) affect and exercise, and whether the maladaptive nature of the exercise (i.e., driven and/or compensatory) moderated these trajectories.

Trajectories of Negative Affect

Unexpectedly, we did not find any significant trajectories of overall NA pre- or post-exercise, and exercise type did not moderate these trajectories. These findings differ from previous studies evaluating trajectories of affect around other compensatory behaviors (Haedt-Matt & Keel, 2015). These findings indicate that exercise likely did not serve to successfully regulate NA on a momentary level, regardless of whether the exercise was non-maladaptive or maladaptive. In the current study, exercise episodes were measured within the interval between two surveys which was several hours in length (i.e., participants were asked, “since the last survey, have you exercised?) and NA recordings were captured at the time of survey response. Thus, NA recordings were likely not often obtained within close proximity of exercise episodes, meaning that any short-term decreases in NA may not have been captured by the time NA was reported at the next signal. While these findings do not completely rule out an affect-regulation function of exercise in binge-spectrum EDs, they do suggest that exercise may function differently than other compensatory behaviors (i.e., not primarily by regulating NA).

Trajectories of Positive Affect

Interestingly, we observed significant quadratic trajectories of PA leading up to exercise engagement. The observed trajectories of PA leading up to exercise engagement suggest that PA is increasing more steeply many hours before exercise episodes, but this upward trajectory of PA decelerates closer to the exercise episode. However, this trajectory of PA pre-exercise was not moderated by exercise type. This pattern may reflect affect-boosting effects of decisions to exercise, as anticipation effects could result in the observed increases in PA leading up to exercise engagement. For example, an individual may feel increased confidence in their ability to control their body shape/weight after making a plan to engage in exercise later that day. The observed trajectories of PA pre-exercise mirror relationships between PA and exercise observed in other ED samples (Karr et al., 2017). Future research should also assess when the decisions to exercise occur and examine impacts of these decisions on affect.

While the overall trajectory of PA and excitement following exercise was non-significant, consistent with hypotheses, post-exercise trajectories of PA were moderated by the type of exercise episode (maladaptive or non-maladaptive). We found that that PA trajectories deflected significantly downward over the hours following maladaptive exercise episodes while PA trajectories deflected upward following non-maladaptive exercise episodes. The magnitude of difference in trajectories of PA following non-maladaptive or maladaptive exercise engagement may be driving the non-significant downward trajectory of PA observed following any exercise engagement. These findings are consistent with theories of engagement in maladaptive exercise (Hausenblas & Downs, 2002; Meyer et al., 2011), and indicate that individuals with binge-spectrum EDs may be less likely to experience the increases in PA which have been found in other populations when engaging in maladaptive exercise. On the other hand, the observed upward trajectory of PA following non-maladaptive exercise engagement lends further support for recent findings showing that non-maladaptive exercise engagement may lead to reduced ED symptoms among individuals with EDs (E. W. Lampe et al., 2021; Mathisen, 2018; Mathisen et al., 2021; Pendleton, Goodrick, Poston, Reeves, & Foreyt, 2002). Alternatively, while it is true that there may be changes in the trajectory of PA as participants move further away from the exercise episode, it may also be possible that other intervening factors may be responsible for the observed changes in PA. While we cannot conclude with certainty that non-maladaptive exercise episodes observed in this sample were adaptive in nature, these findings underscore an urgent need for improved conceptualizations of adaptive, non-maladaptive, and maladaptive exercise in the context of EDs and their overlap/distinctions.

Clinical Implications & Future directions

There are several important considerations for future research and clinical approaches to exercise in EDs. For example, future treatments may include psychoeducation on the differing effects of maladaptive and non-maladaptive exercise on PA trajectories following exercise engagement. Sharing this information with patients may help them to understand why maladaptive exercise engagement can actually be harmful to their recovery by reducing PA post-exercise and contributing to future binge episodes. This psychoeducation may also promote engagement in non-maladaptive exercise by helping individuals better recognize when they are engaging in non-maladaptive or maladaptive exercise, which may serve to build self-awareness that is necessary prior to initiating behavior change. Because non-maladaptive and maladaptive exercise differ in their motivations rather than presence/absence (as in other ED behaviors like purging), it may be difficult to change this behavior in treatment. However, several successful strategies have been implemented within treatments promoting non-maladaptive exercise such as pre-planning exercise in advance and separately from eating episodes (Blanchet et al., 2018; E. W. Lampe et al., 2021; Mathisen, 2018). Importantly, these findings further underscore the importance of research evaluating when, and for whom, exercise serves to regulate affect. Future work should explore other functions of exercise in binge-spectrum EDs such as more long-term relationships and trajectories of other cognitive-affective factors (e.g., dissatisfaction with weight) pre- and post-exercise.

Strengths and Limitations

The current study has several strengths. Most notably, this study builds on previous literature which underscores the importance of robust empirical study of exercise in EDs (Kolar & Gorrell, 2020; Scharmer, Gorrell, Schaumberg, & Anderson, 2020). Due to our use of EMA to capture affect and exercise behavior observations, the current study was able to use a large number of affect and exercise observations. Further, our use of multilevel modeling of affective trajectories allowed for examination of momentary changes in affect in a naturalistic setting. Additionally, the use of EMA allowed us to identify the temporal sequence of affective changes in relation to exercise episodes. Multilevel modeling of affective trajectories allowed us to examine an affect-regulation function of exercise and provided a rigorous test of our hypotheses. Our use of EMA to capture affect and exercise in real time, and in a naturalistic setting, increased the ecological validity of the results and reduced retrospective recall bias. Finally, by using a mixed diagnostic sample (i.e., those with both BN-spectrum and BED-spectrum EDs), we were able to evaluate associations between affect and both non-maladaptive and maladaptive exercise among individuals with binge-spectrum eating disorders, which has previously been understudied.

The current study has several limitations which should be considered when interpreting the results. First, it is possible that measurement error was introduced by using abbreviated measures of affect (e.g., PANAS items). Although this approach has been used in past literature (Engel et al., 2013; Schaefer, Smith, et al., 2020; Smyth et al., 2007), and was used in the current study to reduce participant burden, the use of restricted measures of NA and PA in the current study may have substantially limited construct validity. Second, the current study did not use event-contingent EMA recordings of exercise episodes, which may have reduced the number of recordings overall, and especially reduced the number of affective ratings immediately following exercise engagement. Future EMA studies examining proximal changes in affect following exercise should implement event-contingent reporting in order to maximize the number of available data points shortly after exercise engagement. Third, many participants reported only one episode of maladaptive exercise during the recording period which may have limited our ability to observe within person variability in affective trajectories based on exercise type and may not generalize to individuals with higher levels of maladaptive exercise. Fourth, the current analyses were unable to accommodate multiple exercise episodes within a single day. In order to reduce covariance of affect and exercise, we only included the first exercise episode of each day, however, it may be that individuals who exercise more than once per day may display differing affective trajectories surrounding each exercise episode. Fifth, there is considerable heterogeneity in current conceptualizations of maladaptive exercise (Mond & Gorrell, 2021) and agreed upon definitions of adaptive exercise engagement in the context of an ED are severely lacking. Thus, we are unable to conclude that any non-maladaptive exercise is inherently adaptive in nature. Sixth, despite the fact that EMA allows for the establishment of temporal sequencing of affect trajectories through repeated assessment, we cannot rule out other explanatory factors for the observed relationships. Thus, the current study does not allow for the establishment of causality and results conclusions must be interpreted accordingly. Finally, exercise was measured via self-report and future studies may benefit from objective measures of exercise engagement that can more accurately assess timing, duration, and intensity (e.g., accelerometry). Future research should explore these objective characteristics of exercise episodes as moderators of affective trajectories.

Conclusions

Exercise did not appear to serve individuals in the current study in regulating NA. However, pre-exercise trajectories of PA may reflect positive expectancies around exercise. Post-exercise trajectories of PA suggests that non-maladaptive exercise engagement may actually result in increased PA over the hours following exercise compared to maladaptive exercise engagement. These findings underscore the importance of research evaluating when, and for whom, exercise serves to regulate affect. These findings also point to a need for research examining other momentary drivers of exercise in EDs in order to improve current treatment’s ability to intervene on maladaptive exercise and/or encourage non-maladaptive exercise. Future research should also aim to use more objective measures of exercise (e.g., accelerometry) to increase data validity and availability.

Public Significance Statement.

Individuals with eating disorders (EDs) often engage in maladaptive exercise (e.g., feeling driven, or to “compensate” for eating). Maladaptive exercise has been theorized to help individuals with EDs regulate emotions. This study evaluated affective trajectories pre- and post-exercise and examined whether exercise type (maladaptive or non-maladaptive) changed these trajectories. Pre-exercise trajectories of PA may reflect positive expectancies around exercise. Post-exercise trajectories of PA suggest that non-maladaptive exercise promotes increased PA.

Funding:

This research received no specific grant from any funding agency, commercial or not-for-profit sectors. Drs. Gorrell and Manasse were supported by awards from the National Institute of Mental Health (K23MH126201; K23DK124514). Dr. Smith was supported by an award from the National Institute of Diabetes and Digestive & Kidney Diseases (K23DK128568).

Footnotes

Conflicts of interest: The authors have no conflicts of interest to declare.

Ethics approval: All study procedures were approved by the Drexel University Institutional Review Board.

Consent to participate: All participants provided informed consent.

Consent for publication: All participants provided informed consent for publication of aggregate data.

Code availability: Analysis code is available upon request to the corresponding author.

Availability of data and material:

Data is available upon request to the corresponding author.

References

  1. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (DSM-5®): American Psychiatric Pub. [Google Scholar]
  2. Berg KC, Cao L, Crosby RD, Engel SG, Peterson CB, Crow SJ, … Durkin N (2017). Negative affect and binge eating: Reconciling differences between two analytic approaches in ecological momentary assessment research. International Journal of Eating Disorders, 50(10), 1222–1230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Blanchet C, Mathieu M-È, St-Laurent A, Fecteau S, St-Amour N, & Drapeau V (2018). A systematic review of physical activity interventions in individuals with binge eating disorders. Current obesity reports, 7(1), 76–88. [DOI] [PubMed] [Google Scholar]
  4. Bratland-Sanda S, Martinsen EW, Rosenvinge JH, Rø Ø, Hoffart A, & Sundgot-Borgen J (2011). Exercise dependence score in patients with longstanding eating disorders and controls: the importance of affect regulation and physical activity intensity. European Eating Disorders Review, 19(3), 249–255. [DOI] [PubMed] [Google Scholar]
  5. Brownstone LM, Fitzsimmons-Craft EE, Wonderlich SA, Joiner TE, Le Grange D, Mitchell JE, … Klein MH (2013). Hard exercise, affect lability, and personality among individuals with bulimia nervosa. Eating behaviors, 14(4), 413–419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Centers for Disease Control. (2020). How much physical activity do adults need? Retrieved from https://www.cdc.gov/physicalactivity/basics/adults/index.htm
  7. Chubbs-Payne A, Lee J, Isserlin L, Norris ML, Spettigue W, Spence K, & Longmuir PE (2020). Attitudes toward physical activity as a treatment component for adolescents with anorexia nervosa: An exploratory qualitative study of patient perceptions. International Journal of Eating Disorders. [DOI] [PubMed] [Google Scholar]
  8. Claudino AM, Pike KM, Hay P, Keeley JW, Evans SC, Rebello TJ, … Matsumoto C (2019). The classification of feeding and eating disorders in the ICD-11: results of a field study comparing proposed ICD-11 guidelines with existing ICD-10 guidelines. BMC medicine, 17(1), 93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Davies RR (2015). The treatment of compulsive physical activity in anorexia nervosa lacks a conceptual base. Advances in Eating Disorders: Theory, Research and Practice, 3(1), 103–112. [Google Scholar]
  10. Davis Guller, & Smith. (2016). Developmental trajectories of compensatory exercise and fasting behavior across the middle school years. Appetite, 107, 330–338. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. De Young KP, & Anderson DA (2010). The importance of the function of exercise in the relationship between obligatory exercise and eating and body image concerns. Eating behaviors, 11(1), 62–64. [DOI] [PubMed] [Google Scholar]
  12. Deaver CM, Miltenberger RG, Smyth J, Meidinger A, & Crosby R (2003). An evaluation of affect and binge eating. Behavior Modification, 27(4), 578–599. [DOI] [PubMed] [Google Scholar]
  13. Engel SG, Wonderlich SA, Crosby RD, Mitchell JE, Crow S, Peterson CB, … Lavender JM (2013). The role of affect in the maintenance of anorexia nervosa: evidence from a naturalistic assessment of momentary behaviors and emotion. Journal of Abnormal Psychology, 122(3), 709. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Fairburn C, Cooper Z, & O’Connor M (2014). The Eating Disorder Examination. The Centre for Research on Eating Disorders at Oxford. In: Oxford, UK. [Google Scholar]
  15. Fairburn CG, Cooper Z, & O’Connor M (1993). The eating disorder examination. International Journal of Eating Disorders, 6, 1–8. [Google Scholar]
  16. Haedt-Matt AA, & Keel PK (2015). Affect regulation and purging: An ecological momentary assessment study in purging disorder. Journal of Abnormal Psychology, 124(2), 399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hausenblas HA, & Downs DS (2002). How much is too much? The development and validation of the exercise dependence scale. Psychology and Health, 17(4), 387–404. [Google Scholar]
  18. Haynos AF, Berg KC, Cao L, Crosby RD, Lavender JM, Utzinger LM, … Le Grange D (2017). Trajectories of higher-and lower-order dimensions of negative and positive affect relative to restrictive eating in anorexia nervosa. Journal of Abnormal Psychology, 126(5), 495. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hedeker D, & Gibbons RD (2006). Longitudinal data analysis (Vol. 451): John Wiley & Sons. [Google Scholar]
  20. Hilbert A, & Tuschen-Caffier B (2007). Maintenance of binge eating through negative mood: A naturalistic comparison of binge eating disorder and bulimia nervosa. International Journal of Eating Disorders, 40(6), 521–530. [DOI] [PubMed] [Google Scholar]
  21. Karr TM, Cook B, Zunker C, Cao L, Crosby RD, Wonderlich SA, & Mitchell JE (2017). Examining physical activity and affect using objective measures: A pilot study of anorexia nervosa. The Sport Journal, 20. [Google Scholar]
  22. Kolar DR, & Gorrell S (2020). A call to experimentally study acute affect-regulation mechanisms specific to driven exercise in eating disorders. International Journal of Eating Disorders. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lampe Trainor, Presseller Michael, Payne-Reichert Juarascio, & Manasse (2021). Characterizing reasons for exercise in binge-spectrum eating disorders. Eating behaviors. doi: 10.1016/j.eatbeh.2021.101558 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lampe, Wons O, Taylor LC, Juarascio AS, & Manasse SM (2022). Associations between fear of weight gain and exercise in binge-spectrum eating disorders. Eating and Weight Disorders-Studies on Anorexia, Bulimia and Obesity, 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lampe EW, Forman EM, Juarascio AS, & Manasse SM (2021). Feasibility, Acceptability, and Preliminary Target Engagement of a Healthy Physical Activity Promotion Intervention for Bulimia Nervosa: Development and Evaluation via Case Series Design. Cognitive and behavioral practice. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Latner JD, Hildebrandt T, Rosewall JK, Chisholm AM, & Hayashi K (2007). Loss of control over eating reflects eating disturbances and general psychopathology. Behaviour Research and Therapy, 45(9), 2203–2211. [DOI] [PubMed] [Google Scholar]
  27. LePage ML, Price M, O’Neil P, & Crowther JH (2012). The effect of exercise absence on affect and body dissatisfaction as moderated by obligatory exercise beliefs and eating disordered beliefs and behaviors. Psychology of Sport and Exercise, 13(4), 500–508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Liao Y, Shonkoff ET, & Dunton GF (2015). The acute relationships between affect, physical feeling states, and physical activity in daily life: a review of current evidence. Frontiers in psychology, 6, 1975. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Mathisen TF (2018). A randomized controlled trial of physical exercise-and dietary therapy versus cognitive behavior therapy: Treatment effects for women with bulimia nervosa or binge eating disorder.
  30. Mathisen TF, Sundgot-Borgen J, Bulik CM, & Bratland-Sanda S (2021). The neurostructural and neurocognitive effects of physical activity: A potential benefit to promote eating disorder recovery. International Journal of Eating Disorders, 54(10), 1766–1770. [DOI] [PubMed] [Google Scholar]
  31. Meyer C, Taranis L, Goodwin H, & Haycraft E (2011). Compulsive exercise and eating disorders. European Eating Disorders Review, 19(3), 174–189. [DOI] [PubMed] [Google Scholar]
  32. Mond J, & Gorrell S (2021). “Excessive exercise” in eating disorders research: problems of definition and perspective. In: Springer. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mond JM, Latner JD, Hay P, Owen C, & Rodgers B (2010). Objective and subjective bulimic episodes in the classification of bulimic-type eating disorders: another nail in the coffin of a problematic distinction. Behaviour Research and Therapy, 48(7), 661–669. [DOI] [PubMed] [Google Scholar]
  34. Monell E, Levallius J, Mantilla EF, & Birgegård A (2018). Running on empty–a nationwide large-scale examination of compulsive exercise in eating disorders. Journal of eating disorders, 6(1), 11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Pendleton VR, Goodrick GK, Poston WSC, Reeves RS, & Foreyt JP (2002). Exercise augments the effects of cognitive-behavioral therapy in the treatment of binge eating. International Journal of Eating Disorders, 31(2), 172–184. [DOI] [PubMed] [Google Scholar]
  36. Schaefer LM, Engel SG, & Wonderlich SA (2020). Ecological momentary assessment in eating disorders research: recent findings and promising new directions. Current Opinion in Psychiatry, 33(6), 528–533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Schaefer LM, Smith KE, Anderson LM, Cao L, Crosby RD, Engel SG, … Wonderlich SA (2020). The role of affect in the maintenance of binge-eating disorder: Evidence from an ecological momentary assessment study. Journal of Abnormal Psychology, 129(4), 387. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Scharmer C, Gorrell S, Schaumberg K, & Anderson D (2020). Compulsive exercise or exercise dependence? Clarifying conceptualizations of exercise in the context of eating disorder pathology. Psychology of Sport and Exercise, 46, 101586. [PMC free article] [PubMed] [Google Scholar]
  39. Smyth JM, Wonderlich SA, Heron KE, Sliwinski MJ, Crosby RD, Mitchell JE, & Engel SG (2007). Daily and momentary mood and stress are associated with binge eating and vomiting in bulimia nervosa patients in the natural environment. Journal of Consulting and Clinical Psychology, 75(4), 629. [DOI] [PubMed] [Google Scholar]
  40. Srivastava PS, Lampe EW, Wons OB, Juarascio AS, & Manasse SM (2022). Understanding momentary associations between body dissatisfaction and exercise in binge-spectrum eating disorders. Eating and Weight Disorders - Studies on Anorexia, Bulimia and Obesity. doi: 10.1007/s40519-022-01371-0 [DOI] [PubMed] [Google Scholar]
  41. Taranis L, & Meyer C (2011). Associations between specific components of compulsive exercise and eating-disordered cognitions and behaviors among young women. International Journal of Eating Disorders, 44(5), 452–458. [DOI] [PubMed] [Google Scholar]
  42. Thome J, & Espelage DL (2004). Relations among exercise, coping, disordered eating, and psychological health among college students. Eating behaviors, 5(4), 337–351. [DOI] [PubMed] [Google Scholar]
  43. Udo T, & Grilo CM (2018). Prevalence and correlates of DSM-5–defined eating disorders in a nationally representative sample of US adults. Biological psychiatry, 84(5), 345–354. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Van Hoeken D, & Hoek HW (2020). Review of the burden of eating disorders: mortality, disability, costs, quality of life, and family burden. Current Opinion in Psychiatry, 33(6), 521. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Vancampfort D, Vanderlinden J, Stubbs B, Soundy A, Pieters G, Hert MD, & Probst M (2014). Physical activity correlates in persons with binge eating disorder: A systematic review. European Eating Disorders Review, 22(1), 1–8. [DOI] [PubMed] [Google Scholar]
  46. Watson D, Clark LA, & Tellegen A (1988). Development and validation of brief measures of positive and negative affect: the PANAS scales. Journal of personality and social psychology, 54(6), 1063. [DOI] [PubMed] [Google Scholar]
  47. Wons OB, Michael ML, Lin M, & Juarascio AS (2021). Characterizing rates of physical activity in individuals with binge eating disorder using wearable sensor technologies and clinical interviews. European Eating Disorders Review, 29(2), 292–299. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data is available upon request to the corresponding author.

RESOURCES