Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Sep 1.
Published in final edited form as: Behav Ther. 2024 Feb 13;55(5):950–960. doi: 10.1016/j.beth.2024.01.005

Latent Change Trajectories in Mood During Focused CBT Enhanced for Eating Disorders Are Associated with Global Eating Pathology at Post-Treatment & Follow-up Among Individuals with Bulimia Nervosa-spectrum Disorders: A Preliminary Examination

Elizabeth W Lampe a,b,*, Paakhi Srivastava b,*, Emily K Presseller a,b, Megan L Wilkinson a,b, Claire Trainor a,b, Stephanie M Manasse b,**, Adrienne S Juarascio a,b,**
PMCID: PMC11413876  NIHMSID: NIHMS2017429  PMID: 39174272

Abstract

Bulimia nervosa (BN) is characterized by recurrent loss-of-control over eating (LOC) and inappropriate compensatory behaviors. Although Cognitive Behavioral Therapy (CBT) is efficacious for BN, many patients continue to experience symptoms at post-treatment. One potential driver of this low treatment response may be low mood, which maintains BN symptoms through negative reinforcement. Thus, it is important to understand how mood changes over CBT-E and whether these changes are associated with improved treatment outcomes. Participants (N = 56) with BN-spectrum eating disorders (EDs) received 16 sessions of the focused version of CBT-E. The Eating Disorder Examination was used to measure ED symptoms (Global ED pathology, frequency of binge episodes and compensatory behaviors) at pre- and post-treatment. Latent growth mixture modeling of affective ratings via digital self-monitoring identified latent growth classes. Kruskal–Wallis H tests examined the effect of trajectory of change in mood on pre- to post-treatment symptom change. Latent growth mixture modeling yielded a four-class model that best fit the data representing distinct mood trajectories over the course of treatment: 1) Highest baseline mood, linear improving, 2) Moderate baseline mood, stable, 3) Moderate baseline mood, quadratic worsening, and 4) Lowest baseline mood, quadratic improving. Participants who demonstrated worsening mood over treatment (i.e., individuals in the “Moderate baseline mood, quadratic worsening” class) had significantly higher Eating Disorder Examination (EDE) global scores at post-treatment and follow-up compared to participants with stable mood across treatment. Change in LOC eating frequency and compensatory behaviors across treatment did not significantly differ by mood class. The main effect of mood class or interaction effect between time and mood class on objective binge episodes, subjective binge episodes and compensatory behaviors was not significant. There were no significant differences in global ED pathology at either post-treatment or follow-up for any other class comparisons. These results suggest that certain trajectories of change in mood during treatment are particularly associated with change in pre- to post-treatment EDE global score. If replicated, our findings could suggest that future iterations of CBT-E should target mood early in treatment in order to maximize reductions in global eating pathology.

Keywords: Bulimia Nervosa, Affect, Cognitive Behavioral Therapy

Introduction

Bulimia nervosa (BN), an eating disorder characterized by recurrent episodes of binge eating (i.e., experiencing a sense of loss-of-control over eating) and inappropriate compensatory behaviors (e.g., self-induced vomiting, laxatives, diuretics, driven exercise), is a serious psychiatric condition that is associated with severe health and psychosocial consequences (American Psychiatric Association, 2013). Enhanced Cognitive Behavioral Therapy (CBT-E) is the frontline treatment for BN and targets transdiagnostic maintenance factors of eating disorders such as dietary restraint and overvaluation of shape and weight (Fairburn et al., 2015). Although CBT-E is highly efficacious for some, up to 70% of patients experience inadequate reduction of symptoms at post-treatment (Linardon, 2018; Linardon & Wade, 2018; Olmsted et al., 2015; Södersten et al., 2017). One potential driver of poor treatment response may be low mood, which has been strongly supported in the eating disorders literature as a maintenance factor of BN (Haedt-Matt & Keel, 2011; Stice et al., 1996). For example, negative reinforcement models of BN suggest that individuals engage in binge eating or compensatory behaviors as a method of alleviating or reducing negative affect (e.g., guilt, depression, anxiety) (Haedt-Matt & Keel, 2011, 2015). Alternatively, positive reinforcement models of BN posit that individuals may engage in binge eating for its rewarding properties to extend positive affective experiences (Leslie et al., 2018). Given that affect has been indicated as a key contributor to BN pathology for a subset of patients, it is important to understand how mood changes over CBT-E and whether these changes are associated with improved treatment outcomes in order to improve outcomes for BN.

There is evidence that CBT-E for eating disorders decreases negative mood and increases positive mood on average (S. M. Byrne et al., 2011; C. M. Grilo et al., 2012). However, these studies did not explore potential variability in mood changes over the course of CBT-E. For example, some patients may experience rapid rates of improvement in mood (e.g., individuals with very limited use of mood regulation skills prior to treatment), while others may experience slower rates of improvement (e.g., individuals who attempt but struggle to implement affect regulation skills), remain stable, or even deteriorate (e.g., individuals who gain weight during treatment may temporarily experience more negative mood due to heightened body dissatisfaction). Individuals in each of these groups may have very different treatment outcomes, as individuals who experience improvements in mood may no longer engage in binge eating and/or compensatory behaviors to regulate their mood, while individuals who do not experience improvements in their mood may remain at risk for these behaviors. It is essential to empirically evaluate how different profiles of improvement in mood during CBT-E may impact treatment outcomes.

Some limited previous research supports the role of individual differences in the impact of both negative and positive mood on eating disorder treatment outcomes. Patients receiving guided self-help CBT-E targeting binge eating who reported higher pre-treatment negative mood demonstrated higher post-treatment eating pathology (Masheb & Grilo, 2008a, 2008b). Furthermore, heterogenous trajectories of change during CBT-E has been found to predict treatment response in other constructs besides mood. In particular, trajectories of eating disorder symptom change (e.g., rate of binge eating change, rate of compensatory behavior change) over CBT-E has been shown to differentially predict eating pathology at post-treatment and follow-up (Hilbert et al., 2019; Linardon et al., 2016; Presseller et al., 2022). Therefore, we might expect that variance in trajectories of mood change during CBT-E is also an important predictor of treatment outcomes.

Thus, the current study was a preliminary examination of change trajectories of mood over the course of CBT-E in relation to treatment outcomes among patients with BN-spectrum eating disorders. Patients with once weekly binge eating and compensatory behaviors over the past three months received 16 weeks of the focused version of CBT-E. Affect was measured multiple times per day over the course of treatment using ecological momentary assessment (EMA). The current study aimed to: (1) identify latent trajectories of change in affect during treatment, and (2) examine the association between trajectories of affect change during treatment and post-treatment ED symptoms.

Methods

The current study aimed to: (1) identify latent trajectories of change in mood (i.e., average of affect ratings reported between sessions) during treatment, and (2) examine associations of these trajectories with ED treatment outcomes. All analyses were preregistered with the Open Science Framework and can be found at the following link: https://osf.io/5y7us/. The current study represents a secondary data analysis of participants enrolled in a clinical trial recruiting participants with BN-spectrum eating disorders to complete a course of CBT-E focused version with or without adjunctive digital self-monitoring (A. Juarascio et al., 2021; Juarascio et al., 2022). Participants were eligible if they were between 18–70 years old, had a BMI ≥17.5, reported at least 12 objectively- or subjectively-large binge eating episodes and 12 or more compensatory behaviors in the past three months, and were willing and able to use a smartphone application to track eating patterns and eating disorder behaviors multiple times per day for 16 weeks. Participants were excluded if they were unable to fluently speak, read, or write English, had medical complications that would prohibit safe engagement in outpatient treatment, had a comorbid diagnosis of a psychotic disorder, bipolar disorder, or substance dependence or a diagnosis of intellectual disability or pervasive development disorder, if they had previously completed a full trial of CBT-E for BN, were currently pregnant or planned to become pregnant within the next year, or had a history of bariatric surgery.

Demographics

Participants (N=56, 83.6% female) in this study were on average 38.5 years old (SD=13.8). Of all participants enrolled in the study, 65.4% were Non-Hispanic Caucasian, 9.1% were African American, 7.3% Asian, and 18.1% multiracial or other. The average BMI for participants in the study was 29.5 kg/m2 (SD=6.8). Of the sample, 90.9% met DSM-5 behavioral criteria for BN and 9.1% met criteria for another BN-spectrum eating disorder (i.e., BN with only subjectively large binge episodes).

Procedures

The current study was a secondary data analysis of a trial examining an adjunctive smartphone application for CBT-E for BN-spectrum eating disorders (Juarascio et al., 2022). Participants were recruited from the Philadelphia area (pre-COVID-19 pandemic, n=51) using radio advertising, social media advertising, and community advertising (i.e., flyers) and nationally (during the COVID-19 pandemic, n=5). Interested participants met with research coordinators to provide informed consent and complete a baseline assessment to confirm eligibility. Once eligibility and interest were confirmed, participants received 16 sessions of manualized, individual outpatient CBT-E (focused, hybrid version including mood intolerance module; Murphy et al., 2010) while utilizing an adjunctive smartphone application for electronic self-monitoring. Treatment was delivered by master’s- and doctoral-level students, postdoctoral fellows, and licensed clinical psychologists who received weekly group supervision by a licensed clinical psychologist. Although some participants in the study received just in time adaptive interventions from their smartphone during treatment while others did not, treatment conditions were collapsed for the current study given its exploratory nature and treatment condition was entered as a covariate in analyses. Treatment retention was 89.3% across the entire sample, and participants completed on average 3.01 entries per day (SD = 3.01) across all treatment sessions. Participants were assessed by trained research staff at baseline, mid-treatment, post-treatment, and 3-month follow-up. Further information on the CBT-E treatment and outcomes are published in Juarascio at al. (2022).

Measures

Affect.

Affect was measured using the question “How has your mood been since your last survey”. Participants answered on a Likert scale from 1 (“most sad”) to 5 (“most happy”). Participants completed ratings of affect whenever making an entry in the smartphone application (e.g., when entering a meal/snack, or logging an ED behavior. The average number of affect ratings per day was 5.23 (SD = 3.27). Mean mood ratings were computed for each period (i.e., period between two therapy sessions, mean = 7.2 days, SD = 1.45 days) using all affect ratings reported within this period. We considered utilizing the previously validated method of EMA-style measurement of affect (for example, Wonderlich et al, 2015) to assess affect. However, we decided to use a brief Likert scale to reduce participant burden and promote overall self-monitoring compliance, given that affect ratings were collected as part of electronic self-monitoring records in this clinical trial.

Treatment Outcomes.

The Eating Disorder Examination (EDE) was used to gather information about frequency of binge eating and compensatory behaviors (Fairburn et al., 2014). The EDE also generates a global score of ED psychopathology. Compensatory behavior frequency was computed by summing episodes of self-induced vomiting, laxative misuse, diuretic misuse, driven exercise, fasting, chewing and spitting, and other extreme weight control behaviors (e.g., diet pill use). Frequency of binge eating, compensatory behaviors, and EDE global score were all used as outcome variables. Given that the subjective sense of loss of control over one’s eating is most closely related to distress and clinical impairment in binge-spectrum eating disorders (Colles et al., 2008; Goldschmidt et al., 2016; Niego et al., 1997), we chose to combine objective and subjective binge episodes in frequency counts, as well as to examine objectively large and subjectively large binge episodes separately.

Statistical Analyses

Latent growth mixture modeling (LGMM) was conducted to examine unobserved trajectories of change in mood during treatment using the ‘lcmm’ package in R. As mean mood rating data were not normally distributed, data were pre-normalized using the procedure described in “Pre-normalizing a dependent variable using lcmm” (2015). One- to four-class mixed effects growth models were fitted. LGMM was conducted with random intercepts and random slopes to maximize model fit. Initial start values were based on the one-class model and each model was estimated 100 times. Fixed and random intercepts, linear and quadratic effects of time, and a fixed effect of treatment condition on mood were estimated. Best-fit models were selected by minimizing the Sample-Size Adjusted Bayesian information criterion (SABIC) and the Akaike information criterion (AIC); when the SABIC and AIC disagreed, minimizing SABIC was prioritized. Class membership was assigned using posterior probabilities and entropy was examined as a marker of class separation. Mood classes did not significantly differ in proportion of participants who completed treatments before and during COVID-19 (Fisher’s exact p = .33).

Change in eating pathology (binge eating episodes, compensatory behaviors, and EDE global score) across time was examined using zero-inflated mixed models (binge eating, objective binge eating episodes, subjective binge eating episodes, compensatory behaviors) and linear mixed effects model (EDE global score), with fixed effects of latent mood class and time (pre-treatment, post-treatment, and follow-up) nested within participant; for the objective binge eating episode model, fixed effects were not nested within participant for the zero-inflated model given that nesting within participant prevented model convergence. A negative binomial distribution was used for all zero-inflated models given overdispersion. This statistical approach was a deviation from our planned analytic approach but was utilized given concerns that our planned approach failed to account for potential differences in baseline levels of eating pathology by latent mood class.

Results

Latent Class Identification

Latent growth mixture modeling yielded a four-class model that best fit the data, based on minimizing sample-size adjusted BIC (see Table 2). The four-class model demonstrated good class separation (entropy = 0.86) and posterior probability (mean posterior probabilities ranged from 0.85 to 0.98). The four classes demonstrated distinct baseline mood (means ranged from 2.52 to 3.40 out of 5) and differed in direction and rate of change across treatment (see Figures 1). The first class, labeled “Highest baseline mood, linear improving” demonstrated an average affect rating of 3.40 out of 5, which increased linearly across treatment to an average rating of 3.84 at end of treatment. Class 2, labeled “Moderate baseline mood, stable,” demonstrated an average baseline affect rating of 2.96 and stable mood across treatment, with a mean rating of 3.00 at end of treatment. Class 3, termed “Moderate baseline mood, quadratic worsening,” demonstrated an initial average affect rating of 2.92, which decreased to 2.05 at end of treatment. The fourth class, labeled “Lowest baseline mood, quadratic improving,” reported an average affect rating of 2.52 at baseline, which increased to 3.69 at end of treatment.

Table 2.

Best-fit models of change in mood and their association with post-treatment worsening of symptoms.

Fit Indices and Growth Model Parameter Estimates for 1–4 Class Models
Number of Classes Number of Parameters Max. Log-Likelihood BIC SABIC AIC Entropy Class Membership Proportions Mean Posterior Probability % Posterior Probability > 0.80
1 11 −1143.76 2331.20 2296.65 2309.53 1.00 Class 1: 100.0% Class 1: 1.00 Class 1: 1.00
2 17 −1126.40 2320.29 2266.89 2286.79 0.66 Class 1: 33.96% Class 1: 0.89 Class 1: 0.78
Class 2: 66.04% Class 2: 0.92 Class 2: 0.94
3 23 −1117.84 2326.98 2254.74 2281.67 0.92 Class 1: 73.58% Class 1: 0.97 Class 1: 0.97
Class 2: 13.21% Class 2: 0.97 Class 2: 1.00
Class 3: 13.21% Class 3 0.93 Class 3: 0.86
4 29 −1112.51 2340.07 2249.07 2283.02 0.86 Class 1: 45.28% Class 1: 0.94 Class 1: 0.92
Class 2: 28.30% Class 2: 0.95 Class 2: 0.93
Class 3: 13.21% Class 3: 0.98 Class 3: 1.00
Class 4: 13.21% Class 4: 0.85 Class 4: 0.71
Fixed Effects: Class Membership Model Fixed Effects: Longitudinal Model
Intercept Est. (SE), p Intercept Est. (SE), p Linear time Est. (SE), p Quadratic time Est. (SE), p
Highest baseline mood, linear improving 1.20 (0.53), .02* 0.63 (0.36), .08 0.10 (0.07), .12 −0.001 (.004), .79
Moderate baseline mood, stable 0.78 (0.55), .15 −1.43 (0.58), .01 0.07 (0.09), .46 −0.004 (0.01), .46
Moderate baseline mood, quadratic worsening 0.13 (0.64), .83 −1.77 (1.73) .31 −0.55 (0.35), .11 0.02 (0.02), .29
Lowest baseline mood, quadratic improving Not estimated, reference class −2.51 (1.94), .20 0.71 (0.35), .04* −0.03 (0.02), .15
Open in a new tab

Figure 1.

Figure 1.

Open in a new tab

Predicted trajectories of change in mood during CBT-E for BN

Note: Line designates model-predicted trajectory, shaded bar indicates 95% confidence interval of model-predicted trajectory, and points designate observed mean values within each class. Figure y-axis ranges from −8 to 8 due to pre-normalization of mean mood data. On this scale, −8 corresponds to the lowest possible mean mood rating, 0 corresponds to a neutral mood rating, and 8 corresponds to the highest possible mean mood rating.

Change in ED Symptoms over Time by Class

There was a significant main effect of time (F = 25.39, p <.001) on total binge eating frequency, but the main effect of mood class (F = 1.10, p = .35) and interaction effect of time by mood class (F = 0.28, p = .95) were non-significant. Consistent with this pattern of results, there was a significant main effect of time (F = 19.60, p <.001), but no significant main effect of mood class (F = 0.78, p = .51) or interaction effect between time and mood class (F = 0.65, p = .69) on objective binge episodes. There was a significant main effect of time (F = 7.81, p <.001) on subjective binge eating, but the main effect of mood class (F = 1.36, p = .26) and time by mood class interaction effect (F = 0.62, p = .71) were not significant. Regarding compensatory behaviors, there was a significant main effect of time (F = 20.65, p <.001), while the main effect of mood class (F = 1.58, p = .20) and time by mood class interaction effect (F = 1.13, p = .35) were non-significant. There were significant, main effects of time (F = 95.18, p < .001) and mood class (F = 2.97, p = .04) and a significant interaction effect between time and mood class (F = 3.54, p = .003) on EDE global score (see Table 3). Individual parameter estimates for interaction effects in the linear mixed effect model indicate that the “Moderate baseline mood, quadratic worsening” class demonstrated significantly higher EDE global scores at post-treatment and follow-up compared to the “Moderate baseline mood, stable” reference class (see Table 3).

Table 3.

Mixed models examining the interaction between mood class and time on eating pathology

ED Behavior Predictor Category Predictor Zero-Inflation Model Estimate (p) Conditional Model Estimate (p)
Total Loss of Control Eating Mood Class Moderate baseline mood, stable −3.57 (1.00) −0.19 (.49)
Highest baseline mood, linear improving Reference class Reference class
Lowest baseline mood, quadratic improving −2.66 (1.00) 0.18 (.62)
Moderate baseline mood, quadratic worsening −3.21 (1.00) 0.05 (.90)
Timepoint Baseline Reference class Reference class
Post-treatment 22.73 (.996) −1.30 (<.001)
Follow-up 21.63 (.997) −0.77 (.002)
Interactions Moderate baseline mood, stable * Post-treatment 1.63 (1.00) −0.50 (.25)
Lowest baseline mood, quadratic improving * Post-treatment 2.16 (1.00) 0.58 (.29)
Moderate baseline mood, quadratic worsening * Post-treatment 0.78 (1.00) 0.77 (.13)
Moderate baseline mood, stable * Follow-up 4.29 (1.00) −0.14 (.73)
Lowest baseline mood, quadratic improving * Follow-up 4.65 (1.00) 0.38 (.50)
Moderate baseline mood, quadratic worsening * Follow-up 1.88 (1.00) 0.21 (.66)
Objective Binge Eating Episodes Mood Class Moderate baseline mood, stable −19.54 (.999) −0.30 (.33)
Highest baseline mood, linear improving Reference class Reference class
Lowest baseline mood, quadratic improving −20.35 (.999) −0.11 (.78)
Moderate baseline mood, quadratic worsening −15.00 (.997) −0.19 (.64)
Timepoint Baseline Reference class Reference class
Post-treatment 3.78 (.02) −1.14 (<.001)
Follow-up 3.01 (.07) −1.18 (<.001)
Interactions Moderate baseline mood, stable * Post-treatment 17.56 (.999) −1.22 (.03)
Lowest baseline mood, quadratic improving * Post-treatment 20.43 (.999) 1.20 (.05)
Moderate baseline mood, quadratic worsening * Post-treatment 14.45 (.997) 0.83 (.15)
Moderate baseline mood, stable * Follow-up 20.57 (.999) −0.11 (.84)
Lowest baseline mood, quadratic improving * Follow-up 21.13 0.31 (.62)
Moderate baseline mood, quadratic worsening * Follow-up 14.43 (.997) 0.41 (.45)
Subjective Binge Eating Episodes Mood Class Moderate baseline mood, stable (3) −1.06 (.19) −0.41 (.30)
Highest baseline mood, linear improving (1) Reference class Reference class
Lowest baseline mood, quadratic improving (4) −0.46 (.62) 0.66 (.17)
Moderate baseline mood, quadratic worsening (2) −0.46 (.62) 0.22 (.64)
Timepoint Baseline Reference class Reference class
Post-treatment 0.12 (.88) −1.98 (<.001)
Follow-up −0.24 (.70) −0.44 (.21)
Interactions Moderate baseline mood, stable * Post-treatment 0.23 (.87) 0.65 (.33)
Lowest baseline mood, quadratic improving * Post-treatment 0.46 (.80) −0.71 (.48)
Moderate baseline mood, quadratic worsening * Post-treatment 0.28 (.86) 0.89 (.22)
Moderate baseline mood, stable * Follow-up 1.17 (.28) 0.24 (.65)
Lowest baseline mood, quadratic improving * Follow-up 0.85 (.51) −0.005 (.99)
Moderate baseline mood, quadratic worsening * Follow-up 2.23 (.11) 0.33 (.62)
Compensatory Behaviors Mood Class Moderate baseline mood, stable (3) −3.95 (1.00) −0.23 (.39)
Highest baseline mood, linear improving (1) Reference class Reference class
Lowest baseline mood, quadratic improving (4) −4.16 (1.00) 0.17 (.63)
Moderate baseline mood, quadratic worsening (2) −3.80 (1.00) −0.43 (.22)
Timepoint Baseline Reference class Reference class
Post-treatment 25.53 (1.00) −0.83 (.002)**
Follow-up 22.30 (1.00) −0.45 (.03)*
Interactions Moderate baseline mood, stable * Post-treatment 1.05 (1.00) −0.07 (.85)
Lowest baseline mood, quadratic improving * Post-treatment −3.71 (1.00) −0.17 (.68)
Moderate baseline mood, quadratic worsening * Post-treatment −1.80 (1.00) 0.82 (.06)
Moderate baseline mood, stable * Follow-up 4.33 (1.00) −0.68 (.06)
Lowest baseline mood, quadratic improving * Follow-up 3.63 (1.00) −0.28 (.51)
Moderate baseline mood, quadratic worsening * Follow-up 3.20 (1.00) 0.67 (.11)
ED Behavior Predictor Mixed Factorial ANOVA Model β (p)
EDE Global Score Mood Class Moderate baseline mood, stable (3) 0.44 (.20)
Highest baseline mood, linear improving (1) Reference class
Lowest baseline mood, quadratic improving (4) 0.17 (.71)
Moderate baseline mood, quadratic worsening (2) 0.01 (.98)
Timepoint Baseline Reference class
Post-treatment −1.87 (<.001)***
Follow-up −1.84 (<.001)***
Interactions Moderate baseline mood, stable * Post-treatment 0.22 (.51)
Lowest baseline mood, quadratic improving * Post-treatment 0.28 (.51)
Moderate baseline mood, quadratic worsening * Post-treatment 1.39 (.002)**
Moderate baseline mood, stable * Follow-up −0.14 (.67)
Lowest baseline mood, quadratic improving * Follow-up −0.15 (.73)
Moderate baseline mood, quadratic worsening * Follow-up 1.66 (<.001)***
Open in a new tab

Note: Reference class is included in the table above to designate that a predictor level was considered the reference class within the model and therefore no estimates were computed.

Discussion

The current study examined latent change trajectories of mood during CBT-E and their association with post-treatment ED symptoms among 56 adults with BN-spectrum eating disorders. We identified four latent classes of change in mood across treatment: 1) “Highest baseline mood, linear improving,” 2) “Moderate baseline mood, stable,” 3) “Moderate baseline mood, quadratic worsening,” and 4) “Lowest baseline mood, quadratic improving” (see fig. 1). These heterogeneous trajectories of change in mood across CBT-E demonstrate clear differences in the rate and direction of mood change for patients with BN-spectrum eating disorders. These findings are consistent with extant literature demonstrating different trajectories of change (e.g., quadratic versus linear, improving versus worsening) in other ED symptoms during treatments for EDs (Hilbert et al., 2019; Linardon et al., 2016; Presseller et al., 2022). Further, the identified mood trajectories are also consistent with previous research showing that two distinct negative affect classes including high baseline depression scores with greater overall improvements with treatment and low baseline depression scores with smaller overall improvements with treatment. (S. M. Byrne et al., 2011; C. M. Grilo et al., 2012). While previous studies have examined the improvements in mood class at post-treatment, our study assessed trajectories of change over time during treatments. Our results showed that mood changes observed across treatment were small (e.g., ranging from 0.03-point increase in the stable class to 1.17-point increase in the Lowest baseline affect, quadratic improving class on a 1–5 scale). Although, in this clinical trial, the focused and hybrid version of CBT-E including mood intolerance module was used, the treatment content dedicated for improving mood is relatively small and is covered over two sessions (Fairburn, 2008; Fairburn et al., 2009) Additionally, the moderate baseline quadratic worsening class showed mood deterioration throughout the treatment. The fact that moderate baseline mood worsened was surprising in that negative mood is consistently shown to improve in CBT-E for BN (Susan M Byrne et al., 2011; Carlos M Grilo et al., 2012). Future research should examine factors that may explain differential changes in various baseline mood types (e.g., low, moderate and high) during treatment (Wonderlich et al., 2015). There was a significant interaction effect between time and mood class on EDE global score, indicating that the “Moderate baseline mood, quadratic worsening” class demonstrated significantly higher EDE global scores and post-treatment and follow-up compared to the “Moderate baseline mood, stable” reference class. If replicated in larger samples, these findings may suggest that worsening negative mood during treatment is a risk factor for persistence of ED symptoms at post-treatment and follow-up even after completion of full treatment course. (Goldschmidt et al., 2014; Haedt-Matt & Keel, 2015) Thus, it may be valuable to monitor changes in mood during treatment and optimally introduce in interventions (e.g., mood intolerance strategies from broad version of CBT-E) to maximize impact on global eating pathology.

Surprisingly, we did not observe significant class differences in compensatory behaviors or binge eating outcomes (including all binge episodes, subjective binge episodes only, or objective binge episodes only) at post-treatment. However, non-significant differences by mood class on binge eating at post-treatment are contrary to extant literature linking increased negative affect and binge eating on a momentary level (Goldschmidt et al., 2014; Haedt-Matt & Keel, 2011; Masheb & Grilo, 2008a, 2008b; Srivastava et al., 2021). Our lack of findings regarding the relationship between heterogeneous mood trajectories and binge eating outcomes may suggest that other factors may reinforce binge eating on a longer timescale. For example, recent research supports the role of reward in positively reinforcing binge eating (Steward et al., 2018), such that patients may seek out the rewarding aspects of palatable food when experiencing anhedonia. Future research should examine whether reward to non-food sources changes over treatment and whether these changes affect binge eating outcomes in patients with BN-spectrum eating disorders.

Strengths & Limitations

The current study was the first to examine latent trajectories of change in mood during CBT-E. Strengths of the study were the use of the EMA to measure real-time affect during treatment and identify latent trajectories of change, which promoted ecological validity and minimized recall bias. Another strength of the study was the use of a data-driven analytic approach to identify latent trajectories. The study was limited by the small sample size, limiting power and potentially producing unreliable statistical estimates, particularly inflated effect sizes. Additionally, this study was limited by the narrow focus on individuals with BN. Given that CBT-E is designed to be applied transdiagnostically, the lifetime comorbidity of EDs (Serra et al., 2022) and that over half of EDs fall within the OSFED category (Ward et al., 2019), this narrow focus limits the generalizability of our findings. Future research should explore these associations in larger, more transdiagnostic samples. Because of limited sample size, we also were unable to covary for a number of variables that may impact treatment response, including therapeutic alliance, treatment duration, age, gender, and ethnicity. Future research should seek to replicate our findings while covarying for these potential confounds. While EMA data are likely more accurate than weekly self-report or therapist report, long-term use of EMA demonstrates some decline in compliance (Engel et al., 2016; A. S. Juarascio et al., 2021; Schaefer et al., 2020), which was also apparent in our sample and limits both statistical power and reliability of later affect ratings. Additionally, we conducted bi-polar measurement of affect (i.e., on a scale from totally negative mood to totally positive mood) which precluded observations of both negatively and positively valanced affect at the same survey. Moreover, this measurement approach did not allow for capturing more specific emotional states (e.g., sad, angry, anxious, ashamed, excited, calm) as utilized in other EMA studies (Wonderlich et al., 2015). While the current affect measurement approach was elected to minimize participant burden, it limited our ability to capture true affective experience of participants. Finally, therapy was initially administered in-person but later transitioned to Zoom during the COVID-19 pandemic, which may have impacted treatment response. It is possible COVID might have impacted mood in ways that we cannot systematically investigate, although identified mood classes did not differ in proportion of participants who completed treatment prior to and during COVID-19. Nonetheless, the onset of the COVID-19 pandemic may have contributed to increased symptoms during or following treatment for some individuals due to increased psychological stress and changes to food environments.

Conclusions

Our results show some indications of differential mood change trajectories and their association with improvements in EDE global score during the course of focused version of CBT-E. These results suggest the need to assess and monitor mood trajectories during treatment. If replicated in larger samples, mood regulation skills should be emphasized in treatment to maximize reductions in EDE global scores. Future research should evaluate the role of other factors such as baseline mood on the association between mood trajectories and EDE global scores.

Table 1.

Mean ED behaviors over the previous 28 days from pre-treatment to follow-up by class.

ED Behavior Class Mbaseline Mpost-tx Mfollow-up
Overall Sample 26.00 5.32 8.13
Loss of Control Episodes Moderate baseline mood, stable 21.40 2.67 5.20
Highest baseline mood, linear improving 28.29 3.92 8.71
Lowest baseline mood, quadratic improving 29.14 9.00 9.29
Moderate baseline mood, quadratic worsening 24.86 12.14 11.29
Objective Binge Episodes Overall Sample 18.96 4.06 3.79
Moderate baseline mood, stable 15.47 1.27 1.93
Highest baseline mood, linear improving 23.00 3.21 4.29
Lowest baseline mood, quadratic improving 16.71 8.29 3.00
Moderate baseline mood, quadratic worsening 14.86 8.71 6.86
Subjective Binge Episodes Overall Sample 7.04 1.26 4.34
Moderate baseline mood, stable 5.93 1.40 3.27
Highest baseline mood, linear improving 5.29 0.71 4.42
Lowest baseline mood, quadratic improving 12.43 0.71 6.29
Moderate baseline mood, quadratic worsening 10.00 3.43 4.43
Compensatory Behaviors Overall Sample 32.72 9.26 12.75
Moderate baseline mood, stable 27.40 7.07 4.93
Highest baseline mood, linear improving 35.33 5.75 15.92
Lowest baseline mood, quadratic improving 42.43 13.71 11.29
Moderate baseline mood, quadratic worsening 25.43 21.57 20.14
EDE Global Score Overall Sample 3.31 1.72 1.63
Moderate baseline mood, stable 3.60 1.94 1.62
Highest baseline mood, linear improving 3.16 1.29 1.32
Lowest baseline mood, quadratic improving 3.32 1.73 1.34
Moderate baseline mood, quadratic worsening 3.17 2.68 2.98
Open in a new tab

Note. Loss of control episodes includes both objective and subjective binge episodes.

Funding:

This research received support from the National institute of Mental Health (R34MH116021; K23DK124514).

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.

References

  1. American Psychiatric Association. (2013). Diagnostic and Statistical Manual of Mental Disorders (5th ed.). American Psychiatric Association. 10.1176/appi.books.9780890425596.dsm10 [DOI] [Google Scholar]
  2. Byrne SM, Fursland A, Allen KL, & Watson H (2011). The effectiveness of enhanced cognitive behavioural therapy for eating disorders: an open trial. Behavior Research and Therapy, 49(4), 219–226. 10.1016/j.brat.2011.01.006 [DOI] [PubMed] [Google Scholar]
  3. Byrne SM, Fursland A, Allen KL, & Watson H (2011). The effectiveness of enhanced cognitive behavioural therapy for eating disorders: An open trial. Behaviour Research and Therapy, 49(4), 219–226. [DOI] [PubMed] [Google Scholar]
  4. Colles SL, Dixon JB, & O’Brien PE (2008). Loss of control is central to psychological disturbance associated with binge eating disorder. Obesity, 16(3), 608–614. [DOI] [PubMed] [Google Scholar]
  5. Engel SG, Crosby RD, Thomas G, Bond D, Lavender JM, Mason T, . . . Wonderlich SA (2016). Ecological momentary assessment in eating disorder and obesity research: a review of the recent literature. Current psychiatry reports, 18(4), 37. [DOI] [PubMed] [Google Scholar]
  6. Fairburn. (2008). Cognitive behavior therapy and eating disorders. Guilford Press. [Google Scholar]
  7. Fairburn, Bailey-Straebler S, Basden S, Doll HA, Jones R, Murphy R, . . . Cooper Z (2015). A transdiagnostic comparison of enhanced cognitive behaviour therapy (CBT-E) and interpersonal psychotherapy in the treatment of eating disorders. Behaviour Research and Therapy, 70, 64–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Fairburn Cooper, Doll O Connor, Bohn Hawker, . . . Palmer. (2009). Transdiagnostic cognitive-behavioral therapy for patients with eating disorders: a two-site trial with 60-week follow-up. American Journal of Psychiatry, 166(3), 311–319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Fairburn Cooper, & O’Connor. (2014). The Eating Disorder Examination. The Centre for Research on Eating Disorders at Oxford. In: Oxford, UK. [Google Scholar]
  10. Goldschmidt AB, Accurso EC, Crosby RD, Cao L, Ellison J, Smith TL, . . . Wonderlich SA (2016). Association between objective and subjective binge eating and psychopathology during a psychological treatment trial for bulimic symptoms. Appetite, 107, 471–477. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Goldschmidt AB, Wonderlich SA, Crosby RD, Engel SG, Lavender JM, Peterson CB, . . . Mitchell JE (2014). Ecological momentary assessment of stressful events and negative affect in bulimia nervosa. Journal of Consulting and Clinical Psychology, 82(1), 30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Grilo CM, Masheb RM, & Crosby RD (2012). Predictors and moderators of response to cognitive behavioral therapy and medication for the treatment of binge eating disorder. Journal of consulting and clinical psychology, 80(5), 897–906. 10.1037/a0027001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Grilo CM, Masheb RM, & Crosby RD (2012). Predictors and moderators of response to cognitive behavioral therapy and medication for the treatment of binge eating disorder. Journal of Consulting and Clinical Psychology, 80(5), 897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Haedt-Matt AA, & Keel PK (2011). Revisiting the affect regulation model of binge eating: a meta-analysis of studies using ecological momentary assessment. Psychological bulletin, 137(4), 660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. 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]
  16. Hilbert A, Herpertz S, Zipfel S, Tuschen-Caffier B, Friederich H-C, Mayr A, . . . de Zwaan M (2019). Early Change Trajectories in Cognitive-Behavioral Therapy for Binge-Eating Disorder. Behavior Therapy, 50(1), 115–125. 10.1016/j.beth.2018.03.013 [DOI] [PubMed] [Google Scholar]
  17. Juarascio A, Srivastava P, Clark K, Presseller E, Manasse S, & Forman E (2021). A Clinician-Controlled Just-in-time Adaptive Intervention System (CBT+) Designed to Promote Acquisition and Utilization of Cognitive Behavioral Therapy Skills in Bulimia Nervosa: Development and Preliminary Evaluation Study. JMIR Formative Research, 5(5), e18261. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Juarascio AS, Hunt RA, Lantz Lesser E, Engel SG, Pisetsky EM, Peterson CB, & Wonderlich SA (2021). Enhancing Integrative Cognitive-Affective Therapy with ecological momentary interventions: A pilot trial. European Eating Disorders Review, 29(1), 152–158. [DOI] [PubMed] [Google Scholar]
  19. Juarascio AS, Presseller EK, Srivastava P, Manasse SM, & Forman EM (2022). A Randomized Controlled Trial of CBT+: A Clinician-Controlled, Just-In-Time, Adjunctive Intervention for Bulimia-Spectrum Disorders. Behavior Modification, 01454455221109434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Leslie M, Turton R, Burgess E, Nazar BP, & Treasure J (2018). Testing the addictive appetite model of binge eating: The importance of craving, coping, and reward enhancement. European Eating Disorders Review, 26(6), 541–550. 10.1002/erv.2621 [DOI] [PubMed] [Google Scholar]
  21. Linardon J (2018). Meta-analysis of the effects of cognitive-behavioral therapy on the core eating disorder maintaining mechanisms: implications for mechanisms of therapeutic change. Cognitive behaviour therapy, 47(2), 107–125. [DOI] [PubMed] [Google Scholar]
  22. Linardon J, Brennan L, & de la Piedad Garcia X (2016). Rapid response to eating disorder treatment: A systematic review and meta-analysis. Int J Eat Disord, 49(10), 905–919. 10.1002/eat.22595 [DOI] [PubMed] [Google Scholar]
  23. Linardon J, & Wade TD (2018). How many individuals achieve symptom abstinence following psychological treatments for bulimia nervosa? A meta-analytic review. International Journal of Eating Disorders, 51(4), 287–294. [DOI] [PubMed] [Google Scholar]
  24. Masheb, & Grilo. (2008a). Examination of predictors and moderators for self-help treatments of binge-eating disorder. Journal of Consulting and Clinical Psychology, 76(5), 900. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Masheb, & Grilo. (2008b). Prognostic significance of two sub-categorization methods for the treatment of binge eating disorder: Negative affect and overvaluation predict, but do not moderate, specific outcomes. Behaviour Research and Therapy, 46(4), 428–437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Murphy R, Straebler S, Cooper Z, & Fairburn CG (2010). Cognitive behavioral therapy for eating disorders. Psychiatric Clinics, 33(3), 611–627. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Niego SH, Pratt EM, & Agras WS (1997). Subjective or objective binge: Is the distinction valid? International Journal of Eating Disorders, 22(3), 291–298. [DOI] [PubMed] [Google Scholar]
  28. Olmsted MP, MacDonald DE, McFarlane T, Trottier K, & Colton P (2015). Predictors of rapid relapse in bulimia nervosa. International Journal of Eating Disorders, 48(3), 337–340. [DOI] [PubMed] [Google Scholar]
  29. Pre-normalizing a dependent variable using lcmm. (2015). https://cran.r-project.org/web/packages/lcmm/vignettes/pre_normalizing.html
  30. Presseller EK, Lampe EW, Michael ML, Trainor C, Fan SC, & Juarascio AS (2022). Latent trajectories of symptom change during cognitive-behavior therapy predict post-treatment worsening of symptoms: a preliminary examination among outpatients with bulimia-spectrum eating disorders. Eating and Weight Disorders - Studies on Anorexia, Bulimia and Obesity. 10.1007/s40519-021-01348-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. 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]
  32. Serra R, Di Nicolantonio C, Di Febo R, De Crescenzo F, Vanderlinden J, Vrieze E, . . . Tarsitani L (2022). The transition from restrictive anorexia nervosa to binging and purging: a systematic review and meta-analysis. Eating and Weight Disorders-Studies on Anorexia, Bulimia and Obesity, 27(3), 857–865. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Södersten P, Bergh C, Leon M, Brodin U, & Zandian M (2017). Cognitive behavior therapy for eating disorders versus normalization of eating behavior. Journal of Physiology and Behavior, 174, 178–190. [DOI] [PubMed] [Google Scholar]
  34. Srivastava P, Lampe EW, Michael ML, Manasse S, & Juarascio AS (2021). Stress appraisal prospectively predicts binge eating through increases in negative affect. Eating and Weight Disorders-Studies on Anorexia, Bulimia and Obesity, 1–8. [DOI] [PubMed] [Google Scholar]
  35. Steward T, Menchon JM, Jiménez-Murcia S, Soriano-Mas C, & Fernandez-Aranda F (2018). Neural network alterations across eating disorders: a narrative review of fMRI studies. Current neuropharmacology, 16(8), 1150–1163. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Stice E, Nemeroff C, & Shaw HE (1996). Test of the dual pathway model of bulimia nervosa: Evidence for dietary restraint and affect regulation mechanisms. Journal of Social and Clinical Psychology, 15(3), 340–363. [Google Scholar]
  37. Ward ZJ, Rodriguez P, Wright DR, Austin SB, & Long MW (2019). Estimation of eating disorders prevalence by age and associations with mortality in a simulated nationally representative US cohort. JAMA network open, 2(10), e1912925–e1912925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Wonderlich JA, Lavender JM, Wonderlich SA, Peterson CB, Crow SJ, Engel SG, . . . Crosby RD (2015). Examining convergence of retrospective and ecological momentary assessment measures of negative affect and eating disorder behaviors. International Journal of Eating Disorders, 48(3), 305–311. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES