Mindfulness and acceptance-based behavioral treatment for bulimia-spectrum disorders: A pilot feasibility randomized trial

Abstract

Objective:

Although existing research supports the efficacy of mindfulness- and acceptance-based treatments (MABTs) for eating disorders (EDs), few studies have directly compared outcomes from MABTs to standard CBT.

Method:

Participants (N = 44), treatment-seeking adults with bulimia-spectrum EDs, were screened for eligibility, consented, and randomized to receive 20 sessions of outpatient, individual CBT or MABT treatment. Treatment outcomes (binge eating and compensatory behavior episodes, global ED severity, depressive symptoms, quality of life, emotional awareness/clarity, distress tolerance, values-based decision-making, and emotion modulation) were measured at pre-treatment, post-treatment, and 6-month follow up. Data on feasibility and acceptability are also presented.

Results:

Treatment and assessment retention rates were comparable between MABT and CBT (p range = .51–.73) and between-group differences on acceptability measures were very small (d range = 0.03–0.19). Both conditions produced notable and generally comparable changes in most treatment outcomes at post-treatment (within group d range = 0.06–1.77).

Discussion:

The MABT and CBT conditions demonstrated comparable degrees of feasibility, acceptability, and symptom improvement, suggesting that MABTs warrant further evaluation as ED treatments.

1 |. INTRODUCTION

Mindfulness- and acceptance-based treatments (MABTs) such as Acceptance and Commitment Therapy (ACT), and Dialectical Behavior Therapy (DBT) have become a commonly used treatment technique for eating disorders (EDs). MABTs integrate traditional behavioral therapies with psychological strategies such as distress tolerance, mindful awareness, emotion modulation, and values-based decision making to facilitate behavior change (Baumeister, Heatherton, & Tice, 1994; Forman & Butryn, 2015; Shapiro, Carlson, Astin, & Freedman, 2006; Zimmerman, 2000). MABTs have quickly become one of the most frequently used treatments in clinical practice for EDs (Linardon, Fairburn, Fitzsimmons-Craft, Wilfley, & Brennan, 2017) and preliminary evidence supports the efficacy of MABTs for EDs like bulimia nervosa (BN) and binge eating disorder (BED) (Godfrey, Gallo, & Afari, 2015; Katterman, Kleinman, Hood, Nackers, & Corsica, 2014; Linardon, Fairburn, et al., 2017), with recent meta-analyses showing moderate-to-large improvements in primary outcomes (i.e., global eating pathology, remission, and binge eating frequency; (Linardon, Fairburn, et al., 2017). However, Cognitive Behavioral Therapy (CBT) remains the recommended first-line treatment for BN (American Psychiatric Association, 2006). Further research is needed on MABTs, including to direct comparison of outcomes from MABTs versus CBT.

To date, the vast majority of MABT trials for EDs are either open trials or small randomized controlled trials (RCTs) that compared an MABT condition to a wait-list control, treatment-as-usual, or a minimally active treatment condition. While a small number of trials have shown comparable efficacy for an MABT when compared directly to CBT (Chen et al., 2017; Navarro-Haro et al., 2018; Wonderlich et al., 2014), they were limited by: (a) use of a non-standard length/dose of CBT (e.g., guided self-help CBT, partial hospitalization format; (Chen et al., 2017; Navarro-Haro et al., 2018; Peterson et al., 2020); (b) sample specific characteristics (e.g., including only individuals who also had co-morbid borderline personality disorder; (Navarro-Haro et al., 2018); and/or (c) use of an uncommon form of MABT (e.g., integrative cognitive affect therapy; (Peterson et al., 2020; Wonderlich et al., 2014). To the best of our knowledge, no study has compared an individual, outpatient MABT based on the most widely used MABT treatments (e.g., ACT and DBT), to individual, outpatient CBT.

In the current study, we conducted a pilot RCT that randomized individuals with BN-spectrum disorders (N = 44) to either receive 20 individual, outpatient treatment sessions of MABT that integrated core behavioral treatment skills (e.g., regular eating, reducing dietary restraint) with ACT and DBT principles or 20 individual, outpatient treatment sessions of CBT. MABT retained the key behavioral elements of CBT due to robust evidence that these components are necessary foundations for success, and because MABT skills are aimed at facilitating engagement with these behavioral skills. The goal of this pilot study was to evaluate each condition's feasibility (Primary Aim 1), acceptability (Primary Aim 2), and target engagement (i.e., the degree to which MABT produces change in its putative mechanistic targets; Primary Aim 3), and to obtain preliminary estimates of efficacy in reducing BN symptoms (Secondary Aim 1) and other outcomes such as depression, clinician rated global impairment, and quality of life (Secondary Aim 2) to inform the design of a subsequent, fully powered RCT.

2 |. METHOD

2.1 |. Participants

Adults (N = 44) were recruited from the greater Philadelphia area. Inclusion criteria were meeting DSM-5 criteria for BN or for BN with subjectively-large binge episodes (i.e., all participants experienced at least 12 objectively- or subjectively-large binge eating episodes and 12 episodes of inappropriate compensatory behaviors over the previous 3 months) and age 18 years or older. Exclusion criteria were: medical complications prohibiting outpatient treatment; severe comorbid psychiatric (e.g., psychosis, substance dependence) or intellectual/developmental disorder; current/planned pregnancy; unstable psychiatric medication; history of bariatric surgery; other current eating disorder treatment. Most participants (n = 37, 84.1%) met behavioral criteria for BN; the remaining participants (n = 7, 15.9%) met criteria for Other Specified Feeding or Eating Disorder, BN. Sample characteristics are in Table 1.

TABLE 1.

Participant characteristics, feasibility and acceptability, and treatment outcomes by treatment condition

	CBT (n = 18)		MABT (n = 26)		Comparisons
Participant characteristics	M, SD	Range	M, SD	Range	Group differences (t, p)
Age	35.22	18–60	29.77	18–57	1.43, .16
BMI	27.54, 6.23	17.90–41.30	25.03, 5.70	18.20–38.90	1.38, .17
	n, %	n, %	χ2, p
Gender			.002, .97
Female	16, 88.9%	23, 88.5%
Male	2, 11.1%	3, 11.5%
Race			0.57, .75
White	16, 88.9%	21, 80.8%
Black	1, 5.6%	3, 11.5%
Asian	1, 5.6%	2, 7.7%
Ethnicity			5.17, .02
Non-Hispanic	13, 72.2%	25, 96.2%
Hispanic	5, 27.8%	1, 3.8%
Treatment feasibility and acceptability	Pre	Post	FU	Pre	Post	FU	Group differences (Pre-Post)
Feasibility		n, %			n, %		χ2, p
Drop out from treatment	–	8, 44.4%	–	–	10, 38.5%	–	0.43, .51
Assessment retention	18, 100%	12, 66.7%	9, 50.0%	26, 100%	16, 61.5%	13, 50.0%	0.12, .73†
Acceptability		M, SD				M, SD		Cohen’s d ‡
								[95% CI]
Credibility and expectancya	–	7.17, 1.04	–	–	–	7.14, 1.18	–	0.03 [−0.629, 0.68]
Working allianceb	–	3.06, 0.29	–	–	–	3.01, 0.25	–	0.19 [−0.47, 0.85]
Treatment outcomes	Pre	Post	FU	Pre-Post	Pre-FU	Pre	Post	FU	Pre-Post	Pre-FU
	M, SD	M, SD	M, SD	Cohen’s d §	Cohen’s d §	M, SD	M, SD	M, SD	Cohen’s d §	Cohen’s d §
				[95% CI]	[95% CI]				[95% CI]	[95% CI]
LOC episodesg,¶	27.44, 19.03	3.94, 4.32	2.41, 3.06	1.21 [0.64, 1.78]	1.29 [0.69, 1.89]	27.27, 17.65	3.84, 6.81	2.96, 3.31	1.77 [0.87, 2.66]	2.09 [1.05, 3.12]
Compensatory behaviorsg,¶	35.83, 29.08	7.27, 9.43	6.42, 9.51	1.26 [0.54, 1.99]	1.25 [0.56, 1.93]	31.12, 20.79	5.05, 5.76	5.21, 5.81	1.68 [0.83, 2.53]	1.69 [0.83, 2.55]
Global eating disorder severityh,¶	3.35, 1.11	1.79, 0.96	1.82, 0.88	1.51 [0.72, 2.29]	1.53 [0.71, 2.35]	3.28, 1.13	1.49, 0.96	1.68, 0.951	1.69 [0.89, 2.50]	1.53 [0.71, 2.34]
Depressive symptomsi,¶	20.72, 11.87	14.60, 11.16	12.82, 9.32	0.53 [0.10, 0.96]	0.72 [0.26, 1.18]	22.34, 9.47	10.11, 8.48	10.43, 5.85	1.36 [0.62, 2.11]	1.50 [0.72, 2.27]
Quality of lifej	15.46, 1.96	16.52, 1.69	17.02, 2.23	0.58 [0.01, 1.15]	0.75 [0.07, 1.43]	15.31, 1.74	17.13, 1.283	17.15, 2.063	1.17 [0.50, 1.85]	0.94 [0.32, 1.57]
Global impairmentk,¶	4.31, 0.87	3, 1.23	–	1.21 [0.55, 1.88]		4.60, 1.26	2.88, 1.42	–	1.28 [0.59, 1.97]	–
Emotional awareness
Emotional awarenessc,¶	15.56, 5.03	13.18, 4.25	13.25, 4.01	0.50 [0.13, 0.87]	0.50 [0.03, 0.96]	19.08, 5.21	12.80, 3.93	13.39, 3.81	1.38 [0.69, 2.06]	1.23 [0.57, 1.88]
Emotional clarityc,¶	12.67, 4.99	10.27, 3.29	11.23, 3.18	0.54 [0.08, 0.99]	0.33 [−0.18, 0.84]	14.54, 4.02	10.00, 2.99	10.66, 3.01	1.28 [0.61, 1.95]	1.08 [0.43, 1.72]
Distress tolerance
Distress toleranced	11.16, 3.73	12.64, 4.12	12.63, 3.04	0.30 [−0.21, 0.97]	0.31 [0.10, 1.39]	10.90, 4.24	13.81, 3.53	12.94, 3.21	0.58 [−0.24, 1.10]	0.41 [−0.10, 1.18]
Acceptance and actione,¶	32.17, 9.15	24.61, 10.70	21.19, 4.50	0.76 [0.17, 1.35]	1.47 [0.72, 2.21]	29.35, 10.60	21.08, 8.20	20.54, 6.23	0.87 [0.17, 1.58]	1.01 [0.29, 1.73]
Values-based decision making
lmportancef	62.22, 11.89	57.08, 18.01	58.81, 7.89	0.34 [−0.29, 0.96]	0.34 [−0.30, 0.97]	63.12, 14.53	62.31, 10.41	60.11, 9.39	0.06 [−0.45, 0.58]	0.24 [−0.24, 0.71]
Consistencyf	52.94, 16.945	48.52, 17.98	53.06, 6.99	0.25 [−0.37, 0.88]	0.06 [−0.53, 0.66]	49.42, 15.62	50.85, 10.65	55.27, 11.76	0.11 [−0.45, 0.66]	0.42 [−0.11, 0.95]
Emotion modulation
Goalsc,¶	16.89, 6.55	14.96, 4.77	13.30, 3.35	0.33 [−0.16, 0.82]	0.63 [0.44, 1.76]	16.58, 4.73	11.89, 3.52	12.68, 4.069	1.10 [0.12, 1.15]	0.88 [0.21, 1.56]
Strategiesc,¶	24.06, 7.26	19.55, 6.85	17.10, 5.30	0.64 [0.20, 1.07]	1.07 [0.40, 1.57]	22.92, 8.56	15.48, 5.17	15.56, 6.13	0.98 [0.47, 1.67]	0.97 [0.38, 1.56]

Abbreviations: BMI, body mass index; CBT-E, Cognitive behavioral therapy for eating disorders; CI, confidence interval; FU, 6-months after post-treatment; M, mean; MABT, mindfulness and acceptance-based treatment; pre, pre-treatment; post, immediately post treatment; SD, standard deviation.

^†Group differences in assessment retention were only provided for the post treatment assessment since the percentage of participants that completed baseline and follow up assessments was identical for both groups.

^‡Between-subjects Cohen’s d are presented.

^§Repeated measures Cohen’s d are presented.

^¶Measures in which a decrease in scores indicates an improvement.

^aMeasured by the Credibility and Expectancy Questionnaire; CBT group n = 16, MABT group n = 22.

^bMeasured by the patient’s last obtained rating of the therapeutic relationship as measured by the Working Alliance Inventory (WAI); ABT group n = 22.

^cMeasured by the Difficulties Regulating Emotion Scale (DERS).

^dMeasured by the Distress Tolerance Scale (DTS).

^eMeasured by the Acceptance and Action Questionnaire (AAQ).

^fMeasured by the Valued Living Questionnaire (VLQ).

^gFrequency of behaviors in the prior month as measured by the Eating Disorder Examination (EDE).

^hMeasured by the EDE global score.

ⁱMeasured by the Beck Depression Inventory–II (BDI-II).

^jMeasured by the Quality of Life Inventory (QoLI).

^kMeasured by therapist’s ratings of the Clinical Global Impressions Scale (CGI). Cronbach’s alpha for all measures at all timepoints were acceptable or excellent (αs = .62 – .96).

2.2 |. Procedures

Potential participants completed a phone screen. All eligible participants provided informed consent and all study procedures were approved and overseen by the Drexel University Institutional Review Board. Enrolled participants were randomly assigned to treatment conditions, stratified by global eating pathology, age, and sex. We intentionally over-recruited for the MABT condition to better test the feasibility and acceptability of this treatment approach, yielding 60% allocation to MABT and 40% allocation to CBT. Research assessments occurred at pre-treatment, post-treatment, and at 6-month follow-up.

2.3 |. Intervention

Participants received 20 individual outpatient treatment sessions over 20 weeks. Sessions 1–8 occurred twice weekly, sessions 9–17 once per week, and sessions 18–20 once every 2 weeks.

2.3.1 |. Cognitive behavioral therapy (CBT)

The CBT condition was based on the focused version of CBT-E (Fairburn, 2008). The main components of the focused version of CBT-E include self-monitoring, normalizing eating patterns, reducing dietary restraint, reducing compensatory behaviors, urge management, and decreasing over-evaluation of shape and weight. We elected not to use the broad version of CBT-E with additional modules (e.g., mood intolerance, perfectionism) in order to be able to evaluate whether MABT skills enhance the “base” skills provided in the focused version of CBT-E and to avoid conflict with certain overlapping treatment recommendations in the mood-intolerance module specifically.

2.3.2 |. Mindfulness and acceptance-based treatment (MABT)

The MABT condition retained the core behavioral skills used in CBT but added in mindfulness and acceptance-based treatment components with a specific focus on how these MABT components can be used to facilitate behavior change. MABT components were adapted from DBT and ACT and included content focused on distress tolerance (i.e., willingness to experience aversive internal experiences, such as negative thoughts, emotions, and uncomfortable physical experiences), emotion modulation (i.e., the ability to use healthy coping strategies to down-regulate negative emotions), mindful awareness (i.e., the ability to notice, observe, and label internal experiences, with a particular emphasis on emotional awareness and labeling within this treatment), and values-based decision making (i.e., identifying and clarifying one's life values and making behavioral decisions consistent with one's values). Table S1 describes the content of the CBT and MABT treatment conditions.

2.3.3 |. Therapists

Treatment was delivered by masters- and doctoral-level clinicians. Clinicians received weekly supervision from a licensed clinical psychologist.

2.4 |. Measures

2.4.1 |. Feasibility and acceptability

Feasibility of each condition was assessed using percent retention in study assessments and percent dropout from treatment. Acceptability was measured using the Credibility/Expectancy Questionnaire (Borkovec & Nau, 1972) and end-of-treatment working alliance (Working Alliance Inventory; (Horvath & Greenberg, 1989)).

2.4.2 |. Treatment outcomes

For Primary Aim 3 (target engagement), we evaluated the degree to which MABT produced change in its putative mechanistic variables relative to CBT. Emotional awareness/clarity was measured using the Awareness and Clarity subscales of the Difficulties in Emotion Regulation Scale (DERS), a 36-item self-report measure examining problems with emotion regulation (Gratz & Roemer, 2004). Distress tolerance was measured using the total scores of the Distress Tolerance Scale (DTS; (Simons & Gaher, 2005)) and the Acceptance and Action Questionnaire-II (AAQ-II; (Bond et al., 2011). Values-based decision making was measured using the Current Importance and Consistency subscales of the Valued Living Questionnaire II (VLQ-II; (Wilson, Sandoz, Kitchens, & Roberts, 2010). Emotion modulation was measured using the Strategies and Goals subscales of the DERS (Gratz & Roemer, 2004).

Eating disorder treatment outcomes (Secondary Aim 1) were measured through reductions in loss of control episodes, compensatory behaviors, and global eating disorder severity as measured by the Eating Disorder Examination (Fairburn, Cooper, & O'Connor, 2014). Secondary outcomes assessed (Secondary Aim 2) include depression symptoms (Beck Depression Inventory; (Beck, Steer, Ball, & Ranieri, 1996), overall quality of life (Quality of Life Inventory; (Frisch et al., 2005)), and clinician-rated global impairment (Clinician Global Impressions Scale; (Busner & Targum, 2007)).

2.5 |. Data analysis plan

Data were missing at random, therefore the Markov chain Monte Carlo (MCMC) method was used to compute five iterations of imputed data. Consistent with journal guidelines for pilot clinical trials and based on recommendations from Leon, Davis, and Kraemer (2011), inferential statistics comparing groups were not conducted for target engagement and outcome variables (Leon et al., 2011). Instead, within-condition effect sizes and confidence intervals are reported. To assess differences in treatment feasibility and acceptability, between group Cohen's d were computed (equation (1); Lakens, 2013). To determine within-condition changes from pre-to post-treatment and pre-treatment to follow-up, repeated measures Cohen's d was calculated separately for each condition (equation (9); Lakens, 2013). Ninety-five percent of confidence intervals for all effect sizes were also calculated (Nakagawa & Cuthill, 2007).

3 |. RESULTS

Means, standard deviations, and effect size and confidence intervals (when relevant) are presented in Table 1.

3.1 |. Feasibility and acceptability

Drop-out from treatment and assessment retention rates were comparable in both conditions. Figure 1 displays a flow diagram of participant retention, including reasons for loss to follow-up or treatment discontinuation. Conditions did not differ in treatment dropout or assessment retention. Ratings of credibility and expectancy and working alliance were similar in both conditions and between-group effect sizes were very small.

FIGURE 1.

Flow diagram of participant retention

3.2 |. Target engagement

Changes in the majority of target engagement variables were evident in both treatment conditions. Within-group change effect sizes indicated large increases in emotional awareness in MABT and small-to-medium increases in CBT conditions. Both conditions had small-to-large increase in distress tolerance, and small increase in values-based decision making. MABT had consistently large improvements in emotion modulation, whereas CBT had small-to-large increases.

3.3 |. Preliminary outcomes

Both treatment groups produced large changes in nearly all outcome variables at post-treatment, which were largely maintained at follow-up. Exceptions included depressive symptoms and quality of life changes in CBT, for which within-group changes were medium rather than large.

4 |. DISCUSSION

While there is growing implementation of MABTs for BN spectrum eating disorders in clinical practice, there remains a limited number of trials that directly compare an MABT to CBT (Linardon, Fairburn, et al., 2017). To our knowledge, this is the first study to evaluate an individual outpatient MABT treatment based on ACT and DBT to CBT for BN spectrum disorders. The MABT tested in this trial showed preliminary evidence of feasibility, acceptability, target engagement, and symptom improvement with results broadly comparable to those in CBT. Treatment drop-out rate was comparable across conditions. Therapeutic alliance and treatment credibility were also both high and similar between conditions. These results indicate that the MABT we tested had adequate feasibility and acceptability to support further testing of this treatment in future clinical trials.

Participants in both the MABT and CBT conditions showed large reductions in symptoms that were sustained through the 6-month follow-up with effect sizes similar to those reported for previous CBT outcomes for BN (Hay, Bacaltchuk, Stefano, & Kashyap, 2009; Linardon, Wade, de la Piedad Garcia, & Brennan, 2017) and other MABT studies for EDs (Linardon, Fairburn, et al., 2017). Improvement in secondary outcome variables for both conditions were also large. Participants in the MABT condition experienced the largest improvements in secondary outcome variables such as depression and quality of life. Average depression severity scores on the BDI in both groups were in the moderate range pre-treatment, average scores post-treatment were in the minimal range in the MABT group and in the mild range for the CBT group. Of note, we implemented the focused version of CBT-E in the current trial, which does not contain the Mood Intolerance module in CBT-E (Fairburn, 2008; Fairburn et al., 2009). It is possible that a comparison of CBT-E that includes this module (which draws strategies from MABTs) would lead to superior improvements in depression compared to the focused version of CBT-E.

Lastly, we saw some indication that although MABT constructs broadly improved in both conditions, there was some evidence of superior target engagement in measures of emotional awareness and emotional clarity in the MABT condition. In MABT, strategies were used throughout treatment to increase ability to label internal experiences and choose how to respond to them (e.g., identify the feeling of boredom and choose to engage in an activity consistent with values instead of binge eating). These skills could be particularly important for some patients in practice who have difficulty identifying urges and emotions, or have co-occurring mood symptoms like depression that interact with their ED symptoms. In CBT-E, emotion-based skills are not introduced until the latter part of treatment and for some patients, earlier adoption of emotion-based skills could be useful to augment the effects of behavioral skills that may be distressing such a regular eating. However, further research is needed to identify if MABT-based skills like emotional awareness augment the effect of CBT.

There were several limitations in the current study that should be considered. As this was a pilot trial, our sample was small, precluding the ability to do between group inferential statistics and introducing the possibility of sample specific characteristics. Effect sizes are often unstable in pilot trials and must be interpreted with caution (Leon et al., 2011). Measures of target engagement were primarily self-report and real-time data collection (e.g., ecological momentary assessment) or behavioral measures (e.g., in-lab food paradigms) may have been better able to elucidate changes in processes targeted by acceptance-based versus traditional CBT techniques for BN. Additionally, we did not include measures of treatment fidelity, which could have differed between conditions or between master's and doctoral level clinicians.

Based on this pilot RCT, MABT should be considered as an alternative behavioral treatment for BN and is worthy of future study in a fully powered clinical trial. MABT was just as feasible and acceptable to CBT. MABT skills, such as those to promote emotional awareness, may be particularly beneficial for some patients to augment the effect of traditional CBT skills. Although further research is needed to understand the mechanisms of action of MABT versus CBT for BN, this study provides preliminary support for the use of outpatient MABT for BN.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.