Psychotherapy and Medications for Eating Disorders: Better Together?

Abstract

Purpose:

Eating disorders are prevalent public health problems associated with broad psychosocial impairments and with elevated rates of psychiatric and medical comorbidities. Critical reviews of the treatment literature for eating disorders indicate that certain specialized psychological treatments and specific medications demonstrate efficacy to varying degrees across the different eating disorders but that many patients fail to derive sufficient benefit from existing treatments. This article addresses whether combining psychological and pharmacological interventions confers any additional benefits for treating eating disorders.

Methods:

Critical review of randomized controlled trials (RCT) testing combined psychological and pharmacological treatment approaches for eating disorders with a focus on anorexia nervosa (AN), bulimia nervosa (BN), and binge-eating disorder (BED).

Findings:

For AN, 3 of the 4 RCTs reported no significant advantage for combining treatments; the fourth reported a statistically significant albeit clinically modest advantage. For BN, 10 of the 12 RCTs reported no significant advantage for combining treatments; 2 RCTs found that combining fluoxetine with specific psychological treatments enhanced outcomes relative to medication-only but not relative to the psychological treatments only. For BED, of the 12 RCTs, only 2 (both with antiseizure medications) significantly enhanced both binge-eating and weight outcomes, and only 2 (with orlistat, a weight-loss medication) enhanced weight loss but not binge-eating outcomes.

Implications:

Despite the public health significance of eating disorders, the scope of research performed on the utility of combining treatments is limited. To date, the few RCTs testing combined pharmacologic plus psychological treatments for eating disorders have yielded mostly non-significant findings. Future RCTs should focus on testing additive benefits of medications with relevant mechanisms of action to available effective psychological interventions. In addition, future RCTs testing additive effects should employ adaptive designs that could inform treatment algorithms to enhance outcomes among both responders and non-responders to initial interventions.

Introduction

Eating disorders, which comprise the formal diagnoses of anorexia nervosa (AN), bulimia nervosa (BN), and binge-eating disorder (BED) in addition to atypical or not-otherwise-specified categories, were recently updated by the American Psychiatric Association in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [1] and by the World Health Organization in the 11^th edition of the International Classification of Diseases and Related Health Problems (ICD-11) [2]. Eating disorders are prevalent [3], associated significantly with psychiatric and medical comorbidities [4] and with broad psychosocial impairments including suicidality [5]. Despite the availability of psychological and pharmacological interventions supported to varying degrees by controlled research, naturalistic treatment-seeking by persons with eating disorders is low and the types of help received may not be evidence-based [6].

Critical reviews of the treatment literature for eating disorders have generally converged in supporting certain specialized psychological treatments and specific medications to varying degrees across the different eating disorders [7–9]. The journal’s special issue includes updated critical reviews of psychological and pharmacological treatment literatures specifically targeting AN, and BN, and BED. Our review focuses on a different question which concerns whether combining psychological and pharmacological interventions confers any additional benefits for treating eating disorders. Clinical lore and “expert opinion,” for example, in the American Psychiatric Association’s Practice Guidelines [10], suggests that more intensive and combined treatment approaches might enhance outcomes for complex patients such as those with comorbidities. This clinical question concerning combining treatments has received remarkably little attention [11] and a critical synthesis of the relevant controlled treatment research is needed to inform clinical practice for eating disorders. Thus, we aim to provide a brief state-of-the-art review of randomized controlled treatments (RCTs) for combined psychological and pharmacological treatment of AN, BN, and BED.

Methods

An electronic search was performed for all English-language articles using MEDLINE via PubMed and the Cochrane Library published through August 1, 2020. Registered and ongoing clinical trials for eating disorders were identified via the United States National Institutes of Health web-based registry of private and publicly-supported clinical studies (ClinicalTrials.gov). All randomized controlled trials (RCTs) investigating the effects pharmacotherapy in combination with psychological interventions were considered. Exclusion criteria were: 1) phase I trials (open-label, case series, retrospective cohort designs); 2) non-English language; 3) trials involving individuals under the age of 18 years of age; 4) studies pertaining to non-purging bulimia nervosa, mixed eating disorder samples, or atypical eating disorders, or the ICD-10 category overeating associated with psychological disturbances; and 4) RCTs investigating either monotherapy psychopharmacological treatment or psychological treatment delivered alone (which are reviewed elsewhere in this journal’s special issue).

Results

Characteristics of Studies with Anorexia Nervosa

Although several medications have been tested as part of a multi-disciplinary approach within the context of day-hospital or in-patient treatment programs, few studies were randomized, placebo-controlled, double-blind trials either sufficiently powered or designed to allow for conclusions regarding the effects of pharmacotherapy in combination with a specific psychological-behavioral intervention. Four studies using placebo-controlled designs were identified and included in this review, including two RCTs studying the additive effects of fluoxetine [12, 13] and two RCTs investigating the atypical antipsychotic olanzapine [14, 15]. Fluoxetine has been investigated within the context of a structured behavioral inpatient treatment for AN [12] and within an outpatient setting, in which 12 months of cognitive-behavioral therapy (CBT) was provided with either fluoxetine or placebo following weight restoration [13]. One of the olanzapine RCTs tested intensive day-hospital treatment plus 10 weeks of either olanzapine or placebo [14], the other RCT tested 12 weeks of cognitive-behavioral therapy (CBT) plus either olanzapine or placebo on an outpatient basis [15]. With the exception of the multi-site study by Walsh et al. [13], which included 93 participants, the trials were small, with sample sizes of 30 [15], 31 [12], and 34 [14] participants; three of the four trials were short in duration, ranging from 7 weeks to 12 weeks, and none reported follow-up data. Outcome measures for three RCTs included weight gain and changes in eating-disorder psychopathology and psychological functioning [12, 14, 15]; in the Walsh et al. [13] RCT, the primary outcomes were time-to-relapse and the rate of study completion after having achieved weight restoration.

Characteristics of Studies with Bulimia Nervosa

Table 1 lists 12 published RCTs that tested combination treatments for BN. Most studies examined the addition of pharmacotherapy to CBT delivered in various formats, including individual [16–21], pure self-help [22], guided self-help [23] or group [24] methods. A few studies tested the addition of pharmacotherapy to other forms of “talk” treatments, including nutritional counselling [25], a supportive psycho-dynamically-oriented psychotherapy [21], and an intensive, eclectic inpatient psychotherapy program [26]. Early studies tested tricyclic antidepressants including desipramine [16, 19, 21] and imipramine [27] and later studies focused on fluoxetine [18, 20, 22, 24, 26]. One trial compared the relative efficacy of adding d-fenfluramine, an appetite suppressant, versus placebo to CBT [17]. Two studies used more complex adaptive-type of designs: one study employed a stepped care approach that stratified patients based on initial response to treatment [20] and a second study employed a sequential approach involving a two-stage medication intervention (switching to fluoxetine if desipramine was ineffective or poorly tolerated)[21]. The RCTs ranged from 7 to 20 weeks; two trials reported 6-month follow-up [19, 24] and one trial reported 12-month follow-up [20] data. Sample sizes ranged from 21 [19] to 293 participants [20], with 8 of the 11 trials including fewer than 20 participants per treatment arm. The RCTs included only females except for one study which enrolled one male [26], and all were conducted on an outpatient basis except for one [26].

Table 1.

Study characteristics RCTS investigating combined psychotherapy and pharmacotherapy for BN

Study	N	% F	% W	Age	BMI	Trial (wk)	FUP (mo)	Treatment conditions	Study Design
Agras et al. (1992)[16]	71	100	100	29.6	NA	16	1	SSRI 16 wk SSRI 24 wk CBT + SSRI 16 wk CBT + SSRI 24 wk CBT 24 wk	Flexible dosage desipramine, 16 wk Flexible dosage desipramine, 24 wk Individual CBT delivered in 50-min sessions + SSRI, 16 wk Individual CBT delivered in 50-min sessions + SSRI, 24 wk Individual CBT, 24 wk
Beumont et al. (1997)[25]	67	100	NA	24.2 25.1	22.2 22.2	8	3	Nutritional counselling + placebo Nutritional counselling + SSRI	Intensive nutritional counselling, weekly individual sessions + placebo Intensive nutritional counseling + fluoxetine (60 mg/day)
Fahy et al. (1993)[17]	43	100	NA	23.0 25.0	22.0 22.9	8	2	CBT + d-fenfluramine CBT + placebo	CBT+fixed dose 45 mg/day d-fenfluramine CBT, individual for 8 wks + placebo
Fichter et al. (1991)[26]	40	97.5	NA	26.5 24.6	54.7 kg 56.7 kg	7	---	Inpt tx + SSRI Inpt tx + placebo	Inpatient behavioral psychotherapy + fluoxetine 60 mg/day Inpatient behavioral psychotherapy + placebo
Goldbloom et al. (1997)[18]	76	100	NA	25.8	23.0	16	--	CBT SSRI CBT + SSRI	CBT, individual Fluoxetine fixed dose 60 mg/day FL-CBT *groups stratified by high or low frequency of vomiting and history of AN
Jacobi et al. (2002)[24]	53	100	NA	26.0	20.6	16	6	CBT SSRI CBT + SSRI	CBT, group, 20 sessions, 16 wks Fluoxetine 60 mg/day, 16 wks CBT + Fluoxetine
Leitenberg et al. (1994)[19]	21	100	NA	26.7	NA	20	6	CBT TCA CBT + TCA	CBT 22 individual sessions over 20 wks Desipramine 150 mg/day CBT + desipramine
Mitchell et al. (1990)[27]	171	100	NA	24.4 24.1 22.8 24.3	NA	12	--	Placebo Imipramine Intensive group psychotherapy + PL Intensive group psychotherapy + Med	Placebo Imipramine, up to 300 mg/day Intensive group psychotherapy. Highly structured into 3 phases (preparatory, interruption, stabilization) consisting of psychoeducation, meal planning, cognitive behavioral skills, group meals, relapse Intensive group psychotherapy + imipramine
Mitchell et al. (2001)[22]	91	100	96.7	26.6 26.8 29.3 23.8	NA	16	---	SSRI CBTsh + Placebo CBTsh + SSRI Placebo	Fluoxetine, fixed dose 60 mg/day Self-help CBT + placebo Self-help CBT + fluoxetine, 60 mg/day Placebo
Mitchell et al. (2011) [20]	293	100	84 88	29.5 29.8	23.4 23.5	18	12	CBT Stepped care	CBT, 20 sessions of 50 min, + fluoxetine 60 mg/day offered to those failing to reduce purging by 70% after session 6 (wk 4). If taken, meds continued until 1-year FUP. Therapist assisted self-help, manual + eight 20-minute in-person sessions. Non-responders at session 6 (wk 10) were offered fluoxetine 60 mg/day. At the end of treatment (wk 18), CBT for was offered to those who did not achieve abstinence. Meds continued until 1-year FUP.
Walsh et al. (1997)[21]	120	100	83	26.1 25.8 28.0 26.9 24.3	21.9	16	---	CBT-SSRI CBT-placebo SPT-SSRI SPT-Placebo SSRI	Individual CBT, 20 sessions + desipramine 200–300 mg/day for 8 wks, followed by fluoxetine (60 mg/day) if no improvement CBT + placebo Supportive therapy, 20 sessions + desipramine 200–300 mg/day for 8 wks, followed by fluoxetine (60 mg/day) if no improvement Supportive therapy, 20 sessions + placebo Medication only
Walsh et al. (2004)[23]	91	100	92.3	30.6	NA NA 21.79 22.78	16	---	Placebo Fluoxetine alone Placebo + gSH Med + gSH	Placebo Fluoxetine 60 mg/day gSH manual + 6–8 30-min sessions w/nurse Fluoxetine 60 mg/day + gSH manual +visits

Notes: BN = bulimia nervosa; FUP = follow-up; RCT = randomized controlled trial; SSRI = selective serotonin reuptake inhibitor; CBT = cognitive-behavior therapy; gSH= guided self-help; DSM = Diagnostic and Statistical Manual of Mental Disorders; SPT = supportive, psychodynamically-oriented therapy; TCA = tricyclic antidepressant; BE = binge eating; NA = data not available.

Characteristics of Studies with Binge-Eating Disorder

Table 2 lists 12 published RCTs that tested combination treatments for BED; there are also two ongoing registered clinical trials testing the effectiveness of combination approaches for patients with BED comorbid with obesity (NCT03924193, NCT03045341). The 12 RCTs testing combination treatments for BED ranged in treatment duration from 8 to 36 weeks; all but two studies [28, 29] reported follow-up data which ranged 3 to 24 months after completing treatments. Sample sizes ranged from 52 to 191 participants. Participants were primarily white women, with an average age of approximately 40 years and body mass index (BMI) of 38. Designs and medications tested included unblinded additions of several medications including desipramine [30], imipramine [31], fluoxetine and fluvoxamine [32], and zonisamide [33], and double-blind placebo-controlled addition of various SSRIs, anti-epileptic, and weight loss medications, specifically: fluoxetine [34, 35], orlistat [29, 36–38], topiramate [28], and sibutramine [38, 39] to either CBT or BWL methods.

Table 2.

Study characteristics RCTS investigating combined psychotherapy and pharmacotherapy for BED

Study	N	% F	% W	Age	BMI	Wks	FUP (mo)	Treatment conditions	Study Design
Agras et al. (1994)[30]	108	100	NA	45.0	38.6	36	3	BWL-only CBT/BWL CBT/BWL + desipramine	9 mo (30 sessions) 3 mo (12 sessions) + 6 mo (18 sessions) 3 mo + 6 mo + drug (mean dose 285mg/day)
Laederach-Hoffman et al. (1999) [31]	31	87	NA	35.7 40.7	39.5	8	6	Diet counseling + supportive therapy Imipramine	Bi-weekly (30 min) diet counseling + bi-weekly (15–25 min) supportive therapy + monthly sessions of group (90 min) behavior therapy Fixed dosage of 25 mg three times daily
Ricca et al. (2001) [32]	108	59	NA	25.9	32.3	24	12	CBT CBT + fluoxetine CBT + fluvoxamine Fluoxetine-only Fluvoxamine-only	22 individual sessions CBT + 60 mg/day CBT + 300 mg/day Drug-only, as “b” above Drug-only, as “c” above
Devlin et al. (2005, 2007) [34, 42]	116	78	77	43.0	40.9	20	24	Fluoxetine Placebo CBT + placebo CBT + fluoxetine *All of the above given in combo with group BWL (16 sessions for 5 mo)	Flexible dosage of 60 mg/day Placebo 20 individual CBT sessions As above, “a” + “c” Maintenance treatment included double blind medication for 18/24 months and monthly BWL groups
Golay et al. (2005) [29]	89	91	97	41 41	35.7 37.3	24	None	Orlistat + hypo-caloric diet Placebo + hypo-caloric diet	Fixed dosage of 120 mg three times daily + hypocaloric diet (600 kcal/day subtracted from daily energy expenditure) Placebo + hypocaloric diet
Grilo et al. (2005) [36]	50	88	88	47	36.0	12	3	Orlistat + CBTgsh Placebo + CBTgsh	Fixed dosage of 120 mg three times daily + CBTgsh (CBT self-help manual, plus six 15–20 min individual meetings over 12 wks) Placebo + CBTgsh
Grilo et al. (2005, 2012) [35, 43]	108	78	89	44	36.3	16	6, 12	Fluoxetine –only Placebo CBT + placebo CBT + fluoxetine	Fixed dosage of 60 mg/day Placebo 16 individual weekly (60 min) sessions As above, a + c
Claudino et al. (2007)[28]	73	96	58	35 41	37.4 37.4	21	None	CBT + placebo CBT + topiramate	19 group sessions of CBT CBT + flexible dosage of 200 mg/day
Ricca et al. (2009) [33]	52	83	NA	35 36	39.2 38.4	24	12	CBT CBT + zonisamide	22 individual sessions of CBT for 24 wks CBT + zonisamide 25 mg/day for 7 days, increased as tolerated by 50 mg/day every week to a max of 100 mg/day for BMI < 35 kg/m2 or 150 mg/day for BMI > 35 kg/m2
Grilo & White (2013) [37]	40 †	78	0	46 46	37.2 39.0	16	6	BWL + placebo BWL + orlistat	16 weekly sessions of a culturally-enhanced adaptation of the Diabetes Prevention Program Fixed dosage of 120 mg three times daily + BWL
Grilo et al. (2014) [39]	104	73	45	44	38.3	16	6,12	Sibutramine Placebo shCBT + sibutramine shCBT + placebo	Fixed dosage of 15 mg/day Placebo Self-help CBT + sibutramine (15 mg/day) Self-help CBT + placebo
Grilo et al. (2020) [38, 44]	191	71.2	78.5	48.4	39.0	24	6,12	Standard treatment Stepped care	BWL (1 mo) followed by stratification by response: Rapid respondersa: 1a) BWL + Medb (5 mo) 1b) BWL + Placebo (5 mo) Non-rapid responders: 2a) CBTgsh + Med (5 mo) 2b) CBTgsh + Placebo (5 mo) BWL (6 mos)

Note: BED = binge eating disorder; FUP = follow-up; RCT = randomized controlled trial; BWL = behavioral weight loss; CBT = cognitive-behavior therapy; gsh = guided self-help; LDX = lisdexamphetamine; MDD = major depressive disorder; GAD = generalized anxiety disorder; OCD = obsessive-compulsive disorder; NA = data not available.

^aRapid response defined as 65% reduction in binge eating after 1 month of treatment;

^bSibutramine was dosed at 15 mg/day until market withdrawal and replaced by Orlistat 120 mg/3x day.

^†An additional N=39 obese patients without BED were also randomized to the treatments. The reader is referred to the publication for findings which notably included BED status as a significant predictor and moderator of outcomes [37].

Findings of Combined Treatments for Anorexia Nervosa

Attia and colleagues [12] conducted a 7-week double-blind placebo-controlled RCT testing fluoxetine (60 mg/day) combined with a structured inpatient program for AN. Thirty-one patients were randomized after achieving medical stabilization and reaching 65% of ideal body weight. Attrition rates did not differ significantly (27% for fluoxetine and 25% for placebo) and drop-outs were attributable to premature discharge, worsening depression or severe anxiety, or side effects. Overall, analyses revealed significant improvements in the percentage of ideal body weight over time but no significant differences between treatments (increase from 72.5% to 87.0% with fluoxetine versus 71.8 to 87.4% with placebo). Similarly, significant improvements in measures of psychological functioning were observed over time, but fluoxetine did not confer significant additional benefit relative to placebo to the structured inpatient treatment.

Walsh et al. [13] conducted a two-site (Columbia, Toronto) randomized, double-blind, placebo-controlled trial to examine whether the addition of fluoxetine reduced relapse and enhanced outcomes during the year following successful weight restoration achieved during prior inpatient or intensive outpatient treatment. Ninety-three participants (with a minimum BMI of 19.0) received up to 12 months of outpatient individual CBT and were randomized to either fluoxetine (60 mg/day) or placebo. Exhaustive analyses revealed no significant differences between fluoxetine and placebo conditions in the time-to-relapse, rate of study completion, or the average number of treatment (CBT) sessions attended. At 52 weeks, 57% of the total 93 participants had terminated the study prematurely due to treatment failure, voluntary withdrawal, or staff-initiated withdrawal due to nonadherence or adverse effects. Regardless of the stringency of outcome criteria applied, similar proportions of the fluoxetine versus placebo groups had relapsed by 52 weeks (i.e., 29% vs 27% for the least conservative estimate and 57 vs 55% for the most conservative estimates, respectively). Similarly, none of the clinical outcome measures (including AN severity ratings, depression, and BMI) differed significantly by treatment condition. For instance, a similar proportion of patients in the fluoxetine (65%) versus placebo (57%) conditions met the modified Morgan-Russell criteria for fair or better outcome at termination. These findings represent strong evidence that fluoxetine fails to confer significant benefit over placebo when delivered in combination with CBT for AN following successful initial weight restoration achieved with intensive treatment.

Bissada et al. [14] conducted a double-blind, placebo-controlled, 10-week flexible dose RCT to test olanzapine versus placebo treatments in combination with an intensive day-hospital intervention (supervised meals plus group therapies four days/week) for AN. Thirty-four patients with AN were randomized and 28 (82.3%) completed treatments; olanzapine and placebo conditions did not differ in attrition. A significantly greater proportion of patients receiving olanzapine than placebo (87.5% versus 55.6%, respectively) achieved successful weight restoration criteria, and olanzapine was associated with a significantly more rapid achievement of target BMI (approximately two weeks) than placebo. Olanzapine was associated with significantly greater reduction in obsessional symptoms but not with improvements in compulsive symptoms, depressive symptoms, or anxiety relative to placebo.

Brambilla et al. [15] performed a 12-week multi-site, outpatient-based, placebo-controlled double-blind RCT to test weekly CBT plus either olanzapine or placebo. Twenty-five of the 30 patients completed the RCT. Overall, significant improvements in BMI were observed but no significant differences were observed between the olanzapine and placebo conditions. Similarly, overall improvements in eating-disorder psychopathology scores were observed at posttreatment that did not differ significantly between the olanzapine and placebo. Overall, depression scores improved significantly in both treatment conditions, with olanzapine showing a marginally significant advantage over placebo.

Findings of Combined Treatments for Bulimia Nervosa

Findings of RCTs testing combination treatments with tricyclic antidepressants, appetite suppressants, and SSRI antidepressants are presented in turn below.

Tricyclic antidepressants.

Existing evidence suggests no advantage for adding TCAs to individual CBT or intensive group psychotherapy. In one of the first additive trials for BN, Mitchell et al. [27] investigated imipramine in 171 patients with BN, randomly allocating participants into imipramine only, placebo, a combination of intensive group psychotherapy and imipramine, and intensive group therapy plus placebo. All three active treatment conditions improved binge eating and compensatory behaviors significantly more than the placebo condition; however, the addition of imipramine to group therapy showed no additional effect over adding placebo (except for some significant effects on reducing associated depression and anxiety symptoms).

Agras and colleagues [16], in a RCT with 71 patients with BN randomized to five conditions including a 16 and 24-week treatment duration, found that 16 weeks of individual CBT, and the combination of CBT plus desipramine, were superior to desipramine alone in reducing binge eating and purging. At 24 weeks of treatment, the combined treatment was superior, with binge abstinence rates of 70%, versus 55% and 42% for CBT-alone and medication, respectively. High rates of attrition and relapse affected those receiving medication after both 16 and 24 weeks, indicating the benefit of CBT to prevent relapse following discontinuation of medication.

Leitenberg and colleagues [19], in a RCT with 21 patients with BN, compared the effectiveness of CBT-alone, desipramine-alone, and a combination of CBT+desipramine. The authors suggested that CBT alone appeared more effective than desipramine and that the combination was not superior to CBT alone. Importantly, this study relied on partial analyses performed with just 21 patients because the trial was discontinued due to the high drop-out rate and the ineffectiveness of desipramine; between-group comparisons at post-treatment were not possible.

Appetite Suppressants.

Fahy et al. [17], in an 8-week RCT with 43 women with BN, failed to detect any significant differences between CBT + d-fenfluramine (45 mg/day) and CBT + placebo groups. Overall, a rapid reduction in symptoms occurred during the first four weeks of treatment, but analyses revealed no differences between treatment conditions to suggest any additive therapeutic effect for adding d-fenfluramine to CBT.

SSRI Antidepressants.

Fluoxetine is the sole medication approved for the treatment of BN by the US Food and Drug Administration (FDA). Although fluoxetine has demonstrated efficacy for BN [40, 41], the evidence base regarding the effectiveness of combining fluoxetine with CBT or other psychotherapies is less clear. Within an inpatient setting, Fichter et al. [26] found little benefit to adding fluoxetine to an inpatient behavioral psychotherapy program. Forty hospitalized patients with BN were randomized to fluoxetine or placebo in combination with the intensive inpatient care for a period of 5 weeks (plus a 2-week baseline and wash-out period). Overall, eating-disorder symptoms improved significantly but the fluoxetine and placebo conditions did not differ significantly, suggesting no additive benefit.

Goldbloom and colleagues [18], in a 16-week RCT, randomized a total of 76 women with BN to fluoxetine, CBT, or CBT plus fluoxetine. All three treatment conditions produced clinically-meaningful improvements in bulimia-related outcomes. Analyses comparing treatment conditions revealed that CBT plus fluoxetine was significantly superior to fluoxetine alone, and that CBT alone did not differ from CBT plus fluoxetine. Analyses were limited by a high dropout rate, especially in the combined condition (i.e., 43.8%).

Beumont et al. [25] tested an 8-week trial of intensive, individually-delivered nutritional counseling with and without fluoxetine (60 mg/day) in 67 women with BN. Overall, both treatment conditions were associated with significant acute improvements in binge eating and vomiting frequency, but the fluoxetine and placebo conditions did not differ significantly. Analyses of 20-week follow-up data after finishing/discontinuing the 8-week treatments revealed significantly different patterns: for the fluoxetine plus nutrition group, the proportion of binge-eating abstinent patients fell from 69.6% to 35.7%, whereas for the placebo + nutritional counselling, the proportion of abstinent patients remained fairly constant at roughly around 61%. Noteworthy is the high rate of attrition from the study (50% and 30% for fluoxetine and placebo, respectively) by the 20-week follow-up.

Walsh and colleagues [21], in a complex RCT design, investigated: (1) how a psychodynamically-oriented supportive therapy compared to CBT, (2) whether a two-stage medication intervention, in which fluoxetine was provided if desipramine (first medication used) was found ineffective or poorly tolerated, would improve the outcomes of a psychological treatment, and (3) whether a combination of medication plus psychological treatment (CBT+Med or Supportive+Med) was superior to medication alone. Analyses revealed that CBT was significantly more effective than supportive psychotherapy in improving binge eating and purging. Second, combining CBT with antidepressant medication was superior to medication alone. Third, combining supportive psychotherapy with medication conferred no additional benefit to medication. The study did not include a psychotherapy-only group, and thus, a direct comparison of CBT with and without medication was not possible.

Jacobi et al. [24], in one of the few BN trials with a 12-month follow-up, tested fluoxetine-only, CBT, and their combination. All treatment conditions led to significant improvements in symptoms, yet abstinence rates were highest in the CBT condition, both at post-treatment and at the 1-year follow-up. Adding fluoxetine conferred no greater benefit in the reduction of binge eating and purging than CBT delivered alone. Caution is warranted when interpreting the reported outcomes given the high rates of attrition across the treatment conditions and especially for the CBT condition.

Mitchell et al. [22], tested the effectiveness of self-help CBT and fluoxetine in a RCT with 91 women with BN randomized to one of four conditions: fluoxetine only, placebo, self-help CBT plus fluoxetine, or self-CBT plus placebo. The investigators reported that abstinence rates did not differ significantly across active treatment conditions (16% for fluoxetine, 26% for self-help plus fluoxetine, 24% for self-help CBT plus placebo, and “not available” for placebo) although those who received self-help plus fluoxetine had the greatest reduction in the frequency of vomiting episodes [22].

Walsh et al. [23], in a two-site trial, tested the effectiveness of self-help CBT and fluoxetine in a RCT with 91 women with BN randomized to one of four conditions: fluoxetine only, placebo, self-help CBT plus fluoxetine, or self-CBT plus placebo. This RCT was performed in primary care settings with generalist clinicians delivering the interventions. Attrition was highly problematic, with two-thirds of the participants dropping out of the study; completer rates were 33.3% for the fluoxetine groups and 28.6% for the self-help groups. Rates of remission were poor at 8.8% overall, with no significant differences across the four treatment conditions.

Mitchell and colleagues, in a multi-site trial with 293 patients with BN, compared “standard CBT” to a “stepped-care” approach using an adaptive treatment strategy; both treatment conditions integrated pharmacotherapy with fluoxetine in the case of non-respond [20]. Participants in the “standard CBT” received standard individual (20 sessions) CBT with a subset of patients who were predicted to be non-responders after 6 sessions (defined as less than 70% reduction in purging) having fluoxetine (60 mg/day) added to their treatment. Participants in the stepped-care arm were started with guided-self-help CBT, with fluoxetine (60 mg/day) added for those predicted to be non-responders after 6 sessions, followed by standard individual CBT for six months for those who failed to achieve abstinence by week 18; fluoxetine, if utilized, was continued until the 1-year follow-up. Analyses revealed that the two treatment conditions did not differ significantly on abstinence rates at post-treatment but the “stepped-care” condition was superior to the “standard CBT” at the 1-year follow-up. For the “at-risk” participants (i.e., non-responders at session 6), the stepped-care approach (which included the addition of fluoxetine) enhanced outcomes at the end of treatment. The authors noted that the guided self-help CBT (which was delivered by less experienced clinicians) plus the addition of fluoxetine for non-responders appeared as effective as the traditional individual standard CBT plus fluoxetine. Importantly, the effectiveness of individual components within the sequences (i.e., the addition of fluoxetine) could not be examined because of (1) no comparison condition; and (2) participants were randomized to the overall treatment arms and not to fluoxetine treatments within the arms.

Findings of Combined Treatments for Binge-Eating Disorder

Findings of RCTs testing combination treatments with antidepressant, weight-loss, and anti-epileptic medications are presented in turn below.

Two open-label trials testing the additions of three antidepressants to CBT/BWL treatments for BED reported little-to-no additive effects. Agras and colleagues [30] performed the first additive/sequential treatment study for BED with 108 women randomized to one of three 9-month treatment conditions: BWL, CBT followed by BWL, and CBT followed by BWL with the addition of the antidepressant desipramine (open-label unblinded). The combined/additive condition with despiramine did not significantly enhance binge-eating outcomes although a statistically significant (albeit clinically meaningless) weight loss was observed. Ricca and colleagues [32] performed an open-label comparative trial in which 108 patients (men and women) with BED were randomized to one of five 6-month treatment conditions: CBT, CBT+fluoxetine, CBT+fluvoxamine, fluoxetine, or fluvoxamine. Analyses revealed both antidepressant monotherapies were significantly inferior to CBT conditions (either alone or with addition of antidepressants) and that neither SSRI antidepressant significantly enhanced CBT outcomes.

Double-blind placebo-controlled RCTs have provided further strong evidence indicating little-to-no advantage for adding antidepressant medications to either CBT or BWL. Laederach-Hofmann and colleagues [31], in placebo-controlled double-blind RCT with 31 patients with BED, found that adding imipramine to a treatment with diet plus supportive counseling failed to enhance binge-eating outcomes relative to adding placebo but was associated with statistically significantly greater weight loss (−2.2 kg vs +0.2 kg). Two double-blind RCTs testing fluoxetine in combination designs yielded convergent findings that adding this SSRI antidepressant did not significantly improve outcomes produced by either BWL [34] or CBT [35]. Specifically, Devlin and colleagues [34] found that adding of fluoxetine to BWL failed to significantly enhance reductions in binge-eating frequency or abstinence rates, eating-disorder psychopathology, or weight loss but did produce greater reduction in depression scores relative to adding placebo. In contrast, adding CBT to BWL significantly enhanced binge-eating outcomes (e.g., abstinence rates of 62% vs 33%, respectively).

Devlin and colleagues [42], in their 24-month follow-up analyses, reported that the addition of CBT to BWL – but not fluoxetine – was associated with significantly greater reductions in frequency of binge eating and abstinence rates; neither CBT nor fluoxetine enhanced weight losses whereas fluoxetine enhanced depression outcomes. Grilo and colleagues [35] reported that the addition of fluoxetine did not significantly enhance CBT on any outcome measure, whereas CBT (with either fluoxetine or placebo) was significantly superior to both fluoxetine and placebo (which did not differ significantly from one another) on binge-eating abstinence rates (i.e., 50% and 61% versus 29% and 30%, respectively). Similar patterns of significant differences were reported for frequency of binge eating and reductions in eating-disorder psychopathology and depression; weight loss, however, was minimal and did not differ significantly across treatments.

Grilo and colleagues [43], in their 12-month follow-up analyses after the completion of treatments, reported that CBT plus fluoxetine and CBT plus placebo did not differ significantly from one another but both were both significantly superior to monotherapy with fluoxetine on most outcomes. Collectively, such findings indicate the significantly superior durability of CBT relative to fluoxetine and that adding fluoxetine versus placebo to CBT matters not for longer-term outcomes following treatment completion [43].

Double-blind placebo-controlled RCTs have generally reported little-to-no advantage for adding the two weight-loss medications that haven been tested to date to either CBT or BWL for BED, although some findings suggest they may enhance weight loss, albeit very modestly. RCTs have examined orlistat [29, 36, 37], an FDA-approved weight-loss medication that works locally in the gut to block fat absorption and sibutramine [39], a centrally-acting agent that was FDA-approved and then withdrawn from the market.

Golay and colleagues [29], in a RCT with 89 patients with BED comorbid with obesity, compared adding orlistat versus placebo to a reduced-calorie diet (some guidance but did not appear to be a behavioral (or lifestyle) weight loss intervention). Analyses revealed no significant differences between the addition of orlistat versus placebo to diet on proportion of patients no longer meeting BED criteria (77% vs 71%, respectively) nor on the frequency of binge eating which was reduced substantially in both treatment conditions. In contrast, adding orlistat to diet was associated with significantly greater weight loss than adding placebo (−7.4% vs −2.3%, respectively).

Grilo and colleagues [36], in a RCT with 50 patients with BED comorbid with obesity, found that adding orlistat to guided-self-help CBT was associated with significantly greater binge-eating abstinence rates than adding placebo at post-treatment (64% vs 36%, respectively) but not at 3-month follow-up after discontinuing treatments (52% vs 52%, respectively). In contrast, adding orlistat was associated significantly greater rates of attaining 5% weight loss at both post-treatment (36% vs 8%, respectively) and at 3-month follow-up (32% vs 8%, respectively).

Grilo and White [37], in a RCT with 89 Latinx patients with obesity, 40 with BED and 39 without BED, tested the addition of orlistat versus placebo (double-blind) to BWL treatments delivered in individual sessions over four months with a 6-month follow-up after finishing treatments. Notably, this RCT was performed at a community mental health center serving economically/educationally-disadvantaged Latinxs patients in Spanish in a “real-world” urban setting. Analyses revealed significant reductions in eating-disorder psychopathology, depression, and weight loss. Overall, adding orlistat to BWL did not enhance outcomes; however, BED significantly moderated weight-loss outcomes: adding orlistat to BWL enhanced weight loss those patients without BED but not among those with BED. Analyses with the subgroup of 40 patients with BED revealed binge-eating abstinence rates of 65% at post-treatment and 50% at 6-month follow-up.

Grilo and colleagues [39], in a placebo-controlled double-blind RCT conducted in a primary-care setting with 104 patients with BED comorbid with obesity, tested the effectiveness of self-help CBT and sibutramine, alone and in combination. Patients were randomized to one of four conditions (sibutramine only, placebo, self-help CBT plus sibutramine, or self-CBT plus placebo) delivered for four months and were followed up for 12 months after completing/discontinuing treatments. Analyses revealed that binge-eating abstinence rates did not differ significantly across treatments and the treatments differed little in frequency of binge eating, eating-disorder psychopathology, and depression. In contrast, sibutramine was associated with significantly greater weight loss at post-treatment; however, weight re-gain occurred in patients who had received sibutramine and by 6- and 12-month follow-ups, weight loss was no longer significantly different across groups.

Grilo and colleagues [38, 44] tested the effectiveness of an adaptive stepped-care treatment versus BWL in 191 patients with BED and comorbid obesity. Patients were randomized to either to BWL (standard treatment; n = 39) or to stepped care (n = 152) for six months and were followed up 6- and 12-months after completing/discontinuing the treatments. Within stepped care, after one month of BWL patients were stratified by treatment response. Patients achieving a rapid response (defined as a 65% reduction in binge eating) were randomized to a weight-loss medication (sibutramine or orlistat) or to placebo (double-blind). Nonrapid responders were switched from BWL to guided-self-help CBT plus in addition were randomized to either weight-loss medication or placebo for the remaining 5 months. Analyses revealed that, overall, BWL and stepped care did not differ in binge-eating abstinence rates (74.4% vs. 66.5%), percent weight loss (5.1% vs 5.8%), nor secondary outcomes of depression or eating-disorder psychopathology. Within stepped care, weight-loss medication enhanced certain outcomes. Weight-loss medication significantly enhanced binge-eating abstinence and frequency outcomes among non-responders to BWL and weight-loss outcomes among both responders and non-responders to BWL. Analyses of 12-month follow-up data [44] revealed that overall the binge-eating improvements and weight losses produced by BWL an stepped care did not differ significantly and that within stepped the medication and placebo conditions did not differ.

Two RCTs have tested anti-epileptic medications in combination with CBT for BED [28, 33]. Ricca and colleagues [33], in an unblinded open-label RCT with 52 patient with (subthreshold and threshold) BED, reported that adding zonisamide to CBT for 24 weeks was associated with significantly greater reductions than CBT (alone) in eating-disorder psychopathology, depression, and BMI at posttreatment. At 12-month follow-up, patients who had received zonisamide plus CBT had significantly greater reductions in binge-eating frequency than those treated with CBT alone, along with significantly greater reductions in one of six eating-disorder measures and anxiety (but not depression) scores. Ricca et al [33] also reported analyses suggesting that both treatment conditions had significant reductions in BMI at post-treatment and that CBT, but not zonisamide plus CBT regained weight during the 12-month follow-up.

Claudino and colleagues [28], in a placebo-controlled double-blind RCT with 73 patients with BED, tested the addition of topiramate versus placebo with 21-weeks of CBT. Analyses revealed that topiramate, compared to placebo, was associated with significantly greater binge-eating abstinence rates (84% versus 61%, respectively; using one-week end-point data) and significantly greater weight loss (−6.8 kg vs −0.9 kg). Analyses comparing topiramate and placebo combinations with CBT did not reveal statistically significant differences in reductions in binge-eating frequency, eating-disorder psychopathology, or depression.

Discussion and Conclusions

Since many patients do not derive sufficient benefits from the available monotherapies for eating disorders, the present review evaluated RCTs testing combined pharmacological and psychological treatments to inform clinical practice and future treatment research. In contrast to the public health significance of eating disorders, the scope of research performed on the utility of combining treatments is limited overall and particularly in recent years for AN and BN.

We identified a total of only 28 RCTs testing combined interventions for eating disorders. This RCT literature is characterized by substantial methodological limitations, particularly for the AN and BN trials, included high attrition, small sample sizes, limited diversity of patient groups, and dearth of data following the completion of treatments. Overall, RCTs testing combined pharmacological plus psychological treatments for eating disorders have yielded mostly non-significant findings. For AN, 3 of the 4 RCTs reported no significant advantage for combining treatments and the fourth [14] reported a statistically significant albeit clinically modest advantage. For BN, 10 of the 12 RCTs reported no significant advantage for combining treatments; 2 RCTs [22, 23] found that combining fluoxetine with specific psychological treatments enhanced outcomes relative to medication-only but not relative to the psychological treatments only. For BED, of the 12 RCTs, only 2 (both with antiseizure medications) [28, 33] significantly enhanced both binge-eating and weight outcomes, and only 2 (with orlistat, a weight-loss medication) [29, 36] enhanced weight loss, albeit very modestly, but not binge-eating outcomes.

Future research should examine whether treatments found to be effective in specialty clinics can be effectively delivered in generalist or more diverse setting. Of three relevant RCTs, one found that evidence-based treatments for BED could be effectively delivered even in community clinics served disadvantaged patients [37] but two found that evidence-based treatments for BED [39] and BN [23] were largely ineffective when delivered by generalist clinicians in primary care settings.

Future research should focus on testing additive benefits of medications with relevant mechanisms of action to available effective psychological interventions. For example, there are three FDA-approved weight-loss medications (phentermine/topiramate, naltrexone/bupropion, and liraglutide) that could be considered for the treatment of BED either alone or in combination with psychological/behavioral interventions. Similarly, on-going research is testing the utility of combining LDX [45, 46], the sole FDA-approved medication for BED, with CBT (NCT03924193). It is important to note that at the present time, there are no FDA-approved medications indicated specifically for AN, only one (fluoxetine) for BN, and one (lisexamfetamine) for BED, and as such, any use of other medicines for the treatment of eating disorders is “off-label.”

Future RCTs testing additive effects should employ innovative adaptive designs [38] to test ways to combine and sequence treatments to improve outcomes amongst non-responders and to enhance outcomes amongst responders to initial interventions. Future RCTs should also examine patient factors that might predict or moderate outcomes given the dearth of available empirical data on predictors/moderators of treatment outcomes [47]. For AN, there is no available evidence on patient variables to guide the prescription of specific psychological versus pharmacological versus combined treatments. For BN, one study found that greater depression predicted more rapid response to CBT than to SPT and to antidepressant pharmacotherapy than placebo [48]; such findings might suggest the utility for considering combining treatments in cases with greater depression. For BED, one study found that overvaluation of shape/weight (a specific body-image cognitive feature of eating disorders) significantly moderated treatment outcomes: patients with high overvaluation were significantly more likely to improve if receiving CBT than medication alone [49]. Such findings suggest the utility of combining CBT rather than relying on medication monotherapy if treating patients with BED with high levels of overvaluation of shape/weight. In the only study with data to test for predictors/moderators of combination treatments, Lydecker and Grilo [50]found that psychiatric comorbidity was associated with greater ED psychopathology throughout treatment but did not significantly moderate outcomes attained with CBT, BWL, or combined pharmacological plus psychological treatments. Such findings, which challenge clinical perspectives that comorbidity indicates need for combination treatments, require further research.

Future clinical practice and research also build on a treatment process known as “rapid response” which has been found to be a reliable prognostic predictor of outcomes for BN and BED. Early rapid response to treatment reliably predicts better outcomes for both psychological and pharmacological treatments among patients with BN [48] and with BED [51–53]. Research with BED has found that failing to have a rapid response to initial treatments, particularly to pharmacotherapies [51,52] or to behavioral weight loss [53], is a signal to consider the need to either switch treatments or to add medications [38]. Finally, research should include longer-term studies to determine optimal treatment duration, include follow-up assessments to identify both the differential durability of treatments and periods of risk for relapse [43] and whether to discontinue medications [54].

Table 3.

Main findings from RCTS investigating combined psychotherapy and pharmacotherapy for BN

Study	Treatment condition	Attrition %	Purging frequency	Binge-eating frequency	Depression	Abstinence/Remission
Agras et al. (1992)[16]	SSRI 16 SSRI 24 CBT+SSRI 16 CBT + SSRI 24 CBT 24	a/b)17% c/d)4%	Pre-, 16, 24, 32 wks: 9.7 to 4.7 to 5.0 to 6.2c,d,e 6.3 to 3.9 to 2.9 to 3.4 c,d,e 8.3 to 2.6 to 2.7 to 3.2 a,b 11.7 to 1.2 to 1.7 to 1.1 a,b 10.1 to 1.7 to 2.7 to 2.2 a,b	Pre-, 16, 24, 32 wks: 5.5 to 3.5 to 3.7 to 6.2 c,d,e 5.9 to 3.4 to 2.7 to 3.3 c,d,e 7.5 to 2.4 to 2.1 to 3.2 c,d,e 9.3 to 1.7 to 2.3 to 1.0 a,b 8.7 to 1.5 to 2.8 to 2.5 a,b	NA	Purge abstinence, 16 wks: a/b)33%e c/d)64%e e)48%abcd Binge abstinence, 16 & 32 wks: a/b)35, 42e c/d)65, 70e e)50, 55 abcd
Beumont et al. (1997)[25]	Nutritional counseling + PL Nutritional counseling + SSRI	30% 50%	Pre-, 8, and 20 wks: 7.3 to 2.3 to to 2.3 8.8 to 1.2 to 2.5	Pre-, 8, and 20 wks: 6.1 to 1.2 to 1.9 10.1 to 1.6 to 2.2	HAMD: 11.8 to 6.8 11.0 to 5.3	Binge-abstinence at 8 wks and FUP: 61.5%, 60.9% 69.6%, 35.7%
Fahy et al. (1993)[17]	CBT + d-fenfluramine CBT + placebo	0% 9.4%	At pre- and 16 wks: 6.2 to 3.2 7.8 to 3.8	At pre- and 16 wks: 6.6 to 3.4 5.1 to 2.4	MADRS: 13.1 to 8.7 13.0 to 9.3	No longer fulfilled DSM criteria for BN, total N: 8-wk: 54% 16-wk: 74%
Fichter et al. (1991)[26]	Inpt psychotherapy+SSRI Inpatient psychotherapy + PL	0.0% 0.0%	NA	At pre- and post: 5.6 to 30 8.9 to 6.60	HAMD: 13.3 to 8.3 14.1 to 11.1	EDI Bulimia 10.2 to 3.0 9.9 to 4.0
Goldbloom et al. (1997)[18]	CBT SSRI CBT + SSRI	33.3% 39.1% 56.9%	Percent reduction: 79.2%b 37.4%a,c 82.4% b	Percent reduction: 80% 70% 87%	BDI: 18.4 to 13.8 16.3 to 13.6 14.8 to 7.5	Percent abstinent from vomit + BE: 43% 17% 25%
Jacobi et al. (2002)[24]	CBT SSRI CBT + SSRI	42% 25% 33%	16.4 to 28.5 to 11.3 22.7 to 37.9 to 15.5 15.3 to 14.7 to 21.6 Percent reduction, post-tx: 52% 35% 32%	17.4 to 13.5 to 8.9 20.5 to 34.9 to 12.7 9.0 to 13.5 to 14.8 Percent reduction, post-tx: 42% 46% 50%	BDI: 16.8 to 10.9 16.0 to 14.4 17.2 to 13.9	Percent abstinent from vomiting, binge eating at 1-year FUP: 20%, 40% 13%, 38% 11%, 11%
Leitenberg et al. (1994)[19]	CBT Desipramine CBT + desipramine	14% 57% 28%	Mean frequency vomiting at pre-, end, and 6-mo FUP: 9.3 to 0.3 to 0.9 * 9.2 to 8.2 to 7.2 7.1 to 1.7 to 1.7*	EAT mean score: 41.6 to 14.7* to 12.6* 55.7 to 47.0 to 41.0 62.6 to 26.4* to 36.8	IDD: 27.5 to 8.2 to 8.8* 32.0 to 21.0 to 21.0 36.2 to 18.2* to 19.8	Percent abstinent from vomiting, post-tx: 71.4% (5 of 7) 0% 57.1% (4 of 7)
Mitchell et al. (1990)[27]	Placebo Imipramine Intensive group therapy + placebo Intensive group therapy + imipramine	16.1% 32.6% 14.7% 25.0%	Mean frequency vomiting, pre-post: 10.0 to 9.9 b,c,d 8.6 to 4.7 c,d 13.2 to 1.3 a,b 9.6 to 1.0 a,b	Mean frequency binge eating, pre-post: 8.0 to 7.8 b,c,d 7.3 to 3.7 c,d 9.2 to 1.0 a,b 8.4 to 0.7 a,b	HAMD-D, pre-post: 10.9 to 9.7 b,c,d 11.6 to 7.0 c,d 9.5 to 4.2 a,b 11.0 to 2.3 a,b	EDI global score, pre-post: 68.9 to 64.0 b,c,d 67.4 to 49.6 c,d 60.9 to 28.5 c,d 66.1 to 26.2 b,c,d
Mitchell et al. (2001)[22]	Placebo Fluoxetine CBTsh + Placebo CBTsh + Fluoxetine	5.5% 0.0% 0.0% 5.8%	Percent reduction at endpoint and FUP: 22.8% and 21.8% 52.8% and 46.1% 50.2% and 31.5% 66.7% and 67.7% * Med vs placebo and for gSH vs. no gSH	Percent reduction at endpoint and FUP: 32.4% and 26.9% 50.3% and 43.4% 59.7% and 35.4% 66.8% and 61.6%	HAM-D scores, pre: 10.9 8.85 10.1 8.1 NS time or group effects.	Percent abstinent from vomiting + binge eating. NA 16% 24% 26%
Mitchell et al. (2011) [20]	CBT Stepped care	19% 25%	Frequency binge eating at pre-, post-, and FUP: 27 to 4 to 10 27 to 8 to 3	Frequency compensatory pre, post, and FUP: 44 to 12 to 15 43 to 19 to 5	BDI: 17.9 to 12.3 to 13.3 17.5 to 12.7 to 12.1	Percent remitted (no longer meeting DSM criteria) at end of treatment and FUP: 57% and 44% 52% and 32%
Walsh et al. (1997)[21]	CBT-SSRI CBT-placebo SPT-SSRI SPT-Placebo Medication	34% overall	Frequency: 10.8 to 1.1e 10.8 to 5.6d 10.6 to 5.5 11.9 to 7.5b 10.5 to 3.7a	Frequency: 7.3 to 0.95 e 7.2 to 2.56 d 7.9 to 3.6 6.2 to 3.3 b 8.3 to 2.6 a	BDI: 10.9 to 4.4 11.7 to 6.8 15.9 to 6.7 14.3 to 10.2 14.5 to 8.2	Percent abstinent from vomiting + BE: 48% 20% 9% 14% 21%
Walsh et al. (2004)[23]	Placebo Fluoxetine alone Placebo + gSH Med + gSH	a/b)66.6% c/d)71.4%	17.3 to 12.5 18.3 to 11.2 20.7 to 10.3 19.3 to 17.2	15.5 to 9.9 16.6 to 8.1 20.70 to 10.1 17.9 to 13.9	BDI: 18.4 to 15.9 18.4 to 10.4 19.6 to 17.2 19.7 to 12.5	Percent abstinent from vomiting + BE: a/b)9.5% c/d)12.2%

Notes: BN = bulimia nervosa; FUP = follow-up; RCT = randomized controlled trial; SSRI = selective serotonin reuptake inhibitor; CBT = cognitive-behavior therapy; SPT = supportive psychotherapy; sh = self-help; gSH= guided self-help; DSM = Diagnostic and Statistical Manual of Mental Disorders; ns = not significant; IDD = Inventory to Diagnose Depression, EAT = Eating Attitudes Test, EDI = Eating Disorders Inventory; BDI = Beck Depression Inventory; HAMD = Hamilton Depression Inventory; BE = binge eating; NA = data not available. Subscripts denote significant group differences between treatments.

Table 4.

Main findings from RCTS investigating combined psychotherapy and pharmacotherapy for BED

Study	Treatment condition	Attrition %	% binge abstinent	Binge-eating frequency	Weight loss	Depression	Eating pathology
Agras et al. (1994)[30]	BWL-only CBT/BWL CBT/BWL-D	27 17 23	19, 14 37, 28 41, 32	Binge days, weekly: 4.5 to 1.5 to 2.0 4.4 to 1.2 to 1.7 5.1 to 0.9 to 1.5	Weight loss (kg): 3.7 to 4.2 1.6 to 0 a 6.0 to 4.8 b	BDI: 12.9 to 11.6 to 11.3 13.5 to 12.7 to 8.9 13.7 to 10.8 to 7.8	TFEQ Disinhibition: CBT/BWL-D > BWL-only at wk 24
Laederach-Hoffman et al. (1999) [31]	Counseling Imipramine	6 7	NA	Binge episodes, weekly: 7.1 to 5.3 to 7.2 7.1 to 2.5 to 4.1	Weight loss (kg): +0.2 to +2.2 2.2 to 5.1a	HAM-D: 21.3 to 16.0 to 19.2 22.6 to 9.8 a to 12.6 a	NA
Ricca et al. (2001)[32]	CBT CBT + FLX CBT + FLV FLX-only FLV-only	15 27 22 24 27	NA	Binge episodes, monthly: 18 to 8.0 to 8.0d,e 17 to 6.0 to 7.0d,e 18.0 to 8.0 to 8.0d,e 20.0 to 19.0 to 21.0 20.0 to 18.0 to 18.0	BMI: Significantly reduced at post-tx and FUP for CBT,CBT+FLX and CBT+FLV, but not for the FLX and FLV groups	BDI: 22 to 14 to 14 17 to 11 to 11 22 to 10 to 10 20 to 15 to 16 21 to 14 to 14 Note: Scores were significantly lower at post-tx for all groups, with no significant group differences.	EDE: 3.8 to 3.4 to 3.3 d,e 3.8 to 2.7 to 2.7 d,e 4.0 to 2.7 to 2.6 d,e 3.4 to 3.8 to 3.9 3.8 to 3.8 to 3.8 Note: Significantly reduced EDE total for CBT, CBT-FLX and CBT-FLV but not FLX and FLV groups. Greatest reduction for CBT-FLV. No additional improvement at FUP.
Devlin et al. (2005, 2007)[34, 42]	FLX Placebo CBT CBT + FLX All given in addition to group BWL	31 48 40 25	Main effect for CBT: 62% vs 33%** No main effect for FLX: 52% vs 41%	Monthly, pre-post: 16.4 to 4.9 15.4 to 6.0 17.1 to 3.7 16.1 to 2.2 Note: At post-tx and 2-yr FUP, the addition of CBT to BWL significantly enhanced reduction in binge frequency and remission (p<.001)	Wt (kg) at pre-post: 113.8 to 111.9 113.5 to 111.1 116.5 to 114.6 116.9 to 112.8 Note: At post-tx and FUP, no significant main effects for either CBT or medication assignment.	BDI, pre-post: 14.5 to 7.5 15.6 to 10.6 13.9 to 8.4 16.9 to 6.3 Note: At post-tx and FUP, advantage to FLX vs. placebo (p<05).	At post-tx, no significant treatment effects for TFEQ, BES, or BSQ. At FUP, FLX showed advantage to placebo for TFEQ restraint (p = .03).
Golay et al. (2005)[29]	Orlistat/diet Placebo/ diet	11 29	77 71 Note: % no longer meeting BED criteria	Binge episodes, weekly: 5.4 to 1.0 6.2 to 1.7	% loss, IBW: 7.4% loss 2.3% loss a	BDI, pre-post: 10.8 to 8.2 13.6 to 11.6 a	EDI total: 68.0 to 50.0 64.9 to 58.4 a
Grilo et al. (2005)[29]	Orlistat + CBTgsh Placebo + CBTgsh	24 20	64, 52 36 a, 52	Binge episodes, monthly: 16.4 to 3.2 to 3.4 13.5 to 3.6 to 2.8	% achieving 5% loss: 36%, 32% 8% a, 8% a	BDI: 17.1 to 10.1 to 9.9 20.6 to 14.7 to 14.6	EDE-interview global: 3.2 to 2.1 to 2.2 3.2 to 2.4 to 2.3
Grilo et al. (2005, 2012) [35, 43]	FLX -only Placebo‡ CBT + placebo CBT + FLX	22 15 21 23	22, 4 26, NA 61 a,b, 36a 50 a,b, 27a	Binge episodes, monthly EDE-Q: 17.9 to 10.4 to 10.4 13.2 to 7.2 16.6 to 2.3 to 4.6 a,b 16.5 to 4.3 to 4.6 a,b	Wt loss (lbs) at post-tx and FUP: 4.8 to 1.5 5.0 to 9.8 5.6 to 4.1	BDI: 16.9 to12.6 to 12.9 18.7 to 11.7 16.5 to 7.4a to 11.4 20.2 to 8.3 to 11.2	EDE global: 3.9 to 3.1 to 3.3 3.5 to 2.6 3.8 to 2.1 to 2.7 a,b 4.0 to 2.2 to 2.4 a,b
Claudino et al. (2007)[28]	CBT+ topiramate CBT + placebo	19 28	84 61a	Binge episodes, weekly: 4.2 to 0.0 3.4 to 0.3	Wt loss (kg): 6.8 kg 0.9 kga	BDI: 16.8 to 10.9 15.9 to 9.2	BES: 27.2 to 7.5 26.5 to 8.6
Ricca et al. (2009) [33]	CBT CBT + zonisamide	33 50	NA	Binge episodes, monthly: 5.0 to 2.0 to 3.0 5.0 to 2.0 to 2.0a	BMI: 39.2 to 38.4 to 38.9 38.4 to 36.7 a to 36.5 a	BDI: 19.5 to 14.5 to 17.5 20 to 16 to 16.0	EDE-Q global: 2.8 to 2.6 to 2.7 2.8 to 2.1 a to 2.2 a
Grilo & White (2013)[37] ‡	BWL+orlistat BWL+placebo	30 25	60, 50 70, 50	NA	BMI: 39.0 to 37.9 to 37.6 37.2 to 36.0 to 36.7 a	BDI: 22.9 to 11.4 to 10.3 25.7 to 17.1 to 20.9	EDE global: 2.5 to 1.6 to 1.5 2.7 to 2.0 to 1.9
Grilo et al. (2014)[39]	Sibutramine Placebo shCBT + placebo shCBT + sibutramine	27 15 12 16	39, 19, 19 30, 41, 37 24, 40, 40 23, 50, 42	Binge episodes, monthly: 22.4 to 5.0 to 4.9 to 5.8 21.1 to 5.3 to 5.7 to 6.7 14.6 to 6.4 to 3.6 a,b to 4.9 16.9 to 3.6 to 3.9 to 3.0	BMI: 39.4 to 38.4 c,d to 38.7 to 39.3 39.3 to 39.6 c,d to 38.8to 39.5 36.5 to 35.9 a,b to 35.3 to 35.4 37.8 to 35.6a,b to 36.0 to 36.5	BDI: 12.8 to 7.9 to 9.7 to 10.1 13.6 to 9.6 to 8.7 to 7.5 17.0 to 9.8 to10.3 to 10.5 14.0 to 9.3 to 9.8 to 10.3	EDE global: 2.4 to 1.7 to 1.9 to 1.8 2.6 to 2.1 to 1.8 to 1.8 2.5 to 1.7 to 1.7 to1.6 2.5 to 1.8 to 1.6 to 1.6
Grilo et al. (2020)[38, 44]	BWL standard Stepped care*	2.6 12.5	74.4, 38.2, 44.7 66.5, 33.3, 41.1	Binge episodes, monthly: 17.8 to 1.7 to 1.9 to 2.4 20.2 to 2.7 to 3.8 to 2.4	BMI: 37.5 to 35.7 to 35.2 to 35.9 39.4 to 36.9 to 37.3 to 37.3 Weight loss (%) 5.1 to 5.1 to 3.4 5.8 to 5.2 to 5.0	BDI: 14.0 to 10.0 to 9.1 to 9.4 15.3 to 8.9 to 8.8 to 8.2	EDE global: 2.6 to 1.8 to 1.7 to 1.7 2.7 to 1.8 to 1.7 to 1.6

Note: BED = binge eating disorder; FUP = follow-up; RCT = randomized controlled trial; BWL = behavioral weight loss; CBT = cognitive-behavior therapy; gsh = guided self-help. FLX = fluoxetine; FLV = fluvoxamine; EDE = Eating Disorder Examination-Questionnaire; BMI = kg/m²; BDI = Beck Depression Inventory; BES = Binge Eating Scale; NA = data not available. Subscripts denote significant group differences between treatment conditions.

^†Placebo group was offered treatment at the end of 2×2 trial and were not included in the 12-month FUP.

^‡An additional N=39 obese patients without BED were also randomized to the treatments. The reader is referred to the publication for findings which notably included BED status as a significant predictor and moderator of outcomes [37]

^*Rapid responders were randomized into BWL + Med or BWL + Placebo and non-rapid responders were randomized into CBTgsh + Med or CBTgsh + Placebo.