The Efficacy of Self-Help Group Treatment and Therapist-Led Group Treatment for Binge Eating Disorder

Participants (N=259) were adults recruited from two clinical sites, one in Minnesota and the other in North Dakota. Potential participants were recruited from the community through advertisements as well as referrals from local eating disorder treatment clinics and other health professionals. For inclusion, participants had to meet full DSM-IV criteria for binge eating disorder as assessed by the Eating Disorder Examination (19) and have a body mass index (BMI) ≥25 (kg/m ² ). Exclusion criteria were pregnancy or lactation; lifetime diagnosis of bipolar or psychotic disorder; current diagnosis of substance abuse or substance dependence; medical or psychiatric instability, including acute suicide risk; current psychotherapy; or current participation in a formal weight loss program. Participants who were on a stable dose of antidepressant medication for a minimum of 6 weeks were allowed to participate.

Of the 1,917 individuals screened by phone, 1,519 were ineligible or not interested in participating ( Figure 1 ). Of the 398 who attended initial orientation meetings, 367 were assessed by interview; of these, 108 were excluded or declined to participate, resulting in 259 who were enrolled in the study.

This study was approved by the institutional review boards at both sites. Written informed consent was obtained during the orientation meeting after potential participants received a detailed description of the study.

The primary outcome measure was the frequency of binge eating episodes as measured by the Eating Disorder Examination at baseline, at end of treatment (or end of waiting period for participants assigned to the waiting list condition), and at 6- and 12-month follow-up assessments. Eating pathology was also assessed using the Three-Factor Eating Questionnaire (20) . Additional secondary outcome measures included BMI, Inventory of Depressive Symptomatology–Self-Report score (21), Rosenberg Self-Esteem Questionnaire score (22), and Impact of Weight on Quality of Life–Lite score (23) . Participants completed all of these measures at baseline, at end of treatment or waiting period, and at the 6- and 12-month follow-ups.

Experienced graduate-level assessors blind to participant randomization conducted the interviews. Throughout the study, assessors met in person and communicated between sites by teleconference and e-mail in order to prevent drift. All assessment interviews were audiotaped. Interrater reliability ratings were conducted on a random sample (20%) of Eating Disorder Examination audiotapes. Interrater reliability (based on intraclass correlation coefficients) for the Eating Disorder Examination subscales and global score ranged from 0.955 to 0.982.

After completing the assessment protocol, participants were randomly assigned to one of three active treatments or the waiting list condition by the independent biostatistician (R.D.C.). An adaptive randomization strategy was used in which the probability of assignment to any treatment condition at a given point in time was inversely related to the relative proportion of participants previously assigned to that condition. Assignment sequence was shielded from all investigators, study personnel, therapists, and participants until time of randomization.

Participants randomized to the waiting list condition were informed that they would receive therapist-led group treatment at the end of the 20-week waiting period. Data collected from the waiting list participants after the waiting period were not included in these analyses.

All active treatments consisted of 15 group sessions of 80 minutes’ duration over a 20-week period, with weekly sessions for the first 10 weeks and biweekly sessions for the remaining 10 weeks. The mean group size was six, and the range was two to 11. The content of the three active treatment conditions was identical; only treatment delivery varied. The initial sessions focused on behavioral and cognitive interventions; the middle sessions emphasized techniques to target associated problems, including stress management and body image; and the final two sessions included strategies to prevent relapse (24) . Each 80-minute session was divided into two equal segments, with the first focusing on psychoeducation and the second on homework review and discussion. All participants in active treatment received identical workbooks and homework assignments.

In the therapist-led CBT groups, a doctoral-level psychotherapist provided psychoeducation during the first half of each session and homework review and discussion during the second half. In the therapist-assisted CBT groups, participants watched a psychoeducational videotape (a specific tape was designed for each session) during the first half of each session, and during the second half a doctoral-level psychotherapist joined the group to review homework and lead a discussion. In the self-help groups, participants watched a psychoeducational videotape during the first half of each session and conducted their own homework review and discussion during the second half. Participants in the self-help groups were given comprehensive instructions with detailed guidelines and time allotments for each discussion session. In addition, group members were assigned the role of lead facilitator on a rotating basis. Participants assigned to the waiting list condition received therapist-led treatment at the end of the 20-week waiting period.

Doctoral-level therapists were initially trained by didactic and discussion sessions and led a practice group before administering the treatment for this study. Therapists met regularly with the supervisor (C.B.P.) in person and by teleconference to discuss the implementation of the treatment manual, ensure treatment fidelity, and prevent center drift. All group sessions were audiotaped. Several sessions from each therapist-led and therapist-assisted group (a total of 106 tapes) were reviewed by the supervisors (C.B.P. and S.J.C.) and rated on a 7-point Likert scale (with 7 the highest rating). The overall therapist rating was 6.32 (SD=0.38); subscale ratings were as follows: adherence, mean=6.19 (SD=0.93); comprehensiveness, mean=6.00 (SD=0.86); effective communication, mean=6.58 (SD=0.62); therapeutic technique, mean=6.46 (SD=0.69); and rapport, mean=6.40 (SD=0.55).

Power analysis for this study was based on two previous studies conducted by our research group evaluating the effects of psychotherapeutic and self-help treatment (18, 25) on binge eating. Assuming a two-tailed alpha of 0.05, we calculated that a sample size of 65 per group (260 total) would provide a power greater than 0.99 to detect posttreatment differences in binge eating frequencies between active treatments and the waiting list condition, and a power of 0.88 to detect differences between active treatments (26) .

Baseline demographic and clinical characteristics were compared across treatment assignment and site using analysis of variance for continuous measures and logistic regression for dichotomous variables. Models included main effects for treatment group and site and a treatment group-by-site interaction.

The primary efficacy analysis was based on the intent-to-treat sample. Where posttreatment or follow-up data were missing, the baseline value was carried forward. Three primary outcome variables obtained from the Eating Disorder Examination were evaluated: binge eating episodes in the previous 28 days; binge eating days in the previous 28 days; and abstinence, defined as no binge eating episodes in the past 28 days. Analysis of covariance was used to compare the four groups on binge eating episodes and days at end of treatment or waiting period, controlling for baseline values, site, and gender (based on baseline differences between groups). Because of positive skew, binge eating episodes and days were log-transformed prior to analysis. Pairwise post hoc comparisons between groups were based on covariate-adjusted contrasts using a significance level of 0.008 (0.05/6) at end of treatment and 0.017 (0.05/3) at follow-up assessments. Logistic regression analysis was used to compare groups on abstinence rates at end of treatment waiting period, controlling for site and gender.

Of the 259 participants, 227 (87.6%) were women. The average age was 47.1 years (SD=10.4, range=19–65). Most participants were Caucasian (96.1%), had a college degree or higher (54.9%), were employed full-time (61.0%), and were taking antidepressant medication (78.8%). The average BMI was 39.0 (SD=7.8, range=24.8–72.6).

A total of 69 participants were assigned to the waiting list condition (31 in Minnesota, 38 in North Dakota), 67 to the self-help condition (15 in Minnesota, 52 in North Dakota), 63 to the therapist-assisted condition (18 in Minnesota, 45 in North Dakota), and 60 to the therapist-led condition (37 in Minnesota, 23 in North Dakota). (Table S1, in the data supplement that accompanies the online edition of this article, presents baseline demographic and clinical characteristics by treatment group and treatment site.) Participants from the Minnesota site were older on average (F=6.12, df=1, 251, p=0.014; partial eta-squared=0.024) and had significantly higher restraint scores on the Three-Factor Eating Questionnaire (F=12.21, df=1, 229, p=0.001; partial eta-squared=0.051). The only difference in baseline characteristics between treatment groups was on gender distribution (χ ² =13.97, df=3, p=0.003; Nagelkerke R ² =0.146), where the percentage of women in the therapist-led group (100%) was higher than that for waiting list (81.2%) and therapist-assisted (81.0%) groups.

Of the 259 participants, 192 (74.1%) were assessed at end of treatment or waiting period ( Figure 1 ). A larger number of participants completed the therapist-led (N=53; 88.3%) and waiting list (N=56; 81.2%) conditions than the therapist-assisted (N=43; 68.3%) and self-help (N=40; 59.7%) conditions (χ ² =16.50, df=3, p=0.001). In comparison to those who completed the assessment at end of treatment or waiting period, those who did not complete the assessment were younger (mean=44.4 years compared with mean=47.8 years; F=5.44, df=1, 257, p=0.020; partial eta-squared=0.021) but did not differ on other demographic or clinical characteristics.

Figure 2 presents the percentages of participants who were abstinent from objective binge eating episodes in the previous 28 days at end of treatment or waiting period, at the 6-month follow-up, and at the 12-month follow-up by treatment group; data are shown for both the intent-to-treat sample and those who completed treatment or the waiting period. In the intent-to-treat sample, abstinence rates were as follows: at end of treatment or waiting period, therapist-led group, 51.7%; therapist-assisted group, 33.3%; self-help group, 17.9%; waiting list group, 10.1%; at the 6-month follow-up: therapist-led group, 43.3%; therapist-assisted group, 23.8%; self-help group, 19.4%; at the 12-month follow-up, therapist-led group, 20.8%; therapist-assisted group, 27.0%; self-help group, 25.4%. Posttreatment abstinence rates were significantly different across treatment groups after controlling for site and gender both in the intent-to-treat sample (χ ² =24.19, df=3, p<0.001) and for those who completed treatment (χ ² =22.28, df=3, p<0.001). The therapist-led and therapist-assisted groups had significantly (p<0.008) higher abstinence rates than patients in the waiting list group for both analyses. In addition, abstinence rates were significantly higher in the therapist-led group than in the self-help group in the intent-to-treat sample but not among those who completed treatment. Differences in abstinence rates at the 6-month follow-up approached significance in the intent-to-treat analysis (χ ² =5.55, df=2, p=0.062) and reached statistical significance in the analysis of those who completed treatment (χ ² =6.69, df=2, p=0.035), with the therapist-led group having significantly higher abstinence rates than the self-help group (p=0.013). No differences between abstinence rates at the 12-month follow-up were observed either for the intent-to-treat group or for those who completed treatment. No site differences were observed in abstinence rates.

Table 1 presents average objective binge eating days and episodes at baseline, at end of treatment or waiting period, at the 6-month follow-up, and at the 12-month follow-up by treatment group for the intent-to-treat sample. Significant differences between treatment groups at the posttreatment assessment were found for both objective binge eating days (F=14.97, df=3, 252, p<0.001; partial eta-squared=0.151) and episodes (F=15.29, df=3, 252, p<0.001; partial eta-squared=0.154). Post hoc analyses indicated that 1) the therapist-led group had greater reductions in objective binge eating days and episodes than the self-help and waiting list groups; 2) the therapist-assisted group had greater reductions in objective binge eating days and episodes than the waiting list group; and 3) the self-help group had greater reductions in objective binge eating episodes (but not days) than the waiting list group. No significant differences in binge eating frequencies were found between treatment groups at either of the follow-up assessments. No site differences were observed at end of treatment or waiting period or at follow-up assessments.

Few differences between treatment groups were observed in secondary outcome measures at end of treatment or waiting period ( Table 1 ). The therapist-led group experienced significantly greater reductions than the waiting list group on the Eating Disorder Examination restraint subscale (F=3.46, df=3, 252, p=0.017; partial eta-squared=0.040) and global score (F=4.04, df=3, 252, p=0.008; partial eta-squared=0.046). In addition, both the therapist-assisted and therapist-led groups experienced greater reductions than the waiting list group in disinhibition scores on the Three-Factor Eating Questionnaire (F=5.78, df=3, 230, p=0.001; partial eta-squared=0.070). No differences in secondary outcome measures, including depression, quality of life, and BMI, were observed between treatment groups at either follow-up assessment.