An initial evaluation of a weight loss intervention for individuals who engage in emotional eating

Over 30 % of Americans have obesity (Ogden et al., 2014). Obesity presents an increased risk for a variety of serious medical conditions and is associated with reduced quality of life and psychosocial functioning (Jensen et al., 2013), and being subjected to bias and discrimination (Puhl & Heuer, 2009). Obesity is a heterogeneous condition, both in terms of etiological factors (Marcus & Wildes, 2009) and treatment outcomes (Foreyt & Goodrick, 1993). Randomized trials over the last two decades suggest that behavioral weight loss treatments (i.e., lifestyle interventions targeting caloric intake and physical activity) successfully produce modest weight losses of 7–10 % of initial body weight in 30 weeks (Wadden et al., 2012). However, there is great variability in the response to standard weight loss treatments, with some patients losing little or no weight and others losing at ≥15 kg within 16–24 weeks (Foreyt & Goodrick, 1993; Wingo et al., 2014). This suggests that the standard behavioral treatment approach is not effective for a subset of individuals with obesity, which may be partially attributable to the provision of a uniform treatment “package” to all participants. These uniform treatments do not account for individual differences, including psychological factors and eating behaviors (Brownell & Wadden, 1991).

Emotional eating (eating in response to negative mood states; Arnow et al., 1995) is a potential contributor to individual differences in body weight and responses to weight loss interventions. Although emotional eating is present in a proportion of individuals with binge eating disorder (BED; Greeno et al., 2000), there is also a distinct group of obese individuals who engage in emotional eating but do not meet criteria for BED (e.g., Fischer et al., 2007). High levels of emotional eating are associated with unhealthy weight and poorer weight loss outcomes and greater attrition from weight loss interventions in a variety of diverse samples (e.g., Byrne et al., 2003; Canetti et al., 2009; Cox et al., 2011; Keränen et al., 2009; Delahanty et al., 2013; Neve et al., 2010; Walcott-McQuigg et al., 2002; van Strien et al., 2012). Further, reductions in emotional eating are associated with increased weight loss in observational (Blair et al., 1990) and intervention studies (Teixeira et al., 2010), although the direct causal relationship between emotional eating and weight regulation remains unknown. Nevertheless, emotional eating represents an important behavior in individuals with obesity.

Although behavioral weight loss interventions typically address the role of emotions within the broader context of potential triggers for inappropriate eating, these interventions do not have an explicit, ongoing focus on specific strategies for understanding and modifying emotional eating patterns and their relationship with weight (e.g., see Brownell, 2004; Foster et al., 2005). Alternative strategies for increasing emotional awareness, emotion regulation and distress tolerance, and differentiating between emotional and physical cues for eating, such as those characteristic of mindfulness- and acceptance-based treatments [e.g., dialectical behavior therapy (DBT), Linehan, 1993; Telch et al., 2001; mindfulness-based eating awareness therapy (MB-EAT), Kristeller et al., 2006; acceptance and commitment therapy (ACT), Hayes et al., 1999] may be particularly helpful skills for improving the ability to modify weight and eating patterns in these individuals (see O’Reilly et al., 2014, for review). Two uncontrolled studies of women who were high in emotional or disinhibited eating demonstrated that these types of treatment approaches (i.e., DBT; ACT-informed behavioral weight loss) were associated with decreases in problematic eating and weight (Roosen et al., 2012; Niemeier et al., 2012). Randomized controlled trials of these approaches, or treatments which incorporate strategies consistent with these approaches, have produced reductions in binge eating (e.g., DBT for BED, Robinson & Safer, 2012; MB-EAT, Kristeller et al., 2014) and emotional eating (Robinson & Safer, 2012) in individuals with BED, as well as weight loss in individuals with obesity (i.e., ACT-informed behavioral weight loss, Forman et al., 2013). However, these studies did not directly test these effects in people who are high in emotional eating.

Three small randomized trials of women who eat in response to stress or negative emotions have compared an eating-focused mindfulness- or relaxation based intervention to a wait-list control (Alberts et al., 2012; Daubenmier et al., 2011) or inpatient weight loss program (Manzoni et al., 2008, 2009). Results of these studies were mixed, with two demonstrating greater decreases in emotional eating in the intervention versus control group (Alberts et al., 2012; Manzoni et al., 2008, 2009), but the other finding no group differences in emotional eating (Daubenmier et al., 2011). None of the studies demonstrated group differences in weight loss. These findings provide mixed evidence for the efficacy of mindfulness- or relaxation-based approaches for reducing emotional eating in women with obesity, but the degree to which these strategies provide benefits for weight loss is unclear. In addition, there have been no randomized trials of behavioral weight loss interventions specifically for individuals who emotionally eat.

The current investigation is a pilot study which evaluated the efficacy of a 20-week enhanced behavioral treatment for obesity (EBT), which was specifically developed for individuals who emotionally eat. This intervention integrated skills directly aimed at decreasing emotional eating (e.g., skills for observing, tolerating and effectively managing emotions and decreasing vulnerability to ineffective responses to emotions) with standard behavioral weight loss strategies. This pilot involved a small randomized trial to compare the efficacy of the new EBT to standard behavioral treatment (SBT) for obesity (Foster et al., 2010, “low fat” treatment), with the primary outcomes being changes in weight and emotional eating over 20 weeks. The hypothesis was that participants in the EBT group would experience significantly greater reductions in (1) weight and (2) emotional eating as compared to participants in the SBT group.

Methods

Study design

Participants were randomized to one of the two, 20-week weight loss interventions for individuals who emotionally eat (i.e., SBT or EBT) (Clinicaltrials.gov: NCT02055391). The study was approved by the University’s Institutional Review Board.

Eligibility criteria and recruitment

Inclusion criteria were an age of 21–65, a BMI ≥ 30 and ≤45, and a score within the top tertile on any of the three subscales of the Emotional Eating Scale (EES; Arnow et al., 1995). Because no normative data are available for emotional eating, tertiles were calculated based upon previously published collected data from a similar but independent sample of 217 obese participants seeking weight loss (Goldbacher et al., 2012). Emotional eating status was determined according to the following cutoffs, based upon those tertiles (Goldbacher et al., 2012): EES Anger score ≥23, EES Anxiety score ≥21, or EES Depression score ≥12. Exclusion criteria were: weight loss >10 lbs. within the previous 6 months; type 1 or type 2 diabetes; a history of cardiovascular, cerebrovascular, hepatic, or renal disease; protein wasting disease (i.e., Cushing’s syndrome); pregnancy or lactation; uncontrolled hypertension; medications affecting weight or energy expenditure; history of stroke or seizures; thyroid disease; a current major psychiatric disorder; history of/current diagnosis of Binge Eating Disorder or Bulimia Nervosa [based upon responses to the Questionnaire on Eating and Weight Patterns-Revised (Yanovski, 1993; contained within the Weight and Lifestyle Inventory (WALI); Wadden & Foster, 2006), as well as an investigator-created semi-structured clinical interview based on DSM-IV-TR criterion for BED or BN; American Psychiatric Association, 2000]; a total score ≥30 on the Beck Depression Inventory-II; Beck et al., (1988); current suicidal ideation; current or history of drug or alcohol dependence; current smoking or tobacco use; surgical or device intervention for obesity; history of non-skin malignancy within the previous 5 years; current mental health or weight loss treatment.

Participants were recruited from a major city in the Northeast and its immediate surrounding area via newspaper advertisements, flyers, and physician referrals between January, 2010 and August, 2010. Of the 518 participants who met initial eligibility criteria (based on a scripted phone screen), 298 completed an in-person interview to assess appropriateness for the study and to complete measures of weight, height and blood pressure. Participants also completed the BDI-II to screen for depression; the WALI (Wadden & Foster, 2006), which includes the Questionnaire on Eating and Weight Patterns (QEWP; Yanovski, 1993) to evaluate weight and eating history, eating patterns, and symptoms of binge eating disorder; and the Emotional Eating Scale (EES; Arnow et al., 1995). Informed consent was obtained from all individual participants included in the study. It was obtained prior to completing the interview and assessments, which were conducted by trained masters- or doctoral-level interviewers. Prior to randomization, participants submitted a physical form, completed by their physician, which was reviewed and approved by the study physician. Of the 298 individuals who attended the face-to-face assessment, 200 were excluded because they did not meet the eligibility criteria, four declined to participate, and 15 did not make contact following the face-to-face assessment (see Fig. 1). Following completion of the face-to-face screening visit described above, a computerized random number generator was used to randomly assign the remaining 79 participants to one of the two treatments (n = 39 in SBT, n = 40 in EBT), described below. The project coordinator implemented the randomization, enrollment and treatment assignment procedure. See Fig. 1 for further details regarding participant flow throughout the study.

Fig. 1
figure 1

Participant flow throughout the study

Full size image

Treatment conditions

Participants in both treatment conditions were provided with 90-min group sessions weekly for 20 weeks. The format of each group meeting was consistent across groups, with both groups including weekly weigh-ins, check-ins, new material, and written “skill builders” containing tasks to be completed for the upcoming week (see below for additional details). The group sessions were led by masters- and doctoral-level clinicians.

Standard behavioral treatment (SBT)

The SBT group received a behavioral weight loss intervention, based upon one that has been implemented successfully in previous weight loss trials (Foster et al., 2010). During the first week of treatment, participants were oriented to the treatment and instructed to maintain their typical dietary intake and physical activity patterns. Thereafter, they were provided a caloric intake goal of 1200–1500 kcal/day (women) or 1500–1800 kcal/day (men; Jensen et al., 2013). Physical activity goals were set beginning in week 4, and started at 20 min four times per week gradually increasing to an ultimate goal of 50 min four times a week. All participants were asked to monitor food and caloric intake and physical activity on a daily basis. Participants were also instructed in standard behavioral weight loss strategies, including self-monitoring of energy intake, stimulus control (i.e., identifying and modifying external cues for hunger such as timing, people and places), planning for high risk situations, behavior analysis, identifying and challenging unhelpful cognitions, and relapse prevention. Participants received written summaries of each week’s materials and instructions for setting and achieving goals between sessions, including written “skill builders” which assigned behavioral tasks to be completed between sessions. Participants did not monitor emotions or emotional eating episodes, and group leaders did not review or discuss emotions in any sessions, except within the context of one single week’s topic regarding how thoughts and feelings can affect general behaviors (a common topic for 1–2 weeks of SBT). The session format consisted of a review of the previous week’s skill builders and self-monitoring forms, introducing a new skill, reviewing strategies for implementing the new skill, problem-solving and goal setting, and discussing the next week’s skill builder.

Enhanced behavioral treatment (EBT)

The EBT intervention integrated the SBT program described above with techniques that specifically targeted emotional eating. Participants in this condition were provided with all of the same weekly techniques and information as those in the SBT condition (i.e., the behavioral weight loss intervention based upon Foster et al., 2010) and groups followed the same format described above. In general across all 20 weeks, this treatment condition had an additional explicit focus on understanding and monitoring emotional eating episodes, including emotions before and after each eating occasion, paying particular attention to precipitants, context and consequences (positive and negative) of emotional eating behaviors. With respect to the additional material, the first two weeks were spent reviewing a model of emotional eating and the process and rationale for tracking emotional eating episodes; the remainder of treatment (weeks 3–19) included additional skills to increase emotional awareness and regulation (e.g., mindful awareness of breath, mindful awareness of emotions), and improve the ability to tolerate or decrease negative experiences (e.g., “urge surfing”, self-soothing, relaxation) and make intentional choices (e.g., mindful eating), with the final week (week 20) spent reviewing and discussing all of these skills. See Table 1 for a list of topics covered in each weekly session. These skills were developed, in part, based on mindfulness meditation, relaxation, as well as components of dialectical behavior therapy (DBT; Linehan, 1993) and mindfulness-based cognitive therapy (MBCT; Segal et al., 2002). Exercises were integrated into each weekly meeting, with the previous weeks’ skills being reviewed at the beginning of each meeting and a new skill being taught and practiced as part of the remainder of the meeting. Participants were asked to practice the emotional-eating specific skills between meetings, and were provided with a CD recording of each emotional eating skill to aid in their practice. The CDs did not contain any information about the SBT skills.

Table 1 List of weekly topics for EBT group
Full size table

Measures

Assessments were conducted at baseline and week 20 of the intervention. Participants were compensated $25.00 for completing the assessments at baseline and week 20, as well as for a brief interim visit conducted at 12 weeks (for a possible total of $75.00 across the duration of the study). Individuals who could not attend the 20 week in-person assessment were given the option of completing the questionnaires and returning them via mail (n = 1 EBT; n = 3 SBT). The information from the 12 week visit was not a primary outcome; rather, it was used to facilitate the most complete modeling of data via multiple imputation (MI) in an effort to capture those individuals who did not complete the full 20 weeks.

Weight and height

Weight (in kg) was measured with a calibrated electronic scale with participants dressed in light clothing and without shoes. Height (cm) was measured using a wall-mounted stadiometer at the screening visit. BMI was calculated (weight in kg/height in m2). The primary outcome was weight at 20 weeks.

Emotional eating

The Emotional Eating Scale (EES; Arnow et al., 1995), a 25-item self-report questionnaire that measures one’s desire to eat in response to negative emotions, was used to evaluate emotional eating. The EES assesses a wider range of emotions than other self-report measures of emotional eating [e.g., Dutch Eating Behaviors Questionnaire (EBQ); van Strien et al., 1986], and has been demonstrated to correlate with laboratory measures of food intake in response to emotions (e.g., Schneider et al., 2012). The measure has three subscales assessing eating in response to: anger/frustration, anxiety, and depression. Items are scored on a 5-point Likert-type scale, with 0 = “no desire to eat” to 4 = “an overwhelming desire to eat.” A higher score indicates a stronger desire to eat in response to that category of emotions. This scale has demonstrated adequate reliability in obese binge eaters (Arnow et al., 1995). The total and subscale scores at 20 weeks were the secondary outcomes for this study.

Demographics

A demographic questionnaire, included in the WALI (Wadden & Foster, 2006), was used to assess age, race, gender, and highest level of education at the screening visit.

Statistical analyses

Power analyses

This was an initial, pilot investigation to explore effect sizes that can inform future trials.

Preliminary analyses

Differences between completers (i.e., those who completed the 20 week trial) and non-completers (i.e., those who did not complete the intervention but attended the 20 week assessment), as well as differences between groups at baseline were assessed by using independent samples t tests and Chi square tests for categorical variables

Main analyses

Analysis of Covariance (ANCOVA), as implemented in PROC MIXED (SAS v. 9.3, SAS Institute Inc., Cary, NC), was used to evaluate differences in change from baseline between treatment groups. First change scores were calculated by subtracting the values at week 20 from baseline for every subject. These change scores were then used as the outcome variables in the ANCOVA models with baseline scores as a covariate. The parallel slopes assumption was tested as an interaction effect between baseline and treatment group. To account for missing data, the data were assumed to be missing at random and two procedures in SAS/STAT software were used: PROC MI and PROC MIANALIZE. The MI procedure is a multiple imputation procedure that creates multiple imputed data sets for incomplete multivariate data. T Visual inspection of the pattern matrix revealed no evidence of univariate or monotone missing patterns; therefore, we assumed arbitrary missingness and used the Markov chain Monte Carlo (MCMC) method to create multiple imputations by drawing simulations from a Bayesian prediction distribution (Yuan, 2011; Hammer et al., 2012). The imputation model included three weight measures: baseline, intermediate and study end, and all the psychosocial variables at baseline and study end. There were no obvious missing data patterns; therefore, the Markov chain Monte Carlo (MCMC) method was used to create multiple imputations by drawing simulations from a Bayesian prediction distribution. The “imputers model” had three weight variables (from baseline, week 12 and week 20) and the baseline and week 20 emotional eating variables. Five data sets per model were imputed with MI and then MIANALYZE was used to combine results from the m complete data sets and generate valid statistical inferences for differences in the least squares (adjusted) and unadjusted treatment group means. Multiple regression analyses, controlling for baseline weight and treatment group, were conducted to examine the association between baseline levels of emotional eating and weight change from baseline through week 20. Completer analyses (defined as contributing weight or emotional eating data at week 20) were also conducted. Results were similar across both analytic approaches. An effect size was calculated using Cohen’s d for each group comparison (i.e., SBT vs. EBT). All statistical significance levels were set at p = 0.05, with two-tailed test.

Results

Sample characteristics

Participants were 79, predominantly female (95 %) individuals who endorsed high levels of emotional eating (according to criteria outlined above; M age = 45.6, SD = 10.5 years; BMI M = 36.2, SD = 4.1 kg/m2; 79.7 % African American). There were no significant group differences in age, race, gender, weight, or emotional eating scores at baseline (p’s > 0.10). Sample characteristics are reported in Table 2.

Table 2 Baseline characteristics of study participants
Full size table

Retention

Of the 79 enrolled participants, 55 (70 %) completed treatment at 20 weeks. An additional four individuals (one EBT; three SBT) completed the questionnaires via mail but did not present for a final weight at the 20 week assessment. The 20-week retention rate for the EBT group (n = 33, 82.5 %) was significantly greater than the SBT group (n = 22, 56.4 %; χ2 (1) = 6.35, p = 0.012). Completers and non-completers did not differ in age, race, baseline EES scores, or baseline weight (p’s > 0.10).

Primary outcomes

Body weight

Intention-to-treat (ITT) analyses, adjusted for baseline weight, demonstrated that participants in both groups achieved a significant weight loss at 20 weeks. The adjusted mean change (95 % CI) was −5.77 kg (−7.49, −4.04) in the SBT and −5.83 kg (−7.57, −4.09; within-group p’s < 0.001) in the EBT. Both treatments resulted in a relatively large effect on weight change (Cohen’s d = 1.07 in the SBT; Cohen’s d = 1.09 in EBT). There was no significant difference in weight loss between groups using the ITT (p = 0.96; Cohen’s d = 0.01; see Table 3). Results of the completers’ analysis (n = 22 SBT; n = 33 EBT) similarly demonstrated no significant between group differences in weight change, adjusted for baseline weight [(t (52) = 0.10, p = 0.92; Cohen’s d = 0.03; see Table 4].

Table 3 Imputed mean changes in weight and emotional eating from baseline to week 20 in full sample
Full size table
Table 4 Mean changes in weight and emotional eating from baseline to week 20 in completers only
Full size table

Emotional eating

ITT analyses, adjusted for baseline emotional eating scores, demonstrated significant decreases in all emotional eating scores in both groups; however, there were no significant between-group differences in changes in any of the emotional eating scores (see Table 3). Analyses conducted with completers (n = 25 SBT; n = 34 EBT) similarly demonstrated no significant between group differences in emotional eating score changes (see Table 4).

Exploratory analyses

Emotional eating and weight loss

Post-hoc analyses were conducted to examine whether baseline levels of emotional eating (total, anger, anxiety and depression) were associated with weight loss at 20 weeks. In those who completed the study (n = 55), a multiple regression analysis, controlling for baseline weight and treatment group, demonstrated that greater levels of baseline emotional eating-anxiety scores were associated with less weight loss at 20 weeks [β = 0.21, t (51) = 2.14, p = 0.037, R 2 = 0.11]. None of the other baseline emotional eating scores were associated with weight change. The ITT analyses yielded similar findings.

Attendance

Post-hoc analyses were also conducted to examine the effect of attendance on weight loss at 20 weeks for all individuals who completed the study and provided weight at 20 weeks (n = 55). The multiple regression analysis, controlling for baseline weight and treatment group, demonstrated that greater attendance was significantly associated with greater weight loss [β = −0.58, t (52) = −4.28, p < 0.001, R 2 = 0.28], such that for every session attended, there was an average 0.58 kg weight loss. Attendance accounted for 28 % of the variance in weight loss across both groups. Participants in the SBT condition attended an average of 10.95 (±6.38) groups, whereas participants in the EBT condition attended an average of 13.56 (± 5.57) groups over the course of the treatment; however, this difference did not reach significance (t (77) = −1.95, p = 0.055).

Discussion

This pilot study has several main findings. Participants in the SBT and EBT condition demonstrated significant decreases in weight and emotional eating after 20 weeks, with weight losses that were comparable to average 6-month losses for behavioral weight loss programs (~7–10 kg; Wadden et al., 2012) and among women presenting with similar forms of aberrant eating, such as disinhibited or stress-related eating (Niemeier et al., 2012; Manzoni et al., 2009). However, there were no differences between the SBT and EBT groups in changes in weight or emotional eating. This suggests that both interventions were successful at producing a moderate amount of weight loss and decreasing emotional eating in this sample of individuals who emotionally eat. While there are no previous randomized trials comparing weight loss interventions for this type of population, the absence of group differences in weight loss in the current study is consistent with findings from several brief randomized trials of weight loss interventions for women with other forms of problematic eating (Alberts et al., 2012; Daubenmier et al., 2011; Manzoni et al., 2009). However, a longer randomized trial with overweight women (not selected for problematic eating) provided some preliminary evidence that an acceptance-based weight loss intervention may be more effective for producing weight loss as compared to a SBT (Forman et al., 2013), suggesting that these types of skills may have a more noticeable effect after longer period of time. Thus, it is possible that mindfulness- and acceptance-based skills are more likely to facilitate successful outcomes in weight-maintenance, an area in which standard behavioral weight loss treatments have historically produced less successful outcomes. It would be important to examine the efficacy of “enhancing” standard behavioral interventions for weight maintenance with these types of skills to determine if this promotes better long-term outcomes in individuals who emotionally eat. With respect to emotional eating, previous research has shown that emotional eating in overweight samples decreases in response to brief- or moderate-length standard behavioral weight loss and mindfulness-based interventions (e.g., Keränen et al., 2009; Walcott-McQuigg et al., 2002), although there have been no randomized trials comparing these types of interventions for emotional eating (see O’Reilly et al., 2014 for review).

Inconsistencies in results across studies may partially be due to differences in study design (i.e., the use of an active control group, study duration) or sample (i.e., individuals specifically selected for emotional eating vs. community sample, participant ethnicity, sample size). For instance, although the findings of the current pilot study do not suggest that the EBT skills offered any additional benefit beyond those provided in the SBT condition over 20 weeks, it is possible that a longer trial or larger sample size may have produced different results. Importantly, the current study included a larger proportion of African-American women (79 %) than the previously described studies of similar interventions or samples (e.g., Niemeier et al., 2012; Forman et al., 2013; Alberts et al., 2012; Daubenmier et al., 2011), which may also account for some of the differences between study findings, particularly given potential differences in weight loss outcomes among African American women (Wingo et al., 2014). There is some evidence to suggest that eating in response to emotions such as feeling stressed, overwhelmed and lonely is a relatively common phenomenon in African-American women with obesity (James 2013), and stressors such as discrimination, occupational stress, and caregiving responsibilities contribute to emotional eating and weight control difficulties in African-American women (Befort et al., 2008; Walcott-McQuigg, 1995). Thus, additional research is needed to further examine culturally-specific factors that contribute to emotional eating in African-American women and incorporate culturally-tailored strategies into interventions for this population.

Alternatively, although the specific extra EBT techniques did not produce a greater decrease in emotional eating or weight loss beyond that which was achieved by the SBT skills, it is possible that other mindfulness- or acceptance-based techniques may have been helpful. For example, while the EBT included some components that are closely related to mindfulness and ACT, it did not include an explicit emphasis on other targets of these interventions (e.g., values clarification) that may be important for making sustained behavior changes in the face of difficult immediate experiences, such as emotions (Hayes et al., 1999; Forman et al., 2009, 2013). Notably, the current study held all aspects of SBT (including skills like cognitive restructuring) consistent across both groups, such that the EBT included all of the skills covered in SBT. Thus, it is plausible that the inclusion of these traditional SBT strategies in the EBT condition was sufficient for producing changes in weight and emotional eating, and the additional components of the EBT intervention offered no incremental benefit beyond these SBT skills. Future investigations of weight loss interventions for this population that measure these constructs and deconstruct the active components of weight loss treatments for individuals who emotionally eat may be warranted. Nevertheless, the current findings suggest that individuals who emotionally eat can decrease weight and emotional eating, and that standard behavioral weight loss skills (e.g., self-monitoring, stimulus control, calorie counting, behavior analysis) are effective for producing short-term decreases in weight and emotional eating in this population.

With respect to other predictors of weight loss, post hoc analyses demonstrated that higher levels of eating in response to anxiety at baseline was associated with poorer weight loss outcomes independent of treatment group. This is consistent with previous research on weight using self-report measures of emotional eating (e.g., Canetti et al., 2009; Keränen et al., 2009), and raises the possibility that this reflects eating in response to other experiences that were not specifically assessed in this study (e.g., stress, physiological arousal; James, 2013; Goldbacher et al., 2012). Thus, it may be useful for further research to examine whether there are specific aspects of emotional eating that are more strongly related to weight loss, and to continue to investigate alternative strategies for targeting individuals who eat in response to anxiety and related experiences. The current study also adds to the literature documenting a positive association between attendance and weight loss (e.g., Wadden et al., 2009; Fitzpatrick et al., 2014), in that attendance in the current sample accounted for approximately 28 % of the variance in weight change. Although there were no significant group differences in this association, dropout rates were significantly higher in the SBT (43.6 %) than EBT (17.5 %) condition, and there was some indication of a trend toward higher rates of attendance in the EBT (vs. SBT) condition, suggesting that the emotional-eating specific skills may have contributed to increased attendance and retention. It is plausible that these strategies were more readily applied to factors that may have served as barriers to attendance and treatment completion. However, the post hoc nature of these analyses makes it difficult to draw any definitive conclusions. Nevertheless, given documented inverse associations between emotional eating and retention (e.g., Neve et al., 2010; Walcott-McQuigg et al., 2002), and difficulties with attendance in weight-loss studies, including those conducted with African-American women (consistent with our sample demographics; e.g., Samuel-Hodge et al., 2009), future research may want to examine whether integrating alternative strategies for coping with negative emotions may improve attendance and retention in these samples.

This pilot investigation has several strengths. This is the first randomized study conducted exclusively with individuals who engage in emotional eating which compared a state-of-the art behavioral weight loss treatment to a new, enhanced treatment that integrated additional skills to specifically target emotional eating. In addition, the sample included a large number of African Americans, which is particularly important given the relative dearth of information related to emotional eating, weight loss, and mindfulness-based weight- and eating-interventions in heterogeneous samples (O’Reilly et al., 2014). This study also has several limitations. Because this was a pilot investigation, there was a relatively small sample size and fairly short duration. As such, future studies of individuals who emotionally eat may want to further investigate the effect of these type of skills in weight loss interventions with longer durations and larger sample sizes. Similarly, the limited scope of the study precluded the ability to test the mechanisms through which each intervention affected weight loss and emotional eating; this is an important question for future investigations. In addition, despite the strengths of the EES (Arnow et al., 1995), it is possible that the use of any self-report measure introduces some response bias, and there is new evidence to suggest that one might consider measuring other emotions not captured by the form of the EES that was available for use at the time that this study began (e.g., boredom; Koball et al., 2012). Thus, additional research is needed to further examine emotional eating using alternative forms of measurement. In addition, although the current study screened for the presence of BED and BN, the use of a standardized, semi-structured interview (such as the Structured Clinical Interview for DSM Disorders (SCID; First et al., 2002) would have helped confirm the absence of eating disorders in this sample. Finally, attrition in the SBT condition was relatively high, and significantly higher than the EBT group, although the SBT attrition rates are consistent with some previous behavioral weight loss research (Moroshko et al., 2011). While the intent-to-treat, multiple-imputation analytic strategy is a conservative statistical approach to handling this type of situation (Elobeid et al., 2009), participant drop-out may have still affected the ability to draw definitive conclusions about treatment effects or group differences.

Nevertheless, to date, this is the only randomized trial of a weight loss intervention specifically for individuals who emotionally eat. Results from this sample of individuals who emotionally eat (consisting largely of African-American women) suggest that adding skills to directly target one’s awareness of and ability to tolerate or modify emotions may not improve the short-term efficacy of a standard behavioral weight loss intervention for this type of population. Overall, the findings of the current pilot study indicate that individuals who emotionally eat are able to achieve significant reductions in weight and emotional eating when participating in a standard behavioral weight loss program over 20 weeks, but the longer-term effects remain unknown.