Effects of a Cognitive-Behavioral Intervention Targeting Weight Stigma: A Randomized Controlled Trial

Abstract

Objective:

To test the effects of a cognitive-behavioral intervention for weight bias internalization (WBI; i.e., self-stigma) combined with behavioral weight loss (BWL).

Methods:

Adults with obesity and elevated WBI were randomly assigned to BWL alone or combined with the Weight Bias Internalization and Stigma Program (BWL+BIAS). Participants attended weekly group meetings for 12 weeks, followed by 2 bi-weekly and 2 monthly meetings (26 weeks total). Changes at week 12 on the Weight Bias Internalization Scale (WBIS) and Weight Self-Stigma Questionnaire (WSSQ) were the principal outcomes, with changes at week 26 assessed as secondary outcomes. Other outcomes included changes in mood, body image, eating behaviors, self-monitoring, and weight.

Results:

Seventy-two participants were randomized (84.7% female, 66.7% black, age=47.1±11.5 years) Linear mixed models showed no significant differences between the BWL+BIAS and BWL groups in WBIS changes at week 12 (−1.3±0.2 vs. −1.0±0.2) or week 26 (−1.5±0.2 vs −1.3±0.2). BWL+BIAS participants had greater reductions in WSSQ total scores at week 12 (p=0.03), with greater changes on the Self-Devaluation subscale at weeks 12 and 26 (p≤0.03). BWL+BIAS participants reported significantly greater benefits on measures of eating and self-monitoring. Percent weight loss at week 26 did not differ significantly between groups (BWL+BIAS=−4.5±1.0%, BWL=−5.9±1.0%, p=0.28).

Conclusion:

A psychological intervention for WBI produced short-term reductions in some aspects of weight self-stigma in persons with obesity.

Introduction

Individuals with obesity face strong societal stigma that includes experiences of teasing, bullying, discrimination, social rejection, or other instances of negative or unfair treatment due to weight (Tomiyama et al., 2018). Weight bias internalization (WBI) – also known as weight self-stigma – occurs when individuals with obesity absorb negative, weight-based societal perceptions, including stereotypes, and devalue themselves due to their weight (Pearl & Puhl, 2018). While experiencing weight stigmatization from others leads to negative health consequences (Tomiyama et al., 2018), the psychological and behavioral effects of WBI may be stronger than those of the stigmatizing experiences alone (Pearl & Puhl, 2018). WBI, for example, is associated with adverse mental and physical health outcomes, including depression, anxiety, binge eating, poor body image, reduced physical activity, and greater cardiovascular disease (CVD) risk (Pearl & Puhl, 2018). In addition, higher WBI in adults with obesity is associated with poorer long-term weight loss (Puhl, Quinn, Weisz, & Suh, 2017).

Preventing and reducing instances of weight stigma – for example, by changing public attitudes, providing legal protections against discrimination, and eliminating the social acceptability of weight bias – is a public health priority (Tomiyama et al., 2018). At the same time, interventions to help individuals with obesity cope with weight stigma and reduce or prevent its internalization are also of critical importance. Psychological interventions designed specifically to reduce WBI are lacking (Pearl & Puhl, 2018), with only one randomized controlled trial (RCT) to date (Palmeira, Pinto-Gouveia, & Cunha, 2017). That study tested an acceptance-based group program for weight self-stigma and eating behaviors, combined with individual medical/nutritional visits for weight loss, compared to the latter visits alone. The stigma-reduction intervention significantly improved self-stigma, quality of life, and health behaviors (Palmeira et al., 2017). Of note, weight loss across groups was minimal, and the control condition did not include group treatment (Palmeira et al., 2017).

In addition to acceptance-based approaches, cognitive-behavioral approaches – including skills such as cognitive restructuring and identifying the connections among thoughts, feelings, and behaviors – also may be beneficial for reducing self-stigma (Corrigan, Bink, & Schmidt, 2018; Pachankis, Hatzenbuehler, Rendina, Safren, & Parsons, 2015). Such interventions have been shown to improve psychological well-being and health behaviors in members of other stigmatized groups (such as sexual minorities or individuals with mental illness; Pachankis et al., 2015; Yanos, Lucksted, Drapalski, Roe, & Lysaker, 2015). These benefits may be attributable to the cognitive-behavioral skills learned by group members, as well as to group processes that increase feelings of belongingness and hope for stigmatized individuals (Yalom & Lescze, 2005). In addition, a small study of college students found that a cognitive intervention reduced negative attitudes toward people with obesity (Ciao & Latner, 2011), suggesting this approach may also be effective for reducing self-directed stigma.

Many of the stigma-reduction interventions described have used “identity affirming” approaches that aim to reduce shame and increase acceptance of one’s stigmatized identity (Corrigan et al., 2018; Pachankis et al., 2015). For obesity, which carries significant physical consequences independent of stigma (Jensen et al., 2014), we propose combining an identity-affirming intervention for reducing weight stigma with intensive behavioral weight loss (BWL). This combination diverges from prior studies that have compared the effects of weight-neutral approaches, which promote self-acceptance without weight loss, to weight loss approaches (Carels et al., 2014; Mensinger, Calogero, & Tylka, 2016). Instead, we wished to assess a seemingly contradictory yet dialectical treatment model for individuals with obesity: seeking to reduce the burden of weight-related public and self-stigmatization, while at the same time working toward goals to help patients change their eating, physical activity, weight, and health.

Intensive BWL treatment facilitates a 5–8% reduction in body weight in the short-term and improvements in CVD risk and quality of life (Heymsfield & Wadden, 2017). However, it may not produce substantial reductions in WBI or stigma-related distress (Mensinger et al., 2016; Pearl, Wadden, Chao, et al., 2018). Adults with obesity who seek weight loss tend to have greater psychological distress and higher WBI than those who do not engage in weight management (Friedman & Brownell, 1995; Puhl, Himmelstein, & Quinn, 2018). Adding a weight-stigma reduction intervention to BWL could potentially address this aspect of psychological distress that many persons with obesity report.

The current study investigated the effects of the Weight Bias Internalization and Stigma (Weight BIAS) Program, designed to reduce WBI and help individuals with obesity cope with weight stigma (Pearl, Hopkins, Berkowitz, & Wadden, 2018). The effects of the Weight BIAS program on WBI and other measures of psychological and physical health were tested in combination with standard BWL treatment (BWL+BIAS), compared to BWL alone. The primary hypothesis was that, after 12 weeks, participants in the BWL+BIAS intervention group would report greater improvements in WBI than participants who received BWL alone. We also predicted that participants in the BWL+BIAS group would report greater improvements in psychosocial well-being related to weight and in health behaviors, and that improvements would be sustained at 26 weeks. In addition, we explored differences between the two groups in changes in weight and CVD risk.

Methods

Participants

Participants were men and women, ages 18–65 years old, who were seeking weight loss and had obesity (body mass index [BMI] ≥ 30kg/m2). Participants were eligible if they reported a history of experiencing weight bias (e.g., teasing/bullying, discrimination, or other unfair treatment due to weight) and showed elevated levels of WBI, as indicated by a score of 4.0 or greater on the Weight Bias Internalization Scale (WBIS; Durso & Latner, 2008; described below). A cutoff score of 4 has been used in prior research to indicate “elevated” WBI (Pearl, Hopkins, et al., 2018) and may correspond to the 80–90th percentile of WBIS scores in the general population (Hilbert et al., 2014). Applicants had to confirm in an in-person interview, conducted by a psychologist, that their weight negatively affected how they felt about themselves. These criteria were used to ensure that participants had clinically-meaningful WBI that could benefit from our stigma-reduction intervention.

Exclusion criteria included: type 1 or 2 diabetes; uncontrolled hypertension (blood pressure ≥160/100 mm Hg); a cardiovascular event (e.g., stroke, myocardial infarction) in the past year; any major active kidney, liver, cardiovascular, or cerebrovascular disease; loss of ≥5% of initial weight in the past 6 months; use of medications that significantly affect weight; history of bariatric surgery; women who were nursing, pregnant, or planning to become pregnant; severe symptoms of mood (Beck Depression Inventory-II score ≥29, with clinician discretion), anxiety, or binge eating disorder (eight or more binge episodes per week), or any severity of bulimia nervosa or thought or substance use disorder; and current, active suicidal ideation and/or a suicide attempt within the past year. Participants were not eligible if they had participated in individual or group psychotherapy in the past 3 months (due to the potentially confounding effects of receiving a simultaneous cognitive-behavioral intervention), with the exception of participants receiving counseling for concerns unrelated to mood, self-esteem, or weight (e.g. career counseling or caregiver support). Participants taking anti-depressant medication that did not affect weight were eligible if the dose had been stable for at least 3 months.

Procedures

Participants were recruited by flyers, local media, and physician referrals for a study about weight loss and social experiences related to weight. All participants were screened by telephone and by a subsequent in-person behavioral evaluation, conducted by a psychologist, and a brief medical evaluation, conducted by a nurse practitioner. Prior to the in-person behavioral evaluation, participants completed the WBIS, Beck Depression Inventory-II (Beck, Steer, & Brown, 1996), and the Weight and Lifestyle Inventory (Wadden & Foster, 2006) to assess WBI, mood, weight and psychiatric history, and health behaviors. These topics were reviewed during the behavioral evaluation. Eligible participants were consented and randomized in a 1:1 ratio to one of two treatment groups (described below). Investigators and participants were not blinded to group assignments. Outcome assessments occurred at screening/baseline, week 12, and week 26. All procedures were approved by the institutional review board (IRB).

Intervention

Participants attended 90-minute group meetings held on weekday evenings. Groups of 11–13 participants were led by a psychologist or registered dietitian. Each cohort of participants consisted of two groups (BWL+BIAS and BWL alone) led by the same group leader. Participants received 12 weekly group sessions, followed by 2 every-other-week sessions and 2 monthly sessions (16 sessions over 26 weeks total). The number and timing of treatment sessions were based on established guidelines (Jenson et al., 2014). Participants who could not attend a given group session were offered a brief make-up session (in person or by telephone) with the group leader or a study staff member.

BWL group.

Participants were provided with 60 minutes of BWL treatment, based on the Diabetes Prevention Program and LEARN Program (Brownell, 2004; Diabetes Prevention Program Research Group, 2002). The treatment incorporated culturally-sensitive intervention recommendations (The Look AHEAD Research Group, 2006). A diet of 1200–1499 kcal per day was prescribed for participants < 250 lb, and 1500–1800 kcal for those ≥ 250 lb (Jenson et al., 2014; The Look AHEAD Research Group, 2006). Participants were instructed to eat a balanced diet, to eat regularly throughout the day, and to record their daily food and caloric intake. Weight was measured at every group session. Session topics during the first 12 weeks included self-monitoring, stimulus control, social support, portion sizes, and goal-setting. Standard BWL content on negative thoughts, emotional eating, and body image was also discussed in single sessions, respectively. Promotion of the “thin ideal” was avoided by focusing on health instead of appearance and on modest versus drastic weight loss. Group sessions during weeks 13–26 focused on skills required for weight loss maintenance and relapse prevention.

Physical activity was prescribed at a level consistent with data showing that >250 minutes per week is associated with improved long-term weight loss (Jakicic, Marcus, Lang, & Janney, 2008). Activity prescriptions began at week 2 and gradually progressed to a goal of 150 minutes per week by week 12, and 200–250 minutes per week by week 26. Moderate intensity was prescribed, with an emphasis on walking.

In the standard BWL group, an additional 30 minutes was devoted to discussing recipes and food preparation. A recipe exchange was facilitated, in which each participant was asked to share with the group a healthy recipe for a meal or snack. The recipe discussion served as an educational component of the BWL program that allowed for equal time spent in group sessions across conditions, without giving additional weight loss counseling to the standard BWL group.

BWL+BIAS group.

These participants received the same BWL program described above (without the recipe discussion), combined with a stigma-reduction intervention. In each session, following 60 minutes of BWL, 30 minutes were devoted to a stigma-reduction intervention tested in a previous open-label pilot study (Pearl, Hopkins et al., 2018). Content was adapted from cognitive-behavioral therapy and “third-wave” therapies such as dialectical behavior therapy and acceptance and commitment therapy. Session topics included: psychoeducation about weight and weight bias; challenging myths/stereotypes and cognitive distortions related to weight; the relationship between thoughts, feelings, and behaviors; restructuring negative thoughts and reappraising stigmatizing situations; interpersonal effectiveness skills; increasing self-efficacy; reducing self-criticism; and increasing empowerment, self-compassion, and body and self-acceptance. Acceptance content drew from mindfulness principles of accepting oneself in the present moment without judgment, while allowing for the dialectical possibility of change (Linehan, 2014; e.g., “I can accept and love myself as I am right now, while also working to change my health behaviors and lose weight to improve my health”). Group leaders were trained to address tensions that may arise between weight loss and self-acceptance by validating that these two goals may seem contradictory and encouraging participants to consider how they might achieve both simultaneously (e.g., trying to lose weight without hating oneself and one’s body in the process).

The effects of weight bias (experienced and internalized) on health behaviors were discussed, with an emphasis on helping participants overcome stigma-related barriers to weight management. For example, participants were given strategies to cope with anticipated stigma while being physically active in public spaces (e.g., in a gym), as well as to challenge self-critical beliefs (e.g., that they are lazy) that may lead them to avoid engaging in physical activity. All of the stigma-related topics were presented as skills intended to help participants adhere to their healthy eating and physical activity goals, as well as to improve how they felt about themselves.

Outcome Measures

Primary outcome.

Participants completed the WBIS at screening, week 12 (primary endpoint), and week 26. The WBIS includes 11 items rated on a 1–7 scale (scores averaged, with higher scores indicating greater WBI; Durso & Latner, 2008). Items address weight-related stereotypes (e.g., “I am less attractive than most other people because of my weight”) and self-devaluation (e.g., “I hate myself for being overweight”). The WBIS has strong psychometric properties (Durso & Latner, 2008) and is the most widely used measure of WBI (Pearl & Puhl, 2018). In the current sample, internal consistency was lower than typical at screening (Cronbach’s α=0.65), although consistency improved at weeks 12 and 26 (α’s=0.84).

The 12-item Weight Self-Stigma Questionnaire (WSSQ) was included as a secondary measure of WBI, with items rated from 1–5 and summed, producing a total score and two subscales: Fear of Enacted Stigma (or anticipated experiences of stigma; 6 items; e.g., “People discriminate against me because I’ve had weight problems.”) and Self-Devaluation (6 items; e.g., “I became overweight because I’m a weak person.”; Lillis, Luoma, Levin, & Hayes, 2010). This scale also has strong psychometric properties and is the second most common measure of WBI in the literature (Pearl & Puhl, 2018). In the current sample, internal consistency for the total score was strong (α=0.72, 0.82, and 0.86 at baseline and weeks 12 and 26, respectively), with slightly weaker consistency for the Self-Devaluation compared to the Fear of Enacted Stigma subscale (α=0.66, 0.75, and 0.72 versus 0.77, 0.78, and 0.85). The Fat Phobia scale was also included as a measure of weight stereotype endorsement; participants were presented with pairs of adjectives used to describe people with obesity (e.g., fast versus slow) and used a differential rating scale (1–5) to indicate whether or not they agreed with negative weight stereotypes (α=0.88, 0.91, and 0.92, respectively; Bacon, Scheltema, & Robinson, 2001).

Secondary outcomes.

Participants completed the Impact of Weight on Quality of Life Questionnaire-Lite (IWQOL-Lite; α=0.92, 0.93, and 0.95), which assesses weight-specific aspects of psychological and physical functioning (Kolotkin, Crosby, Kosloski, & Williams, 2001). They also completed: the Patient Health Questionnaire (PHQ-9) to assess symptoms of depression (Kroenke & Spitzer, 2002; α=0.80, 0.80, and 0.85); the Generalized Anxiety Disorder-7 questionnaire (Spitzer, Kroenke, Williams, & Lowe, 2006; α=0.86, 0.91, and 0.93); and the 10-item Perceived Stress Scale (PSS; Cohen & Williamson, 1988; α=0.87, 0.87, and 0.86). The Body Appreciation Scale was included as a measure of body esteem, with items such as “I respect my body” (Avalos, Tylka, & Wood-Barcalow, 2005; α=0.88, 0.93, and 0.93). Other exploratory outcomes can be found in the Supplementary Materials.

Behavioral outcome measures included changes in self-efficacy, as assessed by the Weight and Lifestyle Efficacy-Short Form (WEL; Ames, Heckman, Grothe, & Clark, 2012) and Self-Efficacy for Exercise Scale (SEES; Resnick & Jenkins, 2000). The WEL assesses confidence in one’s ability to overcome challenges that lead to overeating (e.g., “I can resist overeating when others are pressuring me to eat.”), while the SEES assesses confidence to overcome barriers to physical activity (e.g., “the weather was bothering you”). Both scales have strong psychometric properties (WEL α=0.88, 0.88, and 0.92; SEES α=0.91, 0.89, and 0.93).

Changes in self-reported eating were assessed with the Eating Inventory Questionnaire, which includes subscales for dietary restraint, disinhibition, and hunger (Stunkard & Messick, 1985). Dietary restraint assesses behaviors that are encouraged in weight management programs for controlling weight (e.g., “I count calories as a conscious means of controlling my weight.”; α=0.72, 0.79, and 0.79); disinhibition assesses loss of control over eating (e.g., “Sometimes when I start eating, I just can’t seem to stop.”; α=0.73, 0.76, and 0.79); and hunger assesses perceptions of physical hunger (e.g., “I am always hungry enough to eat at any time.”; α=0.82, 0.86, and 0.82). Food records were collected from all participants to identify the number of completed days (i.e., recorded at least two meals per day). Participants also reported their frequency of self-weighing and tracking their food/drink intake, calories, and physical activity.

Duplicate measures of height (with a wall-mounted stadiometer; Veeder-Root, Elizabethtown, NC) were obtained at screening. Weight was measured at screening (with a digital scale; Detecto, model 6800A), baseline (week 1), and weeks 12 and 26. Percentage reduction in baseline weight was calculated at the latter two times. Systolic and diastolic blood pressure were measured in duplicate at screening and weeks 12 and 26 using an automated Dinamap monitor (Johnson & Johnson, XL model 9300) at 1-minute intervals after ≥ 5 minute rest. Waist circumference was measured in duplicate at screening and weeks 12 and 26 to the nearest 0.1 cm with a flexible tension-controlled measuring tape midway between the iliac crest and lowest rib.

Safety and Treatment Acceptability

Study oversight was provided by a Data and Safety Monitor who was not involved in the study. The principal investigator and members of the study team prepared a report and met every 6 months with the Monitor. Changes in participants’ health were assessed at each study contact. Adverse events (AEs) were documented by study staff in consultation with a nurse practitioner and physician, and serious adverse events (SAEs) were reported immediately to the study sponsor, Monitor, and IRB.

Participants in both groups rated (1–7) the BWL treatment, as well as the stigma intervention or recipe exchange, on how helpful they found each component of the program, how much they liked the components, and how much they learned new skills or information from each component. Scores for these three items were averaged separately for the BWL treatment, stigma intervention, and recipe exchange.

Statistical Analyses

Analyses were conducted with the intention-to-treat (ITT) principle. Chi square and analysis of variance (ANOVA) were used to determine whether participants differed on baseline characteristics. Measures with missing items were prorated (all measures had ≤15% missing items). Skewness and kurtosis were examined for continuous outcome measures at all time points, and variables were transformed appropriately to meet assumptions of normality. Linear mixed models were used to compare changes between groups at weeks 12 and 26 on all continuous outcome measures, as well as to examine within group changes over time. Model shape and variance-covariance structure were selected based on model fit criteria (e.g., −2 log likelihood). Piecewise models with a breakpoint at week 12 best fit the data for all outcomes. Based on prior findings, we predicted a 1-point difference on the WBIS at week 12, the study’s primary outcome (i.e., reductions of 1.5 and 0.5 in the intervention and control groups, respectively, with a pooled standard deviations of 1.0; Mensinger et al., 2016; Pearl, Hopkins, et al., 2018; Pearl, Wadden, Chao, et al., 2018). Power analysis with G*Power 2.0 indicated that a sample size of 72 with 20% attrition would give us 95% power to detect significant differences (with a medium effect size) between groups on the WBIS at week 12 with an alpha = 0.05.

In a post-hoc analysis, correlations were used to explore the potential relationship between changes in weight stigma outcomes and percent weight change. All analyses were conducted with SPSS 25 and tested analyses with a significance level of p<0.05. Cohen’s d was computed to determine the effect sizes of between-group differences.

Results

Participant Recruitment and Retention

Figure 1 presents the trial CONSORT diagram. Of the 72 randomized participants, 7 did not attend any group sessions. Thirty BWL+BIAS and 31 BWL participants completed assessments at week 12 and 26 (overall retention rate=84.7%). Among participants who attended at least one group session, 83.7% of group meetings or makeup sessions were completed (mean number of completed sessions=13.4, standard deviation=4.1, range from 1 to 16 sessions; 81.6% attended 10 or more treatment sessions; 47.7% attended all 16 sessions). Attendance did not differ significantly between treatment conditions (86.5% in BWL+BIAS group, 81.1% in BWL group, p=0.40). Of the groups attended, 79.0% of attendance came from group meetings (77.7% in BWL+BIAS group and 80.1% in BWL group, p=0.56), with the remainder from makeup visits completed in person (17.4%) or by phone (3.7%).

Figure 1.

CONSORT diagram.

BMI=Body mass index; WBIS=Weight Bias Internalization Scale; BWL+BIAS=behavioral weight loss combined with Weight Bias Internalization and Stigma Program; BWL=behavioral weight loss alone.

Participants’ Baseline Characteristics

Tables 1 and 2 present demographic and other baseline characteristics for all randomized participants. Participants were predominantly black, female, middle-aged, had class II or III obesity (i.e., BMI of 35–39.9 or ≥40 kg/m2), and had completed approximately 15 years of education. Participant characteristics did not differ significantly between treatment conditions. Participants who did not attend a group meeting or did not complete the week 12 assessment had, on average, 2 fewer years of education than did participants who completed the 12 assessment (see Supplemental Table 1). No other differences were found between participants who did and did not attend at least one group session or complete the week 12 assessment.

Table 1.

Participants’ baseline characteristics, N (%) or Mean ± Standard Deviation

Variable	Total (N=72)	BWL+ BIAS (n=36)	BWL (n=36)	p
Age (years)	47.1±11.5	47.7±11.4	46.6±11.8	0.69
Sex				0.33
Female	61 (84.7%)	32 (88.9%)	29 (80.6%)
Male	11 (15.3%)	4 (11.1%)	7 (19.4%)
Race				0.13
White	21 (29.2%)	7 (19.4%)	14 (38.9%)
Black	48 (66.7%)	26 (72.2%)	22 (61.1%)
Asian	2 (2.8%)	2 (5.6%)	0
Multiracial*	1 (1.4%)	1 (2.8%)	0
Hispanic/Latino/a	6 (8.3%)	4 (11.1%)	2 (5.6%)	0.39
Education (years)	14.9±2.1	14.8±2.2	15.1±2.0	0.61
Weight (kg)	109.8±22.6	112.4±24.0	107.2±21.2	0.34
Height (cm)	166.8±8.5	167.0±8.7	166.6±8.5	0.86
Body Mass Index (kg/m2)	39.3±6.1	40.1±6.5	38.4±5.6	0.25
Waist Circumference (cm)	117.0±14.1	119.1±15.3	114.9±12.6	0.21
Systolic blood pressure+	127.7±10.3	127.4±10.2	128.0±10.7	0.50
Diastolic blood pressure+	73.7±8.5	73.0±9.2	74.3±7.9	0.26

Note. All variables were reported or measured at participants’ screening visits. Education was missing for one participant in the BWL+BIAS group.

^*Participant did not know specific racial background.

⁺Systolic and diastolic blood pressure p values reflect comparisons controlling for use of blood pressure medication at screening.

Table 2.

Participants’ baseline values on key outcome measures, Mean ± Standard Deviation

Variable	Total	BWL+ BIAS	BWL	p
WBIS	5.1±0.7	5.0±0.7	5.2±0.7	0.33
WSSQ Total	38.6±6.7	37.9±6.6	39.3±6.9	0.39
WSSQ-SD	19.2±4.1	19.1±4.2	19.2±3.9	0.93
WSSQ-FNE	19.4±4.7	18.8±4.6	20.1±4.8	0.26
Fat Phobia Scale	3.7±0.6	3.7±0.7	3.7±0.6	0.54
PHQ-9	8.0±4.8	7.5±5.1	8.4±4.5	0.45
GAD-7	5.4±4.3	5.6±4.5	5.2±4.2	0.67
Perceived Stress Scale	17.3±6.4	17.6±6.2	16.9±6.5	0.66
Body Appreciation Scale	2.6±0.7	2.8±0.6	2.5±0.7	0.11
IWQOL-Lite Total	55.3±16.5	56.3±16.7	54.2±16.5	0.61
WEL-SF	40.0±17.5	39.4±17.4	40.7±17.9	0.75
SEE	51.2±20.0	48.3±20.1	54.2±19.9	0.22
Eating Inventory
Dietary Restraint	8.7±3.7	8.5±3.7	8.9±3.8	0.67
Disinhibition	9.7±3.3	9.2±3.5	10.2±3.1	0.19
Hunger	7.1±3.7	7.8±3.6	6.4±3.6	0.10
Self-Reported Weighing	0.9±1.2	0.9±1.2	0.9±1.1	0.83
Track Food/Drink	0.5±0.8	0.4±0.7	0.5±0.8	0.70
Track Calories	0.5±1.0	0.4±1.0	0.6±1.1	0.32
Track Activity	1.2±1.5	1.0±1.5	1.3±1.5	0.38

Note. Total N=72 for the Weight Bias Internalization Scale (n=36 per group). N=71 for all other values (n=36 in BWL+BIAS group and n=35 in BWL group). WBIS=Weight Bias Internalization Scale; WSSQ=Weight Self-Stigma Questionnaire; SD=Self-Devaluation; FNE=Fear of Enacted Stigma; PHQ-9=Patient Health Questionnaire-9, summed scores, 0–27, higher scores indicate more symptoms of depression; GAD-7=Generalized Anxiety Disorder-7, summed scores, 0–21, higher scores indicate greater anxiety; PSS=Perceived Stress Scale, 10-item version, summed scores, 0–40, higher scores indicate greater perceived stress; Body Appreciation Scale: 10 items rated 1–5, scores averaged, higher scores indicate greater body appreciation; IWQOL-Lite=Impact of Weight on Quality of Life-Lite, items rated from 1–5, scores are transformed to a 0–100 scale, with higher scores indicating better quality of life. SEE (Self-Efficacy to Exercise Scale) and WEL-SF (Weight Efficacy Lifestyle Questionnaire – Short Form): 8 and 9 items, scores summed 0–80 or 0–90, respectively; higher scores indicate greater self-efficacy; Eating Inventory items include both true/false and rating scales, with all scores converted to 0 or 1 ratings and summed; Self-reported weighing rated 1–5 (less than once per month through several times per day); other self-monitoring/tracking behaviors rated from 0–4 (from never to everyday).

Weight Stigma

Table 3 (and Supplemental Figures 1 and 2) present estimated changes in WBIS and WSSQ scores (total and subscales), respectively. Participants across groups showed significant reductions on all weight stigma measures. However, changes in WBIS scores did not differ significantly between the BWL+BIAS and BWL groups at week 12 (−1.3±0.2 vs. −1.0±0.2) or week 26 (−1.5±0.2 vs −1.3±0.2). BWL+BIAS participants reported significantly greater reductions in WSSQ total scores than did BWL participants at week 12, with differences in the same direction at week 26. This effect appeared to be driven primarily by differences in the Self-Devaluation subscale at weeks 12 and 26. Fear of Enacted Stigma scores did not differ at week 12 or 26.

Table 3.

Estimated mean change (± standard error) in psychosocial outcomes from baseline to week 12 and week 26

Variable	BWL+BIAS (n=36)	BWL (n=36)	Mean Difference	p	d
WBIS
Week 12	−1.3±0.2***	−1.0±0.2***	−0.3±0.2	0.22	0.31
Week 26	−1.5±0.2***	−1.3±0.2***	−0.2±0.3	0.45	0.16
WSSQ Total
Week 12	−6.0±1.4***	−1.6±1.3	−4.4±1.9	0.03	0.58
Week 26	−7.1±1.5***	−3.6±1.5*	−3.5±2.1	0.11	0.35
WSSQ-SD
Week 12	−3.6±0.8***	−1.0±0.8	−2.8±1.1	0.02	0.48
Week 26	−4.1±0.8***	−1.5±0.8	−2.6±1.1	0.03	0.45
WSSQ-FNE
Week 12	−2.3±0.8**	−0.7±0.8	−1.6±1.2	0.19	0.33
Week 26	−2.9±0.9**	−2.1±0.9*	−0.8±1.3	0.53	0.13
Fat Phobia Scale
Week 12	−0.5±0.1***	−0.2±0.1	−0.3±0.2	0.12	0.31
Week 26	−0.5±0.1***	−0.2±0.1	−0.3±0.2	0.09	0.33
PHQ-9
Week 12	−3.4±0.8***	−2.5±0.8***	−0.9±1.1	0.40	0.16
Week 26	−2.9±0.8***	−2.9±0.8***	−0.03±1.1	0.98	0.01
GAD-7+
Week 12	−1.8±0.8**	−0.7±0.8	−1.1±1.1	0.32	0.26
Week 26	−1.6±0.8**	−0.8±0.8	−0.8±1.2	0.16	0.30
Perceived Stress Scale
Week 12	−3.2±1.2*	−0.1±1.2	−3.0±1.7	0.08	0.34
Week 26	−3.9±1.2**	−1.0±1.2	−2.9±1.7	0.10	0.32
Body Appreciation Scale
Week 12	0.6±0.1***	0.3±0.1**	0.3±0.2	0.13	0.41
Week 26	0.8±0.1***	0.5±0.1***	0.3±0.2	0.13	0.32
IWQOL-Lite Total
Week 12	8.3±2.1***	8.9±2.0***	−0.6±2.9	0.84	0.05
Week 26	12.5±2.1***	14.1±2.1***	−1.7±3.0	0.58	0.13

Note. WBIS=Weight Bias Internalization Scale; WSSQ=Weight Self-Stigma Questionnaire; SD=Self-Devaluation; FNE=Fear of Enacted Stigma; PHQ-9=Patient Health Questionnaire-9; GAD-7=Generalized Anxiety Disorder-7; IWQOL=Impact of Weight on Quality of Life.

⁺GAD-7 scores underwent logarithmic transformation to fit assumptions of normality; statistics for GAD-7 scores are from linear mixed models, and raw change values are shown. Asterisks indicate significance of within group changes from baseline. Significant between-group differences are highlighted in bold.

^***p≤0.001

^**p≤0.01

^*p<0.05

Psychosocial and Behavioral Outcomes

Across both groups, significant improvements were observed for most outcomes, including depression, body image, and quality of life (see Table 3 for means and p values). No differences were found between groups in improvements on any psychosocial outcome (see Supplemental Table 2 for additional outcomes).

Participants across groups showed significant improvements in self-efficacy to control eating but not to exercise (see Table 4). Improvements in self-monitoring behaviors were found in both groups, with significantly greater self-reported improvements in tracking of physical activity in the BWL+BIAS vs. BWL group at week 12. BWL+BIAS participants also reported significantly greater reductions in hunger at weeks 12 and 26 compared to BWL participants. Adherence to recording food intake was good overall, as judged by other studies (Burke et al., 2011), and did not differ between groups.

Table 4.

Estimated mean change (± standard error) in behavioral outcomes from baseline to week 12 and week 26

Variable	BWL+BIAS (n=36)	BWL (n=36)	Mean Difference	p	d
WEL-SF
Week 12	11.3±3.1***	10.2±3.1***	1.1±4.4	0.80	0.05
Week 26	12.0±3.1***	8.9±3.1**	3.1±4.4	0.49	0.14
SEE
Week 12	4.0±4.6	−5.7±4.5	9.7±6.5	0.14	0.28
Week 26	5.8±4.6	−6.3±4.5	12.1±6.5	0.06	0.35
Eating Inventory
Dietary Restraint
Week 12	6.3±0.7***	5.9±0.7***	0.4±1.0	0.69	0.11
Week 26	6.4±0.8***	5.2±0.8***	1.2±1.1	0.29	0.21
Disinhibition
Week 12	−2.8±0.6***	−1.9±0.6**	0.9±0.8	0.28	0.27
Week 26	−3.0±0.6***	−2.2±0.6**	−0.8±0.8	0.34	0.20
Hunger
Week 12	−3.2±0.5***	−1.4±0.5*	−1.8±0.8	0.02	0.48
Week 26	−3.9±0.5***	−1.4±0.5*	−2.5±0.8	0.001	0.66
Days of Food Records
Week 12	52.9±3.4***	55.5±3.4***	2.6±4.8	0.59	0.11
Week 26	85.6±9.5***	95.8±9.2***	−10.2±13.1	0.44	0.20
Self-Reported Weighing
Week 12	0.9±0.2***	1.1±0.2***	−0.2±0.3	0.46	0.16
Week 26	1.0±0.2***	0.6±0.2**	0.4±0.3	0.23	0.25
Track Food/Drink+
Week 12	2.7±0.3***	2.0±0.3***	0.7±0.4	0.12	0.38
Week 26	1.6±0.3***	1.6±0.3***	0.1±0.4	0.90	0.03
Track Calories+
Week 12	2.0±0.3***	1.7±0.3***	0.3±0.4	0.42	0.20
Week 26	1.7±0.3***	1.4±0.3***	0.3±0.5	0.36	0.18
Track Activity+
Week 12	1.6±0.3***	0.8±0.3**	0.8±0.4	0.03	0.56
Week 26	1.3±0.3***	0.9±0.3**	0.5±0.4	0.29	0.21

Note. WEL-SF=Weight Efficacy Lifestyle Questionnaire-Short Form; SEE=Self-Efficacy to Exercise Scale.

⁺Track Food/Drink and Activity scores underwent logarithmic transformation to fit assumptions of normality, and Track Calories scores were transformed with the square root; statistics for these variables are from linear mixed models, and raw change values are shown. Asterisks indicate significance of within group changes from baseline to time point. Significant between-group differences are highlighted in bold.

^***p≤0.001

^**p≤0.01

^*p<0.05

Weight Loss and CVD Risk Factors

At week 12, BWL+BIAS and BWL participants lost a mean of 3.7±0.6% and 4.6±0.6% of baseline weight respectively, which increased at week 26 to 4.5±1.0% and 5.9±1.0%. Groups did not differ significantly at either time (see Table 5). Significant improvements in systolic and diastolic blood pressure were observed across groups at week 12 but not week 26. Changes in CVD risk factors did not differ between groups. In addition, percent weight change did not correlate significantly with changes in WBIS, WSSQ total, and WSSQ subscale scores across groups at weeks 12 or 26.

Table 5.

Estimated mean change (± standard error) in weight and cardiovascular risk factors from baseline to week 12 and week 26

Variable	BWL+BIAS (n=36)	BWL (n=36)	Mean Difference	p	d
Percent Weight Change
Week 12	−3.7±0.6***	−4.6±0.6***	−0.9±0.8	0.28	0.22
Week 26	−4.5±1.0***	−5.9±1.0***	1.4±1.4	0.31	0.26
Waist Circumference (cm)
Week 12	−4.0±0.9***	−4.0±0.9***	0.0±1.3	0.98	0.01
Week 26	−4.3±0.9***	−5.1±0.9***	0.8±1.3	0.54	0.15
Systolic Blood Pressure (mm Hg)+
Week 12	−3.4±1.6*	−5.4±1.6***	2.0±2.2	0.37	0.18
Week 26	−2.6±1.6	−2.4±1.6	−0.2±2.2	0.94	0.01
Diastolic Blood Pressure (mm Hg)+
Week 12	−2.8±1.2*	−5.6±1.1***	2.8±1.6	0.09	0.31
Week 26	−1.4±1.2	−2.1±1.1	0.7±1.6	0.67	0.08

Note. Percent weight change was calculated from week 1 weights; all other values were measured at screening.

⁺Blood pressure analyses control for whether or not participant took blood pressure medication at any time during the study.

^***p≤0.001

^**p≤0.01

^*p<0.05

Safety and Treatment Acceptability

Supplemental Table 3 presents all AEs reported by ≥5% of participants. One SAE occurred during the trial (hypokalemia). The SAE and AEs were not considered to be related to the treatment provided.

Treatment acceptability ratings for the Weight BIAS program were high (6.3±1.0 out of 7). Acceptability ratings did not differ across groups for the BWL component of the program (BWL+BIAS=6.5±0.9, BWL=6.3±1.0, p=0.32), showing consistency across groups. Recipe exchange ratings were approximately 1 point lower than for other components of the program (5.4±1.5), suggesting that it was not a particularly potent component of treatment, as intended.

Discussion

This is the first RCT of which we are aware to test the effects of a combined weight stigma-reduction and behavioral weight management intervention, compared to behavioral weight management alone. Participants in both groups reported substantial improvements in weight bias internalization, as measured by the WBIS, at both weeks 12 and 26, with no differences between groups. At week 12, however, participants who received the combined intervention reported significantly greater improvements on the WSSQ total score than did their counterparts in BWL alone. They also reported significantly greater reductions on the WSSQ self-devaluation subscale at weeks 12 and 26. Some studies have examined differences between the WBIS and WSSQ (Hubner et al., 2016), but more research is needed to understand the distinctions between these two measures of WBI. Reductions in the WSSQ self-devaluation subscale are particularly important, because this scale captures individuals’ tendencies to blame and denigrate themselves for their weight. The combined intervention sought to help participants to both understand and decrease this, at times, relentless self-blame. These findings are consistent with a prior RCT and an open-label pilot study of acceptance-based stigma reduction interventions, which found significant improvements in WSSQ scores (Levin, Potts, Haeger, & Lillis, 2018; Palmeira et al., 2017).

We had not expected BWL alone to produce such large reductions on the WBIS, which were 1.0 and 1.3 at weeks 12 and 26, respectively, compared with 1.3 and 1.5, respectively, for BWL+BIAS participants. Previous studies of behavioral weight control have yielded reductions in WBIS scores of approximately 0.5 points, which we used to power the present study (Mensinger et al., 2016; Pearl, Wadden, Chao, et al., 2018). Two factors may have contributed to the larger-than-expected reductions observed in our study. First, as noted, our participants were required at screening to have a WBIS score ≥ 4, in order to recruit a sample more likely to have clinical distress concerning their WBI. This requirement resulted in the present sample having a higher mean WBIS score at baseline (5.1 points) than in many past weight loss studies (recent means of 3.7–4.3; Mensinger et al., 2016; Pearl, Wadden, Chao, et al., 2018). Thus, past studies may have been limited by a potential floor effect in detecting improvements in WBIS scores.

A second possible explanation is that the process of changing health behaviors and losing weight may have a greater effect on WBI for individuals with elevated levels, since the ability to make these changes inherently challenges weight stereotypes that they have internalized (e.g., being lazy or lacking willpower). Non-specific aspects of the BWL program, such as the group format to reduce participants’ feeling alone in their struggles with weight and increase social support (from other group members and study staff), may also have contributed to reductions in WBI without the specific targeted intervention. Consistent with some but not all past research, changes in WBI did not correlate significantly with changes in weight, suggesting that losing more weight does not necessarily facilitate reducing WBI more and vice versa (Lillis, Thomas, Olson, & Wing, 2019; Palmeira et al., 2017; Pearl, Wadden, Chao, et al., 2018). Future studies which dismantle the components of the BWL and Weight BIAS programs may be useful for understanding which aspects of treatment are most beneficial for patients with elevated WBI.

Participants across both groups achieved clinically meaningful weight losses (of approximately 4.5–6.0%) by week 26. The lack of significant differences in weight loss between groups challenges proposals that reducing weight stigma and promoting body acceptance may impede motivation for people with obesity to improve their health (Snook, Hansen, Duke, Hackney, & Zhang, 2018). The results also did not support the hypothesis that a stigma-reduction intervention may enhance weight loss. These results must be interpreted in the context of several considerations. First, weight loss was a secondary outcome, and this study was not powered to detect these differences. Second, given that participants who received BWL alone also showed reductions in WBI, it is possible that a larger dose of the Weight BIAS program was needed to surpass the benefits of BWL alone. Due to the deeply ingrained nature of internalized weight beliefs, more sessions over a longer period of time may be needed to achieve a sufficient reduction in WBI to affect downstream health and weight outcomes. Further, given the robust weight loss effects of short-term intensive BWL treatment (Heymsfield & Wadden, 2017), the potential effects of a stigma-reduction intervention, combined with BWL, may not emerge until patients face the challenge of weight loss maintenance, when weight regain is common (Jenson et al., 2014). Longer studies with more stigma-reduction intervention sessions would help to clarify these potential questions pertaining to dose-response and long-term outcomes.

Improvements in other outcomes, such as depression, anxiety, body image, and quality of life were also observed across the two groups but did not differ between them. Improvements on these measures were comparable to or greater than those observed in weight loss studies (Pearl, Wadden, Tronieri, et al., 2018) and studies testing weight-neutral programs based on the Health at Every Size (HAES) approach (Mensinger et al., 2016; Ulian et al., 2018). HAES is similar to our Weight BIAS program in its focus on combating myths and misperceptions about weight and reducing self-blame (Bacon, 2010), although WBI is not specifically addressed in the HAES curriculum (Mensinger et al., 2016). However, HAES differs in its focus on intuitive eating and promoting health behavior change without weight loss (Bacon, 2010), while our BWL intervention encouraged modest weight loss through caloric restriction (the Weight BIAS Program content did not directly promote weight loss, though it did acknowledge and support patients’ BWL goals). It is likely that the HAES approach may be suitable for specific patient populations, and weight loss interventions may be more appropriate for others (e.g., individuals with class II or III obesity and/or CVD risk factors; Ulian et al., 2018). Treatment acceptability ratings were high for the Weight BIAS and BWL programs, suggesting that combining weight loss with stigma-reduction was complementary rather than contradictory.

Participants across the two groups reported significant improvement in eating self-efficacy, eating behaviors (e.g., disinhibited eating), and self-monitoring behaviors. BWL+BIAS participants showed significantly greater improvements in tracking of physical activity at week 12 and in hunger at weeks 12 and 26. It is possible that more differentiation between groups in self-efficacy and health behaviors may emerge in the long-term. For example, participants who learned to challenge negative weight stereotypes (e.g., that they are lazy or lack willpower) may have more positive self-regard and self-efficacy when they encounter barriers to long-term maintenance (such as small weight regain) than participants who have not challenged such stereotypes and may be vulnerable to self-devaluation with weight regain. Thus, participants with skills to reduce WBI may be more likely to stick with their behavioral goals in the long-term.

Limitations of the current study include the short-term (3-month) follow-up from the end of the 12-week program, and inclusion only of participants with elevated WBI. Adults who are treatment-seeking, have severe obesity, and/or report greater psychological distress tend to have higher levels of WBI than community samples and those with lower BMIs or without psychopathology (Pearl & Puhl, 2018). Thus, results from the current study could be applicable to a significant proportion of US adults with obesity engaged in weight management. However, the effects of the Weight BIAS program on a general treatment-seeking sample (with varying levels of WBI) are unknown and warrant testing. Weight identity was the primary focus of the stigma-reduction intervention, but future studies might explore issues of intersectionality in this context. This study also was not adequately powered to detect secondary outcomes of psychological well-being, health behaviors, weight loss, and CVD risk reduction. Replication with a larger sample and long-term follow-up is needed to confirm and clarify the potential benefits of combining this stigma-reduction intervention with intensive BWL treatment.

Public Health Significance Statement:

This study demonstrates the efficacy of a cognitive-behavioral weight stigma-reduction intervention, combined with behavioral weight loss.