Toward an Understanding of the Distinctiveness of Body Image Constructs in Persons Categorized with Overweight/Obesity, Bulimia Nervosa, and Binge-Eating Disorder

Abstract

Objective:

We examined distinctiveness of different aspects of body-image disturbance in persons categorized with eating/weight disorders. We compared dissatisfaction with weight/shape, overvaluation of weight/shape, preoccupation with weight/shape, and fear of weight gain – in three study groups of persons categorized with overweight/obesity [O/O], bulimia nervosa [BN], and binge-eating disorder [BED] and examined how each body-image construct relates to clinical measures within and between the study groups.

Method:

1017 community volunteers completed measures of body-image, eating-disorder psychopathology, and depression. Participants were categorized into three study groups: O/O (N=511), BN-purging type (N=167), and BED (N=339).

Results:

Groups differed significantly on the four body-image constructs (medium-to-large effect sizes) with a consistent severity gradient with BN greater than BED greater than O/O. Both within and between groups, the four body-image constructs varied in strengths of association amongst themselves and with clinical measures. Analyses revealed considerable variability in variance accounted for in the clinical measures; distinctive significant patterns observed across the groups included: dissatisfaction with BMI, preoccupation and fear with eating concerns and restraint, and overvaluation with depression.

Conclusion:

Clinical manifestations of body-image disturbances are complex and show important differences across study groups defined as overweight/obesity, BN, and BED. Improved understanding of distinctions between different body-image constructs and their differential salience across different eating/weight disorders is needed to improve case conceptualization and treatment formulation.

Body image(s) comprise complex cognitive schema for how individuals perceive their appearance, which – in turn – can influence their behaviors and psychosocial functioning (Cash & Smolak, 2011). Most generally, dissatisfaction with body image is so widespread that it described for women as a “normative discontent” (Rodin, Silberstein, & Striegel-Moore, 1984). Dissatisfaction with weight/shape is observed across age, sex, racial/ethnic, and weight groups, although it tends to be more common in persons with overweight/obesity (Slevec & Tiggemann, 2011). Amongst eating disorders, disturbances in body-image are conceptualized as core cognitive feature and specific body-image constructs are required criterions for the diagnosis of anorexia nervosa (e.g., “Intense fear of gaining weight or of becoming fat”) and bulimia nervosa [BN] (“Self-evaluation is unduly influenced by body shape and weight”) (American Psychiatric Association, 2013). Body-image disturbance is not a requirement for another eating-disorder diagnosis - i.e., binge-eating disorder (BED) - although it has been suggested as a severity specifier (Grilo, 2013) given consistent findings that it signals greater severity (Grilo, Hrabosky et al., 2008; Grilo, Masheb, & White, 2010; Grilo, White, & Masheb, 2012; Hrabosky et al., 2007) and poorer outcomes (Grilo, Masheb, & Crosby, 2012; Grilo, White, et al., 2013; Sonneville et al., 2015).

There are a number of specific cognitive constructs that reflect different aspects of body-image disturbance (Lydecker, White & Grilo, 2017). Four closely related, yet conceptually distinct body image constructs, are included in the Eating Disorder Examination (Fairburn & Cooper, 1993), a well-established measure of the psychopathology of eating disorders (Berg, Peterson, Frazier, & Crow, 2012): dissatisfaction with weight/shape, overvaluation of weight/shape, preoccupation with weight/shape, and fear of weight gain. Dissatisfaction refers to negative feelings about one’s weight/shape. Overvaluation, which parallels the APA (2013) body-image criterion for BN, refers to when individuals’ self-evaluation is unduly based on their judgements of their own weight/shape. Preoccupation with weight/shape refers to when individuals spend excessive time thinking about their weight/shape and this interferes with their functioning (e.g., being distracted while engaged conversation because thinking about weight/shape). Fear of weight gain refers to an intense fear of gaining weight.

Research has supported the distinction between dissatisfaction and overvaluation across different weight groups (e.g., Masheb et al., 2006; Wade, Zhu, & Martin, 2011; see Grilo, 2013). Newly emerging research on body-image concerns have provided further evidence regarding the overlap and distinctiveness of specific body-image constructs and their specific relations to different eating behaviors and psychopathology (Blechert, Ansorge, Beckmann, & Tuschen-Caffier, 2011; Calugi & Dalle Grave, 2019; Lydecker, White, & Grilo, 2017; Mitchison et al., 2017; Molbert et al., 2018). Mitchison and colleagues (2017), in a study of three body-image constructs among adolescent high school students, found preoccupation was most strongly associated with dietary restraint and binge eating among girls, whereas dissatisfaction, overvaluation, and preoccupation showed similar associations with eating behaviors and psychopathology among boys. Lydecker and colleagues (2017), in a study of four body image constructs with a clinical treatment-seeking sample of 748 adults with BED, found that although constructs were related to one another and showed similarities in their associations with several biopsychosocial variables, the four constructs showed a number of important distinctions. Preoccupation was more strongly associated than other body image constructs with eating concerns and overvaluation was more strongly negatively associated with self-esteem. The four body image constructs were not significantly associated with body mass index (BMI) nor with frequency of binge eating in this group of patients with BED.

Collectively, emerging research (Blechert, Ansorge, Beckmann, & Tuschen-Caffier, 2011; Calugi & Dalle Grave, 2019; Linardon et al., 2018; Lydecker, White, & Grilo, 2017; Mitchison et al., 2017; Molbert et al., 2018) has indicated the importance of finer grained understanding of the complexity of body image disturbances and has added to the literature distinguishing between overvaluation and dissatisfaction with shape/weight (Wade et al., 2011) and attitudinal versus perceptual aspects (Molbert et al., 2018). Generalizability of the recent findings for treatment-seeking adults with BED (Lydecker, White, & Grilo, 2017) and adolescent students (Mitchison et al., 2017) or patients with anorexia nervosa (Calugi & Dalle Grave, 2019; Linardon et al., 2018) to different groups is unknown and represents an important gap in the literature. While studies have explored some differences in body-image constructs between bulimia nervosa and anorexia nervosa (Linardon et al., 2018), possible variations in these specific body image constructs are not yet understood across groups with binge-eating behaviors such as those characterized with bulimia nervosa and binge-eating disorder. Additionally, no study has examined the distinctiveness of these different aspects of body-image disturbance in persons characterized with different eating and weight features. Thus, the present study compared dissatisfaction, overvaluation, preoccupation, and fear of weight gain in persons categorized with overweight/obesity [O/O], bulimia nervosa [BN], and binge-eating disorder [BED] and examined how each specific body-image construct relates to various clinical measures within and between these three study groups.

Thus, this study aimed to extend the literature regarding the significance of different body-image constructs both by considering three study groups (O/O, BN, and BED) and by considering a broader range of clinical variables: physical (BMI), behavioral pathology (binge-eating frequency, purging), cognitive pathology (restraint, eating concerns), and depressive symptoms/levels. Understanding the distinctions between different body image constructs across different eating/weight disorders can inform future research and case conceptualization across different eating/weight disorders and their differential salience can eventually, if replicated and extended using more rigorous methods, refine treatment formulation by guiding targets for intervention. We hypothesize overall that the study group characterized as BN will have greater body image disturbances than the BED study group, which in turn, will have greater disturbances than the O/O study group. We also hypothesize that the four body image constructs will vary in their association with clinical measures and BMI; specifically, we hypothesize greater dissatisfaction will be association with greater BMI, and that the body image constructs will show differential patterns with other clinical measures.

Method

Participants

Participants for this study were 1017 individuals drawn from a larger series of 3,147 potential respondents who initiated an online (internet-based) survey of which 2,619 completed surveys allowing for the creation of the study groups based on the criteria described below (see “Creation of Study Groups”). Respondents failing to meet criteria for Study Groups were excluded, resulting in the final N=1017. Participants were respondents to online advertisements seeking volunteers aged 18 years or older for a research study consisting of internet-based surveys of weight, eating, dieting, and health-related behaviors. Participants included 149 (14.7%) males and 865 (85.3%) females (3 participants did not report gender); race/ethnicity was 79.3% (n= 803) White, 6.5% (n= 66) Hispanic, 6.6% (n=67) Black, 4.1% (n=42) Asian, 3.5% (n=35) “other” (4 did not report race/ethnicity). Mean age was 36.32 (SD=12.23) years. Mean BMI was 33.03 (SD=8.73) kg/m². Of the 1017 participants, 57.7% (n=586) met criteria for obesity (BMI ≥30) and 28.3% (n=288) met criteria for overweight (BMI ≥ 25 and < 30).

Procedures and Assessments

Advertisements were placed on the Craiglist website (an American advertisement internet site). Participants completed an anonymous online survey consisting of demographic information, self-reported height and weight (used to calculate body mass index (BMI; kg/m²)), and self-report questionnaires through SurveyMonkey, a secure online data-gathering platform. The study was approved by the university’s Human Investigations Committee. All participants provided informed consent.

Eating Disorder Examination-Questionnaire (EDE-Q) (Fairburn & Beglin, 1994) assesses the frequency of objective binge episodes (OBEs; defined as feeling a loss of control while eating unusually large quantities of food; this definition corresponds to the DSM-5 criteria for binge eating), subjective binge episodes (SBEs; defined as feeling a loss of control while eating although the quantities not being unusually large), and inappropriate or extreme weight control and purging methods over the past 28 days. The EDE-Q also assesses eating-disorder psychopathology in four domains scored as subscales (Restraint, Eating Concern, Shape Concern, and Weight Concern). In the current study, we examined the variables related to body dissatisfaction (weight dissatisfaction and shape dissatisfaction items), overvaluation (overvaluation of weight and overvaluation of shape items), preoccupation with weight or shape (single item), and fear of weight gain (single item) (see Table 2 footnote for items). Items are rated on a scale of 0 (“none”) to 6 (“marked”). The EDE-Q has good test-retest reliability in community (Mond, Hay, Rodgers, Owen, & Beumont, 2004) and clinical (Reas, Grilo, & Masheb, 2006) studies (Berg, Peterson, Frazier, & Crow, 2012), converges well with the EDE interview (Grilo, Masheb, & Wilson, 2001), and performs well in community studies particularly for purging behaviors (Mond, Hay, Rodgers, & Owen, 2007). It should be noted, however, that recent studies have not supported the factor-structure of the EDE-Q across various study groups (Grilo, Reas, Hopwood, & Crosby, 2015; Machado, Grilo, & Crosby, 2018); this potential limitation is important to bear in mind when considering our analyses.

Table 2.

Means and standard deviations of body image constructs and clinical variables for the three study groups.

	O/O n=511		BN n=167		BED n=339		ANOVA		ANCOVA
	M	SD	M	SD	M	SD	F	Post-hoca	Partial η2	Partial η2 Age	Partial η2 Race	Partial η2 Sex
Dissatisfaction with Weight and Shape	4.34	1.71	5.47	0.92	5.33	1.14	66.44*	O/O<BN, BED	.118	.121	.117	.107
Overvaluation of Weight and Shape	3.30	1.92	5.48	0.82	4.76	1.40	149.44*	O/O<BED<BN	.231	.224	.229	.220
Preoccupation with Weight and Shape	1.16	1.86	4.48	1.88	2.76	2.26	187.77*	O/O<BED<BN	.272	.230	.272	.267
Fear of Weight Gain	2.71	2.53	5.48	1.27	4.53	2.01	129.12*	O/O<BED<BN	.204	.184	.203	.195
Body Mass Index	33.61	7.00	29.66	9.89	33.80	10.03	15.27*	BN<O/O, BED	.029	.025	.029	.029
Binge-eating Frequency	0.00	0.00	14.46	14.12	9.67	7.24	346.89*	O/O<BED<BN	.406	.392	.406	.400
EDE-Q Eating Concerns	1.08	1.13	4.22	1.26	3.05	1.43	493.83*	O/O<BED<BN	.493	.475	.493	.485
EDE-Q Restraint	1.84	1.41	4.00	1.44	2.34	1.51	141.06*	O/O<BED<BN	.218	.201	.218	.211
BDI	12.34	9.08	25.03	10.99	18.58	9.96	107.30*	O/O<BED<BN	.192	.179	.196	.194

Note. Dissatisfaction variable assessed using items “How dissatisfied have you felt about your weight? …about your shape?; Overvaluation variable assessed using items “How your weight (Has your shape…) influenced how you think about (judge) yourself as a person?”; Preoccupation variable assessed using item “Has thinking about shape or weight made it much more difficult to concentrate on things you are interested in?; Fear variable assessed using item “Have you had a definite fear that you might gain weight or become fat?.

^*p<.0001.

O/O = Overweight/Obesity; BN = Bulimia Nervosa; BED = Binge-eating Disorder; EDE-Q = Eating Disorder

Examination-Questionnaire, BDI = Beck Depression Inventory.

^aScheffe post-hoc analyses.

Beck Depression Inventory (BDI) (Beck & Steer, 1987) is a well-established measure of depressive symptoms (Beck, Steer, & Carbin, 1988), and more generally, of negative affect and distress (Watson & Clark, 1984). The BDI taps a broad range of negative affect (for example, it correlates as strongly with anxiety as depression) and is a good marker for broad distress and severity (Beck, Steer, & Carbin, 1988) including BED and BN (Grilo, 2004; Grilo, Masheb, & Wilson, 2001b). In this study group, the BDI had excellent internal consistency with α = 0.92.

Creation of Study Groups

Three study groups were created: participants categorized with overweight or obesity (O/O), participants categorized with features characteristic of bulimia nervosa, purging type (BN), and participants categorized with features characteristic of binge-eating disorder (BED). We note that we categorized these three study groups and while they can be viewed as “probable” they were not diagnosed by clinical interview. The O/O group was created based on BMI≥25 and the absence of eating-disorder behavioral psychopathology operationalized as no binge eating (i.e., neither OBEs or SBEs) and no purging (vomiting, laxative use, or diuretic use). The probable BN and BED study groups were created using frequency criteria of once-weekly or greater for the key behavioral features determined by the EDE-Q for the previous 28 days. Thus, BED required a minimum frequency of once-weekly OBEs without any purging during the past 28 days whereas BN-purging type required a minimum frequency of once-weekly OBEs, once-weekly purging behaviors, and overvaluation of shape/weight (operationalized as four or greater on either EDE-Q item).

Note that the O/O and BED study group were not created with regard to inclusion/exclusion of the overvaluation item thus resulting in a natural unrestricted range of those scores. However, because the BN group required the presence of overvaluation of shape/weight, this creates a restricted range of scores above the clinical threshold for this one specific body-image variable. Thus, certain analyses described below (e.g., patterns of associations between overvaluation and other body-image constructs) may not be relevant within the BN group; they are, nonetheless, reported as they can speak to issues regarding the relevance of overvaluation for diagnoses of BN and BED (see Grilo, 2013).

Statistical Analyses

General linear model (GLM) analysis of variance (ANOVA) was used to compare the three study groups (O/O, BN, and BED) on demographic variables, the four body image constructs (dissatisfaction, overvaluation, preoccupation, and fear of weight gain), and the clinical measures (BMI, binge-eating frequency, EDE-Q Eating Concern and EDE-Q Restraint, and BDI Depression; note that we did not analyze with respect to the EDE-Q Shape Concern and Weight Concern scales because the four body image constructs were included in those scales). When ANOVAs indicated significant overall group differences, Scheffe post hoc tests (a conservative test) were performed to determine which specific groups differed significantly. A parallel series of analyses of covariance (ANCOVAs) was performed adjusting for significant demographic differences between the three study groups. Additionally, partial eta-squared (η²), an effect-size measure, was calculated; these values reflect the proportion of variance in the criterion measure accounted for by group membership (conventions for this effect-size measure are as follows: small (.01 – .09), medium (.10 – .24), and large (≥ .25)). We used partial eta-squared, rather than Cohen’s d for two reasons: (1) Cohen’s d, is based on raw means and is therefore not appropriate when there are covariates in the statistical model whereas partial eta-squared reflects the unique portion of the variance accounted for after adjusting for the covariates; and (2) partial eta-squared can be used when there are three are more than two groups (i.e., O/O, BN, BED).

Correlation coefficients were used to examine associations among the body image constructs and between each body image construct and the clinical measures (BMI, binge-eating frequency, EDE-Q Eating Concern, EDE-Q Restraint, BDI Depression, and – for the BN study group – purging frequency). These correlation coefficients were calculated within each of the three study groups and then the correlations were compared between study groups using Fisher’s r-to-z test (i.e., to test whether the associations differed in magnitude across O/O, BN, and BED study groups). Multiple regression analyses were performed separately for each of the three study groups (O/O, BN, and BED) testing the four body image constructs as independent variables as predictors of variance for each of the clinical measures.

Results

Of the 1017 participants, 511 (50.2%) were categorized with overweight/obesity without ED psychopathology, 167 (16.4%) with features of BN-purging type, and 339 (33.3%) with features of BED. Table 1 summarizes demographic variables across the three study groups and statistical tests of differences, including effect-size measures. ANOVA revealed a significant difference among the three groups on age (albeit one reflecting a small effect-size (partial η² = .009); Scheffe post-hoc tests revealed the O/O group (M = 37.25, SD = 12.54) was significantly older than the BN group (M = 33.92, SD = 11.77)). Chi-square tests of independence revealed the three groups differed significantly in distribution of sex; post-hoc paired-test testing indicated a significantly higher proportion of women in both the BN (93.4%) and BED (88.5%) groups relative to the O/O group (80.6%). Chi-square analysis of race (White versus non-White) revealed significant differences among the groups; post-hoc testing indicated a significantly higher proportion of White participants in the BED group (83.7%) than the BN (76.0%) and O/O (77.4%) groups.

Table 1.

Demographic variables for the three study groups categorized with overweight/obesity, probable bulimia nervosa – purging type, or probable binge eating disorder.

	O/O n=511		BN n=167		BED n=339		Test Statistic	p-value	Post-hoc	Effect Size
Age (M, SD)	37.25	12.54	33.92	11.77	36.09	11.83	4.10a	.017	O/O>BNb	.009c
Female (n%)	410	80.6%	156	93.4%	299	88.5%	20.63d	<.001	OO<BN, BED	.143e
Race (White) (n%)	394	77.4%	127	76.0%	282	83.7%	6.12d	.047	BED>OO, BN	.078e

Note. O/O = Overweight/Obesity; BN = Bulimia Nervosa; BED = Binge-eating Disorder.

^aANOVA,

^bScheffe post-hoc analyses,

^cPartial eta squared,

^dChi-square,

^ePhi.

Table 2 summarizes descriptive statistics and findings from GLM ANOVAs comparing the three study groups on the four body-image constructs, BMI, and the clinical measures. ANOVAs revealed significant differences among the three groups on the four body-image constructs reflecting medium effect sizes for three constructs (dissatisfaction, overvaluation, and fear) and a large effect size for one construct (preoccupation); ANCOVAs adjusting for age, race, and sex indicated these demographic variables did not attenuate the group differences as reflected in the nearly identical partial η² values. Scheffe post-hoc analyses revealed a consistent severity gradient for the body-image constructs, with BN greater than BED greater than O/O, except that dissatisfaction was lower in the O/O group than the BN and BED study groups, which did not differ significantly from each other.

For BMI, ANCOVAs revealed statistically significant overall differences but reflecting small effect sizes; ANCOVAs adjusting for age, race, and sex did not attenuate the differences. Scheffe post-hoc tests revealed that the BN (M = 29.66, SD = 9.89) group had significantly lower BMI than the BED (M = 33.90, SD = 10.03) and O/O (M = 33.61, SD = 7.00) study groups, which did not differ from each other.

For the clinical variables, ANOVAs revealed a consistent pattern of significa nt differences among the three study groups reflecting medium to large effect-sizes. ANCOVAs adjusting for age, race, and sex indicated these demographic variables did not attenuate the group differences as reflected in the nearly identical partial η² values. Scheffe post-hoc analyses revealed a consistent severity gradient for the clinical variables, with BN greater than BED, greater than O/O on binge-eating, EDE-Q scales (Eating Concern and Restraint), and BDI Depression scores.

Table 3 summarizes the correlations among body-image constructs shown separately for the three study groups. Correlations were all significant (at p < .001) within the O/O (r ranged .282 – .649) and BED groups (r ranged .263 – .580). Within the BN study group, correlations were significant (at p < .05) with each other except not for fear of weight gain with overvaluation. Note that correlations with overvaluation were also statistically not significant; however, since overvaluation is required for BN, the restricted range likely accounts for this.

Table 3.

Correlations among body image constructs shown separately for the three study groups.

	Dissatisfaction r	Overvaluation r	Preoccupation r	Fear r
Overweight/Obesity
Dissatisfaction	-	.649a	.282a	.412a
Overvaluation	.649a	-	.413a	.459a
Preoccupation	.282a	.413a	-	.452a
Fear of weight gain	.412a	.459a	.452a	-
Bulimia Nervosa
Dissatisfaction	-	.365b	.284a	.195b
Overvaluation	.365b	-	.165b	.129b
Preoccupation	.284a	.165b	-	.381a
Fear of weight gain	.195b	.129b	.381a	-
Binge-eating Disorder
Dissatisfaction	-	.580a	.263a	.331ab
Overvaluation	.580ac	-	.344a	.375a
Preoccupation	.263a	.344a	-	.381a
Fear of weight gain	.331ab	.375a	.381a	-

Note. Overweight/Obesity and binge-eating disorder: all correlations were significant (p < .001); bulimia nervosa: all correlations were significant (p < .05) except Fear of Weight Gain and Overvaluation (p > .05). Different subscripts are used to highlight correlations that differ significantly (p < .05 for difference between Z scores) among the body-image constructs between the O/O, BN, and BED groups.

Table 3 also summarizes findings of the Fisher’s r-to-z tests used to examine whether the correlations amongst the four body-image constructs differed between the O/O, BN, and BED study groups. We again note that these analyses may not be relevant for comparisons with the BN study group specifically involving the overvaluation construct since it was a requirement to be categorized with BN, thus creating a restricted range above the clinical threshold.

Dissatisfaction and overvaluation were correlated similarly in the O/O (r = .649) and BED (r = .580) study groups, and significantly higher than in the BN (r = .365) study group (perhaps due to the restricted range as BN required this criterion). Dissatisfaction and preoccupation correlations did not differ significantly between the three groups (r range .263 – .284). Dissatisfaction and fear were correlated in a graded fashion from O/O (r = .412) to BED (r = .331) to BN (r = .195); only the difference between O/O and BN groups was significant. Overvaluation and preoccupation correlations did not differ significantly between the O/O (r = .413) and BED (r = .344) groups but both were significantly higher than the BN group (r = .165). Overvaluation and fear correlations did not differ significantly between the O/O (r = .459) and BED (r = .375) groups but both were significantly higher than the BN group (r = .129).

Table 4 summarizes the correlations between the body-image constructs and the clinical measures separately within each of the three study groups. Correlations were all significant (at p < .001) within the O/O (except for preoccupation with BMI) and BED study groups (except for preoccupation and fear with BMI). Within the BN study group, correlations showed a variable pattern: preoccupation and fear were non-significant with BMI, dissatisfaction and overvaluation were non-significant with binge-eating frequency and with EDE-Q restraint, dissatisfaction was non-significant with BDI, and all four body-image constructs were non-significant with purging frequency. (We again note that the non-significant correlations with overvaluation is perhaps due to the restricted range since overvaluation above threshold was required for BN.)

Table 4.

Correlations among body image constructs and clinical variables shown separately for the three study groups.

	Dissatisfaction r	Overvaluation R	Preoccupation r	Fear r
Overweight/Obesity
Body Mass Index	.292***a	.204***a	.028a	.813***a
Binge-eating Frequency	--	--	--	--
EDE-Q Eating Concerns	.378***a	.485***a	.607***a	.444***a
EDE-Q Restraint	.211***a	.268***a	.367***a	.300***a
BDI	.388***a	.447***a	.311***a	.238***a
Bulimia Nervosa
Body Mass Index	.351***a	.196**a	−.008a	− .047b
Binge-eating Frequency	.016a	.106a	.292***a	.197**a
EDE-Q Eating Concerns	.222**a	.191**b	.647***a	.399***a
EDE-Q Restraint	.003b	.070b	.288***a	.296***a
BDI	.153b	.331***a	.175*a	.048b
Purging	.023	.050	.146	.088
Binge-eating Disorder
Body Mass Index	.305***a	.128*a	−.008a	− .016b
Binge-eating Frequency	.118*a	.140**a	.126*a	.182***a
EDE-Q Eating Concerns	.371***a	.397***a	.580***a	.416***a
EDE-Q Restraint	.128*ab	.211***ab	.387***a	.369***a
BDI	.288***ab	.362***a	.202***a	.099b

Note. EDE-Q = Eating Disorder Examination-Questionnaire, BDI = Beck Depression Inventory.

^*p < .05;

^**p ≤ .01;

^***p ≤ .001.

Different subscripts are used to highlight correlations that differ significantly (p < .05 for difference between Z scores) correlations between the body-image constructs and the clinical measures between the O/O, BN, and BED groups.

Fisher’s r-to-z tests, used to examine whether the correlations between the body image constructs and the clinical measures differed between the O/O, BN, and BED study groups, revealed a number of statistically significant differences also summarized in Table 4. Figure 1, using “correlation plots,” schematically depicts the correlations (shown in Table 4 alongside the significance testing) between the body image constructs and clinical measures. In Figure 1, the size of each circle reflects the strength of correlations between each of the four body-image constructs (x-axis) with each of the clinical variables (y-axis) separately for O/O, BN, and BED study groups. As summarized in Table 4 (and visually depicted in Figure 1), several salient significant and distinct patterns in the statistical findings were observed. Notably, fear was most strongly associated with BMI in the O/O group and was strongly associated with EDE-Q Eating Concern and EDE-Q Restraint similarly across all three groups. Preoccupation was most strongly associated with EDE-Q Eating Concern and to a lesser extent with EDE-Q Restraint similarly across all three groups; it showed, however, either weak or non-significant associations with behavioral (binge-eating, purging) and BMI variables. Dissatisfaction was strongly and similarly associated with BMI and EDE-Q Eating Concern across groups, and with BDI Depression in the OO and BED groups, but not in the BN group. Finally, overvaluation was strongly and similarly associated with BDI Depression across groups and with EDE-Q Eating Concern in O/O and the BED study groups but not the BN study group.

Figure 1.

Visual depiction of correlations between body image variables and clinical variables by study group.

Note. O/O=overweight/obesity, BN=bulimia nervosa, BED=binge-eating disorder. Circles depict the magnitude of significant correlations (see size legend); non-significant (NS) correlations are depicted by a dot and correlations that were not tested have been left blank.

Table 5 summarizes the multiple regression analyses evaluating contributing of the four body-image variables to the variance to each of the clinical variables (BMI, binge-eating frequency, EDE-Q Eating Concern, EDE-Q Restraint, and BDI Depression), performed separately for the three study groups (O/O, BN, and BED). The four body-image variables accounted for significant proportions – albeit highly variable – of the variance in each of the (dependent) clinical variables across the three groups, with the notable exception of non-significant findings for purging behaviors. Across the three study groups (O/O, BN, and BED), the four body-image variables accounted for 10.0% to 15.0% of variance in BMI, 4.3% to 10.8% of variance in binge-eating frequency, 14.0% to 21.7% of variance in EDE-Q Restraint, 42.9% to 45.8% of variance in EDE-Q Eating Concern, and 12.5% to 24.1% of variance in BDI. Inspection of the semi-partial correlations both between and within the three study groups revealed considerable variability in the amount of variance accounted for in the clinical variables. Semi-partial correlations had several significant salient and distinct patterns. Notably, dissatisfaction was strongly associated with BMI across groups, preoccupation and fear were both strongly associated with EDE-Q Eating Concern and EDE-Q Restraint across groups, and overvaluation was strongly associated with BDI Depression across groups.

Table 5.

Multivariable analyses of the four body image variables to the clinical features for the three study groups .

		Overweight/Obese			Bulimia Nervosa			Binge-eating Disorder
Clinical Variable	Body Image Construct	R2	Semi-partial R	Semi-partial R2	R2	Semi-partial R	Semi-partial R2	R2	Semi-partial R	Semi-partial R2
Body Mass Index
		10.0%	F(4, 485)=13.41, p<.001		15.0%	F(4, 160)=7.08, p<.001		11.1%	F(4, 323)=10.11, p=.001
	Dissatisfaction		.227*	.052		.310*	.096		.302*	.091
	Overvaluation		.053	.003		.096	.009		−.025	.001
	Preoccupation		−.025	.001		−.067	.004		−.050	.003
	Fear		−.112*	.013		−.119	.014		−.091	.008
Binge-eating Frequency
					10.8%	F(1,160)=4.86, p=.001		4.3%	F(4, 324) = 3.58, p=.006
	Dissatisfaction					−.101	.010		.028	.001
	Overvaluation					.085	.007		.040	.002
	Preoccupation					.243*	.059		.052	.003
	Fear					.094	.009		.115*	.013
EDE-Q Eating Concerns
		46.1%	F(4, 485)=103.68, p<.001		45.8%	F(4, 160)=33.84, p<.001		42.9%	F(4, 324)=60.86, p<.001
	Dissatisfaction		.052	.003		.003	.000		.112*	.013
	Overvaluation		.146*	.021		.060	.004		.070	.005
	Preoccupation		.395*	.156		.504*	.254		.417*	.174
	Fear		.096*	.009		.179*	.032		.142*	.020
EDE-Q Restraint
		16.5%	F(4, 485)=23.92, p<.001		14.0%	F(4, 160)=6.50, p<.001		21.7%	F(4, 324)=22.42, p<.001
	Dissatisfaction		.023	.001		−.107	.011		−.055	.003
	Overvaluation		.053	.003		.037	.001		.041	.002
	Preoccupation		.231*	.053		.202*	.041		.268*	.072
	Fear		.109*	.012		.216*	.047		.227*	.052
BDI Depressive Distress
		24.1%	F(4, 435)=34.46, p<.001		12.5%	F(4, 148)=5.28, p=.001		14.9%	F(4, 293)=12.84, p<.001
	Dissatisfaction		.139*	.019		.037	.001		.099	.010
	Overvaluation		.207*	.043		.275*	.076		.218*	.048
	Preoccupation		.143*	.020		.101	.010		.104	.011
	Fear		−.042	.002		−.003	.000		−.079	.006
Purging Frequency
					2.4%	F(4, 160)=1.00, p=.410
	Dissatisfaction					−.035	.001
	Overvaluation					.036	.001
	Preoccupation					.125	.016
	Fear					.029	.001

Note. EDE-Q = Eating Disorder Examination-Questionnaire, BDI = Beck Depression Inventory.

^*p < .05

Discussion

This study evaluated four different body-image constructs —dissatisfaction, overvaluation, preoccupation, and fear —in persons categorized with overweight/obesity, features of BN, and features of BED. The three study groups differed significantly on each of the four body image constructs (medium-to- large effect sizes) with a generally consistent severity gradient with BN greater than BED greater than O/O. Both within and between the three groups, the four body image constructs varied in their strengths of association amongst themselves and with clinical measures and BMI. The many significant associations between the body image variables and the core and associated features of eating/weight disorders highlight the relevance and importance of body image concerns for clinical assessment and formulation. In addition, our study provides new findings regarding the complexities of body image as reflected in the considerable variability in variance accounted for across clinical variables by the four body image constructs both within and between the three groups.

Overall, persons categorized with BN-purging type had greater body image disturbances and eating disorder pathology than persons categorized with BED who, in turn, had greater concerns than persons categorized with overweight/obesity. This severity gradient is generally consistent with the empirical literature (Friedman & Brownell, 1995; Grilo, Masheb, & White, 2010). Both within and between the three groups, the four body image constructs varied in their strengths of association amongst themselves and with clinical measures. Although most of the correlations were statistically significant, the significant differences in strengths (as reflected in Z scores) suggested important distinctions further supported by analyses revealing considerable variability in variance accounted for across the clinical variables by the four body image constructs both within and between the three study groups. Our findings here for BN=purging type and overweight/obesity are new while our findings for BED (based on a non-clinical community group) are nearly identical to those previously reported by Lydecker and colleagues (2017) for a treatment-seeking series of patients diagnosed with BED using structured and investigator-based diagnostic interviews. Salient distinctive patterns observed across the three eating/weight study groups included: body dissatisfaction was strongly associated with BMI, preoccupation and fear were strongly associated with eating concerns and restraint, and overvaluation was strongly associated with depression. A striking distinctive pattern observed between the BN and O/O groups concerned BMI; specifically, while dissatisfaction was correlated with BMI in both groups, fear of weight gain was very highly correlated in the overweight/obesity group but not in the BN group. The body image variables accounted for less variance in the frequency of purging in the BN study group, binge-eating frequency (in both BN and BED study groups), and BMI (all groups) suggesting that other factors play important roles in maintaining those features.

It is important to emphasize that because overvaluation of shape/weight was a required criterion for the BN study group, the lower correlations with other body image variables (r ranged .129 to .365) and the clinical variables reflects a restricted range. This finding is consistent with the DSM-5 requirement of overvaluation of shape/weight for the diagnosis of BN. Our findings for the BED study group that overvaluation was most highly correlated with eating concerns and depression but minimally with BMI and binge-eating frequency – along with magnitudes – are nearly identical to those reported for BED by Lydecker et al (2017).

Clinically, our findings, which require replication and extension in future studies using more rigorous assessments, suggest the potential utility for clinicians to assess for and to make these distinctions when performing assessments and formulating treatments for patients with these eating/weight concerns. From a cognitive-behavioral perspective (Fairburn, 2013; Fairburn et al., 2003), such conceptualizations are shared with patients at the start of treatment and serve as a template for guiding changes. The shared assessment and formulation serves to help patients understand better the factors that may contribute to maintaining their eating-related psychopathology. This, in turn, facilitates the processes of creating hypotheses to test and to guide behavioral “homework” assignments intended to normalize eating patterns while reducing maladaptive behaviors (Fairburn, 2013).

Our findings, if replicated, might suggest further refinements to psychological treatment models (e.g., Wilson, Grilo, & Vitousek, 2007) for assessing and addressing body-image constructs and their specific relations to patients’ presenting eating/weight concerns. For example, depressive and negative affect is a well-known trigger for overeating and binge-eating (Goldschmidt et al., 2012; Grilo, Shiffman, & Wing, 1989; Grilo, Shiffman, & Carter-Campbell, 1994; Kenardy, Arnow, & Agras, 1996). Our findings that depressive distress is strongly associated with overvaluation across all three groups, with dissatisfaction in the overweight/obese and BED groups, and with preoccupation in the overweight/obese group could suggest that clinicians help patients identify depression not just as a symptom but as a potential trigger. We note that these findings also have potential assessment value and clinical relevance for other interventions, including psychopharmacological methods. For example, McElroy and colleagues (2016) recently reported significant reductions in frequency of preoccupied and obsessive thoughts and binge eating in patients with BED treated with pharmacotherapy. A recent study by Wang and colleagues (2017) found that rumination, a cognitive process perhaps similar to preoccupation, was associated with greater eating disorder psychopathology and weight-bias internalization among patients with BED even after adjusting for overvaluation of shape/weight. Wang et al (2017) speculated that cognitive processes such as rumination might lead patients with obesity and BED to dwell more on weight-based discrimination and bias experiences and to internalize those attitudes. These diverse findings converge in highlighting the potential importance of greater attention to preoccupation with shape/weight in clinical assessment and in future research studies.

As context for the findings, we note potential strengths and weaknesses of the study. One strength includes the relatively large sample size, which allowed for group comparisons and fine-grained analyses. Our study utilized psychometrically established self-report measures of the body image constructs, features of eating disorders, and depression. We highlight our reliance on self-report rather than diagnostic interviews to establish diagnoses; thus, we emphasize that we compared three study groups - termed O/O, BN-purging type, and BED. Although our self-report measures cannot generate clinical diagnoses and may be biased, research has generally supported adequate convergence with validated interview methods (Berg et al., 2011; Berg et al., 2012). The body-image constructs were assessed with just one or two single items from the items from the EDE-Q. Future studies should consider more detailed or potentially rigorous assessments; Conversely, research has recently highlighted some particular strengths and advantages to using clear concrete single-item rather than multiple measures for complex constructs (Bergkvist, 2015; Bergkvist & Rossiter, 2007). We also note that the anonymity afforded by self-report to research volunteers may facilitate honest reporting of sensitive or embarrassing behaviors. Our findings were gathered online from community respondents interested in research and may not generalize to treatment-seeking individuals. Participants were primarily women (86%) and White (80%) and whether persons with different sociodemographic characteristics show similar patterns is unknown. Our findings, which pertain to community volunteers categorized with overweight/obesity, features of BN-purging type, and features of BED complement and extend those previously reported for treatment-seeking patients with BED (Lydecker et al., 2017) and for adolescent students (Mitchison et al., 2017). BMI and the overweight/obese category were calculated using self-reported height and weight. A large body of research has found that while individuals tend to underestimate weight and overestimate height, measured and self-reported values are highly correlated and the magnitude of actual differences tend to be small (Kuczmarski, Kuczmarski, & Nijjar, 2001). Research with persons with BN (Doll & Fairburn, 1998), BED (White, Masheb, Grilo, 2010), and obesity (Ivezaj & Grilo, 2017) has found that they are reasonably accurate reporters of weight/height and that the small magnitude of reporting errors is not systematically related to eating disorder psychopathology (Ivezaj & Grilo, 2017; White et al., 2010). Our study was cross-sectional precluding speculation regarding directionality or causality. Future studies should use prospective (e.g., Tabri, Murray, Thomas, Franko, Herzog, & Eddy, 2015) and experimental designs, including controlled treatment (e.g., Grilo, Masheb, & Crosby, 2012) to further understand their distinctiveness, directionality, and significance.

Highlights:

• Our findings highlight the importance of clarifying the complex nature of body-image concerns in persons with eating/weight disorders.

• Among a community sample of volunteers, the severity of different forms of body-image disturbance were highest in persons categorized with features of bulimia nervosa, followed by those with features of binge-eating disorder, and then by those with overweight/obesity.

• Both within and between the three study groups of persons with eating/weight concerns, different forms of body-image disturbance varied in their strengths of association amongst themselves and with other clinical measures.