Teasing History, Onset of Obesity, Current Eating Disorder Psychopathology, Body Dissatisfaction, and Psychological Functioning in Binge Eating Disorder
Yale Psychiatric Research, Yale University School of Medicine, 301 Cedar Street, P.O. Box 208098, New Haven, CT 06519. E-mail: email@example.com
Objective: The primary goal of this study was to examine associations among teasing history, onset of obesity, current eating disorder psychopathology, body dissatisfaction, and psychological functioning in women with Binge Eating Disorder (BED).
Research Methods and Procedures: Subjects were 115 female adults who met DSM-IV criteria for BED. Measurements assessing teasing history (general appearance [GAT] and weight and size [WST] teasing), current eating disorder psychopathology (binge frequency, eating restraint, and concerns regarding eating, shape, and weight), body dissatisfaction, and psychological functioning (depression and self-esteem) were obtained.
Results: History of GAT, but not WST, was associated with current weight concerns and body dissatisfaction, whereas both GAT and WST were significantly associated with current psychological functioning. Patients with earlier onset of obesity reported more WST than patients with later onset of obesity, but the groups did not differ significantly in GAT, current eating disorder psychopathology, body dissatisfaction, or psychological functioning. Obese women reported more WST than non-obese women, but no differences in GAT or the other outcome variables were observed. Higher frequency of GAT was associated with greater binge frequency in obese women, and with greater eating restraint in non-obese women.
Discussion: Although physical appearance teasing history is not associated with variability in most eating disorder psychopathology, it is associated with related functioning, most notably body dissatisfaction, depression, and self-esteem. Our findings also suggest that the age of onset of obesity and current body mass index status in isolation are not associated with eating psychopathology or associated psychological functioning in adult patients with BED.
Teasing has been identified as a potential component in the formation of attitudes regarding personal appearance (1). Early experiences of being teased may have direct, even long term influences on eating psychopathology (e.g., eating disturbances and body dissatisfaction) and overall psychological functioning (2,3,4). Thompson and colleagues (5) posited that teasing about general appearance as well as teasing specific to weight and size may influence eating psychopathology.
To examine the two targets of teasing, Thompson and colleagues (5) developed the Physical Appearance-Related Teasing Scale (PARTS) to measure individuals’ past experiences of being teased about their weight and size (WST) and physical appearance (GAT). To date, findings regarding the association of WST and GAT to eating psychopathology have been mixed. Thompson and colleagues (5) found that WST was more strongly correlated with higher levels of body dissatisfaction and eating disturbance than GAT in a sample of college students. Another study with college students (6) found that higher frequency of WST and GAT were both significantly correlated with greater body dissatisfaction. In a study of treatment-seeking obese individuals, Grilo and colleagues (7) found that WST, but not GAT, was significantly correlated with body dissatisfaction.
In a community-based case-control design, Fairburn and colleagues(8) examined teasing history as a potential risk factor for binge eating disorder (BED). They ascertained a number of putative risk factors, including childhood teasing, in four groups of females: BED (n = 52), bulimia nervosa (BN; n = 102), general psychiatric controls (n = 102), and healthy controls (n = 104). This design allowed for identification of general risk factors for psychiatric disturbance and specific risk factors for eating disorders (BED and BN). Repeated exposure to negative comments from family members about weight, shape, or eating emerged as one of the few major risk factors that differentiated the BED group from both the psychiatric controls and the healthy controls. In contrast, teasing and bullying about nonappearance or nonweight-related topics did not emerge as a significant risk factor. Teasing about weight, shape, and eating has also been identified as a major risk factor that differentiated women with BN from psychiatric controls or healthy controls (9). Although this study highlighted the unique role of weight- and shape-related teasing as a risk factor for BED and BN, the relationship between teasing and variability in the eating disorder psychopathology, body dissatisfaction, and associated psychological features of BED has not been examined.
In addition to examining teasing in relation to the core and associated features of BED, it may be important to ascertain the timing of the onset of obesity per se. Following Stunkard's (10) initial observations regarding the potential negative sequelae of the onset of obesity during childhood, surprisingly few studies have examined whether early vs. late onset is associated with different patterns of eating disorder psychopathology. Marcus and colleagues (11) reported that early onset of obesity is associated with BED, but the association of early vs. late onset of obesity to variability in specific eating disorder psychopathology was not examined. Among obese persons, two reports (12,13) found that early onset of obesity was associated with greater eating disorder psychopathology, but this finding was not unequivocal (14).
Because teasing appears to be associated with body image dissatisfaction and eating disturbances and has been identified as a potential risk factor for BED, the primary goal of this study was to examine associations among teasing history, current eating disorder psychopathology (i.e., binge frequency, eating restraint, and concerns about shape, weight, and eating), body dissatisfaction, and associated psychological functioning (i.e., depression and self-esteem) in a clinical sample of women with BED. We also examined whether teasing history, current eating disorder psychopathology, body dissatisfaction, and psychological functioning differed with respect to age of onset of obesity (i.e., childhood, adolescent, or adulthood) or current body mass index (BMI) status (i.e., obese or non-obese).
Research Methods and Procedures
Subjects included 115 female adults who were evaluated for outpatient clinical trials and met DSM-IV (15) criteria for BED. Subjects were 21 to 61 years of age (mean = 41.28; SD = 9.57). 88% (N = 101) were white, 7.8% (N = 9) were African American, 2.6% (N = 3) were Hispanic, and 1.7% (N = 2) were self-identified as “other.” A total of 56% (N = 63) were married. All participants completed high school, and 50% were college-educated. Written informed consent was obtained from participants. Mean current BMI (weight in kilograms divided by height in square meters) for the sample was 34.70 (SD = 9.05), and the average reported age of first becoming overweight was 15.47 (SD = 8.81) years.
DSM-IV (15) diagnoses were derived by consensus and based on the independent administration of the Structured Clinical Interview for DSM-IV–Axis I Disorders (SCID-I (16)) and a clinical interview by trained and monitored Ph.D.-level clinicians. The SCID-I interviews were performed by three experienced Ph.D.-level research clinicians with training in the SCID-I and eating disorders. The three SCID-I interviewers for this study were evaluated as part of a larger inter-rater reliability study involving 12 interviewers (17). Inter-rater reliability (calculated using 84 pairs of raters)—as reflected by kappa coefficients—for eating disorder diagnoses was 0.77 for all raters and was 1.0 for the BED diagnosis for the three interviewers in the present study. Diagnoses were further confirmed by relevant portions of the following two self-report measures, which also provided eating disorder psychopathology data:
The Eating Disorder Examination-Questionnair (EDE-Q)
The EDE-Q (18) is the self-report version of the EDE (19), an investigator-based structured interview for the assessment of the core and associated psychopathology of eating disorders. The EDE-Q is comprised of 38 questions based directly on the EDE. The EDE-Q assesses the frequency of the objective binge eating episodes and four subscales: dietary restraint, eating concern, weight concern, and shape concern. The dietary restraint subscale is an admixture of cognitions and behaviors pertaining to dietary restriction. The three other subscales reflect dysfunctional attitudes regarding eating and overvalued ideas regarding weight and shape. The EDE-Q has a number of important strengths relative to other self-report inventories, including clear definitions of specific behavioral and cognitive components of eating disorders and the use of a consistent time frame (20). The EDE-Q has been empirically validated with BED (21,22) and BN (18,23).
The Questionnaire for Eating and Weight Patterns, Revised (QEWP-R)
The QEWP-R (24), employed in the DSM-IV field trials (25), assesses each criterion of BED, including the stipulated 6-month duration. The QWEP-R also includes historical variables relevant to this study, including age at which first overweight (defined as being overweight by at least 10 lbs as a child or 15 lbs as an adult). Historical data (e.g., age of first becoming overweight) was obtained during a clinical interview to confirm the QWEP-R findings.
In addition, the following psychometrically sound self-report measures were administered to assess teasing and associated domains:
The Physical Appearance-Related Teasing Scale (PARTS)
The PARTS (5) is an 18-item measure of the frequency of being teased while growing up. The PARTS has two subscales: weight/size teasing (WST) and general appearance teasing (GAT). Higher scores reflect higher frequencies of reported teasing experiences. Thompson and colleagues (5) reported that the PARTS demonstrated adequate reliability: the WST subscale had an internal consistency of 0.91 and a 2-week test-retest reliability of 0.86, and the GAT subscale had an internal consistency of 0.71 and a test-retest reliability of 0.87. In the current sample, we found good internal consistency with the WST subscale (α = 0.94) and the GAT subscale (α = 0.82) of the PARTS.
Body Shape Questionnaire (BSQ)
The BSQ (26) is a 34-item measure of body image dissatisfaction. The BSQ assesses the frequency of preoccupation with and distress about body size and/or shape. Subjects rate items (e.g., “Have you felt so bad about your shape that you have cried?”) on a scale from 1 (never) to 6 (always). Higher scores reflect greater body image concerns. The BSQ is a widely used instrument in studies of eating and weight disorders (6,7). Body dissatisfaction is frequently found in eating disordered patients and thus represents an important but distinct associated feature (27).
The Beck Depression Inventory (BDI)
The 21-item version of the BDI (28) was employed to assess level of depression. The BDI is a psychometrically established, widely used inventory of the cognitive, affective, motivational, and somatic symptoms of depression. Higher scores reflect higher levels of depression. A score of 16 or higher on the BDI is generally recommended as a cutoff point for symptoms of moderate depression (28,29).
The Rosenberg Self-Esteem Scale (RSES)
The RSES (30) is a widely used 10-item measure of global self-esteem with established reliability and validity (30,31). Subjects rate the items (e.g., “On the whole, I am satisfied with myself.”) on a scale from 1 (strongly agree) to 4 (strongly disagree). Higher scores reflect higher self-esteem.
Means and standard deviations for the teasing, eating disorder psychopathology, body dissatisfaction, and psychological functioning measures for the entire sample are shown in Table 1. To examine the relationship among the variables, Pearson correlation analyses were conducted. GAT was significantly associated with one eating disorder psychopathology variable (EDE-Q weight concern subscale), with body dissatisfaction (BSQ), and with the psychological functioning variables (BDI and RSES). Unlike GAT, WST was not significantly associated with the eating disorder psychopathology or body dissatisfaction variables but was significantly correlated with one psychological functioning variable, RSES (see Table 1).
Table 1. Scores on measures and correlations among teasing, eating symptoms, and psychological functioning variables (N = 115)*
|PARTS|| || || || || || |
|WST||26.96||13.56|| ||0.21†|| ||45§|
|GAT||10.38||5.17||0.21†|| ||45§|| |
| || || || || || || |
|EDE-Q|| || || || || || |
|Body dissatisfaction|| || || || || || |
|Associated psychological functioning|| || || || || || |
Because early experiences of being overweight may influence the development of core and associated features of BED, we re-analyzed the findings by simultaneously controlling for the age of onset of obesity and current BMI status. Nine participants reported never being overweight and thus were excluded from these partial correlation analyses. The associations between GAT and EDE-Q weight concern subscale and BSQ remained significant, and associations between WST and the eating disorder psychopathology and body dissatisfaction variables remained nonsignificant, even after simultaneously controlling for age of onset of obesity and current BMI status (see Ref. (32). The associations between both types of teasing (GAT and WST) and psychological functioning (BDI and RSES) also were significant when controlling for onset of obesity and current BMI status.
A series of hierarchical multiple regression analyses were performed to ascertain the joint and independent contribution of teasing (WST and GAT), BDI, and their interactions to the prediction of BSQ, EDE-Q weight concern, and RSES (i.e., domains significantly associated with teasing in the univariate analyses). For each criterion variable, the predictor variables were entered in three blocks. The first block consisted of the two forms of teasing (GAT and WST). The second block included the level of depression (BDI). The third block included the interaction terms between BDI and both forms of teasing (GAT × BDI and WST × BDI). The results are presented in Table 2.
Table 2. Hierarchical regression analyses conducted with each outcome variable (BSQ, EDE-Q weight concern, and RSES N = 115)
|Body dissatisfaction (BSQ)|| || || || || || |
|Teasing variables (GAT and WST)||0.09||0.09||0.01||GAT||0.28||0.00|
|Depression level (BDI)||0.27||0.18||0.00||BDI||0.45||0.00|
|Interaction terms (GAT× BDI, WST× BDI)||0.32||0.05||0.03||GAT× BDI||−0.86||0.01|
| || || || || || || |
|Weight concern (EDE-Q weight concerns)|| || || || || || |
|Teasing variables (GAT and WST)||0.04||0.04||0.09|| || || |
|Depression level (BDI)||0.16||0.12||0.00||BDI||0.37||0.00|
|Interaction terms (GAT× BDI, WST× BDI)||0.17||0.01||0.85|| || || |
| || || || || || || |
|Self-esteem (RSES)|| || || || || || |
|Teasing variables (GAT and WST)||0.09||0.09||0.01||GAT||−0.22||0.02|
|Depression level (BDI)||0.39||0.31||0.00||BDI||−0.59||0.00|
|Interaction terms (GAT× BDI, WST× BDI)||0.39||0.00||0.97|| || || |
When the teasing variables were entered in block 1, different results emerged across criterion variables: GAT was a significant predictor of BSQ and RSES but not EDE-Q weight concern. WST was not a significant predictor of any of the criterion variables. When level of depression was entered in block 2, identical results emerged across criterion variables: BDI was a significant predictor of BSQ, EDE-Q weight concern, and RSES. When the interaction terms were entered in block 3, interaction term GAT × BDI was a significant predictor of BSQ only, but interaction term WST × BDI was not a significant predictor of any of the criterion variables. Examination of the significant interaction effect of GAT × BDI suggests that, for BED patients with higher BDI scores, elevated BSQ is reported regardless of frequency of GAT. In contrast, for BED patients with lower BDI scores, BSQ is positively associated with frequency of GAT.
To further examine the impact of age of onset of obesity, the overall study sample was divided into three groups based on the women's reported age of first becoming overweight. Because it is possible that the effects of teasing may have a differential impact depending on developmental processes, including the timing or onset of obesity, subjects were divided into three groups: childhood onset (before age 10; n = 26; 24.5%), adolescent onset (ages 10 through 17; n = 47; 44.3%), and adult onset (ages 18 or older; n = 33; 31.1%). ANOVAs were conducted to examine differences among the three groups on the variables of interest (Table 3). The nine participants reporting never being overweight were excluded from these analyses.
Table 3. Comparison of childhood onset, adolescent onset, and adult onset groups*
|Measure||Mean||SD||Mean||SD||Mean||SD||F(df = 2, 105)||P|
|PARTS|| || || || || || || || |
| || || || || || || || || |
|EDE-Q|| || || || || || || || |
| || || || || || || || || |
|Body dissatisfaction|| || || || || || || || |
| || || || || || || || || |
|Associated psychological functioning|| || || || || || || || |
As shown in Table 3, the three groups differed significantly in WST only. Scheffe post hoc tests revealed that WST history differed according to age of onset of obesity. Patients with childhood onset reported greater frequency of WST than did patients with adolescent and adult onset. Patients with adolescent onset reported greater frequency of WST than those with adult onset. Frequency of GAT, current eating disorder psychopathology, body dissatisfaction, and psychological functioning did not differ significantly across the groups. A power analysis suggests that the lack of observed differences between the three study groups is unlikely the result of low statistical power. Our design had approximately an 83% chance of detecting modest to moderate effects (f = 0.30) with two-tailed tests at a level of significance of p < 0.05.
Because prior work (7,14) dichotomized the age of onset of obesity as early onset (under 18 years of age) and late onset (18 years of age and older), we also analyzed the data with two groups according to this age criteria. Again, the nine participants reporting never being overweight were excluded from these analyses. Similar to the findings with three age of onset groups, independent sample t tests (two-tailed) revealed that patients with early onset reported a greater frequency of WST than those with late onset (t(104) = 6.55, p = 0.0001). Group differences for the remaining variables of interest were not observed.
In addition to examining differences in current eating disorder psychopathology, body dissatisfaction, and psychological functioning according to age of onset of obesity, we also investigated this association with respect to current BMI status. The sample was divided into two subgroups based on current BMI status: obese (BMI ≥ 30; n = 76; 66.1%) and non-obese (BMI < 30; n = 39; 33.9%) patients. Independent sample t tests (two-tailed) were conducted to examine differences between obese and non-obese females with BED on the variables of interest. Obese women reported a significantly greater frequency of WST than did non-obese women (t(113) = −2.91, p = 0.004). Obese and non-obese women did not differ significantly in history of GAT (t(113) = 1.50, p = 0.14) or in eating disorder psychopathology/binge frequency (t(113) = −0.68, p = 0.50), eating restraint (t(113) = 1.94, p = 0.06), eating concerns (t(113) = −1.09,p = 0.28), weight concerns (t(113) = 0.14, p = 0.89), and shape concerns (t(113) = −0.71, p = 0.48). The subgroups also did not differ significantly in body dissatisfaction, BSQ (t(113) = −0.88, p = 0.38), or in psychological functioning, BDI (t(113) = −1.15, p = 0.26) and RSES (t(113) = −0.65, p = 0.52).
To further examine differences on the variables of interest between obese and non-obese women, Pearson correlations were conducted separately within each subgroup. In the obese subgroup, GAT was associated with binge frequency (r = 0.26, p = 0.022), BSQ (r = 0.29, p = 0.012), BDI (r = 0.30, p = 0.009), and RSES (r = −0.28, p = 0.013). WST was not significantly associated with the current eating disorder psychopathology variables, BSQ or BDI, but was significantly correlated with RSES (r = −0.23, p = 0.046) among obese women. In the non-obese subgroup, GAT was associated with EDE-Q eating restraint subscale (r = 0.40, p = 0.011), the BSQ (r = 0.34, p = 0.033), and the BDI (r = 0.44, p = 0.006). WST was not significantly associated with any of the variables of interest in the non-obese subgroup.
This study assessed relationships among teasing history, current eating disorder psychopathology (i.e., binge frequency, eating restraint, concerns regarding eating, shape, and weight), body dissatisfaction, and psychological functioning (i.e., depression and self-esteem) in a sample of female patients with BED. Past experiences of GAT were associated with current levels of weight concern and body dissatisfaction, but WST, surprisingly, was not. However, both GAT and WST were associated with psychological functioning. Although past experiences of teasing differed according to age of onset of obesity and current BMI status, current eating disorder psychopathology, body dissatisfaction, and associated psychological functioning did not differ by age of onset of obesity or by obesity status. Correlational analyses conducted separately for obese and non-obese women revealed that, although GAT was similarly associated with body dissatisfaction for both obese and non-obese women, differences in associations of GAT with the eating variables (i.e., binge frequency and eating restraint) were found between groups.
In contrast to two previous studies (5,7), we found that GAT, but not WST, exhibited significant associations with eating psychopathology features. Although these findings for BED require replication, we speculate that sample characteristics may account, in part, for the differences in findings. Grilo and colleagues’ (7) finding that WST, but not GAT, was associated with body dissatisfaction during adulthood was based on a clinical sample of obese females without BED. Those obese, non-BED patients were characterized by lower levels of body dissatisfaction and reported lower frequency of both GAT and WST than the patients in the present BED sample. Thompson and colleagues’ (5) derived their conclusion that WST may serve as a stronger correlate of eating disturbances and body dissatisfaction than GAT from survey responses of a nonclinical sample of college students. Thus we speculate that our findings based on treatment-seeking BED patients—along with those of Fairburn and colleagues (8) that were based on a community, nonclinical sample of BED individuals—suggest that the impact of different forms of teasing vary depending on the severity of eating psychopathology features present.
Although the frequency of past experiences of GAT was inversely related to age of onset of obesity, women with childhood, adolescent, or adult onset of obesity were strikingly similar in their current levels of eating psychopathology and associated psychological functioning. These findings are at odds with Stunkard's (10) initial observations suggesting that early onset of obesity is related to greater levels of eating and weight disturbances in adulthood than late onset of obesity. Prior empirical studies of psychosocial correlates of early vs. late onset of obesity have produced equivocal results. Two studies with obese persons found that earlier age of onset of obesity was associated with greater eating concerns (13) and severity of weight cycling (12). Consistent with our findings are those of Adami and colleagues (14) who also found that persons with early onset of obesity did not differ from those with late onset of obesity in levels of eating disturbance or body dissatisfaction in adulthood. One previous study (11) that examined 17 overweight women with BED found that early onset of obesity was associated with BED, but did not examine whether onset of obesity was related to the specific eating psychopathology features of BED. To our knowledge, the current study is the first to investigate the relationship between age of onset of obesity and variability in current core and associated eating psychopathology features in BED patients.
We also found that obese and non-obese women with BED exhibited similar levels of eating-related disturbances. However, the frequency of GAT, but not WST, was significantly associated with higher binge frequency in obese women and with greater eating restraint in non-obese women. This interesting finding that GAT may be associated with different components of eating-related psychopathology of BED in obese vs. non-obese patients warrants future investigation.
We note several limitations. Our study relied heavily on the use of self-report instruments. Thus, the inherent limitations characteristic of self-report must be considered. Conversely, self-report may facilitate disclosure of embarrassing or uncomfortable material. Also, our findings may not be generalizable to general outpatient (nonspecialty clinic) or community populations (33). Moreover, our findings may not be generalizable to those individuals who seek treatment at weight control programs rather than eating disorder centers. It is possible that such persons may have lower levels of cognitive symptomatology than those observed in our study group.
In addition, our study design does not permit a casual interpretation of the association between being teased and current functioning, most notably body dissatisfaction and self-esteem. However, our findings speak to associations that future research can attempt to delineate better. We note possible alternatives. For example, current negative body image or poor self-esteem may produce a retrospective recall bias regarding the degree of teasing experienced while growing up. Although this possibility cannot be discounted, our finding that patients’ reports of body dissatisfaction were correlated with the frequency of GAT, but not WST, argues against it. That is, patients did not uniformly endorse teasing experiences without making an attempt to be specific.
We also note that our patients reported, on average, moderate levels of depression as measured by the BDI. The co-occurrence of depression may influence reports of current body dissatisfaction, self-esteem, and reports of childhood experiences of being teased about physical appearance. Our hierarchical regression analyses revealed that GAT and level of depression made significant incremental contributions (i.e., main effects) to the prediction of body dissatisfaction and self-esteem. Interestingly, we also observed a significant interaction effect between GAT and level of depression. Whereas patients with higher levels of depression reported elevated body dissatisfaction regardless of GAT, in those patients with lower levels of depression, higher body dissatisfaction was associated with higher GAT. Nonetheless, we cannot discount alternative possibilities such as the presence of cognitive vulnerabilities to depression or to eating disturbances that might have made certain individuals particularly sensitive to social pressures regarding body image or to social commentary or teasing about appearance. Prospective studies are needed to examine these various possibilities.
With these relative strengths and weaknesses as a context, we offer the following clinical implications based on our findings. Clinicians working with patients with BED should consider integrating an assessment of physical appearance-related teasing experiences into an overall comprehensive evaluation. Our findings, and those reported by Fairburn (8), suggest that being teased (or the perception of being teased) is not uncommon in patients with BED. The levels of teasing reported by our BED patients here are higher than those previously reported for obese patients without BED (7). Moreover, we found teasing to be associated with some related aspects of BED, most notably body dissatisfaction, depression, and self-esteem. Certain cognitive errors around body image may have been, in part, learned from childhood teasing experiences. Asking BED patients about teasing experiences may help to elucidate behavioral patterns of social and interpersonal avoidance or cognitions that impact negatively on self-evaluation.
In conclusion, we found that, although only GAT was associated with current levels of weight concern and body dissatisfaction, both GAT and WST were associated with psychological functioning. Childhood onset of overweight was associated with a higher frequency of teasing, but current eating and psychological functioning did not differ by age of onset of obesity nor current BMI status.
Supported by National Institutes of Health Grant DK49587.