Regular Article (CE Activity)
Subjective binge eating with compensatory behaviors: A variant presentation of bulimia nervosa
To determine whether a variant bulimic-type presentation, whereby one meets criteria for bulimia nervosa (BN) except that binge eating episodes are not objectively large (i.e., “subjective bulimia nervosa,” SBN), has comparable clinical severity to established eating disorders, particularly BN.
Treatment-seeking adults with BN (N = 112), SBN (N = 28), anorexia nervosa restricting type (AN-R) (N = 45), and AN-binge/purge type (AN-B/P) (N = 24) were compared.
Overall, SBN could not be meaningfully distinguished from BN. SBN and BN had equivalent eating pathology, depression and anxiety symptoms, low quality of life, impulsivity, Axis I comorbidity, and lifetime psychiatric history, and comparable clinical severity to AN-R and AN-B/P.
Individuals with SBN, differing from BN only by the smaller size of their binge eating episodes, had a form of eating disorder comparable in clinical severity to threshold AN and BN and warranting clinical attention. Health professionals and the community require greater awareness of this variant to optimize detection, treatment-seeking, and outcomes. © 2012 by Wiley Periodicals, Inc. (Int J Eat Disord 2013)
Bulimia nervosa (BN) was conceptualized as a distinct disorder by Russell in 19791 and first appeared in the diagnostic and statistical manual (DSM) in 1980 as one of three officially recognized eating disorders, along with anorexia nervosa (AN) and eating disorder not otherwise specified (EDNOS).2 A widely acknowledged limitation of DSM nomenclature is that EDNOS is heterogeneous, under-researched, and prevalent, capturing at least half of clinical- and community-based cases of eating disorders, in many cases, subthreshold forms of BN or AN.3–5 Research efforts to better characterize subthreshold presentations have ignited with the overarching goal of improving the validity and utility of classification; to parsimoniously inform health professionals of the pathophysiology, course, correlates, comorbidities, and treatment response of categorized illnesses.
A BN criterion under increasing scrutiny is whether the size of a binge eating episode must necessarily be “objectively large”.6–8 Accumulating evidence suggests that the perception of loss of control over eating, regardless of the amount of food consumed, is the critical feature of binge episodes,8–10 and studies related to binge eating have highlighted that size is not always relevant to the perception of a binge eating episode. Women from the community who self-identified binge eating episodes, placed more importance on the loss of control dimension, and classified nearly one-half of subjective bulimic episodes as eating binges, even though intake was not objectively large.11 Food diaries from patients with BN have shown “binges” of <100 kcal and that a third classify episodes involving <500 kcal as binges.8 Regressions predicting pathology from objective binge episodes (OBEs) and subjective binge episodes (SBEs) in BN have shown that the difference in type of eating binge does notaccount for variance in eating pathology, psychiatric symptoms, or Axis I and II comorbidity, indicating that the eating episodes could not be differentiated.10, 12 The labeling of an eating episode as a binge, rather than objective caloric intake, appears to be more predictive of purging among women with BN.13 While size is often relevant, it may not be a requisite feature.
When impairment related to objective and subjective binge eating is compared in community and epidemiologic samples, comparable outcomes emerge. Women with OBEs but not SBEs and those with SBEs but not OBEs had equivalent elevated eating disorder psychopathology, use of extreme weight-control behaviors, low quality of life, psychological distress, and self-identification of an eating problem.7 General population and overweight youth with SBEs versus OBEs are comparable on eating psychopathology and depressive/anxious symptoms, and separately have more symptomatology than healthy controls and youth who overeat with no loss of control.9, 14 A community-based cluster analysis could not differentiate women who purged based on size of binge eating episodes.15 Young adult women in the community who reported OBEs, SBEs, or both, had similar quality of life and more impairment than healthy controls.16 SBEs have been associated with impaired quality of life independent of OBEs.17
On the clinical spectrum, women with BN purging subtype and women who would have met threshold criteria except that they only experienced SBEs (SBN: “subjective bulimia nervosa”) were comparable on most measures of eating and general psychopathology.6 Notable differences were higher impulsivity and a purging rate twice as high for BN, and a lower treatment-seeking rate among SBN. These findings were based on women recruited from the community; no investigations to date have compared clinical presentation in a treatment-seeking population.
The bulimic spectrum “purging disorder” (i.e., recurrent purging after small or normal amounts of food) resembles SBN and has shown similar pathology to BN in a number of investigations (e.g., Ref.18). Evidence suggests that purging disorder is not a prodromal or partial BN syndrome, and has emerged as a separate class from other eating disorders within several latent class analyses.19 Purging disorder is already recognized as a form of EDNOS in the DSM-IV20 and several options for categorization in DSM-5 are under consideration, for example, as a provisional research diagnostic entity. Yet purging disorder is not encompassing of SBN; people with subthreshold BN who use nonpurging compensatory behaviors (i.e., fasting, excessive exercise, appetite suppressants) would not be considered to have purging disorder but could be classified as SBN, and subjective binge eating is not a (proposed) diagnostic criterion for purging disorder.
The purpose here is to investigate in a treatment-seeking sample whether SBN, a subthreshold variant of BN marked by the presence of subjective binge eating only in addition to compensatory behaviors, has comparable clinical severity and correlates to established threshold eating disorders. We hypothesize that individuals with SBN will closely resemble both individuals with BN on measures of eating and general pathology and individuals with AN, except that they will have higher body mass index (BMI) than AN and lower BMI than BN; this latter hypothesis is informed by research linking overeating episodes to BMI.7,10,16 If SBN is highly comparable to established eating disorders and cannot be meaningfully differentiated from BN, this will have important implications for diagnosis and treatment.
The sample (N = 209) comprised individuals with AN-R (restricting type), AN-B/P (binge-purge type), BN, or SBN (i.e., individuals who would have met criteria for BN except that they experienced only SBEs) who were consecutive referrals to a specialist eating disorder service between 2004 and 2011. The service is the only publicly funded youth (16+ years) and adult eating disorder service in Western Australia and is statewide and outpatient-based. The mean age was 25 years (SD = 9), and participants tended to be female (98%), of single marital status (69%; vs. de facto/married), and employed (52%). The average duration of illness was 6.5 (SD = 7.3) years.
The four groups were derived according to outcomes on the Eating Disorder Examination (EDE).21 This resulted in 112 individuals who met criteria for BN, 28 with SBN, and comparison groups of 45 with AN-R and 24 with AN-B/P. For this study, the BN criteria were modified; the frequency threshold of OBEs and purging/compensatory episodes was lowered from at least twice to at least once per week, consistent with research showing concordant clinical severity22 and the proposed change to DSM-5 BN criteria.23 To be classified as SBN, individuals could not have OBEs. Concordant with the amendment to BN criteria, the frequency threshold of SBEs and purging/compensatory methods was at least once per week. The EDE has well-established validity and inter-rater reliability for differentiating type of binge eating episode. Excellent inter-rater kappa values have been reported for both SBEs (K > 0.90–0.91)24–26 and OBEs (K > 0.90–0.99),24–27 although one study of a Spanish translation reported moderate inter-rater agreement for SBEs.28
The service routinely excludes and refers individuals with current acute psychosis, schizophrenia or schizoaffective disorder, alcohol/substance abuse or dependence, suicidality, or a BMI < 14 kg m−2. Participants provided written, informed consent for their data to be used in research. The study was approved by the North Metro Area Health Ethics Committee.
As part of routine intake, all participants attended two to three assessment sessions which included a clinical interview, self-report assessments, and administration of the EDE21 to assist in yielding an eating disorder diagnosis and the Mini International Neuropsychiatric Interview (MINI)29 to diagnose other Axis I disorders. Participants had height and weight (clothed; shoes and outer garments removed) measured by the clinician to determine BMI.
As part of routine intake, all patients are administered the EDE,21 widely considered the “gold standard” for assessing eating disorder behaviors and psychopathology and for facilitating diagnosis. The EDE was administered by a clinician trained in its administration and specializing in eating disorder treatment. The MINI29 was administered to assess DSM-IV Axis I comorbidity. The MINI has good inter-rater reliability and validity and concordance with established, lengthier diagnostic interviews such as the SCID.30–32 The MINI is superior to unstructured interviews at detecting comorbid diagnoses.33 Kappa statistics for interrater reliability of psychiatric diagnoses (i.e., principal and comorbid) range from 0.79 to 1.00,30 with the majority of kappa values (i.e., 70% ≥ 0.90) suggesting outstanding interrater reliability (<0.40 = poor; 0.40–0.59 = moderate; 0.60–0.79 = substantial; >0.80 = outstanding34).
Binge Eating, Purging, and Eating Disorder Psychopathology
The EDE21 measured eating disorder behaviors and psychopathology. Behaviors included OBEs and SBEs, purging episodes (vomiting/laxative/diuretic misuse), duration of exercise, and use of fasting for weight/shape control (avoidance of eating item) over the previous 28 days. Psychopathology was measured with the four subscales (restraint, eating concern, shape concern, and weight concern) and the global scale. The EDE has well-established validity and inter-rater and test–retest reliability for behavioral features and subscales.25, 26, 28, 35, 36
General Psychopathology and Clinical Correlates
General distress was measured with the Depression, Anxiety, and Stress Scales of the Depression Anxiety and Stress Scales (DASS-42).37 The DASS-42 has acceptable reliability and validity in clinical psychiatric and nonclinical samples.37 Self-esteem was measured with the widely used 10-item self-report Rosenberg Self-Esteem Scale (RSES).38 The RSES has good reliability and validity.38–40 Quality of life was measured with the short form Quality of Life Enjoyment and Satisfaction Questionnaire-Short Form (Q-LES-Q SF).41 The Q-LESQ SF consists of the first 14 items of the general activities scale of the original 93-item Q-LES-Q. Items assess physical health, economic status, relationships, living/housing situation, mood, work, medication (if applicable), and overall life satisfaction. The Q-LES-Q SF has good internal consistency and construct and criterion validity.42 Interpersonal difficulties were measured with the inventory of interpersonal problems-32, which has good psychometric properties (IIP-32).43 Perfectionism was measured with the self-oriented perfectionism items from the eating disorder inventory-2 (EDI)44 and impulsivity was measured with the impulse regulation scale of the EDI using the full six-point rating format (i.e., untransformed scores) in both instances, with scale means calculated for interpretability. The EDI is widely used; the perfectionism45 and impulse regulation subscales have acceptable psychometric properties.44
Group differences on descriptive variables were evaluated with an analysis of variance (ANOVA) for age and chi-square test of independence for categorical variables. Group differences on eating pathology, general psychopathology, and clinical correlate measures that used a continuous scale were evaluated with a multivariate analysis of variance (MANOVA). ANOVA with Tukey's HSD (honestly significant difference) was used for post hoc testing; when homogeneity of variance was violated, the Games–Howell modification was used. Group differences on categorical measures were tested with chi-square test of independence. The alpha level was 0.05. To reduce the possibility that the SBN sample was contaminated by emerging AN (i.e., weight loss approaching but not yet at the 85% threshold), the analyses were repeated twice: (1) with the SBN sample restricted to those at “healthy weight” as defined by the World Health Organization according to the most conservative lower limit of healthy (i.e., BMI ≥ 20), and (2) with weight at any point above the AN threshold (i.e., BMI > 17.5). Results were almost identical, hence the second set of analyses are reported.
There were no group differences on age, sex, marital status, or employment status (ps < .05).
Eating Pathology, General Psychopathology, and Clinical Correlates
The MANOVA revealed a significant group difference across measures, F(57, 558) = 7.86, p < .001. Post hoc tests (Table 1) showed that BN had a significantly greater number of total binge eating episodes—SBEs and OBEs combined—compared with SBN and AN-R and a significantly higher BMI than SBN, AN-R, and AN-B/P. SBN could not be differentiated from BN on purging, exercise, fasting, eating concern, shape concern, weight concern, and global eating psychopathology, but had significantly higher restraint than BN and a higher number of SBEs compared with BN and AN-R. The bulimic groups tended to report significantly greater eating pathology than AN-R.
Table 1. Comparison of OBN, SBN, AN-B/P, and AN-R on measures of eating pathology
|OBEs||—||29.05 ± 23.40||18.10 ± 26.63||0.04 ± 0.30||<.001***||OBN, AN-B/P>AN-R||—||—|
|SBEs||26.50 ± 20.49||12.14 ± 18.52||18.25 ± 22.53||5.09 ± 14.25||<.001**||SBN>OBN, AN-R||—||—|
|Total binge eating episodes||26.50 ± 20.49||41.20 ± 29.95||36.35 ± 32.10||5.13 ± 14.24||<.001***||OBN>SBN>AN-R; AN-B/P>AN-R||—||—|
|Purging||47.21 ± 65.72||43.81 ± 44.72||59.10 ± 113.23||0.39 ± 1.19||<.001***||SBN, OBN>AN-R||—||—|
|Exercise (minutes/day)||30.60 ± 50.53||23.24 ± 32.62||13.08 ± 16.45||25.62 ± 37.19||ns||—||—||—|
|Fastinga||0.75 ± 0.44||0.71 ± 0.46||0.71 ± 0.46||0.51 ± 0.51||ns||—||—||—|
|BMI kg/m2||21.07 ± 2.47||23.52 ± 3.42||16.13 ± 1.38||16.04 ± 1.35||<.001***||OBN>SBN>AN-R, AN-B/P||—||—|
|EDE restraint||4.46 ± 0.76||3.93 ± 1.16||4.19 ± 1.32||3.53 ± 1.60||.01||SBN>OBN, AN-R; AN-B/P>AN-R||1.30 ± 1.40b||0–6|
|EDE eating concern||3.61 ± 1.09||3.78 ± 1.25||3.91 ± 1.19||2.79 ± 1.40||<.001***||OBN, SBN>AN-R||.76 ± 1.06b||0–6|
|EDE shape concern||5.05 ± 0.70||4.93 ± 1.07||4.61 ± 1.27||4.08 ± 1.67||.001**||OBN, SBN>AN-R||2.23 ± 1.65b||0–6|
|EDE weight concern||4.66 ± 0.97||4.57 ± 1.20||3.81 ± 1.5||3.37 ± 1.63||<.001***||OBN, SBN>AN-R||1.79 ± 1.51b||0–6|
|EDE global||4.44 ± 0.65||4.30 ± 0.94||4.13 ± 1.02||3.44 ± 1.42||<.001***||OBN, SBN>AN-R||1.52 ± 1.25b||0–6|
SBN, BN, AN-R, and AN-B/P were not significantly different (Table 2) on Axis I comorbidity, anxiety, stress, perfectionism, and interpersonal difficulties, and SBN and BN were not significantly different on quality of life, self-esteem, and impulse regulation, despite differences involving other diagnostic pairings. There was an omnibus group difference on depressive symptoms, although post hoc analyses showed no significant group differences. There were no group differences in lifetime psychiatric history on treatment, involuntary psychiatric hospitalization, and suicidality, although AN had a higher frequency of psychiatric hospitalization and BN had a lower frequency of psychiatric hospitalization. There was an overall difference on self-harm, but post hoc analyses revealed no significant between group differences, only a trend toward greater self-harm in SBN and AN-B/P.
Table 2. Comparison of OBN, SBN, AN-B/P, and AN-R on measures of current psychiatric comorbidity and general pathology and lifetime psychiatric history
|Comorbid axis I disorder, % (n)||68% (19)||46% (52)||46% (11)||38% (17)||ns||—||—||—|
| Depressive disorder, % (n)||57% (16)||37% (42)||25% (6)||29% (13)||ns||—||—||—|
| Anxiety disorder, % (n)||36% (10)||25% (28)||37% (9)||16% (7)||ns||—||—||—|
|DASS depression||21.96 ± 14.70||20.25 ± 11.73||22.79 ± 12.98||15.24 ± 11.19||.03*||ns||5.55 ± 7.48a||0–42|
|DASS anxiety||14.43 ± 7.50||12.51 ± 8.72||12.17 ± 7.93||9.4 0 ± 9.40||ns||—||3.56 ± 5.39a||0–42|
|DASS stress||23.00 ± 10.28||22.14 ± 9.24||21.83 ± 9.09||18.51 ± 9.76||ns||—||9.27 ± 8.04a||0–42|
|Q-LES-Q SF||46.29 ± 21.13||46.59 ± 15.57||38.22 ± 14.58||52.05 ± 18.4||.01*||AN-R>AN-B/P||78.1 ± 13.7b||0–100|
|RSES||20.54 ± 5.47||21.51 ± 5.25||18.67 ± 5.30||23.42 ± 6.04||.01*||AN-R>AN-B/P||31.07 ± 5.15c||0–40|
|EDI perfectionism||3.87 ± 1.40||4.31 ± 1.37||4.53 ± 1.28||4.37 ± 1.52||ns||—||—d||1–6|
|EDI impulse regulation||2.91 ± 0.85||3.06 ± 0.91||3.33 ± 0.74||2.41 ± 0.95||<.001***||OBN>AN-R; AN-B/P>AN-R||—d||1–6|
|IIP||1.48 ± 0.55||1.60 ± 0.58||1.76 ± 0.43||1.43 ± 0.63||ns||—||0.98 ± 0.52e||0–4|
|Lifetime psychological treatment, % (n)f||89% (25)||87% (97)||87% (21)||89% (40)||ns||—||—||—|
|Lifetime psych. hosp., % (n)f||32% (9)||24% (27)||71% (17)||47% (21)||<.001***||>AN; <OBN||—||—|
|Lifetime involuntary psych. hosp., % (n)f||11% (3)||8% (9)||21% (5)||13% (6)||ns||—||—||—|
|Lifetime suicide attempt, % (n)f||36% (10)||24% (27)||33% (8)||16% (7)||ns||—||—||—|
|Lifetime self-harm, % (n)f||57% (16)||32% (36)||54% (13)||27% (12)||.01*||ns||—||—|
SBN could not be reliably differentiated from BN on the basis of eating pathology, clinical markers and correlates, comorbidity, or lifetime psychiatric history. The clinical significance of SBN as a form of eating disorder was apparent, with severity of pathology and functional impairment akin to AN-R, AN-B/P, and BN.
The commonalities among SBN to specific diagnoses recognized in DSM-IV; BN, AN-R, and AN-B/P, were striking, and indicate that presentations characterized by subjective binge eating and compensatory behaviors should be a focus of clinical attention. Individuals with SBN reported eating pathology and functional impairment of clinically comparable levels to individuals with AN and BN, and impairment relative to published normative and community adult samples.43, 45–48 These findings are consistent with Keel et al's study, which found that women with SBN and BN recruited from the community scored similarly on measures of eating and general psychopathology, and also complement accumulating evidence that binge size may be less salient to pathology than loss of control in defining binge eating.6–8, 10 AN had the lowest BMI and BN had the highest BMI of the groups. Previous research has generally shown a positive association between OBEs and BMI,7, 16, 49 likely attributable to higher caloric intake and retention from overeating episodes. Keel et al.'s finding of increased impulsivity in the BN group relative to the SBN group fuelled debate about whether impulsivity is a central feature of BN or an epiphenomenon; we did not reproduce this difference in impulsivity between SBN and BN. This may indicate a nonrobust association or could be attributable to method differences, such as the use of different measures and sampling. In contrast to the study of Keel et al., we did not exclude individuals from SBN on the basis of a lifetime history of OBEs, our BN group included the nonpurging subtype in addition to the purging subtype only, and our SBN group included those who used nonpurging and purging compensatory methods. Also, our SBN group had demonstrably higher clinical severity than the SBN group in Keel et al.'s study, likely due to the clinic- rather than community-based recruitment method.
This study advances previous knowledge and provides the first clinical depiction of SBN as likely to be encountered in clinical practice. Given the clinical significance and the close resemblance of SBN to threshold BN, further research needs to be undertaken to understand whether the course, outcomes, and treatment responses are synchronous; if so, this would provide further evidence that the binge size distinction is nongermane. Currently, SBN presentations fall under the category of EDNOS, yet there is little public awareness of eating disorders beyond AN and BN, so individuals with these symptoms and health professionals may be less likely to identify and intervene. An absence of formal recognition may beget obstacles from insurers and health providers. Keel et al.6 found that individuals with SBN in the community were less likely to have received treatment than those with threshold BN. Research into the course of SBN is necessary, to understand whether the presentation is distinctive, or part of the naturalistic course of other eating disorders. As an example, one could conjecture that some cases of SBN may represent various stages of recovery process of BN: binge size typically diminishes when meal patterns are normalized, but in the face of longstanding appetite dysregulation, eating concern, and rigid food rules, loss of control cognitions may take longer to extinguish. Clinically, a shift from larger to smaller binge sizes has medical implications. A higher or stable weight that had been maintained secondary to larger binges, could rapidly progress to weight loss and underweight if binge size reduces dramatically. Similarly, the clinical course of AN subsequent to weight restoration considered in historical clinical context may just be part of the naturalistic course of AN, yet in a cross-sectional evaluation would appear to be SBN. Understanding the relation between SBN and lifetime history of OBEs, BN, and AN, has implications for screening, identification, and treatment depending on the treatment model used.
Our findings raise important questions about diagnostic classification. SBN could not be meaningfully differentiated from BN, except that the former did not consume an objectively large amount of food during binge eating episodes. Possibly, SBN could be classified within BN nosology by modifying the size criterion; an implication other studies have discussed.7 However, as Mond et al.7 highlight, binge size has become an axiom of binge eating, so this recommendation would be problematic for authorities to accept. Alternatively, the respective presentations could be subtyped within BN,7 which may increase the recognition of SBN; however, it is not clear that this would increase clinical utility. A more cogent justification for subtyping would include indications that the two presentations are different in etiology, course, outcome, or response, but empirical data have not indicated this. Keel et al., have conducted investigations into “purging disorder” (e.g., Ref.50) to characterize presentations distinguished by recurrent purging in the absence of OBEs, and have recommended inclusion of this syndrome as a provisional diagnostic entity into the next DSM. Purging disorder would capture many individuals denoted as SBN in this study, but it would not capture those with non-purging compensatory behaviors (e.g., excessive exercise, fasting).
This study's findings have implications for treatment. Given the pathology and impairment associated with subjective binge eating,7, 9, 10, 12, 14, 16, 17, 49 it would seem important to target more specifically, subjective binge eating within existing cognitive–behavioral interventions. Research is mixed, with some studies finding that cognitive-behavioral therapy (CBT) is slower or less effective at reducing subjective compared to objective binge eating,51–53 but others showing a similar response.54, 55 CBT strategies to reduce OBEs initially highlight the vicious cycle of restriction/deprivation leading to hunger and subsequent over-eating with loss of control (i.e., OBEs). The approach is largely behavioral, with the introduction of regular eating aimed at reducing the maintaining factor of hunger in OBEs. In SBEs, where there is no over-eating, specific strategies may need to be more cognitive, addressing the negative cognitive appraisal of eating even small amounts of food. However, removing negative appraisals and eliminating the “subjective loss of control” is only part of the issue; people with histories of eating symptoms are prone to excess weight gain (overweight and obese) so strategies must not inadvertently promote disinhibition, overeating, and frequently eating small quantities of food.
This study has several limitations. There was no community control, which would help distinguish the clinical severity of the SBN group. Notwithstanding, SBN showed comparable eating pathology, general pathology, and impairment, relative to AN-R, AN-B/P, and BN, and to community data. Second, the operationalization of SBN did not include individuals who reported OBEs. Greater importance was assigned to enhancing the internal validity of study findings, because this would rule out people with a history of bulimia nervosa by removing any potential confounding influence of OBEs, however, an ecologically valid and clinically useful operationalization would be strengthened by inclusion of these presentations. Third, excluding individuals with lifetime OBEs from the SBN sample may have strengthened the findings. Fourth, although the MINI has good convergence with lengthier diagnostic interviews31, 32 and is able to detect comorbidity,33 it is not considered a gold standard psychiatric diagnostic interview. Despite non-significant differences on comorbid profiles it is noted that there were large raw differences between groups on comorbidity. Correspondingly, some cell sizes in categorical analyses were small and results need to be interpreted cautiously. The EDE reliability for distinguishing SBEs from OBEs is imperfect, which may have affected group assignment, although inter-rater reliability is generally excellent and above 0.90. Unfortunately it was not possible to examine cross-sectional inter-rater reliability for this cohort given the study was not conducted prospectively.
In summary, individuals with “subjective bulimia nervosa,” differing from BN only by the smaller size of their binge eating episodes, had a form of eating disorder comparable in clinical severity to AN and BN and in need of clinical attention. The presentation could not be differentiated from BN on a meaningful or pathological basis, nor was it less severe despite the “subthreshold” classification.
Earn CE credit for this article!
Visit: http://www.ce-credit.com for additional information. There may be a delay in the posting of the article, so continue to check back and look for the section on Eating Disorders. Additional information about the program is available at www.aedweb.org