Wonderlich binge_IJED 2009

8/7/2019 Wonderlich binge_IJED 2009

1/19

The Validity and Clinical Utility of Binge Eating Disorder

Stephen A. Wonderlich, PhD1,2*

Kathryn H. Gordon, PhD3

James E. Mitchell, MD1,2Ross D. Crosby, PhD1,2

Scott G. Engel, PhD1,2

ABSTRACT

Objective: This review attempted to

examine the validity and clinical utility of

the DSM-IV binge eating disorder (BED)

diagnosis across a wide range of validat-

ing strategies.

Method: Various electronic databases

(Pub Med, Psych Info) were searched for

terms relevant to the diagnosis of BED

(e.g., binge eating disorder, binge eating)

in order to identify papers. Additionally,

published papers were reviewed in order

to locate additional manuscripts and

papers that were presented at meetings.

Results: The validity and utility of BED

varied substantially according to the vali-

dator chosen. There is reasonable evi-dence that BED can be differentiated

from other existing eating disorders and

is associated with significant impairment

and clinical levels of eating disorder psy-

chopathology. The relationship of BED to

obesity is complex, and in spite of some

positive findings, further research exam-

ining the predictive power of BED,

beyond the simple presence of obesity

and associated psychopathology, in rela-

tionship to clinically relevant outcomes is

needed.

Discussion: Binge eating disorder is

being considered for inclusion in the

DSM-V and various options regarding this

decision are reviewed based upon the

empirical findings in the paper. VVC 2009

American Psychiatric Association.

Keywords: binge eating disorder;

obesity; eating disorders

(Int J Eat Disord 2009; 42:687705)

Introduction

Binge eating disorder (BED) was included in theDiagnostic and Statistical Manual of Mental Disor-ders, 4th edition, text revision (DSM-IV1) as a provi-sional eating disorder diagnosis. To inform the deci-sion-making process for BEDs status in DSM-V, the

present review examines existing empirical evidenceon the validity and clinical utility of BED. Clinicalutility, which may be a broader concept than valid-ity, is a standard by which diagnoses for the DSM-Vshould be evaluated.2 The current review focusesprimarily on the ability of the DSM-IVBED diagnos-tic criteria to predict a broad range of external vali-

dating variables (e.g., family history, biological pa-rameters, quality of life) and, most importantly, vali-dators that reflect high degrees of clinical utility(e.g., clinical course, response to treatment). In apractical sense, this review will examine evidencethat BED discriminates itself from other eating dis-orders and obesity (without BED) on a variety of val-

idators. Although early research focused on the rela-tionship of BED to eating disorder diagnoses suchas anorexia nervosa (AN) and bulimia nervosa (BN),more recent research has tended to emphasize therelationship of BED to obesity, and this recent shiftwill be recognized in this review.

Another recent review of BED research, byStriegel-Moore and Franko3 addressed several con-cepts associated with both validity and clinical util-ity. Their conclusions regarding the inclusion ofBED in the DSM-Vare detailed below. It is first im-portant to note that Striegel-Moore and Franko3

applied criteria established by Blashfield et al.4 to

evaluate whether BED should be considered as adiagnosis in the DSM-V. The criteria outlined byBlashfield et al.4 are as follows:

1. There should be ample literature about theproposed syndrome.

2. The diagnostic criteria should be articulatedclearly, and assessment instruments shouldexist that may be used for determiningwhether an individual meets the criteria.

Accepted 9 June 2009

Supporting Information Table S1Table S12 may be found in

the online version of this article.

1 Department of Clinical Neuroscience, University of North

Dakota School of Medicine and Health Sciences, Fargo,

North Dakota2 Neuropsychiatric Research Institute, Fargo, North Dakota3 Department of Psychology, North Dakota State University,

Fargo, North Dakota

*Correspondence to: Stephen A. Wonderlich, PhD, Department of

Clinical Neuroscience, University of North Dakota School of Medicine

and Health Sciences, 120 8th

Street South, P.O. Box 1415, Fargo,North Dakota 58107-1415. E-mail: [email protected]

Published online 20 July 2009 in Wiley InterScience

(www.interscience.wiley.com). DOI: 10.1002/eat.20719

VVC 2009 American Psychiatric Association. This Article is being

co-published by the International Journal of Eating Disorders and

the American Psychiatric Association.

International Journal of Eating Disorders 42:8 687705 2009 687

SPECIAL SECTION REVIEW ARTICLE


2/19

3. The proposed syndrome should be diagnos-able with a high degree of reliability betweentwo or more assessors.

4. Evidence should be available that the pro-posed syndrome can be differentiated fromother (similar) syndromes.

5. Evidence should be provided regarding thecoherence and validity of the syndrome.

Applying these criteria, Striegel-Moore andFranko3 concluded the following: First, given thatthere were more than a thousand titles includingthe term BED in a PubMed database search in2007, there is an ample literature on this diagnosis.Second, the BED diagnostic criteria outlined inDSM-IVhave been used quite consistently and sev-eral psychometrically sound instruments havebeen developed for the assessment of BED. Theauthors acknowledged continued debate about the

specific criteria of BED, and highlighted the needfor further exploration in this area. Third, there isample evidence that the DSM-IV criteria for BEDcan be reliably applied across diagnosticians.Fourth, the authors concluded that BED can be dis-criminated from BN and obesity, and is therefore adistinct clinical entity. Fifth, with regards to the va-lidity of BED, the authors concluded that empiricalapproaches to classification (i.e., latent class analy-sis) repeatedly identify a class of eating disorderedindividuals that is consistent with BED. Also, theauthors reviewed several indicators of clinical sig-nificance including the high prevalence rates of

BED, considerable evidence of impairment and suf-fering in BED patients, along with elevated rates ofhealth service utilization among BED patients asindications of the clinical significance of BED. Inspite of some inconsistency in the literature, theauthors concluded that BED appears to be a relativelystable and protracted disorder with longitudinal evi-dence of clinical severity comparable to that of BN.

The authors overall conclusion was that a case canbe made for recognizing BED as a formal eating disor-der diagnosis in the DSM-V. However, they also sug-gested that further classification studies were needed,using wider sets of external validators, to examine the

clinical utility of newly derived criteria sets in the pre-diction of course and outcome. Next, we will outlinethe methods and results of the present review.

Method

The present review follows the recommendations of

Striegel-Moore and Franko3 and examines a wide array of

validators. Below, we will review empirical evidence

across a variety of validating strategies, which are gener-

ally consistent with Kendell5 and Kendlers6 recom-

mended validating strategies and include findings in the

following areas:

1. Empirically based studies of classification (e.g.,

latent class analysis and taxometrics)2. Laboratory-based studies of eating

3. Ecological momentary assessment-based studiesof eating in the natural environment

4. Epidemiologic studies

5. Studies of psychiatric comorbidity

6. Studies of medical outcomes

7. Studies of health-related quality of life, life satis-faction, and functional impairment

8. Longitudinal studies of diagnostic status

9. Studies of the effect of BED on treatment outcome

10. Family history studies

11. Studies of genetic polymorphisms12. Studies of brain function in BED

13. Studies of peptides and hormones in BED

14. Studies of the importance of shape and weightevaluation in BED.

To examine the validity and clinical utility of BED, we

conducted a comprehensive literature review on BED

and the validating variables. We searched major com-

puter databases (e.g., MedLine, PsychInfo) and also

reviewed reference lists from published literature. In

addition, we were able to identify recent paper presenta-

tions at meetings. Search terms for the computer-based

searches included binge eating disorder and binge

eating, as well as the key validating variables.

Results

Empirical Classification Studies

Supporting Information Table S1 provides a sum-mary of studies,713 which have used statisticalmethodologies that attempt to identify a class ofeating disordered individuals consistent with theDSM-IV BED diagnostic criteria. Typically, suchstudies employ latent structure models (i.e., latentclass analysis, latent profile analysis) to examineempirical data to identify naturally occurringgroups of eating disordered people based on simi-larities in symptom status, but these approachesdo not help to determine if the distinctionsbetween these groups represent qualitative orquantitative differences. Rarely, researchers haveemployed taxometric analyses following latentstructure models to assess the nature of the bound-

WONDERLICH ET AL.

688 International Journal of Eating Disorders 42:8 687705 2009


3/19

ary between the naturally occurring groups. (Keelet al., submitted for publication).

It should also be noted that not all of the empiri-cal studies of eating disorder classification are rele-vant to BED. For example, Gleaves et al.14 con-ducted a taxometric analysis that had significantimplications for a variety of eating disorder classes,but did not include individuals in the sample whodisplayed prototypic BED behavior. Similarly, Keelet al.15 conducted latent class analyses on a largesample of eating disordered individuals, but indi-viduals with BED were excluded from the sample,thus limiting the inferences relevant to the BEDdiagnostic category.

Several studies have included a variety of eatingdisordered participants, including individuals dis-playing BED-type behavior. A BED-like class wasreliably discriminated from BN in an epidemiologi-cal sample of twins,7 a sample of 238 binge eating

or purging women,11

and a sample of 687 treat-ment-seeking eating disorder patients.13 In con-trast, Wade et al.12 failed to find evidence that BEDcould be reliably differentiated from other eatingdisorder profiles in a sample of 1,002 community-based twins.

Importantly, very few of these studies have exam-ined the boundary between obesity and the relatedconstruct of BED. In other words, among obeseindividuals can a subgroup be reliably differenti-ated based on the presence of binge eating?Williamson et al.10 reported on samples of eatingdisordered, obese, and normal weight individuals

using factor analysis and taxometric analysis.Although these data supported a distinctionbetween BED and other eating disorders, the smallsample size of obese individuals without BEDreduced the strength of inference regarding theBEDobesity relationship. In another study, Mitch-ell et al.9 completed a latent profile analysis ofindividuals seeking treatment and receiving a diag-nosis of eating disorder not otherwise specified(EDNOS). This study produced results in which twoclasses of obese individuals were identified, onewith high levels of eating disorder psychopathologyand the other with low levels of eating disorder psy-chopathology. These results may be interpreted asconsistent with a differentiation of individuals withBED from other obese individuals.

Summary. Empirical approaches to classificationhave provided some support for the validity of BEDespecially in terms of separation from the eatingdisorders. However, limitations in the number ofBED or obese non-BED cases10,12 in these studiesleave the relationship of BED to obesity unclear.

Also, in some studies limitations in the measure-ment of symptoms7,9,13 reduce the strength ofinferences that can be drawn from these data. Inlarge part, these studies tend to be conducted withexisting data sets that were not originally designedto conduct empirical studies of classification. Likeall eating disorder diagnoses, the validity of BED

would be more effectively tested with an empiricalapproach to classification study if it consisted of anassessment of eating disorder symptoms in a sam-ple carefully selected to provide a rigorous test ofdiscriminant validity.

Laboratory-Based Studies of BED

Supporting Information Table S2 provides a sum-mary of a variety of laboratory-based studies1624

examining the eating behavior of individuals withsignificant levels of binge eating, commonly meet-ing diagnostic criteria for BED. Early studies of this

type indicated that individuals with BED consumedmore calories than weight-matched individualswithout BED in several different stimulus condi-tions in the lab (e.g., instructions for binge eating,instructions for non-binge meal consumption).16,17

Using a variety of different paradigms, the funda-mental finding that individuals diagnosed withBED consume more calories than individuals with-out BED was supported in a number of earlyreports1820 and again in more recent studies.2224

Only rarely have researchers failed to find a BEDversus obese non-BED difference in caloric con-sumption in a laboratory paradigm.19

Several of these studies have attempted to exam-ine the possible mechanisms associated with thefinding that individuals with BED tend to consumemore calories. For example, Telch and Agras18

reported that the level of clinical depression wasassociated with a propensity to label an eating epi-sode as an eating binge, regardless of caloric intake,thus potentially implicating negative mood inbinge eating. However, Geliebter et al.20 failed tofind that depression was correlated with caloricintake. Similarly, Sysko et al.22 reported that BEDwas associated with less fullness following foodintake, potentially implying a disturbance in a sati-ety mechanism. However, Geliebter et al.20 failed tofind differences between BED and non-BED partic-ipants in terms of postmeal satiety ratings. Gusset al.21 were intrigued by the fact that BED partici-pants in their study increased their caloric con-sumption only in response to a binge eatinginstruction and not in a typical meal instruction.These authors speculated on the potential disinhi-bitory mechanism that may be stimulated by aninstruction to binge eat. Consistent with this idea,

VALIDITY AND UTILITY OF BINGE EATING DISORDER



4/19

Galanti et al.24 report that test meal intake for bingeeating ranges widely, and is highly correlated withimpulsivity.

Summary. Laboratory studies offer some of thestrongest validity data regarding BED and particu-larly its discrimination from obesity. Like all labora-

tory studies, questions remain about the ecologicalvalidity of such findings. In other words, are BED/non-BED differences in eating specific to the labo-ratory, or do these findings generalize to the naturalenvironment? Studies addressing this concern arereviewed next.

Ecological Momentary Assessment and BED

Supporting Information Table S3 outlines threestudies25,26 and another study has recently beencompleted (Engel et al., submitted for publication)that utilize ecological momentary assessment(EMA) to examine the BED diagnostic construct in

terms of eating behavior in the natural environ-ment. This methodology is an interesting comple-ment to laboratory-based studies in that criticismsregarding ecological validity are presumablyreduced. Two of these studies25,26 had obese-BEDparticipants and obese non-BED participants carrypersonal data assistants for the assessment ofeating disorder behavior, stress, and mood over aperiod of 1 or 2 weeks. Thus, these studies alloweda comparison of reports of binge eating by bothBED and non-BED obese individuals. Of note, inthese reports, participants themselves determinedwhether or not an actual eating binge had occurred

(i.e., self-identified reports). Both studies foundsimilar results: there were no significant differencesbetween BED and non-BED participants in bingeeating episode frequency. Furthermore, the Greenoet al.s25 study included a dietary recall and thesedata similarly revealed essentially no differencebetween the two groups in the caloric intake duringa binge eating episode. However, these studiesrevealed that BED individuals experienced moreemotional distress and perceived loss of controlpreceding binge eating episodes. To the extent thatthese studies imply that BED and non-BED obeseparticipants do not differ in the size or frequency

of binge eating in the natural environment, theypose serious threats to the validity of the BEDdiagnosis.

Recently, Engel et al. (submitted for publication)completed a similar study with obese individualswho varied in terms of BED status, except that thisparadigm used a more objective means of deter-mining whether or not an eating binge hadoccurred. Rather than relying on participantsdetermination of a binge eating episode, these

researchers used the Nutritional Data System forResearch to calculate dietary recalls, along withmomentary assessments of loss of control withhandheld computers, and only considered an eat-ing episode a binge eating episode if it involvedmore than a 1,000 kcal and a self-reported percep-tion of a loss of control. Using these criteria, this

study found that individuals diagnosed with BEDwere more likely than obese non-BED individualsto display binge eating on a daily basis in the natu-ral environment, thus providing support for theBED diagnosis. This study, in spite of having a rela-tively small sample size, provides a more objectiveassessment of the validity of the BED constructwith EMA than previous studies.

Summary. This small set of studies provides mixedsupport for the validity of BED.

Epidemiological Data and BED

Supporting Information Table S4 provides a sum-mary of epidemiologically relevant studies ofBED,2729 which provide data relevant to its con-struct validity. Hudson et al.29 studied 2,980respondents from the National Comorbidity Study-Revised. This sample was weighted for the possibil-ity that respondents may have been more likely todisplay another psychiatric disorder because of thenature of the sample. The data are somewhat lim-ited in value by the fact that the study did not applythe DSM-IV BED diagnostic criteria precisely.Nonetheless, as Supporting Information Table S4

indicates, BED was more common than other eat-ing disorders and had a later age of onset than ei-ther AN or BN. Furthermore, BED displayed a lon-ger duration than the other eating disorders andwas more likely to be associated with obesity. Thisis an important study that demonstrates evidenceof valid distinctions between BED and other eatingdisorders in terms of various clinical characteris-tics, but is limited by the absence of a rigorouscomparison of a group of obese individuals who donot meet diagnostic criteria for BED. Furthermore,as with any large epidemiological sample of adults,measurement may be limited by retrospective

recall-related errors and biases.Two other studies have used large community-

based samples to examine epidemiologic and clini-cal characteristics associated with BED. Striegel-Moore et al.28 studied a sample of 2,046 women inan effort to examine race differences (i.e., Blackversus White) on eating disorder diagnoses.Although this study focused on racial factors, someof the findings are relevant to the current review.For example, although not explicitly tested in a

WONDERLICH ET AL.



5/19

statistical sense, Black participants did not displayany cases of AN, though both BN and BED werewell represented in the Black participants. Overall,White women were more likely than Black womento meet diagnostic criteria for any of the three eat-ing disorders. Also, there was some evidence thatBED was associated with a later age of onset than

other eating disorders. In another study, Pikeet al.27 examined 150 women with BED and 150healthy comparison participants. Women with BEDhad higher body mass index (BMI) scores and werealso more likely to be obese than non-BEDsubjects. Furthermore, among BED subjects, Blackparticipants were more likely to be obese thanWhite participants. Additionally, BED subjectsdisplayed more evidence of eating disorder psycho-pathology, but this was more characteristic ofWhite BED participants than Black participants.Paralleling this finding, BED was associated with

elevated rates of psychiatric comorbidity, but thiswas more characteristic of White women thanBlack women.

Summary. These studies, relying on large commu-nity-based samples, imply that BED differs fromAN and BN in a host of epidemiologic variablesincluding age of onset, racial composition, associa-tion with obesity, and psychiatric comorbidity.

Studies of Psychiatric Comorbidity and BED

Supporting Information Table S5 displays BEDstudies on psychiatric comorbidity.3038 Several

empirical studies suggest that obese individualswho binge eat have higher rates of Axis I and Axis IImental disorders than overweight or obese individ-uals who do not binge eat.3033 Moreover, Telch andAgras34 and Yanovski et al.33 provide evidence thatthe presence of comorbid mental disorders is spe-cifically related to whether or not people engage inbinge eating, rather than their level of obesity.These findings suggest that binge eatings associa-tion with comorbid psychopathology is not merelydue to its correlation with obesity. In addition,compared to BN, BED has been associated withlower levels of comorbid psychopathology across a

number of studies30,36,37 and with equivalent ratesof anxiety and mood disorders in at least onestudy.35

Two recent population-based studies have beenpublished on comorbidity and BED. The first, byGrucza et al.38 consisted of administering a PatientHealth Questionnaire (PHQ) to a population-basedsample that was recruited by random digit dialing.Of the 884 participants, 67 were categorized asBED, based upon their answers to the PHQ. They

found that individuals with BED (with or withoutobesity) had significantly higher rates of majordepression, generalized anxiety disorder, and panicattacks than individuals without BED. Obesity with-out BED was not associated with higher rates ofpsychopathology. A major strength of the study isthat it is population-based, and therefore may

reduce the likelihood of inflated levels of comorbidmental disorders that could exist in clinical sam-ples (i.e., Berksons bias). A notable weakness of thestudy is the method by which the psychiatric diag-noses were made. Individuals self-rated binge eat-ing episodes without being provided with a cleardefinition of objectively large binge eating. Thismay have led to false positives in the BED group.In addition, BED criteria in the study did not matchthe DSM-IVs criteria exactly. The PHQ asks if theperson has engaged in binge eating twice per weekfor 3 months as opposed to the DSM-IVs 6-monthtime period.

Finally, in a large, nationally representative, pop-ulation-based study, Hudson et al.29 comparedindividuals with AN, BN, and BED in terms ofcomorbid mental disorders. They found that, of theparticipants with eating disorders, individuals withBN were at the highest risk for comorbid disorders[94.5%; odds ratio: 4.7 (4.37.5)], followed by indi-viduals with BED [(78.9%; odds ratio: 2.3 (2.63.7)],and that individuals with AN were at the lowest risk(relatively) for comorbid disorders [56.2%; 2.1 (1.22.9)].

Summary. These studies provide evidence that BEDpatients display significant psychiatric comorbidity(roughly comparable to other eating disorders)that cannot be simply explained by the presence ofobesity.

Studies of Medical Outcome and BED

Supporting Information Table S6 lists the fewstudies3945 that have examined medical outcomesassociated with BED. Fairburn et al.39 compared 5-year course and outcome between community-based cohorts with BED (n 5 48) or BN (n 5 102).The BED and BN cohorts did not differ significantly

in terms of 5-year weight gain, although the BEDgroup (4.2 6 9.8 kg) did gain more weight than theBN (3.3 6 10.1 kg) group. A greater proportion ofBED subjects had a BMI[ 30 at follow-up (39% vs.20%, p 5 0.06). However, the proportion of BEDsubjects with a BMI[ 30 was also greater at base-line (22% vs. 12%).

In a large cross-sectional epidemiological studyof 4,651 female primary care and obstetric-gynecol-ogy patients, 245 BED and 44 BN cases were




6/19

diagnosed based upon self-report information.40 Incomparison to patients with no psychiatric diagno-ses, BED was associated with greater health impair-ments, more severe physical symptoms, and higherrates of diabetes. These differences remained aftercontrolling for co-occurring psychiatric diagnoses.However, these analyses did not control for differ-

ences in BMI. Few differences in medical outcomeswere found between BED and BN patients.

Bulik et al.41 examined 5-year medical outcomesamong 59 obese women with binge eating as com-pared to 107 obese women without binge eating ina population-based longitudinal study of Whitetwins. Binge eating at Wave 1 was determinedbased on interview-based response to a lifetimebinge eating question. The obese binge eatinggroup reported significantly greater health dissatis-faction than the obese group at 5-year assessment.Although the rates of all major medical disorders

(hypertension, visual impairment, asthma/respira-tory illness, diabetes, cardiac problems, osteoar-thritis) were higher for the obese binge eatinggroup, none of these differences reached statisticalsignificance. The authors note, however, that theprobability of the binge eating group having ahigher prevalence for all six medical conditionsby chance alone would be p 5 0.015, which isinterpreted as support for a BED-negative healthrelationship.

Reichborn-Kjennerud et al.42 present cross-sec-tional questionnaire data from 8,045 Norwegiantwins on the relationship between medical

symptoms and binge eating in the absence of com-pensatory behaviors. Eating disorder diagnoses andeating-related behaviors were assessed throughself-report responses to nine specific diagnosticquestions. BMI, lifetime occurrence of physicalsymptoms, and disorders were established basedupon self-report. After controlling for BMI, bingeeating was associated with increased risk of insom-nia and early menarche in women, and increasedrisk of neckshoulder, lower back, and chronicmuscular pain, as well as greater impairment dueto physical health in men.

At the recent Eating Disorder and ClassificationConference in Washington D.C., Hudson43 pre-sented a 2.5 year follow-up study comparing medi-cal outcomes in 137 individuals with BED, 139BMI-matched controls without BED, and 51 con-trols without BED not matched for BMI. Therewere no differences between BED and non-BEDgroups in weight change, blood pressure, or indi-vidual metabolic syndrome components (dyslipi-demia, hypertension, IGF/IGT). However, BED wasassociated with greater risk than the non-BED

groups in the development ofany(i.e., at least one)metabolic syndrome component. The BED groupalso reported more frequent health care visits thanthe non-BED groups, a finding that has beenreported previously.45 The analysis of the 5-yearoutcomes is currently underway.

Hasler et al.44 conducted a prospective commu-nity-based study of 591 young adults followed over a20-year period from age 19. They found that bingeeating four or more times in the last year as assessedwith a structured diagnostic interview was positivelyassociated with both increased weight gain andbeing overweight. Binge eating remained positivelyassociated with being overweight even after control-ling for a variety of sociodemographic (e.g., gender,income, education, family history of weight prob-lems), psychopathological (e.g., SCL-90R scores,atypical depression, generalized anxiety disorder),and behavioral (e.g., physical activity) factors.

Finally, although not equivalent to a BED diagno-sis, there is evidence that binge eating and loss ofcontrol eating in children and adolescents conferan increased risk for negative medical outcomes interms of increased body fat, worsening insulin re-sistance, and increased triglicerides,46,47,48 as wellas increases in psychological symptoms49,50 andreduced efficacy of weight loss interventions.48

These findings are important because they mayhelp to ultimately clarify that aspects of a BED di-agnosis are most detrimental to physical and psy-chological health.

Summary. Overall, support for the idea that BED orbinge eating predicts weight gain or negative medi-cal outcomes is both limited and mixed. Many ofthe studies are limited by small sample sizes. How-ever, the largest prospective study of adults44 showsan association between binge eating and weightgain even after controlling for a variety of potentialconfounds. Finally, there is good evidence to sug-gest that BED appears to be associated withincreased health care utilization.

Studies of Health-Related Quality of Life

(HRQOL), Life Satisfaction, and

Functional Impairment

In the following section several studies investi-gating health-related quality of life, life satisfaction,or functional impairment in BED patients arereviewed (see Supporting Information Table S7).Although these constructs are considered by mostto be distinct from each other, we view them asrelated constructs and at least partially overlap-ping. Given the large number of empirical reportsaddressing these constructs, studies were chosen

WONDERLICH ET AL.



7/19

based on the following criteria: (1) number of BEDpatients included in the study, (2) variety in thechosen comparison groups, (3) inclusion of a vari-ety of generic and disease-specific outcome meas-ures, and (4) scientific rigor.

A number of studies have shown that obese BEDparticipants report greater impairments in genericHRQOL, life satisfaction, or functional impair-ment.5155 Masheb and Grilo54 reported that BEDparticipants have considerably impaired scores onthe SF-36 Health Survey compared to those seen inU.S. norms. Additionally, obese BED patients reportlower HRQOL than obese non-BED patients.51,53

Further, women with BED generally report func-tional impairment at similar levels to groups of eat-ing disorder patients.40 Finally, disease-specificHRQOL instruments have shown a similar patternof results: obese BED patients report greaterimpairments than obese non-BED patients in over-

all HRQOL, as well in a variety of specific domainsof HRQOL.55

One interesting study attempted to partial outthe unique contribution of BED on HRQOL afteraccounting for possible confounding differencesbetween obese BED and non-BED groups. Kolotkinet al.52 compared 95 BED patients to 435 obesenon-BED patients and found that the BED groupwas comprised of a higher percentage of females,was younger, more likely to be White, had a higherBMI, and was more psychologically distressed.Importantly, after controlling for these group differ-ences, BED status was unrelated to obesity-specific

HRQOL, suggesting that the diagnosis of BED didnot uniquely predict HRQOL in obese individuals.

Summary. BED patients report more HRQOL andfunctional impairment than a variety of compari-son groups, although the unique association ofHRQOL and binge eating is unclear.

Longitudinal Studies of BED Diagnostic Status

Supporting Information Table S8 depicts a seriesof longitudinal studies39,5658 (Agras et al., submit-ted for publication) that have examined either thediagnostic stability of BED in terms of categorycrossover, or its propensity for recovery or remis-sion relative to other eating disorders. Each of theselongitudinal validators provides meaningful clinicalinformation. In terms of diagnostic stability, theevidence is relatively unclear. Agras et al. (sub-mitted for publication) reported that 5060% ofindividuals with subclinical or full syndrome BEDare best classified as presenting with a BED-likedisorder 12 months following the initial assess-ment. Fairburn et al.39 found much less evidence

that individuals with a BED diagnosis retained thatdiagnosis at 5-year follow-up. Only 9% of the 48individuals diagnosed with BED initially werefound to still meet diagnostic criteria for the syn-drome. Cachelin et al.56 completed a small follow-up study of individuals with BED, and also foundconsiderable variability in maintenance of the BED

diagnosis at 6-month follow-up. Finally, Fichterand Quadflieg58 longitudinally studied a largegroup of various eating disordered patients, includ-ing 60 BED patients. They found little evidence ofdiagnostic crossover between BED and AN. How-ever, there was considerable transition from BED toother forms of EDNOS or BN. Only 1.7% of BEDpatients retained their diagnosis at 2-year follow-up. Thus, these longitudinal data suggest that BEDis highly distinct from AN, and there is little evi-dence to suggest that BED represents a partiallyremitted form of BN.

In terms of recovery rates, another longitudinalvalidator with considerable clinical relevance, Fair-burn et al.39 reported that 85% of their BED casesrecovered within 5 years, and that individuals withBED were more likely to recover than individualswith BN. In a 4-year follow-up study, Agras et al.(submitted for publication) found that 50% of thepreviously diagnosed cases recovered from BED.Similarly, individuals with the subclinical form ofBED recovered in 48% of the cases. Using a verydifferent strategy, Pope et al.57 reported an averageduration of BED of 14.4 years in their family studyof BED probands and their relatives. They con-

cluded that BED is a highly stable disorder that dis-plays more chronicity than AN or BN. However,this study is significantly limited, because numberof years with BED was determined based upon theretrospective recall of participants. Finally, Fichterand Quadflieg58 reported the recovery rates at 2, 6,and 12 years to be 65, 78, and 67%, respectively.

Summary. Although there is variability in the data,it does appear that BED differs from other eatingdisorders in terms of a greater tendency toward re-covery and fluctuation, although this may be em-bedded in a chronic pattern of remission and

relapse. It also appears that those with BED are lesslikely than AN or BN to crossover to another activeeating disorder.

Treatment Studies and the BED Diagnosis

Three types of treatment-related studies areexamined below. The first type of study examinesthe moderating influence of the BED diagnosis onweight loss interventions in obese or overweightsamples. The second type of study compares the




8/19

efficacy of multiple treatments delivered to BEDparticipants and a comparison group (e.g., obese,non-BED) in what may be called a BED-nonBEDtreatment specificity design. A third design com-pares different treatments (e.g., cognitive-behav-ioral therapy [CBT] versus behavioral weight loss)applied to subjects all meeting criteria for BED in

what may be termed a BED multiple-treatmentdesign. Each of these designs will be reviewed anddiscussed below.

Studies Examining the BED Diagnosis

As a Moderator of Behavioral Weight

Loss Treatments

Supporting Information Table S9 provides a sum-mary of studies5974 examining the moderatingeffect of a BED diagnosis on behavioral and surgi-cal treatments for obesity. Evidence that the pres-ence of BED has a significant influence on weight

loss interventions would be an important factor toconsider for clinical utility.

Behavioral Weight Loss. A series of behavioral weightloss or very low calorie diet (VLCD) studies wereconducted before the official DSM-IV BED criteriawere established. Consequently, these studies donot provide a stringent test of the current BEDdiagnostic criteria, but are worthy of brief mention.All three of these studies5961 categorized obeseindividuals into those with binge eating or thosewithout binge eating, and examined the influenceof this classification on behavioral weight loss or

VLCD interventions. All three studies5961

failed todetect any classification-related effect on responseto these treatments. As noted, the measures ofbinge eating used in these studies probably wouldoverlap with the DSM BED diagnosis, but areclearly not isomorphic with that construct. Yanov-ski et al.62 did find that BED impacted several out-comes in VLCD interventions. For example, BEDpatients were more likely to drop out, displayextreme deviations from the weight loss protocol,and also showed less weight loss in the refeedingphase of the VLCD. However compared to non-BED obese patients, BED individuals did not distin-

guish themselves in overall adherence to the proto-col, weight loss at the end of protocol, or weightloss at 1-year follow-up. BED subjects were moreinclined to show substantial weight regain at 3-month follow-up, which was perceived as a nega-tive outcome. However, Yanovski et al.s62 findingswere not supported by two other studies, whichfound that BED was associated with decreasedlike-lihood of dropout from behavioral weight loss pro-grams,63,65 greater weight loss at the end of treat-

ment, and to a lesser degree at the end of 1-yearfollow-ups. In a similar study, Raymond et al.71

compared individuals with BED to individuals withsubclinical BED or no binge eating in a VLCDweight loss program. They found no differences inweight loss at end of treatment, 1-year follow-up,or in the number of sessions completed by either

group. This study is limited by the fact that subclin-ical binge eaters were included in the obese non-binge eater group, which may have obscuredpotential differences.

In a study of the impact of BED on weight loss inan effectiveness design, Pagoto et al.68 examinedthe impact of a clinically derived BED diagnosis onweight loss in a 16-visit outpatient weight loss pro-gram. All patients (n 5 131) who enrolled in aweight loss program were included in the analysis.The sample was on average morbidly obese (aver-age BMI 5 43.08), and the obesity program was

designed to prevent diabetes through lifestyle inter-ventions. There was a significant effect of diagnosison weight loss, with individuals exhibiting baselineBED diagnoses losing substantially less weight(weight change 5 23.10%) when compared toobese individuals without BED (weight change 525.35%). Furthermore, only 16% of the individualswith BED met the program goal of a 7% weightloss, which is substantially less than the total sam-ple percentage of 31%. It is also worth noting thatthese researchers examined the effect of a diagno-sis of major depression on weight loss, and foundthat individuals with major depression lost signifi-

cantly less weight (M 5 25.28%) than individualswithout major depression (M 5 23.29% weightloss). Also, this study reports that BED was notassociated with dropout, whereas major depressionwas. In spite of the value of this study in terms ofevidence that BED moderates weight loss, it failedto examine the predictive power of either BED ormajor depression when controlling for the other di-agnosis. Therefore, the unique predictive power ofBED, above and beyond major depression, was nottested.

As part of the Look Ahead trial, Gorin et al.70

studied 5,145 individuals with Type 2 diabetes. Allindividuals had BMIs greater than 25 kg/m2, andwere randomized to either an intensive lifestyleintervention or a diabetes and support educationtreatment condition. The researchers used theQuestionnaire of Eating and Weight Patterns toassess binge eating and classified 123 subjects as aBED group, whereas 4,222 subjects remained in thecomparison group. In terms of weight loss, the in-tensive lifestyle group had a more favorable out-come than the diabetes support and education

WONDERLICH ET AL.



9/19

condition, regardless of BED status. Across treat-ment conditions, baseline BED diagnosis did notmoderate outcome for either weight loss or reduc-tion of risk factors for cardiovascular disease. Theauthors conducted more intensive analyses ofbinge eating behavior of subjects, and found thatindividuals who retained their BED diagnosis at the

end of treatment lost considerably less weight thanindividuals who met criteria for baseline BED diag-nosis but failed to meet diagnostic criteria at theend of treatment. They concluded that individualswith BED should not be discouraged from enteringbehavioral weight loss programs and, in fact, theynoted the possibility that weight loss programs canimprove binge eating. In other words, this studyconveyed no data to support the role of BED as amoderator of treatment; however, the assessmentstudy for BED represents a weakness, especially fordiagnostic inferences.

Bariatric Surgery. Supporting Information Table S9provides a summary of studies7274 examining themoderating effect of the BED diagnosis on surgicalinterventions for weight loss. Burgmer et al.72

examined a variety of forms of binge eating in asample of bariatric surgery candidates. Theseresearchers failed to find evidence that preopera-tive binge eating predicted weight loss at 12-monthfollow-up, although the number of DSM-IV BEDpatients was insufficient for an optimal test of thediagnostic construct. Similarly, Busetto et al.73 in amore rigorously assessed patient group, failed tofind evidence that BED was associated with differ-

ential weight loss following surgery and at 5-yearfollow-up, but found that BED was associated witha greater likelihood of surgical complications. Inthe only study that provided evidence of BED as asignificant moderator of weight loss following bari-atric surgery, Sallet et al.74 found that 2 years afterRoux-en-y surgery, individuals with BED or sub-threshold BED lost less weight than individuals notmeeting these diagnostic criteria. Thus, the moder-ating significance of BED on weight loss in surgicalpatients is mixed.

Although not directly related to BED, there is

recent evidence suggesting that postoperativebinge eating or so called loss of control eatingpredicts negative outcomes in bariatric surgerysuch as less weight loss or more weight regain.7578

Although again not a precise model for BED, thesefindings may help to clarify those aspects of BEDthat are most predictive of weight gain.

Meta-Analysis of the Moderating Effect of BED on WeightLoss Treatment. In a unique meta-analysis, Blaineand Rodman69 identified samples of obese individ-

uals with BED who had received some form ofweight loss treatment (i.e., psychotherapeutic,drug, surgery). In a creative strategy, they identifiedsamples of obese non-BED individuals in the litera-ture who had received the same fundamental treat-ment and matched them to the BED sample on var-iables including sample size, percentage of females

in the sample, mean age, treatment type, and meansample pretreatment BMI. This meta-analysis isunique in that the investigators were able tocreate comparisons between BED and non-BEDsamples of obese individuals, which were not tech-nically conducted in the same study. Thirteen pairsof matched samples were submitted to meta-ana-lytic analysis that examined the impact of BED onweight loss and decreases in depression. ObeseBED samples lost significantly less weight (averageweight loss 5 1.3 kg) compared to obese non-BEDsamples (average weight loss 5 10.5 kg), but therewere no differences between the groups in the

extent to which depression scores were reducedduring weight loss treatment, although both groupsshowed significant reductions in depression.

The Blaine and Rodman meta-analysis69 isunique and implies that BED has clinical utility interms of its impact on weight loss. However, twokey limitations should be noted. First, although theactual treatments that were matched between sam-ples may be a rough equivalent, they are clearly notcomparable in a highly detailed comparison andshould not be considered equal. Although treat-ments may be roughly comparable in length andfrequency of sessions across studies, the details ofthe procedures cannot be equated. Second, theassessment of BED in this meta-analysis was basedlargely on self report measures such as the BingeEating Scale and the Questionnaire of Eating andWeight Patterns-Revised. Although these measuresmay be correlated with a DSM-IV BED diagnosis,they are not isomorphic with such a diagnosis.

Thus, the idea that BED will moderate the effec-tiveness of weight loss treatments, such as behav-ioral weight loss programs or bariatric surgery, isinconsistently supported in the empirical literatureand the overall effect appears small. This is signifi-

cant because these studies do not indicate that be-havioral weight loss experts or bariatric surgeonsshould modify their treatments for individualswith BED.

BED-nonBED Treatment Specificity Designs. Treatmentspecificity designs may offer a powerful test of thevalidity and utility of a construct, if the designincludes at least two different groups receiving atleast two different treatments. This has long beenconsidered a powerful test of the clinical utility for




10/19

diagnoses.5 In such a design, one is able to deter-mine if a diagnostic group or classification variablemoderates treatment response across all tested treat-ments and whether each treatment tested shows dif-ferential effectiveness across diagnoses. Two studieshave included both multiple diagnostic groups andmultiple treatments in treatment studies of BED (see

Supporting Information Table S9). Porzelius et al.64compared obese individuals with differing levels ofBED in two behavioral weight loss programs. One ofthe programs was more behavioral in nature, whilethe other relied on a more broad-based cognitive-be-havioral strategy. There was a trend for more severebinge eaters to lose more weight in the cognitive-be-havioral intervention than the strictly behavioralintervention and for more moderate binge eaters torespond more favorably to the behavioral program.There was no significant difference between treat-ments for those with less severe binge eating.Although this finding reflects some degree of speci-

ficity, the assessments of BED were made with a self-report instrument for binge eating and consequentlyare somewhat limited.

Nauta et al.66,67 completed a study comparingbehavioral weight loss treatment to a cognitive ther-apy designed to promote weight loss using cognitiveand behavioral techniques to reduce binge eating orovereating. BED subjects differed from non-BEDsubjects in terms of greater decreases in weight andshape concerns over time, increases in self-esteem,and weight regain after treatment. However, therewas not evidence that BED participants showed adifferential treatment response to one treatmentover the other across a broad range of measures.

These two studies suggest that the diagnosis ofBED has a small effect on treatment responseregarding weight loss, but evidence regarding spec-ificity of the treatment types was contradictory. It isimportant to note that the number of studies thathave actually utilized BED-nonBED treatmentspecificity designs is extremely limited, in contrastto a growing number of BED, multiple treatmentdesigns that are reviewed below.

BED Multiple-Treatment Designs. In the eating disor-der literature, there have recently been a growingnumber of treatment specificity studies that com-pared the relative efficacy of different treatmentsfor BED in samples comprised solely of BED sub-jects.7982 (see Supporting Information Table S9).These studies converge to indicate that individualsmeeting criteria for a BED diagnosis show greaterreduction in binge eating and associated psychopa-thology when receiving CBT or IPT than whenbeing treated with treatments that may be viewedas nonspecific, such as fluoxetine or behavioral

weight loss programs and provide useful informa-tion about differential treatment approaches forBED. These findings suggest that BED is nota non-specific class that responds equally to virtually alltreatments (including behavioral weight loss) andthat specialty treatments that target eating disorderpsychopathology are preferred when the desiredoutcome is reduction of binge eating. Thus, BEDmultiple treatment designs provide data supportingthe clinical utility of BED in that the presence ofthe diagnosis predicts a superior reduction in bingeeating when provided with specific treatments(e.g., CBT or IPT). Such information is valuable to apracticing eating disorder clinician who is pre-sented with a BED case and needs to make a choiceabout the most effective treatment.

However, it is important to note that this is a dif-ferent question than that addressed in the previoussections in studies examining the moderating influ-

ence of a BED diagnosis on weight loss treatments.Studies that include at least two diagnostic groupsanswer a different question than treatment designsthat include only one diagnostic group. This issuemay be clarified with an example. Kendell5 makesthe point that therapeutic trials can be useful fortesting clinical validity of psychiatric diagnoses andhighlights the power of multiple diagnostic groupsin treatment studies for the purpose of diagnosticvalidation. He offers the example of differentiatingbetween two diagnoses (i.e., hypochromic anemiaand pernicious anemia) through the demonstrationthat pernicious anemia responds only to vitamin

B12 therapy while the more common hypochromicanemia responds to oral iron therapy, thus provid-ing differential treatment inferences while simulta-neously validating the diagnostic distinctionbetween the two conditions. In this example, notonly are two treatments compared, but they arealso compared across two classes or diagnosticgroups. The inclusion of the diagnostic comparisongroup enhances the strength of inference about thevalidity of the diagnostic distinction by accountingfor treatment outcome variance that can be attrib-uted to diagnostic variability (i.e., two types of ane-mia). On the other hand, consider a similar studyusing only one diagnostic group, similar to theBED, multiple treatment design. If we conducted astudy in which only one class of patients (i.e., per-nicious anemia) was randomized to either B12 ororal iron treatments, we may find that patientsreceiving B12 respond in a superior fashion. Withsuch results, could we infer that pernicious anemiais diagnostically distinct from hypochromic ane-mia? In the absence of a diagnostic comparisongroup, such an inference is not justified. Paralleling

WONDERLICH ET AL.



11/19

this example, assigning subjects with BED to eitherCBT or behavioral weight loss, and demonstrating adifferential treatment response, does not supportthe discrimination of BED from an alternative class(e.g., obesity).

The fundamental point is that these two researchdesigns answer different questions. Designs thatinclude only one diagnostic group, comparedacross different treatments provide a specific typeof information. If one treatment is superior toanother for a given diagnostic class, it is evidenceof clinical utility because the diagnosis informs aclinician to select one treatment over another thathas been shown to be more efficacious. Obviously,this has substantial relevance to the practicing cli-nician who makes diagnoses and then applies whatis believed to be the most effective treatment forthat diagnosis.

On the other hand, multiple diagnostic group

designs help to determine if two putatively differ-ent diagnostic groups respond differently to one ormore treatments and test the validity of the diagno-ses. It is plausible to imagine such a design in theeating disorders in which individuals with BN andindividuals with BED are randomized to eitherinterpersonal psychotherapy or cognitive behav-ioral therapy-enhanced. This hypothetical studywould help us to determine if the BN-BED distinc-tion is valid in terms of treatment response to theseparticular treatments. If there were differencesbetween BN and BED in their response to eithertreatment, it would support the validity of these

diagnoses. If the interaction of diagnosis and treat-ment was significant, it would not only support thedistinction between BED and BN, but would alsohelp us to determine which treatment is mostappropriate for each diagnosis, conferring clinicalutility. Thus, BED-multiple treatment designs offervaluable data regarding clinical utility, while BED-nonBED treatment specificity designs provide use-ful information about validity.

Summary. The influence of BED on weight losstreatments is unclear and limited by a paucity ofrelevant studies. A unique meta-analysis provides

some support for the validity of BED on this issue.Empirical evidence supports the idea that severalspecialty treatments are superior to behavioralweight loss for reducing binge eating and associ-ated psychopathology in BED, which offers clinicalutility.

Family History and BED

Supporting Information Table S10 provides abrief overview of evidence that BED is a diagnosis

that tends to run in families,83,84 which may haveimplications for its etiology. Hudson et al.83 studied300 probands, half of whom met criteria for BED.Subjects were age- and sex-matched, and 888 firstdegree relatives of the probands were blindly inter-viewed for family history information. The dataindicated that relatives of BED probands were sig-

nificantly more likely to carry a BED diagnosis thannon-BED probands relatives. Furthermore, rela-tives of BED subjects displayed higher BMIs andprevalence of obesity even after controlling forproband obesity. In a follow-up study, Javaraset al.84 utilized the same data along with a Norwe-gian twin registry that included heritabilityestimates for binge eating. These researchers esti-mated that the heritability of BED in the familyhistory study was 0.57 and the heritability forbinge eating in the twin registry was 0.39. Thesedata imply that the propensity for BED to run infamilies may be largely accounted by the additiveeffects of genes.

Summary. Early evidence suggests that BED appearsto demonstrate familiality that may reflect geneticinfluences.

Genetic and Psychobiological

Validation of BED

Supporting Information Table S11 provides selec-tive review of information related to genetic andpsychobiological variables in BED subjects. Sup-

porting Information Table S11 provides a synopsisof six different candidate gene studies in BED,which examined several different polymor-phisms.8590 Two studies examined DSM-IV BEDdiagnoses in terms of the MC4R polymorphismand found no evidence that individuals with BEDwere any more likely to carry this polymorphismthan non-BED subjects.85,86 Similarly, Monteleoneet al.87 examined the cDNA 385C polymorphismand failed to find evidence that this genetic abnor-mality is associated with BED, although it didappear to be associated with obesity. Monteleoneet al.88 also failed to find any evidence of the valid-ity of BED in terms of the CLOCK polymorphism,but did find evidence that BED was associated withthe LL genotype or the l allele distribution on the5HTTLPR.89 However, because control subjectswere not obese in this study, it is unclear if thepolymorphism is related to binge eating or obesity.Finally, Davis et al.90 report that BED subjects witha 9 repeat allele of the dopamine transporter genedisplay more appetite suppression in response tomethylphenidate than non-BED individuals with




12/19

the same allelic expression, suggesting that thisgenetic variable may have some influence on BED.

A few studies have begun to examine BEDin terms of brain functioning (see Supporting Infor-mation Table S11).91,92 Karhunen et al.91 reportedthat BED subjects showed greater left hemisphereactivation using SPECT than non-BED control sub-jects, but only when presented with food-relatedstimuli. In this study, BED subjects showedincreased hunger that correlated with cerebralblood flow in the frontal and prefrontal areas.

Geliebter et al.92 used fMRI to study obese andlean women in terms of food-related activation ofspecific brain regions. Obese subjects with bingeeating showed more frontal activation in responseto binge food cues than other groups, but this wasapparently not true of lean subjects with binge eat-ing. Therefore, it is not clear if this brain activationfinding is related to binge eating or obesity. Fur-

thermore, the BED designation was not completelyconsistent with DSM-IV criteria, which limits thedegree of diagnostic validity that can be inferred.

As can be seen in the bottom section of Support-ing Information Table S11, a series of studies haveexamined the validity of BED on the basis of com-parisons of various peptides and hormones.9399

These studies have generally failed to identify areliable difference in particular peptides or hor-mones that is specifically associated with BED. Forexample, a growing number of studies examinedghrelin.9397 Two of these studies failed to find evi-dence that changes in ghrelin levels were associ-

ated with BED, above and beyond the effects ofobesity,93,95 while Geliebter et al.94,96,97 found thatBED was associated with decreased ghrelin, evenwhen subjects were comparable in terms of weight.Furthermore, BED appears associated with lowerbaseline ghrelin and less ghrelin decrease aftermeals. Other studies have examined cortisol inBED subjects98,99 and there was conflicting evi-dence that BED subjects showed differences in cor-tisol levels when compared to weight-matched sub-jects, with one study identifying a difference,98 andthe other failing to find a difference.99

Summary. There are not yet any genetic polymor-phisms or neurotransmitter/peptide findings thatare clearly and consistently related to BED. Studiesof brain function in BED are interesting andpreliminary.

Shape and Weight Concerns in BED

Several recent studies have examined the idea thatweight and shape concerns may help to improve theconstruct validity of the BED diagnosis. As Support-

ing Information Table S12 reveals, BED cases thatdisplay high levels of weight and shape concernappear to have a different profile of psychopathologythan BED cases without similar concerns.100103

Masheb and Grilo100 found that BED subjects(whether obese or not) tended to display lowerrestraint, but similar binge eating frequency and

concern about shape and weight as BN subjects.They concluded that BED cases display shape andweight concerns on par with BN. Three other studiesconverge to suggest that within the BED diagnosticconstruct, the presence of high levels of shape andweight concern help differentiate potential sub-groups of BED cases.101103 In each of these studies,it appears that high degrees of shape and weightconcern are associated with greater levels of psycho-pathology and impairment in comparison to BEDparticipant/patients with lower levels of shape andweight concerns. Furthermore, Grilo et al.103 indicatethat these differences remain the same even aftercontrolling for depression scores.

In an extension of these previous studies, Grilo104

presented preliminary new data examining the pre-dictive significance of overvaluation of weight andshape. Reanalyses of the previous studies105,106

examined the predictive value of overvaluation ofshape and weight and suggested that this cognitivefeature was a significant predictor of outcome, butnot a moderator (interact with specific treatments)in either study. Grilo104 also reported data examiningthree different treatments for BED (i.e., CBT, behav-ioral weight loss, and CBT/behavioral weight loss

combination), and again found that regardless oftreatment condition, overvaluation significantly pre-dicted post-treatment levels of binge eating fre-quency. These data suggest that overvaluation ofshape and weight in BED is not only a predictor ofconcurrent clinical variables, but also appears tohave predictive validity that is clinically meaningful.

Summary. Overvaluation of shape and weight dis-plays evidence of clinical utility across an array ofvalidating strategies and may warrant considera-tion as a diagnostic criterion or specifier.

Discussion

To examine the current status of BED as a mean-ingful diagnosis for the DSM-V, we adopted twoconcepts that have frequently been applied to diag-nostic discussions, but have varied in their defini-tions. The concepts of validity and clinical utilityhave been considered as essentially synony-mous,107 independent,108 or as separate but over-

WONDERLICH ET AL.



13/19

lapping constructs.109 Our conclusions rely heavilyon the ideas of Kendell and Jablensky108 in that weconsider BED to have validity if there is evidence ofa clear boundary between BED and its related orsimilar conditions. We consider BED to have clini-cal utility if it provides nontrivial information aboutthe prognosis and likely treatment outcomes, and/

or testable propositions about biological and socialcorrelates.108 Put differently, utility may depend ontwo things: (1) the quantity and quality of informa-tion in the literature, and (2) whether the implica-tions of that information, particularly about etiol-ogy, prognosis, and treatment are substantiallydifferent from the implications of analogous infor-mation about other related syndromes.

Is BED Valid?

The empirical classification research (i.e., latentstructure models, taxometrics) reviewed previously

highlights that BED is typically identified as a classseparate from AN-like or BN-like classes. Therefore,it seems to display a boundary with other tradi-tional eating disorders in empirical taxonomicstudies. However, studies assessing the boundarywith obesity are limited. As a result, a key boundarytest of validity for BED has not been conducted in asufficient number of studies to lend strong supportto the notion of BED is a diagnosis that is well dif-ferentiated from obesity in empirical classificationstudies. However, research designs that compare atleast two putative diagnostic groups on relevantvalidators may also help to identify possible diag-

nostic boundaries. These studies are summarizedin the next section.

Does BED Have Clinical Utility?

Again, evidence for a distinction between BEDand more traditional eating disorder diagnoses likeAN and BN has been summarized previously,110112

and is reasonably strong with evidence that BEDdiffers from AN and BN in a number of clinicallyrelevant variables, such as recovery rates, diagnos-tic stability, age of onset, gender distribution, BMI,dietary restraint, relative age of onset of dieting andbinge eating, psychiatric comorbidity, and binge

characteristics. Thus, BED appears to distinguishitself from current eating disorders.

However, the comparison to obesity is less clear.This is not to say that all cases of BED necessarilyrepresent a subset of the obese population or thatBED must distinguish itself from obesity in order tohave clinical utility (comparisons to existing eatingdisorders on a variety of other validators may dem-onstrate sufficient clinical utility of the diagnosis).Yet, discrimination from obesity on clinically signif-

icant validators would refute the idea that BEDcannot be discriminated from obesity with concur-rent psychiatric disturbance.113 To the extent thatsuch a model is true, it does not lend support toincluding BED as a diagnosis and implies that reli-ance on existing diagnoses, such major depressionor generalized anxiety, could be used to capture the

essential clinical information. Consequently, evi-dence that BED represents more than obesity withpsychiatric and psychological disturbance becomesimportant in the support for the creation of a neweating disorder diagnosis and is relevant to thisreview.

When compared to obese individuals withoutBED, individuals with the BED diagnosis consumemore calories in various eating episodes, althoughthis is better documented in lab settings than in thefield. There is also evidence that the BED diagnosisis correlated with significant functional impair-

ment, lower quality of life, and psychiatric comor-bidity. Additionally, there is preliminary evidencethat the BED diagnosis captures a phenomenonthat runs in families and seems to be accounted inpart by additive genetic variance. However, there iscurrently little evidence to suggest that BED is cor-related with any particular genetic polymorphism,peptide or neurotransmitter abnormality, or partic-ular pattern of dysfunction in brain function.

Most importantly for clinical classification pur-poses, however, is the question of whether BEDcontributes nontrivial information about prognosisand treatment outcome. The early studies on the

impact of BED on weight loss interventions weremixed and inconclusive. More recent studies con-tinue to be inconclusive, but a controlled meta-analysis provides some greater evidence for thepossible moderating influence of BED on weightloss. There is accumulating evidence that individu-als carrying a BED diagnosis show greater reduc-tion in binge eating and associated psychopathol-ogy with speciality treatments that target suchpsychopathology than with more generic and non-specific weight loss treatment. Thus, the diagnosisinforms treatment planning that is clinicallyrelevant.

Prospective prediction of negative medical orpsychiatric outcomes in individuals with BED isalso clinically useful information, but currentlyrather limited in the empirical literature.Although BED is often correlated with negativemedical status, studies are frequently cross-sectional with weak assessment of either BED ormedical status. The recent Look Ahead trial pro-vided a modicum of evidence that BED is associ-ated with medical parameters associated with




14/19

poor health, but the study by Haslam et al.44

offers more compelling prospective data support-ing this relationship.

Is BED Simply a Marker for Psychopathology?

A variant of the idea that BED is fundamentallyan eating disordered subtype of obesity is that itrepresents a marker for other psychopathology,rather than an eating disorder. Stunkard and Alli-son113 speculated that BED lacks the specific psy-chopathology and stability to warrant its inclu-sion as a true diagnosis and that it may beunderstood as a simple marker for generic, non-specific psychopathology. Based on the premisethat a diagnosis is designed to prescribe thecourse of an illness and a course of action,Stunkard and Allison reasoned that BED shouldnot be considered a viable diagnosis. They sug-

gested that BED is highly unstable and frequentlyremits. Moreover, they indicated that BEDresponds to a wide array of treatments, includingthose that do not target underlying psychopa-thology related to BED, such as behavioral weightloss. They reasoned that in the absence of datasupporting the predictive value regarding a reli-able clinical course and a specific response totreatment, BED should not be considered a dis-tinct eating disorder diagnosis.

Several pieces of data in this review addressStunkards and Allisons previous analysis and offernew findings in that regard. It remains true that

BED is more likely to remit than other eating disor-ders,58 (Agras et al., submitted for publication) butin spite of its remitting pattern, patients perceivethe disorder as stable and enduring, often lastingover a decade.57 Furthermore, the findings that cer-tain specialized treatments targeting specific BEDrelated psychopathology perform better in reduc-ing binge eating than do nonspecific behavioralweight loss treatments79,80 is not consistent withStunkards and Allisons hypothesis. These findingsare inconsistent with the idea that the diagnosis ofBED fails to carry any predictive value regardingcourse or clinical outcomes.

However, Stunkard and Allisons implicationsregarding the overlap between BED, depression,anxiety, and possibly personality traits, such asneuroticism, remain generally untested in the liter-ature. Few studies have examined the predictive va-lidity of BED after controlling for both obesity andsome type of affective disturbance and remainunclear in terms of their implications. On the onehand, there is evidence that the presence of BED,and particularly BED with overvaluation of shape

and weight, predicts eating disorder psychopathol-ogy beyond that accounted for simply by depres-sion.104 Similarly, there is evidence that binge eat-ing predicts long-term medical complications aftercontrolling for levels of affective distress.44 On theother hand, there is evidence to suggest that aftercontrolling for emotional distress and weight, BED

offers little in terms of incremental validity in theprediction of functional impairment.52 However,there is evidence that inclusion of certain criteria,such as overvaluation of shape and weight mayhelp to discriminate BED from depression and dis-tress, and produce a diagnostic construct withmore specific psychopathology.104 Although newdata challenge the original hypothesis of Stunkardand Allison, studies examining the relationship ofBED to other psychopathological constructs areneeded to further test this idea.

Conclusions

The following options may be considered regardingthe future of the BED diagnosis:

1. Retain BED as an example of EDNOS anda disorder in need of further study in theDSM-V.

Arguments in favor of this option would bethat some of the critical prospective or taxo-nomic tests of the validity or clinical utility of

BED have not yet been conducted. It could beargued that such a decision would promptfurther rigorous research that is needed tomore definitively test the diagnostic validityof BED. However, such an option overlookssignificant amounts of research that have al-ready been conducted, which may provideenough information to make an informed de-cision.

2. Include BED as a formal diagnosis in theDSM-V.

This option is supported by the well-documented number of patients who mayhave a disturbance in binge eating behavior,marked psychiatric comorbidity, functionalimpairment, and regularly report to psychiat-ric and eating disorder treatment facilities.Furthermore, there is preliminary evidencethat this is a relatively enduring and impairingcondition, which may have a currently unspe-cified familial basis. Additionally, the salientfeature of the diagnosis, binge eating, is differ-entially responsive to treatments. The primary

WONDERLICH ET AL.



15/19

argument against such inclusion is that criticalstudies testing the predictive value of the diag-nosis regarding clinical outcomes and responseto treatment are limited in number and rely onresearch designs that do not rigorously assessdiscriminant validity from related conditions.

3. Eliminate BED from the eating disorder diag-noses.

This option would be supported if there wasa compelling body of literature, suggestingthat BED cannot be distinguished from othereating disorders, obesity, or another psychiat-ric disorder, such as mood or anxiety disor-ders. It would also be supported if the diagno-sis did not predict other important biological,psychological, or social variables. As noted,the evidence that BED can be discriminatedfrom other eating disorders is substantial.There is also evidence on certain variables

(psychiatric comorbidity, functional impair-ment, eating behavior, and eating disorderpsychopathology) that BED can be discrimi-nated from obesity, but further studies on clin-ical validators are warranted. Additionally, thefinding that some specific treatments result ingreater symptom relief is clinically useful andimportant. Finally, the idea that BED could bereplaced by the application of mood or anxietydisorder diagnoses, particularly in overweightor obese individuals, with no net loss of infor-mation is not sufficiently supported in the lit-erature for serious consideration. Recent treat-

ment specificity findings are not consistentwith a nonspecificity theory and studies exam-ining overlap between mood or anxiety disor-ders and BED are insufficient in number tomake any strong conclusions.

4. Include BED in the DSM-V with overvaluationof shape and weight as a specifier. Support forthis option comes from a small number ofstudies, which have consistently suggestedthat overvaluation of shape and weight hasclinical utility and predicts a variety of out-comes above and beyond obesity, depression,and simple BED. Arguments against this

option would be based primarily on the lim-ited number of studies that have been con-ducted on this construct.

5. Include overvaluation of shape and weight asa criterion in the BED criteria set in theDSM-V. Support for this option rests in thelimited data suggesting that BED patientswith overvaluation regarding shape andweight show the clearest discrimination fromobesity. Arguments against this option would

be that a significant number of individualswith binge eating and possible distress andimpairment would be excluded from the di-agnosis if overvaluation of shape and weightwere required.

Summary

The diagnosis of BED outlined by the DSM-IV TaskForce has received considerable empirical atten-tion. Many of these studies have addressedresearch questions that are relevant to assessingthe validity and clinical utility of this diagnosis.These studies help to examine two key questionsregarding the status of BED. First, is BED distin-guishable from other eating disorders and does itconfer clinically useful information? There hasbeen relatively consistent and strong support for

the identification of a latent class resembling BEDacross a significant number of empirically basedclassification studies, lending support to the notionthat BED can be meaningfully discriminated fromother eating disorders. Furthermore, there is evi-dence that individuals with BED display rates ofeating disorder psychopathology, subjective dis-tress, impairments in quality of life, and psychiatriccomorbidity at a level that is similar to other eatingdisorder diagnoses, implying that BED is character-ized by clinically significant levels of psychopathol-ogy. Moreover, BED displays a clinical course thatdiffers from both AN and BN and diagnostic

crossover between AN and BED is extremely low.Thus, in relationship to other eating disorders, thediagnosis of BED seems to represent a diagnosti-cally distinct entity that carries clinically usefulinformation.

Second, can BED be discriminated from obesityand does the presence of BED confer clinicallyuseful information beyond that associated withsimple obesity or obesity with nonspecific psycho-pathology? There have been relatively few empiri-cal classification studies that have included bothobese BED subjects and obese non-BED subjectsin taxometric analyses or latent class analyses.Thus, a definitive number of studies examiningthe presence of an objective boundary betweenBED and obesity using these methodologies hasnot yet been conducted. However, laboratory-based studies suggesting that individuals withBED clearly consume more calories than obesenon-BED subjects provide some degree of supportfor the content validity of BED. The amount ofsupport for the clinical utility of the BED diag-nosis in relationship to obesity depends on the




16/19

validating variable chosen. There is strong evi-dence that obese subjects with BED experiencemore psychiatric comorbidity, eating disorder psy-chopathology, subjective distress, and impair-ments in quality of life than do obese non-BEDsubjects. Evidence that BED predicts a clinicalcourse associated with negative medical and psy-

chiatric outcomes is limited and would bolster theclinical utility of the diagnosis. The finding thatBED responds to some psychological treatmentsmore completely than to nonspecific treatmentsprovides evidence that the diagnosis has clinicalutility for treatment of binge eating behavior, buttreatment-related evidence supporting the validityof the diagnosis in relation to other diagnosticgroups is rare.

References

1. American Psychiatric Association. Diagnostic and Statistical

Manual of Mental Disorders, 4th ed., text rev. Washington,

DC: American Psychiatric Association, 2000.

2. Walsh BT. DSM-V from the perspective of the DSM-IV experi-

ence. Int J Eat Disord 2007;40:5357.

3. Striegel-Moore RH, Franko DL. Should binge eating disorder

be included in the DSM-V? A critical review of the state of the

evidence. Ann Rev Clin Psychol 2007;4:305324.

4. Blashfield RK, Sprock J, Fuller AK. Suggested guidelines for

including or excluding categories in the DSM-V. Comp Psychol

1990;31:1519.

5. Kendell RE. Clinical validity. Psychol Med 1989;19:4555.

6. Kendler KS. Toward a scientific psychiatric nosology. Arch Gen

Psychiatry 1990;47:699706.7. Bulik CM, Sullivan PD, Kendler KS. An empirical study of the

classification of eating disorders. Am J Psychiatry 2000;157:

886895.

8. Crow SJ, Agras WS, Halmi K, Mitchell JE, Kraemer HC. Full syn-

drome versus subthreshold anorexia nervosa, bulimia nerv-

osa, and binge eating disorder: A multicenter study. Int J Eat

Disord 2002;32:309318.

9. Mitchell JE, Crosby RD, Wonderlich SA, Hill L, le Grange D,

Powers P, et al. Latent profile analysis of a cohort of patients

with eating disorders not otherwise specified. Int J Eat Disord

2007;40:S95S98.

10. Williamson DA, Womble LG, Smeets MAM, Netemeyer RG,

Thaw JM, Kutlesic V, et al. Latent structure of eating disorder

symptoms: A factor analytic and taxometric investigation. Am

J Psychiatry 2002;159:412418.11. Striegel-Moore R, Franko DL, Thompson D, Barton B,

Schreiber GB, Daniels SR. An empirical study of the typology

of bulimia nervosa and its spectrum variants. Psychol Med

2005;35:15631572.

12. Wade TD, Crosby RD, Martin NG. Use of latent profile analysis

to identify eating disorder phenotypes in an adult Australian

twin cohort. Arch Gen Psychiatry 2006;63:13771384.

13. Eddy KT, Crosby RD, Keel PK, Wonderlich SA, le Grange D, Hill

L, et al. Empirical identification and validation of eating dis-

order phenotypes in a multi-site clinical sample. J Nerv Ment

Dis 2009;197:4149.

14. Gleaves DH, Lowe MR, Green BA, Cororve MB, Williams TL. Do

anorexia and bulimia nervosa occur on a continuum? A taxo-

metric analysis. Behav Ther 2000;31:195219.

15. Keel PK, Fichter M, Quadflieg N, Bulik C, Baxter MG, Thornton

L, et al. Application of a latent class analysis to empirically

define eating disorder phenotypes. Arch Gen Psychiatry 2004;

61:192200.

16. Yanovski SZ, Leet M, Yanovski JA, Flood M, Gold PW, Kissileff

HR, et al. Food selection and intake of obese women withbinge-eating disorder. Am J Clin Nutr 1992;56:975980.

17. Goldfein JA, Walsh BT, LaChaussee JL, Kissileff HR, Devlin MJ.

Eating behavior in binge eating disorder. Int J Eat Disord

1993;14:427431.

18. Telch CF, Agras WS. Do emotional states influence binge eat-

ing in the obese? Int J Eat Disord 1996;20:271279.

19. Telch CF, Agras WS. The effects of short-term food deprivation

on caloric intake in eating-disordered subjects. Appetite 1996;

26:221234.

20. Geliebter A, Hassid G, Hashim SA. Test meal intake in obese

binge eaters in relation to mood and gender. Int J Eat Disord

2001;29:488494.

21. Guss JL, Kissileff HR, Devlin MJ, Zimmerli E, Walsh BT. Binge

size increases with body mass index in women with binge-

eating disorder. Obes Res 2002;10:10211029.

22. Sysko R, Devlin MJ, Walsh BT, Zimmerli E, Kissileff HR. Satiety

and test meal intake among women with binge eating disor-

der. Int J Eat Disord 2007;40:554561.

23. Raymond NC, Bartholome LT, Lee SS, Peterson RE, Raatz SK. A

comparison of energy intake and food selection during labo-

ratory binge eating episodes in obese women with and with-

out a binge eating disorder diagnosis. Int J Eat Disord 2007;40:

6771.

24. Galanti K, Gluck ME, Geliebter A. Test meal intake in obese

binge eaters in relation to impulsivity and compulsivity. Int J

Eat Disord 2007;40:727732.

25. Greeno CG, Wing RR, Shiffman S. Binge antecedents in obese

women with and without binge eating disorder. J Consult Clin

Psychol 2000;68:95102.

26. le Grange D, Gorin A, Catley D, Stone AA. Does momentary

assessment detect binge eating in overweight women that isdenied at interview? Eur Eat Disord Rev 2001;9:309324.

27. Pike KM, Dohm FA, Striegel-Moore RH, Wilfley DE, Fairburn

CG. A comparison of Black and White women with binge eat-

ing disorder. Am J Psychiatry 2001;158:14551460.

28. Striegel-Moore RH, Dohm FA, Kraemer HC, Taylor CB, Daniels

S, Crawford PB, et al. Eating disorders in White and Black

women. Am J Psychiatry 2003;160:13261331.

29. Hudson JI, Hiripi E, Pope HG, Kessler RC. The prevalence and

correlates of eating disorders in the national comorbidity sur-

vey replication. Biol Psychiatry 2007;61:348358.

30. Fichter MM, Quadflieg N, Brandl B. Recurrent overeating: An

empirical comparison of binge eating disorder, bulimia nerv-

osa, and obesity. Int J Eat Disord 1993;14:116.

31. Telch CF, Stice E. Psychiatric comorbidity in women with

binge eating disorder: Prevalence rates from a non-treat-ment-seeking sample. J Consult Clin Psychol 1998;66:768

776.

32. Wilfley DE, Friedman MA, Dounchis JZ, Stein RI, Welch RI, Ball

SA. Comorbid psychopathology in binge eating disorder: Rela-

tion to eating disorder severity at baseline and following

treatment. J Consult Clin Psychol 2000;68:641649.

33. Yanovski SZ, Nelson JE, Dubbert BK, Spitzer RL. Association of

binge eating disorder and psychiatric comorbidity in obese

subjects. Am J Psychiatry 1993;150:14721479.

34. Telch CF, Agras WS. Obesity, binge eating and psychopathol-

ogy: Are they related? Int J Eat Disord 1994;15:5361.

WONDERLICH ET AL.



17/19

35. Striegel-Moore RH, Cachelin FM, Dohm FA, Pike KM, Wilfley

DE, Fairburn CG. Comparison of binge eating disorder and

bulimia nervosa in a community sample. Int J Eat Disord

2001;29:157165.

36. Raymond NC, Mussell MP, Mitchell JE, Crosby RD, de Zwaan

M. An age-matched comparison of subjects with binge eating

disorder and bulimia nervosa. Int J Eat Disord 1995;18:135

143.

37. Tobin DL, Griffing A, Griffing S. An examination of subtypecriteria for bulimia nervosa. Int J Eat Disord 1997;22:179

186.

38. Grucza RA, Przybeck TR, Cloninger CR. Prevalence and corre-

lates of binge eating disorders in a community sample.

Compr Psychiatry 2007;48:124131.

39. Fairburn CG, Cooper Z, Doll HA, Norman P, OConnor M. The

natural course of bulimia nervosa and binge eating disorder

in young women. Arch Gen Psychiatry 2000;57:659665.

40. Johnson JG, Spitzer RL, Williams JB. Health problems, impair-

ment and illnesses associated with bulimia nervosa and binge

eating disorder among primary care and obstetric gynaecol-

ogy patients. Psychol Med 2001;31:14551466.

41. Bulik CM, Sullivan PF, Kendler KS. Medical and psychiatric

morbidity in obese women with and without binge eating.

Int J Eat Disord 2002;32:7278.

42. Reichborn-Kjennerud T, Bulik CM, Sullivan PF, Tambs K, Har-

ris JR. Psychiatric and medical symptoms in binge eating in

the absence of compensatory behaviors. Obes Res 2004;12:

14451454.

43. Hudson J.2.5 Year follow-up study of binge eating disorder.

Paper presented at the Eating Disorders Diagnoses and Classi-

fication Conference, Washington D.C., September 2008.

44. Hasler G, Pine DS, Gamma A, Milos G, Ajdacic V, Eich D, et al.

The associations between psychopathology and being over-

weight: A 20-year prospective study. Psychol Med 2004;34:

10471057.

45. Striegel-Moore RH, Dohm FA, Wilfley DE, Pike KM, Bray NL,

Kraemer HC, et al. Toward an understanding of health serv-

ices use in women with binge eating disorder. Obes Res 2004;

12:799806.

46. Tanofsky-Kraff M, Cohen ML, Yanovski SZ, Cox C, Theim KR,Keil M, et al. A prospective study of psychological predictors

of body fat gain among children at high risk for adult obesity.

Pediatrics 2006;117:12031209.

47. Field AE, Skinner H, Corliss H, Horton N. Loss of control eating

as a predictor of weight gain and the development of over-

weight, depressive symptoms, binge drinking, and substance

use. Invited talk at the Classification and Diagnosis of Eating

Disorders Meeting: Scientific evidence regarding eating disor-

der classification for the Diagnostic and Statistical Manual,

5th ed. Arlington, VA, 2009.

48. Tanofsky-Kraff M, Yanovski SZ, Yanovski JA. The significance

of loss of control eating in children and adolescents. Invited

talk at the Classification and Diagnosis of Eating Disorders

Meeting: Scientific evidence regarding eating disorder classifi-

cation for the Diagnostic and Statistical Manual, 5th ed.Arlington, VA, 2009.

49. Tanofsky-Kraff M, Faden D, Yanovski S, Wilfley D, Yanovski J.

The perceived onset of dieting and loss of control eating

behaviors in overweight children. Int J Eat Disord 2005;38:

112122.

50. Neumark-Sztainer D, Hannan P. Weight-related behaviors

among adolescent girls and boys: Results from a national sur-

vey. Arch Pediatr Adolesc Med 2000;154:569577.

51. Dominy NL, Johnson WB, Koch C. Perception of parental ac-

ceptance in women with binge eating disorder. J Psychol

2000;134:2336.

52. Kolotkin RL, Westman EC, Ostby T, Crosby RD, Eisenson HJ,

Binks M. Does binge eat

Documents

Wonderlich binge_IJED 2009