Psychiatry Research xxx (2011) xxx–xxx

PSY-06953; No of Pages 7

Contents lists available at ScienceDirect

Psychiatry Research

j ourna l homepage: www.e lsev ie r.com/ locate /psychres

Binge eating in binge eating disorder: A break-down of emotion regulatory process?

Simone Munsch a,⁎, Andrea H. Meyer b, Vincent Quartier a, Frank H. Wilhelm c

a University of Fribourg, Department of Psychology, 2, Rue de Faucigny, CH-1700 Fribourg, Switzerlandb University of Basel, Faculty of Psychology, Department of Clinical Psychology and Psychotherapy, Division of Applied Statistics in Life Sciences, Missionsstrasse 62a, CH-4055 Basel, Switzerlandc University of Salzburg, Institute of Psychology, Department of Clinical Psychology, Psychotherapy, and Health Psychology, Hellbrunnerstrasse 34, A-5020 Salzburg, Austria

⁎ Corresponding author at: University of Fribourg, Dede Faucigny, CH-1700 Fribourg, Switzerland.

E-mail address: simone.munsch@unifr.ch (S. Munsch

0165-1781/$ – see front matter © 2011 Elsevier Irelanddoi:10.1016/j.psychres.2011.07.016

Please cite this article as: Munsch, S., et al.,Res. (2011), doi:10.1016/j.psychres.2011.0

a b s t r a c t

a r t i c l e i n f o

Article history:Received 13 December 2010Received in revised form 9 June 2011Accepted 7 July 2011Available online xxxx

Keywords:Ecological momentary assessmentEating disorderBinge eatingMoodEmotion regulationAntecedents

Current explanation models for binge eating in binge eating disorder (BED) mostly rely on bulimia nervosa(BN)models although research indicates different antecedents for binge eating in BED. This study investigatesantecedents and maintaining factors in terms of positive mood, negative mood and tension in a sample of 22women with binge eating disorder using ecological momentary assessment during one week. Values fornegative mood were higher and those for positive mood lower during binge days compared with non-bingedays. During binge days, negative mood and tension both strongly and significantly increased and positivemood strongly and significantly decreased at the first binge episode, followed by a slight though significant,and longer lasting decrease (negative mood, tension) or increase (positive mood) during a 4-h observationperiod following binge eating. Binge eating in BED seems to be triggered by an immediate break-down ofemotion regulation. There are no indications of an accumulation of negative mood triggering binge eatingfollowed by immediate reinforcing mechanisms in terms of substantial and stable improvement of mood asobserved in BN. These differences implicate a further specification of etiological models and could serve as abasis for developing new treatment approaches for BED.

partment of Psychology, 2, Rue

).

Ltd. All rights reserved.

Binge eating in binge eating disorder: A brea7.016

© 2011 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

The core feature of binge eating disorder (BED) comprises loss ofcontrol and consumption of large amounts of food (AmericanPsychiatric Association (APA), 1994). Cognitive behavioral therapy(CBT) rationales inBEDare traditionally based on correspondingmodelsfor BulimiaNervosa (BN) and are shown to be the established treatmentfor the majority of BED patients (Vocks et al., 2009). In BN, negativemood has been shown to be an important antecedent by a number ofstudies (Polivy et al., 1984; Agras and Telch, 1998; Waters et al., 2001).According to the affect regulation model, individuals engage in bingepurge behavior to alleviate negative mood (Polivy et al., 1984) or bysubstitution of a less aversive mood state (trade off-theory, Kenardy etal., 1996). Masking theory suggests that rather than decreasing orsubstituting negative mood, binge eating serves as an attribution fornegative mood that masks other problems (Herman and Polivy, 1988).In other words, negative affect can be blamed on binge eating, whichseems to be more controllable to the person than the actual causes ofdistress. The escape theory (Heatherton and Baumeister, 1991) positsthat binge eating represents an attempt to “escape” from distressingself-awareness and to narrow attention to the immediate physicalsurroundings or stimuli (e.g. food). As a secondary effect, thehypothesized shift in awareness impedes higher level cognitive

activities such as inhibition and thus results in the release of previouslysuppressed binge eating behavior (Engelberg et al., 2007).

Recent studies using ecological momentary assessment (EMA) toovercome known limitations of retrospective recall (for an overview,see Shiffman et al., 2008) convincingly demonstrate that, in line withthe affect regulation model, negative mood increases and positivemood decreases before binge eating and vomiting, whereas after BNevents, negative mood decreases and positive mood increases again.Binge-purge behavior in BN thus seems to be reinforcing itself byimproving mood (Engel et al., 2006; Smyth et al., 2007; Engelberg etal., 2007). Another study from Crosby et al. (2009) investigatingpatterns of mood in daily lives of bulimic individuals corroborates thatnegative mood drives bulimic behavior.

Research regarding antecedents of binge eating in BED used to relyonmodels derived from BN, although there seem to be differences withregard to the binge cycle. For example Hilbert and Tuschen-Caffier(2007) comparing BED with BN patients using EMA for multipleassessments duringa two-days span found thatBED individuals not onlyreveal less dietary restraint, they also experience less intense negativemood than BN patients and tend to binge eat also when feeling onlymoderately negative. Further, BED in contrast to BN individuals turnedout to be vulnerable to negative mood in particular when theyconcurrently suffered from high levels of general psychopathology(Hilbert and Tuschen-Caffier, 2007). Together with another naturalisticstudy from Stein et al. (2007) assessing antecedents and consequencesof binge eating at seven intervals during seven consecutive days, thesefindings underline that also in BED negative mood was increased on

k-down of emotion regulatory process? Psychiatry

2 S. Munsch et al. / Psychiatry Research xxx (2011) xxx–xxx

binge eating days compared to non binge periods whereas there wereno such differences for ratings of positive mood. However in bothstudies, contrary to existing emotion regulationmodels, negative moodremained increased when measured immediately after binge eating instudent and patient populations (Hilbert and Tuschen-Caffier, 2007;Stein et al, 2007). The before mentioned studies shed light on possiblydifferent mechanisms driving binge eating in BED but their findingsremain limited as they did not consider longer time intervals than onemeasurement immediately after binge eating. As a consequence, thetime course of mood factors after binge eating in BED remains open.Further limitations of the before mentioned studies concern theapplication of solely the time-contingent sampling method of theStein et al. study as well as the short observationmethod of two days inthe Hilbert et al. study.

In summary, current research about the preceding and maintainingfactors of binge eating in BED indicates distinct processes such as lesspronounced negative mood and a lack of immediate reinforcement interms of a fast and pronounced decrease of aversive mood states afterbinge eating in BED. Thus, research on the concrete cues and reinforcingmechanisms of binge eating in BED will allow to further specifyetiological models and to engage in developing specialized andindividualized treatment options for BED patients.

The present study aims at extending findings of current naturalisticstudies regarding binge cycles in BED and sets out to investigate inmoredetail the binge cycle in BED. Besides negative mood we additionallyincluded potentionally meaningful mood factors such as positive moodand tension. We followed the temporal course of these characteristicsbefore, during, and after binge eating on binge and, for comparison, onnon-binge days in a small sample of overweight to obese female BEDindividuals randomized for participation in a treatment trial for BED. Tominimize influences of retrospective memory recall, participants wereinvestigated using ecological momentary assessment (EMA). Relative totraditionalquestionnaire-basedmethods, EMAreducesbiases associatedwith retrospective recall by shortening the interval between anexperience and its recall. Further the EMAmethod is thought to enhanceecological validity as it is carried outwithin the naturalistic environmentof the participant (Shiffman et al., 2008).

The following research questions were investigated: First, weexamined whether the daily courses of the different aspects of moodvaried between binge and non-binge days. Second, to shed light onbinge cycles in BED we not only examined the pre- but also the post-binge phase. To our knowledge, we are the first to examineconsequences of binge eating not only immediately after but during aprolonged time span during the day after binge eating. Third, accordingto Smyth and colleagues we further acknowledge that the binge eatingevent itself is affect-laden and probably influences estimates recalledimmediately after binge eating. Therefore we analyzed the trajectoriesof mood and tension by including or excluding the 30 min immediatelyprior to and the 30 min following the binge episode (Smyth et al., 2007).Fourth, to account for findings of current literature (e.g., Hilbert andTuschen-Caffier, 2007) we additionally included specific participantcharacteristics related to eating disorder and clinical features, i.e.comorbidity status, BMI, duration of the disorder, depressiveness andseverity of eating disorder pathology, which may all potentiallymoderate the temporal trend of negative mood, positive mood, andtension on binge eating days.

2. Methods

2.1. Participants

Datawere collected from obese individuals with BED presenting for participation ina treatment trial to evaluate the efficacy of a short version of a CBT treatment at theDepartment of Clinical Psychology and Psychotherapy of the University of Basel(Switzerland) (Schlup et al., 2009). The study was approved by the local ethicscommittee of the University Hospital of Basel. Inclusion criteria for the clinical trialincluded being aged between 18 and 70 years, having a BMI between 27 and 40 kg/m2,being free from severe medical conditions such as diabetes, heart disease, or endocrine

Please cite this article as: Munsch, S., et al., Binge eating in binge eatingRes. (2011), doi:10.1016/j.psychres.2011.07.016

disorders and meeting full DSM-IV-TR criteria for BED (American PsychiatricAssociation (APA), 2000) according to a specialized eating disorder interview (seediagnostic assessment below). Of the initially contacted 136 individuals 28 femaleobese individuals with BED fulfilled these inclusion criteria. As individuals participatedin a randomized trial to evaluate treatment efficacy of a short-term CBT approach,individuals were excluded if they were pregnant, participated in a diet orpsychotherapy, received weight loss medications (currently or during the last3 months), had previous surgical treatment of obesity, or met DSM-IV-TR (AmericanPsychiatric Association (APA), 2000) criteria for mental disorders warrantingimmediate treatment, as this may have influenced treatment response. As patientswho never exhibited a binge during the entire week were excluded from all analyses(see results section), the final sample consisted of 22 female obese individuals with anaverage age of 45.5 years (S.D.=12.0, range=21–65), a BMI (kg/m2) of 33.4(S.D.=6.8, range=24.4–55.5), a BDI of 12.8 (S.D.=8.7, range=3–32), an EDE totalscore of 2.43 (S.D.=0.86, range=0.93–4.22), an average number of binges accordingto EDE of 15.6 (S.D.=8.0, range=4–30), and an average age of first manifestation ofBED of 16.6 (S.D.=12.2, range=2–40). Six (27%) participants suffered from anadditional affective or anxiety disorder and one patient (3.6%) from a comorbid mentaldisorder on axis-II. As only one male participant could be recruited, we excluded thesedata from our analyses.

2.2. Measures and procedure

2.2.1. Diagnostic assessmentBED diagnosis and associated eating disorder pathology were assessed using the

Eating Disorder Examination (EDE, Fairburn and Cooper, 1993; Hilbert et al., 2004). TheGerman language screenings for mental disorders on axis-I (Mini-DIPS) (Margraf,1994) and axis-II (SKID-II) (Wittchen et al., 1997) were administered to assess currentand lifetime mental disorders. Interviewers were trained by the principal investigator(S.M.). In cases of discordance of interviewers regarding the diagnoses, the diagnosticprocess was reevaluated using video tapes of the diagnostic interviews.

2.2.2. Daily electronic diary, ecological momentary assessment (EMA)All patients gave written informed consent and were offered free treatment. Data

were collected for seven days before treatment onset using a personal digital assistant(PDA, Palm Tungsten E). According to Smyth et al. (2007) time-contingent assessmentintervals were decreased during the day to adjust for an increased likelihood of bingeeating in the evening. Time intervals were scheduled as follows: The first alarm waspreset individually at one and a half hours after awakening, the second alarm 5 h afterthe first, the third alarm 4 h after the second, the fourth alarm 3 h after the third and thefifth alarm 2 h after the fourth alarm. For event-contingent monitoring participantswere instructed to fill in the questionnaire whenever binge eating occurred.Participants were asked to fill in the questionnaire within a 30-min interval. Pleaserefer to (Munsch et al., 2009) for further details including test-theoretical character-istics of the questionnaires used in this study. Questions were either dichotomous,suggesting a yes or no response, e.g., item 1 “Did you experience binge eating since yourlast entry ?”, or corresponded to a computerized Likert-type or visual analogue scale(VAS) scale. Questionnaires were programmed and displayed using Pendragon Formssoftware (Pendragon Software Corporation, Libertyville, IL, USA; unpublished ques-tionnaire available from the authors). Feasibility of EMA, i.e. practicability (“Did youexperience any difficulties in filling in the electronic diary?”), acceptability (“How didyou feel during the week with the electronic diary entries?”, “Did the electronic diaryalarm go off too often?”, Was your daily routine disturbed?”; correlations among itemsbetween 0.47 and 0.72; mean=7.19, S.D.=2.24, N=20), representativeness (“Did theprevious week correspond to your usual weekly routine?”; mean=7.21, S.D.=2.55,N=17), and signal-compliance (“Was it possible for you to fill in the electronic diary30 min after the signal?”; mean=7.05, S.D.=2.28; Fahrenberg (2006)) were allmeasured on an 11-point Likert-type scale from 0 (“not at all”) to 10 (“yes, exactly”)according to a self-developed exit questionnaire, EXQ (Munsch et al., 2009). EMA-based signal-compliance (filling in of the electronic questionnaire within 30 min afterbeing alarmed) and recording compliance (rate of overall responses to time-contingentsignaling) were all assessed by EMA. Reactivity was registered according to the EXQ(“Did the frequency of binge eating change during the diary period?”, “Did you focusmore on your psychological well-being?”, “Did you benefit from filling in the diary?”,“Did the previous week correspond to your usual weekly routine?”; correlations amongitems were between 0.48 and 0.70; mean of averaged items=5.72, S.D.=3.29, N=16;for detailed information seeMunsch et al., 2009). EMA-based signal-compliance, i.e. theproportion of the number of recordings starting within 30 min after the signal to thetotal number of recordings was 0.87. Participants' mean absolute deviation of entryfrom the scheduled alarms (in min) was 16.0 (median=1.0, S.D.=34.8, N=722).

2.2.3. EMA of binge eating, negative and positive emotionsAfter answering the electronically administered entry question (“did you

experience a binge episode”) with yes, exclusively objective binge eating (OBE, i.e.binge eating defined as consuming unusually large quantities of food with a subjectivesense of loss of control) was assessed according to the German version of the EDE(Hilbert et al., 2004; Munsch et al., 2009) (for a critical discussion of concordance ratesof EMA-based and self-report-based measures in order to assess binge eating in BEDplease see Munsch et al., 2009). Daily course of negative mood, positive mood, and tensionwere assessed on a scale between 1 and 10 using theMood Assessment Inventory (MAI,

disorder: A break-down of emotion regulatory process? Psychiatry

3S. Munsch et al. / Psychiatry Research xxx (2011) xxx–xxx

German version by Feist and Stephan, 2007) developed for ambulatory assessment,which contains five empirically derived subscales: negative mood, positive mood,interest, tension, and sleepiness. Feist and Stephan (2007) reported sufficientcorrelations (r=0.60) of the Negative Mood Subscale with the Beck DepressionInventory (BDI, Beck et al., 1961; Hautzinger et al., 1995) and good test–retestreliability (r=0.70) in a student sample. Based on clinical experience and literature onmood in BED (Stickney andMiltenberger, 1999; Vanderlinden et al., 2001; Binford et al.,2004; Engel et al., 2006; Smyth et al., 2007) we additionally included the followingadjectives in the electronic questionnaire: bored, stressed out, anxious, sad, tense,lonely, and annoyed.

To combine the items of the MAI with the additional items, and to obtain a limitednumber of reliable, valid and interpretable measures of mood, all items were enteredinto an exploratory factor analysis using principal components as extraction andvarimax as rotation method. Items with loadings b0.4 were excluded from furtheranalyses. Based on the Scree plot and the Kaiser criterion (excluding components withEigenvaluesb1.0) we obtained three different factors. These factors represent scalescores i.e. they were computed by taking the mean across all items which loaded highlyon them. The first factor, which explained 10% of the variance was highly correlatedwith the MAI-scale “negative mood” (r=0.92) and was given this label. It containedthe MAI items discontented, depressed, and queasy plus the additional items bored,anxious, lonely, and sad. The second factor, which explained 39% of the variance washighly correlated with the MAI-scale “tension” (r=0.94) and received this label. Itcontained the MAI items calm, nervous, and agitated plus the additional items stressedout, tense, and annoyed. The third factor, which explained 15% of the variance washighly correlated with the two MAI-scales “positive mood” (r=0.92) and “interest”(r=0.94) and was given the label “positive mood”. It contained the MAI itemscheerful/merry, good, and happy of the positive mood scale, and fascinated, interested,and not interested of the MAI interest scale, with no additional items. Factor scoresbased on these three factors were then used as mood-related variables in subsequentanalyses.

2.2.4. Situational context of binge eatingThese were assessed by the two questions “where are you at the moment?” and

“whom are you with?”. Answers were recoded into the two variables “being athome/not at home” and “being alone/not alone“.

2.2.5. Trait specific characteristics moderating the impact of mood and tension on bingeeating

To analyze the moderating impact of participants' characteristics related to eatingdisorder and clinical features on temporal trends of positive or negative mood andtension during binge days, we included the following baseline variables: comorbiditystatus, baseline BMI, duration of BED (years since first manifestation of BED), severity ofeating disorder pathology (measured by the global score, GS, of the EDE), anddepressiveness (measured by the BDI, Hautzinger, 1991).

2.3. Statistical analysis

We used a random intercept model to analyze the data (Pinheiro and Bates, 2000).Random intercept models are special types of linear mixed models in which eachindividual is assumed to have his/her own intercept. This kind of model is suitable forcases where each subject follows his/her own time schedule (i.e. both the number oftime points and the time interval are allowed to vary from subject to subject) which isoften observed in EMA based studies. The distinction between binge days and non-binge days was based on whether at least one daily binge episode occurred or not. Forthe precise model equations please refer to Appendix A.

Model 1 tested whether the daily course of mood factors differed between bingeand non-binge episodes while allowing for a trajectory following a linear and quadraticpolynomial.

Model 2 tested for temporal trends in the mood factors before and after theoccurrence of the first daily binge. Hence we introduced an additional dummy variabledistinguishing between the pre- and post-binge phase during binge days that allowedus to model the temporal trends of these two phases independently. This modelcontained the interactions between time and each of the two dummy variables pre- andpost-binge phase. For the temporal trend during the pre-binge phase we includedpolynomials up to 5th degree as doing so improved model fit. The inclusion ofpolynomials higher than linear during the post-binge phase in contrast did not improvemodel fit andwe therefore only used a linear trend (see Appendix A for the exact modelequation). The variable time was centered to the first binge episode, separately for eachpatient and day. For this analysis we used a subsample covering binge days only. Weomitted days starting with a binge episode as such cases could not have been analyzedusing Model 2 since there would have been no mood factor ratings preceding a bingerating. This concerned 14 binge episodes stemming from 6 persons. We also omitted alldata points including and following the second binge episode within the same day asthe corresponding mood factor values might have been influenced by the first bingeepisode. This concerned 16 binge episodes coming from 12 persons.

According to Smyth et al. (2007) we tested in Model 3 whether trends of values ofmood factors prior to the first binge could still be observed after excluding the values atthe binge themselves. Thus in this model the values covering the 30 min immediatelyprior to and the 30 min following the binge episode were excluded to preventrecordings immediately associated with the binge event influencing the model results.

Please cite this article as: Munsch, S., et al., Binge eating in binge eatingRes. (2011), doi:10.1016/j.psychres.2011.07.016

For the temporal trend during both the pre- and post-binge phase we included onlylinear polynomials as polynomials of higher degree did not improve model fit (seeAppendix A).

Finally, Model 4 tested whether BMI, EDE global score, comorbidity status (y/n),number of years since first manifestation of BED, and depressivenes (BDI) moderatedthe temporal trends of negative mood, positive mood, and tension before and after thefirst daily binge as assessed in Model 2. This model therefore included in addition to theterms listed in Model 2 the moderator (main effect) plus the interaction betweenmoderator and time for the pre- and postbinge phase. Each moderator was tested in aseparate model.

Note that in all fourmodelswe did not include an individual random slope coefficientb1i (Singer and Willett, 2003) as doing so would not have improved model fits. Moodfactors for negativemood and tensionwere both transformed logarithmically (natural) tomeet model assumptions. To analyze these models we used the software SPSS 14.Reported significances are based on an alpha of 0.05 unless otherwise specified.

3. Results

Of the 28 patients 6 never exhibited a binge during the entire weekand were excluded from all analyses. These six patients did not differfrom those reporting one or more binges during the study period withrespect to age, educational level, BMI, BDI, BAI, EDE total score, firstmanifestation of BED (years), and number of binge episodes accordingto EDE (pN0.05 for each t-test performed). For the remaining 22patients filling in the diary five times a day during the entire week atotal of 770 possible time-contingent data entries were possible. Theyactually completed 651 records, corresponding to a compliance rate of85%. In addition, 36 event-contingent data entries were recorded,resulting in a total of 687 data entries of which 75 (11%) concernedbinge episodes. Each patient had on average 0.49 binge episodes perday.

Most binge episodes occurred in the afternoon (52%, 12:00–18:00,N=39) and in the evening (39%, 18:00–24:00,N=29), very fewwereobserved during the night (4%, 24:00–06:00, N=3) and in themorning (5%, 06:00–12:00, N=4). Models 2 and 4 covered only bingedays and thus included 198 records and Model 3 in addition excludedall measurements within 30 min before and after a binge episode andincluded 149 records.

3.1. Situational context of binge eating

The proportion of participants being at home rather than not athome was 67% (N=576) during non-binge periods, 83% (N=29)immediately before a binge episode and also 83% (N=35) during abinge episode. In the same way the proportion of participants beingalone rather than not alone was 46% (N=576) during non-bingeperiods, 72% (N=29) immediately before a binge episode and 63%(N=35) during a binge episode.

3.2. Daily course of negative mood, positive mood, and tension (Table 1)

3.2.1. Model 1Values for negative mood were significantly higher during binge

than non-binge days without showing any particular daily trendneither during binge nor non-binge days. Values for positive moodwere significantly lower during binge than none-binge days,especially later during the day. However, no significant daily trendscould be found. Values for tension did not vary between binge andnon-binge days but increased during the day until the afternoon andthen decreased again, both during binge and non-binge days.

3.2.2. Model 2The average time period between the first measurement in the

morning and the first reported binge episode was 7.23 h (S.D.=3.29).For negative mood, there was a significant curvilinear increaseimmediately before the first binge episode, with particular highrates of increase shortly before the first binge (solid lines in Fig. 1a).The linear post-binge trend was significantly negative. Positive mood

disorder: A break-down of emotion regulatory process? Psychiatry

Table 1Daily course of negative mood, positive mood and tension. Regression coefficients for statistical models 1–3.

ln(negativemood)×1000

ln(positivemood)×1000

tension×1000

β (SE) t β (SE) t β (SE) t

Model 1 β10 intercept: estimated value at 2>h/>pm on a non-binge day 905 (104) 8.74⁎⁎⁎ 6045 (275) 22.0⁎⁎⁎ 1073 (103) 10.7⁎⁎⁎

β11 difference between binge and non-binge days at 2 pm 223 (49.8) 4.48⁎⁎⁎ −422 (170) −2.48⁎ 57.3 (52.9) 1.08β12 linear trend at 2 pm on a non-binge day −0.1 (4.40) −0.02 17.1 (15.1) 1.14 −2.64 (4.67) −0.56β13 quadratic trend on a non-binge day −0.28 (0.88) −0.36 −3.59 (3.00) −1.19 −2.09 (0.93) −2.25⁎

β14 difference in linear trend at 2 pm between binge andnon-binge days

7.12 (6.85) 1.04 −24.5 (23.4) −1.05 5.32 (7.27) 0.46

β15 difference in quadratic trend between binge andnon-binge days

0.22 (1.19) 0.18 −10.3 (4.05) −2.53⁎ 1.19 (1.26) 0.95

Model 2 β20 intercept: estimated value at first daily binge 1379 (105) 13.1⁎⁎⁎ 4579 (355) 12.9⁎⁎⁎ 1251 (119) 10.5⁎⁎⁎

β21 linear trend during pre-binge phase immediately beforefirst daily binge

950 (183) 5.19⁎⁎⁎ −2319 (710) −3.26⁎⁎ 651 (182) 3.59⁎⁎⁎

β22 quadratic trend during pre-binge phase immediately beforefirst daily binge

445 (102) 4.36⁎⁎⁎ −920 (396) −2.32⁎ 316 (101) 3.12⁎⁎

β23 cubic trend during pre-binge phase immediately before firstdaily binge

80.5 (20.5) 3.92⁎⁎⁎ −143 (79.5) −1.79 57.6 (20.3) 2.83⁎⁎

β24 quartic trend during pre-binge phase immediately before firstdaily binge

6.20 (1.71) 3.64⁎⁎⁎ −9.41 (6.61) −1.42 4.43 (1.69) 2.62⁎⁎

β25 quintic trend during pre-binge phase immediately before first daily binge 0.17 (0.05) 3.38⁎⁎⁎ −0.22 (0.19) −1.13 0.12 (0.05) 2.44⁎

β26 linear trend during post-binge phase −55.0 (16.3) −6.86⁎⁎⁎ 173 (62.8) 2.76** −38.3 (16.2) −2.37⁎

Model 3 β30 intercept: estimated value at first daily binge 785 (114) 6.86⁎⁎⁎ 6472 (401) 16.2⁎⁎⁎ 867 (124) 6.97⁎⁎⁎

β31 difference between estimated values 30 min before and 30 minafter first daily binge

−533 (137) −3.88⁎⁎⁎ 1103 (512) 2.15⁎ −308 (136) −2.26⁎

β32 linear trend during pre-binge phase −26.0 (13.5) −1.93 103 (50.5) 2.04⁎ −22.1 (13.4) −1.65β33 linear trend during post-binge phase −47.1 (22.1) −2.14⁎ 50.1 (82.2) 0.61 −25.3 (22.0) −1.15

⁎ pb0.05.⁎⁎ pb0.01.⁎⁎⁎ pb0.001.

4 S. Munsch et al. / Psychiatry Research xxx (2011) xxx–xxx

showed a significant curvilinear decrease before the first bingeepisode, with strongest rates of decrease shortly before the firstbinge, followed by a linear increase during the post-binge phase,which was also significant (solid lines in Fig. 1b). For tension therewas a significant curvilinear increase before the first binge episode,which was also most pronounced shortly before the first binge (solidlines in Fig. 1c). The post-binge trend for tension was significantlynegative.

3.2.3. Model 3If the values up to 30 min before and 30 min after the first binge

were omitted, a different pattern was observed. Values for negativemood and tension now decreased and positivemood increased duringthe pre-binge phase, these trends being significant for positive moodand short off being significant for negative mood (Fig. 1a–c, brokenlines). Trends during the post-binge phase, in contrast, were similar tothose in Model 2 for all three mood factors. Thus negative mood andtension both decreased and positive mood increased, these trendsbeing significant for negative mood only. Note that for each moodfactor predicted values 30 min after the first daily binge weresignificantly higher (negative mood and tension) or lower (positivemood) than those 30 min before it.

3.2.4. Model 4The trait specific factor comorbidity status moderated the

temporal course of tension prior to the first daily binge (not shownin table): patients suffering from comorbid mental disorders hadhigher rates of increase for tension immediately before the first dailybinge than patients without comorbid disorder. Also, patients withhigher depressiveness had lower rates of increase for tensionimmediately before the first daily binge than patients with lowerdepressiveness. The other person specific characteristics, baselineBMI, years since first manifestation of a binge episode, and EDE totalscore did not moderate the course of negative or positive affect andtension before and after the first daily binge.

Please cite this article as: Munsch, S., et al., Binge eating in binge eatingRes. (2011), doi:10.1016/j.psychres.2011.07.016

4. Discussion

The present study is to our knowledge the first to investigate dailycourses of mood and tension experience during binge and non-bingedays and before and after binge eating, thereby covering an extendedtime span in the natural environment of treatment-seeking womenwith BED.

In general the study findings corroborate findings from studies onBN and BED showing that negative mood ratings were higher andpositive mood ratings lower on binge days compared to non-bingedays (Smyth et al., 2007; Stein et al., 2007).

Considering binge days and temporal courses until the first dailybinge we found that positive mood, negative mood and tension allstrongly deteriorated immediately before the first daily binge (seeModel 2), as has been observed in BN (Engelberg et al., 2007; Smythet al., 2007). It must be noted that in general and even during bingeeating values for mood factors, especially regarding negative moodand tension, varied between 2.5 and 4 on a range of 1 to 10 which israther low (Fig. 1). Our analyses further revealed that the temporalcourse of mood and tension was independent of eating disorderseverity or body weight. However, individuals suffering fromadditional mental disorders were prone to higher increases in tensionshortly before the first daily binge than individuals without comorbiddisorders. Also more depressed individuals experienced a lowershort-term increase of tension before the first daily bingecompared toless depressed individuals.

Following the considerations of Smyth and colleagues we excludedthe values covering the time span 30 min before and 30 min after thefirst binge as these measures could be influenced by the affect-ladenevent of recent binge eating per se (Smyth et al., 2007). In a BNsample, Smyth and colleagues found that even after excluding thesemeasures, accumulation of mood deterioration remained a robustpredictor of binge eating. In contrast, in our sample of female BEDindividuals, excluding measures of the one hour's interval resulted ina strikingly different pattern. We even observed a slight improvementof mood up to 30 min before binge eating (Model 3) followed by an

disorder: A break-down of emotion regulatory process? Psychiatry

Fig. 1. Daily course of negative mood, positive mood, and tension before and after the first binge episode. Values for negative mood and tension were backtransformed from ln-transformation. Solid lines denote predicted values from linear mixed models during binge days and refer to statistical Model 2. Broken lines denote predicted values when disregardingvalues at the time of the binge (±30min) and refer to statisticalModel 3. Note that predicted values during non-bingedays are not included. Grey lines denotemeans and95% confidence-limits of observedvaluesandwere obtainedby computingmeans andconfidence limits of all observedvalueswithindefined time intervals. Intervals relative to the timevariate centered atthe first daily binge were: –6 h to –4 h/–4 h to –2 h/–2 h to b0 h/time at first daily binge/N 0 h to +2 h/+2 h to +4 h/+4 h to +6 h.

5S. Munsch et al. / Psychiatry Research xxx (2011) xxx–xxx

abrupt and significant deterioration (Model 2) immediately beforethe binge. These findings seem to contradict affect regulation theoryas hypothesized in BN, where binge eating is supposed to be the resultof an accumulation of negative affect. In contrast, our results indicatethat binge eating in BED might rather be the result of an immediatebreakdown of emotion and impulse regulation caused by suddenincreases of negative affect and tension, and a rapid decrease ofpositive affect. Our results are more in line with the assumption of theescape theory (Heatherton and Baumeister, 1991) in which a shortdecrease in self-awareness is thought to inhibit cognitive control andthus might contribute to triggering binge eating. They are alsoconsistent with Wegner's theory of ironic effects of inhibitory mentalcontrol processes, resulting in a sudden shift toward suppressed

Please cite this article as: Munsch, S., et al., Binge eating in binge eatingRes. (2011), doi:10.1016/j.psychres.2011.07.016

undesired behavior (Wegner, 1994). Further, these findings underlinethe importance of the concepts of affect liability and impulsivity,characterized by a trait-like tendency to experience rapidly shiftingaffective states as risk factors for binge eating in BED and BN (Svaldi etal., 2009; Anestis et al., 2010).

Regarding the long-term course of binge eating, our findingsindicate that after a mood deterioration immediately before a bingeepisode a rather slow but lasting improvement over several hoursfollowing the binge emerged (Fig. 1). Compared to the short-termrelief of distress after binge eating in BN due to both an increase ofpositive and a decrease of negative mood, in BED relief of aversivemood states after bingeing seems to be less pronounced and revealsitself only if longer time frames are considered (Smyth et al., 2007).

disorder: A break-down of emotion regulatory process? Psychiatry

6 S. Munsch et al. / Psychiatry Research xxx (2011) xxx–xxx

This finding does not seem to be astonishing when compared withbinge cycles in BN, as these are normally finished by purging behaviorleading to immediate, even though short-term, relief. As in BEDsuffering might decrease only gradually, this might explain whyprevious studies focussing on the immediate effect of binge eating onmood in BED did not indicate an improvement of mood as a functionof a binge episode (Hilbert et al., 2004; Munsch et al., 2009). Toinvestigate whether the slow improvement of negative and positivemood after binge eating have reinforcing properties as in BN andwhether they contribute to the maintenance of binge eating in BED,more frequent assessment time points after binge eating as well asaccompanying cognitions should be considered. Generally, reinforce-ment is considered more potent with closer time contingeny. Futureresearch should also focus on the investigation of the underlyingbiological mechanisms such as physiological stress levels to furtherinvestigate the different mechanisms in binge cycles between BEDand BN.

Possible criticisms include the lack of representativeness of ourstudy sample as it consisted of individuals fulfilling criteria toparticipate in a randomized treatment trial. As only very few menparticipated, we subsequently had to exclude them from our analyses.Further, the underpowered sample size limits the possibility to detectmoderator effects of mood factors and the sampling time of one weekbefore treatment begin was rather short. Another limitation is that weonly registered objective binge eating episodes even though there arenow considerable data indicating that the only difference betweenindividuals suffering from objective compared to subjective bingeeating is with respect to increased body weight for objective bingeeaters (Mond et al., 2010). We further cannot exclude thatparticipants did not report all occurrences of binge eating, eventhough we did not find any indications for a lack of compliance to theEMA procedure. Participants rated EMA to be an acceptable andfeasible method in their answering to our exit questionnaire. Theyfurther estimated that the one-week assessment period was repre-sentative and that they did not feel that their binge eating patternswere influenced by EMA (Munsch et al., 2009). Nevertheless, evenself-reported EMA remains a retrospective assessment method andmight itself be subject to memory and reporting bias in terms of e.g.mood-dependent recall. Future research could profit from more fine-grained and simultaneous diurnal analyses of the emotional andcognitive characteristics and of shape and weight concern before andafter binge eating using new methods such as automatic soundsampling, which provides observational data with short samplingintervals but nevertheless low subject burden (Mehl et al., 2001;Hilbert et al., 2009).

Overall our findings indicate that binge eating in BED mightrepresent the result of an immediate break-down of emotion andimpulse regulation attempts. After binge eating, in contrast to findingsfrom BN, we found a less pronounced and only slow recovery of moodafter binge eating in BED. With respect to therapeutic implications ofthis result, we ought to develop stimulus control strategies to increasealertness of considered individuals in order to prevent or detect firstsigns of upcoming urges to binge as especially these individuals willhave pronounced difficulties to inhibit binge eating once it hasstarted. Often applied training of response prevention in order tosuppress unwanted behavior during states of high mental load may incontrast further enhance ironic effects of suppression attempts and soincrease the probability of problematic behavior (Wegner, 2009).Response prevention strategies such as the acceptance of stressfulevents or the disclosure of mental states, may then be options to beconsidered in further research and clinical practice.

Appendix A. Supplementary data

Supplementary data to this article can be found online at doi:10.1016/j.psychres.2011.07.016.

Please cite this article as: Munsch, S., et al., Binge eating in binge eatingRes. (2011), doi:10.1016/j.psychres.2011.07.016

References

Agras, W.S., Telch, C.F., 1998. The effects of caloric deprivation and negative affect onbinge eating in obese binge-eating-disordered women. Behavior Therapy 29,491–503.

American Psychiatric Association (APA), 1994. Diagnostic and statisticalmanual ofmentaldisorders: DSM-IV, 4th ed. American Psychiatric Association, Washington DC.

American Psychiatric Association (APA), 2000. Diagnostic and statisticalmanual ofmentaldisorders DSM-IV-TR. American Psychiatric Association Task Force on DSM-IV, 4th ed.American Psychiatric Association, Washington DC.

Anestis, M.D., Selby, E.A., Crosby, R.D., Wonderlich, S.A., Engel, S.G., Joiner, T.E., 2010. Acomparison of retrospective self-report versus ecological momentary assessmentmeasures of affective liability in the examination of its relationship with bulimicsymptomatology. Behaviour Research and Therapy 48 (7), 607–613.

Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., Erbaugh, J., 1961. An Inventory forMeasuring Depression. Archives of General Psychiatry 4 (6), 561–571.

Binford, R.B., PedersonMussell, M., Peterson, C.B., Crow, S.J., Mitchell, J.E., 2004. Relationof binge eating age of onset to functional aspects of binge eating in binge eatingdisorder. The International Journal of Eating Disorders 35 (3), 286–292.

Crosby, R.D., Wonderlich, S.A., Engel, S.G., Simonich, H., Smyth, J., Mitchell, J.E., 2009.Daily mood patterns and bulimic behaviors in the natural environment. BehaviourResearch and Therapy 47 (3), 181–188.

Engel, S.G., Boseck, J.J., Crosby, R.D., Wonderlich, S.A., Mitchell, J.E., Smyth, J.,Miltenberger, R., Steiger, H., 2006. The relationship of momentary anger andimpulsivity to bulimic behavior. Behaviour Research and Therapy 45 (3), 437–447.

Engelberg, M.J., Steiger, H., Gauvin, L., Wonderlich, S.A., 2007. Binge antecedents inbulimic syndromes: an examination of dissociation and negative affect. TheInternational Journal of Eating Disorders 40, 531–536.

Fahrenberg, J., 2006. Assessment in daily life. A review of computer-assistedmethodologies and applications in psychology and psychophysiology, years2000–2005. Retrieved 8 February 2008. From http://www.ambulatory-assessment.org/.

Fairburn, C.G., Cooper, P.J., 1993. The Eating Disorder Examination. In: Fairburn, C.G.,Wilson, G.T. (Eds.), Binge Eating: Nature, Assessment, and Treatment. The GuilfordPress, New York, pp. 317–360.

Feist, A., Stephan, E., 2007. Entwicklung eines Verfahrens zur Erfassung desGefühlszustandes (VGZ)from http://psydok.sulb.uni-saarland.de/volltexte/2007/952/2007.

Hautzinger, M., 1991. The Beck depression inventory in clinical practice. Der Nervenarzt62, 689–696.

Hautzinger, M., Bailer, M., Worall, H., Keller, F., 1995. Beck-Depressions-Inventar (BDI).Hans Huber, Bern.

Heatherton, T.F., Baumeister, R.F., 1991. Binge eating as escape from self-awareness.Psychological Bulletin 110 (1), 86–108.

Herman, C.P., Polivy, J., 1988. Excess and restraint in bulimia. In: Pirke, K.M.,Vandereycken, W., Ploog, D. (Eds.), The Psychobiology of Bulimia. Springer,Munich, pp. 33–41.

Hilbert, A., Tuschen-Caffier, B., 2007. Maintenance of binge eating through negativemood: a naturalistic comparison of binge eating disorder and bulimia nervosa. TheInternational Journal of Eating Disorders 40 (6), 521–530.

Hilbert, A., Tuschen-Caffier, B., Ohms, M., 2004. Eating Disorders Examination: AGerman version of the Structured Eating Disorder Interviews Eating DisorderExamination: Deutschsprachige Version des strukturierten Essstorungsinterviews.Diagnostica 50 (2), 98–106.

Hilbert, A., Rief, W., Tuschen-Caffier, B., de Zwaan, M., Czaja, J., 2009. Loss of controleating and psychological maintenance in children: an ecological momentaryassessment study. Behaviour Research and Therapy 47 (1), 26–33.

Kenardy, J., Arnow, B., Agras, S., 1996. The aversiveness of specific emotional statesassociated with binge-eating in obese subjects. The Australian and New ZealandJournal of Psychiatry 30 (6), 839–844.

Margraf, J., 1994. Mini-DIPS: Diagnostisches Kurz-Interview bei psychischen Störungen.Springer, Berlin.

Mehl, M.R., Pennebaker, J.W., Crow, D.M., Dabbs, J., Price, J.H., 2001. The ElectronicallyActivated Recorder (EAR): a device for sampling naturalistic daily activities andconversations. Behavior Research Methods, Instruments, & Computers 33 (4),517–523.

Mond, J.M., Latner, J.D., Hay, P.H., Owen, C., Rodgers, B., 2010. Objective and subjectivebulimic episodes in the classification of bulimic-type eating disorders: another nailin the coffin of a problematic distinction. Behaviour Research and Therapy 48 (7),661–669.

Munsch, S., Meyer, A.H., Milenkovic, N., Schlup, B., Margraf, J., Wilhelm, F.H., 2009.Ecological momentary assessment to evaluate cognitive-behavioral treatment forbinge eating disorder. The International Journal of Eating Disorders 42 (7),648–657.

Pinheiro, J.C., Bates, D.M., 2000. Mixed-Effects Models in S and S-PLUS. Springer, NewYork.

Polivy, J., Herman, C.P., Olmsted, M.P., Jazwinski, C.M., 1984. Restraint and binge eating.In: Hawkins, C., Fremouw, W., Clement, P. (Eds.), The binge–purge syndrome.Springer, New York, pp. 104–121.

Schlup, B., Munsch, S., Meyer, A.H., Wilhelm, F.H., 2009. The efficacy of a short version ofa cognitive-behavioral treatment followed by booster sessions for binge eatingdisorder. Behaviour Research and Therapy 47, 628–635.

Shiffman, S., Stone, A.A., Hufford, M.R., 2008. Ecological momentary assessment. AnnualReview of Clinical Psychology 4, 1–32.

Singer, J.D., Willett, J.B., 2003. Applied longitudinal data analysis: modeling change andevent occurrence. Oxford University Press, New York.

disorder: A break-down of emotion regulatory process? Psychiatry

7S. Munsch et al. / Psychiatry Research xxx (2011) xxx–xxx

Smyth, J.M., Wonderlich, S.A., Heron, K.E., Sliwinski, M.J., Crosby, R.D., Mitchell, J.E.,Engel, S., 2007. Daily and momentary mood and stress are associated with bingeeating and vomiting in bulimia nervosa patients in the natural environment. Coverof Journal of Consulting and Clinical Psychology 75, 629–638.

Stein, R.I., Kenardy, J., Wiseman, C.V., Dounchis, J.Z., Arnow, B.A., Wilfley, D.E., 2007.What's driving the binge in binge eating disorder? A prospective examination ofprecursors and consequences. The International Journal of Eating Disorders 40,195–203.

Stickney, M.I., Miltenberger, R.G., 1999. Evaluating direct and indirect measures for thefunctional assessment of binge eating. The International Journal of Eating Disorders26 (2), 195–204.

Svaldi, J., Caffier, D., Blechert, J., Tuschen-Caffier, B., 2009. Body-related film clip triggersdesire to binge in women with binge eating disorder. Behaviour Research andTherapy 47, 790–796.

Please cite this article as: Munsch, S., et al., Binge eating in binge eatingRes. (2011), doi:10.1016/j.psychres.2011.07.016

Vanderlinden, J., Grave, R.D., Vandereycken, W., Noorduin, C., 2001. Which factors doprovoke binge-eating? An exploratory study in female students. Eating Behaviors 2(1), 79–83.

Vocks, S., Tuschen-Caffier, B., Pietrowsky, R., Rustenbach, S., Kersting, A., Herpertz, S., 2009.Meta-analysis of the effectiveness of psychological and pharmacological treatmentsfor binge eating disorder. The International Journal of Eating Disorders 43, 205–217.

Waters, A., Hill, A., Waller, G., 2001. Internal and external antecedents of binge eatingepisodes in a group of women with bulimia nervosa. The International Journal ofEating Disorders 29 (1), 17–22.

Wegner, D.M., 1994. Ironic processes ofmental control. Psychological Review101 (1), 34–52.Wegner, D.M., 2009. How to think, say, or do precisely the worst thing for any occasion.

Science 325 (5936), 48–50.Wittchen, H., Zaudig, M., Fydrich, T., 1997. Strukturiertes Interview für DSM-IV. Hogrefe

Verlag Testzentrale, Bern.

disorder: A break-down of emotion regulatory process? Psychiatry