10.1192/bjp.149.4.479Access the most recent version at doi:
1986 149: 479-485 The British Journal of Psychiatry
WH Kaye, HE Gwirtsman, DT George, SR Weiss and DC Jimerson
Relationship of mood alterations to bingeing behaviour in bulimia
Article cited in:
To obtain reprints or permission to reproduce material from this paper, please write
to this article at
You can respond
top right corner of the article or
Receive free email alerts when new articles cite this article - sign up in the box at the
The Royal College of PsychiatristsPublished by
on July 14, 2011 bjp.rcpsych.org
go to: The British Journal of PsychiatryTo subscribe to
British Journal of Psychiatry (1986), 149,479â€”485
Relationship of Mood Alterations to Bingeing Behaviour in Bulimia
day after hospitaladmission.Caloricintake,time spentbingeingand vomiting,and self
and objective ratings of mood indicated that anxiety decreased more frequently and to a
greaterextentthan depression,bothduringand afterbingeingandvomiting.The present
data, obtained in a controlledsetting, tend to confirm previousinformationon binge
episodes obtained by self-report from bulimic patients. Bingeing and vomiting episodes
may provide bulimic patients with a physiological mechanism for temporarily relieving a
dysphoric mood state.
The bulimic symptom
been found to be surprisingly
Printz, 1980; Halmi et a!, 1981; Pyle et a!, 1983;
Johnson el a!, 1984). Bulimia has been hypothesised
to be related to affective disorders.
bulimia(Pyleet a!, 1981;
Hudsoneta!, 1983a) andtheir first-degree
(Hudson et al, l983b) often have disturbances
mood and/ora history of substance
and affective disorders share certain neuro-endocrine
dysfunctions (Gwirtsmanet a!,1983), and the
bulimia symptom complex often responds to anti
depressants (Pope eta!, 1983;Walsh eta!, 1984).
Only a few studies, using patient self-report, have
examined the relationship between bingeingâ€”purging
and mood. The initiation of bingeing appears
associated with weight control (Chiodo & Latimer,
1983) or stress (Strober,
suggest that bingeing may serve to relieve tension
(Johnson & Larson, 1982; Abraham
1982). The relation between bingeingâ€”purgingbe
haviourand the global depression
with bulimia is not clear.
we designed this preliminary study to allow obser
vation and documentation of mood during and after
bingeing and vomiting episodes. Bulimics rated their
mood subjectively while bingeing and vomiting, and
for two hours afterwards. Objective mood ratings
were performed twice, once before and once after
mood was measuredby standardised
objective ratings some time after the bingeing and
vomiting were over.
The mechanisms, biological or otherwise, whereby
food intake alters mood are not known. Studies of
complex in women has
in the pastdecade
attention and has
duce alterations in certain brain neurotransmitters
such as serotonin(Wurtman,
mitter also implicated in mood regulation: therefore
it is important to determine whether different com
ponents of the food consumed, such ascarbohydrate,
influence mood directly.
Caloric intake and time spent in bingeing and
vomiting were documented
mood changes, because we were particularly curious
as to whether improvement
responsible for terminating bingeing and vomiting.
We therefore instructed patients to binge and vomit
until they achieved whatever effect was important to
suggest that carbohydratefoods can pro
1983), a neurotrans
in mood was the factor
wereadmittedtoan NIMH intramuralresearchward:clini
cal data for these subjects are given in Table I. All subjects
gave informed consent for the study. All patients had a
historyof bingeingand vomiting at least once perday for
threemonths priorto admission. Bulimicswho frequently
abused laxatives were excluded. We included one subject
withcurrent anorexianervosa(patient No. 7 in Table I).
All other subjects were between 82% and 118% average
body weight. (Metropolitan
1959). Three of the subjects had met DSM-III
anorexia nervosain the past but werecurrentlyover 82%
average body weight.
Subjectswereinformedof the studyprior to admission
minetheirchoice of bingefoods. Onthedayof admissiona
physical examination, electrocardiogram and laboratory
electrolyte abnormality, but none so severeas to exclude
them from this study. The study was performed on the
secondday of admission,aftermedicalclearance.
KAYE El AL
data for hulimicClinicalpatients
Because bulimics often prefer isolation during bingeing,
we decided to allow them to binge and vomit
a ward room. but with a physician
door. On the day after admission, subjects fasted from
midnight.They were placed alone in a ward bedroom
adjoining toilet facilities. At 9:00 am. an intravenous line
was established. At 10:00am. base-line mood ratings were
obtained and the first blood sample drawn.
then given their first tray of food. A physician entered the
room after each binge and vomit to supervise the self-report
ratings and draw blood. (Our biological findings will be
In an attemptto replicate
patients werepermitted to take as much timeas they desired
to binge and vomit. We tried to supply as many as possible
of the foods that subjects requested. Each subject was
offered a tray of 5000â€”6000 kcal of food. The amount of
food consumed was estimated by the dietitian by methods
described elsewhere (Petersen
instructed to continue to binge and vomit until they had
achieved whatever â€˜¿?desiredeffect' determined the point at
which they stopped at home; but with the exception of one
subject who binged and vomited fivetimes, we terminated
the study after four bingeâ€”vomit episodes even if the
â€˜¿?desired effect' had not occurred.
Our comparison group comprised seven healthy female
controls without personal or family history of psychiatric
illness. Each was given a large (900-1700
et al, 1986). Patientswere
kcal) test meal.
The controls did not vomit, and consumed their entire test
meal at one sitting over one hour.
Mood was quantifiedby several methods.
and all 7 controlsself-rated
factors on a 7-item visual analogue scalebefore eating. This
subjective rating was repeated by the bulimics after each
binge-and-vomit, and at hourly intervals after termination
of the last bingeâ€”vomit cycle(see Figure 2).The same scale
was given to the controls immediately after their test meal
and twicemore at hourly intervals. Two psychiatrists rated
seven of the bulimics on the Brief Psychiatric
(BPRS) (objectiverating)beforethey weregiven any food
and again about 20minutes after the last bingeâ€”vomit cycle.
BPRS ratings for the two raterswere averaged for each
bulimic. The Beck Depression
the Hamilton Depression Scale (Hamilton, 1960)and the
& Asberg, 1979) were administered byoneraterwithintwo
days of subjects completing the bingeing and vomiting
All 12 bulimics
in mood changes and other
Inventory(Beck et al, 1961),
rating scale (Montgomery
The caloric intake per binge and the time spent bingeing,
vomiting, or resting between cycles varied between subjects
MOOD ALTERATIONS AND BINGEINGIN BULIMIA 481
100 t-@-@-@ --0--
o 30 60 90120 150180210240270300
o---o CALM ANXIOUS
Fiu. 1.Time course, caloric intake, and changes in subjective mood ratings during and after bingeing and vomiting. The
horizontal dimension of the striped box indicates time spent bingeing, the vertical dimension indicates caloric intake per
binge (0-5000 kcal). The horizontal dimension of the black box indicates time spent vomiting. Visual analogue mood
ratings for anxiety (open circles) an'4 depression (black dots) are indicated for each patient (100=anxious, depressed;
0 = calm, happy). Patient 3 missed ons. iating.
(Fig. 1). However, every subject who engaged in multiple
bingeing and vomiting cyclesappeared to have a consistent
pattern of caloric intake from binge to binge, and of time
spent bingeing and vomiting.
binge was 30 minutes (s.c. 10) and the mean vomit duration
was 13 minutes (s.c.9). The mean individual binge con
sisted of 3500Â± 1338kcal, and total caloric intake over the
whole cycle of binges had a mean of 9360Â± 5275 kcal.
Macronutrient distribution was 52% Â±10%carbohydrate,
37%Â±9%fat, and 11%Â±2%protein.
The mean duration of each
Before eating, bulimic subjects rated themselves
cantly more depressed, more anxious, lesshungry, and less
confused than controls (Fig. 2). After bingeing,
subjects showed a significantly
greater decrease in anxiety,
and a smaller decrease in hunger, than the controls did after
their meal. Neither group showed .an overall change in
Each bulimic appeared to have a unique pattern of
changes in mood on the self-rating
bingeâ€”vomitcycles (Fig. 1). The great variation between
individuals in patterns of depression and anxiety are
apparent when graphed with caloric intake and time spent
bingeing and vomiting. For example, subjects 6, 7, and 8
eachhadthree cycles of bingeing
had her greatest improvement in anxiety after her second
binge and vomit, but less change in depression; and her
and vomit. Subject 7 had a substantial reduction of both
ject 8 had a step-wise reduction
scales during and after
afterthe third binge
in anxiety and depression
KAYE El AL
though their mood improved these subjects continued
to bingeand vomit. After all thecyclesofbingeing and
vomiting were over, most of these subjects had greater
residual depression than anxiety.
2. Four subjects (numbers1,2,7,9)
in anxiety and/or depression
time they stopped bingeing and vomiting.
3. Two of the twelve subjects (numbers 3and 5)did not
achieve any improvement in mood either during or
after bingeing and vomiting.
had an improvement
of at least 50% at the
-0-- ,@A substantial decreasein anxietyoccurred atsome point
during or after the bingeâ€”vomit cyclesineight of the twelve
bulimics, whereasonly fourhad a substantial reduction
in depression. Base-line self-rating analogue scores for
anxiety and depression were positively correlated (r =
0.643, P=0.02,n= 12), but there was no significant
relation (r=0.480, n= 12)between subjective anxiety and
depression after the end of the last bingeâ€”vomit cycle.
Base-line subjective and objective anxiety scores were sig
but base-line depressionscores were not.
- -- -
TO VOMIT â€˜¿?â€”C-@__--@@@-â€”---.
TO VOMIT _________________
The last seven bulimic subjects to participate
were assessed with the BPRS at 10:00a.m. (base-line) and
again 20 minutes after the end of bingeing and vomiting.
Pairedt-tests revealed a significant
(t= 3.74, P<0.0l)butnot in depression
measures on the 24-item BPRS were significantly reduced:
guilt (t= 3.07, P< 0.05)and tension (t= 4.8, P< 0.01).
in this study
(Fig. 3). Two
Global depression ratings
Relation of caloric intake to mood
There was a weak positive correlation between rate of food
consumption (calories per minute) during the bingeâ€”vomit
period and both initial Beck scores
n= 12) and Hamilton scores (r=0.546, P=0.07,
Rate of caloric consumptionshowed a negative correlation
with the percentage of carbohydrate in the chosen meal
(r= â€”¿?0.591, P.czO.05) and a positive correlationwith the
percentage of fat (r=0.554, P=0.06). There was no rela
during the binge study and the subject's total caloric
consumption, or her relative preference
protein, or fat.
Bulimics had a wide range of depression
Depression Inventory and the Hamilton Depression Scale
(Table I). Montgomeryâ€”Asberg
similar magnitude.There were no correlations
base-line or subsequent scores on the self-rating or BPRS
depression scales(completed on the day of the binge study)
when compared with the Beck, Hamilton,
Hamilton,Beck, and Montgomeryâ€”Asberg
significantlycorrelated with each other at the P< 0.01level.
scores on the Beck
depression scores were of a
FIG.2.Comparison of bulimic patients (n= 12)and healthy
control women (n=7) on visual analogue self-ratings.
Seven factors were self-rated at base-line (â€˜pre-binge'):
ratings were repeated â€˜¿?post-binge' (immediately
last binge and vomit for bulimics,
controls), and again at I and 2 hours â€˜¿?post-binge'.
are indicated by black circles,
circles.â€˜¿?a' indicatesa significant
ANOVA.â€˜¿?b'a significant difference in group x time inter
action (difference in patternof response between groups).
after the test meal for
On the basis
of our clinicalimpressions
if there was more than 50% change from base-line. Because
of heterogeneous response
categorised changes in mood into three patterns:
on the self-ratingscale, we
I. Of twelvesubjects, six(numbers 4,6, 8, 10,II, 12)had
a reduction of 50% or more in depression and/or
anxiety between cycles of bingeing and vomiting: even
MOODALTERATIONSAND BINGEING IN BULIMIA
stressed since this study was done on the day after
hospital admission, the BPRS ratings only indicated
studies, by their very nature,
because they take place in an environment
from that in which bingeing and vomiting usually
occur. It is difficult toknow
mental effects may have altered base-line mood and
responses in our study. The bulimics did volunteer
the information that they might not have begun a
binge in these circumstances ifon their own, but once
in the midst ofbingeingand vomiting they described
their responses as fairly typical.
factors, many subjects had a decrease in dysphoria
with bingeing and vomiting.
ture suggests thatmany
response in their more usual settings.
It is not clear what stimuli cause the subject to
terminate an episode of bingeing and vomiting. Only
4 of our 12subjects had an improvement in anxiety
exactly the time they ended bingeing and vomiting. It
is possible that we may not have rated the most sig
nificant change in mood: some subjects appeared to
be giddy, silly, or somewhat
they volunteered to stop bingeing and vomiting. It is
also possible,as Johnson
suggested, that patientsentering
gramme after years of bulimia
getting relief of symptoms by bingeing and purging.
As with substance abuse, tolerance may develop over
time: bingeing is then needed to maintain
dency rather than to get high. Our bulimics had been
how much environ
In spite of these
have a similar
with base-line) at
intoxicatedat the time
& Larson (1982) have
may no longer
Fio. 3. Brief psychiatric rating scale (BPRS) scores for anxiety and depression in 7 bulimic subjects. Scores just before
patients began bingeing and vomiting (â€˜pre-binge/vomit') are compared
cycle of bingeing and vomiting (â€˜post-binge/vomit').
to scores 20 minutes after they finished the last
The results of this study demonstrate heterogeneous
patterns of mood changes during and after bingeing
and purgingin bulimic
Beumont(1982)found that 34% ofpatients
from anxiety during the cycles of bingeing and 66%
cycles of bingeing had concluded.
subjective reports of decreased
(50% during bingeing,67% after bingeing)were
similar and were confirmed
Johnson& Larson (1982) did not measure
but reported a general worsening
during a binge in terms of greater guilt, shame and
anger: after the bingeâ€”purgeepisodes were over, the
patients-in the Johnson& Larson study were sadder,
drowsier, weaker, and more bored than usual.
Similarly, many of our subjects remained
during and after bingeing and vomiting. The DSM
III states that although
pleasurable, and vomiting
thoughts follow eating binges. While our results
generally support this statement,
reduction in anxietyor tension
after bingeing and vomiting.
described as being associated
by, dysphoria or stress (Abraham
Johnson & Larson, 1982;Strober, 1984). Our study
was performed irrespective of pre-binge mood state.
While it may be assumed
anxiety in our study
of affective state
eating binges may be
often reduces post-binge
we would add that a
often occursduring or
with, or precipitated
KAYE El AL
much earlier in the course ofthe disease would reveal
more consistent mood changes.
This study confirms patient
number ofcalories and proportions
fat and protein consumed
confirming that bulimics do not necessarily consume
high-protein or high-fat foods (Mitchell et a!, 1981;
Abraham & Beumont,1982). Our data tend to sug
gest that the rateof consumption
minute) bears some relationship
eaten and to mood.The bulimics
calories appeared to preferentially
hydrate and to avoid fat-containing
those that consumed the most calories per minute
had the opposite relationship.
rapid eaters preferred fat-containing
any available food is not clear
the bulimic subjectsan unlimited supply of food
to choose from. Food preferencesduring bingeing
might reflect differencesin hypothalamic neuro
chemical control of eating behaviour, since the
choices of fat and carbohydrate
separate systems (Leibowitz,
that consumed the most calories per minute had the
greatest depression ratings on global mood scale. If
such a finding can be replicated
of eatingmay differ between
depressed bulimic subjects.
While bingeing and purging may often produce an
immediate but brief reduction of anxiety and hunger,
the relationship between
mood is less clear. In this study many subjects were
not substantially depressed (according to their global
mood ratings)and only some of those that were
bingeing. Thus bingeing does not appear to be used
to â€˜¿?self-medicate' an underlying
tively, depressioncould be a secondary
tationrelated to chronic
(Johnson-Sabineet a!, 1984). An increase in negative
mood states has been reported on days when patients
Sabine et al, 1984). Johnson
suggested that bingeing and purging might be a form
of substanceabuse, with
and/or vomiting used to improve mood briefly. The
and vomitingmay be followed,
the day, by periods of withdrawal
account for the increased dysphoria
for three or more years. We did not
response, but perhaps
during binges, as well as
as to the
but mayalso consume
to the type of food
who ate the fewest
foods or just ate
since we did not give
may be regulated by
1980). Those bulimics
it might suggest that
that influence the rate
depressed and non
of depression during
& Larson (1982) have
extremes of foodintake
on global mood
scales. Thus bulimics may become trapped in vicious
alternating with rebound,
increase in dysphoria.
It is possible that some bulimics, particularly
without depression or any anxiety disorder,
bingeâ€”vomitcycles to counter a psychological
logical predisposition towards obesity. A few studies
have suggested that bulimics
obese (Pyle et a!, 1981; Fairburn
Similarly, it hasbeen suggested
purging might be a short-term
drive to eat and gain weight (Chiodo
1983). Several studieshave suggested
vulnerability to obesity might be inherent in relatives
of those with normal-weight
Cooper,1984; Mitchellet a!,
study, all three (25%) of the bulimics who had low
global depression scores on admission (numbers 5,8,
and10) had a past history
greater than 115% average body weight, suggesting
that they might have a tendency towards obesity.
This study is, to our knowledge,
istic investigationusing behavioural
appeared that shouldbe borne
investigations. Several authors
Johnson-Sabine et a!, 1984) have suggested
many bulimics do not have a major depressive
order, but ratherhave
tension disturbance:such behaviour
accurately measured by the standardised
we used. We have frequently
bulimics; thus they may be poor witnesses to their
own behaviour. Lastly, â€˜¿?a
some bulimics appeared to be intoxicated
out'. Future studies should employ a broader
of objective behaviouralratings.
In summary,we have demonstrated
that patientswith chronic
anxietythan in depression
bingeing and vomitingepisodes.
bingeing and vomiting appeared
each individual, different
wide range of mood responses.
mechanismsby which bingeing and vomiting
mood states may be expected
treatment of bulimic symptoms.
might easily become
solution to counter a
In the present
of weight equal to or
the first natural
in mind in future
(Walsh et a!, 1984;
would not be
in anxiety' may
in this study
to remain stable for
study of the
to the to contribute
Heatherington assisted inthe reviewand preparation ofthis
NormanRuth Timmonsand Tom
MOODALTERATIONS ANDBINGEING INBULIMIA485
BECK, A. T., WART, C. H., MENDELSON,
Psychiatry. 4, 561â€”571.
CHJ0D0,J.& LATIMER, P. R.(l983)Vomitingasalearnedweight-controltechniquein
Psychiatry, 14, 131â€”135.
FAIRBURN, C. G. &. COOPER, P. J. (1984)The
GWIRTSMAN, H. E., Roy-Bua@â‚¬,
HALMI, K. A., FALK, 3. R. & SCHWARTZ,
HAMILTON, M. (1960)A rating scale for depression.
HUDSON, 3. 1., Pops, H. G., JoN*.s, 3. M. & YURGELUN-TODD,
disorder. Psychiatry Research. 9, 345-354.
S. F. & Bauiiowr, P. J. V. (l982)How patients describe bulimia or binge eating. PsychologicalMedicine, 12,625â€”635.
M.. MOCK, J. & ERBAUGH. J. (1961)An inventoryfor measuringdepression. Archivesof General
bulimia.JournalofBehaviorTherapy and Experimental
clinical features ofbulimia
P., YAGER, J. & GutNu,
nervosa. British JournalofPsychiatry.
in bulimia.R. H. (1983) abnormalitiesAmerican Journalof
E. (1981) Binge-eating and vomiting:A surveyofa collegepopulation.PsychologicalMedicine,II,
ofeatingdisordersD. (1983a)to major affective
,,& (1983b)Family historystudyofanorexianervosa andbulimia. BritishJournalofPsychiazry,142,
JOHNSON, C. & LARSON, R.(1982)
JOHNSON, C., LEwIS, C., Low,
Youth andAdolescence. 13, 15â€”26.
JOHNSON-SABINE, E. C., Wooe, K. H. &WAKEUNG, A.(l984)Moodchangesin
LEIB0wITZ, S. F. (1980)Neurochemical
excretion. In BehavioralStudiesofthe
METROPOLITAN Lwa INSURANcE COMPANY (1959) New weight standards for men and women. Statistical Bulletin ofihe Metropolitan L(fe
insurance Company. 40, 1â€”15.
MITCHELL, 3. E., Pvut,R. L. & EcxvaT, E. D. (1981)Frequency and
M0N'rocmnnty,S. A. & ASBERG, M. (1979) A new depression scale designed to be sensitive to change. British Journal ofPsychiatry, 134,
PrraasaN, R., KAya,W. H. &Gwiwrsat@N,H.E.(1986)Estimationofcaloricintakeforpatientshospitalized
the American DieteticAssociation, 19,490-492.
POPE, H. 0.,HUDSON, 3. 1., JONAs, 3. M.&YE1LGEI.UN-T0DD, D. (1983)
study. AmericanJournal of Psychiatry, 140,554-558.
PYLE, R. L, MITCHELL, 3. E. & EcIUORT, E. D. (1981)Bulimia:A report
,HALv0R50N, P. A.,NEUMAN, P. A.&00FF,G. M. (1983)Theincidenceof bulimiain freshmancollege
students. International Journal of Eating Disorders, 2,75-85.
STANOLER, R. S. & Pitn-@z, A. M. (1980)DSM-III: Psychiatricdiagnosis
S'raosra, M. (1984) Stressful events associated with bulimia in anorexia nervosa: Empirical findings and theoretical speculations. Inter.
nationalJournal of Eating Disorders,3,3â€”16.
W*LsII,B.T., STEwART,J. W., Room, S.P.,G@Dis, M. &GLASSMAN,A. H. (1984)lreatmentofbulimia
Psychiatry. 41, 1105-1109.
WURrMAN,R. 3. (1983) Behaviouraleffectsof nutrients. Lancet. 1, 1145-1
S. ci aI(1984)
high schoolcorrelates in a femalepopulation. Journalof
bulimia nervosa. BritishJournoiofPsychiatry,
of feeding and drinking
Hypothalamus(eds P. 3.Morgane &3.Panksepp), Vol. 3.New York: Mar@l Dekker.
of the hypothalamus: Controlbehaviour and
duration ofbinge-eatingepisodesin patientswith bulimia. American
Bulimia treatedwithimipramine: A placebocontrolleddouble-blind
of 34 cases.Journal of ClinicalPsychiatry,42, 60-64.
in a universitypopulation.American Journoi of Psychiatry, 137,
with phenelzine. ArchivesofGeneral
*Walter H. Kaye, MD, Associate Professor of Psychiatry, Western Psychiatric Institute and Clinic, 3811
O'Hara St, Pittsburgh, Pa. 15213, USA;formerlyStaff Psychiatrist, Laboratory of Psychology and Psycho
pathology, National Institute of Mental Health, 10/4Câ€”110,Bethesda, Maryland 20205, USA
Harry E. Gwirtsman,
Science, National Institute of Mental Health, 10/3Sâ€”231,Bethesda, Maryland 20205
MD, Medical Staff Fellow, Section on Biomedical Psychiatry, Laboratoryof Clinical
David T. George, MD, Medical Staff Fellow, Section on Biomedical Psychiatry,
Laboratory of Clinical Science,
Sandra R. Weiss, BA,Research Assistant, Section on Biomedical Psychiatry, Laboratory of Clinical Science,
David C. Jimerson, MD,Chief, Section on Biomedical Psychiatry, Laboratory of Clinical Science, NIMH