WARNING

This document has been supplied under a CLA Licence. It is protected

by copyright and it may not (even for internal purposes) be further copied,
stored or on-copied electronically without permission, save as
may be permitted by law. The recipient may print out a single copy of
any document received electronically.

morbidity Findings From the MTA Study:

Atomoxetine
in the Treatment
omparing Comorbid
Subgroups

of Binge-Eating Disorder:
A Randomized Placebo-Controlled Trial
N, M.D., STEPHEN P. HINSHAW, P .D., HELENA C. KRAEMER, P .D.,
H

ORA, B.S., JEFFREY H. NEWCORN, M.D., HOWARD B. ABIKOFF, PH.D.,

H, M.D., L. EUGENE ARNOLD, M.D., DENNIS
L. P. CANTWELL, M.D.,
ERS, PH.D., GLEN R. ELLIOTT, M.D., LAURENCE L. GREENHILL, M.D.,
BETSY HOZA, PH.D., WILLIAM E. PELHAM, PH.D., JOANNE B. SEVERE, M.S.,
ANSON, PH.D., KAREN C. WELLS, PH.D., TIMOTHY WIGAL, PH.D.,
AND BENEDETTO VITIELLO, M.D.

Susan McElroy, M.D.; Anna Guerdjikova, Ph.D.; Renu Kotwal, M.D.;

Jeffrey A. Welge, Ph.D.; Erik B. Nelson, M.D.; Kathleen A. Lake, M.S.w.;
Paul E. Keck, Jr., M.D.; and James I. Hudson, M.D., Sc.D.
RccchrJ May 24. 2()()(); ncu/lm/ July. / 9. 2006. From I!J~

ABSTRACT

rch has been inconclusive whether attention-deficit/hyperactivity disorder (ADHD), when comor(oppositional defiant disorder [ODD] or conduct disorder [CD]), with the internalizing disorders
or with both, should constitute separate clinical entities. Determination of the clinical significance

alizing disorder or ADHD + ODD/CD syndromes could yield better diagnostic decision-making,

Objectil'e:
Binge.eating
disorder
(B ED) is
atment outcomes. Method: Drawing
upon cross-sectional
and longitudinal
information
from 579

associ;lIcd with obesity. Alomoxclinc is a high ly

selective norepine phrine reuptake in hibilOr ascit/Hyperactivity Disorder (MTA), investigators applied validational criteria to compare ADHD subsociated wi th weigh t loss. T he purpose of this
bid internalizing disorders and ODD/CD. Results: Substantial evidence of main effects of internalstudy was to eval uate :l1omoxclinc in the treatmorbid disorders was found. Moderate
men! ofevidence
BED. of interactions of parent-reported anxiety and
d on response to treatment, indicating
that
ADHD andsinglc-centcr,
anxiety disorders
(but no
Melhod;children
In thiswithlO-week.
ran-

) with ADHD participating in the NIMH Collaborative Multisite Multimodal Treatment Study of

pond equally well to the MTA behavioral

anddouble-blind.
medication treatments.
Children with ADHD-only
domizcd,
pl:1Ccbo-colltroJlcd,
flex - or

PSyC/lOp/Ul'm<lc%gy R,'uw./I PIlJ/I/'(IIII, Drl'<lritllCI1/ of Ps)'chiml).

UI/her.ily o/Cillci'l/Iali Col/rxc of Ml'dit'i",~ (Drs. MeE!roy. Gllrrdjikow..

Korl<'nl. Wrlgl'. Nelsml. alld K,'ck mid M s. Lake!. <Ill/I Mental HI'"llh
Sl'rl'ire Line and G"""rol Clinical Rtsrm..-h Crmer. Cin, i"n"l; \'rlcr,,"s
Affairs M"dical Cen,..r (Dr. Krck). C;IIdmKJti. Ohio; ulld Illr Drparllllttll
o/Psychiull). Hun'ord Ml'dieal Se/w'll ulld M euml Haspiral, Brlmom.
Muss. (Dr. HudsQlI).
This s",d), ...as s,.ppart~! ill pcm byo gromfmm EIi UII),.
ImlimlOpoIis./m/.
'mJMdr,ol/i'IQ'/C;"! dis.:/u5/11't' a lii I~ found allllt ..,Ill 0{ Ihis ortie!l'.
Carrrspotwillg alllh mill '''llrilll$: SIISQn L . MeEtroy. M.D. ,
PsyeMl'tlflrlll(U'U/og)' R"sl'md, Proi/ram . D<'JNlrllIII'III of P5},e/,iolry.
Ulli\~rlil)' ofCi",:i",KJli CollcSI' of Mfdiei,II'J, 1'.0. 80..-670559. 2JI
Albi'l'/ Sol>ill !Va)'. CilldwlClli. Off 45267.fJ559
{f'lJIai/; sIISllIl.mcriro)'@lIc.<'lI"J.

ib lebest
dose
(40--120
mg/d3Y)
Irial,(with
o utpatients
w ith
thout anxiety disorders) responded
to MTA
medication
treatments
or without behavioral

BED received
moxctinc
or pla-(medwith multiple comorbid disordersDSM-IV-TR
(anxiety and ODD/CD)
respondedato
optimally
to combined

The
outcome
measure
was bingements. Conclusions: Findingscebo.
indicate
thatprim:lry
three clinical
profiles, ADHD
co-occurring
with inter-

eating episode frequency. The primary analysis o f

effic acy was a longitudinal analysis of the intentDD/CD but no anxiety (ADHD + ODD/CD), and ADHD with both anxiety and ODD/CD (ADHD +
IO-[feal sa mple, w ith trcmmc nt -bY-limc imcracufficiently distinct to warrant classification as ADHD subtypes different from pure ADHD with
lion as the effect measure. P:l1icnls wcre enrolled
linical, etiological, and genetics research should explore the merits of these three ADHD classififrom Scplember 2004 through OClobcr 2005.
Child Adolesc. Psychiatry, 2001, 40(2):147158.
Key Words:
attention-deficit/hyperactivity
Results: Conlparcd
with
placebo (N:: 20), disty, comorbidity, treatment, outcomes,
classification,
diagnosis.
alomoxclinc
(N =
20) wa ~ asso<.:imcd with a sig-

y parent-reported anxiety disorders) absent any concurrent disruptive disorder (ADHD + ANX),

nificantly greater r~le of reduction in binge-eating

episode
frequency.
as comorbid
well as in disruptive
bi nge day behavior
freThe
presence of
disorders
quency,
weigh!. body
mass disorder
index. and
scoresoronconduct dis(oppositional
defiant
[ODD]
partment of Psychiatry, Columbia Ihe Clinical Global Impressions.Severity of Illorder [CD]) within children with attention-deficit/
eons; Dr. Hinshaw is Professor of
ness ~cale.
Yale-Brown
Obsessive
Compulsive
hyperactivity
disorder
(ADHD)
has been well estabersity of California at Berkeley; Dr.
Scale
Moditicd
for
Bi
nge
Eming
obsession
sub-et al., 1990;
d University; Ms. Lenora is a medlished for several decades (e.g., see Bird
School of Medicine. Other authors scale, and Three Factor Eating Questionnai re
Hinshaw, 1987). Only in the past decade has it become
tive Group acknowledgment that hunge r sub!icale. The me:m (SO) alomoxeline
apparent
that internalizing
al cndpoim
evalullliondisorders
was 106 (both
(2 1) anxiety and
daily dose
or the Advancement of Childrensmg/day.
depressive
disorders)
also
commonly
co-occur with
Four palienls (N :: 3 receiving atomoxe
I, 1051 Riverside Drive, Unit 78,
ADHD.
Thus, both
clinical
and epidemiological
studies
N
= I receiving
placcbo)
disconti
nued
be
line.
mbia.edu.
consistently
that asfor
many
of adverse
events.shown
TIle reasons
ato as one third of
he American Academy of Childcause have
moxeline
discontinuation
were
increased
children
with ADHD
have
co-occurring anxiety disdepressive symptoms (N '" I), const ipation
(N = 1). and nervou~ n ess (N =- I).
YC H I AT RY, 4 0 : 2 , F E B RU A RY 2 0Ca
0 1 nclusioll: Atomoxetine was efficacious and
147
fai rly well tolcnued in the short-tcnn trentmcnt of
BED.
Clillicall'rials Regislralioll:

ClinicalTrials.gov identifier NCTOOJ27834.

(J Clill Psychialry 2007;68:)90-)98)

J Clin P!ychiatry 68:3, March 2007

inge-eating disorder (BED) is characterized by recurrent bingeeating cpisodes withou t inappropriate compensatory weight loss behav iors.1.2 Inc reasing evide nce indicates Ih .. t BED is an important public health
problem . Its liretime prevalence in the United States general population is estimated to be 3 %.'~ and it is associated with psychopillhology, e~pec ially anxiety and depressive disorders 4 -6; obesity and other types of med ical
co morbidjtl7-~ ; impai red quality of life 'o: and disability.4
Treatment objectives for BED include reduction of
binge eating and associated psychopathology and man
'lgement of comorbid obesi ty Md related medical cond itions,2.l 1-16 The National Institute for Clinical Excellence
gu ide lines recommend the use of cognitive-behavioral
and inte rpersonal therapies and selective serotonin re
uplllke inhibilOr (SS RI) anlidepressants, 11 All or these
treatments, ho wever, have limitations. Both cognitivebehavioral therapy and interpersonallherapy often res ult
in reduced binge eating and associated psychopathology
but usually lire not associated with clinically significnnt
weight losS.11.I4-16 By contrnst, although several SSRls
were associated with stat istically significant reductions in
binge eating and body weight in short-teon, placebocontrolled monotherapy Irials,'1.IJ11-21 nuoxetine was
ineffecti ve for binge eating and weight loss in 2 placebocontrolled stud ies thal compared it with cognilivebehavioral therapy. I. IS OrliSI:1t with eognitive- behavioral
Iherapy,2! s ibutramine, 2,.1~ topiramate,!L!1; and zonisamide 21 have been shown in controlled trials to significantl y reduce both binge eating lIncl body weight; sibutra-

390

HcElroy et al.

mine has also been shown to significantly red uce assoc iHIed depressive symptoms. 2l However, all 4 medications
are associated with problematic side e ffects and relati vely
hig h discontinuation ra tes.H-.\I'I Novel treatments that reduce binge ealing. as well as associated p.~yc hopal hology
and body weight, 1l11d that are well tolerated are therefore
needed for BED.
Several li nes of ev ide nce suggested to us that
ato rnoxetine-a highly speci tic norepine phri ne reuptake
inhibitor "pproved by the U.S. Food and Drug Administration (FDA) for the treatment of attention-deficit/
hyperactivity disorder (ADHD}-might be a useful treatment for BED.)I-J) First. the central norepinephri ne
.~yst em is involved in regulating eating behavi or;14
and atomoxetine has been shown to red uce foo d consumption in several animal mode ls of feeding:ls Second,
atomoxeti ne was associated with anorexia and we ight loss
in the registration clinical trials for ADHD in both children and ad ults. J,-JJ It has also been shown 10 significantly reduce body weight as compared with placebo in a
prel iminary 12-week study of 30 women with obesity
who did not have "major psychiatric disorder or alcohol/
substance abuse within the past year." wrll)~) Third. a
range of antidepressa nts has been reported to reduce
binge eating in BED and the re lated cond ition bulimia
nervosa, including tricyclics wilh primarily noradrenergic
reupta ke inhibiting properties.12.1J..17 Prel imin ary observations suggest atol1loxetine may hnve antidepressant
pro pe rt i e ~-bot h in children w ith ADHD,lI ..lI! and in adu lts
with major depressive disorder_ .w...a Fourth, ntomoxetine is
generally well to lerated. In the A DH D regis tra tion d inical trials, dropout rates for adverse events in both children
and adults were low 5%), and the most commonly reported side effec ts, wh ich were gaslro imestinal and neurologic. were generally described a,~ mild and trnnsicnl. JJ
These observations led us to conduct a singlecenter, random ized. paralle l-group. placebo-control led,
n ex ible-dose study to assess the cffic:lcy and safety of
atomoxetine during a ID-week course of treatment in 40
outpatients with BED. We also compared the effects of
treatment with ntomoxetine and placebo on various metabolic measures. includ ing weight, in this patient group.
METHOD
Patients
Study participants were out patients at the University of
Cincinnati Medical Center who were recr uited by radio
nnd newspaper adve rtisements requesting vol unteers for a
study of a med icntion for binge eating. Pntients were enrol1ed into the study if they met the fo l1ow ing inclusion
criteria: (1) were male or fema le from 18 to 65 years of
age; (2) met DSM-IV-TR criteria for BEDI; (3) weighed
~ 85% of the midpoint of ideal body weight for height
(nccord ing to the Metropolitan HeighlfWeight (abies); and

391

(4) had i!: 3 binge-eating episodes and i!: 2 binge days in the
week before receiving study med ication (confirmed with
prospective d iaries while the patient received single-blind
placebo ru n in; sec outcome measures).
Patients were excluded from part icipation in the study
if they met any of the following criteri a: (1) had concurrent anorexia nervosa or bul imia nervosa (by DSM-IV-TR
crite ria); (2) had a substance use diso rde r (by DS M- IV-TR
criteria) within 6 months of study entry; (3) had a lifetime
history of a psychotic d isorder. a bipolar disorder. or dementia or other cognitive disorder (by DS M-IV-TR criteria); (4) had a personality disorder thal cou ld interfere with
diag nostic assessment. treatment, or compliance; (5) displayed clinically significant suicidaJity or homicidality;
(6) had rece ived cognitive-behavioral or interpersonnl
psyc hotherapy or behav ioral weight management fo r BED
within 3 months of study ent ry; (7) had a cl inically unstable medical ill ness; (8) hnd a history of seizures. including childhood febrile seizures; (9) required treatment with
any drug thnt might adversely interact with or obscure the
action of the study medication; (10) had c1 inicnlly signifi cant laboratory or e lectrocardiogram abnormal ities; (1 1)
hnd received monoamine ox idase inhibitors, tricycl ic antidepressants, lithium, antipsychotics, or nuoxetine within 4
weeks prior 10 ra ndo mization; ( 12) had received other psychoaclive med icution (other than hypnotics. e.g., zolpidem
or zaleplon, as nceded for insomnitl) within 2 weeks
of study mediclltion initintion; (13) had received investiglltionai medications or depot antipsychotics within 3
mont hs prior to randomization; or ( 14) had previously
been treated with atomoxeline. Women were excluded if
they were pregna nt, lactating. or. if fenile, not practicing a
form of medically accepted contraception.
The insti tutio nal review board <lI the Un iversity ofCincinnati Medicnl Center approved the study protocol, ,lI1d
the study was conducted in complinnce with the Declaration of He lsinki. All plltients signed approved written infonned consent forms afte r the study procedures had been
fu lly explained and before an y study procedures were
performed. Patients were enrolled from September 2004
th ro ug h October 2005.
Study Design
111is was a 10-week, outpatient. randomized. doubleblind . parallel-group. n ex ible-dose study cond ucted at the
Universi ty of C incinnati Medical Cenler. The trial consisted of 3 phases: a I- to 2-week screening period which
included a I-week single-blind plncebo run in. duri ng
wh ich patients hnd to display ~ 3 binge episodes and i!: 2
binge days in order to be incl uded; a lO-week doubleblind treatment period: and a I-week treatment discontinuation period. Patients were eva luated at least twice
during the screening period; after I. 2, 3, 4. 6, 8, und 10
weeks d uring the treatment period; and I week after study
med ication discontinuation.

'"

J CHn Psychiatry 68:3, !l1arch 2007

Atomoxetine for Binge-Eating Disorder

The screening evaluation included an interview for demographic and clinical information and medical, psychiatric. and fa mily histories; the Structured Clinical Interview for OSM-IV-TR (SClOt to establish BEO and
comorbid Axis I diagnoses; a physical examination; vital
signs; height and weight; rm electrocardiogram: fast ing
routine blood chem ical and hematological tests; and urinalysis. At this evaluation and each of the following visits. patients were given take-home d iaries in which to
record any binges and. once study medication was initiated, the number of capsu les taken on a daily basis (see
outcome measures). At the last visit of the screening period (the baseline :u;sessment), patients were evaluated to
see if they continued to meet entry criteria. Patients continuing to meet these criteria were enrol led in the treatment period and nmdomly assigned in a I: I ratio to
therapy with atomoxetine or placebo. At each visit following the b3seline visit. patients were assessed for number of binges experienced since the last visit, other outcome measures. medication dose, medication compliance
ascertained by capsule count. adverse events, use of
nonstudy medications. vital signs. und weight.
All study medicat ion Wll S in identical 40-mg
capsules supplied in numbered containers and dispensed
10 patiems accord ing 10 a predeterm ined r:mdomization
schedule. Study medication was begun at 40 mg/day for
the first 7 days. At the beginning of the second week of
treatment, study medication WllS increased, as toleratcd,
to 80 mg/day. AI the beginning of the third treatment
week, study medication was increased. as tolerated. to
120 mg/day. Study medication cou ld be reduced to a
minimum of 40 mg daily because of bothersome side effects at any time during the IQ-weck treatment period. Pntients took aJJ their daily dose of study medication in the
morning. However, if patients preferred. they could take
40 mg or 80 mg in the afternoon.
Outcome ~feasures
The primary outcome measure was Ihe weekly frequency of binge-eating episodes (binge frequency), defined as the mClln number of binges per week in the interva l between vis its (total number of binges in the interval.
divided by number of days in the interval, and multiplied
by 7). Binges were defined using OSM-IV-TR criteria,1
lInd assessed via cl inical interv iew and rev iew of patient
take-home diaries, in which patients recorded binges. d uration of bi nges, and food consumed during binges (so
that binges cou ld be confimled by the research assistant
and physician investigator working with that particular
patient).
Secondary outcome measures were weekly frequency
of binge days (days when the pat ient had I or more
binges): weight (kg); body mass index (BMI, calcu lated
by dividing body weight in kg by height in m1); Clinical
Global Impressions-Severity of Illness scale (CO l-

1 Clin Psychiatry 68:3, March 2007

Severity) and -Improvement scale (CG I - Improvem ent)~l

scores; Yale-Brown Obsessive CompUlsive Scale modified for binge eating (YBOCS-BE)') scores; Three Factor
Eating Questionnaire (TFEQ)4oI scores; and Hamilton Rating Scale for Oepression (HAM_ D)H total scores.
Weight was obtained with the patient in light clothing
without shoes on the same scale zeroed aI each measurement. The YBOCS-BE is a modified version of the YaleBrown Obsessive Compulsive Scaleu used in previous
phannacotherapy studies of BEO,.ll.2s..n (and is avai lable
from the authors on request) that measures obsessiveness
of binge-eating thoughts and compulsiveness of bingeeating behaviors. The TFEQ (also called the Eating Inventory) is 1I self-report questionnaire that measures 3 dimensions of eating behavior thought to be dysregulated in
Obesity: cogni tive restraint in eating (cognitive restraint);
disinhibition of control over eating (disinhibition): :md
perceived hunger (hunger).~4 As done in most previous
BED phannacotherapy studies, response categories were
tabulated based on percentage decrease in frequency of
binges from baseline (the week before treatment initiation) to endpoint (the final week of treatment). These
categories were defined as follows: remission = cessation
of binges; marked = 75%-99% decrease; moderate =
50%-74% decrease; and none ", less than 50% decrease.
In addition, time to recovery was assessed, defined as the
first 4 consecutive weeks during which the patient had no
binge-eating ep isodes.
The following safety measures were assessed: adverse
events. c linical laboTlltory data, physiclll examination
fi ndings. and vi tal signs. Reportable adverse events were
new symptoms or illnesses that occurred during the treatment phase and those that increased in severity compared
with baseline.
Statistical N ethods
The baseline characterist ics of each group were COIllpared by using Fisher exact test for categorical variables
and independent-samples I tests for conti nuous variubles.
The primary efficacy analysis was a longitudinal
lInalysis comparing the rate of change of binge frequency
during the treatment period between groups. The same
analysis was applied to binge day frequency. weight,
BMI, and scores on the CGI-Severity, YBOCS-BE.
TFEQ, and HAM-D scales. The difference in rate of
change was e.~t ima led by random regression methods, as
described in Fitzmaurice et al.ol(, and Gibbons et al.,~7
and as used in several phamlucolherapy stud ies of
BED.IX-ll. l)-" We used a model for the mean of the outcome variable that included tenns for treatment, time, and
treatment-by-time interaction. Time was modeled as a
continuous variable. expressed as the square root of days
since randomization (baseline). For the analyses of binge
fre quency und binge day frequency. we used the logarithmic transformations log ([binges/wk] + I) and log ([binge

392

r-lcElroy et al.

Table 1. Baseline Characteristics of Patients With

Binge-Eating Disorder Randomly Assigned 10 10 Weeks of
Double-Blind Treatment With Atomoxetine or Placebo
Alomoxclinc
(N _ 20 )
Ch~rac le ri slic

Agc. y
Bin{;e frequency (per we ek)
BinBe day frC'lUCncy (pcr wed:)
Ass..s<ment SCOKS
Cli nica l Global I mpres~ ions
~vcri l>' of II1ncu scale
Hamih<m Ral ing Scale for

SD

Mean
43.1
4.2
3.8

10.2
lA
1.1

4.2

OA

Placebo
(N ., W )
Mean
39.2
4.9

3.9

SD
7.7
2. l
l.l

...

0.'

3.3

3.'

2.0

,..

17.4

4.8

17.9

3.1

9.0
8..
106.9
37.3
i6
17

3.1
2. 1
20.2
' .7
80.0
85.0

8.6
9.3
116.6
41.4
17
17

2.0
1.6
30. 1

5.0
45.0

to

~pres.<io"

Ya le-Brown Obsessive Compulsivc

Sc~ le (mollified for binge e~ linB )
Ob'cssions
Compulsions
Wei ghl. kg
Body mass iTMkx. k1l1ml
Female. N
Ca ucasian. N
Ikprusive disorder ui aJ:nosi s. N'
Current
Lifelime

'"

115.0
8S .0
2S .0
50.0

' Includes 16 patienlS Wi lh mnjor dcprc<. ive disorder. 2 with

dyslhymia. and 1 with depressive disorder not otherw ise spec ified.

days/wk + 11). respectively. to normalize the dala and stabil ize lhe vrlriance. To simul lrlneously account for individual differences in initial level of the outcome. rate of
change over time. rind serial autocorrelation (Le., the tendency for correlation among observations 10 decrease
as a function of the amount of time between them). we
used the SAS procedure MIXED version 9. 1 (SAS Institute. Inc .. Cary. N.e). with random intercept and slope
tenns. and a first-order antedependence struClUre for the
residual correlation matrix. 1lle longitudinal analyses
were imenHo-treal, using all available ob.~e rvat ion s from
all lime points from all palients who completed a base line
eVllluation.
Several secondary analyses were perronned. Using lhe
last observation carried forward (LOCF). baseline to endpoint change scores were computed for each measure (on
the logarithmic sCll le for the bingeing measures), and
independem-samples t tests were used 10 compare these
changes betwcen Ihe treatment groups. The CochraneAnnitage exact trend test for 2-by-k ordered lables in
SAS (PROC FREQ) version 9.1 was used to analyze categorical response to treatmen t (as de fin ed in the prev ious
paragraph) for the intem-to-treat and completer groups.
For laboratory measures, including weight. the
mean difference belween endpointllnd base line measures
was computed for each treatment group and then compared using the t test. The correlation between percentage
change in binge frequency and change in we ight was calculated using rank-transfonned data (Speannan rank correlation). Time to recovery (defined as the first 4 consecuti ve binge-free weeks after baseline) WIlS analyzed with a

393

Cox proportional-hazards model fo r the intent-to-treat

popu lation.
All statistical tests and confidence intervals were
2-sided, U = .05.

RESULTS
Overall. 76 patients were screened, 36 were not enrolled
because they chose not to participate (N = 6) or did not
meet entry c riteria (N = 30), and 40 patients who met entry
criteria were randomly assigned to atomoxeline (N = 20)
or placebo (N "" 20). Thirty-three patients (82.5%) were
women. 34 (85%) were white, 5 (12.5%) were African
American, and I (2.5%) was Asilln. Depressive disorders
were the most common co-occurring psychiatric disorders,
occurring in 19 patients (47.5%) as lifetime diagnoses. and
currently in 6 p'ltients ( 15%). There were no significant
d ifferen ces between the treatmenl groups in demographic
or clinical variables at baseline (Table I).
Thirty-nine patients (20 receiving atomoxetine and 19
receiving plucebo) had at leasl I postrandomizalion effi cacy mellsure. Six pmients (30%) in the atomoxetine group
llnd 9 patients (45 %) in the placebo group d id not complete
al l 10 weeks of treatment (Fisher exact p = .5 1). Four patients withdrew from the study because of adverse events
(atomoxetine: N 3; placebo: N I); I withdrew becau~e
of lack of efficacy (placebo); and IQ withdrew because of
nonadherence to the protocol (atomoxctine; N = 3; placebo: N '" 7). The remaining 25 patiems (62.5 %) compleled the 10 weeks of Irelltment (N:: 14 receiving atomoxetine and N = II receiving placebo).
The mean frequency of binges decreased over the study
period in both treatment groups. but more so in the atomoxetine group (Figure I). Mean body weight decreased
over the sludy period in the atol1loxetine group but not in
the placebo group (Figure 2).
The primary efficacy anlllysis using mndom regression
showed Ihat pmiems receiving atomoxeline had a signi fi cantly greater rate o f reduction in binge episodes per week
than patients receiving placebo (Table 2). Atomoxetine
was also associated wilh a sign ificantly greater rate of improvement Ihan placebo for the following variables: binge
days per week, body weight. BMI , llnd scores on the CG ISeveri ty, YBOCS-BE lolal and obsession subscale. and
TFEQ hunger subscales (Table 2). However, Ihere was no
difference in the rate o f change in red uction in YBOCS-BE
compulsion subscal e scores or T FEQ scores for cognitive
restraint or disinh ibilion between the trealment groups.
There was ll lso no difference in Ihe rate of chllnge in
HAM-D scores between the treatmen t groups (Table 2).
In the secondary llnlllysis of baseline-to--endpoint
change scores using LOCF. atomoxetine was associated
with significant decreases in binges per week. binge days
per week. weight. BMI. and scores on the CGI-Severity
and the YBOCS -BE total and obsession scales compared

,Ur NIGhll J] ''I'SI( J Clin Psyc~iatry 68:3, March 2007

.
.

Atomoxetine for Binge-Eating Disorder

Figure I. !'lean Binge Frequency Over 10 Weeks of Treatment

in Patients With Binge-Eating Disorder Randomly Assigned
to Atomoxetine or Placebo
5

Figure 2. !'lean Weight Change in Patients With

Binge-Eating Disorder Randomly Assigned to 10 Weeks of
Double-Blind Treatme nt With AlomOKetine or Placebo
LO

. A~{N.2O)

O~(N . 2O)

,;

rn

0.5
0.0

~ -0.5
E -1 .0

f -1 .5
-2.0
,
c

oL~-=~:::~~--i~
02

46
Week

10

-2.5

Emlpoint

A\OII"IOQIin&{N - 2O)

--3.0 0 Placebo IN . 20)

6
Week

10

Emlpolnt

Table 2. Hean 1>lodel-Based Differences Between Atomoxetine and Placebo Groups in Change From Baseline to Week 10 for
Patients With Binge-Eating Disorder (N _ 40) Randomly Assigned to 10 Weeks of Double-Blind Treatment With Atomoxetine or
Placebo
Endpoint Analys isb
t(df .. S5)
9S % C I
(-0.2910 -0.0 1)
2.20
(-0.30 10 -0.03)
2.37
(-1.90 10 -0.50)
3.48

Lonllitudin~ 1 AnRI~s i s'

Outcome Mc~sun:
BinJ;es/week<
Binge d~ys/wc:ek~
C linical Global lmpn:ss ions-$everi lY
of IIInc~~ Sca te
Hamilton Depression Raling Scale
Yate-Brown Ob.'lCssi"c Compulsive Scale
(modified for bingc eating)
Obsc'sioi15
Comput sion s
Thn:e Factor E:iting Que,tionnaire'
Cognilive restrni11l
Disi nh ibi tion
Hu nger
WeiJ;ht. kg
Body mass index. kJ;1ml

9S""C I
(-0.61 10-0.09)
(-0.6310-0. 18)
<-2.0110-0.22)

Xl(df: 1)

E'

...{lA I
-0.45
- 1.12

S.72
8.75
6.03

.018

.003
.015

Eslimale
-0.16
-0.17
- 1.20

0.58
-4.77

H.33 10 2.49)
(-9.25to...{l.28)

0.36
4.40

.551
.0)7

-O.IS
-5.30

(- 2.1l 10 1.83)
(-9.01 to-1.59)

0. 15
2.S9

.000

_3.04
- 1.82
- 3.54
2.OS
- 1.96
- ) .56
-3.09

(-5.4110 - 0.66)
(-4.2610 0.63)
HI.32 to 1.24)
(-i.3SI 0 !i.54 )
(-4 .9 1 10 0.\)\))
(-7 . 15100.02)
(-5.4610...{l.72)
(-I.S610...{l.20)

6.36
2.15
2. 15
1.42
1.73
3.SS
6.61
5.93

.Ot2
.143
. 142
.2)4
.1 89
.049
.010
.016

- 3.50
- 1.80
- 3.80
2.01
-1.94
- 3.87
- 2.69
...{l.S9

(-5.7310 - 1,27)
(-3.71100.11)
(-9.44 10 1.84)
(-2.471 0 6.49)
(-5 .47 to 1.60)
(- S.56 10 0.S2 )
(-4.811100.4\)
(- 1.66 10-0.12)

3.18

.003

1.91
1.44
0.93

.. "

E>tim~te

- 1.0~

E'
.0)4
.02)

.Oll
.879

.067

.36<

LI D

.287

1.70
2.48
2.34

. 04
.018
.025

'Esti mate is for .uean (week ID minu s b~scline) for ntomoxct inc minll s lIle~n (week ID minu s boseline) for pl~cebo. Tc~t Slali slie is the lreatmentby
time in tcr~~t ion term, which represents the difference in rale of ch~ngc belwec n the atomoxctine Ilnu p1a~cbo llrouP ~ . "'ith time Illodctcd as s4uarc
mOl of days ~incc rJndolllizmion. The c~timate and ils Cl wen: obl"incd by mulliplying the In:almcnt-by-lime inleraction Rnd ils Cl by 112 (112
days in t6 week s) and squaring.
"Eslim~tc is for meoll (weck ID minus base line) for 310moxctiroc minus mean (wcek ID minu~ In.., dille) for placebo. The e.'l im~tc is the test statistic.
which is the me~n difference in (hanlle K Orc (ernlpoim minu!; base linc) between lhe alomoxetinc ~ntI placebo groups.
<Log t"msformation (log lbinges/weekl + I) Wit., useu for analysis ; " alues in table are e ~pn: sscd in the original Kale.
dLog Irnndonn 3tiOl' ( lng lbinge d~ys/v.eck l + I ) wa.' used for am.l y$i ~; "atuc~ ill table arc e~pre ~!ICd in Ihe original sca te.
<Measun:d al ,,eck. O. 4. 8. and 10 on ly.
Significanl al p" .05.

with placebo (Table 2). A m:lrgin;tlly nonsignificnnt

change was oblained for the YBDeS-BE compulsion subscale scores. T here we re no significanl differences between groups in Ihe changes in scores on the TFEQ or the
HAM+D scales.
Regarding global responses, Ihe mean final CG l
ImprovemcnI lit endpoint WllS rated much or very much
improved in 16 momoxetine-tre:l.Ied patients (80%) as
compared with 8 placebo-lrealed pmienls (42%). I.n Ihe
calegorical response analyses, Ihere we re significantly
higher levels of response for aiomoxeline-trealed patients
in both the intcnt-to-lTem and completer groups (Table 3).

J Clin Ps},chiatry 68:3, March 2007

In the intent-to-Ireat population. remiSSIOn of binge

episodes was auained by 70% of atomoxetine-treated
patients at end point compared with 32% of placebotreated palients (Fisher exacl p = .026). AlOmoxetine was
not associated with shortened lime to recovery of binge
eating in Ihe iment-Io-Ireat group (hazard ratio for recovery = 0.62, Xl '" 0.44. p '" .508).
Pntients receiv ing atomoxetine experienced a mean
(SO ) weighl loss of 2,7 (3.7) kg from baseli ne to endpoin!, whereas those receiving placebo experienced a
mean (SO) weight loss of 0.0 (3.2) kg. Among p;tlients
who completed the IQ weeks of treMment, the corre-

394

McElroy et al.

Table 3. Response to Treatme nt Among Patients With

Binge-Eating Disorder R~ndomly Assigned 10 10 Weeks
of Double-Blind Treatment With Atomoxetine or Placebo

Table 4. Adverse Events Reported by:!; 2 Pat ients With

Binge.Eati ng Disorder Receiving Treatment With
Atomoxetine or Placebo

Res~clIIse'

PI~cebo

A tomo~e:ti lle

IN. 19).
N(% )
5 (26)
8 (42)

None
Moderate
Marked
Remi ... ion

Who Completed
10 Week . of Tre atment<

IN. 20).
N{%)

Placebo
(N = II).
N( %)

Alornolle:ti nc
(N = 14).
N{ %)

4 (20)
0(0)

2 (IR)
5 (45)

1(7)
0(0)

0(0)

2 (l0)

6(32)

14 (70)

0(0)
4 (36)

11 (79)

Adverse Even!

Ncrvou snes~

In som nia
Il cmJache
COIlSIipmion
Swemins
Dizziness
Hypertension
Dyspepsia
Somnol cnce
Diarrh ca
Rh initis
Hot nash
Dep ..,,,sion
Abo.Iominal[lain
Uri nary h es itan ~y
Eru"a!ion

2 (1 4 )

from basel ine:: remilisiotl _ c<:ssal io n of binGes: mari;ed '" 7.'1%-99%

n:d uction: moderate = 50%-74 % reduction: none _ less Ihan a 50%
n:ducti on.
~PalicnlS had al lc<t.>t 1 postratl(.[omizal i{)tl cfficm;y mea. urc;
sig nifi,ant diffcre: n,c bct"'cctl s roups (p _ .025. euel
Cochr.mcArmilagc lrend ICS I).
<Si ~nirlC~ nl differencc betwee n ,rou ~ (p _ .0 19 en"
Cochl'lllle -A nn it age tn: nd Ic.I).

1
1
6

35
35
30

4
4

2.

2
2
2
2

2
2
2

"

10
10
10
JO
10
10
10
0
I.
10
more: rrequ ent ly ill the

2
2
' Dry mOlllh (p - .048) oc,um:d
group lhatlthc pl a,cbo &fou p.

sponding we ighl losses were 3.9 (3 .3) kg llnd 0.2 (4.3) kg.
Weight loss since baseline was signiricantl y corre lated
with percentage reduction in binge freq uency at week
10 in the overa ll sample (Spearman p:: -0.49, p = .0 12).
Among patienL~ receiving atomoxetine. lhis correlation
was n01 significant (Spearman p '" -0. 15, P '" .52), but
this result can be att ributed in part 10 the fllCI thal so many
of these patients (1 4 of 20) had no bingeing by week 10.
so Ihat only 7 uniq ue r.lIlked values existed in the sample.
There were no signilic:l1n differen ces belween patients
receiving :UomoxeLine and those g iven plncebo in menn
chll nge from baseline to fin nl visit for the fasting measurements of insu lin (-2.7 and - 1.8 ~u/mL; p = .67), glucose (7.5 and - 1.6 mgldL; p = .30), triglycerides (5 .5 and
- 29.0 mg/dL; p = .2 1). LDL cholesterol (- 0.2 lInd SA
mg/d L; p .61), and toWI cholesterol (-0.7 and - 2.4
mgldL; p = .88).
The mean (SO) daily dose o f atomoxet ine a l endl>oint
evalual ion fo r all 20 pat ients was 106 (21 ) mg. The mean
(SD) daily dose for the 14 patients who completed lhe 10week Iri ,11 wns 109 (19) mg.
Adverse events occurring in at least 2 patients receiving ntomoxeti ne are listed in Tnble 4. Adve rse events
appe;lred to be mo re common over1I w ith atomoxetine
Ihnn with placebo. and lhere were stali stically signilicant
differences belween treatmenl groups in Ihe incidence
of dry moulh. More patients discontinued atomoxeline
(N = 3; 15.0%) for ad verse events than placebo (N == I :
5.0%), but this d ifference in incidence was not statistically significant (Fisher exact p = .60). Adverse events
causing discontinuati on among atomoxet inC!-lreated palients were increased depressive symplOlm (N =: I), constipation (N = I), and nervousness (N '" I ). The adverse
event causing disconti nuation in the place bo-treated patient was increased blood pressure. No patient experienced a seri ous ndverse event during the study.

20

..

Pl acebo (N 20)

", "..,

Dry mo uth'
Nausea

"C3.teSOr~$ defined by the pereentase: dc:<:reasc in binGe frequency

N
4

20

10

3
3
4

"
15

20
10

I
I

S
S

2
2

10
JO
I'

2
I

S
0
10
0
0

~lomoxetifIC

There were no changes in physical ex;unination findings. vital signs, or clinical laboratory values suggestive
of drug-re lated toxicity. There was no evidence of withdrawal symptoms in the 6 patients who d iscontinued ato
moxeti ne prematurely or in Ihe 14 patients who discontinued atomoxctine per protocol.

DISCUSSION
In the primary longitud inal analysis of this mndomized, double-blind lrial in patienls with BED, alOmoxeline was significantly superior 10 placebo in rate of
reduction of binge frequency, binge day frequency, body
weight. BM I, obsessive-compulsive features of bingeeating symptoms, hunge r. and overall severity of illness.
A secondary analysis, change fro m baseline to endpoinl
using LOCF, yieldC!d ~ im i l ar positive findings, wilh atomoxetine being associated with significant decreases in
binge frequency, binge day freq uency, body weighl. BMI,
obsessive-compulsive features of binge-eat ing symptoms,
and overall severity of illness compared with placebo.
AtomoxetinC! was also associated with a significantly
higher level of categoric:! l response in both the endpoint
and eompleter analyses. but not with a shortened time to
recovery of bi nge eating, although the relati vely modest
sample size and length of follow-up may explain this result. The mean weight loss in the intent-to-lrea t group receiving :nomoxeline wns 2.7 kg, a~ compared with 0.0 kg
in the group receiving placebo. There was no significllnt
change in HAM-O scores. but mean HAM-D scores were
low at baseline. Taken together, these fi ndings provide
preliminary evidence for the efficacy of alomoxeti ne
in BED.

395

..

AtoUlolletine (N _ 20)

P~ti e n lS

In1C:n!-to-Treat G rou~~

I '

J Clin~Psych iatrY 68:3, fo.Iarch 2007

Atomoxetine for BingeEating Disorder

The reduction in binge frequency and overall im

provement observed wi th atomoxetine in this s tudy
appea r comparable to res ults reported in stud ies of
cognitive-behavioral therapy.".l4-16 interpersonal therapy. II SSRls. Il.U.I.l l sibutram ine.ll.~4 topiramate, lS.26 zoni.
samide. l7 and orli stat p lus cognitive behavioral therapyl!
in patients wi th BED. The weight loss appears comparable to thal seen for sibutramine,2.,.u topiramate. 2S26
zonisamide?' and orlistat w ith n or wi thout 4M cognitivebehav ioral therapy. However. the high premature d is
cominuation rate also seems comparable 10 thal seen for
the laller treatmems ll.lJ .!5-11,41 and for SSRls. 'H' Appropriately designed controlled compnrison trials are re
quired 10 accurlllely determine ntomoxetine 's efficllcy.
tolerability. lUld safet y relative to other trelltments of
BED, as well as whe re it w ill fit into the grow ing thera
peutic annamcntari um for this condit ion.
The potential mechanism of action of alOmoxetine in
BED is unknown. Since the central norepinephrine system
is involved in the regulation of feeding behavior'4
and atomoxetinc is a highly se lective norepinephrine rcuptake inhibitor,ll..!J one possible mechanism by which
atomoxetinc might reduce binge eating is dec rea~ing ap
pet ite or enhancing satiet y Ihrough its effects on this sys
tem. Decreased binge eating might lead 10 reduced energy
intake and, secondarily. 10 weight loss. Alternatively. II
second possible mechan ism is that lltomoxetine might
reduce binge ellting via its gastroin tcstinal side effects.
A third possible mechanism is that atollloxetine may secondarily decrease binge eat ing through direct weight loss
e ffects. For eXlllllple. compounds wi th noradrenergic
properties have been hypothesizcd to induce weight loss
in part by increasing locomotor activity or enh.mcing
thennogenesis. J.I,49
Several limit:lt ions of this study should be cons idered.
First is that the allrition rate was high, with 37.5% of
patients wi thdrawing before study completion. This feature renders the results heavily dependent on assumptions
regarding missing data:UJ-'IO While the longitudinal analysis, unlike the endpoint analysis. allows that the mis.~
ingness can depcnd on observations obtained before with
drawal (e.g .. a patient who is fnilin g to improve may be
more likely to withdraw), it is nevertheless vulnerab le to
missingness that depends on factors that are not measured
prior to withdrawal (technically. m issi ng not al random or
non-ignorable missingness). Of nOle. although Ihe 37.5%
attrition rate in our study is substantially higher than the
overall altrilion rates of < 5% seen in the initial clinical
trials of ato moxet inc in ADHD. it is comparable to the
30% attrition rMe in the Gadde et al. 12week st udy ol'ato
moxetine in obese women. 36 The 2 mos t common reasons
for dropout while taking atomoxetine in the present study
were protocol nonadherence (15%) and adverse events
(15%). In the Glldde et al. study. I patient (3%) dropped
out for an ndverse event (rapid heartbeat with atomox

J Clin Psychiatry 68:3, March 2007

etine), and the other reasons (N 8; 26.7%) could be

attri buted 10 d ifficulty w ith protocol adhere nce: they
included time constrain ts (N 3 atomoxeti ne. N I
placebo). lost 10 follow- up (N :: I alomoxetine. N = 1 pia.
cebo). reloclltion
= I atomoxetine), and withdrawal
of consent
I place bo). It is thus noteworth y that
attrition is a sign ificant problem in clinical trial resenrch
in both obesitySI and bulimia nervosa. S!,') 2 conditions
related to B ED.~
Research into why att rition "lies are high in phannaco
therapy studies of BED are needed: such studies could
possibly focus on impulsivity, trea tment expectations. and
logistics factors that have been associated wit h attrition in
treatment stud ies o f obesity'~'s, and bu limia nervosa.Sl.!6
Such research might generate information that will lead to
increased completion rates in cl inical trials in persons
with eilting d isorders or obesity.
A second lim itation is Ihat Ihe accuracy of Ihe sel f
report methods used to obtain binge-eating dllta is uncer
la in .~7.ss However, patient diaries were used to enhllnce
patient recall of binges. and randomization and double
blinding should have equalized any patient or investigator
bias in the recording or raling of overeating episodes a.~
binges. A th ird limilation is tha t beCilUse the s tudy group
was small and primarily femllle and white, and Ihe dura
tion of treatment was short (10 weeks). the results may
not generalize to larger groups of persons w ith BED, to
males or African Americans wi th BED, or to longer trcat
ment periods. A fourth limitation is that because persons
with psychotic d isorders, bipolar disorders. substance use
disorders. severe personality disorders, and unstable
medical disorders were excluded, the res ults may not
generalize to BED when it cooccurs with these
condit ions.
In summary. in a IQ-week trial in outpatients with
BED. atomoxetine was found to be superior to placebo in
reducing binge frequency, weight. and severit y of illness.
It was also associated with fairly good tolerability but a
relatively high treatment d iscontinuation rate. Controlled
trials of atomoxetine in larger groups of patienL~ with
BED appcllT wnrranted .

eN =

eN

Drug ,wm<,s; ~101ll0xclinc (Slraltera). nuoxctinc (Prozac and olhcl'll).

tithium (Eskatilh. Ulh<>bid. und olocrs). ortiml (Xcnicat). sibulraminc
(Mcridia). lopiramale (Topamax). zalcpton (Sonala). wtpidcm
(Ambicn). zonisamidc (7.onegmn and OIhe!'!;).

Dr. McEtroy is a eonsulwnl In or member of

the scientific advisory boards of Abboll. EIi Ully. GJa~oSmilhKlinc,
Jnnsscn. Onhn-McNcil. and WyethAycrsl; llnd is a principat or
coinve. ligator on research studies sponsored by ASlrnZcncca,
Brislol-Myers Squibh. E..ai. Eti Ully. ForesI. Nalioll:ltlnslilutcs of
Mentat Hcatlh (N1MIt). Onho-McNcil. Pfiur. SaTKIfi-Symhetabo.
and 50ma"on. Dr. Kolwat is a speakcr for 8r~'Iol- Mycrs 5'1uibb.
Dr. Nelson has received ,,,,nl/research ~U pporl from Eli LiUy and
A~traZcne<:a. Or. Keck is a COnSUllanl 10 or mcmilerof Ihe .cicnliric
advisory boards of Abb<l1l. AstmZcneca. BriSlotMyers 5quibb.
GJa~oSmilhKlinc, Eli Ully. Memory Ph.rmaceU li c~ t s. Neurocrine
Bioseiences. PIi7.cr. and Shire; and is ~ pritlCitl~l Or co-invesligalOr on
Fill,weral disc/rlSllr(','

3%

HcElroy et at.

re~earch studics ~ ponsorcd by Abbot!. Americ~n Diabetes AsslKia

tioll. AstraZeneca. Brislol-MycT!i S<juibb. Gl~xoSmithKline. Eli Lilly.
JanSSCIl. Memory Pharm:xeut icals,. Merd:. NTMH. N3tiona llnstitutc
of Droll. Abuse. Organon, OnhoMcNei l. Plitcr. Stan ley Medical
Research Institute. and UCB Phamla. Dr. Hudson is a consu ltant to
an d has rece ived gmnt!reiiCarch suppon from EH Lil1y and Onoo
McNcil. and serves on the spca kcn; Or advisory board for El i Lilly.
Drs. Guerdjikova :md Wclse and M.<. Lakc repor t no other sign ifican t
co mmercial relationships reievanlto the .<tudy.

REFERENCES
I. American Ps)~hialric AssociatiolL Diasoo.<tic and StatiSlieal Manual of
Mental Disonlcrs. Founh Edition. Tnt Revision. W:Wtingtoo. DC:
American Ps)1:hialric Msoc iation; 2000
2. Walsh BT. cd. The Cum:nt Sta tus of Binge Eating Di~onlcr. lnt J Eat
Di sonl2oo3;34(~urpl):S I- S120
3. StreigclMoore RII, Frol1~o DL. Epidemiol ogy ofbilt!;" eati ng disorder.
[nt J E.lt DiMlnl2003:34(supp l):SI9--S29
4. [[udson J[, !liripi E, I'ojte IIG Jr. (1 al. n.c prevalence and c.me1ates of
eating disonlc:rs in the National Comorbidit y Survey Replication. Bioi
Psychiatry 2007:61:348-358
~. Yanovd:i Sz. Ne150n JE. Dubbc:n KB. et al. Association of binge ealing
disorder and I'-~ychi~tric comorbidity in obese ~ubjecls. Am 1 Psychiatl)'
1993: 150; 1472- 1479
6. Pcterson CB. Mill er KS. Crow SJ. et ~1. Subtype~ of binge eati ng disorder based on psychiatric history. 1111 1 Eat Disnnl 2oo5:3~:273-276
7. Yanovs~i SZ. Binse e~tinJ; disonlcr and obesity in 2003: could treating
~n eming di'\onler have a positive elTt 00 lhe obe~it) cpilk:mic?
[nI J Eoit Di_d 2003;34{wppl):S 117--5 120
8. Jobnwn JG. Spilzcr RL. WiUiams 18W. l!callh problems. imp;linncnt
and illnesses assoeiated wich hulimia 1lCTV0S0I a lld binb~ e~cing disorder
among primary "re and oItelri c g)1T1aecology patients. Psycho! Med
2001 :31: 1455-(466
9. Buli~ CM. RekhbomKj<:nncru d T. Medic"1 morbidi ty in binge eating
disorder. [nt J Eat Disord 2003:34:539-546
ID. Ritgcr E, Wilney DE. Sldn RI, Cl al. A ~()mparison of qua lity of life
in obese indi,idua (. with and " 'i thout binge eating disorder. 1nl J E;!t
Disonl200.'i;31:2 )4-240
11. Wonlk:rlich SA. deZwaan M. Micchell JE, et at ""reholngical and
dietary trea1menu or binge eming disorder: corot"eptual implication .
Int I Eat Di>onl2003:34:S58-573
12. Caner WP, HudSQII JT, LalOO<k JK, ct al. Phamtacologic treatment
of binge eating disorder. Int J E.l1 Disonl2003:34(suppl):S74-S88
13. Appo linario JC. McE(my SL. Pharm~cologic ~pproa<;he" in the treatmem ofbingc emins diso rder. Curr Drug T:''l>et< 2004:5:30 1-307
14. Grilo CM. Mas~b RM. Wilson GT. Efficacy or cognitivc beh,vioral
thcropy and nuo~ctine for the treatment of binge cating disorder:
a r.ondomizcd double blind placelxKontrollcd com p;lriwn. Bioi
Psychiatry 2005:57:)01-)09
15. !)evlin MI. Goldrcin l A. Pct~OV3 E. Cl. al. Cognitive bel13viorul thcr~1'Y
and nl101etine;JS ~djunets t" g""'p behaviorolther:lpy fnr hinge eati ng
disorder. Obe~ Res 2005:13:1077- 1088
16. Gri lo CM, Masheb RM. A ra nd"mizcd controlled conlparisoo of guided
selfhelp cog~iti"e ll<:]mviorJ I tlteropy and bebaviornl we ight loss for
hillge eating di_der.lkhav Res Ther 200~:43:1J-1525
17. National Institute for Clinical Excellence. E;!ting Disonlcrs-Corc
Intc"'. nllom in the Treatment and Ma""gcmcm of AOOfUIa Nervosa.
Bulimia Nervusa. Related Eating Disonlcrs. NICE Clinical Guideline
No. 9. 1.0000" . Engbnd: Nation:t.l lnstitule for Clinical Excellence;

"'"

IS. I(udson JI. McElroy SI.. R~ymond NC. et ut. Fluvonmine in the
treatment of bins. eating disorder: a multice nter pl:tecbocontroUed.
double-blind trial. Alii J Psychiatry I'Ng; 155: 1756-1762
19. McE lroy SI.. CJIUiO LS. Not,oo EB. Ct al. PI:tecbocontrolled tria l of
senrJlinc in the treatm ent of binge eati ng disonle r. Am 1 Psychi:ltry
2000;157:100-1-1006
20. Amold LM. MeElroy SL. ll udsoo 11.et al. A pl~bo ....."1)fItrollcd.
rnndomized tri~1 of nuoxCline in the treatment of binge..e3ting di'"nkr.
I Clin Psy<.:hialT)' 2002;63:1028-1033
21. McElroy SL.lludsoll lI, Mnlhot., S. Cl al. CitalOflrJm in tlte tre>t<nent
of binge ..eating disorder: ~ placebocontrolled trial. J Clin Psychialry

397

2003;64:807-8 13
22. Grilo CM. MlIShcb RM. Sawu SL Cognitive bell;"tviond therapy guided
sel fhelp and orli5tat ror the trell1mcnt of binge ealing disonlcr:
a r.ondomized. doublebli nd. placebo-<;Olllrollcd trial. Bi,,1 PSyclliatry
2005;57:119J.-1201
23. Appolinario JC, Bacaltchuk J, Sich~ri R. cl al. A rnndomi!ed. double
blind. plat:ebo-rotl1rolled ~tudy of sibutromine in the treatment of
binge-eating di sonlc r. Areh Gen Psychiatry 200);60:1109-1 116
24. Mibno W. Petrelln C, Casella A. ct at. Use or sibutramine, an inhibitor
of the reuj1l3~e of seroto nin and noradrcn'lline. in the treatment of binge
cacing disonlcr: n placebo-c:onlrolk:d sludy. Adv n.cr 2005:22:25-31
25. MeElroy SL.Amold LM. Sha" ir.J NA. CI..1. TopiramalC in the treatment
of bingceal'na; disorder associated with obe$ity: a f:lndomized. placcbQ.
controlled trial . Am J Psychiatry 2003;160:155-261
26. McEh"y SL.Iltldson JI, C:lpece JA. et al. Topiran\l<le fo. tbe treatmem
of bi nge eatins disorder associated with obesity: a pla,ebo-C:OIIIrolled
. tudy. Bio( Psychiatry 2007 Jan 26:[epub :\head of printl
27. McElroy SL. Kotwal R. Guenljikova A. ct al. Zonisamide in the treal
ment of binge e~tins di ~orde r wi th obesily: ~ pl~"ebocontfOlltd trial.
J Clin Psychiatry 2006;67: 1897- 1906
28. 10~nni!lcsDcm05 LL. Proieuo 1. McNcil JI. Phmn;KO[her.opy for
obesity. Drugs 2005:65:1391-14 (8
29. McElmy SL. Sharpiro NA, Amold LA. et al. Topiramate in the loog
term treatment ofbinge-eating disonkr associmed ""ith obesity. J ain
P'ycftiatry 2004:6S: 146J.-1 469
30. Sheldon TA. Cu llu m N. Daw5011 D. et al. Wh~t'~ the ev idence that
NICE guidance has been implemented'!: rc~ult $ rrom a nali onal eva lua
tion usint: lime ,;cries analysis. nu dit of p.1ticnls' notcs and interviews.
BMJ 2004:329:999
31. Michel lOO D. Faries D. Wcmicke 1. et al. AtomOl cti,., in the treatment
of children and at.Iolellll< with allentiondefici\JtlypcDClivicy disorder:
a randomized. placebo-<;ollIrolled. doseresponse study. Pcdi:dtics 2001:
108:e83
32. Michclson D. Adler L. SpencerT. et,1. Atorn oxel i,., in :!dullS .... ilh
ADHD: two ramlorn i!ed, placeiJn-<;ontrolled Sludies. Bio( Psychiatry
2003;53: 11 2-120
33. Chrislman AK. Ferl1lo JD. Ma!~owitz IS. Atomoxecine. a novcl treatmenl for attellliondcllcit.ItY]"I!!t:lCliviIY disorder. Pllamt3<"'ollterjpy
2004;24: 1020-1 036
34. Wcllman PI. Modulation of eating by central c3techol:trni,., systems.
CUff Drug Ta'let~ 2005;6:191-199
35. Ge1t1en DR. Orcshroeld L. Tins1cy F. el al . The selective oorepinephrine
reupta1:e inhibitor. LY368975. reduce$ food CQllsumptioo in animal
moods of feeding. J Phmna<:<>i E~p noer 1998;287: 122-127
36. Gadde KM. Yoni , h GM, Wagnc! HR 2nd. et al . AtOmMetine for weight
reduction in obese ...."me n: a prcliminJry r,omtomittd cnn trulled tri,l.
Int 1 Obes 2006;30: 11 38- 1142
37. Bacaltehuk J. Hay P. AllIidcpreSS:lnts v~rsus placebo for people with
bulimia nervosa. Cochr.t~ D.1taba.se Syst Rev 2003:CDOO3391
38. Kratochvil CJ. Newcom JH. Amold LE.. Cl al. Alomoletine alone or
rombincd w;th Ouoxetillc for treatins ADHD with cornorbid dcprusivc
or anxiety symptom5. J Am Acad OriJd Adole;;c Ps)1:hiatry 2005:44:
'115-924
39. Prcti A. Tomolctinc. Curr Opin IlIve.tit: Dru~s 2002:3:272-277
40. Carpenter LL. Milosavljev ic N. Schcctcr JM. et al. Augmentalion with
open-label atomoxetine for partial or non re~]1Onsc to anlidepre.<;sant.
I Clin P!;ychiatry 2005:66: 1234-1 238
41. Fim MB. Spitzcr RL. Gibbon M. Cl al. Structured Oinicallnterview for
DSMJV-TR Alis Tdi$OTllers. Resean:h Version. P:tticnt Ed ition (SCID
lIP). New York. NY: Biomelrics Research. New York SCIlC Psy<.:hiatric
Institute; 2002
42. Guy W. Clinical Olnbal lmpressioos Scale. ECDEU Assessment
Manual for i>sychophammcology. US D<:pt II~JlIh, Education. and
Welfare publication (AD M) 76338. Rockvi llc, Md: NJtiunallosti tutc
of Mcl1lull[c. lth: 1976:2 18-222
43. Gooontan WK. Pric~ LB. Rasmusscn SA. et al. Th( YalcBrown Obses
sive Contpul!ive Scale. 2: VJlidity. Ar,h Gcn Psychialry (989:46:
1012-10 16
44. StunhrdAJ. Messick S. Eating Inventory Manual. SaoAllIonio. Tu:
The Psycholo~ical Corporation. Barcoun Brace &. COntp;tny; 1988
45. Ibmillon M. Development ofa rating SC31e ror primJryd:=pressive
illness. Br J S[J(: Cl in Psycho! 1967;6:278- 2%
46. FilZmauricc GM, Lainl NM. Ware JH. Applied Longitu di na l Analysis.

J Glin Ps)'chiatry'68:3,

March 2007

Atomoxetine for Binge-Eating Disorder

Iloboken. NJ: John \Vik:y & Sons; 20();1

41. Gibbons RD. Hedeker D. Ellin I. el al. Some roncepl ual:md sU\i~t ical
i~sucs in Jn~ ly~is of longiludinal f'Sychi~lric dal a. Areh Gen P~ychi3lry
19 93:~O: 739-750

48. Gol"y A. L:lUrenl-J accard A.lbbicht F. Cl aJ. Effect of orlistJt in obc:.~

p;llic:nt' with binge (al inS tJ i.<onIcr. Obe! Res 2005; 13:170 1-1708
49. Cannon B. Nedergaanl J. 8 rown adipo.o;e ti.ssul:: function and physiologi
cal sisnirlC":lntt. PIJ)'iiol Rev 2004:84:277-359
50. Mallinckrodl CII. S~nccr TM . Dubc S. et aL. A!..<t'SSing and in1erpreti ng
Ircalmenl effects in Inngiludina l c1inicaltrials wil h mis~ing dala. Bioi
Psychiatry 2003:53:754-760
51 . PiSunyer FX. Aronne U. Heshmali IIM. Cl aL. ElTecl of rimomlbanl. a
cann.1binoid' "":")IIar blocker. on weight ~nd ~ardiomc laboiic ris ~ fac
tors in overw(:iglll Of obeS\: patients: R IO-Non h Ameri ....: a randomiud
~roIkd lfi~l . JAMA 2006;295:761-775
52. Ag,ns \VS. Cruw SJ. Halmi KA. et aL. OUlcome prediclors for the roSni.
live beha v;Qf trealmem ofbulintia nel'l'osa: dal~ from 3 mull is;le ' Iud y.
Am J P~ychi3 try 21XXl;I S7: 1302- 1308

J Clin Ps>'C niatry 68:3. r.larch 2007

53. Roouno SJ. HaLm i KA. 5Mhr NP.et al. A placebo-c:orurolled study
ofnuoxel;ne in continue<.! treatment of bulimu ncrvosa afler S~c~sful
aC~IC nuo~etine lrealmenl. Am J Psychiatry 2002;159;96-102
~4. Dalle Grave R. Mc1chionda N. C~lugi S. Cl a!. Con tinuous care in Ihe
trea lment of obe~ ity: ~n observa tionalmulii cenler study. J Inlem Me~
2005:25&:265-273
55. Nkr~oom C. Jansen E. Multens S. et al.l mpu lsiv ity predicts ~atmellt
outcOme;n obese children. 8~ha Res Ther2006 Jul5:lepob3head of
print)
56. Swan.Krcmeier LA. Milche ll JE. Twardowsk i T. et al. Travel disl.;Joce
and all nlion in oUlpal ienl ealing disorders lreatment. 1nl J Eal Disord
2005:3M:367-370
57. Grilo CM. Masheb RM . WilSO<l GT. Diffcn:nl methods far aSKssing
the fealures of ealing di~rders in patienls wilh bioI:<' e;lling disorder:
a replication. Obcs Res 2001;9:4 18-422
~g. le Grd"J:C" D. Goon A. Calley D. et 31. Does momenlary ;messmcn1
deted binge ealing in IlVc rwc ight women thal is de nied at inlcl'I'iew?
Em Eal Disorder> Rev 200 1:9:309- 324

v,

"T

398