\

PERGAMON Journal of Psychiatric Research 22 "0888# 018027

99112845:88 ,08[99 0888 Elsevier Science Ltd[ All rights reserved[
PII] S9 9 1 1 2 8 4 5 " 8 7 # 9 9 9 1 9 X
Neuropsychological assessments before and after treatment in
patients with anorexia nervosa and bulimia nervosa
Christoph J[ Lauer
a\b\
\ Birgit Gorzewski
a
\ Monika Gerlingho}
a
\ Herbert Backmund
a
\
Josef Zihl
a\b
a
Ma: Planck Insiiiuic of Psychiairy\ Kracpclinsirassc 10, D-80804 Munich\ Gcrnany
b
1uawiq Ma:inilians Unircrsiiy\ Insiiiuic of Psycholoqy\ Dcparincni of Ncuropsycholoqy\ 1copolasirassc 13, D-80802 Munich\ Gcrnany
Abstract
In psychiatric patients the identi_cation of cognitive de_cits which predict a poor clinical outcome is important for the development
of speci_c treatment strategies aimed at the amelioration of these impaired cognitive functions to increase the likelihood of full
clinical remission[ However\ such attempts are absent in bulimia nervosa "BU#\ are scarce in anorexia nervosa "AN# and\ furthermore\
provide con~icting results[ In the present prospective study we investigated the neuropsychological demands in 01 patients with AN
and in 03 patients with BU before\ during\ and after a treatment period[ At the initial testing session\ both patients samples showed
similar and impaired performance levels on tasks measuring attentional demands and problemsolving abilities\ while their mnemonic
functions were preserved[ At the _nal testing session\ which took place 6 months thereafter\ the impaired cognitive functions had
improved to a similar degree in the AN and the BU subgroups[ However\ although the eating disorder symptomatology had
ameliorated in parallel\ no direct associations could be established with the initial neuropsychological demands and their recti_cation\
respectively[ On an individual level\ 00 patients initially showed obvious cognitive de_cits[ However\ the clinical characteristics of
this subgroup di}ered not from that found in the 04 {good performers|[ These _ndings indicate that the cognitive functions in the
acute AN and BU are similarly impaired\ but also ameliorate in a similar manner with clinical remission[ Because no associations
were obvious between cognitive and clinical recti_cations\ signi_cant contributions of mediating factors "e[g[\ changes in metabolic
brain turnover and in steroid hormones# are suggested[ 0888 Elsevier Science Ltd[ All rights reserved[
Kcy woras. Eating disorder^ Prospective study^ Neuropsychology^ Attention
0[ Introduction
There is growing evidence that patients with a given
psychiatric disorder in parallel with speci_c cognitive
impairments are more likely to have an unfavorable out!
come than patients with a similar disorder but who per!
formwell on neuropsychological tasks "see] Keefe\ 0884#[
The identi_cation of such cognitive de_cits predictive for
a poor clinical prognosis would allow the development
of additional and speci_c treatment strategies to increase
the likelihood of a favorable outcome[ Furthermore\ if
the patients| cognitive performance are assessed before
and after the treatment period\ a signi_cant contribution
toward unraveling the unclear interdependencies among
cognitive impairments\ primary psychopathology and
secondary psychopathology could then be expected[
Corresponding author[ Klinik Angermu hle\ Department of Sleep
Medicine\ Angermu hle 7a\ D!83358 Deggendorf\ Germany[ Tel[] 38
88 26944 68^ fax] 38 880 26944 86[
In eating disorder patients\ this topic was addressed in
two investigations using a cross!sectional design[ In the
_rst study "Strupp et al[\ 0875#\ which focused mainly on
e}ortful vs automatic information processing\ no di}er!
ences were obvious between acute underweight and
weight!recovered anorexic patients[ Pendleton!Jones et
al[ "0880# found a poorer performance on tasks cor!
responding to attentional focussing\ verbal memory and
visuo!spatial reasoning in their underweight anorexics
compared to weight!restored anorexics^ however\ these
di}erences were subtle and nonsigni_cant[
Prospective studies reported on more or less mildly
impaired skills\ such as short!term verbal and visual
memory\ visuospatial construction:problem solving and
reaction time\ in acute underweight anorexics^ after
weight recovery\ these cognitive de_cits had improved in
most "Hamsher et al[\ 0870^ Kohlmeyer et al[\ 0872^ Small
et al[\ 0872^ Szmukler et al[\ 0881# but not all studies
"Green et al[\ 0885^ Kingston et al[\ 0885#[ In three of
these studies\ the authors attempted to identify predictors
for the clinical outcome "e[g[\ degree of weight!gain\ stab!
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 029
ility of recovered body weight] Hamsher et al[\ 0870^
Small et al[\ 0872^ Szmukler et al[\ 0875#[ However\ the
results reported are con~icting[ Hamsher et al[ "0870#
found their anorexic patients who showed impaired per!
formance on two or more neuropsychological tasks to be
more likely to exhibit an unfavorable outcome "weight
loss# after one year of follow!up than patients with very
mild or no cognitive de_cits[ Small et al[ "0872# reported
the performance in the digit span task\ which measures
immediate "working# memory capacity\ to be a powerful
predictor of weight gain in their anorexic patients[ In
contrast\ Szmukler et al[ "0881# failed to identify any
valuable predictors\ although they had applied a broad
range of neuropsychological tasks[
To our knowledge\ there has been no investigation
published in which a comparable prospective design was
applied to patients with bulimia nervosa[ Cross!sectional
studies\ however\ provide fairly good evidence that the
cognitive performance in bulimic patients is similarly
impaired as in underweight anorexic patients "e[g[ Laessle
et al[\ 0878^ Pendleton!Jones et al[\ 0880^ Cooper and
Fairburn\ 0881^ Perpina et al[\ 0882#[
The purpose of the present prospective study was
twofold[ First\ we monitored neuropsychological task
performance in 01 anorexic as well as in 03 bulimic pat!
ients before\ during and after treatment with the initial
testing session taking place four weeks before treatment
onset[ Our hypothesis was that anorexics and bulimics
would show similar cognitive de_cits before clinical
admission and that these de_cits would be improved simi!
larly in both disorders at post!treatment[ In addition\ we
attempted to identify neuropsychological predictors for
the clinical outcome "anorexic patients] degree of weight
gain^ bulimic patients] decrease in binges per week#[
Because the literature on this issue provides con~icting
_ndings in anorexics and is lacking in bulimics\ no speci_c
hypothesis was formulated[
1[ Methods
2.1. Subjccis
A total of 52 female patients were recruited from the
Therapy Center for Eating Disorders "TCE# at the Max
Planck Institute of Psychiatry[ According to the diag!
nostic criteria of the DSM!III!R "APA\ 0876# 07 patients
presented with anorexia nervosa "AN#\ 22 patients with
bulimia nervosa "BU#\ 7 patients with both anorexia and
bulimia nervosa\ and 3 patients with eating disorders not
otherwise speci_ed "NOS#[ During the 6!month study
period\ 29 patients dropped out "5 patients with AN\ 08
patients with BU\ 3 patients with both AN and BU\ and
one patient with an eating disorder NOS# because they
either dropped the therapeutical schedule "n 04#\ left
the county "n 7# or refused to participate at the _nal
testing session "n 6#[ Although 4 patients with a mixed
anorexic and bulimic eating disorder and 2 patients with
an eating disorder NOS completed all testing sessions\
we considered the size of these two subsamples to be
too small for statistical analyses and\ therefore\ excluded
them from the data analyses[ Thus the _nal study sample
consisted of 15 patients[ Age\ duration of the illness and
years of education were similar in the AN and the BU
"see Table 0#[ As expected\ the AN were of signi_cantly
lower body weight "expressed as ) of ideal body weight^
) IBW# and experienced signi_cantly fewer binges per
week than the BU[
2.2. Siuay acsiqn
The patients were investigated on four occasions which
were closely matched to the therapeutical schedule estab!
lished at the TCE"for detailed description see Gerlingho}
and Backmund\ 0884#[ The _rst testing session "i
0
# took
place at the beginning of a motivational period four
weeks before the onset of any therapy\ the second "i
1
#
took place during the _rst week of therapy\ the third "i
2
#
after 01 weeks of therapy\ and the _nal testing session
"i
3
# took place after 05 weeks of therapy plus 7 weeks
during which the patients were on an out!patient status
with only one therapeutical contact per week[
2.3. Psychonciric asscssncnis
At i
0
\ i
2
and i
3
\ the patients rated themselves on the
Eating Disorder Inventory "EDI^ German version^ Thiel
and Paul\ 0877#\ the Beck Depression Inventory "BDI^
German version^ Hautzinger et al\ 0881# and the Symp!
tom Checklist\ revised version "SCL!89!R^ Derogatis et
al[\ 0865#[ For the purpose of the present analyses and
to reduce the number of psychometric measurements\ a
principal component analysis "PC^ method] stepwise# was
performed at i
0
using the BDI and all subscales of the
EDI and of the SCL!89!R[ This resulted in a _ve factor
solution which explains 66[1) of variance*BDI "Eig!
envalue 00[90# and the four EDI subscales {drive for
thinness| "Eigenvalue] 1[10#\ {bulimia| "Eigenvalue] 0[31#\
{body dissatisfaction| "Eigenvalue] 0[12# and {ine}ec!
tiveness| "Eigenvalue] 0[01#[ The remaining psychometric
scales did not contribute to a signi_cant improvement of
this solution and\ therefore\ will not be further
considered[
2.4. Coqniiirc asscssncnis
2.4.1. Hiicniion
Speed of information processing was measured by the
Letter Cancellation Task "d1^ Brickenkamp\ 0867# and a
Trail Making Test "Zahlenverbindungstest\ ZVT^ Oswald
and Roth\ 0867#[ In the Letter Cancellation Task\ the
subjects were asked to search for the target letter {d| with
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 020
Table 0a
Characteristics of the 15 patients with eating disorders
Anorexia nervosa Bulimia nervosa
"n 01# "n 03#
i
0
i
3
i
0
i
3
Age "years# 08[621[8 * 10[822[0 *
Years of education 00[720[6 * 01[520[8 *
Illness duration "months# 30[3249[2 * 38[3225[3 *
Body weight ") IBW#
a
69[026[7 75[725[7 88[52
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 021
erages\ 3 fruits\ 3 garments\ 3 tools# was verbally pre!
sented in an unstructured manner to the patients followed
by immediate free recall[ The test variable was the number
of correct recalled items after the 4th trial[ Long!term
verbal memory was measured using the reproduction of
the 45!itemstory 13 h after its _rst presentation[ Working
memory capacity for verbally presented material was
assessed by the backward memory span for digits "Wechs!
ler\ 0876#\ for visually presented material by the back!
ward span in the Corsi task "Milner\ 0860#[
2.4.3. Problcn solrinq
The Raven Standard Progressive Matrices Test "Raven
SPM^ Raven et al[\ 0866# was used as a measure of intel!
lectual status[ This task consists of 59 items that di}er
with respect to the level of complexity of the visual!
cognitive problems[ The number of correctly solved items
served as the measure of intellectual capacity[ In addition\
the patients had to complete the Daily!Living Problem
Solving task "von Cramon\ 0877#[ This task requests the
arrangement of _ve daily!living jobs into a _xed time
schedule^ the items of information provided are the times
of opening and closure of the shops\ the duration of the
particular job and the walking distance "min# between
the shops[ This task has one correct solution and the test
variable was whether or not the patients were able to
_nd it[
The Trail Making Test and the Divided Attention Task
were assessed at each of the four testing sessions\ while
the remaining tests were assessed at i
0
and i
3
[ In order to
reduce the possibility of practice e}ects\ the available
parallel forms of the tasks were administered "ZVT\
MVGT\ digit span\ Corsi span\ Free Paragraph Recall#
at the di}erent assessments[
According to the normative data provided for the
respective tasks\ defective performance was de_ned as
values below the third percentile and lower than two
standard deviations "SD# of mean normative data\
respectively^ mildly impaired performance was de_ned as
values below the 05th percentile and below one SD of
mean normative data\ respectively^ the remaining values
were de_ned as average performance[ The respective age
corrected percentile values were obtained from the pub!
lished normative data*Trail Making Test "{average per!
formance|] 0[04 items:s and more#\ divided attention task
"532 ms and less#\ Letter Cancellation task "0[1 correct
items:s and more#\ Digit Span "4 items and more# Corsi
Span "4 items and more#\ Paragraph Recall "16 items and
more#\ Raven SPM "26 items and more#[ The estimation
of the patients| performance on the MVGT was based on
our large data pool of young healthy subjects "not age
corrected^ average performance] 01 items and more#[
2.5. Siaiisiical ncihoas
Beside descriptive statistics "mean2standard devi!
ation#\ a multivariate analysis of variance "MANOVA#\
repeated measurement design\ was performed with group
"anorexic patients\ bulimic patients# as the between!sub!
ject factor and time "i
0
\ i
3
# as the within!subject factor[
Furthermore\ multiple regression analyses "method] for!
ward stepwise# were performed to identify neuro!
psychological predictors for the clinical outcome at i
3
[
Level of signi_cance was set at 4)\ two!tailed[ All com!
putations were made with SPSS for Windows\ 6[9[
2[ Results
3.1. Group conparisons ai t
0
"n40#
In order to evaluate group di}erences between larger
subsamples of anorexic and bulimic patients "n 07 and
n 22\ respectively# and between patients who com!
pleted the study protocol "n 15# and those who dropped
out "n 14#\ a two!factor MANOVA was performed
with the diagnosis "AN\ BU# and {completers|:drop!outs
as the two factors[ Regarding the factor diagnosis\ a
signi_cant main group e}ect was observed "l 9[98\
P9[990#[ Subsequent univariate F tests performed
within the MANOVA revealed that the bulimic patients
had a signi_cantly higher body weight F"0\36# 83[42\
P9[990\ more binges per week F"0\36# 04[93\
P9[990 and higher scores on the EDI subscale
{bulimia| F"0\36# 009[68\ P9[990[ The remaining
variables investigated\ including all neuropsychological
tasks "see Table 0a and 0b# did not di}er signi_cantly
between both subgroups F"0\36# 0[81\ P9[07^ data
not presented[
Regarding the factor {completers|:drop!outs\ the main
group e}ect failed to reach level of signi_cance "l 9[42\
P9[12#[ Instead of this nonsigni_cant group e}ect\ we
compared the neuropsychological variables of these two
subgroups by using univariate F tests[ No signi_cant
group di}erences were found F"0\38# 0[50\ P9[10
with the exception of the Trail Making Test on which the
drop!outs tended to perform slower than the patients
who completed the protocol 62[7219[8 s[ vs 53[5202[9
s^ F"0\38# 2[44\ P9[95[
The main interaction term "diagnosis
{completers|:drop!outs# was far from level of signi_cance
"l 9[55\ P9[69#[
3.2. Clinical ouiconc "t
0
rs t
3
#
Because complete data sets were evaluated at i
0
and
i
3
\ a repeated measurements MANOVA was performed
including these two time points[ The obtained Wilks|
lambda was signi_cant for the main group e}ect
"l 9[08\ P9[90#\ the main time e}ect "l 9[16\
P9[990#\ and the main interaction term "l 9[05\
P9[94#[ Subsequent univariate F tests performed
within the MANOVA revealed that the body weight\ the
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 022
frequency of binges and the EDI subscale {bulimia| were
the parameters that signi_cantly contributed to the group
e}ect and the interaction term\ respectively "Table 0a\
b#[ Signi_cant time e}ects were observed for all clinical
measurements assessed[
3.3. Ncuropsycholoqical ncasurcncnis "t
0
rs t
3
#
Because complete data sets were evaluated at i
0
and
i
3
\ a repeated measurements MANOVA was performed
including these two time points[ The Wilks| lambda
obtained was nonsigni_cant for the main qroup e}ect
"l 9[56\ n[s[# and the main interaction term "l 9[46\
n[s[#^ the main time e}ect was signi_cant "l 9[96\
P9[990#[ Subsequent univariate F tests within the
MANOVArevealed that performances in the Trail Mak!
ing Test\ the Divided Attention Task\ the Letter Can!
cellation Test "correct items:s and ) errors#\ and the
Raven SPM test were all improved at i
3
"Table 1a\ b#[ No
systematic changes were seen regarding the Digit Span\
the Corsi Span\ the short! and long!term verbal memory
and the MVGT[ As indicated by the nonsigni_cant main
group e}ect\ no di}erences were found between the AN
and the BU[ Finally\ the only signi_cant interaction term
observed was regarding the immediate Paragraph Recall
with the number of items recalled decreased in the AN
and increased in the BU^ however\ because the main
interaction term was nonsigni_cant\ this _nding has to
be interpreted with caution[
Regarding the Trail Making Test and the Letter Can!
cellation Test\ inclusion of the _ndings at all four test
sessions did not change the results as mentioned above
as was the case when considering the self ratings "BDI\
EDI# at i
0
\ i
2
and i
3
"data not shown#[
Considering the severity of depressive symp!
tomatology "BDI# as a co!variable\ the results as men!
tioned above did not change[ The only exception was\
that the afore!mentioned signi_cant time e}ect on the
percentage of errors in the Letter Cancellation Test dis!
appeared after controlling for depressive symp!
tomatology[
3.4. Iacniifcaiion of ncuropsycholoqical ana psycho-
paiholoqical rariablcs prcaiciirc for ihc ihcrapcuiic oui-
conc ai t
3
In the AN patients\ the therapeutic outcome was
de_ned as the increase of body weight at i
3
") IBW^
mean delta weight gain i
3
i
0
] 05[7209[1 ) IBW#[ In
order to identify the measurements assessed at i
0
which
would be predictive for the weight gain at i
3
\ a multiple
regression analysis was performed with the neu!
ropsychological measures\ the BDI and the four subscales
of the EDI\ age\ years of education and duration of illness
as the independent variables[ However\ none of these
parameters signi_cantly predicted the observed weight
gain[
In the BU patients\ the therapeutic outcome was
de_ned as the decrease of binges per week at i
3
"mean
delta binges:week i
3
i
0
] 01[3203[3#[ Using the same
independent variables as mentioned above\ this decrease
was signi_cantly predicted by the EDI scale {drive for
thinness| "r 9[67#\ the patients| age "r 9[44# and the
BDI score "r 9[57^ adjusted T
1
9[742\ P9[990#[
However\ none of the neuropsychological assessments
signi_cantly predicted the decrease of binges[
3.5. Iacniifcaiion of rariablcs prcaiciirc for ihc ncuro-
psycholoqical ouiconc ai t
3
In order to identify measurements assessed at i
0
which
would be predictive for the signi_cant improvements on
neuropsychological tasks "delta!values i
3
i
0
#\ multiple
regression analyses were performed with the BDI and the
four subscales of the EDI\ age\ years of education and
duration of illness as the independent variables[ The
decrease in ) errors in the Letter Cancellation Test was
predicted by the scoring on the EDI scale {Ine}ectiveness|
"adjusted T
1
9[19\ P9[94#[ The improvements in the
remaining tasks could not be signi_cantly predicted by
any of the independent variables included in the model[
3.6. `Gooa pcrforncrs ana `poor pcrforncrs ai t
0
Good performers were de_ned as patients who showed
one mildly impaired cognitive performance or none at
all^ this criterion was ful_lled by 04 patients "8 AN\ 5
BU#[ Poor performers were de_ned as patients who
showed mildly impaired or defective performance on two
or more tasks^ this criterion was ful_lled by 00 patients
"2 AN\ 7 BU#[ Calculation of a repeated measurements
MANOVA as mentioned above resulted in a signi_cant
main group e}ect "good performers vs poor performers]
l 9[06\ P9[94# and a signi_cant main time e}ect
"i
0
vs i
3
] l 9[92\ P9[990#^ the interaction term was
nonsigni_cant "l 9[26\ n[s[#[ Subsequent univariate F
tests performed within the MANOVA revealed that the
following measurements signi_cantly contributed to the
main group e}ect] the Divided Attention Task
F"0\14# 00[27\ P9[90\ the Letter Cancellation Test
correct items:sec[] F"0\14# 5[56\ P9[94^ ) errors]
F"0\14# 01[39\ P9[90#\ the digit span F"0\14#
4[04\ P9[94\ the Corsi span F"0\14# 5[39\ P9[94
and the Raven SPM F"0\14# 3[22\ P9[94[ The
remaining measurements\ including the BDI and the EDI
subscales\ did not di}er between both subgroups[ The
signi_cant time e}ect was due to systematic changes in
almost all parameters under investigation "F01[81\
P9[90#\ except for the Digit Span\ the Corsi Span and
the measurements of verbal memory "short! and long!
term recall\ MVGT\ F0[89\ P9[06#[ For signi_cant
subgroup di}erences at i
0
and i
3
see Fig[ 0[
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 023
Table 1a
Neuropsychological assessment in 15 patients with eating disorder "mean2SD#
Anorexia nervosa Bulimia nervosa
"n 01# "n 03#
i
0
i
3
i
0
i
3
Trail making test "items:s# 0[3529[13 0[6329[13 0[3229[21 0[6429[30
Divided attention "ms# 564[7269[8 593[7269[5 564[6272[9 476[6253[3
Letter cancellation test
Correct items:s 0[6229[22 0[7129[17 0[5429[22 0[7429[29
) errors 3[824[8 1[121[9 4[223[3 2[022[9
Digit span "backw[# 4[429[8 4[229[8 4[220[4 4[520[1
Corsi span "backw[# 4[120[4 4[729[8 4[420[4 4[520[1
Paragraph recall
Short!term "items# 31[224[8 28[526[8 28[3209[8 33[027[1
Long!term "items#
a
38[121[5 40[223[1 38[522[3 49[423[7
MVGT "items 4th trial# 04[229[8 04[020[9 03[829[8 03[520[3
Raven SPM "items# 38[423[5 41[423[3 40[022[1 42[822[5
Daily!living problem task 00:0 00:0 7:5 01:1
"completed without:with cues#
Table 1b
Statistical results
MANOVA repeated measurements
Group e}ect Time e}ect Interaction term
F"0# F"0# F"0#
Trail making test "items:s# 9\99 66\71 9\08
Divided attention "ms# 9\94 31\08 9\24
Letter cancellation test
Correct items:s 9\95 10\23 2\45
) errors 9\20 7\87 9\01
Digit span "backw[# 9\92 9\13 0\40
Corsi span "backw[# 9\93 0\59 9\43
Paragraph recall
Short!term "items# 9\12 9\48 6\62
Long!term "items#
a
9\91 2\29 9\48
MVGT "items 4th trial# 9\23 9\90 9\90
Raven SPM "items# 9\79 13\27 9\96
Daily!living problem task n[s[
b
a
n 13^
b
x
1
test[
P9[94^ P9[90^ P9[990[
MVGT] Munich Verbal Memory Test[
In order to identify associations between clinical
improvements "BDI\ EDI\ ) IBW\ binges per week#
and neuropsychological task performance\ the respective
delta values "i
3
i
0
# were correlated for each subgroup
using Spearman rank correlation due to the small size of
the poor performance subgroup "n 00#[ However\ none
of the computed correlation coe.cients reached the level
of signi_cance "4)\ two!tailed#[
3[ Discussion
The major _nding of the present prospective study was
that patients with anorexia nervosa "AN# and bulimia
nervosa "BU# did not di}er on their neuropsychological
task pro_le assessed four weeks before the onset of a
speci_c therapeutic program[ While the mnemonic
capacities were well preserved\ these patients showed mild
to moderate de_cits particularly on those tasks covering
attentional demands and problem solving abilities[ After
seven months "at i
3
#\ the speed of cognitive information
processing and the problem solving skills had similarly
and signi_cantly improved in the AN and the BU[
Although the psychopathology of the patients in parallel
had markedly improved\ we failed to identify any neu!
ropsychological baseline measurements "i
0
# to sig!
ni_cantly predict the amelioration of the clinical symp!
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 024
Fig[ 0[ Mean values "2SD# at i
0
and i
3
of neuropsychological tasks for which the repeated measurement MANOVA yielded a signi_cant main group
di}erence between {good performers| "# and {poor performers| "#[ n[s[] nonsigni_cant^ P9[94^ P9[90[
toms[ On an individual level\ 04 patients "46)# quali_ed
as {good performers| and 00 patients "31)# as {poor
performers| at i
0
[ However\ these two subgroups did not
di}er with regard to any of their demographic and clinical
characteristics or the recti_cations of their neuro!
psychological demands[ Furthermore\ in none of these
subgroups could a predictor of the clinical outcome be
identi_ed[
The present observations of impaired attention and of
de_cits in problem solving skills in acute anorexic and
bulimic patients coincide with several cross!sectional
reports "e[g[ Touyz et al[\ 0875^ Laessle et al[\ 0878^ Pen!
dleton!Jones et al[\ 0880# and with almost all prospective
studies on anorexic patients "Hamsher et al[\ 0870^
Kohlmeyer et al[\ 0872^ Small et al[\ 0872^ Szmukler et
al[\ 0881#[ However\ Kingston et al[ "0885# found the level
of the visuospatial abilities and of the immediate memory
to be de_cient in their anorexics irrespective of whether
these patients had been in the acute underweight or in a
weight gained status[ Green et al[ "0885# reported similar
mnemonic disabilities and discussed their _ndings in
terms of an impaired working memory\ in particular of
the phonological loop which\ in conjunction with the
central executive component and the visuospatial sketch
pad\ forms the working memory system"Baddeley\ 0881#[
In the present study as well as in others "Hamsher et al[\
0870^ Szmukler et al[\ 0881# only a little evidence was
found for such a {speci_c| de_cit[ On the average\ our
anorexics and bulimics performed quite normally on the
tasks measuring the capacity of the phonological loop
"Digit Span#\ of the visuospatial sketch pad "Corsi Span#
and of the short! and long!term verbal memory[ These
_ndings indicate that the function of the working memory
and the processes of encoding and retrieval are not a}ec!
ted in acute eating disorder patients[
Regarding the attentional demands in our eating dis!
order patients\ the _ndings are more complex[ Although
the level of performance increased as a function of time
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 025
in all the tasks predominantly measuring aspects of atten!
tion "Trail Making Test\ Divided Attention Task\ Letter
Cancellation Test#\ only that of the Divided Attention
Task was impaired at i
0
[ Because reaction time is one of
the critical variables of this task\ one might simply attri!
bute this _nding to the general slowing of motor
responses associated with starvation "Maxwell et al[\
0873^ Green et al[\ 0885#[ Alternatively\ the poor per!
formance on this task may be related to an increased
susceptibility to interferences which is a robust _nding in
research on selective information processing in eating
disorders "e[g[ Ben!Tovim et al[\ 0878^ Fairburn et al[\
0880^ Perpina et al[\ 0882^ Long et al[\ 0883#[ In addition\
such an increased susceptibility would well explain our
_nding that a number of patients "16)# were unable to
complete the Daily!Living Problem Task without any
cues[ Therefore\ the impairments on the Divided Atten!
tion and on the problem solving task may depend on a
common underlying de_cit\ that is an insu.cient func!
tioning of selective and ~exible processing of more than
one piece of information[ Although speculative\ this
notion is supported by the _ndings of Kingston et al[
"0885# who demonstrated such attentional de_cits in par!
allel with impaired ~exibility:inhibition functions in their
underweight anorexics[
At the _nal testing session "i
3
# the impaired cognitive
skills were signi_cantly improved in the AN and the BU
subgroups[ These e}ects were most pronounced in the
Trail Making Test and the Divided Attention Task\
which were applied at all four testing sessions^ therefore\
a practice e}ect cannot be excluded[ However\ a similar
strong e}ect was found regarding the Letter Cancellation
Task that was performed only at i
0
and i
3
\ due to the
seven!month period in between\ practice e}ects should
be of less in~uence here[ Aclear estimate of which portion
of the cognitive recti_cations has to be attributed simply
to practice cannot be made in the present study\ because
we had not included a control group[ Nevertheless\ it is
certainly possible\ that the improved attentional abilities
are\ in part\ related to a decrease in the susceptibility to
interferences and an increase of the ~exibility:inhibition
functions as was reported by Kingston et al[ "0885#[ This
is further supported by our _nding that 77) of the pat!
ients completed the Daily!Living ProblemTask at i
3
com!
pared to 62) at i
0
[ Thus\ in the present study as well as
in others "Hamsher et al[\ 0870^ Kohlmeyer et al[\ 0872^
Small et al[\ 0872^ Szmukler et al[\ 0881^ and\ in part\
Kingstone et al[\ 0885#\ the cognitive de_cits which had
been present in eating disorder patients before the onset
of a speci_c treatment program\ vanished after seven
months irrespective of whether the patients su}ered from
an anorexic or a bulimic disorder[
One of the major aims of the present study was to
identify neuropsychological measurements at i
0
that
would predict the clinical outcome[ Hamsher et al[ "0870#
reported that a fairly good posttreatment performance
was indicative for a favorable outcome at the one!year
follow!up\ while a poor performance was not[ Unfor!
tunately\ these authors did not indicate which of the
iniiially assessed neuropsychological parameters had
such a predictive value[ In a second study "Small et al[\
0872# the performance on the Digit Span was identi_ed
to su.ciently predict the degree of weight gain that the
anorexics had reached at post!treatment[ However\ in the
present study and that of Szmukler et al[ "0881# none
of the neuropsychological results quali_ed as a valuable
predictor of the clinical out!come\ although we had also
included the Digit Span as an independent variable in
the multiple regression approach applied[ Given these
con~icting observations\ it appears rather unlikely that
the cognitive impairments as well as their recti_cations
are directly interrelated with the type and the severity of
the eating disorder "see also Strupp et al[\ 0875^ Green et
al[\ 0885#[
At present\ one would suggest that certain cognitive
de_cits are evident in anorexic and bulimic patients\ but
disappear when the eating disorder patients recover[
However\ when looking at an inairiaual level\ this sugges!
tion is relevant in less than half of the patients inves!
tigated so far[ Considering two or more impaired task
performances as a cut!o}\ 24) of patients are reported
to be cognitively a}ected in the acute episode of the eating
disorder*Hamsher et al[\ 0870] 8 patients out of 19 pat!
ients "34)#^ Szmukler et al[\ 0881] 5:07 "22)#^ Kingstone
et al[\ 01:35 "15)#^ present study] 00:15 "31)#[ This
quota drops to 13) at post!treatment*Hamsher et al[\
0870] 6 patients out of 19 patients "24)#^ Szmukler et
al[\ 0881] 4:07 "17)#^ Kingstone et al[\ 0885] 8:35 "19)#^
present study] 5:15 "12)#[ In other words\ 54) of the
patients with an acute eating disorder have well preserved
cognitive skills\ while in the remaining patients displaying
cognitive de_cits "24)# a recti_cation can be observed in
only 00)[ Therefore\ it appears not surprising that the
reports on the cognitive abilities in eating disorders pub!
lished so far do not provide a very consistent picture\
especially as {positive| _ndings depend not only on the
severity of cognitive de_cits but\ more importantly\ on
the number of patients displaying such de_cits\ even
more\ when the respective quota in general is low "about
24)#[ In our opinion\ future research has to focus
especially on those eating disorder patients who indeed
perform poorly on neuropsychological tasks in order to
prevent any masking e}ects due to intermingling patients
with preserved and impaired cognitive skills[ In the pre!
sent study\ the attempt to characterize the {poor per!
formers| on a personal "e[g[ age\ years of education# and
a clinical level "e[g[ BDI\ EDI# as opposed to the {good
performers| yielded completely negative results\ as was
the case in the study by Hamsher et al "0870#[ In addition\
our attempt to identify neuropsychological task variables
reliably predicting the clinical outcome in the {poor per!
formers| yielded completely negative results[
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 026
In the present study as well as in all others\ the ameli!
oration of the cognitive de_cits had run in parallel with
that of the eating disorder symptomatology[ However\
there is little evidence that both processes are closely
interrelated "e[g[ Strupp et al[\ 0875^ Green et al[\ 0885#[
Therefore\ the question arises of whether there is a third
factor mediating between eating disorder symptoms and
neuropsychological demands[ Hamsher et al[ "0870# had
speculated that stress due to hospital admission might
obscure task performance in eating disorder patients[ To
control for such a bias\ we had investigated our patients
four weeks before admission\ but the pro_le of cognitive
de_cits observed was surprisingly similar to that usually
found at or shortly after hospital admission[ Therefore\
it is rather unlikely that the level of performance is a}ec!
ted by unspeci_c stress due to hospital admission[ It can
also be speculated that the impaired cognitive functions
in eating disorder patients are due to secondary psycho!
pathological symptoms\ in particular of depression[
However\ in the present study as well as in numerous
others "Hamsher et al[\ 0870^ Beatty et al[\ 0889^
Pendleton!Jones et al[\ 0880^ Szmukler et al[\ 0881^
Cooper and Fairburn\ 0882^ Green et al[\ 0885^ Kingston
et al[\ 0885# no evidence was found that the severity of
depressive symptomatology had in~uenced the respective
cognitive _ndings in these patients[
One might also speculate that the cognitive de_cits
depend on morphological and functional brain alter!
ations that are present in the acute episode of an eating
disorder\ but tend to normalize with clinical recovery
"e[g[ Krieg et al[\ 0877^ 0878^ Herholz\ 0885^ Delvenne et
al[\ 0885^ Kingston et al[\ 0885#[ Palazidou et al[ "0889#
reported on a signi_cant association between mor!
phological brain alterations "size of the cortical sulci#
and the performance on the digit symbol test^ however\
Laessle et al[ "0878# and Kingston et al[ "0885# failed to
replicate such associations indicating that crude mor!
phological measurements are too insensitive for matching
with cognitive de_cits[ A more conclusive picture can
be expected from "not yet available# studies in which
neuropsychological assessments will be combined with
functional brain imaging[ At least regarding the latter
topic\ several investigations have revealed a relative hyp!
ermetabolism in the caudate nuclei and the inferior fron!
tal cortex associated with a relative hypometabolism in
the parietal and superior frontal cortex in both anorexics
and bulimics "Herholz et al\ 0876^ Delvenne et al[\ 0885\
0886#[ After weight restoration\ the parietal hypo!
metabolism and the inferior frontal hypermetabolism
tended to persist\ whereas the remaining changes had
normalized "Delvenne et al[\ 0885#[ These _ndings appear
to coincide with the assumption that especially the
capacity of the working memory\ which is functionally
{localized| in the frontal and parietal lobes "Shallice\
0871#\ is a}ected in eating disorder patients "e[g[ Green
et al[\ 0885#\ but do also agree with the assumption that
primarily the attentional demands are de_cient in these
patients "e[g[\ Hamsher et al[\ 0870^ Pendleton!Jones et
al[\ 0880^ present investigation#[
Finally\ one has to consider several neurochemical sub!
stances "e[g[\ glucocorticoids and sex hormones# as the
respective mediating factors[ For example\ steroid hor!
mones are known to be speci_cally altered in the acute
eating disorder but normalize with clinical remission "see]
Ploog and Pirke\ 0876^ Pirke et al[\ 0877^ Schweiger\
0880#[ In addition\ these steroid hormones obviously
in~uence human cognitive functions "see] Wolkowitz et
al[\ 0889\ 0882^ Cohen and Pfa}\ 0881^ Squire\ 0881^
Rubinow and Schmidt\ 0885^ Sapolsky\ 0885^ Sherwin\
0885#[ Regarding eating disorders\ we are aware of only
one study in which steroid hormone concentrations "cor!
tisol# and cognitive abilities were evaluated in parallel
"Laessle et al[\ 0881#[ And indeed\ patients with elevated
cortisol levels performed poorer on a vigilance task with
a high load of information processing than patients with
normal cortisol concentrations[ Thus\ although direct
evidence is scarce at present\ there are several potential
candidates that may serve as a factor mediating between
eating disorder symptomatology and neuropsychological
de_cits[
In conclusion\ the present prospective study provides
evidence that anorexia nervosa and bulimia nervosa are
characterized by similar neuropsychological de_cits
mainly of attentional nature which improve with clinical
remission[ Although the amelioration of both the cog!
nitive de_cits and the clinical symptomatology had run
in parallel\ the present _ndings indicate that these pro!
cesses are not directly linked to each other\ raising the
possibility of a third and mediating factor[ We dem!
onstrated that unspeci_c stress due to hospital admission
is unlikely to be such a factor[ Furthermore\ the severity
of secondary psychopathology such as depression also
does not account for the neuropsychological de_cits[
More promising candidates for a signi_cant modulating
factor appear to be the metabolic turnover in certain
brain areas as well as the level of steroid hormones[ How!
ever\ these suggestions are speculative and await testing in
future and prospective studies on patients with anorexia
nervosa and bulimia nervosa[
References
American Psychiatric Association[ Diagnostic and statistical manual
of mental disorders\ 2rd ed[ revised[ Washington\ DC] American
Psychiatric Association 0876[
Baddeley AD[ Working memory[ Science 0881^144]4458[
Beatty WW\ Wonderlich SA\ Staton RD\ Ternes LA[ Cognitive func!
tioning in bulimia] comparison with depression[ Bulletin of the
Psychonomic Society 0889^17]17881[
Ben!Tovim DI\ Walker MK\ Fok D\ Yap E[ An adaptation of the
Stroop Test for measuring shape and food concerns in eating dis!
orders] a quantitative measure of psychopathology< International
Journal of Eating Disorders 0878^7]5706[
Brickenkamp R[ Der Test d1\ 6[ Au~age[ Go ttingen] Hogrefe\ 0867[
C.1. 1aucr ci al. , 1ournal of Psychiairic Tcscarch 33 (1999) 129138 027
Cohen RS\ Pfa} DW[ Ventromedial hypothalamic neurons in the
mediation of long!lasting e}ects of estrogen on lordosis behavior[
Progress in Neurobiology 0881^27]31242[
Cooper MJ\ Fairburn CG[ Selective processing of eating\ weight and
shape related words in patients with eating disorders and dieters[
British Journal of Clinical Psychology 0881^20]2524[
Cooper MJ\ Fairburn CG[ Demographic and clinical correlates of
selective information processing in patients with bulimia nervosa[
International Journal of Eating Disorders 0882^02]09805[
Delvenne V\ Goldman S\ Simon Y\ De Maertelaer V\ Lotstra F[ Brain
hypometabolism of glucose in bulimia nervosa[ International Jour!
nal of Eating Disorders 0886^10]20219[
Delvenne V\ Goldman S\ De Maertelaer V\ Simon Y\ Luxen A\ Lotstra
F[ Brain hypometabolism of glucose in anorexia nervosa] nor!
malization after weight gain[ Biological Psychiatry 0885^39]6507[
Derogatis LR\ Lipman RS\ Covi L[ SCL!89[ Self!report symptom
inventory[ In] Guy W\ editors[ ECDEU assessment manual of psy!
chopharmacology[ Rockville\ MD] National Institute of Mental
Health\ 0865[ p[ 202220[
Fairburn CG\ Cooper PJ\ Cooper MJ\ McKenna FP\ Anastasiades P[
Selective information processing in bulimia nervosa[ International
Journal of Eating Disorders 0880^09]30411[
Gerlingho} M\ Backmund H[ Therapie der Magersucht und Bulimie[
Weinheim] Beltz Verlag\ 0884[
Green MW\ Elliman NA\ Wakeling A\ Rogers PJ[ Cognitive func!
tioning\ weight change and therapy in anorexia nervosa[ Journal of
Psychiatric Research 0885^29]39009[
Hamsher KS\ Halmi KA\ Benton AL[ Prediction of outcome in anor!
exia nervosa from neuropsychological status[ Psychiatry Research
0870^3]6877[
Hautzinger M\ Bailer J\ Worall H\ Keller F[ Das Beck Depression!
sinventar*BDI[ Mu nchen] Huber\ 0881[
Herholz K[ Neuroimaging in anorexia nervosa[ Psychiatry Research
0885^51]09409[
Herholz K\ Krieg JC\ Emrich HM\ Pawlik G\ Beil C\ Pirke KM\
Wagner R\ Wienhard K\ Ploog D\ Heiss WD[ Regional cerebral
glucose metabolismin anorexia nervosa measured by positron emis!
sion tomography[ Biological Psychiatry 0876^11]3240[
Ilmberger J[ Deutsche Version des California Verbal Learning Test[
Mu nchen] Institut fu r Medizinische Psychologie\ Universitat Mu n!
chen\ 0877[
Keefe RSE[ The contribution of neuropsychology to psychiatry[ Amer!
ican Journal of Psychiatry 0884^041^504[
Kingston K\ Szmukler G\ Andrewes D\ Tress B\ Desmond P[ Neuro!
psychological and structural brain changes in anorexia nervosa
before and after refeeding[ Psychological Medicine 0885^15]0417[
Krieg JC\ Lauer CJ\ Pirke KM[ Structural brain abnormalities in pat!
ients with bulimia nervosa[ Psychiatry Research 0878^16]2837[
Krieg JC\ Pirke KM\ Lauer CJ\ Backmund H[ Endocrine\ metabolic\
and cranial computed tomographic _ndings in anorexia nervosa[
Biological Psychiatry 0877^12]26676[
Kohlmeyer K\ Lehmkuhl G\ Poutska F[ Computed tomography of
anorexia nervosa[ American Journal of Neuroradiology 0872^3]326
7[
Laessle RG\ Krieg JC\ Fichter MM\ Pirke KM[ Cerebral atrophy and
vigilance performance in patients with anorexia nervosa and bulimia
nervosa[ Neuropsychobiology 0878^10]07680[
Laessle RG\ Fischer M\ Fichter MM\ Pirke KM\ Krieg JC[ Cortisol
levels and vigilance in eating disorder patients[ Psychoneuro!
endocrinology 0881^06]36473[
Long CG\ Hinton C\ Gillespie NK[ Selective processing of food and
body size words] Application of the Stroop Test with obese
restrained eaters\ anorexics\ and normals[ International Journal of
Eating Disorders 0883^04]16872[
Maxwell JK\ Tucker DM\ Townes BD[ Asymetric cognitive function
in anorexia nervosa[ International Journal of Neuroscience
0873^13]2633[
Milner B[ Interhemispheric di}erences in the localization of psycho!
logical processes in man[ British Medical Bulletin 0860^16]1616[
Oswald W\ Roth E[ Der Zahlenverbindungstest[ Go ttingen] Hogrefe\
0867[
Palazidou E\ Robinson P\ Lishman WA[ Neuroradiological and neu!
ropsychological assessment in anorexia nervosa[ Psychological
Medicine 0889^19]4106[
Pendleton!Jones B\ Duncan CC\ Brouwers P\ Mirsky AF[ Cognition in
eating disorders[ Journal of Clinical and Experimental Neu!
ropsychology 0880^02]60017[
Perpina C\ Hemsley D\ Treasure J\ de Silva P[ Is the selective infor!
mation processing of food and body words speci_c to patients
with eating disorders< International Journal of Eating Disorders
0882^03]24855[
Pirke KM\ Vandereycken W\ Ploog D[ The Psychobiology of Bulimia
Nervosa[ Berlin] Springer Verlag\ 0877[
Ploog DW\ Pirke KM[ The psychobiology of anorexia nervosa[ Psycho!
logical Medicine 0876^06]73248[
Raven JC\ Court JH\ Raven JC[ Manual for Raven|s Progressive Matr!
ices and Vocabulary Scales[ London] Lewis\ 0866[
Rubinow DR\ Schmidt PJ[ Androgens\ brain\ and behavior[ American
Journal of Psychiatry 0885^042]86373[
Sapolsky RM[ Stress\ glucocorticoids\ and damage to the nervous sys!
tem] The current state of confusion[ Stress 0881^0]008[
Schweiger U[ Menstrual function and luteal!phase de_ciency in relation
to weight changes and dieting[ Clinical Obstretrics and Gynecology
0880^23]0806[
Shallice T[ Speci_c impairments of planning[ Philosphical Transactions
of the Royal Society London B 0871^187]088198[
Sherwin BB[ Hormones\ mood\ and cognitive functioning in post!
menopausal women[ Obstretrics + Gynecology 0885^76]19S15S[
Small A\ Madero J\ Teagno L\ Ebert M[ Intellect\ perceptual charac!
teristics and weight gain in anorexia nervosa[ Journal of Clinical
Psychology 0872^28]6791[
Squire LR[ Memory and the hippocampus] A synthesis from _ndings
with rats\ monkeys\ and humans[ Psychological Reviews
0881^1]084120[
Strupp BJ\ Weingartner H\ Kaye W\ Gwirtsman H[ Cognitive pro!
cessing in anorexia nervosa] Adisturbance in automatic information
processing[ Neuropsychobiology 0875^04]7883[
Szmukler GI\ Andrewes D\ Kingston K\ Chen L\ Stargatt R\ Stanley
R[ Neuropsychological impairment in anorexia nervosa] Before and
after refeeding[ Journal of Clinical and Experimental Neuro!
psychology 0881^03]136241[
Thiel A\ Paul T[ Entwicklung einer deutschsprachigen Version des
Eating!Disorder!Inventory "EDI#[ Zeitschrift fu r Di}erentielle und
Diagnostische Psychologie 0877^8]15667[
Touyz SW\ Beumont PJV\ Johnstone LC[ Neuropsychological cor!
relates of dieting disorders[ International Journal of Eating Dis!
orders 0875^4]091423[
Von Cramon DY[ Planen und Handeln[ In] von Cramon DY\ Zihl J\
editors[ Neuropsychologische Rehabilitation[ Berlin] Springer
Verlag\ 0877[ p[ 13752[
Wechsler D[ Wechsler Memory Scale Revised[ New York] The Psycho!
logical Corporation\ 0876[
Wolkowitz OM\ Reus VI\ Weingartner H\ Thompson K\ Breier A\
Doran A\ RubinowD\ Pickar D[ Cognitive e}ects of corticosteroids[
American Journal of Psychiatry 0889^036]01860292[
Wolkowitz OM\ Weingartner H\ Rubinow DR\ Jimerson D\ Kling M\
Berretini W\ Thompson K\ Breier A\ Doran A\ Reus VI\ Pickar
D[ Steroid modulation of human memory] Biochemical correlates[
Biological Psychiatry 0882^22]6335[
Zimmermann P\ Fimm B[ Testbatterie zur Aufmerksamkeitspru fung
"TAP#\ Version 0[91c[ Wu rselen] Psytest\ 0883[