Copyright Blackwell Munksgaard 2007

Bipolar Disorders 2007: 9: 114125

BIPOLAR DISORDERS

Original Article

Neuropsychological decits and functional

impairment in bipolar depression, hypomania
and euthymia
Malhi GS, Ivanovski B, Hadzi-Pavlovic D, Mitchell PB, Vieta E,
Sachdev P. Neuropsychological decits and functional impairment in
bipolar depression, hypomania and euthymia.
Bipolar Disord 2007: 9: 114125. Blackwell Munksgaard, 2007
Objective: To examine whether patients with bipolar disorder (BD)
have subtle neuropsychological decits that manifest clinically as
cognitive and functional compromise, and this study attempted to
determine the pattern of such cognitive decits and their functional
impact across all three phases of BD. We hypothesised that euthymia
does not equate with normal neuropsychological function and that each
phase has a characteristic pattern of decits, with disturbance in
attention and memory being common across all phases of the illness: (i)
bipolar depression psychomotor slowing and impairment of memory;
(ii) hypomania by frontal-executive decits and (iii) euthymia a mild
disturbance of attention, memory and executive function.

Gin S Malhia,b,c, Belinda

Ivanovskib,d, Dusan HadziPavlovicb,d Philip B Mitchellb,d,
Eduard Vietae and Perminder
Sachdevd,f
a

Northern Clinical School, University of Sydney,

Mood Disorders Unit, Black Dog Institute, Prince
of Wales Hospital, cMayne Clinical Research
Imaging Centre, Prince of Wales Medical Research
Institute, dSchool of Psychiatry, University of New
South Wales, Sydney, Australia, eBipolar Disorders
Program, Hospital Clinic, University of Barcelona,
IDIBAPS, Stanley Research Medical Centre,
Barcelona, Spain, fNeuropsychiatric Institute,
Prince of Wales Hospital, Sydney, Australia

Methods: Twenty-ve patients with a diagnosis of bipolar I disorder

underwent neuropsychological testing over a period of 30 months in the
natural course of their illness while hypomanic and/or depressed and/or
euthymic. The results from these assessments were compared with
ndings from neuropsychological tests conducted on 25 healthy controls
matched for age, sex, education and handedness.
Results: Initial analyses revealed modest impairment in executive
functioning, memory and attention in both hypomanic and depressed
bipolar patients, with additional ne motor skills impairment in the
latter. Memory decits, also noted in euthymic patients, were nonsignicant after controlling for confounding variables, although bipolar
depressed patients remained signicantly impaired on tests of verbal
recall. Bipolar depressed and hypomanic patients diered with respect to
the nature of their memory impairment. Depressed patients were more
impaired as compared with euthymic patients on tests of verbal recall
and ne motor skills. Psychosocial functioning was impaired across all
three patient groups, but only in depressed and hypomanic patients did
this correlate signicantly with neuropsychological performance.
Conclusions: The mood-state-related cognitive decits in both bipolar
depression and hypomania compromise psychosocial function when
patients are unwell. In euthymic patients, subtle impairments in attention
and memory suggest that an absence of symptoms does not necessarily
equate to recovery. The possibility of persistent cognitive decits in BD
is an issue of profound clinical and research interest that warrants further
investigation; however, future research needs to adopt more
sophisticated neuropsychological probes that are able to better dene
state and trait decits and determine their functional impact.

Key words: bipolar disorder depression

euthymia mania neuropsychology
Received 17 January 2005, revised and accepted
for publication 2 February 2006
Corresponding author: Gins S Malhi, Academic
Discipline of Psychological Medicine, Northern
Clinical School, University of Sydney, Sydney,
NSW, Australia
e-mail: ginmalhi@gmail.com

The authors of this paper do not have any commercial or other conict of interest with respect to this article.

114

Neuropsychological dysfunction in bipolar disorder

Until recently, bipolar disorder (BD) has been
regarded as an episodic illness with the assumption that between episodes of depression and
mania, patients resume normality of mood and
make a complete functional recovery. Consequently, periods of illness and accompanying
impairments have been generally viewed as wholly
reversible. However, it is increasingly apparent that
this view is somewhat inaccurate and that euthymic
patients, although clinically in remission, continue
to experience functional compromise (14). The
precise nature and extent of this impairment is as
yet unclear and investigations including neuroimaging studies that have attempted in vivo to identify
its biological underpinnings have been less informative than initially hoped. Structural and functional state and even trait (5) abnormalities have
been reported, but a robust and reliable marker of
the illness is yet to emerge. Therefore, the burgeoning neuropsychological literature in BD is of
particular interest, as it implicates subtle neurocognitive decits across each phase of the illness (6,
7) such that even euthymia is neuropsychologically
scarred (1, 610).
The cognitive decits in BD involve attention,
mental exibility, decision making, verbal uency
and memory with some overlap between the two
poles (11, 12). The reported pattern of cognitive
impairment in bipolar depression also overlaps
with that in major depression especially with
respect to memory and executive functioning (13),
and the decits observed in acute mania are shared
across a number of psychoses (14, 15). Of particular interest, however, is the provocative suggestion that euthymic patients may also have subtle
cognitive decits (1618) irrespective of subsyndromal symptoms (19, 20).
The possibility of persistent neuropsychological
decits in BD is of profound importance both
because of its potential as a marker for the illness
and because of prognostic ramications for
patients. A number of recent studies have examined bipolar patients during an episode and when
euthymic, and compared them with healthy
controls. Thus far the ndings are inconclusive
principally because too few studies have been
conducted (21) for the many methodological
issues to be addressed. Many of these investigations have also been limited by cross-sectional
design, small sample sizes and the inclusion of
patients with mixed diagnoses (for example,
bipolar I and II). Confounds of medication and
substance misuse are also dicult to eliminate
(22). Clinically, psychosocial functioning is a
sophisticated construct that encompasses interactions and activities in personal, occupational and

recreational contexts. It is therefore not surprising that few studies have specically examined functionality and its neuropsychological
concomitants.
In this study, we have attempted to characterise
the cognitive prole of BD and examine its
contribution to functional impairment by conducting clinical and neuropsychological assessments
across all three phases of the illness. We hypothesised that bipolar patients would have greater
functional impairment than controls in all phases
of the illness and that this would be associated with
cognitive decits involving in particular attention,
memory and executive function. Specically, bipolar depression would be associated with psychomotor slowing and impairment of memory;
hypomania would be associated with frontal-executive decits; and euthymia would not equate to a
normal neuropsychological prole by virtue of a
mild disturbance of attention, memory and executive function.

Materials and methods

Subjects

Seventeen female and eight male patients aged

1959 years (mean 38.6 11.0 years) with DSMIV bipolar I disorder were recruited from our
Sydney BD clinic, providing a second opinion for
patients currently receiving treatment. Research
psychiatrists (GM, PM), using the Structured
Clinical Interview for DSM-IV (SCID-P) (23)
supplemented by a case note review, made the
diagnoses. Subjects were excluded if they had a
history of ongoing substance misuse, neurological
disease or closed head injury or had an additional
Axis-I or Axis-II DSM-IV diagnosis or a medical
disorder currently necessitating treatment.
Of the 25 bipolar patients assessed, 14 were
assessed during the depressive phase of their illness,
15 when euthymic and 12 when hypomanic. Of the
25 patients, 13 were assessed in one phase only
(7 in euthymia, 4 in depression and 2 in hypomania). Of the remaining 12 patients, 4 were assessed
in all three phases and 8 in two phases of BD (4 in
hypomania and depression, 2 in hypomania and
euthymia and 2 in depression and euthymia).
Patients were not assessed in the same phase of
their illness more than once and several depressed
or hypomanic patients were unable to undergo
more than one assessment because of changes to
their medication, hospitalization or non-response
to treatment. It is important to note that this
necessitates analysis of groups that clearly involve
overlap of subjects.

115

Malhi et al.

Patients were compared with healthy volunteers

recruited via advertisement and matched for age
(2159 years; mean 34.0 13.0 years), handedness (all right handed) and education (1119 years;
mean 14.8 2.1 years). Comparison subjects were
screened for a history of neurological or psychiatric disorder (with SCID-non-patient version) or
for a family history of the same. They were
administered the same clinical and neuropsychological assessments and self-report questionnaires
as patients.
The Prince of Wales Hospital and University of
New South Wales ethics committees approved the
study and all participants provided written
informed consent.
Clinical and neuropsychological assessment

Patients were instructed to monitor their symptoms

and notify researchers of any signicant mood
change. The majority was able to reliably anticipate
their episodes of illness. In addition, researchers
maintained telephone contact with patients and
monitored their mental state throughout the duration of the 30-month study. Assessment involved a
structured clinical interview conducted by a research
psychiatrist (GM) and a series of neuropsychological tests conducted by a research psychologist (BI).
Clinical assessment. Illness severity was assessed
using the 21-item Hamilton Depression Rating
Scale (HAMD) (24), the Montgomery-Asberg
Depression Rating Scale (MADRS) (25), the
CORE measure for psychomotor disturbance (26)
and, if currently manic or hypomanic, the Young

Mania Rating Scale (YMRS) (27). Patients were

characterised as euthymic if they had been in
remission for at least 1 month and scored 9 on
the HAMD and <10 on the YMRS; depressed if
they scored >16 on the HAMD; and hypomanic if
they had a score of >10 on the YMRS.
Psychosocial functioning was assessed using the
Global Assessment of Functioning (GAF) (28)
rating the month prior to assessment. Patients also
completed the following questionnaires: the Beck
Depression Inventory (BDI) (29), the Spielberger
State-Trait-Anxiety Inventory (30), the Eysenck
Personality Inventory (EPI) (31) and the Annett
Handedness Inventory (32) (the latter two were
only administered on the rst assessment).
Neuropsychological assessment. All subjects were
administered a battery of neuropsychological tests
tailored to assess attention, working memory,
learning and memory, executive functioning and
psychomotor speed. To minimise practice eects,
repeat assessments used alternate forms of Digit
Span Backwards, Trail Making Test (TMT), Controlled Oral Word Association Test (COWAT),
Rey Auditory Verbal Learning Test (RAVLT) and
Symbol Digit Modalities Test (SDMT). A summary of these tests is provided in Table 1.
Statistical analyses

Two-tailed paired t tests were conducted to gain

maximal advantage of high subject correlation.
Dependent variables were grouped into cognitive
domains (executive functioning, memory, attention and psychomotor abilities) and variables that

Table 1. Neuropsychological test battery

Estimated premorbid IQ
National Adult Reading Test (NART; 46)

Frontal executive functions

Wisconsin Card Sorting Test (WCST; 43)

Stockings of Cambridge (SOC; 41)

116

Description

Dependent variables

This measure estimates premorbid

intelligence using a list of words,
arranged in order of difficulty.
Pronunciation and knowledge of words
is a reliable measure of both verbal and
general mental abilities (44, 45).

1. Estimated WAIS-III (39) Full Scale IQ

score, derived from NART score.

This computerized version assesses

concept formation and cognitive
flexibility by measuring the ability to
establish, maintain and change set.

1. Total number of categories

completed.
2. Trials to complete first category.
3. Failure to maintain set.
4. Percentage perseverative errors.
1. Average number of moves taken to
solve four move problems.
2. Average number of moves taken to
solve five move problems.
3. Total number of problems solved in
minimum moves.

This test, a component of the Cambridge

Neuropsychological Test Automated
Battery (CANTAB), is a modified
version of the Tower of London task and
assesses spatial planning and motor
control.

Neuropsychological dysfunction in bipolar disorder

Table 1. Continued

Trail Making Test (TMT; 44): Part B

Controlled Oral Word Association Test

(COWAT; 40)

Animal Naming Test (47)

Description

Dependent variables

This test in two parts measures attention,

psychomotor speed and the ability to
shift strategy. Part B assesses set
switching and visuospatial working
memory.
This test which requires the strategy
generation assesses verbal fluency and
is a sensitive measure of phonemic
fluency.
A test of category verbal fluency.

1. Time taken to complete Part B.

1. Total number of words generated.

2. Total number of rule violations.
3. Total number of repetition errors.
1. Total number of words generated.
2. Total number of rule violations.
3. Total number of repetition errors.

Attention, concentration and mental tracking

Stroop Colour-Word Interference Test
(SCWT; 42)

This assesses mainly selective attention 1. Age-adjusted scores of the number of

items read on the word-only trial.
and response inhibition. In particular,
Part 3 of the test generates interference 2. Age-adjusted scores of the number of
items read on the colour-only trial.
and measures the ability to change
3. Age-adjusted scores of the number of
demands and suppress a habitual
items read on the word/colour trial.
response in favour of an unusual one.
4. Interference score.
Trail Making Test (TMT; 44): Part A
This test in two parts measures attention, 1. Time taken to complete Part A.
psychomotor speed and the ability to
shift strategy. Part A assesses focused
attention.
Digit span from the Wechsler Adult
This test assesses auditory attention and 1. Longest digit span backwards.
Intelligence Test Third edition (WAIS-III; 39) short-term retention capacity and the
ability to manipulate information in
verbal working memory.
1. Total number correct.
Symbol Digit Modalities Test (SDMT; 36)
This test requires mental and motor
speed and mental flexibility. It also taps
concentration and visual perception
and scanning.
Verbal learning and memory
1. Recognition score.
Rey Auditory Verbal Learning Test
The RAVLT assesses immediate and
2. Learning slope.
(RAVLT; 35, 37)
delayed-memory span, new learning
3. Total number of words recalled.
and susceptibility to interference and
4. Interference.
recognition memory.
5. Retroactive interference.
6. Delayed recall.
Psychomotor function
Reaction Time (RT) (41)
This assesses simple reaction time,
1. Mean reaction time.
which involves stimulus detection and 2. Mean movement time.
response execution.
Purdue Pegboard (34)
This test assesses finger and hand
1. Total number of pegs placed
dexterity.
correctly, dominant hand.
2. Total number of pegs placed
correctly, non-dominant hand.
3. Total number of pegs placed
correctly, both hands.
1. Time taken to complete using
Grooved Pegboard (33, 38)
This test is also a measure of
dominant hand.
manipulative dexterity; however, the
2. Time taken to complete using nontask requires more sophisticated
dominant hand.
manipulation than that needed in the
Purdue pegboard task and hence
requires more complex visuomotor
coordination.

were highly correlated with each other, and all

other variables within each domain were identied
and removed as they appeared to measure similar
constructs and were thus deemed redundant. This

procedure was performed to reduce the total

number of variables within each domain and hence
enhance statistical power. Potential confounds
such as residual aective symptoms, anxiety and

117

Malhi et al.

neuroticism controlled for by between-group and

within-patient analyses of covariance (ANCOVA)
were severely limited by small sample size. As a
consequence of the latter (small n), p values have
been reported for all comparisons and critical
values were not adjusted for multiple comparisons.
Correlations between GAF and neuropsychological test scores were also performed.

Results

Sample characteristics

Demographics of patients and controls are summarised in Table 2. The mean age of onset of
depression was 24 8.4 years and for mania
25.2 8.6 years. The mean duration of illness
from diagnosis was 15.2 10.8 years and the
mean number of previous depressive episodes was
11.5 8.2 and manic episodes 8.2 7.2. Of the
25 patients, 11 fullled DSM-IV criteria for rapid
cycling BD.
At the time of clinical assessment and neuropsychological testing, 21 patients were on mood
stabilizing psychotropic medications only, with
dosages unaltered over the preceding week. Twelve
patients were taking lithium (mean daily dose
975.1 213.2 mg) with a mean plasma level of
0.77 0.1 mmol/L. Four of these patients and
seven others were taking valproate (mean daily
dose 1436.0 590.4 mg), with a mean plasma
level of 512.5 89.2 lmol/L. Two patients
were taking carbamazepine (mean daily dose
850 636.4 mg), with a mean plasma level of
30.0 14.1 lg/mL, one in combination with
valproate. The remaining ve patients were unmedicated. Importantly, they had been free of
neuroleptic or antidepressant medication for
2 weeks. Sixteen patients had at least one rstdegree relative with an aective disorder (5 BD, 10
unipolar depression and 1 both). One patient had a

family history of schizophrenia and seven patients

had no family history of psychiatric illnesses. Eight
patients had a past history of regular substance
misuse including alcohol, amphetamines and
cannabis.
Clinical assessments

Results from clinical and self-report measures are

summarised in Table 3. Predictably bipolar
patients and control subjects diered signicantly
on clinical scale scores. However, of particular note
is the signicant dierence between scores on the
HAMD (t14 3.57, p 0.003), MADRS (t14
4.93, p 0.000) and BDI (t13 2.82, p 0.01),
between euthymic bipolar patients (means 3.47,
4.33, 8.21, respectively) and between healthy
comparison subjects (means 1.43, 1.22, 3.22,
respectively), indicating residual depressive symptomatology in patients. Relative to controls, hypomanic patients also scored higher on the measures
of depression (HAMD: t11 3.5, p 0.01;
MADRS: t11 2.79, p 0.02; BDI: t10 2.51,
p 0.03), suggestive perhaps of mild mixed mood
states in some patients. Depressed patients scored
signicantly higher on the CORE, reecting greater
psychomotor retardation than matched controls
(t12 8.74, p 0.00); however, euthymic and
hypomanic patients did not dier signicantly
from controls on this measure, although the
dierence for the hypomanic patients did approach
signicance (t10 2.14, p 0.06). In addition, of
note are the signicantly greater state and trait
anxiety levels in the depressed and euthymic
patients (depressed: t8 3.67 and p 0.01, t9
4.06 and p 0.00; euthymic: t13 2.79 and p
0.02, t13 5.42 p 0.00) but not in the hypomanic patients as compared with controls. Finally,
patients scored higher on the Neuroticism scale
of the EPI than controls, irrespective of mood
state (euthymic: t13 7.21, p 0.00; depressed:

Table 2. Demographic variables for bipolar and control groups

Variable

Bipolar group, mean (SD)

Control group, mean (SD)

Age
Education (years)
IQ score (derived from NART score)
Age of onset of depression
Age of onset of mania
Duration of illness from diagnosis (years)
Number of depressive episodes
Number of manic episodes

38.6
14.7
113.8
24
25.2
15.2
11.5
8.2

34.0 (13.0)
14.8 (2.1)
115.4 (9.52)

(11.0)
(2.9)
(7.14)
(8.4)
(8.6)
(10.8)
(8.2)
(7.2)

Mean scores are drawn from the t-test analyses and hence may not include all subjects within each group because of missing data.
Mean scores for the control group are not drawn from the t-test analyses (as the n varies in each analysis) and hence reflect total mean
scores for this group (n 25). NART National Adult Reading Test.

118

Neuropsychological dysfunction in bipolar disorder

Table 3. Clinical variables for bipolar and control groups

Hamilton Depression Rating Scale (HAMD)

Montgomery-Asberg Depression Rating Scale
Beck Depression Inventory (BDI)
Young Mania Rating Scale (YMRS)
CORE
Global Assessment of Functioning (GAF)
State-Trait Anxiety Inventory (State)
State-Trait Anxiety Inventory (Trait)
Extrovert Scale (EPI)
Neuroticism Scale (EPI)
Lie Scale (EPI)

Depressed (D)
Mean (SD)

Euthymic (E)
Mean (SD)

23.08
28.25
27.33

7.85
52.5
53.11
57.2
9.7
17.8
3.9

3.47
4.33
8.21

0.13
86.13
36.43
44.5
11.29
12.86
3.64

(4.25)
(4.81)
(14.99)
(3.24)g
(7.31)h,i
(15.80)k
(15.22)m
(4.72)
(4.211)o
(0.88)

(2.29)a
(2.44)c
(5.81)e
(0.52)
(6.96)
(7.06)l
(8.78)n
(4.83)
(3.94)p
(1.55)

Hypomanic (H)
Mean (SD)
5.5
4.25
8.18
20.67
1.09
66.71
41.55
41.55
11.91
14.82
3.82

(2.88)b
(2.49)d
(4.92)f
(3.52)
(1.45)
(13.34)j
(14.34)
(15.63)
(8.24)
(5.71)q
(2.27)

Control (C)
Mean* (SD)
1.43
1.22
3.22

0.04
92.26
29.13
31.52
12.65
6.30
3.65

(1.95)
(1.98)
(2.70)
(0.2)
(2.92)
(8.17)
(6.43)
(3.6)
(5.03)
(1.72)

*Mean scores for the control group are not drawn from the t-test analyses (as the n varies in each analysis) and hence reflect total mean
scores for this group (n 25). EPI Eysenck Personality Inventory.
a
E versus C: t14 3.57, p 0.003; bH versus C: t11 3.5, p 0.01; cE versus C: t14 4.93, p 0.000; dH versus C: t11 2.79, p
0.02; eE versus C: t13 2.82, p 0.01; fH versus C: t10 2.51, p 0.030; gD versus C: t12 8.74, p 0.00; hD versus E: t17
11.39, p 0.00; iD versus H: t14 2.86, p 0.01; jH versus E: t15 0.04, p 0.00; kD versus C: t8 3.67, p 0.01; lE versus C:
t13 2.79, p 0.02; mD versus C: t9 4.06, p 0.00; nE versus C: t13 5.42, p 0.00; oD versus C: t9 3.49, p 0.01; pE versus C:
t13 7.31, p 0.00; qH versus C: t10 2.40, p 0.04.

t9 3.49, p 0.01; hypomanic: t10 2.40, p

0.04), reecting personality dierences between the
groups although patients did not formally meet the
criteria for personality disorders. However, it is
important to note that confounds of medication
and subsyndromal symptoms, in all three patient
groups, may also have inuenced personality
assessment.
Global Assessment of Functioning scores for
all three groups of patients (depressed
52.5 7.31; hypomanic 66.71 13.34; euthymic
86.13 6.96) were lower in comparison to
controls (92.26 2.92), and compared with
euthymic patients, scores for both depressed
(t17 11.39, p 0.00) and hypomanic (t15
0.04, p 0.00) patients were signicantly lower.
Furthermore, GAF scores of depressed patients
were signicantly lower than the scores of hypomanic patients (t14 2.86, p 0.01). Signicant
correlations between GAF scores and neuropsychological test performance are summarised in
Table 4.

In bipolar depressed patients poor psychosocial

function was associated with poor performance on
the Stroop word-reading task and slower reaction
times, whereas in hypomanic patients poor psychosocial function was associated with poor working memory and poor new learning. In euthymic
patients, there were no signicant correlations
between GAF scores and neuropsychological test
performance, but when the GAF scores for all
bipolar patients were pooled together, poor psychosocial and executive function were signicantly
correlated on the Wisconsin Card Sorting Test,
such that patients with lower GAF scores found it
more dicult to maintain set. However, inference
of causality from such correlations is limited,
especially as many neuropsychological tests scores
are highly correlated.
Neuropsychological assessments

Statistically signicant results from patient versus

control and within-patient group comparisons are

Table 4. Significant correlations between Global Assessment of Functioning (GAF) scores and neuropsychological tests in patients
Neuropsychological
measures

Depressed group
GAF scores (N 10)

0.74 (p 0.015)
Stroop Taska
Reaction Time
)0.72 (p 0.019)
Digit Span backwards

Trail Making Test, Part B

RAVLTlearning

WCSTb

Euthymic group
GAF scores (N 10)

Hypomanic group
GAF scores (N 8)

0.77 (N 8; p 0.027)
)0.80 (N 7; p 0.033)
0.81 (N 8; p 0.016)

Bipolar group (mixed states)

GAF scores (N 28)

)0.42 (N 25; p 0.037)

RAVLT Rey Auditory Verbal Learning Test; WCST Wisconsin Card Sorting Test.
a
Word-only task.
b
Failure to maintain set.

119

Malhi et al.
Table 5. Significant comparisons between patient (depressed, euthymic and hypomanic) groups and the control group
Cognitive domain

Test

Depressed patients

Executive (14 DVs)

COWAT
Animal Naming Test
RAVLT
SDMT
Stroop Task
Trail Making Test, Part A
Purdue Pegboard
Grooved Pegboard

Memory (6 DVs)
Attention (7 DVs)

Motor (9 DVs)

f
g

Euthymic patients

Hypomanic patients
b
e

h
i

DVs dependent variables; COWAT Controlled Oral Word Association Test; SDMT Symbol Digits Modalities Test; RAVLT Rey
Auditory Verbal Learning Test.
a
Impaired ability to generate phonemic words (t12 2.26, p 0.04).
b
Impaired ability to generate semantic words (t11 2.35, p 0.04).
c
Impaired total recall (t12 2.37, p 0.04) and recognition (t12 2.63, p 0.02).
d
Impaired total recall (t14 2.11, p 0.05).
e
Impaired recognition (t11 3.58, p 0.004) and recall (t11 4.84, p 0.001).
f
Completed fewer items (t12 2.90, p 0.01).
g
Read fewer items on the Word Task (t12 2.75, p 0.02), Colour Task (t12 2.33, p 0.04).
h
Slower (t11 2.40, p 0.04).
i
Slower, non-dominant hand (t12 3.37, p 0.01).
j
Slower, dominant hand (t12 2.39, p 0.03).

summarised in Tables 5 and 6, respectively. In

comparison to controls, the majority of dierences
occurred in depressed patients who were found to
be signicantly impaired across all cognitive
domains. They were limited in their ability to
generate phonemic words on the COWAT that
assessed verbal uency (t12 2.26, p 0.04); had
impaired recall (t12 2.37, p 0.04) and recognition (t12 2.63, p 0.02) memory on the
RAVLT; had diminished attention by virtue of
completing and reading fewer items on the SDMT
(t12 2.90, p 0.01) and Stroop Word and Colour Tasks (t12 2.75, p 0.02; t12 2.33, p
0.04) and were slower on motor tasks as assessed
by the Purdue (non-dominant hand, t12 3.37,
p 0.01) and Grooved (dominant hand, t12
2.39, p 0.03) pegboards.
Like depressed patients, hypomanic patients
were impaired on both recognition (t11 3.58,
p 0.004) and recall (t11 4.84, p 0.001) components of the RAVLT, while euthymic patients
were only impaired on immediate recall (t14 2.11,
p 0.05). Compared with controls, hypomanic
patients performed less well when generating
semantic words in the COWAT and were slower
(t11 2.40, p 0.04) in performing Trails A of the
TMT.
Within-patient group comparisons of participants tested in more than one mood state revealed
that recall memory assessed using the RAVLT was
more impaired in depression than in euthymia
(t4 3.64, p 0.02). Compared with hypomanic
patients, on this test depressed patients had greater
retroactive interference (t6 5.46, p 0.002)
although hypomanic patients had poorer

120

long-term memory recall/retention (t6 3.38,

p 0.015). However, as compared with euthymic
patients, both depressed (t4 3.21, p 0.03) and
hypomanic (t5 2.81, p 0.04) patients were
slower.
Within-patient and between-group ANCOVAs
that controlled potential confounds such as
anxiety, neuroticism and residual aective symptoms were constrained by small sample sizes and
the majority of signicant dierences were lost.
In depressed patients, only one memory decit
that was no longer signicant in euthymic and
hypomanic
patients
remained
signicant:
depressed subjects (t5 2.75, p 0.04) recalled
fewer words in total than control subjects. In
hypomanic patients relative to controls, no
dierences remained signicant, although dierences in recall ability (t7 1.88, p 0.096) and
attention (Trail A: t7 2.09, p 0.07) did
approach signicance. In euthymic patients, controlling for depression, anxiety and neuroticism
resulted in a signicant dierence in completing
Part A of the TMT (t9 3.56, p < 0.01). However, it is important to note that these results are
limited by the small numbers of subjects in each
of these comparisons.
Discussion

Functional localisation based purely on neuropsychological impairment is potentially problematic as

there are few strict correlations. In most cases,
neuropsychological decits imply network dysfunctions that may or may not be regionally
localised. However, broad inferences can be made

Neuropsychological dysfunction in bipolar disorder

Table 6. Significant within-patient group comparisons: depressed and hypomanic compared with euthymic

Cognitive domain

Test

Memory
Motor

RAVLT
Purdue Pegboard
Reaction Time

Depressed patients (DPs)

versus euthymic patients

Hypomanic patients (HPs)

versus euthymic patients

a
c

Depressed patients
versus hypomanic patients
b

RAVLT Rey Auditory Verbal Learning Test.

a
DPs demonstrated impaired total recall (t4 3.64, p 0.02).
b
DPs displayed greater retroactive interference than the HPs (t6 5.46, p 0.002) and the HPs had poorer long-term recall/retention
than the DPs (t6 3.38, p 0.015).
c
DPs were slower when using their dominant hand (t4 3.21, p 0.03).
d
HPs were slower (t5 2.81, p 0.04).

with reasonable condence of brain regions most

likely implicated.
Verbal fluency

Initial analyses from our study revealed evidence of

impairment in verbal uency in both depressed and
hypomanic patients but not in euthymic patients,
corroborating the ndings of an earlier study that
specically examined euthymic patients (48).
Therefore, frontal lobe dysfunction in BD may be
mood-state dependent; however, our subsequent
analyses revealed no decits in executive functioning in any of the mood states although patients in
both poles of illness have been shown to perform
poorly on decision making, problem solving and
concept formation tests (14, 49). The initial nding
of a decit in verbal uency is concordant with the
results of an early report that compared depressed
bipolar and unipolar patients (50). In this study,
depressed bipolar patients had a selective decit in
verbal uency, that of semantic categories (50), in
contrast to a more recent study in unipolar
depressed patients in which both semantic and
phonemic categories appear to be aected (51). It is
interesting to note that executive function decits
in euthymic patients have been reported even after
controlling subclinical depressive symptoms (1),
implicating dysfunction of organisational and
planning abilities (5254). A recent prospective
study that examined euthymic bipolar patients
demonstrated enduring neurocognitive impairment
across a broad range of domains (55).
Memory

Mnemonic decits are the most consistently

reported neuropsychological decits in BD (1, 4,
68), and it is therefore no surprise that all three
groups of patients in our study performed poorly
on tests of verbal immediate-recall and that
depressed and hypomanic patients also had

impaired recognition memory. Subsequent analyses that controlled confounds pared these ndings
and revealed that the memory decit was specically one of verbal recall and not recognition,
suggesting a dysfunction of information retrieval
as opposed to encoding. A previous study (56) also
reported poor verbal recall and intact recognition
memory in bipolar depression; however, in our
study because the decit only remained signicant
in the depressed group, it is most likely a staterelated decit rather than a bipolar trait. This is
also borne out to some extent by within-group
comparisons in which depressed patients were
more impaired on recall as tested by the RAVLT
than euthymic patients. In keeping with our nding
that dierences in the performance of memory
tasks between hypomanic patients and controls
approached statistical signicance, depressed and
hypomanic patients when compared directly differed in the nature of their memory decit, such
that bipolar depressed patients had greater retroactive interference, and hypomanic patients had
poorer long-term recall. Diculties in acquisition
and retention in both verbal and non-verbal
domains in hypomanic patients have been reported
previously (49, 57). Such memory and learning
decits implicate temporal lobe and in particular
hippocampal pathology, brain regions that along
with the ventromedial prefrontal cortex subserve
verbal uency (58). Persisting decits in these brain
regions may be responsible for the diculties in
social and occupational functioning reported by
bipolar patients (6, 7, 59, 60). However, the nature
of cognitive impairment in BD especially with
respect to memory and the question of whether
verbal memory decits are a state or trait marker
(1, 6, 7, 48, 61) awaits further clarication.
Attention

In initial analyses, depressed and hypomanic

patients in our study were impaired on tests of

121

Malhi et al.

attention when compared with controls; however,

subsequent analyses revealed no decits for either
of these groups, contradicting the many reports of
attention decits in BD (12). However, the literature reporting evidence of decits in attention has
largely employed tests of sustained attention,
whereas the tasks employed in our study were
measures of focused or selective attention. This
suggests that attention decits in bipolar depression or mania may be specic to sustained rather
than selective attention. Interestingly, the impairment in sustained attention in bipolar depressed
patients is greater than that found in unipolar
depression and is characterised by an inattentive
response pattern that results in errors of omission
(62). In contrast, patients with mania manifest an
impulsive response pattern that results in errors of
commission (63). It is likely that had our samples
been larger, the initial state-related decits may
have remained. This also explains to some extent
the unusual ndings in our euthymic group in
which subsequent analyses demonstrated a significant dierence in focused attention on the TMT
part A as compared with controls. It is, however,
noteworthy that initial analyses almost attained
signicance (p 0.07) and failure to do so may
reect not only methodological limitations but also
the eects of subsyndromal mixed aective symptoms.
Motor function

Motor function was assessed in our study because

of the greater preponderance of melancholia in
BD (64) and the obvious heightened energy
displayed by patients when manic (2). As anticipated, depressed patients performed poorly on
motor tasks as compared with controls, although
this dierence did not remain signicant in
subsequent analyses. However, within-patient analyses partly support a state-related association
as depressed patients also performed poorly in
comparison to euthymic patients and hypomanic
patients were slow to react in comparison to
euthymic patients.
Psychosocial function

In comparison to controls, psychosocial functioning in patients was poorer across all three phases of
the illness but especially in depression and hypomania. Correlating social function and neuropsychological performance in bipolar patients
implicated poor executive functioning; however,
when individual mood states were examined separately, social functioning in euthymic patients did

122

not correlate with any of the neuropsychological

measures. This is extremely important as it suggests that psychosocial impairment in euthymia is
not wholly attributable to cognitive impairment,
although this must be tempered against the fact
that the GAF is a composite assessment comprising symptoms and social adjustment and this may
also blur any association. It is also possible that the
neuropsychological tests administered in this and
many other studies simply fail to capture the subtle
neurocognitive impairment that patients experience in their day-to-day lives. In other words, the
assumption that cognitive testing relates to realworld functioning may well be erroneous (65). The
likelihood remains that patients with BD even
when well have some degree of functional compromise that may reect a persistent neuropsychological decit. Clinically this has important
implications not only for therapeutic intervention
but also for patients daily living. For instance,
patients with BD even when well and functioning
appropriately in terms of work often describe
diculties with decisions that require factoring of
future consequences. A concrete example of this is
the need to balance current expenditure with the
need to pay o debts such as mortgage. This is
important in the context of understanding the
illness and adhering to medication as patients often
stop treatment despite having had catastrophic
consequences in the past.
Such subtle compromise is necessarily dicult to
detect and determine in terms of its immediate and
long-term impact; however, assessing this in the
context of insight into the illness and its potential
consequences, given a number of scenarios, for
example complying with medication versus not
doing so, may provide an indicator as to whether
an individual has a decit.
Clearly, good non-specic bedside markers are
unsatisfactory and future research needs to focus
more on dening the extent and nature of any such
decit. This requires the development of new tests
that are designed to identify subtle decits and are
tailored to the individual. Longitudinal testing
is needed combining neuropsychological tests
with imaging and electrophysiological measures
so as to achieve greater regional and temporal
localisation.
Limitations

The nal neuropsychological data set had some

omissions largely because of patients being too
unwell to complete the entire neuropsychological
test battery or self-report measures. As a consequence, the number of subjects in each analysis

Neuropsychological dysfunction in bipolar disorder

diered according to the pattern of remaining data
and this limited statistical power. Small sample
sizes also prevented more sophisticated data
analysis and limited the statistical power available
to detect signicant dierences as well as the ability
to control for confounding factors in within-subject
analyses. This also limits attempts to co-vary for
other factors. A longitudinal design should have
countered these limitations to some extent by
improving within-subject comparisons and allowing serial assessments to determine whether changes were progressive. However, enrollment and
maintenance of subjects proved to be exceptionally
dicult and this also limited the numbers of
patients in our analyses. To overcome this problem
many previous studies have resorted to using
heterogenous patient samples, for instance, combining patients with bipolar and unipolar depression, including patients with psychoses, and
lumping together bipolar subtypes (66). In this
regard, our patient sample comprised only bipolar
I disorder patients; however, this diagnostic constraint also limited recruitment. Long periods of
observation to determine remission were not possible in many patients because of the frequency of
episodes and longitudinal design of the study. This
possibly increased the likelihood of residual
depressive or hypomanic symptomatology in euthymic patients. With respect to the latter, it is
important to mention that residual manic symptoms were not included in any of the analyses
because the YMRS was not administered to
patients when depressed or euthymic and therefore
the potentially confounding eects of these symptoms cannot be excluded. Furthermore, HAMD
and YMRS scores that demarcated episodes and
dened euthymia were somewhat higher than that
in some other studies because by design patients
were assessed when illness was incipient or when
they had just remitted so as to avoid the eects of
changes in medication.
Despite managing to avoid the acute eects of
antidepressants and antipsychotics, the potentially
confounding eects of mood stabilizers such as
lithium remained. This is important because withholding or withdrawing such treatment is ethically
unjustiable and yet, lithium appears to negatively
aect psychomotor speed and verbal memory (67).
Hence, this is an important limitation not only for
this study but also for the whole eld of BD
research.
Finally, the neuropsychological battery of tests
was determined on the basis of knowledge at the
time the study was commenced. Practical limitations of time, fatigue and potential practice eects
limited the number, sequencing and type of tests

that could be administered. Nevertheless, signicant dierences between the groups were demonstrated, even when controlling for confounding
factors. Detectable dierences, despite limited
power, underscore the need for future longitudinal
investigations.
Conclusion

To our knowledge, this is one of the rst studies to

attempt neuropsychological and psychosocial function assessment across a number of phases of BD
in a cohort of individuals. Despite some methodological limitations, we have identied a pattern of
modest cognitive impairment in both bipolar
depressed and manic patients, which correlates
with psychosocial functioning. Cognitively,
euthymic patients are largely intact although, they
too have poor psychosocial functioning and some
minimal but potentially signicant changes in
attention and memory that warrant further examination. Future studies should ideally recruit larger
sample sizes so that more complete within-subject
comparisons across dierent phases can be conducted, and perhaps attempt to investigate the
rst-onset bipolar patients to capture the emergence of neuropsychological decits.
Acknowledgements
We acknowledge support from the National Health & Medical
Research Council (Program Grant 222308), the New South
Wales Centre for Mental Health and the Australian Rotary
Health Research Fund.

References
1. Ferrier IN, Stanton BR, Kelly TP, Scott J. Neuropsychological function in euthymic patients with bipolar disorder.
Br J Psychiatry 1999; 175: 246251.
2. Goodwin F, Jamison K. Manic-Depressive Illness. New
York: Oxford University, 1990.
3. Chowdury R, Ferrier IN, Thompson JM. Cognitive
dysfunction in bipolar disorder. Curr Opin Psychiatry
2003; 16: 712.
4. Rubinsztein JS, Michael A, Paykel ES, Sahakian BJ.
Cognitive impairment in remission in bipolar aective
disorder. Psychol Med 2000; 30: 10251036.
5. Blumberg HP, Leung H-C, Skudlarski P et al. A functional
magnetic resonance imaging study of bipolar disorder:
state- and trait-related dysfunction in ventral prefrontal
cortices. Arch Gen Psychiatry 2003; 60: 601609.
6. Martinez-Aran A, Vieta E, Reinares M et al. Cognitive
function across manic or hypomanic, depressed, and
euthymic states in bipolar disorder. Am J Psychiatry
2004; 161: 262270.
7. Martinez-Aran A, Vieta E, Colom F et al. Cognitive
impairment in euthymic bipolar patients: implications for
clinical and functional outcome. Bipolar Disord 2004; 6:
224232.

123

Malhi et al.
8. Murphy FC, Sahakian BJ, Rubinsztein JS et al. Emotional
bias and inhibitory control processes in mania and
depression. Psychol Med 1999; 29: 13071321.
9. Clark L, Goodwin G. State- and trait-related decits in
sustained attention in bipolar disorder. Eur Arch Psychiatry Clin Neurosci 2004; 254: 6168.
10. Malhi GS, Ivanovski B, Szekeres V, Olley A. Bipolar
disorder: its all in your mind? The neuropsychological
prole of a biological disorder. Can J Psychiatry 2004; 49:
813819.
11. Martinez-Aran A, Vieta E, Colom F et al. Cognitive
dysfunctions in bipolar disorder: evidence of neuropsychological disturbances. Psychother Psychosom 2000; 69: 218.
12. Bearden CE, Homan KM, Cannon TD. The neuropsychology and neuroanatomy of bipolar aective disorder: a
critical review. Bipolar Disord 2001; 3: 106150.
13. Wolfe J, Granholm E, Butters N, Saunders E, Janowsky
D. Verbal memory decits associated with major aective
disorders: a comparison of unipolar and bipolar patients.
J Affect Disord 1987; 13: 8392.
14. Murphy FC, Sahakian BJ. Neuropsychology of bipolar
disorder. Br J Psychiatry 2001; 41: s120s127.
15. McClellan J, Prezbindowski A, Breiger D, McCurry C.
Neuropsychological functioning in early onset psychotic
disorders. Schizophr Res 2004; 68: 2126.
16. McKay AP, Tarbuck AF, Shapleske J, McKenna P.
Neuropsychological function in manic-depressive psychosis evidence for persistent decits in patients with chronic,
severe illness. Br J Psychiatry 1995; 167: 5157.
17. Kessing LV. Cognitive impairment in the euthymic phase
of aective disorder. Psychol Med 1998; 28: 10271038.
18. Tham A, Englebrekston K, Mathe AA, Johnson L, Olsson
E, Aberg-Wistedt AS. Impaired neuropsychological performance in euthymic patients with recurring mood
disorders. J Clin Psychiatry 1995; 58: 2629.
19. Clark L, Iversen SD, Goodwin GM. Sustained attention
decit in bipolar disorder. Br J Psychiatry 2002; 180: 313
319.
20. Ferrier IN, Thompson JM. Cognitive impairment in
bipolar aective disorder: implications for the bipolar
diathesis. Br J Psychiatry 2002; 180: 293295.
21. Olley A, Malhi GS, Mitchell PB, Batchelor J, Lagopoulos
J, Austin M-PV. When euthymia is just not good enough:
the neuropsychology of bipolar disorder. J Nerv Ment Dis
2005; 193: 323330.
22. Hirayasu Y, Shenton ME, Salisbury DF et al. Subgenual
cingulate cortex volume in rst-episode psychosis. Am J
Psychiatry 1999; 156: 10911093.
23. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for DSM-IV Axis I & Axis II
Disorders (Version 2.0). New York: Biometrics Research,
New York State Psychiatric Institute, 1995.
24. Hamilton M. A rating scale for depression. J Neurol
Neurosurg Psychiatry 1960; 23: 5662.
25. Montgomery P, Asberg B. A new depression scale designed
to be sensitive to change. Br J Psychiatry 1979; 134: 382
389.
26. Parker G, Hadzi-Pavlovic D. Development and structure
of the CORE System. In: Parker G, Hadzi-Pavlovic D, eds.
Melancholia: A Disorder of Movement and Mood. New
York: Cambridge University Press, 1996: 82129.
27. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating
scale for mania: reliability, validity and sensitivity. Br J
Psychiatry 1978; 133: 429435.
28. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th edn. Washington,
DC: American Psychiatric Association, 1994.

124

29. Beck AT, Ward CH, Mendelson M, Mock JE, Erbaugh

JK. An inventory for measuring depression. Arch Gen
Psychiatry 1961; 4: 5363.
30. Spielberger CD. State-Trait Anxiety Inventory. Palo Alto,
CA: Consulting Psychologists Press, 1970.
31. Eysenck HJ, Eysenck SBG. Manual of the Eysenck
Personality Inventory. London: Hodder and Stoughton,
1964.
32. Annett M. The binomial distribution of right, mixed
and left handedness. Q J Exp Psychol A 1967; 19:
327333.
33. Matthews CG, Klove H. Instruction Manual for the Adult
Neuropsychology Test Battery. Madison, WI: University
of Wisconsin Medical School, 1964.
34. Purdue Research Foundation. Examiners Manual for the
Purdue Pegboard. Chicago: Science Research Associates,
1948.
35. Rey A. Lexamen clinique en psychologie. Paris: Presses
Univesitaires de France, 1964.
36. Smith A. Symbol Digit Modalities Test Manual. Los
Angeles: Western Psychological Services, 1973.
37. Taylor EM. The Appraisal of Children with Cerebral
Decits. Cambridge, MA: Harvard University Press, 1959.
38. Trites RL. Neuropsychological Test Manual. Ottawa,
Ontario: Royal Ottawa Hospital, 1977.
39. Wechsler D. Wechsler Adult Intelligence Scales, 3rd edn.
New York: The Psychological Corporation, 1997.
40. Benton AL, Hamsher K. Multilingual Aphasia Examination. Iowa City: University of Iowa, 1976.
41. Fray PJ, Robbins TW, Sahakian BJ. Neuropsychiatric
applications of CANTAB. In J Geriatr Psychiatry 1996;
11: 329336.
42. Golden CJ. Stroop Color and Word Test: A Manual for
Clinical and Experimental Uses. Wood Dale Ill: Stoelting
Co., 1978.
43. Heaton RK. The Wisconsin Card Sorting Test. Odessa,
FL: Psychological Assessment Resources, 1981.
44. Spreen O, Strauss E. A Compendium of Neuropsychological Tests: Administration, Norms, and Commentary. New
York: Oxford University Press, 1998.
45. Lezak MD. Neuropsychological Assessment, 3rd edn. New
York: Oxford University Press, 1995.
46. Nelson HE, Willison JR. National Adult Reading Test
(NART) Second Edition Test Manual. Berkshire, UK:
NFER-NELSON Publishing Company Ltd, 1991.
47. Read DE. Neuropsychological Assessment of Memory
in Early Dementia: Normative Data for a New Battery
of Memory Tests Animal Naming. Victoria: Canada,
1980.
48. Cavanagh JTO, van Beck M, Muir W, Blackwood DHR.
Case-control study of neurocognitive function in euthymic
patients with bipolar disorder: an association with mania.
Br J Psychiatry 2002; 180: 320326.
49. Sweeney JA, Kmiec JA, Kupfer DJ. Neuropsychologic
impairments in bipolar and unipolar mood disorders on
the CANTAB neurocognitive battery. Biol Psychiatry
2000; 48: 674684.
50. Calev A, Nigal D, Chazan S. Retrieval from semantic
memory using meaningful and meaningless constructs by
depressed, stable bipolar and manic patients. Br J Clin
Psychol 1989; 28: 6773.
51. Porter RJ, Gallagher P, Thompson JM, Young AH.
Neurocognitive impairment in drug-free patients with major
depressive disorder. Br J Psychiatry 2003; 182: 214220.
52. Rezai K, Andreasen NC, Alliger R, Cohen G, Swayze V,
OLeary DS. The neuropsychology of the prefrontal
cortex. Arch Neurol 1993; 50: 636642.

Neuropsychological dysfunction in bipolar disorder

53. Goel V, Grafman J, Tajik J, Gana S, Danto D. A
study of the performance of patients with frontal lobe
lesions in a nancial planning task. Brain 1997; 120: 1805
1822.
54. Goel V, Grafman J. Are the frontal lobes implicated in
planning functions? Interpreting data from the Tower of
Hanoi. Neuropsychologia 1995; 33: 623642.
55. Thompson J, Gallagher P, Hughes J et al. Neurocognitive
impairment in euthymic patients with bipolar aective
disorder. Br J Psychiatry 2005; 186: 3240.
56. Ilsley JE, Mooot APR, OCarroll RE. An analysis of
memory dysfunction in major depression. J Affect Disord
1995; 35: 19.
57. Dhingra U, Rabins PV. Mania in the elderly: a 57 year
follow-up. J Am Geriatr Soc 1991; 39: 581583.
58. Fu CHY, Morgan K, Suckling J et al. A functional
magnetic resonance imaging study of overt letter verbal
uency using a clustered acquisition sequence: greater
anterior cingulate activation with increased task demand.
Neuroimage 2002; 17: 871879.
59. Dickerson FB, Sommerville J, Origoni AE et al. Outpatients with schizophrenia and bipolar I disorder: do they
dier in their cognitive and social functioning. Psychiatry
Res 2001; 102: 2127.
60. Zubieta J-K, Huguelet P, ONeil RL, Giordani BJ.
Cognitive function in euthymic bipolar I disorder. Psychiatry Res 2001; 102: 920.

61. van Gorp W, Altshuler L, Theberge DC, Wilkins J, Dixon

W. Cognitive impairment in euthymic bipolar patients with
and without prior alcohol dependence: a preliminary study.
Arch Gen Psychiatry 1998; 55: 4146.
62. Brand N, Jolles J. Information processing in depression
and anxiety. Psychol Med 1987; 17: 145153.
63. Sax KW, Strakowski SM, McElroy SL, Keck J, Paul E,
West SA. Attention and formal thought disorder in
mixed and pure mania. Biol Psychiatry 1995; 37: 420
423.
64. Mitchell PB, Wilhelm K, Parker G, Austin M-P, Rutgers
P, Malhi GS. The clinical features of bipolar depression: a
comparison with matched major depressive disorder
patients. J Clin Psychiatry 2001; 62: 212216.
65. Kibby MY, Schmitter-Edgecombe M, Long CJ. Ecological
validity of neuropsychological tests: focus on the California Verbal Learning Test and the Wisconsin Card Sorting
Test. Arch Clin Neuropsychol 1998; 13: 523534.
66. Mitchell PB, Malhi GS. Bipolar depression: phenomenological overview and clinical characteristics. Bipolar Disord 2004; 6: 530539.
67. Pachet AK, Wisniewski AM. The eects of lithium on
cognition: an updated review. Psychopharmacology 2003;
170: 225234.

125