Journal of Affective Disorders 109 (2008) 286 – 299

www.elsevier.com/locate/jad

Research report
Neurocognitive and clinical predictors of functional outcome in
patients with schizophrenia and bipolar I disorder at
one-year follow-up
Rafael Tabarés-Seisdedos a,b,⁎, Vicente Balanzá-Martínez a,b , José Sánchez-Moreno b,c ,
Anabel Martinez-Aran b,c , José Salazar-Fraile a,b , Gabriel Selva-Vera a,b ,
Cristina Rubio a,b , Ignacio Mata d , Manuel Gómez-Beneyto a , Eduard Vieta b,c,⁎
a
The Teaching Unit of Psychiatry and Psychological Medicine, Department of Medicine,
University of Valencia, Blasco-Ibáñez 17, 46010 Valencia, Spain
b
Ciber en Salud Mental, Instituto de Salud Carlos III, Madrid, Spain
c
The Bipolar Disorders Program, Clinical Institute of Neuroscience, Hospital Clinic of Barcelona, IDIBAPS,
University of Barcelona, Villarroel 170, 08036 Barcelona, Spain
d
The Division of Psychological Medicine, Institute of Psychiatry, De Crespigny Park, London SE5 8AF, UK
Received 8 July 2007; received in revised form 24 December 2007; accepted 24 December 2007
Available online 4 March 2008

Abstract

Objective: Many studies have reported that cognitive ability may be predictive of the functional outcome for patients with
schizophrenia. However, no study has prospectively examined these aspects in schizophrenia and bipolar disorders simultaneously.
The present study attempted to analyze if neurocognition and clinical status predicts the real-life functioning for patients with
schizophrenia or bipolar I disorder, using a longitudinal design.
Method: Forty-seven schizophrenic and 43 bipolar I outpatients were assessed twice with a neurocognitive battery (Executive
Functions, Working Memory, Verbal Memory, Visual Memory, Visual-Motor Processing, Vigilance, Vocabulary and Motor Speed
tasks), clinical scales (the Positive and Negative Symptom Scale, the Hamilton Rating Scale for Depression and the Clinician
Administered Rating Scale for Mania) and functional outcome measures (the Global Assessment of Functioning Scale, the WHO's
Disability Assessment Scale and occupational adaptation level) over a one-year follow-up period. The cognitive performance of the
patients was compared, at baseline and one year later, with that of 25 healthy subjects.
Results: In schizophrenia patients, global functioning one year later was predicted by a composite neurocognitive score and three
specific domain (verbal memory, motor speed, vocabulary). Symptoms appeared to explain less of the variance in functioning. In
bipolar I patients, changes in the composite neurocognitive score over one year, deficits in the visual/motor processing domain,
severity of symptoms (psychotic, excitatory and affective symptoms) and premorbid adjustment at the first assessment were the
variables that better predicted functioning or disability changes over follow-up period.
Conclusions: Although the relationships between cognition, symptoms and functional capacity differ for schizophrenia or bipolar I
patients, neuropsychological performance seems to be a principal longitudinal predictor of functioning in both disorders. Baseline

⁎ Corresponding authors. Tabarés-Seisdedos, is to be contacted at the Teaching Unit of Psychiatry and Psychological Medicine, Department of
Medicine, University of Valencia, Blasco-Ibáñez 17, 46010 Valencia, Spain. Tel.: +34 96 3864744; fax: +34 96 3864767. Vieta, the Bipolar Disorders
Program, Clinical Institute of Neuroscience, Hospital Clinic of Barcelona, IDIBAPS, University of Barcelona, Villarroel 170, 08036 Barcelona, Spain.
Tel.: +34 93 2275401; fax: +34 93 227976.
E-mail addresses: Rafael.Tabares@uv.es (R. Tabarés-Seisdedos), EVIETA@clinic.ub.es (E. Vieta).

0165-0327/$ - see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jad.2007.12.234
 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
neurocognition and cognitive changes over 12 months predicted changes in functioning over the same period, but only in bipolar I patients.
These cognitive domains could be potential neurocognitive endophenotypes (endophenocognitypes) with regard to bipolar I disorder.
© 2008 Elsevier B.V. All rights reserved.
Keywords: Neurocognition; Endophenocognitype; Symptoms; Functional outcome; Follow-up; Schizophrenia; Bipolar disorders
1. Objectives of the study
Over the past 20 years, many studies have reported
that some neurocognitive domains and negative symp-
toms may be predictive of functioning in schizophrenia
(Green, 1996). Unfortunately, little is known about the
association between neurocognition and functional out-
come in bipolar disorders, and how these relationships
are comparable with those found in schizophrenia. Some
pioneer cross-sectional studies indicated that bipolar
subjects with poor functioning are more cognitively
impaired than highly functioning patients, in particular,
on verbal memory measures (Martinez-Aran et al., 2007;
Dickerson et al., 2004), vigilance or sustained attention
domain (Clark et al., 2002), and executive tasks
(Nordenson et al., 2004). Worthy of note is that only
three previous investigations have compared schizo-
phrenic and bipolar patients' neurocognitive and func-
tional outcome in the same cross-sectional study; these
studies had conflicting results (Martinez-Aran et al.,
2002; Dickerson et al., 2001; Laes and Sponheim, 2006).
In spite of the fact that relationships between neuro-
cognition and functional outcome are the type of issues that
can be best approached using a longitudinal design, only
one study of bipolar patients has met this follow-up
criterion (Jaeger et al., 2007), although at least 24 studies in
schizophrenia have (McGurk et al., 2004; Sota and
Heinrichs, 2004; Kurtz et al., 2005; Carlsson et al., 2006;
Jaeger et al., 2006; Siegel et al., 2006; see Green et al.,
2004a for a review). In this systematic review, Green et al.
(2004a) demonstrated longitudinal associations between
cognition and outcome in 18 published longitudinal
studies. Specifically, 12 studies found that neurocognitive
performance at baseline predicts community functioning
months or years later. Even more interestingly, four of these
studies found that cognitive capacity at baseline predicts
change in functioning over time.
The present study analyzes prospectively two cohorts
of schizophrenia and bipolar I outpatients, and a cohort of
healthy subjects. In particular, we examined two questions:
1. Are neurocognition and clinical factors significant
predictors of functioning (and more specifically, of
changes in functional measures) one year later?
287
2. Are these relationships between neurocognition,
symptoms and functional measures different for
schizophrenia and bipolar I patients?
2. Materials and methods
2.1. Subjects
Subjects participating in this observational study were
enrolled in the Valencia Follow-Up Study of Schizo-
phrenia and Bipolar I Disorder (Tabares-Seisdedos et al.,
2003; Balanza-Martinez et al., 2005). Patients who
fulfilled the DSM-IV criteria for either schizophrenia
(n = 52) or bipolar disorder type I (n = 49) (American
Psychiatric Association, 1994) were recruited from those
consecutively attending three different psychiatric out-
patient units in Valencia, over nine months. Diagnoses
were confirmed by the Schedules for Clinical Assessment
in Neuropsychiatry (SCAN-CATEGO) (Wing et al., 1990)
after a minimum of two year's progress of the illness.
All participants were evaluated at baseline (T1) and
one year later (T2). Most patients in this study were
treated in settings that did not offer psychosocial
rehabilitation programs. Only twelve patients (nine
schizophrenic and three bipolar subjects) were partici-
pating in a psychosocial rehabilitation program at
follow-up. Eleven patients were lost to follow-up (five
schizophrenic and six bipolar patients) for the following
reasons: change of area of residence (two schizophrenic
subjects), and refusal to participate (three schizophrenic
and six bipolar subjects). Hence, data from 47 schizo-
phrenic patients and 43 bipolar patients, who had
completed the follow-up at the time this paper was
written and thus conform to the longitudinal sample, are
reported. There was no difference between patients
with and without follow-up assessment for either
neurocognitive, clinical or functional scores at baseline,
suggesting that the longitudinal sample was representa-
tive of the original sample. At each time point, all
patients were medicated by their psychiatrists in a
naturalistic manner. The average daily dose of anti-
psychotic drugs was converted into chlorpromazine-
equivalent (CPZ) units for statistical purposes (Mangell
et al., 1999; Woods, 2003). The healthy volunteers
288 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
(n = 25) were interviewed with the Family History-RDC
interview (Endicott et al., 1978) and the Structured
Clinical Interview for DSM-IV (First et al., 1997).
Written informed consent was obtained from all partici-
pants following an explanation of study procedures. The
Ethics Committee of the University Clinic Hospital of
Valencia approved the research protocol.
2.2. Clinical and functional scales
The clinical evaluation was rated according to the
Positive and Negative Symptom Scale (PANSS) (Kay
et al., 1987), the Hamilton Rating Scale for Depression
(HSRD) (Hamilton, 1960) and the Clinician Administered
Rating Scale for Mania Factor 1 (CARS-M) (Altman et al.,
1994). We used the Positive, Negative, General, and Total
PANSS scores, and the PANSS five-dimension model
(Cuesta et al., 1998). Premorbid adjustment was measured
by means of the Phillips Adjustment Scale (Phillips, 1953).
The assessment of functioning was carried out by
(a) the Global Assessment of Functioning Scale (Amer-
ican Psychiatric Association, 1994), which evaluates
both symptom severity and social adjustment; (b) the
short version of the WHO's Disability Assessment Scale
(World Health Organization, 1988), which is specific to
social adjustment, and was used to analyze five outcome
measures (personal care, occupational functioning,
family functioning, social functioning and total score);
and (c) classification of occupational adaptation, which
was established as “good occupational adaptation” when
patients were working at a good or acceptable level of
functioning most of the time, or as “low occupational
adaptation” if they did not work at all or showed
moderate or severe difficulties in their job most of the
time (Torrent et al., 2006) Higher scores mean better
functioning on the GAF scale, whereas higher scores on
the disability WHODAS scale mean worse functioning.
Trained psychiatrists, blind to the results of the
neuropsychological evaluation, conducted the clinical
interview and rated patients' functioning.
2.3. Neuropsychological measures
All patients and control subjects completed a battery of
11 tests, which were described in two previous publica-
tions (Tabares-Seisdedos et al., 2003; Balanza-Martínez
et al., 2005). These measured eight neurocognitive
domains in the following sequence: 1) Executive Func-
tions/Reasoning and Problem Solving (WCST Categories
and Perseverative Errors; FAS Test from the Controlled
Oral Word Association Test; Trail Making Test, part B;
Color-Word Interference trial of the Stroop Color and
Word Test); 2) Verbal Working Memory (the backward
part of the Digit Span Test from the Wechsler Adult
Intelligence Scale-Revised [WAIS-R]); 3) Verbal Memory
(the Babcock Story Recall Test); 4) Visual Memory (the
Rey–Osterrieth Complex Figure Test); 5) Visual-Motor
Processing/Speed of Processing (Trail Making Test, part
A; Stroop Color and Word Test with two separate tasks:
Word reading trial and Color naming trial; Digit Symbol
Substitution Test [DSST] from the WAIS-R; Semantic
Verbal Fluency from the Category Instant Generation Test
[CIG]); 6) Vigilance (from the Asarnow Continuous Per-
formance Test, the total number of hits minus the total
number the false alarms was calculated); 7) Motor Speed
(Finger-Tapping Test, average of dominant and non-
dominant hand scores in uni-manual and bimanual condi-
tions); and 8) Language or Vocabulary (the Vocabulary
Subtest of the WAIS-R. This test was also used as an
estimate of premorbid intelligence quotient [PIQ]). These
eight domains and tests clustering correspond to six of the
seven domains in the MATRICS consensus neuropsycho-
logical battery (Green et al., 2004b).
Patients' scores on each measure were converted to z
scores, based on data from the normal control group.
Neurocognitive domain scores represented the average z
scores for all available tests for each patient within that
domain. A Composite Neurocognitive Score (CNCS)
was created with the sum average of the z scores for all
eight domains.
2.4. Statistical analyses
To control for the multiple comparison effects, we per-
formed multivariate analysis of covariance (MANCOVA)
for all neurocognitive domains at baseline and one year
later. Age and IQ were used as covariates. Factors included
were group (schizophrenic, bipolar and healthy subjects)
and gender. Univariate analyses of covariance (ANCO-
VAs) on dependent variables from each test, neurocognitive
domain and the CNCS were performed. The factors used
were group (schizophrenic, bipolar and healthy subjects, or
schizophrenic and bipolar subjects) and gender, and the
covariates were age and the IQ. Finally, ANCOVAs with
Tukey post hoc adjustment and t tests were performed.
To compare changes over time for each functional
outcome variable, diagnostic group (schizophrenia, bipo-
lar) × time (T1, T2), ANCOVAs with repeated measures
were performed, with age and IQ as covariates.
Pearson correlations were calculated between neuro-
cognitive measures and clinical variables at baseline, and
functional measures at one year later (GAF and WHODAS
at T2) and over the one-year follow-up period (GAF T2–
T1 and WHODAS T2–T1).
 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299 289

Neurocognitive variables at baseline that demonstrated
a significant association with functional outcome at one
year later at p ≤ 0.006 (p ≤ 0.05, Bonferroni-corrected for
eight comparisons) were entered in the follow-up
regression models. Clinical variables at T1 that demon-
strated a significant association with functional measures
at one year later at p ≤ 0.05 were also entered in the
regression models. Firstly, linear regression analyses with
backward stepwise elimination were performed to assess
the relative contributions of each of the neurocognitive
and clinical variables at baseline to explain the variance in
outcome measures at T2 and over the one-year follow-up
period (T2–T1). Secondly, a logistic regression analysis
was performed to investigate which variable, at baseline,
was the strongest independent predictor of good or low
occupational adaptation group membership at T2. The
results are presented as odds ratios and 95% confidence
intervals. All statistical analyses were processed with the
computer package SPSS, version 14.02 for Windows.
3. Results
3.1. Participant characteristics at baseline and one
year later
Demographic and clinical characteristics for the three
participant groups, at both time points, are summarized
in Tables 1 and 2. Compared with bipolar and healthy

Table 1
Demographic and clinical characteristics of schizophrenic and bipolar I patients and healthy subjects at baseline
Demographic and clinical variables Schizophrenic Bipolar I patients Healthy subjects t-test ANOVA Bonferroni
patients
SZ (n = 47) BP (n = 43) C (n = 25)
Mean SD Mean SD Mean SD t F(p) Post hoc test
Age 33.4 8.2 41.2 11.5 41.4 11.3 8.2⁎⁎⁎ SZ b BP, C
Educational level (years) 9.8 2.9 10.1 4.1 14.3 4.4 12.9⁎⁎⁎ SZ, BP b C
Estimated premorbid IQ 44.2 15.1 47.5 14.2 62.6 10.1 15.0⁎⁎⁎ SZ, B b C
Hand dominance 4.5 1.7 4.9 0.5 4.4 2.3 NS
Age at onset (years) 24.8 6.6 29.3 9.8 − 2.6⁎⁎
Chronicity (years) 8.7 7.3 11.9 9.6 NS
Total episodes 2.4 1.9 5.3 3.5 − 5.0⁎⁎⁎
Hospitalisations 0.6 1.1 1.4 1.7 − 2.5⁎⁎
Length of hospitalisations (weeks) 0.9 1.6 2.0 2.8 2.3⁎
PANSS dimensions
Positive 2.0 1.0 1.3 0.7 3.4⁎⁎
Negative 3.4 1.3 1.5 0.7 8.5⁎⁎⁎
Cognitive 2.5 1.1 1.6 0.6 5.1⁎⁎⁎
Excitatory 1.8 0.9 1.4 0.7 2.1⁎⁎⁎
Affective 2.0 0.8 1.8 1.0 NS
Premorbid adjustment 14.2 6.4 6.8 6.0 5.5⁎⁎⁎

Gender n % n % n % χ2
Female 10 21.3 22 51.2 11 44.0 9.1⁎⁎
Male 37 78.7 21 48.8 14 56.0
Low occupational adaptation 36 76.6 27 62.8 4 16.0 25.2⁎⁎⁎
Prior psychotic symptoms 19 44.2
Positive family history of psychosis 8 17.0 14 32.6 NS
Medication
Lithium 0 0 23 53.5
Carbamazepine 3 6.4 18 41.9
Benzodiazepines 15 31.9 29 67.4
Biperiden 11 23.4 1 2.3
Antipsychotics 47 100 20 46.5
Antidepressants 6 12.8 12 27.9
Mean SD Mean SD t
Chlorpromazine equivalents 862′7 642′2 172′0 276′6 6′5⁎⁎⁎
Number of antipsychotic drugs 1.3 0.6 0.5 0.6 6.2⁎⁎⁎
Total number of psychotropic drugs 2.1 1.4 2.9 1.2 − 2.8⁎⁎
(NS = p N 0.05; ⁎p b 0.05; ⁎⁎p b 0.01; ⁎⁎⁎p b 0.0001).
290 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
subjects, patients with schizophrenia were significantly
younger and showed a lower occupational adaptation
level. Participants with schizophrenia also had worse
functioning at baseline and one year later, and poorer
premorbid adjustment. Schizophrenic patients were
younger at onset of the disorder, and showed more
severe psychopathology as measured by the PANSS and
HDRS scales at both time points. Patients with bipolar I
disorder showed higher mania scores at baseline and one
year later. In schizophrenic patients, the ranges of the
HAMD and the CARS-M scores were 0–20 and 0–3
respectively at T1, and 0–21 and 0–3 respectively at T2.
In bipolar I patients, the ranges of the HAMD and the
CARS-M scores were 0–16 and 0–19 respectively at
T1, and 0–18 and 0–16 respectively at T2. There were
more males in the schizophrenic group. Control subjects
showed more years of education and a higher PIQ than
both diagnostic groups.
At T1, six bipolar patients were clinically depressed,
four were manic/mixed and 33 were euthymic (HDRS ≤ 8
and CARS-M factor 1 ≤ 7). Of the latter group, 27 were
also euthymic at T2. In addition, the bipolar sample's
mean depression and mania scores remained stable at both
assessments (Table 2). Thus, during follow-up, most of
the bipolar patients were judged clinically stable. A
history of psychotic symptoms (defined as the presence of
delusions or hallucinations at any time in the course of
illness) was confirmed by the SCAN-CATEGO for 14
(33%) bipolar I patients. No bipolar patient had a rapid-
cycling illness.
3.2. Neurocognition, clinical status and functional outcome
at baseline and one year later
As expected (Daban et al., 2006), schizophrenic and
bipolar I subjects displayed reduced performance in
overall neuropsychological capacity at T1, compared
with healthy subjects (MANOVA, Pillai's F = 3.2,
p b 0.001; η2 = 0.18) and at T2 (MANOVA, Pillai's
F = 3.7, p b 0.0001; η2 = 0.2) (see Table 2). ANCOVAs
with Tukey post hoc adjustment showed that there were
significant differences between the two patient groups
and the healthy subjects in the CNCS, in all the
neurocognitive domains (except in the sustained atten-
tion domain at T1 and at T2, working memory at T1, and
executive domain at T2), and in the Stroop Color and
Word Test, the Vocabulary Subtest, FAS Test, CIG Test,
Rey–Osterrieth Complex Figure Test, DSST and Finger
Tapping Test at baseline and one year later (see Table 2
for details). The only significant differences among
healthy, schizophrenic and bipolar I subjects were
confined to the Babcock Story Recall Test (immediate
and delayed verbal memory) at both time points, and to
the CIG test (semantic verbal fluency) at baseline but
not at T2. Specifically, schizophrenic patients displayed
a significantly reduced performance in comparison to
normal controls, while bipolar I patients were at an
intermediate level between these groups (Table 2).
Effect sizes were calculated using the On-line
Program for Cohen's d and r (Becker, 2007). Specifi-
cally, effect size (ES) estimates (Cohen's d) were
calculated by subtracting the mean score for healthy
controls from the mean score for patient group divided by
the pooled standard deviation (Cohen, 1988). In all cases,
positive ESs indicate a worse performance for the patient
group compared to controls. When a lower score meant a
better performance, the sign of the ES was reversed. In
schizophrenic patients, ESs ranged from 0.4 (Asarnow
CPT) to 1.9 (CIG and delayed verbal memory) at T1, and
similar range was found at T2. For bipolar I patients, ESs
ranged from 0.6 (Asarnow CPT) to 1.5 (CIG and FAS
tests) at T1, and similar range was found at T2.
To compare changes over one year for each
functional outcome variable, diagnostic groups (schizo-
phrenia, bipolar) × time (T1, T2) repeated-measures
analysis of covariance (ANCOVA) were performed
with age, PIQ and length of illness as covariates. Gender
was also included as an additional inter-subjects factor.
Results showed that time had no significant effect on
GAF score (F = 0.08; df = 1; Ms = 4.4; p = 0.77) and
WHO-DAS scores (F = 1.5; df = 1; Ms = 6.9; p = 0.22).
There was also no significant interaction between
diagnosis and time [GAF score (F = 0.02; df = 1;
Ms = 1.07; p = 0.90); WHO-DAS score (F = 0.09;
df = 1; Ms = 0.4; p = 0.76)].
3.3. Neurocognition and clinical status at baseline as
predictors of functioning at one year later
3.3.1. Schizophrenic patients
3.3.1.1. Correlational analyses. The bivariate correla-
tional analyses are shown in Supplementary Material 1,
which is available with the online version of this article
at http://www.sciencedirect.com. Baseline performance
in all the neurocognitive domains was significantly
associated with global functioning (less in the sustained
attention domain), as measured by the GAF, one year
later. The CNCS at baseline also showed a positive
association with better global functioning, as measured
by the GAF, at T2. Only two neurocognitive domains
(executive and vocabulary) at T1 were significantly
correlated with the WHODAS total score at T2. All
observed correlations were in the expected direction,
 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
Table 2
Functional, clinical and neurocognitive results of schizophrenic, bipolar I and healthy subjects at baseline (T1) and one year later (T2)
Schizophrenic patients Bipolar I patients Healthy subjects
SZ (n = 47) BP (n = 43) C (n = 25)
Baseline 1 year Baseline 1 year Baseline 1 year
Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Functional scales t-test T1 t-test T2
GAF 59.7 (12.4) 61.2 (11.6) 77.3 (12.3) 76.5 (14.5) − 6.7⁎⁎⁎⁎ − 5.5⁎⁎⁎⁎
WHO-DAS total score 5.3 (3.2) 5.6 (2.9) 2.4 (2.4) 2.6 (3.2) 4.9⁎⁎⁎⁎ 4.6⁎⁎⁎⁎
Personal care 0.3 (0.6) 0.1 (0.4) 0.05 (0.2) 0.1 (0.2) 2.4⁎ 1.8
Occupational functioning 2.2 (1.5) 2.6 (1.4) 1.1 (1.1) 1.0 (1.4) 4.0⁎⁎⁎⁎ 5.4⁎⁎⁎⁎
Family functioning 0.9 (1.1) 1.2 (1.1) 0.4 (0.9) 0.8 (1.0) 2.4⁎ 1.7
Social functioning 1.9 (1.6) 1.6 (1.4) 0.7 (1.1) 0.7 (0.9) 3.9⁎⁎⁎⁎ 3.4⁎⁎⁎
Clinical variables t-test T1 t-test T2
PANSS
Positive scale 12.9 (5.1) 13.0 (4.9) 9.8 (4.9) 8.0 (2.2) 2.9⁎⁎ 6.0⁎⁎⁎
Negative scale 21.8 (8.4) 22.0 (8.1) 10.5 (3.9) 10.5 (4.8) 8.0⁎⁎⁎ 8.1⁎⁎⁎
General scale 32.9 (8.5) 33.4 (9.1) 23.9 (7.7) 21.6 (6.3) 5.2⁎⁎⁎ 7.1⁎⁎⁎
Total score 67.6 (18.4) 68.4 (18.7) 44.2 (13.4) 40.1 (11.7) 6.8⁎⁎⁎ 8.5⁎⁎⁎
HDRS 6.4 (4.7) 7.4 (4.9) 4.2 (4.1) 4.7 (5.7) 2.4⁎⁎ 2.4⁎⁎⁎
CARS-M Factor 1 0.4 (0.8) 0.2 (0.3) 2.2 (4.2) 1.3 (2.3) − 2.8⁎⁎ − 3.5⁎⁎⁎

Neurocognitive domains and tests ANCOVA T1 Tukey ANCOVA T2 Tukey

F(p) a Post hoc test F(p) a Post hoc test
Comp. Neurocognitive Score–CNCS − 1.43 (1.1) − 1.46 (1.0) − 1.17 (0.9) −1.27 (0.9) 9.3⁎⁎⁎⁎ c, d SZ, BP b C 13.3⁎⁎⁎⁎ c, d
SZ, BP b C
Executive/reasoning − 2.15 (2.0) − 1.9 (1.9) − 1.63 (1.6) −1.72 (1.7) 5.5⁎⁎ c, d SZ, BP b C NS, c, d
WCST-Categories 3.9 (2.2) 4.6 (2.0) 4.3 (2.1) 4.5 (2.1) 5.8 (1) 5.7 (0.8) NS, d NS, d
WCST-PE 24.7 (21.3) 19.8 (20.6) 19.5 (16.2) 18.2 (15.7) 8.7 (4.9) 7.9 (5.0) NS, c, d NS, c, d
Trail Making B 147.2 (67.6) 134.9 (67.0) 151.4 (87.8) 131.8 (81.5) 92.3 (54.7) 79.2 (32.0) NS, c, d NS, c, d
Stroop 3-(Word-Color Int.) 78.2 (43.0) 68.5 (28.4) 70.0 (27.2) 65.6 (25.8) 48.7 (10.1) 45.4 (9.9) 4.1⁎ c, d SZ, BP N C 3.0⁎ c, d SZ, BP N C
9.4⁎⁎⁎⁎ d 8.2⁎⁎⁎⁎

291
FAS 25.9 (13.6) 29.9 (13.9) 31.3 (13.5) 34.7 (14.1) 50.9 (14.8) 56.2 (15.6) SZ, BP b C d
SZ, BP b C
(continued on next page)
 SZ (n = 47) BP (n = 43) C (n = 25)
Baseline 1 year Baseline 1 year Baseline 1 year
Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD)

292
Working memory −0.83 (1.1) − 0.86 (0.6) − 0.88 (0.7) − 0.78 (0.6) NS, b, d 7.1⁎⁎⁎ b, c, d
SZ, BP b C
Backward Digit Span 3.9 (1.2) 3.8 (1.0) 3.8 (0.9) 3.9 (1.1) 4.8 (1.1) 5.3 (1.8) NS, b, d
7.1⁎⁎⁎ b, c, d SZ, BP b C
Verbal memory −1.74 (1.0) − 1.8 (1.1) − 1.0 (0.9) − 1.1 (1.0) 10.6⁎⁎⁎⁎ d SZ, BP b C 8.9⁎⁎⁎⁎ c, d SZ, BP b C
VM immediate 7.3 (3.1) 8.2 (3.3) 9.5 (2.7) 10.1 (3.1) 12.2 (3.1) 13.7 (3.6) 9.5⁎⁎⁎⁎ c, d SZ b BP b C 8.8⁎⁎⁎⁎ c, d SZ b BP b C
VM delayed 9.7 (4.0) 10.2 (4.2) 12.2 (3.5) 13.1 (4.6) 16.1 (3.4) 16.9 (3.2) 9.2⁎⁎⁎⁎ c, d SZ b BP b C 7.8⁎⁎⁎ d SZ b BP b C
Visual memory −0.9 (0.9) − 1.0 (0.9) − 0.87 (0.9) − 1.0 (0.8) 4.1⁎ c, d SZ, BP b C 5.3⁎⁎ c, d SZ, BP b C
Figure Rey immediate 14.8 (7.7) 17.5 (7.2) 15.1 (7.0) 16.8 (7.1) 21.2 (7.9) 24.6 (7.6) 3.1⁎ c, d SZ, BP b C 5.6⁎⁎ c, d SZ, BP b C
Figure Rey delayed 14.6 (7.6) 17.2 (7.4) 14.8 (7.0) 17.5 (6.4) 21.8 (7.3) 24.8 (7.2) 4.9⁎⁎ b, c, d SZ, BP b C 4.7⁎⁎ c, d SZ, BP b C
Visual motor/processing speed −1.9 (1.7) − 1.8 (1.7) − 1.5 (1.5) − 1.7 (1.6) 7.4⁎⁎⁎ c, d SZ, BP b C 6.4⁎⁎ c, d SZ, BP b C
Trail Making A 55.9 (27.1) 59.7 (22.1) 56.0 (27.3) 49.3 (27.2) 39.4 (14.9) 32.0 (11.1) NS, c, d NS, c, d, e
16.9⁎⁎⁎⁎ d SZ b BP b C 12.8⁎⁎⁎⁎ c, d SZ, BP b C

R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299

CIG 39.9 (11.6) 46.7 (12.3) 46.9 (11.9) 49.2 (11.1) 63.9 (12.6) 69.2 (14.6)
Stroop 1–(Word reading) 26.4 (10.8) 25.9 (10.9) 24.1 (7.19) 24.4 (7.5) 18.7 (3.7) 17.8 (3.3) 3.7⁎ c, d SZ, BP N C 2.9 c, d
Stroop 2–(Color naming) 38.3 (17.7) 34.1 (14.6) 34.7 (14.4) 33.9 (12.5) 24.1 (4.5) 22.9 (4.7) 3.7⁎ c, d SZ, BP N C 3.1⁎ c, d SZ, BP N C
DSST 46.3 (15.1) 47.4 (16.6) 46.8 (19.3) 50.1 (23.4) 69.8 (16.1) 71.6 (16.3) 12.2⁎⁎⁎⁎ c, d SZ, BP b C 10.7⁎⁎⁎⁎ c, d SZ, BP b C
Sustained attention/vigilance −0.86 (2.4) − 1.0 (1.4) − 0.8 (1.4) − 1.2 (1.6) NS, d NS, c, d
Asarnow CPT 20.0 (10.3) 21.7 (4.1) 20.4 (5.9) 21.2 (4.6) 23.7 (4.2) 24.5 (2.8) NS, d NS, c, d
Motor speed −1.2 (0.9) − 1.4 (1.0) − 1.1 (0.8) − 1.1 (1.0) 13.1⁎⁎⁎⁎ b, c, d SZ, BP b C 17.8⁎⁎⁎⁎ d, e SZ, BP b C
Finger-Tapping unimanual condition 64.6 (16.2) 64.2 (14.5) 65.2 (14.5) 67.7 (15.8) 82.7 (17.6) 84.5 (12.1) 9.3⁎⁎⁎⁎ b, c, d SZ, BP b C 15.4⁎⁎⁎⁎ c SZ, BP b C
Finger-Tapping bimanual condition 60.9 (16.7) 57.8 (13.9) 62.5 (15.3) 63.4 (14.3) 81.5 (14.7) 78.4 (16.3) 13.7⁎⁎⁎⁎ b, c, d SZ, BP b C 17.7⁎⁎⁎⁎ b, c SZ, BP b C
Higher scores indicate poorer performance, except for: WCST-Categories, CIG, FAS, VM immediate, VM delayed, DSST and Finger Tapping Tasks, Asarnow CPT, Figure Rey immediate, Figure Rey delayed,
Backward Digit Span Test.
(NS = p N 0.05; ⁎p b 0.05; ⁎⁎p b 0.01; ⁎⁎⁎p b 0.001; ⁎⁎⁎⁎p b 0.0001).
M (SD) of CNCS and cognitive domains in healthy subjects = 0.0 (±1).
a
Main effect of group in ANCOVA.
b
Main effect of gender (p b 0.05).
c
Main effect of age (p b 0.05).
d
Main effect of premorbid IQ (p b 0.05).
e
Main effect of group × sex interaction (p b 0.05).
 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299 293

with higher neurocognitive performance at T1 asso- Table 3

ciated with higher functioning one year later. As shown Relative contributions of cognition and clinical factors at baseline to
explain the variance in outcome measures of schizophrenic patients at
in Supplementary Material 1, no significant correlations one-year follow-up
were found between any of the neurocognitive domains
Number of Predictors at T1 Percent of variance
at baseline and measures of functional change over the regression that entered in the explained (adjusted R2)
one-year follow-up period. analyses and equation
Partial (%) Total (%)
Of the clinical symptoms, PANSS Negative score, dependent
PANSS General score, PANSS Total score, Negative variables at T2
Dimension and Cognitive Dimension at T1 were 1° and 2° analyses Motor speed a 8′8
significantly associated with global functioning one GAF Verbal memory b 23′8 32′6
year later, as measured by the total score on the GAF. 3° analysis PANSS general c 6′9
GAF CNCS⁎ d 18′7 25′6
The remaining clinical variables were not correlated
4° analysis Years of education e 5′5
with any functional variable at T2. All observed WHO-DAS Vocabulary f 14′2 19′7
correlations were in the expected direction, with higher
CNCS⁎ = Composite Neurocognitive Score.
symptom ratings associated with lower global a
[t = 2.62, β = 0.350 (95% CI = 0.95 to 7.24), df = 46, p = 0.012].
functioning. b
[t = 3.92, β = 0.504 (95% CI = 2.69 to 8.38), df = 46, p = 0.001].
With regard to pharmacological factors, only the c
[t = − 2.27, β = − 0.318 (95% CI = − 0.82 to − 0.050), df = 46,
number of antipsychotic drugs at T1 showed a negative p = 0.028].
d
[t = 3.39, β = 0.452 (95% CI = 1.67 to 6.5), df = 46, p = 0.001].
correlation with GAF score at T2: lower numbers of e
[t = − 2.02, β = − 0.329 (95% CI = −0.649 to − 0.002), df = 46,
antipsychotic drugs were associated with better p = 0.049].
functioning. f
[t = −2.93, β = − 0.401 (95% CI = −1.326 to − 0.247), df = 46,
Finally, of the other factors, years of education at p = 0.005].
baseline were significantly associated with global
functioning, as measured by the total score on the domains (first block); and years of education and
GAF and on the WHO-DAS, one year later. The number premorbid adjustment (second block). The variables
of hospitalizations, length of episodes and premorbid that better predicted global disability at T2 were
adjustment at baseline were associated with the GAF vocabulary and years of education [F (2, 46) = 6.65,
score at T2. Premorbid adjustment at T1 was also p b 0.003].
correlated with WHO-DAS at T2, as measured by GAF Finally, occupational adaptation was analyzed by
(T2–T1). logistic regression analysis. The analysis included the
following variables at baseline: sex, age, years of
3.3.1.2. Multivariate prediction of functional out- education, premorbid adjustment, age of onset, number
come. We used four linear regression analyses with of admissions, hand dominance, length of illness,
hierarchical methods (Table 3). In the first analysis, the CNCS, HSRD, CARS-M, PANSS Positive, PANSS
GAF score at T2 was the dependent variable: executive, Negative and PANSS General scores, number of
verbal memory and motor speed domains (first block); antipsychotic drugs, chlorpromazine equivalents (CPZ-
Negative and General PANSS scores (second block); units) and biperiden dose. This regression showed that
number of antipsychotic drugs (third block); and years the CNCS at T1 was the strongest independent predictor
of education, number of hospitalizations and premorbid of good or low occupational adaptation group member-
adjustment (fourth block). The variables that better ship at T2 (χ2 = 7.1; df = 1; p b 0.008). The rerunning of
predicted functioning at T2 were verbal memory and the logistic regression model included all neurocogni-
motor speed domains [F (2, 46) = 12.12, p b 0.001]. The tive domains, and the specific variables that better
second regression analysis was identical to the first, predicted membership of the occupational adaptation
except that the clinical variables PANSS Negative and groups were motor speed domain and HSRD score at
Neurocognitive Dimensions were included. The results baseline (χ2 = 6.7; df = 1; p b 0.009).
were the same as in the first regression model. The third In summary, the correlational and regression analyses
analysis was identical to the first model, but the CNCS suggested that a composite neurocognitive score and
at baseline, was entered. The variables that better three specific cognitive domain (verbal memory, motor
predicted functioning one year later were the CNCS speed and vocabulary) at baseline were strongly
and the General PANSS [F (2,46) = 8.9, p b 0.001]. In associated with general functioning one year later in
the fourth regression analysis, the WHODAS at T2 was schizophrenic subjects. In all cases, poorer perfor-
the dependent variable: executive and vocabulary mances on cognitive tests predicted poorer outcomes.
294 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
Symptoms and other clinical factors, as follow-up
predictors, appeared to explain less of the variance in
functioning. Furthermore, the CNCS, motor speed
domain and HSRD score at baseline were associated
with occupational functioning one year later.
3.3.2. Bipolar I patients
3.3.2.1. Correlational analyses. The bivariate correla-
tional analyses are shown in Supplementary Material 2,
which is available with the online version of this article
at http://www.sciencedirect.com. The performance in
three neurocognitive domains (executive, visual/motor
processing and motor speed) at baseline was signifi-
cantly associated with global functional change over one
year, as measured by the GAF (T2–T1). The CNCS at
baseline also showed a positive association with better
general functioning over one year, as measured by the
GAF (T2–T1). Only the verbal memory domain at
baseline showed a positive association with better global
functioning, as measured by the GAF, at T2. Addition-
ally, change in the composite cognition score over one
year, as measured by the CNCS (T2–T1), was
significantly correlated with the WHO-DAS total
score at T2, GAF score at T2 and better general
functioning over one year, as measured by the GAF
(T2–T1). All observed correlations were in the expected
direction, with a higher neurocognitive performance at
T1 associated with a higher functioning change over the
one-year follow-up period. Even more interestingly,
improvement in the composite cognition score over one
year was associated with better functioning at T2 and
over the one-year follow-up period.
Of the clinical symptoms, PANSS Positive score,
PANSS General Score, PANSS Total score, PANSS
Psychotic Dimension and PANSS Excitatory Dimension
at T1 were significantly associated with global function-
ing, as measured by the total score on the GAF, one year
later. HSRD score, PANSS General Score, PANSS Total
score and PANSS Negative Dimension at T1 were
significantly correlated with WHO-DAS total score at
T2. Interestingly, a lower mania symptom rating at T1,
as measured by the CARS-M, was associated with better
general functioning over one year, as measured by the
WHO-DAS (T2–T1). Positive symptoms, as measured
by the PANSS Positive score and the PANSS Cognitive
Dimension were correlated with global functioning over
one year, as measured by the GAF (T2–T1).
With regard to pharmacological factors, the dose of
lithium at T1 showed a positive correlation with WHO-
DAS total score at T2 and over one year, as measured by
the WHO-DAS (T2–T1) in bipolar I patients treated with
lithium (n = 23). In the bipolar I patients treated with
antidepressant drugs (n = 12), the clomipramine-equiva-
lent units correlated with the WHO-DAS total score at
T2. CPZ units at T1 also showed a positive correlation
with WHO-DAS total score at T2 and over one year, as
measured by the WHO-DAS (T2–T1) in bipolar
I patients treated with antipsychotic drugs (n = 20).
Additionally, the number of antipsychotic drugs at T1
was positively correlated with the WHO-DAS total score
at T2. Finally, of the other factors, age and length of illness
at baseline were significantly associated with global
functioning over one year, as measured by the GAF
(T2–T1). Premorbid adjustment at baseline was associated
with the GAF score at T2. Premorbid adjustment was also
correlated with WHO-DAS total score at T2 and over one
year, as measured by the WHO-DAS (T2–T1).
3.3.2.2. Multivariate prediction of functional out-
come. We used four models of linear regression
analyses with hierarchical methods (Table 4). In the
first model, the GAF (T2–T1) score was the dependent
variable. In the first linear regression analysis: execu-
tive, visual/motor processing and motor speed domains
(first block); PANSS Positive score (second block); and
age and length of illness (third block). The variables that
better predicted functioning change over the one-year
follow-up period were visual/motor processing and
PANSS Positive score [F (2,42) = 8.93, p b 0.001]. The
second regression analysis was identical to the first,
except that the clinical variable of the PANSS
Neurocognitive Dimension was entered. The visual/
motor processing domain was the only independent
predictor of GAF (T2–T1). The third analysis was
identical to the first, but the CNCS at baseline, was
entered. The variables that better predicted functioning
change over the one-year were the CNCS and PANSS
Positive score [F (2, 42) = 5.99, p b 0.005].
Finally, the fourth regression analysis was identical
to the first, but the CNCS (T2–T1) was included in the
first block. The variable that best predicted functioning
change over the one-year was the CNCS (T2–T1) [F (1,
42) = 5.43, p b 0.025].
In the second model, the GAF score at T2 was the
dependent variable. In the first linear regression
analysis: CNCS (T2–T1) (first block); PANSS Positive
and General scores (second block); and premorbid
adjustment (third block). The variables that better
predicted functioning were the CNCS (T2–T1) and
Positive PANSS score [F (2,42) = 12.8, p b 0.0001]. The
second regression analysis was identical to the first
analysis, except that the clinical variables of PANSS
Psychotic, Neurocognitive and Excitatory Dimensions
 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299 295

Table 4
Relative contributions of cognition and clinical factors at baseline to explain the variance in outcome measures of bipolar I patients at one-year follow-up
Number of regression analyses Predictors at T1 that Percent of variance explained
and dependent variables entered in the equation (adjusted R2)
Partial (%) Total (%)
a
1° and 2° analyses PANSS Positive score 9′9
GAF (T2–T1) Visual/motor processing b 17′5 27′4
3° analysis CNCS⁎ c 9
GAF (T2–T1) PANSS Positive score d 10′2 19′2
4° analysis GAF (T2–T1) CNCS (T2–T1) e 9′5 9′5
5° analysis CNCS (T2–T1) f 13′1
GAF at T2 PANSS Positive score g 22′8 35′9
6° Analysis CNCS (T2–T1) h 13′1
GAF at T2 Excitatory dimension i 19′5 32′6
7° Analysis CNCS (T2–T1) j 6′8
WHO-DAS at T2 PANSS General score k 10′8
Premorbid adjusted l 13′5 31′1
8° Analysis WHO-DAS at T2–T1 CARS-M Score m 9′5 9′5
⁎CNCS = Composite Neurocognitive Score.
a
[t = − 2.57, β = −0.334 (95% CI = −1.29 to − 0.15), df = 42, p = 0.014].
b
[t = 3.16, β = 0.416 (95% CI = 1.03 to 4.67), df = 42, p = 0.003].
c
[t = 2.21, β = 0.308 (95% CI = 0.32 to 6.98), df = 42, p = 0.032].
d
[t = − 2.48, β = − 0.298 (95% CI = − 1.34 to − 0.14), df = 42, p = 0.017].
e
[t = 2.33, β = 0.342 (95% CI = 1.445 to 20.206), df = 42, p = 0.025].
f
[t = 2.1, β = 0.267 (95% CI = 0.446 to 23.11), df = 42, p = 0.042].
g
[t = − 3.95, β = − 0.503 (95% CI = − 2.265 to − 0.732), df = 42, p = 0.001].
h
[t = 2.46, β = 0.316 (95% CI = 2.49 to 25.33), df = 42, p = 0.018].
i
[t = − 3.59, β = − 0.461 (95% CI = − 13.9 to − 3.9), df = 42, p = 0.001].
j
[t = − 2.02, β = − 0.301 (95% CI = − 5.8 to −0.003), df = 42, p = 0.05].
k
[t = 2.52, β = 0.359 (95% CI = 0.03 to 0.27), df = 42, p = 0.016].
l
[t = 2.96, β = 0.413 (95% CI = 0.07 to 0.38), df = 42, p = 0.005].
m
[t = − 2.33, β = −0.342 (95% CI = −0.36 to − 0.025), df = 42, p = 0.025].

at T1 were entered. The variables that better predicted
functioning were the PANSS Excitatory Dimension and
the CNCS (T2–T1) score [F (2, 42) = 11.2, p b 0.0001].
In the third regression model, the WHODAS score at
T2 was the dependent variable: CNCS (T2–T1) (first
block); HSRD and General PANSS scores (second block);
number of antipsychotic drugs, CPZ-units and amount of
lithium (milligrams) (third block); and premorbid adjust-
ment (fourth block). The variables that better predicted
disability were the CNCS (T2–T1), PANSS General score
and premorbid adjustment [F (2,42) = 7.3, p b 0.001].
In the fourth regression model, the WHODAS score
(T2–T1) was the dependent variable. Here, the CNCS
(T2–T1) (first block); CARS-M score (second block);
CPZ-units and amount of lithium (third block); and
premorbid adjustment (fourth block). The variable that
best predicted disability changes over the one-year was
the CARS-M score [F (2, 42) = 5.2, p b 0.01].
Finally, occupational adaptation was analyzed by
logistic regression analysis. The analysis included the
following variables at baseline: sex, age, years of
education, premorbid adjustment, age of onset, number
of admissions, hand dominance, length of illness, CNCS,
HSRD, CARS-M, PANSS Positive, PANSS Negative,
and PANSS General Scores, number of antipsychotic
drugs, CPZ-units and lithium. This analysis showed that
the CNCS at T1 was the strongest independent predictor
of good or low occupational adaptation group member-
ship at T2 (χ2 = 7.97; df = 1; p b 0.005). Secondly, the
rerunning of the logistic regression model included all
neurocognitive domains, and the specific variable that
best predicted membership of the good or low occupa-
tional adaptation group was executive/reasoning domain
at baseline (χ2 = 8.1; df = 1; p b 0.004).
To summarize this results in bipolar I subjects, the
correlation and regression analyses suggest that changes
in a composite neurocognitive score over one year (CNCS
T2–T1), the speed of processing domain, Positive and
General PANSS scores, Excitatory dimension, CARS-M
score and HDRS score were the variables at T1 that best
predicted functioning or disability changes over the one-
year follow-up period. Furthermore, the CNCS and
executive/reasoning domain at baseline were associated
with occupational adaptation one year later.
296 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
4. Discussion
The findings of this prospective follow-up study of
schizophrenia and bipolar I cohorts showed that
neurocognition plays a relevant role in the level of
functioning in both disorders. It is important, however,
to note that there are specific differences in the pattern of
relationships between neurocognition and clinical
factors at baseline, and the subsequent outcome, for
both diagnostic groups.
4.1. Neurocognition and clinical predictors of functioning
in patients with schizophrenia
In schizophrenia, the variables at baseline that best
predicted the functioning one year later were the com-
posite neurocognitive score, verbal memory and motor
speed. This composite neurocognitive score was the
strongest predictor of good or low occupational adap-
tation group membership at T2. Finally, the variable that
best predicted disability one year later was vocabulary
domain. The present findings confirm the importance of
verbal memory in the prediction of outcome, but differ
from previous studies that have identified executive
functions and verbal fluency as additional domains
predicting functional outcome (Green et al., 2004b). The
lack of studies that measure motor speed may explain
why this is one of the first longitudinal studies that has
found that this neurocognitive domain is a significant
predictor of outcome (Bilder et al., 2000). Recently,
Milev et al. (2005) found that deficits in verbal memory,
processing speed and attention domains, and the
severity of negative symptoms at intake, are predictors
of worse general and specific functioning years later in
first-episode patients. In this study, negative symptoms
at baseline were related to functional outcome, but were
not entered in the regression models. Similar results
were found in two previous cross-sectional studies
(Harvey et al., 1997; Velligan et al., 1997). Moreover,
the narrowness of negative symptom or functioning
definitions may also explain these results (Milev et al.,
2005). For example, the outcome measures used in
the current study belong to the category of community
outcome measures. These measures exhibit a lower
correlation to neurocognitive function measured by
neuropsychological tests (Penn et al., 1995).
In this study, severity of general psychopathology
was the only clinical significant predictor of subsequent
functional outcome, and the depressive variable was the
only clinical predictor of low occupational adaptation
group membership in a logistic regression analysis. The
relevance of the depressive symptoms, as an important
predictor for functional outcome in patients with
schizophrenia, has been shown recently by Siegel
et al. (2006) in a first-episode and follow-up study.
4.2. Neurocognition and clinical predictors of functioning
in patients with bipolar I disorder
This prospective study showed that the neurocognitive
capacity measure as CNCS at baseline or an improvement
in this score over one year predicted changes in the
functioning over the same period. Furthermore, a deficit
in the visual/motor processing domain at baseline was a
significant predictor of the degree of impairment in
functioning over one year. Finally, impairments in the
composite neurocognitive score or executive domain at
baseline increased the probability of belonging to the low
occupational adaptation group at T2. The particularly
effect of the visual-motor processing domain concurs, in
part, with a recent longitudinal study with bipolar patients.
Jaeger et al. (2007) reported that the attentional domain
(specifically, the Trail Making A, Stroop Word reading
and Color naming Tests were included in the domain) was
significantly predictive of functional recovery 12 months
later in 78 bipolar I and II patients. However, our results
are not consistent with some previous cross-sectional
studies that revealed a relationship between verbal
memory deficits and poor psychosocial functioning in
euthymic bipolar patients (Zubieta et al., 2001; Martinez-
Aran et al., 2007). In our prospective study, verbal
memory at baseline was related to functional status one
year later, but did not enter in linear regression analysis.
On the other hand, the clinical variables that best
predicted functioning or disability changes over the one-
year follow-up period were Positive PANSS score and
mania symptoms, respectively. Previous studies also
reported the negative impact of manic episodes and
positive symptoms on the psychosocial functioning of
bipolar patients (Coryell et al., 1993; Tohen et al., 2003;
Jaeger et al., 2007).
4.3. Limitations of the study
These results must be regarded with caution because of
a number of methodological issues. Firstly, the overall
functioning status was assessed using the GAF scale,
which evaluates both symptom severity and social
adjustment. To avoid this limitation, the rater was in-
structed to use the GAF to measure only psychosocial
functioning in the month before rating, as reported in other
studies (Torrent et al., 2006; Martinez-Aran et al., 2007),
because it was expected that six months after sustained
clinical remission would be the best time to assess
 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
functional recovery. Secondly, there is a measurement
problem with the measures of functional status because
they belong to the category of community outcome
measures. Previous data suggest that several measures are
necessary to understand the complexity of functional
outcomes in schizophrenia (Penn et al., 1995). Finally,
although the participants in this study were enrolled from
three different psychiatric outpatient units, and their
symptom severity was similar to the subjects included in
some studies' large samples (Tohen et al., 2003; Green et
al., 2004a), the extent to which we can generalize these
findings is unknown.
4.4. Clinical implications
The results of this prospective study of schizophrenia
and bipolar cohorts suggest that neurocognition may be
more predictive of functional outcome than clinical
factors in both disorders. Moreover, changes in func-
tioning over the follow-up period were associated with
baseline neurocognitive performance and changes in
neurocognition over the same period, but only in bipolar I
subjects. The lack of this effect in schizophrenic subjects
will probably be related to the short span of time between
the two assessments. Furthermore, changes in neurocog-
nition will become more relevant as a predictor of
functional outcome in the first years of this disease (Rund,
1998). Nevertheless, the neurocognitive improvement
may be a longitudinal predictor of functional recovery in
bipolar subjects. Consequently, these findings suggest
that the level of functioning would benefit immediately
from neurocognition-enhancing interventions by neuro-
cognitive remediation programs, or new forms of
medication (Robinson et al., 2006; McGurk et al.,
2007). This is another specific difference with schizo-
phrenic subjects who possibly need the concurrence of
both treatments to improve neurocognition and function-
ing (Green et al., 2004a). On the other hand, the presence
of changes in neurocognitive capacity predicting changes
in functional outcome might indicate a direct or robust
association between these variables in bipolar I subjects.
However, the intervening variables (e.g., social neuro-
cognition) may act between measures of neurocognition
and functional outcome in schizophrenic patients and
make the translation of neurocognitive performance gains
into changes on some type of functional outcome measure
difficult (Green et al., 2004a).
Future follow-up studies, should determine whether
pharmacological and non-pharmacological treatments
are necessary or sufficient to improve neurocognition
and functioning of people who have schizophrenia or
bipolar I disorder.
297
The identification of predictors of general functioning
and change in functional outcome, by longitudinal stud-
ies, could be particularly important both for our basic
understanding of these disorders (e.g., identification of
potential neurocognitive endophenotypes or endopheno-
cognitypes) and to develop operational criteria for long-
term functional recovery that incorporate neurocognition,
symptoms, and functionality (Harrow and Jobe, 2005;
Liberman and Kopelowicz, 2005; Helldin et al., 2006).
Role of funding source
Funding for this study was provided by Spanish FIS-MSC Grant
(PI051293), GV2005-303 and Spanish Ministry of Health; they had no
further role in study design; in the collection, analysis and interpretation
of data; in the writing of the report; and in the decision to submit the
paper for publication.
Disclosure of potential conflict of interest
Dr. Rafael Tabarés-Seisdedos has acted as consultant, advisor, or
speaker for the following companies: AstraZeneca, Eli Lilly and
Company, Janssen Pharmaceutica, Pfizer Inc.
Dr. Eduard Vieta has acted as a consultant, received grants, or been
hired as a speaker by the following companies: Almirall, AstraZeneca,
Bial, Bristol-Myers-Squibb, Eli Lilly, GlaxoSmithKline, Janssen-
Cilag, Lundbeck, Merck Sharp & Dohme, Novartis, Organon, Otsuka,
Pfizer, Sanofi Aventis, Servier, UCB. He has acted as consultant and
has received grants from the Spanish Ministry of Health, Instituto de
Salud Carlos III, RETICS RD06/0011 (REM-TAP) and from the
Stanley Medical Research Institute.
Dr. Vicente Balanzá-Martínez, Mr. José Sánchez-Moreno, Dr.
Anabel Martínez-Arán, Dr. José Salazar-Fraile, Dr. Gabriel Selva-
Vera, Mrs. Cristina Rubio, Dr. Ignacio Mata, Dr. Manuel Gómez-
Beneyto, declare no conflict of interest.
Acknowledgments
This study was supported by the following: Spanish
FIS-MSC Grant (PI051293 and PI050206), GV2005-
303 , the Spanish Ministry of Health, Instituto de Salud
Carlos III, CIBERSAM, and AstraZeneca.
Appendix A. Supplementary data
Supplementary data associated with this article
can be found, in the online version, at doi:10.1016/j.
jad.2007.12.234.
References
Altman, E.G., Hedeker, D.R., Janicak, P.G., Peterson, J.L., Davis, J.M.,
1994. The Clinician-Administered Rating Scale for Mania (CARS-
M): development, reliability, and validity. Biol. Psychiatry 36,
124–134.
American Psychiatric Association, 1994. Diagnostic and Statistical
Manual of Mental Disorders. DSM-IV, 4th ed. APA, Washington, DC.
298 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
Balanza-Martinez, V., Tabares-Seisdedos, R., Selva-Vera, G., Marti-
nez-Aran, A., Torrent, C., Salazar-Fraile, J., Leal-Cercos, C., Vieta,
E., Gomez-Beneyto, M., 2005. Persistent cognitive dysfunctions in
bipolar I disorder and schizophrenic patients: a 3-year follow-up
study. Psychother. Psychosom. 74, 113–119.
Becker, L.A. On-line Program for Cohen's d and r. http://web.uccs.edu
Accessed Dec 2007.
Bilder, R.M., Goldman, R.S., Robinson, D., Reiter, G., Bell, L., Bates,
J.A., Pappadopulos, E., Willson, D.F., Alvir, J.M., Woerner, M.G.,
Geisler, S., Kane, J.M., Lieberman, J.A., 2000. Neuropsychology
of first-episode schizophrenia: initial characterization and clinical
correlates. Am. J. Psychiatry 157, 549–559.
Carlsson, R., Nyman, H., Ganse, G., Cullberg, J., 2006. Neuropsy-
chological functions predict 1- and 3-year outcome in first-episode
psychosis. Acta Psychiatr. Scand. 113, 102–111.
Clark, L., Iversen, S.D., Goodwin, G.M., 2002. Sustained attention
deficit in bipolar disorder. Br. J. Psychiatry 180, 313–319.
Cohen, J., 1988. Statistical Power Analysis for the Behavioural
Science, 2nd ed. Lawrence Erlbaum Associates, Hillsdale, NJ.
Coryell, W., Scheftner, W., Keller, M., Endicott, J., Maser, J., Klerman,
G.L., 1993. The enduring psychosocial consequences of mania and
depression. Am. J. Psychiatry 150, 720–727.
Cuesta, M.J., Peralta, V., Zarzuela, A., 1998. Psychopathological
dimensions and lack of insight in schizophrenia. Psychol. Rep. 83,
895–898.
Daban, C., Martinez-Aran, A., Torrent, C., Tabares-Seisdedos, R., Balanza-
Martinez, V., Salazar-Fraile, J., Selva-Vera, G., Vieta, E., 2006.
Specificity of cognitive deficits in bipolar disorder versus schizo-
phrenia. A systematic review. Psychother. Psychosom. 75, 72–84.
Dickerson, F.B., Sommerville, J., Origoni, A.E., Ringel, N.B., Parente,
F., 2001. Outpatients with schizophrenia and bipolar I disorder: do
they differ in their cognitive and social functioning? Psychiatry
Res. 102, 21–27.
Dickerson, F.B., Boronow, J.J., Stallings, C.R., Origoni, A.E., Cole, S.,
Yolken, R.H., 2004. Association between cognitive functioning
and employment status of persons with bipolar disorder. Psychiatr.
Serv. 55, 54–58.
Endicott, J., Andreasen, N., Spitzer, R., 1978. Family History-
Research Diagnostic Criteria, 3rd ed. New York State Psychiatric
Institute, New York.
First, M.B., Spitzer, R., Gibbon, M., 1997. Structured Clinical
interview for DSM-IV Axis I Disorders, Biometrics Research
Deparment edn. American Psychiatric Press Inc, Washington DC.
Green, M.F., 1996. What are the functional consequences of neurocog-
nitive deficits in schizophrenia? Am. J. Psychiatry 153, 321–330.
Green, M.F., Kern, R.S., Heaton, R.K., 2004a. Longitudinal studies of
cognition and functional outcome in schizophrenia: implications
for MATRICS. Schizophr. Res. 72, 41–51.
Green, M.F., Nuechterlein, K.H., Gold, J.M., Barch, D.M., Cohen, J.,
Essock, S., Fenton, W.S., Frese, F., Goldberg, T.E., Heaton, R.K.,
Keefe, R.S., Kern, R.S., Kraemer, H., Stover, E., Weinberger, D.R.,
Zalcman, S., Marder, S.R., 2004b. Approaching a consensus
cognitive battery for clinical trials in schizophrenia: the NIMH-
MATRICS conference to select cognitive domains and test criteria.
Biol. Psychiatry 56, 301–307.
Hamilton, M., 1960. A rating scale for depression. J. Neurol.
Neurosurg. Psychiatry 23, 56–62.
Harrow, M., Jobe, T.H., 2005. Longitudinal studies of outcome and
recovery in schizophrenia and early intervention: can they make a
difference? Can. J. Psychiatry 50, 879–880.
Harvey, P.D., Powchik, P., Parrella, M., White, L., Davidson, M., 1997.
Symptom severity and cognitive impairment in chronically hospi-
talised geriatric patients with affective disorders. Br. J. Psychiatry
170, 369–374.
Helldin, L., Kane, J.M., Karilampi, U., Norlander, T., Archer, T., 2006.
Remission and cognitive ability in a cohort of patients with
schizophrenia. J. Psychiatr. Res. 40, 738–745.
Jaeger, J., Tatsuoka, C., Berns, S., Varadi, F., Czobor, P., Uzelac, S.,
2006. Associating functional recovery with neurocognitive profiles
identified using partially ordered classification models. Schizophr.
Res. 85, 40–48.
Jaeger, J., Berns, S., Loftus, S., Gonzalez, C., Czobor, P., 2007.
Neurocognitive test performance predicts functional recovery from
acute exacerbation leading to hospitalization in bipolar disorder.
Bipolar Disord. 9, 93–102.
Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative
syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13,
261–276.
Kurtz, M.M., Moberg, P.J., Ragland, J.D., Gur, R.C., Gur, R.E., 2005.
Symptoms versus neurocognitive test performance as predictors of
psychosocial status in schizophrenia: a 1- and 4-year prospective
study. Schizophr. Bull. 31, 167–174.
Laes, J.R., Sponheim, S.R., 2006. Does cognition predict community
function only in schizophrenia?: a study of schizophrenia patients,
bipolar affective disorder patients, and community control subjects.
Schizophr. Res. 84, 121–131.
Liberman, R.P., Kopelowicz, A., 2005. Recovery from schizophrenia:
a concept in search of research. Psychiatr. Serv. 56, 735–742.
Mangell, L.B., Yudofsky, Sc., Silver, J.M., 1999. Antipsychotic drugs,
In: Hales, R.E., Yudofsky, S.C., Talbott, J.A. (Eds.), Textbook of
Psychiatry, ed 3. Washington, American Psychiatric Press, p. 1064.
Martinez-Aran, A., Penades, R., Vieta, E., Colom, F., Reinares, M.,
Benabarre, A., Salamero, M., Gasto, C., 2002. Executive function
in patients with remitted bipolar disorder and schizophrenia and its
relationship with functional outcome. Psychother. Psychosom. 71,
39–46.
Martinez-Aran, A., Vieta, E., Torrent, C., Sanchez-Moreno, J.,
Goikolea, J.M., Salamero, M., Malhi, G.S., Gonzalez-Pinto, A.,
Daban, C., Alvarez-Grandi, S., Fountoulakis, K., Kaprinis, G.,
Tabares-Seisdedos, R., Ayuso-Mateos, J.L., 2007. Functional
outcome in bipolar disorder: the role of clinical and cognitive
factors. Bipolar Disord. 9, 103–113.
McGurk, S.R., Mueser, K.T., Walling, D., Harvey, P.D., Meltzer, H.Y.,
2004. Cognitive functioning predicts outpatient service utilization
in schizophrenia. Ment. Health Serv. Res. 6, 185–188.
McGurk, S.R., Mueser, K.T., Feldman, K., Wolfe, R., Pascaris, A., 2007.
Cognitive training for supported employment: 2-3 year outcomes of a
randomized controlled trial. Am. J. Psychiatry 164, 437–441.
Milev, P., Ho, B.C., Arndt, S., Andreasen, N.C., 2005. Predictive
values of neurocognition and negative symptoms on functional
outcome in schizophrenia: a longitudinal first-episode study with
7-year follow-up. Am. J. Psychiatry 162, 495–506.
Nordenson, B., Gruber, S.A., Yurgelun-Todd, D.A., 2004. Neurocognition
in bipolar disorder: a review of the current research. Curr. Psychosis
Ther. Rep. 2, 147–152.
Penn, D.L., Mueser, K.T., Spaulding, W., Hope, D.A., Reed, D., 1995.
Information processing and social competence in chronic schizo-
phrenia. Schizophr. Bull. 21, 269–281.
Phillips, L., 1953. Case history data and prognosis in schizophrenia.
J. Nerv. Ment. Dis. 117, 515–525.
Robinson, L.J., Thompson, J.M., Gallagher, P., Goswami, U., Young,
A.H., Ferrier, I.N., Moore, P.B., 2006. A meta-analysis of cognitive
deficits in euthymic patients with bipolar disorder. J. Affect. Disord.
93, 105–115.
 R. Tabarés-Seisdedos et al. / Journal of Affective Disorders 109 (2008) 286–299
Rund, B.R., 1998. A review of longitudinal studies of cognitive
functions in schizophrenia patients. Schizophr. Bull. 24, 425–435.
Siegel, S.J., Irani, F., Brensinger, C.M., Kohler, C.G., Bilker, W.B.,
Ragland, J.D., Kanes, S.J., Gur, R.C., Gur, R.E., 2006. Prognostic
variables at intake and long-term level of function in schizophrenia.
Am. J. Psychiatry 163, 433–441.
Sota, T.L., Heinrichs, R.W., 2004. Demographic, clinical, and
neurocognitive predictors of quality of life in schizophrenia
patients receiving conventional neuroleptics. Compr. Psychiatry
45, 415–421.
Tabares-Seisdedos, R., Balanza-Martinez, V., Salazar-Fraile, J., Selva-
Vera, G., Leal-Cercos, C., Gomez-Beneyto, M., 2003. Specific
executive/attentional deficits in patients with schizophrenia or
bipolar disorder who have a positive family history of psychosis.
J. Psychiatr. Res. 37, 479–486.
Tohen, M., Zarate Jr., C.A., Hennen, J., Khalsa, H.M., Strakowski,
S.M., Gebre-Medhin, P., Salvatore, P., Baldessarini, R.J., 2003.
The McLean-Harvard First-Episode Mania Study: prediction of
recovery and first recurrence. Am. J. Psychiatry 160, 2099–2107.
299
Torrent, C., Martinez-Aran, A., Daban, C., Sanchez-Moreno, J.,
Comes, M., Goikolea, J.M., Salamero, M., Vieta, E., 2006.
Cognitive impairment in bipolar II disorder. Br. J. Psychiatry 189,
254–259.
Velligan, D.I., Mahurin, R.K., Diamond, P.L., Hazleton, B.C., Eckert,
S.L., Miller, A.L., 1997. The functional significance of sympto-
matology and cognitive function in schizophrenia. Schizophr. Res.
25, 21–31.
Wing, J.K., Babor, T., Brugha, T., Burke, J., Cooper, J.E., Giel, R.,
Jablenski, A., Regier, D., Sartorius, N., 1990. SCAN. Schedules
for Clinical Assessment in Neuropsychiatry. Arch. Gen. Psychiatry
47, 589–593.
Woods, S.W., 2003. Chlorpromazine equivalent doses for the newer
atypical antipsychotics. J. Clin. Psychiatry 64, 663–667.
World Health Organization, 1988. WHO Psychiatric Disability
Assessment Schedule. WHO/DAS, Geneva.
Zubieta, J.K., Huguelet, P., O'Neil, R.L., Giordani, B.J., 2001.
Cognitive function in euthymic bipolar I disorder. Psychiatry Res.
102, 9–20.