a prospective 3-year longitudinal study of cognitive predictors of relapse in first-episode...
TRANSCRIPT
www.elsevier.com/locate/schres
Schizophrenia Research
A prospective 3-year longitudinal study of cognitive predictors
of relapse in first-episode schizophrenic patients
Eric Yu-Hai Chena,T, Christy Lai-Ming Huia, Eva Lai-Wah Dunnb,
May Yin-King Miaob, Wai-Song Yeungb, Chi-Keung Wongb,
Wah-Fat Chanb, Wai-Nang Tangb
aDepartment of Psychiatry, University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong KongbDepartment of Psychiatry, Pamela Youde Nethersole Eastern Hospital, Hong Kong
Received 1 December 2004; received in revised form 28 February 2005; accepted 28 February 2005
Available online 6 April 2005
Abstract
Background: Cognitive predictors of relapse have been extensively explored only in few long term longitudinal studies of first-
episode schizophrenia.
Method: This study prospectively followed 93 patients with first-episode schizophrenia, schizophreniform disorder, and
schizoaffective disorder for 3 years after their first-episode illness. Cognitive domains including verbal intelligence, verbal and
visual memory, verbal fluency, and Wisconsin Card Sorting Test performance were investigated as potential predictors of
relapse.
Results: We found that by the first year 21% patients had relapsed, by the second year 33% had relapsed, and by the third year
40% had relapsed. There was a significant difference in the relapse rate between patients with good adherence and patients with
poor adherence to medication regimes. A multiple logistic regression analysis revealed that after controlling for medication
adherence, perseverative error in the Wisconsin Card Sorting Test was the only cognitive function that significantly predict
relapse with an odds ratio of 2.4.
Conclusions: Cognitive flexibility in set shifting is related to tendency towards relapse in first-episode schizophrenic patients.
Other cognitive factors appear not to be related to relapse. Possible mechanisms included the link between prefrontal
dysfunction and sub-cortical dopamine system stability, as well as the effects of executive dysfunction on insight impairment
and adherence behavior.
D 2005 Elsevier B.V. All rights reserved.
Keywords: First-episode schizophrenia; Relapse; Predictors; Longitudinal; Cognitive function; Outcome
0920-9964/$ - s
doi:10.1016/j.sc
T Correspondi
E-mail addr
77 (2005) 99–104
ee front matter D 2005 Elsevier B.V. All rights reserved.
hres.2005.02.020
ng author. Tel.: +852 28554488; fax: +852 28551345.
ess: [email protected] (E.Y.-H. Chen).
E.Y.-H. Chen et al. / Schizophrenia Research 77 (2005) 99–104100
1. Introduction
The high relapse rate that is associated with
schizophrenia remains a significant challenge to
clinicians, patients, and their carers. Relapse preven-
tion is one of the key therapeutic goals in the
management of first-episode as well as chronic
patients (Carpenter, 1996). Although it is clear that
maintenance medication has an overall protective
effect against relapse, the long-term use of medication
carries side effects and psychosocial costs. One of the
challenges in relapse prevention has been the identi-
fication of patients who are at a higher risk of relapse.
Predictors of relapse have mostly been studied in
chronic schizophrenic patients, and the predictive
factors that have been suggested include a younger
age, higher baseline neuroleptic dosage, and a shorter
length of hospitalization (Hershon et al., 1972; Prien
et al., 1969; Rassidakis et al., 1970; Zander et al.,
1981; Gilbert et al., 1995). There are few first-episode
studies that address the predictors of relapse. In
particular, few studies address whether cognitive
function could be utilized as potential predictors for
relapse. In a previous naturalistic study of first-
episode patients, several cognitive domain scores
(language, memory, attention, executive, motor and
visuospatial domains) has failed to predict the time to
relapse (Robinson et al., 1999). Naturalistic relapse
rates depend on prevailing clinical trends regarding
maintenance therapy. For instance, this study was
conducted when it was common practice to actively
offer stable patients the option of discontinuing
antipsychotic medication after 1 year of treatment
(Robinson et al., 1999). Likewise, this study followed
patients for different periods of time, and then used
statistical adjustment to estimate a cumulated relapse
rate. So far the finding has not been replicated.
In the current study, we investigated potential
cognitive predictors of relapse in a naturalistic sample
of first-episode psychosis patients in the 3 years after
their initial psychotic episode. Patients were studied in
a prospective manner with a detailed clinical and
cognitive assessment at baseline and then received
regular assessments during the follow-up period. The
patients in the study were followed up for the same
period of time (3 years after the initial episode) so that
statistical adjustments to estimate survival were not
needed. This study was carried out in a setting in which
clinicians do not spontaneously offer the option of
stopping maintenance medication to patients. Instead,
patients were generally encouraged to carry on with
maintenance medication for a longer period of time.
2. Method
2.1. Sample
Consecutively diagnosed first-episode patients with
schizophrenia, schizophreniform disorder, and schizo-
affective disorder were recruited from the catchment
area in Hong Kong with a population of around 1.3
million. Patients between 18 and 55 years of age who
presented to the outpatient and inpatient psychiatric
units were initially screened for the presence of
psychotic symptoms and subsequently received a
diagnostic assessment from the investigators. Patients
were not included if they had had a previous psychotic
episode (whether treated or not), had a known neuro-
logical condition, significant substance abuse prob-
lems or if there was a history of special school
attendance (usually signifies the presence of a moder-
ate to severe learning disability). The patients in this
study were initially treated with low-dose conventional
anti-psychotics (less than 5 mg of haloperidol or
equivalent). The study was approved by the relevant
institutional review board, and all subjects gave
written informed consent before participation.
The patients were assessed at the point of first
contact and after stabilization of the first psychotic
episode (a mean of 42.6 days after the initial assess-
ment). Subsequent clinical assessments were carried
out every four months for 3 years to determine whether
the patients had had a relapse in the interim period.
2.2. Assessments
2.2.1. Diagnostic assessments
Diagnoses were made according to the DSM-IV
criteria (American Psychiatric Association, 1994)
based on clinical interviews, informant histories, and
medical records. An inter-rater agreement of 86% for
diagnosis was obtained in an independent validation
sample of 38 cases. A project diagnosis was made
after 3 years taking into consideration diagnostic
reviews undertaken at the end of each year in the
E.Y.-H. Chen et al. / Schizophrenia Research 77 (2005) 99–104 101
follow-up period. This approach allowed for change
from the initial diagnosis in the early course of the
illness (Ram et al., 1992).
2.2.2. Cognitive assessments
Forward digit span was rated according to the
standard procedures of the Wechsler Adult Intelligence
Scale WAIS-R-HK (Hong Kong Psychological Soci-
ety, 1989) (Revised Cantonese Version, Hong Kong
Psychological Society, 1989). The sequence started
from 3 digits and increased in length until the subject
was unable to recall the sequence correctly. Executive
dysfunction was assessed by the Modified Wisconsin
Card Sorting Test (MWCST) (Nelson, 1976). Subjects
were asked to sort each of the 64 cards into the correct
category according to a sorting rule which could be
color, shape, or number. After 6 consecutive correct
responses, a change of rule occurred. The number of
perseverative errors was scored. Logical Memory and
Visual Reproduction were performed as described in
the Wechsler Memory Scale Revised (Hong Kong
Psychological Society, 1989) (adapted for Cantonese
speaking patients, C.W. Wong, personal communica-
tion). In the Logical Memory test, subjects were
requested to memorize and recall short narratives for
30 min. The Visual Reproduction subscale of theWMS
involved memorizing abstract line figures. General
verbal intelligence was estimated using the information
subscale from the Wechsler Adult Intelligence Scale
WAIS-R-HK (Hong Kong Psychological Society,
1989) (Revised Cantonese Version, Hong Kong
Psychological Society, 1989). Semantic fluency was
assessed by requesting the patient to name as many
exemplars as possible from the danimalT category in 1
min. A total score was computed by counting the
number of correct items produced (repeated items and
items clearly outside the category were not counted).
2.2.3. Clinical assessments
The symptoms were assessed using the Positive
and Negative Symptom Scale (PANSS) (Kay et al.,
1988) based on clinical interviews and medical
records. The intra-class correlation coefficients were
0.83 for the PANSS positive symptoms subscale and
0.73 for the PANSS negative symptoms subscale. In
addition to the clinical assessment, the investigator
indicated for every visit whether the patient was
receiving maintenance medication, and assessed the
medication adherence based on the accounts of the
patients and informants. Patients were either consid-
ered to be having satisfactory adherence (taking more
than 70% of prescribed medication) or significant
non-adherence (taking less than 70% of prescribed
medication). Relapse was defined by a significant
deterioration in positive symptoms (hallucinations,
delusions, and language disorganization) that led to a
change in pharmacotherapy or hospitalization, and
was assessed for the interim period every 4 months.
2.3. Data analysis
Statistical analyses were carried out using the
Statistical Package for Social Sciences (SPSS) version
12.0. We first described the relapse rate and the
cumulative relapse rate for each year. The relapse
rates of those who were considered to have good
medication adherence were compared with the relapse
rate of those who were judged to have poor adherence.
Patients were then divided into two groups according
to whether they had experienced a relapse in the first 3
years after the first-episode. A multiple binary logistic
regression model was used to determine the relative
contribution of the various cognitive factors towards
relapse over the 3 years. The dependent variable was
whether the patient had a relapse in the 3 year period.
The independent variables were standardized Z-scores
for cognitive function scores. Medication adherence
and symptoms were also entered into the regression
model so that they could be taken into account when
the effects of the cognitive variables were estimated.
The model chi-square and the Hosmer and Lemeshow
Goodness-of-Fit Test were used to evaluate the extent
to which the estimates of the regression model fitted
the data. The Nagelkerke R-square was presented to
indicate the overall strength of association. A classi-
fication table was calculated to summarize the
proportion of cases that were correctly categorized
by the model.
3. Results
3.1. Sample characteristics
One hundred and fifty-three first-episode psychosis
patients were initially recruited into the study. At the
E.Y.-H. Chen et al. / Schizophrenia Research 77 (2005) 99–104102
end of the 3-year follow-up, a longitudinal diagnoses
review suggested that 138 patients had fulfilled the
diagnostic criteria for schizophrenia, schizoaffective
disorder, or schizophreniform disorder. In this sample,
5 patients were deceased, and of the remaining
patients, 93 completed follow-up for 3 years (a
retention rate of 70%). There were 42 men and 51
women, and their mean age was 31.2 (S.D.=9.6)
years. The average educational level was 10.54 years
(S.D.=2.9). The DSM-IV diagnoses for the sample
were schizophrenia (n =75), schizophreniform disor-
der (n =13), and schizoaffective disorder (n =5).
Forty-eight patients were first assessed in a medica-
tion-naRve state, and the rest were assessed within 7
days of starting medication. At clinical stabilization,
the mean dosage of antipsychotic medication was 349
mg chlorpromazine equivalence per day (median 281
mg). The mean dosage of anticholinergic was 3.4 mg
of benzhexol per day (median 4 mg). Two patients
were on risperidone. None of the patients was on
olanzapine or clozapine. Eight patients were taking
depot antipsychotics. Eight patient received concom-
ittent antidepressants and 7 patients received benzo-
diazepines. At the end of the 3-year follow-up period
the mean dosage of antipsychotic medication was 233
mg chlorpromazine equivalence per day (median 150
mg). The mean dosage of anticholinergic was 2.4 mg
of benzhexol per day (median 0 mg). Eleven patients
were on risperidone, 5 patients were on olanzapine,
and 2 patients were taking clozapine. Fifteen patients
were receiving depot injections. Twelve patients
received concomittent antidepressants and 12 patients
received benzodiazepines.
3.2. Relapse in the 3 years after first-episode
psychosis
Sixty percent of patients (n =55) experienced no
relapse in the 3 years of the study. In the first year,
21% had relapsed (n =19). By the end of the second
year, altogether 33% had relapsed (n =31), and by the
end of the third year the cumulative relapse rate was
40% (n =37). Of these, 27% (n =25) had experienced
a single relapse and 13% (n =12) had experienced two
or more relapses. We then divided the patients into
two groups: those with good and those with poor
medication adherence. The cumulative proportions of
relapse in patients with good adherence were 18%
(Year 1), 29% (Year 2) and 36% (Year 3) and relapsed
patients with poor adherence were 29% (Year 1), 42%
(Year 2) and 57% (Year 3). The relapse rate was
significantly higher among patients with poor adher-
ence (Fisher’s exact test, p b0.001).
Further analysis was also carried out restricting to
patients who had a diagnosis of schizophrenia
(n =46). In the first year, 24% had relapsed (n=11).
By the second year 39% had relapsed (n=18), and by
the third year 43% had relapsed (n =20). These rates
were not significantly different from those of patients
with diagnoses of schizoaffective disorder and schiz-
ophreniform disorders (chi-square tests, ns). When the
schizophrenic patients were divided into good and
poor medication adherence groups, we found that by
the first year the relapse rate of patients with good
adherence was 20% and for those with poor adherence
was 36%. By the second year, the relapse rate for
patients with good medication adherence was 31%,
but for those with poor adherence the relapse rate was
64%. At the end of the 3 years, the cumulative relapse
rate for patients with good adherence was 37%, but it
remained 64% for those with poor adherence (Fisch-
er’s exact test, p =0.005).
3.3. Logistic regression of relapse over the 3 years
We employed a logistic regression model to study
the effects of cognitive function performance on
relapse. To allow for possible confounding variables,
a multiple logistic regression model was estimated. In
this model, the independent variable was standardized
Z-scores of cognitive performances (logical memory,
visual reproduction, information, forward digit span,
semantic fluency and perseverative errors of WSCT).
We also included some clinical variables so that they
are accounted for in the model (PANSS positive and
negative symptoms, and medication non-adherence).
The dependent variable was whether patients had a
relapse in 3 years.
The regression model had a chi-square of 24.12
( p =0.004). The Hosmer and Lemeshow Goodness of
Fit Test give a chi-square of 8.29 ( p =0.406). These
suggested that the estimates of the model fitted the
data to an acceptable level. The R-square (Nagel-
kerke) for the regression model was 0.32. The model
correctly classified 90.2% of patients who had not
relapsed and 56.8% of patients who had relapsed
Table 1
Multiple logistic regression model of relapse in the 3 years following first-episode schizophrenia
Cognitive and clinical variables Significance Adjusted odds ratios (OR) 95% Confidence intervals for adjusted OR
Lower Upper
Logical memory 0.342 0.76 0.43 1.34
Visual reproduction 0.433 1.25 0.72 2.18
Information 0.155 1.50 0.86 2.64
Forward digit span 0.248 1.42 0.78 2.56
Semantic fluency 0.400 1.26 0.74 2.13
Perseverative error in WCST 0.027 2.46 1.11 5.45
PANSS positive symptoms 0.617 1.15 0.67 1.99
PANSS negative symptoms 0.543 0.85 0.50 1.44
Medication non-adherence 0.002 7.59 2.12 27.22
Constant 0.001 0.06
E.Y.-H. Chen et al. / Schizophrenia Research 77 (2005) 99–104 103
(overall percentage 76.1%). WCST perseverative
errors and medication non-adherence were significant
predictors of relapse, with odds ratios of 2.46 and 7.59
respectively (Table 1). None of the other cognitive
functions significantly predicted relapse.
We also tested a logistic regression model with
medication adherence as the dependent variable and
the cognitive functions as the independent variables.
None of the cognitive functions was found to be
significant in predicting adherence behavior.
4. Discussion
Our results show that in first-episode patients with
schizophrenia, the presence of executive dysfunction
as reflected in perseverative errors in the Wisconsin
Card Sorting Test is associated with an increased risk
of relapse in the subsequent 3 years. This is in contrast
to visual and verbal memory, measures of verbal
intelligence, and verbal fluency, which do not
significantly predict relapse. Not unexpectedly med-
ication non-adherence is the strongest predictor of
relapse with an odds ratio of 7.6. The finding that
perseverative error is a significant predictor of relapse
in a multiple logistic regression model suggests that it
is important after controlling for the presence of other
cognitive and clinical variables including medication
adherence. Perseverative error has previously been
found to be related to insight impairment (Chen et al.,
2001; Koren et al., 2004; Lysaker et al., 2003; Rossell
et al., 2003), as well as to the likelihood of
discontinuation of maintenance medication after a
relapse (Robinson et al., 2002). It is possible that the
relationship between cognitive set shifting (persever-
ative errors) and relapse are mediated through insight
impairment and adherence behavior. However further
analysis did not find that perseverative error was
directly related to non-adherence in our sample. This
raises the possibility that perseveration may be related
to relapse through mechanisms other than non-
adherence.
Prefrontal cortical activity has been suggested to
modulate the activity of dopamine neurons in the
ventral tegmental area and this relationship has been
suggested to account for the co-existence of positive
symptoms, negative symptoms, and cognitive impair-
ments in schizophrenia (Abi-Dargham, 2004; Brake et
al., 2000; Deutch, 1992; Grace, 2000). Since dop-
amine activity has been suggested to be related to the
proneness to relapse (Lieberman et al., 1987, 1994),
the relationship between deficits in prefrontal cortical
function and increased subcortical dopamine activity
provides a possible account for the association
between perseveration in the Wisconsin Card Sorting
Test and relapse observed in the current study. This
hypothesis could be further tested through functional
and tracer imaging studies.
In this study, we investigated the potential role of
cognitive performance as potential predictors of
relapse in a sample of first-episode psychosis patients
in Hong Kong who were followed up prospectively
for 3 years. It was found that among different
cognitive dysfunctions, perseveration signified an
increased risk of relapse. Whether this relationship is
mediated by cognitive consequences of difficulties in
set shifting, or whether it reflects an underlying
biological trait linked to relapse is a question to be
E.Y.-H. Chen et al. / Schizophrenia Research 77 (2005) 99–104104
resolved in future studies. Effective relapse prevention
efforts could target patients with this risk profile for
education and monitoring. However, it is also
recognized that although a cognitive risk factor of
relapse have been identified, a highly specific
prediction of relapse for individual patients is still
not yet feasible.
Acknowledgements
This work was supported by grant 21500.10202404
from the Research Grants Council of Hong Kong. We
are thankful to Ms Ka Yuet Liu, University of Hong
Kong for her discussions over data analysis. We are
also grateful to the individuals who participated in the
study.
References
Abi-Dargham, A., 2004. Do we still believe in the dopamine
hypothesis? New data bring new evidence. International Journal
of Neuropsychopharmacology 7 (Suppl. 1), S1–S5.
American Psychiatric Association, 1994. DSM-IV: Diagnostic and
Statistical Manual of Mental Disorders, 4th edition. American
Psychiatric Association, Washington, DC.
Brake, W.G., Flores, G., Francis, D., Meaney, M.J., Srivastava,
L.K., Gratton, A., 2000. Enhanced nucleus accumbens dop-
amine and plasma corticosterone stress responses in adult rats
with neonatal excitotoxic lesions to the medial prefrontal cortex.
Neuroscience 96, 687–695.
Carpenter, W.T. Jr., 1996. Maintenance therapy of persons with
schizophrenia. Journal of Clinical Psychiatry 57 (Suppl. 9),
10–18.
Chen, E.Y., Kwok, C.L., Chen, R.Y., Kwong, P.P., 2001. Insight
changes in acute psychotic episodes: a prospective study of
Hong Kong Chinese patients. Journal of Nervous and Mental
Disease 189, 24–30.
Deutch, A.Y., 1992. The regulation of subcortical dopamine systems
by the prefrontal cortex: interactions of central dopamine
systems and the pathogenesis of schizophrenia. Journal of
Neural Transmission. Supplementum 36, 61–89.
Gilbert, P.L., Harris, M.J., McAdams, L.A., Jeste, D.V., 1995.
Neuroleptic withdrawal in schizophrenic patients. A review of
the literature. Archives of General Psychiatry 52 (3), 173–188.
Grace, A.A., 2000. Gating of information flow within the limbic
system and the pathophysiology of schizophrenia. Brain
Research Reviews 31, 330–341.
Hershon, H.I., Kennedy, P.F., McGuire, R.J., 1972. Persistence of
extra-pyramidal disorders and psychiatric relapse after with-
drawal of long-term phenothiazine therapy. British Journal of
Psychiatry 120 (554), 41–50.
Hong Kong Psychological Society, 1989. The Wechsler Adult
Intelligence Scale-Revised (Cantonese version). Hong Kong
Psychological Society, Hong Kong.
Kay, S.R., Fiszbein, A., Opler, L.A., 1988. Positive and negative
symptom scale (PANSS) for schizophrenia. Schizophrenia
Bulletin 13, 21–76.
Koren, D., Seidman, L.J., Poyurovsky, M., Goldsmith, M., Viks-
man, P., Zichel, S., Klein, E., 2004. The neuropsychological
basis of insight in first-episode schizophrenia: a pilot meta-
cognitive study. Schizophrenia Research 70, 195–202.
Lieberman, J.A., Kane, J.M., Sarantakos, S., Gadaleta, D., Woerner,
M., Alvir, J., Ramos-Lorenzi, J., 1987. Prediction of relapse in
schizophrenia. Archives of General Psychiatry 44 (7), 597–603.
Lieberman, J.A., Alvir, J., Geisler, S., Ramos-Lorenzi, J., Woerner,
M., Novacenko, H., Cooper, T., Kane, J.M., 1994. Methylphe-
nidate response, psychopathology and tardive dyskinesia as
predictors of relapse in schizophrenia. Neuropsychophar-
macology 11, 107–118.
Lysaker, P.H., Lancaster, R.S., Davis, L.W., Clements, C.A., 2003.
Patterns of neurocognitive deficits and unawareness of illness
in schizophrenia. Journal of Nervous and Mental Disease 191,
38–44.
Nelson, H.E., 1976. A modified card sorting test sensitive to frontal
lobe defects. Cortex 12, 313–324.
Prien, R.F., Cole, J.O., Belkin, N.F., 1969. Relapse in chronic
schizophrenics following abrupt withdrawal of tranquillizing
medication. British Journal of Psychiatry 115 (523), 679–686.
Ram, R., Bromet, E.J., Eaton, W.W., Pato, C., Schwartz, J.E.,
1992. The natural course of schizophrenia: a review of first-
admission studies [Review] [98 refs]. Schizophrenia Bulletin
18 (2), 185–207.
Rassidakis, N.C., Kondakis, X., Papanastassiou, A., Michalakeas,
A., 1970. Withdrawal of antipsychotic drugs from chronic
psychiatric patients. Bulletin of the Menninger Clinic 34 (4),
216–222.
Robinson, D., Woerner, M.G., Alvir, J.M., Bilder, R., Goldman, R.,
Geisler, S., Koreen, A., Sheitman, B., Chakos, M., Mayerhoff,
D., Lieberman, J.A., 1999. Predictors of relapse following
response from a first episode of schizophrenia or schizoaffective
disorder. Archives of General Psychiatry 56 (3), 241–247.
Robinson, D.G., Woerner, M.G., Alvir, J.M., Bilder, R.M.,
Hinrichsen, G.A., Lieberman, J.A., 2002. Predictors of
medication discontinuation by patients with first-episode
schizophrenia and schizoaffective disorder. Schizophrenia
Research 57 (2-3), 209–219.
Rossell, S.L., Coakes, J., Shapleske, J., Woodruff, P.W., David,
A.S., 2003. Insight: its relationship with cognitive function,
brain volume and symptoms in schizophrenia. Psychological
Medicine 33, 111–119.
Zander, K.J., Fischer, B., Zimmer, R., Ackenheil, M., 1981. Long-
term neuroleptic treatment of chronic schizophrenic patients:
clinical and biochemical effects of withdrawal. Psychopharma-
cology 73 (1), 43–47.