ORIGINAL CONTRIBUTION
Aripiprazole Versus Haloperidol in Combination With
Clozapine for Treatment-Resistant Schizophrenia
in Routine Clinical Care
A Randomized, Controlled Trial
Corrado Barbui, MD,* Simone Accordini, MSc,Þ Michela Nosè, MD,* Scott Stroup, MD,þ
Marianna Purgato, PsycholD,* Francesca Girlanda, PsycholD,* Eleonora Esposito, MD,*
Antonio Veronese, PsycholD,* Michele Tansella, MD,* Andrea Cipriani, MD,*
and the CHAT (Clozapine Haloperidol Aripiprazole Trial) Study Group
Abstract: This multisite study was conducted to compare the efficacy
and tolerability of combination treatment with clozapine plus aripiprazole
versus combination treatment with clozapine plus haloperidol in patients
with schizophrenia who do not have an optimal response to clozapine.
Patients continued to take clozapine and were randomly assigned to
receive daily augmentation with aripiprazole or haloperidol. Physicians
prescribed the allocated treatments according to usual clinical care.
Withdrawal from allocated treatment within 3 months was the primary
outcome. Secondary outcomes included severity of symptoms on the
Brief Psychiatric Rating Scale and antipsychotic subjective tolerability
on the Liverpool University Neuroleptic Side Effect Rating Scale. A total
of 106 patients with schizophrenia were randomly assigned to treatment.
After 3 months, we found no difference in the proportion of patients who
discontinued treatment between the aripiprazole and haloperidol groups
(13.2% vs 15.1%, P = 0.780). The 3-month change of the Brief Psy-
chiatric Rating Scale total score was similar in the aripiprazole and halo-
peridol groups (j5.9 vs j4.4 points, P = 0.523), whereas the 3-month
decrease of the Liverpool University Neuroleptic Side Effect Rating Scale
total score was significantly higher in the aripiprazole group than in the
haloperidol group (j7.4 vs j2.0 points, P = 0.006). These results suggest
that augmentation of clozapine with aripiprazole offers no benefit with
regard to treatment withdrawal and overall symptoms in schizophrenia
compared with augmentation with haloperidol. However, an advantage in
the perception of adverse effects with aripiprazole treatment may be
meaningful for patients.
From the *World Health Organization Collaborating Centre for Research
and Training in Mental Health and Service Organization, Department of
Public Health and Community Medicine, Section of Psychiatry and Clinical
Psychology, †Department of Public Health and Community Medicine, Section
of Epidemiology and Medical Statistics, University of Verona, Verona, Italy;
and ‡Columbia University College of Physicians and Surgeons, New York
State Psychiatric Institute, New York, NY.
Received June 25, 2010; accepted after revision February 11, 2011.
Reprints: Corrado Barbui, MD, Department of Public Health and Community
Medicine, Section of Psychiatry and Clinical Psychology, University
of Verona, Policlinico GB Rossi, Piazzale Scuro 10, 37134 Verona, Italy
(e-mail: corrado.barbui@univr.it).
The authors thank Fondazione Cariverona, who provided a 3-year grant to
the WHO Collaborating Centre for Research and Training in Mental
Health and Service Organization at the University of Verona, directed by
Professor Michele Tansella.
ClinicalTrials.gov identifier: NCT00395915.
Web page of the University of Verona, Section of Psychiatry and Clinical
Psychology: http://www.psychiatry.univr.it.
Supplemental digital content is available for this article. Direct URL citation
appears in the printed text and is provided in the HTML and PDF versions
of this article on the journal’s Web site (www.psychopharmacology.com).
Copyright * 2011 by Lippincott Williams & Wilkins
ISSN: 0271-0749
DOI: 10.1097/JCP.0b013e318219cba3
266 www.psychopharmacology.com Journal of Clinical Psychopharmacology
Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Key Words: antipsychotics, clozapine, schizophrenia, aripiprazole,
combination strategies
(J Clin Psychopharmacol 2011;31: 266Y273)
A pproximately one fifth to one third of patients with schizo-
phrenia derives little or no benefit from monotherapy with
first-line antipsychotics.1 In these treatment-refractory patients,
clozapine has been shown to be the treatment of choice.2Y4
Nevertheless, approximately one third of treatment-refractory
patients have persistent positive and negative symptoms despite
clozapine monotherapy of adequate dosage and duration.2,3 In
these patients, partially responsive to clozapine, augmentation
with haloperidol or other antipsychotic drugs is one of the most
frequently therapeutic options used in clinical practice, although
the background evidence is limited and contradictory.5,6
Despite this paucity of positive results, the need to provide
real-world suggestions for patients who do not have an optimal
response to clozapine has prompted European and American
treatment guidelines to recommend the concurrent prescription
of a second antipsychotic in addition to clozapine in partially
responsive patients, with no indication on which agent should
be prescribed. In the present randomized study, we therefore
compared the relative effectiveness and tolerability of 2 cloza-
pine combination strategies, namely clozapine and aripiprazole,
a recent therapeutic option with promising data in combination
with clozapine,7 versus clozapine and haloperidol, a reference
therapeutic option often used under ordinary circumstances.8
METHODS
Study Design
The Clozapine Haloperidol Aripiprazole Trial (CHAT) is a
multicenter randomized superiority trial. Its rationale, design,
and methods have been described previously.9 The study was
conducted within the framework of a legislation (Decreto Min-
isteriale, December 17, 2004) that regulates the conduct of in-
dependent pragmatic trials in Italy. According to this law, drug
costs were paid by the Italian National Health System, and fees
for submitting the study protocol to the local ethics committees
were waived. Additional support was provided by the World
Health Organization Collaborating Centre for Research and
Training in Mental Health and Service Evaluation, University
of Verona. No support from drug companies was received.
Neither investigators nor patients received any payment or
reimbursement.
& Volume 31, Number 3, June 2011
Journal of Clinical Psychopharmacology & Volume 31, Number 3, June 2011
Consecutive patients with an incomplete response to treat-
ment with clozapine during an appropriate period were randomly
assigned to augmentation with aripiprazole or haloperidol. Pa-
tients were assessed at baseline and after 3, 6, and 12 months
of follow-up. During the study, patients and clinicians were
not blind to pharmacological treatments provided during the
trial. However, outcome assessments based on rating scales were
performed by trained assessors masked to the allocated treat-
ment. The CHAT was undertaken in compliance with Good
Clinical Practice guidelines and the Declaration of Helsinki. The
study protocol was approved by the Italian Medicine Agency
(Agenzia Italiana del Farmaco) and received ethical approval
in each participating site. All phases of CHAT were recorded
following the Consolidated Standard of Reporting of Trials
statement.10,11
Participants
A total of 34 clinical sites across Italy agreed to actively
participate in the study. All sites were community psychiatric
facilities belonging to the Italian National Health System with
a very similar organization of care, and a few of these were
also academic sites with teaching and research responsibili-
ties. Participants meeting the inclusion/exclusion criteria were
recruited during a 28-month period (from September 1, 2006,
to December 31, 2008). Both inpatients and outpatients were
eligible. Participants were included if the following inclusion
criteria were met: (1) currently undergoing treatment with clo-
zapine for the primary indication of schizophrenia (clinical di-
agnosis, guided by the Diagnostic and Statistical Manual of
Mental Disorders, 4th Edition, criteria); (2) undergoing treat-
ment with clozapine for at least 6 months at a stable dose of
400 mg or more per day, unless the size of the dose was limited
by adverse effects; (3) patient has unsatisfactory benefit from
clozapine treatment, as indicated by the persistent presence of
positive symptoms (delusions, hallucinations); (4) when the pa-
tient considered it clinically reasonable to try the combination
treatment with clozapine and aripiprazole or with clozapine and
haloperidol; (5) uncertainty about which trial treatment would
be best for the participant; (6) no medical disorder or condition
contraindicates either of the investigational drugs; (7) agreement
between investigator and patient to discontinue any antipsychotic
drugs other than clozapine (including long-acting antipsychotic
drugs); (8) aged 18 years and older; (9) agreement between inves-
tigator and patient to enter the study; and (10) being a resident
of Italy.
Treatments
To resemble everyday clinical practice, clinicians were al-
lowed to prescribe the allocated pharmacological treatments
(starting dose and dose changes) according to clinical status and
circumstances. No operational rules or previously agreed titra-
tion schedules were to be followed. All dose changes were
recorded. After randomization, treatment had to be taken daily
for 1 year unless some clear reason to stop developed. Before
random allocation, patients were asked to discontinue any anti-
psychotic drugs other than clozapine. Long-acting antipsychotic
drugs needed to be discontinued for at least 2 weeks before
random allocation. Any other concomitant medications were
permitted, including antidepressants, benzodiazepines, mood sta-
bilizers, and any other nonpsychiatric drugs. Routine care outside
the trial continued as usual. During the study, participants were
seen as often as clinically indicated with no extra visits required
for the trial.
* 2011 Lippincott Williams & Wilkins
Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clozapine Combination Strategies in Schizophrenia
Assessments
Before entering the study, patients were asked to provide
written informed consent. Clinical and demographic character-
istics were collected at baseline using a recruitment form (RF).
The RF included sociodemographic and clinical information,
diagnostic information according to the Mini Neuropsychiatry
Interview (MINI) criteria for schizophrenia,12,13 severity of ill-
ness information according to the Brief Psychiatric Rating Scale
(BPRS, 24 items),14 and information on the subjective toler-
ability of antipsychotic drugs (using the Liverpool University
Neuroleptic Side Effect Rating Scale [LUNSERS]).15 The BPRS
consists of 24 items measuring the following dimensions:
positive symptoms, negative symptoms, depression/anxiety, and
disorganization. All investigators received training to use this
rating scale; however, no formal interrater reliability training was
carried out. Assessors were different from the treating physicians
and blind to treatment allocation. The LUNSERS is a self-rated
interview consisting of 51 items that produces a total score that
indicates the burden of adverse effects as perceived by patients
(subjective tolerability). Information on clozapine treatment was
gathered, as well as laboratory and electrocardiography (ECG)
parameters. All laboratory and ECG tests were performed ac-
cording to local usual care. The CHAT did not require additional
or specific patient monitoring. The RF was administered by the
treating clinician before random allocation. Follow-up data were
obtained at 3, 6, and 12 months using a follow-up form (FUF).
The FUF included information on pharmacological treatments
received, severity of illness (BPRS 24), subjective tolerability of
antipsychotic drugs (LUNSERS), and laboratory and ECG par-
ameters. The FUF was additionally completed at any time dur-
ing the 12-month follow-up in case of withdrawal from allocated
combination treatment. Adverse drug reactions and adverse events
were recorded using an ad hoc form.
Randomization
Patients were randomly assigned to 1 of the 2 treatment
groups with an equal probability of assignment to each treatment
(allocation ratio, 1:1). A centralized randomization procedure
was used. The trial biostatistician prepared the sequence of
treatments randomly permuted in blocks of constant size. The
site investigators did not know the block size. The allocation was
stratified by living condition (residential facility vs all the other
living conditions) because this variable was considered a proxy
of severity of illness. The randomization schedule was generated
using STATA software (StataCorp, College Station, Tex).
Allocation Concealment
Patients were randomly allocated by telephone. Recruiting
physicians were asked to contact an administrator at the World
Health Organization Collaborating Centre in Verona (coordi-
nating site) who accessed a computerized system that provided,
after information on the enrolled participant was entered, the
patient’s identification number and the allocated treatment. The
administrator had no access to the randomization lists. This pro-
cedure of randomization was developed to fully conceal treatment
allocation.16
Outcomes
Withdrawal from allocated treatment within 3 months was
the primary outcome. This outcome was selected because stop-
ping or changing antipsychotic combination treatment is a fre-
quent occurrence and major problem in the treatment of patients
with schizophrenia. Pragmatically, combination treatment was
considered withdrawn if (a) clozapine was continued and the
allocated treatment stopped, (b) clozapine was stopped and the
www.psychopharmacology.com 267
Barbui et al Journal of Clinical Psychopharmacology & Volume 31, Number 3, June 2011

allocated treatment continued, (c) both clozapine and the allo-
cated treatment were stopped, and (d) other antipsychotic drugs
were added on a regular basis to the allocated combination
treatment. Combination treatment was not considered withdrawn
if (a) other antipsychotic drugs were occasionally adminis-
tered for emergency purposes (eg, parenteral antipsychotic drug
administration during accident and emergency admission) and
(b) antipsychotic treatment was temporarily stopped (for no
92 weeks in 6 months) for reasons not related to clinical status.
Severity of illness was measured by means of the BPRS 24,14
and the perspective of patients exposed to antipsychotic agents
by means of the LUNSERS.17 In this study report, only 3 months
of outcome data are reported (primary outcome).
Power Analysis for Sample Size Calculation
At the time that CHAT was designed, only 1 antipsychotic
trial used discontinuation by any cause as the primary end
point.18 On the basis of this trial, a withdrawal proportion from
allocated treatment within 3 months (primary study end point) of
25% in the group treated with clozapine plus haloperidol (con-
trol group) was initially hypothesized. Moreover, it was hypoth-
esized that the augmentation with aripiprazole (experimental
group) would show a clinically significant advantage by pro-
ducing a withdrawal proportion of 10%. A sample size of 194
patients (97 in each group) was chosen because it achieves 80%
power to detect a difference of 15% between the 2 withdrawal
proportions. The test statistic used was the 2-sided Z test with
pooled variance. The significance level of the test was targeted
at 5%. Having assumed that 10% of the participants could be
lost within 3 months, or could not provide valid data at month 3,
the target total sample size for CHATwas 216 (=194/0.90) patients
to obtain 194 evaluable subjects.19Y21 The sample size calculation
was performed using PASS software.22
Statistical Analysis
All randomized participants who received at least 1 dose of
the investigational drugs were included in the intention-to-treat
analysis of the primary outcome. The distribution of the socio-
demographic, biometric, functional, and clinical characteristics
of the patients evaluated at baseline and drugs utilization in the
past and at randomization were compared using the Pearson W2
test, the Fisher exact test, the Student t test, and the Wilcoxon
rank sum test, as appropriate. No correction for multiple testing
was performed.
The BPRS total score (at baseline and at month 3) was
computed as the sum of the scores obtained from the 24 items
measuring positive/negative symptoms, depression/anxiety, and
disorganization. The LUNSERS total score (at baseline and
at month 3) was computed as the sum of the scores obtained
from the 39 and 41 items for males and females, respectively,
covering psychological, neurological, autonomic, hormonal,
and other miscellaneous adverse effects, whereas the 10 ‘‘red
herring’’ items were not considered. In case of missing infor-
mation, both the BPRS and LUNSERS total scores were com-
puted multiplying the mean score obtained from the observed
items by the total number of items (eg, the LUNSERS total

TABLE 1. Baseline Characteristics of the Patients According to the Allocated Treatment

Clozapine and Haloperidol Clozapine and Aripiprazole

n %/Mean/Median n %/Mean/Median P*
Females, % 53 32.1 53 37.7 0.541
Age, mean (SD), y 53 41.5 (9.4) 53 40.3 (10.3) 0.549
High school diploma or academic degree, % 53 35.8 51 41.2 0.577
Occupational status, % 51 52 0.846
Employed/sheltered employed, house person, student 33.3 28.8
Unemployed 17.7 21.2
Retired 49.0 50.0
Living in residential facility in the past 6 mo, % 53 34.0 53 35.8 0.839
Length of stay in residential facility,† median (IQR), y 15 2.7 (1.3Y3.9) 14 3.2 (1.2Y6.5) 0.513
Living status, % 52 52 0.138
Alone 7.7 9.6
With relatives 53.9 53.9
With other patients 28.8 36.5
Other 9.6 0.0
Diagnosis of schizophrenia (MINI criteria), % 53 100.0 53 100.0 V
Disease duration, median (IQR), y 52 18 (12Y24) 52 14 (8Y20) 0.076
No. hospital admissions during lifetime, median (IQR) 53 5 (1Y10) 53 3 (1Y8) 0.428
BPRS total score, median (IQR) 53 60 (52Y79) 53 60 (52Y71) 0.395
Past use of clozapine
Length, median (IQR), y 53 5.0 (1.8Y8.6) 52 3.9 (1.7Y8.2) 0.524
Max dose, mean (SD), mg/d 52 483 (158) 51 452 (118) 0.266
LUNSERS total score, median (IQR) 53 22 (16Y32) 53 23 (16Y32) 0.791
*Obtained by the Pearson W2 test, the Fisher exact test, the Student t test, or the Wilcoxon rank sum test.
†
Only for the 18 and 19 patients allocated to clozapine and haloperidol and to clozapine and aripiprazole, respectively, who reported a stay in a
residential facility in the past 6 months.
IQR indicates interquartile range.

268 www.psychopharmacology.com * 2011 Lippincott Williams & Wilkins

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Journal of Clinical Psychopharmacology & Volume 31, Number 3, June 2011 Clozapine Combination Strategies in Schizophrenia

score for a male with k nonmissing items is [sum of the ob-
served scores/k]  39).
The proportion of patients withdrawing from the allocated
treatment within 3 months (primary outcome) was compared
between the 2 groups of treatment by the Person W2 test, whereas
the risk ratio (RR) was calculated using a Poisson regression
model with a robust SE (obtained by the Huber/White/sandwich
estimator of the variance) and no offset.23 The change in severity
of illness (measured by the BPRS total score) and the change in
subjective tolerability of antipsychotic drugs (measured by the
LUNSERS total score) from baseline to month 3 were compared
between the 2 groups of treatment by the analysis of covariance
with the value at baseline as a covariate and robust SEs.
A multivariable analysis was performed to compare the
differential efficacy/tolerability of the 2 treatments adjusting for
the potential confounding effect of the main prognostic factors
measured at baseline (sex, age, living condition, and BPRS and
LUNSERS total scores) and to test the interaction between each
prognostic factor and the allocated treatment. Poisson and linear
regression models with robust SEs (obtained by the Huber/
White/sandwich estimator of the variance) and no offset were
used.23
The statistical analysis was performed using STATA soft-
ware (StataCorp).
RESULTS
Characteristics and Disposition of Patients
Of 129 patients screened for inclusion, 106 were enrolled in
the study and randomly assigned to treatment. The number of
patients recruited in each site ranged from 1 (8 sites) to 8 (1 site),
with a median number of 3 patients per site. With such a total
sample size, the study had 85% power to detect a 20% difference
in the proportion that discontinued the 2 assigned treatments
(25% in the group treated with clozapine plus haloperidol vs
5% in the group treated with clozapine plus aripiprazole). All
106 patients constituted the intention-to-treat population for the
primary outcome. No patients were lost to follow-up, although
the BPRS and LUNSERS were not completed at month 3 for
1 patient, who was excluded from the analysis of these con-
tinuous outcomes. Table 1 shows the baseline demographic and
clinical characteristics of the patients (the full Table is avail-
able as Supplemental Table A, Supplemental Digital Content 1,
http://links.lww.com/JCP/A52). Most patients were males, with a
mean age of 40.3 and 41.5 years in the aripiprazole and halo-
peridol groups, respectively; 35.8% and 34.0% were living in
psychiatric residential facilities, and the median disease duration
was 14 and 18 years in the aripiprazole and haloperidol groups,
respectively. All patients had a diagnosis of schizophrenia at the
MINI interview, and 26.4% in the aripiprazole group and 34.0%
in the haloperidol group had a positive history for alcohol abuse.
At baseline, patients had been receiving clozapine for 3.9 and
5.0 years in the aripiprazole and haloperidol groups, respectively.
Most patients had already received haloperidol in the past,
although only a minority had received aripiprazole in the past.
Effectiveness and Tolerability Measures
After 3 months, the analysis of the primary outcome revealed
no difference in the proportion of patients who discontinued
treatment between the aripiprazole and haloperidol groups (13.2%
vs 15.1%, P = 0.780) (Table 2). Combination treatment was dis-
continued in 7 patients allocated to the aripiprazole group (4
patients discontinued aripiprazole and continued with cloza-
pine monotherapy, 1 patients discontinued both aripiprazole and
clozapine and started fluphenazine, and 2 patients interrupted
aripiprazole for 915 days) and in 8 patients allocated to the
haloperidol group (5 patients discontinued haloperidol and con-
tinued with clozapine monotherapy, 1 patient discontinued halo-
peridol and added aripiprazole, 1 patient discontinued haloperidol
and received clozapine combined with levomepromazine and
clotiapine, and 1 patient was given clotiapine in addition to clo-
zapine and haloperidol). In the aripiprazole group, reasons for

TABLE 2. Withdrawal From the Allocated Treatment Within 3 Months, Change in the BPRS Total Score ($BPRS), and Change
in the LUNSERS Total Score ($LUNSERS) From Baseline to Month 3

Clozapine and Haloperidol Clozapine and Aripiprazole P

Withdrawal from the allocated treatment
No. patients (no. events) 53 (8) 53 (7)
Cumulative incidence (SE) [95% CI]* 15.1% (4.9%) [6.7% to 27.6%] 13.2% (4.7%) [5.5% to 25.3%] 0.780†
RR (SE) [95% CI]‡ 1.00 0.87 (0.42) [0.34 to 2.25]
$BPRS
No. patients 52 53
Adjusted mean change§ (SE) [95% CI] j4.4 (1.5) [j7.5 to j1.4] j5.9 (1.7) [j9.4 to j2.5] 0.523
A regression coefficient|| (SE) [95% CI] 0.00 j1.5 (2.3) [j6.1 to +3.1]
$LUNSERS
No. patients 52 53
Adjusted mean change§ (SE) [95% CI] j2.0 (1.4) [j4.8 to +0.8] j7.4 (1.3) [j10.0 to j4.8] 0.006
A regression coefficient|| (SE) [95% CI] 0.00 j5.4 (1.9) [j9.2 to j1.5]
Comparison between the allocated treatments (reference group: clozapine and haloperidol).
*Binomial exact CI.
†
Obtained by the Pearson W2 test.
‡
Obtained by a Poisson regression model with a robust SE and no offset.
§
Mean $BPRS/$LUNSERS, adjusted for the baseline mean-centered value of the BPRS/LUNSERS total score by a linear regression model with
robust SEs.
||
Difference between the adjusted mean $BPRS/$LUNSERS among the patients allocated to clozapine and aripiprazole, and the adjusted mean
$BPRS/$LUNSERS among the patients allocated to clozapine and haloperidol.

* 2011 Lippincott Williams & Wilkins www.psychopharmacology.com 269

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Barbui et al Journal of Clinical Psychopharmacology & Volume 31, Number 3, June 2011

TABLE 3. Drug Use at Month 3 According to the Allocated Treatment

Clozapine and Haloperidol Clozapine and Aripiprazole

n %/mean n %/mean P*
Dose of clozapine at month 3, mean (SD), mg/d 53 395 (161) 52 421 (142) 0.376
Dose of haloperidol at month 3, mean (SD), mg/d 46 2.8 (1.7) V V V
Dose of aripiprazole at month 3, mean (SD), mg/d 1 15 (V) 48 11.8 (5.1) V
Use of other drugs at month 3, % 53 67.9 53 69.8 0.834
Antidepressants 24.5 18.9
Benzodiazepines 56.6 62.3
Mood stabilizers 17.0 11.3
Use of anticholinergics at month 3, % 53 1.9 53 3.8 1.000
*Obtained by the Pearson W2 test, the Fisher exact test, or the Student t test.

withdrawal included lack of efficacy (4 patients), acceptability
problems (2 patients), and lack of adherence (1 patient); in the
haloperidol group, the reasons were lack of efficacy (4 patients),
acceptability problems (3 patients), and lack of adherence (1 pa-
tient). Drug use at month 3 according to the allocated treatment
is presented in Table 3. The mean daily dose of clozapine was 421
and 395 mg in the aripiprazole and haloperidol groups, respec-
tively. Aripiprazole was administered at a mean daily dose of
11.8 mg, and haloperidol was administered at a mean daily dose of
2.8 mg. There was no difference in the use of concomitant med-
ications between the 2 groups.
The 3-month change of the BPRS total score ($BPRS)
was similar in the aripiprazole and haloperidol groups (j5.9 vs
j4.4 points, P = 0.523), whereas the 3-month decrease of the
LUNSERS total score ($LUNSERS) was significantly higher in
the aripiprazole group than in the haloperidol group (j7.4 vs
j2.0 points, P = 0.006) after adjusting for the baseline value
(Table 2). The differential $LUNSERS in the aripiprazole and
haloperidol groups was confirmed when the 14 subjects with
at least 1 missing item in the LUNSERS total score at baseline
and/or at month 3 were excluded from the analysis (j7.2 vs
j1.8 points, P = 0.012).
Multivariable Analysis
The comparison of the efficacy/tolerability between aripipra-
zole and haloperidol adjusted for the prognostic factors measured
at baseline provided the same results as those obtained in the
main analysis (Table 4). The differential $LUNSERS be-
tween the allocated treatments (beta regression coefficient (A) =
difference between the adjusted mean $LUNSERS in the
aripiprazole group and the adjusted mean $LUNSERS in the
haloperidol group = j5.4 points; 95% confidence interval [CI],
j9.4 to j1.5) was confirmed when the 14 subjects with at least
1 missing item in the LUNSERS total score at baseline and/or
at month 3 were excluded from the analysis (A = j5.5 points;
95% CI, j9.9 to j1.2).

TABLE 4. Withdrawal From the Allocated Treatment Within 3 Months, Change in the BPRS Total Score ($BPRS) and Change
in the LUNSERS Total Score ($LUNSERS) From Baseline to Month 3

Withdrawal From the Allocated $BPRS $LUNSERS

Treatment (n = 106) (n = 105) (n = 105)
Risk Ratio (95% CI) A Regression Coefficient (95% CI)
Allocated treatment (clozapine and aripiprazole 0.77 (0.26 to 2.25) j1.5 (j6.1 to +3.1) j5.4 (j9.4 to j1.5)*
vs clozapine and haloperidol)
Sex (female vs male) 3.60 (1.44 to 8.98)* +3.3 (j1.6 to +8.1) j1.7 (j6.4 to +2.9)
Age† (5-y increase) 0.72 (0.59 to 0.88)* j0.1 (j1.4 to +1.1) +0.4 (j0.7 to +1.4)
Living condition in the past 6 mo 2.10 (0.89 to 4.98) +1.4 (j3.4 to +6.3) j0.9 (j5.0 to +3.2)
(residential facility vs at home)
BPRS total score† (5-point increase) 1.05 (0.93 to 1.20) j1.1 (j1.9 to j0.4)* j0.1 (j0.7 to +0.5)
LUNSERS total score† (5-point increase) 1.11 (1.00 to 1.23) j0.6 (j1.5 to +0.3) j0.5 (j1.5 to +0.5)‡
j2.3 (j3.1 to j1.4)‡,§
Constant j6.1 (j9.8 to j2.3)* j0.9 (j4.1 to 2.2)
Mutually adjusted RRs (obtained by a Poisson regression model with robust SEs and no offset) and A regression coefficients (obtained by linear
regression models with robust SEs) for the association between the allocated treatment, the main prognostic variables measured at baseline, and the
short-term outcomes.
*P G 0.05.
†
Mean centered.
‡
Test for the interaction between the allocated treatment, the LUNSERS total score at baseline, and $LUNSERS: P = 0.011. The adjusted A
regression coefficients for the association between the LUNSERS total score at baseline and $LUNSERS among patients allocated to clozapine and
haloperidol (j0.5) and among patients allocated to clozapine and aripiprazole (j2.3) are reported.
§
P G 0.001.

270 www.psychopharmacology.com * 2011 Lippincott Williams & Wilkins

Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Journal of Clinical Psychopharmacology & Volume 31, Number 3, June 2011
Female sex (RR = 3.60; 95% CI, 1.44Y8.98) and younger
age (RR = 0.72; 95% CI, 0.59Y0.88, for a 5-year increase of
age at baseline) were both associated with a higher risk of treat-
ment withdrawal. In addition, a higher LUNSERS total score
at baseline was positively associated with the primary outcome
(RR = 1.11; 95% CI, 1.00Y1.23, for a 5-point increase of the score
at baseline), even if this relationship did not reach the statisti-
cal significance (P = 0.051).
Higher values of the BPRS total score at baseline were
significantly associated with a greater 3-month decrease of the
BPRS total score (A = $BPRS for a 5-point increase of the score
at baseline = j1.1 points; 95% CI, j1.9 to j0.4). The negative
association between the BPRS total score at baseline and
$BPRS was confirmed when the 12 subjects with at least 1
missing item in the BPRS total score at baseline and/or at
month 3 were excluded from the analysis (A = j1.4 points;
95% CI, j2.1 to j0.7).
A statistically significant interaction (P = 0.011) was found
between the allocated treatments, the LUNSERS total score at
baseline and $LUNSERS. In particular, a 3-month decrease of
the LUNSERS total score was found only among the patients
in the aripiprazole group (A = $LUNSERS for a 5-point increase
of the score at baseline = j2.3 points; 95% CI, j3.1 to j1.4).
The negative association between the LUNSERS total score at
baseline and $LUNSERS in the aripiprazole group was con-
firmed when the 14 subjects with at least 1 missing item in the
LUNSERS total score at baseline and/or at month 3 were excluded
from the analysis (A = j2.4 points; 95% CI, j3.3 to j1.5).
DISCUSSION
In the present randomized trial, 2 clozapine combination
strategies were compared in a representative sample of patients
with treatment-resistant schizophrenia in real-world community
psychiatric services. In treatment discontinuation and psychotic
symptoms, clozapine combined with aripiprazole provided no
additional benefit in comparison with clozapine combined with
haloperidol. By contrast, in tolerability, the addition of aripiprazole
was associated with better patient perception of adverse effects.
The background logic that guided the development of the
present study was based on the need to provide real-world sug-
gestions for patients who do not have an optimal response to
clozapine and for whom clinicians feel the pressing need to add
a second antipsychotic drug.24 We made the choice of compar-
ing 2 active clozapine combination strategies, without a placebo
arm, as we reasoned that clinicians would have been reluctant
to accept the possibility of allocating such difficult-to-treat
patients to placebo. The choice of haloperidol was determined by
prevailing clinical practice, and we designed the trial so that it
would not interfere with the routine care of participants: eligi-
bility criteria were simple, doses of the experimental drugs and
concomitant medications were at the discretion of the attend-
ing physician, and data collection was minimized. Besides, the
primary outcome was treatment discontinuation, a measure that
integrates patients’ and clinicians’ judgments of efficacy, safety,
and tolerability into a global measure of effectiveness that
reflects their evaluation of therapeutic benefits in relation to
undesirable effects.18
Lack of improvement with the addition of aripiprazole
compared with the addition of haloperidol is in line with the
similarly negative studies of previous studies of augmentation of
clozapine with various antipsychotics.5,6 The only randomized
comparison that involved aripiprazole, carried out by Chang et al,25
included 62 patients assigned to aripiprazole or placebo. After
8 weeks of double-blind treatment, aripiprazole augmentation
* 2011 Lippincott Williams & Wilkins
Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clozapine Combination Strategies in Schizophrenia
of clozapine did not lead to a significant improvement of total
symptom severity, but a favorable change in some adverse
effects, including prolactin and triglyceride levels, was observed.
In the present study, we described adverse effects using the
patient viewpoint, and consistent with Chang et al, it was found
that the addition of aripiprazole was associated with a better
perception of adverse effects. Observational case series of
patients exposed to clozapine plus aripiprazole similarly sug-
gested improvements in adverse effects, possibly mediated by
a decrease in clozapine dose when compared with aripiprazole.7
In the present study, a negligible decrease in clozapine dosages
was observed at follow-up in the aripiprazole group, although
aripiprazole dose was similar to that recorded in the study of
Chang et al.25 It is therefore possible that the better perception of
aripiprazole might be related to how this antipsychotic compares
with haloperidol and not to the dose regimen of clozapine.
At the time that CHAT was designed, the CATIE trial was
the only randomized study that used treatment withdrawal as
the primary outcome measure. Subsequently, the results of the
EUFEST study were published.26 The EUFEST study was an
open randomized controlled trial of haloperidol versus second-
generation antipsychotic drugs in patients with schizophrenia
aged 18 to 40 years. In this study, the primary outcome measure
was all-cause treatment discontinuation. In contrast with the
CATIE trial, and similar to the CHAT, the EUFEST study used
treatment discontinuation as primary outcome with an open
study design, in which patients knew which medication they
were taking. In the EUFEST study, a compelling discrepancy
was found between treatment discontinuation that was in favor of
second-generation antipsychotics, and symptom scores mea-
sured using a rating scale administered under blind conditions,
which were not different between competitive treatments.26 The
authors hypothesized that expectations of psychiatrists could
have led to haloperidol being discontinued more often, thus
casting doubts on the use of treatment discontinuation under
open conditions. We note that, in the CHAT study, no discrep-
ancy between the primary outcome measure, recorded under
open conditions, and symptom scores, recorded by blind staff,
was observed. We therefore believe that the pragmatic design of
the present study, including lack of blindness, should not have
hampered the analysis of the main outcomes, considering that
the design was based on a comparison of 2 active interventions,
and physicians used to have no a priori expectations of which
intervention would be better.
Another issue is whether lack of blindness affected the
ratings at the LUNSERS, considering that this scale was self-
rated by patients who knew their drug assignment. We cannot
rule out the possibility that expectations of patients led to
haloperidol being systematically rated as more troublesome than
aripiprazole, although we note that no association between
LUNSERS ratings and antipsychotic drug treatment (first vs
second generation) was found in a recent multicenter study
conducted in 4 European countries.27 It is therefore reasonable
to assume that patients had no a priori expectations of which
intervention would be better tolerated. In addition, if patients had
had strong a priori expectations against haloperidol, then they
would not have consented to be randomly included in a study
with 50% of possibilities to receive haloperidol.
The main limitation of this study is that we failed to reach
the target sample size. Although CHAT is the largest random-
ized comparison so far carried out in a Western country in this
difficult-to-treat population, the CI around the RR point estimate
ranges from the possibility that aripiprazole is appreciably better
than haloperidol to the possibility that haloperidol is appreciably
better than aripiprazole. Another potential weakness is that lack
www.psychopharmacology.com 271
Barbui et al Journal of Clinical Psychopharmacology
of blindness leaves the possibility of performance bias, that is,
investigators and participants might have behaved systematically
differently dependent on the allocated treatment. We note, for
example, that there were some changes in the use of concomitant
antidepressants and benzodiazepines during the study, and this
might potentially have had some influence on outcome mea-
surements. Although we cannot completely rule out this possi-
bility, we note that the involvement of many recruiting sites and
many investigators should have diluted this possibility, making
very unlikely that in different sites different investigators sys-
tematically behaved differently dependent on the allocated treat-
ment. The observed lack of difference in the primary outcome
seems to provide further indirect evidence against the possibil-
ity of performance bias.
The multivariable analysis further reinforced the relation-
ship between allocated treatment and perception of adverse ef-
fects: having adjusted for possible confounders, clozapine plus
aripiprazole was associated with a higher 3-month decrease of
the LUNSERS score compared with clozapine plus haloperidol.
In addition, female sex, young age, and negative perception of
adverse effects were independent prognostic factors of the risk of
withdrawal within 3 months. This is in line with epidemiological
data suggesting that, in real-world settings, age, sex, and adverse
effects are associated with treatment adherence, including adher-
ence to antipsychotic drug treatment.28 Young age is associated
with lower illness insight, and therefore, it is possible that young
people may be more reluctant to adhere to antipsychotic drug
treatment, especially when 2 agents are prescribed concurrently.
Moreover, women perceive antipsychotic drugs as less toler-
able than men,27 and therefore, the relationship between female
sex and treatment discontinuation might be mediated by poor
tolerability.
In conclusion, the results of this study indicate that aug-
mentation of clozapine with aripiprazole offers no benefit with
regard to treatment withdrawal and overall symptoms in schizo-
phrenia as compared with augmentation with haloperidol. How-
ever, the advantage in the perception of adverse effects with
aripiprazole treatment may be meaningful for patients. This fin-
ding needs to be confirmed by the analysis of the 12-month data
from CHAT and needs also to be replicated by additional ran-
domized trials and high-quality observational studies.
ACKNOWLEDGMENTS
The authors thank Donatella Castiglioni for her invaluable
help as study secretary. The authors thank John R. Geddes, pro-
fessor of epidemiological psychiatry at the University of Oxford,
and Clive Adams, coordinating editor, Cochrane Schizophrenia
Group, University of Nottingham, for providing useful comments
and suggestions on the drafted study protocol.
STEERING COMMITTEE
Michele Tansella (Verona, Chair), Tommaso Losavio
(Roma, Co-chair), Corrado Barbui (Verona), Andrea Cipriani
(Verona), Michela Nosè (Verona), Mauro Percudani (Milan),
and Scott Stroup (New York).
CHAT INVESTIGATORS
Accordini S., Aldini F., Appino MG., Artioli P., Barale F.,
Barbui C., Beneduce R., Berardi D., Bertolazzi G., Biancosino B.,
Bisogno A., Bivi R., Bogetto F., Boso M., Bozzani A., Bucolo
P., Casale M., Cascone L., Ciammella L., Cicolini A., Cipresso
G., Cipriani A., Colombo P., Dal Santo B., De Francesco M.,
Di Lorenzo G., Di Munzio W., Erlicher A., Esposito E., Ferrannini
L., Ferrato F., Ferro A., Fragomeno N., Fricchione Parise V.,
272 www.psychopharmacology.com
Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
& Volume 31, Number 3, June 2011
Frova M., Gardellin F., Garzotto N., Giambartolomei A., Girlanda
F., Giupponi G., Grassi L., Grazian N., Grecu L., Guerrini G.,
Laddomada F., Lazzarin E., Lintas C., Malchiodi F., Malvini L.,
Marchiaro L., Marsilio A., Mauri MC., Mautone A., Menchetti
M., Migliorini G., Mollica M., Moretti D., Mulè S., Nicholau
S., Nosè F., Nosè M., Occhionero G., Pacilli AM., Pecchioli S.,
Percudani M., Petrosemolo P., Piantato E., Piazza C., Pontarollo
F., Purgato M., Pycha R., Quartesan R., Rillosi L., Risso F., Rizzo
R., Rocca P., Roma S., Rossattini M., Rossi G., Rossi G., Sala A.,
Santilli C., Saraò G., Sarnicola A., Sartore F., Scarone S., Sciarma
T., Siracusano A., Strizzolo S., Tansella M., Targa G., Tasser A.,
Tomasi R., Travaglini R., Valentini C., Veronese A., Ziero S.
The full list with affiliations is reported in the Web site
(Supplemental List of CHAT Investigators, Supplemental Digital
Content 2, http://links.lww.com/JCP/A53).
AUTHOR DISCLOSURE INFORMATION
The authors declare no conflicts of interest.
REFERENCES
1. Conley RR, Kelly DL. Management of treatment resistance in
schizophrenia. Biol Psychiatry. 2001;50:898Y911.
2. Kane J, Honigfeld G, Singer J, et al. Clozapine for the treatment-resistant
schizophrenic. A double-blind comparison with chlorpromazine.
Arch Gen Psychiatry. 1988;45:789Y796.
3. Rosenheck R, Cramer J, Xu W, et al. A comparison of clozapine and
haloperidol in hospitalized patients with refractory schizophrenia.
Department of Veterans Affairs Cooperative Study Group on Clozapine
in Refractory Schizophrenia. N Engl J Med. 1997;337:809Y815.
4. Mauri MC, Volonteri LS, Dell’Osso B, et al. Predictors of clinical
outcome in schizophrenic patients responding to clozapine.
J Clin Psychopharmacol. 2003;23:660Y664.
5. Barbui C, Signoretti A, Mule S, et al. Does the addition of a second
antipsychotic drug improve clozapine treatment? Schizophr Bull.
2009;35:458Y468.
6. Cipriani A, Boso M, Barbui C. Clozapine combined with different
antipsychotic drugs for treatment resistant schizophrenia.
Cochrane Database Syst Rev. 2009;CD006324.
7. Mule S, Cipriani A, Barbui C. Aripiprazole in addition to clozapine in
partially responsive patients with schizophrenia: a critical review of
case series. Clin Schizophr Relat Psych. 2008;1:341Y347.
8. Sernyak MJ, Rosenheck R. Clinicians’ reasons for antipsychotic
coprescribing. J Clin Psychiatry. 2004;65:1597Y1600.
9. Nose M, Accordini S, Artioli P, et al. Rationale and design of an
independent randomised controlled trial evaluating the effectiveness
of aripiprazole or haloperidol in combination with clozapine for
treatment-resistant schizophrenia. Trials. 2009;10:31.
10. Moher D, Hopewell S, Schulz KF, et al. CONSORT 2010 explanation
and elaboration: updated guidelines for reporting parallel group
randomised trials. BMJ. 2010;340:c869.
11. Schulz KF, Altman DG, Moher D. CONSORT 2010 statement: updated
guidelines for reporting parallel group randomised trials. BMJ.
2010;340:c332.
12. Rossi A, Alberio R, Porta A, et al. The reliability of the
Mini-International Neuropsychiatric InterviewVItalian version.
J Clin Psychopharmacol. 2004;24:561Y563.
13. Sheehan DV, Lecrubier Y, Sheehan KH, et al. The Mini-International
Neuropsychiatric Interview (M.I.N.I.): the development and
validation of a structured diagnostic psychiatric interview for
DSM-IV and ICD-10. J Clin Psychiatry. 1998;59(suppl 20):
22Y33.
* 2011 Lippincott Williams & Wilkins
Journal of Clinical Psychopharmacology & Volume 31, Number 3, June 2011
14. Ruggeri M, Koeter M, Schene A, et al. Factor solution of the
BPRS-expanded version in schizophrenic outpatients living in five
European countries. Schizophr Res. 2005;75:107Y117.
15. Day JC, Wood G, Dewey M, et al. A self-rating scale for measuring
neuroleptic side-effects. Validation in a group of schizophrenic
patients. Br J Psychiatry. 1995;166:650Y653.
16. Altman DG, Schulz KF. Statistics notes: concealing treatment
allocation in randomised trials. BMJ. 2001;323:446Y447.
17. Morrison P, Gaskill D, Meehan T, et al. The use of the Liverpool
University Neuroleptic Side-Effect Rating Scale (LUNSERS)
in clinical practice. Aust N Z J Ment Health Nurs.
2000;9:166Y176.
18. Lieberman JA, Stroup TS, McEvoy JP, et al. Effectiveness of
antipsychotic drugs in patients with chronic schizophrenia.
N Engl J Med. 2005;353:1209Y1223.
19. Chow SC, Shao J, Wang H. Sample Size Calculations in Clinical
Research. New York, NY: Marcel Dekker; 2003.
20. Accordini S. An introduction to sample size calculations in clinical
trials. Epidemiol Psychiatr Soc. 2007;16:299Y301.
21. Cipriani A, Girlanda F, Barbui C. Superiority, equivalence or
non-inferiority? Epidemiol Psychiatr Soc. 2009;18:311Y313.
* 2011 Lippincott Williams & Wilkins
Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clozapine Combination Strategies in Schizophrenia
22. Hintze J. Number Cruncher Statistical Systems. Kaysville, UT:
NCSS; 2004.
23. Zou G. A modified Poisson regression approach to prospective
studies with binary data. Am J Epidemiol. 2004;159:702Y706.
24. Cipriani A, Purgato M, Barbui C. Why internal and external validity
of experimental studies are relevant for clinical practice?
Epidemiol Psychiatr Soc. 2009;18:101Y103.
25. Chang JS, Ahn YM, Park HJ, et al. Aripiprazole augmentation in
clozapine-treated patients with refractory schizophrenia:
an 8-week, randomized, double-blind, placebo-controlled trial.
J Clin Psychiatry. 2008;69:720Y731.
26. Kahn RS, Fleischhacker WW, Boter H, et al. Effectiveness of
antipsychotic drugs in first-episode schizophrenia and
schizophreniform disorder: an open randomised clinical trial.
Lancet. 2008;371:1085Y1097.
27. Barbui C, Nose M, Bindman J, et al. Sex differences in the subjective
tolerability of antipsychotic drugs. J Clin Psychopharmacol.
2005;25:521Y526.
28. Nose M, Barbui C, Tansella M. How often do patients with psychosis
fail to adhere to treatment programmes? A systematic review.
Psychol Med. 2003;33:1149Y1160.
www.psychopharmacology.com 273