1242

Schizophrenia Bulletin vol. 39 no. 6 pp. 12421251, 2013

doi:10.1093/schbul/sbt138
Advance Access publication September 26, 2013
The Author 2013. Published by Oxford University Press on behalf of the Maryland Psychiatric Research Center. All rights reserved.
For permissions, please email: journals.permissions@oup.com
The Effect of Motivational Interviewing on Medication Adherence and
Hospitalization Rates in Nonadherent Patients with Multi-Episode Schizophrenia
EmileBarkhof*
,1
, Carin J.Meijer
1
, Leo M.J.de Sonneville
2
, Don H.Linszen
3
, and Lieuwede Haan
1

1
Department of Psychiatry, Academic Medical Centre, Amsterdam, The Netherlands;
2
Department of Clinical Child and Adolescent
Studies, Leiden University, Leiden, The Netherlands;
3
Department of Psychiatry, Academic Hospital Maastricht, Maastricht,
The Netherlands
*To whom correspondence should be addressed; Department of Psychiatry, Academic Medical Centre, Meibergdreef 5, Amsterdam 1105
AZ, The Netherlands; tel:31-20-8913500, fax: 31-20-8913702, e-mail: e.barkhof@amc.uva.nl
Background: Medication nonadherence in patients with
schizophrenia presents a serious clinical problem.
Research on interventions incorporating motivational
interviewing (MI) to improve adherence have shown
mixed results. Aims: Primary aim is to determine the
effectiveness of a MI intervention on adherence and hos-
pitalization rates in patients, with multi-episode schizo-
phrenia or schizoaffective disorder, who have experienced
a psychotic relapse following medication nonadherence.
Secondary aim is to evaluate whether MI is more effec-
tive in specifc subgroups. Methods: We performed a
randomized controlled study including 114 patients who
experienced a psychotic relapse due to medication nonad-
herence in the past year. Participants received an adapted
form of MI or an active control intervention, health edu-
cation (HE). Both interventions consisted of 58 ses-
sions, which patients received in adjunction to the care
as usual. Patients were assessed at baseline and at 6 and
12months follow-up. Results: Our results show that MI
did not improve medication adherence in previously non-
adherent patients who experienced a psychotic relapse.
Neither were there signifcant differences in hospitaliza-
tion rates at follow-up between MI and HE (27% vs 40%,
P=.187). However, MI resulted in reduced hospitaliza-
tion rates for female patients (9% vs 63%, P=.041), non-
cannabis users (20% vs 53%, P=.041), younger patients
(14% vs 50%, P=.012), and patients with shorter illness
duration (14% vs 42%, P=.040). Conclusions: Targeted
use of MI may be of beneft for improving medication
adherence in certain groups of patients, although this
needs further examination.
Key words: RCT/intervention study/medication
adherence/motivational interviewing/schizophrenia
Introduction
Nonadherence to antipsychotic medication is highly
prevalent in patients with schizophrenia
1
and is associated
with sharply increased readmission rates, more aggressive
incidents, more suicides, signifcant emotional and social
burden for patients and their families, and higher fnan-
cial costs.
24
In the last decades, several interventions have
been developed to improve adherence rates.
5
Recent treat-
ment recommendations promote focusing on specifc tar-
gets that may contribute to nonadherence.
6
Motivational
interviewing (MI) is a client-centered, directive method
for enhancing intrinsic motivation to change by exploring
and resolving ambivalence.
7
MI was originally designed
as a therapeutic approach to treat abuse of alcohol and
other substances, for which it proved to be very effec-
tive.
8,9
Following positive results in other health care
domains with regard to behavior change,
1012
effectiveness
of (adapted) MI for improving medication adherence
has also been studied in patients with psychotic disor-
ders.
5,13
Compliance therapy, an intervention based on
MI and other cognitive approaches, showed substantial
improvements in medication adherence in patients with
schizophrenia,
14,15
although this could not be replicated
in another trial.
16
Studies focusing on a similar interven-
tion (adherence therapy), also containing MI, yielded
mixed results.
17,18
More recently, an individually tailored approach incor-
porating MI proved to be effective in prompting service
engagement and medication adherence.
19
This means that to date evidence concerning the effec-
tiveness of MI as a means to improve medication adher-
ence in patients with schizophrenia is still inconclusive.
The present study was designed to investigate the
effect of MI on adherence to antipsychotic medication

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1243
Effects of Motivational Interviewing on Adherence
in patients, with a multi-episode schizophrenia spectrum
disorder, who were hospitalized or unstable due to medi-
cation nonadherence. These so-called revolving door
patients generally have poor social functioning and an
unfavorable prognosis, and therefore improvement of
their medication adherence is urgently needed.
20
The cur-
rent study explicitly aimed to include this severely dis-
turbed group of patients, who are often either excluded
from studies or are unwilling to participate. To fnd a pos-
sible explanation for the unequivocal results on the effect
of MI on adherence so far, a second aim of the study was
to investigate whether specifc subgroups of patients may
beneft more from this intervention than others. As there
are several factors associated with nonadherence,
1
we
aimed to explore whether known risk factors for nonad-
herence such as cannabis use, illness duration, and route
of medication administration, along with gender and age,
may mediate effectiveness of MI.
Methods
Participants
The study was conducted in 3 mental health care institu-
tions in the greater Amsterdam area. Participants were
selected from inpatient and outpatient facilities for the
treatment of psychotic disorders. Clinicians of the par-
ticipating facilities regularly reviewed their caseloads for
patients that showed a psychotic relapse or a clinical dete-
rioration with nonadherence as a probable cause. Next,
potential participants were invited to receive informa-
tion on the study and to verify the following inclusion
criteria: an age of 1865 years, a clinical diagnosis of
schizophrenia or a schizoaffective disorder confrmed by
the Structured Clinical Interview for DSM disorders.
21

Participants had experienced a recent (<1 y) psychotic
relapse and/or a clinical deterioration, both following
nonadherence to the antipsychotic treatment, resulting in
hospitalization and/or a change on the Clinical Global
Impression Scale, severity of illness.
22
Subsequently,
antipsychotic treatment had to be resumed with at least
some symptomatic improvement, defned as score of 1,
2, or 3 (very much improved, much improved, or mini-
mal improved) on the Clinical Global Impression Scale
for improvement.
22
Participants were required to have an
adequate mastery of the Dutch language and be able and
prepared to give written informed consent. Exclusion cri-
teria were an organic disease with a possible etiological
relation to the psychotic disorder and/or a severe intel-
lectual dysfunction (intelligence quotient <70).
Interventions
Motivational Interviewing. MI is a client-centered,
directive method, through which patients are engaged in
strategically directed conversations about their problems.
It explores personal ideas and ambivalences, eliciting and
selectively reinforcing change talk, by which discrepan-
cies between the present behavior and the patients own
future goals are amplifed. The overall goal is to increase
the patients intrinsic motivation for change.
7
Based on the basic principles of MI, a manual was
designed, incorporating some adaptations directed at
specifc problems of the study group, such as negative
symptoms, specifc positive symptoms (delusions con-
cerning medication) as well as cognitive defcits such as
attention problems. In contrast to the original MI, the
adapted form therefore encompasses a somewhat more
active stance of the therapist, greater fexibility in ses-
sion length, and a more active provision of psychoedu-
cational information, though only after explicit consent
by the patient. MI according to the intervention manual
(available upon request from E.B.) comprised 4 phases.
These were used as a framework, though not in a strict
consecutive order. The phases involved introduction and
engagement; exploring attitudes and beliefs toward treat-
ment, exploring the patients own personal goals, and
the readiness for change. In the next phase, informa-
tion was provided and ambivalences were amplifed along
which favorable attitudes and beliefs toward change were
reinforced. The last phase was committed to evaluation
and consolidation of the motivation to change.
Health Education. The control group was provided
Health Education (HE) sessions. HE comprised indi-
vidual lectures on general health topics (such as healthy
food, physical exercise etc.). Participants were asked to
choose one of these topics for each session. Therapists
delivering HE used an active and didactic attitude with
specifc scripts to ensure that discussing treatment issues
was avoided.
Procedure
Participants were allocated to either the experimental
interventionMIor the control groupHEby means of
a computerized cluster randomization program, produc-
ing a block of codes for every 6 consecutive inclusions,
which were one by one revealed by the coordinating
researcher. The therapists who performed both interven-
tions were psychologists, a psychiatrist, and community
mental health nurses, who were trained by a professional
trainer on MI (eight 4-h sessions) as well as on HE (two
4-h sessions) according to a structured manual, to maxi-
mize skills and minimize contamination effects between
interventions.
Therapy sessions were audiotaped if the patient con-
sented to this, and these tapes were used in monthly
supervised 4-hour sessions for all therapists, in which the
ongoing interventions were discussed, to ensure treat-
ment fdelity. These sessions were focused on delivering
interventions according to the treatment manuals and on
preventing contamination of MI toHE.

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1244
E. Barkhof etal
Within a period of 26 weeks, in both the MI and the
HE group, patients were offered 8 sessions of either MI
or HE. When the therapist judged there were problems to
keep patients engaged in the interventions or in case of
practical barriers, fewer sessions were given, with a mini-
mum of 5 sessions. Less than 5 sessions was counted as
a dropout. The session duration varied between 20 and
45 minutes, depending on the attention span of the par-
ticipant during a session. Therapists were not otherwise
involved in the treatment of participants. Patients were
told they would be allocated to 1 of 2 active interventions
and were not told which was the experimental condi-
tion. Next to the interventions, participants received care
as usual, consisting of functional assertive community
treatment for patients treated on an outpatient basis and
routine clinical care for hospitalized patients.
Before starting the intervention, a baseline assess-
ment (T0) was performed. Participants were interviewed
again after the intervention was completed (T1) and
after 6months follow-up (T2). All assessments were per-
formed by trained psychologists and psychiatrists, who
were masked to which condition a patient was allocated;
a coordinator assigned the interventions and assessments
to different researchers, ensuring that the interventions
and the assessments were never performed by researchers
appointed in the same facility. Data on interventions and
assessments were stored separately.
Participants received a travel expenses compensation
of 5 euro for each intervention session or assessment in
which they participated. The study was approved by the
Medical Ethical Committee of the Academic Medical
Centre, Amsterdam.
Assessments
Demographics. Information on demographic data and
prescribed medication were assessed with the Client
Sociodemographic and Service Receipt Inventory.
23
Primary Outcome. Medication Adherence Medication
adherence was assessed with the medication adherence
questionnaire (MAQ), a 4-item questionnaire with good
levels of internal validity and reliability,
24
covering the
participants report on medication adherence; scores
range from 0 to 4, with higher scores indicating higher
levels of adherence. Furthermore, data concerning medi-
cation adherence as judged by caregivers and treating
physicians were measured by the 5-point adherence item
of the life chart schedule (LCS), with scores ranging from
1 to 5, higher scores indicating higher levels of adherence.
The LCS yields reliable ratings of the long-term course of
schizophrenia.
25
Attitudes toward medication were assessed with the
Drug Attitude Inventory (DAI). The DAI is a self-report
10-item questionnaire, with a score range of 010, higher
scores indicating more belief in the personal beneft of
the medication. The DAI showed a good internal consis-
tency and validity.
26
Secondary Outcomes. Hospitalization Hospitalization
was assessed by the LCS.
25
Based on the LCS data, a
dichotomous variable was constructed indicating whether
patients had been hospitalized, regardless of the number
of admissions.
Psychopathology Severity of psychopathology was
measured with the Positive and Negative Syndrome
Scale (PANSS).
27
Assessors were trained on the PANSS
using original training videos to maximize concordance
between assessors, which was further facilitated by reg-
ular supervised meetings in which videotaped assess-
ments were scored and inconsistencies between assessors
discussed.
Cannabis Abuse Concomitant cannabis misuse was
assessed by means of urine analysis at baseline.
Sample Size
We aimed to fnd a difference on the adherence measures
of 0.5 with a SD of 1.0. To achieve a power of 80% of
detecting such difference with a medium effect size of 0.5
with a 2-sided signifcance level of .05, a sample size of
50 patients in each group was needed. With an estimated
attrition rate of 20%, we aimed to include 120 patients.
Statistical Analysis
To evaluate the effect of randomization on baseline
demographic and disease-specifc parameters, t tests were
performed for the continuous variables and chi-square
tests for the categorical variables. To assess differences
in hospitalization rates, chi-square tests were performed,
including gender, age, cannabis use, medication adminis-
tration route, and illness duration as additional grouping
variables. For this purpose, the continuous variables age
and illness duration were transformed to a dichotomous
variable using the medianvalue.
The main effects of interventions on adherence mea-
sures were assessed using 1-way between-groups ANOVA,
using the baseline values as covariates to adjust for pre-
intervention scores. To assess the infuence of age, gen-
der, cannabis use, medication administration route, and
illness duration on adherence outcomes, these variables
were separately entered as a second fxed factor in 2-way
between-groups ANOVA.
To evaluate the effect of the interventions on the sever-
ity of psychopathology, mixed between-within subjects
ANOVA were performed with PANSS scores as depen-
dant variables.
To detect possible differences between subgroups
on severity of psychopathology, separate independent
sample t tests were performed with the dichotomous
subgroups of the variables gender, age, cannabis use,

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1245
Effects of Motivational Interviewing on Adherence
duration of illness, and medication administration route
as independent variables and PANSS scores as dependent
variables.
Effect sizes of all signifcant results are expressed
in (partial) eta squared, Cohens d, or phi coeffcient,
depending on the analysis.
Because the majority of patients who dropped out of
the intervention-refused follow-up assessments or were
lost to follow-up, we were unable to perform an intention
to treat analysis and decided to perform a per-protocol
analysis instead.
Results
Participants
Four hundred and three patients were referred for par-
ticipation in the study. Of these, 226 did not meet inclu-
sion criteria. Of the 186 remaining patients, 72 refused
to participate. One hundred and fourteen patients were
randomized and allocated to the two treatment arms.
During the intervention, 18 patients dropped out, 10 in
the MI group and 8 in the HE group (see fgure1). The
group that dropped out was younger (M = 30.7, SD =
11.7) than those who completed interventions (M=36.8,
SD=9.8; t=2.35, df=112, P=.020, Cohens d=0.44)
and showed a higher use of cannabis use at trend level
(22.2% vs 10.4%; P = .072, = 0.22). Nevertheless,
these characteristics were equally distributed over both
conditions.
Demographic characteristics and disease-specifc
parameters at baseline did not signifcantly differ between
conditions (see table1).
Interventions
Not all participants received the full 8 sessions of the
interventions. The mean number of sessions in the
MI group was 6.4 (SD = 1.2) vs 6.6 (SD = 0.9) in
the HE group, which was not a significant difference
(P=.60).
Adherence
At both follow-up assessments, there were no signifcant
differences between MI and HE on the two adherence
measures. Likewise, there were no differences in attitudes
toward medication (see table2).
Next, we examined the differences between subgroups
with regard to the effect of MI and HE on adherence.
Patients referred
N = 403
Patients meeting
criteria
N = 186
Informed consent +
randomised
N = 114
Experimental Intervention
(Motivational Interviewing)
N = 55
Refused to
participate
N = 72
Control Intervention
(Health Education)
N = 59
MI
completed
N = 45
HE
completed
N = 51
MI
drop-out
N = 10
HE
drop-out
N = 8
Fig.1. Consort status.

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1246
E. Barkhof etal
At T1, there were no signifcant interactions between
any of the covariates and intervention type. At T2, there
was a signifcant interaction between the MAQ score
and route of medication administration: F(1,59)=4.53,
P=.037,
2
=0.07, and the LCS score and route of medi-
cation administration: F(1,45)=7.36, P=.009,
2
=0.14.
These results suggest that patients using depot medica-
tion show higher adherence rates at 6 months follow-up
on the MAQ and the LCS when they received MI, com-
pared withHE.
Furthermore, there was a trend level interaction
between the DAI score and age group: F(1,49) = 3.93,
P=.05,
2
=0.07, suggesting that patients younger than
35years showed more favorable attitudes toward medica-
tion at 6 months follow-up when they received MI, com-
pared with HE.
Table1. Baseline Characteristics
Total (n=114) Motivational Interviewing (n=55) Health Education (n=59) P
a
Age, mean (SD) 35.9 (10.3) 37 (1.4) 34.7 (1.4) .32
Sex: male, n (%) 91 (80 %) 43 (78%) 48 (81%) .67
Ethnicity: white European, n (%) 53 (47%) 26 (47%) 27 (46%) .87
Duration of illness, years (SD) 7.8 (6.4) 6.8 (5.6) 8.8 (6.9) .11
Number of psychiatric
admissions, mean (SD)
3.8 (4.2) 3.6 (5.3) 4.0 (3.0) .71
Diagnosis .99
Schizophrenia, n (%) 87 (76.3) 42 (76.4) 45 (76.3)
Schizoaffective disorder, n (%) 27 (23.7) 13 (23.6) 14 (23.7)
Medication at baseline, n (%) .48
First-generation antipsychotic 38 (33%) 17 (31%) 21 (36%)
Second-generation antipsychotic 70 (62%) 36 (65%) 34 (58%)
Other (mood stabilizer) 6 (5%) 2 (4%) 4 (6%)
Medication administration route, n (%) .14
Oral 85 (75%) 44 (80%) 41 (70%)
Depot 29 (25%) 11 (20%) 18 (30%)
PANSS total, mean (SD) 71.8 (18.9) 71.9 (20.9) 70.2 (15.8) .64
Cannabis-positive urine sample, n (%) 14
b
(24%) 5 (16%) 9 (33%) .11
Note: PANSS, Positive and Negative Syndrome Scale.
a
Chi-square tests for dichotomous variables; t tests for continuous variables.
b
Available N = 59.
Table2. Effects of Interventions on Adherence Rates
Motivational Interviewing (n=30) Health Education (n=32)
P
a
Mean SD Mean SD
MAQ
b
T0 (baseline) 3.00 1.34 3.13 1.24
T1 (posttreatment) 3.34 0.99 3.13 1.12 .34
T2 (6-mo follow-up) 2.97 1.42 3.38 1.11 .21
DAI
c
T0 (baseline) 6.86 2.18 6.03 2.30
T1 (posttreatment) 6.64 2.50 6.38 1.98 .72
T2 (6-mo follow-up) 6.89 2.39 6.67 2.52 .70
LCS adherence score by
physician and/or caregiver
d
T0 (baseline) 4.45 0.76 4.31 0.96
T1 (posttreatment) 4.32 0.84 4.17 0.81 .74
T2 (6-mo follow-up) 4.33 0.82 4.36 0.71 .83
Note:
a
Univariate analysis with baseline value as covariate.
b
MAQ, medication adherence questionnaire, score ranges 04.
c
DAI, Drug Attitude Inventory, score ranges 010.
d
LCS, life chart schedule, score ranges 15.

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1247
Effects of Motivational Interviewing on Adherence
Hospitalization Rates
As shown in table3, in both groups, 44% of patients were
hospitalized at baseline. At T1 (postintervention), there
was a nonsignifcant difference between the two groups.
At T2 (6 months follow-up), 27% of patients in the MI
group were hospitalized vs 40% in the control group
(P=.19).
Next, we performed analyses concerning specifc sub-
groups. In the group of patients younger than 35 years
(n = 43), 14% of patients were hospitalized during the
follow-up period in the MI group vs 50% in the control
group (
2
= 6.24, df = 1, P = .012, = 0.38). In the
group >35years (n=50), there were no signifcant differ-
ences between conditions (P=.62).
Nine percent of female patients in the MI condi-
tion was hospitalized vs 63% in the control condition
(
2
= 6.12, df = 1, P = .041, = 0.57). Among male
patients (n = 74), there were no signifcant differences
between conditions (P=.81).
In the group with a negative urine analysis on cannabis
use at baseline (n = 40), 20% of patients were hospital-
ized in the MI condition vs 53% in the control condition
(
2
= 4.75, df = 1, P = .041, = 0.35). Among those
with a positive test on cannabis (n = 10), no signifcant
differences were found (P=.48), just as in the group of
patients (n = 43) that refused urine analysis (P = .48).
Among patients with an illness duration shorter than
6years (n=41), 14% of patients were hospitalized in the
MI condition vs 42% in the control condition (
2
=4.21,
df=1, P=.040, =0.33). In the group with a longer
illness duration (n=52), there were no signifcant differ-
ences between conditions (P=.93).
Regarding the route of administration of medication,
there were no signifcant differences between MI and HE
on hospitalization rates in the group of patients on oral
medication (n = 68, P = .27) nor in the group on depot
medication (n=25, P=.38).
Psychopathology
Total PANSS scores showed no signifcant interaction
between intervention type and time (P=.68). There was
a large effect for time, with both groups showing a reduc-
tion in severity of psychopathology (F(2,110) = 5.59,
P=.005 [partial
2
= 0.09]), but there were no differences
between the two interventions (P=.99).
There were also no differences between interventions
on PANSS subscales for positive symptoms (P = .83),
negative symptoms (P = .52), or general symptoms
(P=.87) (see table4).
Focusing on potential treatment effects in terms of
symptoms for specifc subgroups, female patients showed
a larger decrease in general PANSS symptoms in the MI
group ( 7.9, SD = 4.0) compared with the HE group
( 3.4, SD=2.4); t (11)=2.40, P=.035. For the other
variables, there were no signifcant effects on changes in
PANSS scores.
Discussion
This study aimed to investigate the effect of an adapted
form of MI on medication adherence in patients with
multi-episode schizophrenia that were unstable due to
medication nonadherence. Regarding adherence and hos-
pitalization rates, we did not fnd signifcant differences in
adherence over time between MI andHE.
As previous studies incorporating MI in their inter-
vention have shown mixed results,
1419
we hypothesized
that the heterogeneity of the investigated patient groups
might have been of infuence. In addition, we therefore
investigated whether specifc subgroups of patients (dif-
fering on cannabis use, age, illness duration, gender, route
of medication administration) may beneft more from MI
than others.
Besides a small difference in adherence rates favoring
MI over HE in the group who used depot medication on
Table3. Effects of Interventions on Hospitalization Rates
MI (Ratio Hospitalized) HE (Ratio Hospitalized)
2
P
T0: baseline (n=114) 24/55 (44%) 26/59 (44%) 0.002 .963
T1: posttreatment (n=94) 17/45 (38%) 19/49 (39%) 0.01 .921
T2: 6-mo follow-up (n=93) 12/45 (27%) 19/48 (40%) 1.74 .187
Females (n=19) 1/11 (9%) 5/8 (63%) 6.12 .041
Males (n=74) 11/34 (32%) 14/40 (35%) 0.06 .810
Cannabis positive (n=10) 0/3 (0%) 3/7 (43%) 1.84 .475
Cannabis negative (n=40) 5/25 (20%) 8/15 (53%) 4.75 .041
Urine analysis refused (n=43) 7/17 (41%) 8/26 (31%) 0.49 .484
Age 35 (n=43) 3/21 (14%) 11/22 (50%) 6.24 .012
Age >35 (n=50) 9/24 (38%) 8/26 (31%) 0.25 .616
Duration of illness 6 y (n=41) 3/22 (14%) 8/19 (42%) 4.21 .040
Duration of illness >6 y (n=52) 9/23 (39%) 11/29 (38%) 0.01 .930
Depot antipsychotic (n=25) 2/10 (20%) 7/15 (47%) 1.32 .380
Oral antipsychotic (n=68) 10/35 (29%) 12/33 (36%) 1.20 .273
Note: HE, health education; MI, motivational interviewing.

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1248
E. Barkhof etal
the MAQ and the LCS, no other mediators of treatment
effects in terms of medication adherence werefound.
However, we did fnd favorable effects of MI with
regard to hospitalization rates. While cannabis use is
strongly associated with nonadherence in patients with
schizophrenia,
1,28
in our study, MI produced signifcant
lower hospitalization rates for the group who does not
use cannabis, compared with HE. This may imply that
MI is more suitable for improving adherence in patients
who do not use cannabis, and that for patients using can-
nabis specifc additional interventions are warranted. MI
itself may be of value in this respect, because it has shown
effcacy in the treatment of substance abuse, including
cannabis abuse.
11,29,30
Also, hospitalization rates in the MI condition were
found to be lower for patients younger than the median
age of 35 years and patients with shorter illness dura-
tion. A younger age and a shorter duration of illness
are strong risk factors for medication nonadherence.
1

Probably related to this, we found that the group who
dropped out were aged younger, but this difference was
equally distributed over the 2 interventions. This may
imply that although nonadherence in the older group
was less prevalent, it may have been more persistent and
less susceptible for MI. On the other hand, while in the
younger group nonadherence was more prevalent, it
may also have been more amendable, suggesting MI to
be a suitable tool for improving adherence for younger
patients who can be engaged in treatment. The fact that
there were gender-specifc treatment effects of MI result-
ing in reduced hospitalization rates in female patients is
surprising. In most observational studies, gender is not
found to be a risk factor for nonadherence.
1,31
Female
patients with schizophrenia generally show a more favor-
able course of illness, probably due to the fact that there
is a later onset of illness and they show less severe nega-
tive and residual symptoms.
32
Nevertheless, we have
found no correlations with severity of psychopathology
and illness duration between the female responders and
nonresponders. Although none of the female patients
in our study used cannabis and there is less cannabis
abuse in female patients with schizophrenia in general,
33

the gender difference in the effectiveness of MI on hos-
pitalization rates cannot be explained by cannabis as a
confounder, because in the male non-cannabis users, no
differences between interventions werefound.
As hospitalization rates do not reveal the total time
spent in hospital, we compared our results with the dura-
tion of hospitalization between interventions, which
showed virtually the same differences between the sub-
groups of patients (see online supplementary material).
This is an important outcome, because the time spent in
hospital interferes social functioning and hospitalization
is associated with increased costs.
2
The fact that we found differences in subgroups on
hospitalization rates, but not in adherence measures, is
remarkable because adherence is strongly associated with
relapse rates.
3
Apossible explanation for this fnding may
be that there actually was an improvement in adherence,
but that measures used in this study are not reliable or
sensitive enough. In our study, adherence and attitudes
toward medication were measured using the MAQ and
Table4. Psychopathology Rates Across Interventions
MI (n=30) HE (n=32)
Mean SD Mean SD
PANSS total score
a
T0 (baseline) 72.0 17.9 72.0 17.5
T1 (posttreatment) 65.6 22.0 63.5 16.9
T2 (6-mo follow-up) 64.0 30.3 66.2 16.7
PANSS-positive symptoms
b
T0 (baseline) 16.2 5.87 17.2 6.69
T1 (posttreatment) 15.2 6.29 15.0 6.05
T2 (6-mo follow-up) 15.7 8.84 15.9 6.32
PANSS-negative symptoms
c
T0 (baseline) 18.7 5.80 19.1 6.56
T1 (posttreatment) 16.0 5.83 16.4 6.53
T2 (6-mo follow-up) 16.2 7.31 17.3 6.66
PANSS general symptoms
d
T0 (baseline) 37.7 9.74 35.8 8.28
T1 (posttreatment) 35.4 12.98 32.2 7.07
T2 (6-mo follow-up) 32.1 14.33 32.2 7.89
Note: Abbreviations are explained in the frst footnote to tables 1 and 3.
a
Total score ranges 30210.
b,c
Positive and negative subscales score ranges 749.
d
General subscale score ranges 16112.

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1249
Effects of Motivational Interviewing on Adherence
the DAI, which both rely on subjective information of
patients, and the LCS, of which we used information on
adherence by the caregiver and/or a relative. This means
that we have used multiple instruments and did not
only rely on the patient as the sole information source.
However, several authors suggest that all subjective
reports concerning medication adherence are considered
to be relatively unreliable for determining adherence.
6,34

Nevertheless, the information obtained from the pro-
fessional caregiver in the LCS data in the subgroup of
patients on depot medication can be regarded as an excep-
tion to this. Therefore, our fnding that patients on depot
medication show higher adherence rates as assessed with
the LCS when they received MI compared with HE, is
thought to be reliable. Other measures that may provide
more reliable information on adherence are, eg, the level
of medication in blood samples or medication-monitor-
ing devices. However, the latter is expensive and the use
of invasive measures may negatively infuence the willing-
ness of patients to participate in research. Because the
focus of our study was to include the most troublesome
patients we realized that the willingness to participate
would be problematic, so we decided that extra barriers
to inclusion were to be avoided. Another problem with
the adherence measures in our study is that a ceiling
effect of the LCS and the MAQ might have been operat-
ing. Therefore, hospitalization may be regarded as a more
reliable and valid outcome measure to assess medication
adherence in this population. On the other hand, it can-
not be ruled out that the observed reduction in hospital-
ization rates is caused by other factors than medication
adherencealone.
In our study, psychopathology rates improved for both
groups over time, but there were almost no differences
between MI and HE, even when subgroups were ana-
lyzed. Although this may appear incongruent with the
differences found in hospitalization rates, this is consis-
tent with other studies on adherence interventions, which
did not fnd changes in psychopathology even though
improvements in medication adherence or hospitaliza-
tion rates were found.
15,19,35
A possible explanation for
these observations might be that MI promotes a non-
judgemental attitude of caregiver toward the beliefs and
actions of a patient with regard to medication use. This
may result in a more open and trustful relation between
the patient and the caregiver, allowing for a better judge-
ment for additional support of the needs for functioning
in society, preventing rehospitalization.
Strengths and Limitations
Limitations of the present study are that subgroups
were relatively small and patients were not randomized
on these characteristics. Therefore, fndings concerning
subgroups are preliminary and require confrmation in a
future randomized controlled trial.
Secondly, both MI and HE were performed by thera-
pists who were not otherwise involved in the treatment
of patients, to ensure treatment integrity of the interven-
tions. This choice for external therapists renders uncer-
tainty whether in the meantime during the study period,
the treating caregiver(s) might have conducted other
interventions or therapeutic styles and to which extend
this may have infuenced the results.
Thirdly, in our analyses no corrections were applied
for multiple comparisons. Although often the Bonferroni
method is applied, others have argued that the report of
effect sizes in addition to signifcance levels may provide
suffcient information for a correct interpretation of the
results.
36
Fourthly, the proportion of patients that refused to
participate is high (39% of the identifed sample), but is
almost the same as in comparable studies.
17,19
This high
refusal rate (partly) refects the challenging nature of the
population, consisting of unstable patients with often
poor insight, who had recently been nonadherent. These
factors may result in a considerable reluctance to cooper-
ate with a researchtrial.
Finally, the dropout rate during the interventions (18%
in the MI group and 14% in the control group) was con-
siderable, although not unexpected given the severity of
the disorder in the included patients. Also there were
some patients lost to follow-up, resulting in diminished
power. Nevertheless, the attrition rates did not differ
between groups, so this will not likely have infuenced the
results. However, because of the limited power due to a
relatively small sample size, clinical relevant effects may
have been left undetected. This may especially apply to
the considerable difference in hospitalization rate in favor
of the MI condition (27% vs40%).
Strength of this study is that it was designed to include
those patients most in need for improvement of adher-
ence. These troublesome patients are often unwilling to
participate
37
or excluded from studies, because of the
chronicity of their illness or concomitant drug abuse.
Furthermore, the control group received an active inter-
vention (HE), thereby ensuring that the effects are not
infuenced by the amount of attention that was given in
the intervention condition. Another strength is that we
used both subjective (adherence scores) and objective
(hospitalization) outcome measures, the latter represent-
ing a reliable and valid measure of outcome.
Moreover, we were able to identify possible character-
istics of patients for whom MI may be a suitable interven-
tion to improve adherence.
In conclusion, this study shows that an adapted form
of MI does not produce a signifcant effect on medica-
tion adherence or hospitalization, compared with HE
in a group of nonadherent patients with multi-episode
schizophrenia. However, the results provide indications
that MI may yet be suitable for improving adherence in
female patients, non-cannabis users, younger patients,

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1250
E. Barkhof etal
and those with shorter illness duration. Therefore,
targeted use of MI may be of beneft for improving
medication adherence in certain groups of patients,
although this needs further examination. Furthermore,
our fndings underscore the need to focus on specifc
targets that lead to nonadherence and to apply an indi-
vidualized approach for each patient of this challeng-
ing group.
Supplementary Material
Supplementary material is available at http://schizophre
niabulletin.oxfordjournals.org.
Funding
This work was supported by an investigator-initiated and
nonrestricted grant by the Dr. Paul Janssen Foundation.
Acknowledgments
We thank all the participating patients and staff at the
Department of Psychiatry at the Academic Medical
Centre (Amsterdam) and at the mental health institutions
InGeest (Amsterdam, Haarlem) and Arkin (Amsterdam),
and especially the psychologists and community mental
health nurses who contributed to the interventions and
assessments: Rinske Schepers, Frank van der Horst, Dick
Bloemzaad, Anja Lek, Carolien Zeylmaker, and Ronald
van Gool. The authors have declared that there are no
conficts of interest in relation to the subject of this study.
References
1. Lacro JP, Dunn LB, Dolder CR, Leckband SG, Jeste DV.
Prevalence of and risk factors for medication nonadherence
in patients with schizophrenia: a comprehensive review of
recent literature. J Clin Psychiatry. 2002;63:892909.
2. Weiden PJ, Olfson M. Cost of relapse in schizophrenia.
Schizophr Bull. 1995;21:419429.
3. Robinson D, Woerner MG, Alvir JM, et al. Predictors of
relapse following response from a frst episode of schizo-
phrenia or schizoaffective disorder. Arch Gen Psychiatry.
1999;56:241247.
4. Heeg B, Buskens E, Knapp M, et al. Modelling the treated
course of schizophrenia: development of a discrete event sim-
ulation model. Pharmacoeconomics. 2005;23(suppl 1):1733.
5. Barkhof E, Meijer CJ, de Sonneville LM, Linszen DH, de
Haan L. Interventions to improve adherence to antipsychotic
medication in patients with schizophreniaa review of the
past decade. Eur Psychiatry. 2012;27:918.
6. Velligan DI, Weiden PJ, Sajatovic M, etal. Expert Consensus
Panel on Adherence Problems in Serious and Persistent
Mental Illness. The expert consensus guideline series: adher-
ence problems in patients with serious and persistent mental
illness. J Clin Psychiatry. 2009;70(suppl 4):146; quiz 47.
7. Miller WR, Rollnick S. Motivational Interviewing: Preparing
People for Change. New York, NY: Guilford Press; 2002.
8. Bien TH, Miller WR, Tonigan JS. Brief interventions for
alcohol problems: a review. Addiction. 1993;88:315335.
9. Noonan WC, Moyers TB. Motivational interviewing: a
review. J Subst Misuse. 1997;2:816.
10. Harland J, White M, Drinkwater C, Chinn D, Farr L, Howel
D. The Newcastle exercise project: a randomised controlled
trial of methods to promote physical activity in primary care.
BMJ. 1999;319:828832.
11. Burke BL, Arkowitz H, Menchola M. The effcacy of moti-
vational interviewing: a meta-analysis of controlled clinical
trials. J Consult Clin Psychol. 2003;71:843861.
12. Lundahl B, Burke BL. The effectiveness and applicability of
motivational interviewing: a practice-friendly review of four
meta-analyses. J Clin Psychol. 2009;65:12321245.
13. Barkhof E, de Haan L, Meijer CJ, et al. Motivational
interviewing in psychotic disorders. Curr Psychiatr Rev.
2006;2:207213.
14. Kemp R, Hayward P, Applewhaite G, Everitt B, David A.
Compliance therapy in psychotic patients: randomised con-
trolled trial. BMJ. 1996;312:345349.
15. Kemp R, Kirov G, Everitt B, Hayward P, David A.
Randomised controlled trial of compliance therapy. 18-month
follow-up. Br J Psychiatry. 1998;172:413419.
16. ODonnell C, Donohoe G, Sharkey L, etal. Compliance ther-
apy: a randomised controlled trial in schizophrenia. BMJ.
2003;327:834.
17. Gray R, Leese M, Bindman J, et al. Adherence therapy for
people with schizophrenia. European multicentre randomised
controlled trial. Br J Psychiatry. 2006;189:508514.
18. Maneesakorn S, Robson D, Gournay K, Gray R. An RCT of
adherence therapy for people with schizophrenia in Chiang
Mai, Thailand. J Clin Nurs. 2007;16:13021312.
19. Staring AB, Van der Gaag M, Koopmans GT, etal. Treatment
adherence therapy in people with psychotic disorders: ran-
domised controlled trial. Br J Psychiatry. 2010;197:448455.
20. Leucht S, Heres S. Epidemiology, clinical consequences, and
psychosocial treatment of nonadherence in schizophrenia. J
Clin Psychiatry. 2006;67(suppl 5):38.
21. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured
Clinical Interview for DSM-IV Axis I Disorders. New York,
NY: Biometrics Research Department, New York State
Psychiatric Institute; 1996.
22. Guy W. ECDEU Assessment Manual for Psychopharmacology.
Rockville, MD: US Department of Health, Education and
Welfare Publication (ADM), National Institute of Mental
Health; 1976:76338.
23. Chisholm D, Knapp MR, Knudsen HC, Amaddeo F, Gaite L,
van Wijngaarden B. Client Socio-Demographic and Service
Receipt Inventory - European Version: development of an
instrument for international research. Br J Psychiatry Suppl.
2000;177:s28s33.
24. Morisky DE, Green LW, Levine DM. Concurrent and
predictive validity of a self-reported measure of medication
adherence. Med Care. 1986;24:6774.
25. Susser E, Finnerty M, Mojtabai R, et al. Reliability of the
life chart schedule for assessment of the long-term course of
schizophrenia. Schizophr Res. 2000;42:6777.
26. Hogan TP, Awad AG, Eastwood R. A self-report scale pre-
dictive of drug compliance in schizophrenics: reliability and
discriminative validity. Psychol Med. 1983;13:177183.
27. Kay SR, Fiszbein A, Opler LA. The positive and negative
syndrome scale (PANSS) for schizophrenia. Schizophr Bull.
1987;13:261276.

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1251
Effects of Motivational Interviewing on Adherence
28. Kashner TM, Rader LE, Rodell DE, Beck CM, Rodell LR,
Muller K. Family characteristics, substance abuse, and hospi-
talization patterns of patients with schizophrenia. Psychiatr
Serv. 1991;42:195197.
29. Hettema J, Steele J, Miller WR. Motivational interviewing.
Annu Rev Clin Psychol. 2005;1:91111.
30. Smeerdijk M, Keet R, Dekker N, et al. Motivational
interviewing and interaction skills training for parents to
change cannabis use in young adults with recent-onset
schizophrenia: a randomized controlled trial. Psychol Med.
2012;42:16271636.
31. Fenton WS, Blyler CR, Heinssen RK. Determinants of med-
ication compliance in schizophrenia: empirical and clinical
fndings. Schizophr Bull. 1997;23:637651.
32. Ochoa S, Usall J, Cobo J, Labad X, Kulkarni J. Gender
differences in schizophrenia and frst-episode psychosis: a
comprehensive literature review. Schizophr Res Treatment.
2012;2012:916198.
33. Foti DJ, Kotov R, Guey LT, Bromet EJ. Cannabis use and the
course of schizophrenia: 10-year follow-up after frst hospi-
talization. Am J Psychiatry. 2010;167:987993.
34. Kikkert MJ, Barbui C, Koeter MW, etal. Assessment of medi-
cation adherence in patients with schizophrenia: the Achilles
heel of adherence research. J Nerv Ment Dis. 2008;196:274281.
35. Valenstein M, Kavanagh J, Lee T, et al. Using a pharmacy-
based intervention to improve antipsychotic adherence
among patients with serious mental illness. Schizophr Bull.
2011;37:727736.
36. Perneger TV. Whats wrong with Bonferroni adjustments.
BMJ. 1998;316:12361238.
37. Storosum J, van Zwieten B, de Haan L. Informed consent
from behaviourally disturbed patients. Lancet. 2002;359:83.

a
t

L
i
b
r
a
r
y

o
f

t
h
e

L
u
c
i
a
n

B
l
a
g
a

U
n
i
v
e
r
s
i
t
y

o
f

S
i
b
i
u

o
n

N
o
v
e
m
b
e
r

1
3
,

2
0
1
3
h
t
t
p
:
/
/
s
c
h
i
z
o
p
h
r
e
n
i
a
b
u
l
l
e
t
i
n
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m