International Journal of Psychiatry in Clinical Practice, 2013; 17: 2–9

ORIGINAL ARTICLE

Factors associated with relapse in patients with schizophrenia

LUIS SAN1, MIQUEL BERNARDO2,3,4,5, AGUSTÍN GÓMEZ6 & MARIO PEÑA7

1Department of Child and Adolescent Psychiatry, Hospital Sant Joan de Déu, CIBERSAM, Esplugues del Llobregat, Barcelona,
Spain, 2Department of Psychiatry, Hospital Clínic de Barcelona, Barcelona, Spain, 3Institut d’Investigacions Biomèdiques August
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

Pi i Sunyer (IDIBAPS), 4Centro de Investigación Biomédica en Red de Salud Mental (CIBERSAM), 5Department of
Psychiatry and Clinical Psychobiology, University of Barcelona, Spain, 6Research and Clinical Epidemiology Unit, Hospital
Universitario 12 de Octubre/CIBER de Epidemiología y Salud Pública, Madrid, Spain, and 7Medical Affairs Manager,
Janssen Iberia, Madrid, Spain

Abstract
Objective. To assess risk factors for relapse in patients with schizophrenia attended in daily practice. Methods. Patients with
schizophrenia admitted consecutively to short-stay/acute-care psychiatric units over a 6-month period were eligible. Vari-
ables statistically significant in the univariate logistic regression analysis were then subjected to multivariate analysis. Results.
The study population included 1646 patients (67.6% men). In the univariate analysis, low family support, duration of ill-
For personal use only.

ness ⬎ 5 years, number of previous hospitalizations, cocaine and cannabis consumption, and number of different antipsy-
chotic drug classes were risk factors for relapse. In the multivariate analysis, number of previous hospitalizations (odds ratio
[OR] 1.29, 95% confidence interval [CI] 1.21–1.36) and number of different antipsychotics previously used (OR ⫽ 1.13,
95% CI 1.03–1.24) were significant predictors of relapse. The absence of cannabis consumption was a protective factor
(OR ⫽ 0.72, 95% CI 0.58–0.89). Neither adherence to treatment in the previous 3 years nor type of antipsychotic regimen
was significantly associated with relapse. Conclusions. Number of previous hospitalizations and number of different types
of antipsychotic drugs were associated with relapse. Absence of cannabis consumption was a protective factor.

Key Words: Schizophrenia, antipsychotic agents, patient compliance, treatment outcome

Introduction countries have demonstrated that relapse is one of

Schizophrenia is a chronic disabling illness that the most problematic and costly aspects of schizo-
affects about 1% of the population [1]. It is a hetero- phrenia [6–9].
geneous disorder with variable aetiological, prognos- One of the most important predictors of relapse in
tic and treatment response patterns [2]. Its course is schizophrenia is adequacy and adherence to antipsy-
generally long term, with acute psychotic exacerba- chotic medication. Continuous prophylactic antipsy-
tions that may require hospitalization. The chronic chotic treatment reduces the risk of relapse by
and relapsing nature of the illness leads to more approximately 70% [10]. Relapse may be caused by
health care resources to manage than other single discontinuation of antipsychotic medication and there
psychiatric disorder. The main cost of treatment is is a real need to optimize patients’ adherence in clinical
hospitalization as a result of exacerbation of symp- practice. Only a third of schizophrenic patients can be
toms often caused by non-compliance with antipsy- considered to be fully compliant and another third to
chotic medication [3–5]. Therefore, treatment costs be partially compliant, such as reducing the prescribed
may be reduced by the prevention of psychotic symp- dose or only taking the medication from time to time
toms. Numerous studies carried out in different [11]. Even short medication gaps (including periods

Correspondence: Dr Luis San, Department of Child and Adolescent Psychiatry, Hospital Sant Joan de Déu, CIBERSAM, Passeig Sant Joan de Déu 2,
E-08950 Esplugues del Llobregat, Barcelona, Spain. Tel: ⫹34 93 32806349. Fax: ⫹34 93 6009454. E-mail: 12636lsm@comb.cat

(Received 16 September 2011; accepted 18 April 2012 )

ISSN 1365-1501 print/ISSN 1471-1788 online © 2013 Informa Healthcare
DOI: 10.3109/13651501.2012.687452
 Risk factors for relapse in schizophrenia 3
from 1 to 10 days) are associated with an increase in from short-stay or acute-care psychiatric units of the
the risk of hospitalization [12]. Non-compliance with participating public hospitals of the Spanish National
antipsychotic medication is often considered to be the Health Care System over a 3-month period were eli-
most important factor related to relapse. The most gible. Patients were recruited during the second half
commonly identified patient-related factors influenc- of 2006 (index hospitalization). Inclusion criteria
ing compliance are poor insight, cognitive impairment, were as follows: males and females aged 18 years or
psychiatric comorbidity, substance use and duration over, definite diagnosis of schizophrenia or schizoaf-
of untreated illness [13,14]. Treatment-related issues fective disorder according to DSM-IV criteria [17]
that may affect compliance are primarily related to the and duration of disease of at least 2 years. Access to
efficacy and tolerability of antipsychotics [15,16]. the patient’s medical history for the previous 3 years
Environmental factors, such as the degree of family was also required. The requirement to have available
and social support available, are also accurate predic- the patient’s history for the previous 3 years was
tors of adherence. Finally, the principle physician- applied to patients with a duration of disease greater
related factor is the therapeutic alliance between than 3 years. For patients with a shorter duration,
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

patients and health care professionals [6]. the requirement was to have access to the medical
Moreover, not only the risk for a relapse after a history since the time of diagnosis of schizophrenia.
schizophrenic episode remains increased throughout Pregnant and lactating women, women intended to
the patient’s lifetime, but also risk for chronic disorder become pregnant in the next 12 months, and patients
increases with every relapse-readmission event. Pre- currently participating in a clinical trial were excluded
vention of relapse is a major challenge and a primary from the study.
focus in the treatment of patients with schizophrenia The study protocol was approved by the Ethics
both from the clinical and economic perspectives. Committee of Hospital Clínic i Provincial (Barce-
The present observational study was conducted lona, Spain) and performed in accordance with the
to assess risk factors for relapse in a population of principles of the Declaration of Helsinki (version
patients with schizophrenia treated in routine daily 1989) and its amendments. Written informed con-
For personal use only.

practice. The results obtained will further contribute sent was obtained from all participants prior to
to develop strategies to reduce the risk of relapse in enrollment in the study.
the real-world setting.
Study variables
Methods At the time of discharge from index hospitalization,
Study design the following data were recorded in the case report
form (CRF): sex; age; ethnicity; educational level,
This was an epidemiological, muticenter and non- categorized as none, primary education, secondary
interventional study. The main objective of the study education, high school, university degree; history of
was to identify which are the most important risk stressor events in the previous 3 years; family support
factors for relapse in schizophrenia. It was a cross- according to level of contact with the patient in the
sectional study that included both a retrospective daily life and categorized as none, low, medium, high,
part (that retrieved data of the last 3 years before the very high; history of substance use, including can-
current hospitalization) and a prospective part that nabis, cocaine, heroin; current main diagnosis, such
retrieved data during a 12-month follow-up period as schizophrenia, schizoaffective disorder, schizo-
after the index hospitalization. The primary objective phreniform disorder; duration of the disease, catego-
of the study was to assess the main risk factors for rized as ⬎ 2–5, ⬎ 5–10, ⬎ 10–15, ⬎ 15–20, ⬎ 20
relapse. The secondary objective was to determine years; number of hospitalizations in the preceding
risk factors for relapse according to antipsychotic 3 years, categorized as 1, 2, 3, 4, 5, and ⬎ 5; current
drug regimen prescribed at discharge of the index Clinical Global Impression Severity scale (CGI-S) (a
(current) hospitalization. Hospital admission was seven-point scale, 1 ⫽ normal, 2 ⫽ borderline men-
used as the central criterion for relapse indepen- tally ill, 3 ⫽ mildly ill, 4 ⫽ moderately ill, 5 ⫽ mark-
dently of exacerbation of any psychiatric symptoms, edly ill, 6 ⫽ severely ill, 7 ⫽ extremely ill) [18]; main
presence of prodromal signs of exacerbation, reduced reason for the index admission, including worsening
social functioning or behavioural disturbance. of psychotic symptoms because of lack of efficacy or
non-adherence to antipsychotic regimen, intensifica-
tion of psychotic symptoms and deliberated self-
Patients
harm or suicidal ideation, and other causes; history
All consecutive patients with schizophrenia and of pharmacological treatment, including number of
schizoaffective disorder admitted to and discharged different antipsychotic agents, type of antipsychotic
 4 L. San et al.
drugs, such as second-generation or atypical agents Table I. Demographic and clinical characteristics of 1646 patients
included in the study.
and conventional (first-generation or typical) drugs,
and route of administration (depot, long-acting Variable Number (%)
injectable and oral); psychiatrist’s opinion on adher-
Sex
ence to pharmacological treatment rated from 1 to Men 1112 (67.6)
9 (1, 2 and 3: poor; 4, 5 and 6: moderate; 7, 8 and Women 534 (32.4)
9: maximum); and relapse/re-admission at 6 and Age, years
12 months. ⬍ 20 29 (1.8)
20–25 121 (7.4)
25–30 220 (13.4)
30–35 308 (18.7)
Statistical analysis 35–40 277 (16.8)
According to the primary objective of the study to 40–45 274 (16.6)
45–50 170 (10.3)
identify predictive factors for a psychotic relapse, 50–55 109 (6.6)
some of which have an expected low prevalence 55–60 59 (3.6)
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

(⬍ 10%), a sample size of 2000 patients provided 60–65 48 (2.9)

80% power at two-sided alpha level of 0.05 to detect ⬎ 65 31 (1.9)
an odds ratio (OR) of at least 1.4 for a predictive Educational level
None 117 (7.1)
variable to be clinically relevant. To this purpose the Primary education 796 (48.4)
participation of 300 psychiatrists recruiting seven Secondary education 355 (21.6)
consecutive patients each was required. The Kolm- High school 276 (16.8)
ogorov–Smirnov test was used to assess normal dis- University degree 102 (6.2)
tribution of data. Categorical variables are expressed Family support
None 165 (10.0)
as absolute numbers and percentages and continu- Low 395 (24.0)
ous variables as mean ⫾ standard deviation (SD) for Medium 519 (31.5)
normally distributed data and as median and inter-
For personal use only.

High 461 (28.0)

quartile range (IQR) (25th–75th percentile) for Very high 106 (6.4)
variables whose distribution departed from normal- Psychiatric diagnosis
Schizophrenia 1272 (77.3)
ity. Univariate logistic regression analysis for each Schizoaffective disorder 330 (20.0)
variable with “relapse” as the outcome of interest Schizophreniform disorder 44 (2.7)
(dependent variable) was performed. To assess the Duration of disease, years
predictors of hospitalizations, variables associated ⬎ 2 and ⱕ 5 360 (21.9)
with relapse/re-admission were subjected to multi- ⬎ 5 and ⱕ 10 344 (20.9)
⬎ 10 and ⱕ 15 326 (19.8)
variate analysis with a logistic regression procedure
⬎ 15 and ⱕ 20 254 (15.4)
and a forward stepwise selection of P ⬍ 0.10 after ⬎ 20 362 (22.0)
univariate testing. Odds ratio (OR) and 95% confi- Drug use in the previous 3 years
dence intervals (CIs) were calculated. A significance Cannabis 553 (33.6)
level of 0.05 was considered for all the statistical Cocaine 269 (16.3)
tests. The Statistical Analysis Systems (SAS) (SAS Heroin 45 (2.7)
Stressor event in the previous 3 years 679 (41.3)
Institute, Cary, NC, USA) version 9.1.3 for Windows Current Clinical Global Impression Severity
was used to analyse the data. (CGI-S) scale
1 ⫽ normal 2 (0.1)
2 ⫽ borderline mentally ill 4 (0.2)
3 ⫽ midly ill 181 (11.0)
Results 4 ⫽ moderately ill 609 (37.0)
5 ⫽ markedly ill 558 (33.9)
A total of 2067 patients were recruited by 291 inves- 6 ⫽ severely ill 264 (16.0)
tigators throughout Spain. However, 421 patients 7 ⫽ extremely ill 28 (1.7)
were excluded because of missing or incomplete data
for the previous 3 years. Therefore, the study popula-
tion consisted of 1646 patients (67.6% men) with a Schizophrenia was the most frequent diagnosis
mean age of 38.2 (11.1) years (range 18–79 years). (77.3% of the cases) followed by schizoaffective dis-
The age of the patients ranged between 30 and 45 order (20%), and schizophreniform disorder (2.7%).
years in 63.8% of cases. The characteristics of the In almost 60% of the patients, the duration of illness
patients are shown in Table I. Family support was was longer than 10 years. In relation to the CGI-S,
considered “high” in only 28% of patients and “very the most frequent categories were “moderately ill”
high” in only 6.4%. Primary education was the (37%) and “markedly ill” (33.9%). Stressors events
most frequent level achieved (48.4% of patients). were present in 41.3% of patients. Cannabis use was
 Risk factors for relapse in schizophrenia 5
recorded in 33.6% of patients, cocaine in 16.3%, and (SD) number of 2.0 (1.2) different antipsychotic
heroin in 2.7%. drugs (median 2, IQR 1-3). Adherence to antipsy-
Data of previous hospitalizations and treatment chotic treatment during the previous 3 years was
are shown in Table II. The mean (SD) number of rated as poor in 31.3% of cases, moderate in 40.7%
previous hospitalizations was 2.7 (2.2) (median 2, and high in 28%. The mean (SD) adherence rate was
IQR 1-3), with 38.2% of patients admitted one time, 4.6 (2.2) (median 5, IQR 3-6). There were statisti-
21.7% two times, and 15.2% three times. However, cally significant differences in adherence according
10.4% of the patients were admitted for more than to antipsychotic medication taken in the previous
five times. Patients had been treated with a mean 3 years (P ⬍ 0.0001), with better rates for depot atyp-
ical drugs followed by depot typical agents, oral
atypical and oral typical antipsychotics. At the time
Table II. Hospitalizations and pharmacological treatment.
of discharge from the index hospitalization in short-
Variable Number (%) stay/acute-care psychiatric units, atypical antipsy-
chotics were administered to 89.6% of patients
Hospitalizations in the previous 3 years
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

1 629 (38.2) (long-acting second-generation formulations in 50.3%

2 358 (21.7) of them) and conventional antipsychotics to only
3 251 (15.2) 10.4% of patients (depot formulations in 7.3% of
4 144 (8.7) them) (Table II).
5 92 (5.6)
Results of univariate logistic regression analysis
⬎5 172 (10.4)
Number of different antipsychotic are shown in Table III. Differences in relation to age
drugs in the previous 3 years and gender were not found, although age ⬍ 20 years
0 65 (3.9) showed a higher odds ratio as compared with the
1 562 (34.1) other age groups. Family support of any kind was a
2 574 (34.9)
protective factor for relapse as well as no drug abuse
3 255 (15.5)
(either cannabis or cocaine as compared with can-
For personal use only.

4 127 (7.7)
5 39 (2.4) nabis or cocaine consumption). The risk of relapse
⬎5 24 (1.5) was higher for duration of disease longer than 5 years
Adherence (n ⫽ 2937)∗ as compared with disease’s duration between 2 and
Poor 918 (31.3)
5 years. The number of hospitalizations in the previ-
Moderate 1194 (40.7)
Maximum 821 (28.0) ous 3 years (excluding the index hospitalization) was
Type of antipsychotic drugs in the previous a significant risk factor for relapse, as was the number
3 years and adherence† of different antipsychotic classes used in the previous
Atypical antipsychotics 3 years.
Depot injection (n ⫽ 370)
Table IV shows the results of the univariate logis-
Poor 82 (22.2)
Moderate 133 (35.9) tic regression analyses corresponding to the four
Maximum 155 (41.9) groups of antipsychotic drug regimens. Statistically
Oral route (n ⫽ 1730) significant factors associated with relapse included
Poor 586 (33.9) schizophrenia and schizoaffective disorders, number
Moderate 697 (40.3)
of hospitalizations in the previous 3 years, cocaine or
Maximum 447 (25.8)
Conventional (typical) antipsychotics cannabis consumption, and number of different
Depot injection (n ⫽ 262) antipsychotic drugs in the previous 3 years.
Poor 59 (22.5) In the multivariate analysis, number of hospital-
Moderate 112 (42.7) izations in the previous 3 years (OR ⫽ 1.28, 95%
Maximum 91 (34.7)
CI 1.21–1.31) and number of different antipsychot-
Oral route (n ⫽ 571)
Poor 191 (33.5)
ics (OR ⫽ 1.13, 95% CI 1.03–1.24) were indepen-
Moderate 252 (44.1) dently associated with relapse, whereas the absence
Maximum 128 (22.4) of cannabis use was inversely associated with relapse
Pharmacological treatment at discharge from (OR ⫽ 0.72, 95% CI 0.58–0.89).
index hospitalization (n ⫽ 1643) The number of hospitalizations in the previous
Atypical antipsychotics 1472 (89.6)
Depot injection 827 (50.3) 3 years was also a predictor of relapse for treatment
Oral route 645 (39.3) with depot atypical drugs (OR ⫽ 1.29, 95% CI 1.21–
Conventional (typical) antipsychotics 171 (10.4) 1.36), oral atypical drugs (OR ⫽ 1.30, 95% CI 1.20–
Depot injection 120 (7.3) 1.42) and depot typical drugs (OR ⫽ 129, 95%
Oral route 51 (3.1)
CI 1.11–1.51). Number of different antipsychotic
*Total number of treatments assessed. classes was also significant for treatment with depot
†P ⬍ 0.0001. atypical drugs (OR ⫽ 1.13, 95% CI 1.03–1.24).
 6 L. San et al.
Table III. Results of univariate logistic regression analysis.

Comparator Odds ratio

Variable category (95% confidence interval) P value

Age, years ⬎ 65 years 0.240

⬍ 20 5.61 (1.81–17.32)
20–25 1.82 (0.72–4.59)
25–30 2.08 (0.86–5.03)
30–35 2.31 (0.97–5.53)
35–40 2.26 (0.94–5.42)
40–45 1.94 (0.81–4.67)
45–50 2.34 (0.96–5.74)
50–55 2.41 (0.96–6.10)
55–60 2.04 (0.75–5–50)
60–65 1.71 (0.61–4.82)
Female sex Male sex 0.92 (0.74–1.14) 0.448
Family support None 0.008
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

Low 0.74 (0.52–1.07)

Medium 0.61 (0.43–0.87)
High 0.55 (0.38–0.78)
Very high 0.56 (0.33–0.92)
Education level University degree 0.833
None 1.01 (0.59–1.74)
Primary education 1.06 (0.70–1.64)
Secondary education 0.97 (0.61–1.52)
High school 0.92 (0.57–1.47)
Main current diagnosis Schizophreniform disorder 0.007
Schizophrenia 1.12 (0.60–2.12)
Schizoaffective disorder 1.65 (0.85–3.19)
For personal use only.

Duration of disease, years Between 2 and 5 0.044

5–10 1.11 (0.82–1.52)
10–15 1.31 (0.96–1.79)
15–20 1.29 (0.92–1.79)
More than 20 1.56 (1.16–2.11)
Compliance previous 3 years Maximum 0.417
Poor 1.02 (0.79–1.32)
Moderate 1.16 (0.90–1.51)
No. hospitalizations previous 3 years None 1.32 (1.26–1.40) ⬍ 0.0001
Drug use in the previous 3 years Yes
Cannabis 0.64 (0.52–0.79) ⬍ 0.0001
Cocaine 0.69 (0.53–0.90) 0.005
Heroin 1.03 (0.56–1.90) 0.911
No. different antipsychotic drugs used in the The same antipsychotic drug 1.34 (1.24–1.46) ⬍ 0.0001
previous 3 years
Type of antipsychotic treatment in the Long-acting injectable 0.573
previous 3 years
Atypical oral, typical depot/oral 0.94 (0.77–1.15)

Discussion The beneficial effect of family support on the out-

come of schizophrenia has been widely recognised.
In this large population of patients with schizophre-
nia and schizoaffective disorders admitted to short- People with schizophrenia from families that express
stay or acute-care psychiatric units throughout Spain high levels of expressed emotion (EE) (criticism, hos-
because of an episode of relapse and for which data tility or over involvement) have more frequent relapses
for the previous 3 years was available, low family than people with similar problems from families that
support, longer duration of illness, greater number tend to be less expressive of emotions [19]. In a
of previous hospitalizations, cocaine and cannabis sample of 60 Spanish schizophrenic patients studied
consumption and greater number of different antip- to ascertain the relationship between their relatives’
sychotic drugs were risk factors for relapse. In the EE and relapse at follow-up, there was a tendency
multivariate analysis, number of previous hospitaliza- for patients who interrupted their medication or who
tions, cannabis consumption and number of different did not work to relapse more frequently, particularly
antipsychotics previously used were significant pre- among the high-EE group [20]. In a study carried
dictors of relapse. out in Sydney, there was a significant association
 Risk factors for relapse in schizophrenia 7
Table IV. Results of univariate logistic regression analysis according to class of antipsychotic treatment and
route of administration at discharge from current hospitalization.

Odds ratio
Type of antipsychotic (number of patients) (95% confidence interval) P value

Atypical drugs depot (n ⫽ 827)

Main diagnosis 0.005
Schizophrenia vs. schizophreniform disorder 1.21 (0.49–3.02)
Schizoaffective vs. schizophreniform disorder 2.20 (0.85–5.75)
Number hospitalizations in the previous 3 years 1.37 (1.27–1.49) ⬍ 0.0001
Drug use in the previous 3 years
Cannabis (no vs. yes) 0.72 (0.54–0.97) 0.030
Cocaine (no vs. yes) 0.65 (0.45–0.94) 0.020
Number of different antipsychotics in the previous 3 years 1.42 (1.25–1.62) ⬍ 0.0001
Atypical drugs oral route (n ⫽ 645)
Number hospitalizations in the previous 3 years 1.31 (1.21–1.42) ⬍ 0.0001
Drug use in the previous 3 years
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

Cannabis (no vs. yes) 0.60 (0.42–0.84) 0.003

Cocaine (no vs. yes) 0.64 (0.41–0.99) 0.049
Number different antipsychotics in the previous 3 years 1.29 (1.14–1.46) ⬍ 0.0001
Conventional (typical) drugs depot (n ⫽ 120)
Number hospitalizations in the previous 3 years 1.29 (1.11–1.51) 0.013
Drug use in the previous 3 years
Cannabis (no vs. yes) 0.43 (0.21–0.92) 0.028
Number different antipsychotics in the previous 3 years 1.39 (1.05–1.85) 0.020
Conventional (typical) drugs oral route (n ⫽ 51)
Main diagnosis
Schizophrenia vs. schizophreniform disorder 0.16 (0.01–2.01) 0.017
For personal use only.

between returning to a high EE household and both increases as a function of the number of previous
re-hospitalization and relapse [21]. When family admissions [29,30]. In a cohort of 887 schizophrenic
members are involved in treatment, patients are more patients and using the hospital discharge database of
likely to avoid relapse, achieve a higher level of func- the Lazio region, Italy, 44% of the patients were read-
tioning and improve adherence to medication [22,23]. mitted at least once during a 4-year follow-up period
In the present study, the degree of family support was after the first psychiatric admission [31]. In the present
a significant factor for relapse in the univariate logis- series, 38% of patients were admitted to the hospital
tic regression analysis but this variable did not remain one time in the previous 3 years, 21.7% two times, and
significant in the multivariate model. 40% three or more times. Both in the univariate and
Duration of disease was also a risk factor for multivariate analyses, the number of previous hospi-
relapse. Longer duration of untreated psychosis pre- talizations were a significant risk factor for relapse.
dicts poorer outcomes [24]. It has been shown than The association between drug abuse and risk of
delay in the initial onset of treatment for psychosis is relapse has been also demonstrated [32–35]. Sub-
associated with a poorer response to treatment in stance using schizophrenics (dual diagnosis) repre-
general, increased levels of disability, greater resources sent a more disturbed patient group, with greater
and more intensive interventions [25,26]. On the prevalence of suicide attempts [36]. Moreover, most
other hand, most patients who recover from a first of the excess risk of violence in schizophrenic patients
episode of schizophrenia or schizoaffective disorder appears to be mediated by substance abuse comor-
experience psychotic relapse within 5 years [27]. bidity [37]. In a recent study using National Hospi-
Our patients had schizophrenia with a duration of tal Discharge Registry of Spain for 2004, with 16,776
disease longer than 10 years in almost 60% of cases records with schizophrenia eligible for analysis,
and a median of two hospital admissions in the previ- addiction to drugs, alcohol or tobacco was the most
ous 3 years. In the present study, the number of hos- significant problem, and about one-third of cases had
pitalizations in the previous 3 years was a powerful a code indicating substance abuse or dependency
predictor of relapse, not only for the whole study [38]. In our patients, cannabis use was recorded in
population but also for the subanalyses restricted 34% of cases, cocaine in 16% and heroin in 3%. In
to the different classes of antipsychotic drugs. This accordance with the deleterious effects of substance
finding is consistent with data reported in other stud- use in patients with schizophrenia, we found that the
ies. Previous hospitalization is known to be a valid absence of consumption of cannabis or cocaine were
predictor of outcome [28] and the readmission risk protective factors for relapse. Use of cannabis was
 8 L. San et al.
also selected as an independent risk factor in the • A better understanding of risk factors for relapse
multivariate analysis. are necessary for the development of more effec-
Although there were statistically significant differ- tive prevention strategies for patients with
ences in adherence medication according to antipsy- schizophrenia in daily practice.
chotic medication, with the highest percentage of
patients with better adherence among patients treated
with long-acting injectable atypical antipsychotics, Acknowledgements
neither adherence to treatment in the previous 3 years
The study was sponsored by Janssen Cilag Spain.
nor the type of antipsychotic regimen in the previous
The authors thank Content’Ed Net Communica-
3 years was significantly associated with relapse.
tions, S.L., for editorial assistance and Marta Pulido,
However, the number of different antipsychotic drugs
MD, for editing the manuscript.
used in the previous 3 years was consistently associ-
ated with a higher likelihood of relapse. Overall, the
currently available data suggest that new-generation
Statement of Interest
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

antipsychotics have the potential to improve adher-

ence and to reduce relapse rates [39,40]. Although M. Peña is an employee of Janssen Cilag Spain. M.
second-generation antipsychotics are more expensive Bernardo has acted as a consultant to Lilly, Adamed,
than first-generation antipsychotics, they are per- Janssen-Cilag, Rovi, Bristol-Myers Squibb, Otsuka,
ceived to be more effective, with fewer adverse effects, Roche and Pfizer. The remaining authors have none
and preferable to patients [41]. In fact, when patients interest to declare.
were discharged from the short-stay or acute-care
psychiatric units, second-generation antipsychotics
especially long acting injectable formulations were References
prescribed in the majority of patients. [1] Breiser M, Iacono WG. An update on the epidemiology of
We conclude that number of hospitalizations and schizophrenia. Can J Psychiatry 1990;35:657–68.
For personal use only.

number of different antipsychotic drugs used in the [2] Bromet EJ, Fenning S. Epidemiology and natural history of
schizophrenia. Biol Psychiatry 1999;46:871–81.
previous 3 years were significant predictors of relapse.
[3] Davies LM, Drummond MF. Economics and schizophrenia:
Other risk factors included a low level of family sup- the real cost. Br J Psychiatry 1994;25(Suppl):18–21.
port and the duration of the disease over 5 years. In [4] Vázquez-Polo FJ, Negrín M, Cabasés JM, Sánchez E, Haro
contrast, the absence of consumption of cocaine and JM, Salvador-Carulla L. An analysis of the costs of treating
cannabis, in particular cannabis, was a protective fac- schizophrenia in Spain: a hierarchical Bayesian approach.
Ment Health Policy Econ 2005;8:153–65.
tor for relapse. A better understanding of specific risk
[5] Heider D, Bernert S, König HH, Matschinger H, Hogh T,
and protective factors for relapse will contribute to Brugha TS, et al. Direct medical mental health care costs of
the development of more effective prevention strate- schizophrenia in France, Germany and the United Kingdom –
gies for patients with schizophrenia in daily practice. findings from the European Schizophrenia Cohort (EuroSC).
Eur J Psychiatry 2009;24:216–24.
[6] Samalin L, Blanc O, Llorca PM. Optimizing treatment of
schizophrenia to minimize relapse. Expert Rev Neurother
Key points 2010;10:147–50.
[7] Morken G, Widen JH, Grawe RW. Non-adherence to antip-
• Prevention of relapse is a main objective in sychotic medication, relapse and rehospitalization in recent-
the treatment of patients with schizophrenia onset schizophrenia. BMC Psychiatry 2008;8:32.
both from the clinical and health care costs [8] Fitzgerald P, de Castella A, Arya D, Simons WR, Eggleston A,
perspectives. Meere S, et al. The cost of relapse in schizophrenia and schizoaf-
fective disorder. Australas Psychiatry 2009;17:265–72.
• In a multicenter and non-interventional study of [9] Hong J, Windmeijer F, Novick D, Haro JM, Brown J. The
a clinical series of 1646 patients with schizo- cost of relapse in patients with schizophrenia in the European
phrenia attended in daily practice, low family SOHO (Schizophrenia Outpatient Health Outcomes) study.
support, duration of diseases over 5 years, greater Prog Neuropsychopharmacol Biol Psychiatry 2009;33:
number of previous hospitalizations, cocaine 835–41.
[10] Lindenmayer JP, Liu-Seifert H, Kulkarni PM, Kinon BJ,
and cannabis consumption and greater number Stauffer V, Edwards SE. Medication nonadherence and treat-
of different antipsychotic drugs before the index ment outcome in patients with schizophrenia or schizo-
hospitalization were risk factor for relapse. affective disorder with suboptimal prior response. J Clin
• Number of previous hospitalizations, cannabis Psychiatry 2009;70:990–6.
consumption and number of different antipsy- [11] Llorca PM. Partial compliance in schizophrenia and the impact
on patient outcomes. Psychiatry Res 2008;161:235–47.
chotics previously used were significant inde- [12] Weiden PJ, Kozma C, Grogg A, Locklear J. Partial compli-
pendent factors for relapse in the multivariate ance and risk of rehospitalization among California Medicaid
analysis. patients with schizophrenia. Psychiatr Ser 2004;55:886–91.
 Risk factors for relapse in schizophrenia 9
[13] Doering S, Müller E, Köpcke W, Pietzcker A, Gaebel W, outcome in an early psychosis program. Schizophr Res
Linden M, et al. Predictors of relapse and rehospitalization 2001;47:215–22.
in schizophrenia and schizoaffective disorder. Schizophr Bull [27] Robinson D, Woerner MG, Alvir JM, Bilder R, Goldman R,
1998;24:87–98. Geisler S, et al. Predictors of relapse following response from
[14] Owens DC, Johnstone EC, Miller P, Macmillan JF, Crow TJ. a first episode of schizophrenia or schizoaffective disorder.
Duration of untreated illness and outcome in schizophrenia: Arch Gen Psychiatry 1999;56:241–7.
test of predictions in relation to relapse risk. Br J Psychiatry [28] Doering S, Müller E, Köpcke W, Pietzcker A, Gaebel W,
2010;196:296–301. Linden M, et al. Predictors of relapse and rehospitalization
[15] Liu-Seifert H, Osuntokun OO, Feldman PD. Factors in schizophrenia and schizoaffective disorder. Schizophr Bull
associated with adherence to treatment with olanzapine and 1998;24:87–98.
other atypical antipsychotic medications in patients with [29] Munk-Jørgensen P, Mortensen PB, Machón RA. Hospitali-
schizophrenia. Compr Psychiatry. 2012;53:107–15. doi: zation patterns in schizophrenia. A 13-year follow-up. Schiz-
10.1016/j.comppsych.2010.12.003. Epub 2011 Feb 9. ophr Res 1991;4:1–9.
[16] Olivares JM, Rodriguez-Morales A, Diels J, Povey M, [30] Mortensen PB, Eaton WW. Predictors for readmission risk
Jacobs A, Zhao Z, et al. Long-term outcomes in patients in schizophrenia. Psychol Med 1994;24:223–32.
with schizophrenia treated with risperidone long-acting [31] Gaddini A, Franco F, Di Lallo D, Biscaglia L. Hospitalization
injection or oral antipsychotics in Spain: results from the for schizophrenia in acute psychiatric wards of the Lazio
Int J Psych Clin Pract Downloaded from informahealthcare.com by Selcuk Universitesi on 02/10/15

electronic Schizophrenia Treatment Adherence Registry region: a 4-year follow-up study. Epidemiol Psychiatr Soc
(e-STAR). Eur Psychiatry 2009;24:287–96. 2005;14:227–34.
[17] American Psychiatric Association.: Diagnostic and Statistical [32] Gupta S, Hendricks S, Kenkel AM, Bhatia SC, Haffke EA.
Manual of Mental Disorders: DSM-IV. 4th ed. Washington, Relapse in schizophrenia: is there a relationship to substance
DC: American Psychiatric Association, 1994. abuse? Schizophr Res 1996;20:153–6.
[18] Guy W. Clinical Global Impressions ECDEU Assessment [33] Swofford CD, Kasckow JW, Scheller-Gilkey G, Inderbitzin
Manual for Psychopharmacology, revised (DHEW Publ. No. LB. Substance use: a powerful predictor of relapse in schizo-
ADM 76-338). Rockville, MD: National Institute of Mental phrenia. Schizophr Res 1996;20:145–51.
Health. p. 218–22. [34] Swofford CD, Scheller-Gilkey G, Miller AH, Woolwine B,
[19] Pharoah F, Mari J, Rathbone J, Wong W. Family intervention Mance R. Double jeopardy: schizophrenia and substance
for schizophrenia. Cochrane Database Syst Rev 2010 Dec use. Am J Drug Alcohol Abuse 2000;26:343–53.
8;12:CD000088. [35] Soyka M, Albus M, Kathmann N, Finelli A, Hofstetter S,
[20] Montero I, Gómez-Beneyto M, Ruiz I, Puche E, Adam A. Holzbach R, et al. Prevalence of alcohol and drug abuse in
For personal use only.

The influence of family expressed emotion on the course of schizophrenic inpatients. Eur Arch Psychiatry Clin Neurosci
schizophrenia in a sample of Spanish patients. A two-year 1993;242:362–72.
follow-up study. Br J Psychiatry 1992;161:217–22. [36] Dervaux A, Laqueille X, Bourdel MC, Leborgne MH, Olié
[21] Vaughan K, Doyle M, McConaghy N, Blaszczynski A, Fox JP, Lôo H, et al. Cannabis and schizophrenia: demographic
A, Tarrier N. The relationship between relative’s Expressed and clinical correlates. Encephale 2003;29:11–17.
Emotion and schizophrenic relapse: an Australian replica- [37] Fazel S, Gulati G, Linsell L, Geddes JR, Grann M.
tion. Soc Psychiatry Psychiatr Epidemiol 1992;27:10–5. Schizophrenia and violence: systematic review and meta-
[22] Ramírez García JI, Chang CL, Young JS, López SR, analysis. PLoS Med 2009;6:e1000120, doi: 10.1371/journal.
Jenkins JH. Family support predicts psychiatric medication pmed.1000120.
usage among Mexican American individuals with schizophre- [38] Bouza C, López-CuadradoT, Amate JM. Physical disease in
nia. Soc Psychiatry Psychiatr Epidemiol 2006;41:624–31. schizophrenia: a population-based analysis in Spain. BMC
[23] Sellwood W, Tarrier N, Quinn J, Barrowclough C. The family Public Health 2010;10:745, doi: 10.1186/1471 - 2458-10 - 745.
and compliance in schizophrenia: the influence of clinical [39] Leucht S, Barnes TR, Kissling W, Engel RR, Correll C,
variables, relatives’ knowledge and expressed emotion. Psy- Kane JM. Relapse prevention in schizophrenia with new-
chol Med 2003;33:91–6. generation antipsychotics: a systematic review and explora-
[24] Farooq S, Large M, Nielssen O, Waheed W. The relationship tory meta-analysis of randomized, controlled trials. Am J
between the duration of untreated psychosis and outcome in Psychiatry 2003;160:1209–22.
low-and-middle income countries: a systematic review and [40] Peuskens J, Olivares JM, Pecenak J, Tuma I, Bij de Weg H,
meta analysis. Schizophr Res 2009;109:15–23. Eriksson L, et al. Treatment retention with risperidone long-
[25] Barnes TR, Leeson VC, Mutsatsa SH, Watt HC, Hutton SB, acting injection: 24-month results from the Electronic Schiz-
Joyce EM. Duration of untreated psychosis and social func- ophrenia Treatment Adherence Registry (e-STAR) in six
tion: 1-year follow-up study of first-episode schizophrenia. countries. Curr Med Res Opin 2010;26:501–9.
Br J Psychiatry 2008;193:203–9. [41] Naber D, Lambert M. The CATIE and CUtLASS studies in
[26] Black K, Peters L, Rui Q, Milliken H, Whitehorn D, Kopala schizophrenia: results and implications for clinicians. CNS
LC. Duration of untreated psychosis predicts treatment Drugs 2009;23:649–59.