Download Incident users of antipsychotics: who are they and how do

Soc Psychiatry Psychiatr Epidemiol
DOI 10.1007/s00127-015-1129-8
ORIGINAL PAPER
Incident users of antipsychotics: who are they and how do they
fare?
Lone Baandrup1 • Marie Kruse2
Received: 14 July 2015 / Accepted: 22 September 2015
Springer-Verlag Berlin Heidelberg 2015
Abstract
Purpose In Denmark, as well as in many other countries,
consumption of antipsychotics is on the rise, partly due to
increasing off-label use. The aim of this study was to
analyze and quantify the extent of off-label use and
polypharmacy in incident users of antipsychotic medication, and to examine initial antipsychotic prescribing patterns and associated use of mental health care services.
Method Population-based cohort study linking the following Danish national registers: the Central Psychiatric
Research Register, the Register of Medicinal Product
Statistics, and Statistics Denmark.
Results From 2007 to 2012, 154,351 Danish subjects
initiated treatment with antipsychotics. Among these,
71,254 (46 %) subjects had a psychiatric diagnosis recorded. The most frequent diagnoses were reaction to severe
stress and adjustment disorders (10,106; 14 %), depressive
episode (8876; 12 %), and recurrent depressive disorder
(6810; 10 %). We found high antipsychotic discontinuation
rates during the first few months (45 % in 4 months) and
frequent antidepressant co-prescribing from treatment
onset (47 %). Significantly greater likelihood of psychiatric
hospitalization was observed for antipsychotic polypharmacy (HR 1.38; 95 % CI 1.32–1.45), whereas
& Lone Baandrup
[email protected]
1
Center for Neuropsychiatric Schizophrenia Research and
Center for Clinical Intervention and Neuropsychiatric
Schizophrenia Research, Mental Health Center Glostrup,
Copenhagen University Hospital, Nordre Ringvej 29-67,
2600 Glostrup, Denmark
2
Center for Health Economics Research, University of
Southern Denmark, 5230 Odense, Denmark
antipsychotic discontinuation was associated with
decreased hospitalization risk in most off-label conditions.
Conclusions The brief duration of most antipsychotic
prescriptions suggests that antipsychotics are prescribed
more liberally than recommended. As a consequence of the
range of adverse effects associated with antipsychotic drug
use, the documented widespread off-label prescribing
practices warrant careful monitoring for adverse effects
and prompt discontinuation in case of an unfavorable risk–
benefit ratio.
Keywords Antipsychotics Prescribing practice Off-label Register-based
Introduction
An increasing share of antipsychotic medication prescriptions fall outside the main evidence-based indications
comprising schizophrenia spectrum disorders and bipolar
disorder, and can thus be characterized as off-label. A
number of other authority-approved indications for treatment with specific antipsychotic compounds exists (e.g.
Tourette’s syndrome, adjunctive therapy for treatment-resistant depression, irritability associated with autism in
children), with some inconsistency regarding which
specific compounds are approved for which of these minor
indications by, respectively, the US Food and Drug
Administration and the European Medicines Agency.
However, for the remainder of this paper, we will use offlabel synonymously with indications other than
schizophrenia spectrum and bipolar disorder. A US study
from The Veterans Affairs found that 60.2 % of antipsychotic prescriptions were off-label in 2007, with the most
common diagnoses among off-label treated subjects being
123
Soc Psychiatry Psychiatr Epidemiol
posttraumatic stress disorder (41.8 %), minor depression
(39.5 %), major depression (23.4 %), and anxiety disorder
(20.0 %) [1]. In a recent cohort study from UK primary
care it was reported that for most antipsychotic compounds
it was less than half of the people prescribed these medications that had a diagnosis of psychosis or bipolar disorder
[2]. Internationally, the rate of antipsychotic medication
use has markedly increased during the last couple of decades [3], mostly due to newer on-label use (bipolar disorder) and increasing frequencies of off-label use [4, 5].
Currently, schizophrenia spectrum disorders and bipolar
disorder are the main indications for treatment with
antipsychotic medications, based on high-quality evidence
documenting substantial efficacy both in the acute phase
and in relapse prevention [6–8]. In schizophrenia, the high
rate of polypharmacy with antipsychotic compounds has
received considerable attention because of the lack of
evidence of a superior therapeutic effect, as well as evidence of an increased burden of side effects [9, 10].
Polypharmacy in off-label antipsychotic medication use,
however, has not been systematically investigated. A US
national survey among office-based psychiatrists reported a
significant increase in polypharmacy involving antidepressant and antipsychotic medications [11]. This study
went on to conclude that most were of unproven efficacy
and therefore offered only uncertain benefits for quality of
care and clinical outcomes [11]. Thus, there is a need for
studies that describe and analyze off-label antipsychotic
prescribing patterns, including polypharmacy.
The aim of this study was to analyze and quantify the
extent of off-label use and polypharmacy in incident users
of antipsychotic medication. Furthermore, we analyzed the
use of mental health care services in this population,
hypothesizing that the association of prescribing patterns
with use of mental health care services differs across
diagnostic categories.
Methods
Design and materials
Population-based cohort study of antipsychotic prescribing
patterns in adult incident users of antipsychotic medication.
We used population-based national registers that include
longitudinal data on the entire vital population with the
exception of emigrants. We linked data from the Central
Psychiatric Research Register (CPRR), the Register of
Medicinal Product Statistics, and Statistics Denmark at the
individual subject level, using the unique civil registration
number for linkage. The civil registration number is
assigned to all Danish residents at birth or upon
immigration.
123
Data extracted from the national registers
Mental health care in Denmark is divided between psychiatric hospitals (including outpatient clinics), practicing
specialists, and general practitioners. The Danish mental
health care services are public, tax-financed, and free of
charge at the point of receipt, except for pharmaceuticals
and a few other services. The degree of co-payment for
pharmaceuticals varies along the quantity consumed. The
CPRR contains information on all individual contacts
within the psychiatric hospital system in Denmark [12].
From the CPRR, we extracted the International Classification of Diseases 10th Revision (ICD-10) diagnostic
codes (as explained below) together with information on
the number of psychiatric hospitalizations, the total duration of hospitalizations, and the number of visits to outpatient clinics and mental health care emergency rooms
(within the study period from 2007 to 2012). Consequently,
the analyses were restricted to patients with one or more
hospital contacts in the study period. Most patients with
moderate or severe symptoms of a mental disorder are
treated or at least evaluated in the psychiatric hospital
system.
The Register of Medicinal Product Statistics contains
information on all prescription medicines collected outside
hospitals. For each transaction, the following information
is recorded: the civil registration number of the subject
(allowing for linkage with other registers), the Anatomic
Therapeutic Classification (ATC) code [13], the dosage per
sold package, and the date of the purchase. Consequently,
the medication exposure in this study comprised the collected (i.e., purchased) medication at all pharmacies in
Denmark, which is an exhaustive measure of out-of-hospital use of prescription medication. Medication received
during inpatient admissions was not available in the registers and therefore not considered.
The registers within Statistics Denmark comprise data
on a range of variables including socioeconomic data. For
this study, we extracted the length of education and the
marital status in the year of the index prescription to obtain
the best possible comparability between subjects.
Study population, diagnostic classification
and exposure definition
The study population comprised all Danish adult
(C18 years of age) residents who collected a minimum of
one antipsychotic prescription during the years 2007–2012.
Subjects who had collected at least one prescription for an
antipsychotic compound up to 3 years before the observation period were excluded in an attempt to use incidencebased measures. For each included subject, the diagnosis
was defined as the psychiatric diagnosis (coupled with a
Soc Psychiatry Psychiatr Epidemiol
psychiatric inpatient/outpatient/emergency room contact)
given in closest proximity to the date of the index
antipsychotic prescription. Thus, the reported diagnosis
could be on both temporal sides of the index antipsychotic
prescription, whichever was closest in time. We chose this
method of reporting the diagnosis to obtain the most precise picture of the population at the particular time of the
first antipsychotic prescription.
Individuals without any contact to the psychiatric hospital system (and hence without a diagnosis in the CPRR)
were not further analyzed. We traced antipsychotic prescriptions (as defined by relevant ATC codes) across the
observation period divided into intervals of 60 days. We
chose an interval duration of 60 days so that most patients
would fill at least one prescription during an interval, and at
the same time the interval was sufficiently short to capture
changes in prescribing patterns. Therefore, the interval was
approximately twice the average prescription length. The
average prescription length was based on an a priori
assumption of antipsychotic prescription duration and
verified in a subsample of the population. In this subsample, there was available information on defined daily doses
(DDD). The subsample was representative of the entire
population regarding distribution of ATC code, diagnosis,
age and gender. For this subsample, a basic calculation of
the mean DDD per prescription yielded approximately 30,
indicating that the average prescription duration was close
to 1 month.
Within each 60 days’ interval, we defined antipsychotic
medication treatment as continued monotherapy (collecting
a prescription of the same antipsychotic, i.e., same ATC
code, as in the preceding interval), switched (collecting a
prescription of another antipsychotic with a different ATC
code than in the preceding interval), discontinued (not
collecting any prescriptions of antipsychotics within two
successive 60 days’ intervals), or combined with another
antipsychotic, antidepressant or benzodiazepine. Antipsychotic polypharmacy (APP) was defined as collection of
two (or more) different antipsychotic compounds within
two successive intervals of 60 days. Add-on treatment was
defined as collection of two (or more) prescriptions of
antidepressants (AP/AD), benzodiazepines (AP/BZ), or
both (AP/AD/BZ) within an interval of 60 days. Each
subject could be in none, one, or two of these combined
medication groups.
Study medication
Pharmaceutical compounds with the following ATC codes
were included in the study:
•
Antipsychotics N05A (except the following: acepromazine N05AA04, prochlorperazine N05AB04,
•
•
periciazine N05AC01, tetrabenazine N05AK01, and
lithium N05AN01). All formulations were included,
i.e., both oral formulations and long-acting injectables.
Tricyclic antidepressants N06AA, selective serotonin
reuptake inhibitors N06AB, and other antidepressants
N06AX.
Benzodiazepines as anxiolytics N05BA, benzodiazepines as hypnotics N05CD, benzodiazepine-related
drugs N05CF, and clonazepam N03AE01 (traditionally
used as an anxiolytic agent in Denmark despite not
being approved for this indication).
Conventional mood stabilizers (lithium and anticonvulsants) were not considered in this study because ambulatory prescriptions of these medications were anticipated to
be of less relevance in a sample of incident antipsychotic
users. Furthermore, off-label prescriptions of these medications were expected to be much less frequent than for the
psychotropic drugs included in the study.
Statistical analyses
The descriptive part of the study investigated the initial
antipsychotic medication use by defining and examining
the following treatment patterns: continuing monotherapy,
discontinuing, switching or combining.
We conducted three different analyses of the association
between antipsychotic medication use and utilization of
mental health care services.
Firstly, we tested the hypothesis that the number of
mental health care services utilized was associated with
antipsychotic combination regimen (AP/AD, AP/BZ, AP/
AD/BZ, APP) using a linear regression model. The
dependent variable was the total number of contacts within
the psychiatric hospital system, and the antipsychotic
combination regimen was the independent variable. Age,
gender, diagnosis, marital status, length of education, and
type of mental health care contacts as proxy of illness
severity were entered into the model as covariates. In
addition, we added several interaction terms to the model,
to capture the interaction between diagnosis and antipsychotic combination regimen.
As described above, we identified the main psychiatric
diagnosis recorded in relation to a hospital or ambulatory
contact closest in time to the index antipsychotic prescription. A smaller subgroup of the subjects (N = 5596)
were diagnosed in the CPRR with a schizophrenia spectrum (F2) or bipolar disorder (F31) diagnosis farther away
than the closest contact. In a sensitivity analysis, these
supplementary on-label diagnoses were included in the
analysis, in order to investigate how the study definition of
the diagnosis might influence the distribution of diagnoses
and the results of the linear regression.
123
Soc Psychiatry Psychiatr Epidemiol
Secondly, we tested the hypothesis that time to the first
hospitalization (after the index antipsychotic prescription)
was associated with antipsychotic combination regimen,
taking into account the covariates and interactions (as
described above). This hypothesis was tested using duration analysis (Cox proportional hazards regression) with
time to the first hospitalization as the dependent variable
and antipsychotic combination regimen (AP/AD, AP/BZ,
AP/AD/BZ, APP) as the independent variable. Proportional
hazards were tested using the proportionality test option in
the actual regression (proc phreg, SAS).
Thirdly, we investigated whether discontinuation of
antipsychotic treatment was associated with the risk of
subsequent hospitalization. This association was analyzed
using a logistic regression model with hospitalization as the
dependent variable, and antipsychotic discontinuation as
the independent variable, adjusting for gender, age, marital
status, and length of education. We acknowledge that both
hospitalization and discontinuation are associated with
diagnosis, and therefore this last analysis was stratified by
diagnosis.
In the regression models, we used data for the entire
observation period from 2007 to 2012, while the descriptive analysis only presents data for the year of the index
antipsychotic prescription.
Subjects were censored if they were hospitalized for one
or more 60 days’ intervals, because they were not at risk of
antipsychotic exposure (according to the study definition)
during such intervals in which medication was received
from the hospital and, therefore, not recorded in the database. All analyses were conducted using SAS, version 9.3.
Findings
We initially identified a cohort of 154,351 adult subjects
who collected their first antipsychotic prescription within
the observation period (2007–2012). More than half (54 %)
of these subjects did not receive any treatment in the
psychiatric hospital system within the observation period
and therefore did not have a recorded diagnosis in the
CPRR. Thus, the analytical results were based on a sample
of 71,254 subjects.
The distribution of the most frequent diagnoses and
corresponding gender and age in relation to the index
antipsychotic prescription is presented in Table 1. The
ICD-10 diagnostic categories are listed in Table 1 with one
or two digits, depending on which categorization gave the
best picture of the most frequent diagnoses. Rare diagnoses
(having a frequency \1 %) were compiled in a rest group,
summing up to 1.6 % of all prescriptions. For evaluation of
our definition of diagnosis, we identified the number of
subjects with an approved diagnosis for antipsychotic
123
treatment [schizophrenia spectrum disorder (F2) or bipolar
disorder (F31)] given at any point in time, irrespective of
relation to the first antipsychotic prescription. This group of
patients comprised a smaller subgroup of 5596 subjects.
When including these subjects in the respective diagnostic
groups, the frequency of schizophrenia rose to 9.8 %
(compared with 4.3 %) and of bipolar disorder to 5.8 %
(compared with 3.5 %). No other substantial changes were
observed for the diagnostic categories listed in Table 1.
The on-label diagnoses were thus still surpassed by the
more frequent off-label diagnoses of adjustment disorders
and unipolar depressive disorders.
The most frequently prescribed antipsychotic compounds were quetiapine (30 %), chlorprothixene (26 %),
risperidone (14 %), olanzapine (11 %), flupentixol (4 %),
haloperidol (4 %), levopromazine (7 %), and aripiprazole
(3 %). Overall, prescriptions of quetiapine, olanzapine,
risperidone, chlorprothixene and levopromazine accounted
for 71 % of the total number of collected prescriptions.
The descriptive analysis of initial antipsychotic prescribing patterns showed that 45 % of incident antipsychotic medication users discontinued treatment within the
first 4 months of treatment. After the first year, 59 % of the
subjects had discontinued antipsychotic treatment all together. During the first year, there was a constant rate of 10 %
of patients switching to another antipsychotic compound in
each defined interval of 60 days. The transitory nature of
antipsychotic treatment was most pronounced in subjects
with disorders related to severe stress and adjustment; in
this diagnostic group, 55 % had discontinued their treatment within the first 4 months, and 67 % within the first
year (data not shown). When analyzing antipsychotic
combination regimen, we found that antipsychotic medication was prescribed together with an antidepressant in
47 % of the cases. The proportion of antipsychotic–antidepressant co-prescribing rapidly declined, and after the
first year, 21 % of incident antipsychotic users were still
being treated with the same antipsychotic–antidepressant
combination as initially. Likewise, benzodiazepines were
co-prescribed with the antipsychotic initially in 20 % of the
subjects, but this regimen declined during the first year to
9 %. Antipsychotic polypharmacy was observed in\5 % of
incident antipsychotic users, and this share was stable during the first year. In subgroup analyses split on the most
frequent diagnoses, there was some variation in the pattern.
For patients diagnosed with schizophrenia, the proportion
of subjects receiving antipsychotic polypharmacy was
around 14 % initially, and then diminished slightly to 11 %
after one year. For patients diagnosed with depression or
recurrent depressive disorder, the proportion of patients
receiving treatment with antidepressants in addition to
antipsychotics was 70 %, and for additional benzodiazepine
treatment there was a frequency of 20–25 %.
Soc Psychiatry Psychiatr Epidemiol
Table 1 The distribution of ICD-10 diagnoses within the study population of new antipsychotic users with a least one contact to the psychiatric
hospital system
ICD-10 code
F43: Reaction to severe stress and adjustment disorders
Frequency
Males
N
N
%
Age
%
Mean
SD
14.8
10,106
14.2
4728
46.8
40.3
F32: Depressive episode
8786
12.3
3653
41.6
45.2
18.8
F0: Organic mental disorders (except F00)
F33: Recurrent depressive disorder
7295
6810
10.2 3377
9.6 2273
46.3
33.4
77.0
49.1
14.8
18.1
F1: Mental and behavioral disorders due to psychoactive substance abuse
5962
8.4 4223
70.8
40.8
15.5
F00: Dementia in Alzheimer’s disease
4433
6.2 1785
40.3
81.5
7.8
F2: Non-affective psychoses other than schizophrenia
4409
6.2 2095
47.5
44.6
21.0
F4: Neurotic, stress-related and somatoform disorders (except F43)
4188
5.9 1603
38.3
39.0
16.0
F6: Disorders of adult personality and behavior
4174
5.9 1271
30.5
32.1
12.0
F9: Behavioral and emotional disorders with onset usually occurring in childhood and adolescence
3775
5.3 2267
60.1
33.8
16.6
Z codes: Factors influencing health status and contact with health services
3082
4.3 1769
57.4
42.2
19.6
F20: Schizophrenia
3028
4.3 1832
60.5
34.6
14.4
F31: Bipolar disorder
2491
3.5 1045
42.0
44.1
16.2
F3: Mood disorders (except F31, F32 and F33)
792
1.1
342
43.2
47.0
18.3
F7: Mental retardation
763
1.1
425
55.7
38.1
15.9
1160
1.6
499
43.0
29.6
13.6
–
–
–
Summation of diagnoses with frequency \1 %
Total
71,254
100
–
The diagnosis is the one given in closest proximity to the index antipsychotic prescription
Table 2 Linear regression of the combined use of mental health care services (hospitalization, outpatient clinic, and emergency room)
Variable
Adjusted analysisa
Raw analysis
Parameter estimateb
Intercept
95 % confidence limits
Parameter estimateb
95 % confidence limits
4.03
3.93
4.13
1.92
1.64
2.21
AP/AD
-0.19
-0.32
-0.06
-0.26
-0.62
0.10
AP/BZ
0.87
0.67
1.07
2.02
1.59
2.45
AP/AD/BZ
0.54
0.31
0.78
0.39
-0.25
1.02
4.33
4.02
4.63
3.04
2.54
3.55
APP
2
Adjusted R = 0.25
AP/AD antipsychotic–antidepressant combination, AP/BZ antipsychotic–benzodiazepine combination, AP/AD/BZ antipsychotic–antidepressantbenzodiazepine combination, APP antipsychotic polypharmacy
a
The analysis was adjusted for the following covariates: diagnosis, gender, age, marital status, education, type of hospital contacts as proxy of
illness severity, and interaction terms
b
Parameter estimate = 0 corresponds to no effect and thus a 95 % CI not including 0 indicates statistical significance
The adjusted linear regression (Table 2) showed that
utilization of mental health care services was associated
with antipsychotic polypharmacy (parameter estimate 3.04;
95 % CI 2.54–3.55) and antipsychotic–benzodiazepine cotreatment (parameter estimate 2.02; 1.59–2.45). In this
model, we included interaction terms that indicated significant interaction between diagnosis and antipsychotic
medication regimen, i.e., the association of antipsychotic
medication regimen with use of mental health care services
were of different magnitude across investigated diagnoses.
In a sensitivity analysis, we included diagnoses approved
for antipsychotic treatment, i.e., either schizophrenia
spectrum disorders (F2) or bipolar disorder (F31), given at
any point in time and irrespective of relation to the first
antipsychotic prescription. Generally, the results were
stable and the conclusions remained unchanged.
The results of the Cox regression indicated that
antipsychotic combination regimen was associated with
time to the first hospitalization after the index antipsychotic
prescription (Fig. 1). Significantly greater likelihood of
123
Soc Psychiatry Psychiatr Epidemiol
Fig. 1 Title Time to first hospitalization according to antipsychotic
drug regimen (Cox regression analysis). 1 AP/AD antipsychotic–
antidepressant combination; 2 AP/BZ antipsychotic–benzodiazepine
combination; 3 AP/AD/BZ antipsychotic–antidepressant–benzodiazepine combination; 4 APP antipsychotic polypharmacy
psychiatric hospitalization was observed for antipsychotic
polypharmacy (HR 1.78; 1.69–1.86) relative to antipsychotic monotherapy and other combination regimens,
which remained clinically and statistically significant when
adjusting for age, gender, diagnosis, and dominant type of
mental health care service utilized as a proxy of illness
severity (HR 1.38; 1.32–1.45).
The logistic regression of inpatient psychiatric admission as a function of antipsychotic discontinuation and
adjusted for gender, age, marital status and educational
level was conducted for each diagnostic group separately.
Table 3 lists the odds ratios for psychiatric hospitalization
for antipsychotic discontinuers compared with antipsychotic continuers for each diagnostic group. Discontinuation of antipsychotic medication was generally associated
with lower risk of hospitalization across all diagnostic
categories, although this was not statistically significant in
the case of schizophrenia spectrum psychoses (F20-F29) as
well as in dementia and other organic mental disorders
(F00-F09).
Discussion
The main aim of this study was to provide a comprehensive
description of off-label use and polypharmacy in incident
users of antipsychotic medication. We found that incident
antipsychotic medication users represent a highly heterogeneous group in terms of diagnosis and prescribing patterns. The index antipsychotic prescription was most often
issued in relation to adjustment reactions or unipolar mood
disorders. Almost half of all incident antipsychotic
123
medication users ceased their medication use within the
first few months, indicating that most prescriptions reflected only brief antipsychotic medication use or pro necessitate prescriptions that were never renewed. Although the
specific indication for each antipsychotic prescription was
not available in this study, the diagnostic profile indicated
that most initial antipsychotic prescriptions were for
unspecific sedative purposes. This unspecific sedative aim
of antipsychotic use was confirmed by the frequency
analysis showing that sedating medications, like quetiapine, were most frequently prescribed. Chlorprothixene was
also among the most prescribed medications. This lowpotency first generation antipsychotic compound is traditionally used in lower doses in Denmark as an unspecific
sedating agent, especially in circumstances where the
prescriber wishes to avoid the use of benzodiazepines due
to a comorbid substance abuse or anticipated risk of abuse.
However, prescribing of chlorprothixene should be
restricted due to important side effects, in particular anticholinergic side effects.
Prescribing patterns and discontinuation rates have not
previously been investigated with respect to off-label
antipsychotic prescribing, but the observed high initial
discontinuation rates are in line with previous findings in
samples of schizophrenia patients, e.g., the CATIE study in
which 74 % of patients discontinued the study medication
before 18 months [14].
Recently, some second generation antipsychotic compounds were approved for augmentation in treatment-resistant unipolar depression, but the excessive use of
antipsychotic–antidepressant drug combinations in the
present study seems to go beyond such indication. Frequent
prescribing of not clearly indicated antipsychotic–antidepressant combination treatment (36.5 %) was also reported
in a recently published US study of prescription practices
in first episode schizophrenia [15].
We found that antipsychotic polypharmacy was associated with increased use of mental health care services and
increased likelihood of hospitalization when controlling for
possible confounders. We included the type of mental
health care services as a proxy of illness severity, but it is
likely that some residual confounding remained. Antipsychotic polypharmacy is an indicator of severe illness and is
therefore associated with increased risk of hospitalization
and use of other mental health care services, irrespective of
medication status.
In the current study, discontinuation of antipsychotic
treatment was associated with lower risk of hospitalization
for most off-label diagnostic categories. No causal inferences can be made from these observational data because
the subjects discontinuing antipsychotic medication might
have had fewer symptoms and an overall better prognosis,
irrespective of medication use, than the subjects continuing
Soc Psychiatry Psychiatr Epidemiol
Table 3 Logistic regression of hospitalization as function of discontinuation versus continuation of antipsychotic medication, analyzed separately for each diagnostic group
Discontinuation of AP treatment
Adjusted analysisa
Raw analysis
ORb for
hospitalization
95 % Wald
confidence limits
ORb for
hospitalization
95 % Wald
confidence limits
F00–F09: organic mental disorders
1.06
0.97
1.17
0.91
0.82
1.00
F20–F29: schizophrenia spectrum psychoses
0.98
0.85
1.12
0.95
0.83
1.09
F30–F39: mood disorders
0.81
0.74
0.88
0.80
0.74
0.87
F40–F49: neurotic, stress-related and somatoform disorders
0.69
0.62
0.76
0.68
0.61
0.76
No F-diagnosis
0.69
0.56
0.84
0.67
0.55
0.83
F10–F19, F50–F99: substance abuse and other F-diagnoses
0.67
0.59
0.76
0.66
0.58
0.75
AP antipsychotic, F00–F09 organic mental disorders, F20–F29 schizophrenia spectrum psychoses, F30–F39 mood disorders, F40–F49 neurotic
stress-related and somatoform disorders, F10–F19 mental and behavioral disorders due to psychoactive substance use
a
Adjusted for age, gender, marital status, and education
b
OR = 1 corresponds to no effect and thus a 95 % CI not including 1 indicates statistical significance
antipsychotic medication. However, for schizophrenia
spectrum psychoses (F20–F29) as well as for dementia and
other organic mental disorders (F00–F09) there was no
statistically significant difference in hospitalization risk for
antipsychotic discontinuers compared with continuers.
Since antipsychotics have a clearly documented relapse
preventing effect in schizophrenia spectrum psychoses, a
hospitalization sparing effect of continued use of antipsychotics could have been expected. The lack of such an
effect cannot be considered equivalent to a lack of effectiveness of antipsychotics because the study is observational and therefore due to confounding by severity. Thus,
patients with less severe symptoms or shorter duration of
symptoms, and thus a better prognosis, are more likely to
discontinue antipsychotic medication than patients with
severe or long-lasting symptoms. The former group is
much less likely to be hospitalized, and thus a potential
hospitalization sparing effect of continued antipsychotic
medication use in the former group is not strong enough to
appear.
Limitations of this study include the fact that in Denmark, only patients with at least one contact to the psychiatric hospital system are registered with a mental
disorder diagnosis in the CPRR. People receiving mental
health care in the primary health care sector would therefore appear as not having a diagnosis in a register-linkage
study like the present one. Equally, patients with a certain
diagnosis in the CPRR might have had another diagnosis
with the physician treating them in the primary health care
sector. Thus, it is possible that, among the high number of
incident antipsychotic users with a diagnosis related to
stress and adjustment, some of the patients will later be
differently diagnosed in the primary health care sector.
Another limitation pertains to the lack of variables of
illness severity in the registers applied in this study.
Consequently, confounding by indication is an important
issue to bear in mind when interpreting the results. We
tried to diminish this confounder by adjusting for ‘patient
type’ as defined by previous use of mental health care
services. Another limitation concerns that fact that no data
were available on the clinical indication for each specific
antipsychotic prescription, and thus the diagnosis in closest
proximity to the index antipsychotic prescription was used
as a substitute. However, the results of the sensitivity
analysis that included patients with ever on-label diagnoses
confirmed that our choice of definition of the diagnosis
yielded robust and valid results.
The results from this study document high rates of offlabel antipsychotic drug prescribing. Off-label prescribing
lacks clear documentation of a favorable risk–benefit ratio.
Consequently, efforts should be made to limit off-label
antipsychotic prescribing and reserve off-label use for
specific otherwise treatment-resistant conditions. In particular, off-label antipsychotic polypharmacy and antipsychotic–antidepressant combinations seem to be prescribed
excessively, considering the lack of evidence base and the
risk of an increased burden of side effects and drug–drug
interactions.
Acknowledgments LB performed this work as part of a postdoc
position at Center for Neuropsychiatric Schizophrenia Research
(CNSR) and Center for Clinical Intervention and Neuropsychiatric
Schizophrenia Research (CINS), Mental Health Center Glostrup,
Mental Health Services of the Capital Region in Denmark. No
additional funding for this study was obtained.
Compliance with ethical standards
Ethics Since only anonymized data were used, no approval from an
ethics committee was required. The study was approved by the
Danish Data Protection Agency and performed in accordance with
national laws.
123
Soc Psychiatry Psychiatr Epidemiol
Disclosures On behalf of both authors, the corresponding author
states that there is no conflict of interest.
9.
References
10.
1. Leslie DL, Mohamed S, Rosenheck RA (2009) Off-label use of
antipsychotic medications in the department of Veterans Affairs
health care system. Psychiatr Serv 60:1175–1181
2. Marston L, Nazareth I, Petersen I, Walters K, Osborn DP (2014)
Prescribing of antipsychotics in UK primary care: a cohort study.
BMJ Open 4:e006135
3. Verdoux H, Tournier M, Begaud B (2010) Antipsychotic prescribing trends: a review of pharmaco-epidemiological studies.
Acta Psychiatr Scand 121:4–10
4. Alexander GC, Gallagher SA, Mascola A, Moloney RM, Stafford
RS (2011) Increasing off-label use of antipsychotic medications
in the United States, 1995–2008. Pharmacoepidemiol Drug Saf
20:177–184
5. Domino ME, Swartz MS (2008) Who are the new users of
antipsychotic medications? Psychiatr Serv 59:507–514
6. Goodwin GM (2003) Evidence-based guidelines for treating
bipolar disorder: recommendations from the British Association
for Psychopharmacology. J Psychopharmacol 17:149–173
7. Leucht S, Tardy M, Komossa K, Heres S, Kissling W, Davis JM
(2012) Maintenance treatment with antipsychotic drugs for
schizophrenia. Cochrane Database Syst Rev 5:CD008016
8. Leucht S, Cipriani A, Spineli L, Mavridis D, Orey D, Richter F,
Samara M, Barbui C, Engel RR, Geddes JR, Kissling W, Stapf
MP, Lassig B, Salanti G, Davis JM (2013) Comparative efficacy
123
11.
12.
13.
14.
15.
and tolerability of 15 antipsychotic drugs in schizophrenia: a
multiple-treatments meta-analysis. Lancet 382:951–962
Correll CU, Rummel-Kluge C, Corves C, Kane JM, Leucht S
(2009) Antipsychotic combinations vs monotherapy in
schizophrenia: a meta-analysis of randomized controlled trials.
Schizophr Bull 35:443–457
Gallego JA, Bonetti J, Zhang J, Kane JM, Correll CU (2012)
Prevalence and correlates of antipsychotic polypharmacy: a systematic review and meta-regression of global and regional trends
from the 1970s to 2009. Schizophr Res 138:18–28
Mojtabai R, Olfson M (2010) National trends in psychotropic
medication polypharmacy in office-based psychiatry. Arch Gen
Psychiatry 67:26–36
Mors O, Perto GP, Mortensen PB (2011) The danish psychiatric
central research register. Scand J Public Health 39:54–57
World Health Organization Collaborating Centre for Drug
Statistics Methodology (2014) Guidelines for ATC classification
and DDD assignment, 2015. World Health Organization, Oslo
Lieberman JA, Stroup TS, McEvoy JP, Swartz MS, Rosenheck
RA, Perkins DO, Keefe RS, Davis SM, Davis CE, Lebowitz BD,
Severe J, Hsiao JK (2005) Effectiveness of antipsychotic drugs in
patients with chronic schizophrenia. N Engl J Med
353:1209–1223
Robinson DG, Schooler NR, John M, Correll CU, Marcy P,
Addington J, Brunette MF, Estroff SE, Mueser KT, Penn D,
Robinson J, Rosenheck RA, Severe J, Goldstein A, Azrin S,
Heinssen R, Kane JM (2015) Prescription practices in the treatment of first-episode schizophrenia spectrum disorders: data
From the National RAISE-ETP Study. Am J Psychiatry
172:237–248