Keywords
lithium - pharmacoepidemiology - bipolar disorders - polypharmacy - psychopharmacology
Introduction
Lithium is a first-line mood stabilizer comprising the mainstay of maintenance
treatment for bipolar disorders [1]. It is
also frequently recommended as an augmentation strategy in patients with unipolar
depression who have not responded to previous antidepressant treatments [2]. Although there is some experience in
treating other diagnostic groups, such as schizophrenia spectrum disorder, with
lithium [3], the evidence quality is low.
Lithium is one of the very few drugs with suicide-preventive properties, making it
crucial for patients with suicidal ideation [4]
[5]. Long-term lithium
treatment may benefit illness progression through inflammatory modulation [6] and neuroprotective effects [7].
Despite strong evidence for lithium’s efficacy and its recommendation for affective
disorders in various international guidelines, pharmacoepidemiological data from
several countries indicate a consistent decline in lithium prescriptions [8]
[9]. Reasons for this decline include concerns over adverse reactions,
intensive blood monitoring requirements [10], a significant challenge especially in an outpatient setting, and the
rising dominance of second-generation antipsychotic drugs (SGAs) in the treatment
of
bipolar and affective disorders during recent years [8] and of off-label prescriptions of
antidepressant drugs [9].
Understanding the reasons behind the reduced lithium prescriptions is challenging,
especially given the variation in decline rates across different treatment settings
and patient subgroups. For instance, a population-based study showed an early
decline in lithium prescriptions until 2006, followed by an increase in 2010 [11]. Similarly, in a large outpatient
cohort over a 20-year period, Rhee and associates reported a drastic reduction in
lithium prescriptions until 2012, with a subsequent increase by 3.4% between
2013–2016 [8]. In other studies, lithium
prescription remained stable after the first episode of bipolar disorder [12] and even showed a gradual increase over
a 10-year period in Canada [13]. Most
available data focus on bipolar disorder, with less attention on other diagnostic
groups.
Our study aims to provide additional data on lithium use across various psychiatric
disorders over a long period using a large pharmacoepidemiological dataset from
three German-speaking countries. Unlike outpatient studies, where it is uncertain
if
the patients take their prescribed medication, our inpatient study reflects actual
use, and the data represent the “real world”.
Methods
We analyzed prescription data provided by the AMSP (German: Arzneimittelsicherheit
in
der Psychiatrie) program [14]
[15]
[16], an ongoing international multicenter drug safety project. AMSP
collects data on psychotropic drug use as well as reports of adverse drug-induced
reactions (ADRs) from psychiatric hospitals in Germany, Switzerland, and Austria,
between 1994–2017. The number of participating hospitals steadily increased over
time. After 2005, between 50 and 60 hospitals participated annually, including 30–40
from Germany. Data were collected cross-sectionally on two reference days per year,
providing information on drug use, age, sex, and diagnoses of psychiatric
inpatients, independent of treatment duration and prescription purpose. Detailed
methods are described in previous publications (13, 14, 15). In this study, we used
the primary diagnoses determined by hospital staff according to the ICD. Patients
may have had additional diagnoses.
Statistical methods
To capture temporal trends for the use of lithium, we compared lithium use for
different diagnoses and within different diagnostic groups based on the
International Statistical Classification of Diseases and Related Health Problems,
10th Revision (ICD-10) before versus after 2001 (1994–2000 versus
2001–2017) using a chi-squared test (χ
2). The year 2001 was chosen
as a cutoff due to changes in ADR definitions in the AMSP project, although this is
not relevant to our evaluations. An average of around 20 hospitals took part in the
first period and around 50 hospitals in the second. We further divided the years
1994–2017 into three periods (T1: 1994–2001, T2: 2002–2009, T3: 2010–2017) and
performed pairwise comparisons of lithium prescription percentages using χ2 tests.
Additionally, we analyzed yearly prescriptions of antidepressants, antipsychotics,
tranquilizers, hypnotics, lithium, and anticonvulsants, as well as yearly
prescription patterns in patients with various psychiatric disorders (ICD-10:
F0-F9). For time-trend analysis, we used linear regression to estimate the
statistical significance of trends. All analyses were performed using R 4.2.2 with
a
significance level of at least 0.01.
Results
We investigated prescription patterns in 158,384 adult inpatients (54% female, mean
age 47.4±17.0 years) treated with an average of 2.5±1.3 drugs.
Lithium use in different psychiatric diagnostic groups
The results of the comparisons of lithium prescriptions in different psychiatric
diagnostic groups are provided in [Table
1].
Table 1 Lithium prescriptions in different psychiatric
diagnostic groups compared between 1994–2000 vs.
2001–2017.
|
ICD-10
|
Prescriptions filled during 1994–2000
|
Prescriptions filled during 2001–2017
|
p-value
|
|
Lithium-treated
|
Total
|
%
|
Lithium-treated
|
Total
|
%
|
|
F0
|
24
|
1’650
|
1.45
|
79
|
8’501
|
0.93
|
0.07
|
|
F1
|
30
|
1’755
|
1.71
|
162
|
12’661
|
1.28
|
0.18
|
|
F2**
|
748
|
9’736
|
7.68
|
2’228
|
43’910
|
5.07
|
<0.001
|
|
F3**
|
1’209
|
7’211
|
16.77
|
4’976
|
52’009
|
9.57
|
<0.001
|
|
F4
|
28
|
1’586
|
1.77
|
130
|
9’197
|
1.41
|
0.34
|
|
F5**
|
5
|
44
|
11.36
|
9
|
537
|
1.68
|
<0.001
|
|
F6
|
53
|
1’113
|
4.76
|
244
|
6’672
|
3.66
|
0.10
|
|
F7
|
7
|
209
|
3.35
|
33
|
1’051
|
3.14
|
1.00
|
|
F8
|
1
|
27
|
3.70
|
3
|
181
|
1.66
|
1.00
|
|
F9*
|
4
|
33
|
12.12
|
8
|
296
|
2.70
|
0.04
|
|
Total**
|
2’109
|
23’364
|
9.03
|
7‘872
|
135‘015
|
5.83
|
<0.001
|
*statistical significance p<0.05; **statistical significance
p<0.001. Comparing the two time periods, there was a decrease in F2
diagnoses from 41.7% to 32.5% and an increase in F3 diagnoses from 30.8%
to 38.4% (F2: 9´736/23´364 vs 43´910/135´015; F3: 7211/23364 vs
52009/135´015). ICD-10: International Classification of Diseases. The
ICD-10 diagnosis was missing for 5 patients ICD-10; F0: Organic,
including symptomatic, mental disorders; F1: Mental and behavioural
disorders due to psychoactive substance use; F2: Schizophrenia,
schizotypal and delusional disorders; F3: Mood [affective] disorders;
F4: Neurotic, stress-related and somatoform disorders; F5: Behavioral
syndromes associated with physiological disturbances and physical
factors; F6: Disorders of adult personality and behaviour; F7: Mental
retardation F8: Disorders of psychological development; F9: Behavioural
and emotional disorders with onset usually occurring in childhood and
adolescence.
Between 1994–2000 and 2001–2017, lithium prescriptions decreased significantly
across various psychiatric disorders. For schizophrenia spectrum disorders
(ICD-10 F2: F20–29), prescriptions fell from 7.7% to 5.1% (p<0.001); for
affective disorders (F3: F30–F39), from 16.8% to 9.6% (p<0.001). In the other
diagnostic groups, the differences were not significant or were based on very
small numbers.
In three periods (T1: 1994–2001, T2: 2002–2009, T3: 2010–2017), we observed
significant changes in F2 and F3 diagnoses. For F2, rates were 7.4%, 5.3%, and
4.7%, respectively; for F3, they were 16.3%, 10.6%, and 8.6% (p<0.01 for all
comparisons, see Table S1a, supplementary material).
Lithium use in different diagnostic subgroups of schizophrenia spectrum
disorders (F2)
For schizoaffective disorders (ICD10 F25), lithium prescription rates dropped
from 27.8% to 17.4% (p<0.001, [Table
2]). The rates for F25 during T1, T2, and T3 were 26.7%, 18.2%, and
16.2%, respectively. The decline for schizophrenia was also significant, from
3.5% to 2.0%, with rates of 3.8%, 2.4%, and 1.6% over the three periods (see
Table S1b, supplementary material).
Table 2 Lithium prescriptions in different diagnostic
groups with schizophrenia spectrum disorders compared between
1994–2000 vs. 2001–2017.
|
ICD-10
|
Prescriptions filled during 1994–2000
|
Prescriptions filled during 2001–2017
|
p-value
|
|
Lithium-treated
|
Total
|
%
|
Lithium-treated
|
Total
|
%
|
|
F20**
|
256
|
7’431
|
3.45
|
639
|
31’366
|
2.04
|
<0.001
|
|
F21
|
3
|
40
|
7.50
|
2
|
165
|
1.21
|
0.10
|
|
F22
|
5
|
307
|
1.63
|
15
|
1’359
|
1.10
|
0.64
|
|
F23
|
5
|
219
|
2.28
|
25
|
1’912
|
1.31
|
0.40
|
|
F24
|
0
|
5
|
0.00
|
0
|
33
|
0.00
|
NA
|
|
F25**
|
478
|
1’719
|
27.81
|
1’535
|
8’847
|
17.35
|
<0.001
|
|
F28
|
1
|
5
|
20.00
|
6
|
51
|
11.76
|
1.00
|
**statistical significance p<0.001. ICD-10: International
Classification of Diseases; NA: not applicable; F2: Schizophrenia,
schizotypal and delusional disorders; F20: Schizophrenia; F21:
Schizotypal disorder; F22: Persistent delusional disorders; F23: Acute
and transient psychotic disorders; F24: Induced delusional disorders;
F25: Schizoaffective disorders; F28: Other nonorganic psychotic
disorders.
Lithium use in different diagnostic subgroups of affective disorders
(F3)
For affective disorders, lithium prescriptions decreased in bipolar disorder
(ICD-10 F31) from 41.3% to 31.1%, depressive episodes (ICD-10 F32) from 8.1% to
3.4%, and recurrent depressive disorder (ICD-10 F33) from 17.9% to 7.5% (all
p<0.001, [Table 3]). Over the
three periods, rates for bipolar disorder were 40.8%, 31.7%, and 30.0%; for
depressive episodes, 7.7%, 4.2%, and 2.7%; and for recurrent depression, 17.2%,
8.6%, and 6.6%. That means, there was no essential change for bipolar disorder
from 2002 onwards (see Table S1c, supplementary material).
Table 3 Lithium prescriptions in different diagnostic
groups with affective disorders compared between 1994–2000 vs.
2001–2017.
|
ICD-10
|
Prescriptions filled during 1994–2000
|
Prescriptions filled during 2001–2017
|
p-value
|
|
Lithium-treated
|
Total
|
%
|
Lithium-treated
|
Total
|
%
|
|
F30
|
173
|
805
|
21.49
|
88
|
421
|
20.90
|
0.90
|
|
F31**
|
414
|
1’002
|
41.32
|
2’392
|
7’703
|
31.05
|
<0.001
|
|
F32**
|
245
|
3’041
|
8.06
|
646
|
18’995
|
3.40
|
<0.001
|
|
F33**
|
351
|
1’959
|
17.92
|
1’833
|
24’565
|
7.46
|
<0.001
|
|
F34
|
26
|
385
|
6.75
|
13
|
268
|
4.85
|
0.43
|
|
F38
|
0
|
6
|
0.00
|
4
|
30
|
13.33
|
0.88
|
|
F39
|
0
|
13
|
0.00
|
0
|
27
|
0.00
|
-
|
**statistical significance p<0.001; ICD-10: International
Classification of Diseases; NA: not applicable; F3: Mood [affective]
disorders; F30: Manic episode; F31: Bipolar affective disorder; F32:
Depressive episode; F33: Recurrent depressive disorder; F34: Persistent
mood [affective] disorders; F38: Other mood [affective] disorders; F39:
Unspecified mood [affective] disorder.
Lithium use in different diagnostic subgroups of adult personality and
behavior disorders
In personality disorders (ICD-10 F60), lithium prescriptions dropped
statistically, not significantly, from 5.0% to 3.6% (p=0.04, in [Table 4]). For borderline personality
disorder (F60.3), rates decreased from 6.3% to 3.9% (p=0.01). No significant
differences were found across the three periods for F60 diagnoses (table
S1d).
Table 4 Lithium prescriptions in different diagnostic
groups with disorders of adult personality and behaviour compared
between 1994–2000 vs. 2001–2017.
|
ICD-10
|
Prescriptions filled during 1994–2000
|
Prescriptions filled during 2000–2017
|
p-value
|
|
Lithium-treated
|
Total
|
%
|
Lithium-treated
|
Total
|
%
|
|
F60*
|
52
|
1’039
|
5.00
|
204
|
5’736
|
3.56
|
0.04
|
|
F61
|
1
|
45
|
2.22
|
32
|
663
|
4.83
|
0.69
|
|
F62
|
0
|
6
|
0.00
|
3
|
60
|
5.00
|
1.00
|
|
F63
|
0
|
11
|
0.00
|
5
|
124
|
4.03
|
1.00
|
*statistical significance p<0.05. ICD-10: International Classification
of Diseases; F6: Disorders of adult personality and behaviour; F60:
Specific personality disorders; F61: Mixed and other personality
disorders; F62: Enduring personality changes, not attributable to brain
damage and disease; F68: Other disorders of adult personality and
behaviour; F69: Unspecified disorder of adult personality and behaviour;
Note: F60.3*: Emotionally unstable personality disorder, decrease from
6.3% (n=37) between 1994–2000 to 3.9% (n=186) between 2001–2017
(p=0.01). F60: Specific personality disorders; F60.0: Paranoid
personality disorder; F60.1: Schizoid personality disorder; F60.2:
Dissocial personality disorder; F60.3: Emotionally unstable personality
disorder; F60.4: Histrionic personality disorder; F60.5: Anankastic
personality disorder; F60.6: Anxious [avoidant] personality disorder;
F60.7: Dependent personality disorder; F60.8: Other specific personality
disorders.
Yearly prescriptions of various drug groups: antidepressant and antipsychotic
drugs, tranquilizers, hypnotics, lithium and anticonvulsants
From 1994 to 2017, lithium prescriptions decreased from 10.9% to 5.4%.
Antidepressant prescriptions rose from 38.2% to 57.6%. Anticonvulsant
prescriptions increased from 14.5% to 25.7% in 2007, then fell to 20.7% in 2017
([Fig. 1], Table S2).
Fig. 1 Lithium prescriptions from 1994 to 2017 for various groups
of psychotropic drugs. Legend: To study the evolution of psychotropic
prescriptions in our study population, we grouped the prescribed drugs
into antidepressant drugs (ADD), antipsychotic drugs (APD),
tranquilizing drugs (TRD), hypnotic drugs (HYPD), lithium (LI), and
antiepileptic drugs (AEP). For each of these groups, we calculated the
number of patients who received a prescription. The exact values for the
figure can be found in the supplement material (Table S3).
To study the evolution of psychotropic prescriptions in our study population, we
grouped the prescribed drugs into antidepressant drugs (ADD), antipsychotic
drugs (APD), tranquilizing drugs (TRD), hypnotic drugs (HYPD), lithium (LI), and
antiepileptic drugs (AEP). For each of these groups, we calculated the number of
patients who received a prescription. The exact values for the figure can be
found in the supplement material.
Combination of lithium with other drugs
The most common psychotropic drugs combined with lithium were quetiapine (21.1%),
lorazepam (20.6%), and olanzapine (15.2%, [Table 5]). Quetiapine, olanzapine, venlafaxine, and clozapine were
more frequently prescribed with lithium than in the overall study population
(p<0.01).
Table 5 Combination with lithium. N with Lithium, then
percent within Lithium, N total, Percent within total. (N_LI=9940,
N_tot=158’384). Psychotropic Medications (Top 10 in regard of N
(with LI)).
|
Medication
|
N (with LI)
|
Percent (in LI)
|
N (total)
|
Perc (of total)
|
|
Quetiapine*
|
2096
|
21.1
|
27032
|
16.3
|
|
Lorazepam
|
2052
|
20.6
|
30042
|
18.1
|
|
Olanzapine*
|
1513
|
15.2
|
20232
|
12.2
|
|
Venlafaxine*
|
1361
|
13.7
|
15454
|
9.3
|
|
Mirtazapine
|
1172
|
11.8
|
21205
|
12.8
|
|
Valproic Acid
|
1050
|
10.6
|
14116
|
8.5
|
|
Clozapine*
|
977
|
9.8
|
12267
|
7.4
|
|
Diazepam
|
894
|
9.0
|
11837
|
7.1
|
|
Risperidone
|
889
|
8.9
|
18601
|
11.2
|
|
Biperiden
|
674
|
6.8
|
10155
|
6.1
|
*Medications statistically overrepresented in combination with Lithium
(p<0.01).
Lithium prescriptions by most relevant diagnoses
Lithium prescriptions significantly decreased from 1994/1995 to 2016/2017 for
schizoaffective disorders (F25) from 25.4% to 15%, bipolar disorder (F31) from
45.1% to 29.9%, depressive episodes (F32) from 13.9% to 2.9%, and recurrent
depression (F33) from 22.0% to 6.1% ([Fig.
2], Table S3). Relative reductions were 34% for F31, 38% for
F25, 79% for F32, and 72% for F33. Detailed information is in the supplement
(table S3).
Fig. 2 Lithium prescriptions from 1994 to 2017 for various
diagnoses: F25, F31, F32, F33. Legend:This figure shows the evolution of
the percentage of patients prescribed lithium for the clinically most
relevant diagnoses. We aggregated the prescriptions for two consecutive
years and performed a linear regression. The slope of the regression of
the four diagnoses yielded statistically significant results (p<0.01;
Coeff (f): F25: − 0.6 (39) , F31: − 0.7 (36), F32: − 0.4 (35), F33:
− 0.8 (70)).
This figure shows the evolution of the percentage of patients prescribed lithium
for the clinically most relevant diagnoses. We aggregated the prescriptions for
two consecutive years and performed a linear regression. The slope of the
regression of the four diagnoses yielded statistically significant results
(p<0.01; Coeff (f): F25: − 0.6 (39), F31: − 0.7 (36), F32:-0.4 (35), F33:
− 0.8 (70)).
Discussion
Our large-scale data provide valuable insights into inpatient lithium prescription
patterns over more than 20 years. Consistent with previous cohorts of patients with
bipolar disorders [8]
[17]
[18], we found a gradual decline in lithium use. This trend may be due to
the emergence of newer antipsychotic drugs, especially SGAs (which replace lithium),
safety concerns [19], and burdensome
laboratory monitoring [20]. Consequently,
clinicians must be trained to manage lithium-related safety concerns in order to
adhere to current treatment guidelines [1]
[20]. Predicting future
trends in lithium use remains challenging, as its decline has stabilized in some
cohorts [8]
[11]
[13].
Our analysis extends previous findings by showing a decrease in lithium use not only
in bipolar disorder but also in schizophrenia spectrum disorders, such as
schizoaffective disorder. Despite guidelines not strongly recommending lithium for
schizophrenia spectrum disorders, large studies suggest it may reduce
hospitalization risk due to psychosis [21]. The high rates of lithium use in schizoaffective disorders, ranging from
27% at the beginning of the study to 16% in recent years, should be emphasized. The
guidelines emphasize the importance of SGAs in the treatment of the disorder [1], but studies have shown that lithium is
effective for the long-term treatment of schizoaffective disorders [22]
[27]. Lithium use also dropped in patients with borderline personality
disorders, whereas it has remained stable in other countries [23]. A previous study with data from the
AMSP project showed that psychotropic drugs are very frequently used in hospitalized
patients with borderline personality disorder, even though they are not recommended
in guidelines [14].
The very pronounced decline in the prescription of lithium over time in the treatment
of depressive episodes may indicate that augmentation therapy with lithium in
treatment-resistant depression is increasingly being replaced by augmentation with
alternative drugs, especially SGAs. Lithium is effective in the long-term treatment
of recurrent unipolar depression [24]
[25], although lithium also appears to be
increasingly replaced by other options, predominantly antidepressant drugs. In
schizoaffective disorders, first-line treatment includes antipsychotic drugs and not
lithium and anticonvulsants [26]. Thus,
the declining prescription found in the present study is consistent with the current
state of the art [1]
[27]. On the other hand, the large decline
of 37% with respect to prescriptions of lithium in patients with bipolar disorder
(which is the main indication for lithium) is worrisome. As our prescription data
refer to inpatients, we are concerned that the phenomenon of the declining
prescription of lithium may be even more pronounced within the outpatient setting.
However, we found no further decrease in lithium prescriptions in bipolar disorder
in the last two periods, T2 versus T3, from 31.7% to 30% (not significant), i. e.,
from 2002 onwards. This is in agreement with a new paper published while the present
manuscript was being written [28],
reporting no change in bipolar disorder from 2014 onwards, in about 30% of the
patients.
When analyzing the temporal course of the different groups of drugs, the increase
and
then the decrease in prescriptions of anticonvulsants over time is particularly
noticeable. It implies that this group of drugs may have experienced a short-lived
hype in the treatment of psychiatric disorders until the year 2007, subsequently
overshadowed by newer antipsychotics. The distinct increase in antidepressant
prescriptions is in line with an official German drug prescription report [29].
Our findings need to be considered in light of some limitations. First, the lack of
data regarding clinical response to pharmacotherapy and adverse drug reactions may
hamper conclusions on effectiveness and tolerability. Second, we only reported
prescription rates in hospitalized patients, and it may be difficult to extrapolate
these findings to outpatient treatment settings. In fact, patients treated as
inpatients may suffer from more severe symptoms and/or comorbidities [30]. Third, as diagnoses were extracted
from the medical records, including chart reviews, they may be less reliable
compared to structured interviews. Fourth, although we know the diagnoses, we do not
know the exact indication for the prescription of lithium, such as to alleviate
acute symptoms, to augment existing medication, or for prophylaxis. We only use the
main diagnoses; the use of lithium could also be indicated by comorbidity. In
addition to the large data set, a further strength of the study is that
prescriptions correspond to the actual use of the medication, as the data is based
on inpatients.
Conclusions
Our analysis provides a comprehensive review of lithium prescriptions over nearly
25
years across hospitals in Germany, Austria, and Switzerland. We observed a
concerning decrease in lithium prescriptions for bipolar disorder, but also for
other psychiatric illnesses, including schizoaffective psychosis, depressive
episodes, and recurrent depression. This trend underscores the need for clinician
training on lithium use. Further studies should compare the efficacy of lithium and
alternative drugs, such as SGAs, and of combinations with lithium in schizoaffective
disorder, bipolar disorder, and unipolar depression augmentation. This could clarify
whether current prescribing practices are justified by clinical experience.
Ethics approval and consent to participate
Ethics approval and consent to participate
Assessments based on the AMSP database have been approved by the Ethics Committee
of
the University of Munich and the Ethics Committee of the Hannover Medical School
(Nr. 8100_BO_S_2018). This study adheres to the Declaration of Helsinki and its
later amendments. The AMSP program is a continuous observational post-marketing drug
surveillance program and does not interfere with the ongoing clinical treatment of
the patients under surveillance (see Methods: Data source).
Availability of data and materials
Availability of data and materials
The datasets used and/or analyszed during the current study are available from the
corresponding author upon reasonable request.
Authors’ contributions
W.G. and G. Sch. initiated the study. G. Sch., W.G. and M.deB. wrote the main
manuscript. M.deB. conducted the statistical analyses and prepared figures and
tables. N.N. undertook the literature research. J.S., S.T. and R.G. gave suggestions
for improving various versions of the manuscript. All authors proofread the definite
manuscript and reviewed and accepted the final version.
