THE DEnTAL AnD
PHARMAcEuTicAL BEnEfiTs AgEncy
D e pr e s sion
The review of
antidepressants
Authors:
Lead Medical Assessor Anders Wessling
Lead Health Economist Joakim Ramsberg
DE PRESSION
The review of antidepressants
The DenTal anD
PharmaceuTical BenefiTs agency
The review of anTiDePressanTs
Dental and Pharmaceutical Benefits agency
P.o Box 55 [sundbybergsvägen 1],
se-171 11 solna, sweden
Telephone: +46 8 568 420 50
first edition
first printed, December 2008
This report may be ordered from:
[email protected]
2
The review of anTiDePressanTs
Authors:
lead medical assessor, Dr. med. sci., anders wessling
lead health economist, D. Phil., Joakim ramsberg
External experts:
specialist in general medicine, stig andersson
specialist in psychiatry, sten Thelander
Professor of psychiatry, anna Åberg-wistedt
Project group:
marianne aufrecht-gustafsson
Joakim ramsberg
anna märta stenberg
anders wessling, Project manager
Decision-makers:
chairperson in the Dental and Pharmaceutical Benefits agency, axel edling
Professor Per carlsson
senior Physician, eva andersén-Karlsson
senior Physician, rurik löfmark
specialist in general medicine, ingmarie skoglund
senior Physician, gunilla melltorp
Professor rune Dahlqvist
Docent, ellen vinge
former member of Parliament, ingrid andersson
association chairperson, christina Bergdahl
Therapeutic group:
n 06a antidepressants
3
The review of anTiDePressanTs
4
The review of anTiDePressanTs
Contents
1
Why the TLV has carried out a review of antidepressants
2
A Summary
12
3
Depression
3.1.1
3.1.2 3.2 3.2.1 3.2.2 3.2.3
3.3
3.3.1
3.3.2
3.4
3.4.1
3.4.2
3.5
3.6
Depression
Anxiety
Great losses in quality of life
Quality of life
Mortality
Quality of life expressed in quality-adjusted years of life (QALY)
Large costs
Depression
Anxiety
Antidepressants
Antidepressants are not only used for depression
Heavy increase in use
Treatment recommendations for medicines used in depression
Other treatments
17
17
18
19
19
20
20
21
22
22
23
24
25
26
27
4
Treatment effects on depression
4.1
4.2
4.3
4.3.1
Ratings scales and outcome measures
Efficacy in the ”acute phase”
Systematic literature reviews – depression
The SBU – all antidepressants have an effect but it is difficult to document differences in efficacy
NICE – difficult to find differences in efficacy between the medicines
University of North Carolina (UNC) – escitalopram better than citalopram, sertraline and venlafaxine better than fluoxetine
Canadian Coordinating Office for Health Technology Assessment
(CCOHTA) – comprehensive review, no differences between substance groups
Norway – seldom differences between medicines, escitalopram one of the exceptions
Effect after change of medicine
Review finds few well-executed studies and no definite answers
What does Star*D show?
Effect from continued treatment and long-term treatment
Side-effect profiles
4.3.2
4.3.3
4.3.4
4.3.5
4.4
4.4.1
4.4.2
4.5
4.6
9
29
29
30
30
31
31
34
35
36
37
37
38
39
41
5
The review of anTiDePressanTs
6
4.7
4.8
4.8.1
4.8.2
4.8.3
In summary – not worse than, but better?
Own meta-analysis of effect
Methods for meta-analysis Results
Discontinuation in study due to side-effects
42
43
45
46
49
51
51
52
52
52
53
53
54
5
Treatment effect for anxiety
5.1
5.2
5.3
5.4
5.5
5.6
5.7
Panic syndrome
Specific phobias
Social phobias
Compulsive-obsessive syndrome
Post-traumatic stress disorder, PTSD
Generalised anxiety disorder, GAD
Bulimia
6
Is pharmaceutical treatment of depression cost-effective? – literature review
55
6.1
6.2
6.3
6.3.1
6.3.2
6.3.3
6.4
Cost-effectiveness comparisons between medicines from different classes 56
Is there any difference in cost-effectiveness within the different classes of medicines?
57
Which strategy for using pharmaceuticals is most cost-effective
57
Is pharmaceutical treatment cost-effective compared to no treatment?
57
Is long-term treatment cost-effective?
58
Is pharmaceutical treatment cost-effective compared to other treatments? 58
Do the studies support our stance?
59
7
Is the pharmaceutical treatment of anxiety cost-effective?
7.1
7.2
7.4
7.5
7.6
7.7
Panic
Compulsive-obsessive syndrome
Generalised anxiety syndrome
Post-traumatic stress disorder, PTSD
Bulimia nervosa
Lack of health economic studies and data on efficacy
8
Own health economic model
8.1
8.2
8.2.1
8.2.2
8.2.3
8.2.4
8.3
Some points of departure
Costs and effects for twelve-month treatment
Preconditions for the model
Efficacy data for the model
Costs in the model
Results
Costs and effects when the first treatment has not succeeded
62
62
63
64
65
65
66
67
67
68
69
70
71
71
73
The review of anTiDePressanTs
8.3.1
8.3.2
8.3.3
8.3.4
8.4
Preconditions for the model
Effect data for the model
Costs in the model
Results and a commentary
Conclusions from our own health economic analyses
74
75
75
76
76
78
78
79
80
80
81
81
82
83
83
84
84
84
86
87
9
Decisions
9.1
9.2
9.2.1
9.2.2
9.3
9.3.1
9.3.2
9.3.3
9.3.4
9.3.5
9.3.6
9.4
9.4.1
9.5
Preconditions
Pricing corridor
The pricing corridor does not constitute an absolute price ceiling
Arguments for a pricing corridor for antidepressants
Body of knowledge
SSRIs
SNRIs
MAO-inhibitors
NRIs
Alpha-2 antagonists
TCAs
Decisions
Strengths and packages
A number of companies decreased their prices prior to the decision
10
Appendix – Base documentation and decisions
10.1
Appendix – base material for pricing corridor
89
89
11
Appendix meta-analysis
11.1
11.2
Meta-analysis – included studies
Meta-analysis – sensitivity analysis
92
92
96
12
Appendix – review of health economic literature in regard to antidepressants
98
List of references
102
References
102
The following appendices are available at www.tlv.se/depression
– Literature search for meta-analysis
– Cost-effectiveness studies
– Socio-economic costs
7
The review of anTiDePressanTs
8
The review of anTiDePressanTs
1 Why the TLV has carried out a
review of antidepressants
On adopting new reimbursement rules in October of 2002, it was not practically pos­
sible to review all medicines according to the new rules overnight. Therefore, the TLV
is now conducting a review of approximately 2,000 medicines to see if they should
continue to be given reimbursement status in the future. Each of the medicines will be
tried according to the new rules and will either retain or lose reimbursement
Status or be granted restricted reimbursement.
More health for our money
The purpose of the new reimbursement rules is to extract as much health as possible
for every tax crown expended on medicines. We eliminate those medicines that do not
show sufficient effectiveness in relation to what they cost. However, this does not mean
that we aim only to have inexpensive medicines in the pharmaceutical reimbursement
system. If a medicine has positive effects on a person’s health and quality of life, and
on a socio-economic level as a whole, then it may also be expensive.
Three principles for our decisions
In reimbursement decisions for a medicine, we shall evaluate whether or not it is cost­
effective. This means that we weigh the effectiveness of the medicine against its cost.
We also incorporate other principles into our evaluation: the needs and solidarity prin­
ciple, which means that those who have the greatest medical needs shall receive more
of our healthcare resources than other patient groups; and the human value principle,
which means that we must respect the equal value of all individuals.
49 therapeutic groups to be reviewed
In this review we are evaluating medicines in one therapeutic area after another. The
review encompasses a total of 49 groups of medicines and the order in which they are
tried is determined by how large the sales figures were for each respective group in
2003. The medicines that sold the most will be reviewed first. At www.tlv.se/genom­
gang we present the therapeutic groups being reviewed right now, the medicines inclu­
ded in those reviews and more.
Extensive research and groundwork
Before any decision is made, we perform a comprehensive investigation and analysis
of data on medical effect and cost-effectiveness which we request from pharmaceuti­
9
The review of anTiDePressanTs
cal companies in regard to their medicines. We also review the scientific, medical, and
health economic literature available for the group of medicines to be reviewed. In ad­
dition, we sometimes need to construct our own health economic models. We publish
each completed review in a final report. The report documents the existing body of
scientific knowledge for the group in question. We also prepare a synopsis of the re­
port to be printed separately.
Assessment by independent external experts
The assembled knowledge in regard to medical effect and health economic documen­
tation which we present in the final report has been assessed by independent external
medical experts. The report has also been circulated for comments to the SBU (The
Swedish Council on Technology Assessment in Health Care), Medical Products Agency
and the National Board of Health and Welfare. The companies and patient organisa­
tion groups concerned, as well as the county councils’ pharmaceutical reimbursement
group, have also had the opportunity to give input.
What makes a cost-effective medicine?
When we try whether or not a medicine should be granted reimbursement, we shall
evaluate whether or not the medicine is cost-effective or, put more simply, if the
medicine is worth its price. That is if treatment with the medicine costs an amount
of money reasonable for society in relation to the healthcare benefits that the medi­
cine delivers. How large the cost of a medicine is, is therefore not a good measure
of whether or not we are using the right medicine or even a sufficient amount of it.
However what is important, is that the use of a medicine is cost-effective, not just for
healthcare, but for society as a whole. Investigating how cost-effective a medicine is
gives us a foundation for being able to prioritize and therefore use our resources in the
best possible way.
The value derived is balanced against the cost incurred
What then, does it mean, for use of a medicine to be cost-effective? Firstly, it does not
mean that all inexpensive medicines are cost-effective, while more expensive ones are
not. When determining if a medicine is cost-effective, all the expenses associated with
the medicine being used must first be added up. There is, first and foremost, the cost
of the medicine. However, costs can also arise due to the patient visiting a physician to
receive the medicine, if any other additional healthcare assistance is needed, as well as
any side-effects that the medicine may cause.
This total cost is weighed against the benefit that the medicine provides, primarily in
the form of healing, alleviation of pain and increased quality of life for the patient.
One must also consider that use of the medicine may also entail savings in other areas
of healthcare, in that the patient does not need to visit the doctor as often, avoids
hospitalisation, operations, etc. However, all this is still not enough to gain a societal
10
The review of anTiDePressanTs
perspective. We also have to account for whether or not a medicine allows a patient
to work and earn a living and contribute to our common welfare instead of being on
sick leave or even being forced into early retirement. Here benefits go to the individual
in production, and to the state, who then avoids fees for sick leave and early retire­
ment. If the patient is older, it is possible that use of the medicine may lead to the
individual’s being able to take better care of himself or herself and thereby require less
assistance from elderly care services or relatives. This is also a socio-economic benefit
on the plus side of a cost-effectiveness analysis.
Cost-savings are not obligatory
Sometimes the positive effects of a medicine are so great that they entirely compensate
for the medicine’s costs. Then it can be said that the treatment is cost-saving. But we
do not make such high demands to consider use of a medicine to be cost-effective;
in other words that it has a reasonable cost when seen in relation to the effect and
therefore should be reimbursed. That people are healthy, without pain, and able to live
a more normal life by taking a medicine constitutes a great value for which society is
prepared to pay.
11
The review of anTiDePressanTs
2 A Summary
The TLV, Dental and Pharmaceutical Benefits Agency, has completed its review of the thera­
peutic group of medicines called antidepressants. The review was presented in December
2008. This chapter is a summary of the report which is available for download on www.tlv.
se/depression and may also be ordered by mailing to [email protected].
Antidepressant medicines - areas of use
Antidepressant medicines are not only used for treating depression but also for various
anxiety-related symptoms. We deal with the use of the medicines for treating depression
and anxiety but in this evaluation put the emphasis on the treatment of depression. There
are both older and newer medicines in this therapeutic group. Usage is dominated by the
SSRI medicines, which represent two-thirds of usage and comprise a number of well-known
brands such as Cipramil and Zoloft.
As much health as is possible for our tax money
The review of medicines used for treating depression is part of the TLV’s general review of
the entire list of reimbursed medicines. The objective of the review is to ascertain if medici­
nes should continue to be part of the high cost threshold. There are three principles which
shall lay the foundation for our decision for whether a medicine should be reimbursed
using State funds.
The principles are:
cost-effectiveness principle
needs and solidarity principle
human value principle.
In order to judge whether usage of a medicine is cost-effective we weigh the value obtained
against the cost from a societal perspective. We also consider that those in the greatest
medical need shall have more of the resources available within healthcare. The equal value
of all people is also part of our evaluation.
The purpose of our work is to extract as much health as possible for the tax money expen­
ded on medicines. Our reviews free up money which can be used for other urgent areas of
treatment within healthcare.
12
The review of anTiDePressanTs
The fifth review
This is the fifth review we are presenting. The first one was on migraine medicines and was
completed in 2005. Following this we presented the review of medicines against stomach
acid, then one on medicines used for treating asthma, COPD and coughing, and then the
review on medicines against hypertension. You can read more on these and the currently
ongoing reviews on our website at www.tlv.se/genomgang.
What is depression and anxiety?
Depression is a word used in many different contexts and can cover many different mea­
nings. In a medical context the term depression is used to describe a cluster of symptoms
which often occur together, but where the underlying causes may vary. Depression is charac­
terised by extended periods of dysphoria, feelings of meaninglessness and hopelessness.
For doctors to be able to see and diagnose ”depression” it is also necessary for the patient to
have difficulties in his/her working or private life. Many patients suffering from depression
are also afflicted with anxiety. Anxiety can be described as a group on non-specific unplea­
sant symptoms which are similar to the reactions to horror and fear. The threat can be expe­
rienced by the patient as external or internal. Conditions counted as belonging to the sphere
of anxiety are phobias, panic attacks, compulsive thoughts and compulsive actions.
4 to 10 percent meet the criteria for depression
Depression-related illnesses are one of the most common causes of ill health, productivity
losses and the inability to work in the world. Sweden is no exception. These illnesses carry
great losses in quality of life both for those afflicted and their closest circle. It is estimated
that 4 to 10 percent of the adult population in Sweden meet the criteria for depression. The
number of women is twice as high as the number of men. Those affected by depression once
are often at risk again, at least one more time during their lives. Depression can also be con­
nected to severe physical diseases such as cancer and cardiovascular diseases. Many affected
by the disease also have other psychological disorders.
700 000 people treated during 2007
According to the latest figures the annual societal cost for depression and anxiety is approx­
imately 40 billion Skr. The societal costs arising from these conditions are costs for healt­
hcare, including medicines, sick leave, early pension and also premature death. Every year
about 1000 people commit suicide in Sweden, of which a large share is due to depression.
Not even half as many die from traffic accidents each year. Just over 700 000 people were
treated with antidepressants during 2007. According to a study by the Board of health and
Welfare this usage is not unreasonably large, but there is probably over-treatment of certain
patient groups and under-treatment of others. The costs for antidepressant medicines in
2007 were 990 million Skr and society reimbursed these costs to the value of 660 million
Skr, which was approximately 3 percent of the total pharmaceutical reimbursement bill in
Sweden. The decisions in this review will free up 40 million Skr per year.
13
The review of anTiDePressanTs
SSRI medicines dominate
There are different ways of treating depression and anxiety. Medicine is one treatment
alternative and psychological treatment another. During the spring of 2009 the National
board of Health and Welfare will present national guidelines for the treatment of depres­
sion and anxiety. With these national guidelines treatment using pharmaceuticals will be
put into a larger context. Our task is to take a stance on which antidepressants should be
included in the pharmaceutical reimbursement system. We have evaluated the 17 active
substances included in the ATC system’s group for antidepressant drugs, N06A. Up until
the 1980s tricyclics were mainly used as antidepressants. These medicines are effective but
can have uncomfortable side-effects. During the 1980s a new class of antidepressants called
SSRIs were launched. One of the first of these was Prozac, sold in Sweden under the brand
name Fontex. These drugs are also effective but do not have as much uncomfortable side­
effects as the tricyclic drugs. This led to a heavily increased use of antidepressants during
the 1990s. Other substances have also joined the frame such as mirtazapine (Remeron and
Remeron-S), venlafaxine (Efexor) and duloxetine (Cymbalta). The use of antidepressants
has continued to increase since the year 2000, but not at the same rate as during the 1990s.
SSRI drugs stand for the greatest use. In 2006 they made up almost 70% of the volume in
the range of antidepressant drugs.
Our own meta-analysis and health-economics analysis
A great number of clinical studies have been carried out to document the effect of anti­
depressants. The factual material is quite comprehensive but simultaneously difficult to
interpret and too contradictory to answer the question of differences in effect between these
medicines.
To augment the existing reviews we have carried out our own review and analysis, focused
on how many patients recover following treatment with antidepressants. The analysis is
based on information on 20 000 patients treated for depression.
We also needed to carry out a health economics analysis in order to understand how cost­
effective the treatment of depression using antidepressants is in Sweden.
SSRI as first line. Important with a broad range of products
According to most current recommendations the treatment of depression should be started
using SSRI drugs. This concurs with what we have found to be most cost-effective. However,
we can also state that not even half (only 40-50 percent) of the patients achieve a satisfactory
result from the medicine they first try as a treatment. For some there is no effect. For others
the medicine may give an effect but due to issues such as side-effects it is quite common that
patients stop treatment. Changing to another medicine can have an effect, regardless if the
switch is to another medicine within the same drug class or to a medicine in another drug
class. For this reason it is important that there are a number of substances to choose from
within the high cost threshold.
14
The review of anTiDePressanTs
On the whole we see that antidepressants have an effect on treating depression, but there
are differences in efficacy between different substances. In an analysis limited to the four
SSRI medicinal preparations which have generic alternatives, there is in practice no diffe­
rence in efficacy. In our health economics analysis the result indicates a somewhat better ef­
fect for the SSRI substance called escitalopram (Cipralex) than for the other antidepressants
when it comes to how many patients recovered. The model also points to a somewhat lower
societal cost even if the difference is small. The medical efficacy of fluvoxamine (Fevarin)
does not however measure up to its current price.
Expensive original brand drugs lose reimbursement – all substances except for
one remain in the reimbursement system
In order for as many patients as possible to receive satisfactory treatment for depression
and anxiety it is important that there is an opportunity to test one’s way towards the best
treatment result. We have taken this into account and decided that all substances in the
antidepressants group shall continue to be reimbursed, excepting the SSRI substance flu­
voxamine (Fevarin). The price for fluvoxamine is considerably higher than for SSRI copies.
Fluvoxamine’s medical effects and qualities do not motivate the higher price.
In the group of antidepressant medicines there are original brand name drugs whose
patents have expired, and copies which have the same medical effect. Our evaluation shows
that there are price differences between original brand name drug and the copies. We can
also see that doctors prefer to prescribe the more expensive originals despite there being
copies with the same medical effect at a considerably lower price. It is not cost-effective to
reimburse the original brand name drugs whose patents have expired as the copy is much
cheaper. Some of the original drugs result in a treatment cost of 8 Skr per day, whereas the
comparable cost for treatment per day for the cheapest copy is not even 1 Skr.
For this reason we have decided that expensive original brand drugs, which have cheaper
generic copies, shall no longer be reimbursed. We have also decided that the highest price
for a tablet within the SSRI group shall be 3 Skr in the most popular strength in the 100­
tablet package size. Due to this most of the companies chose to decrease their price for
their original brand drugs in order to retain reimbursement status in the future.
These decisions come into force on the 1st of April, 2009.
More expensive medicines can be more cost-effective
As depression and anxiety cause both suffering and decreased quality of life for those af­
flicted with the conditions as well as considerable societal costs, small differences in effect
can motivate the price differences which exist. In relative numbers the price differences may
appear to be large, but in absolute terms, that is in Swedish crowns, they are small. Antide­
pressants are in other words not in the expensive range of medicines. Both our review and
the reviews conducted by others indicate that some new and more expensive medicines may
have a better effect and cost-effectiveness than older medicines.
15
The review of anTiDePressanTs
The healthcare system does not have the information it needs to treat patients
effectively
Depression-related illnesses are so common that they should be treated as a health problem
for society as a whole. Pharmaceutical treatment can contribute to alleviating the symptoms
of depression-related conditions but a precondition for this is a choice of well-researched
antidepressants for treating patients. Another precondition for the best use of pharmaceuti­
cals is a collaboration which functions well between the caregiver and the patient. The cost
for making the wrong decision could be high, in both human and economic terms.
We see gaps in the information needed by those working in healthcare to make decisions
in regard to the treatment of patients – not least in comparing medicines to other forms of
treatment.
Our review gives more knowledge on antidepressants to those in positions of responsibility
within healthcare. But we also realise that there are many areas where research is needed for
depression and anxiety, and that more knowledge on these medicines is needed for patients
to receive an effective treatment.
16
The review of anTiDePressanTs
3 Depression Depression is a word used in many different contexts and can cover many different
meanings. In a medical context the term depression is used to describe a syndrome,
that is a cluster of symptoms which often occur together, but where the underlying
causes may vary.
Depression itself is a central component in the depressive syndrome but other symp­
toms are also part of it. Anxiety commonly occurs for many depressed people, as does
a constant feeling of inner worry and anguish. Furthermore, there is also a decrease in
emotional commitment.
Depression-related illnesses are one of the most common causes of ill health, losses
in productivity and the inability to work in the world. Sweden is no exception to this
and depression causes both direct (health care, medicines) as well as indirect (losses in
productivity, premature death). And the condition carries great losses in quality of life
both for those afflicted by the illness and for their loved ones.
3.1.1
Depression
All-permeating and extended periods of depression, feelings of meaninglessness and
hopelessness characterise the illness known as depression. To be defined as what is cal­
led severe depression, or unipolar depression, the illness must also cause difficulties on
a person’s private life and working life. ([1], p 15).
There are a number of different systems for diagnosing depression and a number of
different types of scales and interview instruments are used to establish the existence
of depression. The most common scales are the Hamilton Depression Rating Scale
(HDRS) and the Montgomery Åsberg Depression Rating Scale (MADRS). The most
common self-assessment scale is the Beck Depression Inventory (BDI) ([1], p 48 ff).
It is estimated that at any given time between 4 and 10 percent of the adult population
fulfil the criteria for depression. The share of women is roughly double the size of men
[1]. For bipolar depression (see more below) the occurrence is 1–2 percent. Here there
is no express difference between men and women. The lifetime risk of being afflicted
by a severe (unipolar, see more below) depression is often estimated at 17–18 percent
[1]. There are studies which indicate that depression in the past years has become more
common and that it starts at younger ages ([1], p 74).
17
The review of anTiDePressanTs
Most people afflicted by depression get at least one more episode later in life [1], [2].
Many people with depression also have other psychological disorders, especially va­
rious anxiety disorders, addictions and personality disorders. Most often the anxiety
episodes first show prior to the depression. Patients suffering from depression often
have physical conditions such as diabetes, cardiovascular disease, multiple sclerosis
and other neurological diseases.
Depression-related illnesses can be divided up into sub-groups based on the type of
illness. Also, a main distinction may be drawn between unipolar disease and bipolar
disease. Patients suffering from unipolar disease often have only one type of illness
phase, which is characterised by depression. Those with bipolar disease also have a
manic illness period which is characterised by an elevated state of mind, expansive
plans, rapid association, worsened judgement and difficulty in setting borders for new
contacts. In more difficult forms delusions and aggressiveness may also occur.
Patients experiencing a manic episode have limited or non-existent self-awareness
of the disease.
3.1.2 Anxiety
The following description of anxiety is from the SBU [3]. Within psychiatry anxiety
describes a group of non-specified, unpleasant symptoms which are similar to the ex­
perience of terror and fear. In the Swedish translation of the diagnostic criteria in the
Diagnostic and Statistical Manual for Mental Disorders, DSM-IV, anxiety is defined as
feelings of discomfort or somatic symptoms of tension prior to an oncoming danger
or accident. The threat experienced may be both outer and inner. Phobias are also part
of anxiety and are defined as irrational fear of a specific event, activity or situation and
resulting in a strong desire to avoid the dreaded event, activity or situation.
Other conditions counted as belonging to anxiety syndrome besides anxiety itself are:
panic attacks – delimited and sudden attacks of intense fear or terror and often
connected with feelings of doom.
compulsive-obsessive thoughts – recurring, resilient and meaningless ideas,
thoughts, images or impulses which are not experienced as consciously produced
rather as ideas which force their way in and penetrate consciousness
compulsive-obsessive actions – repeated behaviour which appears to be objective­
oriented and is carried out in reply to a compulsive thought, in accordance with
certain rules or in a stereotyped fashion
This division into syndromes is based on observation and not on knowledge in regard
to the underlying mechanisms. Neither are there any biological markers which may be
utilized to make the diagnosis. Anxiety syndrome lasts as a rule for a much extended
period of time, with symptoms often lingering for many years. The long-term outcome
18
The review of anTiDePressanTs
from treatment is therefore of great importance. Patients suffering from anxiety also
often have other illnesses, both physical and neurological.
3.2 Great losses in quality of life
Depression is an illness which comprehensively impacts on a patient’s life situation. In
this way it can result in very, large losses in quality of life. A systematic judgement on
exactly which losses in quality of life a disease results in demands evaluation and the
weighing of a number of different factors, where the most important are the diseases’s
mortality, function and independence. Quality-adjusted years of life (QALY) are, at least
in the ideal case, a weighing of all of these factors from the perspective of the patient
and therefore especially interesting when we want to study losses in quality of life.
3.2.1 Quality of life
There is a quite comprehensive literature on quality of life and social function for
depression, but few systematic reviews. Besides the SBU’s report there is a systematic li­
terature overview on life quality for unipolar depression by Papakostas and colleagues
[4] and one on social function and depression by Hirschfeld and colleagues [5].
Pincus and Petit [6] have carried out an overview of the burden of the disease for de­
pression where quality of life is a component.
The unequivocal impression is that depression is for many a seriously disabling condi­
tion. Papakostas and colleagues state that depression is the main cause of disability in
the Western World today. People suffering from depression have disorders in functio­
ning at work which can lead to longer sick leave and exclusion from the labour market,
and in their free time. The stigmatization, that is to say the negative social consequen­
ces of identifying somebody as mentally ill, is believed to play a role in this context. [1].
Relationships, sexuality, marriage and relationships to offspring are affected negative­
ly by depression, as are private finances. People suffering from depression run a greater
risk of divorce than people who are not subject to depression. It is also less likely they
will complete their education. The SBU points out that causality cannot however be es­
tablished completely. There may be the same driving factors (such as difficult upbring­
ing or stress) which lead to both depression and divorce.
The SBU comments that there is surprisingly little information on the drain on relati­
ves when it comes to depression as also those who live with people afflicted by depres­
sion are greatly affected. One group possibly especially influenced are the children
of parents suffering from depression. Being the child of a depressed parent carries a
greatly increased risk of suffering from depression for the child involved. As many as
40 percent of the children of depressed parents will suffer from depression before 20
years of age.
19
The review of anTiDePressanTs
Quality of life and function are often affected more by depression than many other
chronic illnesses. The decrease in function in regard to depression is as large or larger
than eight other chronic and somatic diseases (such as diabetes and osteothrosis)
studied. It is only patients with acute coronary heart disease who are more bed-ridden
than depressed patients. [5].
3.2.2 Mortality
Besides a decline in quality of life people suffering from depression run a higher risk
of early death. They have up to double the risk of early death compared to the rest of
the population [7], although the exact level appears to be a long way from being esta­
blished [8]. This elevated rate of mortality is a result of both suicide and the effect on
the general state of health for the individual mainly in the form of coronary heart di­
sease, which is also mentioned by the SBU in their report. The SBU refers in its review
to a WHO study which states that more than 800 000 people die every year as a result
of suicide. Of these, at least 90 percent have had a mental illness, of which the majority
suffered from depression, before their demise. In Sweden the number of suicides for
the past years has been approximately 1 300–1 500 [9].
The number of suicides in Sweden, as in the rest of the Western World, is declining.
The increased use of antidepressants may be assumed to have contributed to this.
Early research showed that for those who have received hospital care for depression
the risk of suicide is as high as 15 percent, and for especially severe depression is even
higher [10]. More recent research seems to indicate that this figure is too high, at least
for today’s day care patients. The lifetime risk for these patients rather seems to lie
between 2 and 6 percent [11].
3.2.3 Quality of life expressed in quality-adjusted years of life (QALY)
Quality-adjusted life year (QALY) is an established way to describe the dimensions,
quality and duration of health using one single measure. One year with the best pos­
sible state of health is given the value “one”, while a year with a decrease in health gets
a value which is less than one. We have carried out a systematic review of the literature
in regard to quality of life weighting for depression and anxiety [12]. Twenty studies
with around 7400 people were part of this review. The articles were analysed by struc­
ture of study, state of health, degree of severity and method for measuring quality of
life. The results are primarily based on answers from patients with subjective states
of health and to a lesser extent on answers where the respondents relate to an already
described state of health. In order to estimate the weightings twelve different methods
had been used. The most common was standard gamble and EQ-5D. Standard gamble
is a direct method where the utility value is measured directly for the person asked.
EQ-5D is an indirect method where the degree of severity for a state of health is esta­
blished with the aid of the question battery EQ-5D and then collating the utility values
for the state of health from a table. The review showed that the degree of severity for
20
The review of anTiDePressanTs
the illness was positively correlated with the QALY weightings, thereby supporting the
credibility of the measures used.
In theory the optimal weightings are reached by using standard gamble on a real pa­
tient population. The two studies which here would have the most validity are the one
by Wells [13], showing a weighting of 0.94 for mild depression, and the one by Revicki
and Wood [14] showing an average weighting of 0.30 for untreated severe depression
and 0.55–0.63 for moderate depression. From the latter study it can be noted that one­
fourth of the patients saw their existence as worse than death. Amongst patients who
were in remission and were not undergoing maintenance treatment Revicki and Wood
measured a quality of life weighting of 0.86 on average.
No study was identified where this method was used for anxiety.
Two important Swedish studies have emerged since the review was carried out. So­
bocki and colleagues [15] followed patients in Swedish primary care for six months
who were being treated for depression.
The quality of life weightings were measured using EQ-5D. When the patients com­
menced their treatment they had on average a quality of life weighting of 0.47. After six
months the average quality of life weighting was 0.81 for patients who recovered, and
0.57 for those who had not recovered. The number of patients in the study was 447,
all of them were 18 years of age and over. Von Knorring and colleagues [16] followed
approximately 1 000 Swedish patients during their treatment for unipolar depression.
The patients were over 18 years of age (average age of 48.4). The quality of life weigh­
tings were established using EQ-5D. The patients had an average quality of life weigh­
ting of 0.61 when they commenced treatment. After two years the patients who had
responded to the treatment (50 percent decrease in MADRS points) had an average
quality of life weighting of 0.81 compared to 0.70 for those who not responded to the
treatment. In our estimation the studies conducted by Sobocki and colleagues as well
as von Knorring and colleagues are those carrying the most validity for our purposes.
They are further supported by their concurrence with the results from Revicki and
Wood.
3.3 Large costs
Depression and anxiety are illnesses which, besides suffering, result in large costs and
economic losses for both individuals and society. A description of the socio-economic
cost is relevant from the perspective of a decision-making body by showing how large
losses in resources caused to society are.
The costs arising from the illness are usually divided up into direct and indirect. Direct
21
The review of anTiDePressanTs
costs are those which relate to the treatment of the illness such as hospital visits and
pharmaceuticals. Indirect costs are the loss in resources which arise due to the patient
not being able to work (morbidity costs) or dying prematurely (mortality costs). The
SBU details the socio-economic costs in its reports on depression and anxiety. These
details are based on the systematic literature searches up to January 2002 and February
2005 respectively.
We have augmented the SBU’s material with searches up until the 31st of March, 2008.
Here we present a short summary of the results. The entire base documentation is
detailed in a separate appendix which is available on www.tlv.se/depression, and which
also may be ordered from the TLV.
3.3.1 Depression
It is indisputable that depression leads to great socio-economic costs. The latest Swe­
dish studies published report a total cost of 32.5 billion Swedish crowns [17]. Phar­
maceuticals stood for three percent of this cost. The studies conducted in comparable
countries in general arrive at similar costs. In these studies national registers have been
used as base documentation for the costs estimations. Two, recent and well-conducted
Swedish studies , which instead have prospectively followed a group of patients, have
shown costs which in one case amounts to approximately 360 000 over two years, and
50 000 Swedish crowns for six months respectively. The indirect costs stood for 87 and
74 percent respectively, and pharmaceuticals for 5 percent.
Two common denominators in all studies are that the indirect costs are considerably
higher than the direct costs and that the costs for pharmaceuticals are a small part of
the total costs.
3.3.2 Anxiety
In the SBU report on anxiety syndrome studies from 1996 are mentioned estimating
the socio-economic costs for anxiety syndrome at about 20 billion Swedish crowns.
A number of other studies were also referred to which indicate that anxiety syndrome
is connected to considerable costs for care, of which a large part was somatic care.
A smaller number of studies on indirect costs for anxiety syndrome were identified
but the few studies found indicated large economic effects due to a decline in working
capability. Following the SBU report only one study has surfaced which calculates the
cost for anxiety in Sweden. In a model study Andlin-Sobocki and colleagues [18] cal­
culated the socio-economic costs for anxiety in the 25 EU countries as well as Iceland,
Norway and Switzerland for 2004 to 375 billion crowns. For Sweden the cost amoun­
ted to 12 billion crowns. The anxiety syndrome found to be most expensive per patient
was generalized anxiety syndrome.
22
The review of anTiDePressanTs
3.4
Antidepressants
The pharmaceuticals covered by our review are the medicines approved in Sweden and
which are part of the ATC system’s group of anti-depressant substances (ATC code
N06A). All medicines in the group are intended for treatment of depression, but many
are used also for treatment of various anxiety conditions. Furthermore, some of the sub­
stances are used for treatment of bulimia nervosa and some for the treatment of pain.
When we take a stance on continued reimbursement for individual medicines we also
have to take any treatments for the different types of anxiety syndrome into account.
The medicines achieve their effect by affecting the neurotransmitters in the brain and
the central nervous system - norepinephrine, serotonin och dopamin. If we simplify it
greatly then one can say that how a medicine affects the neurotransmitters seems to
have less importance for which effect they have on the disease, but more importance
for the side-effects which may arise. Even though one cannot discern a difference on a
population basis between two medicines with different side-effects profiles in the pro­
portion of patients cured, the side-effects profile can be important for which medicine
is most suitable for an individual patient. Patients can have different ways of valuing
the medicine’s effects and side-effects.
The group antidepressants contains 17 substances which can be divided up into sub­
groups, in part due to the mode of action. The ATC system contains four such sub­
groups. In the fourth sub-group a further division can be carried out in relation to the
mode of action. The medicines which are part of the review and the division which we
are using is shown in Table 1
Table 1. Medicines in the review of antidepressants
TCA – tricyclics
ATC-code
name of substance
Pharmaceutical name
n06aa04
clomipramine
anafranil – novartis + generics
n06aa06
trimipramine
surmontil – sanofi aventis
n06aa09
amitriptyline
saroten – lundbeck Tryptizol – msD
n06aa10
nortriptyline
sensaval – lundbeck
n06aa21
maprotiline
ludiomil – novartis + generics
SSRI – selective serotonin reuptakes
ATC-code
name of substance
Pharmaceutical name
n06aB03
fluoxetine
fontex – lilly + generics
n06aB04
citalopram
cipramil – lundbeck + generics
n06aB05
paroxetine
seroxat – glaxosK + generics
n06aB06
sertraline
Zoloft – Pfizer + generics
n06aB08
fluvoxamine
fevarin – solvay Pharma
n06aB10
escitalopram
cipralex – lundbeck
23
The review of anTiDePressanTs
MAO-inhibitors – monoamine oxidase inhibitors
ATC-code
code name of substance
Pharmaceutical name
n06ag02
moclobemide
aurorix – roche + generics
Alpha 2 antagonists
ATC-code
code name of substance
Pharmaceutical name
n06aX03
mianserine
Tolvon – organon + generics
n06aX11
mirtazapine
remeron – organon + generics
SNRI – serotonin-norepinephrine reuptake inhibitor
ATC-code
code name of substance
Pharmaceutical name
n06aX16
venlafaxine
efexor – wyeth
n06aX21
duloxetine
cymbalta – lilly
NRI – norepinephrine reuptake inhibitor
ATC-code
code name of substance
Pharmaceutical name
n06aX18
reboxetine
edronax – Pfizer
3.4.1
Antidepressants not only used for depression
As earlier reported all medicines in the therapeutic group we are reviewing are approved
for the treatment of depression, but some of them are also approved for the treatment of
various anxiety-related conditions, as indicated in Table 2. We list both approved indica­
tions and areas of use as documented in well-executed, published studies.
Table 2. Approved and well-documented indications for medicines in the group of antidepressants (N06A)
Depres­
sion
n06aa04
n06aa06
n06aa09
n06aa10
n06aa21
n06aB03
n06aB04
n06aB05
n06aB06
n06aB08
n06aB10
n06ag02
n06aX03
24
clomipramine
trimipramine
amitriptyline
nortriptyline
maprotiline
fluoxetine
citalopram
paroxetine
sertraline
fluvoxamine
escitalopram
moclobemide
mianserine
X
X
X
X
X
X
X
X
X
X
X
X
X
Profy­
lax
com­
pulsive­
obsessive
syndrome
generalised
Panic
social anxiety
syn­
fhobia disorder
drome
(gaD)
X
X
X
X
X
X
X
X
X
X
Post-trau­
matic stress
syndrome
(PTsD)
Buli­
mia
X
o
o
X
o
X
o
X
X
X
X
X
X
X
X
X
X
The review of anTiDePressanTs
n06aX11
n06aX16
n06aX18
n06aX21
mirtazapine
venlafaxine
reboxetine
duloxetine
X
X
X
X
o
X
o
o
o
X
X
X
X
X = approved indication
o = use for which there is evidence in published and well-conducted studies.
in our compilation we have not included indications for pain
3.4.2 Heavy increase in use
If you look at sales of antidepressants from a longer perspective over a period of years
it is possible to discern a limited and relatively stable sales volume (Figure 1).
The sales volume increased slowly during the 1980s and transformed into rapid
growth during the 1990s as the SSRI medicines became available. The level of sales has
continued to increase although the pace of this has abated over the past years.
Figure 1. Sales development for antidepressants 1974–2006
Daily doses per 1,000 inhabitants per day
100
90
80
70
60
50
40
30
20
10
Women
All
2006
2004
2002
2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
0
Men
source: The national Board of health and welfare – antidepressants for mental health. studies on practice in
primary care [19]
Sales from the beginning of 2000 have continued to increase and are dominated
heavily by SSRI medicines which in 2007 answered for two-thirds of the sales volume
in DDD (Table 3). The newer substances show a more rapid increase in sales than
the older ones. Between 2000 and 2007 the sales of mirtazapine have tripled, almost
doubled for venlafaxine and sales of the latest medicine to the market, duloxetine, are
increasing rapidly.
25
The review of anTiDePressanTs
Table 3. Sales of antidepressants within the pharmaceutical benefits system – millions of DDD during the years
2000–2007
n06aa – tricyclics
n06aB – ssri
n06aX11 – mirtazapine
n06aX16 – venlafaxine
2000
2001
2002
2003
2004
2005
2006
12.5
12.3
12.4
12.4
12.3
12.4
12.2
11.9
114.3
131.6
143.8
150.1
154.4
156.3
163.8
169.5
7.5
10.2
13.2
15.7
17.3
18.7
20.8
22.8
10.9
14.2
15.9
17.2
18.1
19.0
19.9
20.9
0.0
1.0
3.7
5.7
5.0
4.7
4.6
4.1
4.1
4.0
3.8
3.7
150.0
173.1
189.9
199.4
206.2
211.3
224.2
234.3
n06aX21 – duloxetine
n06a – others
2007
It should be noted that the sales statistics reflect total usage of the medicines, that is
to say not just for depression and anxiety but also for pain. We do not have access to
information on the distribution of sales for the various diagnoses.
The Board of Health and Welfare has researched the use of antidepressants in primary
care [19]. It is estimated that the total prescription level of medicines cannot be seen
as unreasonable in relation to the prevalence of the conditions they are meant to treat.
There are however shortcomings when it comes to how mental conditions are diagno­
sed and documented, possibly leading to both overtreatment of some patient groups
and undertreatment of others, a pattern which is also to be found in international
studies of healthcare consumption [20].
We have ourselves studied the sale of antidepressants to people over 80 years for out­
patients over a period of one year (October 2005–october 2006). Sales here can be expressed in the number of defined daily doses per thousand inhabitants and day (DDD/
TIND). Sales in different municipalities varied from 82 to 282 DDD/TIND and can be
compared to the average for the country of 165 DDD/TIND.
Sales can also be compared to ”normal” prevalence of depression in older people. In
Sweden two more recent studies have been carried out for individuals over the age of
80. Bergdahl and colleagues [21] have carried out a study of 319 individuals in Umeå in
the ages 85, 90 and 95 years old. Pålsson and colleagues at the Sahlgrenska hospital in
Gothenburg have carried out a study of 392 individuals between the ages of 70 and
85 years old [22]. Bergdahl’s study reports a prevalence of depression of 168/1 000 at
85 years old, 341/1 000 at 90 years old and 323/1 000 at 95 years of age, Pålssons
study reports a prevalence of 130/1 000 at 85 years of age. Based on the assumption
that the number of individuals over 80 years of age will more probably reach 85 years
of age than 90 years of age we estimate that the prevalence of depression should be
around 150/1 000 for individuals over 80 years of age.
26
The review of anTiDePressanTs
Consequently the sales figures used do not clash with the estimations used by the
Board of Health and Welfare in regard to the use of antidepressants in primary care. In
some municipalities the estimation is too large and in others too small, but on average
it is fairly reasonable when it comes to the prevalence of depression.
3.5 Treatment recommendations for medicines used in depression
The MPA has produced a treatment recommendation for depression based on a work­
shop held in January 2004 [23]. In the treatment of mild and moderate depression the
MPA found that on a group level there are no proven differences in effect between vari­
ous antidepressants. On an individual level however there are patients which respond
to one substance while not responding to another, both within and outside the same
pharmaceutical class.
In treating more severe depressions the MPA has found that clomipramine and ami­
triptyline are more effective than SSRI substances. For mild and moderate depression
SSRI medicines are recommended as a first line treatment. Mirtazapine and mianse­
rine are presented as an alternative for patients who have difficulty in dealing with
side-effects for the SSRI medicines. Clomipramine and amitriptyline should be limited
to patients with more severe depressions. For those patients venlafaxine also has pro­
bably a better effect than SSRI medicines.
The MPA stresses that a medicine should be chosen based on the makeup of the de­
pression and the side-effects the patient may be sensitive to.
In the treatment recommendations which NICE have produced we can note that
in routine healthcare SSRIs are the first-line treatment as they are as effective as tricy­
clics. The risk of the patient terminating the treatment is also less. For patients who
do not respond to or cannot tolerate treatment with SSRI medicines first it is recom­
mended that a switch to another medicine is administered as monotherapy. Suitable
alternatives for a switch are other SSRI medicines, or mirtazapine, but it is also pos­
sible to consider other alternatives such as moclobemide, reboxetine and lofepramine.
For more severe depression one may also consider tricyclics and venlafaxine.
In its treatment recommendations NICE brings attention to the importance of
understanding the risk of side-effects before prescription of the drugs. For prescrip­
tion of tricyclics one should remain observant of any possible side-effects. The recom­
mendations express the need for considerable caution when prescribing venlafaxine.
The reason for this is the agency's view in regard to the risks posed by an overdose of
the medicine and the increase in blood pressure which venlafaxine can give rise to.
27
The review of anTiDePressanTs
3.6 Other treatments
Depression and anxiety can be treated using various psychological methods with
well-documented effects, as well as using medicines. For acute treatment of mild and
moderate depression psychotherapy is at least as effective as pharmaceutical treatment
[1, 24, 25]. The effect becomes apparent later but lasts for a longer period of time.
A well spaced out and continued psychotherapy decreases the risk of relapse [1, 24].
A majority of the patients seem to prefer some form of dialogue-based psychotherapy
(individual or in a group) to pharmaceutical treatment [26].
For severe depression and psychotic depression medicines and electroconvulsive
therapy (ECT) is more effective than psychotherapy [1].
The effect gained from St. John’s Wort has been scientifically proven for short-term
and milder forms of depression. The side-effects do seem to be milder than for other
antidepressant substances but treatment is not free of risk, mainly due to the risk of
interactions with other medicines [1, 24]. And furthermore, physical exercise has been
shown to have an effect on depression [24].
How the care given is organised is also shown to have an extremely important role
to play. In primarily foreign studies it has been shown that treatment results can be
improved through the use of patient training, telephone-based support to patients and
good access to psychologists and psychiatrists who are trained in psychotherapy over
shorter periods of time [1].
28
The review of anTiDePressanTs
4 Treatment effects on depression
A large number of clinical studies have been carried out to shed more light on the effi­
cacy or effectiveness of antidepressants. Many of them have been published in scienti­
fic publications. The results from these studies have become the subject of a number of
systematic reviews, such as that done by the SBU.
If one looks at the number of studies then the extent of the fact-based material appears
to be quite extensive. When one is looking for the answer to questions on differences
in efficacy which may exist between various sub-groups or individual antidepressants
this seems quite difficult to interpret and contradictory.
A contributing factor here is the use of different measures of outcome in the studies.
It is possible to measure the average degree of improvement in a group of patients,
the proportion of patients who have achieved a certain degree of improvement, or the
proportion of patients who have recovered. Another reason that the studies do not
give definite answers to the issue of efficacy differences between the medicines is the
low number of patients in the individual studies.
4.1
Ratings scales and outcome measures
Psychiatric diagnosis is based on defined criteria. As there are no biological markers
which set the foundation for the diagnosis a number of criteria have been defined
which must be fulfilled in order for a certain diagnosis to be made [1]. Based on the
criteria which have been established and as a support in the diagnosis phase some dif­
ferent types of diagnostic instruments have been produced such as interview templa­
tes.
However, the diagnostic instruments do not measure the severity of the condition.
When one wants to evaluate the results from a treatment it is through changes in the
degree of severity that the results would best be tracked.
This is done with the aid of various rating scales. The two most commonly-used are
the Hamilton Depression Rating Scale (HDRS or HAMD) and Montgomery Åsberg
Depression Rating Scale (MADRS) [1].
Primarily in earlier studies of antidepressants the outcome measures have been the
change in the degree of severity, or alternatively how large a proportion of the patients
who after a certain period of time have achieved a 50-percent decrease in the symptom
29
The review of anTiDePressanTs
points. The share of patients who have achieved a 50 percent decrease is a commonly­
used measure of how many patients have responded to a treatment. In later studies
an outcome measure used more and more often is the number of patients who have
managed to land below a certain threshold value through treatment. The threshold
value is set so that the patient can be said to have recovered (be in remission). For the
HAMD-scale this is currently set to less than or equal to 7 points and for the MADRS­
scale to less than or equal to 12 points [27], [28] . Other values have also been used
previously.
The reasons are many for the increase in using the outcome measure share of recove­
red patients [2]. The main reason is that through the treatment one both wishes to and
has the ability to help the patient recover. Also, it is of great value to the patient and
their close ones that the patient recovers. This results in an increase in quality of life
for both the patient and their loved ones, as well as being of great economical value
for the individual and society as a whole [29]. Furthermore, it has been noted that if
through treatment of a depression episode the patient can recover then the risk of the
patient relapsing or becoming ill again decreases [2].
4.2
Efficacy in the acute phase
Clinical studies aimed at establishing the efficacy of antidepressants are mainly studies
where the medicine being studied is the first-line treatment in one episode of the ill­
ness. For many patients depression is an illness they will be afflicted with once in their
lifetime. For even more it is however an illness with a number of episodes of illness
over a longer, in some cases lifelong, period of time. For the majority of patients taking
part in clinical studies it is therefore not the first instance of the illness being treated.
It is however the initial treatment and can due to this be said to be a treatment in the
acute phase of the illness.
4.3
Systematic literature reviews – depression
The SBU has carried out a comprehensive review of the literature in regard to the tre­
atment of depression-related illnesses. We have augmented the SBU’s review with four
other systematic overviews within the area of depression.
Treatment of depression-related conditions (Behandling av depressionssjukdomar) – sBu [1]
Management of depression in primary and secondary care – National Institute of Health and Clinical
Excellence – NICE [24]
Comparative effectiveness of second-generation antidepressants in the pharmacologic treatment of adult depres­
sion – prepared for Agency for Health Care Research and Quality by RTI International-University
of North Carolina Evidence-based Pratice Center – UNC [30], [31]
efficacy and safety of ssris and other newer antidepressants in depression for adults (Effekt og sik­
kerhet for SSRI og andre nyere antidepressive legemidler ved depresjon hos voksne) – norwegian Knowledge
centre for the health services (Nasjonalt kunnskapssenter for helsetjenesten) – norway [32]
30
The review of anTiDePressanTs
Furthermore, we have noted a systematic review from the Canadian Coordinating Of­
fice for Health Technology Assessment – CCOHTA [33].
4.3.1 The SBU – all antidepressants have an effect but it is difficult to
document differences in efficacy
The SBU has found that all antidepressants approved in Sweden have an effect which
differs from placebo. Despite the SBU review of almost 500 studies there are only a few
which show differences in efficacy for depression between the medicines. The SBU notes that the studies displaying a difference are handicapped somewhat due to methodi­
cal weaknesses. As an explanation for the fact that studies seldom show a difference in
efficacy between antidepressants despite their differing mode of action, the SBU points
out that studies are often too small for differences to be able to emerge. The decrease
in the degree of depression measured using a rating scale, normally HDRS or MADRS,
is the measure which is the basis for the SBU’s comparisons on efficacy. The results of
the comparisons are reported in a number of tables which we have tried to summarise
in one common table (Table 4).
The figures in the table do not indicate the number of studies but the number of com­
parisons made for each pharmaceutical substance respectively. In the left section of
the table the number of comparisons which have not shown any statistically signifi­
cant difference in efficacy have been placed, in the centre columns the comparisons to
placebo are shown. And on the right-hand side of the table we show the comparisons
where differences in efficacy have been found.
For fluoxetine there are as an example seven comparisons to imipramine and five to
clomipramine which have not shown any difference in efficacy. In 69 of a total of 73
comparisons between fluoxetine and an active substance (another antidepressant) it
has not been possible to establish a statistically significant difference in effect. The re­
sults from the other four comparisons are as following: worse than imipramine, better
than imipramine, worse than paroxetine, worse than venlafaxine. In the comparisons
to placebo in five cases it was not possible to establish any statistically significant diffe­
rence, and in 13 cases the effect gained from fluoxetine showed a statistically signifi­
cant better effect than placebo.
If you look at the results as a whole then for the most part (414 comparisons) the
conclusion has been reached that there is no significant difference in effect for two
compared substances, and in only a smaller number of cases (29 comparisons) is a
statistically significant difference was found.
4.3.2 NICE – difficult to find differences in efficacy between the medicines
Within the framework of their remit to produce treatment guidelines for the care of
people suffering from depression in both primary and specialist care in Great Britain
31
Table 4. Number and results of the comparisons reported between two antidepressants, or an antidepressant and placebo from the studies included in the SBU review
4
13 1
3
1
4
2
2
1
5
5
2
7
6
1
8
2
1
6
2
2
2
1
1
2
2
1
3
1
1
1
6
2
2
1
1
1
1
2
1
2
3
5
8
1
2
1
5
2
2
3
1
1
1
3
6
1
1
1
2
2
1
1
1
1
symbol key for results of comparison
= no statistically significant difference found in comparison with comparator substance
> the substance has a statistically significant better result than the comparator substance
< the substance has a statistically significant worse result than the comparator substance
>< contradictory results in the study
2
1
1
1
< n06aX16 venlafaxin
> n06ag02 moclobemide
< n06ag02 moclobemide
>< n06ag02 moclobemide
> n06aB08 fluvoxamine
< n06aB06 sertraline
> n06aB05 paroxetine
< n06aB05 paroxetine
> n06aB04 citalopram
> n06aB03 fluoxetine
> n06aa21 maprotiline
1
2
7
> n06aa09 amitr yptiline
1
4
23 30
4 2
1
2 22
1 4
5
5 13
1 5
3 10
3 5
4 5
3
4 4
2 6
1 5
2 9
>< n06aa09 amitr yptiline
2
< n06aa09 amitr yptiline
1
< n06aa07 lofepramine
7
< n06aa04 klomipramine
= n06aX16 venlafaxine
9
> n06aa02 imipramine
= n06aX11 mirtazapine
5
< n06aa02 imipramine
= n06aX03 mianserine
2
= placebo
= n06ag02 moclobemide
4
> placebo
= n06aB08 fluvoxamine
5
1
= n06aX18 reboxetine
= n06aB06 sertraline
5
1
= n06aB04 citalopram
= n06aB05 paroxetine
1
1
3
17
1 15 1
3
1 13 1
5 1
3
= n06aB03 fluoxetine
= n06aa10 nortriptyline
= n06aa07 lofepramine
= n06aa09 amitr yptiline
5
5
= n06aa21 maprotiline
n06aa02 imipramine
n06aa04 clomipramine 6
n06aa06 trimipramine 5
n06aa09 amitriptyline 7
n06aa10 nortriptyline 1
n06aa21 maprotiline 19 5
n06aB03 fluoxetine
7 5
n06aB04 citaloprame
1
n06aB05 paroxetine
4 4
n06aB06 sertraline
3 2
n06aB08 fluvoxamine
6 5
n06aB10 escitaloprame
n06ag02 moclobemide 4 9
n06aX03 mianserine
6 8
1
n06aX11 mirtazapine
n06aX16 venlafaxine
2 2
= n06aa06 trimipramine
= n06aa04 clomipramine
= n06aa02 imipramine
results from the comparison
1
1
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
The review of anTiDePressanTs
a review of studies on efficacy and tolerance for antidepressants was carried out. Over
300 studies were the subject of the review. The number of studies laying a foundation
for the evaluation of individual medicines or classes of medicines is limited in some
cases.
The measures of effect which NICE primarily uses is the share of patients who achieve
a 50-percent decrease of the symptoms as rated using HRSD or alternatively MADRS,
or how large a decrease has been achieved at the final point in time when the patients
were followed up.
In some cases the outcome is also reported in regard to the proportion of patients who
recovered.
A common feature is that NICE finds few differences between pharmaceutical groups
and individual medicines. It may also be noted that NICE in its comparisons differen­
tiates between statistically significant differences and clinically significant differences.
That a medicine in comparison with others has a statistically higher share of patients
exhibiting a change in symptoms of 50 percent or more does not as a result mean that
NICE considers this to be sufficient to put this particular medicine before others in
clinical practice. Neither does a statistically significant higher proportion of patients
in remission constitute the same, which for us is not clearly the correct approach.
NICE finds that amitriptyline is as effective as other antidepressants, regardless if one
looks at other antidepressants as a whole, or class by class. TCAs as a group have an
effect which is comparable to the effect gained from antidepressants of a different
type. However, they are not tolerated as well and this is true in particular for patients
in out patient care. SSRIs, excluding escitalopram, do not differ in terms of effect from
TCAs or MAO inhibitors.
In comparison with the third generation of antidepressants, which in the review car­
ried out by NICE means mirtazapine, venlafaxine and reboxetine, there is a statistically
significant but not clinically important difference which speaks to the advantage of
these three medicines. For escitalopram NICE has carried out a separate analysis which
shows that escitalopram in regard to alleviation of symptoms has an advantage when
compared to other antidepressants, but however not for patients who have recovered.
According to NICE there is no clinically significant difference in effect between moclo­
bemide and TCAs or SSRIs, nor between mirtazapine and TCAs, SSRIs or venlafaxine.
They note a statistical but not clinically significant difference to mirtazapine’s advan­
tage in regard to patients who have recovered at the end of the period of treatment.
Reboxetine has an effect comparable to TCAs and SSRIs.
Venlafaxine may block the reuptake of both serotonin and norepinephrine. According
to NICE the effect is dependent on the dose and at the dose of 75 mg venlafaxine func­
33
The review of anTiDePressanTs
tions as an SSRI, the twinned effect is achieved at doses of 150 mg and higher. NICE
has due to this carried out different analyses for different doses of venlafaxine. They
do not find any differences in effect between venlafaxine and other antidepressants,
but do note that venlafaxine in some cases decreases the symptom points more than
the comparator medicines. NICE however believes that the difference is not clinically
significant. They believe this is supported by the fact that patients using venlafaxine,
in particular in lower doses, are more prone to stopping treatment due to the side­
effects than patients treated with fluoxetine.
4.3.3 University of North Carolina (UNC) – escitalopram better than
citalopram, sertraline and venlafaxine better than fluoxetine
The UNC has examined if there are any differences in effect between the second gene­
ration of antidepressants. In their case this means the following substances: fluoxetine,
citalopram, paroxetine, sertraline, escitalopram, moclobemide, mianserine, mirtaze­
pine, venlafaxine and reboxetine.
The UNC collected information from the literature on to which degree patients impro­
ved from the treatment and to what extent they responded to treatment. The degree of
improvement was measured as a change in the rating scales used, normally HDRS or
MADRS.
The UNC could use data from direct head-to-head studies as a basis for meta-analyses
of four pharmaceutical comparisons. According to these the treatment effect for esci­
talopram was better than for citalopram in regard to how much the patients improved
and how large a proportion of the patients responded to treatment. There were no
differences in effect between fluoxetine and paroxetine. Sertraline gave a better effect
than fluoxetine – in regard to the change in scale points the difference was numerically
larger, in regard to the proportion of patients responding to the treatment the dif­
ference was statistically significant. Venlafaxine compared to fluoxetine gave a nume­
rically larger effect in regard to improvement and a significantly larger proportion of
patients responding to the treatment.
The report also details indirect comparisons. The exactness in the estimations is
sub-par in some cases, but the results coincide with the results from the head-to-head
studies. The results point to no or small differences in effect between the medicines
compared.
The UNC’s review showed that of the patients who had been treated using second
generation antidepressants for a period of 6–12 weeks, 38 percent did not respond to
treatment and 54 percent had not recovered. The UNC could not find factors which
in a reliable way could predict how individual patients were going to respond to a
treatment. The base documentation for the UNC’s review were 293 of an initial 2000
34
The review of anTiDePressanTs
identified studies. Seventy two direct head-to-head studies served as a foundation for
35 of a total of 66 possible comparisons between 12 second generation antidepres­
sants included in the report (Table 5). Five studies concerned comparisons between
non-SSRI medicines. Only one comparison was part of more than one study. Many
studies were too small to satisfactorily determine statistically significant or clinically
important differences.
escitalopram
1
4
mirtazapine
3
1
2
1
venlafaxine
9
1
2
2
1
1
duloxetine
2
reboxetine
1
venlafaxine
4
1
mitazapine
1
2
mianserine
8
fluvoxamine
moklobemide
sertraline
escitalopram
10
fluvoxamine
1
paroxetine
sertraline
paroxetine
citaloprame
fluoxetine
citaloprame
Table 5. The number of head-to-head studies used as a basis for the UNC’s review
fluoxetine
moclobemide
mianserine
2
1
reboxetine
duloxetine
4.3.4 Canadian Coordinating Office for Health Technology Assessment
(CCOHTA) – comprehensive review, no differences between substance
groups
The review carried out by CCOHTA was according to the SBU one of the most com­
prehensive and thorough reviews ever carried out at the time of the SBU’s review. The
review is built on studies published up until April 1996, but only in regard to substan­
ces which were then available on the Canadian market. For that reason citalopram was
not included in the medicines which were studied.
Based on 117 studies, of which 96 concern comparisons between SSRIs and TCAs and
published up until April 1996, the CCOHTA do not find any statistically significant
difference in effect between the substance groups. They have done separate analyses
based on different doses for both SSRIs and TCAs without finding any differences.
Neither have they found any difference in effect when they differentiated between use
in out-patient and in-hospital care patients.
35
The review of anTiDePressanTs
4.3.5 Norway – seldom differences between medicines,
escitalopram one of the exceptions
The Norwegian Knowledge Centre for the Health Services, as commissioned by the
State medical products agency (Statens legemiddelverk), has gone through the clinical
documentation for the SSRI medicines and other new antidepressants. The docu­
mentation mostly didn’t show any significant differences between the medicines. The
differences as noted were the following:
 escitalopram showed a significantly better effect (response and remission) than
citalopram
 escitalopram showed a significantly lower proportion of treatments being stopped
than citalopram
 citalopram showed a significantly lower frequency of side-effects than venlafaxine
 mirtazapine showed a significantly better effect (remission) than paroxetine
 fluoxetine showed a significantly lower proportion of treatments being stopped due
to side-effects than venlafaxine.
The knowledge centre used very strict selection criteria for the publications which were
to be part of the base material. From a list of over 5 000 articles 226 were examined fully. The base material used for the report was 23 articles. All of them were double-blind
studies where the medicines in question were compared to the others in the treatment
of depression in adults. The articles detail 12 different comparisons of pharmaceutical
substances out of a total of 66 possible ones (Table 6). For most of the comparisons
the base material was considered to be of sub-par quality.
fluoxetine
citaloprame
1
paroxetine
3
sertraline
fluvoxamine
1
escitalopram
moclobemide
4
1
1
mianserine
mirtazapine
venlafaxine
reboxetine
duloxetine
36
1
5
1
1
1
1
duloxetine
reboxetine
venlafaxine
mitazapine
mianserine
moclobemide
escitalopram
fluvoxamine
sertraline
paroxetine
fluoxetine
citaloprame
Table 6. Number of direct head-to-head studies in the Norwegian Knowledge Centre’s review
The review of anTiDePressanTs
4.4
Effect after change of medicine
The medicines which are recommended as a first-line treatment and are used initially
to treat patients with depression are the SSRI medicines [23] . In cases where treatment
with the first medicine does not lead to the desired result one often changes to another
medicine. The effect gained from switching medicine has been the subject of a number
of studies, as well as for systematic reviews where the purpose has been to establish
what knowledge can be used as a basis for how to carry out a change of pharma­
ceutical.
4.4.1
Review finds few well-executed studies and no definite answers
From the evidence there doesn’t seem to be many studies which in a satisfactory man­
ner account for the effect from changing pharmaceutical when the patient has not
responded to the initial treatment [34]. Only eight randomized clinical studies and 23
open studies fulfil the criteria for inclusion in the base material for a newly published
literature review by Ruhe and colleagues. Already in their preparations for the study
the authors have stated that a limitation to solely randomized clinical trials would give
a far too small base documentation, thereby opening up for the subsequent inclusion
of open studies. The review was focused on patients who had not responded to SSRI
medicines. The extent and the reasons for the patients not responding to the treat­
ment varied between the different study populations. The studies could include both
patients for whom the medicine did not have any effect as well as patients who did
not tolerate the medicine, for example due to side-effects. Also there were differences
between the studies in how response was defined and remission. Ruhe and colleagues
identified seven open and three blind studies where patients after a first failure with
one SSRI medicine were treated with another SSRI. The number of patients respon­
ding to the treatment with the second SSRI medicine varied widely from 27–71 percent
of the patients. Also in the six studies where a change was effected to a TCA medicine
or mianserine the variation in the number of patients responding to the treatment was
large (17–49 percent).
The authors identified 13 studies where the patients switched to treatment with mir­
tazapine, nefazodon or venlafaxine. The quality of the studies varied as did the results.
For patients with a less delineated resistance to therapy the proportion of patients who
responded to the treatment was 28–50 percent, for patients with a clear resistance to
therapy it was lower.
Four studies in the review report the result for patients who following treatment with
fluoxetine switch to bupropion (3 studies) or reboxetine (1 study). The proportion of
patients responding to treatment was 26–35 percent for bupropion, in the study for
reboxetine the share was 45 percent. The conclusions reached by the authors were that
the level of knowledge we have today cannot be used as a basis for a clear treatment
strategy for patients who do not respond to a first treatment using an SSRI medicine.
37
The review of anTiDePressanTs
It is only possible to give general recommendations to the effect that switching to all
classes of antidepressants studied is possible and that one should choose between
them on the same basis one chooses medicines for the initial treatment.
The authors point out that a number of limitations exist for the studies included in
the review. For example there were only a few clear criteria for which patients should
be included in the study and in many cases the patients did not commence the second
treatment immediately following halting the first treatment. The need for new and
well-executed studies is huge. The authors state that the study they have carried out is
encumbered with many limitations. A review similar to the one carried out cannot cor­
rect for the dilemma posed by the lack of high quality studies.
The conclusion in the overview is that the result from the STAR*D study (see below) to
a great degree increases the extent of and quality in the knowledge base we have today.
But today’s knowledge level, which was to be used as a basis for clear recommendations
for choosing pharmaceuticals, does not clarify any differences in effect between the dif­
ferent classes of antidepressants. All classes of medicines are therefore possible alterna­
tives when the patient in the initial treatment has not been helped by an SSRI medicine.
4.4.2
What does Star*D show?
Star*D, Sequenced treatment alternatives to relieve depression, is a large study sponsored by
the National Institute of Mental Health in USA. It intended to show treatment results
which were representative for patients being taken care of in regular day-patient care.
The study consists of a series of randomized clinical trials where sequenced treatment
steps were examined for patients who had not improved or not been able to tolerate
their previous treatment.
A simple methodology was developed and used within the framework for the study in
order to improve the quality of the treatment. The methodology was designed to be
used independently of if the treatment was given in regular daily practice or within
the framework of a clinical trial. The method was based on measuring the patient’s
symptom’s and side-effects for each visit with the aid of simple instruments (paper and
pen). The patient’s treatment was tailored based on the results of these measurements
in accordance with established treatment guidelines.
It is quite likely that the use of this methodology explains the achievement of similar tre­
atment results for patients treated in general out-patient care and mental health clinics.
In the first stage of treatment all patients were treated (n = 2 876) with citalopram. The
proportion of patients who recovered (achieved remission) was 28 percent (HAMD).
Patients who did not respond to treatment could choose between additional treatment
or changing medicine. In the latter case the patients were randomised to treatment
38
The review of anTiDePressanTs
with bupropion (n = 239), sertraline (n = 238) or venlafaxine XR (n = 250). The share of
recovered patients now reached 21.3, 17.6 and 24.8 percent respectively (HAMD). The
differences between the results are not statistically significant.
In the third treatment stage patients not responding to the previous treatment stage
could choose between additional treatment or changing medicine. In the latter case
patients were randomized to treatment with mirtazapine or nortriptyline. The share of
recovered patients here was 12.3 and 19.8 percent respectively (HAMD).
In the fourth treatment stage patients were randomised to treatment with tranylcypro­
mine or venlafaxine in combination with mirtazapine. The share of recovered patients
was 6.9 and 13.7 percent respectively (HAMD). The differences between the treatment
results achieved in these two treatment stages are not either statistically significant. As a
whole for the entire study 67 percent of the patients who had remained in the study had
recovered. However, the number of patients who left the study was considerable. Patients
who had reached the treatment objectives could take part in a 12-month long follow-up
study. For these patients the following relapse frequencies were noted: 40.1 percent, 55.3
percent, 64.6 percent and 71.1 percent for those who reached the treatment objectives
(remission) in stages 1, 2, 3 and 4. In all treatment stages the patients who had achieved
the treatment objectives ran a lower risk of relapse than those who had not done so.
4.5
Effect from continued treatment and long-term treatment
The SBU deals with the effects of long-term treatment in its review of antidepressants.
From a practical and research-oriented point of view the SBU finds it expedient to
divide the treatment of depression in three phases:



acute treatment, meaning treatment from the point it is commenced until the pa­
tient is free from symptoms or experiences a strong decrease in symptoms
continued treatment, treatment given after successful acute treatment during the
period the underlying depression is believed to last
prophylactic treatment, treatment aimed at preventing new episodes of depression
Others have questioned whether or not this division is clinically relevant as it is diffi­
cult to know if it is the same depression episode returning, or if it is a new episode [35].
The SBU’s base material for the review of continued treatment and prophylactic
treatment is made up of studies published up until May 2003. In total more than 1000
studies were found of which 123 were included in the review. The SBU finds that there
is evidence that imipramine, amitriptyline, fluoxetine, sertraline, paroxetine, citalo­
pram and mirtazapine have an effect during continued treatment. There is also some
evidence that nortriptyline, maprotiline, fluvoxamine and reboxetine have an effect.
However, according to the SBU there is a lack of evidence from published studies that
39
The review of anTiDePressanTs
clomipramine, trimipramine, venlafaxine, mianserine and moclobemide have an ef­
fect. In regard to the effect of long-term treatment the SBU found that if one accepts
the assumption that all antidepressants have a preventive effect, then a meta-analysis
of good quality shows that treatment with an active substance decreases the risk of
relapse from 41 to 18 percent.
The SBU points out that the interpretation of many studies is complicated due to vary­
ing methodological differences or weaknesses in their execution. The clinically crucial
question “How should one treat patients who suffer a relapse while under treatment?”
almost completely lacks answers in the scientific literature according to the SBU.
In an overview from BMJ Clinical Evidence [36] the authors draw the conclusion that
it is beneficial to continue treatment with antidepressants for one to three years after
the patient has recovered, as this prevents relapse. In the overview it is stated that the
effect of treatment through antidepressants was better than for placebo, independent
of the risk of relapse and how long the treatment had been ongoing before recovery.
According to Canadian guidelines for the treatment of depression all patients shall be
treated for at least nine months – eight to twelve weeks acute treatment followed by six
months maintenance treatment. Patients with an elevated risk of relapse shall in some
cases be treated even longer [35] .
In the overview from BMJ Clinical Evidence [36] the authors note that no individual
medicine is better than others, but also notes that in a study over two years sertraline
did not display a better effect at prevention of relapse than placebo [37]. The patients
in this study were on average 76 years old.
In a leader article in the New England Journal of Medicine [38] relapse prevention is
discussed amongst older depressed patients. There one conclusion reached is that the
effect probably is not influenced by which antidepressant used and that depression is
probably treated for too short a time.
Paroxetine in combination with monthly contact with a caregiver (either for psychoth­
erapy or regular clinical care) was significantly better than placebo in combination
with monthly contact in a study over two years. Like in the sertraline study the patients
in the study were older, an average of around 77 years [39].
In a later study patients were treated with either venlafaxine or fluoxetine during an
acute treatment covering ten weeks and following that six weeks of maintenance treat­
ment without controlling for placebo [37]. Patients who had recovered continued then
on two twelve-month maintenance treatment phases. Here patients were randomly
divided into those who had had venlafaxine in the two first phases to venlafaxine or
40
The review of anTiDePressanTs
placebo, while fluoxetine patients continued with fluoxetine. After the two first phases
(ten weeks + six months) venlafaxine was not better than fluoxetine. During the two
year maintenance treatment venlafaxine had a significantly better effect than placebo
in regard to relapse (approx. 28 percent relapse compared to approx. 47 percent). A
comparison to fluoxetine was not carried out. The patients in this study were 40 years
old on average.
In summary there is good evidence that a number of antidepressants are effective in
maintenance treatment up to one year, while venlafaxine is the only medicine which
has shown a preventative effect during a period of two years. In the light of the body of
literature we find it however probable that also other, but not necessarily all, antide­
pressants have a similar effect. This due to venlafaxine not being better than fluoxetine
at the last measuring point in the study and that data for fluoxetine after two years
were not reported.
4.6
Side-effect profiles
The MPA says in its treatment guidelines that an individualisation of pharmaceutical
choice can be done based on which side-effects the patient can be expected to be more
or less sensitive to. The SBU states in its report that there are differences in the side­
effects profile between different groups of antidepressants, and also between different
substances within a group, and that these differences are less difficult to find than
differences in efficacy. At the same time the MPA points out that the difficulties in
comparing different medicines, partly due to the lack of direct head-to-head studies
and the variance in the way of measuring the prevalence of side effects. The SBU notes
the differences in side-effects profile between SSRIs and TCAs.
The former result more often in nausea, diarrhoea, anxiety, agitation and difficulty
sleeping, the latter to dryness in the mouth, blurred vision, sweating and dizziness. In
a number of comparisons nausea appears to be more common for venlafaxine than
for SSRIs. Venlafaxine can also result in elevated blood pressure. The SBU states that
increased appetite and an increase in weight is more common for patients who take
mirtazapine.The SSRIs to a large extent cause sexual problems. Any larger differences
between the SSRI substances are not to be expected.
The compilation of the most common side-effects for the various pharmaceutical clas­
ses made by the SBU is listed in Table 7.
In the report from the University of North Carolina (UNC) mentioned earlier it was
stated that there is a clear variation on the question of how studies establish the preva­
lence of side effects. Few studies use objective measuring instruments, and neither do
they use the terminology set by the WHO. Most studies combine side-effects repor­
41
The review of anTiDePressanTs
ted by patients with those discovered when the doctor examined the patient. Nausea,
headaches, diarrhoea, tiredness, dizziness, sweating, sexual dysfunction, tremors, dry
mouth and increased weight were side effects often reported. The UNC overall sees no
difference between medicines in regard to which types of side-effects they give rise to,
however they do see a difference in regard to the frequency with which individual side
effects occur. Venlafaxine had in comparison to the SSRIs as a group a higher frequen­
cy of nausea and vomiting. Patients also halted treatment at a higher frequency due to
side effects.
Table 7. Common side-effects for the different groups or types of antidepressants
Side effects
Blurry vision
Diarrhoea
constipation
Palpitations
headaches
nausea
Dry mouth
sexual dysfunction
Difficulty sleeping
Tiredness/doziness
Dizziness
anxiety/agitation
anxiety/nervousness
increased appetite/increase in weight
increased sweating
TCAs
SSRIs
moclo­
bemide
mirtazapine/
mianserine
venla­
faxine
reboxe­
tine
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
In comparison to a number of other antidepressants Sertraline had a higher frequency
of episodes of diarrhoea. Mirtazapine gave a generally higher increase in weight than
other medicines. The same was true for paroxetine in comparison to fluoxetine and
sertraline.
The UNC also notes that antidepressants were often associated with symptoms such as
headaches, dizziness and nausea. For paroxetine and venlafaxine these problems were
comparatively more common while they were least common for fluoxetine. Sexual side
effects were comparatively more common for citalopram, paroxetine and venlafaxine
but less common for mirtazapine.
4.7
In summary – not worse than, but better?
On the whole we can say that antidepressants have an effect on the treatment of de­
pression. There are however a number of studies which have not been able to show that
antidepressants have a better effect than placebo [40]. There are also a large number
42
The review of anTiDePressanTs
of studies which do not show any differences in effect between different antidepres­
sants. If one manages to find studies showing that a certain antidepressant has a better
effect than a comparator medicine then most often one finds studies pointing to the
opposite conclusion. The measure of effect normally used in studies is the decrease
in the degree of depression as measured using one of the rating scales for depression.
The treatment time for which the patients have been followed in the study, not least
in older studies, is relatively short, six weeks or less. As a basis for making reimburse­
ment decisions it is desirable to have studies with longer follow-up time, and where the
medicines are compared in regard to the number of patients who have recovered.
4.8
Own meta-analysis of effect
After having reviewed the existing systematic overviews, and other literature which we have detailed, we believed there was reason to further analyse the data available. It is incumbent on us in our role as decision maker to take all available pharmaceutical treatments into account. Therefore we need to compare them to each other in regard to one or more relevant treatment outcomes. We lacked comprehensive systematic reviews and meta-analysis on the proportion of patients in remission, as the measure of treatment effect, and therefore decided to carry out one ourselves.
The review of the literature which should lay the foundation for the meta-analysis should, in respect to our purposes –

be as comprehensive as possible in regard to which medicines should be included, as we are to evaluate the reimbursement status of all medicines within the group

be so recent that also the latest pharmaceutical newcomers are adequately studied

be as free from publication and selective bias as possible

report the proportion of recovered patients.
There are a number of reviews for individual medicines but they do not meet the crite­
ria above and are thereby not suitable for our purposes. The only studies which fulfil
our criteria appear to be those conducted by Machado and colleagues [41] and the
systematic overview carried out by the Norwegian Knowledge centre of healthcare [32].
There are however obvious weaknesses in the method used in the statistic analysis in
Machado’s study, in particular that the randomization has been broken. Neither is it
very recent, take account of unpublished material nor cover all medicines which are
the subject of our review. The Norwegian study does not include unpublished material
and has, in our opinion, been too strict when excluding studies due to patients leaving
the study prematurely. Altogether this has led to the decision to carry out an own sys­
tematic review and meta-analysis of the literature. Furthermore, we wanted to use
a statistical method which fully utilizes available data.
43
The review of anTiDePressanTs
The primary clinical outcome in the meta-analysis is the proportion of patients who
recover (reach remission). As shown below some different definitions of remission are
used in the studies included.
The criteria for including or excluding a study in the initial systematic review were the
following:

Only direct head-to-head randomised studies (studies with active controls, but a
placebo arm could also exist)

moderate to severe major depressive disorder MADRS 18 or HAMD 15

no psychiatric concurrent illnesses (primarily psychoses and bipolar condition)

patients 18 years or older

study length more than six weeks

therapeutic doses of the medicines studied

the patients were not permitted take medicines which could mar the interpretation
of study data (eg lithium)

the study had to report remission as an HAMD point <= 7 or 8 or as an MADRS
point <=12.
In the final meta-analysis more restricted criteria were used, which is discussed further
in section 3.8.2.
One assessor registered data from the studies and the other checked each entry.
Any disagreements were resolved through consensus decisions.
We extracted information from the studies on the studied treatment, the treatment
compared to, dosage of medicines, who financed the study, the length of the study,
country and patient age, weight, sex, HAMD and MADRS points for the patients when
they joined the study and if the treatment took place in in-patient or out-patient care.
The data on effect and tolerance which we registered was the definition of remission,
number of patients in remission, how many completed the study, left the study due to
side-effects and left the study due to lack of effect.
Besides our own searches we have requested companies send us unpublished material
in particular. We have also gone through the websites where companies can post
unpublished material and the websites www.clinicalstudyresults.org and
www.centerwatch.com/patient/trials.html.
We did not identify many entirely unpublished studies, but we identified a large num­
ber of studies which had been published as summaries for presentations at scientific
congresses. The companies sent us a number of studies and we included 34 of them in
our analysis.
44
The review of anTiDePressanTs
4.8.1
Methods for meta-analysis
In a meta-analysis statistical methods are used to pool and weigh up the results of a
number of different studies. With the addition of data the effect of coincidence on the
result is decreased and such measures as the effect of various treatments can be estima­
ted with greater precision. The risk of drawing erroneous conclusions on the effects of
treatments is thereby decreased.
It is important that a meta-analysis is based on a systematic review of the literature
with objective selection criteria to avoid a bias in the selection of studies. For this
reason the literature searches, in so far as it is possible, should include unpublished
studies and studies where only the abstract from scientific conferences is available.
There are various statistic models to use in a meta-analysis. When a fixed effect model
is used the studies included are treated as if they were the entire population of studies
and that the estimation of effect is homogeneous, as if coming from the one study. But
even if the treatments are similar there may be differences in for example the selection
of patients, severity of the disease, measurement methods, treatment compliance and
other concurrent treatment. Using a random effects model is a way to try to handle the
heterogeneity in such data.
In a random effects model the selection of studies is assumed to be a selection from
the population of studies. The argument can be made however that it is more suitable
to carry out a number of different fixed effect analyses than to pool studies one knows
are heterogeneous.
The statistical methods for weighing up the results from different studies focus mainly
on comparing two treatments. That is to say in the cases where medicine A has been
compared with medicine B in a number of studies, or alternatively where medicine A
has been compared to placebo in a number of studies.
However, such methods of analysis are limited when more than two treatments are
being compared and when two treatments have not been part of direct head-to-head
studies.
Mixed treatment comparison (MTC) is an expansion of the meta-analysis making it
possible to compare two treatments and treatments which have not been part of direct
head-to-head studies. Assume that we are interested in comparing three treatments: A,
B and C. Also assume there are studies which compare treatments A and B and thereby
give a direct comparator treatment effect between the treatments A and B (we will call
it TAB). If there are studies which compare the treatments A and C as well as B and C
then one can also make an indirect comparison between A and B as TBC – TAC = TAB.
This mixed treatment comparison in this fashion makes it possible to assign values to
studies which compare AB, AC, BC and ABC (studies with three arms). So it is possible
45
The review of anTiDePressanTs
to estimate each treatment effect in pairs through considering both the direct and
indirect comparisons without breaking the randomisation.
This approach therefore makes it possible to estimate the effect of all relevant treat­
ments and to combine all information from relevant studies [42], [43].
4.8.2
Results
The results from the literature search are accounted for in section 10.1, where there
is a summary of all available studies. In total 85 studies were included with a total of
almost 20 000 patients in the analysis. The details in the literature search are accoun­
ted for in an appendix available for download at www.tlv.se/depression. It can also be
ordered from the TLV.
The structure of the evidence, that is to say the pharmaceutical comparisons included
in the 85 studies and which formed a basis for our meta-analysis, is indicated in Table
8. Note that some of the medicines included in the comparison are not sold in Sweden.
There is however a purpose in having these studies as long as one of the treatment alter­
natives is interesting as it can then be used for an indirect comparison.
5
1
1
mirtazapine
escitalopram
5
1
1
1
1
1
1
1
1
1
1
1
4
3
1
duloxetine
imipramine
citalolopram
sertraline
amitriptyline
fluvoxamine
clomipramine
dothiepine
lofepramine
milniciprane
maprotiline
nortriptyline
reboxetine
46
1
1
1
reboxetine
2
nortriptyline
3
maprotiline
1
milniciprane
1
1
lofepramine
1
2
dothiepine
citalopram
1
1
clomipramin
imipramine
4
7
fluvoxamine
duloxetine
4
amitriptyline
escitalopram
20
sertraline
mirtazapine
paroxetine
paroxetine
fluoxetine
venlafaxine
venlafaxine
fluoxetine
Table 8. The structure of the available evidence in matrix format
1
1
1
The review of anTiDePressanTs
The figures indicate the number of studies for the comparison in question. Empty cells
indicate that no direct head-to-head studies have been found in the literature.
The results for multiple treatment comparison is accounted for in Table 9. The table,
together with the results detailed in Appendix 10.2, is the main result of our meta-ana­
lysis. Note that it is the probability for recovery, that is remission, which is detailed here.
The meta-analysis estimates the odds ratios but they have been transformed into proba­
bilities. Fluoxetine is the common comparison alternative, as it appears in most studies.
Table 9. Proportion of recovered patients estimated when data (85 studies) from all studies
are compiled in a multiple treatment comparison.
Probability
Pharmaceutical
Substance Recovery
Confidence interval
fluoxetine
venlafaxine
paroxetine
mirtazapine
escitalopram
duloxetine
imipramine
citalolopram
sertraline
amitriptyline
fluvoxamine
klomipramine
dothiepine
lofepramine
milniciprane
maprotiline
nortriptyline
reboxetine
0.4021
0.4568
0.4270
0.4508
0.4756
0.4499
0.4237
0.4050
0.4302
0.3840
0.2677
0.6673
0.4187
0.4245
0.3888
0.3976
0.4691
0.4316
0.3729
0.4189
0.3834
0.3949
0.4264
0.3988
0.3470
0.3507
0.3820
0.3080
0.1776
0.4900
0.2642
0.2692
0.2682
0.1967
0.2976
0.2771
0.4301
0.4931
0.4707
0.5080
0.5251
0.5009
0.4986
0.4614
0.4772
0.4638
0.3706
0.8166
0.5862
0.5949
0.5198
0.6274
0.6446
0.5957
Odds ratios for all comparisons where there are two or more studies and where the
data has been pooled in accordance with the classic method are detailed in Table 10.
As a comparison the odds ratios from the analysis where we exploited the entire set of
data simultaneously are shown. Two things are especially notable in Table 10. Firstly,
that both methods in general concur very well. Secondly, that the precision increases
the more information we add, that is to say the confidence interval shrinks when we
exploit all data.
Escitalopram is the substance whose effect has the highest point values of all, excep­
ting clomipramine. The results obtained for clomipramine is however uncertain. The
model values odds ratios which are then related to a common estimated base line.
Clomipramine occurs only in a small study and has a very high odds ratio in this study
compared to paroxetine. In the rest of the data material paroxetine has a good effect,
47
The review of anTiDePressanTs
but in this study the effect is under-par. The results obtained for clomipramine are
therefore a consequence of the unexpectedly bad results for paroxetine in the study
where the substances are compared. The study should possibly be excluded. If it is
excluded then the results are not changed for the other treatments.
Table 10. Comparison (odd ratios) between the classic method and multiple treatment comparison
Classic method only with paired data
Multiple treatment comparison
utilising all data
Comparison
Odds ratios Confidence interval
Odds ratios
Confidence interval
venlafaxine–fluoxetine
paroxetine–fluoxetine
mirtazapine–fluoxetine
paroxetine–venlafaxine
sertraline–venlafaxine
duloxetine–paroxetine
duloxetine–escitalopram
mirtazapine–paroxetine
citalopram–escitalopram
escitalopram–venlafaxine
escitalopram–paroxetine
1.241
1.142
1.067
0.827
0.827
1.072
0.951
1.352
0.659
1.098
1.152
1.253
1.111
1.227
0.888
0.900
1.101
0.905
1.108
0.754
1.082
1.222
1.122
0.947
0.985
0.772
0.765
0.933
0.765
0.894
0.618
0.917
1.039
1.092
0.780
0.752
0.652
0.628
0.831
0.750
0.977
0.509
0.748
0.586
1.410
1.673
1.514
1.049
1.089
1.383
1.205
1.871
0.854
1.610
2.263
1.390
1.287
1.508
1.017
1.051
1.286
1.059
1.361
0.918
1.266
1.429
Citalopram shows a considerably worse result in our analysis than in Machados. This
is because the studies where citalopram is included have in general had a high effect,
that is to say the comparator alternative has had a high effect, compared to the effect
shown in other studies. This illustrates the problem with breaking the randomisation
in the way in which Machado and colleagues do in their meta-analysis, as the point esti­
mation of citalopram’s effect then becomes too high. In our health economic model the
proportion of patients who recover after three months of treatment in out-patient care
is an important parameter. Following discussions with our expert group we make the
assumption that studies with a length of 8–16 weeks reasonably reflect the effect after
12 weeks, an assumption which can of course be questioned. There are however no stu­
dies with a length of 13–16 weeks in our selection. For this reason studies are presented
which are 8–12 weeks in length. As expected there is a tendency for the average effect
to be somewhat greater when six-week studies are excluded from the meta-analysis, as
more patients have managed to recover when the study has gone on for a longer time.
In addition to the analyses above we have carried out a number of analyses where the
inclusion criteria have been varied (see Table 26 in Appendix 10). We have carried out
separate analyses for patients in in-patient care or out-patient care when wanting to
measure the effect and tolerability of treatment with various antidepressants [44]. In
our analysis the differences were relatively small however.
48
The review of anTiDePressanTs
Table 11. Proportion of recovered patients for three different scenarios; all studies are included,
only studies of 8–12 weeks are included, only studies of 8–12 weeks of patients
in out-patient care are included
Pharmaceutical substance
All studies
8–12 weeks
8–12 weeks in
out-patient care
fluoxetine
venlafaxine
paroxetine
mirtazapine
escitalopram
duloxetine
imipramine
citalolopram
sertraline
amitriptyline
fluvoxamine
clomipramine
dothiepine
lofepramine
milniciprane
maprotiline
nortriptyline
reboxetine
0.4021
0.4568
0.4270
0.4508
0.4756
0.4499
0.4237
0.4050
0.4302
0.3840
0.2677
0.6673
0.4187
0.4245
0.3888
0.3976
0.4691
0.4316
0.42420
0.46530
0.43470
0.45730
0.49030
0.46650
0.41550
0.42440
0.43360
0.44170
0.43430
0.47800
0.44920
0.47660
0.50660
0.48070
0.43500
0.44500
0.45120
0.45780
0.37800
0.42170
0.46910
0.43000
0.48110
In four of the studies where duloxetine was included higher doses of duloxetine than
the indication allows have been used. For this reason we have carried out an analysis
where these studies have been excluded, but where the result has changed only margi­
nally. Further analyses we have carried out concern fixed versus flexible dosages and
studies carried out in Europe versus the rest of the world. We have also carried out an
analysis where we excluded studies where more than 30 percent of the patients leave
the clinical study early.
The results show that our meta-analysis is not especially sensitive to changes in the
inclusion and exclusion criteria, except when we only include studies with a flexible
dosage. Then the difference in effect between escitalopram, mirtazapine and venlafaxi­
ne only becomes apparent starting from the third decimal point.
4.8.3
Discontinuation in study due to side-effects
The number of people discontinuing a study is often used as a measure of tolerability.
It is however not an entirely uncontested measure as patients may stop treatment for
a variety of reasons. A better measure for discontinuation is how many leave a study
due to side-effects. This measure is not either entirely uncontested as the patients in a
study can continue with a treatment as long as they think it is tolerable.
49
The review of anTiDePressanTs
Discontinuation in a study due to side-effects has been studied by Song and colleagues
[45]. In an example they detail the results from a meta-analysis where 15.4 percent of
the patients on SSRIs and 18.8 percent of the patients on TCAs discontinued partici­
pation in the study due to side-effects. MacGillvray and colleagues [44] found lower fi­
gures in a later meta-analysis, 11.6 percent for SSRIs compared to 17 percent for TCAs.
We have carried out a meta-analysis of discontinuation in studies due to side-effects
(Table 12). The largest proportion of discontinuation is shown by clomipramine,
fluvoxamine, amitiptylin and duloxetine, all with point estimations over 14 percent.
The lowest proportion of discontinuation is shown by maprotiline and sertraline with
point estimations under 6 percent. The same applies for milnicipran and dothiepin,
but these medicines are not available on the Swedish market.
Table 12. Discontinuation in study due to side-effects
50
Pharmaceutical substance
Frequency discontinuation*
Confidence interval
flouxetine
0.0778
0.0633
0.0940
venlafaxine
0.1049
0.0807
0.1337
paroxetine
0.1046
0.0743
0.1404
mirtazapine
0.0779
0.0526
0.1104
escitalopram
0.0631
0.0426
0.0880
duloxetine
0.1493
0.1086
0.1985
imipramine
0.1063
0.0600
0.1705
citalopram
0.0904
0.0515
0.1455
sertraline
0.0550
0.0298
0.0898
amitriptyline
0.1446
0.0813
0.2333
fluvoxamine
0.1573
0.0676
0.3024
clomipramine
0.4013
0.1586
0.7126
dothiepine
0.0517
0.0089
0.1486
lofepraminee
0.0955
0.0158
0.2715
milniciprane
0.0467
0.0133
0.1101
maprotiline
0.0007
0.0000
0.0063
reboxetine
0.1063
0.0312
0.2396
The review of anTiDePressanTs
5 Treatment effect for anxiety
Medicines in the therapeutic group of antidepressants are also used for treatments of
various anxiety-related conditions. Also here we can use a literature review carried out
by the SBU. In addition to this we have used a compilation from BMJ Clinical Evidence
where the authors have gone through the body of evidence for the treatment of general
anxiety, panic syndrome, post-traumatic stress disorder (PTSD), compulsive-obsessive
syndrome and bulimia. We have also used results gathered from the Cochrane reviews.
Studying which effects the medicines have in the treatment of anxiety syndrome is a
comprehensive and complex process, both in terms of there being different types of
anxiety syndrome and that different outcome measures can be used in many cases for
the different types of syndrome. Furthermore, the definitions of anxiety syndrome
must be taken into account as they have changed over time, and in some cases there is
a difference between the two systems, ICD and DSM, used to diagnose and classify the
syndrome.
5.1
Panic syndrome
The SBU states in its review that antidepressants have been shown to have a beneficial
effect on panic symptoms, while the effect had on any phobic component of this is
uncertain. The risk of relapse after a completed course of treatment is great, according
to the SBU.
SSRI medicines have not been shown to be more effective nor tolerated better than
tricyclics but the latter are more associated with side-effects.
Treatment with medicines gives a quicker effect than psychotherapy but the latter has
been shown in studies to have a retained effect also after the treatment has been com­
pleted. The risk of relapse for patients who after a year finish their course of treatment
with antidepressants is larger than for those who continue with the treatment.
Patients suffering from panic syndrome often seek help in primary care. The SBU the­
refore sees the small numbers of studies carried out in primary care as problematic.
In BMJ Clinical Evidence the authors state that both cognitive behavioural therapy and
antidepressants from the SSRI and TCA classes are effective in their treatment of the
symptoms of panic syndrome [46]. There is a body of evidence indicating that cogni­
tive behavioural therapy has a better and more long-lasting effect than only treatment
with medicines. The treatment which, at least in the short-term, has been shown to
51
The review of anTiDePressanTs
have the best effect was cognitive behavioural therapy in combination with pharma­
ceuticals.
In a Cochrane overview by Furukawa and colleagues [47] the effect in the short and
long-term for a course of treatment where psychotherapy was combined with antide­
pressants for treatment of panic syndrome (with or without agoraphobia) in compari­
son with only one of these options is detailed. The overview included 21 studies with a
total of 1 709 patients.
The authors draw the conclusion that in the acute treatment phase a combined
treatment is more effective than only antidepressants. For long-term treatment psy­
chotherapy combined with antidepressants is more effective than only treatment with
pharmaceuticals. But treatment only with psychotherapy is as effective as combination
treatment. Treatment only with antidepressants is for this reason not recommended as
a first line alternative.
The authors believe the patient’s preference for combined treatment or only psychoth­
erapy is the most appropriate first line option for the treatment of panic syndrome.
However, it is a fact that there is a lack of psychotherapists. For this reason antidepres­
sants are in practice often the only available treatment alternative.
5.2
Specific phobias
There is a lack of evidence for a beneficial effect from medicines on specific phobias.
The SBU identified only a small number of studies regarding the treatment of these
and only one study on antidepressants.
5.3
Social phobias
Amongst the antidepressants in Sweden which are approved for the treatment of social
phobias it is mainly the substances paroxetine, sertraline, fluvoxamine, escitalopram
and venlafaxine which according to the SBU’s review have a well-documented effect.
An overview by Ipser and colleagues shows that medicines, primarily SSRIs and SNRIs,
are effective for the treatment of social phobias without any particular medicine ap­
pearing to be more effective than others [48]. The results from a number of studies of
maintenance treatment and relapse prevention unanimously support the evidence that
patients should be treated over a long period of time.
5.4
Compulsive-obsessive syndrome
According to the SBU’s review approximately 60 percent of the patients who had been
treated with clomipramine had benefited with an effect from the treatment. For the
SSRI medicines fluvoxamine, paroxetine, citalopram, sertraline and fluoxetine the
52
The review of anTiDePressanTs
proportion was 35–47 percent. The effect lingers as long as the treatment continues at
least for one year.
According to the authors in BMJ Clinical Evidence [49] the SSRI medicines citalopram,
fluoxetine, fluvoxamine and paroxetine have a better effect than placebo and better ef­
fect than TCAs and MAO inhibitors for the treatment of compulsive-obsessive syndro­
me. The authors also state that venlafaxine can have as good an effect as the SSRIs but
that sertraline has not consistently been able to show a positive effect. According to
BMJ Clinical Evidence it is not known which medicine has the best effect or how long
treatment should continue.
5.5
Post-traumatic stress disorder, PTSD
According to the SBU there is strong evidence for the effect of the SSRIs fluoxetine,
sertraline and paroxetine for the treatment of PTSD and for sertraline there is also
evidence when treatment is continued for up to one year.
The authors in BMJ Clinical Evidence [50] draw the conclusion that there is some
evidence to show that fluoxetine and paroxetine are effective in the treatment of PTSD,
while it is unclear if mirtazapine, sertraline and TCAs are. They also write that there is
credible evidence that venlafaxine is not better than placebo for treating PTSD.
A Cochrane overview by Stein and colleagues [51] shows that pharmaceutical treat­
ment gives effective relief for the main symptoms of PTSD. When the pharmaceutical
treatment is compared to placebo the proportion of patients who responded to the
treatment was 59.1 percent and 38.5 percent for placebo. However, the pharmaceutical
treatment was tolerated less well than placebo. The body of evidence does not allow
any conclusions to be drawn on a certain group of pharmaceuticals having a better
effect or being tolerated better than another group. But the authors believe it to be improbable that all of the medicines studied are as effective as each other. SSRI medici­
nes are the most well-documented both in terms of number and the size of the studies.
Therefore the results in terms of effect are more robust for them than for amitryptiline
and mirtazapine. The authors find that the status of the SSRIs as the first line alterna­
tive for pharmaceutical treatment of PTSD is reasonable and motivated.
5.6
Generalised anxiety disorder, GAD
The SBU finds strong evidence that venlafaxine and paroxetine have an effect on the
treatment of generalised anxiety disorder. There is also evidence to indicate that sertra­
line and escitalopram have an effect.
According to the authors in BMJ Clinical Evidence [52] the symptoms of GAD are
53
The review of anTiDePressanTs
decreased by the antidepressants imipramine, paroxetine, sertraline, escitalopram and
venlafaxine compared to placebo, but this has also been shown for benzodiapines,
buspiron and hydroxyzine.
The authors write that comparisons between different antidepressants by and large
have shown a similar effect for the different treatments, but that in one study escitalo­
pram could have had a better effect than paroxetine.
In a Cochrane overview by Kapcinzki and colleagues [53] the authors could not draw
any conclusions on which antidepressant had the best effect. The authors came to the
conclusion that imipramine, venlafaxine and paroxetine have a better effect than pla­
cebo and are tolerated well for treatment of GAD amongst adults. Sertraline showed
a better effect than placebo for the treatment of children and youths suffering from
GAD.
5.7
Bulimia nervosa
In BMJ Clinical Evidence [54] the authors state that fluoxetine, citalopram, desipra­
mine and imipramine are more effective than placebo for the treatment of bulimia
nervosa.
The evidence for MAO inhibitors is considered to be weak, but the authors draw the
conclusion that it is effective to carry out treatments with them. They write that they
do not know if mirtazapine, reboxetine and venlafaxine are effective and neither do
they know if the effect from continuing the treatment for patients who have recovered
is positive.
In a Cochrane overview Bacaltchuk and Hay have also examined if antidepressants are
effective for the treatment of bulimia nervosa [55]. They write that antidepressants
probably have a direct impact on eating behavior but also an effect on the depression
which then indirectly affects the eating behavior.
The authors cannot find any statistically significant difference in terms of effect
between the different pharmaceutical classes which were examined (TCAs, SSRIs and
MAO inhibitors) or between the different substances within the classes. Fluoxetine ap­
pears however to be accepted to a greater degree by patients than other medicines and
is also the most well-documented substance. According to the authors this is a reason
for why fluoxetine is suitable for use as a first treatment of the disease. Only treatment
with antidepressants is however not sufficient. At the end of the studies the majority of
the patients were still sick. The authors do draw however the conclusion that the effect
from the medicines is greater than that derived from placebo.
54
The review of anTiDePressanTs
6
Is pharmaceutical treatment of
depression cost-effective? –
literature review
The SBU’s systematic review of the health economic literature for depression became
the departure point for our literature review [1]. The SBU searched literature from
1975 up until 2001. We have added to this search and included the period up to Octo­
ber 2006.
In the SBU’s systematic review of the health economic literature up until 2001, 150
articles were identified. A large but non-specified number were judged to be irrelevant
after the summaries had been read. In our supplementary search we identified around
400 articles. Having read the summaries we were able to filter out a large number of ir­
relevant articles. Following this 32 articles remained for reviewing. Material submitted
by the companies concerned gave a further 15 articles for reading. After reading the
articles 35 articles remained.
The studies are focused on clarifying differences in cost-effectiveness between and
within various classes of antidepressants. Further more the studies aim at establishing
which strategy for pharmaceutical use is most cost-effective, that is compared to other
treatments, which medicine to use in the first line, second line and so on.
A general conclusion to draw is that higher initial costs can be balanced by lower costs
in the long run. As a whole pharmaceutical costs as one single cost item seem to be of
subordinate importance and even relatively modest improvements in clinical effect can
balance up relatively large differences in pharmaceutical prices. This is also because
the costs in absolute terms are not especially high, from one up to a few tens of crowns
per day.
Below you will find a short summary of our literature search. A more detailed descrip­
tion and discussion on individual studies is available in Appendix 11 and on www.tlv.
se/depression.
55
The review of anTiDePressanTs
6.1 Cost-effectiveness comparisons between medicines from
different classes
The SBU stated that there were a number of model studies showing that SSRIs were
more cost-effective than TCAs but that prospective studies, which in the estimation of
the SBU have a higher value in terms of evidence, did not support this. The SBU found
a couple of model studies showing that mirtazapine was cost-effective in comparison
to both amitriptyline and fluoxetine, and a study which showed that moclobemide was
cost-effective compared to fluoxetine. The SBU details also two model studies showing
that venlafaxine was cost-effective compared to SSRIs and TCAs.
The SBU noted that there was almost a complete absence of health economic data
from Swedish primary care and that most studies had to do with American conditions.
TCAs showed in general worse results in company-sponsored studies than both SNRIs
and SSRIs if costs were considered and especially in terms of efficacy. In the studies
TCAs are to a larger extent associated with side-effects and have also less of a clinical
effect.
Following the SBU’s review a number of studies have been added which have not been
sponsored by companies and which partly give a different picture for comparisons bet­
ween SNRIs, SSRIs and TCAs. The studies are summarised in Table 29 in section 11.
One study, sponsored by the Catalan HTA authority [56], compares fluoxetine to imi­
pramine. The costs for imipramine were lower, at the same time as the clinical effect
was better for patients with major depression and dysthymia. It was only the indirect
costs which differed between the medicines, while the indirect costs were the same.
An Australian government-sponsored study found that TCAs were the most cost-effec­
tive pharmaceutical treatment [57]. A study from the British HTA agency [58] howe­
ver found that lofepramine dominated TCAs and that if one used an SSRI instead of
lofepramine then the cost per QALY was under 100 000 Skr.
A study where mirtazapine is compared to paroxetine found that mirtazapine is domi­
nant, that is to say has a better effect and lower costs. Differences in effect and costs
are however rather small and not statistically significant [59]. The results are suppor­
ted by the studies detailed by the SBU. All studies were financed by Organon.
Venlafaxine has been compared to both TCAs and SSRIs in three cost-effectiveness
studies sponsored by Wyeth [60-62]. Venlafaxine’s higher pharmaceutical costs were
counterbalanced in the studies by lower other costs.
A Swedish study where escitalopram is compared to venlafaxine indicates a lesser advantage for escitalopram [63]. In other studies where escitalopram has been compared
to venlafaxine no significant differences have been found between the substances, in
terms of effect of costs [64-67]. When it comes to effect the substances are very alike,
56
The review of anTiDePressanTs
and in terms of costs there is a very weak trend speaking to the advantage of escita­
lopram. The pharmaceutical prices indicated in the articles are roughly the same as
today’s Swedish prices.
6.2 Is there any difference in cost-effectiveness within
the different classes of medicines?
The SBU identified a study showing that sertraline was less costly than fluoxetine and
a retrospective study which showed that fluoxetine was the least costly of three SSRI
substances. But in general the SBU does not discuss if there is a difference in cost­
effectiveness between medicines within the same class.
In our supplementary search a number of analyses have arisen on medicines within the
same class. In a number of published studies there is evidence showing that escitalo­
pram 10–20 mg is a cost-effective alternative to the double dose of citalopram [64, 66,
68–73].
With one exception all of the studies are financed by Lundbeck. In the one not finan­
ced by Lundbeck [71] the authors start from the assumption that there is no difference
in effect between escitalopram and citalopram, which of course affects the results from
the study.
One study where sertraline is compared to citalopram showed that patients who had
been treated with sertraline had a significantly lower cost for the half-year. A Canadian
study where duloxetine was compared to venlafaxine found that venlafaxine domina­
ted [74].
6.3 Which strategy for using pharmaceuticals is most
cost-effective
The SBU stated in its report that pharmaceutical treatment is seldom valued in rela­
tion to non-pharmacological treatment or no treatment at all. In the few evaluations
of pharmaceutical treatment contra in particular psychotherapy different conclusions
have been reached. Also here, in our supplementary search, we have identified further
studies over and above those identified by the SBU.
6.3.1 Is pharmaceutical treatment cost-effective compared to no treatment?
There are very few studies analysing pharmaceutical treatment in comparison to no
treatment and the SBU does not comment on this.
One study where sertraline is compared to placebo showed that it was cost-effective to
carry out treatments using sertraline. The study did have to do with patients suffering
57
The review of anTiDePressanTs
from specific coronary conditions and was carried out in the USA, and for this reason
we value the relevance as low [75]. But as pharmaceutical costs are only a small part
of the costs for depression we believe it is probable that all pharmaceutical groups
are cost-effective compared to no treatment at all, at least for patients suffering from
moderate to severe depression. Also treatment using cognitive behavioural therapy or
other psychotherapy appears to be cost-effective compared to no treatment at all. This
conclusion is supported by a study from Australia [57].
6.3.2 Is long-term treatment cost-effective?
In some studies long-term treatment with pharmaceuticals aimed at preventing relapse
into depression has been compared to no treatment. In a Swedish study venlafaxine
was analysed as a preventive treatment for two years in comparison to no preventive
treatment. From a socioeconomic perspective the cost per QALY was 18 500 USD [76].
Earlier studies have shown that three years of maintenance treatment with imipramine
is cost-effective compared to both placebo and psychotherapy [77] and that milna­
cipran is cost-effective compared to clinical follow-up with medicines [78]. A study has
shown that maintenance treatment using fluvoxamine [79] is cost-effective compared
to maintenance treatment with TCAs and another that maintenance treatment for
one year with citalopram was cost-effective compared to only acute treatment using
TCAs [80]. Also an older study showed that relapse prevention maintenance treatment
with sertraline was cost-effective, compared to episodic treatment with dothiepin for
patients with a high risk of relapse [81]. A Dutch model study showed that treatment
according to guidelines - acute treatment and then six months of continuing treat­
ment – is not cost-effective, but that maintenance treatment up to a total treatment
time of 21 months can be [82].
As a whole we consider there to be evidence in the literature indicating it is cost-effec­
tive to use an antidepressant over a longer period of time to prevent relapse, as compa­
red to not doing so. It is however not known which medicine is most cost-effective and
how long such a treatment should continue in order to be optimal.
6.3.3 Is pharmaceutical treatment cost-effective compared to other
treatments?
Two studies of moderate relevance and one of low relevance indicate that psychoth­
erapy is cost-effective compared to pharmaceutical treatment, mainly for preventing
relapse [57, 83, 84]. Three studies of low relevance arrive at the opposite conclusion
[85-87].
In its model analysis NICE finds that the cost of treatment using only cognitive behavi­
oural therapy is probably not reasonable if solely the effect is taken into account [24],
but that cognitive behavioural therapy in combination with pharmaceutical treatment
58
The review of anTiDePressanTs
is cost-effective compared to only pharmaceutical treatment. Also Browne arrives at
the result that a combination treatment using cognitive behavioural therapy and ser­
traline is cost-effective compared to each of the treatments separately [88].
6.4
Do the studies support our stance?
The existing health economic literature within the area of depression unfortunately
comprises a rather sub-standard basis for making decisions in regard to reimburse­
ment of medicines in Sweden. For a number of different reasons we judge the overall
usefulness of the material to be of little value.
An overwhelming majority of the information is financed by companies. This is not
a problem in itself but is likely related to the possibly most serious weakness in the
studies; that so few are based on the best data available on clinical effect. The analyses
are based in many cases on one clinical test or a small sample of tests, meaning a risk
of selection bias, that is to say through selection influence the results reached.
When the analysis is based on just one study then relevant information is ignored.
This is a problem in itself, but on top of this there is a risk that the study one has
chosen is structured so that it benefits the sponsoring company’s medicine. The fact
that the majority of the studies are sponsored by companies also means that there is
most likely a bias when it comes to which results are to be published. Both of these
problems have been documented in the literature. Baker and colleagues [89] have
examined the connection between company-sourced financing and published results
and found that company-sponsored studies, to a larger extent than for studies with­
out such sponsorship, arrived at conclusions which might be seen as benefiting the
sponsor. A discernible publication and reporting bias within the area of depression
has been documented by Melander and colleagues [90] and by Hansen and colleagues
[31]. In a meta-analysis the latter have examined company financing of a study related
to how large a proportion of patients in a study recover. The authors found that the
proportion was around five percent higher in studies financed by companies which
marketed the medicine in the study.
Gathering cost data directly from an individual clinical trial for the health economic
analysis can be problematic. The clinical studies are normally structured and dimensio­
ned to show significant differences in clinical effect between various alternatives, but
not to show differences in cost between the treatment alternatives. The latter normally
demands a much greater number of patients in the study, as the distribution of costs is
often uneven with many patients with very low or no costs, and a few with very high.
It is well-known that external validity, even for a well-executed clinical trial, may be in
question. There is often a difference between the effect a medicine shows in the con­
59
The review of anTiDePressanTs
trolled environment of a trial compared to the effect displayed in a real environment.
The age of patients, other concurrent diseases, compliance to treatment and how the
medicine works together with other medicines are examples of what can differ from
the real treatment situation to the controlled trial and can influence the results from
the treatment.
As expected only a few studies are from a Swedish perspective and a majority of the
articles also include only direct costs for medicines and visits to the doctor. We want
to have a societal perspective and take indirect costs into account also. As their share
of total costs can be 75 percent or more [see 91, 92 for examples] this becomes an issue
of consequence. Excluding the indirect costs in a health economic analysis can for
example lead to an underestimation of cost-effectiveness for an expensive substance
which has a better effect than a cheaper substance.
International prices, sometimes many years old, may differ greatly from today’s prices
in Sweden. In particular this applies for medicines where the patent has expired since
the studies were carried out. The decrease in price from a relative perspective can be
huge and in general the largest difference in price between the studies and the Swedish
price in reality can be observed for citalopram, which can be up to 95 percent cheaper
in Sweden. The least difference in price is mainly observed for venlafaxine, where the
price of the studies in some cases is even higher.
Besides the costs for the medicines studied the studies cover many other costs. Costs
cannot be assumed to always apply in Sweden and by extension neither can the re­
sult. That the price of various healthcare resources differ so much between different
countries is obvious, but also indirect costs such as absence from work due to sickness
is country-steered in many cases. In an article by Patel and colleagues [85] based on
Indian cost conditions it can be assumed for example that a daily salary would be com­
parable to a maximum of four USD.
As mentioned above indirect costs are the main cost drivers from a societal perspective.
With a salary level a long way under what is the case in Sweden, the results here are
of limited use for our purposes, at the most. There are a smaller number of database
studies in the selection. These are often retrospective reviews of insurance databases in
the USA. Patients who for a period of time are treated for the first time with a cer­
tain medicine are followed for one or more years, and then compared to a group who
during the same period of time has received another treatment.
A weakness with these types of studies is that the effect side is often badly examined,
many times costs are the only aspect detailed. If the group which received one medi­
cine has lower total costs than the other group then it is assumed to be due to the
medicine leading to lower costs. The causal chain is however seldom established [93].
60
The review of anTiDePressanTs
Other problems with database studies are that the population in some databases is
not representative of the population at large and information on crucial costs may be
missing. The databases are furthermore almost only from the USA with the cost condi­
tions which prevailed there during the study period, making the relevance for Swedish
conditions lower.
61
The review of anTiDePressanTs
7 Is the pharmaceutical treatment
of anxiety cost-effective?
The SBU has published a systematic review of the literature in regard to health econo­
mic aspects of anxiety. The base documentation is made up of articles published from
1996 up until January 2005. The SBU states that to a large extent there is a dearth of
studies shedding light on the cost-effectiveness of various treatments.
We have augmented the SBU’s review with a literature search for the period February
2005 – April 2007 and a review of the reference lists sent in by companies as part of the
base documentation for this review. An evaluation of these resulted in only six articles.
The results from the SBU’s review and our own search are presented below.
7.1
Panic
The SBU identified seven health economic evaluations for the treatment of panic, of
which six included a pharmaceutical. Cognitive behavioural therapy appeared to be
the most cost-effective alternative in the three studies where it was included as an alter­
native. In one of these TCAs were the next most cost-effective for panic and least effect
were the SSRIs. Another study showed that maintenance treatment with imipramine
was cost-effective compared to only acute treatment with imipramine. A study of Ame­
rican insurance data compared treatment with sertraline, paroxetine and fluoxetine
and showed that costs for acute health care and laboratory services were lower for the
six month period after an SSRI treatment had been commenced, than for the six pre­
vious months. When costs for the medicine were counted only sertraline constituted
a definite cost saving. A Spanish study showed that both direct and indirect costs were
considerably lower the year following the introduction of pharmaceutical treatment
and psychotherapy, than the previous year. Total costs decreased from approximately
100 000 USD to approximately 60 000 USD.
Following the SBU’s review two more studies are available. In a Canadian study [94]
only cognitive behavioural therapy was compared to cognitive behavioural therapy in
combination with pharmaceutical treatment. The patients, suffering from panic and
anxiety disorder with agoraphobia, were followed for two years in a prospective cohort
study. The patients were not randomised and they included direct healthcare costs as
well as the patient’s own fees paid. It was found that both treatments were as effective.
However, as only cognitive behavioural therapy is cheaper than the combination with
62
The review of anTiDePressanTs
drugs, only treating with cognitive behavioural therapy is the more cost-effective alter­
native. The relevance of the study is judged to be low and the results should further­
more be interpreted cautiously.
Katon and colleagues [95] compared prevalent treatment of panic to cognitive behavi­
oural therapy in combination with drugs, as a first line with an SSRI. If the patient has
not previously reached the desired treatment outcome with two SSRI treatments then
an alternative antidepressant is chosen. Standard care often consists of treatment with
an SSRI.
The study was carried out in the USA and included only direct healthcare costs. The
combination treatment was somewhat more expensive over a twelve-month period,
about 500 USD, but on average resulted in 60 more anxiety-free days. The cost per
QALY was estimated at between 14 000 – 24 000 USD. The study was sponsored by the
National Institute of Mental Health. We judge the study to be of low relevance.
7.2
Compulsive-obsessive syndrome
No studies for compulsive-obsessive syndrome were identified by the SBU, while in our
search we identified a cost-effectiveness analysis conducted by NICE, where cognitive
behavioural therapy is compared with an SSRI and a combination of SSRIs and cogni­
tive behavioural therapy [96].
Data on effect originate from NICE’s own meta-analyses and systematic review.
A meta-analysis based on 10 studies showed that approximately 43 percent of the
patients responded to a twelve-month treatment with SSRIs, in comparison to
27 percent in the placebo group.
This translates to a treatment effect of 16 percent for SSRIs. The data on efficacy for
cognitive behavioural therapy was based on two published studies and showed that
the average effect of cognitive behavioural therapy was 53 percent. The effect of the
combination treatment was estimated to be somewhat larger than for the treatment
alternatives individually and was assumed to be 63 percent.
The treatment costs for cognitive behavioural therapy are greater than for SSRIs, at
least in the short term. NICE found that treatment where all patients only get cogniti­
ve behavioural therapy costs more and results in less effect than a treatment alternative
where a part of the patients are treated with SSRIs and the rest with the combination
cognitive behavioural therapy and pharmaceuticals. This means that the relevant
comparison is between SSRIs and cognitive behavioural therapy in combination with
SSRIs. The cost for rehabilitating one further patient in this case becomes 2 247 Bri­
tish pounds. NICE states that if at least 0.07 QALYs for each rehabilitated patient are
63
The review of anTiDePressanTs
gained then the combination treatment is cost effective in comparison with SSRIs as
sole therapy.
The model used by NICE utilises a twelve-month perspective. NICE states that if cog­
nitive behavioural therapy has a more lasting effect than SSRIs then this changes the
cost-effectiveness. NICE also draws the conclusion that more value for money is deri­
ved from treating compulsive-obsessive syndrome with cost-effective methods as soon
as the condition is identified, than waiting to treat the person until they have become
more seriously ill at a later stage.
7.3
Social phobia
The SBU identified a study from Great Britain showing that the burden of disease for
social phobias could decrease considerably after making the diagnosis and initiating a
treatment. No further studies have been identified for social phobias.
7.4
Generalised anxiety syndrome
The SBU recounts two studies where the analyses have been made from a healthcare
perspective. Cognitive behavioural therapy, provided by the public sector, was the most
cost-effective treatment, while venlafaxine was as cost-effective as cognitive behaviou­
ral therapy from private care.
In our follow up we have identified four studies. Guest and colleagues [97] carried
out a study financed by Wyeth which was a health economic analysis of venlafaxine
compared to diazepam for treatment of generalised anxiety syndrome in Great Bri­
tain. The analysis was based on a multinational clinical trial where it is shown that the
probability of achieving remission of generalized anxiety syndrome was approximately
30 percent for venlafaxine compared to almost 17 percent for diazepam, and the risk of
relapse was 3.5 percent and 17 percent respectively.
Only direct costs are included and these increased somewhat under venlafaxine,
around 500 Swedish crowns in six months, meaning a cost per further recovered pa­
tient of approximately 5 000 crowns. The study was of moderate relevance.
Panzer and colleagues [98] examine the economic implications of either prescription
of generics (generic step therapy) or free prescription (open formulary) of patented
products within the SSRI class for patients suffering from anxiety. The analysis was in
regard to a hypothetical health plan in the USA with a million members. Only direct
costs were counted. In the generic line fluoxetine, paroxetine (tablet) and citalopram
were included. In the open formulary group sertraline, paroxetine (slow release tablet)
and escitalopram. Efficacy and cost data were gathered from the literature. The me­
64
The review of anTiDePressanTs
asure of outcome was how many patients were treated for 180 days and the number of
treatment changes.
Generic step therapy resulted in lower pharmaceutical costs, but more changes in
therapy type and higher total healthcare costs. The study is of low relevance and who
sponsored the study is not stated.
Shehan and colleagues [99] examined the differences in direct healthcare costs during
a period of six months between patients treated using paroxetine as a slow release ta­
blet and SSRIs in the form of conventional tablets (sertraline, paroxetine, citalopram,
fluoxetine and escitalopram). The study, in relation to patients with some depression
or anxiety diagnosis, was a retrospective database analysis from the USA and compri­
sed more than 140 000 patients, of which 7 percent received paroxetine slow release
tablets. SSRIs in the form of conventional tablets, both as a whole and individually,
entailed higher costs than paroxetine slow release tablets regardless of the diagnosis in
question. We judge the study to be of low relevance. The sponsor for the study is not
indicated.
Jörgensen and colleagues [100] showed in a study sponsored by Lundbeck that escita­
lopram dominates paroxetine for treatment of generalised anxiety syndrome in Great
Britain. The study had a nine-month time horizon and was carried out from a societal
perspective. The data on efficacy came mainly from a direct head-to-head clinical trial
where patients treated with escitalopram had approximately 14 percentage points
greater probability of experiencing a successful treatment than patients treated using
paroxetine. The most important costs were losses in productivity and other indirect
costs, which stood for 95 percent of the total costs. From a societal perspective es­
citalopram is ahead in comparison to paroxetine with a cost saving of 1 400 British
pounds over nine months. From a healthcare perspective the savings are 39 British
pounds. As escitalopram also has a better effect in the study it is the dominant alterna­
tive. One strength is that it follows NICE’s treatment recommendations for generalised
anxiety syndrome. Efficacy data based on one single study is not satisfactory. The study
is of a medium to high relevance.
7.5
Post-traumatic stress disorder, PTSD
No studies have been identified for PTSD.
7.6
Bulimia nervosa
Bulimia nervosa was not part of the SBU’s review of antidepressants. NICE developed
an own health economic model to analyse the cost-effectiveness of cognitive beha­
vioural therapy compared to treatment with an antidepressant (first line fluoxetine)
65
The review of anTiDePressanTs
[101]. The analysis showed that cognitive behavioural therapy is more effective but
costs more. In the analysis it is assumed that 1 000 people are treated with cognitive
behavioural therapy and an antidepressant. Based on a systematic review of the clinical
literature the number of recovered patients is estimated to be 370 for cognitive behavi­
oural therapy and 192 for treatment with an antidepressant.
The costs for cognitive behavioural therapy amounted to 967 000 British pounds
compared to 118 000–238 000 British pounds for treatment with an antidepressant,
depending on the prescriber. The estimation of the cost for every additional success­
fully treated case of bulimia nervosa varied from between 4 807, 4 942 and 4 126 Bri­
tish pounds depending on if the antidepressant was prescribed from primary care, or
in specialist care by a junior or senior psychiatrist. NICE considers it however to likely
be a case of overestimation of cost per gained effect, as potential savings in healthcare
are not calculated at all. Also, NICE thought it very improbable that a combination
treatment of bulimia with cognitive behavioural therapy and an antidepressant would
be cost-effective for the English healthcare system. NICE utilises a healthcare system
perspective in its evaluations in the first instance. In this case not even savings within
this sector were calculated for.
7.7
Lack of health economic studies and data on efficacy
The conclusions reached in the SBU’s review are still on the whole valid in regard to
the lack of health economic evaluations within the area of anxiety. We have in general
assigned a low relevance to the studies which do exist for decisions made by the TLV.
One conclusion which seems to be supported by the literature is however that cogniti­
ve behavioural therapy is a cost-effective treatment compared to pharmaceutical treat­
ment, at least for panic syndrome, generalized anxiety syndrome and bulimia nervosa.
For compulsive-obsessive syndrome it appears that combination treatment with SSRIs
and cognitive behavioural therapy is the most cost-effective.
Two company-sponsored studies have shown that venlafaxine and escitalopram are
cost-effective compared to diazepam and paroxetine respectively for treatment of gene­
ralized anxiety syndrome.
There is a need for additional health economic evaluations within the various indica­
tions for anxiety. The main weakness within the area is however the absence of relevant
and reliable data on clinical efficacy. In the TLV’s view there is no need for an own
health economic model in the area of anxiety to be able to make decisions on the reim­
bursement status of medicines.
66
The review of anTiDePressanTs
8 Own health economic model
It is easy to state that in many cases there are differences in the cost of treatment bet­
ween the antidepressant substances. If the more expensive medicines are to remain in
the reimbursement scheme then the differences in cost of treatment must be motiva­
ted by differences in efficacy.
Our review of the health economic literature has not given us a base documentation
which is sufficient for our evaluation of the cost-effectiveness of the medicines. For
this reason we have chosen for the purposes of the analysis to develop an own health
economic model. We have found that the relationship between the medicine’s effects
on depression and costs therefore ought to be analysed in a health economic model
which to a reasonable degree reflects how we in Sweden have organized the care of
patients suffering from depression.
Primary care is responsible for the first treatments and a large proportion of the
patients receive all of their treatment within primary care. Specialist care takes care
of patients who have not achieved their treatment objectives within primary care, and
patients with more severe and complex sets of symptoms.
8.1
Some points of departure
A comprehensive health economic analysis of the treatment of depression would need
to contain a considerable number of steps and shed light on the acute treatment phase
for patients suffering from different types of depression (unipolar, bipolar, with or
without anxiety etc), maintenance treatment, treatment of relapse, treatment of pa­
tients in different ages as well as multiple other factors. But also other analyses which
are not comprehensive can be of great value for the TLV as decision-maker, especially if
the TLV’s primary remit is limited to reviewing the reimbursement status for antide­
pressant substances.
We have identified a patient group which appears to be pivotal to analyse and where we
believe the health economic documentation to hand is insufficient – Swedish patients
with depression and treated as out-patients. The patients are assumed to have concur­
rent anxiety in many cases, but many other disease-related conditions are excluded,
such as addictions, psychotic illnesses and personality disorders. Similarly bipolar
depression is not included as to a large degree this is treated using medicines not in­
cluded in this review.
67
The review of anTiDePressanTs
Other points of departure have been that our health economic model shall be from
a socioeconomic perspective and use QALYs as a measure of outcome as well as the
objective from the treatment of depression being that the patient recovers.
One must be aware of the limits set by general conclusions reached based on the
results from such an analysis. In particular that if our analysis shows that a medicine
is not cost-effective then it does not mean that there is no cost-effective use of this
medicine in general.
The health economic model gives the opportunity to estimate what a certain clinical
outcome, what for example the share of recovered patients in the first treatment of
patients with depression-means in terms of costs and health. The model aims at descri­
bing in a conventionalised form Swedish primary care during the first twelve months
after the patient has fallen ill. This means that not all antidepressant medicines are in­
cluded as an alternative, as they are not used for safety reasons as a first line treatment,
to give an example. This is for instance the case with TCAs and MAOs.
SNRIs are often not considered to be suitable as a first line alternative for cost reasons
when there are SSRIs available at low prices. However, we do not consider this to be
sufficient reason for excluding a medicine from the analysis, it is on the contrary this
issue which is to be illuminated; is it cost-effective to use a somewhat better but more
expensive antidepressant as a first line treatment in out-patient care?
In order to analyse the cost-effectiveness of the medicines at hand we have constructed
two different health economic models: a twelve-month model for the first treatment
for an episode of depression; and a thirteen-month model for the cases where the first
treatment has not given an adequate treatment result.
8.2
Costs and effects for twelve-month treatment
We have compared costs and effects for first line treatment of moderate to severe de­
pression with fluoxetine, citalopram, paroxetine, sertraline, fluvoxamine, escitalopram,
mirtazapine, venlafaxine, reboxetine and duloxetine in a health economics model for
twelve months of treatment in Sweden. There are a number of treatment steps in the
model but it is not these which should be evaluated and they are therefore the same for
all treatment alternatives. It is only the probability of needing further treatment which
differs.
Some treatment steps and results are based on the American STAR*D study. STAR*D
is further discussed in section 3.4.2. Despite some weaknesses STAR*D was in any case
the study with the highest validity for treatments after the initial treatment had not
given the desired result.
68
The review of anTiDePressanTs
fluoxetine
attempted
suicide
paroxetine
remission
no relapse
maintenance treatment
duloxetine
attempted
suicide
dies
mirtazapine
escitalopram
remission
no remission
no attempted
suicide
reboxetine
venlafaxine
specialist care
relapse
sertraline
fluvoxamine
dies
survives
survives
switch
remission
maintenance treatment
no remission
specialist care
remission
no remission
specialist care
remission
no remission
no remission
no attempted
suicide
switch
citalopram
remission
maintenance treatment
no remission
specialist care
remission
no remission
Figure 2. Structure of the model for 12 months of treatment of depression
8.2.1
Preconditions for the model
The structure of the model is accounted for in Figure 2. The patients in the model are
treated with one of the antidepressants and evaluated for three months. The treatment
objectives are to achieve remission, measured as seven points or lower on the HAMD
scale. Patients achieving remission in the first treatment step in the model do so after
one month of treatment. In the STAR*D study it took on average approximately six
weeks to remission so this could be an overestimation.
Once in remission there is a risk of relapse in depression. A systematic review of obser­
vational studies in primary care [102] refers to two studies with relapse frequencies of
11 and 30 percent respectively, but the latter study was very small. Keller [103] reports
a 13 percent relapse after six months while STAR*D had 33.5 percent Relapse for those
who achieved remission in the first step. Those who suffered a relapse experienced
this on average 4.4 months after remission. NICE uses a twelve month risk of relapse
of 0.55 in its health economics model, but this figure comes from a single study from
1986 and is valid over, as mentioned, twelve months. In our principal analysis we as­
sume the risk for relapse is 11 percent and that relapse occurs after four months. The
risk of relapse is varied in a sensitivity analysis.
If the patient does not experience a relapse then she or he is put on maintenance
treatment for six months following reaching remission, in accordance with Swedish
treatment guidelines.
Patients who relapse, like patients who have not achieved remission in first line treat­
ment, have a certain tendency to carry out a suicide attempt. As reported in section 2.2
there are different estimates of the risk for suicide. According to an overview article the
risk was over 1 percent in only one of 31 different studies [8]. We follow Löthgren here
[63] and base our risk on Khan [104] and set the risk of suicide attempts at 0.031 and
the risk of dying in such an attempt at 0.1. If the patient survives the suicide attempt
then he or she is moved to specialist care.
69
The review of anTiDePressanTs
Patients who relapse after an initial successful treatment and who do not attempt
to commit suicide change treatment (”the switch arm” in the model) to venlafaxine.
This is based on data from STAR*D. Those who, after three months in second line
treatment in primary care, have not achieved remission are moved to specialist care
which we believe, on consultation with our clinical experts, reasonably reflects Swedish
traditions in treatment. Those who achieve remission are assumed to do this after one
month of treatment, which compares well with data from STAR*D, and they are on
maintenance treatment then for six months.
Specialist care, for those who have moved on to this, continues until a total of twelve
months have passed. In our analysis we use the combined average treatment effect for
the third and fourth step in STAR*D and those who have achieved remission are as­
sumed to have done this after a month of treatment, which is comparable to data from
STAR*D. As mentioned we are not concerned here with evaluating the steps themsel­
ves.
8.2.2
Efficacy data for the model
The treatment effects after three months come from our own meta-analysis which has
been detailed in section 3.8 and shown in Table 13. The effect of changing to venlafax­
ine in the second treatment step comes from STAR*D and has been set at 24.9 percent.
We have assumed that the effect of specialist care is the same as the combined average
treatment effect for the third and fourth step in STAR*D. Similarly the average time to
relapse and to the patient recovering for the different steps comes from STAR*D.
Table 13. Treatment effect and monthly cost after three months as well as for switching treatment and specialist care
Treatment
Cost in crowns
per month
Effect
mirtazapine
104
0.4503
escitalopram
433
0.4744
sertraline
49
0.4289
paroxetine
58
0.4259
duloxetine
367
0.4484
reboxetine
306
0.4313
venlafaxine
626
0.4561
citalopram
22
0.4040
fluoxetine
32
0.4017
fluvoxamine
231
0.2670
venlafaxine as second step
626
0,249
specialist care
–
0,249
The treatment effects have been expressed as quality-adjusted life years with the aid of
data from Sobocki and colleagues where quality of life weightings for a patient who
70
The review of anTiDePressanTs
has recovered and one who has not recovered are measured as 0.81 and 0.57 respec­
tively. This means the number of QALYs gained are 0.0675 of one more month as a
recovered patient, while being 0.0475 of a month for not having recovered.
8.2.3
Costs in the model
Costs for the primary care part of the model come from Sobocki and colleagues [29]
and patients are categorized after recovery (remission) or not. The costs are accounted
for in Table 14. We have deducted costs for antidepressant treatment in Sobockis and
colleagues’ figures. Pharmaceutical costs for the ten medicines to be evaluated shall
also be added here. The monthly costs for them are detailed in Table 13. Costs for the
specialist arm in the model are gathered from Löthgren and colleagues [63]. We have
assumed that the same relationship between recovered and non-recovered patients
applies for Löthgren and colleagues as for Sobocki and colleagues and von Knorring
and colleagues [16], ie, that costs are 39 percent lower for recovered patients. Also,
costs for suicide are gathered from Löthgren and colleagues.
Table 14. Costs in SKR for various events in the model.
Event
Costs in Swedish crowns per patient and month
– remission
6 285
– no remission
10 410
Patient in specialist care
– remission
9 637
– no remission
15 964
Costs in Swedish crowns per patient and month
switch of treatment
1 000
suicide – attempt
42 275
suicide – death
4 950
8.2.4
Results
Escitalopram is the treatment which gives the highest number of QALYs. It is also a
treatment which is associated with lower total costs than all other alternatives excep­
ting mirtazapine. All other treatments are more expensive and have a worse effect than
escitalopram, in other words they are dominated.
The cost and effect ratio for escitalopram compared to mirtazapine is approxima­
tely 61 000 Swedish crowns which is normally considered to be a low cost per QALY.
Detailed results are accounted for in Table 15. There it is indicated that the difference
in absolute numbers is small in terms of cost and the number of quality adjusted life
years for most of the treatments.
71
The review of anTiDePressanTs
Table 15. Results from the cost-effectiveness modelling over twelve months with
all alternatives, based on all studies in the meta analysis
Incremental
cost and effect ratio
Treatment
Cost
Effect (QALY)
mirtazapine
162 753
0.6952
escitalopram
162 978
0.6988
61 257
sertraline
163 699
0.6919
Dominated
paroxetine
163 909
0.6914
Dominated
duloxetine
164 094
0.6949
Dominated
reboxetine
164 754
0.6923
Dominated
venlafaxine
164 893
0.6960
Dominated
citalopram
164 979
0.6881
Dominated
fluoxetine
165 154
0.6877
Dominated
fluvoxamine
173 544
0.6672
Dominated
In the results above we used all studies in our meta-analysis to calculate remission
frequencies at three months. If we instead use only studies with a follow up period of
8–12 weeks (see Table 12) then the result is changed somewhat, as seen in Table 16.
Escitalopram still gives most QALYs and now dominates all alternatives, including
mirtazapine.
Table 16. Results from the cost-effectiveness modelling over twelve months with
all alternatives, based on studies of 8-12 weeks follow up
Medicine
Cost
Effect (QALY)
Incremental
cost and effect ratio
flouxetine
117578
0.6770
Dominated
venlafaxine
118030
0.6838
Dominated
paroxetine
117093
0.6787
Dominated
mirtazapine
116014
0.6825
Dominated
escitalopram
115709
0.6879
Dominant
duloxetine
116731
0.6839
Dominated
citalopram
117518
0.6770
Dominated
sertraline
117115
0.6785
Dominated
reboxetine
118517
0.6781
Dominated
Mirtazapine can be seen as a second line alternative. If it is removed from the compari­
son then escitalopram has a better effect and lower costs than all of the other alternati­
ves. Other comparisons which may also be of interest for decision-makers are studying
only SSRI substances without escitalopram, which is of course dominant. The results
are detailed in Table 17 where it indicates that paroxetine has both a better effect and
lower total costs than other medicines in the comparison although the differences are
so small that they are almost non-existent.
72
The review of anTiDePressanTs
Table 17. Results from the cost-effectiveness modelling over twelve months with only SSRIs and
excluding escitalopram, 8–12 weeks
Treatment
Cost
Effect (QALY)
Incremental
cost and effect ratio
paroxetine
117093
0.6787
Dominant
sertraline
117115
0.6785
Dominated
citalopram
117518
0.6770
Dominated
fluoxetine
117578
0.6770
Dominated
If we only analyse SNRI/NRI substances then duloxetine has both a lower cost and bet­
ter effect than the others if we only utilise 8–12 week studies. Venlafaxine on the other
hand has a better effect and higher cost than duloxetine if we use all studies (Table 18).
The differences are small, but the cost and effect ratio is over 600000 Swedish crowns
which can be considered to be quite high.
Table 18. Results from cost-effectiveness modelling over twelve months with only SNRIs/NRIs.
All studies.
Incremental
cost and effect ratio
Treatment
Cost
Effect (QALY)
duloxetine
117 600
0.6800
venlafaxine
118 400
0.6813
615 385
reboxetine
118 200
0.6774
Dominated
As a sensitivity analysis we have let the relapse frequency rise from 11 to 33.5 percent
but this does not affect the result.
Figure 3 shows a cost-effectiveness acceptability curve (CEAC). A CEAC shows the proba­
bility (on the Y axle) that an evaluated treatment alternative is cost-effective (given observed data) at different levels of maximum willingness to pay (on the X axle) in order
to reach one further unit of the resultant outcome, ie in our case QALYs. For all levels
of willingness to pay escitalopram has the greatest likelihood of being cost-effective,
but this finding is rather uncertain.
8.3
Costs and effects when the first treatment has not succeeded
We have compared costs and effects from second line treatment of depression with
bupropion, venlafaxine and sertraline in a simple model based on treatment steps and
treatment results from the STAR*D study (see section 3.4.2). Despite some weaknes­
ses STAR*D was in any case the study with the highest validity in terms of treatment
routes after a first treatment had not given the desired results.
As we only include three medicines the analysis only gives a partial answer to the ques­
tion on which medicine should be used in the first instance. Our main purpose with
73
The review of anTiDePressanTs
Probability treatment is cost-effective
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
200,000
escitalopram
400,000
600,000
800,000
Willingness to pay for a QALY (SEK)
mirtazapine
reboxetine
1,000,000
1,200,000
duloxetine
Others (fluoxetine, venlafaxine, paroxetine, citalopram and sertraline)
Figure 3. CEAC curves for the analysis with all studies of 8—12 weeks follow-up
the analysis is to decide which medicine the second line treatment shall be comprised
of in our own health economic analyses of the first treatment route.
8.3.1
Preconditions for the model
The structure of the model is detailed in Figure 4. The patients in the model are as­
sumed to have been treated with an antidepressant for three months without recovery,
that is to say have reached seven points or lower on the HAMD scale. In the STAR*D
study only the medicine citalopram was used in the first treatment step, but we believe,
with the support of our clinical expert group, that the results are valid to a reasonable
degree even when the patient commenced treatment with another antidepressant.
After the first step the patient is treated using one of the medicines bupropion, venla­
faxine or sertraline for three months.
If the patient recovers then she is treated with the same medicine until possibly rel­
apsing into depression. In such a case treatment is commenced with one of the treat­
ments given in the third step in STAR*D. In our analysis we use the average treatment
effect for the third step as we do not want to evaluate this step itself. A patient who
does not relapse is treated with the original medicine for the entire time period being
modelled, which is 13 months.
A patient who does not recover goes directly to step 3 after three months. The treat­
ment lasts for three months, or the entire time period being modelled if the patient
subsequently recovers or does not suffer a relapse. If the treatment in step three does
not lead to the patient recovering, or if the patient relapses then there is also a step
four.
74
The review of anTiDePressanTs
remission
relapse 3
steg 4
remission
relapse 2
remission
remission
1
no relapse
no remission
no relapse
step 3
no remission
step 4
no remission
bupropion
remission
relapse 3
Failed 1st
treatment
venlafaxine
sertraline
step 4
remission
no remission
no remission
no relapse
step 3
remission
no remission
step 4
no remission
Figure 4. Structure of the model
The average remission frequency from STAR*D is also used here. After the fourth step
the time period being modelled is at an end.
8.3.2
Effect data for the model
Treatment effects for bupropion, venlafaxine and sertraline after three months as well
as for step three and four come from STAR*D and are detailed in Table 19. Similarly
the relapse frequencies from the various treatments come from STAR*D as do the
average time to relapse and to remission for the different steps.
Treatment effects have been expressed as QALYs with the aid of data from Sobocki and
colleagues, where quality of life weightings at remission and non-remission have been
measured as 0.81 and 0.57 respectively which means that the number of QALYs gained
by having a month with remission is 0.0675, while they are 0.0475 for a month without
remission.
Table 19. Treatment effects, costs, relapse frequency, time to relapse and time to remission for bupropion, venla­
faxine and sertraline after three months as well as for step three and four treatments from STAR*D.
Effect
Cost each
per month
Relapse
frequency
Time to
relapse
Time to
remission
sertraline
17.6%
48
47.4%
4 months
1 month
bupropion
21.3%
345
47.4%
4 months
1 month
venlafaxine
24.8%
626
47.4%
4 months
1 month
step 3
13.7%
400
42.9%
3 months
1 month
step 4
13%
400
–
–
2 months
8.3.3
Costs in the model
Costs in the model originate from Sobocki and colleagues [29]. The patients are cate­
gorised according to if they have reached remission or not. Costs are detailed in Table
14. We have deducted costs for antidepressant treatment in the figures from Sobocki
and colleagues. Pharmaceutical costs for bupropion, venlafaxine or sertraline are ad­
ded to this plus in some cases costs for the third and fourth step treatment.
Costs each for these are detailed in Table 19. Steps three and four are assumed to cost
75
The review of anTiDePressanTs
4 percent of the total costs associated with depression, that is to say approx. 400 Skr
per month.
8.3.4
Results and a commentary
Venlafaxine is the treatment which results in most QALYs, but is also the treatment
associated with highest costs. The cost/effect ratio for venlafaxine compared to bupro­
pion is approximately 51 000 Skr which is normally seen as a low cost per QALY. More
detailed results are outlined in Table 20. The difference in absolute numbers is howe­
ver both small in terms of costs and the number of QALYs.
Table 20. Results from cost-effectiveness modelling over 13 months
Treatment
Cost
Incremental
cost
Effect
Incremental
effect
Incremental
cost and effect ratio
sertraline
126 400
bupropion
126 500
100
0.6787
0.6734
0.0053
26 411
venlafaxine
126 800
300
0.6838
0.0051
51 309
We have chosen to halt the analysis after 13 months as we believe that this captures the
principal costs and the effects of the treatment as described here. A greater problem is
probably that the STAR*D study does not describe how Swedish patients are actually
treated in reality in healthcare.
8.4
Conclusions from our own health economic analyses
The differences in price between various antidepressants are rather large in relative
terms, but small in absolute numbers, especially in light of the large costs associated
with the diseases being treated. Like earlier analyses our results indicate that even
relatively modest differences in effect can motivate the differences in price which exist.
Our own meta-analyses, and earlier reviews, indicate that new and more expensive
medicines may have a somewhat better effect on older medicines and through this be
cost-effective.
It should be noted that as the diseases being treated are very common, so common that
they should be treated as a matter of public health safety, the use of the medicines involved is widespread. Due to this it is crucial to possess a high degree of accuracy and
precision in making decisions. The costs involved in making the wrong decision can be
very high in both human and economic terms. Unfortunately we have to say that the
data available on these medicines are, to a large extent, hardly constructed in order to
promote great precision and accuracy in the decisions made.
The term ”depression” as used in everyday clinical practice carries a wider and more
heterogeneous meaning than that used in clinical studies. For this reason much re­
76
The review of anTiDePressanTs
mains unknown in regard to effects and tolerability of the antidepressants when used
on real patients. This of course limits the possibilities for drawing general conclusions
from our results. One example of a big problem is that the differences in effect indi­
cated come from studies which are only 6-8 weeks long, meaning it is altogether a too
short treatment period in which to measure differences in remission frequencies.
77
The review of anTiDePressanTs
9 Decisions
9.1
Preconditions
So that antidepressants which were previously part of the reimbursement scheme shall
be granted continued reimbursement they must fulfil the criteria according to the cur­
rent legislation. As a basis for our decisions we have carried out our own meta-analysis
where we sought to clarify if medicines differed in terms of the proportion of patients
who recover, as well as developed a health economic model to clarify the cost-effective­
ness of medicines used in Swedish healthcare. Both the meta-analysis and the health
economic model concern the treatment of depression.
There are a number of systematic reviews which illustrate the effects and side-effects
for pharmaceutical treatment of depression and anxiety. In supplements to those we
have carried out our own systematic reviews of the literature, a meta-analysis with the
purpose of illustrating differences in effect between the medicines and an own health
economic model to shed light on the socioeconomic effects from the treatment of
depression.
Medicines used to treat depression have come into existence during a longer period of
time, explaining why the views of, and demands made on, clinical and health economic
analyses have had time to change.
The absence of a certain type of clinical or health economic data can therefore not
make up the only basis for which to exclude a medicine from the pharmaceutical reim­
bursement scheme.
In decisions made on which medicines will continue to be part of the reimbursement
scheme we must place considerable weighting on the need for treatment alternatives.
As indicated earlier less than half of the patients who reach their treatment objectives
did so with the medicine with which they had commenced treatment. According to
current recommendations the treatment should, if no special circumstances indicate
otherwise, be commenced with an SSRI. Other medicines may also be an alterna­
tive depending on how the patient can be expected to tolerate side-effects, any other
concurrent conditions or ongoing treatments. Therefore, there is a need for various
pharmaceutical alternatives even when choosing the initial treatment.
Individualisation of pharmaceutical choices can also be done based on the character
of the depression and which side-effects the patient can be expected to be more or
78
The review of anTiDePressanTs
less sensitive to. A medicine with a sedative effect can be of advantage for patients
suffering from agitation or sleeping difficulties, for other patients this effect can be a
disadvantage. How the patient values antidepressant effects contra sexual side-effects
varies, probably depending on life situation.
There is a need for pharmaceutical alternatives not least for the patients who do not
respond to the initial treatment. Patients who do not respond to treatment with an
SSRI can, as shown in the STAR*D study, receive an effect from another SSRI. Apply­
ing a restriction within the group of SSRIs to only one SSRI therefore does not appear
to be possible. The study also shows that switching to a medicine within another class
of antidepressants can give an effect.
In the report from North Carolina and in the comments on the Star*D study it is sta­
ted that today it is not possible to predict how a patient will respond to treatment with
a certain medicine based on clinical, demographic or genetic characteristics. Due to a
high proportion of patients not responding to treatment with one medicine there are
many patients who try many medicines before they find a working treatment alterna­
tive.
Therefore, there is a need for treatment alternatives in the form of range of alternatives
both outside and within the group of antidepressants.
To a great extent pharmaceutical treatment of depression takes place in out-patient
care and primarily in primary care. The assessment for continued reimbursement and
decisions to this effect for antidepressants are for this reason made from a pronoun­
ced out-patient care perspective, which is reflected in the health economic model.
Due to this, but also for other reasons, the material on specialist care used for making
decisions is not as extensive as for the primary care medicines.
9.2
Pricing corridor
Within a pharmaceutical group there are sometimes a number of medicines which
have a similar clinical effect for the average patient. This is the case for large number
of medicines against depression. A narrow interpretation of the cost-effectiveness
principle would mean that many medicines would lose their reimbursement at their
current prices. We do however see that there is reason for the price to vary and there­
fore reason for a pricing corridor.
One reason for this is that within many therapeutic areas there is a need for a range of
medicines to choose from. People can respond differently to a medicine in regard to
effect and side-effects.
79
The review of anTiDePressanTs
A second reason can be that some medicines, which have very similar properties and
completely similar treatment effects, have certain unique properties which can be
of value to a smaller group of patients. Such minor differences are often difficult to
measure and carry out a health economic analysis on. For that reason we cannot expect
to receive any documentation to this effect. A pricing corridor can create space for and
encourage the utilization of such small differences. The size of the pricing corridor
reflects the value we believe a range of choices has within each therapeutic area.
9.2.1
The pricing corridor does not constitute an absolute price ceiling
For this review we are also using a pricing corridor. However we accept a higher price
for a medicine if the company marketing it can show that the medicine is cost-effective
at its higher price. A medicine which is significantly better than others can therefore be
allowed a price over the pricing corridor.
The need for a range of medicines varies between therapeutic groups, meaning that
there is reason to let the size of the pricing corridor vary. It is however not possible to
utilize a fixed formula or calculation method to decide how large the pricing corridor
shall be within the various groups. The size is determined based on a collated evalua­
tion of the following:
1) how great the utility is of having more than one medicine available within the
group in question
2) how important the differences are in the unique properties displayed by the medicines
3) how large a pricing corridor is needed to maintain price competition within the
substitutable groups within the framework for generic substitution.
The TLV has previously presented the results of four reviews: medicines used for trea­
ting migraines, hypertension, excess stomach acid, and asthma, coughing and COPD.
In the review of medicines used for treating excess stomach acid we used a pricing cor­
ridor for the first time.
9.2.2
Arguments for a pricing corridor for antidepressants
For antidepressants we judge the need for a range of medicines to be large. This is also
true in regard to other therapeutic areas. Many patients do not achieve their treatment
objectives with the initial medicine. At the level of the individual patients can respond
very differently to treatments with the same medicine.
In our estimation differences are large in areas such as side-effects between antide­
pressants. Even though the conclusion is that the effects achieved with most of these
medicines are similar at a group level, it is clear that they are different. The medicines
have different modes of action, have differences in approved indications between them
and so on.
80
The review of anTiDePressanTs
For pharmaceutical substances where generics are available there is however no reason
to grant continued reimbursement for products where the prices are significantly hig­
her than the prices for generic products. In our estimation there is a need for at least
two producers of generics in order to maintain healthy price competition in the gene­
ric substitution system. The reason for this is that there is seldom price competition
within the substitution group if it is only the original brand holder and one generic
producer who are supplying the medicines.
For the pharmaceutical substances in question there are a number of products avai­
lable for under 3 Skr. This is the highest price we have been able to accept for medici­
nes which are exposed to generic competition.
Above we have detailed the three criteria we have for introducing a pricing corridor.
When it comes to treatment of depression we have stated there is such a great need for
treatment alternatives that, with one exception, there is no need to question the cur­
rent treatment recommendations or use of the current alternatives.
9.3
9.3.1
Body of knowledge
SSRIs
The SSRI medicines appear to be, on both a national and international level, the
first-line choice for the treatment of depression. The material base we use – literature
reviews and own meta-analyses – indicate there are no great differences in terms of
effect between either the SSRIs or antidepressants as a whole. Fluvoxamine appears,
however, in our meta-analysis as an exception, with a considerably lower proportion
of recovered patients than for the other antidepressants. Our results coincide with a
newly published compilation, based on the material used for the registration of more
recent antidepressants [105]. In a newly published meta-analysis the authors do find
however that fluvoxamine has an effect which is similar to that produced by other anti­
depressants [106]. The frequency of recovered patients calculated for fluvoxamine – 29
percent – matches closely however that in our meta-analysis – 27 percent – which could
indicate that the frequencies in the studies included were low in general.
Escitalopram was the drug in our meta-analysis which displayed the highest propor­
tion of recovered patients from all of the antidepressants. That escitalopram should
have an advantage in terms of effect was found in the review carried out by University
of North Carolina, UNC, and in Norway.
SSRI medicines displayed side-effects mainly in the form of nausea, decreased appetite,
diarrhoea, anxiety, agitation and difficulty sleeping and decreased sexual ability. Ser­
traline has displayed a higher frequency of cases of diarrhoea and paroxetine a higher
frequency of weight increase than other SSRIs. In our meta-analysis of ceasing par­
81
The review of anTiDePressanTs
ticipation in a study due to side-effects sertraline was amongst those with the lowest proportion of this with a point estimation of under 6 percent.
When we used remission frequencies from our meta-analysis in our health economic analyses escitalopram dominates all other antidepressants. In an analysis limited to the four SSRIs for which generic alternatives exist (fluoxetine, citalopram, paroxetine, sertraline) there is in practice no difference between the treat­
ments.
9.3.2
SNRIs
Venlafaxine and duloxetine are two medicines included in the SNRIs group. Venlafax­
ine has been around for a long time on the market and there is comprehensive clinical
documentation on the product. Duloxetine is a fairly new medicine and the clinical
documentation is due to this limited.
In our meta-analysis venlafaxine had a point estimate for proportion of recovered
patients which was lower than for escitalopram but higher than for other SSRIs. The
SBU find venlafaxine interesting but the advantages in terms of effect compared to
more selective substances has appeared only at higher doses in a number of studies
and with an increase in the side-effects potential as a result. However, we do not
discern this result in our meta-analysis. NICE does not find any difference in effect
between venlafaxine and other antidepressants, while UNC notes a difference to the
advantage of venlafaxine in comparison to fluoxetine.
Venlafaxine was one of the three treatment alternatives for patients not receiving any
effect from the first initial treatment (citalopram) in the STAR*D study. Here venla­
faxine gave a numerically better treatment result than the other two (bupropion and
sertraline).
Both the SBU and UNC note that venlafaxine results in a higher frequency of nausea
and vomiting than the SSRI substances. Also the SBU and NICE note that venlafaxine
can cause an increase in blood pressure. In our meta-analysis of the proportion of pa­
tients who ceased treatment due to side-effects the point estimate for venlafaxine was
just over 10 percent.
In Swedish treatment practice venlafaxine is used primarily when a first treatment has
not given sufficient effect. When we analyse this use of venlafaxine in our health eco­
nomic model then it is this medicine which gives the greatest effect (QALYs) but which
also has the highest treatment cost. The cost per QALY, 51 000 Skr, may be considered
low.
Duloxetine is a new medicine and is for this reason not included in the reviews carried
82
The review of anTiDePressanTs
out by the SBU and NICE, but it is in the review carried out by UNC. Some key dif­
ferences in effect between duloxetine and other antidepressants are not noted. In our
meta-analysis the point estimate for proportion of recovered patients for duloxetine
lies just under that for venlafaxine if we observe all study lengths, but just over if we
observe studies with 8–12 weeks follow up time. Therefore we have no data showing
that the antidepressant effect for the products differs.
When it comes to side-effects the UNC notes that no decisive differences between du­
loxetine and other second generation antidepressants were observed.
There are however differences between venlafaxine and duloxetine which should be
noted. Venlafaxine has been approved and is best documented for long-term relapse
prevention. Duloxetine has been approved for the treatment of painful diabetes
neuropathy and in some studies has shown an effect also for pain in association with
depression.
In Swedish treatment practice Duloxetine should be seen as an alternative to venlafax­
ine for patients with depression who have not achieved an acceptable treatment result
in the first treatment. There does not appear to be any decisive differences between
venlafaxine and duloxetine in terms of effect and side-effects, however there is in terms
of treatment cost. The cost of treatment for venlafaxine is today considerably higher,
but venlafaxine will become available next year as a generic and we can therefore expect
that the cost of treatment will become lower than for duloxetine.
9.3.3
MAO-inhibitors
Moclobemide is the only medicine in the group of MAO-inhibitors and should there­
fore be retained in the reimbursement scheme. The studies included in the SBU’s base
documentation are small and mostly do not show any differences in effect either in
comparison with placebo or an active substance. According to NICE however the effect
in the treatment of depression is similar for the other antidepressant pharmaceuticals
(SSRIs and TCAs). The SBU has not reported any documentation or commented on
moclobemide in regard to side-effects.
NICE’s report is succinct and has found that patients tolerate treatment using moclo­
bemide to the same extent as SSRIs. The probability that patients will cease treatment
is lower than for TCAs.
9.3.4
NRIs
Reboxetine is the only medicine in the group of NRIs and should for this reason be
retained in the pharmaceutical benefits scheme. The SBU finds it difficult to evaluate
reboxetine in relation to other antidepressants as the published documentation is limi­
ted. Based on a lesser number of studies NICE finds that reboxetine has the same effect
83
The review of anTiDePressanTs
as other antidepressants in the treatment of depression. Reboxetine was not part of the
review by UNC. In our meta-analysis the proportion of recovered patients for rebox­
etine is at the level of the SSRIs.
Both the SBU and NICE find that they do have the documentation to evaluate rebox­
etine in relation to the other antidepressants in terms of side-effects.
9.3.5
Alpha-2 antagonists
Two medicines are part of the group called alpha-2 antagonists, mianserine and mir­
tazapine. Mianserine, which can result in a serious side-effect, bone marrow depres­
sion, shall according to the indication text only be used in cases where treatment with
traditional antidepressants have given rise to unacceptable side-effects.
Mirtazapine is included in the SBU’s review but the documentation is limited and no
position is taken on the effect of mirtazapine in relation to other antidepressants.
NICE finds in its review that mirtazapine gives a higher proportion of recovered pa­
tients and to a greater degree decreases the symptoms of depression. The conclusion is
that despite this mirtazapine does not differ in terms of effect from other antidepres­
sants. But it can have an advantage in the form of fewer patients who cease treatment
due to side-effects. The UNC finds that mirtazapine has a faster onset than other anti­
depressants but that as studies are completed do not discern any difference in effect.
In our meta-analysis the point estimate for the proportion of patients in remission is
amongst the highest.
In terms of side-effects the UNC notes that comparatively more patients who have
been treated with mirtazapine go up in weight but that fewer are afflicted with sexual
dysfunction.
9.3.6
TCAs
The SBU states that clomipramine and amitriptyline have a somewhat larger effect
than the SSRIs for more severe depressions and depression being treated at a hospital.
The differences in the side-effects profile indicate however greater compliance with
treatment for SSRIs. NICE finds that amitriptyline can have a marginally better effect
than other antidepressants but that this advantage is lost due to an inferior side-effec­
ts profile in turn causing a greater proportion of ceased treatments. This also applies
for other TCAs. The UNC has not included TCAs in its review.
Usage of TCAs is overwhelmingly confined to clomipramine and amitriptyline. Only
part of this use is however in regard to the treatment of depression. The medicines are
established in the treatment of neuropathic pain. Usage of other TCAs is very restric­
ted.
84
The review of anTiDePressanTs
Table 21: The TLV's result from the review of medicines against depression, in summary:
Type of medicine
ATC-code
Origional brand drug
& any generic
Continued
reimbursement
Tca – tricyclics
n06aa04
anafranil + generics
all
n06aa06
surmontil
all
n06aa09
saroten, Tryptizol
all
n06aa10
sensaval
all
n06aa21
ludiomil + generics
all
n06aB03
fontex + generics
only generics
n06aB04
cipramil + generics
all
n06aB05
seroxat + generics
only generics
n06aB06
Zoloft+ generics
all
n06aB08
fevarin
No substances
n06aB10
cipralex
all
mao-monoamine oxidase inhibitors
n06ag02
aurorix + generics
all
alpha 2 antagonists
n06aX03
Tolvon
all
n06aX11
remeron + generics
only generics
n06aX16
efexor
all
n06aX21
cymbalta
all
n06aX18
edronax
all
ssri – selective serotonin reuptake inhibitors
snri – serotonin norephrine reuptake
inhibitors
nri – norepinephrine reuptake inhibitors
9.4
Decisions
All pharmaceutical substances – excluding fluvoxamine – within the group of antide­
pressants will remain within the reimbursement scheme. However, not all products or
all strengths and package sizes of a product will be retained (see Table 21).
Antidepressants for the most part offer a cost-effective alternative in the treatment of
depression and anxiety syndrome.
As many patients do not reach a satisfactory result from treatment with one medicine,
there is an unavoidable need for treatment alternatives. This need applies for both
alternatives in the form of different classes of medicines as well as in the form of dif­
ferent medicines within a class.
Medicines in other groups than SSRIs are not cost-effective as a first line treatment
other than in specific exceptional cases where for example a certain side-effects profile
is desired.
85
The review of anTiDePressanTs
The SSRI medicine fluvoxamine has an effect which is not better than the effect gai­
ned from other SSRIs. At the same time the price for fluvoxamine is higher than the
cheapest generic alternatives for SSRI medicines where the patent has expired. Due to
this fluvoxamine cannot be considered cost-effective for an average patient. There is
of course a need for a number of medicines within the group of SSRIs, but this need
is well met by the various SSRIs which will continue to be part of the pharmaceutical
reimbursement system.
For the pharmaceutical substances where generics are available there is no reason to
continue to reimburse products whose prices are considerably over the price for gene­
rics if these products sell for a comprehensive amount within the benefits system. The
pharmaceutical substances here are:
 N06AB03 - fluoxetine
 N06AB04 - citalopram
 N06AB05 - paroxetine
 N06AB06 - sertraline
 N06AG02 - moclobemide
 N06AX11 - mirtazapine
In order to establish a pricing corridor for these pharmaceutical substances we have
examined the average prices for the most sold strengths in package sizes of 100 tablets
or comparable during the period October 2007– March 2008. The results, presented in
the Appendix Base material for on the pricing corridor, have led to the highest price
per tablet which the TLV accepts is 3 Skr (AUP) in the strength and package size we
have indicated as a reference.
Generics are available for a further three substances:
 N06AA04 - clomipramine
 N06AA21 - maprotiline
 N06AX03 - mianserine
The difference in price between the original brand drug and the generic is however
small and there is therefore no reason to deny continued reimbursement status due to
price differences.
Some of the antidepressants are prescribed within the reimbursement system in other
dosage forms than tablets and capsules such as mixtures and injection solutions, and
are generally used for patients with special needs when it comes to care. The use of
these other dosage forms is very limited. We judge the need for having these alternati­
ves available for the patients concerned here to be so great that these dosage forms may
be allowed remain in the reimbursement system without further evaluation.
86
The review of anTiDePressanTs
9.4.1 Strengths and packages
In accordance with what was applied in the review of hypertension medicines for other
strengths and package sizes the following applies:
1.1
same strength in larger packages – at the most the same price AUP per tablet as the refe­
rence
1.2
same strength in smaller packages – at the most the same price AUP per package as the
reference
2.1
higher strength in the same package size as the reference – at the most the same price AUP
per mg of active substance as the reference
2.2
higher strength in larger package than the reference – at the most the same price AUP per
tablet as in 2.1
2.3
higher strength in smaller package than the reference – at the most the same price AUP per
package as in 2.1
3.1
lower strength in same package size or in smaller package than the reference – at the most
the same price AUP per package as the reference
3.2
lower strength in larger package size – at the most the same price AUP per tablet as the
reference.
9.5
A number of companies decreased their prices prior to
the decision
When we analysed the medicines in this review it was clear that many original brand
drugs were being prescribed at comparatively high prices, despite the availability of
cheaper generic alternative products. The reason for this was that the prescribers to a
large extent resisted the substitution. The result of this is that generic competition has
not become as established as we had expected. It is not reasonable that these medicines
shall continue to be reimbursed, considering that there are similar alternatives avai­
lable at a lower cost. Many of the companies have now lowered their prices for their
medicines in order to retain reimbursement.
This has been achieved through continuous contact with the companies in question.
In the final stage of this we sent out a memorandum for the medicines where we in­
tended to change reimbursement status. If the company wished then we offered them
the opportunity for a deliberation with the TLV’s ruling board. Many companies chose
instead to decrease their prices. The price decreases altogether result in 40 million
Swedish crowns per year being released.
This has meant that most of the medicines can remain in the pharmaceutical reimbur­
sement scheme.
87
The review of anTiDePressanTs
88
ALTERNOVA
Merck NM
Teva Sweden
Sandoz 1
CNSpharma
ACTAVIS
ARROW
Orifarm Generics
H. LUNDBECK
Teva Sweden
Sandoz 1
Sandoz 2
Merck NM
CNSpharma
ALTERNOVA
ACTAVIS
ARROW
Orifarm Generics
H. LUNDBECK
Merck NM
CNSpharma
Sandoz 1
Sandoz 2
ACTAVIS
ARROW
ALTERNOVA
Teva Sweden
Orifarm Generics
H. LUNDBECK
Sandoz 2
Sandoz 1
Teva Sweden
Merck NM
CNSpharma
ACTAVIS
ARROW
ALTERNOVA
Orifarm Generics
H. LUNDBECK
Sandoz 2
Sandoz 1
Merck NM
CNSpharma
ORION PHARMA
ACTAVIS
ARROW
ALTERNOVA
Teva Sweden
Orifarm Generics
H. LUNDBECK
ORION PHARMA
Sandoz 2
Sandoz 1
Merck NM
Teva Sweden
CNSpharma
ACTAVIS
ARROW
ALTERNOVA
Orifarm Generics
H. LUNDBECK
Sandoz capsule
STADAPharm soluble
Teva Sweden capsule
Sandoz soluble
Merck NM capsule
Orifarm Generics soluble
RATIOPHARM capsule
Merck NM soluble
RATIOPHARM soluble
ELI LILLY soluble
Merck NM capsule
Teva Sweden capsule
Sandoz capsule
Orifarm Generics soluble
Sandoz soluble
RATIOPHARM soluble
STADAPharm soluble
RATIOPHARM capsule
Merck NM soluble
SELENA FOURNIER soluble
ELI LILLY soluble
Teva Sweden capsule
Merck NM capsule
Sandoz capsule
Orifarm Generics soluble
STADAPharm soluble
Sandoz soluble
RATIOPHARM capsule
RATIOPHARM soluble
Merck NM soluble
ELI LILLY soluble
Teva Sweden capsule
Sandoz capsule
Sandoz soluble
RATIOPHARM capsule
Orifarm Generics soluble
STADAPharm soluble
Merck NM capsule
RATIOPHARM soluble
Merck NM soluble
ELI LILLY soluble
Teva Sweden capsule
Sandoz capsule
RATIOPHARM soluble
Sandoz soluble
STADAPharm soluble
Orifarm Generics soluble
RATIOPHARM capsule
Merck NM capsule
Merck NM soluble
ELI LILLY soluble
STADAPharm soluble
Sandoz soluble
Teva Sweden capsule
Sandoz capsule
Orifarm Generics soluble
RATIOPHARM soluble
Merck NM capsule
RATIOPHARM capsule
Merck NM soluble
ELI LILLY soluble
The review of anTiDePressanTs
10 Appendix – Base documentation
and decisions
10.1 Appendix – base material for pricing corridor
Number of
tablets sold
Fluoxetine 20 mg 100 st – sold tablets and AUP/tablet
October 2007– March 2008
Tablets sold
Number of
tablets sold
Citalopram 20 mg 98/100 st – sold tablets and AUP/tablet
October 2007– March 2008
Tablets sold
AUP/tablet
500,000
10.0
450,000
9.0
400,000
8.0
350,000
7.0
300,000
6.0
250,000
5.0
200,000
4.0
150,000
3.0
100,000
2.0
50,000
1.0
0
0.0
2007-10
2007-11
AUP/tablet
2007-12
2008-01
2008-02
2008-03
AUP/tablet
3,000,000
9.0
2,500,000
7.5
2,000,000
6.0
1,500,000
4.5
1,000,000
3.0
500,000
1.5
0
0.0
2007-10
2007-11
AUP/tablet
2007-12
2008-01
2008-02
2008-03
89
90
Sandoz 2
Ranbaxy Pharma
ACTAVIS
Sandoz 3
Sandoz 1
IVAX 2
Teva Sweden
Merck NM
KRKA Sverige
Orifarm Generics
Parallel Pharma
PARANOVA
MEDARTUUM
IVAX 1
PFIZER
STADAPharm
ACTAVIS
Sandoz 1
Sandoz 3
Sandoz 2
IVAX 2
Teva Sweden
Ranbaxy Pharma
Merck NM
KRKA Sverige
Orifarm Generics
Parallel Pharma
PARANOVA
MEDARTUUM
IVAX 1
PFIZER
Ranbaxy Pharma
Sandoz 3
ACTAVIS
AVENTIS PHARMA
Merck NM
Sandoz 2
Teva Sweden
IVAX 2
KRKA Sverige
Orifarm Generics
Parallel Pharma
PARANOVA
MEDARTUUM
IVAX 1
PFIZER
ALTERNOVA
Sandoz 1
Sandoz 2
Sandoz 3
ACTAVIS
Ranbaxy Pharma
Teva Sweden
Merck NM
KRKA Sverige
Orifarm Generics
Parallel Pharma
PARANOVA
MEDARTUUM
IVAX 1
PFIZER
Ranbaxy Pharma
ALTERNOVA
Sandoz 3
Sandoz 2
ACTAVIS
Sandoz 1
Merck NM
Teva Sweden
KRKA Sverige
Orifarm Generics
Parallel Pharma
PARANOVA
MEDARTUUM
IVAX 1
PFIZER
Ranbaxy Pharma
Sandoz 2
Sandoz 3
ACTAVIS
Merck NM
Teva Sweden
ALTERNOVA
KRKA Sverige
Orifarm Generics
Parallel Pharma
PARANOVA
MEDARTUUM
IVAX 1
PFIZER
Sandoz 1
Sandoz 2
Tablets sold
2007-10
Tablets sold
Number of
tablets sold
AUP/tablet
2007-11
2007-12
2008-01
2008-02
Sertraline 50 mg 100 st – sold tablets and AUP/tablet
October 2007– March 2008
GLAXOSMITHKLINE
Paroxetine 20 mg 100 st – sold tablets and AUP/tablet
October 2007– March 2008
STADAPharm
Sandoz 1
Merck NM
Sandoz 2
RATIOPHARM
ACTAVIS
GLAXOSMITHKLINE
RATIOPHARM
Sandoz 2
Sandoz 1
Merck NM
STADAPharm
ACTAVIS
GLAXOSMITHKLINE
RATIOPHARM
STADAPharm
Merck NM
Sandoz 1
Sandoz 2
GLAXOSMITHKLINE
Merck NM
STADAPharm
RATIOPHARM
Sandoz 1
Sandoz 2
GLAXOSMITHKLINE
RATIOPHARM
Merck NM
Number of
tablets sold
Sandoz 1
Sandoz 2
STADAPharm
GLAXOSMITHKLINE
RATIOPHARM
Merck NM
STADAPharm
The review of anTiDePressanTs
AUP/tablet
900,000
9.0
800,000
8.0
700,000
7.0
600,000
6.0
500,000
5.0
400,000
4.0
300,000
3.0
200,000
2.0
100,000
1.0
0
0.0
2008-03
AUP/tablet
AUP/tablet
2,000,000
10.0
1,800,000
9.0
1,600,000
8.0
1,400,000
7.0
1,200,000
6.0
1,000,000
5.0
800,000
4.0
600,000
3.0
400,000
2.0
200,000
1.0
0
0.0
2007-10
2007-11
2007-12
2008-01
2008-02
2008-03
2007-10
Tablets sold
2007-11
2007-12
2008-01
2008-02
ARROW
Sandoz
STADAPharm
Sandoz
KRKA Sverige
Teva Sweden
ALTERNOVA
RATIOPHARM
Teva Sweden
MEDARTUUM
ORGANON
STADAPharm
Sandoz
Sandoz
Teva Sweden
KRKA Sverige
ALTERNOVA
ARROW
RATIOPHARM
Teva Sweden
MEDARTUUM
ORGANON
Sandoz
Sandoz
Teva Sweden
ALTERNOVA
STADAPharm
ARROW
KRKA Sverige
RATIOPHARM
MEDARTUUM
Teva Sweden
ORGANON
Sandoz
Teva Sweden
ALTERNOVA
KRKA Sverige
ARROW
RATIOPHARM
STADAPharm
Teva Sweden
ORGANON
KRKA Sverige
STADAPharm
Sandoz
ALTERNOVA
Teva Sweden
ARROW
RATIOPHARM
ACTAVIS
Teva Sweden
PARANOVA
ORGANON
Mirtazapine 30 mg 96/100 st – sold tablets and AUP/tablet
October 2007– March 2008
ROCHE
Moklobemid 150 mg 100 st – sold tablets and AUP/tablet
October 2007– March 2008
ACTAVIS
ALTERNOVA
ROCHE
ALTERNOVA
ACTAVIS
ROCHE
ALTERNOVA
ACTAVIS
ROCHE
ALTERNOVA
ACTAVIS
Number of
tablets sold
ROCHE
Tablets sold
ALTERNOVA
ARROW
Sandoz
Sandoz
ALTERNOVA
Teva Sweden
KRKA Sverige
RATIOPHARM
STADAPharm
Teva Sweden
MEDARTUUM
ORGANON
Number of
tablets sold
ACTAVIS
ROCHE
ALTERNOVA
ACTAVIS
The review of anTiDePressanTs
AUP/tablet
600,000
12.0
500,000
10.0
400,000
8.0
300,000
6.0
200,000
4.0
100,000
2.0
0
0.0
2007-10
2007-11
2007-12
2008-01
2008-02
2008-03
AUP/tablet
AUP/tablet
50,000
5.0
40,000
4.0
30,000
3.0
20,000
2.0
10,000
1.0
0
0.0
2008-03
AUP/tablet
91
The review of anTiDePressanTs
11 Appendix meta-analysis
11.1 Meta-analysis – included studies
Table 1. Studies included in the systematic overview.
Study
schatzberg et al 2006
rudolph & feiger 1999
Tylee 1997
alves 1999
nemeroff 2007
De nayer 2002
costa e silva 1998
Kornaat 1998
s332 (unpublished)
rudolph et al 1998
s606 (unpublished)
s102 (unpublished)
Dierick et al 1996
Keller et al 2007
mehtonen 2002
silverstone 1999
stevens 1997
Tzanakaki 2000
cantillon and Daley 2000
clerc et al 1994
gagiano 1993
Tignol 1993
geretsegger 1994
dewilde 1993
hong et al 2003
versiani et al 2005
amini et al 2005
wheatley et al 1998
Kasper et al 2005
goldstein et al 2002
Beasley 1993
Patris 1996
Ballus 2000
mcPartlin 1998
s349
casabona et al 2002
salinas 1997
Dufour 2001
s632
shelton et al 2006
sir 2005
92
Medicines compared
medic. 1
medic. 2
Setting
medicine 1
medicine 2
N
rem.
N
rem.
flouxetine
flouxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
flouxetine
flouxetine
flouxetine
flouxetine
flouxetine
flouxetine
flouxetine
flouxetine
flouxetine
fluoxetine
flouxetine
flouxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
fluoxetine
flouxetine
flouxetine
venlafaxine ir
venlafaxine Xr
venlafaxine ir
venlafaxine er
venlafaxine er
venlafaxine er
venlafaxine ir
venlafaxine
venlafaxine
venlafaxine
venlafaxine
venlafaxine ir
venlafaxine ir
venlafaxine ir
venlafaxine ir
venlafaxine ir
venlafaxine
venlafaxine ir
venlafaxine ir
venlafaxine er
venlafaxine er
venlafaxine ir
venlafaxine er
venlafaxine er
venlafaxine er
venlafaxine ir
venlafaxine ir
venlafaxine ir
venlafaxine ir
Paroxetine
Paroxetine
Paroxetine
Paroxetine
mirtazapine
mirtazapine
mirtazapine
mirtazapine
escitalopram
Duloxetine
imipramine
citalopram
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
sertraline
sertraline
99
103
170
47
101
67
186
77
23
146
63
26
157
266
50
119
114
54
99
34
45
87
52
50
59
147
18
63
164
33
56
184
41
183
75
57
161
173
40
76
79
20
23
58
19
28
27
112
14
10
49
37
12
71
132
28
33
26
19
17
9
28
42
9
19
16
61
4
16
49
10
12
109
23
98
25
18
82
78
18
37
43
93
95
171
40
96
64
196
79
24
144
64
28
145
781
50
122
102
55
91
33
45
89
54
49
60
145
18
60
170
66
62
173
43
178
80
52
80
180
45
82
79
25
35
60
20
31
38
118
19
9
62
44
14
75
380
30
37
26
22
24
19
26
47
11
22
21
58
7
14
68
28
21
115
14
93
26
20
30
76
18
31
47
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
inpatient
inpatient
inpatient
outpatient
inpatient
Both
outpatient
inpatient
Both
Both
Both
outpatient
inpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
inpatient
outpatient
outpatient
The review of anTiDePressanTs
Definition
Dur.
Dos
Land
remission
stud.
medic. 1
medic. 2
hamD<=7
hamD21<=7
maDrs<=6
hamD<=8
hamD-17<=7
hamD21<=8
hamD17<=8
hamD21<7
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD21<7
hamD17<=7
hamD17<=7
maDrs<=12
maDrs<=12
maDrs<=11
maDrs<=12
hamD-17<=7
hamD-17<=7
hamD-17<=7
hamD-17<=7
maDrs<=12
hamD<=7
hamD<=7
maDrs<=12.
hamD21<=8
hamD<7*
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD17<=7
hamD<=7
hamD<=7
8
8
12
12
6
12
8
8
6
6
6
8
8
10
10
12
12
6
6
6
6
6
6
6
6
8
6
6
8
8
6
8
12
12
8
8
8
12
8
8
8
20-60
20-60
20
20-40
20-60
20-40
20-40
20-40
20-40
20-60
20
20
20
20-60
20-40
20-60
20-40
20-60
20-80
40
20-60
20
20-60
20-60
20-40
20-40
20
20-40
20
20
20-80
20
75-150
75
75-150
75-150
75/150
75-150
75-150
75-225
75-225
75-225
75-225
75
75-150
75-225
75-150
75-150
75-225
75-225
75-375
75
75
75-150
75-399
75-150
75-225
75-150
75-225
75-375
200
20-40
20
20-40
20-40
15-45
15-60
30
15-60
10
40-120
75-300
20
20-40
20
20-40
20-40
20
20-40
20-40
50-150
50-150
Severity
usa
usa
europe
usa
usa
canada
s amer.
Disk.
> 30
%
no
no
no
no
no
no
no
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
moderate
wyeth
wyeth
gsK
Pfizer
wyeth
wyeth
wyeth
whyet
europe
no
mod/sev
wyeth
mod/sev
mod/sev
mod/sev
mDD**
wyeth
wyeth
wyeth
wyeth
no
no
no
no
no
no
no
no
no
no
no
no
no
no
no
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
severe
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
wyeth
gsK
gsK
gsK
gsK
organon
organon
nr
organon
wyeth
lilly
lilly
lundbeck
wyeth
whyet
europe
europe
no
europe
europe
usa
no
Turk./austral. no
mod/sev
wyeth
wyeth
lundbeck
h.lundbeck a/s
Pfizer
Pfizer
europe
europe
europe
europe
europe
s africa
europe
europe
europe
china
europe
iran
europe
usa
europe
austral.
no
no
Sponsor
inclusion
severe
mod/sev
mod/sev
mod/sev
93
The review of anTiDePressanTs
Study
Medicines compared
medic. 1
medicine 1
medicine 2
N
rem.
N
rem.
mehtonen 2000
s402
s414
Bielski et al 2004
montgomery et al 2004
lenox-smith et al 2001
guelfi et al 2001
gentil 2000
venlafaxine
venlafaxine er
venlafaxine er
venlafaxine
venlafaxine
venlafaxine er
venlafaxine
venlafaxine
sertraline
sertraline
sertraline
escitalopram
escitalopram
citalopram
mirtazapine
amitriptyline
75
287
288
100
142
193
75
57
40
116
89
31
99
66
21
33
72
288
294
98
146
198
77
59
27
96
95
35
102
56
29
32
outpatient
inpatient
inpatient
outpatient
outpatient
inpatient
inpatient
outpatient
Perahia et al 2008
venlafaxine Xr
Duloxetine
330
116
318
100
outpatient
hacket et al 1998
hmaT study group a
Perahia et al 2006
goldstein et al 2004
Detke et al 2004
lee 2007
Benkert 2000 et al
schatzberg et al 2002
wade et al 2003
Boulenger et al 2006
Baldwin et al 2006
Danish university 1990
arminen 1994
29060/056/uk gsk
moon & vince 1996
not publ. ser-chn-1
Åberg-wistedt
yoshimura 2007
laurelle 1991
nierenberg et al 2007
Jonas 2006
wade et al 2007
Khan 2007
colonna et al 2005
moore et al 2005
lepola et al 2003
ventura et al 2007
Tignol et al 1998
venlafaxine ir
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
Paroxetine
escitalopram
escitalopram
escitalopram
escitalopram
escitalopram
escitalopram
escitalopram
escitalopram
milnicipran
mirtazapine
(oD)
sertraline
sertraline
citalopram
citalopram
fluvoxamine
venlafaxine
fluvoxamine
Duloxetine
Duloxetine
Duloxetine
Duloxetine
Duloxetine
mirtazapine
mirtazapine
mirtazapine
escitalopram
escitalopram
clomipramine
imipramine
Dothiepin
lofepramine
amitriptyline
sertraline
milnacipran
maprotilin
Duloxetine
Duloxetine
Duloxetine
Duloxetine
citalopram
citalopram
citalopram
sertraline
imipramine
77
87
97
84
85
240
123
120
84
223
156
62
25
59
60
113
177
21
28
274
136
141
136
165
138
155
104
112
38
31
42
31
37
121
42
34
21
114
96
12
11
33
33
54
101
12
10
88
54
74
56
91
75
81
51
36
34
81
93
86
93
238
127
126
93
228
165
56
32
62
62
118
176
21
32
273
126
146
126
174
142
159
107
107
12
23
41
43
47
117
52
48
25
143
93
26
12
32
32
40
91
10
11
101
45
70
44
78
61
68
57
38
inpatient
outpatient
outpatient
outpatient
outpatient
outpatient
Both
inpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
outpatient
Both
sertraline
171
76
168
73
Both
nortriptylin
imipramine
fluvoxamine
amitriptyline
imipramine
reboxetine
55
426
108
179
68
87
25
175
15
96
10
21
54
209
109
186
70
80
26
82
9
99
16
20
outpatient
outpatient
Both
outpatient
inpatient
Behnke et al 2003
wisner et al 2006
Keller 1998
haffmans 1996
Kyle et al 1998
Birkenhäger 2004
schwartz (2002)
*maDrs critera also available
** melancholia
94
medic. 2
Setting
outpatient
outpatient
outpatient
outpatient
Both
inpatient
inpatient
outpatient
The review of anTiDePressanTs
Definition
Dur.
Dos
remission
stud.
medic. 1
medic. 2
Land
hamD<=10
hamD17<=7
hamD17<=7
maDrs<=12
hamD17<=7
hamD-17<=7
hamD21<=7
8
10
10
8
8
12
8
8
75-150
75-300
75-300
75-225
75-150
75-300
75-375
75-150
50-100
50-200
50-200
10-20
10-20
20-60
15-60
50-150
hamD<=7
6
60-120
150-225
hamD17<=7
hamD<=7
hamD<=7
hamD<=7
hamD<=7
hamD17<=7
hamD-17<=7
hamD-17<=7
hamD-17<=8
maDrs<=12
maDrs<=12
hamD17<=7
hamD17<=7
maDrs<=12
maDrs<=12
hamD decr
maDrs<7
hamD<=7
maDrs<=12
hamD<=7
hamD<=7*
hamD<=7*
hamD17<=7
maDrs<=12
maDrs<12
maDrs<=12
hamD<=7*
hamD<=7
6
8
8
8
8
8
6
8
8
12
8
6
12
6
6
6
8
8
6
8
8
8
8
8
8
8
8
8
75/150
20
20
20
20
20
20-40
30-40
20-30
40
20-40
30
20-40
20
20-30
20-30
20-40
10-40
20-40
10
10-20
20
10, 20
10
20
10-20
10
100
100/200
80
80
80
80
60
30-45
30-45
30-45
20
10-20
150
100-200
75
140-210
150
50-150
25-150
50-150
60
60
60
60
20
40
20-40
50-200
100
hamD-17<=7
8
30-45
hamD-17<=7
hamD17<=7
hamD17<=7
maDrs<=12
hamD17<=7
hamD-17<=8
8
12
6
8
6
8
50-200
50-200
30-40
20-40
150-1800
225-375
europe
europe
europe
usa
europa
Japan
Brazil
eur./us/aus­
tral.
europe
usa
europe
usa
europe
usa
europe
europe
eur./canada
europe
Disk.
> 30
%
no
Severity
inclusion
mod/sev
mod/sev
mild/mod
mod/sev
mod/sev
mod/sev
wyeth
wyeth
wyeth
forest
no
no
mod/sev
wyeth
no
mod/sev
lilly
no
no
no
no
no
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
lilly
lilly
lilly
lilly
no
no
no
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
no
no
n.r.
europe
eur./s amer.
europe
europe
europe
usa
europe
no
no
no
no
no
no
no
no
no
no
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
severe
mod/sev
severe
mod/sev
mod/sev
mod/sev
organon
organon
organon
wyeth
organon
Duag
organon
gsK
gsK
gsK
Pfizer
wyeth
gsK
lilly
forest
lundbeck
organon
lundbeck
lundbeck
h.lundbeck a/s
forest
lundbeck
50-150
europe
no
mod/sev
Pierre fabre
25-150
100-300
150-200
50-100
75-450
8-10
usa
europe
europe
europe
europe
no
no
no
no
no
mod/sev
mod/sev
mod/sev
mod/sev
mod/sev
severe
Pfizer
wyeth
lundbeck
lundbeck
solvay-Duphar
europe
europe
china
europe
europe
europe
usa
no
no
no
no
no
no
no
95
The review of anTiDePressanTs
11.2 Meta-analysis – sensitivity analysis
Sensitivity analyses in the meta-analyses where studies with clomipramine were exclu­
ded, studies with shorter follow-ups shorter than 8 weeks were excluded, only studies
longer than 8 weeks on out-patients were included, 4 duloxetine studied with wrong
dosage were excluded, studies carried out outside Europe were excluded and only stu­
des with flexible dosage were included.
Table 2. Sensitivity analyses in the meta-analysis
Medicine
All
studies
All studies
minus clo­
mipramine
8-12 weeks
Outpatient
8-12 weeks
All studies
European
minus 80 mg
studies
duloxetineX
Flexible
dose
fluoxetine
0.4021
0.4054
0.42420
0.43430
0.4015
0.40680
0.38310
venlafaxine
0.4568
0.4591
0.46530
0.47800
0.4560
0.45630
0.42260
paroxetine
0.4270
0.4287
0.43470
0.44920
0.4292
0.42060
0.40280
mirtazapine
0.4508
0.4525
0.45730
0.47660
0.4518
0.41740
0.42640
escitalopram
0.4756
0.4774
0.49030
0.50660
0.4736
0.53340
0.42850
duloxetine
0.4499
0.4517
0.46650
0.48070
0.4454
0.47230
imipramine
0.4237
0.4253
0.41550
0.43500
0.4240
0.39850
0.37610
citalopram
0.4050
0.4063
0.42440
0.44500
0.4034
0.45620
0.36320
sertraline
0.4302
0.4312
0.43360
0.45120
0.4299
0.40660
0.39140
amitriptyline
0.3840
0.3846
0.44170
0.45780
0.3833
0.46140
0.37930
fluvoxamine
0.2677
0.2674
0.2689
0.27810
0.21540
klomipramine
0.6673
0.6725
0.66940
dothiepine
0.4187
0.4190
0.4201
0.41060
lofepramine
0.4245
0.4257
0.4288
0.41490
0.40260
milniciprane
0.3888
0.3883
0.3868
0.36550
0.39110
maprotiline
0.3976
0.3931
0.4015
0.39550
0.38430
nortriptyline
0.4691
0.4688
0.46910
reboxetine
0.4316
0.4356
0.43000
0.37800
0.42170
0.48110
0.4701
0.43120
0.4361
0.40730
4 studies excluded due to too high doses of duloxetine
X
Excluding the clomipramine study has no effect on the result. And excluding the
6-week studies makes the point estimates for effect increase and also if studies for inj­
patients are removed from the analysis then the estimates for effect increase further.
Removing the four duloxetine studied has no effect. If we only include the European
studies then the point estimate increases significantly for, for example, amitrytilin,
citalopram and escitalopram. If we only include studies with a flexible dose then all
point estimates decrease. Then, for example, there is no longer any difference between
escitalopram, mirtazapine and venlafaxine.
In many HTA circumstances a reduction of more than 30 percent in any arm is used as
an exclusion criterion. If we exclude these studies then the level of some point estima­
96
The review of anTiDePressanTs
tes changes, but escitalopram (whose point estimate does not change) still appears to
be the most effective medicine.
Table 3. Results from a meta-analysis of all studies with a reduction
of < 30 percent
Pharmaceutical
substance
Remission rate
Confidence interval
flouxetine
0.4484
0.4030
0.4945
venlafaxine
0.4959
0.4368
0.5565
paroxetine
0.4859
0.4141
0.5581
mirtazapine
0.4850
0.3923
0.5776
escitalopram
0.5200
0.4506
0.5882
duloxetine
0.4743
0.4000
0.5504
imipramine
0.4508
0.3397
0.5678
citalopram
0.4583
0.3816
0.5363
sertraline
0.4651
0.3849
0.5466
amitriptyline
0.3968
0.2885
0.5127
fluvoxamine
0.3047
0.1867
0.4429
dothiepine
0.4669
0.2936
0.6442
lofepramine
0.4731
0.3009
0.6495
maprotiline
0.4378
0.2209
0.6711
97
The review of anTiDePressanTs
12 Appendix – review of health
economic literature in regard to
antidepressants
The SBU’s systematic review of the health economic literature on depression became
the departure point for our literature review[1]. The SBU searched literature from
1975 and onwards until 2001. The search terms used by the SBU were: ”depression”,
”depressive disorder”, ”economics”, ”health economics”, ”economic evaluation”, ”cost­
minimization”, ”cost-effectiveness”, ”cost-utility” and ”cost-benefit”.
We have augmented this with a search covering the time from 2002 up until October
2006. We searched the PubMed and Cochrane databases with combinations of search
terms ”depressivedisorder” [MeSH], ”antidepressive agents” [MeSH], ”cost-benefit
analysis” [MeSH],”cost-effectiveness”, ”economic evaluation”, ”cost-utility”, “pharma­
coeconomic”, “health economic”. We also used the names of all of the various pharma­
ceutical substances included in our review as search terms. Finally, the reference lists
submitted by the companies due to the review were searched by hand.
We applied rather wide exclusion criteria without any geographic limitations. A health
economic evaluation includes many factors which are country-specific and the result
is often not applicable to other countries, but also explicit criteria were used here to
analyse the relevance of the analysis at a later stage, while also wanting to follow the
inclusion criteria applied by the SBU as much as possible. That the SBU in part used
other criteria is because the SBU has a different purpose and another breadth to it. In
order to be included the analysis should contain an economic evaluation, where costs
are compared to effects for one or more of the substances included in the review and
compared to an alternative treatment. All types of evaluations are included such as
cost-effectiveness, cost-utility, cost-revenue and cost minimisation analyses. In order to
take a position on the relevance of the studies we set a minimum requirement that the
studies detailed:
 which costs were included
 the time horizon uses
 the measure of effect used
 the model or calculation used
 the patient population the study was in relation to
98
The review of anTiDePressanTs
In order to judge the relevance of the studies we have used the TLV’s general guide­
lines for economic evaluations as a starting point. In practice not all of the points in
the general advice given are equally important and we have for this reason judged the
relevance of the studies in relation to:
 comparator alternative
 country
 perspective
 time horizon
The purpose of this is to judge to which degree the analysis reflects the cost-effecti­
veness in the care which a patient gets/could get in Sweden today. We have applied
a three-tiered scale for relevance: ”high relevance” (for example an analysis with a
relevant comparator alternative, Swedish, current data from a socio-economic perspec­
tive where the time horizon is relevant for the course of the diseases and treatment),
”medium relevance” (for example a study with a socio-economic perspective on a Eu­
ropean country) and ”low relevance” (for example an analysis carried with placebo as a
comparator alternative). If the comparator alternative or perspective is wrong then the
relevance is set at the most as “moderate”. Studies based on conditions in other Nordic
countries may be judged to have a high relevance, while studies in the developing
world may be judged to have a low relevance. Studies based on conditions in the USA
can at the highest receive a grading of medium-high relevance (see Table 28 for more).
Table 1. Evaluating relevance of health economic studies.
Factor
comparator alternative
country
Perspectiv
Time horizon
relevant
irrelevant
nordic area
europe + industrialised countries (excepting usa)
usa
Developing countries
societal perspective
other perspective
adequate
not adequate
Highest relevance:
high
low
high
high
moderate
low
high
moderate
high
moderate
Sometimes there are differences between studies which cannot be described with the
parameters which have been discussed so far and this means that the evaluation of the
relevance of studies to a certain extent will be based on the experience and understan­
ding of the evaluators.
In Table 29 an overview is given of the studies which have been included over and
above those already identified by the SBU. A more exact review of these studies is
available in an appendix which can be downloaded from the TLV website or may be
ordered from the TLV.
99
The review of anTiDePressanTs
Table 2. Health economic studies covered.
Study
Browne [88]
Comparison
sertraline-iTP-comb.
Result
comb>sert.>iTP
chisholm [86]
Tca-ssri-short-term psychotherapy-comb
PT/Tca or PT/ssri-proactive collaborative
care w/ Tca or ssri
ssri>Tca (proactive w/ssri vs Tca had icer
on i$15 000)
Dardennes [78]
milna as relapse prevention treatment,
medical follow-up without medicines
milna>no pharmaceuticals
Demyttenaere [66]
fernandez [67]
francois, 2002 [64]
francois, 2003 [68]
cita-escita.-venla.
escita-venla
cita-escita-venla-fluox
cita-escita-venla-fluox
escita=venla>cita
escita>venla
escita>venla>cita>fluox
escita>venla>cita>fluox
greenhalgh [107]
ecT-Tca-ssri-snri various strategies
med. strategy 1:ecT; 2:ssri; 3:lithium and ssri
as maintenance treatment had highest net benefit
haby [84]
hemels 2004a [69]
hemels 2004b [70]
Kulp [65]
lenox-smith [108]
löthgren [63]
cBT-ssri
cita-escita
cita-escita
escita-venla
venla-ssri-ami
cita-escita.-venla.
cBT>ssri
escita>cita
escita>cita
escita>venla
venla>ssri>ami
escita>venla>cita
mclaughlin [109]
cita-sertra
sertra>cita
miller [87]
Psychotherapy – antidepressants
antidepressants > Psychotherapy
nice [24]
cBT – aD - combination
combination > aD
aD > cBT
nuijten [79]
fluvox as maintenance treatment – Tca as
maintenance treatment
fluvox>Tca
nuijten [82]
maintenance treatment aD – continued
treatment for 9 monthsr
maintenance treatment aD > continued treatment
for 9 months
o’connor [75]
sertra-placebo
sertr>placebo
Patel [85]
Peveler [58]
Polsky
romeo [59]
scott [83]
serrano-Blanco [56]
fluox-psychotherapy-placebo
ssri-Tca-lofe
Parox – sertra - fluox
fluox-mirta
cogn. therapy as complement to aD
fluox-imi
fluox> placebo >psychotherapy
ssri>lofe>>Tca
Parox = sertra >fluox
mirta>fluox
not cogn. therapy as complement
imi>fluox
sobocki [76]
venla as relapse preventive treatment, no
preventive treatment
venla>no preventive treatment
sullivan [71]
Trivedi [61]
van Baardewijk [74]
cita-escita-venla-fluox-parox-sertra
venla-ssri
Dulox-venla
escita>cita
venla>ssri
venla>dulox
vos [57]
cBT-Tca-ssri-maintenance treatment with
Tca- ssri -KBT
Bibliotherapy>cBT>Tca>ssri for acute treatment
wade 2005a [72]
wade 2005b [73]
wan [60]
cita-escita
cita-escita-venla.
venla-ssri
escita>cita
escita=venla>cita
venla>ssri
aD=antidepressants, ami= amitryptiline, cita= citalopram, dulox=duloxetine, escita=escitalopram, fluox= fluoxetine,
imi=imipramine, parox=paroxetine, PT=psychotherapy, cBT=cognitive behavioural therapy, iTP=interpersonal therapy,
100
The review of anTiDePressanTs
Country
canada
Relevance
low
Comment
no soc. persp.
The world's who regions,
including western europe
low
no soc. persp.
france
low
Belgium
multi
sweden
norway
moderate
moderate
low
moderate
uK
low
australia
austria
austria
Deutschland
uK
sweden
low
moderate
moderate
low
low
high
usa
low
Data from the nineties, no
soc. persp.
uK
low
12 months naturalistic
rcT, no soc. persp.
uK
moderate
france
moderate
The netherlands
moderate
evaluates various
strategies
usa
low
no soc. persp.
cv-patienter
india
uK
usa
uK
uK
spain
low
moderate
low
moderate
moderate
high
sweden
low
wrong comparator
alternative
usa
usa
canada
low
low
high
no soc. persp.
no soc. persp.
no soc. persp.
australia
moderate
no soc. persp.
uK
uK
usa
moderate
moderate
low
severe depression
no soc. persp.
severe depression.
no soc. persp.
children and youths
severe depression
no soc. persp.
no soc. persp.
no soc. persp.
no soc. persp.
claims data
fluvox=fluvoxamine, mirta= mirtazapine, sert=sertraline, lofe=lofepramine,
ecT=electroconvulsive treatment, soc. persp. = socioeconomic perspective
101
The review of anTiDePressanTs
102
The review of anTiDePressanTs
List of references
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
SBU, Behandling av depressionssjukdomar. 2004: Stockholm.
Keller, M.B., Past, present, and future directions for defining optimal treatment outcome in
depression: remission and beyond. Jama, 2003. 289(23): p. 3152-60.
SBU, Behandling av ångestsyndrom. 2005: Stockholm.
Papakostas, G.I., et al., Quality of life assessments in major depressive disorder: a review of
the literature. Gen Hosp Psychiatry, 2004. 26(1): p. 13-7.
Hirschfeld, R.M., et al., Social functioning in depression: a review. J Clin Psychiatry,
2000. 61(4): p. 268-75.
Pincus, H.A. and A.R. Pettit, The societal costs of chronic major depression. J Clin
Psychiatry, 2001. 62 Suppl 6: p. 5-9.
Harris, E.C. and B. Barraclough, Excess mortality of mental disorder. Br J Psychiatry,
1998. 173: p. 11-53.
Wulsin, L.R., G.E. Vaillant, and V.E. Wells, A systematic review of the mortality of
depression. Psychosom Med, 1999. 61(1): p. 6-17.
Socialstyrelsen, E.C., Dödsorsaker 2005, in Rapporter från Socialstyrelsen 2007.
Guze, S.B. and E. Robins, Suicide and primary affective disorders. Br J Psychiatry, 1970.
117(539): p. 437-8.
Nierenberg, A.A., S.M. Gray, and L.D. Grandin, Mood disorders and suicide. J Clin
Psychiatry, 2001. 62 Suppl 25: p. 27-30.
Cerps, C., Assessing preference-based HRQoL in mood disorders and anxiety – a literature
review, in Department of Public Health Sciences, Division of Social Medicine. 2006, Karo­
linska Institutet: Solna, Sweden.
Wells, K.B. and C.D. Sherbourne, Functioning and utility for current health of patients
with depression or chronic medical conditions in managed, primary care practices. Arch Gen
Psychiatry, 1999. 56(10): p. 897-904.
Revicki, D.A. and M. Wood, Patient-assigned health state utilities for depression-related
outcomes: differences by depression severity and antidepressant medications. J Affect Disord,
1998. 48(1): p. 25-36.
Sobocki, P., et al., Health-related quality of life measured with EQ-5D in patients treated for
depression in primary care. Value Health, 2007. 10(2): p. 153-60.
von Knorring, L., et al., Cost of depression: effect of adherence and treatment response.
Eur Psychiatry, 2006. 21(6): p. 349-54.
Sobocki, P., et al., Cost of depression in Europe. J Ment Health Policy Econ, 2006. 9(2):
p. 87-98.
Andlin-Sobocki, P. and H.U. Wittchen, Cost of anxiety disorders in Europe. Eur J Neu­
rol, 2005. 12 Suppl 1: p. 39-44.
103
The review of anTiDePressanTs
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
104
Socialstyrelsen, Antidepressiva läkemedel vid psykisk ohälsa. Studier av praxis i
primärvården, in Uppföljning och utvärdering. 2006.
Wang, P.S., et al., Use of mental health services for anxiety, mood, and substance disorders in
17 countries in the WHO world mental health surveys. Lancet, 2007. 370(9590):
p. 841-50.
Bergdahl, E., et al., Depression among the oldest old: the Umea 85+ study. Int Psycho­
geriatr, 2005. 17(4): p. 557-75.
Palsson, S.P., S. Ostling, and I. Skoog, The incidence of first-onset depression in a popula­
tion followed from the age of 70 to 85. Psychol Med, 2001. 31(7): p. 1159-68.
Läkemedelsverket, Farmakoterapi vid unipolär depression hos vuxna och äldre - behand­
lingsrekommendation. Information från Läkemedelsverket, 2004(5:2004).
NICE, Depression: management of depression in primary and secondary care., in Clinical
Guideline. 2004, National Institute for Clinical Excellence: London.
de Maat, S.M., et al., Relative efficacy of psychotherapy and combined therapy in the treat­
ment of depression: a meta-analysis. Eur Psychiatry, 2007. 22(1): p. 1-8.
Dwight-Johnson, M., et al., Treatment preferences among depressed primary care patients.
J Gen Intern Med, 2000. 15(8): p. 527-34.
Zimmerman, M., M.A. Posternak, and I. Chelminski, Derivation of a definition of
remission on the Montgomery-Asberg depression rating scale corresponding to the definition of
remission on the Hamilton rating scale for depression. J Psychiatr Res, 2004. 38(6):
p. 577-82.
Zimmerman, M., M.A. Posternak, and I. Chelminski, Implications of using different
cut-offs on symptom severity scales to define remission from depression. Int Clin Psycho­
pharmacol, 2004. 19(4): p. 215-20.
Sobocki, P., et al., The mission is remission: health economic consequences of achieving full
remission with antidepressant treatment for depression. Int J Clin Pract, 2006. 60(7):
p. 791-8.
Gartlehner, G., et al., Comparative Effectiveness of Second-Generation Antidepressants in
the Pharmacologic Treatment of Adult Depression. Comparative 2007, (Prepared by RTI
International-University of North Carolina Evidence-based Practice Center un­
der Contract No. 290-02-0016.) Rockville, MD: Agency for Healthcare Research
and Quality.
Hansen, R., et al., Efficacy and Safety of Second-Generation Antidepressants in the Treat­
ment of Major Depressive Disorder. Annals of Internal Medicine, 2005. 143: p. 415­
426.
Helsetjenesten, N.K.f., Effekt og sikkerhet for SSRI og andre nyere antidepressive legemid­
ler ved depresjon hos voksne (pdf, 1MB), in Rapport fra Kunnskapssenteret. 2007.
CCOHTA, Selectiv serotonin reuptake inhibitors (SSRIs) för major depression. Part I: Evalu­
ation of clinical litterature. 1997.
Ruhe, H.G., et al., Switching antidepressants after a first selective serotonin reuptake in­
hibitor in major depressive disorder: a systematic review. J Clin Psychiatry, 2006. 67(12):
p. 1836-55.
The review of anTiDePressanTs
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
Kennedy, S.H., et al., Clinical guidelines for the treatment of depressive disorders. IV. Medi­
cations and other biological treatments. Can J Psychiatry, 2001. 46 Suppl 1: p. 38S-58S.
Barbui, C., et al. Depression in adults (drug and other physical treatments). BMJ Clinical
Evidence 2007 [cited 2007; 1003].
Wilson, K.C., et al., Older community residents with depression: long-term treatment with
sertraline. Randomised, double-blind, placebo-controlled study. Br J Psychiatry, 2003. 182:
p. 492-7.
Reifler, B.V., Play it again, Sam--depression is recurring. N Engl J Med, 2006. 354(11):
p. 1189-90.
Reynolds, C.F., 3rd, et al., Maintenance treatment of major depression in old age. N Engl
J Med, 2006. 354(11): p. 1130-8.
Turner, E.H., et al., Selective publication of antidepressant trials and its influence on appar­
ent efficacy. N Engl J Med, 2008. 358(3): p. 252-60.
Machado, M., et al., Remission, dropouts, and adverse drug reaction rates in major depres­
sive disorder: a meta-analysis of head-to-head trials. Curr Med Res Opin, 2006. 22(9):
p. 1825-37.
Caldwell, D.M., A.E. Ades, and J.P. Higgins, Simultaneous comparison of multiple treat­
ments: combining direct and indirect evidence. Bmj, 2005. 331(7521): p. 897-900.
Ades, A.E., K. Claxton, and M. Sculpher, Evidence synthesis, parameter correlation and
probabilistic sensitivity analysis. Health Econ, 2006. 15(4): p. 373-81.
MacGillivray, S., et al., Efficacy and tolerability of selective serotonin reuptake inhibitors
compared with tricyclic antidepressants in depression treated in primary care: systematic
review and meta-analysis. Bmj, 2003. 326(7397): p. 1014.
Song, F., et al., Selective serotonin reuptake inhibitors: meta-analysis of efficacy and accept­
ability. Bmj, 1993. 306(6879): p. 683-7.
Kumar, S. and M. Oakley-Browne. Panic disorder. BMJ Clinical Evidence 2007
[cited 2007.
Furukawa, T.A., N. Watanabe, and R. Churchill, Combined psychotherapy plus anti­
depressants for panic disorder with or without agoraphobia. Cochrane Database Syst Rev,
2007(1): p. CD004364.
Ipser, J.C., C.M. Kariuki, and D.J. Stein, Pharmacotherapy for social anxiety disorder:
a systematic review. Expert Rev Neurother, 2008. 8(2): p. 235-57.
Soomro, M.G. Obsessive compulsive disorder BMJ Clinical Evidence 2007 [cited
2007; Available from: http://clinicalevidence.bmj.com.
Bisson, J. Post-traumatic stress disorder 2007 [cited 2008; Available from: http://clini­
calevidence.bmj.com.
Stein, D.J., J.C. Ipser, and S. Seedat, Pharmacotherapy for post traumatic stress disorder
(PTSD). Cochrane Database Syst Rev, 2006(1): p. CD002795.
Gale, C. and J. Millichamp. Generalised anxiety disorder. BMJ Clinical Evidence 2007
[cited; Available from: http://clinicalevidence.bmj.com.
Kapczinski, F., et al., Antidepressants for generalized anxiety disorder. Cochrane Data­
base Syst Rev, 2003(2): p. CD003592.
105
The review of anTiDePressanTs
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
106
Hay, P.J. and J. Bacaltchuk. Bulimia nervosa (updated) BMJ Clinical Evidence 2008
[cited 2008; Available from: http://clinicalevidence.bmj.com.
Bacaltchuk, J. and P. Hay, Antidepressants versus placebo for people with bulimia nervosa.
Cochrane Database Syst Rev, 2003(4): p. CD003391.
Serrano-Blanco, A., et al., Effectiveness and cost-effectiveness of antidepressant treatment
in primary health care: a six-month randomised study comparing fluoxetine to imipramine.
J Affect Disord, 2006. 91(2-3): p. 153-63.
Vos, T., et al., Cost-effectiveness of cognitive-behavioural therapy and drug interventions for
major depression. Aust N Z J Psychiatry, 2005. 39(8): p. 683-92.
Peveler, R., et al., A randomised controlled trial to compare the cost-effectiveness of tricyclic
antidepressants, selective serotonin reuptake inhibitors and lofepramine. Health Technol
Assess, 2005. 9(16): p. 1-134, iii.
Romeo, R., et al., The cost-effectiveness of mirtazapine versus paroxetine in treating people
with depression in primary care. Int Clin Psychopharmacol, 2004. 19(3): p. 125-34.
Wan, G.J., et al., Healthcare expenditure in patients treated with venlafaxine or selective se­
rotonin reuptake inhibitors for depression and anxiety. International Journal of Clinical
Practice, 2002. 56(6): p. 434-9.
Trivedi, M.H., et al., Cost and effectiveness of venlafaxine extended-release and selective
serotonin reuptake inhibitors in the acute phase of outpatient treatment for major depressive
disorder. J Clin Psychopharmacol, 2004. 24(5): p. 497-506.
Lennox-Smith, A., et al. A double blind trial of venlafaxine XR vs citalopram in patients
with treatment resistant depression. in British Association for Psychopharmacology meeting.
2001. Harrogate, UK.
Löthgren, M., et al., A cost-effectiveness analysis of escitalopram as first line treatment of
depression in Sweden. Primary Care Psychiatry, 2004. 9(4): p. 153-162.
Francois, C., et al., A Swedish pharmacoeconomic evaluation of escitalopram, a new SSRI:
comparison of cost-effectiveness between escitalopram, citalopram, fluoxetine and venlafaxine
[abstract]. Value in Health, 2002. 5(3): p. 230.
Kulp, W., J.M. von der Schulenburg, and W. Greiner, Cost-effectiveness of outpatient
treatment in depressive patients with escitalopram in Germany. Eur J Health Econ, 2005.
6(4): p. 317-21.
Demyttenaere, K., et al., A cost-effectiveness model of escitalopram, citalopram,and ven­
lafaxine as first-line treatment for major depressive disorder in Belgium. Clin Ther, 2005.
27(1): p. 111-24.
Fernandez, J.L., S. Montgomery, and C. Francois, Evaluation of the cost effectiveness
of escitalopram versus venlafaxine XR in major depressive disorder. Pharmacoeconomics,
2005. 23(2): p. 155-67.
Francois, C., et al., A pharmacoeconomic evaluation of escitalopram, a new selective sero­
tonin reuptake inhibitor. Comparison of cost-effectiveness between escitalopram, citalopram,
fluoxetine, and venlafaxine for the treatment of depression in Norway. The European Jour­
nal of Health Economics, 2003. 4(1): p. 12-9.
The review of anTiDePressanTs
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
Hemels, M.E., et al., Cost-effectiveness analysis of escitalopram: a new SSRI in the first-line
treatment of major depressive disorder in Austria. Curr Med Res Opin, 2004. 20(6):
p. 869-78.
Hemels, M.E., et al., Cost-effectiveness of escitalopram versus citalopram in the treatment of
severe depression. Ann Pharmacother, 2004. 38(6): p. 954-60.
Sullivan, P.W., et al., A comparison of the direct costs and cost effectiveness of serotonin
reuptake inhibitors and associated adverse drug reactions. CNS Drugs, 2004. 18(13):
p. 911-32.
Wade, A.G., I. Toumi, and M.E. Hemels, A pharmacoeconomic evaluation of escitalo­
pram versus citalopram in the treatment of severe depression in the United Kingdom. Clin
Ther, 2005. 27(4): p. 486-96.
Wade, A.G., I. Toumi, and M.E. Hemels, A probabilistic cost-effectiveness analysis of
escitalopram, generic citalopram and venlafaxine as a first-line treatment of major depressive
disorder in the UK. Curr Med Res Opin, 2005. 21(4): p. 631-42.
van Baardewijk, M., P.M. Vis, and T.R. Einarson, Cost effectiveness of duloxetine
compared with venlafaxine-XR in the treatment of major depressive disorder. Curr Med Res
Opin, 2005. 21(8): p. 1271-9.
O’Connor, C.M., A.H. Glassman, and D.J. Harrison, Pharmacoeconomic analysis
of sertraline treatment of depression in patients with unstable angina or a recent myocardial
infarction. J Clin Psychiatry, 2005. 66(3): p. 346-52.
Sobocki, P., et al., The cost-utility of maintenance treatment with venlafaxine in patients
with recurrent major depressive disorder. Int J Clin Pract, 2008. 62(4): p. 623-32.
Kamlet, M.S., et al., Cost utility analysis of maintenance treatment for recurrent depression.
Control Clin Trials, 1995. 16(1): p. 17-40.
Dardennes, R.M., et al., Economic assessment of a maintenance treatment strategy in
prevention of recurrent depressive disorder. Value Health, 2000. 3(1): p. 40-7.
Nuijten, M., et al., Cost effectiveness of fluvoxamine in the treatment of recurrent depression
in France. Pharmacoeconomics, 1998. 14(4): p. 433-45.
Nuijten, M.J., M. Hardens, and E. Souetre, A Markov process analysis comparing the
cost effectiveness of maintenance therapy with citalopram versus standard therapy in major
depression. Pharmacoeconomics, 1995. 8(2): p. 159-68.
Hatziandreu, E.J., et al., Cost utility of maintenance treatment of recurrent depression with
sertraline versus episodic treatment with dothiepin. Pharmacoeconomics, 1994. 5(3):
p. 249-68.
Nuijten, M.J., Assessment of clinical guidelines for continuation treatment in major depres­
sion. Value Health, 2001. 4(4): p. 281-94.
Scott, J., et al., Use of cognitive therapy for relapse prevention in chronic depression. Cost­
effectiveness study. Br J Psychiatry, 2003. 182: p. 221-7.
Haby, M.M., et al., Cost-effectiveness of cognitive behavioural therapy and selective serotonin
reuptake inhibitors for major depression in children and adolescents. Aust N Z J Psychiatry,
2004. 38(8): p. 579-91.
107
The review of anTiDePressanTs
85.
Patel, V., et al., Efficacy and cost-effectiveness of drug and psychological treatments for com­
mon mental disorders in general health care in Goa, India: a randomised, controlled trial.
Lancet, 2003. 361(9351): p. 33-9.
86. Chisholm, D., et al., Reducing the global burden of depression: population-level analysis of
intervention cost-effectiveness in 14 world regions. Br J Psychiatry, 2004. 184: p. 393-403.
87. Miller, P., et al., Counseling versus antidepressant therapy for the treatment of mild to mod­
erate depression in primary care: economic analysis. Int J Technol Assess Health Care,
2003. 19(1): p. 80-90.
88. Browne, G., et al., Sertraline and/or interpersonal psychotherapy for patients with dysthymic
disorder in primary care: 6-month comparison with longitudinal 2-year follow-up of effective­
ness and costs. J Affect Disord, 2002. 68(2-3): p. 317-30.
89. Baker, C.B., et al., Quantitative analysis of sponsorship bias in economic studies of antide­
pressants. The British Journal of Psychiatry, 2003. 183: p. 498-506.
90. Melander, H., et al., Evidence b(i)ased medicine--selective reporting from studies spon­
sored by pharmaceutical industry: review of studies in new drug applications. BMJ, 2003.
326(7400): p. 1171-1173.
91. Bolin, K. and L. Jacobsson, Depression & ångest: de samhällsekonomiska kostnaderna.
Pfizer health economic working Paper 1, 2001.
92. Norinder, A., S. Nordling, and L. Häggström, Priset per tablett dålig indikator i ett
samhällsekonomiskt perspektiv. Läkartidningen, 2000. 97(14): p. 1693-8, 1700.
93. Usman Iqbal, S. and M. Prashker, Pharmacoeconomic evaluation of antidepressants A critical appraisal of methods. Pharmacoeconomics, 2005. 23(6): p. 595-506.
94. Marchand, A., et al., Analysis of the cost and the effectiveness of a psychotherapy for panic
disorder with agoraphobia (PDA) versus a treatment combining pharmacotherapy and psycho­
therapy. Sante Ment Que, 2004. 29(2): p. 201-20.
95. Katon, W., et al., Incremental cost-effectiveness of a collaborative care intervention for panic
disorder. Psychol Med, 2006. 36(3): p. 353-63.
96. NICE, Obsessive-compulsive disorder: Core interventions in the treatment of obsessive-compulsive
disorder and body dysmorphic disorder. 2005, National Institute of Clinical Excellence.
97. Guest, J.F., J. Russ, and A. Lenox-Smith, Cost-effectiveness of venlafaxine XL compared
with diazepam in the treatment of generalised anxiety disorder in the United Kingdom. Eur J
Health Econ, 2005.
98. Panzer, P.E., et al., Implications of an SSRI generic step therapy pharmacy benefit design: an
economic model in anxiety disorders. Am J Manag Care, 2005. 11(12 Suppl): p. S370-9.
99. Sheehan, D.V., et al., Differences in total medical costs across the SSRIs for the treatment of
depression and anxiety. Am J Manag Care, 2005. 11(12 Suppl): p. S354-61.
100. Jorgensen, T.R., et al., Cost-effectiveness analysis of escitalopram compared with paroxetine
in treatment of generalized anxiety disorder in the United Kingdom. Ann Pharmacother,
2006. 40(10): p. 1752-8.
101. NICE, Eating Disorders - Core interventions in the treatment and management of anorexia
nervosa, bulimia nervosa and related eating disorders. 2004, National Institute of Clini­
cal Excellence: London.
108
The review of anTiDePressanTs
102. Gilchrist, G. and J. Gunn, Observational studies of depression in primary care: what do we
know? BMC Fam Pract, 2007. 8: p. 28.
103. Keller, M.B., Long-term treatment strategies in affective disorders. Psychopharmacol
Bull, 2002. 36 Suppl 2: p. 36-48.
104. Khan, A., et al., Symptom reduction and suicide risk in patients treated with placebo in anti­
depressant clinical trials: a replication analysis of the Food and Drug Administration Database.
Int J Neuropsychopharmacol, 2001. 4(2): p. 113-8.
105. Melander, H., et al., A regulatory Apologia--a review of placebo-controlled studies in regula­
tory submissions of new-generation antidepressants. Eur Neuropsychopharmacol, 2008.
18(9): p. 623-7.
106. Omori, I.M., et al., Efficacy, tolerability and side-effect profile of fluvoxamine for major
depression: meta-analysis. J Psychopharmacol, 2008.
107. Greenhalgh, J., et al., Clinical and cost-effectiveness of electroconvulsive therapy for depres­
sive illness, schizophrenia, catatonia and mania: systematic reviews and economic modelling
studies. Health Technol Assess, 2005. 9(9): p. 1-156, iii-iv.
108. Lenox-Smith, A., P. Conway, and C. Knight, Cost effectiveness of representatives of three
classes of antidepressants used in major depression in the UK. Pharmacoeconomics, 2004.
22(5): p. 311-9.
109. McLaughlin, T.P., M.T. Eaddy, and A.N. Grudzinski, A claims analysis comparing
citalopram with sertraline as initial pharmacotherapy for a new episode of depression: impact
on depression-related treatment charges. Clinical Therapeutics, 2004. 26(1): p. 115-24.
109
The review of anTiDePressanTs
110
Cost-effective to reimburse a broad range of antidepressants
in this report the TLV states that depression-related illnesses lead to great suffering for the
people afflicted and to large societal costs. According to our analysis it is cost-effective to retain
a broad range of antidepressants in the swedish pharmaceutical reimbursement system. Our
investigation shows that not even every second patient reaches the therapeutic objectives set for
their first treatment. side-effects also lead to patients not completing their treatments. Many
need to try a number of medicines in order to get a good result. We would also like to underline
the need for more effective antidepressants and more knowledge for the better applications of
these medicines.
Our decisions mean that all medicines bar one will continue to be reimbursed. The substance
fluvoxamine loses its reimbursement status as, in comparison to other similar medicines, it carries a higher price but does not have a better effect. Expensive original brand name drugs, where
there are alternatives in the form of generics, shall no longer be reimbursed. in total 40 million
swedish crowns are released by this review on an annual basis which can be used in other
urgent areas in healthcare.
This report covers the fifth therapeutic group to be evaluated in the reviews
of reimbursed medicines carried out by the TLV. When we received new
rules for the reimbursement of medicines in October 2002 it was not
practically possible to try all medicines overnight against the new
regulations. for this reason the medicines which were reimbursed
in the old system were allowed retain reimbursement pending
review. The TLV is carrying out a review of the approximately
2000 medicines in the high cost threshold to ascertain if
they should be reimbursed or not in the future. The
objective of our work is to extract as much health
as possible per tax crown which is expended on
medicines.
DEnTAL AnD PHARMAcEuTicAL BEnEfiTs AgEncy
P.O. BOx 55 [sunDByBERgsVägEn 1],
sE-171 11 sOLnA, sWEDEn
Phone: +46 8 5684 2050, fax: +46 8 5684 2099
[email protected], www.tlv.se