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
© Copyright 2024 Paperzz