Comments on appraisal consultation document for the appraisal of drugs for the treatment of Alzheimer's disease.

Transcription

1 Comments on appraisal consultation document for the appraisal of drugs for the treatment of Alzheimer's disease. March 2005 The difference in my life was outstanding. I was given my life back, albeit changed, from sitting zombie like all day I started to be glad to wake up. I saw things in colour again, I could make conversation with my family, I could now choose for myself the right clothes to wear. Such joy, such simple things, all courtesy of two little reminyl tablets a day. 1 of 41

2 Contents Page 0. Summary 3 1. Introduction 4 2. The use of patient/carer evidence 5 3. Conventional pharmacoeconomics cannot be applied to dementia 8 4. Criticisms of the model Incorrect assumptions What the model misses out Lack of transparency in the model AD Cost alone is not a sufficient reason for rejection Implications for dementia if draft guidance is not revised Recommendations 38 The Alzheimer s Society is very grateful for the input of the numerous people with dementia and carers who have helped in the development of this submission. We would also like to thank Professor Sube Banerjee, Professor of Mental Health & Ageing and Clinical Director, South London and Maudsley NHS Trust, London and Pete Stephens at IMS Health for their help and advice. 2 of 41

3 0. Summary The Alzheimer s Society believes that the draft guidance needs to be revised substantially. We recommend that Aricept, Exelon, Reminyl and Ebixa should be made available through the NHS as one part of the treatment of some people with Alzheimer's disease. The Alzheimer s Society believes that the draft guidance is fundamentally flawed because: 1. patient/carer evidence has not been given appropriate weight and decisions are not based on outcomes of value to people with Alzheimer s and their carers. 2. the benefits of the drugs on quality of life for both the person with dementia and their carer have not been captured. 3. the model does not reflect the reality of dementia care (including the vast amount of care provided by unpaid carers). We have set out our concerns under five main sections: 1. the wrong model has been used to calculate cost effectiveness and incorrect assumptions have been made within the model 2. essential factors, such as quality of life and benefits to carers, have been excluded 3. there is a lack of transparency about how decisions have been made on cost effectiveness 4. decisions about drug treatments cannot be made on cost alone 5. our concerns about the implications for dementia if draft guidance is not revised Recommendations In its recommendations, the Alzheimer s Society has set out six principles which any new guidance needs to take into account in order to reflect both the reality of dementia care and the outcomes that people with Alzheimer s and their carers value most highly. 1. Any calculation of cost effectiveness should be based on a measure of quality of life that is appropriate for dementia 2. The benefits that the drugs bring for unpaid carers should be recognised 3. A cost effectiveness model should ensure that accurate costs of dementia care are taken into account 4. Any model used should reflect clinical practice 5. It should be explicit how cost-effectiveness is calculated 6. Patient/carer evidence should be given greater weight 3 of 41

4 1. Introduction Alleviating the burden and distress of Alzheimer s disease has important economic, social and humanitarian implications. Drug treatments play an important role in improving quality of life and promoting independence. Guidance issued by the National Institute of Clinical Excellence (Nice) in 2001 had a positive impact on dementia care. We remain optimistic that new guidance can promote appropriate and equitable access to effective dementia treatments. We are stunned by the recent draft recommendations that none of the drug treatments for Alzheimer s disease should be made available on the NHS. With other consultees, including the Faculty of Old Age Psychiatrists, the Alzheimer s Research Trust and the British Geriatrics Society, we remain convinced that these drugs work. Considerable attention has been paid to the draft guidance. The Society has encouraged those affected to take part in the consultation. We believe that the appraisal committee s initial decision was based on flawed and incomplete information. If the sole benefit of the drugs was simply to delay full time care (FTC) by one to three months, we would not, nor would thousands of individuals, have spent so much effort trying to ensure that the draft guidance is revised. The Alzheimer s Society is eager to work with Nice to find a better way of calculating the cost-effectiveness of drugs for Alzheimer s disease. This paper sets out a number of our proposals. 4 of 41

5 2. The use of patient/carer evidence The views of patients or service users, their carers and the public matter to Nice. By working with patients, carers, patient organisations and the public, Nice aims to produce guidance that addresses patient/carer issues, reflects their views and meets their healthcare needs. i We welcome the Nice s commitment to involving patients and carers in its work. The criteria used to evaluate the clinical and cost-effectiveness of drug treatments must draw on the priorities of patients and carers. Role of the Alzheimer s Society The Alzheimer s Society is the UK s leading care and research charity for people with dementia, their families and carers. Dementia affects over 750,000 people in the UK alone. The Alzheimer s Society provides information and support for those with any form of dementia and their carers. It advises professionals working in the field, funds research, runs quality care services, and campaigns for improved health and social care and greater public understanding of dementia. The Alzheimer s Society has valued the opportunity to participate in the Nice appraisal process and has contributed at every stage of the review. We have: Commented on scope Nominated patient/carer representatives Submitted evidence based on a summary of 4,000 people with dementia and their carers, detailing 18 recommendations ii Commented on the assessment report Commented on additional work on cost-effectiveness. We believe that Nice has an vital role in promoting equal access to drug treatments that work and that its work must reflect the outcomes valued by people with dementia and their carers. User-defined outcomes On medication, I feel more confident, trying to maintain independence. I didn t think I could write but I can. I enjoy the day centre, find it uplifting. User-defined outcomes reveal what patients and carers think about a treatment or service in terms of what they find meaningful for their lives. It is of considerable concern that patient/carer views are rarely included in research design, particularly in the design of clinical trials. While the drugs for Alzheimer s disease were approved for improvements in cognition, the changes 5 of 41

6 most frequently reported and highly valued by those directly affected are the positive changes in behaviour and mood. I am able to enjoy music and, with help, simple outings. I enjoy being able to be part of the family again, albeit more limited, and my self-esteem has improved. Within a few weeks my husband had calmed down. He was happy to be left alone and was more stable and content and easier to be cared for. These may seem like minor achievements in the face of a life-threatening, degenerative illness, but they are the benefits that people with Alzheimer s and their carers value and which help to restore their quality of life. Role of people with dementia in consultation Nice has received many letters written by people with a diagnosis of Alzheimer s disease. Without drug treatments, this would not have been possible. One person with dementia told us: I started taking Exelon nearly three years ago, my life has changed. I now cook, clean, able to hold a conversation, write poems and childrens stories, also write this letter to you, which I couldn t do before taking exelon. The quality of my life is great now thanks to exelon. The introduction of drugs for Alzheimer's and their subsequent approval by Nice has enabled people affected to communicate their views to others. It would be a considerable irony if Nice rejects the very drug treatments that enable people with dementia to take part in the appraisal and review of treatments for dementia. Inaccessibility of consultation document The Society welcomes Nice s efforts to involve patients and carers in the consultation process. Members of the public are encouraged to respond to the appraisal consultation document (ACD). However, we have already voiced our concern that the final document is inaccessible to the general public. People with dementia and their carers have told us that they feel unable to comment on the document owing to its technical nature. Many did not understand how a decision had been reached, so did not feel able to comment. One woman wrote: I have, as you suggested, gone to Nice s website and found the whole thing totally intimidating. I would be interested to know how many carers gird their loins and read through the whole thing. 6 of 41

7 As a result, we requested, on behalf of our members, a lay summary that would help carers and people with dementia understand how cost-effectiveness. Unfortunately this was refused. 7 of 41

8 3. Conventional pharmacoeconomics cannot be applied to dementia Dementia is a complex condition. Models used to determine cost-effectiveness for other conditions do not apply; a different approach to health economics is required. With an illness where the drugs do not appear to increase life expectancy, measuring the value of delaying symptoms or improving quality of life in people with dementia should be one of the highest priorities. iii The complexity of dementia arises from: Numerous factors which impact on quality of life in dementia Care provided by unpaid carers Difficulty in putting a value on certain benefits of the drug treatments: for example, being unable to recognise that the woman in your home is your wife of 50 years Stigma attaching to the condition; the effect of the condition on a person s confidence and self-esteem Wider benefits of the drugs for family and friends. It is clear that long term conditions, particularly those that affect older populations, do not lend themselves well to studies of cost-effectiveness, especially where quality adjusted life years (QALYs) are employed. Nice measures cost-effectiveness by calculating the cost of treatment per quality adjusted life year (QALY). This involves giving a score which ranges from a top score of 1 (representing perfect health) to 0 (which represents death) to different health states such as mild dementia and severe dementia. Nice calculates that mild dementia is equivalent to a score of 0.6 (out of a total of 1) and severe dementia to a score of The cost of a drug treatment that maintains a person in a higher health state for a year is then calculated to give a cost per QALY. Nice has calculated that the cholinesterase inhibitors (Aricept, Exelon and Reminyl) have a cost per QALY of 64,000 to 94,000 and Ebixa a cost per QALY of over 50,000. They conclude that this does not represent good value for money. Nice faced the same problem of how to measure quality of life in 2001, stating: Quality of life may be affected for both patients and carers. At present, there is simply inadequate research available to understand the quality of life impact and thus any economic modelling on this issue ought to be regarded cautiously. iv The appraisal did not therefore include a cost per QALY methodology. Southampton Health Technology Assessment Centre (SHTAC) found a similar lack of research evidence and have been appropriately cautious 8 of 41

9 in the presentation of their model. Therefore, the Society does not understand why Nice have been inconsistent in their approach over time. We would contend that, in the absence of random controlled trial (RCT) evidence, Nice should examine how it can use the experience of Alzheimer's Society members and the evidence provided by over 4,000 submitted questionnaires that informed the Society s 2004 submission. In the absence of RCT evidence, we believe more weight should have been given to the experiences of patients and carers. 9 of 41

10 4. Criticisms of the model In its 2001 appraisal, Nice accepted that there was not enough data to model the cost-effectiveness of the drug treatments. For the current appraisal, it has attempted to create such a model. We recognise Nice s desire for a consistent approach to allow a comparison of different treatments over different disease areas. However, there are clearly too many difficulties in calculating a cost per QALY to make a suitably reliable basis for a decision. The SHTAC model does not adequately capture the complexities of dementia. The data available is far too limited to model cost-effectiveness meaningfully. The calculations produced are highly unstable and do not enable definitive conclusions to be drawn. In this section we address our criticisms of the model. These are based around: Incorrect assumptions within the model Factors that have been excluded Lack of clarity related to how the final cost per QALY has been calculated. 4.1 Incorrect assumptions used to calculate cost-effectiveness The Society believes that the model makes the following incorrect assumptions: Dementia stage is a good proxy of quality of life People in full time care can be given a standard quality of life score Global utility scores can be based on the Neumann study Carer quality of life does not change over time The average cost of full time care is 355 a week All those who start drug treatment will remain on it It is wrong to assume quality of life is inevitably linked to disease severity. Mum is on Aricept and Ebixa her memory is very poor now but she is happy, laughs, jokes, does crossword puzzles. She is content and knows she is loved She is funny, mischievous, and has an interest in life. I didn t expect this at all after 5 ½ years. Although his short term memory was still not good, we were able to return to a normal life, travelling around shopping and visiting friends and family, carrying 10 of 41

11 out normal household tasks, in particular enjoying the company of our three young grandchildren playing with them and watching school and dance performances. In fact we got our life back for eighteen months and at times we could almost forget that he had Alzheimer s. It is clear from the Society s previous two submissions to Nice, and from our systematic survey of over 4,000 people with dementia and their carers, that drug treatments improve quality of life for the people who respond to treatment and their carers. The benefits are individual, but are often substantial. There is a paucity of data on quality of life for people with dementia. SHTAC has tried to overcome this deficiency by using severity of disease as a crude proxy measure to produce a cost per QALY. We believe this to be a flawed approach. Any modelling from the currently published data is highly speculative and hence inaccurate. There is likely to be some inter-relationship between severity of dementia and quality of life. However, given the complexity of factors affecting quality of life, using disease severity as a single proxy measure of quality of life as the basis of the health economic analysis is flawed. Moreover, SHTAC expressed concern that the additional analysis fails to highlight the fact that cognitive function alone is a poor indicator of disease progression and health-related quality of life. We would echo their concern: better cognitive skills do not inevitably mean an increased sense of well-being; it is not necessary to have full cognition to be happy. In other illnesses there may be a direct association between quality of life and a clinical variable such as pain or activity limitation. Where an intervention or disorder does not impact on quality of life (relative to mortality) then it may be appropriate to minimise or omit quality adjustment, v but this is not the case in Alzheimer's disease. Supporting evidence on the complexity of quality of life in dementia: Three sources of research (noted below) highlight the limitations of the model s assumption of a linear relationship between dementia severity and quality of life. 1) A recently completed Health Technology Assessment-funded programme has generated a disease-specific measure of health-related quality of life in dementia with robust psychometric properties. vi The DEMQOL system consists of two interviewer-administered instruments, one completed by the person with dementia and one a proxy report of the person with dementia s quality of life completed by the main carer. The development of this system shows how wide ranging quality of life is in dementia, with qualitative work defining five domains: 11 of 41

12 daily activities and looking after yourself; health and well-being; cognitive functioning; social relationships; and self-concept. In fact, social impairment and behavioural and psychological symptoms appear to be more strongly associated with quality of life than with cognition. Data from the final field test of the DEMQOL system vii confirms that quality of life was statistically significantly correlated with higher levels of behavioural and psychological disturbance (Neuropsychiatric Inventory total score and agitation, depression, anxiety, disinhibition and irritability subscales). Quality of life was not statistically significantly associated with cognition. A regression analysis shows that behavioural problems account for 52 per cent of the variance, patient age with 31 per cent and cognition with only 16 per cent. This data suggests that quality of life in dementia is more complex that the domains commonly covered in trials. Simple proxy substitution of a single measure such as dementia stage cannot adequately capture changes in quality of life. 2) The confirmation that quality of life is correlated with behavioural and psychological disturbance is particularly important in the light of new evidence of a very significant impact on aggression, agitation and irritability with Ebixa. viii Aggression and agitation is commonly treated with off-licence drugs, which we know can do harm. Evidence that Ebixa can reduce symptoms of dementia that do impact on quality of life highlights the importance of the drug being available to all people with dementia in the later stages of dementia. 3) The Alzheimer s Society commissioned a re-analysis of original patient data originally published in the BMJ. ix This strongly contradicts any assumption that there is a close relationship between quality of life and dementia severity. This study of almost 600 people in 17 care facilities in three different regions uses a validated method of evaluating quality of life for people with dementia called Dementia Care Mapping. This method records a rating of well-being every five minutes over a six hour period, arranged to produce a score of between -5 and +5. A new evaluation of the data from these 600 people has demonstrated that quality of life scores varied from -3.9 to This huge range in scores emphasises the variability in quality of life for people with dementia in full time care. It directly challenges the model s assumption that all individuals entering a care home have a standard utility health score (of 0.34 on HRQL scale). New analysis of this original data for a sub-group of 129 patients confirms the inadequacy of the assumption that severity is longitudinally linked with quality of life. In a linear regression analysis to determine the variables contributing to well-being, the main association was the amount of time spent engaged in activities (T=7.2 P<0.0001). Other variables in the model included receptive language function (T=2.4 P=0.02) and the degree of impairment of self-care 12 of 41

13 (BARTHEL score, T=2.9 P=0.004). MMSE and dementia stage did not contribute to the model. New analysis of the data from the 107 people from the above study who were followed up for 12 months confirms that quality of life does not necessarily decline over time. In addition, changes in quality of life over one year were not correlated with changes in MMSE (R=0.12, P=0.28), emphasising in the first longitudinal study the absence of a relationship between cognitive decline and decline in quality of life. Conversely, there was a strong relationship between decreased engagement in activities and a decrease in quality of life and a trend towards an associated between decline in self care and decreased quality of life (R=0.19, P=0.05). In light of new developments in quality of life instruments and increasing evidence that suggests quality of life is not inextricably linked to disease severity, we believe that Nice must acknowledge that their model is flawed and look again at whether a more reliable measure of the cost-effectiveness of the dementia drugs can be developed at this stage. Utility scores cannot be based on the Neumann study The report contains no robust explanation of how quality of life weights/utility weights were calculated. The methods of derivation of those weights available are questionable and therefore susceptible to considerable and unquantifiable error. The utility scores from the Neumann et al studies x,xi,xii need to be treated with extreme care. In the study in which the HUI2 and the HUI3 were given to carers to rate the quality of life of people with dementia they cared for there were marked differences in the results of the two scales. The difference appears to be greater than might be expected, even given the differences between the scales, xii suggesting instability in these estimates. This is not surprising given that neither scale has been psychometrically evaluated in dementia (ie we have no idea of these instruments validity or reliability in dementia) and certainly not for proxy use. Neumann et al (1999) in an earlier, less cited paper on the same sample acknowledge the need for caution and for further research into the use of proxy respondents. xi They write: There are limitations associated with the exercise. First, the HUI has not been validated for using in Alzheimer's disease. Second, since the data were not collected as part of the clinical trial itself, it is unclear how the drug itself affects health-related quality of life, or how health-related quality of life changes with disease progression for individual patients. There is considerable evidence to suggest that models that work in other diseases cannot be assumed to work in complex disease systems such as dementia. 13 of 41

14 It is wrong to assume that carer quality of life does not change over time The appraisal fails adequately to assess carers quality of life. We recognise and appreciate the extra work done by Nice s secretariat, which attempted to assess the impact of the drug treatments on carers quality of life, but we believe the approach they took was flawed and does not reflect carer experience. We do not believe that the appraisal committee can come to a decision using a model that does not properly consider the impact of drug treatments on quality of life of carers, as well as the person with dementia. Nice based its decisions on research that appeared to show no change in carer quality of life over disease progression/time. The carer utility scores were based on the same HRQL measures as the ones used for people with dementia. Most of the categories, such as cognition, seem to have very limited relevance for carers and they were drawn from carers who had no experience of drug treatments. It therefore seems to be an inappropriate tool for assessing how the drug treatments can affect quality of life in carers. We are disappointed that carer quality of life, and the drug treatments impact on it, has not been a focus of research. However, in the absence of this, we believe that greater weight in the evaluation should be given to patient/carer evidence, such as the information available from the Alzheimer's Society survey of 4,000 people with dementia and their carers that was presented to Nice in a previous submission, and which clearly demonstrates the benefits in quality of life experienced by the majority of people. ii Evidence to date strongly contradicts any assumption that carer quality of life does not change with progression of disease. In fact, we believe that a very complex relationship exists between carer quality of life and severity of dementia. I found my father s frustration and anger extremely distressing. He cried in the world he didn t understand. He got lost in his own bungalow and was frightened of his own reflection in the mirror. As his physical condition deteriorated, he became more settled as his confusion worsened and he stopped fighting. Over the years as he deteriorated, I found visiting him much easier, as his mental pain was less. A particularly strong theme that emerges is the appreciation of the positive impact the drug treatments have for relationships with family and friends. We would recommend that a more effective method of calculating carer utility, based on what really matters to carers, is developed and used in the appraisal. 14 of 41

15 The following quotes illustrate the benefits of effective drug treatments for carers. No longer does my heart sink every time the phone rings, my own household has returned to normal and I can now get some relaxation time for my own retirement as I am 62 years old. Since taking Aricept his warm character has returned and friends have remarked that Ron has got his twinkle back. To Nice that twinkle is too expensive, but to me it makes my life of caring worthwhile. From a carer s point of view a contented patient equals a very contented carer (the patient is also my wife of 45 years whom I love dearly). The well documented link between carer depression and the prevalence of behavioural and psychiatric symptoms also challenges the assumption that carer quality of life does not change with increasing dementia severity: these symptoms become much more frequent in moderate and severe Alzheimer's disease. We therefore think that model should allow for substantially more change in carer utility between the different disease stages. I have been able to sleep much more soundly at night, prior to starting the rivastigmine my mother tended to wander around at night as her perception was badly distorted I have not had to cope with quite such severe psychotic episodes. It is wrong to assume that everyone in the later stages of dementia is in a care home Equating dementia severity to full time care is flawed. There are other reasons for entry into full time care apart from severity of dementia. There are many people with mild or moderate dementia in care homes. Admission is dependent on a number of complex factors other than severity of disease, including other key symptoms, the family and social support network, and the local availability of services. For example, a third of people with dementia live alone these people are at a high risk of premature entry into full time care. It is unreasonable to assume that any treatment would have an impact on institutionalisation when major family problems or the deficiencies in available services are the main triggers. It is wrong to assume that the average cost of full time care is 355 a week Within the cost model presented by Nice, the cost estimate ( 355 per week) of full time institutional care is a serious underestimate. The SHTAC model did not include the two most intensive types of care, and is based only upon the 15 of 41

16 proportion that is reimbursed by social services. Available care strata in the UK include: Residential care Specialist (EMI) residential care Nursing home care Specialist (EMI) nursing home care Specialist units for people with specific needs NHS continuing care. The last two categories cost more than 1,000 per week. Care for younger people with dementia may exceed 1,500 per week (see 4.2). Specialist residential or nursing home care is usually required when psychiatric and behavioural symptoms are prominent. As these symptoms are a very frequent reason for the need for full time care, people with dementia often require this level of support from the time of first admission to a care facility. In addition, people with dementia may be moved from less to more specialist facilities when their level of need increases, with approximately 12.5 per cent of care facility residents moving to alternative facilities over a 12-month period (figures from Ballard et al, 2002) xiii. Within the model, over the projected treatment period of five years, it is likely that the majority of people will need a more specialised level of care. In addition, 10 per cent of the people moving care facilities require a period of time in psychiatric in-patient facilities (a standard admission would be four to eight weeks) to stabilise the situation before the transfer to a new care facility can be achieved. We therefore suggest that an estimated cost of 0.5 days per annum of psychiatric in-patient care should be added to the per patient per annum cost. SHTAC responded to our criticism that care costs were too low and confirmed that a more reasonable estimate of 520 per week would decrease cost per QALY by about 10,000. We would recommend that a more reliable estimate of the cost of full time care, which takes into account the full range of care settings (including NHS continuing care), is calculated and used. It is wrong to assume that drug treatment is stopped on entry to full time care The existing Nice guidance does not state that admission to a care home is a necessary reason to discontinue treatment with cholinesterase inhibitors. In some situations, where the treatment does not meet the criteria for ongoing benefit, the treatments may be discontinued prior to full time care. In other people, particularly if the reasons for admission to full time care are social, the 16 of 41

17 treatment may well be considered to be giving ongoing benefit. In addition, there is accumulating evidence (eg Holmes et al, 2004) xiv that withdrawal of cholinesterase inhibitors leads to an exacerbation of behavioural and psychiatric symptoms. Stopping treatment at the stressful time of initial placement in full time care would therefore be poor clinical practice, and a review of treatment after six months, when the individual is more likely to have settled into their full time care placement, would represent better clinical practice. However, the model clearly assumes that anticholinesterase treatment will stop on admission to full time care and therefore needs some revision. It is important that people in care homes are not denied access to effective drugs for dementia since people can derive ongoing benefit from drug treatment. The following quotes illustrate how many people can go on benefiting from the treatment while in full time care: My husband has only been on Ebixa for one week and already the change in his condition has been remarkable. Previously, he was unable to go to the toilet unaided. Now he can. He is in a care home and he was not able to walk by himself. Now he is up and about. I am having to buy it privately because the consultant was not willing to prescribe it on the NHS. Last summer, as result of a nasty fall and a lengthy stay in hospital, she became a resident in a care home, where she remains. The difference between my mum and a lot of the other residents with memory problems is marked. She remains confused but alert. We still work on crosswords together and she comes on outings with us from time to time. When we visit the memory clinic, they are always pleased by the result of her tests. Mum was prescribed Aricept. After a couple of months we saw an improvement, she started to remember little things again, and when she had her assessment her scores had gone up, and that happened every time she was assessed. That was in June 2000 and Mum continued to improve. She was able to continue to live on her own until December 2003 at the age of 94, when she went into a lovely rest home in Woolston. She still recognises her family and still has her dignity. I feel so lucky that I got my mum back and have been able to enjoy the last five years thanks to being prescribed Aricept. Measuring only the cost of delay in entry to full time care clearly fails to measure benefits that people with dementia gain, even while in full time care. 17 of 41

18 It is wrong to assume that that all those who start drug treatment will remain on it The Society, along with other commentators such as the Faculty of Old Age Psychiatrists, expressed concern clinical practice. We are unable to understand why the model has not been revised to reflect this. The 2001 Nice guidance states that a further assessment should be made, usually two to four months after reaching maintenance dose of the drug. Following this assessment, the drug should be continued only where there has been an improvement or no deterioration in MMSE score, together with evidence of global improvement on the basis of behavioural and/or functional assessment. The cost-effectiveness model should reflect current Nice guidance and ensure that when using data from the clinical trial, adjustments are made so that only the individuals who benefit and therefore would continue on the drugs are included in the model. Including only those who respond to the drug treatments would lead to a more reliable model and possibly substantially revised cost estimates. SHTAC responded to this criticism by saying that a responder analysis would result in a less favourable cost-effectiveness profile. We find this assumption counterintuitive and cannot accept it unless a responder analysis is undertaken. The results of clinical trials show a mean difference in ADAS-cog across all subjects in the treatment arm compared to controls. This would include nonresponders to treatment who in clinical practice would be taken off the drug treatments after about three months. Only responders would continue on the drug treatments beyond that point. The mean difference in ADAS-cog between responders and controls would be higher than the mean difference between all those in the treatment arm and controls. The model should be based upon the mean difference between responders and controls. It would also have to incorporate the costs of the non-responders, but these would be relatively low because drug costs would be for three months and there would be a reduction in monitoring costs. SHTAC said they do not report data on clinical effectiveness by responder status and that, if such data were available, its use in cost-effectiveness analysis would be questionable because it is not possible to differentiate between potential responders and non-responders at the decision to prescribe treatment or not. We acknowledge that it is currently impossible to know who will respond to the drug treatments before treatment is commenced, but it is clear in a relatively short space of time who is responding and who is not. Those who do not respond are taken off the drug treatments and those who do respond will have a higher mean difference from controls in ADAS-cog score than that reported by 18 of 41

19 clinical trials, which present the mean difference across responders and nonresponders. The model should be revised to reflect this. 4.2 What the model misses out As we have already made clear, the Alzheimer's Society believes that the model used by Nice is based on incorrect assumptions. We also believe that it misses out a number of key aspects of dementia care and the impact of drug treatments, which we detail in this section. However, it is not simply a matter of inputting these missed factors into the existing model. We believe that Nice needs to look at how, and indeed whether, it can develop a more reliable and comprehensive model. We believe the model fails to address the following factors. Reduction in carer time Reduction in patient/carer costs Improved carer health Savings to pre full time care costs Reduced costs in the prescribing of neuroleptics Use of cholinesterase inhibitors in full time care Improved health utility of professional care workers Extra costs faced by younger people with dementia. The model fails to measure the reduction in carer time/burden Is proper account taken, in weighing up the cost effectiveness of the drugs, of how much we carers save the government? I am on duty night and day for every day of each year. Apart from the 12 hours per week at the day centre, for the remaining 156 hours in the week I care for my wife at home. My wife can only, with my help, get about in a wheelchair and is dependent on me for everything. How does one put a cost on this? Let s call it 5 per hour, not very much, but it still amounts to 780 per week. Unpaid carers provide most of the care for people with dementia. The value of services provided by all carers in has been estimated at 57.4 billion. xv Evidence suggests that drug treatment can reduce carer time. We think there is a strong argument for giving weight within the cost-effectiveness analysis to a reduction in carer time, particularly given the increasing recognition of the importance of supporting carers. 19 of 41

20 Research suggests that, following anticholinesterase drug treatment, carers are saving approximately an hour a day of their time. xvi If we value carer time at the minimum wage, then an hour a day of carer time would be a substantial saving over a year. If the draft guidance is not revised, the NHS will effectively be shifting costs onto carers. Carers will be spending an hour a day more caring for their loved one and the NHS will save the cost of the drug treatment. If the carer's time was considered as one of the benefits what the NHS was buying, the use of the drug treatments could be within its cost per QALY threshold. If the deterioration had been accelerated I am not sure that I would have been able to face up to the immense daily problems. Time has allowed me to adjust my life style and manage more positively our time together. As a carer I am convinced that this would not have been possible without the medication. My mother receives little by way of true recognition of the role she plays in taking care of the majority of my father's needs - something you seem loathe to qualify from a true costings point of view in your calculations were it the NHS that was having to provide said care instead. Do not take advantage of the love and duty carers have for their patient - cost it in! The model fails to measure costs of the person with dementia and their carer All of the costs of my mother s care, both at home, with private carer support in addition to family, and subsequently within the care home, have been met by the family. Over ten years this is a significant sum of money for which no NHS support has been forthcoming. This situation is the norm and the debate about the cost to the NHS of medication should be assessed in this light. One cannot help but wonder if the NHS were paying for the true cost of the care of Alzheimer s sufferers, as opposed to the families of sufferers, the calculation of cost benefit when viewed against the enormous costs of such care would not be different. As Nice acknowledges, a large proportion of costs are borne by people with dementia and their families. We are aware that Nice generally adopts a NHS and social services perspective in cost-effectiveness analysis. However, because of the discrimination in the current funding system, which means many people are paying for their own care, we believe that there is a strong case for a departure from usual practice in this appraisal. Nice has a duty to include the costs and benefits that most fair-minded people would deem appropriate. xvii We would like to see further consideration given to cost savings made by carers by the appraisal committee. 20 of 41

21 We are acutely conscious of the fact that if Nice did factor carer costs into its current model, the cost per QALY may actually rise. This perverse result is likely to occur because the model is solely dependent on the cost savings of a delay to full time care (FTC). Including carer costs would undoubtedly increase the costs of pre FTC and the difference in price between pre FTC and FTC would be smaller, thereby making the drugs appear to be less cost-effective. Such a conclusion can only serve to illustrate the unsuitability of the existing model. An alternative model would reflect the true costs and savings at all stages of the illness, as well as the impact of the drug treatments on quality of life of both the person with dementia and their carer. It would thus produce a more realistic cost per QALY. The model fails to measure the impact on carer health The experience of caring for my father was exhausting, mentally distressing. It negatively affected my health beyond his death and its financial implications will impact on me through to my own old age. Nice acknowledges that health benefits to carers can be considered within an appraisal s scope but does not do so within this appraisal. Although we are disappointed by the lack of direct evidence available from clinical trials, we do know that carers make more visits to GPs than others and that carers of people with dementia experience greater strain, distress and higher levels of psychological morbidity than carers of other older people and are prescribed more drugs. It is likely that effective drug treatment results in better carer health. There is a link between carer depression and the prevalence of behavioural and psychiatric symptoms. xviii A drug that reduces behavioural symptoms is likely to have a positive impact on carer health, and carer NHS costs. We believe that a revised model should consider benefits to carer health and resulting reduced NHS costs. Aricept made my mother calmer and independent again. She became less reliant on me, not phoning times a day with the same question, and was able to reason again. She regained her independence and I felt less depressed and worried about her or me being able to cope with the mental anguish of this horrible disease and the effect this had on me and my family too. The model fails to measure the savings to pre full time care costs His improvement has meant that we have not had to use any outside services (except those of the NHS consultant). 21 of 41

22 My mother was diagnosed with this illness three years ago. She was prescribed Reminyl and this has meant that she has remained at home cared for by my father with the occasional visit to the GP. They have not asked for any additional community services nor do they need them. Although the extra work carried out by Nice corrected the SHTAC model s omission of benefits for people who do not reach FTC within five years, it appears to continue to ignore any reduction in services used by people with dementia and carers in the period prior to FTC. It does not allow for the fact that pre FTC costs may be different between patients who are on treatment and those who are not. No data appears to have been presented to Nice on this; it may be that none yet exists (although we are aware of at least one clinical trial that is collecting such information). In the absence of clinical trial data, we believe greater weight should be given to patient/carer evidence. Alzheimer s Society research ii shows that people on drug treatments appear to use fewer services: Receipt of service No antidementia drug treatment (% of total) Received antidementia drug treatment (% of total) P (significance) Memory clinic 147 (11%) 1116 (42%) <0.001 Day care/day centre 880 (63%) 1480 (55%) <0.001 Day hospital 281 (20%) 414 (16%) <0.001 Home care service 707 (51%) 996 (37%) <0.001 Respite care 712 (51%) 906 (34%) <0.001 Admiral nurse 41 (3%) 79 (3%) NS Community 612 (44%) 1230 (46%) NS psychiatric nurse Carers support group 569 (41%) 1136 (43%) NS Specialist equipment 393 (28%) 571 (21%) <0.001 Legal/financial 281 (20%) 636 (24%) 0.01 advice Care home 599 (43%) 588 (22%) <0.001 Transport 531 (38%) 882 (33%) Other 92 (7%) 211 (8%) NS 22 of 41

23 These data suggest that attendance at a memory clinic makes it more likely you will get treatment, and that those who experience drug treatment will need less access to care homes, respite care, day care and home care. Both groups of patients appear to receive similar levels of support from CPNs and carers support groups. Of the 375 respondents who reported that they received none of the services listed, 307 of respondents had experience of the drugs. The Society firmly believes that drug treatments should be provided within the context of a package of good quality care. However, because evidence from carers and people with dementia suggests that they have had to call on fewer community services as a result of the drug treatments, we believe that a more accurate model of cost effectiveness would consider these savings. Responses to Nice s consultation on the ACD also demonstrate the additional benefits from drug treatment in terms of getting more out of care services. We note that these may well be difficult to incorporate into the model, but they should not be ignored. One carer reported that since her mother had been taking Aricept she was now able to leave the house, get in a taxi and go to the day centre. She really enjoyed these days out. It is not clear how this would be dealt with within a costeffectiveness analysis, but it seems perverse for these benefits to result in the drug looking less cost-effective. It does, however, demonstrate that the drugs can help people who respond to treatment get more out of the services that are available. They may be able to interact better with staff, take part in the activities and enjoy the social aspects of the service. We believe that, even in the absence of research, the appraisal committee should consider these benefits. The model fails to measure the benefits of reduced neuroleptic prescribing The likely consequence of withholding drug treatments is an increased use of neuroleptic drug treatments. There is already an accumulating evidence base to support this clinical observation. A new meta-analysis of US and European data from clinical trials with cholinesterase inhibitors reported a significant decrease in neuroleptic use (25 per cent to 9 per cent). xix Therefore, for the model to better reflect reality, the cost savings of a decrease in neuroleptic prescription should be considered within the appraisal. We have estimated that additional costs of prescribing a neuroleptic such as risperidone is 720 per year, with many people being on such a drug for this length of time or longer. The decreased risk of falls, injuries, other side-effects and accelerated care home placement that is likely to be associated with a reduction in neuroleptic prescribing should also be factored into the appraisal. In addition, there are the 23 of 41

24 quality of life implications of increased levels of neuroleptic, which we deal with in more detail on page 36. The model fails to include the use of cholinesterase inhibitors in full time care The existing Nice guidance does not state that admission to full time care is a reason to discontinue treatment with cholinesterase inhibitors. We have already noted that the model incorrectly assumes that drug treatment will stop on entry into full time care (page 16) and therefore needs revision. This incorrect assumption also means that the costs of full time care do not incorporate the costs of drug treatments. There are no readily available data to quantify the continuing use of cholinesterase inhibitors, but we suggest that it should be assumed that 50 per cent of people continue cholinesterase inhibitor treatment for the first six months of full time care and that these costs are added to the cost estimates of placement in full time care. The model fails to include the health utility of professional care workers Care assistants have a particularly demanding job, which is likely to impact upon their HRQL. For example, a re-evaluation of the data from the Margallo- Lana et al study published in the BMJ, xx the mean carer distress score on the Neuropsychiatric Inventory was 1.6, with 31 per cent of professional carers experiencing at least moderate distress. We therefore feel that it is important to make some consideration of professional care worker health utility. Now she is in a residential EMI unit where there are over 20 residents but only three or four staff, so I can imagine their lives would be far worse if Aricept was withdrawn or not prescribed, since I remember very clearly how it was beforehand. The model needs to consider the extra costs faced by younger people with dementia Husband diagnosed at 53, wife was 45 it would be impossible for us to afford to purchase these tablets privately As to cost myself, I have had to give up my job, I am now unable to contribute to a pension for myself. I struggle to maintain our increasing household costs on a decreasing income and cannot begin to count the cost, financially, physically and mentally to myself. The model does not consider the extra costs of younger people with dementia. People with dementia under 65 years of age may have to pay more for appropriate services and who are more likely to lose income if they have to give up work. There may also be additional costs relating to partners who have to give up work to care for their loved one. 24 of 41

25 Two examples of specialist services for younger people with dementia highlight the true costs of care: Bluebirds typically receives younger people (under 65) needing specialised nursing care when EMI homes can no longer cope. The cost on admission including specialised nursing care would be in the area of 1,500 a week (2005) The Rowans is a specialist centre for younger people with dementia and has an occupied bed day cost of 75,105 per annum or per day ( ). These figures suggest that there is a need for the Nice appraisal to consider the extra costs of full time care for younger people with dementia in order to produce an accurate model of cost effectiveness. Indirect costs are also likely to be greater for younger people with dementia. People with dementia may have to stop working or take on less well paid work. Their partners may also have to give up work to care for them. This is the situation that the following people have found themselves in: This cruel disease doesn't just affect one person it has affected everyone we come into contact with. It has forced an end to my husband s career; it has affected my career quite dramatically. It has changed our standard of living and our lifestyle. It has changed everything we have lived and worked for. The Aricept tablets mean that my husband is able to take part in family life for a bit longer. Perhaps you would be good enough to contact my husband and explain to him in detail why you wish to make his life more difficult for him than it already is and will become in the future. The Alzheimer's Society recommends that the model be reviewed to reflect the exceptional cost incurred for younger people with dementia and their carers. 4.3 Lack of transparency in the model Despite our grave concerns about the appropriateness of the proposed model, we have tried to understand how the model was applied and how the stated costs per QALY were identified. We found this very difficult, particularly because of the lack of information made available to us. We appreciate that there may indeed be good reasons which explain the cost per QALY, but it has not been possible for us to know how the final costs were 25 of 41