New drug treatment for Alzheimer’s disease: lessons for healthcare policy

The launch of donepezil (Aricept), a specific treatment for patients with mild or moderate Alzheimer’s disease, attracted intense interest. Clinicians and others were quoted in the media as being optimistic about the drug’s effectiveness, but concerned that NHS funding would be withheld or uneven. However, reaching consensus on the clinical rather than statistical importance of this drug requires open debate, given the relatively small effect sizes and uncertainty over side effects in typical patients. Debate has been hampered because publication of the full results of the main clinical trials has been delayed. This episode highlights several issues of general policy importance that must be resolved if access to the information needed for clinical and other decision making is to be improved.

The launch of donepezil—perceptions and reality

Alzheimer’s disease is a common and devastating condition, and the launch of a specific treatment naturally attracted intense interest. The new drug, donepezil (Aricept), an anticholinesterase agent, was licensed in the United States in December 1996, with reports that it had produced “highly significant improvements in cognitive and clinical global assessments” in randomised trials lasting 30 weeks and had increased the proportion of “treatment successes” by 245%.¹

The drug’s launch three months later in the United Kingdom was greeted with optimism. The lay press quoted eminent clinicians as saying that donepezil “marks a sea change” in management of dementia,² that it should be seen as “a way to alleviate the burden of a terrible disease,” and that it would “give hope to many people and their carers.”³ Several clinicians voiced fears about funding, including a comment that “we must make sure health authorities do not try to hold back.”² One respected journalist even suggested that inability to fund treatments like donepezil could result in a mass move of all but the poor into private medical care, killing off the NHS.⁴

Closer to the NHS, the Health Services Journal quoted a professor as saying that using donepezil would reduce care costs, because it would halve the incidence of Alzheimer’s disease.⁵ The journal also reported that the NHS Confederation (representing trusts and health authorities) welcomed the drug and called for government funding to pay for it. A spokesperson argued that as the drug would mainly bring savings in social care, the estimated annual cost of £200 million could be switched from health to social services budgets. It was largely left to the Alzheimer’s Disease Society to introduce a note of caution into many of the reports by complaining of the unhelpful leaks about donepezil before its launch and suggesting that the drug was likely to play a modest role.⁶

Summary points

• Licensing trials on highly selected patients may provide insufficient information on which to base clinical decisions, especially where effect sizes are small and comorbidity is common

• All trial evidence should be published before new drugs are marketed, and medical journals should not carry advertisements referring to unpublished data

• Communication of benefits and risks should emphasise clinical effect sizes rather than statistical significance

• Claims about effects on populations or services should be based on evidence

• Secrecy surrounding licensing should be ended and data from trials should be available for independent analysis

• Overvaluation of new technology could threaten funding for vital but more mundane care

The little information available from the clinical trials suggests that the Alzheimer’s Disease Society’s caution may have been wise. According to the study’s own clinicians, the additional treatment responses in the highly selected study group were predominantly of minimal clinical importance (see below). This mismatch between perceived benefits and the study data, together with the barriers to obtaining information, raises policy concerns of general importance in health care. These concerns fall into four main categories—namely, the usefulness of the trials required for licensing, the publication of evidence, the communication of overall benefits to individuals, and the assessment of the impact on the population.

Policy concerns

Trials and the evidence of effect size

Clinical trials for drug licensing are designed to establish the efficacy and safety of new compounds. In the case of Alzheimer’s disease, the United States Food and Drug Administration set specific requirements for trials, including the need for appropriate measures of outcome.⁷,⁸ The agency, recognising that changes in psychometric test results alone might be trivial, proposed that clinically useful effects should also be determined by clinical assessment.

Clinical assessment

For people with Alzheimer’s disease and their carers, benefits will depend on improvements in everyday functioning and quality of life.⁹ Changes in the capacity to perform activities of daily living were not considered crucial by American standards, although European draft criteria do include these. As yet, no claims have been made that donepezil changes the outcome of any of these measures, although relevant data were collected. Therefore, the only available assessment of the drug’s clinical benefits, other than the cognitive test scores, is the clinician’s ratings.

“Lack of appropriate information from trials designed for drug licensing is ... not unique to donepezil”

The scale used for clinical assessment was the clinicians’ interview based impression of change (“plus” version; CIBIC-plus), “a semi-structured instrument, examining general, cognitive and behavioural functioning and activities of daily living, rated by a study clinician based upon his/her observations at an interview with the patient, in combination with information supplied by a caregiver.”¹⁰ The seven point scale is ordinal, but the ratings are: marked, moderate, minimal or no change for improvement or deterioration, and it would be absurd to suggest that the notional rating of 3 for marked improvement implies a proportionate change from the 2 for moderate improvement. Despite this, so called statistically significant mean changes in this score (of 0.35 for a 5 mg dose compared with placebo at 24 weeks) have been widely quoted in promotional literature.

Benefits

The American prescribing information leaflet provides the only published breakdown of the CIBIC-plus results.¹⁰ After 12 weeks of treatment (trial A301), the progressive decline in cognition in the placebo group was small and virtually the full effect of the drug (a mean gain of under 3.2 on the 70 point cognitive subscale of the Alzheimer’s disease assessment scale) should have been evident as clinical improvement. Reading from the published histogram, for every hundred patients treated with the 5 mg dose of donepezil for this period, five additional patients showed non-minimal improvements and two avoided non-minimal decline. Results in the group taking the 10 mg dose were similar; here an additional 7.5 showed non-minimal changes. All other changes were rated by trial clinicians as of minimal clinical importance. Put in this way, the benefits of donepezil seem to be of a different order from those implied in the promotional literature, as letters to this journal have pointed out.¹¹,¹²

Side effects

If the benefits seem smaller than claimed, what of the side effects? The American prescribing leaflet proves helpful again: the side effect events cited “reflect experience gained under closely monitored conditions of clinical trials in a highly selected patient population. In actual clinical practice or in other clinical trials, these frequency estimates may not apply.”¹⁰ Given the modest benefits, the frequency and seriousness of side effects in typical patients with mild or moderate Alzheimer’s disease is clearly crucial to a doctor’s assessment of the balance of risk and benefit in individual patients. Alas, few relevant trial data are available to help. This lack of appropriate information from trials designed for drug licensing is, of course, a widespread problem and not unique to donepezil.¹³

Presentation of the information

If a drug has modest effects and uncertain side effects in ordinary patients, the clinicians, patients, carers, and those providing funding will be required to make complex judgements on how the drug should be used. Unfortunately, nine months after launch the main clinical trials referred to in promotional leaflets still had not been published. It is worrying to note that in the United Kingdom the official summary of product characteristics offers no comment on benefits other than that the “tablets are indicated in the treatment of mild or moderate Alzheimer’s disease.” No information is available from the United Kingdom Medicines Control Agency, which currently works in secret (although limited information would have been released if European licensing procedures had been followed). The only systematic report of the outcome of the main trials is, as mentioned, in summary form in the American leaflet, which has not been distributed in the United Kingdom.

“People with dementia are directly affected by any switch in spending from supportive care to new drugs”

Since the launch of donepezil, however, the two companies involved have offered doctors a journal article covering a preliminary trial; conference abstracts; promotional publications, including selected results from unpublished trials; and ball point pens marked with the drug’s commercial name. In addition, advertisements have been published in respected journals, including the BMJ, citing unpublished “data on file.” The promotional pack contained no information on exclusions from the trials and reported outcomes concentrated on “statistically significant” mean changes in cognitive scores and clinician ratings. Failure to detect changes on other measures was not mentioned.

Response to the marketing campaign

The lack of information with which to interpret trial results would not have been damaging if the press, statutory bodies, health professionals, and those marketing the drug had waited until this was available. Lack of published information proved no bar to welcoming donepezil. For example, an editorial in the BMJ,⁹ while mentioning that effect sizes were modest, identified the main issues as ethical or relating to the need for equitable funding. In an impassioned debate on prescribing donepezil, the 1997 conference of the old age section of the Royal College of Psychiatrists was told that the time had come for psychogeriatricians to abandon their former approaches and become neuropsychiatrists, passing the caring roles to others. The motion in favour of prescribing the compound was carried by a large majority.

Population impact

Assessing the population impact of a new technology is complex, and far more information about the benefits and risks of this drug in ordinary patients is needed before responsible estimates can be made.¹⁴ Claims of reduced rates of admission to institutions are highly speculative. Given that needs for institutional care are influenced by many factors, a randomised controlled trial in typical patients and circumstances would be the only way of determining whether the modest net effects reported for donepezil would result in savings in the costs of care.¹⁵

Policy implications

New technology and population aging are the two main determinants of rising healthcare costs, and the unwillingness of most governments to increase funding has resulted in restrictions in access to care. In the United Kingdom, access to support services in the community has reduced,¹⁶ and the right to free nursing care in institutional settings, a core principle at the founding of the NHS, was withdrawn during the past decade.¹⁷ Since over 80% of those in nursing homes have dementia, this group of patients is directly affected by any switch in spending from supportive care to new drugs.¹⁸ Inflated optimism in assessing new drugs for elderly people could thus have negative practical consequences for the very people the drugs are designed to assist.

What can be done?

Among those formulating health care policy, consensus already exists that the evaluation of new healthcare technologies must be improved. However, much of the effort thus far has concentrated on including cost effectiveness studies in drug trials. This has met with little success, partly because of the technical problems involved.¹⁹ The experience with donepezil suggests that more basic issues have not yet been resolved. Action within the current regulations is feasible and should include a requirement to conduct trials on patients who are representative of those for whom the drug will be licensed and marketed, a duty to publish full trial results before marketing drugs, and a duty to ensure adequate presentation of data so that prescribing doctors can judge the size of clinical benefits and risks. In addition, medical journals should not carry advertisements that refer to data outside the public domain. Given the need for meta-analysis and the existence of at least one example of serious misuse of study data by commercial interests, a further step might involve establishing a right of independent access, analysis, and publication of the raw data from trials of marketed drugs.²⁰ Dealing with more fundamental issues, such as the need for cost effectiveness information and the lack of a level playing field in research funding for non-patentable interventions, will require more radical moves.

Conclusions

Assessing the clinical importance rather than the statistical significance of interventions involves difficult judgments by clinicians, patients, carers, and those funding health care. Examination and debate of the relevant data are a prerequisite of adequate decision making, but the launch of donepezil has illustrated a number of shortcomings in current arrangements. Improving access to relevant data on the size of net benefits in patients representative of those for whom the drug is marketed should be the immediate priority in attempts to improve regulation of new technologies. In addition, clinicians, health care organisations, and learned journals should be cautious in their promotion of new technologies before open scientific examination of full study data has been carried out.