The provisional guidance from NICE on drugs for Alzheimers’ disease – donepezil, galantamine, rivastigmine, and memantine – marks a dramatic shift from restricting access to those with moderate disease to recommending these drugs as options for all patients with the disease. What has changed?
No new compelling evidence has emerged. For each of the drugs the draft guidance notes that no new evidence had been located for quality of life, time to institutionalisation, or mortality. While 17 new trials have been reported, the evidence base for efficacy remains essentially unchanged. The draft guidance states, “The committee concluded that the new evidence provided additional support to the conclusions from 2004 that each of the AChE inhibitors offers some benefits in behavioural outcomes, although the nature and extent of behavioural benefits are uncertain owing to the mixed results from the available evidence.”
A key issue with cost implications is whether or not these drugs delay institutionalisation. The committee acknowledged that time to institutionalisation was not generally included as an endpoint in randomised clinical trials. It noted that the AD2000 trial collected data on institutionalisation, “but limited accrual into the trial led to an underpowered conclusion about this outcome” (more on the AD2000 trial below). The committee concluded that although it was clinically plausible that treatment may delay time to institutionalisation, limited direct evidence was available to assess the magnitude of this effect. The draft guidance noted “the absence of a systematic review of observational studies available in this area.”
As a consequence of this, the decision to recommend these drugs was based on the modelling of cost effectiveness. This was the basis on which the committee had recommended against use of donepezil in 2004. That was based on the SHTAC model which showed the cost per QALY for donepezil in mild Alzheimer’s from £55-58k. In the recent review, the Pfizer/Eisai model had donepezil dominating best standard care (improved outcomes and lower costs). The surprise is that the model developed by the assessment group agreed with this.
The assessment group model was a development of the SHTAC 2004 model, but with four key differences according to the committee:
- different sources used to model disease progression
- updated cost estimates of costs and effectiveness
- costs and utilities varied according to time before institutionalisation
- discontinuation of treatment was accounted for.
How plausible are these differences and which is most important?
The data used to model disease progression were UK not US as in 2004. Although the key UK study was published in 2002, the modellers were able to access the individual patient data from this study and from one other London study. The key study was that from Oxford which was based on 92 untreated AD patients followed up from 1988/9 to 1998/9. This assumed that institutionalisation was predicted by age and mini mental state examination score. The Committee acknowledged that the representativeness of this small sample was unknown
Costs were affected by the estimated survival time for Alzheimer’s patients with the new model assuming 3.8 years or just over half the previous models assumed 6.5 years. Both the SHTAC and the assessment group models assumed treatment effects of AChE inhibitors that resulted in similar delays to institutionalisation (2 months in the SHTAC model, 1.4 to 1.7 months in the assessment group’s model). By implication, since the model had only 3 states (pre-institutionalisation, post institutionalisation, and death), the shorter survival time meant less time post institutionalisation.
Inclusion of discontinuation of treatment led to reduced costs. It was assumed that a constant 4% of the original cohort discontinued drug treatment “A constant rate of 4% discontinuation per monthly cycle for all drugs at all doses was assumed. Therefore within 25 months all patients were assumed to have stopped treatment.” (Para 4.2.13). Discontinuation had not been included in the 2004 model. The draft guidance states that, “Clinical specialists advised that discontinuation in clinical practice is between 2% and 5% per year (sic) and considered that a 4% discontinuation rate to be appropriate. In addition, the committee recognised that in sensitivity analyses, variation in the assumed discontinuation rate had the biggest impact on cost effectiveness estimates” (para 4.3.21).
[Note the error in this last quote in describing the discontinuation rate as yearly when it should be monthly as in para 4.2.13 quoted above. A yearly discontinuation of rate of 4% in a population with mean life expectancy of 3.8 years would not make sense. The rest of the documentation makes it clear that this is a monthly discontinuation rate.]
The most important difference thus seems to be the assumed discontinuation rate, which in turn the guidance states was based on the advice of “clinical specialists.” However the Assessment Group based their estimate on RCT reported discontinuations, which were reported for a single point in time. Hence their use of constant rate.
The plausibility of extrapolating discontinuation rates from clinical trials to real life must be questioned. However, Pfizer/Eisai’s model submitted to NICE included a discontinuation rate based on a study of 88 individuals, based on another small study. This shows that 58% stayed on treatment beyond 1 year, which is roughly similar to the 52% one would get with a 4% constant discontinuation rate per month.
A key fact: donepezil is due to lose its patent in 2012. Pfizer/Eisai suggest in their submission to NICE that the price will reduce by 30%. However bigger reductions might well apply depending on the market for generics.
Finally, the disparaging of the AD2000 trial is notable. As a co-investigator of that study, I admit to a possible conflict of interest. The Lancet report of AD2000 stated:
“One objective not achieved by the AD2000 trial was the rapid recruitment of much larger numbers of patients than in previous studies. This shortfall was attributable to several factors including a delayed start because of difficulties in securing placebo and drug supplies from the relevant pharmaceutical companies who opposed the trial, difficulties in extension of recruitment outside the West Midlands region because such centres had to fund the full cost of the trial drugs and patient withdrawal after publication of the NICE guidance in 2001.
To have AD2000 dismissed as underpowered by the agency whose recommendation forced the closure of the trial is – well – annoying.
James Raftery is a health economist with several decades experience of the NHS. He is Professor of Health Technology Assessment at Southampton University. A keen ‘NICE-watcher’, he has provided economic input to technical assessment reports for NICE but has never been a member of any of its committees. The opinions expressed here are his personal views. Comments welcomed.