How does NICE assess drugs?

It works out how much the NHS must spend on them to achieve a defined benefit to patients. They are approved if they cost the NHS less than about £30,000 per quality adjusted life year (QALY).

That means for every £30,000 spent prescribing them, the benefit enjoyed by patients must add up to the equivalent of a single patient living an extra year of good-quality life.

How does it do it?

It often uses computer-based economic models designed by academics, which are fed with the trial data supplied by the companies and work out the cost per QALY. For Aricept the cost per QALY was between £50,000 and £90,000, depending on what assumptions were made about how it is used.

That is miles off target. Why does Eisai think it can change NICE’s mind?

Eisai suspects that the model used in the Aricept appraisal, designed by a group from Southampton, may underestimate the benefits of Aricept. It wants to explore that by feeding in different data. A second issue – not part of the case – is that NICE should take account of benefits to carers, not just sufferers.

Why did NICE object to Eisai and Pfizer seeing the model?

It argued that the model was the intellectual property of the university where it was developed and that the drug companies should be satisfied with outline details of how it worked. It also fears that providing a full working version of the model will slow down appraisals.

How so?

Let’s suppose, as one NICE witness did, that a drug company uses the model backwards. First it fills in the result it wants – a QALY of about £30,000 – and then uses the model to see what starting assumptions are needed to achieve that result. It conceals its method from NICE and simply submits those assumptions to the appraisal.

NICE would then have to spend a lot of time unpicking the application and the amendments that had been made by the manufacturer: that requires, time, expertise and money, as another witness said.

But if a model can produce one result if tweaked by the manufacturer, and another if run by NICE, it cannot be much of a model can it?

Possibly not, but the alternative is worse. Some way must be found for measuring cost-effectiveness and economic modelling is the best we have. The alternative is guesswork, or relying on manufacturers’ claims, which could lead to money being wasted by being spent on ineffective treatments.

What are the odds that Eisai/Pfizer can prove the model misrepresents the benefits of Aricept?

It is hard to say. It was not designed to be wrong. One possibility might be to find a way of defining a patient group that benefits more than average, limiting the drug to that group then recalculating the cost-benefit. That might show, for that group, that Aricept is cost-effective.

Was the court victory against NICE a surprise?

It is a first. Patient groups and the drugs industry have railed against NICE decisions for years but until now nobody has ever pinned them down in court. The lower court verdict was clear cut, so much so that the Alzheimer’s Society dropped out of the action. Only Eisai and Pfizer persisted.

They won but ultimately will it make much difference to patients?

If Eisai/Pfizer can show the appraisal is faulty it could mean that patients who would otherwise be denied Aricept will get it. That is some way off yet – and the drug is no miracle cure otherwise proving its benefits would be a lot easier.

What about other appraisals?

The ruling could have implications for other drug appraisals where economic models have been used. In that case it could reopen some guidance given in the past by NICE that has denied drugs to patients.

Explore Health

• Expert Advice

• Health Features

• Mental Health

• Alternative Medicine

• Child Health

• Health Club