NICE made a near U-turn in its latest draft of guidance on Alzheimer's disease treatments last week. Careful consideration of new evidence, it said, supported use of the drugs to treat patients with moderate Alzheimer's (GP, 27 January).
GPs, pressure groups and patients cautiously welcomed this change to last year's controversial draft guideline that recommended the drugs not be prescribed to newly diagnosed Alzheimer's patients on the NHS (GP, 4 March 2005).
However, NICE is still coming under fire from some quarters with accusations that it is continuing to put its reputation and NHS coffers ahead of patient welfare.
The institute received public criticism in March when its first draft guidance declared that there was no evidence to support the use of donepezil (Aricept), galantamine (Reminyl), rivastigmine (Exelon) or memantine (Ebixa) in Alzheimer's patients.
During the summer, the drug manufacturers were asked to provide new clinical trial data to be analysed by the Medical Research Council's biostatistics unit.
A sub-group analysis tested whether the acetylcholinesterase (AchE) inhibitors were more effective in patients at a particular stage of the disease.
A responder analysis on the same drugs gauged the effect of the drugs in patients who had been identified as responding to treatment.
The sub-group analysis found that the AchE inhibitors had a greater effect on patients with moderate Alzheimer's than on people with mild Alzheimer's, whereas the responder analysis found a beneficial effect in patients with all severities.
The NICE appraisal committee accepted the findings of the sub-group analysis but rejected the responder analysis data after being warned that it contained 'significant selection bias and related uncertainty'.
Professor Clive Ballard, professor of old age psychiatry at King's College London, and director of research at the Alzheimer's Society, said the responder analysis data should not have been rejected. He tried unsuccessfully to find out who gave this advice.
'NICE said it was the MRC biostatistics unit, but it doesn't appear in its report and the unit denies it,' he said.
'The only reason can be that NICE wanted to avoid a complete U-turn to save face.'
NICE chief executive Andrew Dillon maintained that NICE's recommendations were supported by the evidence. Professor Richard Gray, professor of medical statistics at Birmingham University, agreed with NICE's decision to disregard the responder analysis data.
'Responder analysis is not appropriate. You can't measure the response in an individual patient because the scale used isn't very accurate,' he said.
However, he believes that NICE has let politics distract it from clinical evidence, since its recommendations do not tally with the results of his own NHS-funded research that showed a small beneficial effect of donepezil in patients with Alzheimer's of any severity.
'This seems politically driven. It would have been better to focus on better provision of services which the evidence shows work well, like social support,' he said.
Lancet 2004; 363: 2,105-15