Your feature on the introduction of the Alzheimer's drug, donepezil, in the Withington Hospital, Manchester, ('Making introductions', pages 28-29, 25 February) stated that 'evidence of its efficacy in clinical use and its cost-effectiveness can only be obtained from further research and audit after its introduction in clinical practice'.

Unfortunately, the Withington team chose audit rather than research. Audit will never provide information about whether, on average, donepezil is good value for the NHS. The alternative is to enter patients into a proper clinical trial of effectiveness and costs. Only this can provide evidence of whether donepezil improves patients' quality of life and/or delays care requirements. The trials to date, all sponsored by pharmaceutical companies, show improvements limited to cognitive capacity. But being able to spell 'world' backwards for six months longer is of debatable benefit.

Such a clinical trial - 'AD2000'- exists and has recruited its first 100 patients. Funded by West Midlands health authorities and co-ordinated by Birmingham University, it aims to recruit about 3,000 patients nationally and to report in 2000. Several centres outside the West Midlands have already entered the trial, and others are planning to follow, depending partly on financing arrangements for the drug, which costs about£1,000 per patient per year.

AD2000 represents one of the biggest NHS-based clinical trials of a new drug. If the NHS is ever to generate the data it needs for rational policy making, such a trial should be supported. Interested readers should contact the AD2000 co-ordinating office.

James Raftery

Professor of health economics,

Richard Gray

Professor of medical statistics

Birmingham University

Tel: 0121-414 7560