More than 100,000 elderly people with dementia are being sedated with drugs linked to increased risk of pneumonia, strokes and heart attacks.

The first long-term study of a class of drugs called neuroleptics found patients taking them died on average six months earlier than those switched to placebos.

According to the Alzheimer's Research Trust, the drugs are being used to sedate up to 125,000 dementia patients in the UK despite a warning from the Committee on Safety of Medicines three years ago.

Clive Ballard, professor of age-related disorders at King's College London, and lead researcher on the project, said: "We know from short-term studies that these drugs increase chest and other infections as well as ankle swelling.

"It is very clear that even over a six-month period of treatment, there is no benefit of neuroleptics in treating the behaviour in people with Alzheimer's disease when the symptoms are mild." The five-year study involved 165 Alzheimer's patients in UK nursing homes who had been taking neuroleptics for at least three months. About half carried on taking them while the others were given placebos instead.

After two years, only 55 per cent of those taking the neuroleptics were still alive, compared to 78 per cent of those not taking them.

After three years the proportion taking the drugs who were still alive was 35 per cent, compared to 62 per cent of the placebo group.

There are over a dozen drugs in the neuroleptic class. Those being taken by patients in the research were chlorpromazine (Largactil), haloperidol (Serenace), trifluoperazine (Stelazine), risperidone (Risperdal) and thioridazine (Melleril). The last of these was withdrawn by its maker Novartis in 2005.

A spokesman for Janssen-Cilag, the company that makes Risperdal, said: "We have not been able to see the full study.

"All we would say at this point is we work within the license for Risperdal and if you look at the license it is not indicated for this patient population."