Secretin, which had been touted as a new "wonder drug" for autistic children, does not improve their ability to talk or interact with other children, one of the first large-scale studies of the treatment has shown.

The drug first came to the world's attention in 1998 when a mother said her son showed improvement in autistic symptoms after he was given secretin during a diagnostic test for diarrhoea. Doctors published an academic paperreporting that the boy, and two other children, showed increased alertness, eye contact and expressive language within days of receiving a single dose of the hormone.

Parents from all over the world hoped the hormone, which is produced in the intestines, might improve their child's ability to communicate. The drug is not available in Britain but many parents travelled abroad or bought the drug over the internet in the hope that it would help their child.

The new research found, however, that there were no significant changes in either receptive or expressive language among the children it studied.

In Britain, autism is believed to affect about 700,000 people - including those with mild symptoms - with between two and four children per 10,000 borndeveloping the condition. Three times as many boys as girls are affected.

Autism is a syndrome that manifests itself in the first few years of life and involves severe deficits in socially and communication skills. The cause is not known and there is no proven cure.

Behavioural interventions can markedly improve language and social skills. For older people, medication can sometimes help ease obsessive behaviours.

"We began the study because we were concerned that parents and healthcare professional did not have sufficient information to evaluate secretin's potential effects," said Dr Jenifer Lightdale, of the University of California, San Francisco, who presented the findings at the Paediatric Academic Societies meeting in Boston yesterday.

Dr Lightdale conducted the study with Dr Melvin Heyman, chief of paediatric gastroenterology at the University of California, and autism experts from the university's Langley Porter Psychiatric Institute.

Dr Heyman said: "The need for research to see whether the drug was effective was accentuated by reports from autism interest groups that thousands of children had begun receiving secretin in repeated doses."

The researchers gave 20 autistic children, aged 3 to 6, formal language testing before administering secretin intravenously. They then gave them four more tests after one, two, three and five weeks. The children were also videotaped while at play and scored for specific behaviour characteristic of autism.

The findings showed there were no quantifiable changes in the way the children were able to use either words or gestures to express themselves.