Genetic statistics available to the United States public to predict their likelihood of getting Alzheimer's disease are wrong, it has been claimed. If true, this means that people are receiving an inflated estimate of their likelihood of developing Alzheimer's.

But the claim, made at the British Congress of Gerontology in Manchester last week, has been strongly disputed by other academics.

Rudolph Tanzi, associate professor of neurology at Harvard Medical School, claimed that statistics published in US newspapers on the likelihood that certain genes will lead to Alzheimer's are flawed.

The statistics are based on a particular gene, APOE. If someone carries a version of this gene known as APOE-4 then they have an increased likelihood of developing Alzheimer's. The tables predict, for example, that healthy 75-year-olds who have two copies of APOE-4 have an 80 per cent chance of developing it.

ÖйúAƬ

Dr Tanzi said that the "alarming" statistics are based on data from families that already have members with Alzheimer's disease, and other special groups. They could not be extrapolated to the general population.

"We need many many samples before we can give any responsible data back to people," he said. "They need to be stratified according to the age of onset, ethnicity, gender."

ÖйúAƬ

Dr Tanzi revealed data at the Manchester meeting claiming there is no statistically significant association between having two copies of APOE-4 and developing Alzheimer's disease - except in the 61 to 65 age group. This is not such an important group, he said, because most Alzheimer's develops at a later age.

Dr Tanzi said the published statistics were particularly worrying in the light of a test for APOE-4 now available from Athena Neurosciences in the US. The test is only available to doctors who need it to help them decide whether or not a patient who already has cognitive impairment is suffering from Alzheimer's.

Dr Tanzi said he was worried that patients could manipulate doctors into giving them the test and then get the statistics from the newspapers.

But Alan Roses, of Duke University, North Carolina, which licenses Athena Neurosciences to provide the APOE-4 test, said: "The age of onset distribution curves have been confirmed widely . . . if Dr Tanzi has any data to refute the distributions he should publish it.

ÖйúAƬ

"APOE genotyping is widely available internationally for cardiovascular diseases. 'Ordinary people' can get it done anywhere, but any interpretation of the test regarding Alzheimer's becomes the responsibility of the ordering physician and the lab."

In Britain, a consensus published in The Lancet earlier this year concluded that APOE testing in healthy people is not recommended partly because its predictive value is uncertain. Nevertheless people are continuing to ask doctors for it.