Johannesburg — A KENYAN study has revealed that one million people in the country are infected with HIV - two million fewer than previously estimated.

The study by the Kenya Demographic and Health Survey - which was funded and backed by a US government agency, the Centres for Disease Control - suggests that HIV/Aids figures are overestimated.

Surveys in Mali, Zambia and South Africa had previously hinted that the HIV/Aids infection figures could be much lower than predicted.

While most countries use figures based on extrapolations derived from tests of pregnant women - as set out by the United Nations and the World Health Organisation - Kenya conducted a survey that tested 3 000 households for HIV/Aids.

The survey concluded that there were fewer people living with HIV/Aids than projected, and that the figure should hold true for the rest of Africa.

Recently, the renowned South African writer and journalist Rian Malan attacked Aids "doomsters", saying the epidemic was not as huge as they would have us believe.

"They came forth like loonies drawn by a full moon, chanting that Aids was getting worse and worse, 'spinning out of control', crippling economies, causing famines, killing millions, contributing to the oppression of women and 'undermining democracy' by sapping the will of the poor to resist dictators," he wrote in The Spectator magazine.

UNAIDS did not respond to the Kenya survey, but the UN argued in its December report that random population surveys - such as the one carried out in Kenya - risked underestimating infection rates because many people were absent or reluctant to answer questions.

The UN said a panel of experts regularly updated and refined its figures, which put the number of infected people in Africa at 25 to 28.2 million.

South African Aids researchers and experts also stood by their estimates, which put infection at 4.69 million out of a population of 45 million.

Dr Nono Simelela, director of the Health Department's HIV/Aids programme, said that while there was a small margin of error, the figures were as accurate as possible given the "tools we are using".

"We use the best methodology that is currently accepted by the W HO and we consistently say people need to understand that these are estimates because we are not doing incidence, we are doing prevalence," she said.

"We stick to world standards endorsed by WHO and we try to refine it to suit our own circumstances. " Dr Olive Shisana, head of the Human Sciences Research Council's HIV/Aids programme, conceded South Africa's figures were overestimated - one study put prevalence at 5% less than estimated - but she said the overestimation did not make the problem less serious.

"The number of people infected is still large and the resources that are being spent globally are not sufficient," she said.

She said to gauge the true picture, the Health Department needed to do population-based surveys instead of relying on antenatal clinic data, but it was difficult because there was not enough information about the population.

Leigh Johnson, a researcher with the Centre of Actuarial Research , said the Kenyan data was unreliable.

"One has to be careful about extrapolating from Kenya to the rest of Africa. I f our estimates are out, they're not going to be out by that much. "