A new HIV/AIDS survey in Kenya suggests that infection rates in the east African country are significantly lower than what global health organizations have estimated. The new HIV infection study, conducted as part of the Kenya Demographic and Health Survey (KDHS), with funding by the U.S. Centers for Disease Control, concludes that about one million of the country's 30 million people are HIV positive.

That number is significantly lower than figures previously published by UNAIDs, the program run by the United Nations, the World Health Organization and the World Bank. UNAIDS has estimated that as many as two and a half million Kenyans are HIV positive.

The director for the Centers for Disease Control in Nairobi, Kevin DeCock, attributes the discrepancy to different methods in computing the infection rate estimates.

For example, UNAIDS relies heavily on data derived from blood tests of pregnant women, because the women have been sexually active and have not used condoms. These tests, called sentinel surveillance, make the assumption that HIV infection rates among pregnant women are fairly reflective of the general population.

Dr. DeCock said scientists who conducted the Kenya Demographic and Health Survey chose a different method, randomly testing about 3,000 households, including men, in urban and rural areas of the country. He said the KDHS showed similarities and glaring differences with HIV infection estimates derived from sentinel surveillance of pregnant women. "The data in women in KDHS tracked very well. They were slightly lower," he said. "The overall prevalence in sentinel surveillance was 9.4 percent nationally, and the overall prevalence in KDHS was 8.7 percent. And there were a pretty good correlation geographically and across rural and urban differences. So, it was in men that we had a much lower prevalence."

Dr. DeCock said similar studies in Zambia and Mali have also hinted that HIV infection figures in those countries, as well as for the whole of Africa, could be much lower than current estimates.

The chief scientific adviser for the Geneva-based UNAIDS, Katherine Hankins, concedes that by including men and non-pregnant women, household surveys could get a more representative sampling of the population. But Ms. Hankins said household surveys also risk underestimating infection rates. "Overall, we think it is very likely that the household surveys probably underestimate the epidemic a bit, and that is because people can refuse to participate," she noted. "The pregnant women surveys are anonymous, whereas with the households, people are being asked would they like to have an HIV test. So, we have people who refuse to be tested and we have people who are away from their homes when the survey comes through."

Despite some of the drawbacks of the Kenyan survey, Ms. Hankins said UNAIDS is seriously studying the survey results as the organization draws up revised HIV infection estimates for all countries.

UNAIDS will announce the new figures in July at the International AIDS Conference in Bangkok, Thailand.