Kenya study suggests fewer may have AIDS

PRETORIA -- A health survey in Kenya suggests that substantially fewer people there are infected with HIV or AIDS than previously estimated, raising questions among some specialists about whether the extraordinary global infection rates are accurate.

The Kenya survey of 8,561 households in eight provinces, completed in September last year, concluded that an estimated 6.7 percent of adults were infected with HIV, compared to an estimate of 15 percent released by UNAIDS at a Barcelona conference in July 2002.

UNAIDS officials said yesterday that they had revised their estimate of HIV prevalence in Kenya to 9.3 percent this year, based on better information from tests on pregnant women attending public prenatal clinics. They said all previous estimates were based on the best information available at the time, and any claim that past figures overestimated the epidemic were "unfounded."

Kevin DeCock, Kenya director of the US Centers for Disease Control and Prevention, which funded the latest survey, told reporters late last week that HIV prevalence rates in the country were "lower than previously estimated," based on "better, more accurate measurements."

The estimates provide critical information to help countries build an effective response to AIDS. In countries with lower prevalence rates, especially under 5 percent, emphasis is weighted toward prevention. When prevalence zooms into double digits, AIDS specialists say, the response must be an all-out emergency on prevention, treatment, and care.

Global infection rates have also been under scrutiny. At Barcelona, UNAIDS estimated that 40 million were infected at the end of 2001. Late last year, instead of giving one figure, the agency offered a range: Around the world, it said, between 34 million and 46 million were infected.

The largest reason for adjusting the estimate was that household surveys in several countries -- including Zambia and Mali -- found that earlier estimates of HIV prevalence in rural areas were too high.

"We're not pretending these estimates are exact," Dominique de Santis, UNAIDS spokeswoman, said in a telephone interview from Geneva. "There is a lot of uncertainty around the estimate. The range shows a lot of that uncertainty."

In recent years, UNAIDS also has scaled back its estimates of the number of children orphaned by AIDS, from 14 million at the end of 2001 to 13 million last year.

Despite such adjustments, even critics of the estimates say the global epidemic is horrific. They compare it to the situation in 1990, when the number of AIDS orphans was estimated at 1 million, and South Africa had an HIV prevalence of around 1 percent. Today, the adult prevalence is estimated at between 15 and 20 percent (South Africa's studies cite the lower number, UNAIDS the higher number), or between 4.6 million and more than 5 million people.

In Kenya, the new survey estimates that 1.4 million adults are infected; the UNAIDS estimate released in 2002 said that more than 3 million were infected.

"More than 1 million people who are HIV-positive is a serious problem," said Anthony D. MBewu, executive director of research for the Medical Research Council of South Africa, a quasi-public organization that tracks health issues. "Three million is a serious problem. Which number is correct, I don't know. But either way, it's serious."

The most common way to estimate HIV prevalence in the developing world is to test pregnant women at prenatal clinics. The assumption by UNAIDS, the World Health Organization, the US Census Bureau, and other groups, has been that those results roughly translate into the adult prevalence rate.

But in recent years, following at least seven demographic surveys involving tests of women -- both pregnant and not -- and men, that assumption has come under critique. Surveys in rural areas tended to include too small a group. Pregnant women may have slightly higher rates than the general adult population because they had unprotected sex at least once. Women are contracting HIV at higher rates than men, and in some countries a significant percentage of women do not go to public clinics.

"The numbers you should trust are in the representative surveys that include men, women, children, people who are sexually active, everybody," said Olive Shisana, head of South Africa's Human Sciences Research Council program on HIV/AIDS.

Those studies, however, are expensive, and officials from UNAIDS and WHO say they are learning from each survey and have adjusted their statistical modeling because of them.

The household surveys also have problems.

In Kenya, 30 percent of the people surveyed refused to take the HIV test. The survey found that 8.7 percent of the women tested were HIV positive, compared to just 4.5 percent of the men.

"People ask, `Are the population surveys the gold standard?' The answer is no," said Ties Boerma, coordinator for surveillance research and monitoring evaluation in WHO's AIDS department. "Surveys have problems. too. But they are a valuable addition."

For MBewu, the South African medical researcher, the survey has one indisputable benefit, and it's not in the numbers.

"I don't use it for absolute numbers. I use it to plot trends on the epidemic," he said.

In South Africa and several other sub-Saharan countries, one trend is a stabilization of HIV prevalence rates. But another trend is grim: AIDS deaths are on the rise.

John Donnelly can be reached at donnelly@globe.com.