© 2006 The Washington Post Company, April 6, 2006

How AIDS in Africa Was Overstated
Reliance on Data From Urban Prenatal Clinics Skewed Early Projections

By Craig Timberg
Washington Post Foreign Service
Thursday, April 6, 2006; A01

KIGALI, Rwanda -- Researchers said nearly two decades ago that this tiny country was part of an AIDS Belt stretching across the midsection of Africa, a place so infected with a new, incurable disease that, in the hardest-hit places, one in three working-age adults were already doomed to die of it.

But AIDS deaths on the predicted scale never arrived here, government health officials say. A new national study illustrates why: The rate of HIV infection among Rwandans ages 15 to 49 is 3 percent, according to the study, enough to qualify as a major health problem but not nearly the national catastrophe once predicted.

The new data suggest the rate never reached the 30 percent estimated by some early researchers, nor the nearly 13 percent given by the United Nations in 1998.

The study and similar ones in 15 other countries have shed new light on the disease across Africa. Relying on the latest measurement tools, they portray an epidemic that is more female and more urban than previously believed, one that has begun to ebb in much of East Africa and has failed to take off as predicted in most of West Africa.

Yet the disease is devastating southern Africa, according to the data. It is in that region alone -- in countries including South Africa, Botswana, Swaziland and Zimbabwe -- that an AIDS Belt exists, the researchers say.

"What we know now more than ever is southern Africa is the absolute epicenter," said David Wilson, a senior AIDS analyst for the World Bank, speaking from Washington.

In the West African country of Ghana, for example, the overall infection rate for people ages 15 to 49 is 2.2 percent. But in Botswana, the national infection rate among the same age group is 34.9 percent. And in the city of Francistown, 45 percent of men and 69 percent of women ages 30 to 34 are infected with HIV, the virus that causes AIDS.

Most of the studies were conducted by ORC Macro, a research corporation based in Calverton, Md., and were funded by the U.S. Agency for International Development, other international donors and various national governments in the countries where the studies took place.

Taken together, they raise questions about monitoring by the U.N. AIDS agency, which for years overestimated the extent of HIV/AIDS in East and West Africa and, by a smaller margin, in southern Africa, according to independent researchers and U.N. officials.

"What we had before, we cannot trust it," said Agnes Binagwaho, a senior Rwandan health official.

Years of HIV overestimates, researchers say, flowed from the long-held assumption that the extent of infection among pregnant women who attended prenatal clinics provided a rough proxy for the rate among all working-age adults in a country. Working age was usually defined as 15 to 49. These rates also were among the only nationwide data available for many years, especially in Africa, where health tracking was generally rudimentary.

The new studies show, however, that these earlier estimates were skewed in favor of young, sexually active women in the urban areas that had prenatal clinics. Researchers now know that the HIV rate among these women tends to be higher than among the general population.

The new studies rely on random testing conducted across entire countries, rather than just among pregnant women, and they generally require two forms of blood testing to guard against the numerous false positive results that inflated early estimates of the disease. These studies also are far more effective at measuring the often dramatic variations in infection rates between rural and urban people and between men and women.

UNAIDS, the agency headed since its creation in 1995 by Peter Piot, a Belgian physician, produced its first global snapshot of the disease in 1998. Each year since, the United Nations has issued increasingly dire assessments: UNAIDS estimated that 36 million people around the world were infected in 2000, including 25 million in Africa. In 2002, the numbers were 42 million globally, with 29 million in Africa.

But by 2002, disparities were already emerging. A national study in the southern African country of Zambia, for example, found a rate of 15.6 percent, significantly lower than the U.N. rate of 21.5 percent. In Burundi, which borders Rwanda in central East Africa, a national study found a rate of 5.4 percent, not the 8.3 percent estimated by UNAIDS.

In West Africa, Sierra Leone, just then emerging from a devastating civil war, was found to have a national prevalence rate of less than 1 percent -- compared with an estimated U.N. rate of 7 percent.

Such disparities, independent researchers say, skewed years of policy judgments and decisions on where to spend precious health-care dollars.

"From a research point of view, they've done a pathetic job," said Paul Bennell, a British economist whose studies of the impact of AIDS on African school systems have shown mortality far below what UNAIDS had predicted. "They were not predisposed, let's put it that way, to weigh the counterevidence. They were looking to generate big bucks."

The United Nations started to revise its estimates in light of the new studies in its 2004 report, reducing the number of infections in Africa by 4.4 million, back to the total four years earlier of 25 million. It also gradually decreased the overall infection rate for working-age adults in sub-Saharan Africa, from 9 percent in a 2002 report to 7.2 percent in its latest report, released in November.

Peter Ghys, an epidemiologist who has worked for UNAIDS since 1999, acknowledged in an interview from his office in Geneva that HIV projections several years ago were too high because they relied on data from prenatal clinics.

But Ghys said the agency made the best estimates possible with the information available. As better data emerged, such as the new wave of national population studies, it has made revisions where necessary, he said.

"What has happened is we have come to realize that indeed we have overestimated the epidemic a bit," he said.

On its Web site, UNAIDS describes itself as "the chief advocate for worldwide action against AIDS." And many researchers say the United Nations' reliance on rigorous science waned after it created the separate AIDS agency in 1995 -- the first time the world body had taken this approach to tackle a single disease.

In the place of previous estimates provided by the World Health Organization, outside researchers say, the AIDS agency produced reports that increasingly were subject to political calculations, with the emphasis on raising awareness and money.

"It's pure advocacy, really," said Jim Chin, a former U.N. official who made some of the first global HIV prevalence estimates while working for WHO in the late 1980s and early 1990s. "Once you get a high number, it's really hard once the data comes in to say, 'Whoops! It's not 100,000. It's 60,000.' "

Chin, speaking from Stockton, Calif., added, "They keep cranking out numbers that, when I look at them, you can't defend them."

Ghys said he never sensed pressure to inflate HIV estimates. "I can't imagine why UNAIDS or WHO would want to do that," he said. "If we did that, it would just affect our credibility."

Ghys added that studies now show that the overall percentage of Africans with HIV has stabilized, though U.N. models still show increasing numbers of people with the virus because of burgeoning populations.

Many other researchers, including Wilson from the World Bank and two epidemiologists from the U.S. Agency for International Development who wrote a study published last week in the Lancet, a British medical journal, dispute that conclusion, saying that the number of new cases in Africa peaked several years ago.

Some involved in the fight against AIDS say that tallying HIV cases is not nearly as important as finding the resources to fight the disease. That is especially true now that antiretroviral drugs are more affordable, making it possible to extend millions of lives if enough money and health-care workers are available to facilitate treatment.

"It doesn't matter how long the line is if you never get to the end of it," said Francois Venter, a South African doctor and head of Johannesburg General Hospital's rapidly expanding antiretroviral drug program, speaking in an interview in Johannesburg.

But to the researchers who drive AIDS policy, differences in infection rates are not merely academic. They scour the world looking for evidence of interventions that have worked, such as the rigorous enforcement of condom use at brothels in Thailand and aggressive public campaigns that have urged Ugandans to limit their sexual partners to one.

Programs deemed successful are urged on other countries and funded lavishly by international donors, often to the exclusion of other programs.

Rwanda, a mountainous country of about 8.5 million people jammed into a land area smaller than Maryland, has relied on approaches similar to those used in Uganda, and may have produced similar declines in HIV. UNAIDS estimated in 1998 that 370,000 Rwandans were infected, equal to 12.75 percent of all working-age adults and a substantial percentage of children as well. Every two years since, the agency has lowered that estimate -- to 11.2 percent in 2000, 8.9 percent in 2002 and 5.1 percent in 2004.

Dirk van Hove, the top UNAIDS official in Rwanda, said the next official estimate, due in May, would show an infection rate of "about 3 percent," in line with the new national study. He said the U.N. estimate tracked the declining prevalence.

Rwandan health officials say their national HIV infection rate might once have topped 3 percent and then declined. But it's just as likely, they say, that these apparent trends reflected nothing more than flawed studies.

Even so, Rwanda's cities show signs of a serious AIDS problem not yet tamed. The new study found that 8.6 percent of urban, working-age women have HIV. Overall, officials say, 150,000 Rwandans are infected, less than half the number estimated by UNAIDS in 1998.

Bruno Ngirabatware, a physician who has treated AIDS patients in Kigali since the 1980s, said he has seen no evidence of a recent decline in HIV infection rates.
"There's lots of patients there, always," he said.