Two and a half decades into the HIV epidemic, world and national health authorities are still trying to answer one of the most basic questions about it: how many people are infected.
Two recent developments underscore how tough it has been to count victims of a plague so vast. First, the United Nations AIDS agency lowered its worldwide estimate from 39.5 million to 33.2 million cases - a 16 percent reduction.
Then, word leaked out that the Centers for Disease Control and Prevention was likely to upgrade its U.S. estimate of 40,000 annual infections. The agency didn't deny the report but said last week it wouldn't release its new figures or methodology until January, after they have been fully reviewed by experts.
The ups and downs of HIV estimates are more than an academic exercise. The figures influence the amount of money governments and nonprofits spend on treatment and prevention - and where they spend it.
Sizing up the epidemic is a salient issue in Maryland, which has the nation's second highest rate of new AIDS diagnoses, trailing only New York. At the same time, Baltimore has the second highest rate among metropolitan areas, trailing only Miami.
Advocates have warned against any softening of international resolve to fight the scourge. Domestically, the possibility of a higher estimate reinforced complaints about reductions in federal funding for local prevention programs in the last several years.
"If it's correct that the new number is higher, then we need to make sure we do everything we can for HIV," said Dr. David Holtgrave, a social scientist with the Johns Hopkins Bloomberg School of Public Health and former head of HIV prevention programs at CDC.
Despite pledging more to fight HIV oversees, particularly in Africa, the Bush administration has reduced inflation-adjusted spending on domestic prevention by 19.3 percent since 2002, Holtgrave said.
Heather L. Hauck, director of the Maryland AIDS Administration, said declining federal aid has forced her agency to make difficult choices about which groups to target with prevention programs. She said the funding problem was partly to blame for a resurgent epidemic among men who have sex with men - a group that should probably have received more attention.
The AIDS Administration reports that about 2,000 people each year test positive for the human immunodeficiency virus (HIV), which triggers the disease. But that number doesn't reflect people who are infected but have never been tested.
The very nature of the virus makes it difficult for authorities to estimate how many people carry it. People can live symptom-free for many years after they are infected, and many don't seek testing until they feel ill.
Tracking the epidemic in the world's poorest and most populous countries is particularly hard. Some victims never get tested or treated, and if they do, their cases may go unreported. Even in the United States, where testing is more widespread, estimates have been part science, part guesswork.
By any measure, the worldwide epidemic is enormous. The United Nations estimates that 2.1 million people will die of AIDS this year alone, including 1.7 million adults and 420,000 children.
The epidemic remains fiercest in sub-Saharan Africa, where, for example, a quarter of Botswana's adult population and a third of Swaziland's are infected.
But projections of an explosive epidemic elsewhere - in India and China, for instance - have turned out to be exaggerated.
Half of the decline in the worldwide estimate stems from India, where the sheer size of the country makes it hard to arrive at precise figures.
A year ago, the United Nations estimated that 5.7 million people in India were infected - the largest tally for any nation in the world. To arrive at that number, authorities counted infections at certain clinics and hospitals, then extrapolated the results to a nation of over 1 billion people.
But authorities now say they overrepresented infections at prenatal and sexually transmitted disease clinics, as well as public hospitals serving the poor. They also gave too much weight to results from southern India, which has been hit harder than the north.
"It's an imprecise science - we've always known that," said Dr. Thomas Quinn, director of the Johns Hopkins Center for Global Health. "And whenever you go into a country that's got over a billion people, they don't even know how many people live there, never mind knowing how many people are HIV infected."
The Indian government arrived at its new estimate by conducting door-to-door interviews and testing a sample more representative of the whole population.
Not only were the numbers lower, they indicated an epidemic that hadn't broken out into the general population as much as once feared. Instead, HIV appears to be concentrated among risk groups that include prostitutes and their clients, intravenous drug addicts, gay men, migrant laborers and truckers.
Although sub-Saharan Africa remains the region's most afflicted region, the United Nations found declines in a few countries, including Uganda, Kenya and Zambia. Prevention campaigns might have accounted for some of the reduction, but AIDS' high death rate played a role, too.
With millions dead across the region, the virus simply couldn't travel at the same rate. "It was too massive to maintain," Quinn said of the epidemic. "It burns out a population."
Kevin Frost, chief executive officer for amfAR, the New-York based Foundation for AIDS Research, said a lower worldwide estimate shouldn't convince those who fund prevention and research to spend less.
"I don't think anyone's going to say that with 33 million people infected, maybe we can reduce our commitment to fighting the disease," he said. Frost said the new data should help governments and aid organizations direct money to risk groups and regions most in need.
A clearer picture of the U.S. epidemic probably won't emerge until January, when the CDC said it will release its new estimates. A gay newspaper called the Washington Blade, followed by the Washington Post, reported that the agency had substantially increased its estimate of 40,000 annual infections.
Citing unnamed sources, the Post reported that government epidemiologists now believe that 55,000 to 60,000 people are infected annually. Presumably, this would also increase the estimate that 1 million to 1.2 million Americans are living with HIV/AIDS.
Holtgrave, who directed CDC's prevention programs from 1997 until 2001, said the estimate of 40,000 annual infections was never more than an educated guess.
"Forty thousand was really an estimate within a range," Holtgrave said. "But it became gospel a little too much."
In recent years, it became clear that the estimate was too low, he said. Over the course of five years, the 33 states that report HIV cases to the federal government tallied almost 186,500 infections - an average of about 37,300 a year. So it became obvious that infections from all 50 states would surpass 40,000.
A new testing method should also lead to a more realistic sense of the epidemic. Unlike the old one, it can tell whether a person who tests positive acquired the virus recently or years earlier. This will help authorities determine now many positive results should be counted in the annual total.
Colin Flynn, chief epidemiologist at the Maryland AIDS Administration, said the tests should give the state a better picture of the epidemic's size. But it will take another two or three years to know whether the epidemic is increasing, declining or remaining stable.