"Explosive HIV incidence rates" among gay and bisexual young men, particularly among African American men who have sex with men ( MSM ) are "alarming" and "of critical public health importance," said officials from the Centers for Disease Control and Prevention ( CDC ) in a May 31 telephone news conference with reporters.
Mainstream media accounts dutifully relayed the alarmist message without looking too closely at the shaky epidemiology on which it is based. Had they done so, they might have come to the conclusion the news conference was more about politics than about science.
CDC researcher Linda Valleroy led the team that conducted the study. She explained that while it is important to understand the prevalence of HIV, or how many people are infected with the virus, "It is even more important to understand the rate of new infections or incidence." This helps target resources to where they are needed.
In February the CDC released "preliminary data on HIV prevalence" from the Young Men's Survey, an ongoing study since 1994 that looks at gay and bisexual men ages 23 to 29 in six cities...Baltimore, Dallas, Los Angeles, Miami, New York City, Seattle. It showed an overall infection rate of 13 percent, with a "very high" rate of 32 percent among the African American men sampled.
The new data was the second part of that survey, conducted 1998-2000. It used a STARHS assay to determine annual incidence of new infections. STARHS measures two different antibodies to HIV. One antibody appears about a month after infection while the second takes at least three months to develop. If only the first antibody is present, the researcher knows that the infection is relatively recent.
Valleroy said among those they tested, the overall HIV incidence was 4.4 percent. The rate varied tremendously between groups. The rate for white gay men was 2.5 percent, for Hispanics 3.5 percent, and for African Americans 14.7 percent. It means for every 100 African American MSM that they tested, 15 became infected with HIV during the last year. She called these "explosive HIV incidence rates." Valleroy said the 4.4 percent overall incidence rate is similar to that seen in southern and eastern Africa in the early 1990s, immediately prior to the rapid growth of HIV in that region to about 20 percent of the total population.
She said American data from surveys conducted in the early 1990s indicated a 2 to 3 percent overall rate of infection among MSM. A 4.4 percent rate is comparable to what was seen in the U.S. in the mid-1980s at the peak of new infections. There is no comparable earlier data on African American MSM, only the more general figures for all MSM.
"These data, combined with recent outbreaks of STDs among MSM in many cities suggest that HIV incidence may be increasing in this population. And that the epidemic among men who have sex with men in general, and among African American men who have sex with men in particular, is growing worse," she said.
Combing data from several studies, Valleroy concluded that new infections "increase substantially between adolescence and the early 20s. These findings underscore the need to reach each generation of gay and bisexual men early as they enter adolescence."
LET ME COUNT THE WAYS
One underlying principle of statistics is that things tend to even out with larger sets of numbers, while a smaller sample carries greater risk that a few unrepresentative data points can dramatically swing the outcome one way or another. That is exemplified by the "confidence interval ( CI ) ," a range on either side of the statistic where the researcher feels 95 percent certainty of the results. The smaller the range of the CI, the greater the accuracy of the statistic, the larger the CI the fuzzier the math. Larger sets of data have smaller CIs and are more accurate. At first glance the 2,942 young MSM in the survey is an impressive number. But from that large pool, only 38 were identified as recent infections using STARHS. Because white MSM constituted the greatest portion of those in Valleroy's research, their incidence rate of 2.5 percent had a CI of 1.4 to 4.6 percent. Hispanics had a 3.5 percent incidence and a CI of 1.4 to 8.6. But because the number of African American MSM sampled was the smallest of the survey subsets, their incidence of 14.7 percent had a CI of 7.9 to 27.1 percent. That is a huge range.
Valleroy was careful to say, "I must underline that the sample size of African Americans is small and that these findings are preliminary." The CDC did not release other numbers. Valleroy hopes to have all of the data and have it published by the end of the year.
Helene Gayle, who heads up HIV programs at the CDC, said they have "done some statistical corrections" to compensate for the incomplete work, and that affects the confidence intervals.
In an article published by CDC in the June 1 Morbidity and Mortality Weekly Report, the authors were careful to note limitations of the study. Among them are questions as to how representative the sample is of MSM. Gayle said, "The information we have does not suggest men who self-identify would necessarily be at greater risk" for HIV. She suggested the opposite, perhaps those who do self-identify as gay or bisexual are more knowledgeable about the risks of HIV.
CHANGE AND STABILITY
The CDC estimates that 40,000 Americans become infected with HIV each year. It is the same number that they have used since 1992. But if you listen to the different things that the CDC says, the numbers just don't seem to add up.
Men who have sex with men are still the largest group ( 46 percent ) of those with AIDS. The CDC says the rate is explosively high among young Black men who have sex with men and less high among Latinos. CDC does not argue that new infections are falling in gay white men, quite the opposite, citing increases in sexually transmitted diseases, it is worried that new infection are increasing among young gay men.
At the same time we are told women have the highest rate of increase of AIDS infections of any group. Children have always been a very small portion of total infections, while infection through injection drug use continues at a high level.
With all of these claimed increases going on and little talk of decreases, why has the overall estimate of annual infections remained the same since 1992? When pressed on whether that estimate of 40,000 new infections a year needs to be revised upward, Gayle said changes might be coming.
So why did CDC call a news conference? The 20th "anniversary" of HIV was about to occur and CDC did not want to be left out of the flurry of media coverage. It rushed the incomplete survey out before it was ready. Gayle is about to depart for the Gates Foundation, so this was her swan song. And finally, Congress is about to tackle spending bills. Waiting for publication would take too long for CDC to put forward its best case for more money for HIV prevention activities. And that is how science becomes political science.