Tuesday, March 9, 2010

Circumcision may increase HIV spread among gay men

We need data, not headline.

It's unconscionable that since early today, this story has been all the public knows about the referenced study.

Here is what little we are told:

"The findings are based on data from 4,889 men who took part in an HIV vaccine trial begun in 1998; 86 percent had been circumcised. During the three-year study, 7 percent of the men became HIV-positive.

When the researchers accounted for other factors -- including demographics, and HIV risk factors like drug use and having unprotected sex -- circumcision showed no effect on the odds of HIV transmission.

Still, Gust and her colleagues point to some limitations of their study, including the relatively small number of uncircumcised men overall and the small number of uncircumcised men who became HIV-positive during the study -- 43"

So, what can the public think with so little to go on?

4,889 men
86% circumcised
7% HIV infection rate
43 intact men became HIV infected

So that means:

4205 circumcised men (4,889 * .86) = 4,205
684 intact men (4,889 - 4,205) = 684
342 men infected (4,889 * .07) = 342
43 intact men infected
299 circumcised men infected (342 - 43) = 299
7.11% circumcised men infection rate (299 / 4205) = .0711
6.29% intact men infection rate (43 / 684) = .0629

Well, isn't that curious. If you stop and puzzle it out, assuming what little data we're given is at least accurate, you find circumcised men had a 13% increase in HIV infection.

13% increased infection rate of circumcised group (.0711 - .0629) / .0629 = .1304

But it's time-consuming and inconvenient to see this fact. In other words, it was spun to favor the hypothesis of the researchers. The same hypothesis which this data resoundingly fails to confirm.

Since the reporters chose a headline expressed as "circumcision (or not) may (or may not) cut (or not cut, or increase) HIV spread among gay men," it would be more in agreement with the study if the headline were "Circumcision may increase HIV spread among gay men."

If that sounds like an overly strong and misleading title, consider that the actual title ("Circumcision may not cut HIV among gay men") must then be more overly strong and misleading, since it is less true to the data.

The first paragraph of the article establishes relative risk reduction as the standard for this sort of study:

"A number of studies in African nations have found that circumcised heterosexual men were up to 60 percent less likely than uncircumcised men to contract HIV during the study periods."

Given the researcher bias in favor of the hypothesis, and the use of relative risk reduction figures for past studies, does anyone think this wouldn't have been expressed as a 13% reduction in HIV, though statistically insignificant, if it had been in favor of the circumcised group?

We are being spun by the researchers (*) while the data is withheld from the public.

(*) The reporters are not making things better.

No evidence circumcision works = It may work?

It's natural to seize any possible hope for slowing or stopping the spread of HIV. Some researchers have spent careers investigating the hypothesis that less penis will lead to less HIV. But even when evidence emerges that this hypothesis is wrong, it is spun as hope that it may still be right.

Today it is reported that in the latest study of 4,889 homosexual men, about 7% became infected with HIV over three years, regardless of whether they were missing part of their penis.

circumcised and uncircumcised men showed no difference in the risk of HIV infection over three years.

while having unprotected sex with an HIV-positive partner increased a man's risk of infection, there was no evidence that circumcision altered that risk.

The report also makes clear that no previous studies have found circumcision helpful in preventing HIV in gay men, either:

There has so far been no good evidence that circumcision lowers HIV risk among these men.

Despite all the speculation, hope, and expense aimed at discovering that it does, there is no evidence that reducing the penis also reduces HIV infection. There is, however, good evidence that it does not.

But those looking for such a link still have hope. If only they could receive more funding, and perform larger studies, perhaps they may find that if gay men only had their foreskins removed, they might get less HIV.

This hope can be seen in the language in the story, which reflects the views of the researchers who would like more funding to perform more studies. Let's compare what is actually known with what is stated:

evidence that circumcision does nothing --> "may not do much"
evidence that circumcision clearly did not help --> "unclear"
circumcision made no difference --> "may not make much difference"

Is this simply the sort of sensible caution which should always be applied to medical study results? If so, we would expect the same caution to be applied to the evidence which found reduced HIV transmission to heterosexual adult men in Africa. Those findings should be similarly stated with prudent scientific uncertainty. Let's see what the same article says about those studies:

A number of studies in African nations have found that circumcised heterosexual men were up to 60 percent less likely than uncircumcised men to contract HIV during the study periods.

Nope. It's not even-handed caution. Evidence which favors circumcision is stated as a fact. Despite weaknesses in those studies, there's no hedging whatsoever. No admission that they could be wrong. No call for further evidence. But when evidence emerges that circumcision is useless, the headline tells us it still may work, and the researchers tell us it needs more study.

future studies, with larger samples of uncircumcised men, should continue to look at the question of circumcision and HIV transmission among men who have sex with men.

I'm just going to come out and say it. Some of these researchers have a cultural bias. They favor circumcision for reasons unrelated to HIV prevention, and they want more funding to study circumcision. Some of the reporters, too, are culturally biased to accept information favorable to circumcision.