HIV, Africa, circumcision, and PBS Kids

My daughter really likes Cyberchase. She likes it because it is all about how math can solve problems. If you're unfamiliar with the show, a crew of three ordinary kids and their Gilbert Gottfried-voiced toucan-looking cyberbird get to tool around cyberspace helping the beneficent leader, Motherboard, to beat back the nefarious schemes of the evil Christopher Lloyd-voiced Hacker.

It's not that often that kids' shows interact with real life, but yesterday was one of those days. I was reading this study about how men were circumsized for a random trial in Kenya to see how much circumcision cut the risk of HIV infection. It has been recognized for many years that circumsized men in Africa have a lower infection rate from heterosexual sex than do non-circumsized men. But circumcision is a fairly drastic intervention compared to other methods to prevent infection, notably condoms. It's one thing to observe that men who undergo circumcision for cultural reasons have a lower infection rate, it's another to suggest that all men get circumcized as a routine prophylactic measure.

The study reported yesterday was fairly remarkable:

The clinical trial, funded by the National Institute of Allergy and Infectious Diseases and the Canadian Institute of Health Research, enrolled 2,784 HIV negative, uncircumcised men between 18 and 24 years old in Kisumu, Kenya.
Half the men were randomly assigned to circumcision, half remained uncircumcised. All men enrolled in the study received free HIV testing and counseling, medical care, tests and treatment for sexually transmitted infections, condoms and behavioral risk counseling for 24 months.
Study results show that 22 of the 1,393 circumcised men in the study contracted HIV, compared to 47 of the 1,391 uncircumcised men. In other words, circumcised men had 53 percent fewer HIV infections than uncircumcised men.

</p>

The report indicates that all the men in the study tended to reduce their sexual risk behavior, although it doesn't indicate whether that was because of the availability of condoms, medical care, information, or because volunteers were more likely to be health-conscious to begin with.

The conclusion of the report was clear -- based on a 53 percent reduction in HIV infection rate,

Until now, public health organizations have not supported circumcision as a method of HIV prevention due to a lack of randomized controlled trials.
"With these findings, the evidence is now available for donor and normative agencies, like WHO and UNAIDS, to actively promote circumcision in a safe context and along with other HIV prevention strategies," [study PI Robert] Bailey said.

Back to Cyberchase.

You see, in yesterday's episode, Hacker was carrying out a nefarious plot to destroy the Cybrary:

Hacker poses as a virus exterminator. He fools Ms. Fileshare into entering her lair, the Cybrary, with bogus bar graph comparisons. Hacker releases his bugs, but it's Hacker who pays the price with a bout of amnesia. (He's forced to work for Buzz!) But the bugs are still out in the Cybrary, seizing control of various sections. Ms. Fileshare insists the CyberSquad that, if she is to believe them, they must provide an easy way to compare the number of bugs in each Cybrary section.

How could Hacker possibly get hired as a virus exterminator, you wonder? Well, he showed the Cybrarian Ms. Fileshare two graphs -- the number of bugs he eliminated, and the number eliminated by his competitors. Hacker's bar was much higher, and Ms. Fileshare assumed that Hacker had eliminated many more bugs than his competition.

But there was a catch. Hacker's two graphs had different scales. Even though his bar was much higher, it represented only a few bugs. His competitor had a much lower bar, but it had a much smaller scale, so it represented many more bugs than Hacker's graph.

In other words, Cyberchase is teaching my daughter and who knows how many other kids just how people can lie with statistics. A very good thing, I would say.

So what does Cyberchase have to do with HIV infection rates? Well, if you're clever you'll have already figured it out. Look at those numbers again:

Study results show that 22 of the 1,393 circumcised men in the study contracted HIV, compared to 47 of the 1,391 uncircumcised men. In other words, circumcised men had 53 percent fewer HIV infections than uncircumcised men.

</p>

In other words, uncircumcised men in the study had a 3.4 percent infection rate, and circumcised men had a 1.6 percent infection rate. True, that is a 53 percent reduction, and every case of HIV is well worth preventing.

But how many men would sign up for circumcision if they knew that their actual risk reduction was the mere 1.8 percent difference between the two numbers? And how does this reduction compare to other measures, like the consistent use of condoms?

The 1.8 percent reduction puts this quote into another perspective:

The study also evaluated the safety of circumcision in a community health clinic with specially trained practitioners. There were no severe or lasting complications from circumcision. However, 1.7 percent of surgeries resulted in mild complications, such as bleeding or infection.

Granted that HIV infection is much more serious than most instances of complications from circumcision. But the study is assuming that circumcision has no other downsides, which is far from obvious.

The study was stopped before its scheduled completion, because the reduction in risk was significant. But arguably, this reduction would itself reduce as the study went on longer -- men who maintained high-risk sexual activities might see a delay in infection with circumcision, but if those activities continued, they would likely ultimately be infected. In other words, we can't assume that the reduction in risk would project linearly with time. Instead, as time passes, the high-risk population would become increasingly saturated with infections. If this model held, then circumcision would become less and less effective over time.

Increasing the rate of circumcisions might make a big system-wide difference. Every infection has a cascade of economic and social effects, and every prevented infection limits the rate of spread of the epidemic.

But the costs of circumcision very well might exceed the reduced risks -- especially if high-risk sexual behavior is not reduced as well. To me, the study is sufficient to suggest that already-established circumcision traditions should be encouraged, but not enough to suggest that it should be adopted en masse by other populations. It almost certainly shouldn't be given money that would otherwise go to increase consistent condom use.

And consider yourself blessed if, like me, you have a daughter who likes math!