Male circumcision is a hot topic among academics and policy-makers alike. A recent randomized controlled clinical trial suggests that the decrease in transmission rates from circumcision could rival those from putative AIDS vaccines (around a 60% in the odds of contracting HIV).
This is quite impressive, but, as my colleague Brendan here in Cape Town suggests, the public health benefits of a wide-scale circumcision program will likely be smaller than what is seen in clinical trials. The argument is that people will be encouraged to take up circumcision because it will lower their risk of contracting the disease. Men may who choose to get circumcised may then engage in more risky sexual behavior thereafter as a response to lower perceived risks, thus undoing some of the benefits of circumcision seen in the clinical evidence. The take home point is that policymakers should be careful when extrapolating from randomized clinical trials, and that behavioral responses to new public health policies may alter (sometimes radically) the efficacy of new interventions.
On a related note: an interesting piece from 2006 in the Quarterly Journal of Economics examines the effects of the development and roll-out of HAART on HIV incidence rates. The authors find that people on HAART take up more sexual partners as a result of being healthier. While the infectiousness of those on HAART is lower from those who are not treated, the authors contend that this phenomenon increases the risk of an HIV- individual contracting the disease: more HIV+ are engaged in sexual activity. They go on to argue that HAART may lower the welfare for those who are not infected with the virus. (Technical note: to establish causality, the authors use state-level differences in Medicaid generosity towards HAART to predict HAART use, and then use that predicted value in their models for risky behaviors).
8 comments:
randomly assigning individuals to circumcision when we know if prevents HIV infection seems potentially unethical to me. but probably not in comparison to the type of constant gardner stuff probably going on in some other trials.
Since we knew before these 3 large studies were done that most of the half-million US men who had died of AIDS were circumcised at birth, then even undertaking the studies was immoral. Circumcision does not prevent AIDS.
I think the justification for the trials was that we did not have causal evidence of the circumcision effect: there are a lot of reasons why people may get circumcised, many of which be related to risky sexual practices (or the lack thereof) that might influence AIDS. Also on the point of ethics, at least one of the large trials was stopped early because the difference between treatment and control was so significant. (I'll leave James to fill in the details if he has time...
But issue that the issue of circumcision and risk has been around for at least 20 years. People suspected a link between circumcision and HIV/AIDS for at least two decades before going after scientific evidence. This seems irresponsible to me, especially if you believe that circumcision may have an impact on the epidemic. My colleague, Nathaniel Cogley at Yale makes this argument forcefully, and has an interesting piece on the lag between recognition and getting evidence.
Finally, regarding tic's point, the trials do not suggest that those who are circumcised do not get HIV/AIDS. Far from it. They only suggest a reduced risk of transmission. I'm guessing of the half a million men you mention, many of these were gay men, and sexual practices between men and men are supposedly more risky (I think, again James can shed more light on this.) In addition, we would also need to know the rate of transmission in the non-circumcised group to make a meaningful comparison.
Tic,
On your point about circumcision in the US, you must have been referring to a recent study that looked at male-male transmission rates among circumcised v. non-circumcised. I just saw a link to it at Chris Blattman's blog:
http://chrisblattman.blogspot.com/2008/10/circumcisions-hiv-impact-questioned.html
Your point is well taken. Perhaps the extra risk from male-male sex reduces any of the impact of circumcision vis-a-vis male-female sex?
Definitely worthy of further exploration...
Cheers, and thanks for your comment!
Atheen
There are many points to take issue with about the RCTs.
While the number taking part was large, the number infected was small, especially since the trials were cut short - much smaller than the number who dropped out, their HIV status unknown. They were encouraged to be tested elsewhere, as it was deemed "unethical" to tell them their test results within the trials (regardless of the hazard to their partners). Learning you were HIV+ after a painful and marking operation to prevent it would be a powerful inducement not to go back - more so than other outcomes.
The trials were not (of course) double-blinded or placebo-controlled, and the two arms of the trial were not treated identically. Those being circumcised were told to abstain from sex for six weeks afterwards or use condoms if they did not. There is some suggestion that circumcised men who thought they were risk-free were given additional counselling.
It is noteworthy that the same few researchers promoted circumcision before the trials (one of the most prominent, Daniel Halperin, has claimed it's his "destiny" to do so - hardly scientific), held the trials claiming to show circumcision is preventive, harmless, and does not increase risky behaviour or harm sexuality*, attended the closed WHO/UNAIDS meeting promoting circumcision, and now tirelessly work the media.
*Their test was so coarse that virtually all participants claimed near-perfect sex whether circumcised or not - how very different from the rest of us!
For strange cultural reasons, circumcision gets a free ride in the media that any other quick fix would not. Many circumcised men will walk over broken glass defending what was done to them.
I'm intrigued at the strong remarks this topic has roused, especially in a non-HIV circle.
Male to male transmission has a markedly different biology than heterosexual, obviously, so it's not a surprise that some observational studies suggest no benefit to speak of. I think when you have the possibility of being both a recipient as well as a provider of penetrative sex, it complicates things considerably, and that measures aimed at protecting the glans will clearly not extend to other areas of the body.
I'm intrigued by some of the criticisms levied by Hugh. Looking at just one study, the Kenya study by Bailey et al, the lost-to-follow up (for whatever reason) was roughly equal between the two groups-9.1% in the circumcised, 8.2% in those not. This is compared to 21% and 4.2% becoming HIV + respectively in the two groups. While you could certainly conjure a variety of reasons why those missing in the circumcised group were more likely to be HIV + than those in the control group, I think it's a difficult argument to sustain given 2 other RTCs in different settings as well as the range of observational trials that fueled the RTCs to begin with (which were indeed heavily confounded, thus the need for randomization, as Atheen suggested). While I'm not as rabid a proponent as Halperin, I do think his basic point stands well: circumcision provides better protection that any other single-point intervention we have. It is relatively safe, relatively cheap, easily scaled up, and while it may certainly cause disinhibition (a concern with any number of HIV interventions), there is an onus to demonstrate that we cannot control that with intensive counseling.
My two cents.
James: you are right that the number of circumcised and control men dropping out were roughly equal, but one can also imagine that learning you were not HIV+ would be a reason for control group men to lose interest in the trials and drop out. We simply don't know why anyone dropped out, they are a confounding factor. Whatever they are, why should the same reasons for differential dropping out not apply in the other two trials? They were not all that different - or independent: the researchers for all three are known to each other and have published variously together.
The observational studies have problems of their own. In most instances, the circumcised men in those were Muslim, with various sexual prohibitions (and a prohibition on alcohol) that would have restricted risky behaviour.
All studies have been heavily biased towards circumcision. Nobody had been looking for example, for any effects that might make it a risk factor for HIV. (One study has shown that female genital cutting appears to be protective against HIV, but the reseachers bent over backward to try to explain that away.)
The researchers raced to the media with their findings, even before the studies had been published, and policymakers in Uganda have complained that they were not told the experiements were going on.
Post a Comment