Circumcision for HIV needs follow-up

Worries about behaviour change require further research.

The French National AIDS Council (CNS), an independent government advisory body on AIDS issues, last week urged prudence in implementing male-circumcision programmes to reduce the spread of HIV. It cautioned that over-zealous roll out could lead to a false sense of security and exacerbate the problem.

In March, the World Health Organization (WHO) endorsed the promotion of male circumcision as a tool for HIV prevention, alongside condoms and antiretroviral drugs. The move was based on a WHO expert consultation, which concluded that the evidence for the intervention's efficacy was “compelling”. Recent criticism of this view has been received with anger (see 'Cutting criticism').

The strongest evidence comes from three recent studies. In 2005, a study¹ of 3,300 heterosexual men living in and around Orange Farm, South Africa, by France's National AIDS Research Agency, showed that circumcising men reduced the risk of infection by 60%. Similar levels of protection were found in 2007 by American-funded studies in Kisumu², Kenya, and Rakai³, Uganda.

Taken together with results from observational studies, this is “as convincing evidence as one ever gets in public health,” says Helen Weiss, a statistical epidemiologist at the London School of Hygiene & Tropical Medicine.

Circumcision reduces the spread of HIV, but needs to be combined with public education. Credit: M. SAFODIEN/AFP/GETTY

But questions remain about the intervention's applicability as a preventative tool on a large scale, says Willy Rozenbaum of the Pierre and Marie Curie University in Paris, who led the CNS study. Rozenbaum questions the speed with which the WHO has acted, and thinks the organization's endorsement — although detailing the caveats — has been misunderstood in some quarters as saying circumcision is a “miracle solution”.

The impact of circumcision on HIV prevalence will depend on a host of social and cultural factors.

Rozenbaum notes that the effect of circumcision on HIV prevalence in a population will depend on a host of social and cultural factors. There is a real risk, he says, that after circumcision men may have a false sense of security and increase their number of partners, or dispense with condoms. The report also points out that circumcision leaves men more vulnerable to infection if they don't refrain from having sex until the wounds have fully healed and that women may find it more difficult to insist on condom use with circumcised men.

Given these concerns, the CNS calls for implementation schemes to be accompanied by public education as well as research that considers the behavioural and cultural factors that affect HIV transmission. Such research is already under way at various sites. At Orange Farm, a 5-year follow-up study of 30,000 people will look at how the implementation of a real circumcision campaign affects risk behaviours, and overall HIV prevalence in the population.

Tim Farley, an official in the WHO's HIV prevention team in Geneva, says he “absolutely agrees” with the CNS view, and the need for such research. “These are concerns that need to be studied, but they are not a reason to stop,” he says. “We must move ahead in the knowledge that there could be riskier behaviour.”

Box 1: Cutting criticism

A controversial recent publication argues that the evidence from cross-country statistics does not support male circumcision as a key intervention against HIV. But according to several angry AIDS researchers, the paper merely shows that the peer-review system of the journal that published it, PLoS One, failed on this occasion. The study, they say, is flawed and, moreover, concerns a debate over statistical techniques that, in this instance, have been largely superseded by more powerful clinical trials.

“Size matters: the number of prostitutes and the global HIV/AIDS pandemic,” by John Talbott⁴, a former investment banker, was published on 20 June. In a comparison of HIV prevalence across 77 countries, it argues that the effect of male circumcision is overstated, and that the prevalence of prostitution is the dominant factor. Among other things, Talbott takes issue with the way that some previous studies have 'weighted' circumcision figures according to the size of national populations.

The paper adds nothing substantial to our state of knowledge.

Ecological analyses such as Talbott's, which look for correlations between HIV prevalence and various factors, are useful for generating hypotheses, but don't prove anything, says Anne Buvé, an expert in HIV prevention at the Institute of Tropical Medicine in Antwerp, Belgium. In her opinion, the paper consists of “sweeping generalizations that add nothing substantial to our state of knowledge”.

In understanding the effect of male circumcision “we are past this sort of ecological research”, says Buvé, pointing to recent randomized clinical trials showing a major protective effect. Talbott's paper mentions one such study in passing. Talbott declined to comment for publication.

“Our data were never intended to be used in this way,” says Buvé, a co-author on the 2006 paper from which Talbott took the preliminary data that he draws many of his figures on prostitution from⁵. A host of factors leads to under- and overestimates of the number of prostitutes nationally. There are also problems in using the Muslim proportion of the population as a surrogate for the level of circumcision, experts say.

Nonetheless, the press release made available by PLoS One describes the paper as containing “important findings that should impact policy decisions”. It recommends “mandatory” testing and treatment of prostitutes, adding that opposition to such control methods is comparable to “claiming that a rabid dog must be allowed to run free in a neighborhood regardless of how many men women and children he infected and killed”.

These are Talbott's words. It is PLoS One's normal practice to publicize any text supplied by the author as a press release, adding a brief disclaimer at the end.

Talbott's paper is a baptism of fire for the young journal, launched in August 2006. The journal is testing a new publishing model, whereby papers are published quickly after being assessed by peer reviewers on technical soundness alone. The idea is that the papers' reliability and significance will be further ascertained by online comments and discussion by other scientists that are posted on PloS One's website after publication.

“The paper is total drivel, it should have been picked up in the review process,” claims Tim Farley, an official in the World Health Organization's HIV-prevention team in Geneva. “And coming from PLoS One gives [the views in the press release] a public perception of validity. In public health there are severe dangers in such stuff getting through.”

“There are lessons to be learned from all papers that we publish; we are a young journal,” says Chris Surridge, PLoS One's managing editor, adding that its peer-review model is constantly under refinement. “We are feeling our way.”

In this case, the paper was reviewed by a member of PloS One's editorial board and one external referee. The review criteria of PLoS One are that “the data are sufficient to be published and the conclusions not radically overstated by the data,” Surridge explains. “I hope we get some discussion of this paper [in the online post-publication comments].”

Declan Butler and Lucy Odling-Smee