Astonishingly, demand for circumcisions is lower than expected

"Just lie back and think of the average treatment effect"

“But wait, look at the treatment effect results”

Thanks to three randomised controlled trials held in several sub-Saharan countries, it is now fairly clear that male circumcision leads to a sizeable reduction in the probability that a man will go on to contract HIV. The treatment effect of getting snipped is nothing to laugh at: most estimates put it at a 50-60% reduction in risk. I’m not sure that anyone has actually ever bothered to do a cost-benefit analysis, but as a policy it seems like a no-brainer to the global health community, who has been quick to support it as an effective means of battling the AIDS epidemic.

Like with most `new’ technologies which have been shown to be effective in a near-laboratory setting, the next step was to roll-out and scale up the intervention. Circumcision campaigns with very ambitious targets kicked off in several SSA countries, many of them supported by PEPFAR. Several years down the road, it seems that many countries are struggling to get anywhere near their goals, as is the case with Swaziland, where only 20% of the targeted demographic were `reached’.

Surprise, surprise, the biggest issue seems to be that of low demand: men in Swaziland don’t appear to be particularly keen to forfeit their foreskin, for a host of reasons, ranging from personal doubts over efficacy, to concerns over pain, to worries about witchcraft:

“That’s also what I wanted to know, and they wouldn’t tell me – what happens to my foreskin once it is cut off?” said Mduli. 

Health Minister Xaba alluded to this when he told the Times of Swaziland, “Some men feared that the foreskin could end up in wrong hands, being used by some unscrupulous people for their ulterior motives.”

The lack of coverage in the Swaziland campaign seems to be indicative of similar campaigns in Kenya, Uganda and Zimbabwe: there is initially a surge in demand as large groups of exceptionally-enthusiastic men line up to be circumcised (similar to the zeal that the Merry Men in Robin Hood: Men in Tights show before Rabbi Tuckman reveals to them what circumcision actually is), but eventually demand levels off and most of the population remains untreated.

This problem isn’t unique to circumcision – there are a lot of `treatments’ which have shown to be effective using rigorous methods, but have failed to scale due to lack of demand. Unfortunately, most randomistas (rightfully) spent there time trying to establish that treatment effects exist before worrying about treatment demand, but then fail to spend as much time figuring out how these things should be rolled out; although sometimes both questions can – to some extent – be answered at once if we are randomly inducing people to take up a treatment, rather than randomising the treatment itself. These questions are not only important for a policy perspective, but they might also give us more insight as to what the actual real or perceived benefits are once we move out of the experimental setting.

Circumcision is a particularly difficult area in this regard. We know very little about how to increase demand for circumcision. Can you tell me what a man’s marginal willingness-to-sell a foreskin is? It is also a difficult subject to cover dispassionately – I suspect that for the already-circumcised, the marginal willingness-to-pay for a foreskin is closer to zero. For many of the uncircumcised, the marginal willingness-to-pay  keep one’s foreskin is probably very, very high. Given the extensive roll-out of circumcision programmes in southern Africa, there will be some incentives to experiment with ways of creating demand here. For example, Berk Ă–zler at the World Bank has suggested that it might be time to start paying men to get circumcised (or at least  begin compensating them for lost wages).

Yet, this brings us to the second problem with the roll-out of circumcision programmes as well as the different methods campaigns might employ to entice men onto the chopping block. We know that, within a relatively short time frame, circumcision significantly reduces a man’s chance of contracting HIV. Yet we know substantially less about how subsequent behaviour might play a role in overall HIV risk down the road. Many have been concerned about Peltzman effects, the tendency for people to do more of an activity when it becomes (what the medical literature is calling behaviour risk compensation). As some have pointed out, even if circumcised men don’t increase their sexual activity enough to offset the effects of circumcision, the effect of transmission on women (who are not protected by male circumcision) will go unmitigated. There hasn’t been nearly enough work done on the behavioural effects of circumcision, but at least one paper, which I wrote about some time ago, has suggested that Peltzman effects are, on average, not a problem. However, the same paper found a bifurcation in behaviour: men who had been circumcised AND believed in the treatment effect reduced their risky sexual behaviour, but those who had been circumcised and did not believe in its effectiveness increased their risky sexual behaviour.

Consider this result in light of some of the proposed methods to get more men circumcised: methods which do not actually involve changing beliefs, such as paying people to get circumcised, might lead to more men who don’t believe in the power of the snip signing up, precisely the men who will end up increasing their risky sexual behaviour. This is quite scary.

I’ve written before about how I’m deeply sceptical about the success of mass-circumcision campaigns. However, even a sceptic wants things to be done right if they’re going to be done at all. It shouldn’t be so astonishing that most men aren’t dashing to the clinic to receive a circumcision. The global health community will need to do more rigorous research on how to get men into clinics, but at the same time that research should be leveraged to make sure that these campaigns are having their intended impact.

3 thoughts on “Astonishingly, demand for circumcisions is lower than expected

  1. Hugh7

    June 4, 2013 at 8:09am

    Thank you for your skepticism about the circumcision campaigns. You could be a lot more skeptical of the studies on which they are based.

    “most estimates put it at a 50-60% reduction in risk”.

    No, in three trials by interconnected researchers (some with a previous track record of promoting circumcision) with many of the same methodological flaws, a total of 5,400 paid volunteers were circumcised and similar numbers told to wait. The trials were not, perforce, double blinded or placebo controlled, leaving the way wide open for experimenter and experimentee effects. (A dummy operation resembling circumcision would have given at least partial control and greater parity.) The experimental group (only) was told to abstain from s3x for six weeks, or use condoms if they could not, giving them the experience and habit of using condoms. If men showed signs of thinking circumcision had made them immune, they were given additional, intensive counselling. 703 dropped out, their HIV status unknown, 327 of them circumcised.

    After less than two years, a total of 64 of the circumcised men had HIV, 73 fewer than the control groups. That is the total evidence that circumcision has any effect. Contacts were not traced, so there is to proof that all or even any of the HIV was acquired s3xually, let alone heteros3xually. Homos3xuality is stigmatised in all three countries, but especially Uganda where it carries the threat and risk of death, so it would be underreported. Several men who reported no s3xual activity seroconverted. Transmission through contaminated medical or quasi-medical equipment is a distinct possibility.

    There are also problems with the documentation of the studies. The raw data have not been released, and only the Kenyan trial was prospectively registered. The South African trial was registered a year and a half after recruitment ended and seven days before the results were published! The Ugandan one a month after recruitment ended, and a month before publication.

    Women are not just “not protected by male circumcision”. In a subset of the Uganda study, Wawer et al. started to find that circumcising men increases the risk to women, but they cut the study short “for futility” before that could be confirmed. Women are already at much greater risk from heteros3xual transmission than men, so a slight increase caused by circumcision (through abrasion by the keratinised glans on the v@gina, for example) could greatly increase the risk overall.

    In 10 out of 18 countries for which USAID has figures, more of the circumcised men than the non-circumcised have HIV. Some have blamed this on the style of customary circumcision in those countries, but any circumcision should have some effect.

    (And a pox on dirty-minded spambots!)

  2. S Nipit

    June 5, 2013 at 1:18am

    The Gambella region of Ethiopia is an outlier. Twenty-seven thousand circumcisions in a population of less than half a million. What is different there and what can be learned by others?

  3. Eric Djimeu and Annette Brown

    June 5, 2013 at 9:25pm

    Thanks for your interesting blog post highlighting a very real, and as you say, not too surprising, problem with the campaign to increase voluntary medical male circumcision in sub-Saharan Africa. The international donors are well aware of the demand-side challenges, and in fact, there is an effort underway to identify possible solutions and to test them rigorously. 3ie just closed a call for proposals under a grants window that will fund pilot programs to increase the demand for VMMC. Each pilot program will be accompanied by an impact evaluation to determine whether the program has an attributable impact on demand. We expect to announce awards in mid July and to have some actionable findings in about year. (This grants program is funded by the Bill & Melinda Gates Foundation, and more information is available here:

    In addition, there is at least one existing study on the demand for VMMC and the sensitivity to price. It was conducted by Chinkhumba, Godlonton, and Thornton.

Comments are closed.