Our family
« So the afterlife is meant to be considered as a reward? | Main | A day in the park »

Facts and Myths about HIV & Circumcision

Fact: Circumcision protects against HIV infection

There are three tiers of evidence for the protective effect of circumcision against HIV infection. Firstly, there are the epidemiological observations, where rates of HIV in circumcised and uncircumcised populations are compared. Secondly, there are the case-control observations, where rates of HIV in circumcised and uncircumcised individuals are compared. And thirdly, there are the randomised clinical trials, where men are assigned to either circumcision or no circumcision and the effect of future HIV infection is compared. 

We can deal with these in turn. The first are the epidemiological surveys. There are multiple relevant studies, all with similar effects, but one of the best designed is the multicentre study set in four cities in different regions of Africa. These studies show a much lower rate of HIV infection in west and north Africa compared to east and south Africa. The infection prevalence closely mirrors the religious border, with lower rates in Muslim Africa and higher rates in Christian Africa. Despite the glee with which certain Muslim scholars touted this as a representation of increased sexual restraint among Muslims, the multicentre study showed very few differences in sexual activity (number of sex partners, prostitution levels, etc). This is not evidence for the role of circumcision in protection against HIV, but it is very strong evidence that something is different between these two communities and that it has a strong role in protection against HIV.

The epidemiological studies lead multiple groups of researchers to investigate the circumcision hypothesis using case-control experiments (comparing the infection rate in circumcised and uncircumcised men). With dozens of different studies, all of various quality, the best way to assess the results is from the systematic reviews that have been performed. General population studies. This systematic review in 2005 looked at all 36 studies into circumcision that had been performed to date. Among the 18 general population studies, seven studies showed a protective effect and two studies showed a harmful effect (right). The difficulty of general population studies is that the rate of HIV infection is low enough that it can be difficult to control for bias and to generate enough statistical power. High-risk studies, by contrast, tend to have higher HIV rates and to have less bias in risk factors, often leading to additional statistical power. Among the 18 high-risk group studies, 13 studies showed a protective effect and no studies showed a harmful effect (left). High-risk population studies Thus, over all, of the 36 studies, 20 showed a significant protective effect, 2 showed a significant detrimental effect, and 14 had insufficient power to draw a conclusion. For anyone who is used to looking at research on inbred mice or the like, this data looks very noisy, however with the immense variation of behaviour and exposure among the human population this type of noise is typical, and the noise in the results are comparable to that observed in condom use studies. One interesting observation is that the studies where circumcision status was actually assessed by the trial nurses showed a stronger (beneficial) effect than the studies where circumcision status was self-reported, other studies show that the error-rate in self-reporting circumcision can be as high as 10%. Overall, the average effect of these trials is a 60% protective effect on HIV infection, again comparable in scope to the average effect observed in condom use studies (~80%).

The key criticism of any case-control study is that there may be confounding effects. When these confounding effects are known (eg number of sexual partners) they can be controlled for, but when confounding effects are unknown they cannot be controlled for. It is therefore always theoretically possible that there is some unknown confounding effect that has a strong correlation with circumcision and is protective against HIV infection. The only way to control for this possibility is to have a randomised clinical trial, where HIV-negative men enroll and are then randomly assigned to either the control group or the circumcision group. In this ideal experiment any confounding factors will be randomised to the two different groups and the effect only of the treatment can be identified. This randomised clinical trial design is the exact experiment performed by three independent groups.  

The study by Auvert et al enrolled a total of 3,274 uncircumcised men in South Africa, tested for HIV and assigned half to be circumcised. The group then followed up both cohorts for HIV status, condom use, sexual activity and so forth, and found a 60% protective effect for HIV infection among the circumcised group. Condom use, sexual activity and the like were nearly identical among the two groups, normalisation for these factors resulted in a 61% protective effect. The study by Bailey et al enrolled uncircumcised 2784 men in Kenya, tested for HIV, assigned half to be circumcised and again followed up for HIV infection and behaviour change. Again, no changes in sexual behaviour were observed and the risk of HIV was reduced by 60%. Finally, the study by Gray et al, with a similar design in Uganda, enrolled 4996 uncircumcised men and found a net protective effect of 60%. All three trials were independently run along best practice guidelines with blinded tested, yet all three trials found the identical effect of 60% protection - which was also the average protective effect observed in the case-control studies. Together, with multiple independent lines of evidence pointing towards the same result, the effect can be considered conclusive, to the point that it is now considered unethical to conduct more clinical trials as it would mean withholding treatment to the control group - the same way that we cannot ethically conduct more clinical trials on proven vaccines.

Several reoccurring criticisms come up for these three clinical trials. The most common objection is that all three trials were stopped early. This is true, however they were specifically stopped early by the ethical board review because the results were clear early on. It is now a built-in feature to clinicial trials that intermittent review will take place and the trials will be halted if adverse events surpass a particular level (so that excess participants are not exposed to the treatment) or if the protective effect surpasses a particular level (because it is considered unethical to withhold the treatment from the control group at this point). This is not a unique feature of the circumcision trials and is an agreed upon compromise between getting perfect scientific results and treating the participants of the trial in an ethical manner. The other main criticism that is raised is that the control group for circumcision is not like a traditional placebo - the trial doctors are blinded but the participants are not, and those assigned to the circumcision group may drop out at higher rates, creating a bias. While theoretically possible, each of the three studies investigated this possibility by looking at the drop-out rate. For example the Gray study found that out of 4996 enrollments, only 37 dropped out (24 in the circumcision group and 13 in the control group), not enough to create any substantial bias.

One further comment. The direct protective effect of circumcision is only known to protect men during vaginal intercourse. It is also likely to protect men during anal intercourse, but this has not been studied. It provides little to no direct protection to the woman, however mathematical modelling suggests that when the take-up of circumcision reaches 50% the "herd immunity" effect would reduce HIV infection among females and uncircumcised men by 25-30%. While not exactly a "silver bullet", this would make an impact to millions of people within southern Africa, where existing circumcision rates are low.


Myth: the foreskin must be functional or it would have been eliminated by evolution

This myth comes in two flavours. The first is that because it is present it must be functional, the second that if it actually was detrimental in HIV infection it would be selected against. As to the first, evolution does tend to result in the loss of anatomical features with no function, however there are strong exceptions for sexually dimorphic features. Thus a male nipple has no function, but there is strong sexual selection to keep the female nipple. In the absence of selection to create a suppressive pathway against nipple development in males, the useless male nipple is maintained. In all likelihood the foreskin is similar to the male nipple, as the male relic of the female labium. As to the second argument, it must be remembered that evolution is responsive, not predictive. Prior to the entry of widespread HIV infection there would have been no evolutionary pressure against the foreskin. If HIV had been a common infection for millions of years, the continued existence of a foreskin would indeed be a mystery, but this is not the case.


Myth: HIV protection is just a matter of cleanliness

A commonly stated myth about circumcision is that the HIV protection is simply due to the ease of keeping the circumcised penis clean and that good hygiene would replicate the effect. As a starting hypothesis, this is not an unreasonable model to test. A prediction of this model would be that circumcision would protect against a broad range of sexually transmitted infection (STIs), as the "cleanliness hypothesis" would not predict any special status for HIV protection. The ability of circumcision to protect against multiple STIs has been tested in epidemiological studies and randomised trials and so far no effect has been reliably measured for any STI other than HIV. It is possible that for some rare STIs in addition to HIV there may be protective effects and also possible that there is a weak effect against HSV, but to date the evidence against the "cleanliness hypothesis" is very strong - the protective effect of circumcision does not appear to be due to differential cleanliness and is almost certainly due to the unique biological properties of HIV outlined below.


Myth: there is no known biological mechanism to explain the protective effect of circumcision on HIV

The gold standard for incorporating a technique into evidence-based medicine is success in randomised clinical trials, as already proven for circumcision. However medical researchers prefer to understand the mechanism of protection for any intervention, as it allows optimisation or replacement with simpler strategies. It is sometimes claimed that there is no plausible mechanism by which circumcision protects against HIV, however a review of the literature demonstrates that the known biology of HIV suggests an optimal infection route via the foreskin. Unlike some sexually transmitted viruses, like HPV, that are able to directly infect through the skin, HIV is an exceptionally poor virus at crossing the epidermal barrier - purified HIV placed on the skin will remain safely external. Instead, HIV has to rely on two different mechanisms to breach the epidermal barrier - microabrasions and cellular trafficking.

Microabrasions are small tears in the skin barrier which expose the inner tissue and blood to the environment, allowing a direct passageway for HIV to enter. One common cause of microabrasions are sexually transmitted diseases, which often form small ulcers to allow increased shedding. These ulcers allow the reverse infection of HIV, which is why the transmission rate of HIV increases 100-fold with coinfection with other sexually transmitted infections (such as HSV-2). This accounts for the recent data suggesting that anti-HSV-2 treatment programs may reduce HIV spread. The skin of different organs is more or less prone to microabrasions. The mucosa of the anus is the thinnest, followed by the vagina, followed by the oral cavity, followed by the penis, which correlates with the increasing risk of HIV acquisition per sexual act (anal receptive > vaginal receptive > oral receptive > insertive).


The second mechanism is that of cellular traffic. HIV infects through the CD4 receptor, using the coreceptors CCR5 and CXCR4. The expression pattern of these receptors limits the infection of HIV to CD4 T cells, macrophages and dendritic cells. Typically, these cells are found in circulation (which is why intravenous injection of HIV in contaminated blood provides the highest efficiency infectious route), but activated CD4 T cells and naive dendritic cells also circulate into the tissue. In the skin, the top layer is a keratinised barrier with dead cells, with the living tissue deep below this layer. The mucosa is quite different - as a functional interface it requires living cells to directly border the environment. While most of these cells are epithelial in origin, and hence not infected by HIV, dendritic cells lie just below the surface. The reason for this is the role of dendritic cells in antigen sampling, ironically a defense mechanism against common mucosal pathogens. Critically, these dendritic cells do not only lie just below the surface, but the also push thin dendrites through the epithelial cell barrier so that they directly contact the surface (right). We even know exactly how the dendritic cells form these dendrites, as a key paper in Science demonstrated that dendritic cells that lack the chemokine receptor CX3CR1 still home to the epithelial cell surface, but they are unable to produce the dendrites that penetrate to the surface (left).

With regards to circumcision, the key risk is the region of the inner foreskin, which has more in common with the mucosal surface of the vagina than with the keratinised surface of the rest of the penis. During an erection the inner foreskin of the uncircumcised penis is exposed (right), creating a region of relatively thin tissue that does not exist on the surface of an erect circumcised penis. This is the tissue that is thinner and populated by surface level dendritic cells, so it is also the tissue which is most prone to microabrasions and to cellular trafficking via infected dendritic cells. In the circumcised penis this tissue is absent, with the region covered in a thicker layer of non-mucosal skin. It is therefore likely that the biological mechanism of circumcision protection is simply the removal of this mucosal surface during intercourse.


Fact: Condoms are more protective than circumcision

The protective effect of circumcision on HIV is around a 60% lifetime protection. For single event condom use the protective factor is around 99% (with a 1.6% slippage factor), which results in an 80% protection rate. When condom usage is accompanied by sex ed classes on how to use a condom correctly the lifetime protection rate goes up to 95%. Clearly a correctly used condom is more protective than circumcision.

However, it is important to note that this does not mean that circumcision has no added value. In the randomised control trials men were still advised to wear condoms, but as you might expect 100% condom usage was not achieved (total condom usage was the same in both groups). The protective effect of circumcision in these trials is therefore an additive effect on top of typical condom usage. Public health is an experimental science and it needs to differentiate between the ideal effects of a treatment and the actual effects of implementation. For example, assuming that all sex was consensual (clearly not the case), voluntary abstinence would block the transmission of HIV. The ideal effect is therefore 100% protection. What happens when abstinence advice is rolled out as a campaign? Absolutely nothing. Circumcision may provide little additional protection when combined with ideal condom use, but in terms of public health what matters is that it provides substantial protection when combined with actual condom use.


Myth: Religious circumcision originated because of the health benefits

A number of religious supporters have lept upon the scientific evidence for the protective effects of HIV as support for ritual religious circumcision. They tout the proposition that the religious tradition of circumcision is validated by the scientific evidence, which therefore validates other aspects of their religion. This is by far an overly generous idea, for several important reasons:

1. While the western world tends to think of circumcision as the removal of the entire foreskin, anyone familiar with men would not be surprised to find out that religion has found many weird ways to manipulate the male penis. For example, there is the dorsal slit circumcision, where the foreskin is cut only along one side of the penis, leaving it flapping below. In some places it is then common to create a hole in the free foreskin and fold it back over the penis, sometimes called the "cowboy cut" as the result looks a little like a cowboy hat. While most of these traditional circumcisions have not been tested for protective effects, based on the biological mechanism of HIV protection, it is highly likely that only the full foreskin removal will result in substantial protection.

2. HIV only originated within the past 100 years, so any protective effect of circumcision would have been non-existent at the time these practices originated. As circumcision has little to no protective effect to other STIs, there is currently no scientific basis on which to claim the practice was beneficial at the time they originated.

3. Ritual circumcision in traditional contexts is highly dangerous. These surgical operations were carried out in non-sterile circumstances by untrained religious leaders. This stands in stark contrast to the modern non-surgical approach to circumcision, where typically a band is used to cut off blood circulation to the foreskin so that it falls off - in exactly the way the umbilical cord is removed, leaving behind the belly-button. While modern (secular) circumcision has extremely low rates of complication (on the order of 1.5% minor events such as swelling, 0% severe events), traditional/religious circumcision can have much higher rates of complication (with adverse event reports of over 10% reported, including severe events). The cost-benefit ratio of religious circumcision was therefore almost certainly a net negative, while the cost-benefit ratio of modern secular circumcision produces a net positive.


Myth: Circumcision reduces the pleasure of sex

This is a very common myth used in opposition of circumcision, often accompanied by an anecdote of some man they know who has a "botched" circumcision and now has pain during sex (anecdotes of uncircumcised men who have pain during sex are duly ignored). Fortunately, in science we can actually go beyond anecdotes and look at some hard data on sexual pleasure.

Who needs a peer-reviewed study when you have a placard and a website?Firstly, what are the effects of circumcision on subsequent adult sexual pleasure?

* when 1410 American 18-59 year old men were asked if they had "trouble achieving sexual gratification" in the past 12 months, around 45% of men reported sexual dysfunctional, with slightly higher rates in uncircumcised men. This small decrease in sexual dysfunctional in circumcised men remained significant even after controlling for variables such as race, age and sexual preference.

* the same study found that circumcised men had a more varied sexual practice, with more masturbation and oral sex, inconsistent with a hypothesis that sex is less enjoyable to circumcised men.

* Payne et al directly tested the sensitivity of circumcised and uncircumcised penises by measuring the response to touch on the ventral and dorsal surfaces. No difference was observed in sensitivity between the two groups.

* most studies are performed on men circumcised as infants, with relatively few men being circumcised as adults. The recent push for adult circumcision in Kenya has allowed a survey of men before and two years after adult circumcision (with a randomised control group). No increase was observed in sexual dysfunction and most men actually reported an increase in sexual pleasure (64% said their penis was "much more sensitive" and 55% said it was "much easier" to reach orgasm). A Ugandan study found that men circumcised as infants were more likely to have earlier and more promiscuous sex than uncircumcised men.

Not all studies find such strong results as the Kenyan survey, which suggests a strong increase in sexual pleasure. Indeed, the three randomised clinical trials for HIV protection found no change in sexual behaviour. Thus the conservative reading of these studies would be that there is no decrease in sexual pleasure among circumcised men, whether circumcised as infants or adults. The only plausible exception may be within the group of men who have religious-traditional (non-modern) circumcision, where relatively little study has been performed.


Myth: Circumcision is the male equivalent of female genital mutilation

Female genital mutilation is the practice of scraping away part or all of the external genitalia of a woman, typically the removal of the clitoris and labium. While it is euphemistically called "female circumcision" it has almost nothing in common with male circumcision. Sexual dysfunctional, while not ubiquitous, is increased in women who have been genitally mutilated, and sexual pleasure is generally decreased. Multiple health risks are associated with the practice, especially increased risk of complication and even death during childbirth. Female genital mutilation is not protective against HIV, and may even increase the risk of HIV infection, either during the mutilation procedure itself or due to additional tissue damage during sexual intercourse. Male circumcision should never be compared to female genital mutilation, a procedure that is more akin to penectomy.


Do parents have a right to circumcise an infant, or should they wait until he can make his own decision in adulthood?

The Declaration of the Rights of the Child upholds the right of children to have autonomy as individuals. This does not, however, preclude parents making decisions in the interest of the child, as a child cannot be considered to be truly autonomous. There are multiple examples that are widely accepted for parents making decisions for a child - such as in the area of education. The best comparison to infant circumcision is that of vaccination: both confer protection to infectious disease, both are irreversible and both result in a small chance of minor side-effects (such as swelling for a few hours to days). While this provides a basis for parental right to circumcision, it does not provide an unrestricted mandate - the least damaging form of intervention must be used (ie, non-surgical sterile circumcision over religious circumcision) and the benefits need to be placed in context to alternative (eg, if a vaccine for HIV is successfully generated the rationale for circumcision will be lost, just as the eradication of smallpox eliminates the rationale for the smallpox vaccine - a procedure with more complications than infant circumcision).

Another version of this objection, with somewhat more validity, is that since HIV is generally a sexually transmitted disease the protective effects do not kick in until the child reaches adulthood and has sex, at which time he can decide for himself. Well... perhaps, although it would be naive to assume that all men wait until they are 18 to have sex. Even if you were to wait until the age of 16, the surgical advice for adult circumcision is to have no sexual intercourse or masturbation for at least two weeks following the procedure. That may be quite a hard sell to a 16 year old boy, while being entirely irrelevant to an infant. Again, the best comparison is to vaccination. We have available an outstanding vaccine against human papilloma virus (HPV), which provides substantial (but not 100%) protection against cervical cancer in women who catch HPV. As HPV is a sexually transmitted disease you could advocate that this procedure should also be delayed until the age of 18, but with such mild side-effects as tenderness for a couple of days why not vaccinate all children as young as possible? Several Christian groups object to the HPV vaccine of girls on the basis that it will create "moral hazard" and promote promiscuous sex, but there is no actual evidence to suggest that girls are refraining from sex due to a fear of cervical cancer, and no evidence to suggest that the vaccine changes the rate of sexual activity.

References (11)

References allow you to track sources for this article, as well as articles that were written in response to this article.
  • Response
    Response: hiv infection
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: Hiv Symptoms
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: Hiv infection
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: common symptoms
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: United Cash Loans
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: Judge Ray Harding
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: pills
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: sneak a peek here
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Adrian Liston - With a Graph and a Map - Facts and Myths about HIV & Circumcision
  • Response
    Response: 5

Reader Comments (6)

Very interesting post. Quite a heavy case control study though, giving men a circumcision. If the same study woud be done in Western Europe I doubt that the researchers would find enought 'volunteers'. :)

July 4, 2011 | Unregistered CommenterSarah


If the same study woud be done in Western Europe I doubt that the researchers would find enought 'volunteers'.

That is certainly true. A 60% protection is worth a lot more at the individual level when rates of HIV as as high as they are in Africa. Also, the 60% protection figure is based on the HIV epidemic, where most transmission is sexual. In parts of Europe the contribution of intravenous drug use to HIV prevalence is much higher, and of course circumcision won't change the rik of getting HIV through a used syringe.

July 4, 2011 | Registered CommenterAdrian Liston

I have to admit that I would have thought that circumcision would cause more AIDS because you create a wound, a scar and a local inflammatory reaction. This reaction could leave more dendritic cells in the scar tissu. A quick pubmed scan learned me that dendritic cells rather flee away from the scar (J Cutan Pathol. 2008 Aug;35(8):752-6.). So again something learned today :)

July 4, 2011 | Unregistered CommenterS.

Interesting but I am curious why there isn't a cite page. Was this actually published in a respectful journal or is this just a blog. Can I find this anywhere else? These are some pretty big accusations and I can spot some flaws in the article and studies but if this is just someon es opinion blog than it is expected.

May 31, 2012 | Unregistered CommenterDanielle

You can use the hyperlinks to get directed to the 35 peer-reviewed scientific studies that I summerised here. When I write for scientific journals I use a reference list, when I write for the general public I use hyperlinks.

June 1, 2012 | Registered CommenterAdrian Liston

Interested to know your circumcision status and your views on neonatal circumcision.

July 6, 2014 | Unregistered CommenterDavid Cale

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>