Consistent condom use in anal sex stops 70% of HIV infections, study finds

Only one-in-six men reported 100% condom use during three to four years of follow-up
Dawn Smith of the US Centers for Disease Control (CDC) presenting at CROI 2013.
This article is more than 11 years old. Click here for more recent articles on this topic

An analysis by Dawn Smith of the US Centers for Disease Control (CDC) reported at the 20th Conference on Retroviruses and Opportunistic Infections (CROI 2013) on 4 March has provided the first estimate of the efficacy of condoms in preventing HIV transmission during anal sex since 1989. It found condoms stop seven out of ten anal transmissions – the same efficacy found by the 1989 study.

However, it also found that sometimes using condoms is not effective at preventing HIV infection, and that long-term 100% condom use is a minority behaviour: only one-in-six gay men actually managed to maintain it over the three- to four-year time frame of the analysis.

One ongoing problem in assessing the effectiveness of different HIV prevention methods is that anal sex is under-studied. We do not have enough data on rectal viral loads and their effect on transmission, or on whether HIV treatment reduces transmission via anal sex as well as it does for vaginal sex.



How well something works (in a research study). See also ‘effectiveness’.


A serodiscordant couple is one in which one partner has HIV and the other has not. Many people dislike this word as it implies disagreement or conflict. Alternative terms include mixed status, magnetic or serodifferent.


Receptive anal intercourse refers to the act of being penetrated during anal intercourse. The receptive partner is the ‘bottom’.


Insertive anal intercourse refers to the act of penetration during anal intercourse. The insertive partner is the ‘top’. 


The last part of the large intestine just above the anus.

We are also unclear about to what extent condoms actually prevent HIV transmission in anal sex. This last fact may seem surprising, given that condoms have been recommended since the mid-1980s as the only effective HIV prevention method for gay men who have anal sex.

In fact, there is only one large study in gay men, dating from 1989. In this study of 2914 gay men, HIV incidence among those who said they used condoms 100% of the time was 70% lower than in men who did not use them at all. There has been one small study in the era of antiretroviral treatment (ART), which found an efficacy of approximately 75%.

Amongst all men having anal sex, men who said they used condoms 100% of the time were 70% less likely to acquire HIV than men who never used condoms.

These are somewhat lower than efficacies computed for vaginal sex, which is in the order of 80 to 85%, and may reflect both that HIV is at least ten times more easily transmitted via anal than vaginal sex, and also that condoms may be more likely to fail during anal sex.

So a new study estimating condom efficacy in anal sex is very overdue and, given the need to compare condom efficacy against newer, biomedical prevention methods, very useful.

The studies analysed

Researchers from the US Centers for Disease Control retrospectively analysed condom use and HIV infection data from two different studies of US HIV-negative gay men: the VAX004 study, the first efficacy trial of an HIV vaccine, conducted between 1998 and 1999, and EXPLORE, one of the largest studies ever conducted of a behavioural intervention in HIV, conducted between 1999 and 2001.

There were some differences between the trial populations. Men in EXPLORE were twice as likely to be black and 50% more likely to have a college or university degree, and were younger, with fewer than half of the men in VAX004 aged 35 or under but nearly two-thirds in EXPLORE.

There were 1323 out of 3102 men in EXPLORE who said that they had had at least one episode of unprotected sex with an HIV-positive partner (43% of the men in the trial) and 2167 out of 4264 in VAX004, or 51% of all men in that trial. Altogether then, the CDC studied 3490 men who had had serodiscordant unprotected anal sex out of a total of 7366. There were 154 HIV infections in men in VAX004 and 71 in EXPLORE.

The total follow-up period in VAX004 was three years, and four years in EXPLORE. Participants were tested for HIV every six months and asked whether they had always used a condom for anal sex in the previous six months, sometimes used one, or never used one.

Condom efficacy with 100% use

Amongst all men having anal sex, men who said they used condoms 100% of the time were 70% less likely to acquire HIV than men who never used condoms, and 68% less likely than men who said they sometimes used them.

Condom efficacy was consistently higher in EXPLORE:  it was 86% for all anal sex, 87% for receptive anal sex and 76% for insertive anal sex compared with 59, 63 and 55% in VAX004.

Why the big difference? One possibility is that because EXPLORE was a behavioural intervention, it may have helped participants use condoms better and have fewer ‘accidents’ than in VAX004, which monitored condom use but did not intervene.

These figures are derived by comparing HIV incidence in men who said they always used condoms with men who never used them. What about the men who sometimes used them?

Condom efficacy with intermittent use

This analysis also shows that sometimes using condoms is no better than not using them at all. Overall, men who said they sometimes used condoms were only 4.4% less likely to acquire HIV than men who never used them. This difference was statistically insignificant; the margin of uncertainty means that, statistically, the 'true' efficacy of 'sometimes' versus 'never' using condoms could be anything between 29% fewer infections to 29% more infections, which is as good as saying that intermittent condom use essentially has zero efficacy.

There were 26% fewer infections in EXPLORE in men who used condoms intermittently than in men who never used them, but this was not statistically significant.

In VAX004 ‘sometimes’ using condoms was, if anything slightly less effective than never using them, with nearly 10% more infections in ‘sometimes’ versus ‘never’ users. How could this be? In the 1989 study, gay men who said they 'sometimes' used condoms were no less than 70% more likely to acquire HIV than men who said they never used them. The researchers at the time hypothesised that this because men who never used condoms might be more likely to be in monogamous relationships.

The researchers are now going to do further research to split ‘sometimes’ into different frequencies of use to find out below which level condom use ceases to be protective.

Consistency of condom use

How consistent was consistent condom use? Over the whole length of the studies, not very.  Two-thirds of men reported using condoms 100% of the time for at least one six-month slot during the two trials: but only 16.4% reported using them in every single six-month slot, i.e. truly consistently. Conversely, while only 5% of men reported never using condoms for the whole length of the studies, 40% reported never using them for at least one six-month period.

In the same session, Bob Grant, lead investigator of the iPrEx PrEP trial, showed an interesting slide showing selected ‘condom careers’ in individual trial participants and showed a whole variety of different use patterns over time, with the only consistent factor being no consistency.

There are a couple of important considerations to apply the CDC study’s conclusions. One is that social desirability bias almost certainly means that men's reported use of condoms was higher than it really was. This would mean the figures would tend to underestimate condom efficacy in the men who really did use condoms 100% of the time.

On the other hand, since only a minority of men in the studies did use condoms 100% of the time, these computed efficacy figures for condoms have to be divided by the fraction who actually did use them, resulting in lower effectiveness than this in actually preventing HIV on a population level.


Smith D et al. Condom efficacy by consistency of use among MSM: US. 20th Conference on Retroviruses and Opportunistic Infections, Atlanta, abstract 32, 2013. 

View abstract 32 on the conference website.

A webcast of the session in which this research was presented, HIV prevention: ARV, counseling, contraception, and condoms, is available on the conference website.