Risks and precautions


The three sexual acts referred to throughout this section are called Act 1, Act 2 and Act 3.

  1. Act 1 is when an HIV infected man is INSERTIVE in anal intercourse with an HIV uninfected man.
  2. Act 2 is when an HIV infected man is RECEPTIVE in anal intercourse with an HIV uninfected man.
  3. Act 3 is when an HIV infected man is insertive in ORAL intercourse with an HIV uninfected man.

For HIV to be transmitted during sex between men, a man with infectious HIV and a susceptible man without HIV need to perform a sexual act which allows the transfer of virus or viral particles from the infected to the uninfected man.

The number of new infections is the outcome of many different factors working together. Our current theory of the causes of HIV incidence includes the following.

The profile of infectivity among MSM with HIV

  • the number of men with HIV infection (both diagnosed and undiagnosed);
  • the profile of viral load in the body fluids (semen, anal mucus, blood) of the infected population, which is influenced by:
  • the proportion of infected men on successful viral suppressive treatment and
  • the proportion with an STI co-infection, which are both influenced by:
  • frequency of HIV and STI testing;

Sexual mixing and sexual behaviours

  • the frequency of sex between HIV infected men and HIV uninfected men, which is influenced by;
  • the rate of new sexual partner acquisition and;
  • the extent of concurrent regular sexual partnerships (over-lapping or open relationships).
  • the proportion of those sexual sessions that feature:
  • anal intercourse with the infected partner insertive (Act 1);
  • anal intercourse with the uninfected partner insertive (Act 2);
  • oral intercourse with the infected partner insertive (Act 3);
  • the proportion of those risk acts that feature condoms and the extent of condom failure when they are used;
  • the volume of body fluid transferred during risk acts and condom failure events, specifically semen during Act 1 and Act 3 and anal mucus during Act 2.

The profile of susceptibility among MSM without HIV

  • the proportion of uninfected men using nitrite inhalants during Act 1;
  • the proportion of uninfected men who have a foreskin during Act 2;
  • rectal trauma in the uninfected partner prior or during Act 1, and rectal trauma in the infected partner prior to or during Act 2.
  • the prevalence of other STIs in both partners, specifically:
  • the proportion of uninfected men who have a rectal STI during Act 1;
  • the proportion of uninfected men who have a genital STI (or a history of genital STI) during Act 2;
  • the proportion of uninfected men who have an pharyngeal STI during Act 3; all of which are influenced by:
  • frequency of STI testing;
  • the proportion of uninfected men who take PEP following exposure.

Not all of these factors are equally important to HIV incidence. The attributable risk each factor contributes to incidence is the proportion of infections that would not occur if the factor were not present. It need not be the case that factors with a higher relative risk also make a larger contribution to attributable risk. For example, a co-factor that had a high absolute risk but which is very uncommon in the population may contribute less to new infections than a factor with a smaller relative risk but which is very commonly done or present. There is very little data about the attributable risk associated with different factors contributing to the HIV epidemic. This is a major impediment to programme planning.

Strategic Research Goal: We will stimulate and/or carry out research that increases our understanding of the relative and especially attributable risks associated with each of the factors contributing to HIV incidence among men who have sex with men (MSM) in England.

Below we consider the evidence for the contribution of each of these factors, and the potential for them to be population level targets for HIV health promotion programmes.

Number of sexually active MSM with HIV

HIV comes from people with HIV, so the number of people living with HIV is related to the number of new infections occurring. To date the number of MSM living with HIV in England has only ever gone up. Since the only way for the number of men with HIV to go down is through them dying or leaving the country, we are not attempting to reduce the number of men living with HIV by removing HIV infected men from the population.

The number of men with diagnosed HIV has continually increased since the first diagnoses were made in the early 1980s.This graph shows how the number of MSM living in England with diagnosed HIV has increased each year, having more than doubled in the last ten years.

fig3.2msmhivAnti-retroviral (ARV) therapies have meant that men with HIV live healthier lives than previously. Healthier lives can include more sex, so it is likely that men with diagnosed HIV have sex with uninfected men on average more often now than was occurring before ARVs were introduced around 1995. This may mean more casual sero-discordant sexual sessions and more sexual sessions between men in sero-discordant relationships.

The majority of men who acquire HIV do so after having tested HIV negative at least once and the majority of men with undiagnosed HIV still believe themselves to be HIV uninfected (Williamson et al. 2008).

The average length of time spent undiagnosed influences the length of time men go without ARVs, which influences both their infectivity and their future health. All other things being equal, more men with HIV means more new HIV infections.

Infectiousness of MSM with HIV – the profile of viral load

The probability of transmission occurring during homosexual HIV exposure is a function of the amount of virus that is passed from the infected to the uninfected partner. The amount of virus transferred is a function of both the amount of body fluid passed and the concentration of virus in that fluid. If the amount of body fluid is held constant then the probability of transmission is a function of the concentration of virus (or of viral particles) in the body fluid being passed from the partner with HIV (Fisher et al. 2009). When HIV-infected men have anal intercourse or insertive oral intercourse with HIV-uninfected men, those with higher viral load will be more likely to pass on HIV if they do not use a condom or if the condom fails.

Viral load is a measure of the amount of virus in a body fluid, usually expressed in RNA copies per millilitre of fluid on a logarithmic scale. The specific body fluid involved in sexual HIV transmission varies by the type of sexual act. During Act 1 (anal intercourse with the infected partner insertive) and Act 3 (oral intercourse with the infected partner insertive) it is pre-ejaculatory fluid and semen. During Act 2 (anal intercourse with the uninfected partner insertive) it is anal mucus and rectal blood. The viral load in these four body fluids can vary within the same person at any point in time and across time. Viral load in each fluid varies with disease stage (how long someone has had the virus), anti-retroviral treatment and other infections.3.3increasedviralload

More viral particles (that is, a higher viral load) results in greater infectiousness. In HIV sero-discordant vaginal intercourse, viral load is the most important predictor of transmission. The graph illustrates the relationship between viral load and infectivity in vaginal intercourse between heterosexual couples of mixed HIV status with no treatment (Attia et al. 2009). Viral load in the HIV positive partners was measured on a ‘log-scale’ then grouped into six groups. The rate at which the positive partners passed their infections to their HIV uninfected partners is shown up the left hand side in infections per hundred person years. Each log increment in plasma viral load is associated with an increase in transmission rate ratio of 2.45 (95% confidence interval, 1.85 to 3.26, see Quinn et al. 2000). All other things being equal, higher viral load in the HIV infected population means more new HIV infections.

There is disagreement over the risk associated with a very low viral load (see Bernard 2008). A body fluid with very little virus in it may require an unfeasibly large amount of body fluid to be passed to cause infection. Men with very little virus in their body fluids may not be infectious. This has led some HIV specialists to state that people with HIV whose viral load is consistently undetectable (that is, below the level at which tests can detect it) are not sexually infectious.

This idea was first made public by a group of Swiss scientists and physicians in what is known as The Swiss Statement (Vernazza et al. 2008). However, there is disagreement about whether people with HIV in this position cannot pass on the virus or that it will be a very uncommon event.

A systematic review of studies looking at the relationship between viral suppression in plasma and semen (Kalichman et al. 2008) found that (in the absence of sexually transmitted infections), plasma and seminal viral loads are moderately well correlated in men receiving antiretroviral treatment. For individuals, anti-retroviral therapy has the potential to reduce the risk of transmitting HIV, although it may not eliminate that risk.

Isolated episodes of infectious HIV RNA have been measured in the semen of HIV positive men with undetectable viral load, sometimes at high levels, indicating that an undetectable viral load in plasma does not always mean undetectable viral load in semen (Sheth et al. 2009).

The Swiss Statement only applies to people with HIV on treatment. It does not say that people with HIV who are not on treatment and have an undetectable viral load are also unable to pass their infection. However, it is clear that among people with HIV not on treatment, higher viral load is associated with higher probability of onward transmission if exposure occurs.

Modelling of the effects of the Swiss Statement suggests that substituting treatment for condom use altogether carries a risk of increasing HIV incidence if even a small proportion of individuals do not have fully suppressed viral load, and if transmission can occur between men even when viral load levels are very low. In a scenario in which people with fully suppressed viral load did not use condoms during 100 episodes of vaginal or anal intercourse each year with a steady partner, in which 85% of people had fully suppressed viral load at any time and condoms were used in 80% of sex acts, HIV incidence would increase fourfold in HIV-discordant male couples over ten years (Wilson et al. 2008). For this reason we think there is good reason to not transfer the Swiss Statement about vaginal intercourse to the case of anal intercourse.

Disease stage of untreated HIV infection


The length of time men spend with undiagnosed HIV is related to how frequently men at risk for HIV test for it. Increasing the rate at which men at risk test is a desirable goal for the population of MSM in order to minimise new HIV infections.

Whether men choose to take anti-retroviral therapy is also a choice that is only possible if men with HIV have had their infection diagnosed.

The natural history of HIV infection results in viral load being very high in the first few months following infection, dropping to a much lower level for a ‘latent’ (or asymptomatic) period and increasing with the onset of illness and later stage HIV disease. If HIV transmission probability is a function of viral load then people will be more infectious in the earlier and later stages of infection.

An early review (in 1994, so before anti-retroviral combination therapy) suggested that men in the initial stage of HIV infection were between 100 and 1000 times more infectious than the following asymptomatic phase, and then between 10 and 100 times more infectious once AIDS develops (Jacquez et al. 1994). More recent studies have suggested the difference is large but not this large.

A recent review of HIV sero-conversion during serodiscordant vaginal intercourse estimated the per-contact risk to be 9·2 (95% CI 4·5-18·8) and 7·3 (95% CI 4·5-11·9) times larger when the infected partner is in the early and late phases of HIV infection respectively, than during the asymptomatic phase (Boily et al. 2009).

The first six months following infection (known as recent, primary or acute infection) has therefore been identified as the period during which onward transmission is most likely should exposure occur.

Among MSM in the UK, men with recent HIV infection have been shown to be 3 to 4 times more likely to have recently passed on their infections than men with untreated chronic infection (Fisher et al. 2009).

Not all MSM with HIV are diagnosed. About a quarter of MSM with HIV are yet to be diagnosed and are not in touch with clinical services. In order to benefit from ARVs (see next section) men with HIV must have had their infection diagnosed. HIV testing needs in the population of MSM not diagnosed with HIV are therefore central both to the profile of viral load among men with HIV and consequently their health and their HIV infectivity.

Anti-retroviral treatment for HIV (ARV)


The proportion of men with detectable viral load influences the rate of HIV transmission. Fully suppressive anti-retroviral therapy will reduce the risk of HIV transmission. To minimise new HIV infections, increasing the proportion of men eligible for treatment who receive anti-retroviral treatment that is fully suppressive is a desirable goal for the population of HIV-infected MSM. Advances in drug treatments have reduced the average infectiousness of men with diagnosed HIV. Antiretroviral treatment works by suppressing the ability of HIV to replicate, thereby forcing down viral load.

Driving viral load down below the level at which tests are able to detect it is desirable for the health of people with HIV because it means people have a very low risk of developing AIDS and also a low risk of developing resistance to ARVs.

There is ecological evidence from two populations with somewhat similar epidemiological profiles to England - San Francisco and British Columbia - demonstrating that an increase in the proportion of individuals eligible for treatment who take anti-retroviral drugs is strongly correlated with a decline in new HIV diagnoses over time. Both settings have a higher frequency of HIV testing than England, and both studies use new HIV diagnoses as a surrogate for HIV incidence.

In the case of San Francisco the profile of viral load in the community as a whole was correlated with the decline in new HIV diagnoses. All available viral load measurements of HIV-diagnosed persons receiving care in the city were averaged over each of the years 2004-2008, and correlated with new HIV diagnoses in the period. The analysis showed that mean and total community viral load were strongly correlated with new diagnoses, with an increase in the uptake of antiretroviral treatment among eligible persons and with an increase in the proportion of patients with fully suppressed viral load (Das et al. 2010).

In British Columbia an increase in the proportion of medically eligible persons receiving anti-retroviral treatment correlates strongly with significant declines in new HIV diagnoses in the periods 1996-99 and 2004-2008. Conversely a stabilisation in numbers on treatment due to later treatment initiation and treatment interruptions in the period 2000-2003 was strongly correlated with a stable rate of new HIV diagnoses in the province during the same period. These trends occurred despite increases in sexually transmitted infections after 1998, an indicator of increases in unprotected sex, and despite an increase in the total numbers tested. The trends in each period were also correlated with declines in the median population viral load (Montaner et al 2010).

The most relevant studies in this area have been conducted with heterosexuals. While the reduction in infectiousness during anal sex cannot be quantified, it is likely to be substantial. A study with 3,381 HIV discordant heterosexual couples in several African countries calculated that treatment reduced the transmission risk by 92%. There were 103 HIV transmissions, but 102 of these were from a partner not taking HIV treatment (Donnell et al. 2010). A 92% reduction in risk is greater than the reduction in risk given by universal consistent condom use.

Of course, not all MSM diagnosed with HIV are on ARVs and with undetectable virus. Across the UK in 2010, 82% of MSM seen for HIV care were using ARVs (Health Protection Agency 2011). This suggests up to a fifth of MSM with diagnosed HIV are not on ARVs, nor are any of the men with undiagnosed HIV. Among 758 MSM with diagnosed HIV using an East London clinic in 2004/5, 71% were on ARV and 61% had an undetectable viral load (Elford et al. 2007).

Other sexually transmitted infections (STI)


The length of time HIV infected men spend with heightened infectivity due to STIs is related to how frequently they acquire STIs and how quickly they have them diagnosed. To minimise new HIV infections, increasing STI screening among more sexually active men is a desirable goal for the population of HIV infected MSM.

Even if plasma viral load is undetectable, there can be a spike in seminal or anal mucus viral load if another STI is acquired (Sadiq et al. 2005). So men with undetectable (plasma) viral load may pass on HIV by acquiring another STI and are then being involved in sexual exposure. Among MSM with HIV in Brighton, those diagnosed with another STI were 6 times more likely to have passed on their infection than those without an STI diagnosis (Fisher et al. 2009). Men with diagnosed HIV are more likely to acquire other STIs than are men without diagnosed HIV (Dougan et al. 2007). As elsewhere in the world, in England there have been outbreaks of syphilis, lymphogranuloma venereum (LGV) and sexually transmitted hepatitis C, concentrated in HIV infected MSM.

Most men with diagnosed HIV have more sexual partners than sexual health screens (Ferrand et al. 2008). Increasing the rate of STI screening among MSM with diagnosed HIV has great potential for reducing harm. At the population level further sexually transmitted infections among men with HIV means more new HIV infections.

The profile of plasma viral load in the HIV infected MSM population3.3.4hivbrighton

Although men with recent infection may be much more infectious and much more likely to pass on their infection, at any point in time they are in the minority of men with HIV.

Among men with diagnosed HIV attending a Brighton clinic (Fisher et al. 2009), almost half of all men’s time is spent with a detectable viral load (that is, above 50 copies). This suggests that the population of men with diagnosed HIV have the infectious potential to pass on their virus.

The group of men with currently undiagnosed HIV will almost certainly have a very different profile of viral load, with men weighted toward the higher end of viral load.

Frequency of sex between HIV infected and uninfected men


The number of sexual partners men have between each STI screen influences the rate at which STIs and HIV are passed on. More partners per screening results in more infections.

To minimise new HIV infections, reducing the number of partners between screens is a desirable goal for the population of MSM.

Sex between infected and uninfected men are the events within which all MSM sexual HIV transmissions occur. They include sex between uninfected men and both diagnosed and undiagnosed infected men, both between regular and casual sex partners.

In addition to the prevalence of HIV in the population, the overall parameters in the population influencing the number of HIV sero-discordant sexual sessions are: the rate of new partner acquisition; the pattern of mixing between infected and uninfected men; the proportion of men in HIV sero-discordant regular sexual relationships (not necessarily recognised as discordant by the men in the relationship); and the frequency of sex between men in such relationships.

In the HIV uninfected population of MSM as a whole, HIV acquisition is more common in men with higher numbers of sexual partners (Fox et al. 2009). This is both because the probability of engaging in sex with a man with HIV increases with increasing numbers of partners, and because the probability of engaging in acts that can cause HIV transmission also increases. Multiple sexual partners on their own may not be associated with HIV acquisition but multiple anal intercourse partners, especially multiple unprotected anal intercourse partners are. However, men with higher numbers of sexual partners are also more likely to acquire and pass on other STIs which increase susceptibility to infection (in uninfected men) and infectivity (in infected men). Therefore men with multiple ‘safer’ sexual partners can increase the risk other men pass and acquire HIV by picking up and passing on other STIs.

The UK National Survey of National Sexual Attitudes and Lifestyles (Mercer et al. 2005) found that the number of men having sex with other men increased between 1990 and 2000. However, there was not a significant increase in the rate of new partner acquisition (that is, the rate of partner change).

Knowledge and disclosure of HIV statuses


Telling sexual partners about one’s HIV status can influence the sexual choices that are made. Asking a sexual partner about their HIV status can increase knowledge of HIV sero-concordancy but may also result in the wrong perceptions especially for men who think they are HIV uninfected.

Increasing sharing of HIV status between sexual partners is a desirable goal for the MSM population to minimise new HIV infections.

In order to enjoy sex with minimum risk of onward HIV transmission, some men with diagnosed HIV purposefully choose sexual partners who are also living with HIV, especially if their sexual preference is for unprotected anal intercourse (Bolding et al. 2005).

Sexual partners who know or believe themselves to have the same HIV status often choose to engage in sexual acts that they would not engage in if they did not know they had the same status. In order to acquire this knowledge both partners must be tested for HIV and to share their HIV test results with each other. Sharing knowledge about one’s HIV status is therefore a behaviour directly related to sexual risks and precautions and in order to do this men must have tested for HIV and if HIV negative to have done so since their last HIV risk.

Sexual acts able to transmit HIV

We are concerned with three sexual HIV exposure behaviours. Each involves an HIV infected and an uninfected man. We recognise infected men being insertive in UAI with uninfected men as a different risk act (transferring a different body fluid) from uninfected men being insertive in UAI with infected men. There is no current data to tell us what proportion of sexual sessions feature each of these risk acts. Nor do we have current data about the frequency with which the risk acts occur and how they are spread through the population. Most data consists of the proportion of men who had engaged in a particular behaviour over a specified time period.

In addition to the number of men with HIV and the frequency with which sex occurs with men without HIV (see above), the frequency of exposures is a result of the frequency of anal and oral intercourse, the frequency with which condoms are used for these acts, and the frequency with which those condoms fail.


When men have sex together, the type of sexual acts they engage in vastly influence the likelihood STIs and HIV are passed on if present in either or both partners.

Both receptive anal intercourse and insertive anal intercourse carry risk and the proportion of sessions that include anal intercourse influences HIV incidence. To minimise new HIV infections, it is desirable that the overall proportion of sexual sessions that feature anal intercourse be lower.

Risk Act 1: infected partner is insertive in anal intercourse with uninfected partner

Receptive unprotected anal intercourse (RUAI) continues to be the most common route for HIV acquisition during sex between men in England (and conversely insertive UAI (IUAI) is the most common way in which it is passed on) (Macdonald et al. 2009).This is because it is a fairly common act and has a high per-contact probability of transmission relative to other acts.

Both IUAI in HIV infected men and RUAI in uninfected men are common behaviours. Almost half of all men with diagnosed HIV engaged in some IUAI in the last year (Hickson et al 2006). Not all of these will have done so with uninfected men but many did do so: 18% of men with diagnosed HIV said they had engaged in IUAI with a known HIV uninfected partner and 29% had done so with a partner of unknown status. Many of the 37% who indicated they had IUAI with other men with HIV based their judgements of their partners’ statuses on insufficient information and wishful thinking (Bourne et al. 2009).

Conversely, 33% of never tested men and 38% of men last tested negative had RUAI in the last year. Again not all of these men will have done so with HIV infected partners but some did, and many had opportunity to do so: less than 1% of untested men and less than 2% of tested negative men had RUAI with men they knew to have HIV, while 19% of both had done so with a partner of unknown status. In addition, because of undiagnosed infection, many of the men who thought they had RUAI with HIV uninfected partners were in fact being exposed to HIV.

At the level of the population, the proportion of men that engage in Act 1 increases with increasing numbers of sexual partners. The higher the average number of sexual partners in the population, the larger the proportion of the population will be involved in exposure.

The per-contact probability of transmission is very strongly associated with the co-factors described below. These co-factors create differences in transmission probability that are so large that it is unhelpful to attempt to give an estimate for the per-contact probability for Act 1 per se. It depends on the seminal viral load of the infected partner, whether he ejaculates into the rectum of the uninfected partner and whether the uninfected partner is using poppers (see below).

The risk of pos-insertive sdUAI (Act 1) in the absence of ejaculation may be similar to that for pos receptive sdUAI (Act 2), the difference arising because ejaculation of HIV infected semen into the body is possible in Act 1 but not Act 2.

Risk Act 2: infected partner is receptive in anal intercourse with uninfected partner

It is possible for men to acquire HIV through the penis during insertive anal intercourse with an infected man. Act 2 is common among MSM - 18% of men with diagnosed HIV indicated having had RUAI with known uninfected partners in the last year and 36% had done so with men of unknown status. Correspondingly, less than 1% of never tested and 3% of last test negative had IUAI with men known to have HIV, while 20% and 23% respectively had done so with men of unknown status. Some men are willing to take a risk with Act 2 but not Act 1 knowing it is less likely to result in transmission. However, the overall contribution of Act 2 to the HIV epidemic among MSM in the UK is sufficiently large for it to be picked up as an independent risk factor in a relatively small case control study (Macdonald et al. 2007).

Risk Act 3: infected partner is insertive in oral intercourse with uninfected partner

It is possible for HIV transmission to occur during Act 3 in the absence of ejaculation. HIV has been found in pre-ejaculatory fluids (Ilaria et al. 1994) and case studies have reported HIV acquisition during receptive fellatio without ejaculation (Lane et al. 1991 & Edwards et al. 1995). However, this appears to be a very rare event. Ejaculation in the mouth increases transmission probability considerably (see below).

A Spanish study followed 96 HIV uninfected women who were in monogamous sexual relationships with HIV positive male partners, and who always used condoms for vaginal or anal intercourse and had no condom failure (del Romero et al. 2002).

Between them the women were estimated to have performed 8965 acts of fellatio of which 34% (3048 acts) included ejaculation into the mouth. None of the women acquired HIV. It is worth noting however that few if any of the women’s partners were in the early (highly infectious) stage of HIV infection.

Fellatio is an almost universal behaviour among MSM and far more uninfected men are receptive in unprotected oral intercourse than unprotected anal intercourse. Condom use for fellatio has not been measured for some time in the UK MSM population and is assumed to be rare.

Condom use and condom failure


When men have anal intercourse, whether or not they use a condom influences the likelihood HIV and other STIs will be transmitted if present.

To minimise new HIV infections, it is desirable that there is an increase in the proportion of anal intercourse events that feature condoms.

A condom used during anal and oral intercourse is an effective but not infallible way to block body fluid transfer during all three sexual risk acts. Because condoms block body fluid transfer but may not wholly cover an infected skin area, during anal intercourse between infected and uninfected partners condoms reduce the transmission risk of STIs that are transmitted by genital fluids (eg. gonorrhoea, chlamydia and HIV) more than of STIs that are primarily transmitted by skin-to-skin contact (eg. herpes, human papilloma virus / warts virus and syphilis).

If they remain intact and on the penis throughout intercourse condoms provide 100% protection against HIV. However, the extent to which condoms reduce HIV transmission probability is dependent on the extent to which they fail (break or come off) during intercourse.

Under ideal conditions in vaginal intercourse, condoms can have a failure rate of 2%. This would reduce the risk of transmission to 1/50th of the act without a condom and is the best performance estimate. A meta-analysis of comparisons between HIV sero-discordant couples who always used condoms with those who never used them found consistent condom using couples experienced an HIV incidence of 13% that of couples never using them. In other words, consistent condom use was 87% effective at preventing HIV infections during sero-discordant vaginal intercourse (Davis et al. 1999).

At the event level, under domestic conditions among gay couples who are sexually familiar with each other, 6.5% occasions of condom use for anal intercourse result in the condom breaking or slipping off (Golombok et al. 2001). The rate of condom failure may be higher in protected intercourse between new partners in non-domestic environments. It may also be higher between sero-discordant partners than between HIV uninfected partners because men with diagnosed HIV are more likely to have experienced condom failure than men without HIV. Inconsistent condom users (who are more likely to be HIV infected) are also more likely to experience condom failure than those who are consistent condom users.

At the population level there is a larger increase in HIV risk between no AI and AI with a condom than there is between AI with a condom and AI without a condom. A single city USA study (Seattle, 2001-2007) looked at the proportion of MSM clinic attenders who tested HIV negative and then tested HIV positive within a year, by their sexual behaviour in the 12 months preceding their positive diagnosis (Golden et al. 2008). The data therefore come from men repeat testing who are likely to be very sexually active.

The unit is the year preceding the visit (some men contributed more than one year to the data). Overall, 3.1% (89/2912) of tests following a year featuring UAI were positive, 1.5% (28/1827) of tests following a year featuring protected but not unprotected AI were positive, and zero tests following 410 years with no anal intercourse were positive. This suggests that a larger proportion of the risk associated with protected AI is eliminated by avoiding AI, than is the proportion of risk associated with UAI eliminated by using a condom. Avoiding AI is a better way of reducing HIV risk than is using condoms if only HIV risk (and not the quality of men’s sex lives) is taken into account.

Change over time in unprotected anal intercourse (combined Acts 1 and 2 without condoms)

It is well documented that sexual risk acts betweef3.7.1confuain men became increasingly frequent in gay communities across the globe from 1996 onwards. This trend may have increased in the UK until the early 2000s before reversing more slowly.

This graph shows the odds ratio for engagement in UAI with a partner of unknown HIV status or known to be discordant, for the years 1997 to 2006 compared with 1996 (Wilkinson et al. 2007; Sexual Health Survey of Gay Men: London 2006. London, UCL). The odds of engaging in this risk behaviour rose significantly higher in 1998 than 1996 (the black bar is all above 1) and has stayed significantly higher since. The overall trend across the decade appears to be a rise followed by a slower fall.

Body fluid transfer

HIV is passed by the body fluid of an infected partner entering the body of an uninfected partner. The amount of body fluid along with the concentration of virus in the fluid affect transmission. Body fluid transfer varies by sexual risk act.

Pre-cum and semen in Act 1 and Act 3


When men have unprotected intercourse or fellatio (and when condoms fail during use), whether or not they ejaculate into each other’s bodies influences the likelihood some STIs, including HIV, are transmitted if present.

To minimise new HIV infections, it is desirable that the proportion of intercourse events that result in ejaculation into the body be reduced.

Because ejaculation increases the amount of virus carrying fluid transferred into the body of the uninfected partner, ejaculation into the body makes HIV transmission more likely than without it. In the UK men who engaged in RUAI to ejaculation with men not known to be HIV uninfected were 2.5 times more likely to acquire HIV than men who do this but not to ejaculation (Macdonald et al. 2007). The unit here is individual men and some (if not many) of the sexual partners men had RUAI with will not in fact have had HIV. The effect should be stronger considering only men known to have had discordant RUAI with and without ejaculation.

Accordingly, a study in Sydney found that HIV uninfected MSM who engaged in RUAI to ejaculation with known HIV infected partners had an HIV incidence 6 times that of men who had RUAI with known infected partners but who avoided ejaculation (Jin et al. 2009).

Again the unit here is individual men. The size of the effect of ejaculation on the probability of a single event of Act 1 (between infected and uninfected partners) may be higher again (because some of the uninfected men who engaged in sdRUAI to ejaculation may also have done so not to ejaculation, diluting the strength of the comparison group).

Ejaculation into the mouth during Act 3 appears to greatly increase transmission risk, with transmission being very rare in its absence. Ejaculation into the mouth is also associated with a higher risk of throat infection with other STIs in MSM50 which may then render men more susceptible to oral HIV infection.

Rectal mucus and blood in Act 2

Remarkably little is known about anal mucus and what determines the amount of anal mucus that may be transferred to the penis during Act 2. Increased production of mucus may be a response to irritation or stimulation through anal intercourse itself.

A study in Washington that measured viral load in different body fluids from the same men found that the viral load in rectal secretions was much higher than that in blood which in turn was higher than that in semen (Zuckerman et al. 2004). The differences between viral load in rectal secretions and in semen were much larger among men not on HIV treatments.

The use of sex toys and ano-brachial insertion (fisting) can cause trauma resulting in bleeding, thereby increasing the amount of infectious body fluid present, which if in the infected partner prior to Act 2 could increase transmission risk. There may also be blood present in faeces.

Susceptibility of men without HIV

A variety of factors related to the susceptibility of the uninfected partner can influence the probability of transmission when sexual exposure takes place. It is possible some men are naturally immune to HIV but there is no way of knowing who, so this is of no practical value for precautionary choices. If this were the case it might be thought to apply to all three exposure routes. Other factors are specific to the exposure routes.

Act 1 risk facilitator: anal trauma

The cells of the rectum are vulnerable to HIV and trauma is not necessary for infection to occur. However, ano-brachial insertion and the use of sex toys can cause colonic perforation and other damage (Cohen et al. 2004). Anal trauma prior to anal exposure to HIV may also increase the probability of HIV transmission occurring, although there is little empirical evidence to demonstrate this. As with anal mucus, the properties of the rectum and its vulnerability to HIV have not been substantially studied.

Act 1 risk facilitator: poppers


When HIV uninfected men are receptive in unprotected anal intercourse with HIV infected men, whether they are using poppers influences the likelihood HIV is acquired.

It is desirable that the proportion of such exposures that feature poppers be reduced.

Poppers is the street name for various kinds of liquid alkyl nitrite which are sold in small bottles and inhaled as a recreational drug. They effect a drop in blood pressure, a rise in heart rate, a head rush and a relaxation of smooth muscle (French & Power 1997). Of the three types of nitrites in circulation, one (amyl) is controlled by the medicines act, another (butyl) is banned due to its carcinogenic properties and the third (iso-butyl) is uncontrolled.

The use of poppers by the uninfected partner during Act 1 has been demonstrated to increase by approximately 3-fold the probability of HIV transmission occurring (Macdonald et al. 2007, Buchbinder et al. 2005). Which of the three types of poppers were involved is unknown. The mechanism by which transmission probability is increased during poppers use is also unknown but may be a consequence of vasodilation or longer, rougher intercourse.

Poppers are the most commonly used drug after alcohol among MSM the England (Hickson et al. 2010) and are particularly popular during receptive anal intercourse. Poppers are also frequently used for being fisted and using larger sex toys, multiplying any risk associated with anal trauma. The ratio of the three types of poppers in current use is unknown.

Act 2 risk facilitator: foreskins

As well as the urethra being susceptible to HIV infection, the cells of the foreskin are also vulnerable. Therefore a foreskin increases transmission probability specifically during Act 2 (HIV uninfected man insertive in anal intercourse with HIV infected man) and circumcision reduces that probability. Foreskins are also associated with other STIs which are themselves facilitative factors for HIV transmission.

Experimental trials have found that HIV uninfected men with foreskins who are randomly selected for circumcision are less likely to acquire HIV than are controls (with similar sexual behaviour) who are not circumcised. These experiments show that the foreskin has a casual role in HIV transmission to the penis. This has been demonstrated among predominantly heterosexual men in Africa and may be generalisable to MSM in Britain. A meta-analysis of observational studies of sero-conversion in heterosexuals suggests a foreskin increases the per contact probability of HIV transmission during unprotected vaginal intercourse with an HIV infected female partner by a factor of at least 2 (Boily et al. 2009).

However, large samples of MSM show no difference in HIV prevalence between men with a foreskin and those who have had it removed (Millett et al. 2008). This suggests any attributable risk of foreskins in the MSM epidemic is small. No study has yet demonstrated a significant difference in HIV acquisition between MSM with and without a foreskin.

Act 3 risk facilitator: oral trauma

Oral trauma and ulcerative mouth conditions increase transmission probability during Act 3 (Scully & Porter 2000). As risk factors multiply, ejaculation by an infected man and oral trauma in the uninfected man will provide a higher probability of HIV transmission than either facilitator alone. In the absence of both these facilitators the risk of transmission during Act 3 appears to be very low (Page-Shafer et al. 2002; Baggaley et al. 2008).

Act 1, Act 2 & Act 3 risk facilitator: other STIs


The number of STI screens that occur before new sex partners have sex influences the rate at which STIs and HIV are passed on. More screens per new partnership results in fewer infections.

Increasing the number of screens that occur before new partners have sex is a desirable goal for the population of MSM.

As well as other STIs making HIV infected men more infectious, STIs can make HIV uninfected men more susceptible to HIV. A concurrent ulcerative STI in the uninfected partner increases HIV transmission probability during Act 1 if the infection is rectal, during Act 2 if it is genital and during Act 3 if it is oral.

Among MSM clinic attenders, men diagnosed with an STI are more likely to also have undiagnosed HIV than men not diagnosed with an STI (Health Protection Agency 203.9.5gonnor09).

Gonorrhoea in particular makes a detectable independent contribution to HIV acquisition among MSM in England - MSM clinic attenders diagnosed with gonorrhoea are four times more likely to be diagnosed with HIV than MSM not diagnosed with gonorrhoea (Macdonald et al.  2007). Gonorrhoea diagnoses rose in MSM until 2006 but have since dropped (see graph), although MSM continue to be the gender / sexuality group with the highest incidence of gonorrhoea in the UK (Health Protection Agency 2010). It is therefore probably the single most significant other infection for HIV prevention programmes.

Most HIV uninfected MSM have more sexual partners than sexual health screens but the average number of sexual partners between STI screens is poorly documented. Increasing the rate of STI and HIV screening among MSM not diagnosed with HIV has great potential for reducing harm.

Act 1, Act 2 & Act 3 risk reducer: PEP


When HIV uninfected men are sexually exposed to HIV, whether or not they swiftly seek and use PEP influences the likelihood they acquire HIV infection.

It is desirable that the proportion of men in this position using PEP increases.

Taking anti-HIV drugs as soon as possible following sex with a risk of HIV exposure can reduce the risk of HIV acquisition. The drugs then need to be taken for a month (Benn et al. 2011). This framework supports the recently updated UK Guideline for the use of post-exposure prophylaxis (PEP) for HIV following sexual exposure, by the British Association for Sexual Health & HIV (BASSH). This new Guidleine on PEP is summarised and contextualised in a recent Making it Count Briefing Sheet on PEP.

Only a proportion of men sexually exposed to HIV will acquire it. It is not clear by how much the risk of acquisition would be reduced if all the men exposed took PEP within 72 hours of HIV exposure (and then followed the course of drugs correctly). However, it may approach 100%. The reduction in risk will get smaller as the time between exposure and onset of PEP increases and will eventually reach the same risk of acquisition as if PEP had not been taken.

Only a small proportion of men sexually exposed to HIV access PEP and those who do so also have risk occasions when they do not access PEP (Sayer et al. 2009). For those who do access PEP it can mean the difference between becoming HIV infected or not on that exposure occasion. This makes it an important personal health service. However, PEP has very limited utility either as a long-term risk reduction tactic for an individual or as a public health intervention covering a significant proportion of the population at risk.

Absolute risks of transmission

The absolute risk of HIV transmission under any given set of circumstances is how likely the event is to occur. As noted above, the absolute risk during any specific act depends on both the viral load of the infected partner and a number of other specific co-factors. It is therefore simply not possible to say, for example, how risky fellatio on an HIV infected partner is, because it is not the sexual act itself that is most important in determining the risk.

However, a few studies (using a variety of methods) have provided estimates for the per contact transmission probabilities of sexual acts between infected and uninfected men averaged across all the co-factors prevalent in the population providing data at the time of the studies.

An early study of gay couples in Boston (DeGruttola et al. 1989) estimated per contact risk of pos-insertive sdUAI to ejaculation (no PEP, averaged across disease stage, other STIs and poppers use were not measured) to be in the range 0.5% - 3.0% (that is 1-in-200 to 1-in-33). Similarly, a prospective US cohort study conducted between 1992 and 1994 (Vittinghoff et al. 1999) estimated the per contact risk of pos-insertive sdUAI (ejaculation not specified, no PEP, averaged across disease stage, other STIs and poppers use which were unmeasured) to be in the range 0.24% - 2.76% (or 1-in-417 to 1-in-36, 95% confidence interval), a range similar in magnitude to the Boston study. More recently, a cohort study among gay men in Sydney (Jin et al. 2010) estimated the per act risk of pos-insertive sdUAI to ejaculation to be in the range 0.48% - 2.85% (95% CI), again very similar to the previous two studies.

The per contact risk of transmission applies to only one event. What appears a very small risk can become important because many people do it. A large number of people taking a small risk can result in more negative outcomes than a small number of people taking larger risks. In addition, men’s sex lives consist of a large number of events over a long period of time. What appears to be a very small risk (from a single event) can add up to a large risk if that act is repeatedly engaged in. People are generally very poor at appreciating the summative effect of numerous small risks over a period of time.

The incidence of infection over a year (the proportion of men who sero-convert) from a single route (eg. IAUI only) is a function of both the number of times the act occurs and the probability of transmission when the act occurs.

Population profiles, individual choices and sexual lifestyles

In order to get the number of HIV infections currently observed in England, a much larger number of exposures is required, which themselves occur in an even larger number of risk acts. The risk acts vary both in terms of what men know about their sero-statuses (whether they are concordant or not and whether or not the infected partner is more infectious than the uninfected partner is susceptible) and also in terms of what acts they engage in (pos-IAI, pos-RAI, pos-IOI), with what transmission facilitators (ejaculation, poppers) and transmission obstacles (condoms, PEP).

It is the volume of risk overall that determines HIV incidence. How this volume is distributed is not documented. For example, how many times men are sexually exposed to HIV before they sero-convert (an indicator of how concentrated the volume of risk is in the same group of men) is unknown.

Men who have sex with men are as diverse as the general population. The contexts in which HIV transmission occurs during sex between men are as varied as the contexts in which sex occurs. In global terms, England is a relatively easy country for MSM to live in. In eighty countries in the world homosexuality is illegal and in five it is punishable by death (Ottosson et al. 2010). In many others men known or thought to have sex with men are subject to a higher level of day-to-day discrimination than in England. Large cities with extensive gay scene, such as London and Manchester, therefore attract MSM from across the globe.

Characteristics of men passing on HIV

Men who have themselves been recently infected are thought to be disproportionately likely to pass on HIV (National AIDS Trust 2008). However, since these men are in the minority of men with HIV, their proportionate contribution to overall incidence is unclear.

Sexual risk is common in men with diagnosed HIV. In a sample of MSM with HIV attending a London clinic, 20.2% had UAI with an unknown or known discordant partner in the past 3 months (Elford et al. 2007) and as noted earlier many men with diagnosed HIV spend some time with detectable viral load.

One reason often given for wanting to reduce the length of time spent with undiagnosed HIV is that men are less likely to sexually expose their infection once the infection has been diagnosed. There is a measurable downward impact of HIV diagnosis on sexual risk behaviours among men with undiagnosed HIV. Although this reduction may not be sustained, if it occurs early in infection and at a time of high HIV infectivity it will be more advantageous to incidence than later diagnosis.

Less reduction in risk at diagnosis has been associated with having more sexual partners before diagnosis and a concurrent STI at diagnosis (both indicating an overall riskier sexual behaviour pattern), as well as drug use, ketamine in particular (Fox et al. 2009).

However, men with diagnosed HIV are more likely to be involved in UAI with partners of unknown status than are men with undiagnosed infection (Williamson et al. 2008). Many men with diagnosed HIV use sex on premises venues such as saunas and backrooms (and are more likely to do so than other men (Hickson et al. 2009) and unprotected intercourse in these settings is not uncommon (Bourne et al. 2009).

Characteristics of men picking up HIV

Behaviourally and causally, picking up HIV is associated with larger numbers of intercourse partners, less use of condoms, and accepting more ejaculation into the body. In today’s MSM cultures, these behaviours are in turn associated with using saunas and backrooms, a larger number of sexual partners overall, a greater use of ‘party’ drugs and (for a minority of men) sex work (Fox et al. 2009).

The factors associated with acquiring HIV are the same as those associated with passing it on, both before and after diagnosis. Behaviourally and in terms of sexual lifestyle, men who are most likely to pick up HIV and men most likely to pass on HIV are not distinguishable. They are the same men at different points in time.

Demographically, approximately half of the men who pick up HIV in England are under 30 years of age (based on the average age of diagnosis and the estimated average length of time spent undiagnosed). Black MSM (African and Caribbean) appear twice as likely to pick up HIV as other ethnic groups. Men with lower education are more likely to pick up HIV than men with higher education.

Characteristics of the relationship between men passing and picking up HIV

HIV is being passed in England during both casual sex between men and during sex between regular partners (Elam et al. 2008). Men with diagnosed HIV pass their infection both to men who know they have HIV and to men who do not know. Of men acquiring HIV, 28% had RUAI with a man they knew to have HIV around the time of their infection (Macdonald et al. 2007). However, it is difficult to be precise about the relative contribution of each of these contexts. Many men who acquire HIV had multiple opportunities to pick it up.

The distribution of risk acts between men who do and do not acquire HIV (for example, the profile of the number of exposure events MSM experience before HIV acquisition occurs) is not documented. It is clear however, that it is not only men acquiring HIV who are involved in exposure. Given the low probability of transmission occurring during a single exposure, there are likely very many exposures occurring to generate the current number of infections. It may be only under certain circumstances does exposure result in transmission, and identifying and altering those circumstances may be a more effective route to reducing incidence than attempting to eliminate all exposures.

The incidence of HIV among MSM is a property of the population and is the outcome of thousands of men making millions of choices in interaction with each other. These choices cover:

Patterns of use of clinical services: How frequently men screen for STIs and HIV, their use of treatments if diagnosed with an infection, and their use of PEP if exposed to HIV.

Patterns of sexual partnerships: How frequently men chose to have sex with new partners and whether or not they choose to form closed (exclusive) or open (non-exclusive) regular sexual relationships.

Patterns of disclosure and sexual negotiation: The extent of information sharing about HIV and STI diagnoses between sexual partners and the moderation of sexual acts based on this information.

Patterns of sexual behaviour: Whether men choose to engage in oral and/or anal intercourse, whether or not they use a condom and lubricant, which orifices (if any) ejaculation occurs in, and the use of poppers.

Elsewhere we consider these patterns in the populations from the perspectives of men making choices in their day-to-day lives and how we can approach influencing these choices. In particular we consider the incentives individuals have for making precautionary and risky choices.

Page last updated: 26 March 2012