Compare the two charts below from the San Francisco Department of Public health’s 2004 HIV Epidemiology Report (2004):

The first shows that unprotected anal sex (any episode over the previous six months) was practised by about a third of HIV-negative men consistently from 1998 to 2004, with very little change. In contrast the proportion of HIV-positive men having unprotected sex went up during the same period from 38 to 52 per cent.

The second chart, however, shows that the high and (in positive men) increasing levels of unprotected sex did not translate into increasing levels of unsafe sex. The amount of potentially serodiscordant sex – the true measure of the degree of HIV risk among a community – has declined from a peak in 2001 among both positive and negative men.

Similarly, in the UK, a decline in serodiscordant unprotected sex may be starting to happen too, and amongst gay men and among positive men the decline appears to be more dramatic. The annual survey of gay men using London gyms (Elford 2005b) between January and March each year, in which so far more than 5,000 gay men have participated, found last year that the amount of serodiscordant unprotected sex increased between 1998 and 2002, but has remained stable since then or slightly declined among HIV-negative men, and has declined significantly (p=<0.05) in HIV-positive men, from a peak of 41 per cent in 2002 to 20 per cent in 2005.

How can gay men be having more unprotected sex while at the same time having less sex of the kind that could spread HIV?

Clearly, by increasing the proportion of unprotected sex they have with men with the same HIV status.

The phenomenon by which HIV comes to be brought under better control by people seeking out sex partners with their own HIV status has become called Serosorting, and is usually applied specifically to situations where unprotected sex is being sought.

Serosorting may be one reason why an increase in STIs in gay men, especially syphilis, has not led to a concomitant increase in HIV infections. This has shaken an assumption previously used by many HIV epidemiologists – that increases in STI rates can be used as surrogate markers or predictors of increases in HIV. This has found not to be the case. In the US huge increases in syphilis in gay men have not coincided with equally big increases in HIV. (Conversely, in certain African countries like Zimbabwe, STI rates have gone down while HIV incidence has not, because more HIV infections are now occurring within marriages and fewer within casual encounters (Gisselquist 2003).

This was first noticed in 2003 when two US cities, Seattle and San Francisco, noticed that HIV incidence among gay men attending for HIV tests was starting to decline even though syphilis rates had increased 25-fold (Buchacz 2004). At the time this was partly put down to a lot of syphilis being spread via oral sex, and particularly among gay men, as they were unaware of this transmission route. However only 25 per cent of syphilis, it was thought, was being spread orally. Syphilis is also more contagious than HIV so can spread more rapidly through a connected network of sexual partners.

But Dr Jeffrey Klausner, San Francisco Health Department’s Director of STI Prevention, commented at the eleventh Retrovirus Conference where these findings were presented that there was a lot of evidence that much of the lack of a rise in HIV was due to serosorting.

This phenomenon – of concentrating the unprotected sex you have with people of the same HIV status – was previously thought to be rare.

Serosorting is one example of ‘negotiated safety’. It was thought, as an HIV prevention strategy, to be quite uncommon, particularly in the UK. Research conducted in 1996 showed that only 16 per cent of men questioned were having unprotected anal sex in a relationship were doing so in the light of knowledge about both partner's HIV status, compared with 60 per cent of men in Australia.

However serosorting may be an important way by which HIV-positive men attempt to reduce the risk of their passing on HIV to others. In one US study (Lightfoot 2005) of young US gay men (most of them black or Latino) aged 15-24, about 34 per cent of youth with multiple partners and 28 per cent with one primary partner had unprotected anal intercourse.

However they were overwhelmingly more likely to have unprotected sex with other partners perceived to be HIV-positive than with partners whose HIV status was negative or unknown.

In some cases the perception was probably accurate. The estimated number of instances of unprotected sex was 32 times greater among youth in a committed relationship if their regular partner was thought to be positive than if he was thought to be negative but at high risk of having other STIs (e.g. in an open relationship), and 18 times more if he was thought unlikely to have STIs (e.g. if thought monogamous). In at least some of these cases, the HIV status of the partners must have been known, rather than just a perception.

However, it is the word ‘perceived’ that has made HIV prevention workers very wary of encouraging serosorting as an HIV prevention technique. Attempting to guess a partner’s HIV status, many studies have shown, is doomed to failure, especially as many so-called ‘negotiations’ do not take place with words. In a backroom or anonymous sex situation, people’s HIV status is deduced by their status. The lack of insistence on a condom is assumed by a positive man to be a sign that the other is positive (“He must be positive or he wouldn’t let me do that”). A negative man makes the opposite assumption (“If he was positive he would protect me and use a condom”).

The 2002 UK Gay Men’s Sex Survey Out and About demonstrates the false assumptions gay men make about other’s status very clearly – and underlines the disincentive many HIV-positive men have to disclose.

A summary of the survey’s findings on Aidsmap says:

“Of the participants whose most recent HIV test was negative, two-thirds (65.3 per cent) said they would expect an HIV-positive man to disclose his status before having sex. Even more men who had never tested for HIV had the same expectation (77 per cent). In contrast only just over a third of HIV-positive men expected that a partner would disclose their HIV status.

44 per cent of HIV-negative or untested men said they would not want to have sex with the man who’d just disclosed his HIV status to them – and this rose to 56 per cent of men who had never had an HIV test."

Report authors Sigma Research comment: “Expectations that men with HIV will tell a prospective sexual partner their HIV status are still widespread. Over a third of men not tested [HIV] positive both expected a positive partner to disclose their status prior to sex and would not want to have sex with them if they did.

“In this climate, it is difficult to see what incentive men with HIV have for disclosing their status.”

In addition, significant numbers of gay men do not know their HIV status. In the year 2000 Gay Men’s Sex Survey (Time for More), 44 per cent of respondents had never tested for HIV, but of those only 1.3 per cent thought they were ‘probably or definitely positive’.

Even if a lot of the others are correct that they are negative, this is still only an eighth of the 10.9 per cent prevalence reported from another survey of gay men in UK cities (Dodds 2004). Of the HIV-positive men in this survey, who were tested anonymously with an oral saliva test, a third (32.5 per cent) were unaware of their infection and over one in five (21.2 per cent) said their last HIV test was negative.

In these circumstances, how can making safer-sex decisions on the basis of a partner’s HIV status possibly be a way of not being infected or infecting someone? The idea that serosorting or making other safer-sex decisions on the basis of a partner’s HIV status can possibly contribute to HIV prevention is a challenge to orthodox HIV prevention approaches, and makes many people deeply uneasy:

• As above, people may be making decisions on the basis of assumptions or inaccurate information.
• It relies on disclosure – something still practiced by a minority of gay men with HIV. In one unpublished survey by Gay Men Fighting AIDS, only 20 per cent of respondents said they always disclosed their HIV status (positive or negative) to partners, forty per cent said they sometimes did, and forty per cent said they never did.
• It is seen as diluting the ‘use a condom’ message and providing ways for men to rationalise unsafe behaviour.
• ‘Serosorting’ men can do things that at first look like ways of wilfully increasing their HIV risk rather than reducing it. For instance, gay men who advertise for ‘bareback’ sex may seem to be nothing other than irresponsible. If however ‘bareback’ is a code for ‘HIV-positive’, or if bareback discussions lead to disclosure of HIV status, the net result may be a paradoxical isolation of HIV within a specific group.
• Even if this is the case, HIV-positive men having unprotected sex together are still vulnerable to other STIs, to hepatitis C infection, and possibly (especially during the first three years of infection) to infection with a second strain of HIV (‘superinfection’).
• All ‘negotiated safety’ unprotected sex strategies between men of differing HIV status are likely to involve a considerably greater risk of HIV infection than protected sex does.

However there is indirect evidence that a lot of HIV-positive men, in particular are at least attempting to restrict their unprotected sex to other HIV-positive men. The debate around this kind of ‘negotiated safety’ centres on this dilemma:

• Should prevention messages concentrate on the fact that these practices still involve considerable HIV risk, and should therefore be discouraged?
• Or, since the gay men (and heterosexuals, though negotiated safety among heterosexuals is less well studied) who use these strategies to minimise HIV risk are unlikely to be persuaded back to consistent condom use, should prevention messages encourage behaviours that enable them to happen, such as disclosure of HIV status? 

In one analysis of the ongoing survey of gay men attending London gyms (Elford 2001), the authors detected two very different strategies being adopted by HIV- positive and negative gay men to avoid infection.

In HIV-negative men what the authors call “concordant UAI” and we call serosorting was mainly restricted to main partners. Over one in four (28.6 per cent) practised it with their primary partner and only five per cent with casual partners.

In HIV-positive men concordant UAI was equally practised with main partners (22.2 per cent) and with casual partners (20.6 per cent).

These unprotected sexual encounters were those restricted to ones where men were fairly sure their partner was of the same HIV status. If all men had an equal tendency to be uncertain or to be making decisions on the basis of assumption, the authors argue, one would expect the lower proportion of unprotected encounters with casual partners to apply to both HIV-negative and positive men. The significantly greater difference can only be explained, at least in part, by the fact that it is possible for HIV-positive men to disclose their status with certainty:

“Seroconcordance among negative men can only be established with confidence if both men test for HIV together. For this reason it is difficult for HIV-negative men to establish concordance with a casual partner.”

"On the other hand, HIV-positive men can establish concordance, be it with a casual or regular partner, simply by mutual disclosure. This requires no confirmatory test.”

They add a caution: “Although seroconcordant UAI among positive men carries no risk of HIV transmission to an uninfected person, it raises the possibility of reinfection and drug resistance for the men themselves.”

But is this disclosure actually taking place?

The internet as disclosure venue

Concern has been expressed in recent years that the huge increase in sexual encounters being arranged on the internet may be facilitating an upsurge in unprotected sex. This concern has been fuelled by the rise in the number of explicit ‘barebacking’ sites.

In a more recent study by the same group of researchers (Bolding 2005), four groups of gay men (internet chatroom users, London gym users, HIV-positive men attending clinics, and HIV-negative men attending for an HIV test) were asked about their patterns of unprotected sex and internet use.

HIV-positive men in the clinic and gym samples who used the internet to look for sex were significantly more likely to report unprotected anal sex with men of the same HIV status than other men (p < 0.05).

The investigators also established that in both the clinic and gym samples, HIV-positive and HIV-negative men who used the internet to find sex were more likely to report non-concordant unprotected anal sex with a casual partner than other men (p < 0.05).

So far, this looks as if the internet is facilitating increased levels of unsafe sex.

But the investigators found that internet users were also more likely to have concordant unprotected sex, i.e. to ‘serosort’.

In all samples, HIV-positive men who looked for sex through the internet were significantly more (p <0 .05) likely to report concordant unprotected anal sex with a partner they met on-line rather than off-line. For example, 10 per cent of the clinic sample reported concordant unprotected anal sex with a man they met on-line, and only four per cent said they had had concordant unprotected sex with a man met off-line.

What’s more, the investigators found that amongst the HIV-positive clinic sample, men said that they were more likely to disclose their HIV status to men met on-line (24 per cent) than men met off-line (14 per cent, p < 0.001).

They also found that the apparent causal link between internet use and serodiscordant unprotected sex was an artefact. When asked directly about how they met partners they subsequently had unprotected sex with, there was no evidence that gay men, whether HIV-positive, negative or untested, were more likely to meet partners for discordant unprotected anal sex on-line rather than off-line.

For example, among HIV-negative men in the internet sample, 10 per cent reported non-concordant unprotected anal sex with men met on-line only, 11 per cent with men met off-line only, and six per cent with men met on- and off-line. "In fact," note the investigators, "for HIV-negative men in the clinic and gym samples, the reverse pattern was seen; they were more likely to report non-concordant unprotected anal sex with a casual partner met off-line."

In other words internet meets were more likely to result in unprotected sex which was concordant: serodiscordant unprotected sex was at least as likely to occur during casual encounters.

"What is new about this study is that we can establish whether the excess risk of HIV and sexually transmitted infections seen among gay men who looked for sex through the internet actually occurred with the men they met on-line", write the investigators.

"In our study, HIV-positive men who looked for sex through the internet were more likely to meet other HIV-positive men with whom they had (concordant) unprotected anal intercourse on-line rather than off-line ", note the investigators. They add "Men who looked for sex through the internet were no more likely to meet their non-concordant unprotected anal intercourse partners on-line than off-line. This was seen for HIV-positive, HIV-negative and never-tested men alike."

The investigators suggest that the internet may provide a safe space for HIV-positive men to disclose their health status, ‘thus facilitating "filtering" or "serosorting’ of sexual partners’.

This is not the only study that suggests the internet is being used by HIV positive men as a safe place in which to disclose and negotiate the level of sexual safety they want.

A study presented at the 2005 Retrovirus Conference (Chiasson 2005) of users of 14 US-based gay websites found that 28.5 per cent of men had unprotected sex during their last encounter.

However twice as many instances of unprotected sex happened after casual offline encounters that after internet meets.

And online meets, however, were more likely to involve discussion of HIV status. More than half of the men who had met online had discussed their HIV status before sex as opposed to a third of partners who met offline.

The authors comment: “The large number of men on-line and the diversity of their risk and ways of meeting partners show that the Internet provides a unique opportunity for far-reaching behavioural interventions.”

Viral load and negotiated safety

HIV status is not the only thing that can be disclosed by men attempting to minimise their HIV risk. Several recent studies have found that gay men are questioning each other about their HIV viral load in order to try and establish if they are infectious.

In a study from San Francisco (Goldhammer 2005) 78 per cent of 507 gay men questioned were familiar with the term `viral load` and one third (111 of the total sample) had discussed viral load with a partner of a different HIV status during the previous year in order to make decisions about which sexual practices to engage in.

Of those who had discussed viral load, more than half estimated that they used viral load disclosure to guide sexual decision-making in at least 70 per cent of their sexual encounters.

In another study from Sydney (Van de Ven 2005) researchers asked 119 men who were in an HIV-serodiscordant regular relationship about whether they used viral load as a basis for their decisions on condom use. Twice as many (39.4 per cent) reported unprotected anal intercourse when the partner’s HIV last viral load test was undetectable as when it was detectable (20.8 per cent).

HIV-negative couples and negotiated safety

Negotiated safety agreements are also used by HIV negative couples to attempt to reduce the chance of HIV entering the relationship. In one study (Guzman 2005) of 76 HIV negative gay men with HIV negative steady partners, 17 per cent of the men did not practice anal sex and 22 per cent maintained 100 per cent condom use in all anal sex, inside or outside the relationship.

Another 11 per cent had unprotected sex within and outside the relationship and had not negotiated any rules prohibiting it.

But 39 of the men (51 per cent) had some sort of negotiated safety agreement in place in their relationship. Nineteen (25 per cent) had unprotected sex with each other but had negotiated total monogamy. Three (four per cent) disallowed anal sex with partners outside the relationship but allowed other sex. And 16 (21 per cent) allowed anal sex outside the relationship as long as it was always protected.

This left six (eight per cent) who had protected sex within the relationship but allowed unprotected sex outside it – a stance protecting their partner but not themselves.

However these negotiated safety agreements were often broken. Eleven (14 per cent of the whole group, 29 per cent of those who had an agreement) had broken it in the previous three months. However they were less likely to break it if there was a requirement that they ‘must always tell’ if it had been broken; only 18 per cent of those who had an agreement had broken it if the rule was always to say if they had.

Disclosure: an HIV-positive-controlled safer sex strategy

There are two necessary conditions for any of these attempts to reduce HIV risk. Men have to know their HIV status. And disclosure (of HIV status, or in the case of the HIV-negative couples, HIV risk behaviour) has to happen.

The recent upsurge in studies of negotiated-safety behaviour has led to the asking of a question. In most of these situations the partner (or partners) with HIV are the ones in possession of the knowledge that makes a difference to behaviour.

Drives to get more high-risk people to test for HIV have been based on the assumption that, once they test positive, people will moderate their behaviour. In fact, as we have seen, HIV-positive people end up having more unprotected sex rather than less. This has led to anguished questioning among HIV prevention experts and a lot of hostile media and intensified stigma against the perceived irresponsible behaviour of people with the virus.

If, however HIV-positive people are at least attempting to inform and protect their sexual partners, should more HIV prevention money be directed towards enabling them to do so? After all, HIV- positive people form 50 per cent of any risky sexual encounter and 100 per cent of those with the knowledge to reduce its risk.

References

Bolding G et al. Gay men who look for sex on the internet: is there more HIV/STI risk with online partners?AIDS 19: 961-968, 2005.

Buchacz K et al. Trends in Primary and Secondary Syphilis and HIV Seroincidence among Men Who Have Sex with Men in San Francisco, 1998-2002. 11^th Conference on Retroviruses and Opportunistic Infections, San Francisco, abstract 882004, 2004. 

Chiasson MA et al. A Comparison of On-line and Off-line Risk among Men Who Have Sex with Men. 12^th Retrovirus Conference, Boston, abstract 168, 2005. 

Chinouya, M. & Davidson, O. The Padare Project: Assessing health-related knowledge, attitudes and behaviours of HIV-positive Africans accessing services in north central London. African HIV Policy Network, February 2003.

Dodds JP et al. Increasing risk behaviour and high levels of undiagnosed HIV infection in a community sample of homosexual men. Sex Transm Infect 80: 236-240. 2004.

Elford J et al. High-risk sexual behaviour among London gay men: no longer increasing. AIDS, 2005 (in press).

Elford J et al. High-risk sexual behaviour among London gay men: no longer increasing. AIDS, 2005 2005b (in press).

Gisselquist D et al. Let it be sexual: how health care transmission of AIDS in Africa was ignored. Int J STD AIDS 14(3): 148-161, 2003.

Goldhammer H et al. Beliefs about viral load, sexual positioning and transmission risk among HIV+ men who have sex with men (MSM): Shaping a secondary prevention intervention. 2005 National HIV Prevention Conference, Atlanta, USA, presentation W0-D1201.

Guzman R et al. Negotiated safety relationships and sexual behaviour among a diverse sample of HIV-negative men who have sex with men. JAIDS 38(1): 82-86, 2005.

Lightfoot M et al. The Influence of partner type and risk status on the sexual behaviour of young men who have sex with me living with HIV/AIDS. JAIDS 38(1): 61-68, 2005.

See http://www.aidsmap.com/en/news/930CCABC-A29F-4CCC-B7A9-52D93402E765.asp

Van de Ven P et al. Undetectable viral load is associated with sexual risk taking in HIV serodiscordant gay couples in Sydney. AIDS 19(2): 179-184, 2005.

Weinhardt L et al. HIV Transmission Risk Behavior among Men and Women Living with HIV in 4 Cities in the United States. JAIDS 36(5): 1057-1066, 2004.