Is serosorting working, or even possible?

There was a lot of discussion about serosorting at the Sixteenth International HIV Conference in Toronto last week, but not a lot of agreement about what it meant, whether it had an effect on HIV incidence, or whether it should be promoted as a behaviour.

Firstly, what is it? ‘Serosorting’ has been used by prevention researchers to mean four different things:

• Trying to restrict sex to people with your own HIV status.
• Trying to restrict unprotected sex to people with your own HIV status.
• HIV-positive people trying to restrict sex to other people with HIV.
• HIV-positive people trying to restrict unprotected sex to other people with HIV.

In the latter two cases, the rationale for saying that serosorting can only be something HIV-positive people do is because they know something definite and can disclose it: people who are “HIV-negative” are only definitely negative from the last time they tested until the next time they have unsafe sex.

In one of the last sessions of the conference, on men who have sex with men (MSM) and HIV prevention, Professor Jonathan Elford of the City University in London said: “I think serosorting among men who think they’re negative is not effective.”

This was for two reasons: firstly, data from gay African-American men in the United States show that they test for HIV as frequently as other gay men, but the incidence in the community is so terrifyingly high (in one study quoted by Ron Stall of the University of Pittsburgh, it was 15% a year) that two-thirds of HIV- positive black gay men remain unaware of their status; and, secondly, that people with HIV, unfortunately, do lie.

One HIV-positive audience member said he had declared his status as negative in an internet profile “and got far more hits than when I told the truth, which shows that guys who say they’re negative are believed.” He said that internet site owners should be discouraged from requiring men to declare their HIV status as this only created a false sense of security.

There is plenty of evidence that serosorting exists as a behaviour. Jonathan Elford presented a poster showing that serosorting was common not only among HIV-positive gay men in the UK but also among Africans. He interviewed 1,687 people attending HIV clinics in north-east London, 704 heterosexual Africans (a third of them men) and 758 MSM (17% of them from ethnic minorities).

He found that 53% of gay men, 62% of African women, and 72% of African men were in a current relationship. About 40% of the women and the gay men said their partner also had HIV and about 50% of the African men (definition 3 above).

When it came to unprotected sex (definition 4 above) 57% of the gay men and about 30% of the Africans said they had unprotected sex with their main partner if they were also HIV-positive. If their partner was HIV-negative, only 14% of the gay men and African women said they had unprotected sex and only 7% of the African men. This means that the men were 75% less likely to have unprotected serodiscordant sex than seroconcordant sex, and the women 50% less likely – which may say something about men’s greater power to determine condom use.

A Danish study (Cowan) compared the proportion of gay men who would end up having seroconcordant sex if it was left to chance with the proportion who actually did.

The researchers interviewed 4,857 gay men in 2000-2002 of whom 11% said they had HIV. They found that the HIV-positive men had a higher frequency of unprotected sex than HIV-negative men, “but,” as the poster presenters commented, “HIV-positive men are rarer: so it is more difficult to create a concordant relationship.”

If the 4,857 men were presumed to be one “pool” (the only ones available to each other as sex partners), then random mixing would mean that 11% of the HIV positive men and 68% of the negative men would have seroconcordant relationships.

In fact when it came to HIV-positive men, 23% had relationships with other with HIV-positive men, 22% were in a known serodiscordant relationship, and 55% did not know their partner’s status.

For HIV-negative men the figures were 48% in a seroconcordant relationship, 4% in one known to be serodiscordant, and 48% with a partner of unknown HIV status.

The most disappointing aspect of this study was the low rates of disclosure and knowledge of HIV status between partners. Sixty per cent of the gay men (negative or positive) said they did not disclose their status to sex partners. “Disclosure is not as much in demand as we expect,” was the researchers’ comment.

Does serosorting work? In some American cities, evidence of a “disconnect” between rising rates of sexually transmitted infections but steady or declining rates of HIV incidence has been seen as evidence that serosorting is working.

A team from San Francisco (Truong) confirmed that there had been a significant increase in sexually transmitted infections among gay men. Between 1998 and 2003, the number of cases of rectal gonorrhoea (a very good indicator of unprotected anal sex) almost doubled from 158 to 311 and, as in many other US and European cities, there was a substantial increase in the incidence of early syphilis from only eight cases in 1998 to 314 cases in 2003, though cases appear to have levelled off since then.

Data provided from three different and independent sources revealed that there was a significant increase in the amount of reported unprotected anal sex during this time (p

HIV testing data showed that the incidence of new infections peaked in 1999 at 4% amongst men attending anonymous testing sites and 5% amongst men using confidential HIV testing at sexual health clinics. However, by 2004, HIV incidence amongst gay men in San Francisco had levelled off.

True, HIV incidence is notoriously hard to establish as one has to distinguish the genuinely recent infections among all new diagnoses. But the investigators concluded that, while there’s only tenuous evidence of any decline in HIV incidence, it does appear to be plateauing after several years of increase.

Audience members questioned whether in fact lower-to-steady incidence rates might be more due to increased antiretroviral uptake and therefore lower average viral load in HIV-positive gay men; and also whether the serosorting was deliberate: were gay men actually disclosing their status more often, or were they just getting better at guessing?

Two things are necessary for serosorting: a high rate of HIV testing, and disclosure of status. A number of different posters looked at rates of disclosure by HIV-positive people all over the world.

A poster from Uganda (King) found that of 1,092 HIV-positive clients of the country’s largest AIDS NGO, TASO, 45% of them men, 42% of them had been sexually active in the past three months and of these 69% had disclosed their HIV status to their partner. Encouragingly, higher rates of disclosure were associated with higher rates of condom use, showing that the two activities are linked.

In qualitative interviews with a subset of 47 TASO clients, 24 said disclosing had had positive outcomes both in terms of safer sex and in strengthening partner and family relationships; only four said they had experienced negative consequences including blame, separation and violence.

Researchers said that “assisted disclosure” was a common and indeed preferred method of disclosure by clients: people would ask friends of the same sex as their partner to disclose on their behalf, and this was so commonly used that TASO was instituting a so-called ‘assisted disclosure’ programme.

In a similar survey in Cape Town, South Africa (Simbayi) the Population Research Council recruited over 1,000 people with HIV and asked them about their sexual behaviour. They were in the main new to living (knowingly) with HIV: the average time since diagnosis was two years. Encouragingly, 50% were taking antiretrovirals.

Ninety per cent had been sexually active in the last three months, with 40% of men reporting more than one sex partner but only 18% of women. Unprotected sex was very common and, despite the fact that 58% did disclose their HIV status to their partner – a somewhat lower proportion to the Uganda study - and there was little evidence of serosorting, with high rates of unprotected sex regardless of partner’s status. However those who did disclose their status had less unprotected sex than those who did not.

Professor Simbayi said that the most frequently cited reason for nondisclosure was fear of the information spreading to the wider community, and that both nondisclosure and high risk sexual behaviour was relayed to previous HIV-related discrimination experiences such as being dismissed from work.

In a nationwide French study of 1,187 HIV-positive people in steady relationships (Bouhnik), Bruno Spire and his team found that of the gay men only 5% had not disclosed to their partner but that 10% were unsure of their partner’s HIV status and that 17% had had potentially unsafe sex with them. Among heterosexual men, 5% had not disclosed, only 5% did not know their partner’s HIV status but a quarter - 25% - had potentially unsafe sex with their partner. And among heterosexual women, 9% had not disclosed their status, 19% did not know their partner’s status and 33% reported potentially unsafe sex.

Spire commented that amongst gay men disclosure was predictive of unprotected sex and that men who disclosed were less likely to have unsafe sex. But among heterosexuals, any influence of disclosure was overwhelmed by the fact that poor living conditions and poverty were much stronger predictors of unprotected sex.

Finally, in an upbeat presentation, Fiona Percy-de Korte reported on high rates of disclosure among recipients of antiretrovirals in the Botswana National Treatment Programme.

Among this cross-sectional survey of 275 patients in two districts, she said that 90% of ARV recipients had disclosed their HIV status to family members, 71% to their spouse or partner, 54% to friends, 26% to a church leader, and 20% to their employer.

Seventy-three per cent said they always used a condom during sex. And 48% said they had reduced their number of sexual partners since diagnosis. But 39% said that receiving ARVs had resulted in sexual disinhibition of some sort – either reduced condom use or more partners.