New directions in HIV prevention: serosorting and universal testing

The use of HIV testing and information about HIV status as an HIV prevention tool remains a complex and controversial area of debate, largely due to issues of trust – trust in confidentiality of information, trust that healthcare providers will not test without consent, and trust that partners are telling the truth about their status. For all these reasons prevention experts have shied away from addressing the topic in the developed world, despite the fact that HIV testing is considered an essential ingredient in the prevention mix in low-income countries. At last month’s Thirteenth Conference on Retroviruses and Opportunistic Infections in Denver, it was clear that the issue of HIV testing’s role in prevention cannot be avoided any longer.

A contentious issue in the world of HIV is whether – given that, once tested, HIV-positive people do in general have less unsafe sex – a drive for HIV testing to be universal and ‘normalised’ is a way forward in prevention.

Two symposium contributions at the conference exposed this as a very ‘live’ issue, especially in the USA. In one, Tom Coates of the University of California, San Francisco, looked largely at drives to normalise HIV testing in Africa and contrasted Africans’ general agreement with normalising HIV testing (82% of people in Botswana, for instance, think that the routine testing introduced by President Festus Mogae is 2004 was a good thing) with their individual distress at contemplating a positive result. In one qualitative survey, one said: “I have a dream of having children; if I test positive my dream will be shattered.” Another said: “My father will chase me away from the house and call me Satan.”

But in a talk immediately after Golden’s serosorting presentation, Timothy Mastro of the Centers for Disease Control (CDC) got a considerably rougher ride from the audience, and session chair Jeff Klausner extended the Q&A session after his presentation saying that “It’s not often we get the most prominent members of the CDC in one room to answer these questions.”

Mastro said that a CDC study showed that HIV-positive people reduced the amount of unprotected serodiscordant sex they had by 68% after diagnosis. This led them to believe that the 25% of people who did not know their HIV status in the USA contributed about 50% of infections.

He cited the startlingly high prevalence and incidence figures among gay men and particularly black gay men in cities other than San Francisco in the report quoted above. In a large sample of gay men in five US cities 25% of gay men had had HIV and 48% were unaware of their infection; 46% of black gay men were positive and 67% did not know it. Late testing was also common: 45% of AIDS diagnoses were among people who had been diagnosed less than twelve months previously.

He said that HIV testing in the USA had not been increasing in recent years despite the fact that that the CDC had launched its Advancing HIV Prevention strategy in 2003 to make voluntary HIV testing a routine part of medical care.

He said that only about one in 500 visits to hospital emergency departments involved an HIV test despite the fact that, when tested, rates of previously undiagnosed HIV among A&E patients varied from 1.3% to 3.2%.

In Dallas, Texas, which adopted opt-out testing in its sexually transmitted infections' (STI) clinics back in 1997, the proportion of patients tested for HIV increased by over 50% (from 78% to 97%) in one year and the number of positive tests had gone up 60% from 168 to 268.

He showed a notice from a Dallas STI clinic which said that “All patients seen in this clinic will be tested for gonorrhoea, syphilis, Chlamydia and HIV.”

He said that before opt-out testing had been adopted in pregnant women in the United Kingdom only 35% had tested because they feared it indicated high-risk behaviour whereas 88% accepted opt-out testing.

“We think the need for extensive pre-test counselling is less because it’s 2006 and people now have a high level of knowledge about HIV,” he said.

After two studies published in the New England Journal of Medicine last year found that routine screening would also be cost-effective, the CDC had decided to revise its HIV screening guidelines and would be recommending routine, voluntary screening for all persons aged 13-64 in healthcare settings, not based on risk, and annual HIV testing for people with risk behaviour. Pre-test counselling would not be required. ‘Health care settings’ includes all hospital in-patient and out-patient departments and community clinics as well as STI clinics. An exception would be made of prisons, where it was recognised that receiving an HIV diagnosis created profound difficulties both for inmate and institution.

This is quite a radical proposal which would require revising a myriad of local regulations.

Mastro was faced with a battery of questions from questioners with numerous concerns after his presentation . Among them were:

• How would the CDC move from recommendation to implementation? It had recommended names-based reporting years ago but it had taken the threat of the removal of federal funds for states to move towards this.
• If the information given in pre-test counselling and discussion around informed consent is removed, where are patients going to get any option to talk about HIV and harm reduction?
• How real is the ‘voluntary’ nature of the testing when the photo he showed said ‘You will be tested for HIV?’
• How is the opting-out process to work and how will it be recorded? Without adequate recording, patients could say they were tested without consent.

Dr Matt Golden of the University of Washington, Seattle Center for AIDS and STDs (sexually transmitted diseases, US expression) looked at a controversial ‘prevention activity’ adopted by gay men: serosorting, which he defined as “The practice of preferentially choosing sex partners, or deciding not to use condoms with selected partners, based on their disclosed, concordant HIV status.”

Five studies conducted from 1992 up to 2005 had found that gay men were between 2.5 to 9.1 times more likely to have unprotected anal intercourse with partners they knew had the same HIV status as themselves than with partners of differing HIV status.

Data from Golden’s own clinic found that HIV-positive patients were particularly likely to serosort. Forty and 49 per cent of his HIV-positive patients, respectively, had unprotected receptive and insertive sex with HIV-positive partners but only 3% and 6%, respectively, with HIV-negative partners.

In his HIV-negative patients 31% and 37%, respectively, had unprotected receptive and insertive sex with HIV-negative partners, and 19% and 15% respectively had unprotected receptive and insertive sex with HIV positive partners – less, though still a surprisingly high figure.

“Where the whole system breaks down,” however, Golden commented, “is where the other partner is of unknown status.” Here partners were almost equally likely to have unprotected insertive sex regardless if their own status or if the partner’s was unknown. In the case of receptive sex, there was some evidence that positive gay men were attempting to adopt ‘strategic positioning’. HIV-positive men were somewhat more likely (31% vs 24%) to have unprotected receptive rather than insertive sex with partners of unknown status; conversely HIV-negative men were somewhat less likely (16% vs 22%. Golden did not say whether any of these differences reached statistical significance.

Golden then investigated whether serosorting was actually reducing the number of serodiscordant partners that gay men had, regardless of condom use. The answer was yes. In a population like Seattle where 15% of gay men have HIV (not dissimilar to London), if gay men chose partners completely at random, and if they all had the mean number of partners rather than a few having many and many having a few, you would expect 54% of gay men to have at least one serodiscordant relationship per year (with the figure obviously lower for people with few partners and higher for those with many).

In fact about 35% of gay men had had at least one serodiscordant partner, so serosorting appeared to be reducing the number of serodiscordant relationships by about 40 per cent, though Golden also suggested some of this was due to the fact that gay men tend to have sex with men fairly near their own age, and that because young men are less likely to have HIV than older men, some of this concordance was purely due to age similarity. Golden also found that 13-18% of gay men were ‘exclusive serosorters’, i.e. only had unprotected sex with seroconcordant partners.

Is serosorting actually protective? When it comes to HIV-negative men, Golden found that the rate of new HIV diagnosis among patients who had unprotected sex but tried only to do it with same-status partners (2.6%) was intermediate between men who had unprotected sex regardless (4.1%) and men who attempted always to use condoms (1.5%). Adjusting for the number of partners, whereas condom use was 76% effective in preventing new HIV infections, serosorting was about 40% effective.

As a ‘control’, Golden also looked at the rate of STIs and in this case, as you would expect, there was no difference in the STI rates between serosorters and non-serosorters.

Was serosorting increasing? Golden showed data from San Francisco and London which suggested that the proportion of unprotected sex that was discordant, especially as practised by HIV-positive men, was decreasing, but said he had not seen the same pattern in Seattle.

And Susan Buchbinder from the San Francisco Department of Public Health, commenting in an earlier prevention seminar, said that unfortunately the latest assessment from SFDPH was that, based on a number of key indicators in a number of studies there was no decline in seroincidence among gay men, contrary to a study using a detuned assay to detect incidence which was reported on Aidsmap last year. This did not preclude the possibility that serosorting could drive down infection rates, she added.