A Canadian study which compared the HIV status of gay men’s recent sexual partners with what would be expected if they chose partners regardless of status has found that HIV-negative men who used PrEP were nearly twice as likely to have HIV-positive partners as those who did not.
It also found that men using PrEP were more likely to have sex with other PrEP users than with men not using PrEP. Conversely, non-users were more likely to have sex with non-users.
It also found that HIV-positive men were still far more likely to have partners who also had HIV than men they knew were HIV-negative, though it’s not clear how much of this is due to choice, and how much to being discriminated against.
Linwei Wang of St Michael's Hospital, Toronto and colleagues looked at HIV status and PrEP usage among 1179 gay and bisexual men who completed a survey of gay men in Montreal called Engage-Montreal in 2017 and 2018. This was not an internet survey; it was enrolled via Respondent-Driven Sampling (RDS), in which men already recruited ask their contacts to join the survey. It included trans men, but all respondents had to have had sex with another man (cis or trans) in the last six months.
The survey asked two different questions about the HIV status of partners. Firstly it asked men about the number of all partners in the past six months who had had HIV, or whose HIV status was unknown, regardless of what kind of sex they had had (oral or anal).
Secondly, it asked a more specific question about the last five partners men had had, including whether they’d had oral or anal sex, and whether they had known the HIV status of their partners before they’d had sex.
The reason for asking these two types of question was for sensitivity analyses – in other words, to cross-check one set of answers against another to get a more accurate picture of people's behaviour.
HIV-negative men were also asked about PrEP in relation to their most recent sexual partner – whether the respondent had been using PrEP and whether their partner was using it too.
These figures were then compared with the proportion of partners with HIV that would be expected if all respondents (both HIV-positive and negative) had selected their partners regardless of their HIV status – in other words, the proportion of partners of different HIV statuses they’d have purely by chance. The numbers of available positive and negative partners was not only based on HIV prevalence in Montreal, but also on other factors such as their age distribution.
A similar calculation was made for the chances of an HIV-negative respondent having a partner taking PrEP.
Results: HIV status and serosorting
The average age of the 1137 respondents was 34 and 18.5% defined as something other than gay (bisexual, queer etc). Thirty-eight per cent defined themselves as being of an ethnicity other than French or English Canadian.
Eighteen per cent said they had HIV while 9% did not know their HIV status, so just over three-quarters knew they were HIV-negative, or had been last time they tested. People who did not know their HIV status had had fewer sex partners in the last six months (three) than people who did (five, for both HIV-positive and negative men).
Of the HIV-negative men, 13.5% had used PrEP in the past six months. Of the HIV-positive men, 96% were on antiretroviral therapy (ART), of whom 92% were virally suppressed.
It is not surprising that respondents did not know the HIV status of 44% of their sexual partners in the last six months. This could underestimate the proportion of HIV-negative partners if untested partners were largely negative, or underestimate the number of HIV-positive partners if they did not disclose due to fear of rejection. However, since there was no way of knowing the true HIV status of these partners, they were not included in the analysis.
Leaving aside the partners of unknown status, if sexual partners in the survey met purely by chance, then it was estimated that 76% of people’s partners would be HIV negative and 24% positive.
The HIV-positive respondents reported considerably more HIV-concordant partners than this: 66% (of those whose status they knew) instead of the expected 24%. In other words serosorting is still very much alive as a phenomenon among gay men with HIV.
HIV-negative men are serosorting too: instead of the 76% of partners who one would expect to be HIV-negative by chance, 88% of the partners of HIV-negative men also did not have HIV.
Results: PrEP use, serosorting and ‘PrEP matching’
PrEP use, however, is starting to make a difference to this pattern among HIV-negative men. For a start, it seems to be encouraging discussion of HIV status: in HIV-negative men taking PrEP, the proportion of partners whose HIV status they did not know was only 31%, versus nearly 50% of the partners of men not on PrEP.
In addition, among HIV-negative men taking PrEP, the proportion of their partners who had HIV (17%) was almost twice the figure for men not taking PrEP (9%).
If HIV-negative men had met partners by chance, the proportion of their partners using PrEP would be 28.5%. However, among respondents who were using PrEP themselves, this proportion nearly doubled, to 51%. There is thus clear evidence of 'PrEP sorting', or 'PrEP matching' as the authors call it, among HIV-negative men. This finding is similar to a previous one (Grov) which found that US men taking PrEP reported that 41% of their partners were also taking PrEP, versus 28% of partners of men not taking PrEP.
The authors point out that, while 'PrEP matching' doubly benefits men taking PrEP, it is a lose/lose situation for men not taking PrEP, who are less likely to benefit from their partners’ protection from HIV conferred by PrEP, as well as not having any of their own. If PrEP use becomes even more widespread, HIV transmission could become concentrated in men not taking it.
Limitations and unknowns
This study has several limitations. As mentioned above, the HIV status of those with unknown status (either respondents or partners) cannot be estimated. Also, while on a population level this study seems to indicate a shifting of sexual risk behaviours and a degree of relaxation about serosorting, it cannot say whether the apparent 'PrEP matching' was deliberate, or whether partner selection was influenced by other factors such as assertiveness or health literacy.
The most obvious gap, however, is the question about HIV-positive men and their viral suppression status: is knowledge of viral load and 'U=U' also leading to a relaxation of serosorting among HIV-positive men and their partners? The study could not compare serosorting among men with undetectable and detectable viral loads as there were only 33 of the latter, which would have limited the statistical power.
However, the finding that the majority of men with HIV still have other positive men as partners suggests that they continue to face rejection from HIV-negative men, that they are still unconvinced of the U=U message themselves, or that they are unconvinced that partners will believe it. As the authors comment, even if HIV-negative men became perfectly relaxed about having positive men as partners, if HIV-positive men still serosorted, it would seem like negative men still were serosorting too.
In general though, this study bears out the authors’ hypothesis, that “With the scale-up of PrEP…serosorting and related patterns of ‘Who has sex with whom’ may be changing.”
Wang LW et al. Population-level sexual mixing by HIV status and pre-exposure prophylaxis use among men who have sex with men in Montreal, Canada: implications for HIV prevention. American Journal of Epidemiology, online ahead of print, 15 October 2019 (open access).
Grov C et al. Prevalence of and factors associated with the use of HIV serosorting and other biomedical prevention strategies among men who have sex with men in a US Nationwide Survey. AIDS and Behavior 22:2743-55. 2018.