Evidence against

In a setting where HIV testing is lower than it is in the USA, serosorting may make a lower contribution to HIV-incidence reduction, because people are more likely to have inaccurate knowledge of their own and others’ HIV status. A team of researchers in the Netherlands,¹ where about 70% of gay men have tested for HIV as opposed to 90% in the USA, tried to compare risk factors in a group of 281 men classed as ‘lower risk’ and 232 men classed as ‘higher risk’, based on numbers of partners and condom use. Behaviour was assessed by surveying the two groups of men during 2004 and 2006.

The men classed as lower risk were from the Amsterdam Cohort Study – an ongoing study of mainly gay men – which is estimated to have had a relatively stable annual HIV incidence of 1.24% from 1999 to 2005. Men classed as higher risk were recruited from the Amsterdam STI Outpatient Clinic population, in which annual HIV incidence is estimated to have risen in the same period to 3.75%. In the high-risk group 46% were HIV-positive compared to 7% in the lower-risk group.

In both groups, men who discussed their HIV status with their partners were more likely to have unprotected anal sex, and some degree of serosorting was practised in both groups. Regarding the HIV-positive men specifically, 72% in the lower risk group had unprotected sex with other HIV-positive partners and 22% with partners who were HIV-negative or of unknown status, while in the high-risk group the figures were 82% and 36% respectively. However, the fact that HIV incidence in the high-risk group was so much higher led the researchers to conclude that their serosorting behaviour did not compensate for their lower levels of condom use.

In a German gay internet and magazine survey already discussed,² exclusive serosorting behaviour amongst gay men (only ever having unprotected sex with men of the same HIV status, definition ‘B’) was associated with a five times higher risk of having had an HIV diagnosis in the last 18 months than men who had used condoms and/or monogamy as their method of reducing HIV risk. Serosorters even had a 60% greater risk of having a recent HIV diagnosis than men who had no protection strategy at all.

However, the significance of this result was compromised because the survey failed to establish whether the serosorting behaviour happened pre-diagnosis with HIV-negative men (definition ‘F’) or post-diagnosis with HIV-positive men (definition ‘D’), or both. It could not, therefore, establish whether the serosorting was an attempt to avoid HIV infection or an adaptation to it.