The group studied were 815 participants in the Amsterdam Cohort Study, which is a cohort of HIV-negative gay men that has been running ever since 1984 and requires participants to visit a clinic run by the Public Health Service of Amsterdam and report on their sexual behaviour in the previous six months. For the purposes of the present study, men who participated between May 2007 and January 2017 were included. Average length of follow-up was 5.34 years with a median number of 12 visits. The age of participants ranged from 18 to 70 with a median age at first visit of 35.6.
The four indicators of sexual risk behaviour most strongly associated with subsequent HIV infection were number of casual partners with whom participants had receptive anal sex, number of receptive condomless anal sex acts with casual partners, number of condomless sex acts as the insertive partner with people of positive or unknown HIV status, and any anal intercourse during group sex situations. These were combined into a “sexual risk score”. This score correlated closely with the likelihood of becoming HIV positive and ranged from zero to 4.87 with a mean risk score of 0.53. The highest risk score equated to a 56% probability of acquiring HIV within a year of first observation.
Statistical analysis found that the study participants fell into the three risk groups detailed above and, from averaging the changes in behaviour over time of individual members of the three groups, plotted an average ‘risk career’ for each group. Note that this technique deliberately smooths out individual-level fluctuations in risk over time, with the aim of pinpointing ‘signatures’ of risk that could indicate intensified help and support.
Annual HIV incidence in the lower-risk majority was 0.85%, and 4.7% of this group eventually caught HIV. In the risky-when-young group annual HIV incidence was 4.01%, and 20.5% eventually caught HIV. In the risky-with-age group incidence was 4.46% a year and 25% of them caught HIV. However, because 90% of the men were in the lower-risk group, the actual number of infections in the three groups respectively was 35, five, and nine.
Men in the three groups did not differ by educational level, ethnicity (75% were of Dutch origin), or average age.
But they did differ in three respects. The first was age at sexual debut, which was 18.4 in the lower-risk majority, notably younger in the risky-when-young group (15.5) and somewhat older in the risky-with-age group (19.6).
Secondly, the proportion of men in the lower-risk majority with a steady partner when they entered the study was 63%. The risky-when-young group were less likely to have a steady partner when they entered the cohort (39.5%) and the risky-with-age group were much more likely (90%). Interestingly, by the time they left the study, the proportion of men with a steady partner tended much more to cluster around the average (66%, 58% and 65% of the three groups respectively had steady partners by the end).
Thirdly, as we reported above, the one thing the two higher-risk groups had in common was that their use of recreational drugs, erection drugs and poppers was higher throughout the follow-up period than that of the lower-risk majority.
Thirty-seven per cent of the lower risk majority reported using recreational drugs or poppers at their first visit and this figure was unchanged at their last visit. Their use of erection enhancers increased modestly from 18% to 25.5%.
In contrast 70% of the risky-when-young group reported recreational drug use at their first visit and this had only declined to 65% at their last visit despite their sexual risk score falling to no higher than average. In the risky-with-age group, 58% reported recreational drug use, 75% poppers and 35% erection enhancers even at their first visit when their sexual risk score was a median of zero. Drug use in this group had increased to very high levels by the time of their last visit with 85% reporting recreational drug use, 90% poppers and 70% erection enhancers.
As we reported, there was a wide variety of ages in the cohort and therefore number of years since sexual debut. This enabled the researchers to relate risk to age and thereby compute ‘risk careers’ for each group over a much longer timespan than the five-and-a-third years average follow-up.
In the lower-risk majority the sexual risk score rose gradually from 0.3 to 0.6 over a timespan of 50 years since sexual debut. The researchers attribute this to a rise in condomless sex that started as soon as antiretroviral therapy became available.
In contrast the sexual risk score in the ‘early risk’ minority would fall more than tenfold using the same projections from the data, and the risk score in the ‘late risk’ group would rise at least 16-fold, though the age ranges and therefore times since sexual debut that could be covered were smaller.
In the risky-when-young group, the average sexual risk score fell from 3.0 to 0.2 over a period of 25 years. And in the risky-with-age group, the risk rose from near zero to 3.5 over a period of 16 years.
There are limitations to this study, most notably the fact that men dropped out of the study if they became HIV positive, so we don’t know the effect that HIV diagnosis has on sexual risk behaviour. As already noted, it did not record PrEP use or the viral suppression status of HIV-positive partners, though there was no PrEP use in the Netherlands until the last year of this study. Another limitation is that this was a group of gay men with, on average, a high educational level who were motivated enough to attend a twice-yearly behavioural study. And finally only casual sex was examined as a risk factor: sexual behaviour with steady partners was not looked at.
Nonetheless, the study does offer researchers a chance to, as they say, “identify individuals before they increase their risk behaviour”, and also identify people whose risk behaviour might decline over time.
“This is especially important for interventions such as PrEP,” they write, “where life-long use is not desirable and might not be feasible.”