What
figures we have for HIV incidence follow this trend: there was a sharp fall in
incidence in many populations after the first phase of the epidemic; it
increased somewhat, in step with the introduction of antiretroviral combination
therapy, in the late 1990s, but amongst most groups now appears to have reached
a steady state.
HIV
incidence increased among gay men in many areas. For instance, according to
UNAIDS, HIV incidence (new diagnoses) in gay men increased from 0.6% a year in
the late 1990s in Vancouver to 3.7% a year in 2000; from 1.16% in 1996 in
Madrid to 2.16% in 2000.
In
other areas, however, an increase in incidence was not noted. In the UK,
for instance, annual HIV incidence amongst gay men attending STI clinics was
3.0% in 1995 and 2.45% in 2001. Within London it
remained static at about 3% a year (3.02% in 1995 and 3.06% in 2001); outside London, it actually
declined from 2.96% in 1995 to 0.97% in 2001, though this was not statistically
significant.19
Similarly,
in the San Francisco study cited above,15
despite an 87.5% increase in unsafe sex amongst gay men, annual HIV incidence
amongst men seeking care at the STD clinic appeared to be declining: from 7.3%
in 1994 to 4.7% in 1999, though this was not significant (p=0.26). At a
non-clinical anonymous community testing site it increased from 2.2% in 1996 to
4.2% in 1999 and this was significant (p=0.015), but more recent survey20
HIV incidence showed signs of a community-wide decline (see HIV treatment as preventionfor more on
this).
There
are multifactorial reasons why a decrease in condom use may not always lead to
an increase in HIV incidence. Firstly, HIV incidence is very difficult to
measure unless large populations are surveyed and sensitive incidence assays
are used. An observed increase in the proportion of people who are diagnosed in
early infection, for instance, may be due to more people coming forward for
testing and testing more frequently, so that HIV infections are detected
sooner.
Secondly,
as we point out in Serosorting, sexual harm reduction and disclosure, an increase in unprotected sex is
not necessarily an increase in unsafe sex, if people are preferentially
choosing people of their own status as unprotected-sex partners. Even in the
Dodds study cited about from the late 1990s,14
although the proportion of gay men who had unprotected sex increased from 33 to
46% between 1996 and 1999, the proportion of those who had unprotected anal sex
with people of opposite or unknown HIV status – the true measure of unsafe sex
– did not increase, and the proportion who (by self report) only had it with
men of their own HIV status remained static at 38 to 42% of all the men who had
unprotected sex.
Thirdly,
even if there is an increase in the community prevalence of HIV, as long as
there is also an increase in the proportion of people with HIV in that
community who are on treatment and have an undetectable viral load, HIV
incidence will not increase and may even start to decline.
In
is important to add, however, that HIV incidence can start to increase in real terms in high-risk communities
whenever any of these factors, which serve to hold back incidence despite
increasing rates of unprotected sex, starts to have a weaker influence.
Recently,
for instance, rising HIV incidence has been noted in young gay men in Amsterdam.21
The
Amsterdam Cohort Study is a longitudinal cohort study of HIV infection among a
group of initially HIV-negative gay men, recruited at regular intervals ever
since 1984, at an average age of 29, and tested for HIV every six months
thereafter. For this study, the average length of follow-up was six years.
Out
of 1627 men included in this study, 215 acquired HIV during follow-up. Incidence
was calculated by assuming that the time of HIV infection was midway between a
positive test result and the previous negative test. HIV incidence was
initially 7.4% a year in 1985 but fell to 1.3% by 1990 and stayed at that level
until around 1995-97. Since then it has slowly increased to 2% a year, though
the increase does not pass the test of statistical significance (p = 0.1).
However,
among gay men aged 30 and under, annual HIV incidence, which had been 0.9% in
1997, increased to 3.8% by July 2009. This result was due to a sudden doubling
in the level in 2009 (it had been about 2% in 2004-8) and the increase observed
is statistically significant (p = <0.01).
Sexual
risk behaviour mirrored the incidence pattern, with an increase in the
proportion of men reporting unprotected sex with casual partners from 12% in
1992 to 30% in 2008.
The
strongest factors associated with becoming HIV positive were unprotected sex
with casual partners (relative risk [RR] 4.74), having over five partners (RR
2.5), having gonorrhoea (RR 5.76), and having had no tertiary education (RR
2.11). It was estimated that three-quarters of new infections were due to sex
with casual partners and only a quarter in steady partners.
However,
among older gay men, unprotected sex with steady partners was more significant,
and three times as many infections were acquired from primary partners in
over-50 year olds between 2003 and 2008 than in the late 1980s. The incidence
pattern in gay men aged 25 and gay men aged 50 were mirror images of each
other: for younger gay men, the year 1995 featured the maximum risk of
infection by a steady partner, but by 2009 this risk had declined considerably.
Conversely, in 50-year-olds the maximum risk from a steady partner was reached
in 2005.
The researcher commented that the Netherlands
had a lower HIV testing rate than many other developed countries: although 88%
of sexual health clinic attendees test for HIV, but rates are much lower in gay
men who do not attend sexual health clinics.