The role of HIV testing in HIV prevention

It might seem strange to propose HIV testing as a prevention method. After all, it could be argued that the role of HIV testing is to find out whether prevention has failed.

This, however, is to ignore the contribution people with HIV can make to HIV prevention.

Quite apart from the benefits of early testing for the individual in terms of being able to take treatment before significant immune damage has occurred, there is abundant evidence that, once diagnosed, people with HIV reduce their risk behaviour. Maximising the uptake of HIV testing and detecting early infections should therefore, in theory, lead to a significant decline in HIV infections.

It has been known since the early 1990s that people diagnosed with HIV modify their risk behaviour. Initially this fact tended to be ignored by researchers who were more interested in people with HIV who continued high-risk behaviour, in studies whose primary aim was to quantify that figure so that prevention programmes could be directed at these people.

But it was eventually realised that the majority of people with HIV tended to reduce their risk behaviour and that because of this, ensuring that as many people knew their status as possible could be an effective prevention method.

To take a recent example: a study of recently-diagnosed gay men in the USA (Gorbach 2006) interviewed 153 men an average of 4-6 weeks after diagnosis and then again three months later. It found that between the first and second interviews nearly half (47%) of the 91% of men who remained sexually active, reported that they had reduced the number of sexual partners they had had in the previous three months, a third reported no change and a fifth reported that they had increased their number of partners. The average number of partners within three months went down from 7.9 to 5.2, a 34% decrease.

Nonetheless, 54% of the newly-diagnosed HIV positive men had unprotected anal intercourse (UAI) during the three months before the second interview. In the past this might have been regarded as a failure of prevention methods. However, the researchers found that, of the men who did have UAI, the majority switched to HIV positive partners if they had unprotected sex. In other words, there was considerable evidence of serosorting.

At the baseline interview the men reported that nearly half (48%) of their UAI partners had been HIV negative, a third of unknown status and less than a fifth (18%) HIV positive.

Three months later the proportion of UAI partners who were negative had declined to 30% of the total (a 37% decline). The proportion that was positive had nearly tripled to 52% of the total (a 188% increase). And the proportion of unknown status had halved, to 18%. This is strong evidence both for disclosure (since fewer partners were of unknown status) and for serosorting.

This evidence of behaviour change is nothing new. In a 1994 survey (Wenger 1994) half of a group of HIV positive men had UAI with HIV positive partners but only a sixth with HIV negative ones – exactly the same proportions as in the 2006 survey. The difference was that in this survey 41% had UAI with partners of unknown status, but by 2006 this proportion had more than halved. Although obviously we are talking about two different groups of gay men who cannot be compared directly, this seems to provide some evidence of both increased knowledge of status and increased willingness to talk and ask about it.

The Centers for Disease Control (CDC) in the USA(CDC 2003) has calculated that on average people with HIV, once diagnosed, reduce their risk behaviours (i.e. potentially serodiscordant unprotected sex and needle-sharing) by 70% a year after diagnosis. A meta-analysis of eleven US studies found that there was a 53% reduction in unprotected sex post-diagnosis, and a 78% reduction in unsafe (i.e. unprotected and serodiscordant) sex after diagnosis.

Some of the reduction in risky sex may be ascribed to ‘post-diagnosis shock’: many people, following an HIV diagnosis, may re-evaluate and try to change the risk behaviour that led to them acquiring HIV; the diagnosis may have deleterious effects on their self-image and libido; and they may experience a strong determination not to infect others and subject them to the same shock. Does the reduction in unprotected and unsafe sex persist, however?  

Few studies have followed the sexual risk behaviour of people diagnosed with HIV for more than a year post-diagnosis. One that did (McClelland 2006) was a seven-year prospective study (1993-2000) of 1600 high-risk women in Tanzania, all of them initially HIV negative. The average length of follow-up was more than five years in this study. During the study 265 women seroconverted, an HIV incidence of 7.7% a year. The average post-diagnosis follow-up time was 3.8 years.

There was a 44% reduction in incidents of unsafe sex amongst the women post-diagnosis, which was maintained over time. However sex in general and unsafe sex tended to decline as the women got older anyway. Adjusting for age, the researchers found a long-lasting decline in unsafe sex of 31% post-diagnosis, independent of age.

The proportion of women who maintained 100% condom use increased from 59% to 67% post-diagnosis. The proportion who had more than one sexual partner in the previous week declined from 20% to 9% and the proportion who had more than two sexual encounters in the previous week declined from 27% to 16%. All these differences were maintained over time and were independent of age.

The difference HIV diagnosis may make could be even larger than this. The CDC used this figure and what it knew or suspected about the relative infectiousness of people with HIV and different disease stages to calculate that as many as a two-thirds of HIV infections could be transmitted by people unaware of their HIV status (CDC 2003). This takes into account not only the behaviour change in people with HIV but the fact that because a proportion will be on treatment, their average infectiousness will decrease.

In 2004 Holtgrave and Anderson(Holtgrave 2004) used these figures to claim that more than one in nine of people in the USA who do not know their HIV status pass on their virus in a year compared with only one in 58 of people who do know their status – over six times as many.

This depends on assumptions about the proportion of infections passed on by people in primary infection. Phylogenetic studies in various locations (Hughes 2008; Brenner 2008; Yerly 2008) have estimated that about 50% of HIV infections are passed on by people who have been infected for less than a year. Frequent testing in high-risk groups could pick up more recently-infected people, the majority of whom would hopefully reduce their sexual risk behaviour post-diagnosis.