Wayne Johnson's meta-analysis in gay men

There are so many HIV-intervention studies conducted that an attempt to review all systematic reviews, meta-analyses and meta-reviews is beyond the scope of this section. However, Wayne Johnson’s¹ review of HIV-prevention interventions with gay men in the USA is interesting, partly because it is about the group that still has the highest HIV incidence in a number of countries, and partly because it supports a number of Albarracin’s conclusions - but equally interestingly does not support others.

Johnson sifted through 99 HIV-prevention studies conducted in the USA to find nine that specifically targeted gay men (and comments on the fact that this, the highest-incidence group in the USA, only had 10% of studies devoted to it). The date of studies ranged from 1989 to 1998 and Johnson comments that he would have had at least twice the data to work with if he had included later studies. Altogether, 2270 gay men participated in the interventions.

The bottom-line effect of the interventions was a reduction of 26% in instances of unprotected anal sex relative to baseline. Because only 32% of gay men reported unprotected sex before the interventions (a big contrast to Albarracin’s populations, where only 32% did use condoms) this represents an absolute decrease in unprotected sex acts of 8.5%.

Many of Johnson’s findings mirror Albarracin’s, although he divided up his interventions in a slightly different way (with some interventions using a number of methods), as follows:

• Training in interpersonal skills such as safer-sex negotiation, disclosure and communication.
• Training in personal skills such as self-management, decision making and stress management.
• Programmes designed to enhance self-esteem or community pride.
• Programmes designed to increase the social acceptability of condom use by means such as peer-leader endorsement and outreach by peer volunteers (similar to Albarracin’s ‘normative’ studies).
• Programmes designed to enhance responsibility by the use of means such as behaviour contracts and agreements.

Like Albarracin, Johnson found that interventions that included interpersonal-skills training produced the most clearly favourable effects. Behaviour contracts and responsibility agreements produced the weakest effects.

In contrast to Albarracin, Johnson found that the three interventions that took place in the community worked better than ones that involved small-group training, though these worked too. Two of these were similarly designed studies which recruited community ‘opinion leaders’ as disseminators of information and advice about safer sex in gay bars and clubs.

Also in contrast to Albarracin, interventions generally worked better with younger people, and significantly worse with men over 33. Johnson comments that this may also be because older men tended to have baseline higher levels of condom use. It may also be because the majority of interventions that targeted young people in Albarracin’s meta-analysis took place in schools – and it has proved particularly difficult to deliver sex-education programmes of proven efficacy in schools, as witnessed by the failure of a carefully designed programme in Mexican schools.²

One of Johnson’s most important findings was to identify significant ‘antagonisms’ between particular methods. For instance, studies that included personal-skills training and self-esteem boosting were effective as long as they did not also include arguments for the acceptability of condom use or safer sex. Similarly, programmes for young people worked particularly well if they had low baseline levels of condom use, as long as they did not include behavioural contracts. This is valuable preliminary work towards finding out which components of an ideal HIV-prevention programme work well together, and which do not.