HIV treatment optimism and high-risk sexual behaviour among gay men

Stolte et al. [1] examined the relationship between HIV treatment optimism and high-risk sexual behaviour in a cohort of 146 gay men in Amsterdam whom they interviewed every 6 months for 2.5 years. The strength of their investigation is that they were able to monitor changes in sexual behaviour at an individual level in relation to beliefs about new treatments recorded before the change occurred. Consequently, they were in a position to consider cause and effect. They found that most men in the cohort were quite realistic about highly active antiretroviral therapy (HAART). Only a minority were optimistic, as reported in a number of other studies [2–6]. However, perceiving HIV or AIDS to be less of a threat in the light of new treatments predicted a change to high-risk behaviour over the following 6 months. The associated odds ratio was 1.60 [95% confidence interval (CI) 1.16, 2.22]. The authors wrote ‘although causality is difficult to establish even with longitudinal data, the findings in this study are supportive of the hypotheses that perceiving less HIV/AIDS threat since HAART is a cause of the change to high risk behaviour'. They concluded that the decreased threat of HIV/AIDS since the advent of HAART explains at least part of the increase in risk behaviour and sexually transmitted diseases seen at the population level among gay men since HAART became available. Their paper marks an important advance. The odds ratio derived from this longitudinal study allows us to estimate, at a population level, how much of the increase in high-risk behaviour can be attributed to HIV treatment optimism. We can do this using standard epidemiological methods that estimate the proportion of the total population risk that is attributable to a given factor (known as the ‘population attributable risk') [7]. The population attributable risk (AR) is a function of both the proportion of the population with the factor as well as the relative risk (or odds ratio) associated with that factor. The population AR is calculated using the formula: AR = [p (RR − 1)]/[p (RR − 1) + 1], where p equals the proportion of the population with the factor and RR is the relative risk associated with the factor. In the Amsterdam study, the authors gave median scores for optimism rather than percentages. However, inspection of the median scores suggests that between 15 and 25% of men in their study could be classified as optimistic (i.e. they scored 5, 6 or 7 on the ‘perceiving less HIV/AIDS threat’ scale) as has been reported elsewhere [2–6]. In the Amsterdam cohort, optimistic men were 1.6 times more likely to switch to high-risk sexual behaviour than other men. Using these estimates for p (15%) and the RR (1.6), the population AR works out to be 8% [0.15 (1.6 − 1)/0.15 (1.6 − 1) + 1] with a 95% CI of 2–15%. In other words, 8% of the change to high-risk behaviour in the Amsterdam cohort can be explained by HIV treatment optimism when 15% of the men are optimistic. The remaining 92% of the change can not be explained in this way. Increasing the proportion who are optimistic from 15 to 25% increases the population AR from 8 to 13% (95% CI 4–23%). Nearly 90% of the change in risk still remains unexplained. The longitudinal study by Stolte et al. [1] demonstrates that, even when a causal association exists, the contribution of HIV treatment optimism to the overall increase in high-risk sexual behaviour remains extremely modest at a population level. This is not entirely surprising. Numerous studies have demonstrated that only a minority of gay men are optimistic in the light of HAART [2–6]. This low proportion exerts a strong influence on the magnitude of the population AR [7]. It appears, therefore, that the findings from the Amsterdam cohort study do not differ substantially from those recently reported in London [8], Glasgow [9] and Sydney (P. Rawthorne, personal communication) based on behavioural surveillance time series data. Namely, that at a population level, HIV optimism is unlikely to explain the recent increase in high-risk sexual behaviour among gay men. Regardless of this debate, we all agree that priority should now be given to research that helps us better understand the factors that underlie high-risk sexual behaviour among gay and bisexual men at risk of HIV infection.