Abstract
The effectiveness of population-level intervention for HIV elimination is influenced by individual-level variation in sexual behaviour. We assess within-person changes in the frequency of condomless anal sex with two or more partners (CLS2+), estimate the transition probabilities and examine the predictors of transitions among a prospective cohort of HIV-negative gay, bisexual, and other men who have sex with men (GBMSM). Participants were recruited through one of three sexual health clinics in London and Brighton (July 2013 to April 2016) and self-completed a baseline paper questionnaire in the clinic. During follow-up, they were invited to complete four-monthly questionnaires twice a year and subsequent annual online questionnaires once a year (March 2015 to March 2018). We used Markov chain models to estimate transition probabilities from 'higher-risk' (CLS2+) to 'lower-risk' (no CLS2+) and vice versa, and to assess factors associated with transitions between different sexual risk levels. Among 1,162 men enrolled in the study, 622 (53.5%) completed at least one online questionnaire. Higher-risk behaviour was reported in 376/622 (60.4%) men during online follow-up. Overall, 1,665/3,277 (37.5%) baseline and follow-up questionnaires reported higher-risk behaviour. More than 60% of men (376/622) reported higher-risk behaviour at least one period during the follow-up, while 39.5% of men (246/622) never reported CLS2+ during the follow-up. In the next four months, the estimated probability of continuing higher-risk behaviour among men who reported higher-risk behaviour was 78%. Calendar time, recent HIV tests, PrEP and PEP use were the predictors of staying in higher-risk behaviour, while less stable housing status was associated with switching to lower-risk behaviour. Among men who reported lower-risk behaviour, the probability of engaging in the same behaviour was 88%. Recent HIV tests, PrEP and PEP use, recreational drugs, chemsex-associated drug and injection drugs, and bacterial STIs diagnosis were the predictors of switching to higher-risk behaviour. Our results indicate that at any one point in time, the majority of GBMSM are at low risk for HIV acquisition, although many experience short periods in which they are at higher risk. Markers of transitions can be utilized to identify which GBMSM are likely to increase or decrease their risk, thus helping the timing of HIV prevention interventions.