Maximum impact of HIV prevention measures targeted at injecting drug users.

OBJECTIVE: To examine whether the decrease in HIV incidence and injecting risk behaviours is ongoing in Amsterdam, and to study the determinants of injecting risk behaviours.
DESIGN: Prospective open cohort study (1986-1997) amongst injecting drug users (IDU).
METHODS: HIV incidence was studied using Poisson regression analysis. Trends in injecting risk behaviours were determined using 6645 visits of 879 participants, stratified by HIV serostatus and number of the follow-up visit. Generalized estimating equations were used to account for repeated measurements, and additive model structures were used.
RESULTS: A large initial risk reduction (1986-1991) occurred concerning borrowing and lending of used needles, multiple needle use and frequent injecting. However, except for frequent injecting, the rate of behaviour change significantly decreased, and from 1991-1993 onwards there was no substantial further risk reduction. HIV incidence followed a similar pattern. Injecting risk was much lower at follow-up visits. The following determinants of borrowing and lending among both HIV-positive and negative IDU were largely comparable: young age, non-injecting heroin and cocaine use, frequent tranquillizer use, multiple needle use, injecting with others, and irregular use of needle-exchange programmes.
CONCLUSIONS: The observed residual risk, given the extensive HIV prevention measures in Amsterdam, indicates that further prevention is difficult. Because this risk was associated with an HIV incidence of 3-4% per year, HIV prevalence is likely to be high for many years. Therefore, prevention measures should be maintained because of the possibility of behavioural relapse associated with recent availability of potent HIV treatments. To prevent an HIV epidemic amongst a new generation of drug users, prevention of injecting itself is warranted.