Objective: To explore the risk of a future rise of HIV prevalence in p
opulations of injecting drug users (IDU) with low HIV prevalence but c
ontinuing risk behaviour, and to study the potential influence of prev
ention measures on HIV incidence. Methods: A stochastic simulation mod
el was used to describe a network of longterm buddy relationships in a
population of IDU. HIV transmission took place when borrowing injecti
ng equipment from an infected buddy or stranger. The probability of tr
ansmission depended on the duration of infection. Individuals remained
in the population on average for 10 years. Two surveys amongst IDU in
The Netherlands containing information about risk behaviour were used
to estimate model parameters. We investigated the effect of different
prevention strategies. Results: Below a threshold sharing frequency t
he epidemic never takes off; above the threshold there is a large stoc
hastic variation in prevalence. After reduction of risk behaviour, HIV
prevalence decreases very slowly. Reducing sharing with strangers is
more effective than reducing the overall sharing frequency. Prevention
focused on new IDU greatly reduces HIV incidence. Reduction of sharin
g frequency in HIV-positive IDU has no significant influence on HIV in
cidence at HIV testing rates of 10 and 50% per year, if infectivity is
highest during primary infection. Conclusions: A stabilization of HIV
prevalence does not exclude the possibility of a future rise. Predict
ions about the future course of an epidemic are inherently uncertain.
The effect of prevention programmes on HIV prevalence only becomes vis
ible on a long time-scale. Social networks of IDU play an important ro
le in transmission dynamics and success of prevention. (C) 1998 Lippin
cott-Raven Publishers.