Background:  We examined the effect of partially effective HIV vaccines, and subsequent changes in risk behavior, in a population at high risk for heterosexually transmitted HIV.

Methods:  We developed a mathematical model to simulate the effect of various low-efficacy preventive HIV vaccination scenarios in Soweto, South Africa. We calculated annual rates of movement between population groups defined by gender, disease stage (HIV^­, HIV⁺, AIDS, and death) and vaccination status. Input data included:  a sexually active, ART-naïve population of 820,000 men and women; initial HIV prevalence of 30%; vaccination coverage of 75%; exclusive male negotiation of condom use; and varying contact rates and per-partner infectivity rates, depending upon disease stage. We explored changes in vaccine efficacy and post-vaccination condom use by assessing HIV prevalence and total HIV infections prevented over a 10-year period.

Results:  With no change in risk behavior, a 40% effective HIV vaccine would avert 83,000 infections in Soweto and reduce future HIV prevalence from 33% to 23%. A 30% and 20% effective vaccine would avert 64,000 and 43,000 infections and reduce HIV prevalence to 25% and 28%, respectively. Changes in risk behavior would have a significant influence on the benefits of these vaccination programs. With a 40% effective vaccine, a 25% increase in condom use among vaccinated individuals would instead avert 104,000 infections and reduce the HIV prevalence to 20%. However, if condom use decreased by 25%, a 40% effective HIV vaccine would avert only 61,000 infections and reduce the HIV prevalence to 25%. For simulations that used vaccines with <43% efficacy and caused any decrease in condom use post-vaccination, some combinations of vaccine efficacy and risk behavior change could even worsen the epidemic. A 30% effective HIV vaccine with a resulting 75% decrease in condom use, for example, would cause an additional 20,000 infections and increase the HIV prevalence to 36%.

Conclusions:  Even modestly effective HIV vaccines can confer enormous benefits in terms of HIV infections averted and decreased HIV prevalence. However, these findings are sensitive to assumptions regarding the influence of vaccination on subsequent risk behavior. For South Africa and other countries with similar epidemic profiles, programs to reduce risk behavior may be important components of successful vaccination programs.