A tenofovir-gel vaginal microbicide with the same efficacy as that seen in the CAPRISA 004 trial could be cost-effective in the context of the South African HIV epidemic even if only one-in-four sexually active women used it.
If the gel was available to 25% of sexually active women, the cost per disability-adjusted life year (DALY) saved would be $104. With very high coverage (available to 80% of women), the cost per DALY saved would be $74. This compares unfavourably with the cost per DALY saved by condom social marketing, but is superior to the cost of certain other prevention measures such as the adoption of sterile needles for all injections.
The findings come from a mathematical model by Brian Williams and his team from the World Health Organization, which has also been used to estimate the cost-effectiveness of male circumcision as an HIV prevention method.
Williams and his team assumed in the model that the efficacy of tenofovir gel was 54% for women who used it at least 80% of the time, as observed in CAPRISA 004. They also assumed that the cost per two applications (one before and one after sex) was US60¢, the same as in CAPRISA, though they point out this is likely to fall substantially if the microbicide is mass-marketed, especially as 54¢ of this cost is that of the applicator and wrapping.
They projected HIV prevalence, incidence and related mortality from the beginnings of a national microbicide roll-out in 2012, through to full availability in 2015, and on to 2030, and plotted what was likely to happen given three different availability levels of 25, 50 and 80%, and three different usage levels of 25, 50 and 90% of all acts of intercourse.
Other assumptions included in the model were: that HIV incidence would tend to fall over the next few years anyway (as is being observed for the maturing southern African epidemic); that the average treatment and care cost of a single HIV infection was US$8396: and that the average survival time per untreated HIV infection was eleven years.
Given these assumptions, they found that with 25% coverage the microbicide would avert an average of 500,000 new infections and prevent an average of 290,000 deaths over the next 20 years. However, the 95% confidence intervals in the model’s output are very wide: the actual number of infections averted could be anything between 40,000 and 770,000 and of deaths prevented 20,000 to 440,000.
With 80% availability, the average number of infections averted would be 2.3 million and the number of deaths prevented 1.3 million. With this availability prevalence would fall from 16 to 12% by 2026, annual incidence from 1.65 to 1.35% by 2020, and annual mortality from 1.6 to 1.2% by 2026, if the gel was used in 50% of sex acts by the women who had it. If it was used in 90% of sex acts, prevalence would decline to 8.5%, annual incidence to less than 1%, and annual mortality to 0.8%.
If the gel was only available to 25% of women, prevalence would only fall from 16 to 15%, annual incidence from 1.65 just under 1.6% and annual mortality from 1.6 to 1.45%.
The model does not account for age, and the researchers point out that if microbicides were targeted at young women aged 16 to 30, this would prevent the majority of infections at reduced cost.
Williams BG et al. Epidemiological impact of tenofovir gel on the HIV epidemic in South Africa. JAIDS, early online publication, doi: 10.1097/QAI.0b013e3182253c19, 7 June 2011.