Background: Despite the importance of HIV testing for controlling the HIV epidemic, testing rates remain low in many countries. Growing emphasis on HIV treatment-as-prevention will require increases in the frequency and coverage of testing to place individuals on treatment rapidly after diagnosis, and new waves of testing will need to reach populations who have not previously tested. Home based testing is one potentially promising approach for supporting this goal. However, testing rates during home visits can be low, particularly in settings with high HIV prevalence, and there is little existing evidence on the effectiveness of interventions to increase testing rates for home based visits. This study tests the hypothesis that unconditional gift vouchers can be used to increase consent rates for home-based HIV testing.
Methods: This study uses data on 18,478 men and women who participated in the 2009 and 2010 Africa Center HIV surveillance cohorts in rural KwaZulu-Natal, South Africa. In 2010, HIV prevalence was 24%, and 41% consented to test for HIV. All residents aged 15 and older are eligible for the HIV surveillance. A subset of households in the 2010 cohort were provided with a gift in the form of a food voucher worth 50 South African Rand (5 USD) which was not conditional on consenting to test. We compare the change in HIV test consent rates between 2009 and 2010 for residents in households that received a voucher in 2010, with the change in HIV test consent rates for residents in households that did not receive a voucher in 2010. Our approach corrects for unobserved confounding using a quasi-experimental difference-in-differences design.
Results: Allocation of the voucher to a household in 2010 increased the probability of household members consenting to test in 2010 by 29 percentage points (risk difference 95% CI 23 – 35; p<.001). We also find an attenuated effect of treatment on consent rates in 2011. Rates of consent to test by intervention group (being a member of a household which received the voucher in 2010) are shown in figure 1. We estimate the cost of the program at 7 USD per additional HIV test obtained.
Conclusions: The provision of gift vouchers to surveillance participants is likely to be a cost-effective tool for raising consent rates for home-based HIV testing. In addition, by persuading respondents who would ordinarily refuse to test in the absence of the voucher effect, this approach can potentially be used to assess the extent of selection bias in the population.
Figure 1: The intervention group is defined as being a member of a household which received the gift voucher in 2010, while the control group is defined as being a member of a household which did not receive the voucher in 2010. Mean consent to test is the number of residents in each group who consented to take a HIV test, divided by the number of residents who were eligible for participation and were successfully contacted by the surveillance surveyors in each group.