Background In the first wave of a large population-representative study of older adults in rural South Africa, the 'Health and Aging in Africa: A Longitudinal Study of an INDEPTH Community in South Africa (HAALSI)' study, large gaps in knowledge of HIV status were found among participants, coupled with low rates of recent HIV testing and high rates of sexual behaviour that can lead to acquisition and transmission of HIV. In general, older adults should test for HIV at least once per year (HIVAge.org), but few do so, in part because current HIV testing programmes have traditionally focused on young and middle-age adults and HIV testing facilities are often difficult to access for older populations. Only 17% of older adults in the HAALSI community reported having tested for HIV in the past 6 months.
Home testing can provide a powerful approach to bring the health systems 'function' of HIV testing to older adults in rural Africa. In the routine health system, community health workers, such as the South African community care givers, could provide this 'function' on a regular basis. Home rapid HIV testing and HIV self-test kits have been intensively studied in many populations. HIV self-test kits have recently become available as over the- counter products in South Africa, after the South African Pharmacy Council (SAPC) lifted a ban on their sale in pharmacies in December 2016.
To date, however, no study has compared head-to-head different 'forms' of home testing for HIV. In this study, the investigators aim to establish the comparative effectiveness and cost-effectiveness of three different options of home HIV testing among older adults in rural South Africa: (1) home delivery of HIV self-test kits, (2) home HIV rapid testing, and (3) both home delivery of HIV self-test kits and home HIV rapid testing.
Overall, home HIV testing is likely attractive for older adults because it allows testing without having to visit a health care facility. Home delivery of HIV self-testing kits has the additional advantage that it allows HIV testing in a place and at a time of one's own choosing. Home approaches to HIV testing are particularly policy-relevant at this time, because South Africa and many other countries in sub-Saharan Africa are currently expanding in re-designing their community health worker programs, through which home HIV testing could be provided on a routine basis.
Preliminary evidence in this and in other populations have suggested that HIV testing and linkage to HIV treatment and care improve health care utilization for other non-HIV chronic care and reduce risky sexual behavior. The investigators will thus not only assess the effects of the different home HIV testing offers on HIV testing but also assess comparative effects on healthcare utilization, noncommunicable disease endpoints, and sexual behaviors.
Research Aims
This study thus aims to:
1. Establish the comparative effectiveness of three options for home HIV testing for older adults in rural South Africa
2. Identify spill-over effects of the different options for home HIV testing to other adults in the families and communities of the older adults receiving the testing options
3. Establish the comparative cost-effectiveness of the three options for home HIV testing among older adults.
Study Design Participants in Wave 2 of the HAALSI study will be offered one of three options (randomly assigned 1:1:1) for home HIV testing: (1) three oral HIV self-test kits with a brief introduction to HIV self-testing (home HIV self-testing arm), (2) home rapid HIV testing and counselling (home rapid HIV testing arm), and (3) three oral HIV self-test kits with a brief introduction to HIV self-testing plus home rapid HIV testing and counselling (home HIV self-testing and HIV rapid testing arm). The brief introduction to HIV self-testing will include encouragement of both self-use and onward distribution to household and community members.
The HAALSI participants were randomized (sample \~3,600; 1,200 individuals in each of the three arms). Randomization has occurred during routine HAALSI fieldwork visits, using a pre-programmed randomization process that is integrated into the tablet-based HAALSI computer-assisted personal interview (CAPI). This randomization was completed during a three-month period between May and July 2019.
Outcome Assessment The primary outcomes (1-2) and a group of secondary outcomes (3-21) will be assessed by telephone at 9 months after randomization. Based on the timing of randomization, primary data collection will take place from February through May 2020. Then, both the primary and an expanded set of secondary outcomes including several biologically-measured secondary outcomes (1-25) will be assessed at 18-24 months, during the subsequent, planned cohort-based data collection (HAALSI Wave 3).
Power Calculation In the first wave of the HAALSI study, the proportion of people having tested for HIV in the past 6 months (which was about 17%) was measured. Assuming uniform distribution of testing over time, it would be expected that about 3% of older adults in the HAALSI population tested in the past month. Based on the HAALSI 1 data, the proportion of older adults who ever tested for HIV is 65%. Given these estimates for the endpoints in the control arm, this study is powered to detect a 3 percentage point difference in HIV testing in the past month and a 6 percentage point difference in ever testing for HIV across the three arms. For this power calculation the investigators further assumed, 20% loss to follow-up, 80% power and a significance level for rejecting the Null hypothesis of 0.0167. This significance level results from the adjustment of the standard 0.05 significance level for multiple hypothesis testing - each arm compared to each other arm - using the Holm-Bonferroni method.
Data Analyses The investigators will measure risk ratios for all the binary outcomes in this study, using modified Poisson regression implemented in generalized linear models with Poisson distribution and log link. Furthermore, the effect sizes for count outcomes using generalized linear models with negative binomial distribution and log link will be measured; the investigators will measure effect sizes for continuous outcomes using generalized linear models with normal distribution and identity link. All estimations will be adjusted for clustering of outcomes at the household level as well as for baseline values of the endpoints. Baseline values of our primary and secondary endpoints will be assessed during the baseline visit before the HAALSI participants receive the trial exposure.
