This is the first study to report acceptability of home based HIV counselling and testing in rural South Africa. The intervention achieved a high uptake (75.0%) of home-based counselling and testing in a remote rural area. This is encouraging given that the baseline rate of testing in this area was only 32.0%
 and that HIV-related stigma in this community is anecdotally high. We surmise that the intervention was successful because of the extensive mobilisation strategies used to introduce the project to local leaders and communities, high quality training which ensured that counsellors were confident offering their services, and community members’ trust in the counsellors’ skills and confidentiality.
The migratory patterns of this setting
 may partially explain why we primarily met females in the home, and correspondingly, why they comprise the majority of those who tested. Furthermore, those with a marital partner largely tested alone, presumably because their partners were away. These circumstances may have critical implications. For example, females testing in the absence of male partners or authority figures may lead to potential negative social consequences
 and could also impact disclosure, prevention, and care-seeking behaviour for those who test HIV-positive. While these issues were not formally studied within the context of this intervention, we did not receive any adverse reports following testing. However, further research is encouraged to inform future implementation and scale-up.
The gender stratified levels of uptake in our context are highly consistent with findings from other sub-Saharan African settings
[15, 17, 21]. Although we reached a limited number of men overall, it is very encouraging that 70.0% of the males met did accept testing. Considering that only 17.0% of adult men had ever tested at baseline
, it is unlikely that such a large proportion would have sought clinic-based services in the absence of HBHCT. Strategies to better reach men who are present in the community may include offering testing outside conventional working hours and during holiday periods, and employing male counsellors.
We found a relatively low level of sero-discordance among couples testing together. Though this is in contrast to findings from studies involving household members of TB or HIV-positive clients
[16, 17, 27], it is within the range commonly found among general populations
[28–30]. In our setting, couple testing was optional and it is possible there may have been self-selection of those with less risk. Furthermore, relatively few clients with a partner actually tested together, and only 7.5% of our HIV-positive clients tested with a partner. Thus our rates of discordance for this area may be underestimated.
The intervention reached a broad range of age groups. The majority of testers were in the reproductive age group of 14–49 years. This is promising since this population is generally the most sexually active and at risk of HIV acquisition. However, since high HIV prevalence and mortality are reported among older adults in sub-Saharan Africa
[31, 32], it is also important that our intervention reached this population. Additionally, in this traditional community the elderly serve as role models, and their participation may have encouraged others to do so as well.
It was also encouraging that the intervention reached a large proportion of people who had never previously tested or who had not tested recently, suggesting that increased access does lead to increased utilization, and that the HBHCT model may have mitigated other common barriers such as fear of stigma, lack of transport, or financial constraints. HBHCT also successfully identified a substantial proportion of previously undiagnosed HIV-positive clients, an important first step in ensuring timely access to care and treatment.
Our overall HIV prevalence of 9.5% is lower than expected when compared to the provincial HIV prevalence of 25.8% among those aged15-49 years in Kwazulu-Natal
. This may be explained in part by the fact that we tested a substantial number of clients who were well beyond the conventionally reported reproductive age range and that many clients who declined testing reported being HIV-positive. To test these theories, we recalculated HIV-prevalence for those aged 15–49 years, including refusals and classifying those who self-reported an HIV-positive status as HIV-positive, and classifying all others as HIV-negative. This resulted in an HIV prevalence of 14.3%, which is still well below the provincial rate. However, this may be explained in part by the fact that some of those who refused testing for other reasons or who were not met in the home, could have been HIV-positive. Further, it could be the case that this particular rural sub-community is less heavily affected by the epidemic than others in the province.
Males and females who were reached by the HBHCT intervention had distinctly different profiles. Compared to males who tested, a greater proportion of females who tested were from smaller households, older, married or widowed, HIV-positive, and had tested previously. These populations are likely to have distinctly different needs and concerns. For example older widowed women may need greater support related to acceptance of status while young single men may need more information about practising safer sex.
Very few people refused HBHCT because they preferred a clinical setting. This offers strong evidence of the HBHCT model’s acceptability in this community. Reasons for refusal varied significantly by age and gender. While these particular differences may be specific to our setting, they underscore the importance of better understanding target clients, and the potential need to tailor intervention messages and counselling for different sub-groups.
As previously stated, a sizeable proportion of non-testers reported their reasons for refusing as either being HIV-positive (18.5%) or already knowing their HIV status – negative or undisclosed (22.6%). The former fall out of the target group, and it is possible that even some clients in the latter group were HIV-positive but chose not to disclose their status. Combining these two groups, it means that 41.1% of non-testers were aware of their HIV status. This is comparable to the proportion of testers who reported ever having taken an HIV test prior to HBHCT (42.6%). However, amongst that group, over half (58.3%) reported having tested over a year ago. If the same applies for non-testers, knowing one’s status may not necessarily be a valid reason for refusal, particularly since the majority of those claiming to be aware of their status were in the 14–49 age group (80.3%) and likely to be sexually active. It is also possible that clients may have offered this reason to mask their true concerns or fears about testing. This implies that counsellors may need to give more attention to assessing clients’ reported reasons for refusal and further encouraging testing when appropriate.
From the sample of clients who refused testing, those who merit particular attention are the 34.1% who reported feeling scared or unready for a test. It is possible that their anxieties may stem from self-knowledge of risk behaviour, fear of the testing outcomes, stigma, or other reasons that were not easily voiced. To better motivate such clients it is important that we unpack and address the specific reasons behind their apprehension. The other group warranting closer attention is those who do not feel at risk of HIV infection. In our setting, this was most commonly reported by those in the 50–64 and 65+ age groups. As mentioned previously, given that HIV-prevalence and mortality are high among older age groups in sub-Saharan Africa
[31, 32], this may be a false perception that future interventions should address.
The findings of our study must be interpreted in light of important limitations. First, we do not have comparable background information on non-testers, which limited our ability to examine personal characteristics that are predictive of HBHCT uptake. Second, we did not ask people why they tested, and thus may have overlooked “hidden” influential factors. Third, we did not explore whether the HBHCT model might have led to unintended social consequences – for instance, clients feeling pressured to test or to disclose their status to those who also tested within the home, or women feeling un-empowered to share their results with partners not present at the time of testing. These potential outcomes are important and must be further studied. Fourth, since our study participants primarily included females, our findings may not be reflective of a more gender-balanced population. Fifth, we acknowledge that the provision of a t-shirt for people who agreed to test could have led some people to accept testing who otherwise might not have; however it is doubtful that a t-shirt alone would have been an influential factor for the majority of participants. Finally, the study was only conducted in one rural sub-district among those aged 14+ years, thus the findings may only be generalized to similar settings and adult populations.
This study also has several strengths. First, it involves a relatively large sample size. Second, it was conducted in an area with poor infrastructure and constraints typical of many rural areas in Africa, thus providing insight about the operational issues and challenges that may arise when scaling up in similar settings. Third, for the first time, we provide insight about the reasons for refusal of home-based testing, which may be as important as predictive socio-demographic background characteristics more commonly reported by other studies. Further, the gender and age-specific analyses offer an important understanding of client differences and their implications for programming.