To our knowledge, this is the first study to assess the feasibility and acceptability of home-based HIV testing in a refugee settlement in sub-Saharan Africa. We found that by visiting homes up to three times, we encountered 90% of eligible adults noted to be living in those households, confirming this approach is feasible. Furthermore, 75% of eligible adults encountered participated in HIV testing and received their results, reflecting the acceptability of home-based HIV testing among this population (Table 1). This study demonstrates that home-based HIV testing in Nakivale Refugee Settlement is possible and is a strategy that should be considered for other refugee settlements in Uganda and in nearby countries in sub-Saharan Africa.
There were numerous reasons to suggest that going to homes and encountering individuals at home might be a challenge in this context. Nakivale is geographically large, covering 71-mile2 [26], and we were not certain if our study team would successfully reach all the households in these remote villages while carrying HIV testing supplies. However, by allowing for 1–2 h of driving in the morning and late afternoon and equipping each team of research assistants with backpacks with all necessary research and HIV testing supplies (i.e. electronic tablet, gloves, HIV test kits, cleaning supplies), our teams moved about with ease. Further, though the refugee settlement is expansive and refugees are initially given large plots of land, the dwellings in each village are close to one another, often with only approximately 3 m of land separating homes. This is because many refugees choose to live near small village trading centers rather than on their land; they frequently travel 1–2 h walking to cultivate their land allotment. The layout of the villages, with homes mostly lined up along both sides of a main street, made it possible for research staff to move quickly between homes and to draw maps of each village that enumerated households to facilitate multiple visits when necessary. By returning to homes up to three times, we encountered most adults living in each household.
Aware that many refugees in Nakivale previously experienced violent conflict and sometimes gender-based violence as a weapon of war [36,37,38,39], we had concern that research staff might not be welcomed into the villages or into homes in the refugee settlement. Further, given the proximity of neighboring homes, we thought that fears regarding potential lack of confidentiality of HIV test results might compound this problem. This was not the case. As we moved door-to-door, people living in the village volunteered information about how many people lived in each home, how old their neighbors were, and which homes were abandoned. This acceptance of our research staff in the villages may have been facilitated by the demographics of our research staff (individuals from the DRC, Rwanda, and Burundi), all refugees from similar countries of origin as those in the villages and able to communicate in the native language of many of our participants. Many of the adults who we did not encounter were reported to be far away for an extended period, working elsewhere in Uganda.
We identified differences comparing home-based HIV testers to clinic-based testers living in Nakivale, some of which are potentially explained by the structure of the refugee settlement. Home-based testers were older, with a difference in median age of 2 years. It may be that younger individuals move from the village to live closer to “basecamp”, the social and economic center of Nakivale. In basecamp there are informal stores based out of windows of homes (including electronic stores for charging cellular phones), small motels providing room and board for travelers, merchants with tarps spread on the ground selling goods (i.e. jerrycans, washing bowls, shoes, and maize), and a small selection of restaurants and bars. Similarly, Ugandan nationals living in Nakivale likely choose to live closer to the center of the settlement where commerce is most active and where Nakivale Health Center is located. It is likely for this reason that home-based testing had a higher proportion of refugee vs. Ugandan national participants compared to clinic-based testers (93% vs. 79% respectively). It may also be that refugees are less willing to access clinic-based services such as HIV testing, which are provided mostly by Uganda national staff who are often not facile in their native language. Not surprisingly, home-based testers were more likely to live further from a health clinic. This may be because of the remote location of the 3 villages where we conducted the pilot, though there are 4 health clinics distributed around Nakivale. More likely, given the scarcity and cost of transportation in Nakivale, clinic-based HIV testers simply lived closer to and more often accessed clinic.
Home-based testers lived in Nakivale longer than clinic-based testers, with more than half reporting living in Nakivale ≥5 years (62% vs. 41%), and a minority living in Nakivale for < 1 year (2% vs. 12%). This likely reflects how villages are established in Nakivale. As refugees are transported from transition camps elsewhere in Uganda, they are placed in new villages mostly with others from the same country of origin. By sampling villages further from basecamp, we may have unknowingly selected participants who had spent more time in Nakivale rather than new arrivals that could have been settled in villages closer to basecamp. It also may be that those with less time in Nakivale have ailments that bring them to clinic more frequently or that those who arrived more recently must depend more on clinic services than well-established refugees who have a larger network of family and friends on which to rely.
More home-based testers reported they had no schooling (25% vs. 21%) and less reported attending some secondary school and above (11% vs. 20%). This may be because of the limited number of primary schools scattered around the settlement and the reality that many children living in the villages are expected to help with essential chores such as cooking and farming [6]. Additionally, there is only one secondary school in Nakivale, which is located 1–3 h walking distance from the villages where our pilot took place. More likely, however, refugees did not attend school in their country of origin because of war and unrest. Additionally, individuals with more education probably sought medical care at a health center more often. Despite less schooling, home-based testers were more likely to achieve ≥75% correct answers on a 4 question HIV knowledge test (70% vs 60%), perhaps reflecting the success of prior HIV prevention campaigns in the settlement.
Even when encountered at home, it was uncertain if home-based HIV testing would be acceptable to this unique population living in the refugee settlement. This was particularly concerning given the reality that some refugees have experienced potentially traumatic events and may have posttraumatic stress disorder (PTSD) [40,41,42]. However, 75% of individuals encountered accepted the home-based HIV testing intervention, with the large majority (95% of participants tested) doing so during the first or second visit. Many even anticipated our visit to their home and planned to be present, telling us they usually leave to work their land during the day but instead stayed home when they heard about the home-based HIV testing services from others in the village. Given the low proportion of remaining eligible individuals willing to test, and the yield of HIV-positive results declining with each subsequent visit, limiting home-based HIV testing interventions to two home visits appears to be sufficient in this setting. While we originally thought HIV testing would be more acceptable to individuals at home alone as they would have more privacy, the odds of participating in HIV testing instead increased with each additional person home. It may be that refugees sought support from their family or permission from their partner prior to testing, and therefore could proceed with testing when those additional people were present. A campaign to notify individuals of upcoming home-based testing in their village and to invite them to be home at a specific time may facilitate uptake of HIV testing [43, 44]. Though we intentionally conducted repeat visits at different times of day, we did not find a time of day when more people tested for HIV.
While home-based testing in the refugee settlement was acceptable for most, there may be subsets of the population that found testing at home less acceptable. Other than sex, the characteristics of those who were eligible but did not test were not identified and therefore cannot be evaluated. In Village 3, comprised mostly of Rwandans, only 70% of individuals encountered where willing to test for HIV and no individuals tested HIV-positive. This was unusual considering 17% (56/330) of those found to be HIV-positive in our clinic-based HIV testing study in Nakivale were Rwandan [7]. It may be that higher risk individuals declined participation or avoided being encountered at home.
Given financial constraints, and the need to prioritize health expenditures to maximize benefits, consideration of the cost-effectiveness of various home-based HIV testing strategies in humanitarian settings is crucial. A cost-effectiveness analysis comparing home-based and facility-based testing strategies in rural South Africa found that home-based testing led to higher uptake and was less costly per client tested [45]. While knowledge of HIV status is important in HIV prevention, with resource constraints it may be more useful to evaluate HIV positivity. Though HIV-positive test results were not significantly different between home-based and clinic-based testers in Nakivale (1.9 vs 3.4%, p = 0.27), it is intuitive that HIV testing outside of the clinic may have a lower diagnostic yield. However, a cost-effectiveness study in Uganda comparing home-based and facility-based HIV testing which measured effectiveness as number of HIV sero-positive clients identified found that home-based testing was the least costly strategy both for the number of clients tested and for the number of positive clients identified [46]. Finally, another study in Uganda compared four HIV testing strategies (stand-alone, hospital-based, household-member, and door-to-door), and found home-based HIV testing strategies reach populations with low rates of prior testing and identified people with HIV with higher CD4 cell counts [9]. Stakeholders caring for humanitarian-affected populations will need clear goals to select appropriate HIV testing strategies for their target population. Multiple testing strategies may be needed to meet UNAIDS goals, improve the health of individuals, and curb transmission of disease in this unique setting.
This study should be viewed in the context of certain limitations. There may be an increased risk of coercion to test or to disclose one’s status to family members or others in the household with home-based testing. Despite this possibility, our study team observed the opposite. Individuals may be less likely to link to HIV clinical care following home-based testing [20, 47], as newly diagnosed individuals cannot easily be personally escorted to HIV clinic for immediate linkage as can be achieved during clinic-based testing. Follow-up data collection was beyond the scope of this pilot study; therefore, we were unable to evaluate linkage to care after home-based testing based on these data. The study was performed in three distinct geographic villages in Nakivale, representing participants from three primary countries of origin. Findings may not be generalizable to other country groups or to refugee settlements with people from different countries of origin. However, as of the end of 2016, there were nearly 1 million refugees in Uganda and numerous refugee settlements in the southern half of Uganda with similar population demographics [48, 49]. It is important to note that we do not have complete data to present sero-concordance among household members; the numbers we presented are an estimate based on existing data. Further, we did not design our data collection tools to record couples, and therefore we are unable to evaluate sero-concordance among couples. Finally, the study design comparing home-based testers from early 2014 and clinic-based testers from 2013 did not account for potential secular trends that may have effected HIV testing behaviors.