Feasibility and acceptability of home-based HIV testing among refugees: a pilot study in Nakivale refugee settlement in southwestern Uganda

This pilot was conducted in Nakivale Refugee Settlement, a 71-mile² settlement established in 1960, located in rural southwestern Uganda, and managed by the United Nations High Commissioner for Refugees (UNHCR) and the Ugandan government [26]. At the time of this study in 2014, Nakivale was home to 68,000 refugees, many of whom had lived in the settlement for generations; 52% of the settlement’s population was from the Democratic Republic of the Congo (DRC), 17% from Somalia, 15% from Rwanda, and 15% from Burundi [29]. The settlement is divided geographically into 79 villages with an average of 800 to 1000 people per village [26], and with villages largely comprised of refugees from the same country of origin. Refugees are provided plots of land of 50 m by 100 m on which to build their own home and grow food or raise animals [26]. Ugandan nationals also live in and around Nakivale and access health services in the settlement. Medical Teams International (MTI), the non-governmental organization responsible for health-related activities in Nakivale, oversees the medical care provided at the four health facilities in the settlement. At these health centers, HIV testing is offered free of charge and is conducted using serial rapid HIV tests outlined in the Uganda national guidelines [30]. At the time of this study, the serial HIV test algorithm started with Determine HIV-1/2 Ag/Ab Combo [31], confirmed positive specimens using HIV 1/2 STAT-PAK Assay [32], and used Uni-gold HIV as a tiebreaker when needed [33]. Additionally, antiretroviral therapy (ART) is available free of charge in these locations and, at the time of this study, was offered to those with a CD4 ≤ 350/mm³ or World Health Organization (WHO) stage III/IV based on 2010 WHO guidelines [34]. Data from a routine clinic-based HIV testing study conducted at Nakivale Health Center in 2013 showed the new HIV diagnoses frequency was 3.3% in the standard of care period and 4.5% in the intervention period (P > 0.5) in Nakivale [25], and the Uganda AIDS indicator survey reported HIV prevalence in the surrounding region of rural Uganda to be 7.3% [35].

From February–March of 2014, research assistants spent three weeks conducting home visits in Nakivale. Each week they visited a distinct village (labeled Village 1–3 for the purposes of this study). Village 1 is considered by locals in Nakivale to be Congolese, Village 2 Burundian and Village 3 Rwandan. Despite this perception, each village has people from various countries of origin including Ugandans. Research assistants were fluent in English, Kiswahili, Kinyarwanda, and Runyankore, had prior experience in HIV counselling, and had been trained in rapid HIV testing by the Center for Disease Control and Prevention (CDC) in Uganda. Working in teams of two, research assistants visited consecutive homes covering all dwellings in each village. Each home received a total of three separate visits, on alternative days of the week at different times of day, unless all adults were present at an earlier visit or unless the research assistants were asked not to return. Research assistants gathered information from those at home and from neighbors about how many individuals lived in each home and the approximate age of each person to evaluate potentially eligible participants. HIV counseling and testing were offered to all adults in residence who met eligibility criteria. Research assistants consented participants and offered an HIV test and an oral survey in one of the four languages noted above. Survey information, which was collected directly onto an electronic tablet, included demographic information (i.e. sex, age, refugee status, years in Nakivale, country of origin, relationship status, education), HIV knowledge, HIV testing history, and approximate travel time to clinic. Research assistants used serial rapid HIV tests as per Ugandan national guidelines [30]. The HIV tests used included Determine™ HIV-1/2 as a screening test [31], HIV 1/2 Stat-Pak™ Assay as a confirmatory test [32], and Uni-gold ™ HIV as a tiebreaker; these are the same HIV tests used at Nakivale Health Center at the time of this study [33]. Participants identified as HIV-positive were referred to the nearest health center’s HIV clinic with instructions to attend clinic within 1 week.

To protect participant privacy during data collection, an attempt was made to find a confidential location in or around the homes – participants were asked where they would feel safe taking the test and survey. To protect data confidentiality, survey data were de-identified and stored in electronic, pass-code protected tablets and on computers with anti-virus software. No individual identities were used in any reports or publications that resulted from this study.

Three distinct villages in Nakivale were selected for the home-based HIV testing pilot. This sampling plan was designed to include different villages representing predominant country groups in the settlement: Congolese, Burundi, and Rwandan. Additionally, Ugandan nationals reside throughout Nakivale, and we anticipated – based on findings from our clinic-based Nakivale study -- that our sampling plan would include nationals. Though Somalis were also a predominant country group in Nakivale (17% of settlement population) [29], < 4% of individuals identified as HIV-positive in our routine clinic-based HIV testing study were from Somalia [7]; therefore, we did not conduct home-based HIV testing in a Somali village in this pilot study. Eligibility criteria included: adults ≥18 years of age, capacity to give informed consent, and the ability to speak English, Kiswahili, Kinyarwanda, or Runyankore, similar to the inclusion criteria for our previous study in Nakivale [25]. The clinic-based HIV testing comparison cohort participated in routine HIV testing at Nakivale Health Center III in 2013 [25].

This study was approved by the Makerere University School of Health Sciences Institutional Review Board (Kampala, Uganda; Ref No 2012–020), the Uganda National Council of Science and Technology (Kampala, Uganda; HS 1167), and the Partners Human Research Committee (Boston, MA, USA; 2010P001963/BWH). Written consent was obtained from all surveyed and tested study participants.

Over the 3-week period, we visited 319 homes in three villages. Most households consisted of ≥2 eligible people and 9% of the households were abandoned (Table 1). For feasibility, we encountered 507 (90%) of the 566 eligible individuals reported to be living in these homes within three visits. For acceptability, 378 (75%) of the 507 encountered individuals participated in home-based HIV testing and received their test results. The proportion of those at home who tested for HIV varied by village (Village 1: 89%, Village 2: 64%, Village 3: 70%; p < 0.001). The majority (480/507, 95%) of the individuals reached were encountered during visits 1 and 2. There was a small decrease in the proportion willing to test from visit 1 to visit 2 (77 to 73%), and a steep decline on visit 3 (44%). Three individuals at home during visit 1 did not agree to testing initially, but agreed to and consented during a subsequent visit—two of these individuals tested during visit 2 and one during visit 3. All others tested during the visit in which they were first encountered at home. Of the 378 who tested for HIV, 7 (1.9%) were diagnosed HIV-positive (Village 1: 4 [2.5%], Village 2: 3 [2.9%], Village 3: 0 [0%]). Among 233 households with > 1 eligible individual, 119 households had more than one person participate in HIV testing and 114 of those households (96%) had sero-concordance identified in the study within the household.

There was no significant difference in sex distribution among home-based and clinic-based testing participants (56% vs. 53% female respectively, p = 0.20), but home-based testers were slightly older (median age 30 [IQR 24–40] vs. 28 [IQR 22–37], p < 0.001), were more often refugee than Ugandan national (93% vs. 79%, p = <0.001), and had lived in Nakivale longer (median time of 6 years [IQR 3–8] vs. 4 years [IQR 1–8], p < 0.001, Table 2). Of the participants who were tested, more home-based testers were from the Democratic Republic of the Congo (35% vs. 25%) or from Burundi (24% vs. 15%). A greater proportion of home-based testers were married/living together (77% vs. 63%). Fewer home-based testers had some secondary school or above education than clinic-based testers (11% vs. 20%). Home-based HIV testers had a higher proportion that scored ≥75% correct on the 4 question HIV knowledge test (70% vs. 60%) but no significant difference in the proportion with previous HIV tests (78% vs. 74%) or in the proportion with HIV tests within the past year (40% vs. 36%). Home-based HIV testers were more likely to live ≥1 h from a health clinic (74% vs. 52%). The yield for HIV-positive test results was not significantly different between home-based and clinic-based testers (1.9 vs. 3.4%, p = 0.27).

We examined the association between a limited number of available factors and willingness to test in home-based HIV testing (Table 3). Time of visit was not significantly associated with willingness to test (p = 0.50). There was also no difference in willingness to test between females and males (p = 0.66). However, as the number of eligible individuals at home when testing was offered increased, willingness to participate increased (70% when no one else present, 73% when only one other person was present, and 86% when more than one other person was present). For each additional person present, the odds of testing increased by over 50% (OR: 1.52, 95%CI 1.12–2.06, p = 0.007).