Scaling up HIV self-testing in sub-Saharan Africa: a review of technology, policy and evidence

Purpose of review HIV self-testing (HIVST) can provide complementary coverage to existing HIV testing services and improve knowledge of status among HIV-infected individuals. This review summarizes the current technology, policy and evidence landscape in sub-Saharan Africa and priorities within a rapidly evolving field. Recent findings HIVST is moving towards scaled implementation, with the release of WHO guidelines, WHO prequalification of the first HIVST product, price reductions of HIVST products and a growing product pipeline. Multicountry evidence from southern and eastern Africa confirms high feasibility, acceptability and accuracy across many delivery models and populations, with minimal harms. Evidence on the effectiveness of HIVST on increased testing coverage is strong, while evidence on demand generation for follow-on HIV prevention and treatment services and cost-effective delivery is emerging. Despite these developments, HIVST delivery remains limited outside of pilot implementation. Summary Important technology gaps include increasing availability of more sensitive HIVST products in low and middle-income countries. Regulatory and postmarket surveillance systems for HIVST also require further development. Randomized trials evaluating the effectiveness and cost-effectiveness under multiple distribution models, including unrestricted delivery and with a focus on linkage to HIV prevention and treatment, remain priorities. Diversification of studies from west and central Africa and around blood-based products should be addressed.


INTRODUCTION
Adult HIV incidence has largely remained static in sub-Saharan Africa, with an estimated 25.5 million people living with HIV (PLHIV) and 1.3 million new infections annually [1]. Despite increased availability of provider-initiated and community-based HIV testing services (HTS), only 76% of PLHIV in eastern and southern Africa and 42% of PLHIV in western and central Africa are aware of their serostatus [2].
HIV self-testing (HIVST), a process in which individuals collect their own specimen, perform the test and interpret the results, can provide complementary coverage to standard HTS and reach undiagnosed PLHIV and individuals with high ongoing HIV risk [3]. Reactive results need to be confirmed through additional testing by a trained provider, whereas nonreactive results should prompt linkage to prevention services, including voluntary medical male circumcision (VMMC) and preexposure prophylaxis (PrEP) if indicated [3].
The field of HIVST is rapidly evolving. This review summarizes the current technology, policy and evidence landscape for HIVST in sub-Saharan Africa, with priorities for scaled implementation outlined. Current developments in technology and policy were assessed through HIVST.org, a relational map hosted by WHO [4]. Recent and ongoing observational studies and trials were identified through PubMed, ClinicalTrials.gov, ISRCTN and the Pan African Clinical Trials Registry. The review was conducted through September 2017.

TECHNOLOGY UPDATE
Products for HIV self-testing available in sub-Saharan Africa are listed in Table 1. Products use either oral-fluid or finger-prick blood samples and take between 5 and 7 steps and 1 and 45 minutes to provide results [5]. Ideal products should be easy to use and interpret for optimized accuracy, with clear instructions-for-use (IFU) that are available in local languages and understandable at low literacy and education levels [6]. Most HIVST products in development are repurposed professional-use rapid diagnostic tests (RDTs) [5], with innovation limited to packaging and IFU modifications and single-use parts (e.g. pipette, lancet, etc.) [7]. Most commercially available RDTs for self-testing are secondgeneration assays, which detect immunoglobin G antibodies but not immunoglobin M antibodies (third generation) nor viral antigens (fourth generation) and nucleic acids. Compared to later generations, second-generation tests require a longer window period of 28 days between infection and test positivity [5].
In low and middle-income countries (LMIC), price per self-test currently ranges from US$2 to 3 for public sector procurement and US$8-16 retail in the private sector [5]. In 2017, unit costs for Ora-Quick HIV Self-Test were reduced under a temporary donor agreement, enabling government or charitable purchase for US$2 in 50 LMICs [8]. This agreement has important implications for HIVST market development, with potential for price reductions to increase demand and facilitate a competitive market for HIVST. Alternatively, it could discourage manufacturers from investment in innovations that improve usability and precision but increase costs, underpinning the need for incentives to further product development [7].

POLICY UPDATE
International and national policy to guide HIVST implementation, as well as regulatory and quality assurance systems, are integral to scale-up in sub-

KEY POINTS
HIVST is moving towards scaled implementation, with the release of WHO guidelines, WHO prequalification of the first HIVST product, price reductions of HIVST products and a growing product pipeline.
Multicountry evidence from southern and eastern Africa confirms high feasibility, acceptability and accuracy across many delivery models and populations, including adolescents, men and female sex workers, with reassuringly minimal harms.
Randomized trials evaluating the effectiveness and costeffectiveness under multiple distribution models, including unrestricted delivery and with a focus on linkage to HIV prevention and treatment, remain priorities.
Diversification of studies from west and central Africa and around blood-based products should be addressed. Saharan Africa. In 2016, WHO released guidelines recommending HIVST based on evidence of increased uptake and frequency of testing, especially among underserved and high-risk populations [3]. This prompted a shift in national policies supportive of HIVST, from four countries in 2015 to 40 countries at the time of the review, of which 15 countries are from sub-Saharan Africa (Fig. 1) [9]. Despite the changing policy environment, few countries are implementing HIVST at scale and only Kenya has released full operational guidelines [10]. Regional and national regulatory systems for HIVST are also generally poorly developed in sub-Saharan Africa, with unregulated and low-quality products available for purchase in certain markets [11]. To inform procurement decisions, sub-Saharan African countries tend to rely on approvals issued from founding members of the Global Harmonization Task Force or WHO [5], which prequalified its first HIVST product for LMICs in 2017 [12]. Approved products are also now available through limited pharmacies and retailers in Kenya and South Africa [13,14]. However, given the numerous potential outlets for HIVST through the private sector, countries will need to reassess their regulatory and postmarket surveillance systems to ensure adequate consumer protection from ineffective devices, without presenting undue barriers to bona fide manufacturers.

Readiness and preferences for HIV self-testing
Successful implementation of HIVST hinges on both provider readiness to deliver HIVST and public demand once HIVST services have been established. Early studies in sub-Saharan Africa reported high interest in HIVST among the general population [15][16][17], couples [18], high-risk populations [19], healthcare providers [20][21][22] and policy stakeholders [23,24] [25,30] (Table 2). Principal motivations for self-testing include convenience and associated time and cost savings; control over the testing process; privacy and confidentiality; and ease-of-use and   38]. Specific values and preferences for HIVST varied by population group. In Kenya, there was high interest among the general population for feebased, pharmacy models because of enhanced accessibility [34]. Alternatively, individuals who valued access to professionalized care preferred to obtain HIVST kits from health facilities [25,31]. Young people in Malawi and Zambia preferred low-cost, home delivery and valued HIVST for providing greater discretion around their sexual Men expressed interest in distribution through lodges and bars because of flexible service hours amid their work obligations [25]. Secondary distribution of self-tests to male partners of antenatal care clients similarly provided added convenience and privacy compared to clinic-based couples testing, wherein male attendance was often stigmatized [26 & ]. Sex workers valued HIVST as a way to demonstrate trust to primary partners and inform sexual decision-making with commercial sex clients [30], whereas HIV-uninfected individuals on PrEP found that HIVST reduced anxiety in between scheduled testing at clinics [35 && ]. Willingness to purchase HIVST kits varied across studies (  29]. Commonly reported concerns included the possibility for coercive testing and inability of individuals to self-test accurately, psychologically cope with reactive results and successfully link to confirmatory testing and care [17,20,[23][24][25]26 & ,27 & , 28,31,34,37,38]. User-suggested solutions to tackle potential adverse effects included clear, locally translated IFUs, coupled with development of effective training and educational materials for providers and accessible in-person and hotline-based support for self-testers [25,27 & ,31,33,37,38]. A recent study among Malawian men found that time-limited behavioral motivators, specifically financial incentives (US$3 and $10) and phone call reminders, were also acceptable to facilitate linkage to prevention and care [26 & ].

Performance and usability of HIV self-testing
The public health impact of HIVST is contingent on user ability to self-test and confidence in the results to access further HIV services. Blood-based self-tests have generally performed higher than oral fluidbased self-tests, with sensitivity of 96.2-100% versus 80-100% and specificity of 99.5-100% versus 95.1-100%, but have rarely been evaluated in sub-Saharan Africa [3]. Oral fluid-based self-tests have achieved good sensitivity (93.6-100%) and specificity (99.1-100%) [15,16,19,39] by rural and urban Africans with a demonstration or provider supervision. In a recent blood-based study in South Africa, 96.4% of 224 participants correctly performed the self-test and interpreted the results under direct assistance [36 && ]. Studies evaluating unassisted oral fluid-based self-tests have attained sensitivity of 66.7-90.0% and specificity of 95.2-100% [19,22,29]. Performance can depend on literacy level and previous exposure to HIV testing, but can be optimized through a demonstration-of-use [40][41][42]. Across studies, self-testing was often described as easy, with few reported errors [6,15,19,20,22,29,35 && ,36 && , 38,39,43]. Common missteps included incorrect specimen collection and use of the buffer solution and early reading or misinterpretation of results [15,19,29,36 && , 38,39]. Visual stability of selftests was inconsistent [44,45].

Uptake of testing
Evidence on effective delivery models to increase testing coverage among underserved populations and testing frequency among high-risk populations is critical for informing HIVST implementation.
A foundational study in Malawi reported that community-based HIVST implementation led to high uptake, particularly among women and adolescents, in a high prevalence setting [16]. High demand for HIVST has been subsequently shown for pharmacy [34] High uptake was also achieved under secondary distribution models. In a cohort of 280 pregnant and postpartum women and sex workers, 75-91% of participants across groups reported distributing kits to their primary partner and 80% of sex workers also distributed to commercial sex clients [47 && ]. Among partners of sex workers who received a reactive result, 90% were clients, underpinning the ability of HIVST to reach high-risk individuals. A sister randomized controlled trial (n ¼ 570) observed higher coverage of self-reported partner testing among pregnant and postpartum women when given HIVST kits for secondary delivery compared to partner invitation letters for clinic testing (90.8   ]. Couples testing and partner status disclosure were also more likely. A number of ongoing randomized trials are examining the effectiveness of HIVST on recent and lifetime testing and positivity. The Self-Testing Africa (STAR) Initiative is conducting trials of community-based delivery of self-tests among the general population in Malawi (NCT02718274) and Zambia (NCT02793804). Other notable trials are evaluating direct and secondary distribution among young women and sex workers (NCT02827240; NCT02846402) and their partners (NCT03135067; NCT03162965).

Linkage to prevention and care
Scalable strategies for confirmation of HIV status and continuation into the HIV prevention or care cascade are needed to maximize individual and public health benefits of HIVST. Studies have described suboptimal linkage to care following reactive results, though they were not designed to assess nor statistically powered on linkage [16,46 && ,47 && ,48]. In Malawi, linkage to care was 56.3% among community-based self-testers [16]. A study in Kenya reported linkage to confirmatory HTS was two of four among partners of pregnant and postpartum women and 26 of 41 among partners of sex workers [47 && ]. In a Kenyan trial, two of eight sexual partners linked to care in the HIVST arm compared to three of four partners in the clinic testing arm, as reported by proxy [46 && ].
To enhance linkage after self-testing, an early trial in Malawi found that the offer of home-based confirmatory testing and antiretroviral therapy (ART) initiation led to a three-fold increase in population-level ART demand compared to referral to facility-based care [48]. Interventions to facilitate timely linkage to care are also being investigated in Zimbabwe (PACTR201607001701788) and Malawi (ISRCTN18421340), with preliminary results from the latter study reporting significant benefits on linkage to VMMC and ART using financial and nonfinancial incentives [49].

Cost and cost-effectiveness
Cost and cost-effectiveness estimates, which are highly context-specific and dependent on the delivery model and prevalence of undiagnosed HIV, are vital to inform national HIVST policy and implementation. Mathematical modeling from Zimbabwe suggests that HIVST has potential to be cost-effective, contingent on delivery to highburden settings with low coverage of HIV testing; reductions in delivery costs through less resource-intensive implementation and cuts in HIVST unit costs; and improvements in linkage to prevention among HIV-negative individuals [50]. More epidemiological and economic data from HIVST programs are needed to parameterize costeffectiveness estimates at a national level [51].
Ongoing trials mentioned in this review are aiming to address this evidence gap. Cost data are available from previous research in urban Malawi, with evidence that community-based HIVST has potential to be cost-effective in high-burden settings [52]. The mean cost per individual tested through community-based HIVST (US$8.78) was comparable to facility-based HTS (US$7.54) [53 && ], but higher per HIV-positive individual identified (US$97.50 versus US$25.30-US$76.14). Self-testers incurred almost no costs, compared to US$2.93 for facility testers, and were diagnosed at earlier stages. A follow-on analysis reported no differences in economic and quality of life outcomes between self-testers and facility testers one year after ART initiation [54]. Maintaining a strong focus on implementation design to maximize cost-effectiveness will be important, especially as the prevalence of undiagnosed HIV declines.

Social and behavioral impact
Broader social and behavioral effects of HIVST, specifically around social harms and sexual risk-taking, are important to evaluate and consider for scaled implementation. Reassuringly, the incidence of serious social harms related to HIVST has been uniformly low, with no cases of suicide and few reports of intimate partner violence to date [16,18,35 && ,46 && ,47 && ]. Coercive testing has been reported by couples but highlights a complex dynamic, where pressure to self-test is balanced by the sense that primary partners are entitled to know each other's status [18]. Evidence on sexual risktaking is limited, with one study in Kenya finding that a cohort of prepartum and postpartum women and female sex workers were significantly less likely to have sexual intercourse (18 versus 62%, P < 0.01) and more likely to use condoms (100 versus 44%, P < 0.01) if their partners received reactive compared to nonreactive self-test results [47 && ]. Ensuring that self-testers are able to assess and avoid social harms and understand risks of serosorting, as well as developing systems for identification and management of social harms, are key considerations for implementers.

Research gaps and priorities
safety of HIVST among target populations, but has also revealed gaps that require immediate prioritization (Table 3). Most studies have taken place in southern and eastern Africa, with geographic diversification from west and central Africa needed. There is also limited evidence on blood-based self-tests, which use less costly specimen collection devices and are more sensitive compared to oral fluid-based tests, and are therefore vital to facilitate price reductions in HIVST products and for use among high-risk populations. Research on how implementers can improve HIVST performance by optimizing IFUs and developing clear support materials through systematic assessment and adaptation remains a priority across products.
The majority of current evidence comes from small-scale observational studies, with findings emerging from a number of ongoing randomized trials and economic evaluations. Evidence on effective and affordable HIVST delivery models for increasing testing coverage among underserved populations and testing frequency among high-risk populations are necessary for country decisionmaking. Recent and ongoing trials have largely evaluated community-based and partner-delivered HIVST, revealing a gap among a wider range of distribution models. With scale-up in mind, unrestricted distribution through public and private sectors, as well as interventions to minimize linkage delays among hard-to-reach populations, should be evaluated. Studies should also ensure alignment of epidemiological and economic metrics with inputs required for mathematical modeling to inform national cost-effectiveness estimates, with the aim of effectively and efficiently reaching undiagnosed PLHIV. Finally, secondary effects of HIVST, including sexual risk-taking and potential for efficiency gains for providers through task-shifting, are relatively unknown and require further investigation.

CONCLUSION
HIVST is moving towards scaled implementation, with the release of WHO guidelines, WHO prequalification of the first HIVST product, price reductions of HIVST products and a growing product pipeline. Multicountry evidence from southern and eastern Africa confirms high feasibility, acceptability and accuracy across many delivery models and populations, including adolescents, men and female sex workers, with reassuringly minimal harms. Evidence on the effectiveness of HIVST on increased testing coverage is strong, while evidence on demand generation for follow-on HIV prevention and treatment services and cost-effective delivery is emerging. Despite these developments, HIVST delivery remains limited outside of pilot implementation. Important technology gaps include increasing availability of more sensitive HIVST products in LMICs. Regulatory and postmarket surveillance systems for HIVST also require further development. Randomized trials evaluating the effectiveness and More evidence needed on effectiveness and cost-effectiveness under multiple delivery models, including unrestricted distribution through public and private sectors and strategies to minimize linkage delays. Limited studies on BST and from west and central Africa Secondary effects of HIVST, including on sexual risk-taking and healthcare efficiency, are unknown BST, blood-based self-testing; HIVST, HIV self-testing; LMIC, low-income and middle income; OFST, oral fluid-based self-testing; PQ, prequalification; SSA, sub-Saharan Africa.
cost-effectiveness under multiple distribution models, including unrestricted delivery and with a focus on linkage to HIV prevention and treatment, remain priorities. Diversification of studies from west and central Africa and around blood-based products should be addressed.