Examining the Use of HIV Self-Testing to Support PrEP Delivery: a Systematic Literature Review

Purpose of Review HIV self-testing (HIVST) has the potential to expand access to and uptake of HIV pre-exposure prophylaxis (PrEP) delivery. We conducted a systematic literature review to understand the evidence on HIVST use for PrEP delivery. Recent Findings After screening 1055 records, we included eight: three randomized trials and five values and preferences studies. None measured PrEP initiation. Most studies occurred in Sub-Saharan Africa (7/8) and included different populations. One trial found that HIVST use between quarterly clinic visits as part of an adherence package with biofeedback slightly increased adherence; the other two trials found that HIVST use between or in lieu of quarterly clinic visits had no significant or non-inferior effects on adherence. HIVST to support PrEP delivery was acceptable, feasible, and preferred. Summary HIVST use for PrEP continuation largely resulted in similar outcomes to standard-of-care delivery and was perceived acceptable and feasible. Further research is needed to optimize HIVST use within PrEP programming. Supplementary Information The online version contains supplementary material available at 10.1007/s11904-022-00617-x.


Introduction
Pre-exposure prophylaxis (PrEP) use has been increasing steadily since it was first recommended by the World Health Organization (WHO) for HIV prevention in 2015 [1]; however, many individuals that could benefit from PrEP lack access [2]. Common barriers to PrEP initiation and continuation in high HIV prevalence settings include long wait times at often overcrowded healthcare facilities [3] and the need for frequent clinic visits for HIV testing and PrEP refills which lead to high client opportunity costs [4], as well as stigma associated with PrEP access and use [5]. Simplified and novel models of PrEP delivery are needed to increase the reach and access of PrEP to populations at HIV risk not currently engaged in PrEP care [6].
HIV self-testing (HIVST), which has been recommended as an effective HIV testing approach since 2016, has the potential to simplify and support PrEP delivery [7]. Currently, the WHO recommends that individuals using PrEP should test for HIV every 3 months to detect potential breakthrough infections [8], and that HIVST can be used as a way to maintain PrEP programs in the context of the COVID-19 pandemic [9]. Currently, PrEP initiation is generally linked to clinic-based HIV testing services, followed by quarterly testing that accompanies PrEP refills. HIVST could enable individuals taking PrEP to test routinely, by replacing or complementing existing testing intervals with providers, which could potentially increase access and adherence to PrEP services [10]. Additionally, HIVST could enable greater differentiated service delivery for PrEP initiation and continuation in new settings beyond the bounds of traditional healthcare facilities, such as at private pharmacies or during at-home visits [11].
We sought to understand the existing evidence on the use of HIVST to support PrEP delivery to inform policy making. In this review, we specifically sought to understand how HIVST has been used to support PrEP initiation and continuation by identifying studies that measured the effectiveness of, values and preferences for, and economic outcomes associated with these interventions.

Methods
We followed guidelines from the Cochrane Collaboration [12] and PRISMA [13] for the completion of this systematic literature review. Our review protocol is included in Appendix 1 (Supplemental material) and published on PROSPERO (ID CRD42022296937). geographic restrictions on our database, trial registry, or abstract searches, and only included English publications and peer-reviewed papers. We reviewed the reference list of included studies for additional publications and contacted experts in the field to identify any additional articles not found through other search methods. We continued to search for trials we found that did not have outcomes reported until a month before we submitted this manuscript for publication and added any identified study results to the review.

Study Selection
We included studies for extraction that focused on the use of HIVST for PrEP initiation or continuation. This included studies that used HIVST in lieu of traditional HIV testing services (HTS) as well as in addition to HTS at different time points. We defined PrEP initiation as the process of being prescribed and dispensed PrEP for the first time or after a long break (i.e., restarting PrEP). We defined PrEP continuation as continuing PrEP use following initiation, which included measurements of retention and drug refills as well as adherence (e.g., self-reported PrEP use, detectable drug levels in blood/hair/urine, pill counts). We identified three potential types of PrEP continuation: (1) as prescribed (e.g., often every 3 months, could be linked or not necessarily linked to PrEP refills), (2) as needed (e.g., for stopping and restarting as periods of HIV risk change), and (3) as desired (e.g., between PrEP visits and refills). Studies that used HIVST to support any of these types of PrEP continuation were included in our review.
Additionally, we only included studies for extraction that measured or planned to measure effectiveness, values and preferences, or economic outcomes related to HIVST use for PrEP delivery. Studies that measured effectiveness outcomes directly compared a group, facility, geographic area, or population that used HIVST to a group, facility, geographic area, or population that used only standard HTS without HIVST for PrEP delivery. These studies also included at least one of the primary outcomes: (1) PrEP initiation or (2) PrEP continuation. If studies met all the criteria but did not present comparative data, we still included them and presented them as case studies. Studies that measured values and preferences outcomes included qualitative or quantitative data on client and provider opinions, perspectives, values, and preferences related to PrEP initiation and continuation supported with the use of HIVST at clinics or in the community. Finally, studies that measured economic outcomes presented primary data on the costs, cost-effectiveness, cost-utility, or cost-benefit of using HIVST to support PrEP initiation or continuation.
Two authors (AK and CK) reviewed all titles and abstracts to ensure reliable application of the inclusion criteria. Authors AK and AR independently screened all studies at the full-text level and noted reasons for exclusion. We resolved any disagreements on inclusion via group consensus following discussion. We used Covidence (Melbourne, Australia) to collaboratively screen titles/abstracts and facilitate the full-text review.

Data Extraction and Analysis
We collaboratively developed a structured data abstraction form in Covidence. Thereafter, two authors (AK and AR) extracted the following information from each study: author; title; year; study location; study design; study population (e.g., men who have sex with men); sample size; a description of the intervention (e.g., HIVST for PrEP initiation or continuation); a description of the comparison group (if available); and a summary of key findings. For effectiveness outcomes, we included effect size estimates, confidence intervals, and significance level. For values and preference outcomes, we included a descriptive summary of key findings. Authors AK and AR assessed the risk of bias for each study according to guidance by the Cochrane Collaboration and used a risk-of-bias visualization (robvis) tool [14] to create figures that reflect these outcomes (Appendix Figs. 1-4). To resolve any differences in data extraction and risk of bias assessment, we reviewed any disputed studies together until we achieved total group consensus (AK, AR, AW, CJ, CK, KN, and KO).
We summarized the key findings of the review using a modified Developing and Evaluating Communication Strategies for Support Informed Decisions and Practice Based on Evidence (DECIDE) Evidence to Decision (EtD) framework to summarize our results. The DECIDE EtD framework includes criteria that have been determined as necessary for informing evidence-based recommendations to key stakeholders such as health professionals, policymakers, patients, and the general public [15,16].

Results
Our search yielded 1055 unique studies and after screening titles, abstracts, and full manuscripts, we identified eight for inclusion: three randomized controlled trials (RCTs) measuring effectiveness outcomes and five studies measuring values and preferences (Fig. 1). We identified two of the three included RCTs in December 2021, four months after completion of the initial review in August 2021, while reviewing relevant study protocols identified in the initial review (Appendix Table 1).
We describe the characteristics of the studies included in Table 1. The majority of studies were in sub-Saharan Africa (86%) and in different populations (e.g., transgender people, sex workers, HIV serodifferent couples). All included studies focused on HIVST-supported models for daily, oral PrEP delivery, and used a combination of both oral-fluid and blood-based HIVST. All the RCTs reported the effect of these models on PrEP continuation; no studies reported on the effect of HIVST-supported PrEP delivery on initiation. Among the studies that reported on HIVST use for PrEP continuation, four (50%) reported on continuation as desired (with HIVST for testing between refill visits) and six (75%) reported on PrEP continuation as prescribed (with HIVST for refilling). Like PrEP initiation, no studies reported HIVST use for PrEP continuation as needed (with HIVST to inform stopping and restarting PrEP).

Effectiveness Outcomes
In the three RCTs that measured effectiveness outcomes, HIV self-tests were used in a unique way to support PrEP continuation ( Table 2). The Empower study used oral-fluid HIV self-tests to support as-desired interim testing between quarterly clinic-based PrEP visits for sex workers in Uganda [17••]. The JiPime-JiPrEP study used both oral-fluid and blood-based HIV self-tests to support semiannual clinicbased PrEP visits with six-month PrEP dispensing and HIVST for as-prescribed interim testing at three months for HIV serodifferent couples and women singly enrolled in Kenya [18••]. The PrEP-PP study delivered oral-fluid HIV self-tests as part of a PrEP adherence promotion package with counseling and enhanced adherence biofeedback to support as-desired interim testing between quarterly clinicbased PrEP visits for postpartum women in South Africa [19••].
In the identified protocols for ongoing RCTs and case studies (Appendix Table 1-2), HIV self-tests were additionally used for continuation as prescribed by testing while waiting for clinic-based PrEP refills in Kenya [20] and continuation as desired by testing between quarterly clinic visits in Uganda [21] and Kenya [22].
Adherence All three RCTs reported on the effect of HIVSTsupported PrEP delivery models on PrEP adherence, as measured by electronic adherence monitoring (EAM) (Empower) [ And finally, the JiPime-JiPrEP trial found that six-month PrEP dispensing supported interim as-prescribed HIVST at three months and resulted in non-inferior PrEP adherence (61% adherence) compared to three-month PrEP dispensing with clinic-based rapid diagnostic tests (RDT) at six months (57% adherence; risk difference [RD] 2.4%, one-sided 95% CI lower bound −5.1%) [18].

Retention and Refilling
The Empower study and JiPime-JiPrEP trial also measured the effect of HIVST-supported PrEP delivery models on PrEP retention and refilling, respectively [17••, 18••]. The Empower study found no effect of HIV self-test delivery at quarterly clinic-based PrEP visits on retention at 12 months [17••], while the JiPime-JiPrEP trial found that HIVST for interim testing between biannual clinic-based PrEP visits was non-inferior compared to quarterly clinic-based HIV testing at six months [18••].
HIV Testing Outcomes All included RCTs had slightly different HIV testing outcomes. In the Empower study, almost all participants (99%) that received the intervention used more than one HIV self-test between PrEP clinic visits by 12 months [17••]. In the JiPime-JiPrEP trial, any HIV testing in the past six months was high (>80%) among participants that received six-month PrEP dispensing supported with interim HIVST and non-inferior compared to quarterly clinic-based Table 1 Descriptive characteristics of studies included in our review, N=8 1 Abbreviations: WHO, World Health Organization; PrEP, pre-exposure prophylaxis; HIV, human immunodeficiency virus; STI, sexually transmitted infections 1 Data are shown for eight included manuscripts and conference abstracts, which represented seven unique PrEP trials, delivery programs, or values and preferences studies (because some studies reported findings for the same population or the same PrEP program) 2 There was no overlap in randomized trials and values and preferences studies 3 The frequencies and percentages in this category sum to greater than the total number of studies because several studies included more than one study population  3 Blood-based 1 (33%) 1 (20%) Oral-fluid based 3 (100%) 4 (80%) Not mentioned 0 1 (20%) Primary outcome: HIVST for PrEP 3 Continuation as desired (e.g., between visits/refills) 1 (33%) 3 (60%) Continuation as prescribed (e.g., often for refills) 3 (100%) 3 (60%) Continuation as needed (e.g., for stopping and restarting) 0 0 Initiation 0 0 HIV testing at six months [18••]. In the PrEP-PP study, HIV self-test delivery with PrEP adherence counseling increased partner testing in the past month compared to adherence without HIV self-test delivery at one month [19••].
HIV Incidence In all three RCTs, HIV incidence was very low; only one participant seroconverted in the Empower study [17••] and no participants seroconverted in the JiPime-JiPrEP study [18••] and in the PrEP-PP study [19••]. None of the RCTs were powered to measure this outcome.

Value and Preferences Outcomes
In the five studies that measured values and preference outcomes, the models of HIVST-supported PrEP delivery were all unique (

• Economic empowerment:
HIVST-supported PrEP helps eliminate unknown client HIV status as a barrier to the provision of condomless sex with clients (associated with higher client fees).

• Relational empowerment:
HIVST restored trust in partners' fidelity upon being reunited after a separation. This trust, in combination with condomless sex made possible by PrEP use, restored intimacy, empowering partnered relationships.

• Empowerment of sexual health:
HIVST-supported PrEP enabled SWs to take control of their HIV prevention efforts and avoid the stigma of public clinic visits.

Evidence Summary
We summarize the key findings from this systematic literature review, including risk of bias (in Table 4). We found that the overall risk of bias across all included studies was low to moderate (Appendix Figs. 1-4). For the three RCTs, most bias was attributable to missing outcomes and for the five case studies, bias was primarily attributable to confounding and selection of participants. In this table, we clarify that no studies were identified that explored HIVST use for PrEP initiation or for stopping and restarting PrEP.

Discussion
This systematic review found that HIVST use for PrEP continuation results in similar adherence and retention outcomes compared to standard-of-care PrEP delivery. In addition, the use of HIV self-tests was perceived to be feasible and acceptable. Most of the included studies were conducted in sub-Saharan Africa and across different populations. In the RCTs, HIVST was used to support PrEP continuation as

Risk of bias 1
Low to moderate risk of bias.

Balance of benefits vs. harms
HIVST-supported models of PrEP delivery resulted in similar PrEP continuation outcomes to standard-of-care PrEP delivery with no adverse events.

Effectiveness
Adherence: HIVST-supported PrEP delivery models achieve similar (no effect/non-inferior effect) adherence compared to standard clinic-based PrEP delivery (Empower & JiPime-JiPrEP). HIVST use for PrEP continuation may increase adherence among single women (JiPime-JiPrEP) or as part of an adherence promotion package with biofeedback (PrEP-PP). Retention: HIVST-supported PrEP continuation achieves similar levels of PrEP retention and refills (no effect/non-inferior effect) compared to standard clinic-based PrEP delivery (Empower & JiPime-JiPrEP). HIV testing: HIVST use during PrEP continuation was high when used for interim or as prescribed testing in all three RCTs. HIV incidence: HIV incidence was very low across all three RCTs with only one person who seroconverted. Social harms: No adverse events or harms reported by using HIVST for PrEP continuation.

Values and preferences
Feasibility. Participants completed HIVST between PrEP clinic visits and returned to the clinic for scheduled follow-up. These models reduced the overall number of PrEP clinic visits and saved time. Acceptability. HIVST-supported models of PrEP delivery reduced anxiety associated with PrEP clinic visits and were preferred to clinic-based or provider-delivered PrEP.

Economics/costs
No cost data on models of PrEP delivery supported by HIVST were identified.

Equity
HIVST has the potential to increase equity by enabling more people to access or continue using PrEP.

Research gaps
There are no studies on HIVST use for PrEP initiation or for stopping and restarting PrEP. More evidence is needed in this research area to better understand if individuals interested in PrEP can safely initiate or re-initiate PrEP as needed themselves with the assistance of HIVST. Additionally, no information on the costs of these models is available at the moment. Risk of bias was measured using RoB2 for RCTs and ROBINS-I for non-RCTs desired with testing between scheduled clinic visits [17••, 19••], and as prescribed with testing in lieu of scheduled visits [18••]. From these trials, we found that HIVST-supported models of PrEP delivery may increase adherence when delivered as part of an adherence package with biofeedback [19••] or may result in non-inferior adherence when used as a replacement for clinic visits [18••], compared to standardof-care quarterly clinic-based PrEP delivery. None of the studies in this review directly evaluated the use of HIVST to support PrEP initiation. This review highlights evidence for HIVST may be best used to simplify models of PrEP continuation rather than enhance existing models. For example, in the Empower study, where HIVST was additionally delivered with PrEP at scheduled quarterly clinic visits, this did not significantly impact PrEP continuation (e.g., retention and adherence) [17••]. And while a significant increase in PrEP adherence was observed in the PrEP-PP study, which tested an enhanced model of PrEP delivery that included HIVST, this effect was relatively modest and it was difficult to determine how much of the effect was attributable to HIVST delivery because the model also included adherence biofeedback and counseling [19••]. While this did add costs and complexity to some aspects of PrEP delivery, the ability for HIVST with counseling and adherence support to increase the number of people taking PrEP and reduce HIV incidence may be worth the investment. In contrast, a simplified model of PrEP delivery that reduced the number of clinic-based PrEP visits in half by replacing quarterly clinic-based testing with athome HIVST, as tested in the JiPime-JiPrEP trial, resulted in non-inferior PrEP continuation (e.g., HIV testing, PrEP refilling, and adherence) compared to standard-of-care delivery [18••] with potential cost and time savings to both PrEP clients and providers (the costing analysis for this trial is forthcoming).
HIVST-supported models of PrEP delivery reflect a movement towards self-care and community-based care that may have benefits beyond just PrEP initiation and continuation. For example, if these models are used to simplify PrEP delivery, like in the JiPime-JiPrEP trial [18••], this could result in health systems savings and reduce the client burden associated with PrEP continuation. Additionally, access to HIVST in between PrEP refills could have psychological and emotional benefits, including feelings of empowerment associated with knowledge of HIV status and ability to control their HIV risk by overcoming physical and mental barriers to intimacy [27], as well as motivation to use PrEP to remain HIV uninfected [26•]. Such benefits could be particularly useful in populations with notably low adherence, such as adolescent girls and young women [28][29][30]. And while it was not explored explicitly in any of the studies identified in this review, PrEP clients with fluctuating HIV risk could potentially use HIV self-tests to stop and restart PrEP as needed with the appropriate guidance and instruction. Finally, in the era of COVID-19 and other potential future pandemics, HIVST-supported models of PrEP delivery could help decongest clinics, allowing healthcare provider to focus on serving sick patients, and mitigating potential disruptions to PrEP delivery [11].
In this review, we did not identify any completed, ongoing, or planned studies evaluating the use of HIVST to support PrEP initiation. We believe this is primarily attributed to two factors: (1) in many settings, PrEP still requires a prescription from a certified healthcare provider [31], and (2) concerns still remain around the sensitivity and specificity of HIV self-test kits, especially when it comes to diagnosing acute HIV infections [32]. First, if a healthcare provider is at risk of losing their license, they may be unmotivated to provide PrEP based on clients' self-reported HIVST results. However, new online models of PrEP delivery are in development that may help address this by enabling individuals self-testing at home to upload an image of their self-test results, which providers can interpret and validate with the help of artificial intelligence algorithms [33], similar to models of COVID-19 testing that have been developed during the pandemic [34,35]. Second, because HIV self-tests may detect HIV later than standard facility-based testing algorithms used for PrEP prescribing in many settings [36], concerns remain that someone living with HIV might be incorrectly diagnosed as HIV-negative and prescribed PrEP, which could lead to potential drug resistance [37]. Ongoing modelling studies suggest, however, that because acute and breakthrough HIV infections while using PrEP are rare, population-level PrEP drug resistance likely remains similar with or without HIVST use to support PrEP delivery. As a result, the potential benefit of HIVST enabling more people to access PrEP and avert new infections may outweigh likely this potential low-level risks of drug resistance [38].
The review had limitations that are important to note. First, the design of HIVST-supported models of PrEP delivery varied across the included studies, some using HIVST to support PrEP continuation as desired or as prescribed and some including HIVST as part of an intervention package with other things (e.g., adherence biofeedback). These differences made it difficult to draw direct comparisons between the studies and at times determine the effect that HIV self-test delivery alone had on study outcomes. Second, the measurement and timing of study outcomes differed across the included studies. Some studies measured PrEP adherence (i.e., TFV levels) using intracellular blood samples, while others used urine samples and some studies measured PrEP adherence at one month, while others measured it at six and 12 months, which again made comparisons across studies challenging. Third, feasibility and acceptability are multifaceted concepts for which the field of implementation research has yet to establish standard, validated metrics and definitions; thus, we captured diverse assessments for these outcomes in the review. Fourth, the review was limited to HIVST use for PrEP initiation and continuation and did not include the many interventions that have used HIVST to support linkage to PrEP services [39] or facilitate HIV testing among sexual partners [40][41][42] or peers [43]. Fifth, we only included English and peer-reviewed publications, and thus may have missed implementation projects that have used HIVST to support PrEP delivery but have not published these findings. Finally, most of the included studies took place among HIV priority populations in sub-Saharan Africa, which may limit the generalizability of the findings to other populations and settings.

Conclusion
The existing evidence on HIVST-supported models of PrEP delivery is limited but suggests that HIVST has the potential to simplify PrEP delivery and that HIVST-supported delivery models are acceptable and feasible among diverse populations. It is possible that research on these models is limited because the WHO [11] and many countries [44,45] currently do not recommend HIVST use for PrEP initiation and continuation outside of temporary adaptations to maintain essential services during the COVID-19 pandemic. Researchers should work in collaboration with policy makers, regulators, and implementors, to better understand their concerns around HIVST-supported models of PrEP delivery so they can generate evidence that fills these important research gaps and can be used to guide implementation studies and inform policy. In the HIV prevention world, there is a growing recognition that differentiated service delivery models can help decentralize prevention services and reach populations not engaged in traditional care models. Countries should consider integrating HIVST into PrEP delivery to potentially increase the resilience of service delivery and patient-centered care during an era of pandemics and achieve and maintain global HIV prevention targets.