Willingness to use and ease of use

Most studies found that high proportions of (>70%) to their study population were willing to use HIV-ST but there was some variation across key populations. For example, willingness was consistently high among female sex workers (FSW) ranging from 72% [142] to 95% [30]. In contrast, rates in MSM were more varied, ranging from 51 [73] to 99% [77] (Fig 3). We noted that the two studies (conducted in Kenya [129] and in Brazil [73]) that reported low willingness to use rates included individuals who have never tested for HIV (self-test or usual testing). This inclusion may have led to the lower measure of overall willingness. One study in South Africa [48] found that the odds of being willing to use HIV-ST was two times higher among ever testers compared to never testers. Although, we also found a study in Tanzania [86] that did not find significant differences in the willingness to use rates of ever and never testers.

Fig 3

Acceptability results.

Nearly all studies concluded that HIV-ST was perceived easy to use [23, 28, 30–32, 35, 38, 40, 42, 44, 46, 47, 49, 57, 61, 63, 68, 70, 74, 77, 80, 82, 83, 85, 127, 130–134, 141]. Most studies also found that HIV-ST was preferred over standard testing (i.e. facility-based performed by a trained health care worker (HCW)) [22, 32, 39, 43, 46, 52, 61–63, 72, 80, 127, 130, 132, 136, 140].

Willingness to pay

Willingness to pay varied widely ranging from 21% [30] to 96.5% [133]. Most studies showed willingness greater than 50% across key populations. Some of the low rates are from populations with limited income such as students in Tanzania and Central African Republic [30, 75] and FSW primarily working in poor neighborhoods in Central African Republic [30]. The remaining study with low WTP [79] is on MSM in China where free HIV testing is available. Their sample also included a lot of people who have never tested which may affect WTP and the way they asked the WTP question (“Willing to purchase an oral self-test kit (in the next 6 months)?”) seemed to capture both willingness to try and pay (Fig 3).

Observational studies measuring uptake

Observational studies showed that with just market availability and no widespread program by the State or national scale actors, HIV-ST diffused among key populations. These surveys measured prevalent use of HIV-ST among HCW from Ethiopia [138], Kenya [21], and South Africa [31], among MSM in China [59, 170, 176, 178, 192], and the general population in Malawi and Zimbabwe [37]. Studies measured uptake either by asking about current use or reported uptake rates after offering the test kit for use during data collection. In the latter type, we found that those who adopt after being offered seem to represent a certain segment of the target population. For example, MSM in China who use gay apps were found to be more likely to use HIV-ST [193]. Women in community trial sites in Malawi [172] and Zambia [171] were more likely to participate and use HIV-ST. These two studies also showed significant differences in age and sexual behavior of HIV-ST users vs non-users. For example, in the Malawi trial, they found men, but not women, who recently practiced condom-less sex were significantly more likely to perform HIV-ST than those who did not.

Overview of interventional studies

Interventional studies have focused on evaluating the impact of implementation strategies on HIV-ST uptake or on overall HIV testing rates. The most commonly studied strategy was distribution of self-testing kits through community members, volunteers or peers [44, 61, 67, 74, 78, 85, 124, 125, 140, 146, 151, 159, 162, 194] or intimate partners (usually female) [28, 47, 49, 68, 144, 147, 152, 156, 161, 165, 166]. All except Tun et al. [74] used direct distribution of kits to users. Tun et al. [74], instead, asked peers in Nigeria to invite target users to visit a center to claim and use an HIV-ST kit. The other approaches for increasing uptake were distribution via online platforms [58, 66, 84, 87, 149, 157, 158, 160, 167], HIV-ST information or promotion campaigns [38, 67, 79, 150, 153, 155, 168], and offering HIV-ST after contact with health services (e.g., PreP, pharmacy visits, HIV testing) [25, 127, 141, 143, 145, 147, 154, 159, 164]. Many studies did not have controls that allowed for proper assessment of impact on HIV testing rates (Fig 4).

Fig 4

Harvest map of interventional studies on HIV-ST adoption.

Distribution using peers, partners or community members

The utilization of peers, intimate partners, or community members was mostly successful in increasing uptake. In quasi-experimental studies, there was high uptake (≥80%) after direct distribution by peers [43, 78, 146, 151] or partners [28, 144, 152]. Nearly all randomized trials [28, 47, 124, 144, 146, 147, 156], except for one study in Zimbabwe [67], showed significant positive effects on overall HIV testing rates in the interventions arms. Also, Choko et al. [152] noted that in Malawi, distribution via partners with financial incentives can improve actual clinic follow-up by male partners.

The non-significant results in the Zimbabwe study [67] might be due to the comparison group. The Zimbabwe study had three factors in the intervention (price of kit, type of kit distributor, and promotional messaging). They found that needing to pay even just $0.50 significantly reduced uptake. In rural sites, they compared community health workers to retail stores as distributors and did not find any significant difference in uptake. They also found that type of messaging did not significantly affect uptake.

Distribution via online platforms

Online platforms were a promising approach to educate about and/or distribute HIV-ST, but only one of nine studies included control groups that enabled some form of evaluation on testing rates. Tang et al. [149] demonstrated through a stepped wedge trial in China that a crowd-sourced online distribution coupled with a social media campaign can significantly increase the rate of HIV self-testing. Seven papers [66, 84, 87, 157, 158, 160, 167] showed that online distribution is feasible and led to the recruitment of HIV-ST users but no evaluation was done to assess performance compared to other distribution strategies.

A pertinent issue in online distribution is that HIV-ST users recruited online might not complete the counseling and testing process. One study in Thailand [58] demonstrated that a purely online approach to counseling and testing with HIV-ST had HIV testing completion rates comparable to the conventional in-person model. Another study in Brazil [87] also showed high rates of getting counseling among online users who tested positive.

Other distribution strategies using education or promotion

Seven papers that used education or promotion campaigns through means other than peers or partners had mixed results. SMS announcements from HIV clinics in Kenya regarding HIV-ST availability significantly increased HIV testing among FSW [153] but not among truck drivers [155]. In China, inclusion in an SMS group that pushes HIV-related messages also increased HIV-ST uptake and overall HIV testing rates [150]. However, one study in Zimbabwe [67] found that merely adding promotional messages to voucher distribution did not improve HIV-ST uptake. Another study also found that a hospital-based information campaign among HCW in Kenya [38] led to high uptake among information session HCW attendees but not among non-attendees. Finally, two studies focused on comparing types of recruitment. One study in China [79] showed that online recruitment’s impact on HIV-ST uptake was comparable to peer recruitment. Another China-based study [168] showed that social media key opinion leaders may be more effective than community-based organizations in recruiting HIV-ST users especially if they approach first-time HIV-ST users.

Some studies tested offering HIV-ST during contact with health services (e.g., routine visit, circumcision, pharmacy visits) and these had mixed results as well. Offering HIV-ST during a visit to an HIV testing facility seemed to increase HIV testing coverage. In Kenya, truckers who were randomized to the arm that can choose HIV-ST had 1.5x higher odds of accepting HIV testing compared to those randomized to just standard HIV tests [154]. Similarly, two other randomized trials in Malawi and South Africa [25, 147] that offered HIV-ST during facility visits or community events showed significantly higher HIV testing rates. A quasi-experimental study in Eswatini [164] showed an increase in HIV-ST uptake. Another quasi-experimental study done in Malawi, Zambia, and Zimbabwe [159] demonstrated the feasibility of distribution even in male circumcision sites, but they had no comparators for proper assessment. Integrating HIV-ST with pre-exposure prophylaxis (PreP) services also seemed helpful. In a sub-study of an open-label trial in Kenya [127], majority (93.2%) of recruited PreP users reported using HIV-ST at least once during the study. Offering HIV-ST during regular pharmacy visits in Kenya [141], however, did not have a high yield (only 35% of invited agreed to participate). Finally, in a community trial in Zambia [145], residents in communities that had access to HIV-ST (through community health workers) had higher odds of knowing their HIV status on follow-up than residents in control communities.

Appropriateness and qualitative results

Appropriateness is defined as “the perceived fit, relevance, or compatibility of the innovation or evidence-based practice for a given practice setting, provider, or consumer; and/or perceived fit of the innovation to address a particular issue or problem” [6]. We found that all studies that touched on appropriateness used a qualitative design. While Proctor’s framework differentiates between appropriateness, acceptability, and adoption, we found that qualitative themes for these outcomes relating to these three outcomes overlapped so we opted to present them together in this section.

Positive perceptions on HIV-ST

Qualitative studies echoed findings of surveys described previously, HIV-ST was perceived to be easy and convenient, often in comparison to standard testing [57, 78, 88, 91–93, 95, 97, 99, 100, 102, 103, 105, 107, 108, 111–114, 120, 122, 124, 127, 131, 133, 137–140, 183, 184, 191, 195–197]. HIV-ST allows them save time and money as they would not need to wait in a line at a testing facility, skip work, or spend on transportation [20, 57, 88, 92–94, 99, 100, 102, 107, 111, 113, 114, 116, 119, 122, 126, 131, 137, 140, 183, 184, 191, 196].

HIV-ST also had potential to reduce stigma and anxiety associated with testing [9, 20, 45, 91, 95, 98, 99, 103, 104, 107, 112, 114, 118–120, 122, 126, 127, 131, 138, 184, 198]. Compared to standard facility-based testing, HIV-ST was perceived to be better in protecting confidentiality [20, 57, 88, 91, 95, 98, 100, 103, 104, 108, 112–116, 118–120, 122, 124, 126, 127, 129, 131, 137–140, 183, 184, 195–197, 199] especially in settings where people doubt the ability of HCW to protect confidentiality [20, 57, 99, 100, 104, 113, 124, 126, 137]. It was a common belief that HIV-ST would be useful for increasing HIV testing rates in certain populations, such as men or young people [45, 48, 78, 90, 91, 95, 98–100, 102, 106, 116, 118, 120, 137, 139, 184, 190, 191, 196, 199].

Relatedly, HIV-ST was believed to empower users and promote autonomy [20, 91, 92, 95–100, 102, 104, 105, 118, 122, 124, 137–139, 184, 189, 190, 200, 201] since HIV-ST allowed the user to retain full control of their own health information. The self-test option may also address the feeling of coercion associated with standard testing since some patients felt that they could not refuse if asked to undergo HIV testing because they perceived nurses to be in a position of power [20, 91].

Considerations for designing HIV-ST programs

Studies showed several considerations for the design and implementation of HIV-ST programs in LMIC settings to improve acceptability and appropriateness. Proper kit design for the target population was often mentioned. Design considerations included the instructional materials [90, 98, 100, 103, 107, 114, 120, 124, 131–133, 187, 190, 201–203] and the type of fluid (blood or oral fluid) [90, 92, 100, 102, 104, 106, 108, 110, 114, 116, 119, 124, 127, 131, 132, 137, 139, 184, 191, 196]. Instructions needed to be clear and easy to understand. Some papers suggest making sure that there are clear instructions on how to link to care or seek support and include a number that users can easily call. Other studies showed that videos or mobile apps can serve as supplemental forms of education materials. Cognitive interviews was shown to be very useful in helping identify issues with instruction materials and optimize prior to roll-out [203]. In terms of specimen type, blood might be more trusted than oral fluid-based tests in some settings or by some subgroups. However, users might also have more difficulty with performing blood-based tests, particularly in using the lancet.

Counselling, especially post-test counselling, was deemed necessary in HIV-ST programs [9, 20, 45, 78, 89–92, 98, 100, 101, 103, 106, 108, 109, 111, 113, 114, 116, 119, 120, 127, 129, 132, 137, 139, 140, 184, 190, 196, 201]. Relatedly, two commonly reported negative features of the HIV-ST: the lack of access to immediate counselling [45, 57, 91, 92, 98–101, 103, 104, 106–109, 113, 114, 116, 118–122, 124, 126, 127, 132, 135, 138–140, 184, 186, 190, 195, 197] and the lack of linkage to care (e.g., confirmatory testing, anti-retroviral therapy) [45, 89–91, 98–100, 103, 104, 106, 108, 109, 113, 116, 121, 128, 129, 140, 186, 190, 196]. These two issues were voiced more often in the context of positive HIV-ST results. A commonly reported belief was that individuals who test positive would be more likely to resort to self-harm in the absence of counselling or post-test support [45, 88, 89, 91, 93, 98–101, 106, 107, 109, 113, 114, 116, 120–122, 129, 131, 132, 139, 140, 186, 190, 195, 196]. Some studies, however, suggested that there was no need to use traditional face-to-face facility-based counselling; telephone or online counselling could be sufficient [20, 57, 90, 91, 98, 106, 108, 111, 112, 119–121, 132, 184, 190, 201].

Selecting the appropriate mode of kit distribution was important to ensure accessibility and adoption [57, 78, 90, 92, 99, 104, 105, 108, 110, 112, 119, 122, 137, 139, 141, 163, 184, 189, 190, 202, 204]. Modes of distribution described include: (1) direct delivery by community health workers, peers, partners, and sex workers, (2) Home delivery, and (3) Pick up at health facilities or community-based organizations or other sites (e.g., health clubs, markets, private pharmacies, vending machines). Regardless of the mode, the venue for distribution needs to be accessible and convenient. Venues need to ensure that the user privacy is protected, and the discretion is maintained. Peer or community distributors need to be perceived as trustworthy individuals, preferably coming from the potential user’s own network. Users also need to be informed about storage strategies, especially if their homes have limited spaces for private testing.

Issues related to partner distribution

Partner distribution is an often-used strategy for distributing HIV-ST kits. Several papers explored issues of acceptability and appropriateness [94, 95, 99–102, 105, 107, 114, 118, 119, 121, 122, 127, 183, 184, 186, 189, 191, 196] and all except one [121] were done in African countries. Female partners find this strategy acceptable because it allows them to engage their partners in matters of sexual health and see it as a good way to encourage HIV testing and even couple testing (testing at the same time). Trust and relationship quality play important roles in the perceived acceptability or success of this strategy. Some females feel that it can improve their relationship and trust in their partners, but some fear that it can have the opposite effect. Males might view even view the request to use HIV-ST as a test of their fidelity.

Relatedly, the safety of female distributors is an important consideration in program design. Female distributors were believed to be at risk of physical violence, abandonment, and relationship dissolution. A qualitative study in Malawi [95] found that “women feared being branded as unfaithful and a possible domestic violence” if they bring up doing HIVST together. A similar study in Uganda also documented that women fear their partner’s reaction the test although, in this study, none of the participants reported serious adverse events after offering the kit to male partners except for angry reactions among those who lied about the purpose of the test [102]. Results of quantitative studies on safety and social herms are reported in the safety section below.

Studies seem to agree that to mitigate these risks (e.g., negotiation skills training) must be included in the HIV-ST program if partner distribution is adopted as a strategy. A few papers [114, 183, 184, 186, 191] outlined strategies used by female distributors on how to engage their partners. Examples are: (1) selecting the appropriate time and place for introducing self- or couple testing, (2) developing a distraction free environment for the discussion and testing, (3) mentioning positive experiences and/or benefits of HIV-ST, and (4) reassuring partners that it is a couple’s activity and not meant to test their commitment.

Policies and other issues

Modification of the policy environment to facilitate acceptance and adoption of HIV-ST seemed to be necessary for proper HIV-ST implementation. Policy issues included regulation or policies related to HIV-ST [45, 89, 97–100, 103, 108, 120, 123, 186, 190, 196, 200], pricing and financing [90, 98, 99, 103–105, 108, 118, 119, 123, 128, 131, 132, 137, 139, 187, 196, 202], and experience of frontline healthcare workers [141, 205]. There were calls for strict access to kits and putting in place protection against fake tests. One paper found that policy-makers from three African countries thought that policy should be “in line with WHO guidance and adapted to local conditions” [190]. Nearly all studies reported preference for free or heavily subsidized kits to ensure acceptability and adoption.

Relatedly, widespread availability of HIV-ST can also lead to coercive testing in the home and workplace and, in turn, threatens appropriateness. For example, parents can force their children to undergo an HIV-ST or a boss could require HIV testing of employees in the workplace or a spouse can coerce their partner during couples testing [94, 97, 100, 101, 116, 184, 190, 192].

Other reported considerations when designing HIV-ST programs included (1) adequate follow-up [100, 113, 118, 120, 127, 135], (2) establishment of peer support networks [78, 104, 107, 122, 140, 199] and, (3) proper education on HIV-ST use [93, 98–100, 102, 106, 108–110, 112, 116, 120, 122, 127, 131, 133, 139, 140, 190, 196, 201, 206]. Education, both at the individual and community level, ensures proper technique and addresses any concerns on test accuracy. Supervised testing may be more appropriate, especially for new users since they can provide additional education opportunities and emotional support. Despite overwhelming positive attitudes towards HIV-ST, a few studies [45, 100, 120] pointed out that HIV-ST will not improve testing coverage if unfavorable values towards HIV testing are not changed. Education to modify perceptions on HIV testing in general would then be needed. Education may also address potential problems like improper waste disposal [99, 119, 132, 190, 202], safety concerns (e.g., use of lancet as a weapon) [99, 137], psychological results of inaccurate tests [93, 190], sexual disinhibition due to negative results [100], and misconceptions regarding HIV or HIV-ST [90, 107, 110, 114, 116, 122, 132] (e.g., How can they trust oral testing if HIV cannot be transmitted via saliva?).

User errors and readability

Errors in performing HIV-ST were common. Studies reported error rates in different ways: some report the number of people with at least one error, others calculate an average or useability index, and some report error per step. Error rates of 10% or higher were commonly reported in studies that used unsupervised HIV-ST [24, 30, 40, 42, 46, 53, 57, 63, 77, 131, 132, 136, 142, 194] although error rates ranged widely from 1.6% to 93%. In these unsupervised HIV-ST studies, participants were observed but no guidance was provided outside of the education materials that were part of the kit simulating an unsupervised test. Meanwhile, supervised HIV-ST [23, 133, 173, 207] had lower error rates ranging from 0% to 19.5% (Fig 5). Gaining experience with HIV-ST did not lower error rates as seen in studies in South Africa, Brazil and Peru that compared error rates on the initial introduction and after months of unsupervised use [42, 77].

Fig 5

Feasibility results.

Requests for assistance was also reported [23, 30, 61, 127, 130, 133, 136, 142, 153, 155, 173, 208]. Studies on supervised HIV-ST reported higher assistance requests with rates ranging at 18.1% [130] and 41.5% [23] while studies on unsupervised ST ranged from 10% [136] to 23.6% [23] (Fig 5).

Seventeen studies performed readability assessments in which they asked participants to read one or more HIV-ST kit results but not necessarily their own [30, 32, 46, 53, 57, 70, 77, 80, 83, 133, 134, 136, 142, 173, 177, 194, 209]. Most studies reported >90% correct reader rate but this seemed to be dependent on the type of result (Fig 5). Inconclusive and weakly positive results commonly led to incorrect reading by the users.

Test performance

Despite issues of performance and test result interpretation, HIV-ST, as used and read by study participants, was found to have good (albeit variable) diagnostic accuracy. To better summarize the results, we pooled results using meta-analysis methods for diagnostic studies. In 17 studies [23, 24, 32, 40, 57, 79, 85, 125, 133, 134, 136, 142, 174, 175, 177, 207, 210], sensitivity ranged from 66.7% to 100%, while specificity ranged from 81.3% to 100%. Pooled results from testing unsupervised oral HIV-ST and found that it is specific (98.7%, 95%CI: 97.6% to 99.3%) and fairly sensitive (90.6%, 95%CI: 84.9% to 94.3%) (Fig 6). User-readings usually agree with standard tests. Eight of ten studies with data [32, 132, 134, 136, 142, 175, 194, 207] reported high (≥90%) concordance between results as read by users compared to reference testing (e.g., done by a healthcare worker or using standard test). Similarly, seven of nine studies with kappa statistics showed high agreement between user readings and standard results [24, 30, 70, 133, 142, 175, 207, 209].

Fig 6

Diagnostic accuracy of HIV-ST.

(A) Sensitivity (all studies), (B) Specificity (all studies), (C). Pooled Sensitivity and Specificity (Only studies using Unsupervised Oral HIV-ST). Notes: AUC–area under the curve, SN–sensitivity, SP–specificity, SROC–summary receiving operator characteristics, UN–unsupervised.

Finally, we found two studies that tested interventions to reduce user errors. A group in South Africa [211] developed a mobile app to aid HIV-ST users. While they reported only 8.7% user errors (at the low range compared to observational studies above), there was no control group to properly assess effectiveness. Another group performed a randomized trial involving Tanzanian youth [212] and they found that providing instructions using a graphic booklet and video was better than the graphic booklet alone. The booklet with video group had lower errors, better instruction comprehension, and higher intent to seek care.

Cost-effectiveness of HIV-ST

Despite higher program implementation costs, offering HIV-ST with standard testing can still be cost-effective (compared to standard testing alone). Maheswaran et al. [204] found that HIV-ST can be cost-effective in Malawi since it lowered societal cost by reducing patient opportunity costs. Cambiano et al. [219] found that adding HIV-ST can lead to cost-savings for the Zambian healthcare system. They also found that keeping the price of kits low ensured the cost-effectiveness of the program. Nichols et al. [218] found that aside from maintaining kit prices low, ensuring high linkage rates is also important to ensure that HIV-ST programs are cost-saving.

HIV-ST cost-effectiveness is also affected by the target population and HIV prevalence. Cambiano et al. [220] found that an HIV-ST program in sub-Saharan Africa is likely to be cost-effective for FSW and adult males but not for young people. Their results suggest that HIV-ST works best in communities where the prevalence is high and implementation costs can be maintained low. Korte et al. [156] echoes these findings in their CEA of distributing HIV-ST to males via their female partners in Uganda where they found that adding HIV-ST to existing testing programs would be best in areas in high prevalence.

The design and policies of the larger HIV program where HIV-ST is integrated also affect cost-effectiveness. For example, Maheswaran et al. [221] found that HIV-ST as a complementary service of standard testing can be cost-effective from a public provider perspective if positive individuals are managed following the 2015 WHO anti-retroviral treatment guidelines but not if the 2010 guidelines were used. Meanwhile, Johnson [222] showed that adding HIV-ST to offering a home-based HIV testing program in South Africa is cost-saving, but pairing HIV-ST with antenatal care partner testing programs is only cost-effective.

Penetration and sustainability

Penetration and sustainability are organization or setting-level constructs that refer to degree of institutionalization and maintenance of services. Both are late phase outcomes that require some implementation program to be in place prior to measurement. One measure of penetration is how many of the eligible users are using HIV-ST. Some studies that measured uptake could be considered as studies of this form of penetration [21, 37, 59, 82, 126, 170, 175, 176, 182, 192] with one study in Malawi using a more robust measure of population uptake by including questions in the national demographic health survey [37] rather than using convenience samples. We consider this as a passive type of penetration since HIV-ST uptake was not necessarily due to actual HIV-ST promotion programs and more attributable to market availability. We see in China (one of the countries where HIV-ST is widely available through e-commerce sites [223]), that the uptake rate of ever using HIV-ST can be at <20% [175] if measured outside of HIV-ST implementation programs.

There were also interventional studies that may affect penetration. We see from a cluster randomized trial of HIV-ST distribution by community health workers in Malawi [78] that community uptake takes some time but can reach high or maximal levels as long as the intervention is sustained. Further analysis of this trial’s data [172] demonstrated, however, that high community uptake might be driven by people already reached by conventional and cheaper HIV testing services. The authors point out the need for demand creation activities so underserved groups would also be reached. Relatedly, a trial in Zambia [171] showed that secondary distribution with household distribution may be useful for targeting men and those older than 30 years old but tracking outcomes and minimizing costs would be challenging. Finally, a cohort study in China [180] showed that initiating HIV-ST might lead to increased engagement in facility-based testing which demonstrates HIV-ST’s potential for improving the reach and long-term engagement with general HIV testing services.

We found only one study in Zimbabwe [125] that clearly described the organizational experience of routinization and scaling up of HIV-ST services and thus measures Proctor’s penetration and sustainability. Scaling up of HIV-ST services in the program occurred only after three years of formative assessment and implementation. The program involves distribution of HIV-ST kits in 7 clinics and 12 outreach sites for FSW. The report showed that sustained implementation of HIV-ST. FSW were able to use the kits for themselves or distribute to their partners and clients. They were able to use routine data to monitor HIV-ST uptake including how many take the test on-site and how many agree to do secondary distribution. They also found that with adequate support, FSW were able to use HIV-ST accurately and did not encounter issues of linkage to care for that testing positive. The program was also able to maintain activities to routinely obtain feedback from FSW and the feedback subsequently informed their next steps.

Linkage to care

There was a mix of approaches to measuring confirmatory testing. Most studies verified receipt of testing by either having all HIVST users undergo confirmatory testing as part of the protocol or by looking at clinic records with only a third (12/36, 33%) relying solely on report by study participants. From the studies with available counts, we found that rates of confirmatory testing were often high [23, 24, 29, 57, 61, 66, 68, 74, 77, 81, 84, 125, 133, 140, 145, 147, 156, 160, 162, 164, 168, 174, 178, 180, 207, 224, 225]. Only two studies [81, 144] reported low rates of confirmatory testing. One [144] was a trial of HIV-ST distribution through female partners in Kenya that showed only 25% (2/8 positives) of those with positive results in the HIV-ST intervention arm sought confirmatory testing compared to 75% (2/4 positives) confirmatory testing rate in the control arm. The other was a pilot trial of distribution of HIV-ST kits to male clients of brothels in Indonesia [81] where they detected two HIV positive results among the 188 who agreed to show their results but on follow-up they did not find evidence of receiving additional testing.

Similarly, most studies that reported linkage to care of confirmed positive cases found more than 50% linkage rates [23, 29, 57, 74, 77, 78, 84, 133, 144, 151, 152, 207] with only two reporting <20% linkage rates [81, 156]. One study with low rates [81] was the pilot study in Indonesia described above. The other was a trial of HIVST distribution via female partners in Uganda that reported low linkage of males who tested positive.[156] Most studies also reported 50% or higher rates ART initiation [24, 25, 29, 61, 84, 133, 140, 151, 152, 160, 162, 207, 224–226]; only two studies (in China [168] and in South Africa [174]), both of which were focused on assessing strategies to improve uptake among males, reported <30% initiation rates. Both studies had systems in place to provide counseling and follow-up to those who tested positive but still observed low rates.

Two interventional studies in Malawi and Zambia [152, 227] allowed comparing linkage rates of HIV-ST recipients to those under standard care. In both studies, linkage was lower in the HIV-ST arm confirming concerns raised previously described in the appropriateness section.

Safety and coercive testing

The occurrence of adverse events like coercive testing, self-harm after positive test results, and interpersonal violence seemed to be low across the studies that included these metrics [25, 47, 78, 85, 96, 97, 100, 125, 140, 145, 147, 150, 156, 162, 165, 169, 176, 178, 180, 181, 183, 188, 190, 192]. Some studies reported no serious adverse advents [78, 127, 144, 152]. Coercive testing was experienced by 2% to 9% of users with perpetrators being partners and healthcare workers and occurred via deception or threat of harm. While self-harm is a major concern based on qualitative studies, this does not seem to occur widely. An active community surveillance system in Malawi did not detect suicide due to HIV-ST [78] although a study in China reported that suicidal ideation and violence among positive testers was common [178]. Interpersonal violence (including sexual abuse) was an issue in distribution programs and was documented in qualitative and quantitative studies. Occurrence of this in HIV-ST distribution studies [47, 68, 78, 85, 156, 165, 169, 188] varied widely ranging from 0% to 10.5%. There were also other forms of harm reported in studies. In a pilot trial in Uganda [85], two of 19 distributors reported facing hostility from potential users. Marriage dissolution was raised by participants in a qualitative study in Malawi [95] but the report of the trial where the qualitative study was nested did not report occurrence of this event.[78] In a Zambian trial [145], there were 13 reported social harm events among 13,267 participants in the HIV-ST arm and these events included two cases of couple separation, three cases of emotional distress, and one cases of threat of suicide. In trial in Uganda [146], there were only two reported events among 632 HIV-ST users involving one case of verbal abuse and one case of emotional distress from a false positive result.