Protocol, rationale and design of SELPHI: a randomised controlled trial assessing whether offering free HIV self-testing kits via the internet increases the rate of HIV diagnosis

The United Nations (UN) 90–90-90 targets aim by 2020, that 90% of all people living with HIV (PLWH) are diagnosed, that 90% of people diagnosed with HIV are on ART, and that 90% of those on ART have a suppressed viral load [1]. The first target (90% diagnosis) remains the key challenge with global estimates of 47% of PLWH being unaware of their infection. Knowledge of one’s own HIV status and accessing ART benefits health on both an individual and population level. People who are unaware of their status are estimated to contribute disproportionally to new transmissions (between 60 and 80%) [2]. In the 2017 PHE (Public Health England) report on HIV, it was estimated that 10% of gay/bisexual men who have sex with men (MSM) were unaware of their HIV status. Although this percentage has decreased since 2010, testing is often less frequent than current recommendations. For example, UK guidelines [3] currently recommend annual HIV testing for MSM, and three-monthly testing for those considered ‘at higher risk’ (a definition that includes condomless (CL) anal sex with a new partner, diagnosis of new STI or chemsex drug use). In UK MSM, an estimated 28% have never tested for HIV and only a quarter of men at ‘higher risk’ of HIV infection test even 6-monthly (27%) [4‐6]. Late diagnosis of HIV also remains a problem in the UK; 32% of all MSM diagnosed with HIV in 2016 (663/2,096) had CD4 counts below 350 mm³, which is associated with greater morbidity and mortality than those who are diagnosed earlier in the course of infection. All guidelines now recommend that PLWH commence ART at diagnosis as this has been shown to be beneficial to individual health even at high CD4 counts [7].

HIV diagnoses in MSM in London have fallen since 2016 and this is thought to be a combination of increased rates of HIV testing, rapid initiation of ART when HIV is diagnosed (which reduces transmission risk through sex to almost zero) and increasing use of pre-exposure prophylaxis (PrEP) [8]. Rates of HIV diagnosis in MSM outside London have also reduced but to a lesser degree [9]. Expanding ways for MSM to test for HIV outside of traditional settings (such as GUM clinics) has been a focus for over a decade and there is now a national self-sampling service. This involves an individual taking their own test sample which they post back to the relevant laboratory for testing and are subsequently contacted with the result.

A further approach is to offer HIV self-testing (HIVST) where the person not only takes their own blood/saliva sample but also processes it themselves using a self-testing kit, and obtains the results immediately. A potential advantage of HIVST is that, removing structural and social barriers to testing and increasing associated privacy and convenience, may lead to increased testing [10, 11]. HIVST is also an opportunity for prevention synergies with the availability of PrEP (which requires frequent testing), and for harm reduction strategies, such as sexual partner screening. The WHO now incorporate HIVST into its global HIV testing guidelines as a ‘supplementary’ or ‘additional’ option [12] and has described it as “an empowering and innovative way to help achieve the first of the United Nations 90–90–90 treatment targets” [13].

HIVST also has a number of potential challenges, although empirical data are lacking. Firstly, a person who has a reactive HIVST requires confirmatory HIV testing to link to care which relies on the individual seeking more traditional testing as a gateway to care and support. It remains unknown what proportion of individuals who obtain a reactive result on HIVST link to care in a timely manner [14]. A further issue may be the potential for social and emotional harms from a reactive test in the absence of counselling, or coercion to test from a partner. HIVST may also be a missed opportunity for STI screening and advice about risk management due to fewer visits to GUM clinic settings, which may put MSM at increased risk of other STIs. HIVST kit accuracy is also an area of concern, as the window period is prolonged in comparison to 4th generation tests (antibody and P24 antigen test) and the sensitivity is relatively low, particularly with oral fluid HIVST in early infection or in breakthrough infections on tenofovir-based PrEP as antibody levels may be low. This is particularly important during early infection, when the risk of onward transmission is markedly increased.

Globally, SELPHI is the largest RCT evaluating the offer of free HIVST kits via the internet, and has uniquely been designed to assess the impact of this intervention on HIV diagnosis with linkage to clinical care and thus access to early ART.

Another moot design point was whether sending testing reminders should be an intrinsic part of the intervention or whether this should be considered as a separate intervention. We considered a factorial design but eventually decided that the HIVST kits and testing reminders should constitute a single “package”, reflecting the likely promotion of HIVST kits during implementation and to maximise the chance of demonstrating an effect. A residual concern is that the regular 3-monthly questionnaires will act as a reminder for participants in the nRT group to seek an HIV test and reduce the difference between the groups. However, we expect that a large proportion of participants in this group will opt out of receiving emails or ignore the follow-up surveys. This highlights the importance of determining the study primary endpoint from an independent national surveillance database to mitigate potential selective survey completion bias. A similar linkage to the PHE national dataset has been conducted to identify additional infections in the long-term follow-up of participants in the PROUD trial of HIV pre-exposure prophylaxis [28]. Reassuringly, of the 32 participants who were diagnosed in one of the PROUD clinics, only 1 (3%) was not identified in the national surveillance database [29]. The sample size calculation for SELPHI factored in a linkage failure rate of 10%. Another consideration is the delay in the centralisation and reconciliation of reports of HIV diagnoses from clinics and laboratories across the country, and a finalised dataset for a given calendar year is usually not available until June of the following year.

SELPHI is expected to complete in 2020 although the results of Randomisation A should be available earlier than this. As well as the main randomised comparisons, we are undertaking qualitative sub-studies of SELPHI participants, process evaluation and developing cost-effectiveness models which will be informed by the results of the trial.

With expanded prevention and treatment options available for people living with HIV, the need for testing and diagnosis is more important than ever. The reality of shrinking NHS resources and radical cuts and changes to sexual health service provision will require increased innovation with cost saving services such as e-health and postal services. This shift will increase the relevance of results from this trial.