Articles

Free HIV self-test for identification and linkage to care of

previously undetected HIV infection in men who have sex
with men in England and Wales (SELPHI): an open-label,
internet-based, randomised controlled trial
Alison J Rodger, Leanne McCabe, Andrew N Phillips, Fiona C Lampe, Fiona Burns, Denise Ward, Valerie Delpech, Peter Weatherburn,
T Charles Witzel, Roger Pebody, Peter Kirwan, Michelle Gabriel, Jameel Khawam, Michael Brady, Kevin A Fenton, Roy Trevelion,
Yolanda Collaco-Moraes, Sheena McCormack, David Dunn

Summary
Lancet HIV 2022; 9: e838–47 Background High levels of HIV testing in men who have sex with men remain key to reducing the incidence of HIV.
See Comment page e811 We aimed to assess whether the offer of a single, free HIV self-testing kit led to increased HIV diagnoses with linkage
Institute for Global Health to care.
(Prof A J Rodger FRCP,
Prof A N Phillips PhD, Methods SELPHI was an internet-based, open-label, randomised controlled trial that recruited participants via
F C Lampe PhD ,
Prof F Burns PhD, T C Witzel PhD)
sexual and social networking sites. Eligibility criteria included being a man or trans woman (although trans women
and UK MRC Clinical Trials Unit are reported separately); being resident in England or Wales, UK; being aged 16 years or older; having had anal
(L McCabe BSc, D Ward BSc, intercourse with a man; not having a positive HIV diagnosis; and being willing to provide name, email address,
M Gabriel MSc, Y Collaco-Moraes date of birth, and consent to link to national HIV databases. Participants were randomly allocated (3:2) by computer-
PhD, Prof S McCormack PhD,
Prof D Dunn PhD), University
generated number sequence to receive a free HIV self-test kit (BT group) or to not receive this free kit (nBT group).
College London, London, UK; Online surveys collected data at baseline, 2 weeks after enrolment (BT group only), 3 months after enrolment, and
National Infection Service, UK at the end of the study. The primary outcome was confirmed (linked to care) new HIV diagnosis within 3 months
Health Security Agency,
of enrolment, analysed by intention to treat. Those assessing the primary outcome were masked to allocation. This
London, UK (V Delpech FPHM,
P Kirwan BSc, J Khawam BSc); study is registered with the ISRCTN Clinical Trials Register, number ISRCTN20312003.
Department of Public Health,
London School of Hygiene & Findings 10 111 participants (6049 in BT group and 4062 in nBT group) enrolled between Feb 16, 2017, and
Tropical Medicine, London, UK
March 1, 2018. The median age of participants was 33 years (IQR 26–44 years); 9000 (89%) participants were White;
(P Weatherburn BSc); NAM,
London, UK (R Pebody MA); 8118 (80%) participants were born in the UK; 81 (1%) participants were transgender men; 4706
Department of Sexual Health (47%) participants were university educated; 1537 (15%) participants had never been tested for HIV; and 389
and HIV, King’s College (4%) participants were taking pre-exposure prophylaxis. At enrolment, 7282 (72%) participants reported condomless
Hospital, London, UK
anal sex with at least one male partner in the previous 3 months. In the BT group, of the 4511 participants for whom
(M Brady BM); Department of
Health and Social Care, London, HIV testing information was available, 4263 (95%) reported having used the free HIV self-test kit within 3 months.
UK (Prof K Fenton PhD); HIV Within 3 months of enrolment there were 19 confirmed new HIV diagnoses (0∙31%) in 6049 participants in the BT
i-Base, London, UK (R Trevelion) group and 15 (0∙37%) of 4062 in the nBT group (p=0∙64).
Correspondence to:
Prof Alison Rodger, Institute for
Interpretation The offer of a single, free HIV self-test did not lead to increased rates of new HIV diagnoses, which
Global Health, University College
London, London, NW3 2PF, UK could reflect decreasing HIV incidence rates in the UK. Nonetheless, the offer of a free HIV self-testing kit resulted in
alison.rodger@ucl.ac.uk high HIV testing rates, indicating that self-testing is an attractive testing option for a large group of men who have sex
with men.

Funding UK National Institute for Health and Care Research.

Copyright © 2022 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND
4.0 license.

Introduction progressing to global targets for elimination of HIV

In the UK, from 2015 onwards, HIV incidence in men transmission. In the UK, despite the decline in incidence,
who have sex with men (MSM) began a rapid decline.1–4 MSM remain at highest risk of HIV, and although testing
Combination prevention was central to this decline,1,4 rates have increased over the past decade, the number of
through expansion of HIV testing, early initiation of diagnoses that are late (CD4 count <350 cells per µL)
antiretroviral therapy (ART) after HIV diagnosis, and remains too high. An estimated 4200 MSM are living
increased use of pre-exposure prophylaxis (PrEP).5,6 with undiagnosed HIV in the UK.2 Early diagnosis of
Globally, knowledge of HIV status through testing HIV and initiation of ART also has population-level
remains key to attaining UNAIDS 95-95-95 goals7 and benefits because effective HIV viral suppression in

e838 www.thelancet.com/hiv Vol 9 December 2022

 Articles

Research in context
Evidence before this study Added value of this study
Before this trial, almost a quarter of men who have sex with The SELPHI trial remains the largest HIV self-testing trial
men (MSM) in the UK living with HIV were estimated to be implemented in a high-income setting and one that was also
unaware of their HIV infection and disproportionally fully internet-based. The SELPHI trial, and associated formative
contributing to onward transmission. Clearly, innovative HIV work, has generated a substantial body of evidence supporting
testing strategies were required. HIV self-testing had the development, implementation, and evaluation of HIV self-
the potential to increase initial and repeat testing rates due to testing among MSM and transgender people in high-income
confidentiality and convenience, and they needed to be settings. The trial addressed key questions regarding the
evaluated in a European setting. Formative qualitative potential of HIV self-tests for increasing HIV testing uptake
research done to inform the trial indicated a role for HIV self- without reducing STI testing or linkage to HIV care. For high
testing. We found that MSM preferred blood-based self-tests, prevalence settings, this trial provides useful evidence on how
delivered through the postal system and with clear links to self-testing offered through an internet-based platform could
further support included. Our 2020 meta-analysis comparing be used to increase HIV testing among MSM.
HIV self-testing to standard HIV testing approaches for key
Implications of all the available evidence
populations included ten randomised controlled trials, seven
Because the HIV response in high-income countries is
of which were done with MSM and small numbers of
increasingly focused on the elimination of new transmissions,
transgender people. We found that HIV self-testing increases
HIV self-testing plays a key role in expansion of HIV testing to
HIV testing uptake and frequency, without adverse effects on
facilitate timely HIV diagnosis and access to care. Questions
condom use. HIV self-testing detected greater numbers of
remain as to how HIV self-testing is best situated alongside
positive results than standard testing services, including mail
multiple testing opportunities before widespread
and online delivery approaches, for MSM and transgender
implementation and, in particular, how to support HIV self-
people. However, these data on positive HIV self-tests often
testing provision in a way that responds to existing inequities
relied on self-report, and often had low survey completion
(related to ethnicity, migration status, geography, gender and
rates. In addition, HIV self-testing led to worse linkage to care
sexual identity, health status, and digital literacy) while
for key populations overall, but these results were not
improving health.
statistically significant for MSM and transgender people.

people living with HIV eliminates sexual transmission to The motivations that influence testing behaviours of
other individuals.8 It is important that MSM are MSM, particularly those who engage in high-risk
diagnosed as soon as possible after acquiring HIV behaviours, are complex. However, one factor that affects
infection because up to 80% of all HIV transmissions are HIV testing is the influence of structural barriers to
estimated to derive from individuals who are obtaining a test, including time constraints or
undiagnosed.9 National and international guidelines for geographical distance to clinics. A further barrier to
recommended frequency of HIV testing in MSM are testing could be concerns about disclosure of sexual
similar in the USA, Australia, and the UK, where annual practices and sexual activity, and perceived stigma,16–20
HIV testing for sexually active MSM is recommended, particularly in MSM who do not identify as gay. With
and testing every 3–6 months is recommended for HIV self-testing, the person not only takes the sample
individuals who have frequent condomless anal sex, have but also immediately processes it themselves, so only
multiple sexual partners, or partake in sexualised drug they are aware of the result. Increased ease of access to
use.10–12 HIV testing is a key attribute of HIV self-testing, but
However, although the rates of HIV testing in MSM there is little evidence about whether increasing access to
living in the UK are increasing, they remain this modality increases HIV diagnosis rates in MSM.
suboptimal.13,14 In particular, rates of testing are low in The aim of the first stage of randomisation in the
MSM at increased risk of HIV infection through SELPHI randomised controlled trial was to assess if the
condomless anal sex with multiple partners. Our offer of a single free HIV self-testing kit to MSM led to
published analysis of baseline characteristics in the increased diagnosis of HIV infections and linkage to
SELPHI trial found that less than 60% of MSM who care. Data for trans women were collected as part of this
had two or more recent condomless anal sex partners trial; however, their data are reported separately.
had tested for HIV in the 6 months before enrolment.15
Factors associated with lower than recommended Methods
rates of HIV testing were lower levels of formal Study design
education and living in Wales or the northeast of SELPHI was an open-label, internet-based randomised
England, which indicates potential geographical controlled trial with a two-stage randomisation
barriers to testing.15 (appendix p 8) that has been described in full previously.21 See Online for appendix

www.thelancet.com/hiv Vol 9 December 2022 e839

 Articles

All trial processes, including recruitment, took place to the BT group and who met further eligibility criteria.
online, although participants had to be resident in Due to the large number of participants, a simple
England or Wales, UK, to receive delivery of the test kit. approach that randomly assigned participants using a
The main rationale for the first stage of randomisation completely automated random number generator was
was to evaluate the role of self-testing in detecting used. Given the nature of the intervention, participants
prevalent (possibly long-standing) HIV infections. In were not masked to intervention allocation. Those
this Article, we report the procedures and results of the assessing outcomes were masked to allocation, whereas
first stage of randomisation. Results from the second those analysing the data knew the group assignment.
stage of randomisation will be published separately. The
For the full study protocol see full study protocol is available online. Ethics approval Procedures
http://www.selphi.org/ was granted by the UCL Research Ethics Committee Participants in the nBT group were offered additional
application/
files/3516/4154/9192/SELPHI_
(REC Number 9233/001). information on how to undertake HIV testing through
Protocol_V4.0_17Dec19.pdf Extensive formative work was conducted before the routine services including how to access a nearby clinic
trial to explore the acceptability of HIV self-testing as part of standard of care. Participants randomly
among MSM and assess preferences for types of HIV assigned to the BT group were offered a free HIV self-
self-testing kits. The formative work informed the design test kit (BioSURE, Waltham Abbey, UK) immediately
of specific intervention components to boost engagement after randomisation. This kit incorporates an antibody
in care for individuals who received a reactive (ie, a immunoassay detecting HIV 1/2 antibodies (from
positive) HIV self-test result, and to provide risk- approximately 28 days after infection) and requires a
reduction information and signposting to HIV testing whole blood sample from a finger prick. The HIV self-
services for those randomly assigned to not receive a testing kits were posted directly to the participants by the
baseline HIV test. This formative work also informed the manufacturer.
design of supportive mechanisms within the intervention Participants who ordered a kit were contacted 2 weeks
to reduce the risk of adverse emotional reactions or other later asking whether they had received and used the kit,
types of harm following a reactive HIV self-test in the the result of the test, and, if reactive, if they had been to a
randomised controlled trial.17,18 clinic to have the result confirmed. Participants who
reported not receiving a kit or receiving a faulty kit were
Participants sent a replacement. All participants (in both the BT and
Participants were recruited through sexual and social nBT groups) received an online survey 3 months after
networking sites including Grindr, Hornet, Recon, enrolment that asked questions about testing for HIV
Scruff, and community Facebook webpages, using and other sexually transmitted infections (STIs) and
tailored advertising targeted to a broad spectrum of MSM about sexual behaviour since enrolment. Participants in
and transgender people.22 Eligibility criteria for the BT group were asked again about the use of the
participants included age 16 years or older; residence in self-test kit. If participants did not complete the online
England or Wales; being a man (including transgender survey that was sent at 3 months they were sent a
men) or transgender woman; having ever had anal reminder 2 weeks later. A final online survey was sent to
intercourse with a man; not being known to be HIV all participants between April 25 and May 9, 2019. This
positive; being willing to provide name, date of birth, and final survey largely asked the same questions as the
a valid email address; and providing consent to link to survey that was sent at 3 months, although with different
the UK national HIV surveillance databases held by timeframes. Additional questions about any potential
Public Health England. Very few transgender women harms from self-testing were also included but will be
were recruited and their data have been described reported elsewhere. Participants who completed the final
separately;23,24 these data are not included in the analysis survey were offered a free BioSURE self-testing kit.
presented in this Article. Data on gender were collected Survey invitations and responses were securely managed
through a self-reported survey at enrolment. Participants by Demographix, an online research company.
were asked the question “how do you describe yourself?”,
and the options for answers were: man, trans man, Outcomes
woman, trans woman, non-binary, or other. The primary outcome was a confirmed HIV diagnosis
within 3 months of enrolment, with date of diagnosis
Randomisation and masking defined as the date of the first confirmatory test at a
The first stage of randomisation took place at enrolment. clinic. Data on HIV diagnoses were primarily obtained
Following online consent, eligible participants were from linkage to national HIV surveillance databases
randomly allocated (in a 3:2 ratio) to the offer of a single (appendix p 2), which are maintained by Public Health
free baseline HIV self-test kit (baseline test [BT] group) or England. From Oct 1, 2021, Public Health England was
no offer of a baseline HIV self-test kit (no baseline test replaced by the UK Health Security Agency (UKHSA).
[nBT] group). The second stage of randomisation took Linkage was performed by UKHSA staff masked to the
place at month 3 and was open to participants allocated randomised allocation using a deterministic, hierarchical

e840 www.thelancet.com/hiv Vol 9 December 2022

 Articles

algorithm followed by a manual review of putative size, many statistically significant associations were
matches. Matches were classified as definite (exact found, even when the size of the effect was modest. It is
matching on several fields, including date of birth, and therefore more informative to focus on estimates and
no conflicting criteria) or partial (matching on some confidence intervals than on p values. The time to HIV
fields, but not all). Consistency between HIV diagnoses diagnosis was examined using a Kaplan-Meier plot. In
reported in follow-up surveys and those recorded in this analysis, participants randomly assigned to the offer
UKHSA databases was cross-checked. Participants who of reminders to complete HIV self-tests every 3 months
reported a positive self-test in the survey that was sent and free HIV self-test kits in the second stage of
out 2 weeks or 3 months after enrolment or in the final randomisation were censored on the date of this
survey who did not link with the UKHSA databases were randomisation, because this group were offered regular
contacted by a study clinician to verify that they had self-testing kits. Other participants were censored on
linked to care. Participants who withdrew from the trial March 31, 2019, by which time almost all HIV diagnoses
but did not ask for their data to be removed were included in the UK should have been reported to UKHSA.
in the UKHSA linkage. The primary analyses included Analyses were done with Stata (version 16.0). The
all UKHSA diagnoses (ie, definite or partial) and verified SELPHI trial is registered with the ISRCTN Clinical
self-reports. Trials Register (ISRCTN20312003).
Secondary outcomes, which were primarily assessed
using the survey sent 3 months after enrolment, were Role of the finding source
overall frequency of HIV testing irrespective of testing The funders had no role in study design, data collection,
modality, frequency of STI screening, and frequency of data management, data analysis, data interpretation, or
condomless sex. Additional information on the use of conduct of the study. The funders had no role in
kits and HIV testing in the BT group was obtained from preparation, review, or approval of the manuscript, or in
the survey sent 2 weeks after enrolment and from the the decision to submit the manuscript for publication.
data provided by UKHSA. Other secondary outcomes
were markers of recent of infection at the time of HIV Results
diagnosis (eg, CD4 count or antibody avidity assays) in 10 719 participants were randomly assigned to the BT or
participants for whom data were available, and diagnosis nBT group between Feb 16, 2017, and March 1, 2018
of a new STI. These other secondary outcomes are not (figure 1). Of those, 648 were later deemed ineligible
discussed in this paper because there were too few (figure 1), nine asked for all of their data to be withdrawn,
primary outcomes for these secondary analyses to be of and 24 were transgender women whose data are reported
interest, Furthermore, there was no difference in STI elsewhere.23,24 These exclusions left 10 111 participants in
testing between groups, which lessened interest in the analysis dataset. Of these participants, 6049 (60%)
findings on STI diagnoses. were allocated to BT and 4062 (40%) to nBT. 262 (3%)
participants subsequently withdrew or unsubscribed
Statistical analysis from further contact but were assessed for the primary
The target sample size of 10 000 was determined by the outcome.
number of self-test kits that could be acquired within the The baseline characteristics and associations with
study budget. To assess whether this sample size previous HIV testing behaviour in all participants
provided adequate statistical power we considered (randomised groups combined) have been described.16
plausible HIV seroprevalence values between 1∙5% and Median age was 33 years (IQR 26–44), 89% participants
2∙5% and various diagnosis rates in the BT and nBT were White, 80% born in the UK, and 47% university
groups.21 Statistical power was acceptably high (>90%) educated (table 1). Only 1% participants were transgender
when the absolute difference in diagnosis rates between men. 17% participants had completed HIV tests in the
the two groups was at least 30%. 3 months before enrolment, but 15% had never tested.
Analyses of the primary outcome were performed The most recent HIV test was at a sexual health clinic
using the intention-to-treat principle, including for 60% of participants, with a self-sample test for
participants in the BT group who did not order a kit when 17% of participants, a self-test for 7%, and other methods
offered and those randomly assigned to the nBT group for 16%. In terms of numbers of condomless anal
who accidentally received a kit due to a duplicate sex partners in the 3 months before enrolment,
enrolment. Participants were only excluded if they were 3330 (33%) participants reported only one partner,
determined to be ineligible after randomisation (figure 1) 2943 (29%) participants reported two to four partners,
or asked for all their data to be removed. Sensitivity and 1009 (10%) participants reported five or more
analyses of the primary outcome involving looser and partners. At the time of enrolment, 389 (4%) participants
stricter criteria for an HIV diagnosis were also performed. were taking PrEP. Baseline characteristics were
Comparisons of outcomes between randomised groups reasonably balanced over the two groups.
used χ² tests for categorical data and Mann-Whitney 3895 (64%) of 6049 participants in the BT group
U tests for ordinal data. Because of the very large sample completed the survey sent 2 weeks after enrolment. The

www.thelancet.com/hiv Vol 9 December 2022 e841

 Articles

survey 3 months after enrolment was completed by those who either completed the survey sent 2 weeks
4041 (67%) participants in the BT group and 1566 (39%) after enrolment (BT group only) or completed the survey
of the 4062 participants in the nBT group. Higher rates sent 3 months after enrolment, or who had a confirmed
of completion were identified for several baseline HIV diagnosis in this time period. 4263 (95%)
characteristics: participants who were older, those who participants in the BT group reported having used the
were better educated, and those who had tested for HIV SELPHI self-test kit. Another 105 participants reported
more recently (appendix pp 3–4). Additionally, race and accessing another HIV test (including three who used a
ethnicity and the number of condomless anal sex non-SELPHI self-test kit), giving a total of 4368 (97%) in
partners were associated with completion of this survey the BT group who had received any HIV test within
in the nBT group. The final survey was sent a median of 3 months of enrolment. This proportion is much higher
19 months (IQR 17–22) after enrolment, and was than that observed in the nBT group (670 participants
completed by 1695 (28%) participants in the BT group [43%]; table 2). Of the 3722 participants in the BT group
and 1069 (26%) participants in the nBT group. had used the SELPHI HIV self-test kit and responded to
5996 (99%) of 6049 participants in the BT group further questions, 892 (24%) reported having had an
accepted the offer of a free HIV self-test kit, but of these, additional HIV test within 3 months of enrolment.
224 (4%) participants reported not having received the Among the 1566 participants in the nBT group who
kit. Information on HIV testing within 3 months of completed the 3-month survey, the proportion with
enrolment was available for 4511 participants in the BT an HIV test between enrolment and 3 months
group and 1574 participants in the nBT group, based on (670 participants [43%]) was twice as high as the

20 019 individuals assessed for eligibility

5824 ineligible
347 not interested
545 had previous HIV diagnosis
1339 never had anal sex with a man
338 not a man, transgender man,
or transgender woman
61 younger than 16 years
1032 not living in England or Wales
317 planning to move from England
or Wales
311 did not consent
1438 duplicate email
96 error in survey processing
14 195 eligible

3403 did not complete enrolment

10 792 randomly assigned

4339 randomly assigned to receive no kit (nBT group) 6453 randomly assigned to receive self-testing kit (BT group)

257 randomly assigned in error 391 randomly assigned in error

205 duplicate enrolments 329 duplicate enrolments
11 did not provide a postcode 12 did not provide a postcode
41 previous HIV diagnosis in national database 2 reported previous HIV diagnosis
11 transgender women not included in this analysis 48 previous HIV diagnosis in national database
9 participants asked for complete data removal 13 transgender women not included in this analysis

4062 included in the analysis of the primary outcome 6049 included in the analysis of the primary outcome

221 withdrew or unsubscribed 41 withdrew or unsubscribed

Up to 3841 included in the analysis of the secondary outcomes Up to 6008 included in the analysis of the secondary outcomes

Figure 1: Trial flow diagram for the first stage of randomisation

Excluded participants could only be included in one ineligibility category or be categorised as having been randomly assigned in error. Due to the nature of the
ascertainment of the primary outcome (linkage with national databases), all randomly assigned participants are included in the analysis of the primary outcome.
Participants who had been randomised in error, who were transgender women, or who asked for complete data removal were excluded from this analysis.

e842 www.thelancet.com/hiv Vol 9 December 2022

 Articles

proportion who had an HIV test in the 3 months before

BT group (n=6049) nBT group (n=4062) Total (n=10 111)
enrolment (324 participants [21%]; p<0·0001).
For participants who reported taking a non-SELPHI Median age, years (IQR) 33 (26–44) 33 (26–44) 33 (26–44)

HIV test in the 3 months after enrolment, sexual health Gender

clinics were the most common location or type of testing Cisgender men 6002 (99%) 4028 (99%) 10 030 (99%)
in both the BT and nBT groups, although self-sampling Transgender men 47 (1%) 34 (1%) 81 (1%)
(ie, when a person collects their own test and then sends Country of birth
it to a laboratory for analysis) was also popular (table 3). UK 4849 (80%) 3269 (80%) 8118 (80%)
13% of participants in the nBT group and 7% of Other 1200 (80%) 793 (20%) 1993 (20%)
participants in the BT group who took a non-SELPHI test Highest educational qualification
in the 3 months after enrolment reported that their last University 2854 (48%) 1852 (46%) 4706 (47%)
HIV test was a non-SELPHI self-test. Higher education 645 (11%) 474 (12%) 1119 (11%)
Of the 4449 participants who reported having used the School 2264 (37%) 1536 (38%) 3800 (38%)
self-test kit in surveys sent 2 weeks, 3 months, or None 136 (2%) 104 (3%) 240 (2%)
approximately 19 months after enrolment (final survey), Data missing 150 (2%) 96 (2%) 246 (2%)
4378 (98%) participants obtained a non-reactive Race or ethnicity
result, 14 (<1%) participants obtained a reactive result, White 5347 (88%) 3653 (90%) 9000 (89%)
and 55 (1%) participants obtained no result (ie, no lines Asian 181 (3%) 126 (3%) 307 (3%)
appeared or there was another problem with the test). Of Black 100 (2%) 61 (2%) 161 (2%)
the 14 participants with a reactive result, ten participants Mixed 214 (4%) 99 (2%) 313 (3%)
either reported a positive confirmatory clinic result or Other, unknown, or undisclosed 207 (3%) 123 (3%) 330 (3%)
linked to the UKHSA database (implying a positive Time since last HIV test
confirmatory result), and one participant reported a <3 months 989 (16%) 706 (17%) 1695 (17%)
negative confirmatory clinic result. There were no 3–12 months 2250 (37%) 1471 (36%) 3721 (37%)
reported false negatives, although false negatives would >1 year 1813 (30%) 1248 (31%) 3061 (30%)
have been difficult to ascertain through the surveys. A Never 929 (15%) 608 (15%) 1537 (15%)
study clinician attempted to contact the remaining Data missing 68 (1%) 29 (1%) 97 (1%)
three participants, but was not successful. Median time Location of last HIV test*
between enrolment and linkage to care for participants Sexual health clinic 3030/5052 (60%) 2059/3425 (60%) 5089/8477 (60%)
reporting using the SELPHI self-test kit was 9 days
Other NHS or clinical setting 430/5052 (9%) 279/3425 (8%) 709/8477 (8%)
(IQR 6–12). One participant who reported a reactive self-
Self-sample 826/5052 (16%) 554/3425 (16%) 1380/8477 (16%)
test result did not link to care until 333 days after
Self-test 336/5052 (7%) 220/3425 (6%) 556/8477 (7%)
enrolment, and therefore did not meet the criteria for the
Elsewhere 340/5052 (7%) 242/3425 (7%) 582/8477 (7%)
primary outcome.
Data missing 90/5052 (2%) 71/3425 (2%) 161/8477(2%)
A total of 34 (0∙3%) of all 10 111 participants had a
Last STI test
confirmed HIV diagnosis within 3 months of enrolment
<3 months 896 (15%) 631 (16%) 1527 (15%)
(table 2). There was no evidence of a difference
3–12 months 1814 (30%) 1225 (30%) 3039 (30%)
between the two groups (p=0·64), with 19 (0∙31%) of
>1 year 2090 (35%) 1386 (34%) 3476 (34%)
6049 participants in the BT group versus 15 (0∙37%)
Never tested 1223 (20%) 797 (20%) 2020 (20%)
of 4062 participants in the nBT group diagnosed with
Data missing 26 (<1%) 23 (1%) 49 (<1%)
HIV (risk difference –0∙1%, 95% CI –0∙3 to 0∙2). This
finding was unchanged in sensitivity analyses Number of CAI partners in previous 3 months

(appendix p 5). An additional 51 participants had an HIV None 1716 (28%) 1112 (28%) 2828 (28%)

diagnosis after the 3-month time period (figure 2), again One 2000 (33%) 1330 (33%) 3330 (33%)
with no difference between the randomised groups. No Two to four 1745 (29%) 1198 (29%) 2943 (29%)
clear differences in baseline characteristics were evident Five or more 587 (10%) 422 (10%) 1009 (10%)
for participants with and without a confirmed HIV Data missing 1 (<1%) 0 (0%) 1 (<1%)
diagnosis (appendix p 6), including previous HIV testing PrEP use
history, although the number of events was too small for Currently taking PrEP 241 (4%) 148 (4%) 389 (4%)
statistical comparison. Not taking PrEP 5802 (96%) 3913 (96%) 9715 (96%)
The proportion of participants who had an STI test Data missing 6 (<1%) 1 (<1%) 7 (<1%)
between enrolment and 3 months was slightly lower in Data are n (%) unless otherwise specified. CAI=condomless anal intercourse. NHS=National Health Service.
the BT group than in the nBT group (table 2). In an PrEP=pre-exposure prophylaxis. STI=sexually transmitted infection. *Denominator is the number of participants who
exploratory analysis, we observed that in participants had previously tested for HIV for whom data were available.

who did not have any HIV test within 3 months of Table 1: Baseline characteristics of participants randomly assigned (in a 3:2 ratio) to the offer of a single,
enrolment, STI testing was uncommon in both the BT free HIV test at enrolment (BT group) or to no offer of a test at enrolment (nBT group)
group and the nBT group (table 4). In contrast, among

www.thelancet.com/hiv Vol 9 December 2022 e843

 Articles

participants who did have an HIV test, STI testing was condomless anal sex partners, the proportion reporting
more frequent in the nBT group than in the BT group. an STI test seemed higher than in participants overall
Among participants who had had HIV tests but not STI and was slightly higher in the nBT group than in the BT
tests, most had used a self-sampling or a self-testing kit, group (table 2).
including non-SELPHI tests (appendix p 7).
Participants in the BT group were slightly more likely Discussion
to report one or more condomless anal sex partners in In over 10 000 MSM enrolled in the SELPHI HIV self-
the 3 months after enrolment than those in the nBT testing trial, we observed no significant difference in
group (table 2). Of those reporting one or more HIV diagnoses between the men randomly assigned to
received self-testing kits and those who were not. The
BT group (n=6049) nBT group Risk difference p value low prevalence of new diagnoses probably reflects the
(n=4062) (95% CI) major national declines in HIV infections in MSM in the
Primary outcome UK, which occurred after the study was planned.2
Confirmed HIV diagnosis* 19/6049 (0∙31%) 15/4062 (0∙37%) –0∙1 % (–0∙3 to 0∙2) 0∙64 Rates of HIV testing were very high in participants
Secondary outcomes† assigned to receive an HIV self-test and who completed
Reported any HIV test 4368/4511 (97%) 670/1574 (43%) 54% (52 to 57) <0∙0001 a survey at 2 weeks or 3 months, with 97% reporting
Reported >1 HIV test 940/4368 (22%) 125/670 (19%) 3% (0 to 6) 0∙10 that they had done any HIV test within 3 months of
Reported use of SELPHI or 4266/4511 (95%) 89/1574 (6%) 89% (88 to 90) <0∙0001 enrolment (an HIV self-test in 95%). Of participants in
non-SELPHI self-test kit the BT group who had used the SELPHI self-test kit and
Reported any HIV test in 742/756 (98%) 222/325 (68%) 30% (25 to 35) <0∙0001 responded to further questions, a quarter reported
participants who tested having had an additional HIV test after the self-test,
<3 months before enrolment
which could indicate that they wanted further reassurance
Reported an STI test 903/4028 (22%) 397/1563 (25%) –3% (–5 to 0) 0∙018
of their negative HIV status. A further explanation for
Reported an STI test and one 663/2542 (26%) 281/927 (30%) –4% (–8 to –1) 0∙013
or more CAI partners
additional testing could also be that the SELPHI trial
Number of CAI partners ∙∙ ∙∙ ∙∙ 0∙010‡
itself had raised awareness of HIV testing, prompting
some participants to screen after a subsequent potential
None 1497/4039 (37%) 639/1566 (41%) –4% (–7 to –1)
exposure within the follow-up period. The proportion of
One 1292/4039 (32%) 451/1566 (29%) 3% (1 to 6) ··
participants who had tested for HIV (by any test) within
Two to four 931/4039 (23%) 328/1566 (21%) 2% (0 to 5) ··
3 months of enrolment was significantly higher in the
Five or more 319/4039 (8%) 148/1566 (9%) –2% (–3 to 0) ··
BT group (97%) than in the nBT group (43%). However,
Data are n/N (%) unless otherwise specified. In view of the unequal allocation to single, free HIV test at enrolment (BT among the 1566 participants in the nBT group who
group) or to no offer of a test (nBT group), comparison of percentages is more interpretable than absolute numbers.
p values were calculated using χ² tests. The denominators for each of the secondary endpoints differ because of missing completed the survey that was sent 3 months after
data or because only a subset of participants were analysed. CAI=condomless anal intercourse. STI=sexually enrolment, the proportion who had an HIV test between
transmitted infection. UKHSA=UK Health Security Agency. *Data on HIV diagnoses were obtained from linkage to baseline and 3 months (43%) was twice as high as the
national UKHSA surveillance databases: 29 participants (18 in BT group, 11 in nBT group) were a definite match, four
participants (one in BT group, three in nBT group) were a partial match, and one participant (none in BT group, one in
proportion of participants who had an HIV test in the
nBT group) had a verified self-report only. †Outcomes are for tests taken or partners reported by participants who 3 months before enrolment (21%). This finding suggests
completed the 3-month survey; denominators are those who answered the relevant question in that survey. The that participation in the SELPHI trial, even without an
2-week survey was only used for use of SELPHI test kit. UKHSA linkage used for primary outcome and HIV testing
offer of a free HIVST kit, might have led to increased
(confirmed cases only). ‡p value refers to comparison of no vs one or more CAI partners.
participant testing rates. As we found that HIV testing
Table 2: Outcomes at 3 months after enrolment in participants in the BT group who completed a follow-
up survey was close to 100%, the HIV diagnosis rate in
all participants randomised to the BT group (0∙31%) is
BT group nBT group Total
probably close to the true proportion of individuals with
Used SELPHI kit Did not use SELPHI kit undiagnosed HIV.2 Applying this infection rate to the
Took a non-SELPHI test within 892 124 661 1677 nBT group would result in approximately 13 expected
3 months of enrolment infections, which is lower than the 15 infections
Location or type of non-SELPHI test diagnosed in this group, suggesting that all, or almost
Sexual health clinic 486 (54%) 88 (71%) 287 (43%) 861 (51%) all, HIV infections in the nBT group were also
Other clinical setting 53 (6%) 4 (3%) 38 (6%) 95 (6%) diagnosed. One possible reason for the high rate of HIV
Self-sample 169 (19%) 13 (10%) 201 (30%) 383 (23%) diagnosis in the nBT group is that participants might
Self-test 66 (7%) 9 (7%) 89 (13%) 164 (10%) have enrolled in SELPHI because they were considering
Community service or setting 33 (4%) 4 (3%) 20 (3%) 57 (3%) testing for HIV, and when they were not randomly
Elsewhere or unknown 85 (10%) 6 (5%) 26 (4%) 117 (7%) assigned to receive an HIV self-test kit within the trial
Data are n or n (%). Participants in the BT group were offered of a single, free SELPHI HIV self-test at enrolment. (ie, they were assigned to the nBT group), these
Participants in the nBT group were not offered a test at enrolment. individuals decided to test elsewhere. All participants
randomly assigned to the nBT group were offered
Table 3: Location or type of last non-SELPHI HIV test within 3 months of enrolment
information on accessing HIV testing through the trial

e844 www.thelancet.com/hiv Vol 9 December 2022

 Articles

materials, which could have facilitated HIV testing 100 nBT group
elsewhere. BT group
1·5
However, with an estimated total population of
700 000 MSM in the UK,25 UKHSA estimates that there
are 1760 MSM with undiagnosed HIV, giving a diagnosis

Confirmed HIV diagnosis (%)

rate of 0∙25%.2 With an overall diagnosis rate in this
1·0
SELPHI trial of 0∙34%, this difference would imply that
we were not particularly effective in enrolling individuals
who were most likely to have an undiagnosed HIV
infection. It is well recognised that people who consent to 0·5
take part in a clinical trial are not necessarily representative
of the background population, with particular under-
representation of minority ethnic groups.2,26 Additionally,
log-rank p=0·40
because stigma is associated with both HIV and being 0
gay or bisexual in some populations, under-representation 0 3 6 9 12 15 18 21 24
in this trial might have been an even greater issue. One of Time since randomisation (months)
Number at risk
the main reported benefits of HIV self-testing is its nBT group 4062 4047 4042 4037 4035 3106 1637 420 296
BT group 6049 5704 4770 4765 4760 3666 1890 544 370
capacity for complete privacy compared with all other
modes of testing.16,18 Taking part in a trial that required Figure 2: Kaplan-Meier plot of time, in months, until confirmed diagnosis of HIV for participants in the BT
providing their name, address, and permission for group and those in the nBT group
The vertical dashed line at 3 months shows when the primary endpoint was analysed.
linkage to the UK surveillance database and contact from
the study team probably deterred some individuals
from taking part. Roll-out of HIV self-testing would have Completed Did not complete Total
to take in to account the effect of including notification of any HIV test any HIV test
linkage to care through the national surveillance database BT group n=3845 n=179 n=4024
against the need for individuals testing to have absolute Completed an 894 (23%) 9 (5%) 903 (22%)
privacy, confidentiality, and autonomy. STI test
As we have reported previously,15 MSM entering the Did not complete 2951 (77%) 170 (95%) 3121 (78%)
trial had a low rate of HIV testing in the 3 months before an STI test
enrolment. This rate was particularly low among men nBT group n=659 n=903 n=1562
with at least two condomless anal sex partners in the Completed an 369 (56%) 27 (3%) 396 (25%)
STI test
previous 3 months,15 who would be viewed as being at
Did not complete 290 (44%) 876 (97%) 1166 (75%)
higher risk of HIV infections and who should, according
an STI test
to current UK testing recommendations for MSM, be
Data are n (%). Participants in the BT group were offered of a single, free HIV self-
testing at least every 3 months. Despite this low rate of
test at enrolment. Participants in the nBT group were not offered a test at
testing, only 0∙31% in the BT group and 0∙37% in nBT enrolment. STI=sexually transmitted infection.
group tested positive for HIV in the 3 months after
enrolment. It is not clear why men entering the trial had Table 4: Joint analysis of HIV and STI testing between enrolment and
3 months for all participants
such a low undiagnosed prevalence of HIV. In the
formative work it was clear that MSM viewed HIV self-
tests as having limited utility when testing in response health systems by removing the need to attend the
to specific risk events (due to concerns about the window clinical facilities to test for HIV, thereby saving both
period and absence of immediate clinical support in the time and costs for patients and health-care providers.
event of a reactive result), except in the case of substantial Another potential use for HIV self-testing is that it
structural barriers to other testing opportunities. HIV allows for regular testing, which is necessary for men on
self-tests were considered to have use when seeking PrEP. However, the HIV self-tests that are currently
reassurance of ongoing HIV negative status and thought available have lower levels of sensitivity and specificity
to be useful if testing to satisfy norms and expectations than are required for this purpose.27 However, we also
of others (eg, peers, friends, or clinical staff) around found that participants in the BT group who did not use
regular testing.17,18 Therefore, men who took part in the the SELPHI HIV self-test kit had higher rates of
SELPHI trial could have accurately judged themselves to attendance at sexual health clinics and were less likely to
have been at low risk of HIV despite ongoing risk use self-sampling than participants in the BT group who
behaviours and might have enrolled in the trial to access did use the SELPHI HIV self-test kit. This outcome
HIV self-tests to confirm their low risk and to meet the implies that the participants in the BT group who did
normative expectations of their peer group. This not use the kit did not like the self-managed processes,
outcome also indicates a further key benefit of HIV self- which reinforces the need for provision of a range of
testing, which is that it has the potential to support services, including clinical face-to-face services, to

www.thelancet.com/hiv Vol 9 December 2022 e845

 Articles

enable patient choice. In addition, although there is a endpoint, and caution in interpretation is needed. The
benefit in being able to test outside sexual health direction of potential biases is difficult to predict,
services, the key loss with HIV self-testing is testing although there is probably a link between the likelihood
data, which has been effective in the UK context in of survey completion and positive health behaviours,
helping to understand shifts in patterns of infection. such as seeking out testing for HIV and STIs. In
It is clear from this trial and from other studies25 that addition, the requirement to provide personal details
most individuals who HIV self-test (even in high and agree to data linkage could have deterred some who
incidence populations) will test negative. Therefore, might have valued the complete anonymity of HIV self-
delivery of HIV self-testing interventions must be testing. A further potential area of concern is that the
designed to also facilitate uptake of HIV prevention SELPHI intervention was an online intervention, which
interventions, such as condom distribution and access to meant that participants needed to be sufficiently
PrEP, by individuals who test negative. Only 4% of digitally literate, health literate, and aware of HIV to
participants were using PrEP at enrolment in this trial, want to participate. Some people who are part of groups
which might have partly reflected the perceived that are disproportionately affected by HIV are likely to
limitations of self-tests for PrEP users; however, the be excluded from this kind of intervention. For example,
number of people using PrEP is likely to increase and another study showed that, even with optimised self-
any HIV self-testing implementation programme will sampling packs and processes, people with mild
need to build in access to such HIV prevention initiatives learning difficulties or low health literacy, or both, found
as part of the intervention. taking blood samples and using kits correctly to be
Consideration must also be given to encouraging challenging.30 Furthermore, we did not recruit many
uptake of STI testing in people who opt for HIV self- Black men or many transgender men, who are both at
testing because a frequently raised concern about the increased risk of HIV, which could make our results
provision of HIV self-testing is that it could lead to fewer less generalisable.
visits to sexual health clinics and reduced testing for There are also several key strengths of our study,
STIs. However, in our trial, participants in both BT and including the large sample size, the high degree of
nBT groups who completed a follow-up survey had acceptability of the intervention, and the linkage with the
similar rates of STI testing in the 3 months after national HIV database, which ensured that the primary
enrolment: in the BT group, the rate of STI testing was endpoint was a hard public health endpoint of linkage to
22% and in the nBT group, the rate of STI testing was 25%. care and not self-reported testing.
Future interventions must address both risks of HIV and In summary, we found that the offer of a single, free
other STIs and it is likely that the future of HIV self- HIV self-test at enrolment to MSM who enrolled in a
testing will be linked to self-testing for other STIs. large online self-testing randomised controlled trial did
HIV self-testing has further inherent challenges not lead to increased rates of confirmed diagnosis of
including linkage to care.28 However, in our trial, the prevalent HIV infections in a 3-month time period. The
median time between enrolment and linkage to care for absence of an increase in confirmed diagnoses of HIV
participants reporting positive results with the self-test could, however, largely reflect relatively low rates of
kit was only 9 days (IQR 6–12). The SELPHI intervention undiagnosed infections because of a decreasing
had been structured to ensure that very clear information incidence rate of HIV in the UK.1,2 Nonetheless, the offer
(developed through the formative work and more of a free test did result in a much higher rate of HIV
extensive than available commercial kits) was provided to testing within the 3-month time period, without a
participants with the HIV self-test kit about the necessary reduction in STI testing, indicating that HIV self-testing
next steps, including confirmatory testing if participants is an appealing HIV testing option for a large group of
tested positive on the HIV self-test.22,29 This emphasis on men.
clear information might have facilitated the rapid access Contributors
to linkage to care. AJR, ANP, FCL, SM and DD conceived the study and obtained funding.
There were several limitations to our study. The trial AJR drafted the manuscript and wrote the final version of the paper.
LM and DD analysed the data. VD, JK, and PK undertook the linkage of
participants were, by the fact they chose to enrol in the study data to national HIV surveillance databases maintained by
trial, interested in HIV self-testing. However, we UKHSA. PW, TCW, KF, RP, MG, JK, MB, RT, YC-M, FB, DW, PK, and
recruited a substantial proportion of men who, based on VD contributed to the study design, interpretation of the data, or writing
self-reported risk behaviours, were at a high risk of HIV of the paper, or a combination of all three. LM and DD take responsibility
for the integrity of the data and for the accuracy of the data analysis.
infection. Another limitation of the study is the low, but All authors reviewed and approved the final manuscript. All authors had
typical, completion rates of surveys. Survey completion full access to all the data in the study and had final responsibility for the
rates were especially low in the nBT group, in which decision to submit for publication.
only 39% completed the survey that was sent 3 months Declaration of interests
after enrolment. The low rate of survey completion MB has received speaker’s fees from Gilead and Bristol Myers Squibb.
is likely to have introduced bias into the analysis of FB and TCW have received consultancy fees from Gilead. All other
authors declare no competing interests.
the secondary endpoints, although not the primary

e846 www.thelancet.com/hiv Vol 9 December 2022

 Articles

Data sharing 14 Logan L, Fakoya I, Howarth A, et al. Combination prevention

The policy of the SELPHI Core Management Group is to make deidentified and HIV: a cross-sectional community survey of gay and bisexual
participant data available to any researcher who submits a scientifically men in London, October to December 2016. Euro Surveill 2019;
robust proposal, provided data exchange complies with information, 24: 1800312.
governance, and data security policies in all the relevant countries. Our 15 Rodger AJ, Dunn D, McCabe L, et al. Sexual risk and HIV testing
policy includes the replication of findings from published studies, although disconnect in men who have sex with men (MSM) recruited to
the researcher would be encouraged to work with the main author of the an online HIV self-testing trial. HIV Med 2020; 21: 588–98.
published paper to understand the nuances of the data. Enquiries should 16 Figueroa C, Johnson C, Verster A, Baggaley R. Attitudes and
be addressed to DW (denise.ward@ucl.ac.uk) in the first instance. acceptability on HIV self-testing among key populations: a literature
review. AIDS Behav 2015; 19: 1949–65.
Acknowledgments 17 Witzel TC, Weatherburn P, Rodger AJ, Bourne AH, Burns FM.
The SELPHI study was funded by the UK National Institute for Health Risk, reassurance and routine: a qualitative study of narrative
and Care Research under its Grants for Applied Research Programme understandings of the potential for HIV self-testing among men
(RP-PG-1212-20006). The views expressed in this paper are those of who have sex with men in England. BMC Public Health 2017;
the authors and not necessarily those of the UK National Health Service, 17: 491.
National Institute for Health and Care Research, or Department of 18 Witzel TC, Rodger AJ, Burns FM, Rhodes T, Weatherburn P.
Health. DD and SM were supported by the Medical Research Council HIV self-testing among men who have sex with men (MSM)
(MC_UU_12023/23). We acknowledge and thank all SELPHI study in the UK: a qualitative study of barriers and facilitators,
intervention preferences and perceived impacts. PLoS One 2016;
participants who generously participated in this research. We also
11: e0162713.
acknowledge the contributions of members of the PANTHEON
19 Flowers P, Riddell J, Park C, et al. Preparedness for use of the rapid
Programme Grant Steering Committee: Ann Sullivan (Chair),
result HIV self-test by gay men and other men who have sex with
Paul Fisher, and Nuala McGrath.
men (MSM): a mixed methods exploratory study among MSM
References and those involved in HIV prevention and care. HIV Med 2017;
1 Girometti N, Delpech V, McCormack S, et al. The success of HIV 18: 245–55.
combination prevention: the Dean street model. HIV Med 2021; 20 Flowers P, Knussen C, Li J, McDaid L. Has testing been
22: 892–97. normalized? An analysis of changes in barriers to HIV testing
2 Public Health England. Trends in HIV testing, new diagnoses and among men who have sex with men between 2000 and 2010 in
people receiving HIV-related care in the United Kingdom: data to Scotland, UK. HIV Med 2013; 14: 92–98.
the end of December 2019. 2020. https://assets.publishing.service. 21 Gabriel MM, Dunn DT, Speakman A, et al. Protocol, rationale and
gov.uk/government/uploads/system/uploads/attachment_data/ design of SELPHI: a randomised controlled trial assessing whether
file/959330/hpr2020_hiv19.pdf (accessed May 10, 2022). offering free HIV self-testing kits via the internet increases the rate
3 O’Halloran C, Sun S, Nash S, et al. HIV in the United Kingdom: of HIV diagnosis. BMC Infect Dis 2018; 18: 531.
towards zero 2030. 2019. https://assets.publishing.service.gov.uk/ 22 Witzel TC, Gabriel MM, McCabe L, et al. Pilot phase of an
government/uploads/system/uploads/attachment_data/ internet-based RCT of HIVST targeting MSM and transgender
file/965765/HIV_in_the_UK_2019_towards_zero_HIV_ people in England and Wales: advertising strategies and
transmissions_by_2030.pdf (accessed May 10, 2022). acceptability of the intervention. BMC Infect Dis 2019; 19: 699.
4 Brown AE, Nash S, Connor N, et al. Towards elimination of HIV 23 Witzel TC, Wright T, McCabe L, et al. Impact and acceptability of
transmission, AIDS and HIV-related deaths in the UK. HIV Med HIV self-testing for trans men and trans women: a mixed-methods
2018; 20: 74–76. subgroup analysis of the SELPHI randomised controlled trial and
5 McCormack S, Dunn DT, Desai M, et al. Pre-exposure prophylaxis to process evaluation in England and Wales. EClinicalMedicine 2021;
prevent the acquisition of HIV-1 infection (PROUD): effectiveness 32: 100700.
results from the pilot phase of a pragmatic open-label randomised 24 Wright T, Nicholls EJ, Rodger AJ, et al. Accessing and utilising
trial. Lancet 2016; 387: 53–60. gender-affirming healthcare in England and Wales: trans and non-
6 Molina JM, Capitant C, Spire B, et al. On-demand preexposure binary people’s accounts of navigating gender identity clinics.
prophylaxis in men at high risk for HIV-1 infection. N Engl J Med BMC Health Serv Res 2021; 21: 609.
2015; 373: 2237–46. 25 Marcus U, Hickson F, Weatherburn P, Schmidt AJ. Estimating
7 UNAIDS. 90-90-90, an ambitious treatment target to help end the size of the MSM populations for 38 European countries by
the AIDS epidemic. 2014. http://www.unaids.org/sites/default/ calculating the survey-surveillance discrepancies (SSD) between
files/media_asset/90-90-90_en.pdf (accessed on Dec 21, 2021). self-reported new HIV diagnoses from the European MSM internet
8 Rodger AJ, Cambiano V, Bruun T, et al. Risk of HIV transmission survey (EMIS) and surveillance-reported HIV diagnoses among
through condomless sex in serodifferent gay couples with the MSM in 2009. BMC Public Health 2013; 13: 919.
HIV-positive partner taking suppressive antiretroviral therapy 26 Erves JC, Mayo-Gamble TL, Malin-Fair A, et al. Needs, priorities,
(PARTNER): final results of a multicentre, prospective, observational and recommendations for engaging underrepresented populations
study. Lancet 2019; 393: 2428–38. in clinical research: a community perspective. J Community Health
9 Phillips AN, Cambiano V, Nakagawa F, et al. Increased HIV 2017; 42: 472–80.
incidence in men who have sex with men despite high levels of 27 Brady M, Rodger A, Asboe D, et al. BHIVA/BASHH guidelines on
ART-induced viral suppression: analysis of an extensively the use of HIV pre-exposure prophylaxis (PrEP) 2018. HIV Med
documented epidemic. PLoS One 2013; 8: e55312. 2019; 20 (suppl 2): s2–80.
10 Workowski KA, Bolan GA. Sexually transmitted diseases treatment 28 Witzel TC, Eshun-Wilson I, Jamil MS, et al. Comparing the
guidelines, 2015. MMWR Recomm Rep 2015; 64: 1–137. effects of HIV self-testing to standard HIV testing for key
11 Templeton DJ, Read P, Varma R, Bourne C. Australian sexually populations: a systematic review and meta-analysis. BMC Med 2020;
transmissible infection and HIV testing guidelines for 18: 381.
asymptomatic men who have sex with men 2014: a review of the 29 Witzel TC, Weatherburn P, Bourne A, et al. Exploring mechanisms
evidence. Sex Health 2014; 11: 217–29. of action: using a testing typology to understand intervention
12 British HIV Association, British Association for Sexual Health performance in an HIV self-testing RCT in England and Wales.
and HIV, British Infection Association. IV Association/British Int J Environ Res Public Health 2020; 17: 466.
Association for Sexual Health and HIV/British Infection Association 30 Middleton A, Pothoulaki M, Woode Owusu M, et al. How can we
Adult HIV Testing Guidelines 2020. 2020. https://www.bhiva.org/ make self-sampling packs for sexually transmitted infections and
file/5f68c0dd7aefb/HIV-testing-guidelines-2020.pdf (accessed on bloodborne viruses more inclusive? A qualitative study with people
May 10, 2021). with mild learning disabilities and low health literacy.
13 McDaid LM, Aghaizu A, Frankis J, et al. Frequency of HIV testing Sex Transm Infect 2021; 97: 276–81.
among gay and bisexual men in the UK: implications for HIV
prevention. HIV Med 2016; 17: 683–93.

www.thelancet.com/hiv Vol 9 December 2022 e847