This mixed-methods study follows a data integration approach termed by Moran-Ellis et al. as following a thread [32]. As such key areas of inquiry were identified from quantitative RCT data; these were then used to guide the focus of the analysis of the in-depth interviews. This has allowed us to generate additional nuance in responding to questions about feasibility and acceptability.
RCT study procedures
The pilot was designed to test the RCT recruitment strategy and the procedures in place for the full online trial. The pilot also tested the likely acceptability of the intervention, especially the usability of the chosen HIVST kit, its delivery mechanisms and the support offered for its use. Full details of the RCT methods can be found in the published protocol [33].
Eligible participants were men (cis and trans) and trans women; reporting lifetime anal sex with a man; not known to be HIV positive; aged 16 years and older; resident in England or Wales; willing to provide name, date of birth, postal and email address; consent to linkage with surveillance and clinic databases and not previously enrolled to the study.
The recruitment strategy utilised adverts placed in geo-location social-sexual networking applications (apps) (Grindr, Growlr, Scruff & Hornet) as well as targeted Facebook advertising. Free advertisements were placed on the Facebook page of a transgender focused clinical service. Recruitment sources were chosen based on previous experience, and through consultation with voluntary sector organisations. Grindr was chosen as it has the largest market share in the UK, with Hornet targeting a similar group. Growlr caters to a largely older sub-group of MSM, while Scruff is ostensibly most used by hirsute MSM and their admirers. Some adverts targeted a national audience, while others took a city or regional approach. Messaging was devised drawing learning from earlier formative work [11, 34], and with participant and public involvement (PPI) representatives. Key themes regarding barriers and facilitators to recruitment were identified, and two members of the study team met with PPI co-chairs to develop specific advertising messages. Adverts focused on all COM-B domains: capability was addressed through promoting ease of HIVST use; opportunity was addressed through highlighting the HIVST kits were available at no cost; and motivation was enhanced through highlighting privacy and appealing to altruism to take part in a study. Some messages specifically highlighted trans eligibility. Advertisements appeared as sponsored posts, as direct inbox messages, as pop-up messages and as banners.
Participants were directed to a registration survey requiring informed consent and confirming eligibility, and then to an enrolment survey via email. Ineligible participants (and those not randomised to HIVST) were offered additional information on HIV testing. The enrolment survey asked additional demographic and behavioural questions. Participants were randomised 60:40 to receive an HIVST kit (baseline test [BT]) or to no kit offer (no baseline test [nBT]). Kits were distributed by post, directly to the given address by the test manufacturer (BioSure™).
Two weeks after enrolment, participants randomised to receive the HIVST kit were emailed an online follow-up survey asking if the kit had been used (and if not, why not), what the test result was, and whether further care was accessed. Two reminders were sent.
Three months after randomisation a survey was emailed to all participants asking for information on testing and risk behaviour in the intervening period. Two reminders were sent. Participants randomised to BT were also asked questions about their experiences with HIVST. They ranked on a 5-point scale their agreement with statements related to acceptability and usability of the kit: 1) the instructions were easy to use; 2) performing the test was simple; and 3) my overall experience was good.
Intervention development
The intervention being trialled was linear. The recruitment messages being tested were both part of the intervention and trial process, but in a scaled-up intervention delivering HIVST these would be adapted accordingly. A brief HIV risk assessment was conducted through behavioural questions in the enrolment survey. The kit and accompanying sleeve were then delivered and two weeks later a follow-up survey asked about kit use and the test result. Those who reported not receiving a kit had a new delivery arranged. These components (advertisement, risk assessment, kit and two-week follow-up) were defined as the intervention as all were theorised to increase engagement with HIV testing through COM-B channels.
Formative work was central to intervention development. Focus groups with MSM and key informant interviews identified specific barriers to uptake and use of the HIVST which the SELPHI intervention development was attentive to ameliorating, using COM-B. These efforts were also used in developing appropriate messaging for advertising. Figure 1 provides a visualisation of intervention components with a description of the intervention functions and the COM-B domains they seek to affect.
Anticipated concerns regarding ease of use, coded in COM-B as capability (physical), were addressed in advertisements (see Fig. 2 for examples). This combined intervention approaches described as persuasion and education in COM-B [20], enhancing motivation by minimising concerns regarding ease of use and highlighting privacy and convenience. Issues concerning lack of knowledge in using HIVST were also identified. The COM-B model codes this as psychological capability; the behaviour change wheel then suggests intervention functions such as education, training and enablement might be useful [20]. This was alongside an observed preference for additional supportive information beyond what was provide in the original BioSure™ kit. This necessitated the development of a sleeve over the box holding the kit to provide support information (education), as well as behavioural support (enablement) which was highlighted in the two-week follow-up survey. The sleeve also provided signposting to a free telephone helpline and website for HIV and sexual health information (enablement).
In order to increase engagement with HIV testing generally, a risk assessment was included in the enrolment survey to provide a reflective experience examining personal risk. It was theorised that this approach (persuasion) can increase motivation (reflective and automatic) [20].
Formative research also identified issues with the instructions and packaging of an earlier iteration of the kit, both of which reduced motivation to access HIVST and capability when doing so. The kit instructions were reformatted by the manufacturer before implementation began, effectively addressing this issue. This intervention component is theorised as training (the imparting of skills) in the COM-B system [20].
The broader HIVST literature identifies support issues as a key concern in HIVST delivery, a concern also identified in our formative work [8, 11, 35]. This informed the provision of enhanced support information via our kit sleeve produced in collaboration with our community advisory group co-chairs (see Additional file 1). The two-week follow-up survey was also designed to counter this concern. If a participant reported a positive result here, they were directed to a page providing information on how to find their local HIV clinic. This same page was linked to from the three-month survey. Additional information about receiving a positive result was provided on the SELPHI website.
Data handling, generation & analysis
Data pertaining to advertising reach was recorded for all adverts, then pooled according to platform. The click conversion rate (proportion of those clicking on the advert who subsequently registered) was calculated. Eligible and ineligible registrations as well as the number of successful randomisations were tabulated. Registration conversion was calculated by deriving the proportion of eligible registrations who filled in the enrolment survey and were subsequently randomised.
Baseline demographic and behavioural profiles were tabulated overall and by recruitment source. Variables considered were age (both continuous and 10 year bands), gender, sexual orientation, ethnicity (recoded from standard UK ethnicity codes into white, Asian, black & other), highest educational qualification (low: GCSEs and below; medium: A-levels or equivalent, higher education below degree level; high: degree or higher), HIV testing history (tested in preceding 12 months; tested more than 12 months ago; never tested), and condomless anal intercourse (CAI) in the preceding 3 months. Participant demographic and behavioural characteristics were compared between recruitment sources using chi-squared tests or a Kruskal-Wallis test for age.
Responses to the 2-week survey were summarised by proportion who completed the survey, proportion who received the kit, and the proportion who subsequently used the kit.
Kit use was summarised again from the 3-month survey alongside acceptability variables pertaining to instructions, simplicity of test performance and overall experience.
Qualitative data
A qualitative study was undertaken with 10 cis-gender MSM participants during the pilot in order to examine intervention acceptability in greater depth. Participants were sampled purposively from those randomised to receive an HIVST kit. Sampling aimed to be diverse with regard to testing history: whether an individual had tested in the 12 months before joining SELPHI; not tested in the preceding 12 months; or never previously tested for HIV. Efforts were made to ensure sample diversity with regard to demographic features, especially education.
A semi-structured interview topic guide was developed to explore questions from formative research [11, 34], including issues related to capability, HIVST potential, anticipated responses and acceptability and mapped onto COM-B. The guide covered HIV testing history, motivations for joining and experiences of the SELPHI RCT, questions related to using HIVST and emotional responses.
Interviews were conducted over the phone or through Skype, and participants were electronically given a £30 incentive. All interviews were audio recorded and transcribed.
A thematic framework was developed for analysis, fusing the approaches described by Braun and Clarke [36] and Richie and Spencer [37]. This inductive process involved familiarisation with the transcripts and drawing out emerging themes. These themes were arranged into groups, with higher-level themes emerging from sub-themes, both organised hierarchically, and again mapped onto COM-B to better elucidate how acceptability of intervention components related to the behaviour change domains. The framework was piloted on two transcripts, refined, and applied to all remaining transcripts. We draw data from across this framework and report themes by COM-B domain for simplicity of interpretation.
Ethical approval for the RCT and qualitative sub-study were provided by MRCCTU and LSHTM (refs: 11945 & 9233/001). SELPHI is registered with the ISRCTN (ref: ISRCTN20312003). All RCT participants provided online written consent. Qualitative sub-study participants provided verbal recorded consent at the time of interview.