See Appendix 2 for completed Consolidated Standards of Reporting Trials (CONSORT) 2010 checklist for reporting of pilot or feasibility trials. Also see our protocol paper20 for a detailed description of planned methods.
Methods
Aim
The aim of the fRCT was to assess whether and how it would be possible to carry out a future definitive RCT of the Wrapped intervention.
Design
A two-arm, 1 : 1 allocation, parallel-group randomised feasibility RCT of Wrapped (intervention) plus usual care (basic information on STIs and condom use) compared to usual care alone (control).
Objectives
The primary objectives of the fRCT were to estimate the following parameters for planning a definitive RCT:
- The rate of recruitment of eligible participants
- The rate of participant follow-up for the definitive RCT primary outcome measure (chlamydia positivity at 12 months measured using biological samples)
The secondary objectives for the fRCT were as follows:
- Estimating the rate of participant follow-up for the definitive RCT secondary outcome measure chlamydia cumulative incidence
- Identifying whether the level of deprivation of the final sample is representative of online STI self-sampling users
- Identifying whether differential retention occurs across groups (gender, ethnicity, sexual identity, deprivation, randomised groups, chlamydia diagnosis at baseline)
- Estimating chlamydia positivity at 12 months in the intervention and control groups (to support sample size calculation for the definitive RCT)
- Estimating the rate of attrition at 3 months, 6 months and 12 months, and ways of minimising this
- Determining the feasibility and participant acceptability of all primary and secondary outcome measures (including health economic)
- Identifying which recruitment message(s) results in the highest rate of recruitment
- Identifying and removing intervention ‘friction points’ to minimise attrition and maximise future intervention dose
- Identifying costs and resource use associated with the intervention for healthcare services and users (to inform the design of a future definitive RCT with economic evaluation)
- Measuring contamination of intervention effect in the control group
- Identifying possible adverse effects of the intervention, for example, participants’ inadvertent disclosure that they are sexually active and/or testing for a STI, initiation or increase in the consumption of pornography
Deviation from protocol
The above list of objectives is an amended version. With agreement of our Data Monitoring and Ethics Committee (DMEC) and Study Steering Committee (SSC) (see Ethical approval and research governance), and approval of the National Institute for Health and Care Research (NIHR) and National Health Service (NHS) research ethics, a correction was made to the second primary objective (June 2022) to read ‘chlamydia positivity’ instead of ‘cumulative incidence of chlamydia’. New secondary objective also added: ‘Estimating the rate of participant follow-up for the definitive RCT secondary outcome measure chlamydia cumulative incidence’.
Ethical approval and research governance
Protocol for the fRCT was approved on 16 November 2020 by NHS East Midlands – Leicester Central Research Ethics Committee (reference 20/EM/0275), registered with the ISRCTN (ISRCTN17478654) and later published.20 An independent SSC and DMEC provided oversight and ensured adherence to standards of best practice. Research sponsorship provided by the University of Hertfordshire. Trial management and statistical oversight by Brighton and Sussex Clinical Trials Unit.
Public and patient involvement
The Wrapped PPI group (as described in Chapter 2) also supported delivery of the fRCT. Their roles included working with the research team to draft the participant information and consent statements to ensure that they were clear and easy to understand, and to iteratively test the surveys prior to use to maximise content validity, readability, clarity and comprehensiveness. Two members of the PPI group also supported monitoring of the fRCT by serving as members of the SSC.
Study population and setting
The study population included young people requesting a test for chlamydia via a web-based STI self-sampling service, namely Freetest.me or SH.UK (both operated by Preventx Ltd). Preventx provides a STI self-sampling service to 70 local authority areas across England (free at the point of use for most young people). Users residing in one of the five geographical areas (selected to represent English demography) were invited to participate, namely East Sussex, Kingston-Upon-Thames, Northamptonshire, Somerset and Warwickshire.
Eligibility criteria
People aged 16–24 years living in one of the above five geographical recruitment areas were eligible. People who satisfied these requirements were ineligible if they had no internet access or reported having sexual preferences which meant that they were unlikely to have penetrative sex (penis in vagina or anus) over the course of the study period.
Participant timeline
Participation was for a period of 12 months. See Table 3 for an overview of participant activities and timing.
| Activity | M0 | M3 | M6 | M12 |
|---|---|---|---|---|
| Read participant information | ✓ | |||
| Complete consent | ✓ | |||
| Complete survey | ✓ | ✓ | ✓ | ✓ |
| Self-report chlamydia test result | ✓ | |||
| Complete chlamydia self-sample | ✓ | ✓ |
To recompense for the time spent completing activities, participants received payment in the form of vouchers – see Appendix 3, Table 40 for the schedule of payments (refer to first wave).
Recruitment and randomisation procedure
The study advert was placed on the ‘thank you’ page of freetest.me/SH.UK viewed by users following an order of a chlamydia self-sampling kit (see Report Supplementary Material 8 – example advert). Every 2 weeks, the advert changed (see Chapter 2, Stage 4: Confirmation of strategies for the fRCT for the finalised set of adverts). A hyperlink took users to the study web page (located on REDCap, secure data capture and management platform hosted on University of Hertfordshire server)22,23 for participant information, eligibility assessment and consent. Each individual’s freetest.me/SH.UK order number was appended to the hyperlink uniform resource locator (URL) and captured by REDCap. All participants were required to provide informed consent before joining the study. Randomisation of participants to trial groups (1 : 1 allocation) was managed within REDCap on completion of the baseline (M0) survey. Stratification across groups, using randomly permuted blocks (ethnicity, sexual identity, deprivation), was performed to balance participants across the trial arms (see Appendix 3, Table 41 for groupings). Participants and those assessing outcomes were blind to allocation. Prior to randomisation, manual checks were performed to ascertain that each individual was unique by cross-checking identifying information (name, e-mail address, postal address, mobile phone number, freetest.me order number) with participants already in sample. Checks were also made to ensure that the baseline survey was answered with reasonable care (i.e. no evidence of random responding or completion under the expected minimum time). Those not meeting these criteria were not randomised.
Challenges with recruitment and deviation from protocol
Initial per protocol recruitment was planned from four areas offering freetest.me: Bromley, East Sussex, Somerset and Warwickshire. Just prior to starting recruitment, Bromley changed their STI self-sampling provider. Recruitment from the three existing areas was initiated while a replacement area was sought. At week 7, recruitment from East Sussex and Somerset was paused to enable service migration from freetest.me to SH.UK (unplanned migration in response to COVID-19). At week 13, recruitment began from a new replacement site, Kingston-Upon-Thames. At week 17, recruitment was resumed via East Sussex and Somerset. At week 28, recruitment began via a further replacement site Northamptonshire. To address slow recruitment, an amendment to the protocol was sought from the NIHR and NHS research ethics committee to increase total e-voucher payment from £65 to £85 (commenced week 8) and again to £100 (commenced week 11). See Appendix 3, Table 40 for revised schedule of payment (refer to second and third waves).
Measures
Survey completion (months 0, 3, 6 and 12)
All survey completion was via REDCap. Baseline (M0) survey collected information on: contact details (name, postal address, e-mail address, mobile phone number), demographic characteristics (age, sex, gender identity, ethnicity, sexual identity), recent STI diagnosis/treatment, sexual well-being, condom use, condom-use intentions, multilevel behavioural determinants targeted by the intervention, health-related quality of life [using the EuroQol-5 Dimensions, five-level version (EQ-5D-5L) and the Short Form questionnaire-12 items (SF-12)],24,25 and public sector resource use. Except for contact details and demographic characteristics, all items were repeated within surveys at 3 months (M3), 6 months (M6) and 12 months (M12). An additional item on adverse events was included within all follow-up surveys. The M3 survey included items to collect information on participants’ engagement with the intervention. The M12 survey included items to identify evidence of contamination between arms and to assess use of Wrapped products (intervention group participants only).
Completion of the M0 survey was scheduled for immediately after consent was provided. All other survey completion was triggered by e-mail. Participants not completing a survey were sent up to three reminders (delivered by SMS text message). If no response was received, one e-mail and then one telephone call was made to prompt completion. During the telephone call, participants were given the option to verbally respond to a reduced set of survey questions. If there was no response to the telephone call, a paper-based version of the survey was posted out with a stamped addressed return envelope. Alongside was a letter informing participants that as an alternative to completing the paper-based survey, they could access the online version using a Quick Response code provided or by searching for the link already sent to them via e-mail. Those not providing any data 20 days after the initial survey was sent out were classed as non-responders. Surveys continued to be sent to non-responders at subsequent time points unless a participant asked to be withdrawn from the study.
Chlamydia testing (months 0, 3 and 12)
Ten days after randomisation, participants were sent an SMS text message with a link to a single-item survey (on REDCap) to self-report the result of their baseline chlamydia test (the test requested by participant from Preventx immediately prior to study sign-up). Participants were asked to record the result, and any treatment received if applicable. The option to record that the self-sampling kit or result was ‘not yet received’ or ‘not yet returned’ was also available. This could be selected up to two times which triggered a repeat of the SMS text message 10 days later. A single SMS text message reminder was sent to participants who did not respond. Those not providing these data 4 weeks after the first SMS contact was made were classed as non-responders.
Chlamydia infection at M3 and M12 was measured using self-sampling kits sent to participants directly by the research team. One week before posting out the kits, participants were sent an SMS text message to confirm their postal address. Kits contained the usual user instructions to collect a sample and return by Freepost to Preventx for processing. Participants were sent up to three SMS text message reminders to complete and return a self-sample. If the self-sample was still not returned after this point, one e-mail and one phone call were used to prompt return. Those not returning their kit 30 days after posting out were classed as a non-responder. Non-responders were sent test kits at subsequent time points unless they asked to be withdrawn from the study. Participants with a positive test result at M0, M3 or M12 received one item within SMS survey 2 weeks post result to record whether infection had been treated. If treatment was incomplete, or the participants did not respond, they were sent one reminder SMS to obtain this information. Each local NHS trust was informed of any positive cases (for participants residing in their local authority area) so that appropriate treatment and follow-up could be provided.
Intervention access, content and tracking
Participants were directed to the intervention content by a hyperlink sent in an e-mail (e-mail triggered by completion of randomisation). All participants were free to interact as much or as little with the respective materials (participants assigned to the intervention condition were required to register and to login on repeat visits). An e-mail reminder to access the content was sent to participants who had not accessed the material after 2 weeks.
Wrapped website
See Table 4 for an overview of Wrapped. Also see logic model (see Report Supplementary Material 9) for theoretical rationale, and the development paper26 for a detailed description of intervention content reported in line with Template for Intervention Description and Replication (TIDieR) guidance.
| Component | Description | Tailoring |
|---|---|---|
| Sample pack | Box containing 12 condoms, two sachets of lube and condom-application instructions. Single order. Posted to user. | For everyone: to enable users to test out different types of condoms/lubes and identify preferred fit/feel |
| Condom ordering | A free discrete condom postal service. Bundles of 6 or 12 condoms plus up to three sachets of lube. Maximum one order per month. | For users who report finding it hard to access/afford condoms |
| Condom carrying | Choice of two small wallet designs with a discrete, hidden compartment for condoms/lube. Single order. Posted to user. | For users who do not routinely carry condoms |
| Condom demo | Video of step-by-step instructions on how to correctly put on a condom, giving tips on how to do this in a way that is enjoyable/pleasurable for self and partner. Unlimited access. | For users who are not confident putting on condoms or who find that condoms interrupt the flow of sex |
| Discussing condoms | A series of videos in which young people discuss talking to partners about condom use. Unlimited access. | For users who find condom use difficult to discuss |
| Real life | Videos of three real couples (two straight, one gay) discussing and demonstrating condom use. Unlimited access. | For those who report associating condom use with reduced pleasure or enjoyment (available to 18 + years only) |
Comparator website
Basic information on STIs and condom use (akin to that provided by NHS choices; www.nhs.uk/live-well/sexual-health/#have-safer-sex) was presented via a Wordpress website created for this study. Wrapped branding was used on the site to create an equivalent intervention ‘experience’. Identical information was duplicated within the Wrapped website.
Analytics data
Use of intervention and control materials was tracked using Matomo analytics package. A unique study identifier (ID) for each participant was appended to intervention URL to enable individual-level use to be measured.
Sample size
At least 60 participants per arm are recommended to estimate proportions with good precision.27 This was inflated to allow for estimated 25% non-return of initial chlamydia self-test sample (based on freetest.me statistics; 60/0.75 = 80) and a further estimated 30% dropout by 12 months (80/0.7 = 114 per arm).
Progression criteria
- Proportion of Preventx users recruited to the feasibility trial is sufficient to obtain the sample size required for the definitive RCT.
- 60% of participants (those randomised) followed up for the definitive RCT primary outcome measure at 12 months.
- Index of Multiple Deprivation quintile distribution for the final sample comparable to that of individuals in the sampling pool (freetest.me/SH.UK users in the partner local authority areas).
- Adverse events are judged as sufficiently infrequent and/or serious to cause concern.
Progression decision to be determined by the DMEC and SSC based on achievement of the criteria or convincing evidence that any one criterion was amenable to sufficient improvement.
Deviation from protocol
The above criteria are an amended version. The amendment was made in June 2022 with agreement of the DMEC/SSC and with approval of NIHR and NHS research ethics. Criterion 1 was amended to enable the target recruitment rate to be determined at study end. Criterion 2 was amended on the advice of the DMEC/SSC that 80% was too ambitious. Criterion 3 was amended as Preventx was only able to provide data on IMD quintile (as opposed to index of multiple deprivation [IMD] score) for the sampling pool. Criterion 4 was amended as the DMEC/SSC requested that the wording reflects the subjective nature of this assessment. See Appendix 4 for original criteria and further details on rationale for the changes.
Analysis
Planned descriptive analyses are presented below in line with each research objective (RO).
Primary outcome measures
- Percentage of participants recruited to the feasibility RCT (assessed as the number of participants randomised) out of those in the sampling pool (those shown the study advert via freetest.me/SH.UK) (RO 1).
- Percentage of participants with valid primary outcome measure (chlamydia positivity measured using biological samples) at 12 months (RO 2).
Secondary outcome measures
- Percentage of participants with outcome data required to measure cumulative incidence of chlamydia (measured using M3 and M12 chlamydia self-sample, and the relevant item within M3, M6 and M12 surveys) (RO 3).
- Distribution of IMD quintile among those in sampling pool compared to that of the final sample (RO 4).
- Percentage of participants with valid primary outcome measure at 12 months by group (gender, ethnicity, sexual identity, deprivation, randomised groups, chlamydia diagnosis at baseline) (RO 5).
- Percentage of participants with a positive chlamydia test result at 12 months in the intervention and control groups (RO 6).
- Attrition curves comparing the percentage of participants in the trial at M3, M6 and M12 plotted for the intervention and control arms (dropout attrition) (RO 7).
- Percentage completion of survey items required to measure primary and secondary outcomes at main trial (including self-report of chlamydia result at baseline, results from biological samples, demographic information, self-report of condom use, and data needed for cost-effectiveness and cost–utility analyses) (RO 8).
- Percentage of participants randomised as a direct result of each of the different adverts employed (RO 9).
- Percentage of participants in the two arms that do not achieve pre-determined intervention ‘goals’, and the bounce rate for home and content pages (RO 10).
- See Chapter 4 for planned health economic analysis (RO 11).
- Percentage of participants in the control group who report (at M12) any exposure to Wrapped (RO 12).
- Percentage of participants who report (at M12) having experienced an adverse event during the study (RO 13).
Deviation from protocol
In June 2022, with agreement of DMEC/SSC, and with approval of NIHR and NHS research ethics, it was agreed that all quantitative analysis should be performed and reported for the full sample (n = 230) and a ‘restricted sample’ where relevant (with comparisons made to examine whether these groups differed across any important demographic characteristics or fRCT outcomes). The full sample is defined as all randomised participants. The restricted sample is defined as randomised participants who report both (a) baseline chlamydia test result, and (b) if positive, whether the full course of prescribed antibiotics was taken (indicating that the individual was infection free). The decision to present data on both samples was made with the acknowledgement that participants at full trial would be required to meet these additional criteria (and therefore that the restricted sample provides a better approximation of the nature of the sample at full trial). The rationale for requiring these additional criteria to be applied at full trial was that data on baseline chlamydia positivity (primary outcome) would be needed to ensure a balance across groups, and that the number randomised and infection free at baseline would be needed to calculate chlamydia cumulative incidence (secondary outcome).
Results
Data were downloaded from REDCap and analysis performed using International Business Machines (IBM) Corporation Statistical Package for the Social Sciences (SPSS) 28 (SPSS Statistics for Windows, IBM Corp., Armonk, NY, 2022).
Recruitment took place over 31 weeks between 22 March and 18 October 2021. The study advert was displayed to individuals residing in the selected local authority areas. As outlined in Challenges with recruitment and deviation from planned protocol, the number of participating local authority areas varied throughout the recruitment period from one to five. In total, 11,413 individuals were presented with the study advert (either on freetest.me or SH.UK) over the recruitment period. How many of these clicked on the link and visited the project page is unknown as Preventx was unable to supply this data. Four hundred and eighty-eight of those shown the advert completed the eligibility questions. Of these, 36 (7.4%) did not meet the inclusion criteria and 12 (2.6%) reported that they were unable to commit to the study. Two hundred and ninety-four individuals went on to provide consent. Of these, 247 (84.0%) completed the baseline survey. Prior to randomisation, manual checks were performed to establish that each respondent was unique and had completed the baseline survey with sufficient care. Seventeen individuals were excluded at this point, leaving 230 available for randomisation. Recruitment stopped once the target of 230 participants was reached. In total, 115 participants were allocated to the intervention and 115 to the control (Figure 1).
FIGURE 1 Participant flow diagram. Reproduced from Newby et al.21 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure above includes minor additions and formatting changes. (Figure not reproduced here; see the original article.)
Withdrawals
Three participants withdrew from the study, two in the intervention arm and one in the control arm. None of these individuals requested that data they had provided up to that point be withdrawn. One of these participants self-reported the STI test result at baseline (negative result) and requested withdrawal 7 months after randomisation. Another self-reported the STI test result at baseline (negative result), returned the M3 and M6 surveys, and requested to be withdrawn 7 months after randomisation. The third participant withdrew 8 months after randomisation having provided no data other than at the M0 survey. All three participants requested withdrawal via e-mail in response to receiving a non-responder e-mail or voicemail message. In line with the study ethics protocol, participants were not asked to provide a reason for their decision, but one participant volunteered that, ‘it’s not something I can be a part of at this time’.
Reasons for losses to follow-up
Most losses to follow-up at the M3, M6 or M12 surveys were due to non-completion. Some contact problems were however experienced. One M6 survey was lost to follow-up due to an e-mail bounce back; this participant was sent a letter by post asking for an alternative e-mail address, but no reply was received. Surveys that remained incomplete following a series of e-mail prompts triggered postage of a paper copy. Two M12 surveys were known to be lost to follow-up due to incorrect postal address (these surveys were posted back to the research team by the occupants stating that the participant no longer lived at that address). Unlike with the paper surveys, chlamydia self-sampling kits were not posted out with a return address (deliberate strategy to ensure discrete delivery) and consequently there were no returns. The number of STI results that were lost to follow-up at M3 and M12 due to incorrect postal address is therefore unknown. Finally, five chlamydia self-sampling kits returned by participants for testing were identified as invalid at M3 (one due to degraded sample and four due to incomplete information submitted with the sample), and two at M12 (one due to degraded sample and one due to incomplete information submitted with the sample).
Demographics of the sample
The demographic characteristics of participants at baseline (full sample) are presented in Table 5. At randomisation, participants were stratified according to self-reported deprivation, ethnicity and sexual identity. See Appendix 3, Table 41 for categorisation of these variables for stratification purposes. The distribution of participants across the intervention and control groups by demographic characteristics appeared broadly balanced with three exceptions. The observed proportions suggested imbalances across groups for bisexual participants (lower proportion in the intervention group) and across two age groups (20 years: lower proportion in the intervention group; 21 years: higher proportion in the intervention group) albeit comparisons across broader age groups (16–19 years and 20–24 years) and median age indicated balance.
| Characteristic | Intervention (n = 115) | Control (n = 115) | Total (n = 230) |
|---|---|---|---|
| n (%) | n (%) | n (%) | |
| Age | |||
| 16 | 1 (0.9) | 2 (1.7) | 3 (1.3) |
| 17 | 2 (1.7) | 1 (0.9) | 3 (1.3) |
| 18 | 4 (3.5) | 7 (6.1) | 11 (4.8) |
| 19 | 5 (4.3) | 8 (7.0) | 13 (5.7) |
| 20 | 15 (13.0) | 26 (22.6) | 41 (17.8) |
| 21 | 27 (23.5) | 10 (8.7) | 37 (16.1) |
| 22 | 16 (13.9) | 21 (18.3) | 37 (16.1) |
| 23 | 21 (18.3) | 23 (20.0) | 44 (19.1) |
| 24 | 24 (20.9) | 17 (14.8) | 41 (17.8) |
| Median age (IQR) | 22 (2) | 22 (3) | 22 (3) |
| Age group | |||
| 16–19 | 12 (10.4) | 18 (15.7) | 30 (13.0) |
| 20–24 | 103 (89.6) | 97 (84.3) | 200 (87.0) |
| Ethnicity (stratification factor) | |||
| White | 98 (85.2) | 95 (82.6) | 193 (83.9) |
| Mixed or multiple ethnic groups | 6 (5.2) | 10 (8.7) | 16 (7.0) |
| Asian or Asian British | 1 (0.9) | 4 (3.5) | 5 (2.2) |
| Black, African, Caribbean or Black British | 8 (7.0) | 5 (4.3) | 13 (5.7) |
| Other ethnic group | 2 (1.7) | 1 (0.9) | 3 (1.3) |
| Gender | |||
| Female | 79 (68.7) | 86 (74.8) | 165 (71.7) |
| Male | 32 (27.8) | 27 (23.5) | 59 (25.7) |
| Non-binary or gender fluid | 3 (2.6) | 2 (1.7) | 5 (2.2) |
| Other | 1 (0.9) | 0 | 1 (0.4) |
| Gender not as assigned at birth | |||
| 4 (3.5) | 5 (4.3) | 9 (3.9) | |
| Sexual identitya (stratification factor) | |||
| Heterosexual | 79 (68.7) | 72 (62.6) | 151 (65.7) |
| Gay | 11 (9.6) | 4 (3.5) | 15 (6.5) |
| Bisexual | 23 (20.0) | 38 (33.0) | 61 (26.5) |
| Other | 2 (1.7) | 1 (0.9) | 3 (1.3) |
| Financial situation while growing up (stratification factor) | |||
| Very comfortable – I had everything I needed and more | 10 (8.7) | 8 (7.0) | 18 (7.8) |
| Comfortable – money was never an issue | 24 (20.9) | 26 (22.6) | 50 (21.7) |
| Fairly comfortable but it was necessary to keep an eye on money | 37 (32.2) | 35 (30.4) | 72 (31.3) |
| Things were OK but money was tight sometimes | 27 (23.5) | 29 (25.2) | 56 (24.3) |
| Money seemed to be a problem a lot of the time | 13 (11.3) | 13 (11.3) | 26 (11.3) |
| It was always a struggle to get the basics (food, clothes, heating) | 4 (3.5) | 4 (3.5) | 8 (3.5) |
| IMD quintile | |||
| 1 (most deprived) | 18 (15.7) | 16 (13.9) | 34 (14.8) |
| 2 | 19 (16.5) | 18 (15.7) | 37 (16.1) |
| 3 | 24 (20.9) | 31 (27.0) | 55 (23.9) |
| 4 | 29 (25.2) | 28 (24.3) | 57 (24.8) |
| 5 (least deprived) | 25 (21.7) | 22 (19.1) | 47 (20.4) |
| Relationship status | |||
| Single | 44 (38.3) | 45 (39.5) | 89 (38.9) |
| In a casual relationship with one person | 24 (20.9) | 21 (18.4) | 45 (19.7) |
| In casual sexual relationships with two or more people | 14 (12.2) | 13 (11.4) | 27 (11.8) |
| In a serious relationship with one person | 32 (27.8) | 29 (25.4) | 61 (26.6) |
| In a serious relationship with more than one person | 0 | 1 (0.9) | 1 (0.4) |
| Combining serious and casual relationships | 1 (0.9) | 4 (3.5) | 5 (2.2) |
| Other | 0 | 1 (0.9) | 1 (0.4) |
| Missing | 1 (0.4) | ||
See Appendix 5, Table 42 for demographic characteristics at baseline for the restricted sample. Here, the distribution of participants across groups also appeared broadly balanced with, as before, the exception of bisexual participants and those aged 20 and 21 years (same trends as observed for the full sample). In addition, there also appeared to be less females and more males in the intervention group than in the control group.
Primary objectives
RO 1: estimate the rate of recruitment of eligible participants
Over the 31-week (217 days) recruitment period, 230 participants (full sample) were recruited to the study, at a recruitment rate of 1.06 participants per day (230/217), representing 2% of the total sampling pool (230/11,413). For the restricted sample (i.e. those recruited who met our additional criterion of self-reporting their M0 chlamydia test result, and treatment if relevant), the recruitment rate was 0.80 participants per day (173/217), representing 1.5% of the sampling pool (173/11,413).
RO 2: estimate the rate of participant follow-up for the definitive RCT primary outcome measure (chlamydia positivity at 12 months measured using biological samples)
Table 6 presents the number and proportion of participants who completed and returned a valid chlamydia self-sample at M12 within both the full and restricted samples. Confidence intervals have been provided for the total sample. As is evident, the response rate was higher within the restricted sample.
| Intervention | Control | Total | |
|---|---|---|---|
| Full sample | |||
| Total number of participants in the sample | 115 | 115 | 230 |
| Number of participants who provided M12 self-sample | 81 | 70 | 151 |
| Percentage (95% CI) | 70.4– | 60.9– | 65.7(59.1 to 71.8) |
| Restricted sample | |||
| Total number of participants in the sample | 84 | 8 | 173 |
| Number of participants who provided M12 self-sample | 72 | 59 | 131 |
| Percentage (95% CI) | 85.7– | 66.3– | 75.7(68.6 to 81.9) |
Secondary objectives
RO 3: estimate the rate of participant follow-up for the definitive RCT secondary outcome measure chlamydia cumulative incidence
Cumulative incidence is defined as the proportion of a closed population at risk that develops the first occurrence of an event of interest within a given period of time.28 With respect to this study, the denominator for this calculation is the number randomised who are known to be infection free at baseline. The numerator is the number of this group who test positive at any point during the 12-month study period (as identified using M3 or M12 chlamydia self-sample, or the relevant item within M3, M6 or M12 surveys). Of those known to be infection free at baseline (n = 173), 114 participants completed all measures used to identify an instance of infection during the study period. The rate of follow-up for chlamydia cumulative incidence is therefore 65.9% (114/173). Note: Very few participants who completed M3, M6 and M12 surveys had missing data on the relevant items used to assess chlamydia infection (see RO 8). Absence of data is therefore largely due to non-completion of entire surveys or failure to return (valid) self-samples for testing (see RO 7 for response rates).
RO 4: identify whether the level of deprivation of the final sample is representative of online STI self-sampling users
To identify whether the level of deprivation of participants in the final sample at 12 months (determined as those who returned a chlamydia self-sampling kit at M12) was representative of service users in the wider sampling pool, the proportion of individuals in IMD quintiles 1 (most deprived) to 5 (least deprived) in each population was compared for both the full and restricted samples. These data are presented in tabular (Table 7) and visual (Figure 2) formats below. For completeness, data on the proportion of participants across these groups at baseline are also included in Table 7 and Figure 2.
| IMD quintile | Sampling pool (n = 11,413) | Baseline sample (n = 230) | M12 follow-up full sample (n = 151) | M12 follow-up restricted sample (n = 131) |
|---|---|---|---|---|
| N (%) | N (%) | N (%) | N (%) | |
| 1 (most deprived) | 1418 (12.4) | 34 (14.8) | 24 (15.9) | 21 (16.0) |
| 2 | 2136 (18.7) | 37 (16.1) | 22 (14.6) | 19 (14.5) |
| 3 | 3110 (27.2) | 55 (23.9) | 31 (20.5) | 29 (22.1) |
| 4 | 2770 (24.3) | 57 (24.8) | 36 (23.8) | 32 (24.4) |
| 5 (least deprived) | 1979 (17.3) | 47 (20.4) | 38 (25.2) | 30 (22.9) |
FIGURE 2 Bar chart displaying the proportion of individuals by IMD quintile within the sampling pool, the baseline sample, the full sample at M12 and the restricted sample at M12. (Figure not reproduced here; see the original article.)
These data indicated that the M12 full and restricted samples were broadly representative of the sampling pool across IMD quintiles with two exceptions. For IMD 3, it appeared that participants were less well represented at M12 than in the sampling pool. For IMD 5 (least deprived), there appeared to be a higher relative proportion of participants at M12 than in the sampling pool. In both cases, the difference between the proportion of participants at M12 and in the sampling pool was less pronounced for the restricted sample than for the full sample. These same patterns were reflected in comparisons of M12 data to baseline, that is, there appeared to be some evidence of a continued slight under-representation of IMD 3 and a slight over-representation of IMD 5. Again, these differences were less pronounced for the restricted sample than for the full sample.
Additional analysis was performed to compare the relative proportions of individuals in the sampling pool and at baseline with those in the full and restricted samples at M12 across age, gender, ethnicity and sexual identity (see Appendix 5, Tables 43–46 and Figures 6–9). For age, participants were grouped as 16–19 years and 20–24 years as inspection across individual ages indicated differing trends across these broader categories and grouping aided making comparisons. Within both the full and restricted samples, some differences in the relative proportions were observed as follows. Compared to those in the sampling pool, there appeared to be a lower relative proportion of 16- to 19-year-olds and a higher relative proportion of 20- to 24-year-olds at M12. Participants in these age groups at M12 did however appear to represent those at baseline. For gender, participants at M12 broadly represented those in the sampling pool and at baseline. For ethnicity, there appeared to be a slight under-representation of white participants, and a slight over-representation of black participants, at M12 compared to the sampling pool. Participants at M12 did however appear to represent the baseline sample across all ethnic categorisations. Participants at M12 appeared to represent participants at baseline across all sexual identity categorisations (Note: Comparison to the sampling pool for this characteristic was not possible as data on sexual identity for the sampling pool was unavailable).
RO 5: identify whether differential retention occurs across groups (gender, ethnicity, deprivation, sexual identity, randomised groups, chlamydia diagnosis at baseline)
Note: Examination of differential retention across demographic groups (comparison of participants at M12 to baseline) has already been reported above alongside findings relating to RO 4. This section continues our examination of differential retention, presenting findings related to retention by randomised groups and by chlamydia diagnosis at baseline.
The proportion of participants in each of the randomised groups retained in the study at M12 (measured as the proportion of participants in the initial baseline sample who went on to provide a valid chlamydia self-sample at M12) was calculated. Table 8 displays these data for both the full and restricted samples.
| Intervention | Control | Total | ||
|---|---|---|---|---|
| n/n (%) | n/n (%) | n/n (%) | 95% CI | |
| Lower limit | Upper limit | |||
| Full sample | ||||
| 81/115 (70.4) | 70/115 (60.9) | 151/230 (65.7) | 59.1 | 71.8 |
| Restricted sample | ||||
| 72/84 (85.7) | 59/89 (66.3) | 131/173 (75.7) | 68.6 | 81.9 |
As can be observed in Figure 3, overall retention appeared to be higher for the restricted sample than for the full sample, and higher for the intervention group than for the control group in both samples.
FIGURE 3 Overall retention by randomised group (full and restricted sample). (Figure not reproduced here; see the original article.)
To investigate differences between the intervention and control groups further, the relative proportions of participants at M12 across the randomised groups were compared for the different demographic characteristics. Table 9 presents this for the full sample (see Appendix 5, Table 47 for the restricted sample).
| N (%) with valid sample | Percentage difference | 95% CI | |||
|---|---|---|---|---|---|
| Intervention (n = 81) | Control (n = 70) | Lower limit | Upper limit | ||
| Age | |||||
| 16–19 | 6 (7.4) | 9 (12.9) | −5.5 | −15.2 | 4.2 |
| 20–24 | 75 (92.6) | 61 (87.1) | 5.5 | −4.2 | 15.2 |
| Gender | |||||
| Female | 56 (69.1) | 53 (75.7) | −6.6 | −20.8 | 7.6 |
| Male | 22 (27.2) | 16 (22.9) | 4.3 | −9.5 | 18.1 |
| Other | 3 (3.7) | 1 (1.4) | 2.3 | −2.6 | 7.2 |
| Ethnicity | |||||
| White | 69 (85.2) | 60 (85.7) | 0.5 | −11.8 | 10.8 |
| Mixed or multiple ethnic groups | 2 (2.5) | 4 (5.7) | −3.2 | −9.6 | 3.2 |
| Asian or Asian British | 1 (1.2) | 2 (2.9) | −1.7 | −6.3 | 2.9 |
| Black, African, Caribbean or Black British | 7 (8.6) | 3 (4.3) | 4.3 | −3.4 | 12.0 |
| Other ethnic group | 2 (2.5) | 1 (1.4) | 1.1 | −3.3 | 5.5 |
| IMD quintile | |||||
| 1 (most deprived) | 12 (14.8) | 12 (17.1) | −2.3 | −14.0 | 9.4 |
| 2 | 11 (13.6) | 11 (15.7) | −2.1 | −13.4 | 9.2 |
| 3 | 15 (18.5) | 16 (22.9) | −4.4 | −17.4 | 8.6 |
| 4 | 22 (27.2) | 14 (20.0) | 7.2 | −6.3 | 20.7 |
| 5 (least deprived) | 21 (25.9) | 17 (24.3) | 1.6 | −12.3 | 15.5 |
| Sexual identity a | |||||
| Heterosexual | 54 (66.7) | 41 (58.6) | 8.1 | −7.3 | 23.5 |
| Gay | 8 (9.9) | 4 (5.7) | 4.2 | −4.3 | 12.7 |
| Bisexual | 17 (21.0) | 25 (35.7) | 14.7 | −29.0 | 0.0 |
| Other | 2 (2.4) | 0 (0) | 2.4 | −1.0 | 5.7 |
For both the full and restricted sample, the distribution of participants across demographic characteristics appeared broadly balanced with some limited exceptions. These exceptions mirrored those observed at baseline, namely a lower proportion of bisexual participants in the intervention group than in the control group (full and restricted samples), and a lower proportion of females and a higher proportion of males in the intervention group than in the control group (restricted sample only).
Also examined was the retention of participants by chlamydia self-report at baseline (see Table 10). This showed that retention among participants who reported positive and negative results at baseline was similar. Retention was observed as markedly lower among those who did not report a result at baseline; this further exemplifies the rationale for excluding this group at baseline.
| Test result | Participants at baseline with each test result | Participants with each result at baseline who went on to provide a valid chlamydia self-sample at M12 | ||
|---|---|---|---|---|
| N | N (%) | Lower 95% confidence limit | Upper 95% confidence limit | |
| Positive | 8 | 5 (62.5) | 24.5 | 91.5 |
| Negative | 165 | 126 (76.4) | 69.1 | 82.6 |
| Not reported | 57 | 20 (35.1) | 22.9 | 48.9 |
| Total | 230 | 151 (65.7) | 59.1 | 71.8 |
RO 6: measure chlamydia positivity at 12 months in the intervention and control groups (to support sample size calculation for the definitive RCT)
Table 11 presents data on test positivity at M12, broken down by intervention and control group, for both the full and restricted samples.
| Intervention | Control | Total | |
|---|---|---|---|
| Full sample | |||
| No. in sample | 115 | 115 | 230 |
| No. of positive cases reported | 2 | 3 | 5 |
| Test positivity | 1.7 | 2.6 | 2.2 |
| Restricted sample | |||
| No. in sample | 84 | 89 | 173 |
| No. of positive cases reported | 1 | 3 | 4 |
| Test positivity | 1.2 | 3.4 | 2.3 |
Sample size calculation for definitive RCT
As indicated in the title of RO 6, our plan had been to use the derived estimates of chlamydia positivity in the intervention and control group to support a sample size calculation for a future full trial. In practice, with the agreement of our SSC and DMEC, other sources were prioritised. This was because we lacked confidence in the estimates obtained due to the small sample size (n = 173) and the COVID-19 context of the study, which may have had a downwards impact on population levels of chlamydia positivity (see the Discussion section for commentary on the possible impact of the COVID-19 pandemic on the observed levels of chlamydia positivity). Instead, extensive consultation and document review were used to estimate the parameters necessary for the sample size calculation as follows (see Report Supplementary Material 10 for progression criteria report which includes further details on this process. Note: Changes in national data on chlamydia positivity across test settings have been observed since the progression report was produced, and the sample size calculation presented below has been updated to reflect the most recent 2023 data). Chlamydia positivity in the control group is estimated at 10% based on data reported for the year 2023 in UK Health Security Agency STI data tables for 16- to 24-year-olds testing for chlamydia via the internet.29 We consulted multiple sources to determine the target level of chlamydia positivity for the intervention group. These included (1) existing evidence concerning chlamydia positivity among young people in the general population (range estimates at 4–6%30–32), (2) level of chlamydia positivity among young people testing via community-based general practices (6%29) and (3) consultation with 13 sexual health commissioners and sexual health service managers across five English regions about what level would reflect a clinically meaningful reduction from that in the control group (consensus was between 4% and 6% in the intervention group, i.e. a difference of between 2% and 4% between the groups). Based on advice from an independent expert group, we decided that the study should be powered to detect a difference of 3.5% (i.e. 10% in control arm and 6.5% in the intervention arm). A power calculation using G*Power estimated that 2706 participants would be required to detect this difference, thus requiring a trial sample of 3574 participants [based on 75.7% retention (see RO 2 above); 2706/0.757]. With a recruitment rate of 1.5% (see RO 1 above), a sampling pool of 238,266 users is required to achieve this target (3574/0.015). The number of 16- to 24-year-olds using Preventx for chlamydia self-sampling varies annually, but the most recent data put this figure at 585,000 users (personal communication) indicating that this would provide a sufficiently large sampling pool at full trial to make recruitment over 12 months a feasible proposition.
RO 7: identify the rate of attrition at 3 months, 6 months and 12 months, and ways of minimising this
Unless a participant withdrew from the study, he or she continued to be invited to complete research measures regardless of his or her levels of prior completion. One hundred and thirteen participants (49.1%) completed all seven research measures. Attrition was explored by examining the proportion of participants who completed each of the study measurements. As can be seen in Table 12 and Figures 4 and 5, completion was higher for the surveys than for the chlamydia self-sampling kits, and furthermore completion was higher overall for participants in the restricted sample than in the full sample.
| Measure | Proportion of participantsa completing study measure | |
|---|---|---|
| Full samplen/N (%) | Restricted sample n/N (%) | |
| M0 survey | 230/230 (100) | 173/173 (100) |
| Self-report result of chlamydia test | 173/230 (75.2) | 173/173 (100) |
| M3 survey | 192/230 (83.5) | 158/173 (91.3) |
| M6 surveyb | 191/230 (83.0) | 157/173 (90.8) |
| M12 survey | 192/227 (84.6) | 157/171 (91.8) |
| Valid chlamydia sample at M3 | 148/230 (64.3) | 130/173 (75.1) |
| Valid chlamydia sample at M12 | 151/227 (66.5) | 131/171 (76.6) |
FIGURE 4 Bar chart displaying proportion of participants who completed each study measure (full sample). (Figure not reproduced here; see the original article.)
FIGURE 5 Bar chart displaying proportion of participants who completed each study measure (restricted sample). Note: Attrition is not presented using attrition curves as planned. As participants were invited to complete all measures (regardless of prior completion), this skewed the presentation of results. (Figure not reproduced here; see the original article.)
Measure acceptability was also discussed during the nested qualitative study. Participants who also took part in qualitative interviews found the measures acceptable (see Chapter 5, Subtheme 2.1: process of joining the study straightforward and as expected).
RO 8: determine the feasibility and participant acceptability of all primary and secondary outcome measures (including health economic)
Data presented above on the overall completion of research measures (RO 7) provide an indication of participant acceptability. Missing data on survey items that would be used in a definitive trial to evaluate primary and secondary outcomes were also assessed as a further measure of acceptability. There were no missing data for any of the items collecting demographic information; participation in the study was dependent on completion of the baseline survey and these items were set as mandatory fields. Table 13 displays the number and percentage of participants who completed each survey measure who had missing data on the remaining items. All survey items had fully complete or near-complete data, with the latter defined as < 5% missing data.
| Item | n/N (%) of participantsa with missing data | |
|---|---|---|
| Full sample | Restricted sample | |
| Chlamydia infection | ||
| Self-report of any chlamydia infection (including whether treated, if applicable) within M0 survey | 0/230 (0) | 0/173 (0) |
| Self-report of any chlamydia infection (including whether treated, if applicable) within M3 survey | 0/192 (0) | 0/158 (0) |
| Self-report of any chlamydia infection (including whether treated, if applicable) within M6 surveyb,c | 0/191 (0) | 0/157 (0) |
| Self-report of any chlamydia infection (including whether treated, if applicable) within M12 survey | 0/192 (0) | 0/157 (0) |
| Condom use | ||
| Self-report of condom use at M0 | 1/230 (0.4) | 1/173 (0.6) |
| Self-report of condom use at M3 | 1/192 (0.5) | 1/158 (0.6) |
| Self-report of condom use at M6b,d | 0/191 (0) | 0/157 (0) |
| Self-report of condom use at M12 | 0/192 (0) | 0/157 (0) |
| Cost-effectiveness and cost–utility data | ||
| EQ-5D-5L at M0 | 0/230 (0) | 0/173 (0) |
| EQ-5D-5L at M3 | 3/192 (1.6) | 3/158 (1.9) |
| EQ-5D-5L at M6e | 3/191 (1.6) | 2/157 (1.3) |
| EQ-5D-5L at M12 | 2/192 (1.0) | 2/157 (1.3) |
| SF-12 at M0 | 2/230 (0.9) | 2/173 (1.2) |
| SF-12 at M3 | 3/192 (1.6) | 3/158 (1.9) |
| SF-12 at M6f | 8/191 (4.2) | 6/157 (3.8) |
| SF-12 at M12 | 2/192 (1.0) | 2/157 (1.3) |
RO 9: identify which recruitment message(s) results in the highest rate of recruitment
As described in the Methods section, participants were recruited to the study via adverts displayed to Preventx service users, each of which contained a different set of value propositions. Over the course of the recruitment period, 13 different adverts were presented. Table 14 presents the proportion of Preventx users that was randomised to the study following the presentation of the different messages. Given that the incentive amount offered to participants changed over the course of recruitment, this information is also provided.
| Advert number | Advert text | Length of time advert displayed (in weeks) | Incentive amount as communicated in advert and/or participant information (£) | Percentage (n/N) of service users recruited following presentation of advert |
|---|---|---|---|---|
| 1 | Help us make a positive change to young people’s sexual health: make a difference by joining our study and get paid for your time! | 2 | 65 | 1.9 (19/954) |
| 2 | Take part in a study to improve sexual health and earn up to £65 in Amazon vouchers, find out how you can make a difference today! | 2 | 65 | 2.0 (19/941) |
| 3 | Your views are important: join our study to help improve sexual health for young people while being paid! | 2 | 65 | 1.8 (17/941) |
| 4 | Make a difference to improve sexual health for young people – join our study and earn up to £65 in Amazon vouchers for your time! | 1 | 65 | 3.5 (9/254) |
| 5 | Interested in sexual health research? Take part and get up to £85 in vouchers. | 3 | 85 | 3.5 (20/577) |
| 6 | Help us make a positive change to young people’s sexual health: make a difference by joining our study and get up to £100 in vouchers. | 2 | 100 | 1.1 (5/460) |
| 7 | Take part in a study to improve sexual health and earn up to £100 in vouchers, find out how you can make a difference today. | 2 | 100 | 1.6 (7/446) |
| 8 | Your views are important: join our study to help improve sexual health for young people and get paid up to £100 in vouchers! | 2 | 100 | 2.6 (13/491) |
| 9 | Make a difference to improve sexual health for young people – join our study and earn up to £100 in vouchers for your time. | 3 | 100 | 2.0 (25/1282) |
| 10 | Want to make a difference to young people’s sexual health? Take part in our study and get up to £100 in vouchers for your time. | 2 | 100 | 1.6 (15/922) |
| 11 | We need your help! Take part in a study to improve young people’s sexual health and earn up to £100 in vouchers for your time. | 2 | 100 | 0.9 (8/944) |
| 12 | Interested in sexual health research? Take part and get up to £100 in vouchers. | 3 | 100 | 2.1 (28/1347) |
| 13 | Shown to service users aged 20–24:Interested in sexual health research? Take part and get up to £100 in vouchers. | 3 (plus 1 day) | 100 | 2.8 (39/1401) |
| Shown to service users aged 16–19:Interested in sexual health research and aged 16–19 years? Want to earn up to £100 in Amazon vouchers? | 1.3 (6/ 453) |
Also see Appendix 5, Table 48 for recruitment by age group for each combination of message/incentive offered. As can be seen from both tables, there was no clear pattern in terms of which voucher amount(s) or advert(s) displayed resulted in higher recruitment.
Also see findings from the nested qualitative study (Subtheme 2.1) views on the adverts used.
RO 10: identify and remove intervention ‘friction points’ to minimise attrition and maximise future intervention dose
Please note, in this section, data on the full sample are presented.
Engagement with the intervention and control material was measured using data from Matomo analytics, surveys, and the product ordering and dispatch logs held within the Wrapped website. For a small number of participants in the intervention condition (4/79; 5%), website engagement data reported by Matomo analytics were inconsistent with the logs held within Wrapped. Specifically, this related to occasions where a product was known to have been dispatched to the participant but Matomo had no record of the participant engaging with the Wrapped website. In the following section, the proportion of individuals who successfully accessed the Wrapped website and went on to use allocated components is reported. While not recorded by Matomo, these four individuals are reported as ordering products as recorded in the website logs and as successfully accessing the website (as this is a precursor to ordering). Whether these individuals engaged with their allocated video components is unknown.
Wrapped website
Of the 115 participants randomised to the intervention group, 34 (29.6%) did not click on the link provided to gain access to the Wrapped website. Two participants (1.7%) clicked on the link and created a user profile but did not complete the required tailoring questions to gain access to the site. Observation of user recordings suggests that while one of these individuals chose not to complete these questions, the other may have experienced an on-site technical error. The remaining 79 participants (68.7%) successfully completed this registration process and landed on the homepage. See figure 1 in Report Supplementary Material 11 for acquisitions flow chart. Table 15 presents the demographic characteristics of participants who did and did not gain access to the Wrapped website. Across most demographic characteristics, access levels were broadly in line with the overall level. There was some limited evidence of higher access among participants in the younger age group (16–19 years) and among those in deprivation quintile 5 (least deprived), and of lower access among those in deprivation quintile 2. The small number of participants across these subgroups means that the findings should be treated with caution.
| Accessed Wrapped N (%) | Did not access Wrapped N (%) | Percentage difference | |
|---|---|---|---|
| Age | |||
| 16–19 (n = 12) | 10 (83.3) | 2 (16.7) | 66.6 |
| 20–24 (n = 103) | 69 (67.0) | 34 (33.0) | 34.0 |
| Gender | |||
| Female (n = 79) | 53 (67.1) | 26 (32.9) | 34.2 |
| Male (n = 32) | 22 (68.8) | 10 (31.3) | 37.5 |
| Other (n = 4)a | 4 (100) | 0 (0) | 100 |
| Ethnicity | |||
| White (n = 89) | 70 (71.4) | 28 (28.6) | 42.8 |
| Mixed or multiple ethnic groupsa (n = 6) | 3 (50.0) | 3 (50.0) | 0.0 |
| Asian or Asian Britisha (n = 1) | 1 (100) | 0 (0.0) | 100 |
| Black, African, Caribbean or Black Britisha (n = 8) | 4 (50.0) | 4 (50.0) | 0.0 |
| Other ethnic groupa (n = 2) | 1 (50.0) | 1 (50.0) | 0.0 |
| IMD quintile | |||
| 1 (most deprived) (n = 18) | 12 (67.7) | 6 (33.3) | 34.4 |
| 2 (n = 19) | 11 (57.9) | 8 (42.1) | 15.8 |
| 3 (n = 24) | 17 (70.8) | 7 (29.2) | 41.7 |
| 4 (n = 29) | 20 (69.0) | 9 (31.0) | 37.9 |
| 5 (least deprived) (n = 25) | 19 (76.0) | 6 (24.0) | 52.0 |
| Sexual identity a | |||
| Heterosexual (n = 79) | 53 (67.1) | 26 (32.9) | 34.2 |
| Gay (n = 11) | 8 (72.7) | 3 (27.3) | 45.5 |
| Bisexual (n = 23) | 16 (69.7) | 7 (30.4) | 39.1 |
| Other (n = 2) | 2 (100) | 0 (0) | 100 |
When participants landed on the homepage, they were presented with blocks representing each of their assigned components. All participants were assigned the sample pack by default, but all other assignments were in response to individuals’ self-identified barriers. Table 16 presents the proportion of participants who were assigned each Wrapped component. Eight individuals (10.1%) ‘bounced’ off the homepage, that is, they left before visiting relevant pages for any of their assigned components. The remaining participants went on to visit at least one of their assigned components. Figures 2–7 in Report Supplementary Material 11 display the user journeys through each of these components with achievement of a pre-determined goal as the endpoint. The pre-determined goals were as follows: ordered sample pack, placed at least one order for condoms, ordered condom carrier, watched condom demo video (any duration), watched one of discussing condom videos (any duration), watched one of the real-life videos (any duration). Table 16 presents how many of the participants, assigned each component, achieved these goals.
| Component | Proportion of participants assigned N (%) | Proportion of participants assigned to each component who achieved the pre-determined goal N (%) |
|---|---|---|
| Sample pack | 79 (100) | 48 (60.8) |
| Condom ordering | 70 (88.6) | 37 (52.9)a |
| Condom carrying | 63 (79.7) | 31 (49.2) |
| Condom demo | 70 (88.6) | 7 (10.0) |
| Discussing condoms | 59 (74.7) | 4 (6.8) |
| Real life | 66 (83.5) | 9 (13.6) |
The table and figures together show that the product-related components (sample pack, monthly condom/lube packs, condom carrier) were more likely to be visited than the video-related components (condom demo, discussing condoms, real-life). While some individuals visited their assigned product-related components and did not place an order, most did, and few were lost during the ordering process indicating little to no friction at this point. Of the small numbers of participants who visited-video related components, lower proportions of participants also went on to achieve the goal (i.e. watched one or more of the videos) than for the product pages. This suggests that the descriptor on the homepage for these components, and the video stills and associated captions on each of the component pages, were not attracting individuals to access and play the videos. Furthermore, few of the videos played were watched in full, suggesting that either these were not sufficiently engaging and/or perceived as irrelevant or too long.
Three information and support pages are available to all Wrapped website users. The content of these pages is a duplication of that provided within the control website. Of those who accessed Wrapped, 10.1% (8/79) visited the sexual health information and support page, 5.1% (4/79) visited condom information page and 13.9% (11/79) visited STI information.
Participants were asked about their experience and use of the intervention within the fRCT surveys. At M3, participants were asked how useful they found the website and how they would rate the design. Of all those who had accessed Wrapped and responded to these questions (n = 68), 70.6% (n = 48) reported finding the site very or extremely useful, and 80.9% (n = 55) rated the design as good or very good.
At M12 survey, participants assigned to the intervention group were asked about their use of the sample pack and condom carrier. Table 17 summarises this feedback.
| Use of Wrapped products | Proportion of participants who used each product (as described in the table) out of all those who placed an order n/N (%) |
|---|---|
| Tried condoms in sample pack | 35/39 (89.7) |
| Tried lube in sample pack | 32/39 (82.1) |
| Looked at leaflet inside the sample pack | 36/38 (94.7) |
| Used box (that sample condoms delivered in) to store condoms | 26/39 (66.7) |
| Used carrier for condoms | 10/24 (41.7) |
| Attached the carrier (used for condoms) to keys or bag | 7/10 (70.0) |
Control website
Of the 115 participants randomised to the control group, 76 (66.1%) clicked on the link and visited the site. Of these, 14 (18.4%) bounced off the home page, that is, they did not visit any of the content pages. The proportion of those visiting the site who accessed each content page is as follows: 57.9% (44/76) visited information and support, 63.2% (48/76) visited condom information and 63.2% (48/76) visited STI information. Of those who visited and responded to the relevant M3 survey items, 31.1% (19/61) reported the website as very or extremely useful, and 72.1% (44/61) rated the design as good or very good.
RO 11: identify costs and resource use associated with the intervention for healthcare services and users (to inform the design of the definitive RCT and the future economic evaluation)
See Chapter 4 for findings of health economic analysis.
RO 12: measure contamination of intervention effect in the control group
Of the total number of participants in the control group who completed the M12 follow-up survey (n = 96), 42 (43.3%) reported exposure to Wrapped, that is, they indicated that they recognised one or more images from the intervention website (presented to them as screenshots within the M12 survey). This level of contamination was unexpected and therefore investigated further.
One potential route to contamination that was explored was a participant telling another person(s) about the study, them going on to enrol and be randomised into a different group, and then these individuals sharing experiences of differing intervention content. There is some limited evidence that this occurred. The research team was contacted twice during the study by participants asking how to refer a friend to the study. These participants were told that the study was only for people who were self-testing for STIs. Whether these participants went on to instruct their friends on gaining access via Preventx is unknown. There was also a group of three house-sharing participants (one in the intervention group and two in the control group) who may have signposted each other to the study, albeit this could not be verified as this would have led to a breach of participant confidentiality. There were also two occasions where participants in the control group e-mailed the research team to ask where to ‘register’ on the control website, one adding that this was for the purpose of ‘ordering condoms’. As the control website does not require registration, and condom ordering is not a feature, this indicated that details of the intervention website had been shared with them.
While there was the above evidence of friendship pairs/group participation, there was nonetheless no evidence that the intervention website had been directly accessed by anyone outside of the intervention group. Access to the intervention website was strictly controlled during the study. Each intervention group participant was provided with a unique URL (appended with their participant number) and was required to enter valid log-in credentials at each access. Examination of the intervention website log and data analytics records showed that the intervention website had only been accessed by individuals with one of the assigned participant numbers (i.e. no rogue/fake numbers). The sharing of log-in details was also considered unlikely due to the limited number of repeat visits by participants (other than by those who were using the condom-ordering component). A further possibility considered was that participants in the intervention group had shown the website to others in the control group (i.e. allowing them to physically view their screen). There is no means of verifying the occurrence of this but given that product ordering is restricted by design (either one per user or limited in quantity/frequency), and that there were few repeat views of videos, this scenario seems unlikely to have occurred.
Given that there is limited evidence that contamination of the control group did occur, the most likely explanation for the level of contamination reported in the M12 survey is that the question designed to detect this was not reliable. This question asked participants if they recognised ‘some or all of the images’ within three screenshots of the intervention website. Given that consistent branding (website name, logo, colour palette, icons) was used across both the intervention and control websites, it is possible that participants interpreted the question literally (i.e. they recognised the branding and therefore ‘some of the images’), or that the branding gave the screenshots a feeling of familiarity.
RO 13: identify possible adverse effects of the intervention
Within all follow-up surveys, participants were asked to self-report any adverse events related to their participation in the study. There were four fixed response items, three for anticipated adverse events, and one for ‘other problem’ which if endorsed routed participants to an open-ended question asking for further information. Two of the anticipated events (inadvertent disclosure that sexually active or testing for a STI) were considered a potential risk for all participants, predominantly as a result of receiving chlamydia self-sampling kits by post; a risk that was further enhanced for intervention participants if they chose to order products from the Wrapped website. The third anticipated event, increase in use of pornography, was only considered as a risk to those in the intervention group as a result of exposure to content aiming to eroticise condoms.
In line with the study protocol, all adverse events were reported to the NHS ethics, the sponsor, the SSC and the DMEC. On each occasion, all groups were satisfied that appropriate measures were being taken to manage these risks. Any participant who reported an adverse event was contacted by e-mail to offer support. On one occasion, a participant replied to the research team, reflecting on the event. Six months before the end of data collection, on the advice of the SSC and with ethical approval, an additional open-ended question was asked from all participants who indicated that they had experienced any adverse event. This invited them to provide details on why they thought that taking part in the study had led to this outcome; one participant provided an open-ended response to this at M12.
Table 18 presents each of the event types, the number of reported instances across all time points, and any qualitative comments received via e-mail or survey. Most adverse events that were reported concerned disclosure of STI testing. Although one individual reported that participation in the study had led to an increase in their use of pornography, this participant was allocated to the control group. Participants were contacted and asked if they would be willing to provide further information, but none was received.
| Adverse event | Number of instances reported | Qualitative comments |
|---|---|---|
| Led to someone finding out I was having sex when I did not want them to know | 1 | None |
| Led to someone finding out I was testing for a STI when I did not want them to know | 9 | E-mail response on offer of support: ‘It’s OK this hasn’t caused me too much trouble, there is no need for any discussion, it wasn’t a serious matter and there is no reason to change anything that you currently do’Open-ended response in M12 survey: ‘It has made some uncomfortable conversations, but they have also been positive in showing that I am trying to be safe’ |
| Led to an increase in my use of pornography | 1 | None |
| Other problem | 0 | None |
Review of progression criteria
On completion of data analysis, separate meetings were convened with the DMEC and with the SSC to review the progression criteria. A report was sent to members in advance setting out evidence against each criterion (see Report Supplementary Material 10). At each meeting, this evidence was presented followed by questions and discussion. A section of the meeting was closed (research team not present) to enable members to discuss their views in confidence. DMEC members cast their votes during this closed section of the meeting. The SSC cast their votes after the meeting via a Microsoft form. Both groups were quorate and agreed unanimously that the progression criteria had been met.