Behavioural intervention to reduce sexually transmitted infections in people aged 16 24 years in the UK: the safetxt RCT

Article history

The research reported in this issue of the journal was funded by the PHR programme as project number 14/182/07. The contractual start date was in December 2015. The final report began editorial review in February 2021 and was accepted for publication in February 2022. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PHR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Disclaimer

This report contains transcripts of interviews conducted in the course of the research and contains language that may offend some readers.

Copyright statement

Copyright © 2023 Free et al. This work was produced by Free et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Free et al.

Scientific background

Young people aged 16–24 years bear the heaviest burden of sexually transmitted infections (STIs) such as chlamydia and gonorrhoea and their long-term adverse health effects, including ectopic pregnancy and subfertility. 2,3 The risk of adverse health effects increases with repeated infections and reinfection rates following treatment are high, at up to 30% for chlamydia and 12% for gonorrhoea at 1 year. 4 Those with a STI are more likely to acquire further STIs and HIV (human immunodeficiency virus), if exposed. Inequalities in sexual health persist; STIs are positively associated with lower educational levels5 and living in more deprived areas. 3,6,7

Sexual and reproductive health (SRH) is defined as ‘a state of physical, emotional, mental and social wellbeing in relation to all aspects of sexuality and reproduction, not merely absence of disease, dysfunction or infirmity’ [© Copyright World Health Organization (WHO), reproduced with permission]. 8 High STI rates among young people also reflect broader aspects of poor sexual health, such as a lack of knowledge, skills or confidence in how to carry out safer sex behaviours and how to communicate with partners about sex or safer sex behaviours. 9

Development of the safetxt intervention

A description of the intervention and its theoretical rationale has been published (Figure 1). 32

FIGURE 1.

Safetxt theoretical model. BCT, behaviour change technique.

The intervention was informed by the capability, opportunity and motivation model of behaviour (COM-B). 51 According to this model, behaviours are influenced by an individual’s capability, opportunity and motivation to carry out the behaviours. Capability, opportunity and motivation for safer sex are influenced by the context: the individual’s knowledge, self-efficacy, intentions, attitudes and skills; the relationship’s quality and intimacy and the degree of trust in the relationship; and social factors, such as gender norms and roles. The opportunity to carry out safer sex behaviours is also influenced by broader socioeconomic and structural factors, which were not targeted in this intervention and so do not appear in the theoretical model.

The intervention employs educational, enabling and incentivising behavioural change functions to increase capability and motivation to carry out safer sex behaviours and, hence, increase correct STI treatment, partner notification, condom use and self-testing and partner testing for STIs before sex without a condom, leading to reduced STIs. The specific aspects of capability that the intervention was hypothesised to alter were knowledge about how to prevent STIs and correct treatment of STIs, condom use knowledge and self-efficacy, condom negotiation self-efficacy and partner notification self-efficacy. Increased knowledge regarding the health consequences of STIs and altered attitudes to partner notification were hypothesised to increase motivation for safer sex practices. The messages were developed based on behaviour change theory; evidence-based behaviour change techniques; the content of effective face-to-face safer sex interventions; the factors known to influence safer sex behaviours; the views of 82 young people collected in focus groups; and a questionnaire completed by 100 people aged 16–24 years. We ensured that the young people included in the focus groups and completing the questionnaire were from diverse socioeconomic and ethnic groups and included those living in cities, towns and rural areas of the UK (London, Manchester, Cambridgeshire and Norfolk).

Messages were written and adapted based on young people’s preferences as expressed in focus groups. Participants preferred messages with a non-judgemental and credible tone, short messages written in a positive style and those providing practical information regarding what needed to be done, why and how. They wanted no more than four messages per day and wanted the message frequency to reduce within the first 2 weeks. Positive approaches to sexuality and reproduction should recognise the part played by trust and communication, as well as pleasurable sexual relationships, in promoting well-being and enabling people to fulfil their sexual and reproductive health and rights. 52 The scope of the safetxt intervention was limited as some content was considered inappropriate for delivery by text message. Effective face-to-face interventions include content regarding gender roles, sexual pleasure and relationships, but this content, when adapted for delivery by text message, was considered too personal, intrusive and unacceptable, and so messages containing this content were removed from the intervention. Many effective face-to-face interventions include reviewing behavioural goals as behaviour change techniques, but these were considered too intrusive for delivery by text message. For some behaviours, instead of action plans, only suggestions of when, where and how risk reduction behaviours could be carried out were considered acceptable.

In a qualitative study with young people, coding and analysing the interviews led to the development of a theoretical framework for the mechanism of action, including how the intervention is hypothesised to work in increasing safer sex behaviours. 53 Prerequisites for an acceptable intervention from young people’s perspectives included that the content be simple and engaging and resonate with personal experiences. The tone was perceived to be trustworthy, friendly, professional, enabling, non-judgemental and non-pressured. The findings from the interviews suggest that the intervention could work by providing information, skills, reminders and reinforcement to young people via a channel that is convenient and accessible to them. For example, the texts appeared to help by providing new knowledge on how to put a condom on or having a reminder text facilitating condom use, and breaking down assumptions about how chlamydia infection is transmitted. The messages could also work by allowing recipients to reflect on their behaviour. The information participants received in text messages that chlamydia is common, that one may not know that they have it and that it is easily curable was said to have allayed fears and reduced stigma, which in turn increased participants’ confidence in telling a partner about an infection and/or helped them talk to their partner about the importance of protecting themselves against STIs, by, for example, being given words that they could use when having these discussions or sharing the actual texts. Messages were also shared with friends and siblings, initiating conversations about sexual health. The fact that this was done in a way that reduced stigma and was not pressured or judgemental assisted the communication with others. Social support for safer sex behaviours was increased through enhanced conversations with partners as well as sharing information and skills with others such as friends and siblings. There was less suggestion from the interviews that, aside from reduced stigma, other attitudes had shifted. Stigma associated with STIs can result in young people not accessing appropriate care and services,54 and therefore its inclusion in our health promotion intervention addressing sexual health was considered key.

Based on recipients’ feedback, the intervention was refined further for the main trial. We ensured that messages were relevant to men who have sex with men and women who have sex with women by seeking their views on the intervention in further interviews or a focus group. For example, we ensured that the pronouns used were gender-neutral. Additional content was included to provide examples of how others have negotiated condom use in ongoing sexual relationships and post-exposure prophylaxis (PEP) for HIV transmission and infection prevention. Participants reviewed new messages and reported that they were relevant, easy to understand and acceptable. We demonstrated that interactive support via text message is particularly acceptable in the area of sexual health intervention. 32,53,55

The safetxt intervention delivered in the randomised controlled trial

The intervention delivered in the RCT aimed to increase safer sex in three ways: (1) encouraging participants to correctly follow STI treatment instructions, including informing partner(s) about infection; (2) promoting condom use with new or casual partners; and (3) encouraging participants to obtain testing for STIs prior to unprotected sex. Participants in the intervention group received regular messages delivered by text message in community settings in accordance with a predetermined schedule.

Over the first 10 days, participants are sent messages targeting engagement with the intervention, taking treatment, avoiding sex for 7 days after treatment and telling partner(s) about an infection. These messages provide non-judgemental, non-stigmatising information about STIs. They provide suggestions about when, where and how to tell partner(s) about an infection and examples of how others have told partners, covering a range of different types of relationship (e.g. casual, long term).

Messages then target condom use and testing for STIs before having sex without a condom with a new partner. The topics covered are risk assessment, instructions on how to use condoms, positive aspects of condom use, tips on preventing condom problems and examples of how others have resolved condom use problems. Participants are prompted to think about their own success in achieving safer sex strategies, risks they have taken and what they could do differently in the future. Messages include advice regarding testing for STIs before having unprotected sex with a new partner. Participants are sent links to support for those concerned about partner violence, and web-based information regarding contraception, alcohol and sexual risk, how to use a condom and general communication about sex. The messages provide social support for safer sex behaviours and acknowledge participants’ experiences.

The intervention employs educational, enabling and incentivising behaviour change functions and 12 behaviour change techniques. Here we provide a list of each behaviour change technique used and example text messages from the message set for women who have sex with men and were diagnosed with chlamydia:

• information about health consequences of behaviour [e.g. ‘You can make sure you don’t get another infection (1) by getting the person you are having sex with treated, (2) by using condoms every time you have sex and (3) by you and your partner getting tested before sex without a condom and by having another test in 3 months’; or ‘If Chlamydia is left untreated or you keep getting it again, this can affect your ability to get pregnant. One way to avoid this is by getting a check-up before sex with new partners.’]

• instruction on how to carry out the behaviour [e.g. ‘To treat the infection, take the tablets and then don’t have sex (oral, vaginal and anal) for 7 days while the infection clears’; or ‘One reason a condom may split is because there is air trapped inside. To prevent this, hold the tip of the condom between your forefinger and thumb and roll it down, making sure there are no air bubbles’]

• demonstrations of risk reduction behaviour (e.g. ‘There are lots of other ways of telling the person you are having sex with that they need treatment. Here are some examples of how some people started the conversation: “I said that if I didn’t respect you I wouldn’t be telling you this. It’s awkward to tell people but it’s not right not to, is it? They may not know. You can’t just let them walk round with an infection”.’)

• social support (e.g. ‘If you can’t face telling them, you can ask the clinic to contact them for you and they won’t mention your name’); emotional support (e.g. ‘Here are how others felt when they found out that their test was positive: “I never thought I’d get chlamydia. I’ll use a condom in the future or get a check-up with them first”.’)

• social rewards (e.g. ‘You made the right decision to get a test.’)

• non-specific incentives (e.g. ‘Regular check-ups & check-ups with new partners means infections can be treated before they cause problems.’)

• encouragement to add objects to the environment (e.g. ‘Having condoms with you makes it more likely you’ll use one. Find a time to put a few in your purse. You could also keep a supply in places where you have sex.’)

• anticipated regret (e.g. ‘You can’t tell if someone has an infection just by looking at them. Ask yourself if having sex without a condom is worth taking the risk.’)

• reframing (e.g. ‘I said using a condom was about respecting each other.’)

• problem-solving [e.g. ‘If he says condoms aren’t comfortable you could try a different brand. Some guys find they can feel more with thinner condoms (which are still safe).’]

• action-planning techniques (e.g. ‘Think about when, where and how you will talk to them about condoms and how you could start the conversation.’). 56

The information on safer sexual practices is in accordance with existing guidelines. 57

The messages are tailored according to gender and sexual orientation. Women who have sex with men only (WSM), MSM, men who have sex with men and women (MSMW) and women who have sex with men and women (WSMW) are sent messages about how others had negotiated condom use. WSM, WSMW and men who have sex with women (MSW) are sent messages about emergency contraception. MSM and MSMW are sent messages about PEP. Women who have sex with women only (WSW) are not sent messages about condom use. The information provided is specific to the STI diagnosed. This tailoring results in different numbers of messages being sent to those of different gender and sexual orientation.

The core message sets include 74 messages for WSM and WSMW; 42 messages for WSW; 69 messages for MSW and 79 messages for MSMW; and 76 messages for MSM. Recipients can request additional messages on specific topics. Participants are sent four messages per day for days 1–3, and then one or two messages per day for days 4–28, two or three messages per week for month 2 and two to five messages per month for months 3–12 (see Appendix 1).

Patient and public involvement

Patients and the public were involved in all phases of this study. Prior to developing the intervention, we discussed possible safer sex interventions with young people based at Southwark Further Education College. Those aged 16–24 years were invited to take part in a discussion group by one of the research team at the end of a class or by a researcher approaching students in the college canteen. Five discussion groups were convened in a private room, each comprising 2–10 participants (total of 25 participants). The participants were enthusiastic about receiving information and support via mobile phone. Patient views informed the intervention development through their participation as research participants in designing the content of the intervention. The views of the target audience were collected in nine focus groups, which informed the tone, style, frequency, duration and content of the intervention. Some research participants wrote some sections of the messages. King’s College Hospital has an active user group of young people wishing to contribute to SRH research. We met with a group of 14 patient representatives to seek their views on the trial design. Their views influenced our follow-up procedures. They asked that materials be posted by normal (not recorded) delivery in a coloured envelope and wanted to receive a text message saying that the materials had been posted so that they could look out for the package. We modified the patient information, questionnaires and consent procedures based on feedback from this group. A patient representative was included in the Trial Steering Committee (TSC); it provided comments on all trial materials and provided advice during the trial. A limitation of our patient involvement work was that we did not have a patient advisory group providing advice throughout the trial. At the time of writing this report we have an active group of five patient representatives making plans about how they can be actively involved in disseminating the trial results widely once they are published.

Aim

The primary objective of this trial was to quantify the effects of the safetxt intervention on chlamydia or gonorrhoea infection at 1 year compared with a control group receiving usual care and messages about trial participation.

The secondary objectives were to determine the effects of safetxt on partner notification and condom use at 4 weeks and on condom use and STI testing at 1 year. To explore which intervention components are effective, data were collected on the theoretical constructs on the pathway to behaviour change influenced by the intervention components. We planned to establish the cost-effectiveness of the intervention.

The intervention development work was funded by the UK National Institute for Health and Care Research (NIHR) Health Technology Assessment programme as project number 10/93/04. This trial was funded by the NIHR Public Health Research programme as project number 14/182/07.

Study design

Safetxt was a parallel-group, individual-level, randomised superiority trial in which care providers and outcome assessors were blinded to allocation. The trial used a 1 : 1 allocation ratio and was designed to establish the effects of a safer sex intervention delivered by text message on the incidence of chlamydia and gonorrhoea infection over 12 months.

Study setting

Participants were recruited from 92 sexual health clinics across the UK. Full details of the clinics are provided in Acknowledgements. The clinics were in cities, towns and rural areas of England, from Cornwall in the south-west to Kent in the east and Northumberland and Cumbria in the north, and in East Lothian in Scotland.

Participants

Recruitment

Randomisation

Randomisation was computer-based, independent, automated and remote from the recruiting clinic, and carried out in a 1 : 1 allocation ratio.

Treatment groups

Outcomes

Assessment of outcomes

Follow-up procedures

The study team based at London School of Hygiene & Tropical Medicine Clinical Trials Unit collected follow-up data between 20 May 2016 and 28 February 2020. We collected follow-up data at two time points: 4 weeks and 1 year post randomisation. Questionnaires were sent to participants at 4 weeks and 1 year, and a STI test kit was sent at 1 year. The current contact details of non-responders were checked with the person nominated by participants at randomisation for contact in the event of non-response.

An unconditional £5 cash incentive was sent with questionnaires at 4 weeks and 1 year, and a conditional £20 cash incentive was sent after the STI test kit had been returned. We notified participants by text message 1 week before each time point (i.e. 3 and 51 weeks) and asked them to respond if they had changed their address.

We contacted non-responders at 4 weeks by post, text message, e-mail and telephone at several time points until they returned their questionnaire. At weeks 5 and 8 they were contacted by e-mail and at weeks 6, 10, 12 and 14 they were contacted by telephone, text message or post. At week 14, they were contacted by post for a response to the key questions only (i.e. ‘Did you tell the last person you had sex with before you tested positive to get treatment?’ and ‘Was a condom used the last time you had sex?’).

The primary outcome was assessed using chlamydia and gonorrhoea tests collected by post at 1 year and clinic records of completed tests during the 12-month trial follow-up period. STI test kits were posted (in P650 standard packaging) to respondents. Directions in the pack were for participants to provide a vaginal swab (women), a urine sample (men), an oral swab (men who have sex with men) or anal samples (men who have sex with men) and then place this in the packaging before posting it to the laboratory using the prepaid and addressed envelope. The test kits were identified by laboratory number only, rendering the laboratory staff blind to the participant’s allocation. The results of the STI testing were reported on the secure trial laboratory site by laboratory code. Laboratory codes and participant identification numbers were linked at the trial co-ordinating centre. Self-reported data were collected by post or any method to which the participant had agreed at enrolment (i.e. mobile phone or e-mail).

Non-responders were contacted at 1 year by post, text message, e-mail and telephone at several time points until they returned their questionnaire and/or STI test kit. At week 55 they were contacted by telephone and e-mail. At weeks 57, 59 and 61 they were contacted by telephone, text message and post. At week 62 they were contacted by telephone, text message and post for a response to the key questions only [i.e. ‘Have you been diagnosed with chlamydia, gonorrhoea or non-specific urethritis (NSU) after joining the study?’, ‘Where did you get tested?’ and ‘Where did you get treated?’].

We included data from participants for whom there was evidence of a STI test conducted elsewhere (i.e. a test result not obtained through the safetxt STI postal test kit). Clinic records were checked by clinic staff to confirm self-reported data regarding STI tests, STI diagnoses after joining the trial and partner attendance for treatment. We also included data from tests conducted at the clinic at which participants had been recruited. If participants were tested elsewhere, they sent evidence of their STI test results to the study team for confirmation that the test had been carried out at another service. Alternatively, we asked participants for permission to contact their service provider to confirm that they had been tested there.

Data management

Data were held on a secure system and password protected. Any data on paper were locked in a cabinet in a room that was locked unless staff were working in the room. Access to the building is only by London School of Hygiene & Tropical Medicine identification cards. All trial procedures were in accordance with the principles of Good Clinical Practice. The essential documents of the sponsor/trial organisers and investigators will be retained for at least 10 years after the completion of the trial. In accordance with the London School of Hygiene & Tropical Medicine’s retention requirements, primary research data will be retained for 10 years following the completion of the trial. In accordance with the Data Protection Act 2018,63 personally identifiable data about participants were not kept longer than necessary and were deleted within 3 months of the end of the trial. If a patient withdrew from the trial, attempts were made to contact them to determine if they were still happy for any further data collected from clinics to be used. If no contact could be made with the patient, we assumed that they had withdrawn from the whole trial and did not want their data to be used.

Data monitoring, quality assurance and preparation

All data systems were set up with checks to alert the trial assistants if the data being entered were illogical, inconsistent or incomplete. A random sample of 10% of questionnaire data entered at 1 and 12 months was double checked in March 2017. As no errors were found, a further random sample of 100 participants’ 1- and 12-month data were checked in March 2020. Again no errors were found. STI tests were conducted in accredited laboratories and the results data were entered manually into the database directly. All positive test results were double checked. Melissa J Palmer prepared and coded the data for analysis (see the statistical analysis plan64 for full details).

Sample size

Analysis

Background

Achieving full recruitment and high follow-up in RCTs remains a significant challenge. In 2017, a review found that just 56% of RCTs published in the Health Technology Assessment journal recruited to their final target. 79 This leaves the remaining trials underpowered, reporting less precise results than planned, which could lead to potentially beneficial interventions not being implemented due to lack of clear evidence of their effects. 80 Achieving high follow-up is even more challenging for studies that collect data on sensitive topics, such as sexual health. 81,82 Failure to achieve high follow-up can reduce trial power and is a threat to trial validity, as those lost to follow-up may differ from those followed up. 83 Failure to fully recruit or achieve high follow-up may also waste valuable time and resources, or additional resources may be needed to fully recruit. 80

There is little research on increasing recruitment success in clinical trials that focuses on interventions targeted at recruiters. 80 The few studies that are available found that sites who had site initiation visits recruited more participants80 and that having all staff present at site initiation visits is deemed beneficial for trial progress. 84 Effective planning for the trial is consistently noted as an important factor of successful recruitment. 85,86 Communication is frequently raised as important for recruitment success; sites that receive additional communication strategies recruit more participants on average and in less time. 80 Regular communication through a variety of methods, tailored to individual sites and used to convey a variety of messages, is highlighted as important for trials to recruit successfully. 84,85,87 Monitoring is also frequently highlighted as important for recruiting effectively. Constant monitoring enables bottlenecks in recruitment to be identified88 and flexible and rapid solutions to recruitment issues to be provided. 86

A wide range of approaches to increasing trial follow-up and follow-up for postal, e-mail and telephone questionnaires have been evaluated in RCTs. Achieving high follow-up on sensitive topics such as sexual health, especially among young people, is challenging; for example, in the UK, response rates to the Natsal-3 sexual health survey were 57.7%,89 and the response rate to a previous survey and trial involving STI postal tests among young people was 31.5%90 and 45.5%, respectively. 91 We previously reported our approach to developing trial follow-up procedures that were used in our pilot trial to achieve over 80% follow-up with 200 young people. 92

In this chapter, we detail our approach to recruitment and document how we adapted our follow-up procedures during the safetxt trial. We provide practical examples of our strategies to share our learning about trial recruitment and follow-up.

Methods

Results

Our highest-recruiting month saw 360 participants recruited, exceeding our peak monthly target of 350 participants. Based on recruitment strategies discussed at the site initiations, 88% of clinics screened pre-booked appointments and 60% of clinics identified patients from walk-in appointments. The strategy used depended on the type of appointments the clinic ran, with many clinics having both types available. Sixty-nine per cent of clinics planned to recruit patients when they contacted them with test results. The clinics that did not use this approach did not for a number of reasons, including because their results text messages could not be adapted, or that staff contacting patients with test results were not trained in Good Clinical Practice. Many sexual health clinic services were reconfigured or altered during the recruitment period, meaning that the recruitment strategies used at some clinics changed during the course of the trial.

Although we saw increases in recruitment figures following our interventions (Figure 2), the time it took from implementing an intervention to it having an effect on sites varied depending on the type of intervention. For example, in May 2017 we sent refreshed recruitment packs to all sites and increased contact with sites that were finding it hard to recruit. It was time-consuming to put together the recruitment packs and post these to 53 sites. Contact with lower recruiters was carried out in accordance with a schedule of increasing regular contact over several weeks. As a result, the impact from such interventions was not seen immediately but was delayed until a couple of months later.

FIGURE 2.

Recruitment progress and when recruitment interventions were implemented.

Interventions were ongoing and cumulative to strategies already employed. For example, after certificates had been introduced in May 2017 these were available to and requested by sites throughout the remainder of the recruitment period.

A total of 5457 participants (88%) completed the 4-week questionnaire (intervention, 2710/3123, 87%; control, 2747/3125, 88%). A total of 4675 participants (75%) provided data for the primary outcome (intervention, 2329/3123, 75%; control, 2346/3125, 75%), with 4871 out of 6248 (78%) providing data for the 12-month questionnaire.

Discussion

Our dynamic collaborative approach to recruitment (Figure 3) combined activities deemed important for recruitment success in clinical trials into one cyclical approach, involving good communication and joint working with recruiters and recruiting sites, constant evaluation and new approaches to overcome recruitment challenges, from the outset of opening recruitment sites and throughout the trial recruitment period. Recruitment progress was constantly evaluated, interventions designed to address recruitment challenges arising from these evaluations were implemented, and recruitment progress was continuously evaluated. The initial site set-up was crucial to planning successful recruitment, anticipating recruitment barriers and building a strong relationship foundation with recruiting sites. At the core of the approach was regular, engaging, responsive communication, which we see as key to successful recruitment.

FIGURE 3.

A dynamic and collaborative approach to trial recruitment.

Follow-up in the safetxt trial was the responsibility of the trial co-ordinating centre. Important elements of our approach were learning from participants about barriers to follow-up, monitoring follow-up, developing new approaches to increase follow-up based on monitoring and participants’ experience, and employing evidence-based follow-up methods. At least 40% of our trial population changed address at least once and, therefore, keeping track of these changes was an essential element of achieving postal STI test follow-up among young people in this trial. It is important to note that any one of the recruitment or follow-up interventions mentioned is likely to have a small effect on its own; rather, it was the cumulative effect of continued efforts that led to success.

Evidence from specific trials can be difficult to apply to other trial settings and participant populations. 79,80 For example, it is especially challenging to achieve full recruitment and high follow-up in research with young people and about sexual health, such as the safetxt trial population. Therefore, the techniques explained in our approach may need to be tailored specifically to other trial settings and patient groups, if implemented. Although some of the interventions or elements of our approach have been previously described, recruitment and follow-up in other trials could benefit from adopting our dynamic, collaborative approach to recruitment and co-ordinated approach to follow up.

Our approach is limited in that we did not test our recruitment or follow-up techniques in such a way that cause and effect could be reliably established. However, we believe, based on our improvement in recruitment rate and recruiting to target, that our approach was successful in aiding recruitment. Walters et al. ’s79 review showed that just 56% of trials met their recruitment target, and multicentre trials achieved an average recruitment rate of 0.86 patients per centre per month. We met our recruitment target and exceeded this recruitment rate, demonstrating our approach’s success in supporting recruitment.

Recruitment, randomisation and exclusions

Between 1 April 2016 and 23 November 2018, 20,476 young people were assessed for eligibility in 92 sexual health services across the UK (Figure 4). Of these, 14,217 were excluded before randomisation (7316 were eligible but declined and 6901 were eligible and approached by text message or e-mail but did not respond). Informed consent was provided by and baseline data were submitted through the trial database system for 6259 participants. Eleven participants were excluded after randomisation because of duplicate randomisations. This resulted in 6248 participants included in the trial, with 3123 randomised to the intervention group and 3125 randomised to the control group. A total of 281 out of 6248 (4.5%) participants withdrew from the trial after allocation: 134 out of 3123 (4.3%) in the intervention group and 147 out of 3125 (4.7%) in the control group.

FIGURE 4.

The CONSORT (Consolidated Standards of Reporting Trials) flow diagram.

Baseline characteristics

Table 3 reports the baseline characteristics of the trial participants. The mean age of participants was 20 years, and 63% (3942/6248) were aged 20–24 years. Sixty-five per cent (4067/6248) of participants were female and 78% (4864/6248) were white British/other white background. Half (3091/6195) of participants were from the two most deprived Index of Multiple Deprivation (IMD) quintiles. 95 Baseline sociodemographic characteristics were similar between the treatment groups.

Loss to follow-up

Follow-up was completed between 1 May 2016 and 28 February 2020. A total of 5457 participants (88%) completed the 4-week questionnaire (intervention, 2710/3123, 87%; control, 2747/3125, 88%). A total of 4675 participants (75%) provided data for the primary outcome (intervention, 2329/3123, 75%; control, 2346/3125, 75%).

Primary analysis results

Secondary analyses

Subgroup analysis

There was no evidence that the effect of the intervention was different among participants in any of the prespecified subgroups (Figure 5).

FIGURE 5.

Primary outcome by prespecified subgroup.

Additional trial data

A small proportion of participants reported that they knew someone else in the study (intervention, 137/2414, 5.7%; control, 141/2453, 5.8%). Overall, the proportion of intervention group participants who reported that their messages were read by another participant was 1.5% (37/2414) and the proportion of control group participants who reported that they read another participant’s messages was 1.3% (32/2453; Table 7).

Additional non-prespecified analyses

Path analysis

Had the intervention been effective in reducing STIs, we intended to carry out a path analysis to identify which (if any) of the intermediate outcomes mediated this effect. However, in the absence of a positive intervention effect on the primary outcome, we examined whether or not there was evidence for a mediatory role of the intermediate outcomes in the intervention effect on the secondary outcome of condom use at first sex with a new partner (the secondary outcome on which the intervention had the greatest positive effect).

Two intermediate outcomes, ‘knowledge’ and ‘correct condom use self-efficacy’, were positively associated with the intervention. However, only ‘correct condom use self-efficacy’ was associated with condom use at first sex with a new partner, and so this was the only potential mediator of the intervention effect on condom use.

We used a path analysis to estimate the effect of the intervention on condom use at first sex with a new partner through a direct path, and specified an indirect path via the intermediate outcome ‘correct condom use self-efficacy’. The results indicated that the indirect path accounted for a small proportion of the overall effect of the intervention on condom use (total effect coefficient 0.109, p = 0.016; total indirect path coefficient 0.015, p = 0.015). Therefore, there is evidence that correct condom use self-efficacy was a partial mediator of the effect of the intervention on condom use at first sex with a new partner.

Pooled analysis with the safetxt pilot trial data

The pooled analysis of the safetxt main trial and pilot trial on the incidence of chlamydia/gonorrhoea showed substantial heterogeneity (I² = 55%).

To explore heterogeneity, we conducted one additional subgroup analysis.

We conducted a pooled analysis with all of the main trial and pilot trial data from participants diagnosed with a STI at baseline (where the intervention group had been allocated to receive content targeting partner notification, condom use and STI testing). The pooled OR was 1.12 (95% CI 1.02 to 1.24; n = 6915; I² = 0%) (Figure 6).

FIGURE 6.

Pooled analysis of main trial and pilot trial: incidence of gonorrhoea or chlamydia infection (objectively assessed at 12 months) among participants diagnosed with a STI at baseline. a, Incidence of chlamydia/gonorrhea at 12 months, subsample of participants diagnosed with an STI at baseline; b, incidence of chlamydia/gonorrhea at 12 months. df, degrees of freedom. M–H, Mantel–Haenszel.

Fixed-effects model

The remaining subgroup of pilot trial participants was participants aged 16–24 years who reported having two or more partners and unprotected sex in the last year but no STI at baseline. These participants were sent the safetxt content targeting STI testing and condom use only. The effect of the safetxt content targeting STI testing and condom use only on the incidence of chlamydia/gonorrhoea was 2/58 (3.8%) in the intervention group and 8/53 (15.1%) in the control group (OR 0.20, 95% CI 0.04 to 0.99). 32

Aim

The aim of this analysis is to describe young people’s views of the safetxt text intervention and control messages as expressed in open feedback comments.

Methods

On the final sheet of the 12-month questionnaire was an open-ended question, ‘Did anything good or bad happen as a result of being involved in the study or receiving the text messages? Please describe’, followed by a blank space in which participants could write a response.

Two researchers (AG and SB) not involved in the development of the safetxt intervention or previously involved in the safetxt trial independently coded the free-text comments and categorised data by theme. Anasztazia Gubijev used Microsoft Excel^® 2019 (Microsoft Corporation, Redmond, WA, USA) and Sima Berendes used NVivo version 12 (QSR International, Warrington, UK). Both researchers initially took a purposive sample of 12% (n = 390) questionnaires and checked these for consistency of analysis. Once this was complete, Anasztazia Gubijev and Sima Berendes independently analysed all of the remaining free-text comments in all questionnaires. The findings were compared with the previously published findings from qualitative interviews with participants in the pilot trial. 53

We compared the themes identified in analysis of the open feedback comments with the themes identified in the qualitative research. We described where the themes were similar and where new themes or new aspects of the themes emerged.

Results

Fifty-six per cent (n = 3526) of participants provided comments in the open feedback section of the 12-month questionnaire, 72% of those who completed a 12-month questionnaire. In total, 51% of intervention group participants and 53% of control group participants left open feedback comments. The comments varied in length. Most were only a few sentences long, with some participants providing longer feedback. Many participants in both the intervention and control group simply stated ‘no’, ‘nothing’ or ‘not applicable’.

Participants across all sociodemographic backgrounds provided open feedback comments, and the characteristics of respondents were similar to the characteristics of safetxt trial participants (Table 8).

Relationship status and intervention usefulness

Some participants in a ‘relationship’ reported that the intervention content was not relevant to them but would be useful if they were single or for those with more casual partners. This was true for various aspects that the intervention targets, including condom use and STI testing:

Due to being in a relationship I only have the 1 partner, however I feel the information via texts would be useful for someone single.

22 years, WSM, intervention

I was more frequently reminded of protection options regarding sexual intercourse and the risk of STDs/STIs. However, these were of limited use to me as I am in a monogamous relationship.

19 years, WSM, intervention

However, it is important to note that a change in relationship status could mean that the intervention could still be useful in the future:

I continued to still not use condoms . . . my boyfriend doesn’t like them as he feels like they don’t feel as good. However, if I was single I would definitely 100% use condoms with anyone I have sex with.

17 years, WSM, intervention

A suggestion for change was to include more varied text messages:

The text messages were a good reminder to get checked, however I have been with the same person since the survey started. I think the texts should be more varied, as after a while I’d skip over them. Maybe more of the texts that had interesting facts.

23 years, MSW, intervention

Although participants reported that the study was less relevant to those in a relationship, receiving the text messages enabled important information to be passed on to peers:

Although I, myself did not use protection due to only being with my long term partner. This study has helped me pass on vital info to my friends to make sure they stay safe when sleeping with multiple people

18 years, WSM, intervention

There were similar comments from control participants in a ‘relationship’:

The study reminds you and makes you think about safe sex, If I were to be single it would definitely be more effective and make you more careful with new partners.

23 years, MSW, control

Discussion

The open feedback comments were consistent with the findings from qualitative research previously published, in that recipients liked the tone and convenience of messages, which were described as reassuring and helpful. 53 The messages reportedly increased knowledge, reduced stigma, enabled participants to tell their partner about their infection more confidently and increased condom use and STI testing. In contrast to the previous qualitative research findings, whereas some participants reported that the messages frequency was about right or too few, others reported that there were too many messages or there were too many messages to start with and then too few in the longer term. There were also a few reports of concerns regarding the confidentiality of messages.

Participants’ reports also encompassed new themes or new aspects of the themes previously described. Some reported greater awareness and concern about their sexual health, with an impact on decisions regarding having sex only with people they ‘knew’ or could trust. Some reported greater agency and confidence in asserting their needs or wishes in relation to choosing whether or not and with whom to have sex. One reported that the study had given them the confidence to start a new relationship after their STI. Reduced embarrassment contributed to greater agency, resulting in participants being able to talk with partners and others about sex and safer sex practices, including condom use. Some reported a reduced sense of ‘being the only one’ diagnosed with a STI and, hence, less stigma regarding having a STI.

A number of control group participants reported that an impact of participating in the study was that their embarrassment had reduced because they had realised that they were not the only person to have a STI and changed their behaviour even though they had not received the intervention. Participants attributed this to the impact of taking part in the study; the impact of the monthly control group message about trial participation, reminding them of their STI and the importance of safer sex; or the impact of having a STI.

Strengths and limitations

A strength of the findings is that free-text comments were provided by over 3000 participants and analysed by two researchers not previously involved in the intervention development or trial. Other than asking about whether or not anything ‘good or bad‘ had resulted from taking part in the study, the topics for feedback were not predefined. A strength of our analysis was the triangulation of results with the previous qualitative research findings about recipients’ experiences of the intervention. It is reassuring that the findings were highly consistent with the previous qualitative findings. The response rate was 52%, so there may be non-response bias. The experience of those not leaving a free-text comment may be different from that of those who completed this section. However, participants across all sociodemographic groups provided comments, and the characteristics of respondents were similar to the characteristics of trial participants. It is not possible to blind participants receiving a behavioural intervention, and this could introduce bias in the ascertainment of feedback. All free-text feedback was brief, optional and completed at the end of involvement in the trial, so it is was not possible to explore participants’ views in depth or to follow up on feedback.

Discussion in relation to the existing literature

The overwhelmingly positive feedback from the free-text comments to this text messaging intervention is in line with findings from the pilot study. 53 Positive approaches to sexuality and reproduction should recognise that trust and communication, as well as pleasurable sexual relationships, should play a part in promoting well-being and enabling people to fulfil their sexual and reproductive health and rights. 52 The open feedback comments suggested that recipients perceived the intervention to have benefits in broader aspects of sexual health such as confidence and agency in communicating about sexual health and condom use with partners and others.

Consistent with other trials of behaviour change interventions, feedback from control group participants suggests that there is likely to be a strong Hawthorne effect, whereby the control group messages unrelated to safer sex and participation in the trial influenced attitudes and behaviour. In line with other research, the experience of having a STI also altered behaviour. 96–98 A full discussion of the open feedback comment findings in relation to the trial findings is provided in Chapter 7.

Conclusion

Overall, according to recipients’ views expressed in open feedback comments, the safetxt intervention had a positive impact on many aspects of broader definitions of positive sexual health.

Introduction

As the trial results failed to show a beneficial effect of the intervention on STIs, we did not complete a full cost-effectiveness analysis. This chapter presents the methodology for the proposed cost-effectiveness analysis, alongside the parameters gathered and the model schematic. The model was programmed in Microsoft Excel 2019 and is available on the trial website (https://safetxt.lshtm.ac.uk/publications/).

Methods

Model structure

This model considers the impacts of an England-wide roll-out of this intervention for 1 year, with the parameters detailed in Table 9. We developed a gender-stratified flow chart model simulating the number of eligible individuals we expect either to be reinfected with chlamydia or gonorrhoea or in whom their original chlamydia or gonorrhoea infection has persisted. We then estimate the proportion of individuals who will, as a result of their persistent/reinfection, suffer from sequelae associated with chlamydia and gonorrhoea (Figure 6). This model aims to explore the impact of the safetxt intervention on reducing the proportions of individuals with reinfections or persisting infections over the period of 1 year.

TABLE 9 - Model parameters

IVF, in vitro fertilisation; PID, pelvic inflammatory disease; SD, standard deviation.

Parameter description	Assumed value and distribution	Data source(s)	Notes and assumptions
Proportion with reinfection or persistent gonorrhoea infection at 1 year (intervention group)	∼U(12.9, 21.0)	safetxt	From safetxt trial data, examining gonorrhoea infection at follow-up compared with gonorrhoea at baseline in the intervention group (range is based on 95% CI of the data)
Proportion with reinfection or persistent gonorrhoea infection at 1 year (control group)	∼U(11.5, 18.9)	safetxt	From safetxt trial data, examining gonorrhoea infection at follow-up compared with gonorrhoea at baseline in the control group (range is based on 95% CI of the data)
Proportion with reinfection or persistent chlamydia infection at 1 year (intervention group)	∼U(18.4, 22.0)	safetxt	From safetxt trial data, examining chlamydia infection at follow-up compared with chlamydia at baseline in the intervention group (range is based on 95% CI of the data)
Proportion with reinfection or persistent chlamydia infection at 1 year (control group)	∼U(16.1, 19.5)	safetxt	From safetxt trial data, examining chlamydia infection at follow-up compared with chlamydia at baseline in the control group (range is based on 95% CI of the data)
New diagnoses of gonorrhoea in those aged 16–24 years over 1-year period	Females: ∼U(9472–14,208)	Mitchell et al.99	New infections in those aged 15–24 years in England 2019 (assumed uniformly distributed with a ±20% uncertainty range)
	Males: ∼U(11,047–16,571)
New diagnoses of chlamydia in those aged 16–24 years over 1-year period	Females: ∼U(69,787–104,681)	Mitchell et al.99	New infections in those aged 15–24 years in England 2019 (assumed uniformly distributed with a ±20% uncertainty range)
	Males: ∼U(37,475–56,209)
Proportion of those aged 16–24 years who own a personal mobile phone	U(0.98–1.00)	Free et al.58	Almost complete coverage of mobile phone access in this age group
Number of partners traced (intervention)		safetxt
Number of partners traced (control)		safetxt
Number of contacts needed to be traced to prevent one secondary infection	Females: ∼U(1.12–1.68)	Althaus et al.100	Reference value is for chlamydia among individuals aged < 25 years considering all partnership types. However, assumed to be equal for both chlamydia and gonorrhoea (no range given in reference, so assumed a ±20% uncertainty range uniformly distributed)
	Males: ∼U(1.62–2.42)
Proportion of gonorrhoea infections that lead to symptoms associated with uncomplicated infection (‘symptomatic infection’)	Females: ∼U(0.25, 0.50)	Farley et al.,101 Institute of Medicine,102 Satterwhite et al.103	Range of values across three studies. Values from Farley et al.101 (lower end values of ranges) appear to be assumed values or drawn from Wiesner and Thompson.104 Values from the Institute of Medicine102 (middle values of ranges, not directly used) obtained from testing, symptom questionnaire and medical record follow-up in a study of individuals in the USA aged 18–29 years. Values from Satterwhite et al.103 (upper end values of ranges) estimated by a committee of experts in the USA
	Males: ∼U(0.50, 0.85)
Proportion of chlamydia infections that lead to symptoms associated with uncomplicated infection (‘symptomatic infection’)	Females: ∼U(0.20, 0.30)	Farley et al.,101 Institute of Medicine,102 Satterwhite et al.103	Range of values across three studies. Values from Farley et al.101 (lower end values of ranges) appear to be assumed values or drawn from Wiesner and Thompson.104 Values from the Institute of Medicine102 (middle values of ranges, not directly used) obtained from testing, symptom questionnaire and medical record follow-up in a study of individuals in the USA aged 18–29 years. Values from Satterwhite et al.103 (upper end values of ranges) estimated by a committee of experts in the USA
	Males: ∼U(0.11, 0.75)
Proportion of those with gonorrhoea and chlamydia infection who develop PID	Females: ∼U(0.13, 0.19)	Price et al.105	Estimate obtained from an evidence synthesis of observational studies and randomised controlled trials. Values are for chlamydia but assumed to be similar for gonorrhoea
Proportion of those with PID who develop chronic pelvic pain	Females: ∼U(0.14, 0.22)	Yeh et al.106	Estimate based on value and range used in an economic modelling study
Proportion of those with PID who develop ectopic pregnancy	Females: ∼U(0.022, 0.032)	Low et al.107	Estimate obtained by meta-analysis of values from two studies of women who have ever had a positive chlamydia test. The first study107 was done in Swedish women up to age 35 years
Proportion of those with PID who develop tubal factor infertility	Females: ∼U(0.054, 0.080)	Low et al.107	Estimate obtained by meta-analysis of values from two studies of women who have ever had a positive chlamydia test. The first study107 was done in Swedish women up to age 35 years
Proportion of those with gonorrhoea or chlamydia infection who develop epididymitis	Males: ∼U(0.00, 0.02)	Institute of Medicine,102 Adams et al.108	Estimate based on value estimated by a committee of experts in the USA103 and value used in previous modelling study108
Health state utility value for asymptomatic gonorrhoea infection	1	Institute of Medicine102	It is assumed that there is no direct health loss associated with asymptomatic gonorrhoea infection
Health state utility value for symptomatic gonorrhoea infection	Females: ∼U(0.77, 0.93)	Institute of Medicine102	Values were estimated by a committee of experts in the USA based on ‘mild’ symptoms in women and urethritis in men, both with outpatient treatment (assumed uniformly distributed with a ±10% uncertainty range)
	Males: ∼U(0.76, 0.92)
Health state utility value for symptomatic chlamydia infection	Females: ∼U(0.81, 0.99)	Institute of Medicine102	Values were estimated by a committee of experts in the USA based on ‘mild’ symptoms in women and urethritis in men, both with outpatient treatment (assumed uniformly distributed with a ±10% uncertainty range)
	Males: ∼U(0.76, 0.0.92)
Health state utility value for PID	Females: ∼N(0.87, SD 0.26)	Jackson et al.,109 Smith et al.110	Mean time trade-off estimate for inpatient women without previous PID. Considered to be best available estimate among those found in a previous systematic review
Health state utility value for chronic pelvic pain	Females: ∼N(0.79, SD 0.29)	Jackson et al.,109 Smith et al.110	Mean time trade-off estimate for women without previous PID. Considered to be best available estimate among those found in a previous systematic review
Health state utility value for ectopic pregnancy	Females: ∼N(0.87, SD 0.26)	Jackson et al.,109 Smith et al.110	Mean time trade-off estimate for women without previous PID. Considered to be best available estimate among those found in a previous systematic review
Health state utility value for tubal factor infertility	Females: ∼N(0.84, SD 0.29)	Jackson et al.,109 Smith et al.110	Mean time trade-off estimate for women without previous PID. Considered to be best available estimate among those found in a previous systematic review
Health state utility value for epididymitis	Males: ∼U(0.46, 0.78)	Zwart et al.111	Taken from an economic modelling study
Duration of health loss associated with symptomatic gonorrhoea infection (years)	Females: ∼U(0.019, 0.080)	Institute of Medicine,102 Satterwhite et al.103	Range of values across two studies. Values from Satterwhite et al.103 (lower end values of ranges) estimated by a committee of experts in the USA. Values from Farley et al.101 (upper end values of ranges) appear to be assumed values or drawn from Wiesner and Thompson104
	Males: ∼U(0.019, 0.080)
Duration of health loss associated with symptomatic chlamydia infection (years)	Females: ∼U(0.062, 0.092)	Institute of Medicine102	Values estimated by a committee of experts in the USA (assumed uniformly distributed with a ±20% uncertainty range)
	Males: ∼U(0.015, 0.023)
Duration of health loss associated with PID (years)	Females: ∼U(0.024, 0.036)	Jackson et al.109	Modal value used by previous economic modelling studies as reviewed by Jackson et al.109 (assumed uniformly distributed with a ±20% uncertainty range)
Duration of health loss associated with chronic pelvic pain (years)	Females: ∼5 years with a lag of 5 years until development	Jackson et al.;109 lag estimate from Institute of Medicine102 and Yeh et al.106	Modal value used by previous economic modelling studies as reviewed by Jackson et al.109 (assumed uniformly distributed with a ±20% uncertainty range)
Duration of health loss associated with ectopic pregnancy (years)	Females: ∼U(0.062, 0.092) with a lag of 5 years until development	Jackson et al.;109 lag estimate from Institute of Medicine102 and Yeh et al.106	Modal value used by previous economic modelling studies as reviewed by Jackson et al.109 (assumed uniformly distributed with a ±20% uncertainty range)
Duration of health loss associated with tubal factor infertility (years)	Females: ∼10 years with a lag of 5 years until development	Jackson et al.;109 lag estimate from Institute of Medicine102 and Yeh et al.106	Conservative value taken from those used by previous economic modelling studies as reviewed by Jackson et al.109 (assumed uniformly distributed with a ±20% uncertainty range)
Duration of health loss associated with epididymitis (years)	Males: ∼N(0.014, 0.022)	Institute of Medicine102	Values were estimated by a committee of experts in the USA (assumed uniformly distributed with a ±20% uncertainty range)
Treatment cost of chlamydia, £	∼U(28.49–39.77)	Adams et al.108	Includes receiving results, treatment and partner notification; costs scaled from 2012 data (assumed uniformly distributed with a ±20% uncertainty range)
Treatment cost of gonorrhoea, £	∼U(108.04–134.45)	Adams et al.108	Includes receiving results, treatment partner notification and follow-up appointment; costs scaled from 2012 data (assumed uniformly distributed with a ±20% uncertainty range)
Treatment cost per case of PID, £	Females: ∼U(164.30–246.46)	Aghaizu et al.112	Includes clinician time, consumables and treatment. Based on scaled cost from £163.00 mean cost in 2008 (assumed uniformly distributed with a ±20% uncertainty range)
Treatment cost per case of chronic pelvic pain per year, £	Females: ∼U(2206.39–3695.36)	Armour et al.113	Based on study in Australia, which has comparable costs to the UK. Average per-person, per-year costs for health care were estimated to be international US$2528–4234 for total health costs in 2017. We translate to UK currency and scale to 2020 costs. We assume on average that chronic pelvic pain will start exactly mid-way throughout the year in which it is contracted
Treatment cost per case of ectopic pregnancy, £	Females: ∼U(1119.94–1679.90)	Thomas and Cameron114	Based on scaled cost from £1228 mean cost in 2008 (assumed uniformly distributed with a ±20% uncertainty range)
Treatment cost per case of tubular infertility, £	Females: ∼U(699.55–6295.97)	NICE115	Costed on average as one round of IVF, scaled from £3123 in 2013. However, as the proportion of women with tubular infertility who will undergo IVF is unknown and the average number of IVF treatments given to couples in the UK is 1.8, we assume a uniform distribution and include a wide ±80% to encapsulate a large degree of uncertainty
Treatment cost per case of epididymitis, £	Males: ∼U(164.30–246.46)	Adams et al.116	Based on scaled cost from £142 mean cost in 2004 (assumed uniformly distributed with a ±20% uncertainty range)
Cost of texts (per person), £	Option 1: 2.52	safetxt	Option 1 is based on the safetxt trial SMS provider costs, which at the cheapest are 3.5 pence per text with, on average, 72 SMS messages sent to each participant (£2.52 in total). Option 2 utilises existing infrastructure within the NHS, which provides free SMS text messaging to patients such as MJOG or Accurx
	Option 2: free
Cost of maintenance (yearly), £	Option 1: 300.00	safetxt	Option 1 is based on the cost of the virtual mobile number given by the safetxt trial provider at £25 per month (£300 in total). Option 2 utilises existing infrastructure within the NHS, which provides free SMS text messaging to patients
	Option 2: free

The input population consists of the expected number of individuals in England who will meet the eligibility criteria over the next year. This is estimated from looking at the number of chlamydia and gonorrhoea infections diagnosed in 2019 among those aged 15–24 years using data from Public Health England. 99 The model then uses trial data to estimate the proportions of this population who will be reinfected with chlamydia or gonorrhoea and whether this will be a symptomatic or an asymptomatic infection. The model then examines the existing literature to estimate the number of individuals who will develop common sequelae. In females, this consists of pelvic inflammatory disease, which can in turn progress to chronic pelvic pain, ectopic pregnancy and tubal factor infertility. In males, the only sequelae considered is epididymitis (Figure 7).

FIGURE 7.

Flow chart model showing disease progression in (a) females and (b) males. Solid arrows represent possible health state transitions. Dashed arrows represent the additional impacts of partner notification and number of secondary cases prevented.

To each disease state we attached a cost relating to treatment and considered the duration of the induced states and their associated health utilities measured in quality-adjusted life-years (QALYs). We then estimated the costs averted and health benefits gained from implementing the intervention. Although the model considers the intervention running for only 1 year, we estimated the effects of the intervention on an individual over longer periods of time due to the long delay involved in the development of some sequelae and their long-term impacts.

As a separate measure, we also included the functionality to estimate the expected number of secondary infections (infections occurring in the subsequent partners of those who were contact traced from an index case) likely to be prevented by the increased partner notification driven by the intervention. We used an estimate of the expected number of secondary infections prevented using findings from a previous study, which estimated these values based on the number of partners traced for those 16–24 years of age and further stratified by gender. 100 Finally, we considered the costs of the intervention, including the cost of the intervention per individual in terms of the cost of text messages and maintenance of the virtual mobile number. We performed 1000 individual model projections, sampling parameters from within their range of uncertainty to ascertain a 95% credibility interval for our results. The primary outputs were total costs, QALYs and incremental cost-effectiveness ratios (ICERs).

Next steps

The model has been developed to the point of a functional draft. If the cost analysis had gone forward, this would have been fine-tuned for costs of treatment and explored with a clinical team. Some of the costs, such as for treating tubal factor infertility, are complex and difficult to fully determine. Alternative parameter distributions could also be applied to parameter estimates that throughout we found to be highly uncertain. For many parameters we made use of uniform distributions as we could only obtain a point value from the literature or were faced with a significant range between different sources in the literature. The model could also easily be adapted to be UK focused by scaling up the population included proportionately or using further data from the devolved nations. Furthermore, the model could readily be extended to 10,000 or 100,000 parameter samples if required. Finally, further sensitivity analyses could also be undertaken, examining alternative costings, probabilities of disease transmission or differences in reinfection/persistence rates for chlamydia and gonorrhoea.

Strengths and limitations

Our trial had many strengths. The design, conduct and reporting of the trial followed CONSORT (Consolidated Standards of Reporting Trials) recommendations. 117 We ensured allocation concealment by using computer-based randomisation remote from the recruiting clinic site. Baseline prognostic factors were well balanced between the groups. Data collection and laboratory and statistical analyses were carried out blinded to treatment allocation. The primary outcome was known for 75% of participants in both the intervention and the control group. It is challenging to achieve high follow-up for objective or self-reported outcomes in trials of sexual health interventions with young people. Although we used evidence-based methods and developed our own trial specific procedures to achieve higher follow-up for laboratory-assessed chlamydia and gonorrhoea than other trials with similar populations,91,118 some potential for bias remains. Our primary analysis used multiple imputation methods because MICE is recognised as a way of reducing bias in and increasing the precision of trial results. 28,67 Losses to follow-up are not likely to have resulted in significant bias, as the complete-case, additional and sensitivity analyses making different assumptions regarding missing data all showed similar results. The higher than anticipated event rate in the control group improved the precision of the trial results. Highly sensitive and specific NAAT PCR (polymerase chain reaction) tests were used to assess chlamydia and gonorrhoea infection. Our recruitment across the UK and across sociodemographic groups, combined with no evidence of heterogeneity of effects in subgroups, suggests that the results are generalisable across the UK. The primary analyses were on an intention-to-treat basis.

Our trial also had some limitations. It is possible that involvement in behaviour change trials confers a Hawthorne effect, whereby the behaviour of the control group alters as a result of participating in the trial. Trial follow-up procedures could be experienced as a form of behavioural monitoring that can alter behaviour. However, the follow-up procedures at the end of the trial would have been unlikely to have influenced the acquisition of STIs during the trial, and the follow-up procedures were the same for both groups, so this would not result in differential bias. Many control group participants reported in the open feedback section of the 12-month questionnaire that the monthly messages reminding participants to let the LSHTM trial team know about changes in address or contact details indirectly acted as reminders regarding safer sex, which may have reduced the intervention effect estimates measured in the trial. However, text message reminders alone are known to have only small effects on sexual behaviour. 33,34 As > 40% of participants in our trial changed their contact details or address at least once, it would not have been feasible to have achieved high follow-up without monthly reminder messages. Our conclusion, that the safetxt intervention content confers a small benefit on some safer sex behaviours but may increase numbers of partners and STIs, remains valid. Changes in self-reported behaviour in both groups between baseline and follow-up are notable; for example, the proportion of participants reporting two or more partners in the preceding year was 83.9% in the intervention group and 82.7% in the control group at baseline and 56.5% in the intervention group and 54.7% in the control group at the 12-month follow-up. These self-reported changes are also consistent with changes in behaviour reported following STI diagnosis in the wider literature. 96–98 Participants were not blinded to their allocation, which could have affected the self-reported outcomes; however, this would not have influenced the STI test results, as analyses were completed at an independent laboratory and the data were entered by a blinded researcher. Clinic data regarding partner attendance for treatment were incomplete as not all clinics collect these data or do so consistently. We randomised 11 people twice and excluded them from the analysis, but this would not have influenced our results. If other participants obtained a new mobile phone number and used a false name and date of birth, then they could have been randomised twice. It is not very plausible that participants would have gone to this effort to join our trial. If this occurred, it could reduce the power of the trial to detect an impact of the safetxt intervention. Our sample size calculation allowed for 2% of control participants viewing intervention messages, which was based on the safetxt pilot trial. If we assume that all control participants who reported that they read another participant’s messages read intervention messages, this would mean that there was 1% contamination (32/3125), so our trial has greater power to detect a difference in infection between the groups than planned.

Interpretation of the results

The safetxt intervention was hypothesised to reduce the incidence of chlamydia and gonorrhoea by increasing condom use with new partners, increasing testing before sex with new partners and increasing partner notification. According to the theory of change, the safetxt intervention would achieve this by increasing knowledge regarding STI, increasing self-efficacy in how to use condoms and self-efficacy in communication about condoms with partner(s), and increasing self-efficacy in telling partner(s) about their infection. The intervention did not aim to influence the number of partners or number of new partners. The intervention worked partly in the way hypothesised. The intervention increased self-efficacy in how to use condoms and knowledge but there was no difference in other intermediate outcomes. Overall, the modest increases in condom use and no change or suggestion of slight changes in other behaviours targeted did not result in a reduction in chlamydia and gonorrhoea infection.

There is a suggestion of slight increases in the incidence of chlamydia and gonorrhoea at 12 months in the intervention group compared with the control group, and there are a number of possible explanations for this. First, we considered whether or not these results could have occurred as a result of small chance imbalances in sexual behaviour between the intervention and control groups at baseline. Post hoc analyses adjusting for the number of partners in the previous 12 months at baseline did not alter the effects of the intervention on the primary outcome, namely the incidence of chlamydia and gonorrhoea, or the effects of the intervention on the proportion of people with two or more partners. It is not likely in a large trial such as ours that the effect would be due to other unmeasured differences in sexual risk behaviours between the groups at baseline. Second, we explored whether or not the effect could be due to follow-up data not being MAR as our primary analysis assumes. Additional, non-prespecified analyses, making a range of different assumptions about STI rates in those lost to follow-up in the intervention and control groups, yielded similar results to our primary analysis. Third, we explored whether or not the results could be because those in the control group were more likely than those in the intervention group to have been tested at a site other than their recruiting clinic between randomisation and follow up and, therefore, were less likely to have had a positive test result confirmed in our data set. There is no evidence to support this. We asked all participants reporting a positive STI test where they had been tested, and we sought data from all testing sites, not just the clinics involved in trial recruitment. Furthermore, the proportion of positive tests confirmed by testing sites (clinics, general practitioner surgery, online service or other) was the same in the intervention and control groups. Finally, we conducted a post hoc per-protocol analysis to explore the intervention effect among those who received the whole intervention. In this per-protocol analysis, there were slightly higher odds of chlamydia and gonorrhoea infection at 12 months than in the primary intention-to-treat analysis. The participants in the intervention group in the per-protocol analysis were slightly younger than those in the control group, but the per-protocol analysis adjusted for the same variables as the primary analysis, including age. The participant characteristics were otherwise similar. The consistency of the results in the primary, complete-case, additional and sensitivity analyses, combined with the slightly increased effect size in the per-protocol analysis, adds to the weight of evidence suggesting that the intervention may have slightly increased the incidence of chlamydia and gonorrhoea at 12 months.

The slightly higher proportions of the intervention group having a new partner and having two or more partners at 12 months is likely to have contributed to the increased incidence of chlamydia and gonorrhoea. However, our understanding of the mechanism of action for the unanticipated effect on sexual partnerships and increased STIs is limited as the trial was not designed to explore this. Our prior qualitative research and the analysis of open feedback comments provide little direct evidence from participant reports regarding why this may have occurred. There was one comment that the intervention ‘gave me the confidence to engage in a new sexual relationship with a new partner without worrying about unwanted consequences’ as well as many comments about how the intervention had increased caution about sexual relationships.

It is not likely that the approaches to promoting condom use in safetxt resulted in increased STIs as these approaches were adapted from and similar to the content of face-to-face interventions that increase condom use and reduce STIs. 12–14,16,17 Our analysis exploring heterogeneity in the pooled effect of the pilot trial and main trial data (see Chapter 4, Pooled analysis with the safetxt pilot trial data) is consistent with this view. This analysis suggests that the effects of safetxt content about condom use and STI testing on the incidence of chlamydia/gonorrhoea remain uncertain, but may differ from the effects of content that targets condom use, STI testing and partner notification among those diagnosed with a STI.

To promote partner notification, the intervention was designed to provide non-stigmatising, non-blaming STI information and examples of how others notified partners. We reviewed our qualitative research findings published in 2016 to consider plausible mechanisms for the unanticipated effects found. 53 The first possible mechanism relates to recipients’ reports in our qualitative research that the intervention reduced stigma about having a STI. In both groups, the proportion of participants with two or more partners is lower at the 12-month follow-up than at recruitment. However, if there was less stigma regarding STIs in the intervention group, then this may have resulted in a slightly higher proportion of participants in the intervention group with two or more partners at 12-month follow-up, in turn influencing STI infections. This is consistent with other research demonstrating that lower levels of stigma measured in individuals are associated with a larger number of partners, although in this research lower stigma was also associated with reduced rates of adolescent pregnancy. 119 Lower levels of societal stigma towards MSM measured at a country level have been associated with higher levels of both precautionary behaviours and country-level HIV prevalence,120 with the highest unmet need for HIV treatment in countries that have high levels of societal stigma towards MSM. Previous research has also demonstrated that lower levels of stigma regarding sex, STIs and sexuality are associated with higher levels of precautionary behaviours, such as testing for or obtaining treatment for HIV/STIs and using PEP and emergency contraception, and lower rates of unplanned pregnancy, whereas experiences of stigma have negative impacts on emotional and mental well-being. 121–126 In keeping with this wider research evidence, we note that levels of STI testing in clinic were slightly higher in the intervention group (1549/3123, 49.6%) than in the control group (1477/3125, 47.4%). Levels of STI in the postal samples collected at 12 months as part of the trial procedures were similar in both groups (170/3123, 5.44%, in the intervention group and 165/3125, 5.28%, in the control group), and almost all of the excess STIs in the intervention group were diagnosed and treated in clinics during the 12 months’ follow-up.

Some other trials have reported on interventions that inadvertently increased STI rates through increasing the number of partnerships. 127 In the safetxt intervention group, the balance of risk and precautionary behaviours may have resulted in slightly increased STI rates compared with the control group.

Positive approaches to sexuality and reproduction should recognise the part played by trust and communication, as well as pleasurable sexual relationships, in promoting well-being and enabling people to fulfil their sexual and reproductive health and rights. 52 Our qualitative research and open feedback from participants suggest that, from young people’s perspectives, many aspects of positive sexual and reproductive well-being were met by the safetxt intervention, such as impacts on confidence, agency, communication and precautionary behaviours. However, ethically, this should be achieved without increasing STIs and risks to physical health.

The long-term effects of the safetxt messages on condom use at 12 months were larger than those reported at 12 months in RCTs of single sessions of face-to-face counselling, telephone counselling, videos or other interactive digital interventions such as websites. 15,19,128–130 Some intensive face-to-face behaviour change interventions involving multiple weekly group meetings achieve reductions in STI and larger increases in condom use at 12 months than the safetxt intervention, but these have not proven practical for widespread implementation. 12–14,16,17 Our previous systematic review showed that messaging interventions probably increase STI testing (at any point in time, not specifically prior to sex with new partners) in general populations of young people,37 but our trial shows that testing before sex with new partners was not increased. STI testing prior to sex with new partners requires planning in relation to the timing of first sex, which may not reflect how many sexual relationships start. The intervention may have had little or no effect on partner notification or correct treatment of STI, as levels of partner notification and correct treatment in the control group in our trial were already high and high in comparison with partner notification levels reported in the UK. 26,27 In the UK, most clinics have partner notification pathways that both the intervention and the control group would have received, so it is possible that the effect of messages would be different in a setting where partner notification pathways are less developed.

Implications for health-care services

The safetxt intervention did not reduce STIs and may have slightly increased STIs, so we cannot recommend the implementation of the safetxt intervention as delivered to participants and evaluated in this trial in the NHS.

The components of the safetxt intervention promoting condom use were effective, and providers could consider implementing these. The safetxt condom promotion content was developed based on the content of face-to-face interventions that increased condom use and reduced STIs. The safetxt condom promotion content had larger effects at 12 months on condom use than web, video or telephone counselling or single sessions of counselling evaluated in RCTs. 15,18,128,131–134 Based on a cost of 5 pence per message, the condom promotion content costs £1.80 per person. Text messaging services are embedded in many primary care and NHS services. It would be relatively straightforward and low cost for health-care providers to send messages targeting condom use and condom use skills to at-risk 16- to 24-year-olds who opt to receive such messages.

Implications for further research

Our unexpected findings highlight the importance of evaluating novel health communication interventions in well-powered RCTs to reliably establish their effects, especially in the complex area of sexual behaviour. We have limited insight into the unanticipated impacts of the safetxt intervention identified in our RCT. These occurred despite earlier uncontrolled evaluations suggesting high acceptability and positive behavioural impacts of the intervention. The elements of the safetxt intervention targeting partner notification may have inadvertently increased sexual partnerships and the incidence of chlamydia and gonorrhoea, the mechanism of which could be explored in future work. We conducted some further qualitative interviews at the end of participants’ involvement in the trial that may provide further insights into the mechanism of action of the intervention. These are being analysed and will be reported in a separate publication.

New approaches to increase partner notification and safer sex are needed. Future interventions aiming to reduce the harmful effects of stigma in the area of sexual health should consider the potential impacts of interventions on sexual partnerships in their development phase. Further research to identify ways of achieving the benefits of stigma reduction without increasing STI risk is needed. The importance of RCTs in reliably evaluating the effects of interventions cannot be overemphasised.

Future research should evaluate the effects of blended interventions, as their effects on behaviour and STI outcomes could be larger than those of text messages or other digital media alone. Safetxt and other interventions delivered by automated text message have small impacts on behaviour. 28,34

We were not able to include all of the elements we would have liked in the safetxt intervention, as some content was not considered acceptable for delivery by text message (e.g. content on relationships). 32 A blended intervention with content delivered through a range of media, including some interactions with a real person, would have allowed us to include all of these elements. Such a blended intervention may also be feasible to implement in the NHS and other health-care services.

The use of artificial intelligence (AI)-facilitated interventions could be explored. AI could address some of the limitations that our intervention delivered by text messages had for some participants. For example, an intervention using AI-assisted learning could generate a more tailored intervention, allowing the most relevant content likely to result in the largest behaviour change to be shown to participants, taking into account their relationships and other contextual factors. Such an intervention could also be more responsive to changes in individuals’ circumstances over time, enhancing its effectiveness.

Conclusions

The safetxt intervention did not reduce the incidence of chlamydia and gonorrhoea; instead, there were slightly more STIs in the intervention group. The intervention increased knowledge, self-efficacy regarding how to use condoms and condom use at 12 months. Providers could consider implementing the content of the safetxt intervention targeting condom use. Even when uncontrolled evaluations suggest benefits, randomised controlled trials are essential for evaluating health communication interventions, which can have unanticipated effects.

Funders

The NIHR Public Health Research programme funded the trial but did not have any role in the design, conduct, analysis, interpretation or writing up of the results.

Trial Steering Committee

Professor Pippa Oakeshott (chairperson), Professor Caroline Free (chief investigator), Dr Andrew Copas, Dr Michael Brady, Professor Michael Ussher and Colum McGrady (patient representative); the observers were Rosemary Knight and Kimberley Potter. As the risk of harm from the intervention was considered low, the analysis was conducted once, at the end of the trial. There was no separate Data Monitoring and Ethics Committee. The TSC took on the ethical responsibility for the trial and monitored adverse events.

Trial Management Group

Professor Caroline Free, Ona McCarthy, Melissa Palmer, Irrfan Ahmed, Kimberley Potter, Lauren Jerome, Rosemary Knight, Megan Knight, Zahra Jamal, Faran Dhaliwal, Rebecca Matthews (December 2015–August 2016) and Rebecca Swinson (August 2016–August 2017)

Trial Statisticians

James Carpenter, Tim Morris and Phil Edwards. George Ploubidis provided oversight of the analysis of intermediate measures conducted by Melissa Palmer.

Collaborators

Caroline Free, Graham Hart, Julia Bailey, Ian Roberts, Rebecca French, Phil Edwards, Susan Michie, Katy Turner, Paula Baraitser, John Cairns, Karen Devries, Ford Hickson, Kaye Wellings, Ona McCarthy and Melissa Palmer.

Contributing trusts and health boards

Contributions of authors

Caroline Free (https://orcid.org/0000-0003-1711-0006) (Professor of Primary Care and Epidemiology, General Practitioner, Honorary Consultant in Sexual and Reproductive Health, trials, mixed methods research, digital and behavioural interventions) led and takes responsibility for the development of the intervention, the trial design, conduct and analysis and the writing up of the research.

Melissa J Palmer (https://orcid.org/0000-0003-3937-8070) (Assistant Professor, SRH, Quantitative Research) prepared data for the TSC, wrote the statistical analysis plan, analysed the quantitative intermediate measures, prepared tables, contributed to trial recruitment, was a member of the Trial Management Group throughout, developed trial recruitment and follow-up materials, wrote up the intermediate measures analysis and contributed to the report.

Kimberley Potter (https://orcid.org/0000-0002-4273-207X) (Trial Manager) was responsible for the day-to-day management of the tria and the setting-up, co-ordination and communication with trial sites. She was joint lead of the Trial Management Group, wrote the first draft of Chapter 3, Follow-up and commented on the final report.

Ona L McCarthy (https://orcid.org/0000-0002-9902-6248) (Assistant Professor, SRH, Mixed-Methods Research) co-developed the intervention, intermediate measures, and trial recruitment and follow-up procedures and materials. She contributed to the trial design and writing the statistical analysis plan, was a member of the Trial Management Group and contributed to writing the report.

Lauren Jerome (https://orcid.org/0000-0002-6959-1529) (Assistant Trial Manager) contributed to trial recruitment and follow-up, developed recruitment and follow-up procedures and materials, liaised with trial sites, was a member of the Trial Management Group throughout, wrote drafts of Chapter 3 and commented on the report.

Sima Berendes (https://orcid.org/0000-0001-7000-868X) (Research Fellow, SRH) analysed open feedback data, conducted the updated systematic review of existing evidence and wrote the first draft of that section in Chapter 1, and contributed to the writing of the report (Chapter 1, Chapter 5 and revision and editing of various other sections).

Anasztazia Gubijev (https://orcid.org/0000-0003-4317-5866) (Research Fellow, Qualitative Research, SRH) analysed the open feedback comments, wrote the first draft of Chapter 5 and contributed to the updated systematic review of existing evidence.

Megan Knight (https://orcid.org/0000-0003-2902-4818) (Trials Assistant) conducted trial recruitment and follow-up, developed trial follow-up procedures and liaised with trial sites, and was a member of the Trial Management Group throughout.

Zahra Jamal (https://orcid.org/0000-0002-3817-6795) (Trials Assistant) conducted trial follow-up, developed trial follow-up procedures, worked with the patient and public involvement group on trial results dissemination and was a member of the Trial Management Group.

Farandeep Dhaliwal (https://orcid.org/0000-0002-1106-5884) (Trials Assistant) conducted trial follow-up, developed trial follow-up procedures and was a member of the Trial Management Group.

James R Carpenter (https://orcid.org/0000-0003-3890-6206) (Professor of Statistics) wrote the statistical analysis plan and analysed the data.

Tim P Morris (https://orcid.org/0000-0001-5850-3610) (Senior Research Associate in Statistics) commented on the statistical analysis plan and analysed the data.

Phil Edwards (https://orcid.org/0000-0003-4431-8822) (Associate Professor in Statistics) wrote the statistical analysis plan in the grant application, commented on the statistical analysis plan and final report and contributed to whole-team meetings.

Rebecca French (https://orcid.org/0000-0002-1962-5022) (Associate Professor) contributed to the intervention development, supervised the analysis of open feedback comments and contributed to whole-team meetings.

Louis Macgregor (https://orcid.org/0000-0003-3611-2453) conducted and wrote up the health economics analysis.

Katy ME Turner (https://orcid.org/0000-0002-8152-6017) designed, supervised and co-wrote Chapter 6 and contributed to whole-team meetings.

Paula Baraitser (https://orcid.org/0000-0002-3354-6494) (Consultant in SRH, Mixed Methods Research) co-developed the initial concept of safetxt, contributed to intervention development, contributed to whole-team meetings and commented on the final report.

Ford CI Hickson (https://orcid.org/0000-0003-0395-374X) (SRH, Mixed Methods Research) led the adaptation of the safetxt intervention for men who have sex with men, contributed to whole-team meetings, and contributed to interpreting the mechanism of unanticipated effects of the intervention.

Kaye Wellings (https://orcid.org/0000-0003-1053-8640) (SRH, Mixed Methods Research) contributed to intervention development and whole-team meetings, and commented on the research application, protocol and final report.

Ian Roberts (https://orcid.org/0000-0003-1596-6054) (Clinical Trials) commented on the trial design and protocol, contributed to whole-team meetings and commented on the final report.

Julia V Bailey (https://orcid.org/0000-0002-5001-0122) (Senior Lecturer, SRH Clinician and Researcher, Mixed Methods, Digital Interventions) contributed to intervention development, commented on the application and final report, and contributed to whole-team meetings.

Graham Hart (https://orcid.org/0000-0001-9676-6577) (SRH) commented on the concept, application and final report, and contributed to whole-team meetings.

Susan Michie (https://orcid.org/0000-0003-0063-6378) (Health Psychology, Behaviour Change) contributed to the intervention design and whole-team meetings and commented on the final report.

Tim Clayton (https://orcid.org/0000-0002-1266-3288) (Clinical Trials) commented on the application and protocol, provided clinical trials advice and commented on the final report.

Karen Devries (https://orcid.org/0000-0001-8935-2181) (Professor of Social Epidemiology, Trials, Intervention Design, Epidemiology) contributed to intervention development, and commented on the application, protocol and final report.

Publications

McCarthy OL, French RS, Baraitser P, Roberts I, Rathod SD, Devries K, et al. Safetxt: a pilot randomised controlled trial of an intervention delivered by mobile phone to increase safer sex behaviours in young people. BMJ Open 2016;6:e013045.

McCarthy O, French RS, Roberts I, Free C. Simple steps to develop trial follow-up procedures. Trials 2016;17:28.

Free C, McCarthy O, French RS, Wellings K, Michie S, Roberts I, et al. Can text messages increase safer sex behaviours in young people? Intervention development and pilot randomised controlled trial. Health Technol Assess 2016;20(57).

French RS, McCarthy O, Baraitser P, Wellings K, Bailey JV, Free C. Young people’s views and experiences of a mobile phone texting intervention to promote safer sex behavior. JMIR Mhealth Uhealth 2016;4:e26.

Berendes S, Palmer M, Gubijev A, McCarthy O, Wilson E, Free C. Sexual Health Interventions Delivered by Mobile Technology: Protocol of a Systematic Review of Randomised Controlled Trials. CRD42020172323. York: PROSPERO; 2020.

Free C, McCarthy OL, Palmer MJ, Knight R, Edwards P, French R, et al. Safetxt: a safer sex intervention delivered by mobile phone messaging on sexually transmitted infections (STI) among young people in the UK – protocol for a randomised controlled trial. BMJ Open 2020;10:e031635.

Berendes S, Gubijev A, McCarthy OL, Palmer MJ, Wilson E, Free C. Sexual health interventions delivered to participants by mobile technology: a systematic review and meta-analysis of randomised controlled trials. Sex Transm Infect 2021;97:190–200.

Free C, Palmer MJ, McCarthy OL, Jerome L, Berendes S, Knight M, et al. Effectiveness of a behavioural intervention delivered by text messages (safetxt) on sexually transmitted reinfections in people aged 16–24 years: randomised controlled trial. BMJ 2022;378:e070351.

Ethics statement

Ethics approval was provided by the NHS Health Research Authority – London – Riverside Research Ethics Committee (REC reference 15/LO/1665) and the London School of Hygiene & Tropical Medicine (reference 10464).

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review. We will make data available to the scientific community with as few restrictions as feasible.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data are vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the PHR programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the PHR programme or the Department of Health and Social Care.

Treatment and 7 days’ abstinence after treatment

Telling partner (after initial diagnosis)

How others felt after a positive sexually transmitted infection test

Preventing reinfection, and information on specific sexually transmitted infections (depending on type of sexually transmitted infection participants had at baseline)

Condom use

Sexually transmitted infection testing

Contraception

Talking about sex

More information after reinfection/new sexually transmitted infection