Sexual health promotion for young people delivered via digital media: a scoping review

Sex and relationships education in schools

In the UK, there is a wide variation in sex and relationships education (SRE) policy and content because each country has its own SRE framework. While SRE is mandatory in schools in England,24 Northern Ireland25 and Wales,26 it is not mandatory in Scotland.

The provision of SRE to young people in England is based on national-level SRE guidance. 24 However, curriculum content is determined locally for local authority-run schools or individually for academies, in keeping with the national sexual health policy. Although SRE is mandatory in primary schools in England, it currently focuses only on biology, including puberty, reproduction and fertility. In secondary schools it is mandatory to provide at least basic information on biological aspects of human growth and reproduction, STIs and HIV as part of the National Curriculum for Science. 24 However, broader aspects of SRE, including relationships, sexual and domestic violence, same-sex relationships and sexuality, come under the personal, social, health and economic (PSHE) education curriculum and are not mandatory. Independent schools (including private schools and government-funded academy schools) do not have to follow the national curriculum and are not governed by the same statutory obligations as maintained schools. Parents in any school have the right to withdraw their children from all or part of any sex education except teaching on the biological aspects of human growth and reproduction, which are specified in the National Curriculum for Science. This means that while sex education in some schools is excellent, in others children are not offered the information, skills and resources necessary to learn to enjoy safe and pleasurable sex lives.

In Wales, SRE is a compulsory part of the basic curriculum in all secondary schools. 26 The Welsh government recommends that primary schools should have a graduated programme of SRE tailored to the age and emotional maturity of the children. 27 PSHE has been a compulsory part of the basic curriculum in both primary and secondary schools since 2003. In Scotland there is no statutory requirement for schools to teach sex education. In 2001, the Scottish Executive published a circular on sex education in Scottish schools, which encouraged all schools to provide sex education within a comprehensive programme of PSHE and religious and moral education. 28 In Northern Ireland, relationships and sexuality education (RSE) is included on a statutory basis in the school curriculum. 25 However, while policy documents are positive about providing comprehensive RSE, in practice there is variable provision in schools.

Across the UK, some schools provide a very well-planned comprehensive SRE curriculum, while in others there is very little provided. Parents or guardians are entitled to withdraw their children from all or part of a planned sex education programme, meaning that many young people receive only minimal SRE.

In recent years there has been a growing call for all children and young people to receive high-quality SRE. 4,29,30 An Office for Standards in Education, Children’s Services and Skills (Ofsted) report has called for improvement in SRE because over one-third of the primary schools in England did not address issues related to physical and emotional changes during puberty and in secondary schools, and many SRE programmes did not address the impact of relationships, sexuality and pornography on understanding of healthy sexual relationships and safety. 31 Although pupils had received education on sexual risk and staying safe, they had not developed, for example, assertiveness and negotiation skills. There are also calls to include internet and media literacy in curricula. 24,32 In addition, teachers may lack expertise in teaching about sex and sexuality, and the assessment of pupils’ learning of PSHE may be poor in comparison with other subjects. Experts and researchers in sexual health have also called for the National Institute for Health and Care Excellence (NICE) to resume its work on the modernisation of SRE and to embrace a holistic perspective to focus on relationships, communication and consent in addition to biology. 33 Access to information technology (IT) in schools varies across the UK, with some schools providing IT both within the school and for pupils to access from home, and others providing only minimal IT access. 34

Sexual health promotion in clinic and community settings

The provision of sexual health services for young people varies across the UK. Young people may find it hard to access clinic services because of their concerns about privacy and confidentiality and fears about staff being critical or unfriendly. 35 The convenience of clinic locations and opening hours is particularly important for young people, who may face restrictions on where they can go and when. 35

Sexual health promotion forms a core component of care in sexual health clinics (e.g. as a part of a health advisor role), but the quality and intensity of advice or support is variable. The UK Guidelines on Safer Sex Advice recommend evidence-based behaviour-change interventions for those at higher risk of STIs (including young people), focusing on enhancing communication skills and increasing motivation to adopt safer sexual behaviours. 36 Techniques such as motivational interviewing are recommended to increase motivation to adopt safer sexual behaviours, and the guideline recommends that detailed and tailored information on safer sex should form part of all sexual health consultations (in settings such as general practice as well as sexual health clinics). The NICE guidance on one-to-one interventions to reduce STIs and under-18 conceptions recommends that clinicians assess STI risk when the opportunity arises, for example when someone attends for contraception or registers as a new patient. 37 The guidance recommends offering advice on STIs and contraception as part of a structured discussion, especially for vulnerable young people who may be at higher risk of poor sexual health, such as those from disadvantaged backgrounds and those who are in or leaving care.

It is difficult to ensure consistent, high-quality delivery of face-to-face, one-to-one behaviour-change interventions by clinic staff, and interventions are costly and time-consuming. Clinics are often overstretched and there is not necessarily funding available to train and mentor staff to ensure the maintenance of high standards of intervention (such as skills in motivational interviewing). In addition, sexual health as a topic can be difficult for both patients and practitioners to broach. 38

Outside school and clinic settings, many localities have outreach sex education for young people, but the funding of these programmes is often local and short term, and provision varies across the UK. In community settings, group interventions can be effective, but these tend to be available for specific groups, such as MSM only. Charities such as the Terrence Higgins Trust and Gay Men Fighting AIDS (acquired immunodeficiency syndrome), have embraced digital media for sexual outreach work, including HIV prevention within populations at higher risk of HIV, such as MSM and African migrant groups.

Young people wish to receive sexual health advice from parents, teachers and doctors, as well as online,39 but at the moment sexual health programmes in schools, clinics or the community do not adequately meet young people’s needs. Face-to-face interventions are expensive to provide and it is difficult to ensure that interventions are delivered as intended. Digital interventions have the potential to usefully supplement and augment face-to-face teaching and/or health-care provision.

Digital health: key policy events

There have been important policy commitments to the integration of digital communications into health-care systems. In 2002, the Wanless Report Securing Our Future Health: Taking A Long-Term View detailed the spiralling costs of health-care and the importance of self-care, and recommended improving the use of IT in the health service to improve quality and productivity. 40 This was followed by Securing Good Health for the Whole Population,41 which recommended the development of internet and telephone services to make self-care advice freely available. From 2005, the Connecting for Health programme was introduced to implement the NHS national programme for IT to create a single, central electronic care record for patients and to connect general practitioners (GPs) to hospitals. However, after enormous expenditure, this programme was abandoned after encountering insurmountable technical and security problems. The Innovation, Health and Wealth initiative from 2012 provides policy support for innovation of all types within the NHS, including a roll-out of telehealth and support for electronic care planning. 42 Under this programme, there are new Academic Health Science Networks, a new legal duty for Clinical Commissioning Groups and the NHS Commissioning Board to promote innovation, a Specialised Services Commissioning Innovation Fund, investment in the Small Business Research Initiative, Innovation Challenge Prizes, new networks with industry and financial incentives (Commissioning for Quality and Innovation). The 3millionlives programme (www.3millionlives.co.uk) is a Department of Health initiative to implement telehealth and telecare services for the 3 million people with long-term conditions and/or social care needs.

The Department of Health’s Digital Strategy: Leading the Culture Change in Health and Care43 aims to put patients in control of their health and care information, to make services more convenient and more joined up and to promote more transparent sharing of information. The policy proposes implementing the Digital First strategy for health and care to deliver innovation (such as bringing together information and services from across the NHS, public health and social care into a single integrated customer service platform, developing an active community of digital professionals within the health and care system and providing a digital workspace to share best practice, case studies and digital knowledge). The principle of Digital First is that it aims to reduce unnecessary face-to-face contact between patients and health-care professionals, using technology to make access to health care more flexible and convenient for patients and at a lower cost [e.g. online, by telephone, by e-mail or via Skype™ (Microsoft Corporation, Redmond, WA, USA)]. Ideas include pre-assessment or triage (online or by telephone), online booking of appointments, automated appointment reminders, remote follow-up and electronic communication of results and hospital letters. 44

The proposals in the Department of Health’s Power of Information45 strategy aim to enable patients (in England) to access their GP health records and to book appointments and request repeat prescriptions online, and gives a commitment to making hospital and other records available online. Patients will be able to access information about health and social care services and will also be able to give feedback online. There will be national standards to ensure that locally developed IT systems can exchange information effectively and securely. The strategy also aims to encourage technology use by both patients and staff. The NHS England’s Five Year Forward View46 sets out a much greater commitment to disease prevention and breaking down boundaries between services to provide better co-ordinated and more patient-centred care. The Personalised Health and Care 2020 policy paper47 acknowledges the role of IT in empowering people to take charge of their own health by providing access to medical records and to accredited health and care apps and digital information. It sets out a commitment to facilitating digital information on a person’s health and care by 2020 for all NHS-funded services, and access to data on services and treatment outcomes. The National Information Board is setting the strategy and direction for the health and care system on IT and information, and is monitoring progress (www.gov.uk/government/organisations/national-information-board). The Research and Innovation in Health and Social Care Policy of March 201348 promotes health research and use of new technologies for the development of more effective treatments for NHS patients, and programmes such as Code4Health have been established for knowledge sharing and skill building, to facilitate the development and implementation of digital solutions for health and care. 49

Sexual health policy and digital health

In England, changes brought about by the Health and Social Care Act 201250 have led to significant alterations in the landscape of sexual health provision, which impact on the potential for use of digital technologies in the statutory sector. Councils have acquired commissioning responsibilities for sexual health under the umbrella of Public Health and, at the same time, greater local autonomy for decisions about how services are configured. Local strategy is driven by Joint Strategic Needs Assessments and services are commissioned from a diverse community of providers by local authorities. National-level policy guidance, support for local health communities and data analyses are provided by Public Health England to drive, but not determine, local decision-making. For local authorities, bringing different areas of health improvement, such as alcohol, drug misuse services and sexual health, under the same roof as other departments, such as education, may mean greater interdisciplinary collaboration. The changing context has brought new opportunities to address the wider determinants of sexual health and for aligning priorities across the health and social care sectors. However, primary care, a significant sexual health provider, is commissioned separately, and different sexual health providers may find themselves competing for rather than collaborating for a local authority tender to provide services.

In Scotland, young people are a priority group in health policy: reduction in unintended pregnancy, harm prevention and integrated approaches to sexual health are important target areas. Relatively high rates of HIV among MSM mean that this has greater priority than some other areas, such as chlamydia screening. Health and social care are integrated, with common budgets and integrated approaches to health care and education. Health care is commissioned almost exclusively from NHS providers, and that has facilitated national-level initiatives such as the National Sexual Health Electronic Patient Record, enabling more integrated approaches to care and sharing across services.

In Wales, care is commissioned from a single provider, with the allocation of resources for sexual health promotion devolved to local regions, much like in England, leading to the potential for variation in provision. Particular national health priorities for young people are safeguarding, identifying and preventing sexual exploitation and addressing domestic violence.

National-level ambitions for sexual health are laid out for England, Scotland, Wales and Northern Ireland (A Framework for Sexual Health Improvement 2013;51 The Sexual Health and Blood Borne Virus Framework 2011;52 The Sexual Health Wellbeing Action Plan for Wales 2010–2015;53 Progress and Priorities 201454). The importance of peers and relationships, self-esteem, culture and freedom from coercion in sexual relationships represent key underlying principles for the strategy documents. Wider influences on behaviour, such as drugs and alcohol, as well as the particular needs of risk groups, such as those with learning disabilities or mental health problems, are also highlighted. In England, the sexual health of young people also remains a policy priority, with two out of the three sexual health indicators in the Public Health Outcomes Framework (2012)55 being young-person specific – reducing chlamydia among 15- to 24-year-olds and reducing teenage pregnancy. Gender-based violence has become an increasing national policy priority both politically and across education and health policy settings.

Until recently, IT in government policy has been seen as facilitating communication, data storage, data retrieval and analysis, and ‘innovation’ is conceptualised as changes in the way that services are provided or new treatments for health problems. The main focus of policy initiatives nationally has been on electronic records (increasing access to information for health providers and patients), electronic communications (e-mail, mobile phones or videoconferencing) and ‘big data’ (collecting and analysing data sets of patient information), rather than on digital innovations for self-directed health promotion. IT is seen as an important plank in saving NHS resources by reducing the frequency and costs of face-to-face interactions with health providers. 47 Digital interventions could also help to shift the emphasis from the diagnosis and treatment of conditions to self-care and prevention. 56

Policy at national level promotes the use of technology and social media for sexual health education and promotion, with a focus on prevention through behaviour-change intervention. 4 The National Strategy for Sexual Health and HIV57 from 2001 mentions the use of multiple channels for helplines (e.g. digital TV and the internet), and encourages the development of technology in clinical decision support systems and computer links between clinical sites to facilitate good practice and shared care. The 2008 Review of the National Strategy for Sexual Health and HIV58 does not mention digital media, but the British Association for Sexual Health and HIV (BASHH) and British HIV Association guidelines on safer sex advice36 recommend that digital interventions be considered as an alternative or adjunct to face-to-face interventions.

Digital interventions in clinic settings

Figure 1 gives examples of the ways in which sexual health information can be communicated digitally before, during, after or separately from a journey through clinical care.

FIGURE 1.

Opportunities for digital interventions connected with clinical care.

As shown in Figure 1, digital media interventions may enhance patient access to information and/or services. Many digital health tools are already widely used, for example health information online (e.g. the NHS Choices Livewell website: www.nhs.uk/LiveWell/Pages/Livewellhub.aspx), search tools to locate services (e.g. www.nhs.uk/Service-Search/Sexual-health-information-and-support/LocationSearch/734), online symptom checkers (e.g. www.nhs.uk/symptomcheckers/pages/symptoms.aspx), electronic appointment booking (e.g. www.chelwest.nhs.uk/services/hiv-sexual-health/clinics/west-london-centre-for-sexual-health/book-an-appointment), text reminders for appointments,74 online test-kit ordering (e.g. www.checkurself.org.uk for chlamydia testing and www.tht.org.uk/sexual-health/About-HIV/HIV-postal-test for HIV self-testing) and results by text or e-mail. 75

Digital interventions for sexual health may enhance the reach of clinical services, for example automated text messages to encourage HIV testing,76 or an online facility to notify sexual partners of the need for STI testing and treatment (e.g. www.gmfa.org.uk/pn) and online support groups or discussion forums (e.g. www.tht.org.uk/myhiv).

Digital innovations may enhance clinical care, for example electronic triage (with the option of appropriate testing kits being dispensed in the clinic) or electronic history-taking and/or risk assessment prior to an appointment. Electronic decision aids for clinicians are available (e.g. websites to check for HIV drug interactions, such as www.hiv-druginteractions.org), as are those for patient use (e.g. ‘My Contraception Tool’: www.brook.org.uk/contraception/my-contraception-tool). 77 Clinician IT systems may also provide links to patient information, which can be passed on to particular client groups.

Digital media (especially mobile phones) can also be used for medication adherence, for example to remind people to take HIV medication78 or the contraceptive pill. 79–81 Text messages or voice messages via mobile phone can be used to support contraceptive choice, uptake and adherence. 82–84

Shared online clinical records and online medical consultation or therapy have not so far been widely available within the NHS, partly because of concerns about confidentiality, data security and medical indemnity for services provided outside routine NHS pathways of care. In contrast, the private sector offers many paid-for services online, including medical advice by e-mail and diagnosis, although these services may disclaim responsibility for the accuracy of information or diagnoses. Online ‘Q&A’ forums can offer an avenue for advice on sexual problems (e.g. the BASHH ‘Health Unlocked’ forum at https://healthunlocked.com/#bashh).

Digital interventions for sexual health promotion may encourage people to use sexual health services, for example text messages to promote STI testing,76,85–87 to increase discussion of sexual health with a health-care professional87 or to enhance the successful treatment of STIs. 88 Mobile phones and websites are also feasible ways to reach young people to encourage the use of condoms for safer sex. 87,89–91 Online outreach by health educators on social networking sites can target particular populations, such as young people92 or MSM, and national health promotion campaigns often have online components (e.g. www.facebook.com/NationalHivTestingWeek).

There are, therefore, a wide range of interventions which can facilitate convenient access to elements of sexual health services or which are designed to enhance aspects of care (such as adherence to medication). However, such digital innovations for sexual health are deployed to a variable extent across the UK, and most innovations associated with health services are designed to enhance the treatment of health problems rather than to address the prevention of sexual ill health. 56

Digital media interventions in schools

Figure 2 gives examples of the ways in which sexual health information can be communicated digitally before, during, after or separately from school lessons.

FIGURE 2.

Opportunities for digital interventions in schools.

In schools, there are opportunities for digital interventions either in or out of the classroom: IDIs can be self-directed93 or teacher-facilitated during lessons. 94 Many schools have IT systems which could be used to broadcast health promotion messages on whiteboards or to pupils’ phones, for example. 71 Schools may also use online programmes for learning and submitting schoolwork, and could allow access to health promotion interventions from home within these systems, with the use of passwords ensuring access to appropriate content. There are also online sexual health resources for schoolteachers, which could help to address some of teachers’ embarrassment or reservations about teaching SRE.

Summary: opportunities for digital interventions for sexual health

There are many possibilities for digital media interventions in clinics or schools (and online), but most of the currently available innovations either do not provide tailored advice or help or are costly in terms of health service staff or health educators. However, self-directed, tailored programmes could be offered in conjunction with other types of digital communications or health tools which are already available and in use.

Definition of sexual health and sexual health promotion

Interventions included in this review were those designed to promote sexual health, defining sexual health promotion as providing individuals, groups and communities with tools to make informed decisions about their sexual well-being. Sexual well-being/sexual health can be thought of as

a state of physical, emotional, mental and social well-being in relation to sexuality [. . .] Sexual health requires a positive and respectful approach to sexuality and sexual relationships, as well as the possibility of having pleasurable and safe sexual experiences, free of coercion, discrimination and violence. 5

Definition of interactive digital interventions

Digital media comprises technology such as the computer and the internet, mobile phone, compact disc, video, games and interactive television. Any of these modes of delivery may be used for sexual health education. We used the following definition of IDIs: digital media programmes that provide sexual health information and one or more of decision support, behaviour-change support and emotional support for health issues. 97 Programmes must be ‘interactive’, meaning packages that require contributions from users (e.g. entering personal data, making choices) which alter pathways within programmes to produce feedback that is personally relevant to users of the programme. Users may interact with programmes as members of a small group, as well as individually. Our main focus of interest is standalone IDIs accessed directly by users and not composite IDIs which have face-to-face elements as part of their mechanism of action (e.g. input from teachers or health professionals).

Focused review literature search

We conducted a focused review of the literature on IDIs for sexual health, focusing on all types of evidence concerning young people, sexual health and digital interventions in high-income countries over the past 10 years. Particularly where evidence was thin, we also drew on evidence from other (low-income) countries, other age groups, other health conditions and digital interventions that did not strictly meet the definition of ‘interactive digital intervention’. We term such interventions ‘digital interventions’.

Cochrane literature review

We conducted a separate (but overlapping) systematic review of the literature in order to update the Cochrane review ‘Computer-based interventions for sexual health promotion’,97 selecting randomised controlled trials (RCTs) of IDIs for sexual health promotion and extracting and synthesising data in meta-analyses where appropriate. The inclusion criteria for these two literature searches were slightly different (Table 1).

Patient and public involvement

We set up a project user panel of young people aged 16–24 years. The panel members (two male and one female) had previously used the internet for sexual health advice or information, or had previously attended a sexual health clinic. They were recruited via posters and pamphlets put up in three London sexual health clinics and via e-mail sent to Sexpression, a network of students in UK universities involved with SRE. They attended project meetings and received £30 per session of approximately 1–2 hours (including preparation time), and were reimbursed for their travel. Panel members commented on the acceptability of the proposed study procedures and the clarity and language used in research materials such as topic guides and information leaflets. They also helped with the recruitment of young people to focus group discussions by informing young people from their social network about the study. One panel member also commented on the clarity and content of a summary document for circulation to the public.

Target population and setting

The target population for this scoping review is young people aged 13–24 years, to include children of school age (to cover sex education in school) as well as those aged 16–24 years (who are at higher risk of STI acquisition and unplanned pregnancy). This review focuses on the impact and potential of digital media for sexual health promotion for young people, focusing on the UK (England, Scotland, Wales and Northern Ireland).

Database searching, selection and categorisation of studies

We searched for all types of evidence on intervention design and implementation, digital research methodology to evaluate IDIs, data on costs, cost-effectiveness and health economic outcomes and future potential for IDIs. We conducted a systematic search of 17 databases, including Cochrane Central Register of Controlled Trials (CENTRAL), Database of Abstracts of Reviews of Effects (DARE), MEDLINE, EMBASE, Cumulative Index to Nursing and Allied Health Literature (CINAHL), British Nursing Index (BNI) and PsycINFO (for a full list of databases, see the Cochrane review97). We searched for all types of evidence, that is quantitative evidence, qualitative evidence, theoretical papers, health economic analyses and editorials from 2002 to 2013 (April), which addressed sexual health and IDIs according to the definitions of the concepts specified above. A 10-year period was chosen because technological advances and patterns of use of digital technology change rapidly, rendering earlier interventions less relevant to future health promotion. We included only those studies that were published in English, owing to the cost and time involved in translating materials.

A total of 17,802 unique citations were identified. SW downloaded all of the citations identified by the search into Reference Manager software (Thomson ResearchSoft, San Francisco, CA, USA). Ten per cent of the titles and abstracts were independently double screened by three researchers (SW, SM and Salmaan Khan) as a quality check to establish concordance in screening of citations. Following this, the three researchers each screened one-third of the whole database to identify eligible studies.

The literature of core interest for the scoping review comprised any type of study concerning IDIs for sexual health promotion for young people in high-income countries. We also considered studies of DIs for sexual health from low-income settings; studies of digital interventions for sexual health which did not meet the definition of IDI; studies reporting research methodology; health economic evaluations of non-DIs or non-sexual health IDIs; and digital interventions for health promotion other than sexual health. These papers were considered supplementary literature.

All of the citations on the Reference Manager database were classified as follows:

• ‘unclear’ – potentially relevant study (i.e. needs further consideration on reading full text)

• yes, core citation

• yes, supplementary citation

• ‘no’ – does not meet (above-mentioned) inclusion criteria.

Eligible studies were further classified for their relevance to the research topics of interest:

• A: RCTs of effectiveness of IDIs

• B: research methodology

• C: intervention design, development and implementation

• D: health economic literature

• E: future potential for digital interventions

• R: relevant systematic reviews.

The full text of studies classified as ‘yes, core’ and ‘yes, supplementary’ were obtained and read by SW or SM to confirm eligibility and to check the preliminary categorisations. Two researchers screened full texts of papers if there was any doubt about eligibility or categorisation. The results of the search are shown in Figure 4.

FIGURE 4.

Results of searches conducted for the focused review.

Internet search (horizon scanning)

As there can be long delays before the publication of research papers, we conducted a targeted search using the Google search engine (Google Inc., Mountain View, CA, USA) to locate media articles, conference proceedings, web logs (blogs), videos of opinion leaders and interest group discussions for views on the future potential of digital interventions for (sexual) health promotion. As with the focused search, materials of core interest were those concerning young people, digital media and sexual health, but we also collected materials on technological innovations in health fields other than sexual health.

Search for key studies in related fields

To address some of the research questions, we conducted targeted searches for evidence beyond the database of studies on digital/computer-based interventions for sexual health, for example locating key studies on digital interventions in other health promotion fields, such as alcohol, smoking and drug use; sexual behaviour change; human–computer interaction; health economics; and sex education.

Data extraction and analysis

We immersed ourselves in the literature, reading all core and supplementary papers. We extracted data on study characteristics and the key findings of core studies using a Microsoft Excel^® sheet (Microsoft Corporation, Redmond, WA, USA), making notes on themes arising from papers. We synthesised the results of RCTs of digital interventions for sexual health promotion, as these provide the best evidence of effectiveness, but did not attempt to synthesise the results of non-randomised quantitative studies. We had planned to conduct a formal meta-synthesis of qualitative themes; however, we identified 873 published papers which were relevant to this scoping review (348 of which were of core relevance in terms of sexual health, digital media and young people) and we also drew on non-peer-reviewed evidence sources, such as online discussion groups and blogs. It was not possible or desirable to formally assess the quality of this number and diversity of sources of evidence. We summarised and synthesised the findings of studies narratively, discussing emergent themes and their significance in the team and with key informants.

We recorded and tabulated details of intervention characteristics, target populations and settings.

We searched for possible ‘front-running’ interventions (i.e. IDIs suitable for implementation in the UK), judging evidence of intervention effectiveness and cost-effectiveness, interventions’ suitability for young people in the UK and the feasibility of implementation in a UK context. We also searched for IDIs which have been developed but have not been evaluated in RCTs, judging their suitability as ‘front running’ according to whether or not they had been designed according to ‘best practice’ principles (see Chapter 3). We judged interventions by (a) whether or not they had a clear theoretical underpinning, including clear proposed mechanisms of behaviour change reflected in the content design, and (b) whether or not intervention content was likely to address users’ needs (i.e. users were involved in intervention development in a meaningful way).

We developed a facility for intervention developers or researchers to make details of digital interventions publicly available in a searchable database located on the Sexual Health Research Network website (www.sexualhealthnetwork.org.uk). The database summarises the key characteristics of the IDIs that we located, with details of authors or developers, interventions themselves (if available) and links to any available evaluations.

Types of studies included and timeline

We conducted a systematic literature search to locate published and unpublished RCTs which test the effectiveness and mechanism of action of IDIs for sexual health promotion. This work entailed updating the previous Cochrane review,97 which has been renamed ‘Interactive digital interventions for sexual health promotion’, to reflect changes in the way that interventions can now be accessed.

We used Cochrane Collaboration methods to search for eligible RCTs, evaluate their quality and extract and synthesise data. Full details of the methodology used for Cochrane review searches and meta-analyses are available in the published review. 97 In brief, we updated our comprehensive search strategy, adding new relevant search terms which capture changes in terminology within the e-health field. We searched electronic databases (including CENTRAL, DARE, MEDLINE, EMBASE, CINAHL, BNI and PsycINFO), databases of grey literature for unpublished work, trials registers and reference lists of published studies, and contacted study authors. We had no language restrictions. Details of the full search strategy are available as appendices to the Cochrane review97 and on request from the authors. To update the previous review, we ran the search from 2007 to April 2013, locating 15,655 unique citations concerning digital media and sexual health.

Cochrane review inclusion and exclusion criteria

We drew on the definitions of sexual health and IDIs as for the focused review, selecting only RCTs of IDIs for sexual health which exactly met these definitions. We included RCTs of the following types of study:

• IDIs compared with minimal exposure (e.g. usual practice or waiting list) or non-interactive forms of education (e.g. written information, non-interactive computer packages)

• IDIs compared with face-to-face sexual health educational sessions

• studies that compared two or more types of IDI, in order to compare the effects of different designs of intervention

• any type of economic evaluation of IDIs.

Citation screening and data extraction

SW downloaded all the citations identified by the search into Reference Manager software and deleted duplicate references. Two review authors (JB and SW) independently screened all citations, obtained and appraised the full texts of any candidate papers and extracted data using the Cochrane Consumers and Communication Review Group data extraction pro forma. 97 Study authors were contacted to clarify questions about study methodology and to obtain missing data. The quality of RCTs was judged using the methods recommended by the Cochrane Collaboration. Information about study and intervention characteristics, risk of bias and numerical outcome data were entered into Review Manager software (Cochrane, The Nordic Cochrane Centre, Copenhagen, Denmark), double-checking the accuracy of data entered.

Data analysis

We analysed and present separately the results for studies that compare IDIs with minimal intervention (group 1), those that compare IDIs with face-to-face sexual health education (group 2) and those that compare two or more different designs of IDI (group 3). Separate meta-analyses were conducted for type of outcome (cognitive, behavioural, biological or economic), selecting outcomes to combine using concepts derived from a theoretical pathway for sexual behaviour change100 (including sexual health knowledge, self-efficacy, intention/motivation, sexual behaviour and biological outcomes). In the full Cochrane review, we present the results of data syntheses for studies involving any age group and setting. In this review, we present the results of data syntheses for RCTs of IDIs for sexual health involving young people (< 25 years of age) in high-income settings only.

Interviews and focus groups with professional experts and stakeholders

We purposively selected stakeholders who were sexual health youth workers, e-health or sexual health researchers, clinicians, educationalists, policy-makers, intervention developers and digital technology and media experts nationally and internationally. We searched online to map e-health expertise in the UK, and selected expert key informants on the basis of information about their expertise which was available in the public domain (e.g. organisations’ or academic websites or contributions to online debates). We selected stakeholders on the basis of their professional position, experience and expertise. 102 We made contact by e-mail initially and sought permission to interview either face to face or by telephone, using semistructured topic guides to seek their views on digital intervention design, development, evaluation, implementation, health economic issues, digital research methodologies and future potential of digital interventions for sexual health. We conducted a total of 28 interviews with expert, professional stakeholders, including e-health and sexual health researchers, policy-makers, educationalists and software developers and digital media experts. We also conducted a workshop on digital media for sexual health promotion, attended by sexual health nurses, teachers and young people from across the UK, and a focus group with researchers working in e-health research (sexual health, alcohol harm reduction, diabetes and other health conditions) (see Appendix 1).

Focus groups with young people and parents

We conducted six focus groups with young people between the ages of 13 and 24 years to seek their views on their use of technology, the usefulness of digital interventions for sexual health promotion and future potential. We also conducted a focus group discussion with parents of young people to seek their views on the use of digital interventions for sexual health promotion for young people in schools and other settings. We used semistructured topic guides to seek young people’s and parents’ views on the role and potential of digitally mediated sexual health information, the design and delivery of digital interventions for sexual health promotion and future potential.

We recruited 42 young people to participate in six focus groups. We sampled purposively by age [young people aged 13–15 years (n = 21); 16–19 years (n = 12); and 20–24 years (n = 9)] and by gender (22 female and 20 male). We recruited the younger age groups (13–15 years) from schools, with the help of teachers. Those aged 16–24 years were recruited from a college by handing out flyers about the research with the help of a sexual health outreach worker and college staff, and via Twitter (Twitter Inc., San Francisco, CA, USA) and snowball techniques (with young people asking their friends). School staff sought consent from the parents of school children in advance. Focus groups were conducted in school/college settings or on University College London premises. All participants were recruited from London. Participants were from different ethnicities, with an equal representation of young people from white backgrounds and non-white backgrounds. Participants were offered a cash incentive of £15 for participating in focus groups.

We held a focus group with parents of teenagers, recruited with the help of schoolteachers (n = 9: one male and eight female parents). All participants provided written consent prior to the discussion and received a small token of gratitude for their time and contributions (£15).

Expert symposium

We held an expert symposium with 21 participants (including e-health and sexual health researchers, policy-makers, educationalists and software developers) to discuss areas of controversy concerning intervention design, development and implementation, health economics, digital research methodologies and the place and future potential for digital interventions for sexual health promotion (see Appendix 1).

We integrated stakeholder perspectives (young people’s, parents’ and experts’ views) with the findings from our literature review, producing a draft report of findings to circulate to expert symposium participants in advance.

Delegates were asked to:

• comment on and debate emergent findings (the draft scoping review report)

• debate contentious topics and clarify the reasons for lack of consensus

• make recommendations for investment and implementation where there is strong evidence for the effectiveness of interventions

• make recommendations on future research needed to resolve uncertainties on optimal methods for intervention evaluation

• make recommendations on future research needed to resolve uncertainties on the effectiveness and mechanism of action of interventions and effects on health inequalities

• make recommendations for the integration of digital interventions with current sexual health promotion policy and practice in the UK

• make recommendations for the integration of digital interventions with current practice in schools and health-care settings in the UK

• comment on the implications of digital intervention implementation for training and skill development (of researchers, educationalists, health-care workers and young people)

• make recommendations on commissioning priorities

• comment on the future potential for digital interventions for sexual health promotion in the UK

• generate ‘blue-sky’ innovative ideas to set an agenda for possible future research and development.

Youth symposium

We held a young people’s symposium with 23 participants to gather young people’s views on sexual health IDI design, aims, targeting and implementation in schools, clinics and online. We recruited young people aged 16–24 years from sixth-form colleges by handing out flyers about the research with the help of a sexual health outreach worker and college staff, and via Twitter and snowball techniques. Twenty-three young people participated in this symposium (see Appendix 1). Of these 23 participants, three were recruited via Twitter (i.e. they saw the Twitter feed and contacted the researcher organising the symposium), seven had previously participated in a focus group discussion, one was a user panel member, five were recruited through snowballing and eight were recruited with the help of the college staff. They majority of the symposium participants (n = 15) were male and eight were female. Participants were predominantly from non-white backgrounds. They were offered £40 for participation in the half-day symposium. We invited young people to debate the following topics and to make recommendations for intervention design, development and implementation:

• the role of IDIs for sexual health for young people online

• the role of IDIs for sexual health for young people in schools

• the role of IDIs for sexual health for young people in health-care settings

• the future potential of digital interventions for sexual health promotion in the UK

• generation of ‘blue-sky’ innovative ideas to set an agenda for possible future research and development.

Scope of the literature

Studies that contributed to our current understanding of best practice in intervention design for sexual health promotion IDIs for young people incorporated a range of both qualitative and quantitative designs, as well as descriptive papers. We drew on the literature across the following categories:

• effectiveness studies of any designs, incorporating those included and not included in the systematic review

• studies of the preferred use of different technologies by young people, their main sexual health concerns and how different platforms are used by young people to find sexual health information that they need

• evaluations that identify factors associated with and that may be used for quality assessment of existing sexual health resources for young people

• user and expert views undertaken prior to development or as part of formative evaluation to inform the design and development of particular interventions

• descriptive papers relating to processes in intervention design and development, including concept development, description of theoretical constructs, operationalisation, prototype development and formative and process evaluation.

We reviewed a diverse range of evidence of variable quality. For many sources of evidence, formal assessment of quality was not appropriate (e.g. ideas discussed in online discussion forums). We did not conduct formal quality appraisal, in order to preserve a broad focus on relevant issues and to cover the full scope of this field across a variety of disciplines.

Needs assessment

A needs assessment should be conducted at the outset of intervention development to identify epidemiological need and particular target populations and behaviours to be addressed. There is a pressing need to design interventions that are effective in responding to the high rates of STIs, acquisition of HIV and unintended pregnancies. A number of IDIs designed to improve the sexual health of young people were identified in this review. However, almost all were developed outside the UK (mainly in the USA), which limits their external validity. We identified only one digital intervention for young people that has been developed and evaluated in the UK (Sexunzipped). 39,109

Use of Behaviour-Change Theory

User perspectives

Stakeholder perspectives

Best practice in design

A comprehensive approach to intervention development involves synthesis of evidence about the mediating factors of sexual behaviour, user and expert input about the barriers and facilitators to the target behaviours, identification of relevant contextual factors and a structure for planning behavioural change mechanisms. 126,131 Few reported studies describe how these separate components can be linked together and operationalised to form the final product. One of the challenges for developing interventions is of retaining fidelity to the behavioural constructs while remain appealing and meaningful to the target audience and adapting for a particular context (see Chapter 6). We have proposed a model of best practice in intervention design (see Appendix 2) and in Box 4 we describe three studies which exemplify elements of good practice.

Assessing the quality of interventions

We have described good practice for intervention design. However, it is a challenge to keep pace with innovations in the commercial, voluntary and statutory sectors, given the time and costs of a rigorous approach to design. At present, there are inadequate mechanisms to control the quality of digital interventions. Health websites or mobile phone apps can be developed and released quickly, but without any assurance that information is accurate or that a programme has a positive and not a negative impact on health. Hundreds of new apps are produced each week, and these are currently unregulated. There are many more mobile phones apps for smoking cessation, alcohol use and exercise167 than for sexual health. These apps are rarely based on strategies and BCTs that are known to be effective,168 and some may cause adverse effects. For example, of 148 apps which claimed to estimate blood alcohol level, most gave inaccurate estimates. 169 This study found that alcohol apps were often used by young people in ways that facilitated risky behaviour. It is, therefore, important to know whether or not digital innovations are used in the way intended and do not have unanticipated adverse effects.

In terms of effectiveness of freely available websites, one study reviewed 21 publicly available interactive safer sex websites. 170 These websites targeted mainly safer sex messages and condom use for teenagers. The types and number of interactive features were highly variable, as were the identified Behaviour-Change Theories that could be applied. The threat of HIV and STIs, a discourse more likely to stigmatise young people,149 was more commonly featured than promotion of positive reinforcing messages and social norms. Most contained some skills and self-efficacy-building content. Interactive features that were likely to increase enjoyment and, therefore, potentially engagement, were variable in number; however, the number and type of interactive features that optimise behaviour change is not known.

If digital interventions meet the criteria for ‘medical devices’, they are subject to extensive, detailed product standards and controls on quality. Software that is intended to be used for the diagnosis, prevention, monitoring, treatment or alleviation of disease, injury or handicap or control of conception, can be considered a medical device. 171 Although some IDIs could meet the definition for being medical devices (e.g. if they include a symptom checker which suggests a diagnosis), most would not. Devices that simply provide information would not usually be considered medical devices.

It is very hard for users to evaluate the quality of digital resources. The general public trust the NHS brand, and there have been discussions about ‘kite marking’ online resources (websites) that have been assessed as medically accurate. There are debates for and against trying to do this,172 and it is practically very difficult, as there is such a profusion of resource online. It is particularly hard to guarantee medical accuracy in the era of Web 2.0, in which content is often posted by general users rather than by experts (interview, e-health researcher); websites may sometimes provide moderators who pre-approve comments made on bulletin boards, for example, but this is resource intensive. Mobile phone apps which are included in the NHS Health Apps Library have been reviewed to ensure that they are relevant to people living in England, use information from a verifiable or a trusted source (such as NHS Choices), comply with the UK Data Protection Act 1998,173 and are unlikely to cause harm to a person’s health or condition. However, there are a number of challenges. It is impossible to keep up with the pace of release of new health apps, the intervention content may change after any approval process and it is not clear who should meet the cost of approvals and quality assurance.

In an attempt to address issues of quality, Keller et al. 174 used a content analysis approach to assess HIV/STI prevention websites for teenagers. In their study, coding tools were developed and applied to assess communication, information provision and usability (such as currency, authority, readability and objectivity) of the websites. This is the only study that we identified that addressed the concept of developing agreed standards for sexual health promotion digital interventions. The NCSP highlights that the hugely variable quality of publicly available sites for ordering testing kits is, similarly, an issue. As this field has expanded, so the NCSP has developed a quality standard to assess whether or not a digital intervention meets this standard (interview, manager, NCSP). However, it may be that the quality criteria for users differ from those that are perceived by professionals. 175 Although the regulation and quality control of intervention content is very important, this should be proportionate to any potential risks; for example, digital interventions which offer prevention advice or ‘find-a-service’, but not medical diagnosis or treatment, could be designated ‘low risk’, so that innovation and implementation is not inhibited by cumbersome, time-consuming approvals procedures.

Security, privacy, confidentiality and ethical concerns about online interventions

The online environment can facilitate enhanced privacy but, on the other hand, it can facilitate the instant, wide sharing of personal information. For example, those seeking advice on a sexual problem via an online agony aunt or a self-help website may choose to remain anonymous by creating a username. On the other hand, dating apps, such as Grindr for MSM, for example, can allow instant access to photographs, location, sexual preferences and drug use, for the purposes of arranging sex. There is sometimes a mismatch between young people’s behaviour online (where young people often freely share personal details on social networking sites and have little concern about their privacy or data security)70 and adult concerns about privacy, confidentiality and data security.

There is a legal obligation in the UK to safeguard children under the age of 18 years who are at risk of harm. 176 However, it is not clear how this legislation should be enacted in an online context, for example what responsibilities a webmaster has if a child discloses sexual abuse online, especially if they have used a username and their personal details are not known. It can be difficult to determine the maturity of users, who may also be free to lie about their age and could be exposed to age-inappropriate material. 96,177 These issues raise the question of how and by whom interventions would be monitored in the longer term, and who should meet the cost. There is currently little public discussion of the ethics of automated data collection, for example the implications and potential risks of surveillance via geolocation, etc. There is debate about who owns the data collected by apps and websites, and users may not even know that such data have been collected. There are particular considerations about the potential sensitivity of sexual health-related data online or saved locally on mobile phones. For example, an internet browsing history or a sexual health app on a mobile phone could easily be seen by someone else. The pace of intervention development has outstripped the regulatory framework, but it is not clear what framework would be appropriate, particularly given that the internet is accessible across national borders.

Statistical analyses

Where appropriate, we pooled the results of RCTs using a random-effects model, which gives an estimate of the average intervention effect. A random-effects model is more conservative than a fixed-effects model, as it allows for statistical heterogeneity rather than assuming that differences between studies are owing to chance alone. 97 We used standardised mean differences (SMDs) for continuous outcomes and odds ratios (ORs) for dichotomous outcomes. These measures allow combination of outcomes which may have been measured using different scales, giving averages or ratios which are adjusted by trial size. We comment on the size of SMDs using Cohen’s rules of thumb, judging 0.2 to be a ‘small’ effect, 0.4 to be a ‘moderate’ effect and 0.8 to be a ‘large’ effect. 178

Aims and design of interventions

Eight of the 19 included studies focused on HIV prevention67,110,111,127,141–143,183 and three focused on STI prevention, including HIV. 68,112,113 Two studies addressed both STIs/HIV and unwanted pregnancy117,181 and two studies focused on the prevention of unwanted pregnancy. 179,184 One of these studies focused on sexual pleasure and well-being as well as safer sex. 181 One study focused on promoting positive sexual communication and preventing sexual assault,180 one focused on sexual health knowledge and decision-making166 and one focused on responsible sexual behaviour. 63 One study had a focus wider than sexual health, promoting also mental health and the reduction of alcohol and drug use. 182

All of the interventions were digital and administered either via a computer or online – we located no completed trials of interactive (i.e. tailored) digital interventions delivered by mobile phone. Two interventions were delivered using a CD-ROM. 67,113 Some studies provided feedback based on knowledge tests or behaviour assessment tailored according to answers given. 67,110,112,117,182,184 Tailoring (individual feedback) was also by gender and ethnicity,111 gender,112 ethnicity and sexual experience,113 risk profile,127 adolescent health norms,182 stage of change,143 health behaviours or relationship outcomes and communication skills,166 perceived threat of HIV/AIDS and perceived self-efficacy to refuse sex. 141 Tailored content was also provided through story scenarios or rehearsal of the communication and decision-making process63,68,183 and following virtual decisions180 (e.g. a virtual baby following a decision to have unprotected sex)179 videos and modelling exercises for skill development142,143 and a role-playing game for decision-making skills. 141

Theoretical framework of intervention designs

Interventions were informed by various theoretical frameworks, including the Four-Step Model of Assertive Responding;67 SCT;142,183 Gender and Power Theory;142 Decision-Making Process Model;63 Cognitive Restructuring of Irrational Beliefs;184 Mental Models and Decision Theory;68 Information Processing Model of Social Competence;180 Extended Parallel Process Model;117,141 Elaboration Likelihood Model of Persuasion;166 TRA and SoC;143 ‘Fluency-Based’ Computer-Assisted Instruction;127 AIDS Risk-Reduction Model;112 TPB and SCT;111 Integrated Model of Behavioural Prediction;113 Practice in Decision-Making;179 IMB Skills Model;110 and Integrated Behavioural Model. 109 Some studies drew on a combination of theories. 63,68,110

Consumer and stakeholder involvement in intervention development

Several studies involved stakeholders in the process of development of the intervention, including involving community institutions in programme development179 and expert opinion. 68,179,180 Several studies also involved user groups in the development of interventions, using various methods, such as focus group discussions,39,68,111,117,143,180 surveys,117 pilot testing127,183 and semistructured interviews. 141,181

Delivery of interventions

Most interventions were delivered to individuals but one was delivered to small groups. 179

Exposure to interventions

The duration and intensity of interventions varied across studies. Some studies involved one single session interaction with the intervention,67,113,141,143,166,179,183 while other studies involved multiple sessions or access to interventions over a period of time. 63,68,110–112,117,127,142,180,184

Participants and geographical settings

Some studies were conducted only among girls and some included both boys and girls (see Table 2). The majority of the studies were conducted in the USA, while two were conducted in the Netherlands, one was conducted in Singapore and one was conducted in the UK.

Study settings

Participants were recruited from the following settings: clinical or treatment settings (including one special service programme for previously pregnant adolescents),179 a sexual health clinic,111 emergency department,143 a health-care centre,68 a family planning clinic113 and a community treatment centre for substance abuse. 127 Some studies recruited from education settings: colleges110,112,179,180 and schools. 63,117,141,166,184 One study recruited from social service agencies67 and another relied on referrals from recruited individuals. 151 Two studies recruited via social networking sites or online. 111,142 Two studies recruited participants from multiple settings. 142,185

Types of trial design

The majority of the studies were two-arm individual-level RCTs; however, four were three-arm RCTs,68,112,141,183 two were two-arm cluster RCTs67,117 and one was a three-arm cluster RCT. 63

Follow-up periods

Studies varied in length of follow-up period: 2 weeks or less post intervention;67,113,141,143,166,179,183,184 3 weeks to 5 months;63,110–112,117,127,142,180,181 and 6 months post intervention. 68

Outcomes

Outcome data were available from 18 trials but we excluded the trial conducted by Roberto et al. 117 from meta-analysis because only two schools were randomised to either intervention or control group and, therefore, it is not possible to adjust for clustering effects. For full details of outcome data selection and synthesis please see the methods section of the Cochrane review. 97 No studies reported economic outcomes. Similarly, no studies measured potential harms apart from reporting any deterioration in measured outcomes.

Group 1: studies comparing interactive digital interventions with minimal intervention

For this analysis we combined data from 14 studies that compared the effect of IDIs with minimal interventions. Minimal interventions were defined as non-interactive interventions or non-sexual health comparators, such as waiting lists, leaflets, books or IDIs on non-sexual health topics. 97 Three studies used ‘no intervention’ as a comparator;63,182,183 two studies used waiting list control groups;67,180 and two studies compared IDIs plus standard clinical care with standard care only. 113,127 In three studies the comparator was a ‘placebo’: a computer-delivered nutrition tutorial110,142 and an IDI about career planning. 184 One study compared IDIs with leaflets or books with the same content68 and three used a non-interactive website with generic information. 111,112,181

Group 2: studies comparing interactive digital interventions with face-to-face interventions

Four studies compared IDIs with non-digital face-to-face sexual health interventions. Three studies used a lecture with same sexual health content63,180,183 and one study used face-to-face counselling. 143

Group 3: studies comparing two different interactive digital intervention designs

Three studies compared different designs of IDIs for sexual health interventions. One study compared the effect of messages conveying high threat of HIV together with messages which boost self-efficacy, with high-threat-only messages. 141 Two studies compared tailored IDIs with non-tailored digital interventions for sexual health promotion. 112,166

Comparison 1: are interactive digital interventions effective compared with minimal interventions?

Please see the Glossary for explanations of the statistical terms used.

Do interactive digital interventions improve sexual health knowledge?

Standard deviations (SDs) were calculated for Di Noia et al. 67 and Evans et al. ,183 allowing us to combine data from seven studies which reported sexual health knowledge outcomes. Data were obtained from authors for Bailey et al. 181 and Downs et al. 68 Data suitable for analysis were not available for Alemi et al. ,179 Kann63 and Marsch et al. 127 Meta-analysis shows a statistically significant positive effect on sexual health knowledge, with a SMD of 0.54 [95% confidence interval (CI) 0.17 to 0.92]. This is a moderate effect size using Cohen’s criteria178 and shows that IDIs for young people have a significant, moderate effect on sexual health knowledge. As shown in Figure 6, the I²-statistic was 92%, indicating considerable heterogeneity in the results (i.e. variation in outcomes between studies).

FIGURE 6.

Comparison of IDIs with minimal intervention: knowledge. df, degrees of freedom.

Do interactive digital interventions improve self-efficacy with respect to sexual health?

We combined data from eight studies which reported data on self-efficacy (Figure 7), with SDs calculated for Di Noia et al. 67 and Evans et al. 183 Data were obtained from authors for Bailey et al. 182 No data were available for Marsch et al. 127 Combining outcomes from studies gave a SMD of 0.11 (95% CI 0.02 to 0.20). This is a small effect size using Cohen’s criteria,178 showing that IDIs for young people have a small but significant effect on self-efficacy. The I²-statistic was 0%, suggesting that there was no statistical heterogeneity.

FIGURE 7.

Comparison of IDIs with minimal intervention: self-efficacy. df, degrees of freedom.

Do interactive digital interventions increase safer-sex intentions?

We combined data from four studies that measured intention on continuous measurement scales (Figure 8) (SMD 0.09, 95% CI –0.01 to 0.19), which show a non-significant effect on intention. Ito et al. 113 measured intention as a dichotomous outcome, showing an OR of 5.88 (95% CI 0.60 to 57.30). No data were available for Marsch et al. 127 The I²-statistic was 0%, suggesting that there was little statistical variation.

FIGURE 8.

Comparison of IDIs with minimal intervention: intention. a, Data from authors; and b, SD calculated from the F-statistic (F = 2.80). df, degrees of freedom.

Do interactive digital interventions have an effect on sexual behaviours?

We combined data from four studies that measured sexual behaviour as dichotomous outcomes (numbers of events) (Figure 9). The outcomes combined were condom use68,181,182 and number of participants who did not experience sexual victimisation. 180 Data were obtained from authors for Bailey et al. 181 and Downs et al. 68 The OR following meta-analysis was non-significant (OR 1.18, 95% CI 0.96 to 1.44), showing no demonstrable effect on sexual behaviour (measured as number of events). The I²-statistic was 0%, suggesting that there was no statistical heterogeneity.

FIGURE 9.

Comparison of IDIs with minimal intervention: sexual behaviour (dichotomous). a, Condom use during intercourse ‘always’; b, number of participants who did not experience sexual victimisation; c, condom use at last vaginal and/or anal sex; and d, condom use in past 3 months, every time with partner, only for sexually active group. df, degrees of freedom.

Five studies measured sexual behaviour on continuous measurement scales, with all studies measuring condom use. Data were obtained from authors for Bull et al. (clinic and online samples). 111 The combined SMD was 0.19 (95% CI –0.04 to 0.43), which shows no demonstrable effect on sexual behaviour (Figure 10). The I²-statistic was 71%, suggesting that there was considerable heterogeneity in study results.

FIGURE 10.

Comparison of IDIs with minimal intervention: sexual behaviour (continuous). a, Online sample: change in the proportion of condom-protected sex acts in 60 days; b, clinic sample: change in the proportion of condom-protected sex acts in 60 days; c, condom use in the past 30 days, only those sexually active since baseline; d, condom use in past 30 days; and e, condom use with current partner. df, degrees of freedom.

To increase the power to detect an effect, we combined dichotomous and continuous outcomes for sexual behaviour by expressing SMDs as ORs. 97 This allowed meta-analysis of outcomes from eight studies that measured sexual behaviour. The combined OR was 0.28 (95% CI 0.01 to 1.61), which shows a significant effect on sexual behaviour (Figure 11). The I²-statistic was 54%, suggesting that there was moderate heterogeneity in study results.

FIGURE 11.

Comparison of IDIs with minimal intervention: sexual behaviour (combined). a, Condom use ‘always’; b, number of participants who did not experience sexual victimisation; c, condom use at last vaginal and/or anal sex; d, online sample: change in the proportion of condom-protected sex in 60 days; e, clinic sample: change in the proportion of condom-protected sex in 60 days; f, condom use in the past 3 months, every time with every partner, only for sexually active subgroup (data from author); g, condom use in past 30 days; h, proportion of condom-protected sex acts in last 90 days; i, condom use with current partner. df, degrees of freedom; SE, standard error.

Do interactive digital interventions have an effect on biological outcomes?

Data for biological outcomes were available from authors for three studies. 68,111,181 As shown in Figure 12, the OR was 1.18 (95% CI 0.78 to 1.80), indicating no demonstrable effect on STI diagnoses.

FIGURE 12.

Interactive digital interventions vs. minimal intervention: biological outcomes. a, Self-reported STI; b, clinic sample: history of STIs; c, online sample: history of STIs; and d, laboratory chlamydia test. df, degrees of freedom.

Comparison 2: are interactive digital interventions as effective as face-to-face sexual health interventions?

We combined outcomes from group 2 studies to address the question of whether or not IDIs are as effective as face-to-face interventions, taking studies that compared IDIs with non-digital, face-to-face sexual health education. These trials are equivalence trials, in that it is a positive finding if there is no difference in effectiveness between IDIs and face-to-face interventions.

Comparison 3: how do interactive digital interventions work?

Are interactive digital interventions effective?

We found that IDIs have statistically significant effects as follows: a moderate effect on sexual health knowledge (SMD 0.54, 95% CI 0.17 to 0.92), a small effect on self-efficacy (SMD 0.11, 95% CI 0.02 to 0.20) and a positive effect on sexual behaviour (OR 1.28, 95% CI 1.01 to 1.61), but no significant effects on safer sex intentions (SMD 0.09, 95% CI –0.01 to 0.19) or biological outcomes (OR 1.18, 95% CI 0.78 to 1.80). IDIs, therefore, seem to positively influence some modifiable determinants of sexual behaviour (knowledge and self-efficacy but not intention). There were no data on adverse effects.

Are interactive digital interventions as effective as face-to-face interventions for sexual health?

Data were available from only one study183 on the effectiveness of face-to-face interventions compared with IDIs. The results of this study suggest that IDIs may be as good as or better than face-to-face interventions for sexual health knowledge acquisition (SMD 0.51, 95% CI 0.11 to 0.90) and intention (SMD 0.46, 95% CI 0.06 to 0.85), but not for self-efficacy (SMD 0.38, 95% CI –0.11 to 0.77). There are no data with which to draw conclusions about effects of IDIs on sexual behaviour, biological outcomes or adverse effects.

How do interactive digital interventions work?

The existing evidence on this topic is limited and focuses on the effects of tailoring. Data from one study shows no evidence of effect on knowledge improvement or message elaboration. 166 Another study provides no evidence of effect of tailoring on self-efficacy, intention or sexual behaviour. 112

Interactive digital intervention effectiveness and mechanism of action

• We located 19 studies which were RCTs of IDIs for sexual health promotion for young people.

• Interventions were evaluated in variety of settings (schools, colleges, health-care settings and online).

• Trials evaluated IDIs for heterosexual young people and young MSM.

• The majority of RCTs were conducted in the USA, with two in the Netherlands and one in the UK.

• Interactive digital interventions are effective tools for learning about sexual health.

• Interactive digital interventions have small but significant effects on self-efficacy, and sexual behaviour.

• There is not enough evidence to be sure of the effects on biological outcomes or to be sure of longer-term impacts.

Recommendations for research

Robust RCT evidence is needed in order to:

• assess the impact of IDIs on biological outcomes and on sexual well-being (e.g. pleasure and relationship and sexual satisfaction)

• evaluate the effects of IDIs in comparison with face-to-face sexual health services or teaching

• evaluate the mechanisms of action of IDIs (e.g. optimal theoretical design, including BCTs, and core components)

• assess optimum exposure, dosage and timing for interventions

• assess the effectiveness of alternative models of intervention delivery (e.g. direct vs. facilitated access, different technological platforms)

• assess longer-term impacts.

Implications for policy and practice

• Interactive digital interventions developed in other countries could be adapted for young people in the UK.

• Interactive digital interventions are effective tools for learning about sexual health.

Expert interviews

Three experts in health economics and either sexual health promotion interventions or public health were interviewed using a topic guide. Interviews were one to one with a researcher who took notes and then transferred the conversation into a document that was used to inform the content of the chapter. The views of the expert interviewees will be referred to as health economist (HE) 1, HE 2 and HE 3.

Literature review

Economic evaluations of sexual health promotion delivered by digital media were sought as part of the main systematic review (see Chapter 2 for detailed methodology). In addition to the criteria for inclusion (sexual health promotion delivered by digital media), to be an economic evaluation studies had to report costs and consequences of an intervention collected alongside a trial and/or in a decision-analytic model. Systematic reviews and cost of illness studies were excluded. Additional papers were identified through hand-searching references. Papers were checked for inclusion by two researchers. A total of 10 papers were identified as being potentially relevant. Of those 10, one paper was a cost-effectiveness analysis of a non-interactive digital media intervention for sexual health. 186 As so little directly relevant literature was identified, papers which did not fit the inclusion criteria have been used to inform the discussion (i.e. guidance documents on economic evaluations of public health interventions, papers of relevance to economic evaluation of digital interventions and economic evaluations in the sexual health field).

Economic evaluations in public health

• Economic evaluations are the comparative analysis of costs and consequences of two or more options and provide decision-makers with information about which option might result in the best outcomes given a budget constraint or limited resources.

• Cost–utility analysis, the incremental cost per QALY gained, is the type of economic evaluation recommended by NICE in the UK to allow for the comparison of results across disease areas and programmes of work.

• In public health there is recognition that cost–utility analysis might not be the most appropriate option, as QALYs are potentially too narrow to capture all of the outcomes of interest, in particular non-health outcomes (such as wider behavioural, environmental, education and social functioning outcomes).

• RCTs are generally considered the gold standard in determining the effectiveness of an intervention. However, they may not capture all of the information necessary to adequately assess the cost-effectiveness of an intervention. In these instances, decision-analytic modelling (collating information across a range of published sources) is recommended.

• Mathematical models and published data are needed to overcome some of the limitations of primary data collection.

Economic evaluations of sexual health promotion interventions

• Cost–utility analysis (composite mortality and morbidity) may not be suitable for assessing sexual health interventions as it may be hard to capture the main outcomes of interest, especially as the events, such as STIs, may be rare and costs and outcomes may occur a long time after an intervention.

• Cost-effectiveness analysis, expressed as cost per STI case detected or cost per pregnancy avoided, may be more relevant for assessing sexual health interventions. However, it is not clear what value should be placed on outcomes or how the results compare with cost-effectiveness analyses for other health-care programmes.

• Decision modelling of long-term outcomes allows for the collation of a wider range of information over a longer time horizon. When modelling sexually transmitted disease in particular, careful consideration needs to be given to the type and structure of the model used.

• The cost and effectiveness of medical treatments for diseases is constantly changing, with more effective but more expensive treatments regularly becoming available. As a result, economic evaluations can quickly become out of date in their prediction of costs and QALYs.

• High-quality data on the effectiveness of an intervention at preventing new infections are not always readily available, unless STI surveillance systems are in place.

Cost-effectiveness of interactive digital interventions for sexual health

• Digital media interventions provide a potentially low-cost way to access a large audience. This is mostly because of their scalability, as the main cost is in the initial development of the intervention.

• The high cost of treating STIs, in particular chronic conditions such as HIV, means that a cheap intervention needs to prevent only a few cases before it becomes cost saving.

• In some situations, particularly if the uptake of the intervention is low, cost-effectiveness may be determined by the number of high-risk people accessing the intervention as much as the magnitude of effectiveness of the intervention.

• Targeting those at higher risk of sexual ill-health may be most likely to make an intervention cost-effective.

• Although there is limited evidence, effective sexual health IDIs are likely to be cost-effective.

Recommendations for research

• The costs of development, implementation and maintenance of IDIs should be recorded and reported.

• Evaluations and trials of new interventions should include an appropriately designed economic evaluation to increase the evidence base in this area.

• Work is needed to establish how best to capture and measure outcomes in economic evaluations of sexual health IDIs (e.g. the SQOL scale compared with the generic health-related quality of life scale).

• Work is needed to establish the best ways to collect, value and apportion costs and consequences of IDIs in sexual health over longer time horizons, for example how to capture long-term health costs and consequences of new cases of STIs and how to correctly apportion the initial costs of development and implementation over a longer time horizon and a large number of users.

• The role of sexual health IDIs in reducing health inequalities requires further research.

Implications for policy and practice

• Although there is limited evidence, effective sexual health IDIs are likely to be cost-effective.

Frameworks

We refer mainly to the RE-AIM framework206 for structuring discussions and identifying the key issues and gaps in the literature that might be important for implementation of sexual health promotion IDIs in clinics, schools or for self-access through the internet. RE-AIM considers reach, efficacy, adoption, implementation and maintenance of interventions as the necessary and important elements contributing to impact. In this context, these concepts are taken to mean the following:

• Reach – how well it reaches its target audience. For IDIs this is influenced by factors that affect both access to and engagement with the intervention.

• Efficacy of the intervention on targeted outcomes – this is addressed in Chapter 4.

• Adoption – the types of setting in which interventions are implemented, the barriers and facilitators to their adoption in different settings and consideration of how they can be adapted appropriately.

• Implementation – how well and consistently the intervention is delivered in a particular setting (health and education) by staff and how this should be facilitated and overseen. We also refer to Normalisation Process Theory,207 a conceptual framework that seeks to determine the mechanisms through which an intervention might become embedded within educational and health-care settings.

• Maintenance – the factors influencing sustainability of interventions over time at individual, community and organisational levels. At an individual level this includes the degree to which behaviour change can be sustained over time. At an organisational level, this addresses how the use of interventions can be maintained over the longer term, what support would be needed including technical and financial and how institutionalisation of an intervention can be predicted.

In applying these frameworks, we found that the literature was weighted more heavily towards design and effectiveness studies, to a lesser extent towards reach and engagement and was very limited with respect to furthering understanding about the challenges of implementing sexual health IDIs in practice. We were, therefore, inclusive in our approach, incorporating a diversity of study designs and publication types for the narrative synthesis. We included any publication that gave insights into issues for implementation, an approach that precluded formal quality review of each paper. We have also drawn on expert interviews and symposia and user focus group discussions to supplement the literature where appropriate. Data were coded and analysed descriptively in relation to the RE-AIM and Normalisation Process Theory frameworks and triangulated with findings from the literature.

Reach

Digital interventions have the flexibility to be implemented either in a defined setting, such as a school or clinic, or through self-access wherever people are located, via the internet or, for example, a smartphone app. While most evaluation studies have been conducted in school or clinic settings, as most interventions are internet based, there is flexibility about how they might be delivered in practice.

The proportion of the target population reached by the intervention is fundamental to its impact, with a potential trade-off between effectiveness and reach. The population impact of an intervention with relatively low effectiveness can be mitigated if it reaches relatively high proportions of its target population. 144 However, little is known about the factors that might influence the reach of interventions in a real-world context and how this might differ for different target populations. Measures of recruitment to evaluation studies can be only a proxy measure for how well implementation might work (especially as incentives have often been used for research recruitment). In addition, studies from the USA strongly predominate in the literature, limiting the generalisability and learning for UK education and health contexts.

Adoption

Sexual health promotion IDIs may be implemented in schools, clinics or any other setting either to complement existing services or to replace elements of the sexual health promotion currently delivered face to face. Aims, likely impacts on current ways of working and anticipated benefits all need to be conveyed to frontline staff in advance in order for adoption to be effective. 207 Some adaptation of the intervention to local context is necessary for this process to be successful;228 however, this can also create tensions where retaining fidelity to theory may be at odds with the needs of context: the ‘local adaptation-fidelity’ debate. 229 It also assumes an understanding of which of the intervention ‘ingredients’ are active and must, therefore, be preserved in this process of adaptation. Initiatives arising from stakeholders ‘on the ground’ in clinic or educational settings have the advantage of being grounded in existing contexts and constraints, but are less likely to be underpinned by Behaviour-Change Theory (interview, e-health researcher).

In a clinical setting, there is huge potential for digital interventions to replace the simpler elements of clinical care, but there is anxiety among stakeholders about how this might threaten important functions of the face-to-face consultation [service providers, SASH (Studies in Adolescent Sexual Health) workshop]. ‘Buy-in’ from implementing staff is important for the effective implementation of technology, and a significant cultural change is still needed in both clinical and educational settings (interview, voluntary sector provider).

Implementation

We did not identify theory-based sexual health promotion IDIs from the literature that had actually been implemented in original or adapted form in practice in either a research or a real-world health or education setting. However, there are pockets of innovation in this area in both statutory and voluntary sector settings, and there is increasing motivation among some policy-makers and people on the front line for exploring this area further.

Maintenance

Ongoing management and financing digital interventions after implementation has been underexplored. Websites need continuous maintenance and review. Interventions incorporating a social networking element require posts to be checked for accuracy and relevance, as well as provision of moderation of discussions. One approach to moderation via social networking sites has been the use of the peer-led HIV prevention education via Facebook that has been adopted in the HOPE (Harnessing Online Peer Education) study, training peer educators in methods likely to be effective. 218

Small, regular updates of digital interventions are needed to incorporate new facts and research. The look and feel of a website also dates rapidly, meaning that more major revisions are required. These maintenance issues are costly in terms of financing and time and present a challenge for a low-resource environment. 89

Advantages of digitally mediated research methods

Digitally mediated research methods have several advantages compared with face-to-face methods.

Challenges of digitally mediated research methods

Several factors may threaten the validity of data collected using digitally mediated research methods.

Recruitment

Recruitment of young people to online studies is feasible offline (e.g. via health-care settings83,88), at community events249 or entirely online. 66,119 Online trial procedures can be completely self-directed/automated, for example accessing trial information, establishing eligibility, obtaining informed consent and randomisation to experimental arms. There is a possibility of losing a high proportion of potential participants during all of these stages. 66 However, several factors can enhance recruitment in online trials: exclusivity (i.e. intervention website access limited to research participants), user-centred design (i.e. involvement of user groups to develop online trial recruitment materials), findability (routing the website through well-known and respected sites), media advertisement and piloting. 99

Randomisation systems

Robust methods of randomisation in trials are essential to minimise allocation bias. Automated randomisation and concealment of allocation to study arms is technically easy, and can be seamlessly achieved as part of an online trial procedure. 111,139,185 Automated methods can also enhance the blinding of research personnel. Trials of sexual health interventions delivered by mobile phone have used paper-based data entry combined with remote randomisation systems, or web-based data entry systems which automatically link to a remote randomisation system. The randomisation system can then automatically send text or voice messages according to the group allocation. 83,88

Fidelity of digital interventions

It is important to know whether or not an intervention was delivered as intended. Fidelity can be defined as ‘the methodological strategies used to monitor and ensure the reliability and validity of behavioural interventions’. 229 There are different components of fidelity; content fidelity refers to whether or not the intervention components were delivered as prescribed, whereas process fidelity refers to the consistency with which intervention content was delivered. 250 Using robust study designs to avoid contamination between the study arms and standardised delivery of an intervention can enhance fidelity.

Digital interventions may be self-administered, with private, individual access which can facilitate controlled access to an intervention and reduce the risk of contamination between study arms. However, contamination between study arms can occur, for example if participants show each other text messages216 or websites. 237

The standardised format and medium of delivery of IDIs to trial participants ensures fidelity in administration of digital interventions, although, in practice, actual delivery is less controllable. A challenge with self-administered online interventions and short message service (SMS)-based interventions is to ensure that an appropriate dose is delivered and that the intervention is completed. Adequate exposure means engaging with interventions for long enough (or a sufficient number of times) to be able to process the information. 206 A lack of engagement can affect intervention fidelity251 and underestimate the potential efficacy of an intervention.

Among adolescents who do not have a specific health problem they wish to address, it is a challenge to sustain their interest and, therefore, their exposure to digital interventions. A review of studies showed that theoretical foundation of the intervention, perceived personal relevance to the user, perceived effectiveness, persuasive technologies, credibility and social networking also influence engagement with online interventions. 252 Self-regulatory strategies, such as inviting users to commit to revisiting the intervention (if required), by indicating a time when the user might like to revisit, can be used to enhance exposure. 253 As described in Chapter 3, design features of interventions can also influence engagement. Thus, it is important to conduct usability testing and involve users in the development of online intervention content and to conduct feasibility testing to avoid technical challenges and ensure smooth access and navigation. Various strategies, such as e-mail reminders to log in for booster sessions111 or return for follow-up sessions,66 resending intervention messages90,120 or giving financial incentives,68 have been used to enhance exposure to interventions.

Differential usage of interventions by trial participants can influence the validity and reliability of the trial results. Eysenbach254 suggests that researchers should report raw drop-out proportions at different points in time, as well as ‘usage half-life’, which indicates after how much time 50% of the participants will have stopped using interventions. Analytics software can automatically record visits to individual webpages185 or activities, such as writing or responding to posts, updating medication adherence data, viewing articles, etc. 255 To ensure exposure, willingness to return to the intervention site at least twice during the study was used as one of the eligibility criteria for study participation. 111 However, with self-administered digital interventions, it is hard to know how they are used in practice; young people in particular may be multitasking while programmes are open. Defining exposure/usage of digital interventions can, therefore, be a challenge. There is a need to define what constitutes usage/adherence to digital interventions and to develop mechanisms for reliable measurement.

Retention in randomised controlled trials of digital interventions

Trial feasibility is described as the percentage of participants recruited and retained at each assessment point. 256 Low levels of retention in trials can affect the power of the study and external validity. A common challenge in online trials is the high proportion of dropouts before completion of the trial. 230,247,254 However, some online trials have shown higher retention rates, for example 90% at 1 month,255 72% at 3 months185 and > 70% in a trial with a follow-up period of 1 year. 239 In a trial of an IDI delivered by automated voice message, 86% follow-up was achieved for self-reported data at 4 months. 83 A trial of a digital intervention delivered by text message to increase safer sex behaviours achieved 93% follow-up for self-reported data at 1 month and 86% follow-up for chlamydia tests at 3 months. 88 Eysenbach has hypothesised that dropout from a trial usually follows the pattern of non-usage of an intervention,254 although this was not substantiated in a trial of a website for alcohol harm reduction, in which greater adherence to the intervention was not necessarily associated with greater retention, especially in the intervention arm, and factors associated with intervention usage and retention in the trial were not similar. 252 There is a need to conduct further research to better understand factors influencing retention in trials of digital interventions and engagement with digital interventions. Various strategies can be used to enhance retention in trials of digital interventions, as follows.

Nature of outcomes measured for sexual health interventions

The choice of outcomes to measure should reflect the theoretical design of an intervention; in other words, it should reflect the aims of the intervention and capture the behaviours and health outcomes targeted. It is also important to measure the modifiable determinants of behaviour which are specified in hypothesised pathways of action (such as knowledge, self-efficacy, perceived susceptibility and intention), to understand possible mechanisms of action of interventions. 266

Researching sexual behaviour is challenging because sex and sexuality are embedded in webs of often contradictory social significance. 267 Qualitative research gives access to deeper understandings about sexual behaviour, but does not yield quantifiable estimates of the effect of interventions. Capturing a ‘true’ picture of sexual beliefs, attitudes and behaviour is challenging with quantitative tools because standardised questionnaires strip away the complexity and context. 268 Qualitative research can be conducted online (e.g. interviews by instant message or group discussions in chat rooms), but data from these may be less rich than those which are attainable from face-to-face or telephone interviews. Quantitative data collection (online) is cheaper, quicker and more convenient than qualitative data collection, but qualitative understandings are necessary to be confident that research findings are meaningful. 269

Sexual behaviours are often complex. For example, ‘condom use’ involves many steps: a condom user needs the knowledge of how to use condoms correctly, the confidence to use it without adversely impacting on sexual performance or pleasure, the motivation to use it, to obtain and carry condoms and skills to negotiate with partner(s). ‘Condom use’ is, therefore, not just one behaviour, but a series of behaviours requiring the involvement of others. In addition, ‘condom use’ needs to be understood in context; for example, condoms may be seen as less relevant for a monogamous heterosexual couple using the contraceptive pill and with negative STI screens. For these reasons, composite sexual health outcomes which take into account a number of factors may be more meaningful. 270

There is a lack of standardised outcome measures, even for behaviours such as condom use or self-efficacy. Few studies capture factors which are important to young people (e.g. measures of sexual well-being), and there is little agreement about how best to capture dimensions of social and emotional well-being, although researchers are turning their attention to these issues. 21 Interventions may target several health behaviours (e.g. alcohol consumption, drug use and sexual behaviour)120 but this may present challenges for measurement, as outcome questionnaires may need to be long and spurious associations may be found simply by chance if large numbers of outcomes are measured.

Behavioural and biological sexual health outcomes may not be appropriate for all age groups. For example, interventions may aim to impact on condom use, but young people may not be experiencing penetrative sex with sufficient frequency to capture changes in condom use. Measuring the determinants of behaviour might be more appropriate (e.g. self-efficacy or intention), and/or over much longer follow-up periods. Age-appropriate outcomes should, therefore, be used in research with young people.

Reliability of outcome measurement

Many outcomes in sexual health are, by their nature, subjectively experienced and so are necessarily self-reported, for example self-efficacy, intention, regret, relationship and sexual satisfaction, etc. Some behavioural outcomes can be measured objectively. For example, a study examining the effectiveness of daily text reminders for adherence to contraceptive pills used both daily diaries and an electronic monitoring device to measure adherence to pills and found that the latter was a better predictor of (lack) of adherence. 81 In another study, actual condom use was established by checking bins in a brothel. 271 However, methods such as these may be expensive to administer, may pose ethical challenges and may not be feasible practically.

Biological outcomes are the ‘gold standard’ for measuring impacts on health,68 but STI events are relatively rare for most young people and it may be hard to capture acute infections such as chlamydia at single time points, as they may have been acquired but also treated by the time of follow-up. 185 Capturing incident STIs over a given study period through clinic records is likely to be more accurate, but there can be problems in identifying or tracking patients through medical records because of concerns about confidentiality and data security.

Optimal follow-up periods

Most RCTs of IDIs that we located measured the effect of IDIs on outcomes either immediately post intervention or by 3 to 4 months (see Chapter 4). One study followed up participants for 6 months. 68 Measurement at different time points for different outcomes is appropriate; an intervention may have an immediate impact on knowledge, self-efficacy and intention,63 but behaviour change will take longer to emerge (especially if sexual contact is a relatively rare event for a young person) and impacts on biological end-points are likely to take even longer. For example, it could take years to realise the impact of a school educational programme on STIs or unwanted pregnancy. The impacts of interventions could be amplified over time (e.g. as skills are practised) or could diminish: repeated measurements would help to understand the impact of interventions on different outcomes over time,272 and to understand impact on determinants of behaviour change. As noted above, the follow-up periods for educational interventions for school-aged children may need to be long in order to capture changes in behaviour. 94

Measurement reactivity

The detailed measurement of sexual health outcomes in both an intervention arm and a control arm can prompt reflection about sex and sexual health. 237 A study conducted among MSM showed that the mean number of unprotected anal sex partners in both the study arms declined over time during the 12-month follow-up period. 239 It is difficult to know if this observed decline was caused by real decline in both the study arms (which can potentially suggest measurement reactivity), regression to the mean, or if it was caused by other coincidental factors. Collecting only minimal baseline data prior to allocation to study arms could be one way to minimise measurement reactivity.

Evaluation paradox

There is also an ‘evaluation paradox’; if participants know that there is doubt about the effectiveness of an intervention, this can lead to an underestimation of its effects and, hence, a reduced confidence about following advice given during the intervention. 273

Advantages and disadvantages of digitally mediated research methods

• Research administration via technology such as mobile phones and laptops can be convenient for participants and for researchers.

• Online recruitment can be quick and relatively cheap.

• The internet can help to reach hard-to-reach, stigmatised and at-risk populations.

• Audio-visual features can help to reach participants with poor literacy.

• Digitally mediated surveys facilitate research of sensitive issues, such as sexual health, by offering privacy during data collection.

• Digitally mediated (private) data submission encourages honest disclosure, enhancing the internal validity of data collected online.

• Convenience sampling is used in online research: as the sampling frame is not known it is hard to assess sample representativeness (and, therefore, the generalisability of results).

• Risk of multiple enrolment in online studies can be minimised by using multiple personal identifiers to enhance internal validity of data.

• Researchers should take adequate measures to ensure the confidentially and security of data collected and stored digitally.

• Study participants should be provided with adequate information about the risks of participating in online research.

Evaluation of effectiveness

• Automated (digital) randomisation and allocation methods can minimise allocation bias.

• Digital interventions, by their nature, offer standardised content, minimising inconsistencies in delivery of intervention content.

• Ensuring exposure to digital interventions as intended can be a challenge. Various digital methods, such as reminders via e-mail, text message, etc., can be used to improve exposure to digital interventions.

• Reliable measures to assess exposure to digital interventions need to be developed.

• Retention in trials of digital interventions conducted solely online is a challenge. Strategies such as offering incentives and using multiple methods of contact can enhance retention in online trials.

• Evaluation of digital interventions is vital for assessing any potential adverse effects.

Measuring sexual health outcomes

• The choice of outcomes measured should reflect the theoretical design of an intervention, including hypothesised pathways of action.

• Few studies capture factors which are important to young people (e.g. measures of sexual well-being).

• Age-appropriate sexual health outcomes should be used in research with young people.

• Composite outcomes may be more accurate (e.g. contexts for condom use, including duration and type of partnership, history of STI testing, use of other types of contraception, etc.).

• Suitable outcomes to capture the costs and cost-effectiveness of IDIs are needed.

• Long-term follow-up data are needed to assess any intermediate impacts on determinants of behaviour change and the persistence of intervention effects.

• Biological outcomes are the ‘gold standard’ for measuring impact on health, but STI events are relatively rare and may be hard to capture at single time points.

• Study designs should allow for the effect of potential measurement reactivity (the effect of measurement itself).

Implications for research

• Mechanisms for intervention evaluation could be planned at the intervention development stage, for example baseline and follow-up questions and user views on an intervention in practice.

• Online (quantitative) research methods can be convenient for participants and researchers, and relatively cheap.

• Digitally mediated data collection (e.g. by mobile phone or by computer) can enhance participant privacy and data accuracy.

• Randomised controlled trial procedures can be conducted entirely online (recruitment, informed consent, registration, random allocation, data collection and automated follow-up).

• Robust measures are needed to guard against multiple enrolment in online studies and to ensure the confidentiality and security of digitally collected and stored data.

• Multiple strategies can tackle the challenge of retaining participants in online studies over time (e.g. incentives, reminders and different methods of contact).

• Adverse effects of interventions should be actively monitored before and after implementation of an intervention.

• Qualitative evaluations are needed to understand sexual behaviour, to evaluate how complex interventions work and to assess engagement and implementation in practice.

Quality of the body of evidence

We located a vast quantity of literature through our searches: 348 ‘core’ papers and 525 ‘supplementary’ papers, as well as online sources. We appraised all RCTs of IDIs according to Cochrane Collaboration quality criteria: only two of the 19 included RCTs were categorised as being at low risk of bias in terms of method of randomisation, concealment of allocation and/or loss to follow-up. For some studies, papers did not contain sufficient information to judge quality; in these cases we contacted authors, but contact was not always possible, especially for studies conducted 15 or more years ago. The uncertainty about the quality of the studies included in the meta-analyses reduces confidence in the results.

There were few studies (and few randomised participants) contributing to analyses of IDIs in comparison with face-to-face comparators, such as a lecture or counselling, and analyses of IDIs in comparison with a different design of digital intervention. Decision-makers may wish to know (a) whether or not interventions are effective and (b) whether or not interventions can usefully replace or supplement face-to-face interventions. 260 There is evidence that digital interventions for sexual health are effective (especially for knowledge acquisition), but evidence on whether or not IDIs could be as good as face-to-face interventions is lacking. We did not conduct meta-analyses of studies of composite, human-facilitated IDIs (e.g. teacher-facilitated IDIs), but examples of these are given to demonstrate a variety of possible models of design and delivery.

This scoping review is a broad review of evidence, differing from a systematic review by including primary data (the opinions of young people, experts and evidence from sources such as the internet). In the rapidly evolving field of digital interventions, many relevant debates are conducted online rather than published in peer-reviewed journals. We extracted relevant themes from diverse sources of evidence and synthesised ideas narratively instead of carrying out a formal qualitative synthesis of data. This was to capture ideas which could offer fresh insight into issues that were not necessarily captured by formal qualitative studies. It was not necessary to find a particular viewpoint reflected in multiple pieces of evidence for it to lend insight into relevant issues.

Generalisability

This review focuses on the UK, but the findings are applicable to other settings, especially countries with the resources and infrastructure that can support digital technology. Many of the issues raised are of relevance for digital interventions for health issues other than sexual health.

Evidence on best practice for digital intervention design and development

We identified many good examples of IDIs for sexual health promotion, particularly from the USA. Good practice for IDI design and development includes (1) developing an understanding of the target population and their behavioural needs, (2) targeting the modifiable mechanisms of the desired behaviour change through empirical research with users, (3) selecting BCTs that match user needs, (4) implementing techniques in forms that are engaging and promote long-term interest/use among users and (5) ensuring that interventions are feasible and sustainable within an implementation context. Young people should be involved at all stages, including needs assessment, the design stages and prototype testing and to inform delivery and implementation in different settings so that the relevance of both the actual intervention and how it is delivered is maintained. The views of other stakeholders may also address important sexual health needs not identified by young people themselves and help to optimise implementation.

Most identified IDIs focus on reducing sexual risk behaviour and increasing condom use, with few interventions addressing issues that are important to young people, such as sexual pleasure and relationships, or cofactors that may influence risk behaviour, such as alcohol and mental health. There are also some notable gaps for risk groups such as young women after pregnancy, looked-after young people (in institutional care), young people experiencing sexual and domestic violence, LGBT youth and young people with learning difficulties. Promising interventions that have already been developed could be adapted for specific target groups and evaluated in different UK settings.

There is rapid innovation in the development and design of digital interventions in the commercial and service sectors, as well as in the academic sector. The guiding principles for design and the drivers for intervention development vary widely between sectors. More collaboration is needed to capitalise on the design and software skills of the commercial sector, the knowledge of users and stakeholders and the theoretical expertise and evaluation skills of academia. There is a need for mechanisms to assess whether or not existing interventions meet defined quality criteria for intervention content quality and to assess potential risks.

Evidence on effectiveness and mechanism of action of digital interventions for sexual health

We located 19 studies which were RCTs of IDIs for sexual health promotion for young people. We extracted data and (where possible) synthesised the findings from these studies to assess the effectiveness of IDIs. IDIs were delivered in a variety of settings (schools, colleges, health-care settings and online) and targeted heterosexual young people as well as young MSM.

Evidence on methods for economic measurement, analysis and modelling in sexual health

There is very limited health economic evidence which relates directly to digital interventions for sexual health promotion, so we draw on evidence and guidance regarding (non-digital) sexual health promotion and (non-sexual health) digital interventions.

Sexual health promotion interventions are likely to be cost-effective if the target groups have a high prevalence of STIs and/or if the intervention is relatively cheap. Once developed, the ongoing costs of IDIs can be relatively low and targeting large numbers of people can, in theory, be relatively cheap and easy. However, the level of uptake of and engagement with an intervention and the characteristics of target populations might be more instrumental in determining the cost-effectiveness than intervention efficacy alone.

Cost–utility analysis is the type of economic evaluation recommended in the UK (calculating the incremental cost per QALY gained). However, this may not be the most suitable type of economic evaluation for sexual health intervention evaluation if it does not capture all of the costs and consequences of interest. Cost-effectiveness analyses (e.g. reporting results as cost per STI case detected or cost per pregnancy avoided) may provide more useful information to a decision-maker in a sexual health context. Decision modelling can potentially capture a wider range of information than a RCT as well as long-term impacts of an intervention beyond the duration and scope of the latter. As most of the costs and benefits of sexual health promotion come from potentially rare events (such as the prevention of cases of STIs or prevention of unintended pregnancies), it is likely that large, observational data sets will play an increasing role in capturing this information.

Further research and consensus is needed on how best to cost intervention development, implementation and maintenance; how to measure health and well-being outcomes, particularly long-term outcomes, in the sexual health promotion field; and the best ways to conduct economic evaluations of digital media interventions for sexual health promotion.

Evidence on implementation of sexual health interactive digital interventions for young people

The impact of a sexual health promotion IDI will be determined by its reach (proportion of the target population reached), efficacy, adoption (within the target setting), implementation (how well it is delivered) and maintenance (sustainability) – RE-AIM.

Sexual health IDIs delivered in static settings, such as a clinic or the classroom, have a captive audience which enables interventions to be delivered with high fidelity over a defined period of time. In contrast, online interventions can allow private and convenient access, and reach populations who may not be linked into mainstream services, but require the user not only to find the intervention but also to stay with it. Mixed delivery through complementary routes (in static settings and online) is most likely to maximise the proportion of the target population gaining access.

The reach of IDIs could be enhanced by linking sexual health promotion interventions with existing digital initiatives, such as STI testing, or with trusted branded websites or commonly used social networking sites. Face-to-face recruitment and facilitated engagement (e.g. with teachers or clinicians) also encourage young people to access interventions and are more likely to lead to continued engagement. More research is needed on how social networking sites, mobile phones and gaming can be harnessed for sexual health promotion.

The introduction of any new intervention into an existing setting requires frontline staff and management to believe that there are additional benefits compared with the current arrangements. Using the local knowledge and intelligence of local staff is vital for both successful intervention development and successful implementation. An effective intervention usually requires some adaptation for local contexts, but care is needed in identifying and preserving the core components so that effectiveness is maintained. Technical support, moderation/monitoring and updating material are further challenges for implementing sustainable digital interventions.

There are few national policy levers to drive implementation of sexual health promotion IDIs in practice and responsibility for health and education is increasingly devolved to local health care and local authority commissioning groups. However, an increased emphasis on local cross-sectoral working means that there may be more opportunity for shared initiatives and shared (financial) risk.

Evidence on optimum research design and outcome measurement

Digital platforms offer quick, convenient and relatively cheap methods for conducting sexual health research. Recruitment via the internet offers opportunities for reaching hard-to-reach, stigmatised populations, although convenience sampling makes sample representativeness more difficult to assess. Online recruitment to trials allows self-registration, online consent, automated randomisation to trial arms, automated follow-up and online data collection, which can potentially reduce the cost of conducting trials. Using digitally mediated research methods to administer surveys (e.g. computer-assisted self-interviews or mobile phones for data collection) can enhance confidentiality. Requesting several participant identifiers (such as address, telephone number and date of birth) can help to reduce the risk of multiple enrolment. Robust measures are needed to ensure security and confidentiality of data collected using digital methods.

Retention in studies which use digitally mediated research methods can be a challenge (e.g. online trials and longitudinal surveys). However, using multiple retention strategies, such as offering incentives, sending reminders via text and e-mail and appealing to the altruism of participants, can enhance retention. Digital interventions, by their nature, offer consistent content, which can be standardised as intended. There is a need to develop mechanisms to ensure the standardisation of exposure to digital interventions to measure intervention use.

Age-appropriate sexual health outcomes should be used in research with young people. If interventions address multidimensional aspects of sexual well-being and other health issues, such as substance use and mental well-being (as users would like), outcome evaluation should also reflect these broader concepts of health. There is a trade-off between producing a specific intervention with clear (narrow) aims and producing an intervention which addresses the complexity of sexual health. The theoretical design of the intervention and the hypothesised mechanism of action will dictate which outcome measures are the most suitable.

It is difficult to capture impacts on health (e.g. STIs or pregnancy) as these events are relatively rare, especially in younger age groups. It is important to measure determinants of behaviour change (such as knowledge and self-efficacy) to capture shorter-term impacts and to understand how interventions work. Adequately powered, longer-term studies are needed to assess the impact of digital interventions among young people. Qualitative process evaluations are needed to evaluate how complex interventions work and to assess engagement and implementation in practice.

In conclusion, digitally mediated research methods are acceptable and feasible for recruitment and administration of sexual health research, and there is increasing evidence on how best to ensure good-quality online data and how to maximise retention in studies.

‘Front-running’ interactive digital interventions suitable for young people in the UK

We did not find ‘front-running’ IDIs which are suitable for implementation in the UK (judging on the basis of intervention effectiveness, cost-effectiveness, suitability for young people in the UK and the feasibility of implementation in a UK context).

We identified five IDIs for young people with RCT evidence of effectiveness which were all developed and evaluated in non-UK settings: four in the USA110,142,183,274 and one in the Netherlands. 112 All of these IDIs focused on HIV prevention, mostly recruiting in educational settings. However, all of these studies were very small scale, randomising around 100 people or fewer into each arm of RCTs, and so there is less confidence about the strength of evidence. There are no available data on cost-effectiveness for these interventions. Many of the interventions identified in our search were underpinned by Behaviour-Change Theory and involved users in development, but it is difficult to judge the quality of these interventions as it is not known exactly how to deploy theory effectively, or which models and mechanisms for user involvement in intervention design are the most effective (see Chapter 3). The choice of a suitable intervention will depend on the needs of the target audience and on the choice of setting for implementation; as we have noted, ‘young people’ are not a homogenous group, and it is unlikely that any one intervention could meet the specific needs and preferences of different target populations in different settings. We identified a large range of IDIs which utilise imaginative multimedia and interactive features with different populations of young people in different settings (see Chapters 3 and 4), which could be adapted for particular target groups of young people in the UK (see Sexual Health Research Network database: www.sexualhealthnetwork.org.uk).

Intervention design and development

• Collaboration with target populations is essential in order to design accessible, attractive and engaging IDIs which meet young people’s needs and wants.

• Implementation and engagement with interventions should be considered from the outset, with interventions developed in collaboration with all stakeholders.

• Sexual health promotion IDIs could be adapted from existing IDIs or developed afresh to make them relevant for young people in the UK.

• Collaborations which link young people’s views, public health expertise, the dynamism and creativity of industry and rigour offered by academia could help to optimise intervention design.

• Protocols and examples of ‘best practice’ in IDI design and development and models for interdisciplinary collaboration could usefully be shared.

Interactive digital intervention aims and target populations

• IDIs are effective for knowledge acquisition and could address young people’s need for sexual health information in clinics, in schools and online.

• Young people express a preference for integrated models of sexual health which focus on sexual well-being as well as disease and physical health.

• Sexual health promotion IDIs could form part of health promotion interventions that address other linked health issues, such as alcohol, mental health and smoking.

Interactive digital intervention evaluation

• Mechanisms for intervention evaluation could be planned at the intervention development stage, for example baseline and follow-up questions or user views on an intervention in practice.

• Online (quantitative) research methods can be convenient for participants and researchers, and relatively cheap.

• Digitally mediated data collection (e.g. by mobile phone or by computer) can enhance participant privacy and data completeness.

• Randomised controlled trial procedures can be conducted entirely online (recruitment, informed consent, registration, random allocation, data collection and automated follow-up).

• Robust measures are needed to guard against multiple enrolment in online studies and to ensure the confidentiality and security of digitally collected and stored data.

• Multiple strategies can tackle the challenge of retaining participants in online studies over time (e.g. incentives, reminders and different methods of contact).

• Adverse effects of interventions should be actively monitored before and after implementation of an intervention.

• Qualitative evaluations are needed to evaluate how complex interventions work and to assess engagement and implementation in practice.

• Tiered ethical committee approval processes could facilitate more rapid evaluation of health promotion interventions with low risk of harm to health.

Interactive digital intervention implementation

• Interactive digital interventions could be offered in conjunction with currently existing digital systems, for example online service locators, online services (e.g. for STI testing), clinic results systems or dating websites.

• Education in internet and media literacy could help to reduce the risks of exposure to harmful online content.

• NHS endorsement of sexual health IDIs would be valuable in increasing awareness of interventions and public trust.

• Closer cross-sectoral working between health and social care and voluntary and commercial sectors could help to share any financial risk.

• A quality assurance system could include criteria for assessing the accuracy of content. A tiered system could apply for interventions with different possible risk (e.g. health promotion websites vs. medical devices for diagnosis or treatment).

• A quality assurance system could help users to judge possible risks associated with online data collection (e.g. privacy, confidentiality and informed consent for the use of data).

• Budgets need to include funds for marketing, maintaining and updating IDIs.

Can interactive digital interventions have a long-term impact on sexual behaviours, sexual well-being and biological outcomes?

• Effect of IDIs on sexual behaviour, biological outcomes (pregnancy, HIV and other STIs) and sexual well-being (e.g. pleasure and relationship and sexual satisfaction).

• Effectiveness of digital interventions for particular problems, including intimate partner violence, repeat abortion and sexual problems.

• Effects of IDIs on inequalities in health.

• Longer-term outcomes (6 months or longer).

What component features of an interactive digital intervention are most likely to be associated with beneficial outcomes in populations in need?

• The best designs for IDIs, for example theoretical underpinning including BCTs and core components.

• Effectiveness of interventions for particular groups, for example school-aged children, young MSM, disadvantaged youth and BME groups.

• Effectiveness of interventions which target multiple health problems, including drug and alcohol use and mental health, as well as sexual health.

• Optimum intervention design to maximise young people’s interest and engagement.

• Optimal settings and modes of delivery of IDIs for different populations.

• Understanding the potential of features such as social media and gamification for health promotion.

• Optimum exposure, dosage and timing for interventions.

How can interactive digital interventions for sexual health best be delivered to populations in need?

• The effectiveness of alternative models of intervention delivery (e.g. direct vs. facilitated access, different technological platforms).

• The best methods and costs of reaching, engaging and retaining young people with interventions (particularly reaching populations in greatest need).

• How to target and reach neglected at-risk populations (e.g. young people post pregnancy, looked-after children, young people with learning difficulties).

• Barriers and facilitators to implementing IDIs for sexual health promotion online, in clinics and in school settings.

What are the optimal methods of economic measurement for sexual health?

• Appropriate sexual health-related costs and consequences to use in economic models and analysis, and in particular what outcomes should be used.

• How best to proportion the cost of a digital intervention (development, implementation and maintenance) over the longer term in order to calculate the mean cost per person.

• How best to quantify health outcomes that occur a long time after the digital intervention.

What are the costs and cost-effectiveness of interactive digital interventions for sexual health?

• The costs of different models of implementation.

• The costs of disseminating and maintaining digital interventions for sexual health.

• Which digital media interventions might represent the best value for money in sexual health promotion?

• The role of digital media interventions in reducing health inequalities.

• Best ways of measuring sexual well-being outcomes (including social and emotional well-being).

• Optimal outcome measurement instruments to capture the complexity of sexual behaviour.

• Best ways of measuring sexual behaviour (e.g. condom-use outcomes).

• Optimal outcome measures for young people which are appropriate for their stage of sexual development.

What are the best ways of enhancing data validity in online research?

• Assessment of potential recruitment bias with online recruitment.

• Best methods of minimising potential fraud or dishonesty with online recruitment.

• Best methods of assessing and enhancing data validity.

• Potential size of measurement reactivity (change prompted by measurement alone).

• Best methods of enhancing retention in trials of digital interventions for sexual health.

• Best methods of assessing meaningful exposure to IDIs.