The clinical effectiveness of individual behaviour change interventions to reduce risky sexual behaviour after a negative human immunodeficiency virus test in men who have sex with men: systematic and realist reviews and intervention development

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/77/03. The contractual start date was in August 2014. The draft report began editorial review in August 2015 and was accepted for publication in May 2016. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

none

Copyright statement

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Flowers et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

The health problem

Men who have sex with men (MSM) experience significant inequalities in health and well-being, and a disproportionate burden of ill health in relation to sexual health, mental health and substance use. 1 MSM is the group at highest risk of acquiring a human immunodeficiency virus (HIV) infection in the UK. In 2013, it represented 54% of all new diagnoses in the UK. Data from Scotland concerning 2005–9 showed a relatively stable incidence rate among MSM of around 15.3/1000 person-years. The most recent available UK data1 show that, across the UK, 3250 new HIV diagnoses in MSM were reported in 2013. In terms of diagnosis rates, given that there are an estimated 43,500 MSM living with HIV in the UK, this is equivalent to 59 out of 1000 MSM aged 15–59 years. Moreover, an estimated one in five, or 7200, HIV-positive MSM in the UK remain undiagnosed2 and approximately 1000 are diagnosed late each year (i.e. within 3 months of diagnosis they report a CD4 count of < 350 cells/mm²). 1 Delayed diagnosis is associated with poorer health outcomes3,4 and treatment response, increased mortality and health-care costs and increased levels of onward transmission. Given that men living with HIV who are taking effective antiretroviral therapy are highly unlikely to transmit HIV,5 it is clear that undiagnosed infection, particularly primary infection (when individuals are extremely infectious), is responsible for most new infections. 6 Mathematical modelling suggests that increased testing, linkage to care and early treatment could reduce HIV infection incidence in MSM. 7 Furthermore, currently we know that most undiagnosed infections in the UK have been acquired recently. 1 Indeed, the proportion of new diagnoses associated with recent transmission increased in some parts of the UK between 2011 and 2013, from 23% to 30%. Therefore, there is evidence that ongoing new infections are a particular challenge for HIV infection prevention. HIV infection testing rates among MSM overall have increased8 and 83% of sexually transmitted infection (STI) clinics report HIV infection testing coverage of at least 80% among MSM patients. 1 Increasing the frequency of testing to every 3 months for men at increased risk of HIV infection is recommended in the UK and other national guidelines9,10 and is highly likely to be cost-effective at the incidence observed in UK MSM genitourinary medicine (GUM) clinic attendees. 11 Together these factors highlight the considerable opportunities for prevention of onwards transmission among MSM following HIV infection testing.

In the last few years there have been major changes in HIV infection prevention in the UK. These changes are complex. They represent the interplay of a range of issues including factors relating to the diversification of HIV infection prevention initiatives and concomitant biomedical and wider cultural changes. In terms of cultural changes to HIV infection prevention, innovations within social media (particularly geospatial social media network applications or ‘app’-based platforms) have led to changes in patterns of sexual mixing between men. Social media platforms, like the internet platforms that preceded them, facilitate seroadaptive approaches to HIV infection prevention, reducing the risk of onward transmission (wherein risk of onward transmission of HIV is minimised by HIV status disclosure and subsequent condomless sex occurring between men of the same HIV status). However, sexual mixing facilitated by online communication is also responsible for increases in a range of other negative health outcomes among MSM. For presumed HIV-negative men, the effectiveness of seroadaptive behaviours is dependent on accurate assessment of their own and their partners’ HIV status, limited by the high rate of seroconversion and high infectivity of undiagnosed newly positive men. For men living with HIV there is a rising incidence of a number of STIs, with a significant rise in the ratio of bacterial STIs observed in HIV-positive compared with HIV-negative men in recent years. 12 Certain STIs, including hepatitis C virus and lymphogranuloma venereum, are seen predominantly, but not exclusively, in MSM living with HIV,13,14 and recent outbreaks of syphilis and enteric infections, including shigellosis,15,16 across the UK are associated both with high proportions of HIV-positive MSM, sexualised drug use and the use of geospatial social networking apps. 17 These observations suggest a complex picture of dense sexual networks of HIV-positive MSM12 and populations of men at high risk, who are currently HIV uninfected or undiagnosed. 18 Social media applications are also implicated within the rise of what has been described as ‘chemsex’, in which recreational drugs are taken in an explicitly sexual context19 and the easy, or ‘instant’, organisation of sex parties. There is emerging evidence that sex parties organised online facilitate the transmission of a number of STIs, including HIV. In the light of these recent and significant changes to the sociocultural contexts of sex between men, particularly the increases in condomless sex,20 HIV infection prevention and concomitant risk-reducing interventions must address the shifting and inter-related psychosocial and sociocultural contexts of HIV transmission.

Biomedical changes to HIV infection prevention, some of which are underpinning recent cultural changes, have recently been used effectively to reduce the chance of HIV transmission. These biomedical approaches can be thought of in two distinct ways: treatment as prevention (TasP), in which antiretroviral drugs, when taken effectively by individuals living with HIV, reduce transmission by reduction in both individual and population viral loads;21 and pre-exposure prophylaxis (PrEP), in which HIV-negative men take antiretroviral drugs to reduce the chance of exposure to HIV resulting in established infection. In these ways, antiretroviral drugs can be used alongside condomless sex to reduce the risks of HIV transmission. Recent UK22 and French23 trials have demonstrated that PrEP can reduce HIV transmission among MSM by up to 86% and also serve to illustrate the very high incidence of HIV infection in MSM eligible for the trial [9.1% per annum in the Pre-exposure Option for reducing HIV in the UK: immediate or Deferred (PROUD) study], who represent a sizeable minority of MSM attending GUM clinics. Moreover, recent population surveys among MSM suggest growing PrEP awareness and widespread acceptability among those men who would benefit most from using them. 21 Future behaviour change interventions must incorporate and reflect these biomedical developments within HIV infection prevention. It seems likely that the most effective HIV infection prevention interventions will combine mechanisms of action that encompass a combination of biomedical, psychological and social approaches.

Alongside these pharmaceutical interventions, and of particular relevance to the current report, there have also been significant changes in access to, and the monitoring of, the fundamental technologies associated with HIV infection testing (e.g. self-sampling and self-testing). The range of contexts in which testing can take place has changed profoundly since the widespread availability of antiretroviral drug therapy. HIV infection testing continues to take place within traditional clinical settings (e.g. GUM clinics), yet following guidance from the National Institute for Health and Care Excellence (NICE) in 2011,24 provision is also recommended in a range of community settings. With changes in legislation in April 2015, instant-result home HIV infection testing is now possible, with tests delivered via the internet or purchased over the counter and results available within a few minutes to the person testing. 25 Equally, HIV self-sampling has become relatively commonplace and has high acceptability; samples are self-collected and returned for laboratory-based testing, with the results given to individuals at a later date. 26,27 Test results are often delivered electronically using mobile phone technology for non-reactive results and recalling those people with reactive test results for further testing.

Finally, in terms of key changes in the infection prevention landscape, the commissioning and implementation of HIV infection prevention services has also changed in key ways. In terms of the very ethos of HIV infection prevention, there has been a widespread recognition of the wider psychosocial and sociocultural determinants of HIV infection risk and sexual ill health among MSM in the UK. 28–30 This acknowledgement of the multiple and distal determinants of HIV risk, and its syndemic qualities, again merits a pluralistic approach to HIV infection prevention. Equally, the way HIV infection prevention is commissioned across the UK has changed, with a diversification of approach between the nation states. In Scotland, for example, HIV infection prevention is commissioned within the NHS at the local health board level, but in England HIV infection prevention is commissioned both by local authorities and at a national level by Public Health England (see Significance to the NHS).

In summary, HIV transmission among MSM in the UK remains an ongoing problem, with significant implications for health and social care. Our understanding of HIV transmission risk itself has changed fundamentally, from simple concepts such as, in the past, condom use per se to more contemporary understandings that necessitate simultaneous consideration of condoms, HIV status, use of antiretroviral drugs, viral loads, sexual position, disclosure, the social context of sex and patterns of sexual mixing (e.g. social media and sex parties). Equally, there has been a growing recognition of the complexity of factors that are associated with HIV transmission and the concomitant need for HIV infection prevention interventions to address this complexity (so focusing on the distal as well as proximal determinants of HIV transmission). It is vital that interventions are developed which reduce HIV transmission and take account of the wider health inequalities associated with those vulnerable to HIV infection. Innovations in biomedical approaches to HIV infection prevention (such as TasP or PrEP) demand further interprofessional and interdisciplinary understanding in the context of the psychosocial and sociocultural factors that determine their effective use. There are clear pharmacological mechanisms of action implicated with the use of antiretroviral drugs for HIV infection prevention (in PrEP and TasP it is the action of drugs on viral replication). However, it is psychological, social and cultural factors that shape decisions to initiate, maintain and terminate the adoption of these particular prevention approaches. Their effective use demands a concomitant investigation of the psychosocial determinants of their use, and the provision of psychosocial interventions, to maximise the opportunities such approaches will bring.

Significance to the NHS

There is limited information concerning lifetime costs of HIV infection and wider considerations of the economic and social burden of HIV infection. However, recent estimates of health-care costs associated with HIV infection suggest an estimated mean lifetime cost of treating one person as £360,800. 31 Therefore, if around 3000 men are newly infected each year, new direct lifetime costs relating to their treatment alone will amount to around £1.1B. If generic drugs replace patented drugs, estimated mean lifetime costs could fall to £179,000. Critically, these costs do not include social costs associated with living with a HIV infection, or the associated cost savings associated with reduced onwards transmission via access to effective treatment (e.g. through TasP). Costs are associated with wider NHS burden.

Current guidelines for human immunodeficiency virus infection prevention among men who have sex with men

There is no single contemporary UK-wide guidance relating to HIV infection prevention among MSM, although the European Centre for Disease Prevention and Control has recently published comprehensive STI and HIV infection prevention guidance. 32 At the level of individual nation states, and some local areas, a range of key documents exist which suggest both broad and specific approaches to HIV infection prevention.

In Scotland, the Good Practice Guidance on HIV Prevention in Men who have Sex with Men (MSM): Health Protection Network Scottish Guidance33 details the importance of combination prevention as an overall perspective within which to understand the potential combination of HIV infection prevention approaches. Combination prevention details how biomedical interventions, such as PrEP, must be thought about alongside wider structural interventions that aim to address the multiple and concurrent determinants of HIV infection risk-related behaviour. In turn, these wider interventions addressing the distal determinants of HIV infection risk (e.g. reducing homophobia or hegemonic masculinity) should be considered alongside the role of behavioural interventions (such as those being examined in this review). Behavioural interventions focus on how the determinants of sexual risk are experienced by the individual within their own intrasubjective world. As such, they focus on addressing the proximal determinants of HIV infection risk. The Scottish guidance details the importance of the full range of prevention approaches. It details the provision of condoms, post-exposure prophylaxis and HIV infection treatment for people living with HIV as a prevention. It explicitly suggests that focused brief or intensive behaviour change interventions are provided, and based on the acquisition of interpersonal skills and increasing motivation to adopt safer sexual behaviours. It also recommends that these interventions are theoretically informed. It goes on to suggest that these behavioural interventions are provided by staff who have gained competency in their provision through training. Finally, the guidance details the importance of interventions that address the wider social and cultural determinants of HIV-related risk.

At a fundamental level the combination prevention approach recognises that no single approach to HIV infection prevention will be sufficient to end HIV transmission and harnessing the synergistic effects of multiple prevention strategies as impact on each other promises to have the greatest effect on reducing HIV transmission. The effectiveness of approaches which address the distal determinants of HIV risk at the societal level, for example, is likely to be boosted by individual-level behaviour change interventions that focus on specific subpopulations. In turn, individual-level behaviour change interventions, targeted at those most at risk, are likely to be boosted by effective biomedical approaches to HIV infection prevention. Critically, the logic of the combination prevention approach also exposes the dangers of any singular approach to HIV infection prevention. If biomedical approaches, such as PrEP, for example, do not engage with psychological factors (such as risk perception) they may amplify health inequalities. Equally, if individual-level interventions do not engage with the wider social context (such as health literacy or cultures of masculinity) they too may exacerbate inequities within sexual health.

In England, HIV infection prevention guidance has focused almost solely on biomedical interventions to date and has not addressed behavioural- or structural-level interventions, or the interplay between them. The NICE HIV infection testing guidance24 was intended to increase the effectiveness of testing as the major approach to HIV infection prevention among MSM in England. NICE’s recommendations included the promotion of HIV infection testing in a range of settings, promoting testing in specialist sexual health services, testing in primary and secondary care, outreach and point-of-care testing and promoting repeat testing. Again, as mentioned in The health problem, developments in new technologies (e.g. self-testing, PrEP and TasP) have rendered this guidance as outdated.

More recently, at regional level, in 2013 London Directors of Public Health and London Councils collaborated on HIV Prevention Needs Assessment for London. 34 The authors note major limitations with regard to existing evidence with which to shape the needs assessment, including the lack of information concerning cost-effectiveness of HIV infection prevention interventions, a lack of focus on theories of behaviour change and a lack of focus on contemporary issues regarding HIV risk. The accompanying evidence synthesis is briefly described in Table 1; however, the consortium makes a number of broad recommendations for commissioning of HIV infection prevention services, again intended to be tailored to the specific local context. The report recommends that:

[W]hen commissioning HIV prevention interventions, commissioners should consider a ‘combination’ approach, capitalising on the multiple available prevention interventions now available, that are evidence-based and focused on knowledge, skills and behaviours as well as access to high quality services. These interventions should be targeted to the right populations, delivered at sufficient scale to maximise their impact, and should address both primary and secondary prevention.

HIV Prevention Needs Assessment for London, p. 4434

In addition, the report outlined a number of key areas that should provide a focus for prevention, including HIV infection testing, condom promotion and provision, harm reduction approaches to drug treatment, public and patient engagement, and the use of the digital media and associated technologies.

HIV Prevention England has been the national HIV infection prevention programme in England. Funded by Public Health England, it delivers a nationally co-ordinated programme of HIV infection prevention work with UK-based black African people and MSM. It aims to increase HIV infection testing and support behaviour change (such as increased and correct condom use) to prevent HIV transmission. It is designed to complement locally commissioned infection prevention in areas of higher prevalence, including behavioural change interventions (www.hivpreventionengland.org.uk).

In 2007, NICE produced public health guidelines entitled Prevention of STIs and under 18 conceptions (PH 3) (www.nice.org.uk/guidance/ph3). 35 MSM is one of the key target populations of this guidance, which recommends risk assessment and delivery of one-to-one structured discussions with those identified as being at high risk. Aimed primarily at health professionals trained in sexual health, the guidelines recommend that the discussions be structured on the basis of behaviour change theories; addressing factors that can help reduce risk-taking; and improving self-efficacy and motivation. The session should last at least 15–20 minutes, with the number of sessions depending on individual need. The range of suggested behaviour change techniques (BCTs) are based on Conner and Norman’s Predicting Health Behaviour. 36

Currently, the NHS England’s HIV Clinical Reference Group is formulating policy in relation to HIV infection prevention utilising the biomedical infection prevention interventions of both TasP and PrEP. It is unclear at present whether or not this policy will include the integration of behavioural and other psychosocial approaches.

In summary, there is no contemporary overarching UK guidance that describes and recommends effective HIV infection prevention interventions for MSM in the UK. In Scotland, there is some guidance specifically concerning HIV infection prevention among MSM; however, this is outdated. In regional areas, such as London, there are complementary approaches providing some guidance as to the provision of HIV infection prevention. English HIV infection prevention guidance has focused particularly around HIV infection testing,24 with some guidance on behaviour change, and awaited policy on biomedical interventions such as antiretroviral drug therapy. Again, given the pace of change within the HIV infection prevention field, this guidance is now outdated and is currently being refreshed.

Current service provision

Historically, and currently, HIV infection prevention has been delivered by a broad range of stakeholders. HIV infection prevention has its origins in the ‘grass roots’ gay community organisation-led initiatives, which gradually over time have become professionalised (see Flowers37 for an overview of these changes). However, the delivery of HIV infection prevention services remains distinctly heterogeneous, with third-sector organisations actively involved [e.g. Gay Men Fighting Acquired immune deficiency syndrome (AIDS)], and commissioned by NHS/local authorities, as well as being provided by professional practitioners (e.g. sexual health advisors) through NHS services.

The relationship between infection prevention guidance and the commissioning and provision of prevention services remains complex in the UK. As detailed in Current guidelines for human immunodeficiency virus infection prevention among men who have sex with men, there is a lack of overall, contemporary guidance and little evidence of a co-ordinated approach to HIV infection prevention among MSM. This represents a fundamental challenge for standardising HIV infection prevention, assessing its effectiveness and ensuring the implementation of the best evidence-based initiatives, their tailoring to local contexts and epidemics, the routine collection of robust data and the collation of, and access to, cumulative knowledge.

It is only within the NHS that attempts to standardise HIV infection prevention and guarantee minimum standards of prevention have been made. Critically, these do not specifically address the MSM population and do not reflect many of the contemporary issues outlined in The health problem. Of particular note are NHS Healthcare Improvement Scotland’s Standards for Human Immunodeficiency Virus (HIV) Services38 and the UK National Guideline on Safer Sex Advice. 39 Of note, the latter, while providing evidence-based recommendations, has a remit that covers a range of STIs (i.e. it is not solely concerned with HIV infection prevention). Moreover, it also addresses sexual health interventions delivered within GUM settings alone.

The health-care improvement standards are applicable to all NHS boards with responsibility for delivering HIV services in Scotland. This includes prevention directly provided by, or secured on behalf of, NHS boards. They apply to primary, secondary and tertiary care settings. The standards provide detail of general recommendations for HIV infection prevention; prevention standards for people living with HIV; and, finally, service provision which addresses behavioural interventions. Audits or assessments of compliance with the standards are not available. The standards state that ‘NHS boards provide access to HIV risk reduction behaviour change interventions’. 38

The UK National Guideline on Safer Sex Advice39 focuses on safer sex advice and behavioural interventions that are delivered within UK GUM clinics. Similar to the health-care improvement standards it does not assess, or provide, MSM population-specific guidance relating to the provision of behaviour change interventions. The guidance focuses on six key recommendations, which provide a further level of detail concerning the scope and delivery of behavioural interventions to reduce onwards transmission of HIV. Recommendations include the need for intensive multisession, evidence-based behaviour change interventions targeting individuals and focusing on skills acquisition, enhancing communication skills and increasing motivation to adopt safer sexual behaviours should be available directly or by referral in all GUM clinics. They suggest that motivational interview techniques should be used as part of an intensive course of risk reduction counselling in MSM at high risk of HIV infection. Brief (15–20 minutes) evidence-based behaviour change interventions targeting individuals and focusing on skills acquisition, enhancing communication skills and increasing motivation to adopt safer sexual behaviours [using techniques such as motivational interviewing (MI)] should be provided as part of routine care of those at elevated risk of STIs and HIV infection in GUM clinics. The delivery of safer sex advice, including condom demonstration, based on the characteristics of effective brief behaviour change interventions, should be part of the routine care of all those at continued risk of infection/transmission in GUM clinics. The provision of accurate, detailed and tailored information on safer sex should form part of all sexual health consultations. MI should be provided by clinic staff who have gained competency in its provision through training.

Desai et al. 40 report and audit the provision of behavioural interventions following the implementation of the UK guidance relating to safer sex. 39 The authors reported that, in an assessment of 24 sentinel GUM clinics, there was a low level of offer and uptake of behaviour change interventions to higher-risk MSM. They note that reasons for this poor compliance may relate to patient factors (e.g. only around 42% of MSM accepted the offer of any behaviour change intervention). However, the lack of offer, and uptake, may also relate to clinic staff’s perception of a lack of training or resource shortage. Many of these factors are also explored in Chapter 7. Interestingly, Desai et al. 40 also examined differences between men deemed at risk and those deemed not to be at risk. They explored whether or not these risk assessments related to the offer of behavioural interventions. They found that significantly more MSM reporting unprotected anal intercourse (UAI) were offered behavioural interventions, especially MI.

Overview of current possible approaches to human immunodeficiency virus infection risk reduction

Current service provision shows there is no definitive guidance relating to the approaches that may, or may not, be useful in relation to behaviour change interventions among MSM. In this section we provide a brief overview of potential interventions, which are sometimes offered within prevention and testing services.

Cognitive–behavioural therapy (CBT) is an intervention originally developed to treat depression. The goal is to change unhelpful thinking (i.e. cognitions) and related behaviours. It involves several phases including assessment, reconceptualising thinking and behaviour, and developing and practising skills to change behaviour. Behaviour change is achieved through a number of exercises both during the therapy sessions and between sessions (‘homework’). The number of sessions can vary, but CBT can be delivered in as few as six sessions.

Motivational interviewing was developed by Miller and Rollnick,41 originally to counsel clients with alcoholism,41 and has been recommended in national guidelines for safe sex (e.g. in the UK)39 and by the US Centers for Disease Control and Prevention (CDC). It is defined as ‘directive, client-centred counselling . . . for eliciting behavior change by helping clients to explore and resolve ambivalence’. 41 The goal is to increase client motivation so that change is achieved by the individual rather than imposed externally. It incorporates counselling strategies such as assessing and reflecting readiness to change, using decisional balance exercises to increase motivation for change, and building self-efficacy for change by identifying perceived barriers. Judgemental statements about any particular behaviours are avoided in order to reduce the risk of increasing resistance or defensiveness. An individual’s freedom to choose, as well as their responsibility for their own choices, is emphasised.

Motivational enhancement therapy (MET) is an adapted version of MI. It is a systematic approach designed to produce rapid, internally motivated change using MI techniques.

Commonly, MI and MET interventions are brief and provided over one or two sessions. They can be delivered as a freestanding intervention and can be delivered prior to, or integrated with, other treatments (e.g. with the use of PrEP).

In summary, although standards of HIV infection prevention are available, they are not available UK-wide and are not aimed at specific populations (e.g. MSM). There is limited and outdated evidence concerning the implementation of such guidance within English GUM settings and this shows poor compliance with the recommended evidence-informed standards.

Existing evidence syntheses

Several systematic reviews have attempted to summarise existing literature and provide guidance on HIV infection prevention among MSM (Table 1). These include overview of systematic reviews,55 narrative reviews of primary studies34 and meta-analyses. 56

As outlined at the start of Chapter 1, there are particular complexities with regard to HIV infection prevention, and the pace and scope of its changes. At the end of the 20th century, effective treatments for HIV infection became widely available, profoundly changing the nature and meaning of being infected with HIV from a ‘death sentence’ to a chronic, manageable condition. 60 As a result of this loss of the ‘perceived severity’ of HIV infection,61 evidence concerning the effectiveness of behaviour change interventions is not transferable from the pre-treatment era. Several existing systematic reviews straddle this watershed moment in which effective treatment for HIV infections became widely available in the developed world. 46,47,50,53,57 These studies mix knowledge concerning behaviour change interventions, designed to reduce HIV infection risks with high perceived severity, with those that are designed to reduce HIV infection risks with lower perceived severity. Equally, systematic reviews of behavioural interventions exist for a range of populations, but do not solely focus on MSM42,45,48,49,51,54 and this limits the direct translation of findings to this specific population.

Table 1 also shows the ongoing methodological problems associated with the existing evidence. These relate to issues such as conflicting and limited evidence:43 on the one hand, the heterogeneity of populations included in studies and, on the other hand, where population criteria is more tightly specified, the small number of studies available for review. 48,62 Many of the studies available suffer from high or unclear risk of bias. 62 Primary studies often suffer from high attrition,48,53 a lack of statistical power,48 limited follow-up periods,48 heterogeneity of intervention content,54 heterogeneity of cultural context regarding implementation,49 differential attrition between intervention and control, a lack of information regarding cost-effectiveness,34 a lack of focus relating to theories of behaviour change,34 a lack of focus regarding the complexity and changing nature of HIV risk-related behaviours,34,55 a lack of inclusion of qualitative and mixed-methods studies,55 and the heterogeneity of outcome measures. 58

However, overall, these evidence syntheses do suggest cautious support for the efficacy of behavioural interventions among MSM. 34,43,49,55,56,58,59,62 Moreover, they suggest that evidence-based interventions have larger effects than non-evidence-based interventions. 48

Rationale for current evidence synthesis

The existing evidence is insufficient to direct future policy and practice, or to determine which specific candidate interventions to reduce HIV transmission should be tested in future trials. The findings of existing reviews tend to be difficult to translate to a contemporary UK context for a number of reasons. For example, their broad operationalisation of behavioural interventions typified by a tripartite distinction of modality of delivery as ‘individual level’, ‘group level’ and ‘community level’ is not helpful. These distinctions are often used in reviews to compare approaches to determine the best use of resources rather than to provide an evidence base that might be useful in delivering the synergistic effects of combinations of interventions. Given both the heterogeneity of the existing reviews and the key changes to HIV infection prevention outlined earlier in the report (see Table 1), further broad-level evidence synthesis, or reviews of reviews, are unlikely to demonstrate any additional contribution to shaping HIV contemporary infection prevention policy and practice. They have proved excellent in directing HIV infection prevention in broad ways, but they have not delivered in terms of the specificity of the cumulative knowledge they represent, for example detailing the particular content of recommended interventions.

There is a clear need to extract useful knowledge from the existing evidence through focusing more specifically in the available literature on a number of key dimensions.

Population factors

The heterogeneity of the MSM population (in terms of sociodemographic, psychosocial and behavioural factors) demands a review that is as inclusive as possible with regard to the MSM population. This will enable the most inclusive use of the literature and maximise the applicability of the reviews findings to UK populations.

While operationalising an inclusive approach to MSM, however, demographic factors should not limit the potential of studies to be included in the review. However, given the focus on behavioural interventions, a number of psychosocial and sociocultural factors are important in shaping inclusion and exclusion criteria for systematic reviews. Some populations of MSM have clear psychological and social needs that are distinct from those of other MSM and make the transfer of knowledge between samples problematic. For example, some populations of MSM have unique material needs and/or life experiences that merit distinct consideration, such as incarcerated populations, those in residential care and those who are homeless. Equally, for some populations of MSM, the social context of sexual conduct (e.g. the influence of power and gender) may be so distinct from that of other populations of MSM that there may be challenges in terms of the transferability of knowledge from intervention trials in these populations to wider populations of MSM, for example those engaged in commercial sex work, those who are victims of sexual or domestic abuse or violence or those MSM who are transgender. We fully acknowledge that several of these factors are likely to be episodic or infrequent in the MSM population and that interventions aimed at the general population of MSM will also reach them. However, the reverse of this logic is more problematic, transferring evidence from studies that only recruit commercial sex workers or victims of sexual assault to the general MSM population may fundamentally distort the applicability of findings.

Other cultural changes including, but not restricted to, the constantly evolving change in interaction through social media, changes in supply and use of established recreational drugs, the constant stream of newly available new psychoactive substances, the rise in ‘chemsex’19 and the increasing prevalence of intravenous drug use among the general population of MSM, necessitates a systematic review capable of shaping current HIV infection prevention policy and practice that reflects these population trends. Such a review would have to refine inclusion criteria that maximised the use of available literature yet which did not assume the transferability of knowledge between evaluations among distinct populations, such as those receiving intensive interventions for their drug use. For these last populations, drug use may well be the primary of focus of behaviour change rather than HIV risk-related behaviour per se. In this way, a systematic review is needed that includes studies sampling MSM who engage in drug use episodically, but excludes those who were recruited into trials solely because of their problematic drug use.

An additional critical factor that shaped the scope of the original commissioning brief is the issue of HIV infection status. It is widely acknowledged that the HIV infection prevention needs of HIV-positive men are distinct from those of HIV-negative or untested MSM and merit a particular focus for HIV infection prevention, and subsequent systematic reviews and meta-analyses. 45,63,64 HIV-positive and HIV-negative men differ on a number of psychosocial, behavioural and sociodemographic variables. 65–67 Engagement in HIV risk-related behaviour differs according to HIV status on a fundamental level: for the HIV-positive person risks are focused on potentially infecting another person (i.e. relatively little impact to self), and for the HIV-negative person risks are focused on the potential for them to become infected (i.e. relatively large impact on self). Interestingly, to date, no single evidence synthesis has focused only on the HIV-negative population of MSM. Once more there are a number of complex issues, both practical and conceptual, that relate to how such a focus on the HIV-negative population could be implemented. These include (1) the meaning and significance of HIV-negative status, (2) the accuracy of knowledge regarding HIV status and (3) the role of HIV infection testing in shaping perceptions of HIV infection status and HIV-negative identity. Given their complexity, each of these issues is introduced in the following sections.

The meaning and significance of HIV-negative status have arguably changed in recent years. Changes in the psychosocial and sociocultural context of HIV infection prevention suggest that seroadaptive behaviours are likely to be increasingly important in HIV infection prevention. Moreover, they will be increasingly important in light of the prominence of biomedical approaches to infection prevention. Historically, HIV-positive status has been clearly associated with a HIV-positive identity. HIV positivity has a long history and significance as a ‘master identity,’ often to the detriment of those living with HIV. 68 In contrast, HIV negativity has not had the same historical significance and meaning. However, the social significance of HIV-negative status is arguably increasing. The issue of eligibility for PrEP, for example, or the increasing importance of seroadaptive behaviours, increases in the prominence of HIV infection testing per se and recency of HIV infection testing (behaviours of relevance only to those who are HIV negative), and increases in the use of the social media, which often ask for people to be explicit about their HIV status, have potentially led HIV-negative status to become an identity in new ways. Moreover, identity can represent a focus for intervention, as identity is central to a whole group of BCTs. In this way, interventions that focus on the HIV negative and enable them to stay HIV negative represent a core, and as yet unexplored, focus for reducing onwards infection.

However, HIV-negative status is different from HIV-positive status in that the latter is not amenable to change. HIV-negative status and any concomitant identity are only as accurate as the last HIV infection test results that people have received. There is variation in the ability of different testing technologies to identify recent HIV infection, with older tests accurately reflecting HIV status only around 3 months prior to a testing event. In this way, depending on the specific testing technology, the timing of HIV risk-related behaviour preceding the test and the time that has elapsed since the most recent test, the accuracy of a HIV infection status may vary considerably.

The HIV infection testing process represents a complex intervention its own right. Some forms of delivery of the testing process necessitate a focused interaction with a health professional, representing a teachable moment. Teachable moments are those events or sets of circumstances in which individuals may be particularly responsive to behaviour change interventions. Although the concept lacks internal coherence, it is intuitively plausible. Moreover, it has a resonance with concepts such as ‘cues to action’ from the health belief model,61 and also chimes with frameworks such as ‘stages of change’, which also provide a heterogeneous account of people’s readiness and willingness to engage with behaviour change interventions. Lawson and Flocke69 describe a teachable moment as something cocreated through interaction, and highlight the importance of communication between health professional and patient. 69

Like the HIV infection testing process, receiving negative HIV infection test results also represents a teachable moment. It can be thought of as a BCT in and of itself (i.e. ‘biofeedback’70,71) and, if delivered in a health professional context, also can be thought of as including the BCT ‘credible source’ in which verbal communication is likely to occur in favour of behaviour change. 70 Equally, there is literature that suggests the ongoing receipt of negative test results in the light of ongoing high-risk behaviour represents reinforcement of perceived invulnerability and unrealistic optimism. However, it could be argued that the receipt of negative test results represents less of a teachable moment than the testing process itself, as if the patient has been concerned and anxious about their risk-related behaviour, it may be that they are so relieved at hearing negative results that they lack the cognitive capacity to engage with the offer of a brief behaviour change intervention.

This complexity means that any systematic review with a focus on HIV-negative MSM must achieve a difficult balance in setting inclusion criteria. The most reliable HIV-negative populations are those who take part in interventions delivered immediately after receipt of a HIV-negative test result. However, very few trials focus on recruiting this particular population alone. From a more psychological perspective, it makes sense to transfer knowledge to HIV-negative populations of MSM who think and act as if they are HIV negative, while acknowledging that many may actually have seroconverted since their last test result. At the same time, and more pragmatically, it is important to consider whether or not knowledge gleaned from studies in which the vast majority of MSM are HIV negative (e.g. by self-report), can usefully be transferred to populations of MSM who have better knowledge of their HIV status.

In relation to these issues and advice from the project management and advisory groups, the current review sought to exclude studies that included HIV-positive men alone, and to exclude studies in which 33% or more of the population were HIV positive.

Rationale for a focus on behaviour change techniques

In recent years there has been an explosion of interest in clarifying, and examining, the specific content of behavioural interventions. To date, much of this work has focused on individual-level interventions and represents a focus on reflective behaviour change interventions, rather than those that are oriented to more non-reflective processes of facilitating behaviour change. BCTs are described as ‘the smallest component compatible with retaining the postulated active ingredients that is, the proposed mechanisms of change, and can be used alone or in combination with other BCTs’. 70 The focus on BCTs represents an attempt to develop a common language of intervention content that is concerned with the granularity of intervention detail. There is growing evidence of the feasibility and value of using BCT taxonomies to review behaviour change interventions,71–75 in order to shape the specificity of future candidate interventions and policy practice. Analyses that examine the content of interventions at this detailed level are likely to yield results that are similarly detailed. To date, 93 individual BCTs have been identified, and these individual techniques are clustered into 16 distinct groups of BCTs. Further information on the details of BCT is available at www.ucl.ac.uk/behaviour-change-techniques (accessed 26 November 2016).

In contrast to the granularity of the BCT approach, the recent NICE guidelines, Individual-Level Behaviour Change. External Evidence Review 1: Review of Current NICE Guidance and Recommendations,76 highlight the value of a less myopic perspective and note that it is highly likely that there are synergistic effects of individual BCTs working together and in specific social contexts. They suggest the importance of analysing BCTs in conjunction with the potential moderating effect of other variables (such as mode of delivery and intervention intensity). To date, there has been very little domain-specific developments of BCTs in relation to HIV infection risk reduction and no specific examination of the literature regarding MSM and HIV risk-related behaviours (one study77 focused on general populations and condom use).

Rationale for a focus on the role of theory

The appropriate use of theory within behaviour change is central to the effective design and evaluation of complex evaluations. 78 A trend towards increasing use of theory in behavioural interventions has been reported. 79 There is no clear consensus regarding whether or not the use of theory in behaviour change interventions enhances their effectiveness. Some authors have indeed found a positive correlation between theory and effectiveness. 80–83 However, others have found that the relationship is not so clear. 84,85 There are clear arguments regarding the reasons why the correct use of theory should improve effectiveness. 78,86,87 For example, outlining a hierarchy of outcomes and selecting a primary outcome or selecting the use of appropriate BCTs should improve intervention effectiveness. Equally, detailing the key mechanisms of behaviour change and enabling intervention evaluation designs to assess the role of moderation and mediation factors in tracking intervention effectiveness at an individual level should improve intervention effectiveness considerably.

There are also key reasons why the relationship between theory and effectiveness can be obscured, for example tokenistic or poor operationalisation of theory or, indeed, the inappropriate choice of theory. In relation to these issues, a systematic review is needed that has a distinct focus on understanding the role of theory within interventions, how well, and in which particular ways, theory has been operationalised. Equally, there is a clear need to assess how the patterning of effectiveness is associated with theory use.

Rationale for a focus on modes of delivery

Given the rise of the social media in shaping the sexual cultures of MSM outlined at the start of Chapter 1 and the increasing prominence of digitally delivered sexual health interventions,88 there is a demonstrable need for a clear focus on the ways that interventions are delivered. It is also important to assess who delivers interventions and whether or not effectiveness is patterned by the type of interventionist available, as this may have implications for treatment fidelity.

Rationale for expert event

It is vital to extract the most useful evidence relating to cost- and clinical effectiveness of interventions. However, in order for this evidence to be useful in NHS settings, it is also important to assess the transferability of this evidence to contemporary UK settings and to identify the factors that are critical to facilitating the implementation of potential interventions in the near future.

Aims

• To determine the clinical effectiveness of behaviour change interventions to reduce risky sexual behaviour after a negative HIV infection test in MSM.

• To develop a candidate intervention suitable for the UK context.

Objectives

• To map existing evidence on behaviour change interventions to reduce risky sexual behaviour after a negative HIV infection test in MSM.

• To evaluate the clinical effectiveness of behavioural change interventions to reduce risky sexual behaviour after a negative HIV infection test in MSM.

• To identify the most effective component or components within behavioural change interventions in reducing risky HIV infection-related behaviours:

  • model of delivery

  • number and type of BCTs

  • theory-congruent clusters of BCTs.

• To use post hoc realist review methods to further synthesise the evidence in order to develop a candidate intervention.

• To examine the sequential presentation of intervention components in order to develop a candidate intervention.

• To develop a candidate intervention and associated intervention manual.

• Organise and host expert events to enable the translation of findings and further develop a candidate intervention.

Eligibility criteria

Identification of studies

Search strategies were designed to capture the four relevant concept areas: (1) HIV, AIDS, or sexually transmitted disease (infection); (2) prevention research methods (e.g. intervention, education, counselling and evaluation); (3) sex risk-related behaviours and biological outcomes; and (4) target population of MSM. All relevant keywords were combined with appropriate search filters. All searches were carried out between January 2000 and December 2014. We chose to restrict searches from 2000, as the widespread availability of highly active antiretroviral therapy renders the translation of findings from earlier periods of time problematic. English-language restrictions were applied. The MEDLINE search strategy is shown in Appendix 2.

Synthesis of results

Included studies

The searches identified 3500 records (see Appendix 3 for details of database sources and references obtained). After duplicates were removed, we reviewed 2571 records (Figure 1). Following reviewing of titles, abstracts and full texts, 11 studies meeting the inclusion criteria were included in the review. 96–106

FIGURE 1.

The PRISMA flow chart for study selection. CCT, controlled clinical trial.

Number and type of studies excluded

A total of 92 full texts were excluded, as they did not meet the inclusion criteria. The reasons for each decision are given in Appendix 3. Articles were excluded for not being a RCT or other suitable study, as specified in our inclusion criteria;107–136 not having the sufficient percentage of MSM in the study population;137–145 not including a sufficient percentage of HIV-negative participants;146–156 or not being an individually focused intervention, instead focusing on groups or structural factors. 157–167 Other reasons for exclusion, such as interventions not being brief, reporting a pilot trial, reporting non-HIV-specific data (e.g. if only alcohol-related outcomes were provided in a study assessing alcohol as well as HIV risk-related behaviours) or data were unobtainable despite our attempts to gain information from authors, were captured in our ‘other’ category. 99,168–197

Ten RCTs, reported in 11 studies, met the inclusion criteria and were included in the final review. 96–106 A summary of the characteristics of the included studies are shown in Table 2. All studies were conducted in the USA with the exception of one study, which was conducted in China. 103 These studies evaluated the efficacy of a number of online and face-to-face interventions in the reduction of HIV risk-related behaviour.

Recruitment methods and eligibility criteria

Table 2 shows that the participants across the trials were recruited through a variety of sources. These include recruitment of those who attended clinics for testing,99–101 and recruitment via community outreach,98 at gay bars103,105 and through distribution of advertising material and fliers. 102,105 A number of studies, in particular those evaluating online interventions, recruited participants online using banner advertisements on relevant websites. 96,97,102–104 One study also used respondent-driven sampling after initially recruiting at relevant commercial gay venues. 103

Eligibility criteria were similar across all studies; all participants were MSM who engaged in HIV risk-related sexual activity within a specified time period. However, the definition of HIV infection risk and the associated time period varied. The number of individuals randomised across the trials ranged from 143 to 3029. The mean age of the participants was reported in five trials; the ages ranged from 16 years to > 61 years. Three trials specifically targeted ‘young’ adult MSM: 18–24 years,97 18–39 years96 and 18–29 years. 105 The majority of participants across the trials were self-identified as ‘white’ or ‘Caucasian’. In one trial, a separate analysis on the subgroup of ‘men of colour’ was undertaken. 99 The trial conducted in China103 did not report ethnicity within the Chinese population. Most studies recruited only HIV-negative participants. In one study, HIV status for 16% of the participants was unknown. 96 In another study, 17% of participants were HIV positive and 8% were of unknown HIV status. 104 Similarly, one study included 21% HIV-positive and 9% HIV-unknown status participants. 106 Socioeconomic status was not reported explicitly in any study and other markers of socioeconomic status are inconsistently reported (income, educational attainment and employment status). However, most studies note that their samples tend to over-represent men with higher socioeconomic status. One notable exception is Eaton et al. ,102 who highlight the relatively high numbers of men with low income and unemployment within their study (specific details of educational attainment are not reported).

Outcome measures

All studies sought to minimise the risk of HIV transmission. As Table 2 shows, within the studies multiple outcomes are described, and discernible hierarchies of outcomes are often unclear. A priori primary outcomes are rarely listed. Carpenter et al. ,96 for example, do not detail any primary outcomes per se, Parsons et al. 105 only detail provide concerning ‘outcomes of interest’ and Eaton et al. 102 provide multiple outcomes. Similarly, Picciano et al. 106 do not clearly state primary outcomes. All other studies provide some detail of primary outcomes; some explicitly refer to secondary outcomes.

Primary outcomes are as follows: incidence of UAI at a 3-month follow-up (inferred);96 change in counts of risky sexual behaviour over 3 months;97 number of UAI events; number of UAI partners and number of UAI events with three most recent non-primary partners;98 UAI in the preceding 90 days with a man who is not the primary partner and who is of unknown or discordant HIV status;101 change in number of episodes of UAI within a 90-day period;100 change in number of episodes of UAI with non-primary non-concordant male partners in previous 90 days;99 and number of unprotected anal sex acts with HIV-positive/unknown status partners (inferred). 102 Hao et al. 103 report a range of primary outcomes, including episodes of UAI with any male sex partners in previous 6 months; episodes of UAI with regular male sex partners in previous 6 months; and episodes of UAI with casual male sex partners in previous 6 months. 103 Hirshfield et al. 104 utilised HIV infection testing, serostatus disclosure and UAI incidence at 60-day follow-up. 104 Parsons et al. 105 utilised UAI with a casual partner (overall and under the influence of drugs/alcohol) and number of days of drug use (inferred). It can be inferred that Picciano et al. 106 utilised a range of primary outcomes including percentage of participants reporting monogamy; percentage of participants reporting negotiated safety relationships; percentage of participants reporting using condoms for all anal sex occasions; and percentage of participants reporting abstaining from anal sex. 106

Therefore, UAI was the most common outcome utilised; however, definitions also differed among studies. These differences related to addressing the more complex aspects of measuring HIV risk, for example sex role within unprotected sex or partner type. Some studies incorporated aspects of the relationship context of unprotected sex; for example Dilley et al. 99–101 specifically measured the number of UAI sex episodes with non-primary non-concordant male partners in the previous 90 days. Similarly, Hao et al. 103 differentiated between UAI with regular and casual male sex partners. Carpenter et al. 96 and Eaton et al. 102 assessed HIV status of sexual partners to further differentiate the context of UAI.

In terms of secondary outcomes, it can be inferred that Christensen et al. 97 address sex shame from baseline to immediately after testing. Hao et al. 103 report change in HIV and syphilis incidence, whereas Eaton et al. 102 reported condom use self-efficacy and HIV risk perceptions. Eaton collected data on alcohol and substance use among participants using the Drug Abuse Screening Test and Alcohol Use Disorders Identification Test measures. Picciano et al. 106 also collected data on drug use using the Drug Abuse Screening Test. Dilley et al. 100 collected data on incidence of drug use using the Addiction Severity Index.

The follow-up periods ranged from 1 week to 12 months across the trials: 1 week,106 1 month,102 2 months,104 3 months,96–102,105 4 months,106 6 months,99–101,103,105 7 months,106 9 months,105 10 months106 and 12 months. 99–101,105 Incentives were provided in all studies, with the exception of Hirshfield et al. 104

No measures relating to economic aspects of the interventions were described.

Nature of interventions and comparators

Table 3 shows that the interventions evaluated varied considerably in their overall and specific content, and in relation to the ways that distinct intervention components were sequentially delivered. Many were, broadly speaking, brief counselling-based interventions, often incorporating major elements of MI. 96,103 MET was evaluated in one study106 and represents an adapted version of MI. Personal cognitive counselling (PCC) was used in several studies and combined techniques drawn from MI. 98–101 PCC uses structured questions to encourage clients to scrutinise and change their ‘self-justifications’ (e.g. thoughts, attitudes or beliefs), which are understood to facilitate their engagement in high-risk sexual behaviours.

Five studies evaluated interventions based on novel interactive visual and/or multimedia components. 96,97,102–104 These include the use of a multimedia presentation using simulated peers;96 intervention based on ‘Socially Optimised Learning in Virtual Environments’,97 an interactive game incorporating avatars to engage in decision-making and rehearse behavioural skills in a number of virtual scenarios involving sexual decision-making situations and a specially designed graphic novel which was intended to facilitate sexual decision-making. 102 The graphic novel visually depicted a narrative of a man and his journey to seroconversion. Video-based components were also included in two studies evaluating multicomponent interventions. 103,104

In the majority of the studies, behavioural change interventions were compared with a ‘usual care’ intervention. These were usually brief standard counselling-based interventions based on local CDC guidelines. 99–103 Other studies used knowledge or information-based interventions;98,104,106 for instance, Coffin et al. 98 delivered HIV infection testing followed by information on HIV transmission and testing; and Carpenter et al. 96 delivered a stress reduction intervention. One study used a waiting list control. 97

Apart from the three online interventions, most of the interventions were delivered in a clinical setting98–101,103 and two were delivered at a research site. 102,105 Most interventions were delivered as single/one-off sessions except Parsons et al. 105 and Picciano et al. 106 Dilley et al. 101 included a diary intervention in two of the study arms. The duration of each session ranged from 16 minutes103 to a maximum of 8 hours and 25 minutes delivered over a number of sessions106 (although it should be noted that duration of actual intervention delivery was not recorded).

Proximity to testing

Table 4 describes how the studies included within the systematic review are patterned according to the HIV infection testing process.

Excluding Hirshfield et al. 104 and Picciano et al. ,106 all interventions can be seen as utilising HIV-negative status. Dilley et al. 101 delivered their intervention after testing but before the result was available. Coffin et al. ,98 Dilley et al. ,100 and Hao et al. 103 all delivered their interventions after the patient received a negative test result (thus encompassing any effects of the testing process itself and providing the most accurate HIV-negative status and potential HIV-negative identity).

Behaviour change techniques utilised within interventions and controls

The range of BCTs utilised in both interventions and controls is summarised in Table 5. Across the selected studies, at least one BCT from 15 of the 16 groups of BCTs were included. Only BCTs associated with grouping 14, ‘scheduled consequences’, were not represented in the data set. In total, 32 BCTs out of a possible 93 were utilised in the selected studies. The intervention used in Hirshfield et al. 104 used the fewest BCTs (n = 4), while the one-to-one counselling-based interventions in Dilley et al. 100,101 utilised the most BCTs (n = 11 each). Examining the BCTs used across interventions and controls, the most commonly used BCTs were ‘social support (unspecified)’98–106 and ‘information about health consequences’, which were present across eight studies. 96,97,99–104,106 Other BCTs that were relatively common were ‘problem-solving’96,98–102,105 and ‘pros and cons’. 96,97,102,103,105,106

The control interventions included fewer BCTs overall and, hence, fewer groupings were represented. BCTs associated with the groupings ‘associations’, ‘repetition and substitution’, ‘comparison of outcomes, ‘rewards and threat’, ‘scheduled consequences’ and ‘covert learning’, were not represented. In total, 14 out of a possible 93 BCTs were used in the control interventions across all the included studies. Note that Christensen et al. 97 used a waiting list control. Many of the control interventions were described as ‘usual care’ brief counselling-based interventions and this was reflected in the BCT coding. The most common BCTs used in the controls were ‘information about health consequences’ and ‘social support (unspecified)’, which were present in eight and seven studies, respectively. BCTs used in active control groups, such as usual care, may well be important active ingredients within interventions.

Behaviour change techniques unique to interventions when removing behaviour change techniques shared with control groups

Table 5 shows the range of BCTs employed only in the intervention groups (i.e. disregarding BCTs also used in the corresponding control groups). It provides definitions of the BCT concerned and illustrative examples taken from the primary studies. Table 6 illustrates the distribution of distinct BCTs present in interventions when removing BCTs used in both the intervention and control groups. It shows that a broad range of unique BCTs has been utilised within the interventions. Only 3 of the 16 groups of BCTs are not represented here; these are BCTs associated with ‘scheduled consequences’, those with a focus on ‘covert learning’ and those associated with ‘antecedents’.

The most common groups of BCTs used relate to ‘goals and planning’ (across all selected studies, a BCT from this grouping is coded as present 12 times), ‘identity’ (BCTs from this grouping are represented nine times across all studies), ‘social support’ (appearing a total of seven times across all studies) and ‘comparison of outcome’ (again appearing a total of seven times across all studies).

In terms of the total numbers of individual BCTs used in interventions, removing those used also in control groups leaves a relatively small range of different BCTs. Christensen et al. 97 utilised the most unique BCTs in their intervention (11), while Parsons et al. 105 used nine, both Carpenter et al. 96 and Picciano et al. 106 used eight, Hao et al. 103 used six and Coffin et al. 98 used five, as did Dilley et al. 101 Hirshfield et al. 104 and Dilley et al. 100 used four BCTs and Eaton et al. 102 used three. The most commonly used unique BCTs were ‘pros and cons’,96,97,102,103,105,106 ‘goal-setting’,96,101,105,106 ‘social support (emotional)’,96,103,105,106 ‘framing and reframing’98–101,105,106 and ‘incompatible beliefs’. 100,101,103,104

The use of theory within selected interventions

Table 7 shows that three studies were based on a single theory: Eaton et al. 102 and Dilley et al. 100,101 It shows that theories and models of behaviour were used to explain or predict behaviour in 9 out of 10 studies. A range of theories was reported. Christensen et al. ,97 Coffin et al. 98 and Hirshfield et al. 104 all reported using social cognitive theory. 198 Carpenter et al. 96 and Picciano et al. 106 use the information–motivation–behavioural skills model. 199 The theory of planned behaviour200 is mentioned by Christensen et al. 97 and Hao et al. 103 The health belief model is also used by Hao et al. 103 Eaton et al. 102 utilised a conflict theory of decision-making. 201 In addition, Christensen et al. 97 utilised a theory of shame reduction, which posits shame as a major predictor of negative health behaviours. 202 Two further clusters of the use of theory are worthy of mention.

An additional cluster of studies98,100,101 also used what is described as self-justification theory. 203 Although self-justification is not a formal theory, it is clearly used to explain or predict behaviour and merits some discussion here. Its basic premise rests on an imagined difference between ‘online’ thinking (e.g. heat of the moment) and ‘offline’ thinking (e.g. in the cold light of day). It introduces the importance of self-dialogue concerning the attribution of justifying unwanted behaviour. Through rehearsing these self-justifications in offline thinking, techniques and skills can be garnered that help people solve problems with their risk-related behaviour.

A further cluster of studies address stages of change theory: Carpenter et al. ,96 Coffin et al. 98 and Hao et al. 103 Carpenter et al. 96 used ‘stages of change’ as a motivational tool with all participants. Coffin et al. 98 do not specify how they used the theory. Hao et al. 103 also used ‘stages of change’ as a means of motivating and tailoring the intervention to the individual, although, again, in this study the theory was used to shape the intervention for all participants. In no study was ‘stages of change’ used to select participants for receipt of the intervention per se.

Risk of bias

Table 8 provides details of the risk of bias. Despite three requests to all authors, only two authors provided protocols or intervention manuals. 99–101,104 These documents were utilised alongside the published papers to glean further information regarding bias.

In all other studies, information was garnered from published papers only. Generally, most studies were considered to be fair or good, with the study by Dilley et al. 99,100 being considered as fair, scoring low risk of bias on all items except ‘other sources of bias’. Eaton et al. 102 and Picciano et al. 106 were the studies judged to be at least risk of bias, with the risk of bias on most of the domains being scored as either high or unclear.

Most studies demonstrated adequate sequence generation and allocation concealment. They were all relatively good at addressing incomplete data and reasonably free of selective outcome reporting.

The main domain on which studies either scored ‘no’ or ‘unclear’ was whether or not participants and/or personnel had been adequately blinded. Of particular note, within individual behaviour change interventions it is often obvious to participants that they have been assigned to the intervention arm rather than the control, particularly when the control is ‘usual care’ or a null comparison. Likewise, it is often not possible to blind those delivering or administering face-to-face interventions. Although such studies would score ‘no’ based on the given criteria, it could be argued that such lack of blinding is unavoidable and, hence, any bias is also unavoidable and should not be considered as impeding the quality of the study.

When studies scored ‘no’ or ‘unclear’ for ‘other sources of bias’, this was mainly because of baseline imbalance in the outcome measure or in factors related to the outcome measure, possibly a result of the small sample size and/or low participation rate reported.

Intervention fidelity

Table 9 shows the treatment fidelity. Summary scores of treatment fidelity ranged from 20%102 to 68%;105 therefore, no intervention had a high (> 80%) treatment fidelity rating overall. Studies that scored higher in design96,100,103 gave detailed accounts content of interventions, and the number and length of contact for both treatment and control interventions. Dilley et al. 100 and Parsons et al. 105 both scored very highly for interventionist training, as they provided considerable detail on how providers were trained, how their skills acquisition was assessed, how training was monitored and what was expected of providers. This domain was not relevant for three studies,96,97,104 as they were delivered online. Parsons et al. 105 was the only study that scored fairly highly for intervention delivery, providing details on how the methods ensured and assessed that the intervention was delivered as specified and adhered to by participants, that there were plans in place to assess the delivery of the active ingredients and proscribed components of the intervention and that there was a manual available. Most studies scored fairly poorly on intervention receipt (some scored 0%), which meant that they did not provide detail on how well participants understood the intervention, whether or not they had the skills to be able to comprehend the intervention or if the interventions took into account multicultural factors in relation to participants. None of the studies scored for intervention enactments, which detail how intervention skills are assessed beyond treatment.

Assessment of effectiveness

Overall, the trials included in this review reported positive findings and suggested that behavioural change interventions are effective in reducing risky sexual behaviour in HIV-negative MSM. However, these trials are heterogeneous, in terms of participant characteristics, the nature of the interventions evaluated, the length of follow-up and definition of risky sexual behaviour. Therefore, the estimated treatment effects in these studies varied substantially.

Risk-related behaviour outcomes

As noted in Chapter 1, HIV risk and attempts to measure it are complex and imperfect. However, all 10 trials measured behavioural outcomes related to HIV risk reduction (Tables 10 and 11). These include UAI,96–106 number and type of sexual partners102,106 and HIV disclosure. 104 UAI was the most commonly reported behavioural outcome relating to HIV risk; however, definitions of UAI differed among the studies, for example with respect to specificity of UAI (in general or specific to receptive or insertive UAI), partner type (casual or regular) or period of recall. Moreover, some studies also reported multiple measures of UAI.

In an attempt to draw comparisons across the studies, we had to provide some consistent way of looking across these varied measures of HIV risk. We decided to present the findings of the most risky type of UAI reported within individual studies. This was based on contemporary knowledge of HIV transmission that shows that receptive unprotected intercourse with casual partners of HIV-unknown or HIV-positive status carries more risk than insertive unprotected sex. Similarly, unprotected sex with a regular partner (in which case status disclosure is more likely to occur) carries less risk than that with a casual partner, and unprotected sex with another HIV-negative man carries no risk of HIV transmission at all. Although we acknowledge that unprotected sex with a HIV-positive person with an undetectable viral load carries very little risk of transmission, no studies reported here examined the viral load of positive partners of the participants.

We were able to express the reduction of UAI risk as risk ratios based on data from seven trials98–101,103–106 and as standardised mean difference in four trials. 96,99,100,102,106 Two trials reported outcomes as both rates and means. 100,106 One trial did not report quantitative findings regarding change in UAI, even though it was one of the outcome measures evaluated. 97 The majority of trials reported outcomes at 6 months; when data were available, we presented outcomes at this time point. For trials without a 6-month follow-up, we presented data at the follow-up period closest to 6 months.

With the exception of the trial by Hirshfield et al. ,104 who found no difference in sexual behaviour between the intervention and comparator arms at 4 months, all other trials reported a reduction in UAI (see Table 10). In particular, three trials reported a statistically significant reduction in UAI risk in the intervention arm. 99,100,103,105 All trials, with the exception of that by Picciano et al. ,106 reported a reduction in mean UAI episodes in the intervention arm (see Table 11).

Other reported key outcomes

In general, the trials reported that ‘other’ outcomes were better in the intervention arm than in the control arm (Table 12). However, these are not primary outcomes and the trials were typically underpowered for the detection of statistically significant differences. Trials that evaluated outcomes at multiple time points suggest the possibility of greater risk reduction over time. However, even in the longest trial participants were followed up for only 12 months. In addition to behavioural outcomes, one trial also reported cognitive outcomes,102 expressed as condom use self-efficacy and HIV risk perceptions.

Exploratory meta-analysis

As an exploratory analysis, we pooled the UAI risk ratios from the trials (see Tables 10 and 11). Overall, there is a statistically significant reduction in the risk of UAI (risk ratio 0.75, 95% CI 0.62 to 0.91). However, there is much clinical and methodological and statistical heterogeneity (I² = 57%, τ² = 0.04) among these studies, and this pooled estimate must be interpreted with caution. Additional exploratory a priori subgroup analyses were also carried out. Based on the mode of delivery of the interventions, the results suggested that face-to-face delivery may be more effective than interventions delivered via the telephone or online (Figure 2). Based on the proximity to HIV infection testing and implementing the interventions, the results suggest that immediately after testing may be associated with greater treatment effects (Figure 3).

FIGURE 2.

The distribution of risk ratios.

FIGURE 3.

The distribution of standardised mean difference.

In order to explore the relationship between behavioural change components and potential summary effect size, further subgroup analyses were also carried out. To illuminate the differential role of BCTs within interventions, interventions were divided in two: those interventions dominated by BCTs from the ‘goals and planning’ and ‘identity’ groups and those whose BCTs were more heterogeneous. Interventions containing ‘goals and planning’ and ‘identity’ BCT components were potentially more effective than those with heterogeneous BCT components (Figure 4). Similarly, in terms of the measure of theory-congruent BCTs, a score of ≥ 4 was potentially more effective than those with score of < 4 (Figure 5). There was no apparent trend in relation to number of BCTs.

FIGURE 4.

Risk ratios of UAI stratified by BCTs. RR, relative risk/risk ratio.

FIGURE 5.

Risk ratios of UAI stratified by theory-congruent BCT scores. RR, relative risk/risk ratio.

We also explored the feasibility of undertaking indirect comparisons across the interventions by anchoring on the ‘control’ group. However, on reviewing the evidence, it was clear that control groups are heterogeneous across the studies, and cannot be considered as a common connecting treatment. Furthermore, it was clear that such an analysis with the existing data would violate the fundamental assumptions on consistency and transitivity. We therefore followed the criteria set out in Chapter 2, which concluded that going beyond the stratified pairwise meta-analysis would not add value to our understanding of the existing evidence.

It may be less advantageous to utilise interventions that are poorly targeted, delivered only via digital and other electronic media, and adopt a range of theoretical perspectives. For example, Hirshfield et al. 104 and Picciano et al. 106 were the only researchers to include HIV-positive and HIV-negative men (see Table 9). They employed media-based interventions that were not delivered face to face (Figure 6), not delivered immediately after testing (Figure 7) and based on a relatively heterogeneous range of behaviour change theories (see Figures 4 and 5). They also reported relatively low effect sizes. This indicates that there may be important attributes of intervention, which potentiate impact on risk-related behaviour. This is explored in greater depth in the next chapter.

FIGURE 6.

Risk ratios of UAI stratified by mode of delivery. RR, relative risk/risk ratio.

FIGURE 7.

Risk ratios of UAI stratified by proximity to testing. RR, relative risk/risk ratio.

Introduction

The previous chapter highlighted how exploratory meta-analyses suggested that brief behavioural individual-level interventions benefit MSM in terms of reducing HIV risk-related behaviour. However, the intervention effect is difficult to interpret because of the range of uncertainties outlined earlier (e.g. the limited number of studies included within the review, variations in outcome measures, recall periods and other aspects of study design such as discrepancies in sample size/power, the nature of the comparator condition, the proximity of interventions to the HIV infection testing process and heterogeneity in terms of mode of delivery).

An original objective of the funded research was to organise and host an expert event to enable (1) the synthesis and (2) the translation of the review findings to develop a candidate intervention ready for manualisation. Given the limitations of the meta-analytic perspectives possible in the study and the resulting complexity of synthesising study findings, alternative approaches to identifying a candidate intervention suitable for manualisation have been adopted.

A range of qualitative approaches are available to synthesise complex and heterogeneous evidence. For example, approaches such as metasynthesis, metaethnography, metanarrative and other forms of thematic analysis are often used in health research. These approaches to qualitative synthesis often aim to identify patterns of findings across a range of studies (metasynthesis, metaethnography and narrative synthesis in particular) rather than seeking to identify a particular the causal mechanism that underpins given interventions. Realist synthesis is particularly useful when data are both quantitative and qualitative. Finally, realist synthesis adopts a particular theory-driven approach to the identification of mechanisms in individual interventions and studies, and then seeks to focus on recurrent relationships between concepts (known as demiregularities) rather than focusing its synthesis on individual concepts, as many other approaches do. 204

‘Theory’ here does not relate to the formal theoretical frameworks or models used within some disciplines to empirically explain variance in outcome measures. Instead, theory in this approach is understood as a heuristic and explanatory device, whereas the articulation of theory in this approach may reflect existing theoretical assumptions and frameworks, or it may not. For this reason, the approach adopted here reflects the explanatory-based content of primary studies and builds theory (sometimes called a ‘programme theory’) that facilitates the establishment of possible causations and inferences about why certain outcomes are generated in particular contexts. 205–207 In this way, ‘realist synthesis’ is a theory-driven methodology to evaluate complex interventions, ascertaining what works, for whom, and in what contexts. 208

Unlike conventional reviews, realist synthesis is explanatory rather than summative. Our meta-analyses have highlighted that individual-level behaviour change interventions among MSM are highly heterogeneous, presenting major challenges in drawing conclusions about how they work. Realist synthesis identifies and articulates ‘programme theories’ (which detail why and how an intervention works) through focusing mainly on the context and mechanisms in which complex interventions work or do not work. In a realist synthesis data are used to iteratively improve programme theories and build a refined framework capable of shaping future interventions. Within a selected domain, it provides an opportunity to gain novel perspectives not occluded by disciplinary assumptions and dogma.

We conducted a post hoc realist synthesis of the 11 included papers, with the aim of confirming, or refuting, key aspects of interventions related to reducing HIV risk-related behaviour. This required equal attention to the papers that reported a successful intervention and those that reported that the intervention added little to the effects of a control arm; therefore, articulating fully the mechanisms within the intervention components to understand how they worked or did not work.

Realist synthesis begins with an exploration of existing theories or models in order to provide an initial framework as a reference point for data extraction and interpretation. This initial framework can then be consulted, refined and tested using data from the review, iteratively refining key ideas suitable for future intervention development.

In order to develop the initial framework, several health behaviour change models and theories were integrated. The expertise and knowledge of members of the research team was combined with the key assumptions reflected in the formal theories utilised within the review, including the information–motivation–behavioural skill model,199 social learning theory,198 the theory of planned behaviour200 and the health belief model. 209 In this way an initial framework was developed which provided a comprehensive scheme of understanding how interventions work and why. Given the individual-level nature of interventions included within the review and the hypothesised mechanism of action implicit within their underlying approaches, such a framework is psychological and primarily concerned with the individual and their mental world. In other words, it is concerned with illustrating behaviour change within the individual and in relation to their thoughts and feelings.

Our initial framework (Figure 8) illustrates the assumed personal resources and characteristics that clients bring to the intervention. Given that the interventions included within the review were motivational in nature, clients were also assumed to have, or to develop, intentions to change behaviour. In many psychological models, intentions together with attitudes and beliefs are expected to predict whether or not the intervention ‘works’. Features of the intervention are assumed to activate, promote or support necessary changes in behaviour that lead to desired outcomes. Individuals’ belief in their own capacity to enact change is also widely assumed to be an important aspect of behaviour change and is reflected in a range of concepts, such as perceived behavioural control, health locus of control or, most commonly, self-efficacy. As a result, within the initial framework, self-efficacy was specified as a possible moderator; specifically, its presence may increase the influence of the intervention in delivering behaviour change.

FIGURE 8.

Initial framework.

Analytical methods and synthesis

Realist synthesis involves the extraction of contexts, mechanisms and outcomes, and the interpretation of the relationships between them. These entities are inferred from studies through a sustained period of examination.

Contextual factors affect whether or not the intervention works, and can include environmental factors, intervention design or personal characteristics. Mechanisms are usually hidden processes that explain why the intervention works. Contexts activate mechanisms. Outcomes include both intended and unintended results. We fully acknowledge that several of these factors are likely to be episodic or infrequent in the MSM population, and interventions and, in particular, mechanisms that required a large degree of interpretation and abstraction to identify were discussed with another reviewer (PF), who approached emerging patterns in the data and the resultant theory with social science expertise and extensive knowledge of sexual health interventions. The primary reviewer (HH) was experienced in realist review methodology and in research about health-care interventions. Her background was in improvement science, with little experience of either health psychology or the domain of sexual health. This offered an advantage in that the process of data extraction, interpretation and abstraction was free from a priori psychological theoretical, conceptual or disciplinary assumptions. This allowed us to identify where the model developed in line with existing theory and where it departed from our advanced conventional understanding about sexual health behaviour change.

Realist synthesis involves several iterative stages: familiarisation, individual context, mechanism and outcome (C-M-O) configurations, patterns across C-M-O configurations, the development of general statements and the development of explanatory theories that can then be used to formulate hypotheses in a final refined framework.

Following the construction of an initial framework, a familiarisation phase commenced, to enable the primary reviewer to understand the patterns of intervention characteristics (sample, setting, theory-based components, interventions and BCTs) and outcomes. The reviewer read and re-read the papers, in consultation with summary data from the meta-analysis, to extract possible contexts (which included any important features of the intervention itself), mechanisms and outcomes (which may themselves be interlinked). Extracted information included study setting and population, participant demographics and baseline characteristics, descriptions of the intervention and control conditions, study methodology, intervention completion rates, outcomes and suggestions by the authors about possible mechanisms of action of the chosen intervention strategies. Such information was extracted from the introductions, methodology and descriptions of the interventions (and in cases when the intervention did not demonstrate effectiveness, extractions were also taken from the control/usual care arm) and from the results and discussion sections of each paper.

All potentially important information that the reviewer considered to be contributory to the refinement of our initial framework was categorised as being initially a context, mechanism or outcome. These categories were then used to draft C-M-O configurations for each paper, describing how each contextual factor interacted with mechanisms to produce outcomes (those measured by the original authors, as well as unintended outcomes that were specified in our initial framework). This took the form of diagrams with interlinking arrows between each C-M-O, along with explanatory notes to justify its categorisation. Between 4 and 15 individual, whole or partial C-M-Os were extracted from each paper, although many of the papers contained similar configurations. Appendix 4 summarises the C-M-Os across all papers. However, it should be noted that this table depicts only summary data to the end point of this process; it does not, and cannot, capture the iterative process of exploring the relationships between C-M-O variables, which sometimes involves adjustments to their categorisation (e.g. mechanisms can became contexts, or contexts can became mechanisms).

Then, C-M-O configurations were explored for patterns, across the interventions detailed within the studies, and by constant comparing and contrasting variables between studies, seeking confirmatory and contradictory findings. This allowed comparison of which mechanisms were activated across different contexts. In this way, explanatory theories concerning statements about what works, for whom and in what circumstances were formulated. This process was conducted in tandem with continual revisitation of the initial framework, which was devised to guide analysis and interpretation for the synthesis (see Figure 8).

General statements or hypotheses were derived from utilising processes of abstraction, with special attention paid to areas of confirmation or divergence from the initial framework. The aim of the synthesis was to refine the initial framework explaining how the interventions detailed within the review worked, with a specific focus on the mechanisms and features of the interventions, either explicit or implicit, which may have had casual effects. Throughout the process of realist synthesis there was a gradual developmental focus from the specific to the general (in other words, analysis and interpretation changed in scope from specific statements about specific interventions in specific papers, to more global statements about how individual behavioural interventions work more broadly). Theming and organisation of these statements, using the initial framework as a guide, resulted in three explanatory theories, which articulated the underlying processes by which interventions are expected to operate.

A distilled collection of features and associations was identified. This is presented in Table 13.

Explanatory theories

The theming and organisation of statements generated the formation of three key explanatory theories (which capture and articulate the underlying and general logic about the processes that create outcomes). 210 These each provide explanations about how individual HIV risk-related behaviour change interventions work. Each contributes directly to the refinement of the final explanatory framework (Figure 9).

FIGURE 9.

Refined framework.

The only assumptions that structure the realist review were that interventions should be face to face and one to one after testing, based in some kind of counselling format, targeted at reducing the risks for high-risk clients, and not aiming to achieve ‘reach’ (i.e. increase engagement in testing or counselling).

Introduction

Using post hoc realist review methods, the previous chapter highlighted intervention components that were indicative of increasing the effectiveness of individual behaviour change interventions among MSM through the delivery of a refined framework and accompanying explanatory theories. The refined framework suggested three explanatory theories relating to interactions of C-M-Os. Together the three theories provide useful direction for developing future interventions in this field. This chapter is concerned with further aspects of intervention development and specification.

The post hoc realist synthesis suggested that ideal interventions should include a number of key elements. It suggested that interventions should start with the client’s explication of past risk-related behaviour to objectively quantify and acknowledge previous risk. This should be followed by supportive and skilled facilitation that manages the emotional vulnerability of the client. The interventionist should be non-judgemental and should distinguish between the client’s identity and his health-compromising behaviour. Within this interactive element, the interventionist creates an educationally sensitive moment utilising the client’s emotional state. The interventionist should focus the client on past thoughts, feelings and behaviour concerning risk and explore alternative patterns of thought and behaviours which are associated with risk reduction. These alternative thoughts should address plans and goals for the future. Opportunities to rehearse thinking styles and behaviours are likely to enhance the intervention through the growth of perceived control and perceptions of self-efficacy. Throughout the interventions the reduction of negative internalised emotions is likely to be important in enhancing effectiveness. Positive, affirmatory interactions should replace negative feelings. Interventions should end with action planning, as this appears important in the transfer of new skills to future sexual practices.

This chapter further examines the available evidence to enhance the development of a candidate intervention. It moves beyond the commissioned research to deliver additional tasks. It focuses on two supplemental areas: (1) further examination and interpretation of the components of interventions as delivered in sequence, to enable the specification of detail required to create an intervention manual; and (2) reassessing the included studies to maximise the potential transferability of available evidence, by critically appraising the social contexts of original studies.

In terms of the first key area, the realist synthesis suggested the importance of attending to the sequential and temporal elements of the interventions. From the client’s perspective these are likely to be important, as they shape the client’s personal journey through the intervention. They reflect the client’s exposure to the range of BCTs and aspects of intervention modality. An evidence synthesis collated for recent NICE guidelines,76 for example, also highlights the value of perspectives that focus on a wider consideration of intervention content, over and above that provided by the BCT framework in isolation. The evidence synthesis suggests that it is highly likely that there are synergistic effects of individual BCTs working together with cumulative effect. Along these lines, in order to further develop and specify a candidate intervention, it was felt that supplemental analysis should focus on these sequential and potentially cumulative aspects of intervention content and delivery.

The second key area examined here is concerned with a fuller examination of the historical and social context of the interventions. The realist review focused on the microcontext of the intervention as it was delivered. However, in order to further develop an intervention and pre-empt issues relating to the feasibility of potential future studies, it was felt that a wider consideration of cultural and historical issues may provide further useful information and guidance in developing a candidate intervention.

The chapter concludes with the development of a matrix to synthesise the findings from the preceding chapters. It provides a transparent account of how the different types of evidence were utilised within the development of a candidate intervention. The chapter ends with a presentation of the candidate intervention defined by a ‘mock’ intervention manual.

Method

With regard to the sequential and potential synergistic effects of intervention components, no formal pre-existing assessment tool could be found. Data extraction focused on trying to detail these complex, sequential and potentially cumulative aspects of intervention content. First, each intervention component in turn was noted. Following this, each study was described in relation to the diversity of components used, the incorporation of tools or discrete devices within the intervention, whether or not there was a sense of increasing complexity within the delivery of the intervention across constituent components, how interventions both began and ended, whether or not there was a pattern of increasing personalisation across the intervention and the degree to which within the whole intervention there was a focus on the intrasubjective (i.e. the client thinking about himself presented a major focus for intervention). In addition, given the project team’s experience in working with this population, three additional areas of intervention delivery were examined within analysis of the studies: whether or not there was evidence of how the intervention as a whole was imbued with gay identity and community; whether or not it reflected cultural competency with regard to the sexual cultures of MSM (i.e. a set of congruent behaviours, attitudes and policies that come together within intervention delivery to reflect and affirm the sexual cultures of MSM); and the degree to which processes of social influence may have played a tacit, yet still important, part within the research process. The data were examined for patterns and trends in order to provide further direction in the development of the candidate intervention.

One reviewer (PF), with particular expertise in the sociocultural aspects of gay men’s sexual cultures, led the assessment of the 11 included studies in relation to factors that may be important with regard to the transfer of evidence to contemporary UK contexts. As no formal tool exists, data relating to cultural and historical contexts were systematically collated regarding the temporal, national, regional and local contexts of research. Such sites were also assessed in relation to their association with gay communities and associated cultures and any information highlighting the role of gay communities within the intervention, either explicit or implicit, was noted. The data were examined for patterns and trends in order to provide enhance the feasibility of delivering a candidate intervention.

To synthesise the evidence from preceding chapters and enable the development of a candidate intervention, key findings were entered into a single table (Table 14). This matrix of findings was examined to ascertain patterns in the findings produced by the diverse approaches to evidence synthesis. This matrix enabled a sense of the commonality of findings across the diverse approaches, yet also provided an illustration of where findings relate to only one approach and, indeed, where no available evidence was presented.

Results

Introduction

The previous chapters have detailed the best available evidence presented as a narrative review, meta-analyses, realist synthesis and the synthesis of findings to date. In the previous chapter, these were combined in order to develop a candidate intervention, which can be thought of as optimal and evidence informed. In this chapter this intervention is further refined through expert events, which sought to consider its translation into the existing service provision in the UK. Medical Research Council guidance on the development, evaluation and implementation of complex interventions notes that insufficient consideration of intervention development, piloting and assessment of issues of implementation can result in weak interventions that are difficult to evaluate and implement. 78

Methods

Experts in both Scotland and England were invited to take part in an intensive full-day workshop, split into three sessions:

1. systemic barriers to, and facilitators of, implementation

2. psychosocial barriers to, and facilitators of, implementation

3. contextual barriers to, and facilitators of, implementation.

One event was held in Glasgow and another in London to include perspectives on the different health systems in Scotland and England. Table 17 provides an illustration of the diverse expertise utilised within the expert events.

Results

Dual theoretical synergy matrix of key barriers to, and facilitators of, intervention implementation

Having identified the major systemic and psychosocial barriers to, and facilitators of, intervention implementation through the NPT and TDF sessions, respectively, and the contextual factors specific to this intervention in particular, the final stage of the DTS analysis combined the results to assess how the individual (i.e. psychosocial) issues are embedded in the systemic and contextual. Similarities and differences by domain were mapped to create a DTS matrix. Figure 10 is a visual representation of the matrix and shows the overlap in the systemic, psychosocial and contextual issues that need to be addressed in optimisation of the intervention. This process identifies the elements of the intervention that will have to be refined and operationalised prior to feasibility testing in order to take forward a candidate intervention that is likely to be implementable in the future.

FIGURE 10.

Dual theoretical synergy matrix of systemic, psychosocial and contextual issues to be addressed in optimisation of the ‘How to stay HIV negative’ intervention.

Key recurrent issues that were evident across all three levels were resources, training, and staff roles. Issues that crossed the systemic and psychosocial levels were institutional support, completing priorities, patient eligibility, intervention delivery, intervention length, feedback and supervision/appraisal processes. Staff competency was identified as an issue at the systemic level (in both the primary clinical setting and community settings), whereas staff skills and enthusiasm (presence or absence) were identified as potential issues at the psychosocial and community levels. Further key issues at the systemic level were the presentation of the intervention (branding), location of clinics and client population in clinics. Further issues at the psychosocial level were knowledge, stress on staff, staff confidence, client readiness, client expectations and negative emotions. Finally, at the contextual level, security and loss of personalisation were suggested issues for delivering the intervention via digital technology, and the less restrictive environment was suggested as a potential facilitator in the community settings. In summary, these issues demonstrate that optimisation of the intervention requires focus on institutional support, intervention delivery, intervention content and training.

Gaining institutional support, at all levels within organisations, will be key to future testing of the intervention, and there was generally thought to be a need to provide greater justification for the intervention, both in terms of the evidence informing the intervention and its potential value (i.e. in terms of transferable skills) to organisations involved. Although there was a belief that staff would be enthusiastic and welcome the opportunity to try out a new tool, there would be a need for further stakeholder consultation to identify the key individuals (commissioners and practitioners) that would be required to support any trial. Requests for more evidence and piloting demonstrate the importance of linking between the current evidence generated by the systematic review and the need to explain why the candidate intervention should be evaluated before being rolled out in practice. The limited evidence base for the candidate intervention suggests a pilot evaluation might first be required. Resources, or rather lack thereof, was a key issue throughout the expert events, and was deemed to be particularly important in gaining institutional support.

There were concerns that resource constraints would limit engagement with the intervention and careful thought will need to be given to intervention delivery, in terms of where it is provided and by whom, as a result. Who should deliver the intervention generated competing views and this merits further attention, although health advisors were identified as one of the staff groups likely to already have many of the skills required. This does have resource implications, and peer educators and/or outsourcing to the third sector were suggested as alternatives. Alternative means of delivery could be explored and were recognised to be useful for geographically remote access. Delivery in community settings (by third-sector organisations and/or the NHS) was thought to be a possibility and a less restricted environment, but the same issues about the capabilities of staff to deliver the intervention would apply. Additional concerns about security issues, technology becoming outdated and the loss of the personalised aspects of the intervention if delivered through apps, Skype™ or online links may limit their potential. There was further concern about the delivery of the intervention immediately following a HIV-negative test result. In practice, most people are not given this result face to face and any such change to clinical structures and work practices would have significant resource implications.

The content of the intervention would also need to be more clearly defined, particularly in terms of what the visual aid or graphic novel would entail. This is required, first, to help support understanding of the purpose of the intervention and to construct the value of it (key elements of whether or not someone would be willing to support and implement the intervention) and, second, to ensure that staff can be fully trained in its delivery (to facilitate intervention fidelity). A particular issue raised by some of the experts was the intervention’s branding and the acceptability of the ‘How to stay HIV negative’ tagline. Branding would be one aspect to explore in further consultation with potential users, and gay men in particular, prior to assessing the feasibility of the intervention. Content could also be affected by resource decisions. The suggested length of the intervention (30–50 minutes) was deemed to present too much pressure on existing resources (time, staff and financial).

Training was suggested to be a key issue. The intervention would be reliant on individual staff remembering to deliver it and feeling capable in identifying and recruiting eligible men when faced with competing demands of their time. Accreditation, prompts/reminders, training refreshers and appraisals (possibly via e-learning) were additionally suggested. Beyond training, it was suggested that supervision, continual assessment, feedback and updates (and an ongoing measure of competency) of, and for, those delivering the intervention would be key to ensuring fidelity and ongoing engagement with the intervention.

It was suggested that training would be required in terms of HIV literacy, MSM, behaviour change, self-efficacy and cultural context, in addition to intervention content and delivery, to ensure that staff had the skills required to deliver the intervention. Work is also required to address how the nuances of contemporary risk management strategies (i.e. the impact of other biomedical HIV infection prevention such as PrEP on these) would affect the screening and identification of potential clients for the intervention. The grounding of the intervention in MI skills and techniques was discussed as a barrier to, and facilitator of, behaviour change, but widespread training in this was seen to be a facilitator, particularly as those trained already could easily be up-skilled to deliver the intervention. This is clearly dependent on who delivers the intervention and the varied capabilities of different staff groups would have to be addressed in the training, as would concerns with delivering particular aspects of the intervention, for example challenging risk-related behaviours, addressing the extent to which clients were ready to change behaviour and dealing with ‘emotionally charged’ situations. Such an intensive, psychological intervention was feared to involve additional personal stresses and consequences for staff involved; however, the experiences of other professional groups facing similar challenges could be learned from and incorporated into training.

Conclusions

To address the complexity and challenges of transferring knowledge to an increasing range of settings and intervention delivery modalities, the expert events were designed to simultaneously identify issues to be addressed in intervention manualisation and enable the optimisation of the intervention prior to feasibility testing. The DTS method identified the major systemic and psychosocial barriers to, and facilitators of, intervention implementation and the benefit of the method is in the assessment of how the individual (i.e. psychosocial) issues are embedded in the systemic and contextual (of this particular intervention). This synergistic consideration of what might facilitate or impede the intervention highlights the central elements of the intervention that will have to be refined and operationalised prior to feasibility testing.

The DTS analysis highlights that gaining institutional support and clearly defining and refining the intervention content and delivery model will be key to optimisation. In turn, extensive training will be required to ensure intervention fidelity. We have demonstrated the intervention’s fit with existing service provision in the UK, including that provided by the NHS, private providers and other sectors such as voluntary organisations, including acceptability by staff and across varied geographical areas (e.g. urban and rural settings). We explored if the intervention could be delivered in a way that encourages access and uptake, for example in community settings or uses new technologies (such as digital media). None was immediately ruled out, but each presented their own challenges to operationalising the intervention. The clinical setting, with access to more specialist services such as STI treatment and psychological support if required, was deemed most appropriate. Cost and resourcing were recurrent issues and there was concern that the intervention would not be able to be delivered within existing budgets. It was clear that the health economics of the intervention would have to be included as part of its evaluation. In the light of concern that such behaviour change interventions would not be prioritised over biomedical prevention, such as post-exposure prophylaxis after sexual exposure and PrEP, it will be important to demonstrate how the two can complement each other in the future, as will evaluating the fit of the intervention if testing service delivery models change because of greater use of self-sampling and testing for HIV.

Most importantly, none of the experts thought that the candidate intervention would be impossible to implement or evaluate. Instead, the DTS method has highlighted how the intervention can be best optimised in order to take forward a candidate intervention that is most likely to be implementable in the future.

Main findings

Overall, the trials included in this review (n = 10) reported positive findings and suggest that individual-level behavioural change interventions are effective in reducing HIV risk-related behaviour in HIV-negative MSM. Overall, there was a statistically significant reduction in the risk of UAI (risk ratio 0.75, 95% CI 0.62 to 0.91). However, there is much clinical and statistical heterogeneity among these studies. As a result, this pooled estimate must be interpreted with caution.

With regard to intervention content, when removing BCTs shared with controls, the most common groups of BCTs used within interventions related to ‘goals and planning’, ‘identity’, ‘social support’ and ‘comparison of outcome’. In terms of individual BCTs the most commonly used were ‘pros and cons’, ‘goal-setting’, ‘social support (emotional)’, ‘framing and reframing’ and ‘incompatible beliefs’. All interventions were based on theory to some extent, although there was considerable variation in the type and numbers of theories utilised. There was some evidence of the use of theory-congruent BCTs.

Exploratory stratified meta-analyses suggest that face-to-face delivery of interventions may be more effective than interventions delivered via telephone or online, that interventions delivered immediately after testing may be associated with greater treatment effects, that interventions containing BCTs from the ‘goals and planning’ and ‘identity’ groups may potentially be more effective than those with more heterogeneous and less common BCT components and, potentially, that interventions that have more theory-congruent BCTs may be more effective than those that contain fewer theory-congruent BCTs. Again, these results must be interpreted with caution.

To illuminate review findings further, and provide direction for intervention development, a post hoc realist review was conducted. It suggested the explication of behaviour, to both quantify and mentally re-enact sexual risk, is an important step in client’s recognising their risky sexual behaviour. Furthermore, it suggested that interventions can be enhanced by supportive, skilled facilitation. Skilled interventionists can create educationally sensitive moments that balance the clients vulnerability with insights into their potential to change behaviour and patterns of associated thought. Clients should be supported in the development of plans and goals. The realist review also suggested that opportunities to rehearse planned behaviours and thinking styles are likely to enhance the intervention. Throughout interventions, the reduction of negative, internalised emotions such as shame is likely to be central to the exploration and future planning of new skills and behaviours. These findings are the result of a sustained and interpretative activity and, again, are limited by the small number of trials included, the subjective nature of this approach (conducted by a single reviewer) and the inferred causality that structures this approach.

In order to develop a more detailed candidate intervention, an additional analysis of the review data was conducted. This focused on an attempt to assess intervention content as a whole rather than in terms of its individual components. It examined patterns and trends within intervention content addressing the number, order and relative purpose of intervention components. This suggested that most of these effective interventions comprised multiple components, delivered with increasing complexity and often focusing on a single developmental or narrative trajectory. Interventions were marked by the first intervention component being different than subsequent components. Moreover, the majority of interventions began with a personal assessment of risk, often conducted with little initial interaction with the interventionist. This assessment of personal risk usually included exposure to relative risks faced by peers. Similarly, the majority of interventions ended with a component that focused on the future, often with the client setting goals or agreeing an action plan. The majority of interventions were personalised and tailored to clients’ individual needs, and most also had a distinct focus on intrasubjective elements with the aim of getting clients to take responsibility for their behaviour change and to identify potential ways to achieve it. The majority of interventions could also be seen to demonstrate cultural competency with respect to the sexual environments and the concomitant dilemmas faced by MSM at the time of their implementation. Once more the results of this supplemental analysis should be treated with caution. They too rely on a single reviewer and a novel methodological approach.

Finally, the review details issues of transferability and overall acceptability regarding the candidate intervention developed within this review. Multidisciplinary experts (n = 24), including MSM (n = 9), agreed that the intervention was possible to implement and evaluate. They highlighted key ways that the candidate intervention could be optimised.

Wider considerations for evidence synthesis regarding behaviour change

Beyond the primary goal of addressing the project’s aims and objectives, the team identified additional issues that may be useful for other researchers conducting similar projects to reflect on and discuss. These relate to considering the scope and implications of search strategies with regard to choices of outcome measures in relation to their locus and specificity, inclusive approaches to mapping the content of interventions and the relative role of theory within utilising BCT taxonomies. We discuss each of these issues in turn.

A behavioural focus for outcome measures

In developing search strategies for evidence syntheses regarding behaviour change interventions, the relative importance, or hierarchy, of outcome variables utilised may differ from those employed within efficacy and effectiveness studies. Considering both the locus, and specificity, of outcome variables is important to yield the most useful knowledge with regard to understanding behaviour change. With regard to locus of outcome variables, those that relate to behaviour change per se are arguably more important and indeed more useful than outcomes that relate to the consequences of behaviour (reflecting a much more complex causal chain). This is true across a range of health conditions. For example, not everyone who smokes will die of lung cancer and useful knowledge relating to smoking cessation per se offers insights into behaviour change and forms a good locus for cumulative knowledge. However, a behavioural locus of outcomes for developing cumulative knowledge for behaviour change is perhaps particularly important within infectious disease when infection also relates to factors outwith the control of the individual. Population-level issues, such as incidence, prevalence and the density of social networks associated with transmission, impact on exposure to a disease sometimes irrespective of individual actions. So depending on the epidemiological context, people engaging in the same risk-related behaviours may have very different chances of being exposed to a disease. For the current study, for example, if we had developed only cumulative behavioural knowledge from outcome data relating to reduced HIV transmission (affecting only a very small fraction of those people who failed to change their behaviour), sample sizes within primary studies would have had to have been far larger and we would have lost valuable data relating to factors associated with behaviour change across the larger population (i.e. those who did not change their behaviour, but were not exposed to HIV). Therefore, HIV transmission data as primary outcome data reflect exposure (a combination of population incidence and prevalence and patterns of sexual mixing) in addition to the role of behaviour change. Therefore, in this context, it may be useful to prioritise behaviour change outcomes for behaviour change evidence syntheses rather focusing on the consequences of behaviour change.

Similarly, it is important to consider the specificity of outcome measures employed within evidence synthesis concerning behaviour change interventions. It is assumed that if studies do not provide data regarding a behavioural locus but do address consequential outcomes of behaviour change, the latter must be directly and indisputably related to the behavioural domain that interventions are manipulating. Problems may arise, however, when consequential outcomes are infrequent (e.g. the example of HIV transmission), and researchers may be tempted to use more frequent, highly correlated proxy outcomes (such as bacterial STIs, which are much more common than HIV transmission). When doing so it is vital to consider the specificity of proxy measures in relation to the behavioural domain that is being manipulated through interventions. For example, the behavioural domain of HIV transmission is distinct from that of many bacterial STIs. The use of PrEP to reduce HIV transmission may actually increase STI transmission, or in more behavioural terms, adopting unprotected oral sex (low HIV transmission risk) rather than anal sex (high HIV transmission risk) to reduce HIV transmission may not change STI-related outcomes (as oral sex is associated with high STI transmission risk), but would change HIV-related outcomes (as HIV transmission through oral sex is very uncommon). In this way, it may be useful for researchers to ensure a direct congruence between the behaviours that are being changed by interventions and the outcomes that are reported, particularly if they are consequences of behaviour change rather than behaviour change itself. This modelling of outcomes may well inform decisions regarding the locus of outcomes and should be core to developing search strategies in relation to choice of outcomes.

Inclusive attempts to map intervention content

The work embodied within this report has also integrated a range of approaches to mapping intervention content; the BCT taxonomy, realist review approaches and ways of looking at the synergistic content of diverse intervention components. All of these approaches are encompassed within the broad field of implementation science yet each stems from a separate epistemological position. Each has its relative merits and weaknesses. Here we discuss our reflections on the varied approaches we adopted and discuss the ways in which they might help health researchers.

With regard to utilising the BCT taxonomy72 to map the content of interventions, the strengths of this approach focus on its granularity (e.g. discrete, replicable and irreducible techniques) and the development of a common and highly specific language to describe particular BCTs. This perspective enables researchers to look across diverse interventions and identify commonality of intervention components irrespective of the way that original authors describe that content. In this way it can compensate for disciplinary, national, cultural or historical trends in reporting intervention content and represents a foundational moment in building cumulative knowledge for behavioural insights. This approach is grounded within a positivist epistemology and makes the most of ideas of cumulative knowledge to conduct effective implementation science. Its weaker aspects reflect its very strengths. Weaknesses include the inability to examine the effects of the context in which its granular components are delivered, for example the ways in which the ordering of BCTs may lead to amplifications (e.g. boosting each others effectiveness) and neutralisations (e.g. cancelling each other out), or indeed cumulative or additive effects (where it is the effectiveness of BCTs delivered sequentially) that makes the most effective active ingredients.

Our original goal was to develop methods to systematically identify patterns in the BCTs used in our interventions, to enable network meta-analysis95 grouping interventions together on the basis of the patterning of BCTs in order to give novel explanatory purchase regarding effectiveness. In this way we had hoped to address the contexts of the BCTs as embedded within interventions. However, owing to the small number of studies identified within the current project, the analysis lacked the necessary statistical power. Nevertheless, this approach may well prove very fruitful for other behavioural evidence syntheses in which larger numbers of intervention studies are available. There is also scope for methodological innovation in coding patterns of BCTs across what appear to be heterogeneous interventions in order to preserve the benefits of their granularity but also address their sequential patterning. In this way analytic techniques, such as network meta-analysis, can deliver comparisons of interventions reflecting the degree of similarity of intervention content.

The second approach to mapping intervention content used within the project was realist review and synthesis. 205–210 It represents an alternative inductive epistemological approach that also delivers useful knowledge regarding the role of intervention content within particular contexts. Its iterative and explanatory focus enables knowledge generation in ways that are not constrained by a priori assumptions and that focus primarily on explanation. Strengths of the approach are to deliver a grounded explanatory theory from a given data set, rather than through the lens of an ‘off the shelf’ theoretical framework drawn from any given disciplinary tradition or indeed within the highly specific framework suggested by the BCT taxonomy. Realist analysis also allows for the identification of latent factors through reflexivity, interpretation and abstraction that might inform how a specific intervention works to produce specific outcomes. The realist review approach may be useful in providing a useful counterpoint to the more rigid frameworks suggested by formal theory or indeed the highly specific framework of the approaches such as the BCT taxonomy. 72 Whereas realist review and synthesis gains explanatory purchase with regard to novel perspective in relation to explaining intervention effectiveness, its weaker aspects relate to the transferability of its insights across domains. Its ability to contribute to cumulative knowledge is thus limited, but counterbalanced, by its context-bound relevance.

The final approach related to our earlier attempt to use network meta-analysis to examine the patterning of individual BCTs to examine some aspects of the intervention contexts. It focused on describing and interpreting potential synergies across descriptions of intervention components. It addressed the ways intervention content, delivery and context combined. For the current project with its focus on intervention development, this particular approach proved very useful for creating a brief mock intervention manual. It looked at normative patterns in the flow of interventions and enabled a sense of how the findings of the BCT taxonomy could be combined with insights from the realist synthesis. This approach uses inductive reasoning to describe normative patterns within intervention flow. Although it does contribute to cumulative knowledge, it does not do so with the rigour and specificity of the BCT taxonomy approach.

Theory and mapping intervention content

Across a range of disciplines the meaning and function of theory can differ dramatically. Overall, there is little conceptual coherence regarding ‘theory’, ‘models’ and ‘frameworks’. At times these terms appear interchangeable and dependent on disciplinary context. Theory is probably most often understood as representing a heuristic device, or explanatory tool, that suggests or illuminates a common pattern in a given phenomenon. In this way theory can represent the way that a phenomenon works for most people, most of the time and can give useful direction for the identification of trigger points for intervention or the targeting of resources. However, in other fields theory can relate to the formal specification of a predictive model often pitched in terms of a single discipline-specific focus, for example in terms of cultural factors (sociology), structural factors such as sociodemographics (epidemiology) and cognitive or affective factors (psychology).

The epistemology of the current study is rooted within implementation science and the primacy of applied research. Throughout we have adopted a diverse range of methods reflecting our pragmatic approach. For the team an inclusive and interdisciplinary approach to theory has also been adopted within the current project: making the most of generic social scientific approaches such as realist synthesis, as well as making the most of contemporary health psychology and its increasingly practice-oriented nature. Following an almost exclusive disciplinary focus on building and refining psychological theory at the cognitive level for the last 40 years, there has been a notable trend to enhance the practical application of psychology within the health domain. This shift does not represent a rejection of the cognitive realm, or a neo-behaviourist turn, but does represent a focus on practical application. Moreover, while a trend towards increasing use of theory within behavioural interventions has been reported,69 there is no consensus regarding whether or not the use of theory within behaviour change interventions enhances their effectiveness. Some authors have found a positive correlation between theory and effectiveness. 70–73 However, others have found that the relationship is not so clear. 74,75

The BCT taxonomy approach to evidence synthesis is not contingent on theory: it can work independently of any theoretical framework. As such, and unlike other approaches such as intervention mapping,217 it does not focus on theory or theoretical constructs per se. In some ways it is ‘agnostic’ to theory and instead, its primary focus is on the techniques themselves and for evidence synthesis how the use of techniques relates to effectiveness. However, it is worth noting that it is entirely possible to ‘reverse engineer’ theory from this end point and it would be entirely reasonable to expect some resonance with more formal theory.

Limitations and strengths

The review and its findings are limited by factors associated with its primary studies. Despite an inclusive operationalisation of population criteria, very few trials were found which could be included within the review. In the primary studies there was considerable methodological and statistical heterogeneity. In terms of assessing intervention content and delivery, findings are limited by an over-reliance on published papers rather than intervention manuals or detailed protocols. There was a lack of process evaluations, which would have further illuminated intervention effects and exposed potential confounding factors. Relatedly, there was limited evidence of intervention fidelity being assessed rigorously within the trials. Again, these findings highlight the need for caution when interpreting these results. Similarly, in the chapters on realist synthesis (see Chapter 4) and intervention elements (see Chapter 5), sole reviewers took the lead with data extraction and analysis and no independent analysis took place. A realist synthesis would also typically search and select studies based on more qualitative criteria (depth and richness of intervention description). However, as this was a post hoc realist synthesis, it was limited to those studies included under meta-analysis criteria and, therefore, cannot fully fulfil its potential in synthesising all available and appropriate evidence. Compensatory approaches were adopted, however, including by the reviewers’ approach to including all outcomes reported in the studies, and not just those of interest to the meta-analysis (UAI).

The strengths of the review relate to both its detailed appraisal of intervention content and its innovative use and development of methodology. In relation to intervention content, for the first time within this field the review has provided detailed descriptions of the content of interventions in terms of BCTs employed. In this way it has illustrated the viability of examining evidence at this degree of granularity. In terms of methodological innovation, as a useful counterpoint to approaches that focus on the granularity of intervention content, the current review used a post hoc realist synthesis. This approach offered perspectives regarding potential factors that may be associated with increased effectiveness which were not limited by a priori assumptions and beliefs. Similarly, an approach to assessing intervention content was developed which charted the episodic and sequential components of the interventions. Finally, at the experts’ events, which focused on issues of potential implementation, a DTS was developed which integrated insights from existing approaches used within implementation science. Although these innovations have yielded new and useful knowledge, their findings should again be interpreted with caution given their novelty.

Assessment of factors relevant to the NHS and other parties

No UK-based evidence exists which details the effectiveness of one-to-one brief behavioural interventions among MSM. Expert events demonstrated considerable demand and enthusiasm for such evidence. Expert events demonstrated the candidate intervention’s fit with existing service provision in the UK, including that provided by the NHS, private providers and other sectors such as voluntary organisations, including acceptability by staff and across varied geographical areas (e.g. urban and rural settings). Intervention delivery was explored and experts concluded that the candidate intervention could be delivered in a way that encourages access and uptake, for example in community settings or using new technologies such as digital media. No settings were immediately ruled out, but each presented challenges to operationalising the intervention. The clinical setting, with access to more specialist services such as STI treatment and psychological support if required, was deemed most appropriate. Cost and resourcing were recurrent issues and there was concern that the intervention would not be able to be delivered within existing budgets. It was clear that the health economics of the intervention would have to be included as part of its evaluation. In the light of concern that such behaviour change interventions would not be prioritised over biomedical prevention, such as post-exposure prophylaxis after sexual exposure and PrEP, it will be important to demonstrate how the two can complement each other in the future, as will evaluating the fit of the intervention if testing service delivery models change because of greater use of self-sampling and testing for HIV. Findings from the expert events highlighted a number of issues that would enhance the acceptability and feasibility of such interventions. They highlighted the importance of institutional support and clearly defined and refined intervention content and details of delivery. Experts also noted that extensive training would be required to ensure intervention fidelity.

Implications for service provision

Despite the limited evidence base and potential concerns about the transferability of knowledge, the current review does suggest the effectiveness of one-to-one, brief behavioural interventions for MSM who are HIV negative. Moreover, it provides clear direction to how best to develop and deliver such interventions. For example, exploratory stratified meta-analyses suggested face-to-face delivery of interventions may be more effective than interventions delivered via telephone or online, and those delivered immediately after testing may be associated with greater treatment effects. With regard to intervention content, the most commonly used BCTs were identified and the BCT groups from which interventions should be drawn were identified. Similarly, we concluded that an intervention should deliver a demonstrable sense of cultural competency, be multicomponent and be composed of sequential elements; it should begin with a risk assessment exercise, include a normative peer reference point and the use of discrete tools, and end with a future-facing element.

These findings and many others are directly applicable to intervention design. However, the breadth and detail of the conclusions of the meta-analysis and review findings and the expert event, and the heterogeneity of the studies analysed, mean that the direct application of the findings to clinical practice at this stage is unlikely to be possible. The use of the findings to develop a candidate intervention for use in a further clinical trial or trials, as suggested in our recommendations, would be the means of realising the benefits of this research in clinical practice. Evaluation of a candidate intervention should necessarily include operational outcomes (such as time taken, location and how it would fit within existing HIV infection testing clinical pathways, for example when many HIV infection test results are no longer delivered face to face) in addition to staff factors (training required, incentives for engagement and retention in the use of the intervention, measuring competencies to deliver it) and subsequent financial factors, operating within a budget-constrained setting. Opportunity costs will also need to be evaluated. The move, in England, of commissioning to local authorities with a specific public health mandate may foster support for more preventative interventions such as this. Many of these topics were raised during the expert events (see Chapter 6), and should inform both the candidate intervention and its evaluation.

Suggested research priorities

UK-based RCTs of brief one-to-one behaviour change interventions among HIV-negative MSM who are at high risk should be conducted. Such trials should be accompanied by comprehensive process and economic evaluations.

Prior to such trials, a comprehensive package of further intervention development work could encompass (in sequential order):

1. Further intervention development work:

   Primarily qualitative implementation research focusing on intervention development is needed to specify the content, and duration, of interventionist training, and the ways to address the barriers and facilitators specified within this report (approaches such as the ‘behaviour change wheel’) may be appropriate.

   Mixed-methods research focusing on intervention development is needed to ensure that brief individual behaviour change interventions are complementary to biomedical interventions such as PrEP.

   Primarily qualitative research focusing on intervention development is needed to ensure that brief individual behaviour change interventions can be offered within the diversity of settings and across the range of HIV infection testing technologies available.

   Experimental quantitative and complementary qualitative research focusing on intervention development is needed to validate the proposed mechanisms of behaviour change suggested by the candidate intervention.

2. The acceptability and feasibility of individual behaviour change interventions:

   Primarily qualitative methods research addressing the acceptability and feasibility of brief individual behaviour change interventions among MSM across diverse national and local settings is needed.

3. Trials of individual behaviour change interventions:

   Following a demonstration of the acceptability and feasibility of an individual behaviour change intervention that builds on the work described here, multisite trials across a range of regions in the UK should be conducted.

   Interventions must be well-designed with particular attention paid to intervention content, theoretical base and mode of delivery. Intervention content should be reflected within intervention manuals that detail the episodic sequential elements of their composition in addition to details regarding BCTs employed. Such trials must be well-designed, appropriately powered with particular attention paid to issues of retention and attrition.

   Mixed-methods process evaluations should be conducted to illuminate a range of issues including intervention fidelity, the role of confounders, mechanisms of action and issues concerned with local implementation.

   There is a need for modelling and behavioural economics studies to determine the relationship between behaviour change effect sizes and likely public health impact across a range of outcome measures.

4. Additional methodological work:

   Finally, mixed-methods research focusing on methodological development is needed regarding ways of assessing the sequence and potential synergistic and cumulative effects of intervention components.

Contributions of authors

Paul Flowers (Professor of Public Health Psychology) initiated the review and helped to design the methodology, extracted data, conducted analyses and results interpretation, led with PPI in Scotland and drafted the report.

Olivia Wu (Professor in Health Technology Assessment) helped to design the methodology, conducted the meta-analytic work and drafted sections of the report.

Karen Lorimer (Senior Research Fellow, Sociology) helped to design the methodology, led literature searches and data extraction and drafted sections of the report.

Bipasha Ahmed (Research Fellow, Health Psychology) conducted the BCT analysis, theory coding analysis and intervention fidelity analysis, conducted analyses and results interpretation, and drafted the report.

Hannah Hesselgreaves (Lecturer in Qualitative Methods for Health Research) conducted the post hoc realist review and drafted sections of the report.

Jennifer MacDonald (Research Fellow, Health Psychology) conducted the BCT analysis, theory coding analysis and intervention fidelity analysis, supported the meta-analysis, conducted analyses and results interpretation, and drafted sections of the report.

Sandi Cayless (Research Fellow, Nursing) located references, extracted data, assisted with analyses and results interpretation, and drafted sections of the report.

Sharon Hutchinson (Professor of Epidemiology and Population Health) helped to design the methodology, contributed to risk-of-bias assessment, supported analysis and results interpretation, and drafted sections of the report.

Lawrie Elliott (Professor of Nursing) helped to design the methodology, contributed to analysis and results interpretation, and drafted sections of the report.

Ann Sullivan (Consultant Physician in HIV and GUM) helped to design the methodology and drafted the report.

Dan Clutterbuck (Sexual Health and HIV Consultant) helped to design the methodology and drafted the report.

Michael Rayment (Consultant Physician in HIV and GUM) helped to design the methodology, led on PPI in London and supported the expert event.

Lisa McDaid (Programme Lead for Social Relationships and Health Improvement) initiated the review and helped to design the methodology, led and conducted the expert events and drafted the expert event sections.

Data sharing statement

Data can be obtained from the corresponding author.

Disclaimers

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

Not a randomised controlled trial/behavioural randomised controlled trial

Chesney MA, Koblin BA, Barresi PJ, Husnik MJ, Celum CL, Colfax G, et al. An individually tailored intervention for HIV prevention: baseline data from the EXPLORE study. Am J Public Health 2003;93:933–8.

Bowen A, Williams M, Daniel C, Clayton S. Internet based HIV prevention research targeting rural MSM: feasibility, acceptability, and preliminary efficacy. J Behav Med 2008;31:463–77.

Bull SS, Lloyd L, Rietmeijer C, McFarlane M. Recruitment and retention of an online sample for an HIV prevention intervention targeting men who have sex with men: the Smart Sex Quest Project. AIDS Care 2004;16:931–43.

Carrico AW, Flentje A, Gruber VA, Woods WJ, Discepola MV, Dilworth SE, et al. Community-based harm reduction substance abuse treatment with methamphetamine-using men who have sex with men. J Urban Health 2014;91:555–67.

Centers for Disease Control and Prevention. Compendium of Evidence-Based Interventions and Best Practices for HIV Prevention: Personalized Cognitive Risk Reduction Counseling (With Optional Sex Diary). 2013. URL: www.cdc.gov/hiv/prevention/research/compendium/rr/pcrrc.html (accessed February 2015).

Chiasson MA, Shaw FS, Humberstone M, Hirshfield S, Hartel D. Increased HIV disclosure three months after an online video intervention for men who have sex with men (MSM). AIDS Care 2009;21:1081–9.

Fisher HH, Patel-Larson A, Green K, Shapatava E, Uhl G, Kalayil EJ, et al. Evaluation of an HIV prevention intervention for African Americans and Hispanics: findings from the VOICES/VOCES Community-based Organization Behavioral Outcomes Project. AIDS Behav 2011;15:1691–706.

Fletcher JB, Rusow JA, Le H, Landovitz RJ, Reback CJ. High-risk sexual behavior is associated with post-exposure prophylaxis non-adherence among men who have sex with men enrolled in a combination prevention intervention. J Sex Transm Dis 2013;2013:210403.

Frasca T, Balan I, Ibitoye M, Valladares J, Dolezal C, Carballo-Diéguez A. Attitude and behavior changes among gay and bisexual men after use of rapid home HIV tests to screen sexual partners. AIDS Behav 2014;18:950–7.

Guo W, Wu Z-Y, Song A-J, Poundstone K. Impact of HIV/sexually transmitted infection testing on risky sexual behaviors among men who have sex with men in Langfang, China. Chin Med J 2013;126:1257–63.

Hosek SG, Green KR, Siberry G, Lally M, Balthazar C, Serrano PA, et al. Integrating behavioral HIV interventions into biomedical prevention trials with youth: lessons from Chicago’s project PrEPare. J HIV AIDS Soc Serv 2013;12:3–4.

Hosek SG, Lemos D, Hotton AL, Fernandez MI, Telander K, Footer D, et al. An HIV intervention tailored for black young men who have sex with men in the House Ball Community. AIDS Care 2015;27:355–62.

Huan XP, Tang WM, Babu GR, Li J, Zhang M, Liu X, et al. HIV risk reduction counseling and testing on behavior change of MSM. PLOS One 2013;8:e69740.

Jones KT, Gray P, Whiteside YO, Wang T, Bost D, Dunbar E, et al. Evaluation of an HIV prevention intervention adapted for Black men who have sex with men. Am J Public Health 2008;98:1043–50.

Kasatpibal N, Viseskul N, Srikantha W, Fongkaew W, Surapagdee N, Grimes RM. Effects of Internet-based instruction on HIV-prevention knowledge and practices among men who have sex with men. Nurs Health Sci 2014;16:514–20.

Katz D, Golden M, Hughes J, Farquhar C, Stekler J. Acceptability and Ease of Use of Home Self-testing for HIV among MSM. Proceedings of the 19th Conference on Retroviruses and Opportunistic Infections, Seattle, WA, USA, 5–8 March 2012.

Landovitz RJ, Fletcher JB, Inzhakova G, Lake JE, Shoptaw S, Reback CJ. A novel combination HIV prevention strategy: post-exposure prophylaxis with contingency management for substance abuse treatment among methamphetamine-using men who have sex with men. AIDS Patient Care STDs 2012;26:320–8.

Mikolajczak J, Kok G, Hospers HJ. Queermasters: developing a theory- and evidence-based internet HIV-prevention intervention to promote HIV-Testing among men who have sex with men (MSM). Appl Psychol 2008;57:681–97.

Mimiaga MJ, Reisner SL, Pantalone DW, O’Cleirigh C, Mayer KH, Safren SA. A pilot trial of integrated behavioral activation and sexual risk reduction counseling for HIV-uninfected men who have sex with men abusing crystal methamphetamine. AIDS Patient Care STDs 2012;26:681–93.

Operario D, Smith CD, Arnold E, Kegeles S. The Bruthas Project: evaluation of a community-based HIV prevention intervention for African American men who have sex with men and women. AIDS Educ Prev 2010;22:37–48.

Reback CJ, Grant DL, Fletcher JB, Branson CM, Shoptaw S, Bowers JR, et al. Text messaging reduces HIV risk behaviors among methamphetamine-using men who have sex with men. AIDS Behav 2012;16:1993–2002.

Rhodes SD. Hookups or health promotion? An exploratory study of a chat room-based HIV prevention intervention for men who have sex with men. AIDS Educ Prev 2004;16:315–27.

Rhodes SD, Vissman AT, Stowers J, Miller C, McCoy TP, Hergenrather KC, et al. A CBPR partnership increases HIV testing among men who have sex with men (MSM): outcome findings from a pilot test of the CyBER/testing internet intervention. Health Educ Behav 2011;38:311–20.

Shoptaw S, Klausner JD, Reback CJ, Tierney S, Stansell J, Hare CB, et al. A public health response to the methamphetamine epidemic: the implementation of contingency management to treat methamphetamine dependence. BMC Public Health 2006;6:214.

Shoptaw S, Rotheram-Fuller E, Landovitz RJ, Wang J, Moe A, Kanouse DE, Reback C. Non-occupational post exposure prophylaxis as a biobehavioral HIV-prevention intervention. AIDS Care 2008;20:376–81.

Wilkerson J, Danilenko G, Smolenski D, Myer BB, Rosser BR. The role of critical self-reflection of assumptions in an online HIV intervention for men who have sex with men. AIDS Educ Prev 2011;23.

Williams M, McCoy HV, Bowen A, Saunders L, Freeman R, Chen D. An evaluation of a brief HIV risk reduction intervention using empirically derived drug use and sexual risk indices. AIDS Behav 2001;5:31–43.

Ye S, Xiao Y, Jin C, Cassell H, Blevins M, Sun J, et al. Effectiveness of integrated HIV prevention interventions among Chinese men who have sex with men: evaluation of a 16-city public health program. PLOS One 2012;7:e50873.

Zhang H, Wu Z, Zheng Y, Wang J, Zhu J, Xu J. A pilot intervention to increase condom use and HIV testing and counseling among men who have sex with men in Anhui, China. J Acquir Immune Defic Syndr 2010;53(Suppl 1):S88–92.

Zule WA, Poulton WE, Coomes CM, Mansergh G, Charania M, Wechsberg WM, Jones HE. Results of a pilot study to reduce methamphetamine use and sexual risk behaviors among methamphetamine-using men who have sex with men (MSM) not currently in treatment. J Psychoactive Drugs 2012;44:351–8.

Majority of population not human immunodeficiency virus negative

Blas MM, Alva IE, Carcamo CP, Cabello R, Goodreau SM, Kimball AM, Kurth AE. Effect of an online video-based intervention to increase HIV testing in men who have sex with men in Peru. PLOS One 2010;5:e10448.

Bowen A, Horvath K, Williams M. A randomized control trial of Internet-delivered HIV prevention targeting rural MSM. Health Educ Res 2007;22:120–7.

Elford J, Bolding G, Sherr L. Peer education has no significant impact on HIV risk behaviours among gay men in London. AIDS 2001;15:535–8.

Hightow-Weidman LB, Pike E, Fowler B, Matthews DM, Kibe J, McCoy R, Adimora AA. HealthMpowerment.org: feasibility and acceptability of delivering an internet intervention to young Black men who have sex with men. AIDS Care 2012;24:910–20.

Hightow-Weidman LB, Smith JC, Valera E, Matthews DD, Lyons P. Keeping them in ‘STYLE’: finding, linking, and retaining young HIV-positive black and Latino men who have sex with men in care. AIDS Patient Care STDs 2011;25:37–45.

Kalichman SC, Cain D, Weinhardt L, Benotsch E, Presser K, Zweben A, et al. Experimental components analysis of brief theory-based HIV/AIDS risk reduction counseling for sexually transmitted infection patients. Health Psychol 2005;24:198–208.

Menza T, Jameson D, Hughes J, Colfax G, Shoptaw S, Golden M. Contingency management to reduce methamphetamine use and sexual risk among men who have sex with men: a randomized controlled trial. BMC Public Health 2010;10:774.

Mikolajczak J, Breukelen VG, Kok G, Hospers H. Evaluation of an online HIV-prevention intervention to promote HIV-testing among men who have sex with men: a randomised controlled trial. Neth J Psychol 2012;67:21–35.

Morgenstern J, Bux DA Jr, Parsons J, Hagman BT, Wainberg M, Irwin T. Randomized trial to reduce club drug use and HIV risk behaviors among men who have sex with men. J Consult Clin Psychol 2009;77:645–56.

Shoptaw S, Reback CJ, Peck JA, Yang X, Rotheram-Fuller E, Larkins S, et al. Behavioral treatment approaches for methamphetamine dependence and HIV-related sexual risk behaviors among urban gay and bisexual men. Drug Alcohol Depend 2005;78:125–34.

Young SD, Holloway I, Jaganath D, Rice E, Westmoreland D, Coates T. Project HOPE: online social network changes in an HIV prevention randomized controlled trial for African American and Latino men who have sex with men. Am J Public Health 2014;104:1707–12.

Population not men who have sex with men or fewer than 33% men who have sex with men

Carey MP, Senn TE, Vanable PA, Coury-Doniger P, Urban MA. Brief and intensive behavioral interventions to promote sexual risk reduction among STD clinic patients: results from a randomized controlled trial. AIDS Behav 2010;14:504–17.

Gray RH, Kigozi G, Serwadda D, Makumbi F, Watya S, Nalugoda F, et al. Male circumcision for HIV prevention in men in Rakai, Uganda: a randomised trial. Lancet 2007;369:657–66.

Johnson AP, Macgowan RJ, Eldridge GD, Morrow KM, Sosman J, Zack B, et al. Cost and threshold analysis of an HIV/STI/hepatitis prevention intervention for young men leaving prison: Project START. AIDS Behav 2013;17:2676–84.

Metcalf C, Douglas J, Malotte C, Cross H, Dillon BA, Paul SM, et al. Relative efficacy of prevention counseling with rapid and standard HIV testing: a randomized, controlled trial (RESPECT-2). Sex Transm Dis 2005;32:130–8.

Metcalf CA, Malotte CK, Douglas JM, Paul SM, Dillon BA, Cross H, et al. Efficacy of a booster counseling session 6 months after HIV testing and counseling: a randomized, controlled trial (RESPECT-2). Sex Transm Dis 2005;32:123–9.

Metsch LR, Feaster DJ, Gooden L, Schackman BR, Matheson T, Das M, et al. Effect of risk reduction counseling with rapid HIV testing on risk of acquiring sexually transmitted infections: the AWARE randomized clinical trial. JAMA 2013;310:1701–10.

Neumann MS, O’Donnell L, Doval AS, Schillinger J, Blank S, Ortiz-Rios E, et al. Effectiveness of the VOICES/VOCES sexually transmitted disease/human immunodeficiency virus prevention intervention when administered by health department staff: does it work in the ‘real world’? Sex Transm Dis 2011;38:133–9.

Warner L, Klausner JD, Rietmeijer CA, Malotte CK, O'Donnell L, Margolis AD, et al. Effect of a brief video intervention on incident infection among patients attending sexually transmitted disease clinics. PLOS Medicine 2008;5:e135.

Wolitski RJ. Relative efficacy of a multisession sexual risk reduction intervention for young men released from prisons in 4 states. Am J Public Health 2006;96:1854–61.

Not individual-level intervention

Duan Y, Zhang H, Wang J, Wei S, Yu F, She M. Community-based peer intervention to reduce HIV risk among men who have sex with men in Sichuan province, China. AIDS Educ Prev 2013;25:38–48.

Hidalgo MA, Kuhns LM, Hotton AL, Johnson AK, Mustanski B, Garofalo R. The MYPEEPS randomized controlled trial: A pilot of preliminary efficacy, feasibility, and acceptability of a group-level, HIV risk reduction intervention for young men who have sex with men. Arch Sex Behav 2015;44:475–85.

Hosek SG, Siberry G, Bell M, Lally M, Kapogiannis B, Green K, et al. The acceptability and feasibility of an HIV preexposure prophylaxis (PrEP) trial with young men who have sex with men. J Acquir Immune Defic Syndr 2013;62:447–56.

Imrie J, Stephenson JM, Cowan FM, Wanigaratne S, Billington AJ, Copas AJ, et al. A cognitive behavioural intervention to reduce sexually transmitted infections among gay men: randomised trial. BMJ 2001;322:1451–6.

Ko NY, Lee HC, Hung CC, Chang JL, Lee NY, Chang CM, et al. Effects of structural intervention on increasing condom availability and reducing risky sexual behaviours in gay bathhouse attendees. AIDS Care 2009;21:1499–507.

Lau J, Tsui H, Lau M. A pilot clustered randomized control trial evaluating the efficacy of a network-based HIV peer-education intervention targeting men who have sex with men in Hong Kong, China. AIDS Care 2013;25:812–19.

McKee MB, Picciano JF, Roffman RA, Swanson F, Kalichman SC. Marketing the ‘Sex-check’: evaluating recruitment strategies for a telephone-based HIV prevention project for gay and bisexual men. AIDS Educ Prev 2006;18:116–31.

O’Donnell L, Stueve A, Joseph HA, Flores S. Adapting the VOICES HIV behavioral intervention for Latino men who have sex with men. AIDS Behav 2014;18:767–75.

Safren SA, Thomas BE, Mayer KH, Biello KB, Mani J, Rajagandhi V, et al. A pilot RCT of an intervention to reduce HIV sexual risk and increase self-acceptance among MSM in Chennai, India. AIDS Behav 2014;18:1904–12.

Tobin K, Kuramoto S, German D, Fields E, Spikes PS, Patterson J, Latkin C. Unity in diversity: results of a randomized clinical culturally tailored pilot HIV prevention intervention trial in Baltimore, Maryland, for African American men who have sex with men. Health Educ Behav 2013;40:286–95.

Williams ML, Bowen AM, Timpson SC, Ross MW, Atkinson JS. HIV prevention and street-based male sex workers: an evaluation of brief interventions. AIDS Educ Prev 2006;18:204–15.

Rejected for ‘other’ reasons

ClincalTrials.gov. Online HIV Prevention Vignette Evaluation. 2005. URL: https://clinicaltrials.gov/ct2/show/NCT00249327 (accessed February 2015).

Grantome. Brief Introductory Therapy for Opioid Dependence. 2006. URL: http://grantome.com/grant/NIH/R01-DA013108-06 (accessed February 2015).

ClincalTrials.gov. Effects of Contingency Management for Methamphetamine Abstinence on Post-Exposure Prophylaxis for HIV in Men Who Have Sex With Men (MSM). 2009. URL: https://clinicaltrials.gov/ct2/show/NCT00856323 (accessed February 2015).

ClincalTrials.gov. Reducing HIV Risk Among Episodic Substance Using Men Who Have Sex With Men (SUMSM). 2011. URL: https://clinicaltrials.gov/ct2/show/NCT01279044 (accessed February 2015).

Bedoya C, Mimiaga M, Beauchamp G, Donnell D, Mayer K, Safren S. Predictors of HIV transmission risk behavior and seroconversion among Latino men who have sex with men in Project EXPLORE. AIDS Behav 2012;16:608–17.

Bourne C, Knight V, Guy R, Wand H, Lu H, McNulty A. Short message service reminder intervention doubles sexually transmitted infection/HIV re-testing rates among men who have sex with men. Sex Transm Infect 2011;87:229–31.

Brown EL, Wald A, Hughes JP, Morrow RA, Krantz E, Mayer K, et al. High risk of human immunodeficiency virus in men who have sex with men with herpes simplex virus type 2 in the EXPLORE study. Am J Epidemiol 2006;164:733–41.

Colfax GN, Santos GM, Das M, Santos DM, Matheson T, Gasper J, et al. Mirtazapine to reduce methamphetamine use: a randomized controlled trial. Arch General Psychiat 2011;68:1168–75.

Das M, Santos D, Matheson T, Santos GM, Chu P, Vittinghoff E, et al. Feasibility and acceptability of a phase II randomized pharmacologic intervention for methamphetamine dependence in high-risk men who have sex with men. AIDS 2010;24:991–1000.

Davidovich U. Effect Study of a Theory-Based Internet Intervention on Safe-Sex Practice: The Effect of a Theory Based Tailored Intervention Online on the Reduction of Risk Behavior for HIV Transmission Among Men Who Have Sex With Men - a Randomized, Single-Blind, Active (Waiting-List) Controlled Trial. 2003. URL: http://clinicaltrials.gov/show/NCT00378885 (accessed February 2015).

Davidovich U, de Wit JBF, Stroebe W. Using the Internet to Reduce Risk of HIV Infection in Steady Relationships: A Randomized Controlled Trial of a Tailored Intervention for Gay Men. In Davidovich E. Liaisons Dangereuses: HIV Risk Behavior Prevention in Steady Gay Relationships. Amsterdam: Roel & Uitgeefprojecten; 2006. pp. 95–122.

Dilley JW, Woods WJ, Sabatino J, Lihatsh T, Adler B, Casey S, et al. Changing sexual behavior among gay male repeat testers for HIV: a randomized, controlled trial of a single-session intervention. J Acquir Immune Defic Syndr 2002;30:177–86.

Dilley JW, Loeb L, Marson K, Chen S, Schwarcz S, Paul J, et al. Sexual compulsiveness and change in unprotected anal intercourse: unexpected results from a randomized controlled HIV counseling intervention study. J Acquir Immune Defic Syndr 2008;48:113–14.

Eaton L, Kalichman S, Kenny D, Harel O. A reanalysis of a behavioral intervention to prevent incident HIV infections: including indirect effects in modeling outcomes of Project EXPLORE. AIDS Care 2013;25:805–11.

Eaton LA. The Development and Outcomes of an Intervention to Address Serosorting among HIV Negative Men Who Have Sex With Men. Connecticut, CN: University of Connecticut; 2009.

Khosropour CM, Johnson BA, Ricca AV, Sullivan PS. Enhancing retention of an internet-based cohort study of men who have sex with men (MSM) via text messaging: randomized controlled trial. J Med Internet Res 2013;15:156–66.

Koblin B. A behavioral intervention to reduce sexual risk among African-American men who have sex with men. 2008. URL: https://clinicaltrials.gov/ct2/show/NCT00691496 (accessed February 2015).

Koblin B, Chesney M, Coates T. Effects of a behavioural intervention to reduce acquisition of HIV infection among men who have sex with men: the EXPLORE randomised controlled study. Lancet 2004;364:41–50.

Lau JTF, Lau M, Cheung A, Tsui HY. A randomized controlled study to evaluate the efficacy of an Internet-based intervention in reducing HIV risk behaviors among men who have sex with men in Hong Kong. AIDS Care. 2008;20:820–8.

Mikolajczak J. Evaluation of an online human immunodeficiency virus (HIV)-prevention intervention to promote HIV-testing among men who have sex with men. 2006. URL: www.controlled-trials.com/ISRCTN52832382?q=&filters=conditionCategory:Infections%20and%20Infestations,recruitmentCountry:Netherlands&sort=&offset=6&totalResults=51&page=1&pageSize=10&searchType=basic-search (accessed February 2015).

Morgenstern J, Irwin T, Wainberg M, Parsons J, Muench F, Bux D, et al. A randomized controlled trial of goal choice interventions for alcohol use disorders among men who have sex with men. J Consult Clin Psychol 2007;75:72–84.

Mustanski B, Garofalo R, Monahan C, Gratzer B, Andrews R. Feasibility, acceptability, and preliminary efficacy of an online HIV prevention program for diverse young men who have sex with men: The Keep It Up! Intervention. AIDS Behav 2013;17:2999–3012.

Mustanski B, Ryan D, Sanchez T, Sineath C, Macapagal K, Sullivan P. Effects of messaging about multiple biomedical and behavioral HIV prevention methods on intentions to use among us MSM: results of an experimental messaging study. AIDS Behav 2014;18:1651–60.

Outlaw AY, Naar-King S, Parsons JT, Green-Jones M, Janisse H, Secord E. Using motivational interviewing in HIV field outreach with young African American men who have sex with men: a randomized clinical trial. Am J Public Health 2010;100:S146–S51.

Picciano JF, Roffman RA, Kalichman SC, Rutledge SE, Berghuis JP. A telephone based brief intervention using motivational enhancement to facilitate HIV risk reduction among MSM: a pilot study. AIDS Behav 2001;5:251–62.

Read S, Miller L, Appleby P, Nwosu M, Reynaldo S, Lauren A, et al. Socially optimized learning in a virtual environment: reducing risky sexual behavior among men who have sex with men. Hum Commun Res 2006;32:1–34.

Rosser B, Oakes J, Konstan J, Hooper S, Horvath K, Danilenko G, et al. Reducing HIV risk behavior of men who have sex with men through persuasive computing: results of the Men’s INTernet Study-II. AIDS (London, England) 2010;24:2099–107.

Santos GM, Coffin PO, Vittinghoff E, DeMicco E, Das M, Matheson T, et al. Substance use and drinking outcomes in personalized cognitive counseling randomized trial for episodic substance-using men who have sex with men. Drug Alcohol Depend 2014;138:234–9.

Shoptaw S, Reback C, Yang X, Rotheram-Fuller E, Peck J, Larkins S. Differential Outcomes in a Randomized Trial of Behavioral Drug Therapies for Reducing Drug Use and Sexual Risk Behaviors Among Gay and Bisexual Male Methamphetamine Abusers in Los Angeles. Abstract presented at The XIV International AIDS Conference, Barcelona, Spain, 7–12 July 2002.

Tan JG, Cheng JQ, Lu ZX. Evaluation of effects of combination intervention model to men who have sex with men. Chin J Prev Med 2012;46:732–5.

Wolfers M, de Wit J, Hospers H, Richardus J, de Zwart O. Effects of a short individually tailored counselling session for HIV prevention in gay and bisexual men receiving hepatitis B vaccination. BMC Public Health 2009;9:255.

Background

The intervention is based on an evidence synthesis of trials that all focus on brief, individual, behaviour change interventions delivered in a range of settings (n = 11, 10 trials). The systematic review showed that these interventions overall had small medium-term effects, although some were significant when compared with their controls. Some of these interventions have shaped prevention policy within the USA. There have been no UK trials of these kinds of interventions.

The ‘How to stay HIV negative’ intervention is delivered in all sites that offer HIV infection testing. It is delivered within a single session lasting between 30 and 50 minutes. It is not an intervention for the ‘worried well’ or for those with multiple and complex needs. It should be thought of as one of a suite of interventions available to clients. ‘How to stay HIV negative’ is only offered to MSM who meet all of the following criteria:

1. are seeking or accept an offer of HIV infection testing

2. have received a HIV-negative test result

3. report condom-less sex with two or more partners of HIV-positive or HIV-unknown status within the last year

4. feel ready to change their risk-related behaviour.

Theory behind the intervention and proposed mechanism of action

‘How to stay HIV negative’ is not based on a single theory and draws on a range of theoretical perspectives. Its direct theoretical antecedents include the information–motivation–behavioural skill model199 and the health action process theory. 218 These both draw on Bandura’s theory of self-efficacy198 and the theory of planned behaviour. 200 Relevant constructs relating to each component are:

1. The role of information in terms of potentially increasing knowledge concerning risk-related behaviour and the role of descriptive norms in relation to the visual aid depicting other MSM’s risk-related behaviour.

2. How motivation is assessed through the disclosure of salient attitudes, norms and attributions.

3. Uses predictors from the health belief model61 such as ‘perceived susceptibility’, ‘perceived benefits and barriers’, and ‘cues to action’ (through the comparative task).

4. Motivation to change is assessed and reinforced in order to focus on the client’s readiness to change and prepare them for the subsequent move from motivational to action phases. 218

5. Utilises self-efficacy198 via building a sense of the client’s capability in reducing risk in the future and resonates with the planning and rehearsal of the self-regulatory processes required to implement action plans. 218

Research findings

‘How to stay HIV negative’ is adapted from the interventions included within our systematic review. Each of these effective interventions adopted a multicomponent approach that included many of the key elements detailed in the following five points. Commonalities in these studies suggested the importance of BCTs drawn from the problem-solving group of the BCT taxonomy, and suggested action planning, pros and cons and incompatible beliefs may be of particular use.

Key characteristics of ‘How to stay HIV negative’

Conduct ‘How to stay HIV negative’ in the context of HIV infection testing and counselling.

Complete the intervention in one 30- to 50-minute session.

‘How to stay HIV negative’ can be delivered by anyone having undergone ‘How to stay HIV negative’ training: a 2-day intensive training course. Within this course they must demonstrate a level of competency in effective delivery.

Procedures

Six steps of ‘How to stay HIV negative’

Differences/similarities from other interventions

Although the intervention does provide knowledge concerning the range and complexity of contemporary HIV risk management, this is not its primary concern. If clients are lacking basic HIV literacy, then they should be offered HIV literacy interventions instead of ‘How to stay HIV negative’.

‘How to stay HIV negative’ draws on many of the central ideas of MI.

‘How to stay HIV negative’ is not a client-centred, or client-led, approach focusing on the client’s feelings. Its focus is to proactively get clients to think about their behaviour in new ways and generate their own solutions to the problems they face.

‘How to stay HIV negative’ is not an approach based on reducing client short-term stress and anxiety. Practitioners may need supervision and support about their desire to contain client’s short-term feelings, but these are not the focus of ‘How to stay HIV negative’.

‘How to stay HIV negative’ is not focused on the practitioner providing the client with prepackaged easy solutions to their risk-related behaviour. It is about providing the client’s self-generated insights and skills for sustainable behaviour change within complex demanding and ever changing environments.