Face-to-face interventions work best says Gus Cairns
Background to the analysis
Devising a modern behavioural intervention
It also found evidence that the best way to conduct such interventions was face-to-face, i.e. not via the internet, telephone or phone apps, and that immediately or shortly after HIV testing was an ideal “learning moment” to conduct them.
Interventions that helped participants set goals for themselves, and ones that helped them understand and restructure self-justifying or contradictory thinking, were the ones most likely to result in behaviour change. Programmes worked better if they involved participants feeling differently about themselves and their behavioural risk, rather than receiving new information. Interventions worked better if they were based explicitly on a theory of behaviour change.
However, while the analysis did find positive evidence for such interventions producing behaviour change, the writers also conclude that the HIV prevention “landscape” has changed radically in the last few years. The studies were published between 2002 and 2014 and probably gathered their data at least two years before their publication date. None included as measurable outcomes serosorting or seropositioning (i.e. basing condom use or sex role on a partner’s perceived status), pre-exposure prophylaxis (PrEP) use, decisions based on partners’ viral load, or the use of negotiated safety arrangements, even though the authors tried to find studies that measured these.
There was just one outcome measure that was sufficiently universal across the studies for the authors to make a statistical generalisation about study effectiveness: whether the interventions reduced condomless anal sex. Even though this was measured in various ways (number of condomless acts, number of condomless sex partners, whether they occurred with primary or casual partners, the HIV status or assumed status of the partner, and so on) and was measured over different time periods, the general reduction in condomless sex acts after the eleven interventions was 25%.
Background to the analysis
Compared with the comprehensive effectiveness of both PrEP and immediate HIV treatment as prevention, this may sound like a small reduction in HIV risk. But, as lead author Paul Flowers of Glasgow Caledonian University told aidsmap.com: “Behaviour change interventions boost and complement biomedical technologies rather than competing with them. Getting people on to PrEP and getting people to test regularly is what behaviour change interventions can and should be doing.”
He explained that the impetus behind this meta-analysis of behaviour-change interventions was that he and other academics were involved in writing two sets of new prevention guidelines, for the British Association for Sexual Health and HIV (BASHH) and for Health Protection Scotland.
They were confronted with the lack of an up-to-date evidence base that would indicate how and when to use such interventions to help people at risk of HIV reduce their risk. As well as much of the research being out of date, the field suffers from the fact that most of the research has been done in the US, where effective behavioural interventions are collated at a site run by the US Centers for Disease Control and Prevention (CDC).
Because of this, they were commissioned by the UK’s National Institute for Health Research (NIHR) to investigate the evidence for what worked and then devise an intervention based on that evidence, which could be 'transferable' enough to produce positive effects on other behaviours such as PrEP use or testing.
There have been a fair number of meta-analyses of behaviour change interventions published. Flowers and colleagues counted 19 published since 2000. These vary by the publication date of the studies they included, by whether they only included specific populations, and by whether they only used certain methods. Because of this, they vary enormously in size, from a systematic review of three studies using cognitive behavioural therapy (CBT) in US gay men who inject drugs, to a huge 2005 synthesis of 354 studies that were published between 1988 and 2003.
Flowers and colleagues drew tight criteria for inclusion in their meta-analysis. They only included studies in which at least 60% of participants were HIV-negative men who have sex with men; they only included studies published since 2000, as they wanted to exclude studies conducted in the days before effective antiretroviral therapy (ART) became available; and they only included studies of brief interventions, which meant five sessions or fewer. The latter was because the commissioning call from NIHR required the evidence for brief interventions.
They found eleven studies in all, which varied greatly in their methodology. Some were online and/or used novel intervention strategies: a graphic novel, a telephone intervention, or an informational video added to counselling sessions. Others were face-to-face. Six out of the eleven studies used a version of Personalised Cognitive Counselling (PCC), an adaptation of CBT for single or brief interventions, designed to be used at the time of or close to HIV testing. In addition to the six PCC-based interventions, there were two peer-delivered interventions where the basis of the intervention is not stated. Four were delivered at an HIV testing appointment (three after the negative result was given, one while waiting for the result); two were given during or after testing to both HIV-positive and negative test recipients; and four, primarily the online interventions, were given to people who recorded themselves as being HIV-negative and not at a testing appointment. The follow-up period over which the effect of the intervention was measured varied from two to ten months, with six studies using six months.
As we said above, the eleven studies produced an overall reduction in condomless sex (measured in various ways) of 25% (95% confidence interval, 9% to 38%). Six out of the eleven produced statistically significant reductions.
This agrees well with the second-largest meta-analysis of behavioural interventions, which included 102 studies and was published in 2013: this found a 27-30% reduction in condomless sex among its study recipients, which included HIV-positive and HIV-negative gay and heterosexual men. The largest meta-analysis, the 354-study one mentioned above, found a 38% reduction in condomless sex among recipients of ‘active’ prevention methods, i.e. ones with a counselling component, and 19% in ones without a counselling component such as videos and information sessions.
Flowers’ findings were therefore broadly in line with these. The problem is, however, that the behaviours that interventions may wish to change are now different. The crucial question then, is whether the skills and rethinks taught by the studies involve the kind of learning that might also encourage people to seek and adhere to PrEP, test for HIV regularly, and so on.
As clues to devising an intervention that could reinforce these behaviours as well as condom use, Flowers’ team did some sub-analyses of factors associated with significant reductions in condomless sex.
They found that two methods were associated with significant reductions. The first was using goal-setting and action-planning as part of the method, to get recipients to set behavioural targets. The second was using methods that drew attention to contradictions and justifications in participants’ thinking and thus helped them to feel more understanding, positive and capable about their ability to change. Taken together, these methods were 34% effective. Other methods such as information-giving, social support and an emphasis on threat or danger were not associated with effectiveness.
Interventions delivered via telephone or online were not effective. In contrast interventions delivered face-to-face were 34% effective. If the intervention was delivered immediately after receiving a test result, the average effectiveness was 36%; delivered longer after the result, interventions were not effective.
Although these factors were not quantified, the study found also that interventions were effective if they were delivered by professionals, but ones with clear and contemporary knowledge of the gay scene and MSM sex.
Devising a modern behavioural intervention
As a result, Flowers and colleagues drew up a specification for a suggested modern behavioural intervention for gay men.
It would include:
An initial ‘peer-oriented visual aid’ which would not just talk about health risk but also address the complexity of modern HIV risk and talk about emotions and feelings, serosorting, condomless sex, PrEP, treatment as prevention and drug and alcohol use.
This would segue into a one-to-one counselling session that would:
Focus initially on a single event seen by the client as risky or unhelpful (not using a condom, missing a PrEP dose, etc)
Refer back to the visual aid to put the client’s actual risk and the processes that led to taking a risk (emotional need, intoxication, being too busy, etc) in perspective
Invite the client to think about weighing up the pros and cons of different behaviours/strategies and how they might do things differently
Help the client draw up an action plan to help reduce risk or encourage health-seeking behaviours for the future.
Paul Flowers and colleagues are hopeful that this intervention could be the subject of a study of a behavioural intervention truly adapted to the new world of HIV prevention.
“We definitely think it is time for studies to embrace diverse outcome measures – it’s long overdue. We can’t speak for NIHR but we think it’s worth considering post-test interventions to increase frequent testing among those who need it and to encourage the consideration of PrEP. There will be challenges as HIV testing diversifies into self-testing and so on and we’ve written another paper on that.
But an HIV test represents a ‘teachable moment’ for some people and we want to help them sustain positive changes they might be motivated to make at that moment.”
Flowers P et al. 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. Health Technology Assessment 21(5): DOI 10.3310/hta21050. See full report here.