Are randomised, controlled trials appropriate in HIV prevention?

There has been considerable debate amongst health promotion professionals and researchers over what constitutes evidence of effectiveness in health promotion. One of the chief disagreements has arisen over the question of the research methodology that should be used to measure effectiveness. Oakley and colleagues have argued that the gold standard measure for effectiveness is a randomised controlled trial (RCT). However, this view has been criticised as a relic of biomedical research which isn't appropriate for the assessment of behavioural interventions (Fraser).

An obvious parallel can be drawn with HIV treatment research. In a treatment environment where there is just one new drug available and little previous treatment experience, it's easy to recruit and randomise people and judge the effects of the drug, providing you have an agreed endpoint, or outcome measure. But when you have disagreements about the endpoint's validity, even this form of simple study becomes difficult to interpret.

The situation becomes much more difficult when the standard of care has improved greatly, and many drugs can now be combined to treat a disease. It becomes a lot more difficult to find people who have never been exposed to anti–HIV drugs. It's unethical to give people just one drug. It must be added to the standard of care, and the study can't last very long because of the medical and commercial urgency of proving that new drugs work. So we don't know whether the drugs produce long–term changes, or whether the regimens will need to be changed frequently.

Substitute HIV prevention intervention for drugs in this scenario and it's easy to understand why randomised controlled trials are inappropriate in some circumstances (although the circumstances in which they may be appropriate are discussed in Outcome measures below).

Researchers are adapting to the new virology and the new treatment environment by designing new sorts of trials and being much more flexible about the kinds of evidence that will be needed to license drugs and develop effective treatment strategies. There is no reason why the HIV prevention field needs to ignore these lessons or to adopt a methodology simply because it is a medical orthodoxy. In reality, one of the big attractions of the randomised controlled trial in a conservative culture may be its capacity to slow down the diffusion of innovations and reduce spending on experimental approaches. It contributes to the rationing of scarce resources, but it may not be the best way of determining how prevention money should be spent.

Stifling innovation

There is also a danger that researchers and practitioners will be unwilling to draw inferences from study results, and will persist in advocating conservative solutions to prevention problems rather than following paths which are plausible. This is exactly what happened in the field of HIV treatment, where some researchers and purchasing authorities have insisted on obtaining evidence from one or several clinical endpoint trials in order to validate particular treatment strategies, rather than being prepared to extrapolate from existing evidence as to which treatment strategies might be biologically plausible. Yet the history of HIV treatment shows that since the advent of antiretroviral therapy, it is those who have advocated biologically plausible therapy who have tended to get things right. For instance, triple combination therapy was advocated on the basis of biological plausibility over two years before clinical trials proved that it was superior to dual therapy or monotherapy.

Inadequate service provision

This leads us to another danger of the evidence–based approach as it is currently being pioneered. This is the possibility that health authorities will only choose to fund the approaches which are validated by randomised controlled trials or the effectiveness reviews cited above. Lucas has argued that the approaches cited as most effective by such reviews should only be seen as the minimum contents of a package of prevention measures. They should not become a prescription for purchasing.

Nor do such reviews give any guidance on the allocation of resources between different types of programme. This isn't possible without some concomitant analysis of outcome measures. Whilst a programme which increases people's ability to talk about safer sex may have been proven by an RCT, how effective is the intervention at changing infection rates? Ultimately, resource allocation must be judged according to the capacity of different elements of a prevention programme to exert the best possible effect on new infections. At the moment we have no data which can allow us to judge which measures will have the most effect on this outcome. We can only make informed guesses.

Currently available effectiveness reviews have another weakness. They don't tell commissioners or providers what sort of agency is best suited to carry out particular types of interventions. Of course, you can make a guess, but it helps if you understand the models of behaviour change which underlie a particular intervention package. For example, a community mobilisation approach proposed by a local health promotion agency needs to be considered as an example of a social diffusion intervention. What does social diffusion theory in general tell us about the likely nature of the agents best placed to bring about change in a community or group? Is a local health promotion agency best placed to develop this type of programme effectively? Is the community mobilisation approach likely to be the best method of allocating scarce resources, or might other methods of diffusion have a higher contact rate (and a speedier diffusion process)?

In summary, the current limitations of effectiveness reviews suggest the need for better efforts to define the aims and objectives of prevention efforts. A more outcome–focussed set of aims and objectives would allow purchasers and providers to think through the implications of different interventions with more attention to establishing realistic measures of efficacy.

For example, take a town with a large gay population which has a very limited range of prevention activities. This town is not close to any other major gay centre, so it's realistic to use several inter–related outcome measures to measure the overall effectiveness of prevention activities. These would be gonorrhoea reports as short–term indicator, and over the longer–term, HIV incidence in a cohort recruited from the local gay population.

The aim of local prevention activities would be to effect a reduction in both these indicators within a given budget. But what should the package of local prevention activities consist of? To define this package a purchaser could take one of several approaches:

• Fund several different outcomes: number of condoms distributed; number of gay men attending counselling sessions for partners having unprotected sex; number of workshop sessions provided for men having risky sex; number of peer educators/volunteers trained according to the overall aims of the local HIV prevention strategy.

Evaluation of projects funded under this strategy would follow two tracks: process investigation of service delivery, and self–reported behavioural and cognitive changes amongst a sample of gay men recruited from users of these services. Waiting list control groups for some of these interventions could also be required as additional back–up.

This strategy takes account of several different models of the ways in which people change their behaviour, and seeks to deploy these in a mesh of prevention activities to be targeted at the local gay community

• However, it doesn't help purchasers judge where they will get the most value for money. What is the factor leading to most new infections in the locality? Is it unprotected sex in relationships? Is it unprotected sex in saunas and sex clubs? Or are gay men reporting a large number of unprotected contacts whilst using recreational drugs? Whilst these factors can be extrapolated from the research literature, local action research and ethnography will be a vital precursor to funding a cost–effective local strategy. The other vital element during this phase of commissioning is the maintenance of existing levels of service.

Local needs assessment need not reproduce national behavioural data, but it can often give an idea of particular local factors which are facilitating new infections – or more likely, local opportunities for exploiting social networks. For example, a local needs assessment need not investigate what sort of risk–taking activity is going on amongst local men, but it will need to map local gay networks.