The role of behavior change programs in HIV prevention strategies: Lack of evidence or lack of impact?


This page in:

With funds devoted to HIV/AIDS declining, there has not been a better time, at least in the past decade or so, to optimize the use of the limited resources between treatment and prevention. In the wake of HPTN 052, which found that treating HIV-infected partners in mostly heterosexual HIV-discordant couples can practically eliminate HIV transmission to the HIV-negative partner, the debate over whether and how much of the funds should be devoted to prevention has become even more salient (I wrote about this earlier here). If the study findings are replicated elsewhere for other populations, it is clear that treating people living with HIV is also prevention to the extent that they’re sexually active with HIV-negative individuals.

However, it is also clear that governments, even richer ones let alone those in Africa, are not going to adopt early treatment regimes anytime soon. Without going into detail, there are good reasons for this: the most infectious people are those in the late stages of the disease, i.e. eligible for treatment anyway; we don’t know the long-term effects of treating people with ARVs; starting treatment earlier does not provide substantial benefits to the treated person (he/she would be taking medication for his/her sexual partners); and practicality of ‘test and treat’ for the entire population. While these get sorted out, we still need to try to prevent people from becoming infected with HIV.

What do we know so far on what works in HIV prevention? We have more or less figured out prevention of transmission from mother to child. This does need to become a regular part of prenatal care, but it won’t affect sexual transmission rates. We now also have good evidence that voluntary medical male circumcision works (longer term studies have not yet found any significant disinhibition effects, meaning the protective effect is persisting over the longer term), but many countries need to roll it out to reach the masses. There are both supply and demand-side issues that need to be tackled and I will be writing more about this in the upcoming weeks. Vaccine development continues, so does research on the effectiveness of pre-exposure prophylaxis (remember the microbicidal gel trials successful in Phase 2 but failing in Phase 3?), neither of which have yet been proven to be effective. Treating other STIs, such as HSV-2, does not seem to be effective either – even though there are biological reasons to think it should be.

Moving away from the biomedical prevention methods to behavior change programs, the picture is dismal. A recent review by Nancy Padian and colleagues finds not a single intervention showing a decline in HIV incidence as a result of a behavior change intervention. The authors point to the fact that lack of statistical power to detect meaningful declines may have been a factor in these studies. Regardless, this puts the funding agencies in a bind: what to do with interventions that try to promote abstinence, condom use, being faithful – the ABC approach that was touted heavily until recently? We have no evidence that they work. Yes, it is possible that there are some approaches that have not been tried which might work, but should we keep banging our heads trying to make ABC work? Or, is it time to start looking elsewhere?

To the proponents of behavior change interventions, there is a mirage: there has been some decline recently in HIV incidence in Sub-Saharan Africa. DHS data are also showing some signs of (self-reported) behavior change, especially among young people. So, the thinking goes, these two things must be linked. Perhaps, but that does not mean that the causality went from behavior change programs to behavior change to decline in HIV incidence. It could also be that it went from high HIV incidence to behavior change to lower HIV incidence – with behavior change programs making implementers feel good but doing nothing to prevent new infections.

That behavior change might be happening in response to the intensity of the HIV epidemic and the competing disease risk gets support from a paper by Emily Oster just published in the Journal of Health Economics. Using an IV strategy, she finds that increased HIV rates lead to lower rates of unprotected sex with multiple partners and that this response varies by non-HIV life expectancy. She finds no evidence that lack of knowledge about HIV is a factor in behavior change.

Given that biomedical solutions are still distant, there is no doubt that we’ll continue to try other prevention strategies – including behavior change interventions. However, we owe it to ourselves to (a) not keep trying the same things; and (b) design evaluations properly (i.e. pointing to the channels rather than combination packages) and with sufficient power to either show effects or convince us of that the null effect is real. Funding agencies should apply extra scrutiny to proposals with behavior change interventions.

Of course, our behavior responds to all sorts of other things, such as income, aspirations, competing disease risks, equity, etc. These are domains that have not been traditionally in the HIV prevention field. Perhaps they deserve a harder look, at least a place on the table, and I plan to talk more about the potential of these ‘structural’ factors in the upcoming weeks.



Berk Ozler

Lead Economist, Development Research Group, World Bank

Join the Conversation

Berk Ozler
February 03, 2012

Hi Antonio,

Good question. The most relevant, of course, would be seeing people sick, number of funerals you attend or see people attend of young adults, etc. Of course, early in the epidemic people would not know what the deaths were due to, but this is unlikely to be the case almost 30 years after the first case.

That said, information campaigns might be useful. However, such campaigns usually give average, rather than disaggregated numbers which may be less useful for any given individual. You should read Dupas (2011) who provided informatuon on prevalence rates disaggregated by age and sex to teenagers and saw their sexual behavior change. People usually do overestimate (in surveys) the prevalence in their own area, as well as their own likelihood of being HIV positive. But, with increased testing, this is also likely to be changing.

Jason Kerwin
February 02, 2012

A minor quibble: there are actually two peaks of infectiousness in the course of HIV. One occurs, as you say, late in the disease when people develop what we typically call AIDS. But the other occurs early, just 4-6 weeks after initial infection. This is unlikely to be relevant for "test-and-treat"-style approaches since that initial peak in infectiousness happens before the immune system has produced large numbers of antibodies and hence is in the "window period" where people still don't test positive for the virus. However, it would be an important factor for the potential effectiveness of a more radical approach to tackling the virus, which is to try to treat *everyone*, irrespective of serostatus - or at least the entirety of certain high-risk groups.

Gregg Gonsalves
February 03, 2012

Hi, Berk. Great piece. I think part of the weakness in the data on behavioral interventions is due to a historical accident; it was psychologists who appeared early on in the epidemic to shape prevention programs and many of the most common interventions are based on one-on-one or group counseling. This is about as far from a public health approach as you can get. It's like sitting down with drivers of automobiles on-by-one and talking with them about road safety.
I also think that the reaction to questions about behavioral research effectiveness get deflected in comments about "other kinds of evidence" being just as important as that from trials with hard endpoints. Often proponents of a greater investment in behavioral interventions, such as economists Mead Over and Bill Easterly, simply gloss over the problems in the field, in their own attempts to push back against further roll-out of ART, which both men do not see as a public good to be supported by the state. It would be good to know if HPTN 052 changed their minds.
What we do need is new ways of thinking about HIV prevention and a rigorous adherence to evidentiary standards--I am pleased this is the case you are making as well.
It's vital we figure out how to reduce new HIV infections, but we'll need a minor revolution in the field to move us forward.

February 03, 2012

In relation to Oster's paper, mentioned in your post, I am wondering how people know about increased infection rates - are these perceived or due to campaingns of information? Thank you for this interesting blog.

Matt Greenall
February 03, 2012

"I doubt how many people would consent to it"

There have indeed been many examples of presumptive STI treatment programmes for sex workers; they are still going on in many countries. Effectively, coercion is often a feature since those who fail to comply face penalties, or have to work in even riskier, more hidden environments to avoid detection. There are other reasons to worry about the concept of presumptive treatment of highly stigmatized groups such as sex workers. If clients are led to believe that condoms are no longer necessary, because all sex workers are on ART, what impact might that have on sex workers ability to negotiate how they have sex and whether condoms are used?

Berk Ozler
February 02, 2012

Hi Jason, thanks for the comment. You are absolutely correct that there are indeed two acute infection periods: just after being infected and later when people are actually becoming sick. As we cannot actually test for acute early infections, I left this more 'techy' point out of the post. Perhaps, we now have a very discerning readership who expect more nuanced detail rather than less, which, of course, is great.

Presumptive treatment of everyone in a sub-group, say commercial sex workers, is an interesting idea that is unlikely to happen until the price of treatment goes down dramatically, and we know there are no harmful effects. (In fact, I think this was done among sex workers in some countries with respect to the treatment of other STIs.) Even then, I doubt how many people would consent to it...