The reversal of the global gag rule by the Obama administration during the past few weeks means a new dawn for talking about HIV/AIDS prevention activities. It is therefore great timing that a new working paper on the effectiveness of HIV/AIDS prevention health education has been released.

In an update to an much earlier paper on the same topic, Pascaline Dupas from UCLA economics presents the results of a randomized experiment in Kenya that investigates the effectiveness of health prevention education efforts among primary school aged Kenyan teenagers.

Her study actually investigates the effectiveness of two somewhat different prevention interventions. The first, which is basically the strategy adopted by the Kenyan Ministry of Health was based on a strategy of teaching training (training of trainers type set up) focusing on abstinence message (all sex is risky, so avoid it). The second, was a school based, trainer to student type intervention that focused instead on the relative risk of different types of sexual behavior (some sex is riskier than others, so avoid risker sex). She finds that students in schools treated with the latter intervention have a 28% reduced rate of pregnancy (a proxy for unprotected, and hence riskier, sex). She also finds that girls in the treated schools report less sex with older, riskier sexual partners (older partners are more likely to be infected with HIV and are more likely to be wealthier and thus can demand more unprotected sex), and more sex with same age partners (this is also confirmed by younger boys who report more sexual activity). She finds that the government strategy was ineffective and thus prevention efforts may be more effective at reducing the spread of HIV (and potentially unwanted pregnancies) if they focus instead of harm reduction rather than abstinence only messages.

First off, I think this study is great, and always have, as it does shed a lot of light on the effectiveness of health prevention activities in general. However, I think her conclusions may go a bit far. She compares two different messages that are not just different but that are delivered in very different delivery settings. It could be that the delivery mechanism of the teaching training model is simply ineffective and that one needs to think about going straight to the students. We don’t know. Delivery mechanisms are key.

Second, I think she dismisses too much the potential side effects of promoting additional sexual activity among same age partners. Yes, it is true, for the time being this is lower risk, but at what point does this have other effects? The boys in the relative risk treatment group reported a 50% increase in sexual activity and an increase in the number of sexual partners. She also acknowledges in the paper that what happens to the older men who are no longer engaging in as much sex with younger girls? Are they engaging more in sexual activity with commercial sex workers? Are they targeting even younger girls? What are the long-term impacts of such shifts? These spillovers should be considered.

Picky points aside, I think the findings of this study are very significant as it does suggest, if nothing else, that teenagers may be more responsive to prevention messages that involve harm reduction rather than abstinence messages. It also goes to show how important well designed research can be in advancing the evidence base for what we do in global health.

Share on Facebook

2 Responses to “Moving forward with HIV prevention efforts”

  1. guy says:

    Hey Karen,

    I can’t remember how I stumbled across your blog, but I’m impressed with the high info:noise ratio. A couple of quick notes: I think you mean “presents” not “prevents” in the second paragraph above. And the link to the smoking cessation post a couple of posts ago sends you through a Harvard sign-in page rather than direct to the NEJM.

    Oh, and slightly more constructively, are you following the CGD’s HIV/AIDS Monitor progrom on HIV and HRH (

  2. tiantian says:

    Very nice post. But I think Dupas did make an effort to emphasize that the differences between the two delivery mechanisms might have created noise in the data in the conclusion. This problem of noise introduced by means of delivery seems to be rather generic than limited to the Dupas study. Most RCTs of this type collaborate with well-known NGOs to carry out the evaluation.Then the authors tend to compare the effects of some measure (A) delivered by the NGO to those of the pre-existing more common measure (B) (RR VS TT in Dupas’s study) and conclude A is better than B. The alternative conclusion could be that those NGOs are simply more passionate about what they do and are better prevention delivery agents than the government workers. I would love to hear your thoughts on what could be done to minimize the noise introduced by such difference in human capital and delivery efficacy. Thanks.

Leave a Reply

Analytics Plugin created by Web Hosting