Syndicate content

Ending Stigma: Lessons from the AIDS Epidemic -- Guest post by Laura Derksen

This is number 9 in our series of posts by students on the job market this year.

As HIV continues to spread in sub-Saharan Africa, so does stigma. Many go to great lengths to hide their HIV status, get tested at clinics far from home to avoid being seen, and put off medical care until it's much too late. This has devastating effects. While life-saving medication is now provided for free in most parts of Southern Africa, there are still over one million AIDS deaths every year. Reluctance to seek treatment also has a negative externality. Antiretroviral drugs slow the spread of HIV dramatically, but a “treatment for prevention” strategy won’t work if people don’t seek treatment.

What causes stigma? What can we do about it?

Many existing policies, such as door-to-door testing and private HIV counselling, help people cope with stigma by improving privacy. Papers by Rebecca Thornton  and Muthoni Ngatia find that monetary incentives give test seekers plausible deniability: “I couldn’t possibly have HIV, I only got tested for the money”, but that stigma still plays a role in the decision to test. All of these policies seem to be extremely effective at increasing HIV testing. However, people will still face stigma if they’re caught seeking treatment, and it might be hard to provide enough privacy for people to seek medication every month or two without being found out.
Another option would be to identify the root causes of stigma, and to reduce stigma itself. This should lead to a permanent increase in both HIV testing and treatment. In my Job Market Paper, Love in the Time of HIV: Theory and Evidence on Social Stigma and Health Seeking Behavior, I show that destigmatizing HIV is possible, and increases HIV testing, making it a potentially important tool in efforts to combat the spread of disease.

I use an information experiment to detect and reduce one obvious type of HIV stigma: statistical discrimination between sexual partners.  The decision to get an HIV test is often observable. You might run into friends or family at the health clinic, and in small communities, word gets around. Because HIV testing is the first step toward starting treatment, those who are likely to be infected have the most to gain from testing. So, HIV testing is a signal of risk. If people are afraid of contracting HIV, they might reject partners who’ve been seen hanging around the clinic, in the HIV testing queue, or taking antiretroviral drugs.

This logic ignores a key fact, a fact which is not widely known. Someone who’s been tested and treated for HIV is actually a pretty safe sexual partner. Recent medical research shows that antiretrovirals have a huge public benefit: they reduce per-act HIV transmission by approximately 96% (Cohen et. al., 2011). This should reverse the signal of an HIV test. In a region with high HIV prevalence, a person who is tested and treated is actually safer than a person who’s never been tested.

What happens when communities learn that the treatment for HIV prevents transmission?

In order to find out, I ran a randomized information experiment in 122 villages in Malawi. A community health meeting took place in every village, but in the control group, only information on the (well-known) private benefits of antiretroviral drugs was discussed: the drugs save lives and reverse the symptoms of AIDS. In the treatment group, villages received information on both the private and public benefits of the drugs. In particular, they learned that the drugs reduce HIV transmission by 96%.

Health educators used an infographic to explain the effect of antiretroviral drugs on HIV transmission risk.

I obtained administrative data on HIV testing rates from registers at all health facilities in the study area, and digitized data covering the three months before and after the intervention. I was able to observe the treatment status of an HIV testing client by looking at the home village recorded in the register. I also conducted a survey five months after the intervention to measure beliefs about the effect of antiretrovirals, and attitudes towards those taking the drugs.

The intervention was very successful in shifting beliefs about the public benefit of antiretroviral drugs, as measured by five different survey measures. The private benefits of the drugs, on the other hand, were well known in both treatment and control villages. Information on the public benefit of antiretroviral drugs increased the HIV testing rate. The magnitude of the effect is large, significant and almost immediate. The intervention caused a 60% increase in HIV testing among the target population (ages 15 to 49, non-pregnant) within three months of the intervention, from a base annual testing rate of 6.5% in the control group. The effect size is similar and significant for both men and women.

Various potential mechanisms come to mind, but my results point to a reduction in stigma between sexual partners as the first order explanation. Survey respondents in treatment villages are more likely prefer a sexual partner who is taking antiretroviral drugs to one who has never been tested, and more likely to say that a person taking the drugs might find a new partner. I also see a strong relationship between community-level beliefs about the public benefit of drugs and the decision to test nearby (a person who seeks an HIV test near his home village is more likely to be seen by friends and neighbors). Finally, a person’s HIV testing decision seems to depend on his perception of the community’s beliefs about antiretroviral drugs, and not his own beliefs. If altruism drove the results, one would expect the opposite.

A reduction in stigma should have long lasting effects on both HIV testing and treatment, and ultimately slow the spread of HIV in the population. As more data comes in, I will be able to eventually test the long-run predictions empirically.
Risk compensation is a natural concern, and the survey data does show a small increase in unprotected sex among those who’ve been tested. However, couples would have to increase their risk taking 25-fold to offset the preventative effect of antiretroviral drugs, so, at least in terms of new HIV infections, the benefit of the intervention seems to outweigh the cost.

My experiment doesn’t tell us what the optimal information intervention should look like. I would argue that because HIV test seekers face stigma from members of their community, a good policy intervention should involve community outreach as opposed to information at health clinics. We can’t rely on information sharing to shift beliefs over time, because stigma could make people reluctant to discuss HIV and its treatment. What’s more, an HIV-positive person who expounds the benefits of antiretroviral drugs may be suspected of having an ulterior motive. In the absence of an information campaign, social stigma allows incorrect beliefs to persist.

My results suggest that informed communities make different and often better choices.  Health campaigns can sometimes be fear-based, and aim to scare the public into safer sex practices by hiding key facts about HIV transmission. This may do more harm than good; misinformation reinforces stigma, which keeps HIV testing levels low. Efforts to reduce stigma, in the context of HIV and more generally, often focus on behavior change directly. When stigma is based on statistical discrimination by a misinformed public, providing new, precise information may be a more effective strategy.

Laura Derksen is a PhD Candidate at the London School of Economics


This is really interesting work. One question: That infographic makes having sex with an HIV+ person NOT on ART look REALLY scary. But my understanding is that the risk of transmission per encounter is extremely low. (Please correct me if I've got that wrong.) Is there a risk that such an information campaign could INCREASE stigma for people who aren't tested (but suspected)?

Submitted by Berk Ozler on
Hi David,

I thought about the same thing when I was first reading this post. However, I think this is still the more sensible way to go. If you tried to first fill in 0.007 of an HIV positive person not on ARVs (say, a guess on tranmission probability per sex act), then fill in 4% of that for a person on ARVs, ther graph would not really work. You could guess the chances of transmission over a year having unprotected sex with the same person by guessing a frequency of sex, but that would still get you to, say, filling in half the person. So, I think of the image in the infograph as a married couple who will, in the long run, almost certainly turn from HIV-discordant to HIV-concordant without ARVs. I am sure Laura has put in a lot of thought on the design and can say more...

Submitted by Laura Derksen on

Thanks! Yes, you're right, the actual transmission rate is pretty low (somewhere between 0.1% to 1% per act depending on various factors). I didn't include anything about the absolute transmission rate in the information campaign; the stick figures in the infographic were simply described as being "more contagious" or "less contagious". Beliefs about the absolute rate were the same in treatment and control groups post-intervention. This might be explained by the fact that people greatly overestimate the risk of transmission, so even if the infographic did make HIV+ partners look really scary, it didn't shift their prior beliefs.

Submitted by Laura Derksen on

Also, as Berk pointed out, over the course of a marriage or years of unprotected sex in a high-prevalence region the probability of transmission starts to approach 1...

Good points all around, and instructive that beliefs on absolute rates were the same post-intervention. I've always wondered about the fact that everyone seems to overestimate their probability of infection from a single encounter: No one seems excited to design a public health intervention to correct THAT info. I guess it would mean more spread but potentially higher utility (from encounters).

Submitted by Laura Derksen on

Actually, Jason Kerwin finds that extreme overestimates of HIV risk can lead to even more risk taking... "rational fatalism"

Submitted by Arianna on

Only few of those who test start ARVs right away...

Great paper! Right now, only a relatively small percentage of men and non-pregnant women who test positive actually start ARVs right away though, and even in those cases it usually takes between 6 and 14 weeks to start ARVs after testing. Were people thinking that testing=starting ARVs right away? Otherwise, the magnitude of the effects seems too large to be explained by the stigma between sexual partners story. But very interesting.

Submitted by Laura Derksen on

Thanks! It does usually take a few weeks or months to start ART after testing positive, and that's actually the reason I don't use take-up as a short-run outcome. I don't think this is inconsistent with my stigma story. The idea is that some people would really like to get tested and treated for their own benefit, but don't do it because if they are seen getting tested -- or even worse, treated -- they will be permanently blacklisted in the marriage/dating market. If people see this attitude starting to change (the survey suggests that it did), testing might increase pretty quickly, even if the dates come later. I'm really curious about what will happen over the longer run. Let me know if I've misunderstood your comment!