Post-AIDS

J points to an article in the June 2008 Prospect Magazine entitled "The Plague is Over, Let's Party" by Elizabeth Pisani, an epidemiologist. Pisani notes the curious state of affairs in a "post-AIDS" world. On the one hand, the advent of highly active antiretroviral treatments (HAART), particularly the discovery of the protease inhibitor, has effectively halted AIDS progression. The time to development of AIDS now appears to either have no upper bound, or is so high that we haven't reached it yet. As I like to tell people, Magic Johnson still doesn't have AIDS. If caught early enough and if treated aggressively with HAART, an HIV-positive person can remain healthy and keep his/her viral count low without seeing the infection progress to the advanced stage of autoimmune system failure. See the 2006 HIV Surveillance Report, Table 7, for a short window showing AIDS cases falling.

That AIDS is theoretically no longer a fatal disease, but rather a chronic disease, seems to be an unambiguous good, but Pisani notes how this has also led to the unexpected changes in behavior (unexpected, at least, to epidemiologists, but I'll get back to that in a second). The complexities of post-AIDS is that while AIDS mortality and even AIDS rates are plummeting, HIV rates are rising. Both the number of new HIV cases and the HIV incidence rate is on the rise in the United States and other industrialized countries. See Table 8 from the 2006 HIV Surveillance Report showing a rise in HIV cases for 2003-2006. What is going on? Well, one of these explanations is innocuous and the other is more problematic. First, the decrease in AIDS mortality and AIDS progression translates into simply more and more people each year with HIV who would've previously perished pre-HAART. AIDS kills its host, but the protease inhibitor stops the replication of the HIV virus in the host's body, thus halting the progression to AIDS and therefore causing mortality to fall. These people who previously would be dead are alive, which translates into increased HIV prevalence in the population. Because these people are also healthier now post-AIDS, they are able to have healthy sex lives, which increases the spread of HIV throughout the sexual network.

Interestingly, HIV is a rather low infectivity STD, meaning the per-act probability of infection is fairly low compared to other sexually transmitted infections. The chance of an HIV-negative female contracting HIV from an infected HIV-positive male, while having unprotected intercourse, is 0.002 - that that is, if the female had 1000 unprotected sexual encounters with her infected male partner, the cumulative risk of infection at the end would 0.86 (86%). If she always wore a condom, it'd be only .18 (18%), assuming a condom failure rate of 10 percent per exposure. Laumann, Gagnon and Micahels' The Social Organization of Sexuality: Sexual Practices in the United States, chapter 11 for more. Gonorrhea, on the other hand, has an infectivity rate of as high as 0.50, meaning it only takes a few acts of unprotected sex before the uninfected person contracts it. After 10 acts of unprotected sex with an infected partner, a female will be infected with gonorrhea. If they used a condom consistently (assuming a 10 percent failure rate), she'd have .99408 probability of infection after 100 sexual encounters.

Thus it's not surprising that whereas gonorrhea is extremely common and in both homosexual and heterosexual networks, HIV has historically been concentrated among high risk groups. High risk groups tend to have multiple partners, thus increasing the probability of transmission, and also tend to engage in unsafe practices, like unprotected sex and anal intercourse. Certain sexual practices on the African continent may be partly to blame for the greater prevalence of HIV in heterosexual networks, such as higher rates of concurrency and polygamy, as well as possibly the practice of herbally drying vaginal lubrication and create more dangerous conditions during vaginal intercourse. But in the US, despite what you may have heard, HIV is still a predominantly homosexual infection - see Table 9. This is important because, as the article notes, most of the increase in HIV now appears to be among homosexual males, as it appears homosexual males infected with HIV are now surviving with HIV, are also healthy and therefore able to engage in sex lives, have much higher numbers of partners over times than heterosexuals, are more likely to engage in anal insertive and anal receptive intercourse, and finally and maybe most important, appear to be engaging in riskier sex in response to the advent of HAART. This last point was not obvious to the epidemiologists whose models, as the article notes, were always based on the assumption that sexual behavior wouldn't change. Pisani notes:

"mathematical modelling showed that if behaviour stayed the same and infected people got treated, HIV transmission would trickle to a halt. But the models didn't factor in the possibility that you need less virus to get infected in the more dangerous act of anal sex. They also forgot that humans don't behave like mathematical models."

Or maybe these models just wrong and useless for predicting the effect of the drugs on the course of the epidemic. These models were flawed explanations of sexual behavior because they explicitly assumed human beings did not respond to incentives. That is, even though the risk of AIDS mortality was rising, epidemiologists assumed that that change in "price" wouldn't change people's behaviors. But we know this is wrong. Here's just a few papers showing this is wrong.

1. Ahituv, Hotz and Phiipson, "The Responsiveness of the Demand for Condoms to the Local Prevalence of AIDS, Journal of Human Resources, Vol. 31, No. 4 (Autumn, 1996), pp. 869-897.

2. Chesson, Dee and Aral, "AIDS Mortality May Have Contributed to the Decline in Syphilis Rates in the United States in the 1990s." Sexually Transmitted Diseases. 30(5):419-424, May 2003.

3. Cornwell and Cunningham, "Did Abortion Legalization Cause Second Generation Gonorrhea Rates to Fall? An Application of Donahue-Levitt (2001), 2007 Working Paper.

4. Mechoulan, "Risky Sexual Behavior, Testing
and New HIV Treatments," Forum for Health Economics & Policy, Volume 10, Issue 2 2007 Article 5.

5. Lakdawalla, Sood and Goldman, "HIV Breakthroughs and Risky Sexual Behavior", Quarterly Journal of Economics August 2006, Vol. 121, No. 3, Pages 1063-1102.

Those are just a few of the papers showing a causal relationship between AIDS mortality and falling STD risk and STD incidence, as well as showing a causal relationship between HAART and increased STD risk, particularly among gay US males, both HIV-positive and HIV-negative. It's not clear to what degree the rise in STD risk is due merely to more men surviving and living with HIV as healthy men, and therefore causing increased spread of HIV, and how much is actually due to men now facing a lower price of being infected taking fewer precautions at all. Mechoulan (2007) shows that when the protease inhibitor was discovered, it basically immediately led to decreased condom usage among gay males, both HIV-negative and HIV-positive, whereas Lakdawalla, Sood and Goldman focuses on just the HIV-positive population of individuals receiving HAART treatment, who later more than doubled the number of partners they had over some short period of time as a result. Either way, the causal link is very strong.

To answer J's question, I think the answer is that unless we find a cure, greater availability of the HAART regiment will lead to simultaneously lower AIDS prevalence in the population, including lower AIDS mortality, and higher HIV prevalence and HIV risk in the population. This complicates matters when thinking about the appropriate policy response to HIV epidemics.