Myth of the Russian AIDS Apocalypse

Many analysts of an unreasonably gloomy (or Russophobic) bent delight in raising the specter of an AIDS mortality crisis sometime in the next few years, indulging in fantasies of Russia as a dying, blighted wasteland populated by nihilistic, promiscuous druggies. In 2002 Vadim Pokrovsky, well known government anti-AIDS crusader, predicted the number of infected would rise to 3-5mn in “a few years“; by 2005, “we could be talking about five-million being infected, and these are realistic, even conservative figures” and tens of thousands would be dying by 2007. Prominent doomer demographer Nick Eberstadt modeled a 10% HIV prevalence rate by 2025 under a “severe scenario” and 2% under the lowest “mild scenario”. The World Bank (Ruhl et al) predicted a range of 3.21% (3.2mn cases) to 7.26% (5.3mn cases) by 2010, which is still much lower than US governmental National Intelligence Council estimates which project truly apocalyptic figures from 7.0% (5mn cases) to 11.2% (8mn cases). In other words the demographic, economic and geopolitical collapse of Russia is imminent.

To be fair a few years back I too was very concerned about these trends and apparent apathy of the Russian government. Then I became a bit more relaxed as the exponential tidal wave of new infections promised by Pokrovsky and the other prophets of doom failed to materialize and overall anti-AIDS spending ratcheted up to 300mn $ in 2006 and more than 500mn $ in 2007. Furthermore, a bit more research showed that these scenarios of African-style STD oblivion simply don’t stack up neither in theory nor in practice.

In 2007 Pokrovsky believed that there were “as many as 1.3mn” people infected with AIDS, very far from the multi-million rates he was predicting just five years ago, and not a catastrophic increase from “expert estimates” of 0.8mn in 2000. Russian government data shows that the percentage of pregnant women testing HIV positive reached a plateau in 2002 and tended down ever since. The models used by Eberstadt and co. are themselves critically flawed, because according to the international research program Knowledge for Action in HIV/AIDS in Russia, they assume that “the epidemic would be essentially heterosexual in nature and follow trends observed in sub-Saharan Africa”, which is “not borne out by current surveillance data from Russia”. (They are also not borne out by the slightest acquaintance with comparative development and sociology. Few Russians are malnourished and hence have greater immune resistance, their medical equipment tends to be sterilized and it is socially unacceptable for them to have many partners or engage in anal sex; all this cannot be said for sub-Saharan Africans).

I will now look justify my bold claims with more details. Firstly, let’s go through the 2008 Russian Progress Report on AIDS to the UN to get the key statistics.

As of year-end 2007, some 416,114 people were registered as HIV-positive in Russia, which means HIV prevalence amongst the total population stood at 0.3% (in practice, twice or thrice more because many cases go undetected). Infection rates were much more “concentrated” amongst high-risk groups like injecting drug users (IDUs) and to a much lesser extent amongst sex workers, male homosexuals and prisoners. There are great variations in prevalence between regions, from already concentrated epidemics amongst IDUs in some Siberian cities to very low rates in rural and conservative regions.

Although the first case of HIV was uncovered back in 1987, the disease remained relatively dormant for the first decade. However, it exploded from around 1996, when the major vector of infection shifted to the usage of non-sterile instruments for intravenous drug injection (83% of registered cases). The epidemic at this stage was concentrated amongst young men. In recent years the role of heterosexual sex began creeping back up, indicating that it was spreading from IDUs to the general population and women (e.g. men getting infected from drug-injecting sex workers and then spreading it to their usual partners). By 2007 some 34% of infections were through heterosexual sex (including 63% of women, whose share of new cases is increasing).

The government has vastly ramped up spending on AIDS prevention from a paltry 33mn $ in 2005, to 254mn $ in 2006 and 445mn $ in 2007. The latter two years saw the implementation of the “The Project for the Prevention of HIV, Hepatitis B and C and Diagnosis and Treatment of HIV”, which involved spending more money on funding anti-retroviral therapy (ART) and informational campaigns aimed at prevention. By 2007, the percentage of registered people in advanced stages of the infection receiving ART rose to 93%; some 87% of HIV positive pregnant women were getting ART to limit the chances of spreading the infection to their babies; 39% of sex workers, 17% of practicing homosexuals and 24% of IDU’s got covered by HIV prevention programs.

82% of schools were providing some form of HIV/AIDS education by 2007, so it is no longer correct to say Russian society ignores this problem. Nonetheless, people remain relatively uninformed so far with only 34% of young people displaying accurate knowledge about AIDS untainted by myths. 7% of under-15 year olds had sex and 15% of 15-49 year olds had sex with two or more partners in the last year. This is not particularly promiscuous by international standards (in the US, 14% of under-15 year olds are no longer virgins, while 14% of 20-59 year olds had sex with two or more partners in the last year). According to the survey, amongst the high risk groups 92% of sex workers, 60% of male homosexuals and 37% of IDUs used condoms the last time they had sex.

Ever since the first case was uncovered in 1987, a huge surveillance program was set up to monitor the disease in the last years of the Soviet Union. It tests some 20-25mn  people per year, which makes it probably one of the most comprehensive systems in the world. After remaining flat and very low until 1998, it spiked in 2001 and has since settled down to a lower level.

Over period to end 2007, some 21,959 HIV positive people died; in 2007 the number of deaths and new cases fell due to wider availability of ART.

Possibly the most important graph is above, which shows HIV prevalence amongst pregnant women (and as such is a good proxy for infection rates amongst the low-risk and overall population). As we can see the figures took off after the IDU explosion in 1996 and accelerated sharply, before hitting a slowly declining plateau by 2002. This is important because it implies that there is a mostly linear relationship between infection rates amongst high risk groups like IDUs and the general population, and that purely sexual transmission does not have a critical momentum of its own. (PS – the reason the number of HIV tested pregnant women fell during the 1990’s was because of the post-Soviet fertility collapse).

Which is not to say that there do not exist ominous signs in the other direction elsewhere. From 2002 to 2007, AIDS cases due to heterosexual contact increased from 18% to 34% of cases and is now the major vector for transmission to women (IDU remains by far the biggest for men). There is a rising incidence of new HIV cases amongst sexually active homosexuals (1.1% HIV positive in 2007 amongst those who sought testing, against 0.5% in 2006). However, there was a decrease in new HIV infections amongst registered IDUs (6% in 2001, 2% in 2007 were new cases of HIV).

My impression is that the epidemic is now more or less controlled amongst the highest risk group, IDUs. There is also evidence that it is not seeping into the general low-risk population. However, the situation is getting worse amongst those who live in the nether regions between these two worlds, by which I mean the partners of men sleeping with drug-injecting sex workers and the girlfriends of IDUs. This would imply that the HIV prevalence rate will continue going up for perhaps another decade and will peak at no more than 1% of the population. This is realistic because as mentioned the material conditions for the mass spread of AIDS simply do not exist and in any case the government is now making lively efforts to provide ART (and thus reduce chances of transmission, i.e. rate of diffusion) and educate people on the topic. While researching this post, I even found a (mildly erotic) children’s poetry book on AIDS from the town of Kaluga!

Now that we know a bit more of the context, let us now turn to the Knowledge for Action in HIV/AIDS in Russia report. This excellent publication has a whole chapter devoted to “modeling the HIV epidemic in Russia”, published in October 2006.

They start off by saying that although there is a risk of a concentrated epidemic amongst IDUs and sex workers transferring to a slower but deeper amongst the general popular, so far this possibility remains conjecture. They criticize the work of the likes of Eberstadt and the NIC thus (my emphasis):

The future of the emerging HIV epidemic in Russia is difficult to project, owing in part to the varying quality of both HIV surveillance and data describing on prevailing patterns of risk behavior. In order for HIV infection to become generalized within a population, sexual transmission must become the main route of transmission (because the other routes influence fewer people)…Despite these problems, demand from policy makers and the media often leads to speculative projections of the future of emerging HIV epidemics and their economic impact, such as those projections by Eberstadt and the US National Intelligence Council.

The key difference between the lower prevalence projections and the very high projections is the assumption about future heterosexual transmission. The higher projections assume that the epidemic will be essentially heterosexual and will follow the trends observed in Sub-Saharan Africa. This seems unrealistic based on current knowledge of the situation in Russia, where the epidemic is still predominantly concentrated in injecting drug users.

As we can see below, there is a huge divergence between models built on differing assumptions.

The sub-Saharan models of Eberstadt, NIC and the World Bank (Ruhl et al) quickly lead to demographic (and economic, geopolitical, etc) catastrophe. The Transnational Family Research Institute (TFRI) works from different assumptions, namely that “the size of the behaviorally define high-risk groups will not grow; prevalence in high-risk groups is close to stabilization; the epidemic proceeds mainly by spreading to rest of population through sex contacts”, which results in a peak of 1.2% HIV prevalence in 2015 assuming that this figure was 0.6% in 2002 (as estimated by UNAIDS).

The Knowledge for Action writers then create their own model, which uses the available data on Russian IDU injecting and sex habits and different assumptions about the patterns of sexual mixing between low and high risk groups, IDU life expectancies and the effect of sex work on future HIV prevalence. (The report also has all the differential equations used for the model, in case you want to play around with the model).

The degree of sharing needles and syringes observed in the IDU data do not produce an epidemic in the model. As already noted this may be because there is no epidemic in the regions or because behavior has changed, or it may be a result of the survey not capturing the high risk individuals. With the rate of sharing estimated by our Russian colleagues an HIV epidemic is observed. The epidemic is concentrated in the high risk population and is driven by IDU transmission with sexual transmission causing a small HIV epidemic in the low risk population.

Under their model the epidemic becomes endemic amongst IDUs, but otherwise affects the general population little.

The model parameters specified in this way with current prevalences of infection in risk groups generated an epidemic concentrated in injecting drug users.

The model parameters specified in this way with current prevalences of infection in risk groups generated an epidemic concentrated in injecting drug users.

Despite the high levels of prevalence amongst IDUs, they are a small part of the population and as such overall prevalence of HIV does not exceed 1% under these conditions.

The overall prevalence does not exceed 1% of the population.

The overall prevalence does not exceed 1% of the population.

IDU’s dominate in absolute numbers too, albeit substantial numbers of low risk women continue getting infected. (Thus the current trajectory of increasing heterosexual transmission, particularly to women, is not surprising).

There are numerous infections in low risk women who are the sex partners of clients of injecting drug using sex workers.

There are numerous infections in low risk women who are the sex partners of clients of injecting drug using sex workers.

There are a number of factors which can either improve or exacerbate the situation. For instance, expanding ART to more patients will substantially reduce new cumulative HIV infections. Effectiveness would be further increased by reducing the use of unclean needles (e.g. by handing out sterilized syringes). If starting in 2010 some 50% of patients were to be given ART and if there was a 70% reduction in the use of unclean needles, then peak infection will come in at 0.7% of the population round about 2015, instead of nearly 1.0% in 2020.

I will now quote their (condensed) conclusion in extenso:

The future of the emerging HIV epidemics in Russia is difficult to project, mainly as a result of the varying quality of surveillance data and information on prevailing patterns of risk behavior. In this chapter we have explored previously published projections for Russia and have used models to explore epidemic trajectories using model parameter estimates derived from the behavioral surveys and other primary research undertaken during the course of the Knowledge Programme…

Previous projections of the future of the HIV epidemic in Russia give results so varied that they are very difficult to interpret. As well as lack of information about risk behaviors, there is also considerable uncertainty over both the size and turnover of high risk populations. The models used here indicate the potential importance of both the size and turnover of high risk population. The model of IDUs and the general population shows an epidemic concentrated in the high risk IDU population. Prevalence in the low risk general population is driven by high risk individuals who cease their high risk behavior and return to the low risk population, bringing with them a higher probability of being HIV positive. Further, the more extensive model developed in collaboration with Russian partners indicates that small changes in parameters describing transmission can have a considerable impact on the HIV epidemic.

The behavioral surveys undertaken in this Knowledge Programme provide some information on both sexual and injecting drug use behaviour in the general population and the harder to reach IDU population. However, as shown in an exploration of possible sexual and sharing partner change rates, considerable heterogeneity in behaviour is observed and characterizing this heterogeneity is problematic. The rates of needle sharing reported in the IDU behavioral survey were not sufficient to produce an HIV epidemic in our models. This may indicate that there will only be a limited HIV epidemic in this population because of a lack of risk behavior, it may also indicate a change in behavior as a result of greater awareness of risk. Alternatively, it may be a result of the IDU survey not capturing the highest risk individuals. As this result is in contrast to that found in other studies further exploration of the distribution of risks of this high risk population is needed…

Finally, predictions of the HIV epidemic produced by the models in this chapter are relatively reassuring. In one sense they are possibly also conservative given that the rates of sharing used to parameterize the models were considerably higher than that reported in the IDU behavioral survey. However, we should still be concerned as many of the parameters used are poorly estimated and model results show that small variations in parameters may have a considerable effect…

So let’s get this straight. Predictions of a sub-Saharan scale die-off due to AIDS are completely without merit from a theoretical, practical or even intuitive perspective. One of the more serious (and transparent!) works on it acknowledges that even its base case of around 1% HIV prevalence in 2020 might be too pessimistic. Finally, the Russian government itself has woken up to the crisis and is lavishing immense resources on the problem.

Speaking of which, at least Pokrovsky has an excuse for wild exaggeration – as a key figure in the Russian government on AIDS policy, he’s allowed to lobby for more funds. Cutting the number of people with AIDS is a noble goal. If doom-mongering about how Russians are going to be dropping like flies a few years down the road if nothing is done is what it takes, so be it. The other analysts don’t have a good excuse.

Available en français at Le Kremlin contre le sida by Alexandre Latsa.