Hot on the heels of a (probably wrong) paper on ivory poaching that I criticized a few days ago, Vox reports on a paper that claims schools that give away condoms have higher teen pregnancy rates. Ooh look, a counter-intuitive finding! Economists love that stuff, right? This is a bit unfortunate for Vox since the same author has multiple articles from 2014 about rapidly falling birth rates that are easily explained by the fact that teenagers are really good at using contraceptives. So which Vox is correct, 2014 Teens-are-pregnancy-bulletproof Vox that cites national pregnancy and abortion stats, or 2016 give-em-condoms-and-they-breed-like-rabbits Vox that relies on a non-peer-reviewed article by economists at NBER? Let’s investigate this new paper …

The paper can be obtained here. Basically the authors have found data on school districts that did or didn’t introduce free condom programs between 1989 and 1993, and linked this with county-level information on teen birth rates over the same period. They then used a regression model to identify whether counties with a school district that introduced condom programs had different teen pregnancy outcomes to those that didn’t. They used secondary data, and obtained the data on condom distribution programs from other journal articles, but because population information is not available for school districts they used some workarounds to make the condom program data work with the county population data. They modeled everything using ordinary least squares (OLS) regression. The major problems with this article are:

  • They modeled the log of the birth rate using OLS rather than directly modeling the birth rate using Poisson regression
  • Their tests based on ratios of teen to adult births obscures trends
  • They didn’t use a difference in difference model

I’m going to go through these three problems of the model, and explain why I think it doesn’t present the evidence they claim. But first I want to just make a few points about some frustrating weaknesses in this article that make me think these NBER articles really need to be peer-reviewed before they’re published.

A few petty complaints about this article

My first complaint is that the authors refer to “fighting AIDS” and “AIDS/HIV”. This indicates a general lack of familiarity with the topic: in HIV research we always refer to the general epidemic as the HIV/AIDS epidemic (so we “fight HIV/AIDS”) and we only refer to AIDS specifically when we are referring to that specific stage of progression of the disease. This isn’t just idle political correctness: patterns of HIV and AIDS differ widely depending on the quality of notification and the use of treatment (which delays progress to AIDS), and you can’t talk about AIDS by itself because the relationship of AIDS and HIV prevalence depends highly on the nature of the health system in which the disease occurs. The way the authors describe the HIV epidemic and reponses to it suggests a lack of familiarity with the literature on HIV/AIDS.

This sloppiness continues in their description of the statistical methods. They introduce their model as follows:

Condom model

But on page 10 they say that the thetas represent “county and year dummies” and that the Tc represents “county-specific trends”. These are not dummies. A “dummy” is a variable, not a parameter, and “dummies” for these effects should be represented by an X multiplied by a theta. In fact the theta and Tc are parameters, and in any kind of rational description of a statistical model this model is written wrong. It should be written with something like ThetacXc where Xc is the dummy[1].

This kind of sloppiness really offends me about the way economists describe their models. This is a simple OLS regression of the relationship between the log of birth rate and some covariates. In epidemiology we wouldn’t even write the equation, we would just list the covariates on the right hand side. If anyone cares about the equation, it’s always the same and it’s in any first year textbook. You don’t make yourself look smart by writing out a first year sociology equation and then getting it wrong. Just say what you did!

So, with that bit of venting out of the way, let’s move on to the real problems with the article.

Another model without Poisson regression

The absolute gold standard correct method for modeling birth rates is a Poisson regression. In this type of equation we model counts of births directly, and incorporate the population as an offset. This is a special case of a generalized linear model, and it has a special property that OLS regression does not have: the variance of the response is directly related to the magnitude of the response. This is important because it means that the uncertainty associated with counties with small numbers of births is not affected by the counties with large numbers of births – this doesn’t happen with OLS regression. Another important aspect of Poisson regression is that it allows us to incorporate data points with zero births – zero rates are possible.

In contrast the authors chose to use an OLS regression of the log of the birth rate. This means that there is a single common variance across all the observations, regardless of their actual number of births, which is inconsistent with the behavior of actual events. It also means that any counties with zero births are dropped from the model, since they have no log value. It also means that there is a direct linear relationship between the covariates on the right hand side of the model and the outcome, whereas in the Poisson regression model this relationship is logarithmic. That’s very important for modulating the magnitude of effects.

The model is, in fact, completely inappropriate to the problem. It will give the wrong results wherever there are rare events, like teenage births, or wherever there are big differences in scale in the data – like, say, between US counties.

Obscuring trends with a strange transformation

I mentioned above that the article also uses the ratio of teen to adult births (in age groups 20-24) to explore the effect of condom use. Figure 1 shows the chart they used to depict this.

Figure 1: The weird condom diagram

Figure 1: The weird condom diagram

 

Note that the time axis is in years before and after implementation of the program. This is a highly deceptive figure, because the schools introduced condom programs over 4 years, from 1989 to 1993. This means that year 0 for one school district is 1989, while for another it is 1992. If teen births are increasing over this period, or adult births are decreasing, then the numbers at year 0 will be rates from four different years merged together. This figure is the mean, so it means that four years’ worth of data are being averaged in a graph that only covers ten years’ worth of data. That step at year 0 should actually occur across four different points in time, within a specific time trend of its own, and can’t be simplified into this one diagram.

Note that the authors only show this chart for the schools that introduced a condom program. Why not put a similar line, perhaps in a different color, for school districts that didn’t? I suspect this is because the graph would contradict the findings of the model – because either the graph is misrepresentative of the true data, or the model is wrong, or both.

This graph also makes clear another problem with this research: the authors obviously don’t know how to handle the natural experiment they’re conducting, since they don’t know how to represent the diverse start points of the intervention, or the control group.

Lack of a difference in difference model

The authors include a term for the effect of introducing condom distribution programs, but they don’t investigate whether there was a common effect across condom distribution and non-condom distribution regions. It’s entirely possible that school districts without condom distribution programs also saw an increase in teen pregnancies (1989 is when MTV came out, after all, and all America went sex crazy. It’s also the year of Like a Prayer, and Prince’s song Cream was introduced in 1991. Big things were happening in teen sexuality in this period, and it’s possible these big things were way bigger than the effect of government programs.

Statistics is equal to any challenge, though[2]. We have a statistical technique for handling the effect of Miss Calendar grooving on a wire fence. A difference-in-difference model would enable us to identify whether there was a common effect during the intervention period, and the additional effect of condom promotion programs during this period. Difference-in-difference models are trivial to fit and interpret, although they involve an interaction term that is annoying for beginners, and they make a huge difference to the interpretation of policy interventions – usually in the direction of deciding the intervention made no difference. Unfortunately the authors didn’t do this, so we see that there was a step change in the intervention group, but we don’t see if there might have been a similar step change in the control group. This effect is exacerbated by having county-specific time trends, since it better enables the model to adapt to the step in the control group through adaptively changing these county-specific trends. This means we don’t know from the model if the effect in the intervention group was really confined to the intervention group, and how big it really was.

The correct model

The correct model for this problem is a Poisson regression modeling teen births directly with population as an offset, to properly capture the way rates change. It would be a difference-in-difference model that enables the effect of the condom programs to be extracted from any general upward or downward steps happening at that time. In this model, figure 1 would be replaced by a spaghetti plot of all the counties, or mean curves for intervention and control not rescaled to ensure that the intervention happens at year 0 for all intervention counties, which is misleading. Without doing this, we simply have no evidence that the condom distribution programs did what the authors claimed. The ideal model would also have a further term identifying whether a condom program did or didn’t include counselling, to ensure that the authors have evidence for their claim that the programs with counselling worked better than those without.

I’m partial to the view expressed that counselling is necessary to make condom programs work, but Vox themselves have presented conflicting evidence that teenagers are perfectly capable of using condoms. Given this, explicitly investigating this would have provided useful policy insights. Instead the authors have piled speculation on top of a weak and poorly-designed statistical model. The result is a controversial finding that they support only through very poor statistical modeling.

The correct model wouldn’t have been hard to implement – it’s a standard part of R, Stata, SPSS and SAS, so it’s unlikely the authors couldn’t have done it. It seems to me that this poor model (and the previous one) are indicative of a poor level of statistics and research design teaching in economics, and a lack of respect for the full diversity of statistical models available to the modern researcher. Indeed, I have a Stata textbook on econometrics that is entirely OLS regression – it doesn’t mention generalized linear models, even though these are a strong point of Stata. I think this indicates a fundamental weakness in economics and econometrics, and leads me to this simple bit of advice about models of health and social behavior prepared by economists: they’re probably wrong, and you shouldn’t trust them.

I hope I’m wrong, and Vox don’t keep vexing me with “explainers” about research that is clearly wrong. I don’t hold out much hope …


fn1: for those digging this far, or who often stumble across this horrible term in papers they read, a “dummy” is just a variable that is either 0 or 1, where 1 corresponds to the event of interest and 0 to not the event of interest. In epidemiology we would just say “we included sex in the model”. In economics they say “we included a dummy for sex.” This is just unnecessary jargon.

fn2: Except the challenge to be fun.

Save

He likes the smell of new viruses in the morning

He likes the smell of new viruses in the morning

This week the journal Science reports a new study finding HIV first emerged in Kinshasa (now the Democratic Republic of the Congo) in the 1920s – not the 1970s or 1980s as previously suspected. The disease was likely introduced to Kinshasa through bush-meat, but spread rapidly across the Congo through mobile workers moving on Belgian-built train networks. At that time the region was a Belgian colony, and labourers were moving across large areas of the country as they moved to and from the capital and large mining areas in the hinterland. The article also reports that Kinshasa itself had a large and active sex industry in support of he transient labourers, and this may have helped to spread the disease. It’s an interesting story of virology, archaeology and globalization.

What I find fascinating about this story is that HIV took hold in the 1920s, but wasn’t identified as a disease until the 1980s, despite the presence of medical and public hygiene programs in Kinshasa, the growth of tropical medicine as a discipline, and the presence of major militaries in the area during both world wars (most notably the Force Publique, a force of some tens of thousands of black Congolese soldiers led by white Belgian officers). Typically the military establishment pays careful attention to hygiene and to STIs, especially since the work of Florence Nightingale, but somehow during all this period they missed HIV as a disease. In fact, this new research suggests that the success of the entire discipline of Tropical Medicine should probably be reassessed.

The reason that HIV was not identified is, I think, quite simple: it has a very long asymptomatic period, up to 12 years, and it does not manifest through a single set of coherent symptoms, like measles or flu, but through a complex of opportunistic infections. The case definition for AIDS is complex and depends on a list of AIDS-defining conditions that have few commonalities, so it is extremely hard for a doctor seeing these cases in disparate people to identify a single underlying condition. Instead the symptoms are treated, and the patient dies. From the point of view of a doctor in 1920s Belgian Congo, finding an underlying cause would be almost impossible. First the doctor might see a soldier with recurrent herpes, then a miner with a rare and untreatable cancer, then a sex-worker with repeated bacterial infections. Some of these people might have got the disease sexually, some through infected needles during a vaccination drive, perhaps the soldier might have exchanged blood in a fight – 10 years ago. It’s just not possible to identify a cause in this case, or to see a common pattern.

So why do we even know about the existence of HIV at all? It was first identified in 1984, but if it had been around since the 1920s it should surely have been identifiable in the modern era, at least since the program to eradicate smallpox, when modern public health was really beginning to come to terms with infectious disease. Why so late? I think it was identified because of a stroke of luck: a group of cases in the USA that all happened in gay men, and with a disproportionate number of Karposi’s Sarcoma (KS) cases. KS is usually limited to elderly southern European men, and so its presence in young American men was highly unusual. But the real trigger was that it occurred in gay men. Its presence in gay men meant that they were all visiting the same small number of gay-friendly clinics, and they were definably different to other men. They all shared a single common factor: their sexual identity. Of course all those patients in the Congo also shared a common sexual identity but nobody thinks of heterosexuality as a defining characteristic. It’s a background property, a default setting. Whereas homosexuality is a definable strand of difference. I think this coincidence set people thinking, first because a small number of doctors saw all the cases, the diseases these cases were experiencing were very unusual for men of their age and race, and they all shared a different sexuality. This of course tripped the doctors into thinking that they must have a common condition, and that it must be related to their sexuality. This in turn sparked a search for a common cause, probably infectious, and in 1987 HIV was identified. Had HIV instead spread into America through heterosexual carriers those carriers would not all have gone to the same doctors and the disease would not have been linked to their sexual identity. This link is essential for HIV because the symptoms occur so long after the transmissive act that it is not possible to connect them without a symbolic link. Without the sexual link, doctors would not have considered an infectious cause of the range of AIDS-defining conditions they were witnessing, and they would not have sought a virus. Had the Morbidity and Mortality Weekly Review reported on a sudden rash of deaths due to Karposi’s Sarcoma, there might have been discussion, but occurring in only heterosexual people widely separated in the community, an infectious cause might not have been considered. This is especially likely since KS is just the first manifestation of AIDS, and not necessarily the killer – people travel through different trajectories of opportunistic infections to their eventual (horrible) death, and in the absence of deaths, given KS is not notifiable, it would probably simply never have come to anyone’s attention – or would have taken so long to be noticed that HIV would have been entrenched in the wider community before it was identified, if it were identified at all.

So I guess we have the unfortunate sacrifices of a significant proportion of gay men in one generation in the USA to thank for our discovery of HIV. By the time the full scope of the disease and its origins were understood, HIV was already out of control in Africa, to the point where it was causing major social and economic problems, and it’s possible to imagine real economic and social collapse happening in some parts of Africa if the disease hadn’t been identified for another 10 or 15 years – especially if by the time of its identification the rich countries were also burdened with a generalized epidemic and facing their own public health (and potentially economic) emergencies.

Which leads to a horrible speculation about the past. Would human society have survived if HIV had emerged 500 or 1000 years earlier? With death following a pattern similar to non-communicable disease and old age, no coherent virological or bacteriological principles, and the point of infection distal from the point of symptom onset, it would have been almost impossible for human society to identify the existence of the disease, let alone its cause. Worse still, HIV is transmitted from mother to child, with very high mortality rates in children, so it would have spread rapidly over generations and had huge mortality rates. Once widespread the disease is economically highly destructive, since it forces communities to divert adult resources to caring for sick adults who should be in the most productive part of their lives. In the absence of a known cause it would simply be seen as “the Scourge,” but in the absence of well-kept statistics on life expectancy and mortality rates, it might be difficult for societies to realize how much worse their health was than previous generations.

In that period there were other diseases – like the Black Death – that had an unknown transmission mechanism, but these were identified as diseases and (mostly erroneous) methods put in place to prevent them, with of course the final method being case isolation and quarantine, a technique that usually has some success with almost all diseases. But these diseases differ from HIV in that there is a rapid progression from symptom onset to mortality and the symptoms are visible and consistent, making the Black Death clearly definable as a disease, which at least makes quarantine possible. With a diverse range of symptoms, a long period from symptom onset to death (often 2-3 years) involving an array of different infections, in a society where death from common infectious diseases was normal, people just would not notice that they were falling prey to a single, easily preventable disease, so even quarantine or case isolation would be unlikely to be implemented. Another difference between HIV and the Black Death is the long asymptomatic phase of HIV guarantees its persistence even though it has a nearly 100% case fatality rate; whereas the Black Death spread through communities so fast that it soon burnt out its susceptible population, leaving a community with some immunity to the disease. HIV is not so virulent, or so kind.

I think if HIV had spread from Africa 500 years earlier, it’s possible that the majority of the human race would have died out within a century or two, leaving whole continents almost empty of people. I guess the Indigenous peoples of the “new” world would have escaped the scourge, leaving the earth to be inherited by native Americans, and most of Europe and Africa to fall to waste and ruin. It’s interesting to think how different the world might have been then, and also chilling to think how vulnerable our society was in the past through ignorance and happenstance. A salutary lesson in a world where we live ever closer to nature, but where many societies still have health systems that are too fragile to handle the challenge presented by relatively preventable diseases like Ebola virus. The Science paper also presents a timely reminder of the importance of being prepared for the unexpected, and the dangers of complacency about the threats the natural world might offer up to us in future …

Every time you criticize Mandela a fairy puppy dies

Every time you criticize Mandela a fairy puppy dies

Nelson Mandela’s passing was only a few days ago and already the left-wing press and counter-press have managed to come up with a wide range of criticisms of someone who should, ostensibly, be their hero. Slavoj Zizek, that staunch opponent of anything modern in the left, is recycling the old claim that Mandela simply changed the skin colour of the overlords; Counterpunch is leading the charge to claim that he was just a neo-liberal friend of the rich, and black people didn’t benefit from the ANC at all; the Guardian managed to give a thoroughly negative review of his funeral, with the cherry on the icing being their focus on Obama rather than, you know, the South Africans who Mandela led; and they even managed to give Simon Jenkins a go at criticizing the coverage of Mandela’s death. I can’t decide which part of Simon Jenkins’s article is worst – the fact that he paraphrases the title of a profoundly important book about the holocaust in order to criticize coverage of a hugely liberating figure; or the fact that he is writing it at all, given that he is a confirmed HIV denialist and was directly involved in promoting HIV denialist science, which cost South African blacks so many lost lives and chances.

Now, I’ll be the first to admit that I’m no fan of hagiography and I’m happy to criticize my heroes, but I would have thought that in this case someone as profoundly important as Mandela could be given a week or two before the critical analysis of his legacy began. I mean, he only just died and the left – which historically was most broadly supportive of him – have been really quick to start pissing all over his legacy. I guess it’s largely the British left I’m quoting here, but over at Crooked Timber’s comment thread on Mandela a wide range of commenters seem to have joined in with this “he didn’t immediately undo all the economic wrongs of apartheid so he was bad” chorus, and a lot of the commenters there must be American. It makes me a bit uncomfortable, especially since those on the right who were famously opposed to Mandela at the time (people like Bush, the entire Israeli government, etc) have largely refrained from resurrecting their criticisms of the time. Surely if his opponents of the time don’t feel it’s right to say anything bad about him for a week or so, it might be worth one’s while to stow it for a bit?

One of the main threads of left-wing criticism of Mandela appears to be that he didn’t do much to reduce inequality, and we see various strengths of this argument ranging from “he blew a chance” through “he let down his communist allies” to the extreme “he just swapped racial oppression for economic oppression” or “swapped one set of overlords for another” type arguments. I think there are two huge flaws in these opinions (aside from their obviously terrible timing): the first is that the data from within South Africa is not so clearly supportive of the conclusion that Mandela (and more broadly the ANC) have failed to do anything about inequality; and the second is that progress on inequality and the related left-wing complaint of a failure to rein in neo-liberalism’s negative effects needs to be judged against the context of progress in the rest of the world over the same period, and against the backdrop of HIV in South Africa.

What does the data say on inequality in South Africa?

My first complaint with criticisms of these claims is that the data on inequality in South Africa is not being well assessed, and that the broader development issues South Africa faced are not being considered. Let’s consider that second complaint first. In the Counterpunch article I linked to above, for example, Patrick Bond writes critically:

the sustained overaccumulation problem in highly-monopolised sectors continued, as manufacturing capacity utilization continued to fall from levels around 85 percent in the early 1970s to 82 percent in 1994 to below 80 percent by the early 2000s

This seems hugely unfair to me. I don’t know a great deal about South Africa, but I’m guessing that “manufacturing capacity utilization” in the 1970s was highly dominated by the use of cheap, exploitable labour who had no rights and no capacity to control the extent to which they were “utilized.” Furthermore, the sanctions of the 1980s would have further restricted the ability of South African industry to modernize in a way that would improve capacity utilization, and by the time their investments were up and running in the early 2000s they faced … China. This capacity utilization also looks pretty favourable when compared to the USA, where in 2009 it was 64%. It doesn’t seem to me that this claim is fair.

We’ll come back to this problem of context and comparison with the USA later, but for now let’s look at the data. It’s true that South Africa has a terrible level of inequality, with a Gini index of between 0.6 and 0.7 depending on how you measure it. The world bank suggests that there has been an increase in inequality (measured using the Gini), with Gini values in 1995 of 57 and in 2010 of 63. That’s not a big change, though – this UNU working paper shows that World Bank estimates of the Gini coefficient in 1995 showed a wider range of values than the entire change recorded by the World Bank between 1995 and 2010. There is no clear method for calculating variance in Gini coefficients, and not enough data generally to establish what that variance might be, so whether or not the change from 1995 to 2010 is significant is hard to know.

The story becomes even more complicated than that when you consider the data challenges in nations like South Africa, and look at more nuanced research into inequality in South Africa. It’s difficult to believe that data on black South Africans collected before 1995 was really very good or complete, so the true depth of inequality in apartheid South Africa is hard to be confident about. Furthermore, assessment of wealth in low income countries is not so simple as simply calculating income – it is typically done through assessment of consumption expenditure. This is done because poor people in low income countries tend to underestimate or misreport their income, and much of their wealth can be tied up in informal markets and means of exchange (e.g. they have land and pigs but little money). Measures of Gini in South Africa based on consumption expenditure tend to be different to those based on income, and measures of wealth based on consumption are not readily available in earlier years. Furthermore, the Gini is a very poor measure of inequality – not only is uncertainty usually not calculated, but it doesn’t give any meaningful distinction between different types of inequality, and I seriously doubt it’s linear. For example, a change in Gini index from 0.35 to 0.40 may have a very different meaning to a change from 0.57 to 0.63. I don’t think any realistic work has been done on how useful the Gini index is for either within- or between-nation comparisons.

However, there is some recent research available on inequality in South Africa that paints a more nuanced picture. This research, from the University of Stellenbosch, suggests that poverty – measured in absolute and relative terms – has declined in South Africa, and that inequality within racial groups has increased while inequality between racial groups has decreased. In fact, according to this report:

  • The proportion of households with children reporting any form of hunger has declined by 15% in the past 6 years
  • The share of black people in the middle class has increased from 11% in 1994 to 22% in 2004
  • Poverty headcount rates have declined from a peak of 53% in 1996 to 44% now, a record low
  • Income growth over the period 1994-2010 has been approximately similar amongst whites and blacks
  • The proportion of total income earned by black people has grown from 33% to 39%, while amongst whites it has declined from 55% to 48%
  • Within-race inequality contributed only 39% to inequality in 1993 and now constitutes 60%

The report also points out that World Bank Gini coefficients don’t properly adjust for household size, and household-weighted Gini coefficients were 0.67 in 1993 and are 0.69 now. They write:

A decomposition of the Theil index shows that the decline in income inequality between race groups throughout the period offset the rising inequality within groups. This trend of falling inter-racial inequality coupled with rising intra-racial inequality is also a continuation of a phenomenon first observed in the 1970s (Whiteford & Van Seventer 2000). Note that these estimates of the population Gini are near the upper end of South African Gini estimates, although they remain smaller than those calculated by Ardington et al. (2005) using the 2001 census. The trends in inequality derived from the AMPS data are likely to be more reliable than the estimated levels, as the levels may be more affected by the nature of the data (household income estimates in income bands based on a single question).

The Gini coefficients shown here are higher than those often reported. The reason for that is that many Gini calculations use the weighting for the household, without multiplying that by the household size, as should be done: Larger households have more members, and this should be considered in calculating inequality. The Gini coefficients here are thus the correct ones, and much higher than those reported by among others the World Bank, which are based on inappropriate weights. The Gini coefficient of 0.685 reported for 2006 would have been only 0.638 if the more common, but incorrect, weights were used.
This report overall paints a picture of small but noticable reductions in inter-racial inequality, and reductions in the levels of absolute poverty seen before the end of Apartheid. It’s pretty modest, but overall it seems safe to say that South Africa may reduce inequality slightly and cannot be said to have significantly increased it. This claim may seem weak, but when we compare it to the rest of the world and consider it in its proper context, it’s important.
Considering South Africa’s economic changes in the global and regional context
In economics there is a simple method for assessing the effect of an intervention called the Difference-in-Difference model. In this model you compare the actual change in the group that received an intervention with the counter-factual that would be expected if they haven’t; you estimate the counter-factual from a control group measured before and after the intervention. In this case the intervention is the end of apartheid, and the control group is other countries. Consider, for example, how income inequality has changed in South Africa and the USA since the 1970s. According to the World Bank, the top decile of income earners in South Africa control 58% of all income in 2010. Research from Stanford University puts the equivalent number in the USA at 50%, but look at the curve: since the 1970s the share of income held by the top decile of US income earners has increased from about 35% to 50%. Income inequality has increased rapidly in many high income countries under the influence of various forms of neo-liberalism and/or trade liberalization. For example in the UK the Gini coefficient has increased from 0.35 to 0.41 since 1990, a much larger (proportionate) increase than observed in South Africa. Seen against the backdrop of international changes brought about by major international movements, it appears that Mandela and the ANC have managed to resist many of the worst changes that have swept through the industrialized west. It is this international context that is missing from the Counterpunch article linked above, where they provide critical statistics about South Africa’s industrial and economic performance without any comparison to overseas, where equal or far worse changes have occurred in the same time frame. The industrial economies of much of the rich west have been hollowed out by a mixture of ponzi economics and the rapid growth of Asia; South Africa seems to have escaped the worst of some of these changes, and though things clearly aren’t pretty in the economic statistics that South Africa presents, it is also clear that they have got vaguely better and certainly not much worse, against a backdrop of really challenging international and domestic changes.
The domestic changes also need to he emphasized when assessing Mandela’s legacy. He inherited a corrupt one party state with a political system built on state violence against a powerless minority, crippled by years of sanctions and sitting on the silent time bomb of HIV. While the ANC’s response to HIV was terrible, it’s worth noting that the first 10-20 years of growth of HIV happened under apartheid, and it’s really hard to believe that since HIV was identified in 1984 the white regime was doing much to prevent its spread. Against a backdrop of revolution, poverty, discrimination and chaos, what kinds of interventions did de Klerk have in place? And even after Mandela took the reins, most of Africa was still unaware of how to deal with HIV and just how terrible it was going to become; much of the context of the epidemic that unfolded subsequently had already been set and although the response could have been better handled since 1994, and certainly since 1997, it’s not unreasonable to suppose that even a really pro-active intervention would have failed under the circumstances. As a result of this epidemic, life expectancy in South Africa has collapsed, and South Africa is one of the countries facing serious economic challenges because of the epidemic. I have written before about how terrible this epidemic can be for societies suffering it, and challenged readers to consider alternative futures where it arose as a generalized epidemic in the USA or Europe. Does anyone think that the USA would have experienced the same economic growth and changes since 1994 if it had suffered the epidemic the way South Africa is? This needs to be considered when criticizing health spending and economic growth in South Africa.
Given this context, I can only summarize by saying that Mandela and the ANC did okay in handling inequality. Obviously not as well as anyone would have liked but also much better than, say, the US under (lefty) Clinton or the UK under (lefty) Blair. So I think leftists should perhaps be a little more circumspect in their criticisms of Mandela and the ANC’s legacy. Perhaps it would be good if they took a week to laud his obvious achievements, and to read the literature.
A final note about iconoclasm in coverage of Mandela’s death
In case you thought the mainstream left was alone in being a little too quick to criticize Mandela, spare a thought for the lunatic right. The National Review Online editors’ piece on Mandela was remarkable, managing to wrap a vicious and angry rant inside a thin shell of flattery; and its commenters still complained that the NRO has become too left wing in its coverage, and that Mandela committed genocide. But perhaps best is the efforts of the white power losers from the OSR blogosphere, who hate-bombed a thread about Mandela on Dragonsfoot with complaints about his terrorism and white genocide. Remember that next time an OSR blogger drifts over here to complain about my criticisms of Tolkien … still, the OSR being caught up in 1986 I guess they haven’t worked out that apartheid is over.
Anyway, I don’t feel like this has been a great week for the mainstream left media, such as it is, and I hope that some of Mandela’s critics on the left will find this piece and consider a slightly more nuanced understanding of what he did in power. I also hope that people will start using slightly better measures of inequality than Gini indices … but that’s a post for another day!

In October my master’s student had her work on modeling HIV interventions in China published in the journal AIDS, with me as second author. You can read the abstract at the journal website, but sadly the article is pay-walled so its full joys are not available to the casual reader. This article is a sophisticated and complex mathematical model of HIV, which incorporates three disease stages, testing and treatment separately. It is based on a model published by Long et al in the Annals of Internal Medicine in 2010, but builds on this model by including the effects of methadone maintenance treatment, and doesn’t include an injecting drug use quality of life weight. It also adds new risk groups to the model: Long et al considered only men who have sex with men (MSM), injecting drug users (IDU) and the general population, but we added commercial sex workers (CSW) and their clients, who we refer to as “high risk men.” Thus our mathematical model can consider the role of both injecting drug users and sex workers as bridging populations between high-risk groups and the general population, an important consideration in China.

The HIV epidemic in China is currently a concentrated epidemic, primarily among IDUs in five provinces, and amongst MSM. The danger of concentrated epidemics is that they give the disease a foothold in a country, and a poor or delayed response may cause the epidemic to jump to the rest of the population – there is some suggestion this may have happened in Russia, for example. The Chinese authorities, recognizing this risk, began expanding methadone maintenance treatment (MMT) in the early 2000s, but it still only covers 5% of the estimated 2,500,000 IDUs in China. Our goal in this paper was to compare the effectiveness of three key interventions to prevent the spread of this disease: expanded voluntary counseling and testing (VCT); expanded antiretroviral treatment (ART); and expanded harm reduction (MMT and needle/syringe programs); and combinations of these interventions. VCT was assumed to reduce risk behavior and expand the pool of individuals who can enter treatment per year; ART was assumed to reduce infectiousness; and harm reduction to reduce risk behavior. Costs were assigned to all of the programs based on available Chinese data, and different scenarios considered (such as testing everyone once a year, or high-risk groups more frequently than everyone else).
The results showed that all the interventions considered are cost-effective relative to doing nothing; that some of the interventions saved more money than they cost; and that the most cost-effective intervention was expanding access to ART. Harm reduction was very close to ART in cost-effectiveness, and would probably be more cost-effective if we incorporated its non-HIV-related effects (reduced mortality and crime). The Chinese government stands to reap a long-term benefit from implementing some of these programs now, through the 3.4 million HIV cases averted if the interventions are successful (there are a lot of “ifs” in that sentence).This is the first paper I’m aware of that compares ART and harm reduction head on for cost-effectiveness, though subsequently some Australians showed in the same journal that needle/syringe programs (NSP) in Australia are highly cost-effective as an anti-HIV intervention. This is also the most comprehensive model of HIV in China to date, and the first to conduct cost-effectiveness analysis in that setting. I think it might be the first paper to consider the detailed structure of risk groups in a concentrated epidemic, as well. There are obvious limitations to the conclusions that one can draw from a mathematical model, and some additional limitations on this model that are specific to China: the data on costs was a bit weak (especially for MMT) and of course there are questions about how feasible some of the interventions would be. We also didn’t consider restricting the interventions to the key affected provinces, which would have made them much cheaper, and we didn’t consider ART or VCT interventions targeted only at the high-risk groups, which would also have been cheaper. For example, legalizing sex work and setting strict licensing laws might enable universal, quarterly HIV testing and lead to the eradication of HIV from this group within 10 years, but we didn’t include this scenario in the model because a) legalization is not going to happen, b) enforcement of licensing laws is highly unlikely to be effective in the current context in China, and c) data on the size and behavior of the CSW population is probably the weakest part of our model, so findings would be unreliable. Despite the general and specific limitations of this kind of modeling in this setting, I think the results are a strong starting point for informing China’s HIV policy. China seems to have a very practical approach towards this kind of issue, so I expect that we’ll see these kinds of policies implemented in the near future. My next goal is to explore the mathematical dynamics of these kinds of models with the aim of answering some of the controversial questions about whether behavioral change is a necessary or effective part of a modern HIV response, and the exact conditions under which we can hope to eliminate or eradicate HIV. Things are looking very hopeful for the future of HIV, i.e. it’s going to be eliminated or contained in most countries within our lifetime even without development of a vaccine, and that’s excellent, but there is still debate about how fast that will happen and the most cost-effective ways of getting there: hopefully the dynamic properties of these models can give some insight into that debate. This article is a big professional achievement for me in another way. It’s extremely rare for master’s students to publish in a journal as prestigious as AIDS (impact factor over 6!), and my student’s achievement is a reflection of her amazing talent at both mathematics and English, and a year of intense work on her part, but I like to think it also is a reflection of my abilities as a supervisor. There were lots of points where we could have let things slide on the assumption that master’s students don’t publish in AIDS; but we didn’t, and she did. I like to think the final product reflects well on both of us, so read it if you get the chance!

I’m in Nagasaki this week to attend the 86th Annual Meeting of the Japanese Society for Infectious Diseases, where I have presented the results of my work building a mathematical model of the HIV Epidemic in Japan. The model is currently submitted to a journal so I can’t give any detail about it here, but I can present a chart I used in the conference presentation, that is based on publicly available data from the Ministry of Health, Labour and Welfare. This chart shows the number of new cases of HIV/AIDS notified to the government annually, divided into three main transmission modes (Figure 1).

Figure 1: Annual new cases of HIV/AIDS in Japan by transmission category, 1985-2010

In this Figure, “same sex contact” means “homosexual contact,” since there’s no such thing as a case of HIV transmitted by same sex contact between women. From Figure 1 it should be pretty clear that while the epidemic appears to have peaked and even beginning to decline in the heterosexual population, amongst men who have sex with men (MSM) it is growing rapidly. Now, there are some caveats on such a conclusion in Japan: testing rates are quite low so it could be that these “new” cases are actually old cases that have only just been identified, for example, but it would be a strange world indeed if the entire slope of that line were due to remnant cases finally coming to light. So, it’s reasonable to conclude with some confidence that the HIV epidemic is growing rapidly amongst MSM in Japan. Currently prevalence is probably low, but that was the case in Australia back in 1985, and prevalence amongst MSM in Australia now is probably above 5%.

This comparison is noteworthy because Figure 1 makes it look like Japan’s experience of HIV is Australia’s 20 years ago, and if the epidemic continues to follow Australia’s trend, HIV will spread rapidly through Japan’s gay community. Of course there are big differences in HIV treatment and prevention now compared to 20 years ago, and very few people die of AIDS in Japan because of the combination of low prevalence and good treatment. But the rapid increase amongst MSM shown in Figure 1 suggests that prevention efforts to date haven’t been working, and it would be best if something could be done to prevent the further spread of the disease.

Another minor concern (raised in my presentation, actually) is that MSM in Japan tend to be less open than in the rest of the developed world, making them even harder to study but also raising the possibility that they marry and have at least some sexual contact with women. Sexuality in Asia is, in general, more fluid than in the West and less constrained by categories and boundaries, so the idea seems superficially plausible. If this is true though, it means that there is a small risk that the epidemic won’t be contained within the gay community forever. Unfortunately, no one knows the extent of this overlap in Japan, and no one knows how much injecting drug use is happening here, so it’s hard to make judgments about how such behavior might affect the future of the epidemic. This is what my mathematical modeling is (partially) aiming to do, and although I won’t reveal the results here the future is not pretty for MSM if the epidemic is allowed to continue. Even without the benefit of a mathematical model, it’s pretty easy to see from Figure 1 that Japan needs to improve interventions amongst MSM, primarily by increasing rates of voluntary testing and targeting a test-and-treat prevention strategy at this community. Given the current low prevalence of HIV, even a relatively unsuccessful test-and-treat program will probably be sufficient to contain the epidemic (though the international evidence suggests that it takes a very rigorous and probably unrealistically well implemented program to eliminate the disease). It remains to be seen whether such a targeted approach will be tried here, but hopefully my work will be one tiny step towards encouraging such a change.

Not someone you want to go bowling with...

Twenty-five years ago today the Grim Reaper appeared on Australian television to warn us about the dangers of HIV. You can see the ad through this article about the anniversary. I was 14 at the time, and it was truly terrifying. I think it did its job, and scared Australians into sexual responsibility, though now that we have treatments and testing and the like, people may be beginning to become complacent again. Although it now seems a bit hammy, I think it also compares favourably with British health and safety adverts – it’s not as tacky, and makes its point much more succinctly and believably. I particularly like the nod to the holocaust when the narrator says it could kill more people than world war 2 – a nice touch, very understated but very effective.

There was some controversy at the time, because some people interpreted it as likening gay men to the Grim Reaper (at that time it was largely a disease of gay men), but unlike in the USA there was a much better relationship between government, health workers and gay activists, and the controversy didn’t damage the ad’s effectiveness. Of course now people think that the kinds of things being said in this advert were hyperbolic or alarmist, because Australia has largely escaped the problem of HIV – another complaint made at the time was that this ad was overdoing it, and would contribute to that general suspicion people have that government health messages are just intended to scare us. But take one look at the situation in Africa and it’s clear that Australia dodged a very, very scary epidemic, and with our large drug-using population it was possible that HIV could have crossed to the heterosexual population by the early 1990s. It didn’t, and we can thank Australia’s early and very impressive response for our very lucky escape. Part of that response was this cute guy with his scythe and his slightly tatty cape, and we Australians should all be thankful for whatever small part he played in keeping us safe. So, thanks and … happy birthday Grim Reaper! If you get laid at your party, remember that prevention is the only cure we’ve got!

Many Australian public hospitals maintain a ban on circumcision. The Royal Australian College of Phsyicians (RACP) recommends against circumcision, and it is now so unpopular that 80% of Australian boys are uncircumcised. However, a committee of public health experts has recommended reversing this ban and moving to encourage circumcision, on the basis of the many health benefits of circumcision. These health benefits have been known for a long time, but the medical fraternity have responded to strong public pressure in advising against non-medical circumcision, because it is seen as a form of child abuse. Indeed, in America recently there were efforts to ban the practice on the basis that it is a form of child abuse.

This would all be a largely irrelevant debate about men’s willies and the aesthetics thereof, except that circumcision has now been shown to be an extremely effective tool in the battle against HIV. The Bill and Melinda Gates Foundation and various international aid agencies are scrambling to fund circumcision programs in Africa, where they believe that this simple and harmless procedure can significantly reduce the transmission of one of the world’s nastiest diseases. Recently at a training course in the UK I met people involved in this process and saw examples of the kinds of non-surgical devices being used to circumcise adult men (it’s a kind of ring, and after just 1 or 2 days the whole process is over and you can go back to work).

Australia is undoubtedly contributing financially and organizationally to this effort as part of its increasing aid contributions to Africa. But isn’t there something wrong here? Circumcision is essentially banned in government run hospitals in Australia, is on the nose in the USA and is frowned upon in the UK, yet these same countries are recommending circumcising the entire African continent. It’s morally unacceptable child abuse in Australia but an acceptable public health intervention in Africa? How does this kind of attitude differ from previous eras when population control was conducted through sterilization, often not clearly explained to the recipients? Is it just another example of aid-as-imperialism? How is this different to a country where abortion is strictly illegal (say, Ireland) funding abortion-based population control programs in Africa? And how can the Australian aid community (or its public health activists) criticize Chinese aid programs while we’re doing something like this?

It’s also worth remembering that historically white colonialists have been extremely uncomfortable about black men’s sexual fecundity (and their mythically enormous willies). Yet here we are – advocating chopping the end off of those massive, fecund members in the interests of stopping a disease which (apparently) cannot be stopped in Africa through behavioral change alone. Even though in Australia we have it under control through – you guessed it – behavioral change. The public health double standard is disturbingly close to the sexual insecurity …

Don’t mistake me here – HIV is a desperate situation and circumcision a minor operation that I don’t think we should shy away from as a control technique. Furthermore, I think the western bans on circumcision are silly and of the same character as the AMA’s opposition to boxing. It’s not liberal. But this hypocrisy, in which doctors won’t tolerate it in Australia or the UK but will support funding for its widespread use in Africa, reeks to me of cultural imperialism. If you won’t tolerate it here, don’t do it there.