13 February 2004

Postscript to Response to Rian Malan's Confusion about HIV/AIDS Statistics in Noseweek and The Spectator

Rian Malan has responded to my response to his articles in Noseweek and The Spectator (see the noseweek website).

This Postscript makes some comments on his response. It is written primarily to put the record straight on further deceptions by Malan and is only for the truly interested. However, for the less interested, it is worth noting that in my original response I wondered if Malan's errors were due to incompetence or lack of integrity. The balance of evidence now shows that Malan's errors were dishonest.

Firstly though, an important point about scientific debate. Martin Welz, the editor of Noseweek takes umbrage at my comment that Malan's articles contained hardly any verifiable references, were littered with serious errors, unattributed quotes, unnamed science journals, unnamed experts and misrepresentations. Welz points out that Malan's pieces were magazine articles, not publications in scientific journals. He asks "Why is it that so few people are able to conduct a civilised discussion about AIDS". However, the critical question is: How do we conduct a scientific discussion on AIDS? - not how to be nice to each other. A fair and proper scientific discussion can only take place if all parties agree to supply references, attribute quotes correctly, do their best not to misrepresent facts etc. And yes Malan's pieces were magazine articles, but they caused tremendous confusion and were scientifically flawed. There is a good reason that neither Malan nor the less sophisticated denialists (i.e. those that deny that HIV causes AIDS, as opposed to Malan, who denies that there's a massive epidemic) such as Anthony Brink, Anita Allen, Sam Mhlongo and David Rasnick can get their views published in reputable scientific journals and have to settle for the popular press: their arguments collapse under careful scientific scrutiny.

In his response Malan says "that Geffen actually agrees with me, at least on the big issues." So let's make it clear about which big issues we disagree on:

1. Malan contends that too much is being spent on the HIV epidemic at the expense of other diseases. My view is that much more money is needed to implement treatment and prevention programmes in order to save the lives of millions of people in the developing world. My view is also that the focus on the HIV epidemic is creating greater awareness of and political action on inequities between health care in rich and poor countries, as well as awareness of and better responses to other diseases primarily affecting the poor, such as TB and malaria. Of course I acknowledge that there are groups and individuals profiteering off the AIDS epidemic and wasting valuable resources, but the same goes for almost any high profile issue affecting human development.
2. Malan's articles clearly demonstrate that he doubts the scale and seriousness of the global HIV epidemic. But there is overwhelming evidence demonstrating the opposite.
3. Malan's articles argue that South African AIDS statistics are grossly exaggerated. In so doing he makes incorrect arguments that misrepresent both Stats SA and ASSA.
4. Malan suggests that AIDS demographers have deliberately exaggerated their estimates. Undoubtedly, as in every complex field of human endeavour, there are those that deliberately exaggerate, as well as those, like Malan, that deliberately understate. But his inclusion of the ASSA researchers in the category of deliberate fudgers is without evidence.

There are many lesser points of disagreement, already explained in my original response.

Here I cover only the main points in Malan's response to my original article.

Making Up Numbers and Attributing them to Stats SA

In response to my detailed argument about his incorrect Stats SA calculation upon which his argument for this country rested, he admits that he knowingly misrepresented the Stats SA report, which raises serious questions about his integrity, though we are told he "had his reasons". He then tries to justify himself again by questioning why doctors "lie" and do not write down HIV (or something similar) on the new death certificate forms. Firstly, it is not a lie, for a doctor to write down the direct cause of death, such as an opportunistic infection. Secondly, doctors I have spoken to (who treat many HIV patients) are still not comfortable with writing HIV on the confidential part of the death certificate and from practical experiences they have legitimate concerns about the confidentiality of the forms.  Thirdly, at least one respected and influential HIV doctor has, a while ago, publicly called on doctors not to write HIV on death certificates so as to prevent possible breaches of confidentiality. Fourthly, many doctors correctly do not presume HIV if the patient was not tested (in the public sector the doctor writing the death certificate sometimes would not have even met the live patient). It is one thing to examine a trend based on a large sample of TB deaths (as found in the Stats SA report) and draw conclusions about what percentage are HIV-related; it is an entirely different matter to determine that an individual TB death was HIV-related.

There is no doubt that many doctors frequently do not write HIV on death certificates. As a "telling vignette" (of much more durability than Malan's sample of coffin makers drawn from the Yellow Pages),  I have questioned four doctors, three of whom treat very large numbers of HIV patients. All of them seldom wrote HIV on death certificates in the period of the Stats SA report. Now only one of them does so regularly - the one with the smallest number of HIV patients! Malan fails to see that doctors frequently not writing HIV on death certificates is a separate matter from mortality trends in the Stats SA report which clearly demonstrate a growing epidemic that can only be due to HIV.

And frankly, as one doctor put it to me, "Malan has no idea of the psychological complexity surrounding this epidemic."

The table below lists the combined proportion of death due to TB, HIV and influenza and pneumonia, the three biggest defined natural causes of death, for the years 1997 to 2001 as calculated using the figures from the Stats SA report (influenza and pneumonia are one category in the report). Once one understands that TB and influenza and pneumonia frequently occur as HIV opportunistic infections,  the trend in mortality showing that we have a large HIV epidemic that is becoming an AIDS epidemic can be seen.

Year     Proportion
1997    15.6%
1998    18.6%
1999    21.7%
2000    24.8%
2001    26.3%

Incidentally, Malan offers no response to the point that his entire calculation, which he attributed to Stats South Africa, is invalid because the 12% sample taken by Stats South Africa was recorded deaths in their possession, not all deaths.

Getting it Wrong on Medical Schemes

With regard to medical schemes, Malan concludes that he cannot really prove his case and neither can I. Nonsense. Malan's contention was that the ASSA medical scheme estimate is exaggerated. My contention was (and is) merely that given the available evidence the ASSA estimate is plausible because it matches the limited available evidence. I do not vouch for more than that. When more evidence becomes available, ASSA medical scheme estimates might very well have to be revised up or down, but nothing Malan has written should compel the ASSA modelers to change their estimates. Malan's reduced confidence in his argument that ASSA was overestimating is not insignificant. In a letter to the Cape Times (17 October 2003) he asserted that we have very good estimates for the middle and upper classes and then went on to make his argument that HIV estimates in medical schemes  are exaggerated. He obviously hoped by showing that the medical scheme estimates are wrong, the credibility of estimates for the general population (which is predominantly not middle-class) would be undermined. But he is again wrong; we do not have particularly good prevalence statistics on the middle and upper classes; our understanding of prevalence among the poor is much better. Once this is understood, the real conclusion he wanted his readers to draw from the medical scheme estimates dissolves.

Taking Quotes of Context to Malign Scientists

The remainder of Malan's response is uninteresting (albeit frequently wrong or disingenuous), except for one point. It is irrelevant to the debate on the size of the epidemic, but perhaps it says much about Malan's integrity, or lack thereof.

In Malan's description of the MRC Mortality report he quotes a `boffin' who says that further reference to actual death reports `will be of limited usefulness.' in order to demonstrate that the MRC researchers had more faith in their models than the actual data. In my response I asked where he found this unreferenced quote, because the MRC researchers commented on the `usefulness of establishing a rapid mortality surveillance system' and suggested that this system should become more formalised (in contradiction of Malan's contention). Malan responded by pointing out that the boffin quote is in the very same report. But here is the quotation with its surrounding text:

"[A]lthough ASSA is in the process of developing an improved model (ASSA2000), the latter model is not yet fully calibrated. When it is, it will be calibrated, inter alia, on the reported deaths and hence comparison with the reported deaths will be of limited usefulness."

Malan has taken the quote entirely out of context. ASSA600, the model used in the MRC report, was not calibrated on the death data, whereas ASSA2000 is calibrated on the death data. One can therefore gain useful insights into the validity of the ASSA600 model by comparing the estimated deaths with the recorded deaths, but there is little point in comparing the ASSA2000 deaths with the actual deaths on which the model is calibrated (though there is value in comparing the ASSA2000 model with future death data that were not used in the calibration). The quotation when taken in content does not in any way imply that models can replace research on the raw data. Surely Malan realised this? Was he unable to resist the temptation to unfairly malign the MRC researchers?