MATHEMATICAL AND STATISTICAL LIES ABOUT HIV/AIDS
Posted by Henry Bauer on 2007/12/02
I had some trouble settling on a title for this story. I tried “Incompetent…”, “Statistically illiterate…”, “Innumerate…”. But really, when numbers are put out that are misleading, that make or imply unwarranted claims, numbers that are plainly misleading, “lies” seems apt enough. The choice might also be justified by reference to the well known saying about “Lies, damned lies, and statistics”.
* * * * * *
As Oigen commented (11/29, to “HIV/AIDS: NUMBERS THAT DON’T ADD UP”), HIV/AIDS estimates for China are dropping even faster than elsewhere, from “nearly 1 million” to 840,000 in 2004 to 650,000 in 2005 and now 223,501, according to a Nov. 29 AP report from Beijing, “China reports sharp drop in HIV/AIDS”. “Experts have said the figures are probably accurate because they are in line with a change in the way data are collected.”
No sooner said, though, than experts at the same official sources said something different: same date of 29 November from Beijing, “China increases estimate of people with HIV”:
“The number of people estimated to be living with HIV in China has risen to 700,000, says a report released Thursday by the United Nations and the Chinese government. The government had previously estimated that 650,000 people were living with HIV. . . . There were 50,000 new cases in 2007, mainly among intravenous drug users and sex workers [emphasis added] . . . . ‘China’s HIV epidemic remains one of low prevalence overall, but with pockets of high infection among specific sub-populations,’ says the 38-page report, which was to be released officially on Saturday, World AIDS Day. ‘A number of core challenges remain.’ Those include the need for better advocacy and education, improved treatment and care, and more focused education and discrimination reduction . . . . The report also notes that the number of HIV cases officially reported still remained at 223,501–far lower than the estimated total in part because of people’s reluctance to seek testing in China. The officially reported figure includes those who developed AIDS and those who died from the disease. HIV gained a foothold in China largely due to unsanitary blood plasma-buying schemes and tainted transfusions in hospitals.”
Among the points deserving of comment:
1. The glaring incompetence as well as gall of that “1” in 223,501–as though they know it’s not 223,502 or 223,500.
2. That “Experts have said the figures are probably accurate” even as (presumably other) “experts” give a much larger number.
3. Asserting as known what cannot be known.
Official reports, press releases, and media commentary are rife with these examples of incompetence in the most rudimentary practices of mathematics and statistics. An elementary point in writing numbers is that the meaningful digits–those not zero–represent a claim as to accuracy. “223,500” claims that it is known that the actual value lies between 223,450 and 223,549; in other words, that it is known to within ±50, which is a few parts per 1000, a fraction of a percent. Given that over the space of 5 years, the supposed number of HIV/AIDS cases has been changed by a factor of more than 4, such a claim of now being accurate to less than a percent would be laughable–were the claim not so much more absurd, that the actual value is 223,501 and therefore lies between 223,500.5 and 223,501.49.
Unfortunately, this cannot be written off as the result of poor reporting or of non-peer-reviewed press releases, because the same blatant error is committed in one the flagship medical journals, JAMA: “The estimated number of US cases of HIV/AIDS among MSM by year of diagnosis in the 33 states and US dependent areas with confidential named-based HIV reporting increased from 16 167 in 2001 to 18 296 in 2005” (Jaffe, Valdiserri, & De Cock, 298  2412-4). So even though these are estimates and not actual counts, the authors offer this ridiculous array of “significant” digits. “16,200” and “18,300” might be acceptable, though “16,000” and “18,000” would reflect better the uncertainty of these estimates. This is more than a small blot, not only on the authors but on JAMA’s peer-reviewing as well; if the peer reviewers don’t pick up on this sort of point, what confidence can one have in their assessment of more complicated matters?
Remarkably, astonishingly, sadly, this sort of frank incompetence is far from uncommon in the HIV/AIDS literature. My book cites several instances of similar blunders by the Centers for Disease Control and Prevention–incorrect use of “significant figures” notation (pp. 110-1, 192), claims of correlation not borne out by the data (p. 110 ff.), and assertions that correlation proves causation (pp. 194-5), which students of statistics are warned against in their first course, if not their very first lecture.
That unnamed experts assert as probably accurate, numbers that are clearly unreliable, underscores how uncritically the media handle matters of medicine and science. Unnamed “experts”, incessantly cited in all sorts of connection by the media, never deserve to be taken or offered as authoritative. When it comes to HIV/AIDS, examples abound of incompetent and unjustified statements even by named “experts”, even in the peer-reviewed literature, disseminated uncritically by the media; for instance, reporting (on and about 28 November) on the piece in JAMA referred to above:
“ ‘Lack of HIV prevention efforts among MSM fueling increase in new diagnoses, JAMA Commentary says’ (http://allafrica.com/stories/200711281011.html, citing Kaisernetwork.org).
A lack of HIV prevention efforts and an increase in risky sexual behaviors among men who have sex with men are fueling an increase in new HIV diagnoses among the group . . . the number of HIV/AIDS cases among U.S. MSM increased by 13% — from 16,167 to 18,296 — between 2001 and 2005”.
Those ridiculously precise numbers are reported without comment. Moreover, that a lack of prevention efforts and increases in risky behavior are among the reasons is sheer speculation and should be identified as such, not reported as fact.
This is no mere quibble. Failing to make clear that these are guesses strengthens the implication that these cited experts know what’s going on and are to be believed. This is misleading in an important sense.
Several of the cited assertions might appropriately be greeted by asking, “How do you know that?”
— While reported cases are about 225,000, the “real” number is 700,000.
— Better advocacy, education, and “discrimination reduction” would change matters.
— HIV gained a foothold in China via contaminated blood.
Just posing the question, “How could you know?”, can reveal how shaky the assertions are:
— The validity of the method used to estimate the “real” numbers could only be checked against actual counts–and if actual counting were feasible, estimates wouldn’t be needed in the first place.
— The belief that HIV is sexually transmitted is the sole reason for believing that behavioral interventions can make a difference. The JAMA article cites studies suggesting that they can, yet has to admit that “How well behavioral interventions work over time or when translated from research into practice is not known”.
— Those recipients of contaminated blood in China must have engaged in very vigorous sharing of infected needles to amplify their foothold into some 700,000 infectees. Or, if instead one blames them for exceptionally vigorous sexual promiscuity, it raises the conundrum that it is sex workers and drug users who are the main groups of HIV-positives in China: Do the sex workers “work” primarily with drug abusers? And do they thereupon transmit HIV among themselves without benefit of clients? Sex workers earn more, the more clients they have; where are all the infected clients?
This is all quite typical of HIV/AIDS discourse. Unqualified assertions are put out even though they are based on presumptions and not on observations or facts, and even when they are frankly implausible.
* * * * * *
One really cannot keep up with the changes in official statements. “China increases estimate of people with HIV”, above, revealed that the “50,000 new cases in 2007 [were] mainly among intravenous drug users and sex workers [my emphasis]”. But “Sex now the main cause of HIV in China” (from Henry Sanderson, AP, Beijing, November 30): “Sex has overtaken drug use as the main cause of HIV infections in China . . . according to experts and a report released this week”.
* * * * * *
As to “experts”, the authors of the cited JAMA article are anything but neophytes. Jaffe is described as department of public health director at Oxford University, De Cock as head of HIV-AIDS at the World Health Organization, and Dr. Ronald Valdiserri as chief consultant to the public health strategic health care group of the U.S. Department of Veterans Affairs. Jaffe and De Cock have been in the HIV/AIDS field from its beginning, and both contributed significantly to the wrong track that the field took. Jaffe was lead author on a 1983 study* that revealed a high rate of previous illnesses and drug abuse among AIDS patients, clear support for a lifestyle explanation for AIDS; yet later he went along with the HIV story. De Cock’s early studies in Africa are described in much less than complimentary detail in “AIDS, Africa and Racism” (Chirimuuta & Chirimuuta, 1989).
* Jaffe et al., “National case-control study…”, Annals of Internal Medicine, 99  145-51]