Wednesday, October 31, 2012

Antivaccine Post Hoc, Ignorant Reasoning

In a recent edition of BMJ Case Studies, Deirdre Therese Little and Harvey Rodrick Grenville Ward from Australia published a report in which they suggest that a 16-year old girl suffered early menopause as a result of Gardasil, the HPV vaccine. Several anti-vaccine blogs including the Population Research Institute and Gaia Health and  have picked up on the story and run with it, presenting it as further evidence of the dangers of vaccines. Unfortunately, all these people miss the obvious point that the entire case against Gardasil rests on a Post Hoc fallacy in which a temporal correlation is assumed to be a causal relation and an Argument from Ignorance. Let's take a look at the actual case study and the reporting on it to see how this entire case is based on fallacious reasoning.

In the summary of their case study published on the BMJ website Little and Ward write:
Although the cause is unknown in 90% of cases, the remaining chief identifiable causes of this condition were excluded. Premature ovarian failure was then notified as a possible adverse event following this vaccination.
See what little and ward do here. They begin by noting that 90% of the cases of premature menopause have no known cause.  They then conclude that, because they eliminated the 10% of known causes the early onset menopause must have been caused by the vaccine. This is a great example of the Argument from Ignorance. We have no idea what causes this condition, therefore it must have been Gardasil. In addition, the reason for assuming Gardasil must have been the cause is that about a year before she became menopausal, the woman had received Gardasil. So, we have a Post Hoc fallacy combined with an Argument from Ignorance to reach the conclusion that Gardasil caused the amenorrhoea (fancy medical jargon for early menopause). Let's look now at how several antivaccine websites pick up and amplify these errors.

First, Population Reasearch Institute describe the case study as follows:
Dr. Deirdre Little, the Australian physician who treated the girl, has published a complete account in the British Medical Journal. (BMJ Case Reports 2012; doi:10.1136/bcr-2012-006879) Her report explains that the girl's menstrual cycles were regular until she received the Gardasil vaccination in the Fall of 2008. By January 2009, her cycle had become irregular. Over the course of the next two years, her menses became increasingly scant and irregular, until by 2011, she had ceased menstruating altogether.
Look at the reasoning here. In the Fall of 2008 the women gets the HPV vaccine. She then stops menstruating in 2011. We are looking at a roughly three year gap between the event and its supposed cause. I am sure that there were many things that this young girl did in 2008. Why are these not considered as causally related to her amenorrhoea? We need more than a remote temporal correlation to establish a causal relation between these two events. Later in the same post Mosher and Crnkovich write:
While Dr. Little could not confirm that Gardasil caused the destruction of the girl's reproductive system, she was able to rule out all other possible causes. The circumstantial evidence implicating Gardasil is strong.
As we know from the original case study, 90% of these cases have no known cause. Thus, by eliminating all known causes, we are still left with a gaping hole of ignorance. To try and fill that hole by asserting that is must have been the Gardasil is clearly an Argument from Ignorance. Furthermore, as I noted above, the circumstantial evidence implicating Gardasil is actually incredibly weak, amounting to nothing more than Post Hoc reasoning.

We see a similar error in a post by Heidi Stevenson at Gaia Health. She writes:
The girl has been thoroughly examined and tested. There is no known explanation other than the series of three Gardasil vaccinations she had.
This is, again, a classic Argument from Ignorance. Stevenson is arguing that because we eliminated a bunch of known causes it had to have been Gardasil. But, as I noted above, one can't base a conclusion on a lack of knowledge or information. Given that we know so little about what actually causes amenorrhoea, the most that we can conclude is that we don't know.

As a critical thinker, it is very disturbing for me to see such an elementary logical error used as the basis for a case study that was then accepted by a fairly reputable medical journal. These kinds of basic reasoning errors would be grounds for failure if they were committed by my students, yet here are actual medical practitioners committing them in a reputable journal, and these errors are then being amplified by various ideologically motivated bloggers. This point to a sad state of affairs for medicine and science.

h/t to Orac

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.