Amongst the many anti-vaccination memes meandering through cyberspace is one that claims that the FDA has “admitted” an association/causal link between the Tripedia vaccine for diphtheria, tetanus and pertussis and sudden infant death syndrome (SIDS).
Lets leave aside the fact that SIDS deaths are, by definition, unexplained (i.e. the cause of death remains unknown even after autopsy), and examine the evidence presented in this meme.
Here’s a website boldly proclaiming the FDA’s admission of this “shocking fact”:
On the website, they provide a link to the Tripedia vaccine insert and ask the question – “How many doctors administering these vaccines have actually taken the time to read the insert?” Before asking whether or not the author of the article has actually taken the time to read the insert, I want to briefly consider the issue of causation.
The FDA reports all adverse events that occur within a specific period of time following the administration of a vaccine (or other drug) during clinical trials and thereafter (those that are reported to them). Does this mean that all these reported events are caused by the vaccines?
Now, as David Hume definitively demonstrated way back in the 18th Century – the confusion of correlation (co-occurrence) with causation is a basic malfunction of intuition. Briefly, Hume states that the following of event A with event B, even if this occurs so frequently that the two events might be considered “constantly conjoined”, does not in any way demonstrate a causal link between the two. This seems obvious, but it’s actually counterintuitive because we quickly become so used to seeing B follow A that we begin to infer the imminence of B each time we observe A. Our brain automatically creates the impression of a casual link.
In modern science we’re very familiar with this phenomenon; so familiar, in fact, that we have a maxim that warns the uninitiated about its existence – we like to shout “correlation does not imply causation” from the top of our lungs every time we see people conflating the one with the other. In order to avoid confusion as much as possible, we use statistics to tease apart correlation and causation. Statistics are boring so I’m not going to get into them, but generally we use them to demonstrate that the chances of a given result having occurred by chance (i.e. of it being the product of correlation) are less than 5% or less than 1% before we (tentatively) accept a causal link between two events.
So, the FDA does indeed report that SIDS has occurred after the immunisation of babies using the Tripedia vaccine, but does this actually demonstrate a casual link between the two events? It doesn’t, because no statistics have been done to assess whether or not the apparent association could have occurred by chance alone. There’s a fairly good reason why no statistical analyses were conducted in this case though; lets have a look at what it actually says about SIDS in the Tripedia insert:
“Causes of deaths included seven SIDS, and one of each of the following: enteritis, Leigh Syndrome, adrenogenital syndrome, cardiac arrest, motor vehicle accident, and accidental drowning. All of these events occurred more than two weeks post immunization. The rate of SIDS observed in the German case-control study was 0.4/1,000 vaccinated infants. The rate of SIDS observed in the US open-label safety study was 0.8/1,000 vaccinated infants and the reported rate of SIDS in the US from 1985-1991 was 1.5/1,000 live births.”
So, is this an “admission of a shocking fact”? How many people claiming it is have actually read the insert I wonder? It says that the average rate of SIDS observed post-immunisation, between the German and American studies, was 0.6/1000. Meanwhile, the rate of SIDS in the US general population from 1985-1991 was 1.5/1000, or 0.9/1000 greater than amongst the test cohort. Oh my! That is a “shocking fact” – immunisation with the Tripedia vaccine reduces the risk of SIDS!!
Oh. Damn. Correlation does not imply causation!
Disclaimer: the above article is an opinion piece and does not imply anything it does not state directly.