Mark Blaxill and Dan Olmsted: Merrily confusing correlation with causation for polio

I’ve been following the anti-vaccine movement for nearly a decade now, first as a regular on the Usenet newsgroup and then, beginning almost seven years ago, blogging away. Along the way, somehow I stumbled into the role of countering the pseudoscience, misinformation, and nonsense promoted by the anti-vaccine movement. It’s dangerous misinformation, too. For instance, in the U.K., misinformation claiming that the MMR vaccine somehow contributes to autism, a lie based on the work of Andrew Wakefield, has led the MMR uptake rate there to plummet. As a result measles, once vanquished back in the 1990s thanks to the MMR has come roaring back. Here in the U.S., there are areas where there are pockets of low vaccine uptake, and as a result vaccine preventable diseases are making a comeback in some areas. We can only hope that the relatively high vaccine uptake rates elsewhere prevent those pockets from enlarging, spreading, and metastasizing.

Over the last decade or so, if there’s one thing I’ve learned over all that time covering the anti-vaccine movement, it’s that it’s generates some masters of cherry picking and obfuscation. I’ve also learned that the favored topic of the anti-vaccine movement is to conflate correlation with causation, thus trying to lure readers who might not have a scientific background into confusing correlation with causation. The most infamous example of this technique is, of course, thimerosal in vaccines. Apparent autism incidence has risen considerably since the early 1990s, largely due to expanded diagnostic criteria, diagnostic substitution, and more intensive screening. The vaccine schedule was also expanded during the 1990s, with an increasing use of vaccines containing thimerosal as a preservative. Because thimerosal contains mercury, it provided a convenient “bogeyman ingredient” that anti-vaccine activists could blame for the apparent “autism epidemic.” In 2011 (actually, as early as 2005), it is obvious that mercury in vaccines is not a cause of any sort of “autism epidemic,” given that thimerosal was removed from vaccines by the end of 2001 and autism incidence has not declined. If mercury in vaccines were a major cause of autism, one would expect that removing the thimerosal from vaccine would lead to a rapid decline in autism incidence. We’d even expect that this decrease would be evident within three to five years, given that autism is commonly diagnosed at around age three or four. No matter how the anti-vaccine movement tries to parse, this has not happened.

What’s more interesting (well, sometimes) is to watch anti-vaccine activists apply the same sort of flaws in science and reasoning to other aspects of vaccine science. I just saw a whopper of an example of this in what is now a four-part “investigation” on the anti-vaccine crank blog Age of Autism. It’s written by Mark “Not A Scientist, Not A Doctor” Blaxill and Dan “Where are the autistic Amish?” Olmsted. Blaxill is rather amusing because, alone of bloggers, he loves verbosity even more than Orac does. Where Orac might sometimes use 3,000 words where 1,500 might serve, Blaxill would use 6,000 words to say exactly nothing. This might explain why the example of mistakenly confusing correlation with causation that I’m about to discuss was mercifully divided into four parts; that is, four parts so far. I’ve been watching from afar, waiting to see how many parts Blaxill would throw in, and this might represent premature blogging, but, hey, it’s Friday, and I rarely post on the weekend anymore; so it’s now or wait until next week, and I don’t want to wait until next week.

So let’s take a look at the not-so-Glimmer Twins’ “expose” on the polio vaccine:

The five parts thus far can best be viewed as a massive confusion of correlation with causation, which is, as far as I’m concerned, the sine qua non of anti-vaccine arguments, such as they are. First, B&O (Blaxill and Olmsted) begin with an ominous prediction:

Polio is the iconic epidemic, its conquest one of medicine’s heroic dramas. The narrative is by now familiar: Random, inexplicable outbreaks paralyzed and killed thousands of infants and children and struck raw terror into 20th century parents, triggering a worldwide race to identify the virus and develop a vaccine. Success ushered in the triumphant era of mass vaccination. Now polio’s last hideouts amid the poorest of the poor in Asia and Africa are under relentless siege by, among others, the Bill & Melinda Gates Foundation. Eradication is just a matter of time, and many more illnesses will soon meet the same fate.

But based on our research over the past two years, we believe this narrative is wrong – and wrong for reasons that go beyond mere historical interest. The misunderstanding of polio has warped the public health response to modern illnesses in ways that actually make them harder to prevent, control, and treat.

Polio has always bothered anti-vaccinationists. The reason, of course, is because it is unequivocally a success story for vaccines that demonstrates the power of vaccination to alleviate suffering. Anti-vaccine activists think they are the master of narrative, whereby they tell anecdotes and stories about children seemingly regressing right after vaccines, but opposing that narrative is an even more powerful narrative about the elimination of iron lungs that were so common in the 1940s and 1950s. In an amazingly short period of time, after the introduction of a vaccine for polio, gone were the iron lungs. Gone were the terrifying outbreaks of polio. Gone were the yearly summer warning signs about swimming pools. Most people viewed this as a good thing.

B&O apparently do not.

No, it’s not that they want children to be paralyzed due to polio. What they don’t like is that the success of vaccination campaigns against polio gave scientists confidence that, yes, vaccines can make a difference in the health of children. Of course, we already had the example of smallpox, where vaccination ultimately led to the only complete eradication of a disease in history that I’m aware of. B&O are not pleased about this:

But based on our research over the past two years, we believe this narrative is wrong – and wrong for reasons that go beyond mere historical interest. The misunderstanding of polio has warped the public health response to modern illnesses in ways that actually make them harder to prevent, control, and treat.

The reality, we believe, is that the virus itself was just half the epidemic equation — necessary but not sufficient to create The Age of Polio. Outbreaks were not caused solely by poliovirus – the microbe was an ancient and heretofore harmless intestinal bug — but by its interaction with a new toxin, most often innovative pesticides used to treat fruits and vegetables.

And thus the groundwork for a beautiful case of confusing correlation with causation is laid.

This confusion begins with a description of the invasion of gypsy moths in the mid-19th century. At the time, gypsy moths were a real problem, and a frantic effort to find ways to control them was undertaken. What worked in killing these moths, it turned out, was adding lead to arsenic compounds commonly used to kill other insect pests. This new mixture was effective in eliminating gypsy moth larvae and was apparently deployed on a large scale in New England in 1893 and 1894. Confusing correlation with causation like the masters they are at doing that, B&O opine:

Fortunately – or so it seemed – a scientist working for the commission quickly found a solution. Adding lead to arsenic proved lethal to the larvae, and the new compound was sprayed on trees in and around Boston starting in 1893. It quickly proved its value against not just gypsy moths but all manner of agricultural pests. In fact, it worked better against codling moths, the source of the proverbial “worm in the apple.”

“In the case of insects which do not readily yield to Paris Green, a different substance, used with great success by the Gypsy Moth Commission, with which it originated, may be applied,” wrote George H. Perkins, state entomologist of Vermont in his annual report for 1893, published in early 1894. “This is arsenate of lead; sodic arsenate 29.93%, lead acetate 70.07%, are mixed in water, from which arsenate of lead is soon formed.”[ii]

Something else of note happened in 1893 in the Boston area. Two doctors used to seeing sporadic cases of paralysis in infants became concerned when the small caseload suddenly increased, to 23. There had only been six in the same September-November time span the year before.

Does this sound familiar? It should. Fast forward 100 years, and B&O did exactly the same thing with mercury in vaccines and autism, and they did the same sloppy job. This time around, they are trying to claim that polio epidemics and outbreaks were not due solely to the polio virus itself, but to an unholy combination of the polio virus and arsenic-based pesticides. They make the case explicit by listing a whole bunch of agricultural areas where outbreaks occurred 1893 and 1910 and speculate that some combination of the polio virus and lead arsenate-based pesticides was responsible for the paralysis that polio caused. They even go so far as to spin an elaborate tale about Franklin Delano Roosevelt and his battle with polio. Because FDR was 39 at the time he contracted polio, his was an unusual case. In particular B&O blather on and on about how Roosevelt happened to be in the world’s commercial blueberry capital at harvest time and ominously intone that it must have been more than just the polio virus. It must also have been the lead arsenate used as a pesticide, implying hilariously:

None of this shows FDR eating a mound of fresh blueberries treated with lead arsenate in August 1921, but it seems more probable than not.

Yep, it was the blueberries plus polio that paralyzed Roosevelt, not just polio alone! Why? Because lead and arsenic can cause neurological damage. Seriously. That’s B&O’s argument. Never mind that lead and arsenic poisoning don’t produce a syndrome like polio. Never mind that polio has probably been known for hundreds of years, possibly as far back as ancient Egypt. At the very least, polio was known back in the 18th century. Moreover, polio wasn’t confined to areas where agriculture might bring people into contact with pesticides, lead-arsenate-based or otherwise. It swept through cities as well, including New York City, where 2,500 cases were reported in 1907. In 1916, there were over 2,000 deaths from polio in New York City alone. (Yep, I can read Wikipedia, too.)

In fact, some of the worst epidemics of polio occurred in the 1940s and 1950s, when lead arsenate pesticide use was on the wane. So how do B&O explain that inconvenient observation? Easy. they pivot and switch to a different pesticide. Yes, we’re talking DDT, baby! DDT was introduced in the late 1940s and soon supplanted lead arsenate-containing pesticides. Just as B&O borrowed from the mercury playbook, in which they try to claim that mercury poisoning produces symptoms just like autism and now tried to claim that, because lead arsenate can cause neurological damage, it must have been a combination of lead arsenate pesticides plus polio that really caused those epidemics of infantile paralysis (and hence vaccines didn’t save us). Now B&O pivot seamlessly to DDT, amazing the world at how two such different chemicals can somehow interact with the polio virus in the same way in order to produce paralysis. It’s magic:

In 1949, the same year as the Life article, Drs. Morton S. Biskind and Irving Bieber published “DDT Poisoning – A New Symptom With Neuropsychiatric Manifestations” in the American Journal of Psychotherapy. “By far the most disturbing of all the manifestations are the subjective reactions and the extreme muscular weakness,” they reported.[ii]

In subsequent papers and testimony, Biskind linked DDT directly to cases of poliomyelitis – including a Dec. 12, 1950, statement to the Select Committee to Investigate the Use of Chemicals in Food Products, United States House of Representatives.[iii] He quoted another doctor that “wherever DDT had been used intensively against polio, not only was there an epidemic of the syndrome I have described but the incidence of polio continued to rise and in fact appeared where it had not been before.

Amazing how two different chemicals with entirely different mechanisms of action apparently cause the same thing. The problem, of course, is that, just as mercury poisoning doesn’t cause a syndrome that looks like autism, neither DDT nor lead arsenate produce syndromes that look like polio. DDT, for instance, does not cause paralysis. In fact, its mechanism of neurotoxicity at high doses is fairly well known. The symptoms it causes tend to be mild at anything other than very high doses. In the case of acute exposures to DDT, the weakness and paralysis can occasionally (but unreliably) be observed, tend to be of only distal extremities (hands and feet), and require very high exposures to DDT, exposures far above residues that could possibly be left on fruits and vegetables, even at the height of DDT use. At intermediate exposures, the effects tend to me more along the lines of parasthesias and nausea. Remember, too, that by “low,” I still mean pretty hefty doses relative to residues on foods. Finally, in a very superficial way, if you really, really want it to and squint hard enough at the comparisons, you can make some of the symptoms of DDT poisoning sound a little bit like symptoms of polio, but you have to be as–shall we say?–creative as Age of Autism flacks were when they “mapped” the symptoms of autism to those of mercury poisoning as a means of “proving” that autism is due to mercury poisoning from vaccines.

In fact, in part 5 (and you knew it was coming), B&O try to pull mercury into the mix of lead arsenates and DDT as well. More importantly, they seem implicitly to recognize that their graph of U.S.-only outbreaks with time in part 2 was incomplete; so they fill it in with outbreaks from elsewhere in the world in a graph here:


Note that they put the line at 1893 representing the introduction of lead arsenate insecticides in Boston, but this makes no sense from a scientific or epidemiological standpoint. If they really wanted to make the case, they would have to try to correlate the introduction of lead arsenate insecticides with polio outbreaks for each and every location that they examine. Lumping them together on a graph like this might look persuasive to people with no scientific background, but epidemiologists and scientists would laugh uproariously or shake their heads in disbelief (or both) at such amazing ignorance. After all, what is the rationale for placing a line at 1893? Certainly there is no obvious break point or change in slope of the line that began at 1893; one could just as plausibly, if not more so, place such a line back around 1880 based on the curve alone. However, the line at 1893 draws the eye to it and produces an illusion of a definite break point where the line is–and B&O put that line there based on a historical event, not on any detection of a true break point in the curve. It’s bad math of a sort that Mark and David Geier used five years ago, as was so well explained by Mark Chu-Carroll.

B&O might as well try to correlate the number of pirates with global warming or the introduction of compact discs in the U.S. with the rise in autism. Or the introduction of thimerosal in vaccines with the “autism epidemic.” Oh, wait. They just did.

Truly, B&O are a one trick pony.

So why on earth did B&O engage in such napalm-grade burning stupid, trying to claim that somehow pesticides mixed with the polio virus to produce epidemics of infantile paralysis. The reason, it seems to me, is simple. Their claims otherwise notwithstanding, B&O are profoundly anti-vaccine. Unfortunately for them, the elimination of polio is a major success story for vaccines and vaccine science. Vaccination against polio eliminated polio in the United States in short order, as it does wherever high levels of vaccine uptake can be achieved. It’s an iconic example of how effective a public health measure vaccines are and how many lives can be saved, not to mention cases of paralysis prevented. They have to tear it down if they are to convince people that vaccines are harmful and cause autism. After all, to anti-vaccine loons like B&O, no matter how much they deny it or try to claim otherwise, it’s all about the vaccines. It’s always been about the vaccines. It always will always be about the vaccines. Also, if they can “prove” that some environmental influence like pesticides somehow interacted with a virus to cause epidemics, then, in their mind apparently, they can claim plausibility that vaccines somehow cause autism.

I think. B&O’s “reasoning” is quite unclear on this other than their desire to try to discredit the polio vaccine.

And if it takes some conflation of correlation with causation, either intentional or not, so be it. Whether B&O know what they are doing or not, I really don’t know. If they do know what they’re doing as far as statistical and mathematical analysis go, the only possible conclusion is that they are intentionally trying to deceive. If they don’t know what they’re doing, then the conclusion must be that they’re hopelessly scientifically and mathematically illiterate. I suppose they could be a little of both, but personally I favor the conclusion that they’re hopelessly scientifically and mathematically illiterate. Even “better,” B&O promise more to come:

But none of these reached the scale, or occurred with the frequency, of poliomyelitis outbreaks after 1893 and the invention of lead arsenate. This leads to the second test of our theory – once lead arsenate and DDT disappeared from the scene, why did poliomyelitis outbreaks continue, albeit in diminished fashion?

Stay tuned.

Unfortunately, calling B&O’s crackpot idea about pesticides and polio a “theory” or even a “hypothesis” is stretching the term beyond recognition. More importantly: Does anyone want to speculate how they’ll explain this one? My guess: They’ll try to correlate the introduction of mercury-containing vaccines to areas of the world where polio outbreaks still occur.

ADDENDUM: I can’t believe I didn’t find this in all my Googling, but it’s been pointed out in the comments below that B&O cribbed their entire “hypothesis” from…! In fact, the article is an even more exquisite example of confusing correlation with causation and cherry picking three insecticides to add together to give a curve that looks like the incidence curve of polio in the middle part of the 20th century. It almost might be worth a post itself!

Mr. Blaxill and Mr. Olmsted, you do remember Scopie’s Law, don’t you?