This article was originally published at the NeuroLogica Blog and is re-published here with permission.
By Steven Novella, MD
A new study just published in the New England Journal of Medicine, Early Thimerosal Exposure and Neuropsychological Outcomes at 7 to 10 Years, does not support a correlation between mercury in vaccines and neurological damage. It adds to the growing evidence that vaccines are safe and they do not cause neurological disorders. This study did not look at autism (a study that will be published next year looks, again, at vaccines and autism), but the mercury-causes-autism crowd are still unhappy with the results.
I have been following this issue closely for several years. Although my awareness of the issue goes back much farther, I started to seriously research the claim that the MMR vaccine, or that thimerosal in other vaccines, causes autism while researching an article on the topic for the New Haven Advocate. As a physician (a neurologist) and a skeptical activist I knew I had to get this issue right. I certainly did not want to falsely stoke the flames of public fear, nor did I want to cast myself in the role of denier.
Early on in my research I really did not know which way I was going to go with the issue. Should my bottom line be that there is real reason for concern here, that there is nothing to the claims, or that we really don’t know and will have to just wait for further research? But after reading through all the claims on both sides, and all the research, it was an easy call – vaccines, and specifically the MMR vaccine and thimerosal, do not cause autism, and the alleged autism “epidemic” is likely just an artifact. Those claiming there is a connection were drowning in conspiracy thinking, logical fallacies, and blatant pseudoscience. Meanwhile every piece of reliable clinical data was pointing in the same direction – no connection.
But still, while I was confident in my final conclusion, a small connection between mercury and autism that eluded the existing data could not be ruled out. Or perhaps there was an angle to the whole story that we were missing but would later come to light. I have had enough experience with scientific medicine to be humble in the face of the complexity of medicine and biology. The only rational position is to remain open to new data and new ideas. On this issue there was sure to be more studies in the future – and the ultimate test of the thimerosal-autism connection, the removal of thimerosal from the vaccine schedule, was yet to be seen. So I confidently plunked my nickel down and waited for the future to unfold.
Everyone likes being right, and sometimes this desire clouds our judgment. I have learned, therefore, how to cheat, which is to say how to always be right. All you have to do is say that your position is based upon the existing data, but is contingent upon the results of future studies. In other words, the “right” position is to change your final answer to accommodate new evidence as it comes in. Therefore the only “wrong” answer is to stick to your original position despite new evidence that contradicts it.
OK – so this is just restating how science is supposed to work, but it is amazing how many people forget to cover their behinds with this simple rule. Usually this is because they are not doing science – they are taking an ideological position, and ideology is inflexible. This is a huge advantage for science over ideology, and why, when science and ideology clash, science is almost always right.
In the case of vaccines and autism, since writing my first major article on the topic, the data has come in all consistent with my original position (so I get to be doubly right). Removing thimerosal from vaccines did not decrease the incidence of autism (or decrease the rate of increase in new diagnoses). And several new major studies came out all showing no association – not counting the utter crap being produced by the Geiers.
Enough had happened to warrant an update, so I jumped at the chance when Ken Frazier of the Skeptical Inquirer asked me to write an article on the topic. My article will be out in the next issue, along with a couple of smaller ones on the same topic, so take a look.
Alas, as is often the way in the world of science, my paper is outdated before it even goes to print. This new study, of course, will not be covered in the article. But that’s what blogs are for – instantaneous news and analysis.
Actually, several of my fellow science bloggers have already beat me to the punch. They cover the article, and the response to the article by the mercury crowd, in great detail, so I will not duplicate it here.
Orac at Respectful Insolence goes over the press release of A-CHAMP (one of the mercury militia) that attempts to dismiss the study. He shows that, while the study (like all studies) has its weaknesses, it does add significantly to the body of evidence showing that vaccines are safe. The complaints of A-CHAMP are either wrong, overblown, or inconsistent with their prior positions and shows that they are just trying to tear down the study at all costs.
For the record, I agree with Orac that the study is good enough that it’s conclusions add to the cumulative data on this topic, and that the weakest element of the study was the 30% compliance rate. In other words, only 30% of the subjects that they looked at for the study made it into the final analysis. This opens up the door for selection bias. Orac is probably correct that this bias would likely overestimate a correlation, not underestimate it, but you can never be sure with such things.
I will add that the 30% figure is not as bad as it first seems. The study reports:
Of 3648 children selected for recruitment, 1107 (30.3%) were tested. Among children who were not tested, 512 did not meet one or more of the eligibility criteria, 1026 could not be located, and 44 had scheduling difficulties; in addition, the mothers of 959 children declined to participate. Most of the mothers (68%) who declined to participate in the study and provided reasons for nonparticipation cited a lack of time; 13% reported distrust of or ambivalence toward research. Of the 1107 children who were tested, 60 were excluded from the final analysis for the following reasons: missing vaccination records, 1 child; missing prenatal records, 5; missing data regarding weight, 7; and discovery of an exclusionary medical condition during record abstraction, 47. Thus, 1047 children were included in the final analyses. The exposure distribution of the final sample was similar to the exposure distribution of the initial 3648 children selected for recruitment in the study.
So about 40% of those that did not make it into the final analysis simply could not be located – this is unlikely to represent a bias. But this is a small point.
Isles from Left Brain/Right Brain points out that Sallie Bernard (a believer in the mercury hypothesis) was consulted on the study design and execution and did not criticize its methods until after the results came back negative. That kind of behavior instantly sacrifices all your credibility in the science club.
Kristina Chew at AutismVox reiterates what I did above – that the mercury-causes-autism ideologues are always asking for more studies, but refuse to change their position when the studies they ask for come out. They are waiting for studies that show what they want them to show – we call that “cherry picking.”
The other side is busy too. David Kirby (a bad journalist who desperately wants to be a bad scientist), who wrote the book Evidence of Harm, published an article online in that absolute rag, The Huffington Post (to be fair, they also published this piece by Arthur Allen defending the paper’s conclusions). Kirby repeats all the same points in the A-CHAMP press release, but he emphasizes the fact that the study found some neurological symptoms that were higher in the subjects who received more thimerosal. Kirby proceeds to completely misinterpret the significance of this.
The study looked at 42 different outcomes, and set the p-value for significance at 0.05. A vague concept of basic math should be sufficient to see that some outcomes will reach significance by chance alone. The researchers arguably should have adjusted their statistics to account for the fact that they were looking at 42 variables – but instead they just looked at all the outcomes that were significant to see if there were any patterns or trends. What they found was that there were a few scores that were worse among those exposed to more thimerosal, but there were also a few scores that were better. There was a random distribution of positive and negative effects that essentially average out to no net effect. It’s all just noise. (Is there a small signal hiding in the noise? There could be – scientists have to be honest about that. But that doesn’t mean there is.)
What Kirby does is not just really dumb, it’s despicable. He cherry picks all the negative (meaning bad) neurological outcomes and pretends that the study shows a correlation (it doesn’t, when you look at ALL the data). He then tries to dismiss the positive (good) outcomes as absurd. He mockingly writes:
If they (the CDC) really mean that thimerosal increases IQ levels in males, then sign me up for a double-dose flu shot this year.
No, David, they don’t mean that. Not by any stretch of the imagination. It takes incompetent statistical analysis or the blindness of ideology to write something so ridiculous. What the CDC means is that the study does NOT show that thimerosal increases IQ, nor that it causes motor tics, or improve motor skills, or decrease language skills, or anything else. The study showed no correlations because it all averaged out as noise.
This is, by the way, the same mistake that astrologers make (remember that crusty pseudoscience?). They look at many variables then cherry pick the outliers. At best what this study might show is a possible correlation, but any such possible correlation would have to be corroborated by a later study (with fresh data) that looked specifically at that one variable.
So the pattern that I found when I first started looking at this issue – that all the reliable data was on the side of no correlation between vaccines or mercury and autism or neurological disorders – continues to hold up to new data. The other pattern I noticed – that those promoting a correlation were relying on bad science, logical fallacies, and ad hoc conspiracy claims – also continues to hold.
In the last few years every new study showed no correlation, and the mercury militia responded with abject nonsense and dismissal. This cycle seems to be repeating itself over and over, and this latest study is no exception.