By Mark Blaxill
It's tricky business to report on scientific events in which you've played a direct role, but there's a new study coming out on Monday about trends in autism rates in California and their connection with falling thimerosal exposure. It's important and timely for me to comment on the findings. But first I need to go back to the beginning of this discussion, since it's one that I started.
On July 16 2001, I gave a brief presentation to a meeting of the Immunization Safety Review Committee of the Institute of Medicine at the Meeting on "Thimerosal-Containing Vaccines and Neurodevelopmental Outcomes" held in Cambridge MA. I won't repeat the entire story, but if you're interested, David Kirby provides an engaging account of this session in his book Evidence of Harm (pages 175-181) and you can find the audio files of the meeting at the following link HERE.
During a fifteen minute time slot I made a number of comments and answered one interesting question. Looking back from today, six and half years later, here's a quick synopsis of what I said.
"It's a fact that US infants were exposed…to sharply higher amounts of mercury via thimerosal-containing vaccines (TCVs) starting around 1990. And that the timing of the increases in autism rates that are measurable at all and the increases in infant mercury exposures via those vaccines are associated. I make no claim about causality; I…simply observe the association."
I went through a quick overview of some of the autism time trend data (I later published an epidemiology review on this subject entitled "What's Going On? The Question of Time Trends in Autism" in the peer-reviewed journal Public Health Reports). I also sketched out some rough analysis on the broad contours of mercury exposure via TCVs in US children based on the CDC vaccine coverage statistics. Next, I brought these two analyses—autism rates and mercury exposures--together.
"I will confess that this is kitchen table research. Not well funded and the data sets are certainly vulnerable along a number of dimensions [Note: I never attempted to publish this work, although CDC consultant Paul Stehr-Green later published it for me, in order to attack it]. But I've taken the California data set, which is the one continuous data set over time; it's a case count data set. I've converted it into birth cohort prevalence [by birth year] by using the census information on births in California. And I've also taken a weighted average mercury exposure based on the coverage rates and the average rate of mercury that would be in those vaccines (with a number of adjustments). If one overlays those two trends…one observes in fact that there is an association in time. I would be the first to confess to you that this is not a perfect and squeaky clean data set. This is simply the data I could find."
At the end of the presentation Chairperson Marie McCormick allowed only one question. It came from one of the panel members and it was an important one.
Panelist's question: "If you believe that there is a direct connection between the incidence of autism and these vaccines…should we in the next three years begin to witness a rapid decline in the number of cases of autism?"
My answer: "We have a running experiment going on right know, so we'll know, or we will have a better idea actually [emphasis added] in a number of years."
Natural experiments, like the running one in California, are not the same as laboratory experiments. There are a lot of things you can't control and confounding exposures you can't predict. But nonetheless, the running experiment in California, one I first pointed out back in 2001, has been a very important one. And the evidence that emerges from the natural laboratory demands notice and respect, just like every other bit of reliable evidence.
The results from the experiment? The autism numbers in California have been going up. Straight up. Without any visible deceleration, let alone decline, for the entire period since then.
So what are we to make of all this? And, since it was I who first raised this natural experiment, what do I make of all this?
First of all, I'd emphasize one feature of the autism-thimerosal hypothesis that is often overlooked. As a causal hypothesis for the increasing autism rates, it is a highly parsimonious hypothesis. By that I mean the causal theory was specific, testable and extremely simple to evaluate. (The Oxford English Dictionary defines the "law of parsimony" as "the logical principle that no more causes of forces should be assumed than are necessary to account for the facts.") From the point of view of children's health, it would have been wonderfully good news if the most parsimonious form of the hypothesis had been proven correct. The single causative agent would have been removed, autism rates would have plummeted, future families and children would have been spared the heartbreak so many of us have suffered, and the childhood immunization program (not to mention global manufacturing and its associated emissions) could continue forward secure in the knowledge that a tragic mistake had been discovered and corrected with limited future repercussions, excepting a crucial but simple question of justice for the injured children and their families.
As it turns out, the problem is more complicated than that.
Tomorrow, a study by Robert Schechter and Judy Grether will be published in the Archives of General Psychiatry. They report what I have known for several years now: that the rates of autism in California (properly adjusted for birth year and age at time of data collection) have gone steadily upward. For the most part, they conduct the analysis in the right way and use simple and accurate displays. Their analysis is robust; I've done the same calculations they do and get the same result. They've even adopted my display technique in their first two figures (without crediting me, unlike Dan Hollenbeck's excellent work at FIGHTING AUTISM, which described this approach as "the gold standard"). They do cite my IOM presentation and make one valid criticism of my "kitchen table" analysis. Appropriately, they make no effort to dismiss these increases as an artifact of better diagnosing. The evidence is what it is.
Sadly, alongside the study by Schecter and Grether the journal's editors have published a triumphalist editorial by Eric Fombonne, the chief epidemic denier. I've made my views on Fombonne's work very clear, and his editorial makes every unscientific, propagandistic leap that one would expect from someone who cares so little about autistic children and so much about defending his past errors. Read Schecter and Grether's paper, but throw Fombonne's in the trash.
As someone who has actively pursued, and yes promoted, the autism-thimerosal hypothesis, where do I think this leaves us? The subject is too complex to take up in a single blog post, but here are some of the most important implications:
• The rising rates of autism have truly become a national emergency. The rates reported by Schecter and Grether are NOT for all autism spectrum disorders, but what California parent Rick Rollens has consistently described as "full syndrome autism." For children born in the late 1990s, even these rates had risen to more than 1 in 250 children. If that's not a public health crisis, I don't know what is.
• The continued increases in autism rates provide strong evidence against the idea that early thimerosal exposure, and only thimerosal exposure, is causing the increased population rates of autism. It doesn't say thimerosal is safe, it simply says that explaining the tenfold increase in autism rates won't be easy. And for anyone who thinks clearly and compassionately about the children and families, that's not a good thing.
• Put another way, the epidemiological analysis doesn't prove that thimerosal exposure cannot cause individual cases of autism. It simply provides evidence that it's unlikely thimerosal can be the sole cause in all cases.
• The evidence regarding the connection with autism and mercury exposure more generally hasn't changed. The plausible theories regarding the biology of excretion and toxicity of mercury compounds and their different effects on the developing immune system, gastrointestinal system and brain haven't been affected a single bit.
• The evidence that many of us have seen connecting thimerosal exposure and increased population risk of numerous neuro-developmental disorders doesn't go away. The conduct of the CDC investigation into the adverse effects of thimerosal doesn't look one bit better in retrospect.
• Most importantly, and this is a point that escapes Fombonne's Lilliputian mind entirely, the evidence from the California natural experiment doesn't exonerate the broader childhood immunization program. Quite the opposite, it brings the overall escalation in the vaccine program even more strongly under suspicion. While thimerosal was removed from hepatitis B, Hib and DPT/DTaP vaccines (but not from influenza vaccines), the vaccines themselves didn't go away. Indeed the total count of vaccine doses has gone from 15 to 45 by the first grade and this increased exposure is strongly associated with the increases in autism rates.
• The importance of intensive investigation into the environmental causes involved in autism becomes even more important. We need better evidence, deeper understanding and effective treatment ideas now more than ever.
• And the questions of the natural history of autism that my AOA colleague Dan Olmsted has so vigorously pursued becomes more important than ever. Thimerosal may not do all the explanatory work some of us hoped it might do, and autism won't be as simple to eliminate as acrodynia, but the close association in place and time with ethyl mercury's commercial introduction and the beginning of The Age of Autism give it a special role as a model molecule in the biology of autism.
It's also worth pointing out that none of us really anticipated the strength of the confounding factors that we'd see as thimerosal was phased out of some childhood vaccines. We didn't know that the CDC would turn around and reverse the ban on infant thimerosal exposure but would instead continue to expose infants to dangerous amounts of mercury in flu shots required for both pregnant women and infants. We didn't know that manufacturers would simply substitute the mercury with another toxic metal, aluminum, in even greater amounts. We didn't consider the rising environmental mercury burden from coal burning in China. None of this nullifies the results of the natural experiment in California; it simply exposes the different motivations of the different sides of the debate. One side wants to protect their programs, the other wants to figure out why so many children are sick.
It's important to develop compelling theories and even more important to test them. When you have conflicting evidence, you need to deal with all the contradictions and complexities involved and do your best to make sense of all the conflicts; you can't simply toss out the evidence you don't like. When evidence doesn't fit your theory, your theory has to adapt to fit the evidence, not the other way around. To the extent that we may have emotional and/or reputational stakes in certain theories, that may make dealing with conflicting evidence hard for some. But intellectually speaking, it's all pretty easy. Evidence trumps theory. It's as simple as that.
So although for years I hoped that the numbers in California would turn down, that hasn't happened. And therefore, the only intellectual commitments I can make right now with any confidence include some of the following beliefs. Autism rates have exploded in short period of time. Something new and terrible has happened to a generation of children and environmental causes provide the only reasonable explanations. Mercury exposure is bad for children's brains and there's a lot of evidence (with more to come) that both vaccines and mercury exposure are part of the autistic mix. And anyone who thinks the causation problem is simple is sadly misguided, even more so when they use negative evidence as an occasion to celebrate.
As someone with a restless appetite for a theory to guide the life, treatment and prognosis of my daughter, that's an uncomfortable place to be. Quite honestly, I've lost a lot of sleep in the last several years trying to make sense of all this conflicting data. In some of my darker moments, I have found support in this quotation from my favorite philosopher, Karl Popper. Popper is best known for his ideas on the importance of beating up scientific theories with falsifying evidence. But he's less well known for the respect with which he treats those who offer bold conjectures on scientific problems. Here's one of his bits of advice for the ambitious theorizer.
"He who gives up his theory too easily in the face of apparent refutations will never discover the possibilities inherent in his theory. There is room in science for debate: for attack and therefore also for defense...But do not give up your theories too easily--not, at any rate before you have critically examined your criticism."
So I, for one, haven't given up on the mercury-autism theory. Far from it. It's just that we all have a lot more work to do to discover the possibilities inherent in the theory
Mark Blaxill is Editor At Large for Age of Autism and is a director of SafeMinds. This post represents his personal views and is not intended as the official position of SafeMinds. A formal SafeMinds response to the Archives of General Psychiatry study is forthcoming.