Epidemiology is called the ‘basic science of public health,’ but its contribution to this goal is constrained by a preoccupation with supposedly universal exposure-disease relationships that impedes consideration of the contexts in which exposures occur. … [Because] exposure-disease relationships are not self-contained, homogeneous or independent phenomena, they constitute an inadequate object of epidemiological science (Wing, 1994, Medicine and Global Survival, 1, 74-86).
When it comes to the great Autism-Vaccine debate, I don’t know about you but I feel as though I’ve been watching an aggressive never-ending tennis match that’s given me one hell of a crick in my neck. On the one side, parents and like minds arguing tooth-and-nail that vaccines cause autism; on the other side, the CDC, doctors, and the high-functioning autistic community just as fervently arguing there’s no connection. And here I sit in the middle along with other relatively calm confused people, watching both sides talk themselves blue in the face, vainly attempting to convince the opposing side they’re wrong. (As an aside, how often does that approach actually work? Just curious.)
Very truthfully I can tell you that I don’t fall with either side. I don’t assume vaccines share links with autism etiology, nor do I discount it a priori.
“–But wait!” you say to me. “What do you mean ‘a priori’? Studies have already proven that vaccines share no links to autism etiology, they are not harmful but instead disease-preventative inoculations, and there need be no further research on the topic.”
Well, you can just call me a scientific stickler I guess, because to me epidemiological studies generally don’t answer questions but raise them and are meant to lead lab-based science in the directions it needs to go to answer those questions. (Mind you, this isn’t a criticism of the method but merely an insistence that it has its place but is often misused and over-stretched.) We’ve done plenty of autism-vaccine epidemiological studies but where are the animal studies, the in vitro work, the basic gold standards of our biological science which are meant to immediately follow? Sadly, there is a current dearth of those kinds of studies.
I want you to take a second to refer back to the quote I’ve provided above. In a nutshell, it describes the gross limitation of epidemiological studies of disease-exposure scenarios, which is what any link between autism and vaccines hypothetically is: an exposure which triggers or worsens a disease.
In science, when we have an idea we propose two hypotheses. One, the alternative hypothesis, states our primary prediction, in this instance that “Vaccine exposure shares a relationship with autism.” Our other hypothesis is called the null hypothesis, which would state that “There is no relationship between vaccine exposure and autism.” We subsequently design an experiment to test our hypotheses. Should our results prove statistically insignificant we say, “Our results fail to reject the null hypothesis.” This doesn’t mean that we’ve disproven our original alternative hypothesis, au contraire. In science, all we can say in this instance is “We still don’t know.” And if we wish to have our answer it requires further testing. The logic of this lies in the fact that our capacity to measure a true effect is often very dependent on the means we choose to measure it, which many times may be completely inadequate and provide us with a false negative. Therefore if we concluded from negative results that our alternative hypothesis isn’t true, our conclusions would frequently be wrong, hence the statistical caution.
Some researchers say that a hypothesis test can have one of two outcomes: you accept the null hypothesis or you reject the null hypothesis. Many statisticians, however, take issue with the notion of ‘accepting the null hypothesis.’ Instead, they say: you reject the null hypothesis or you fail to reject the null hypothesis.
Why the distinction between ‘acceptance’ and ‘failure to reject?’ Acceptance implies that the null hypothesis is true. Failure to reject implies that the data are not sufficiently persuasive for us to prefer the alternative hypothesis over the null hypothesis [ref].
A number of epidemiological studies have been carried out and many have failed to reject the null hypothesis. These results have generally been taken in support of the null hypothesis. As hopefully I’ve begun to make clear, this is bad bad science.
Granted, in defense of these studies, if there truly is no relationship between vaccines and autism, then you could go on performing epidemiological studies ad infinitum with a hypothesis that can never be proven false. So, in reality, there must be a limit at which point we say that so many failures to reject the null must indicate a dubious link.
However we come back to the problem of the methods in which we choose to study an issue. As Wing (1994) wrote, epidemiological studies are a poor means for delineating exposure-disease relationships because at such a macroscopic level of analysis important details are completely washed away, details which can make the difference between statistical significance or insignificance. And while some animal studies have been performed to date, the results of these studies (those at least which appear halfway reliable) have not been used to inform our opinions but rather the larger epidemiological ones. Below is a listing of some of the major epidemiological studies on the issue to give you some examples.
Kaye et al. (2000). BMJ, 322, 460.
Chen et al. (2004). Psychological Medicine, 34, 543-553.
Dales et al. (2001). JAMA, 285, 1183-1185.
Honda et al. (2005). Journal of Child Psychology and Psychiatry, 46, 572-579.
Geier & Geier. (2004). Medical Science Monitor, 10, PI33-39.
Contrary to media representation, there have been a couple animal studies which have addressed some of these issues. For instance, Hewitson’s group has focused their efforts on primate studies, looking first at the effects of thimerosal-containing Hepatitis B vaccination on various rhesus macaque neonatal behaviors, finding that in exposed infants rooting, snouting, and sucking behaviors were significantly delayed following exposure [ref]. The group went on to study aspects of brain develop in primate neonates exposed to the complete vaccination schedule utilized for American children from 1994-1999, finding that amygdala volumes were overall decreased as well as the binding capacity of diprenorphine [ref]. –Mind you, it is concerning that Andrew Wakefield is attached to both these studies, considering he has proven himself scientifically untrustworthy. However, Hewitson is a good and honest researcher and for that reason I am willing to give these studies a chance.
Other studies have also been performed utilizing animal models. Hornig et al. (2004) for instance exposed autoimmune disease-sensitive mouse pups (SJL/J strain) to thimerosal-only and thimerosal-containing vaccines, and then compared their development to identically-exposed autoimmune disease-insensitive strains (C57BL/6J and BALB/cJ), as well as within-strain controls. The researchers found no differences between the thimerosal-only and thimerosal-containing vaccine groups and therefore condensed them into one. According to the researchers, thimerosal-exposed SJL/J mice exhibited general hypoactivity, although spatial memory and gross motor coordination were insignificantly affected. Sadly, the most interesting “results” of this study were based upon poor stereological methods, nullifying whatever conclusions these scientists had derived (i.e., increased cell packing density within the hippocampus). Rather than using anatomical boundaries by which to determine cell density of the hippocampus and its subregions, the researchers used basic geometry to divide the region, a fatally flawed approach. For those seasoned in the histological study of neuropathology, such approaches are horribly biased and can lead to false results. What makes this even sadder is the fact that no groups have attempted to replicate this work, leaving a good idea to simply rot on the scientific shelves.
There is a dearth of good science addressing whether vaccinations play some role in the etiology of cases of autism. Severe autoimmune reactions following vaccination are noted in the literature (see Guillame-Barré syndrome following influenza vaccination, for instance) and so it is clear that inoculations are not necessarily innocuous but designed to promote an immune reaction. In short, it’s why they’re successful as vaccines. But for as much public attention as this debate has drawn, it hasn’t drawn the right kinds of scientific attention, that truly in-depth unbiased analysis which hard questions like this seem to require.
There have been plenty of epidemiological studies. And those people who wish to believe that vaccinations share no relationship with autism have been eager to twist the commonly-used phrase “These results do no appear to support a link between autism and vaccines” into “These studies prove there is no relationship between autism and vaccines.” I can promise you, these statements are not equal. The former suggests that further in-depth study is necessary; the latter implies that questions have been answered and no more research is needed. On the flip side, proponents of autism-vaccine links have been all too eager to use shoddy research as a foundation for their arguments as well. Both sides have been misusing science for personal agenda, which does an extraordinary disservice to the autistic people both groups are professing to help.
Simply on a humanitarian stance, I prefer that vaccines share no relationship with autism. Who after all would want something like this to occur and forever change a child’s life? However, enough parents have reported curious circumstances regarding their children’s behaviors immediately following vaccination that concern is warranted. On the other hand, we also should be not assuming that there is a causal link before appropriate studies have been performed, especially considering how vital herd immunity is to keeping many diseases far worse than autism at bay. While much may be at risk should a relationship become apparent, much is at risk through the refusal of childhood inoculations as well. This is a matter which must be handled delicately, cautiously, and calmly. Emotion has its place, but it often doesn’t make us rational judges but can blind us to important evidence.
I reiterate that I am not saying I believe autism and vaccinations share a relationship with one another, nor am I saying I believe they do not. What I’m trying to bring light to is the fact that the science which has been used to address this question, in my perception, has been woefully inadequate in its capacity to answer it. We need more lab-based studies, ones which look at exposure and then gauge, not only changes in animal behavior, but changes in brain development. And studies, I would stress, which comprise scientists who are well-trained and offer reliable results; not the slapdash methodology which comes from inexperience. The benefit of animal studies is the reduction of variables which must be accounted for, offering the researcher a better means to judge cause and effect in minute detail. Epidemiology offers no such benefit.
While my area of study is indeed autism, the topic of this post doesn’t fall under my purview of expertise and so if I have failed to report on important issues, I invite you to comment below as I’m always eager to learn more. I also ask that anyone preferring to comment, given the heated nature of the debate that surrounds this topic, to please post respectfully towards me and any other posters. I realize it can feel like life-and-death for some, to which I offer my utmost empathy, but congenial communication will gain you more listeners. Thanks in advance and thanks for reading.