I’ve been thinking about this for the past few days. I recently read an interesting article published last week in the journal PLoS One. I am by no means an expert in autism or epidemiology, but here is my take. In this study Durkin and colleagues show that there is a link between socioeconomic status and the prevalence of autism spectrum disorder. The authors took surveillance data from the CDC which evaluates prevalence of autism throughout the country. Overall they evaluated data from over 550,000 eight-year olds in a two-year period and found about 3500 individuals that would qualify as autistic. Using census data they sorted these cases into three groups based on socioeconomic status. Surprisingly they found that there was a very significant correlation between prevalence of autism and socioeconomic levels such that wealthier areas had relatively more cases of autism (see Figure 1). They further show that this gradient is present across all races and ethnicities examined. One possibility is that this effect may be due to the fact that wealthier communities might have better access to health and diagnostic services. In support of this claim the authors state that in fact the gradient is steeper if you only take into account cases where there was a previous diagnosis of autism before the surveillance data was collected. Meaning that children from wealthier families were more likely to receive a diagnosis during early childhood than children from poorer families. This would lead to diagnostic bias and may explain the correlation between wealth and autism. It also means that autism may be underdiagnosed in young children from socially disadvantaged communities, denying them the possibility of early intervention and treatment. Which is another reason for making health care accessible to all.
That being said, the authors then looked at the group of children that had not previously received a diagnosis of autism but were picked up in the surveillance data. In this case any diagnostic bias should disappear. What they found was that even in this group there was a significant correlation between autism prevalence and socioeconomic status. The authors conclude by saying that “factors associated with socioeconomic advantage might be causally associated with the risk for developing autism.” Which totally baffles and surprises me. I guess that one possible factor that may lead to this correlation might be maternal age, as socioeconomically advantaged women tend to have children at an older age. Or it could be something to do with differences in the type of prenatal care received that we don’t fully understand. However the authors mention in passing something called the “hygiene hypothesis”. Now I’m not sure how much actual evidence there is for supporting this hygiene hypothesis, but what it states is basically that the more you are exposed to a variety of diverse pathogens during early childhood, the less likely you are to develop chronic immune disorders such as allergies, asthma or inflammatory bowel disease. Some have used this view to explain higher prevalence of these disorders in developed countries or correlations between socioeconomic status and immune disorders. The reason this is interesting is that there are some well-established correlations between chronic abnormal regulation of immune system function and several neurodevelopmental disorders including autism. Many molecules which are active in the immune system are also known to regulate normal signaling in the brain and this may explain the association between chronic immune disorders and abnormal brain development. This is of particular interest to me since part of my research involves looking at the role of some of these immune molecules during early brain development. It is important to mention that when I say abnormal immune function I am not referring to the immune reactions which result from vaccination. Immune reactions to vaccines are short-term, and it has been very clearly established that there is no link between vaccines and neurodevelopmental disorders, including autism. What I am talking about is long-term alterations in the levels of various immune system molecules including regulators of inflammation and the so-called MHC markers. So wouldn’t it be interesting if the link between socioeconomic status and autism prevalence was somehow related to the increased prevalence of immune disorders in economically advantaged groups?
Of course this is just a hypothesis, which has no actual proof. In fact I can find several holes in it. For example there is no evidence that someone from a lower socioeconomic status, living in the United States would be necessarily exposed to more pathogens than someone with a high socioeconomic status. I mean, wealthy kids eat as much dirt as poor kids. If you don’t believe me go to a playground in a fancy neighborhood and in a poor neighborhood and observe the kids. They get exposed to all sorts of junk either way. But by making hypotheses scientists are then able come up with ideas for experiments that would either prove or disprove them. What I like about this paper is that it made me think a little outside the box, to come up with ideas and associations that I would not otherwise had come up with. And this is something that is important to do. I have several colleagues who do basic research into the mechanisms of autism and other neurodevelopmental disorders. And everyone is doing very different research, ranging from population-wide human genetic studies, to studying molecules which regulate the communication between brain cells to looking for structural abnormalities using brain scanners. And all of these individuals are doing great science and are very careful with their results and interpretations. Yet despite the fact that they are all studying the same disorder, it’s almost as if they live in totally different, internally-consistent worlds. So in essence, they are all “correct” and all their results are “true”, but if you look at them together then their theories start to look a little more shaky. At least not so consistent with each other.
It’s a little like the movie Rashomon by Japanese director Akira Kurosawa. For those of you who don’t know Kurosawa, he is one of my favorite movie directors and you should see all his movies, particularly the Seven Samurai, which is beyond great. But anyway, Rashomon is a murder mystery taking place in medieval Japan. A woman has been raped by a bandit in the woods and her husband, a samurai, has been stabbed and killed. The movie tells the story of the bandit’s trial and three witnesses recall the events. The first witness is the bandit, the second is the woman, the third is the dead samurai (channeled through a medium). Although the basic facts are the same –the bandit raped the woman and the samurai was killed with a knife– the three stories differ from each other. At the end of the movie we hear a fourth version, from a woodcutter who witnessed the entire thing as he hid in the woods. For some reason he had chosen not come forward during the trail even though his story would resolve all the inconsistencies between the three other stories. Presumably, his version, as a disinterested party, represents the actual events. However at the very end we become suspicious of even his story since we start to suspect that the reason he did not come forward was because he has actually stolen the murder weapon, a bejeweled knife, after witnessing the events. So even he is not a totally disinterested party.
So what’s my point? My point is that even the best hypotheses and ideas in science always have some spin, and that this spin will introduce bias. And it takes a study that makes everyone think outside of the box to bring these various hypotheses together and separate the spin from the facts. Good scientists will gladly let their pet hypotheses evolve or be discarded if evidence shows up to disprove them. Bad scientists will keep looking for specific data supporting their views while ignoring overwhelming amounts of data that don’t support their views. As far as this relationship between socioeconomic status, chronic immune disorders and autism, we’ll see how that one plays out.
Further reading (some may require subscription or a trip to a university library):
Durkin MS, Maenner MJ, Meaney FJ, Levy SE, DiGuiseppi C, et al. (2010) Socioeconomic Inequality in the Prevalence of Autism Spectrum Disorder: Evidence from a U.S. Cross-Sectional Study. PLoS ONE 5(7): e11551. doi:10.1371/journal.pone.0011551
H. Okada, C. Kuhn, H. Feillet and J.-F. Bach (2010) The ‘hygiene hypothesis’ for autoimmune and allergic diseases: an update. Clin Exp. Immunol. ;160(1):1-9.
Cohly HH, Panja A. (2005) Immunological findings in autism. Int Rev Neurobiol.;71:317-41. Review. PubMed PMID: 16512356.