The Spectrum of Autism Research: From Brilliant to Just Plain Stupid

This week, several research papers on autism spectrum disorders were published by press conference, a practice universally acknowledged as hiding one’s research.

One was a brilliant but extremely short paper pre-published in the prestigious Proceedings of the National Academy of Sciences (PNAS) by a team of molecular biologists researching genes associated with autism spectrum disorders. This paper may very well explain why some people are predisposed to developing autism, and the gene in question (MET, a tyrosine kinase receptor) explains the range of neurological, behavioral, and digestive symptoms associated with autism spectrum disorders. It’s a lovely piece of science. More on this paper below.

The other paper was produced by a triumvirate of rogue economists who tried to apply economic statistical theory to autism, a subtle biological condition, and made some of the most astoundingly stupid conclusions that I’ve ever seen. Because they did no empirical science (despite their claims in their paper, which I read, all 67 pages of it,) none of their findings can be disproved in the scientific sense of the word. These three men (Michael Waldman and Sean Nicholson of Cornell, and Nodir Adilov of Indiana University) wound through an amazingly vapid intellectual argument to arrive at what might be the exactly wrong conclusion.

Here’s what those idiots did: they selected a few counties (in California, Oregon, and Washington,) and looked at precipitation records for those areas between 1972 and 1989. That’s right, rain. They found that autism rates incrementally increased more in rainy counties than in less rainy counties.

They then inquired about cable television subscription rates in these counties and found that cable television subscription rates also increased, an astounding finding that one could have figured out by being alive and conscious. In a later test involving counties in California and Pennsylvania, they correlate those areas with the most cable TV subscriptions to county autism rates.

So, they blamed the increase on autism on increased TV viewing.

Here’s only a few of the problems with that data:

Autism rates increased substantially during the time period studied. If the authors had used any increasing variable, they would have found the same correlation.

The Dow Jones Industrial Index increased from 900 to 2300 in that time period, which positively correlates with the rise of autism rates.

The per capita income in the U.S. grew from around $15,000 to $23,000 in that time period. Autism thus correlates with increased income.

Childhood obesity increased from around 7% to about 11% in the same time period.

The feral and domestic cat population in the U.S. doubled from 30 million in 1970 to 60 million in 1990, which positively correlates with the increase in autism diagnoses.

In 1970, only 24% of mothers with children under the age of two worked even part-time. In 1984, that number had nearly doubled to 46.8%. Today, 55% of women with children under one year old work full-time. Consider that statistic. Now consider this line of reasoning:

In 1970, there were very few two-income families. Increase in working women correlates to an increase in two-income families. Two-income families can better afford to subscribe to cable television, which explains the increase in cable TV rates.

However, when women work, they do not stay home with their tots and watch TV. The tot goes to daycare, where most of the time the tot is engaged in structured play and activities, not television watching. Thus, TV-watching by tots may have decreased during the time that autism rates increased, suggesting that TV has a protective effect on autism.

That line of reasoning is every bit as well supported as the line of reasoning in the Cornell-Purdue paper. What’s more, it incorporates data that the three men never thought of: whether or not the televisions were actually watched.

In addition, the three men who wrote this article fail to account for excellent research that shows that the neurological damage associated with autism begins before birth. Their conclusions actually should state that pregnant women watching television causes autism.

This paper is an excellent example of non-scientists (they’re all economists!) trying to apply non-scientific principles (economic statistics) to science and then even forgetting to apply their own rules. Rarely in economics does one find monocausal patterns. There is almost never one cause that produces one effect. This specious attempt to blame autism on television viewing by infants is an example of using the wrong tools and doing the job badly.

The molecular biology paper, on the other hand, produced by Daniel B. Campbell, James S. Sutcliffe, Philip J. Ebert, Roberto Militerni, Carmela Bravaccio, Simona Trillo, Maurizio Elia, Cindy Schneider, Raun Melmed, Roberto Sacco, Antonio M. Persico, and Pat Levitt, is a exquisite piece of statistical modeling that explains nicely why some people are more prone to developing autism spectrum disorders. The candidate gene, MET, is a tyrosine kinase receptor, which means that it is a telephone for conducting signals inside a cell, and MET signaling can result in the cell reproducing, changing, or dying. MET participates in brain growth and maturation, immune function, and repair of the digestive system. Children with autism often have symptoms of disturbances in some or all of these systems. This research ties together these disparate symptoms and explains why children with neurological symptoms often have diarrhea or immunological problems.

There are two common alleles (forms of the gene) for the MET gene. In this case, the change in the gene results in more or less of the protein (and thus the tyrosine kinase receptor) being produced in the cells. Remember that each person has two copies of a gene (one from your mom and one from your dad,) so you have two flips of the proverbial gene coin.

The C allele is associated with an increased risk for autism. The C allele makes less of the MET protein. In this case, less is bad.

The G allele is protective against autism. The G allele means that more MET tyrosine kinase receptor protein is made and thus there is more met signaling.

The relative risk of being diagnosed with autism was 2.27 times higher if you have two copies of the C allele (the CC genotype) and 1.67 times higher if you have one of each allele (the CG genotype, heterozygous) compared with having two of the protective G alleles (the GG homozygous genotype).

There you have it: an excellent autism paper and piece of scientific-sounding poop. I desperately hope that people won’t go bonkers over restricting television viewing for their kids because they believe that, otherwise, the kids will get autism. There are a lot of reasons to moderate television viewing, but fear of autism isn’t one of them.

I also hope that parents of autistic kids won’t beat themselves up over their television viewing because they erroneously think that they caused their kids’ autism. That statistics paper should only be used for potty training.