What do Autism and Milli-Vanilli have in common?

Very little as you might have guessed, but recent popular media articles with clever titles like "Autism: Blame it on the Rain" I get the feeling I'm going to be just as disappointed as when I learned those guys were lip syncing the whole time.  Anyway, the articles go on to talk about an epidemiological study that shows a correlation between counties in west coast states that get a lot of rainfall and increased incidences of autism.  And as I suspected, I am disappointed, not in the research that inspired the popular press articles, but that most of the popular press articles seem to gloss over the fact that this study is VERY PRELIMINARY.  The authors of the study are not saying there is a definite link between living in a rainy area and the chances that your kids have autism.  Here's what I mean.  Before anyone gets too agitated and starts thinking of moving to the Arizona desert, let's consider one of the most famous adages in science: "correlation is not causation".  What that means is: just because two things are correlated doesn't mean they have anything to do with one another.  A great example of this comes from the venganza.org website, home of the church of the flying spaghetti monster, which has a figure correlating the decline in the number of pirates with the global increase in average temperature. 

Does this mean that we should assume global warming is responsible for the decrease in piracy we've seen since the 1800s?  No, it is simply a correlation, and there is no good evidence to suggest that increases in the average yearly temperature have any affect on how many pirates there are.  Or that pirates somehow cool the globe, and their absence is what has caused the world to get warmer. Given what we know of history and climate science, it is much more likely that the increases in technology and industry since the 1800s led to increased air pollution (and increased levels of greenhouse gases) which has led to steadily increasing average temperatures. Conversely, technology also allowed for improved national navies and merchant marines that have had more success in capturing and imprisoning or killing pirates, or evading them.  Additionally, the development of air travel meant less and less people and valuables were transported by sea.  As the cost to benefit ratio of being a pirate became more costly than beneficial, less and less people opted for it as a career.  (And of course, if we look at the recent increase in pirates coming out of Somalia, we see no decrease in temperature to account for this absurd relationship, but the dire circumstances in Somalia and the millions of dollars in ransom these pirates have been able to amass suggests there is something to the cost/benefit idea.)  Anyway, the fact that the only thing this study shows is a correlation with no evidence of causality is not the only strike against an argument for relating rainfall to autism.  Apparently, there were several issues with the methodology of this study that make its results sketchy at best.  An article at scienceline.org actually does a great job of explaining everything, but just to give the highlights here:

Despite the statistically significant findings, a careful reading of Waldman’s article published in the November issue of Archives of Pediatrics and Adolescent Medicine results in more questions than answers.

Could the rain-autism correlation be an artifact of another relationship? Traveling through Oregon or Washington, for example, a distinct trend emerges: the further east, the more rural and dry. It may be that urbanized counties do a better job of reporting autism cases, and that these simply happen to be the same counties with more rainfall.

Teasing out other causes, like ruralness, is difficult in an epidemiological study. This is one reason why Waldman’s research design is the “weakest type of epidemiological study,” according to Irva Hertz-Picciotto, an environmental epidemiologist at the University of California, Davis, also uninvolved with the study. Rather than examining exposures of individuals, ecological studies compare whole groups, which means information that could help decipher the true cause-effect relationship is lost.


Further, the grouped data in Waldman’s study came from a combination of sources — from state agencies to regional health centers — leading Hertz-Picciotto to suspect the numbers are not comparable within or between states. “Often the [agency or center] will put the children into the categories that they know they can provide services for,” Hertz-Picciotto says, “even though the categories may not be the best description of the child’s diagnosis.”

All of that aside, does this study tell us anything important?  Despite my somewhat rough treatment of it here, the answer is actually yes.  The finding that the incidence of autism may be correlated to the amount of rainful an area receives could lead to the discovery of some other factor that is related to both autism and rainy days, and thus, provide an explanation for the correlation.  For example, there is a psychological disorder called Seasonal Affective Disorder (SAD) which is more commonly known as the "winter blues".  When the days are shorter, and we are exposed to less natural light, and it is hypothesized that people with SAD produce less melatonin and serotonin (a hormone and neurotransmitter, respectively, the former most notably characterized as helping to regulate sleep/wake cycles, while the latter is the neurotransmitter most often enhanced by antidepressant medications).  It may be possible, that children who have a genetic predisposition to autism spectrum disorders may be sensitive to getting less light in a similar fashion as people who have SAD, these kids may have brains that produce less serotonin or melatonin, and this may, in turn affect the development of their brains.  Alternately, it may be that kids who fall on the spectrum may become more antisocial or have a more difficult time with vocal development because they are forced to stay indoors more, watching tv and playing video games rather than interacting with other kids.  Both of these ideas are completely speculative and are not substantiated by any hard evidence.  But it is important in science not to discount any ideas until they have been tested, and disproven.  And the benefit to the study mentioned in these articles is that it gives us a reason to hypothesize about the role of light exposure or time spent isolated indoors as factors that could affect autism, and better still, reason to test these ideas so that we will one day know for sure whether or not, and to what extent, where you live, or how you live, affects the chances your kids will be autistic or not.