This Just In: Autism Is Caused by [Fill in the Blank]
Parents are bombarded with stories about autism research. Headlines and somber-voiced announcers declare that new research has found that autism is linked to a smorgasbord of things: mom's age, dad's age, grandfather's age, living near freeways, living near farms, prenatal stress, premature birth, fertility treatments, obese mothers, flu during pregnancy, having babies too closely together, and so on.
How do we make sense of this?
We live in a world of instantaneous news. Research may be slow, taking months or years to complete, but the results of a study may be sent via a news release and appear on our computers, cell phones and televisions within minutes or hours.
Sometimes these short articles lack the hedging, caveats and alternate explanations included in the longer scientific papers on which the news reports are based. And, as a consumer, it's easy to read that autism is "linked to" something and assume that means "caused by."
Correlation versus causation
Many studies look at correlations, or associations, between two things. When factor X goes up, factor Y goes up, or down. In other words, X is correlated or linked to Y. Does that mean that X is causing Y?
Discovering a correlation can be the first step in proving a cause and effect. Decades ago, some scientists noticed that when cigarette smoking increased, so did lung cancer.1 Upon that foundation, researchers constructed studies that eventually proved that smoking causes cancer.2
But in many preliminary autism studies, correlation does not equal causation, at least not yet. More studies may need to be done – and their research methods and results carefully analyzed – before anyone can prove X causes Y.
"In order to establish cause and effect, you need to have a well-designed experiment or a boatload of observational studies," explained statistician Deborah J. Rumsey Ph.D.3
California health officials discussed the correlation-causation issue in a fact sheet about an autism study published in 2013. The study found that, around the time of their children's births, "mothers of children who were later diagnosed with autism were twice as likely to work in a job that involved chemicals (14%) as mothers of the comparison children without autism (7%). This was a statistically significant difference."4, 5
But, as the fact sheet explained, this correlation does not necessarily mean that the chemicals cause autism. This preliminary study had a few shortcomings. Only a small percentage of the mothers worked in jobs with chemical exposures, such as bus drivers, landscapers, flight attendants and hospital nurses. It is harder to draw conclusions about large numbers of people from a study that examined relatively few.
Also, the data available to the researchers did not include details about the women's jobs, work history and actual exposures. They did not have information about other factors that could have contributed to the autism link, such as, in the case of nurses, whether anesthetics or infections could be behind the higher autism risk. The researchers could not reach definite conclusions, beyond calling for more research on this topic.
Rain + TV = autism?
Sometimes correlational studies can be difficult to interpret.
Take the story of television and autism. In 2006, a Slate.com story, "TV Might Really Cause Autism," reported on research by three professors from Cornell and Indiana universities.6 The professors used the unusual trifecta of weather data, cable television subscriptions, and autism diagnostic data to arrive at the conclusion that television viewing is linked to autism. Bloggers and the media found this information fascinating and followed the story, which stoked the "mommy guilt" of anyone who left on Sesame Street while cooking dinner.
In this instance, the professors were economists, not autism experts. They did not explain exactly how, in biological or psychiatric terms, TV could trigger autism.
They based their research on an analysis of three sets of statistics: the rise in autism cases in the last few decades, the increase in cable television subscriptions during the same period, and data on weather precipitation in several states. Rainy and snowy weather keeps youngsters indoors, where they are more likely to watch TV. Regions with more precipitation had more cases of autism, according to their analysis. In addition, as cable television subscriptions increased, so did cases of autism. Therefore, television viewing may be triggering autism in susceptible children, the researchers concluded.
They ventured a step further, straddling the line between correlation and causation, in their working paper, "Does Television Cause Autism?"7
"Our precipitation tests indicate that just under forty percent of autism diagnoses in the three states studied is the result of television watching due to precipitation, while our cable tests indicate that approximately seventeen percent of the growth in autism in California and Pennsylvania during the 1970s and 1980s is due to the growth of cable television," they stated. They acknowledged they had not definitively proven their point, as they had not measured children's actual viewing habits.
Correlational studies cannot control other factors that could influence autism rates. It can be hard to know exactly what, if anything, connects separate sets of statistics. So it's not surprising that others looked at the same numbers and drew different conclusions.
Another economist, Steven D. Levitt, offered this explanation: "My theory: when it rains a lot, parents watch more TV, see more shows about autism, and this leads them to seek out a diagnosis of autism for their kids. They have the same kids, it is just that TV makes them believe that their kids are autistic."8
A Vanderbilt University geneticist, Pat Levitt, had a different take on why autism cases might cluster in rainy areas – and it had nothing to do with television. "How do you know, for instance, that it's not mold or mildew in the counties that have a lot of rain?" Levitt told Time.com.9 Might poor indoor air quality be the bridge between rain and autism diagnoses?
Or is it something completely different?
Statistical associations can be explained in different ways, which is why people should be careful about drawing conclusions from them. So when you read that something is linked to autism, don't assume that it is causing autism, at least not until the study is examined and its results confirmed by other, preferably large, studies.
Stay tuned for the bigger picture.
- Peace, L. R. (1985) A Time Correlation Between Cigarette Smoking and Lung Cancer. In Journal of the Royal Statistical Society. Series D (The Statistician). Vol. 34, No. 4, pp. 371-381.
- Stout, D. (1996, October 18). Direct link found between smoking and lung cancer. The New York Times.
- Rumsey, D. J. (2011) Statistics for Dummies. Hoboken, NJ: Wiley Publishing Inc.
- Windham, G.C., Sumner, A., Li, S.X., Anderson, M., Katz, E., Croen, L.A. & Grether, J.K. (2013) Use of birth certificates to examine maternal occupational exposures and autism spectrum disorders in offspring. Autism Res. 2013 Feb;6(1):57-63. View abstract.
- "Questions and Answers about a Study of Parental Occupational Exposures and Autism in Offspring." Center for Autism and Developmental Disabilities Research and Epidemiology (CADDRE), CA Department of Public Health, Richmond, CA.
- Easterbrook, G. (2006, October 16) TV Really Might Cause Autism. Slate.com.
- Waldman, M., Nicholson, S. & Adilov, N. (2006) Does Television Cause Autism? National Bureau of Economic Research Working Paper No. 12632.
- Levitt, S. D. (2006, October 17) TV causes autism? I doubt it. Freakonomics website.
- Wallis, C. (2006, October 20) Does Watching TV Cause Autism? Retrieved from Time.com.