Can living in rainy areas really cause one-third of autism cases?

At first, this study sounded like it might have been published in the Journal of Spurious Correlations or an entry for the Spurious Correlations Contest and would provide a note of levity.

A few years ago, three economists had self-published a paper reporting a correlation between rising rates of cable television subscriptions and autism in Northwest coastal areas since the 1970s. That was funny because, of course, anything that has increased since the 1970s could be said to spuriously correlate with autism; nor had they made any effort to see if the autistic children had even watched more television.

One section of that original paper has just been republished in Archives of Pediatric and Adolescent Medicine, a journal of the American Medical Association. For real. It reported a correlation between higher amounts of rain and snowfall during 1987-2001 in Northwest coastal areas and autism rates among school-aged children in 2005.

That’s it.

Yet, the authors concluded that this association supports their hypothesis of an environmental trigger for autism and went on to speculate on “a number of possibilities of what such an environmental trigger might be.” The first was television and video viewing because “it seems plausible” that it’s associated with precipitation and could possibly be associated with “more serious health problems such as autism.” They went on to put forth even more fanciful and worrying possibilities of unseen environmental exposures, as we’ll look at in a minute, and concluded that this paper proved further research into this link was warranted.

This isn’t funny anymore. If this is what now constitutes medical research in a peer-reviewed journal and is actually being taken seriously by the medical community and major universities, we are in serious trouble.

A critical look at this paper is warranted, in honor of Steve Allen, who coined the term “Dumbth” to describe his “literally daily frustration with the degree of goofola thinking, speech and behavior that had become dominant.” He noted back in the 1990s that most students lacked even a basic grasp of science and ability to reason, and he lamented the dramatic and continuing erosion of American education and intelligence. Many of those students are today’s scientists, educators and doctors.

There was a press release…

Autism is an extremely politically and emotionally charged issue and countless interests are capitalizing on parents' fears of an epidemic and of some invisible, unknown danger that could be endangering their babies and children. Autism can be, and has been, linked to just about every aspect of childhood from watching cartoons to playing in a playground. It’s a cruel, sad fact that any health problem for which science doesn’t yet have answers becomes easy prey for quackery. Having an autistic child is hard enough for parents without unsupported scares being thrown at them. Most importantly, parents want medical research to find a cure for true autism, not waste resources chasing spurious correlations or using them to further political-environmental agendas.

Yet, not only were the population-based correlations in this paper unsupportable and implausible in themselves, as we’ll see, the authors failed to actually measure among autistic children any of the exposures they alluded to as being a possible link. Without that, this paper offered no science or evidence at all.

Qualifying statements that “our results are clearly not definitive evidence” do not absolve professionals from the responsibility to publish creditable research in medical journals — let alone excuse issuing a press release to the public which greatly overstated the science, knowing full well most media would use it. The press release issued by Johnson School at Cornell University, of lead author Michael Waldman, Ph.D., headlined saying: “AMA Journal Publishes Study Showing Evidence of a Major Environmental Trigger For Autism.” There was nothing in the press release to even suggest these findings were tentative or speculative, no effort to moderate scares, or to do anything but build fears of some hidden environmental exposure causing autism in children. It told the public that this peer-reviewed study’s positive findings “suggest that as many as 30 percent or more of autism diagnoses may be due to an environmental trigger or triggers related to levels of precipitation where the children live… and solidifies the need for further research focused on identifying what the exact environmental trigger might be.” By suggesting a causal role and pulling a figure out of thin air, predictably, specious headlines followed, like MSNBC news: “Can rain trigger autism?”

Regardless of what motivated a study of this caliber to warrant a publicity department, or even to be published at all, it held no useful value for healthcare providers or the public. Yet, the AMA journal made an odd move to publish an accompanying editorial by epidemiologist Dr. Noel Weiss at the University of Washington in Seattle, opining why the editors were to be lauded for their decision to publish this paper. He said that its observed correlation “may well not lead to any insights into the etiologies of autism” but went on to commend the editors for their decision to publish it, “despite the uncertain ultimate contribution of this work and the possibility (likelihood?) that nonprofessionals are going to misinterpret and misuse it.” [The paremphesis was his.]

He went on to say that the primary audience “is not the practicing pediatricians and certainly is not a member of the public at large” because we are unable to understand or take away any practical information from it. He wrote that we are just “eavesdropping” on the exchange of information between real scientists. Instead, this paper was for investigators, epidemiologists interested in doing other observational studies.

If this paper represents the quality of scientific information exchanged among scientists, that’s a bigger story than the paper itself. But I believe that there are still practicing doctors and consumers out there who don’t just blindly believe media marketing and are capable of understanding science and seeing the fallacies of logic when presented the information.

Television and autism

This study, published in the November issue of Archives of Pediatric and Adolescent Medicine, reportedly found a link between precipitation in California, Oregon and Washington and the prevalence of autism, reported by counties. For those journal access-challenged, this study is a reprint of sections of an earlier paper the authors had published on the Johnson Graduate School of Management at Cornell University website in 2006, when they were trying to make a case that television causes autism. That paper is available free here. (See page 17 for the statistical machinations to create the variable they used to define average annual precipitation for each county; and pages 22 -31 on how they then linked rain between 1972 and 1989 in those counties to rates of autism.)

That original’s paper’s hypothesis was that television viewing by children triggers autism. Briefly, they first noted that autism has been rising since the 1970s and the proportion of households in the U.S. with televisions and VCRs has also increased. This established their hypothesis of a link between television and autism.

To get around the problem of variable national data, they said, they then focused on California autism rates by birth year*, which had also steadily increased in coastal counties — Figure 1.

They then looked at cable television subscription rates in those counties — Figure 2.

Low and low and behold, those had increased, too. That confirmed their hypothesis of a link between television viewing and autism. I don’t mean to scare you, but this is what qualifies as peer-reviewed science. They didn’t actually look if the children with autism had actually watched television prior to developing autism, but that’s beside the point. They could have chosen anything in our society that’s increased since the 1970s and made equally spurious correlations… and fabricated equally science-sounding explanations.

For instance, the number of exotic pets owned by U.S. families has been rising, just as have autism rates, increasing 75% just since 1992 according to the American Academy of Pediatrics. Just weeks ago, the APA even issued warnings to parents in the journal Pediatrics, that these nontraditional pets— rodents, reptiles, monkeys and more — expose children to dangerous diseases and infectious agents not typical. Exotic pets, therefore, could be a risk factor for autism.

It’s impossible not to notice that all of the speculative explanations for the environmental links to autism — from television viewing to environmental toxins — are currently popular scares. How likely would this study have received the media coverage, let alone funding for further research, if it had explored less politically correct correlations?

● For example, since the 1970s, children have been eating less fat and more fruits and vegetables, with dark leafy greens and broccoli increasing by about 370%. This parallels increases in autism and anti-vegetable interests (if there was such a thing) could be using these equally impressive trends to scare parents about Popeye the Sailor Man.

● Wine consumption in the United States has increased in parallel to autism rates, too. To overcome variabilities in national figures, we can even look at autism rates in Northwest coastal areas which also correlate with the growth of the coastal wine industry, as reported by the University of California, Davis.

● Smoking rates have dropped since the 1970s, with fewer women smoking during pregnancy and fewer parents smoking around children. Yet, cigarette smoking has, in fact, been shown in research for at least decade to be related to dramatically lower incidences of Parkinson’s disease and many researchers believe there are neuroprotective compounds in tobacco leaves. Nicotine is even believed to have a beneficial role when used to self-medicate by schizophrenics to activate reduced low-affinity alpha-7 nicotinic receptors, maybe it has some role in the neurological triggers of autism. Smoking cessation could be another risk factor for autism.

The point being, these authors could have pulled out any trend over recent decades and reported a correlation to autism. But without actually linking any autistic children to these exposures, in utero or early childhood, they’re just as spurious.

Autism and rain

For this paper, the authors noted that precipitation was higher along the coastal areas of Oregon and Washington than on the other side of the Cascade Mountains, where the eastern parts of the states are drier. They also noted that autism rates reported by those coastal counties were higher.

Proof! Rainfall is linked to autism. [They had included California in their original analysis but had to drop California, reporting “the California data continues to show no evidence of a positive correlation between precipitation and autism.”]

But precipitation in select Northwest coastal counties and autism rates is a spurious correlation. It’s not supported by figures from anywhere else. Didn’t anyone even notice that there’s no link between the states reporting the highest autism rates and those with the highest annual amounts of precipitation?

Mississippi and Louisiana, for example, have the 2^nd and 3^rd lowest reported cases of autism, yet are two of the wettest states in the country [see precipitation map at the top of this article].

The authors proceeded to create scary scenarios to explain their spurious correlation between rainfall and autism. They first suggested that children living in rainy states don’t spend as much time outdoors in the sunshine and that perhaps autism could be related to vitamin D deficiency. Not only did they not measure a single vitamin D level among the autistic children compared to children without autism, but dietary links to autism were disproven years ago, as have early beliefs that parental practices were responsible, according to the National Institute of Mental Health at the National Institutes of Health. The authors then speculated that children in rainy areas spend more time indoors and could be exposed to indoor chemicals and household cleaning products that might have some subclinical health danger. These fears are popular with clinical ecologists promoting environmental illnesses, and are equally unsupported in the medical research to date on the contributing causes of autism. [See sidebar below: “What is known about the causes of autism.”]

Finally, they speculated that rain could be bringing air pollutants to the ground level and increasing the children’s exposure to chemicals and pesticides that could trigger health problems. This is the most spurious correlation of all. Not only are the levels of heavy metals and toxins detected in our bodies lower than in previous generations and in centuries of human history (including mercury, for which the science has also exhaustively shown in our food and vaccines to have no link to autism); but by virtually every measure, the air we breathe in the United States is cleaner than it’s been since the Environmental Protection Agency began monitoring air quality back in 1970.

The states with the highest reported autism rates are also not the highest agricultural regions of the country, dispelling logical links to pesticide exposures. Iowa, for example, is primarily agricultural (think Iowa corn and pork), yet has the fourth lowest reported autism rates in the country. States with the highest autism rates also have no correlation to states with the highest levels of air pollutants. Even looking at the EPA data of the air pollution particulate counts in specific counties in Oregon (available from the American Lung Association), shows no correlation to the counties in the Waldman paper with the highest and lowest autism rates. Jackson county, for instance, had low autism rates but was given a grade of “F” for air pollution by the EPA; Columbia and Benton counties had the highest rates of autism according to the Waldman report, but were given “A” grades for clean air.

It’s impossible to find any empirical support for any of the scary unseen environmental dangers speculated in this paper. So, what might more reasonably explain this spurious correlation?

Asking the right questions

When we don’t ask the right questions, not surprisingly, the our search for answers will take us in unproductive directions and lead to conclusions that aren’t likely to be supported in clinical intervention trials. Have rates of autism actually skyrocketed and increased 16-fold since 1970s?

Looking at the Waldman paper, most all of the correlations between the precipitation in the coastal counties and autism rates were in Oregon. First of all, according to the National Oceanic and Atmospheric Administration report, Population Trends Along the Coastal United States: 1980-2008, those are the same counties that are the most populated and have seen the greatest population increases. The counties on the eastern side of the Cascade Mountains are more rural and less populated, and just happen to be drier. The most populated counties are also where most of the states’ agencies and developmental services are located.

Remember, definitions are everything. The reported autism rates are based on the number of children enrolled in these developmental disability services, and the reports determine how much federal special education funding is available to those districts. Notice how Oregon and Minnesota, for example, have the highest reported autism rates in the country? As Dr. James R. Laidler, M.D., explained in detail in the article, How "Educational Assessments" Skew Autism Prevalence Rates, it’s a matter of the criteria used to diagnose autism in children. Under the federal Individuals with Disabilities Education Act (IDEA), school districts have been assessing autism in children, and medical diagnosis and educational assessments use very different criteria. Further clouding the issue, he said, every state uses its own assessment standards and procedures, and Oregon and Minnesota use much broader criteria than other states. So, you’ll notice in his paper that autism rates reported by those states have been considerably higher than anywhere else in the country.

Lending additional caution in interpreting autism trends, according to the National Center on Birth Defects and Developmental Disabilities at the Centers for Disease Control and Prevention (CDC), the diagnostic criteria for autism has undergone repeated changes since 1956, expanding to include increasing behavioral symptoms and syndromes. Autism is a spectrum of developmental disorders and definitions among them have also changed over the years.

JFS previously examined in detail [here] a 2005 study in the journal of the American Psychological Society, explaining how to create public perceptions of an epidemic that are completely disconnected from the scientific evidence. Researchers at the University of Wisconsin-Madison and the University of Montreal-Quebec in Canada had examined the evidence behind frightening claims of an epidemic of autism. They found that more people were fitting the criteria for autism — not because there was more autism — but because the diagnostic criteria had been broadened, public awareness had increased, and greater lengths were being taken to identify cases. Autism didn’t exist as a diagnosis until the 1940s, they said, when a constellation of differences that had always existed in the population came to be named “autism.” But it wasn’t until 1980 that autism was first added to the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) using six mandatory criteria. But over the years those criteria have undergone several changes, each one widening the definition to include more subthreshold symptoms and milder variations. The most recent DSM-IV has 16 criteria, only half of which need to be met to be diagnosed autistic, and mild variants now account for most of the currently diagnosed cases, they found.

The U.S. Department of Education’s 2002 Report to Congress on IDEA (Individuals With Disabilities Education Act) said that the number of students with autism in America’s schools jumped an alarming 1,354% from the school year 1991-92 to 2000-01. But prior to the 1991–1992 school year, there was no child count of students with autism because autism did not even exist as an IDEA reporting category. It was also an optional category for reporting, only required in more recent years.

Professor Waldman and colleagues, however, negated that more active case identification and changes in diagnostic criteria, as well as differences in reporting methods and population characteristics, could exclude the possibility of a true increase that could be caused by the environment. This belief is popular among lay press, but not widely shared by medical researchers.

The Association for Science in Autism Treatment has left available an article by Dr. John W. Jacobson, Ph.D., which had taken a critical look at speculations that more children have autism than ever. While there are incentives for claiming autism is more prevalent, he concluded, “it is very difficult to support this assertion with scientific, epidemiological evidence.”

We can say the same thing about rainfall triggering autism.

© 2008 Sandy Szwarc

Sidebar: What is known about the causes of autism

According to the National Institutes of Mental Health at the NIH, research into the causes of autism spectrum disorders has advanced in tandem, with no single cause for autism. With the emergence of new brain imaging tools — computerized tomography positron emission tomography, single photon emission computed tomography, and magnetic resonance imaging — scientists can now study the structure and functioning of the brain. These, along with postmortem studies of brain tissue samples, have shown “that many major brain structures are implicated in autism, [including] the cerebellum, cerebral cortex, limbic system, corpus callosum, basal ganglia, and brain stem.” These abnormalities suggest that autism results from disruption of normal brain development early in fetal development.

Recent neuroimaging studies have revealed abnormal brain development already evident in the infant’s first months of life, notes the NIMH. “This growth dysregulation hypothesis holds that the anatomical abnormalities seen in autism are caused by genetic defects in brain growth factors.” Other research is focusing on the role of neurotransmitters such as serotonin, dopamine, and epinephrine.

The evidence to date, they note, most points to genetic factors as playing a prominent role in the causes of autism. Twin and family studies have suggested an underlying genetic vulnerability to autism syndromes. Genetic research is just beginning, but has already identified a number of genes that play a role. Research led by Daniel Campbell at the Department of Pharmacology at Vanderbilt University in Nashville, TN, for example, showed a genetic association of a common C allele in the promoter region of the MET gene in 204 families with autism.

As the Association for Science in Autism Treatment explains, research is looking if there is something that might spark autism to develop in those with a genetic predisposition, such as environmental, infectious, immunological, and other conditions or events. But it’s increasingly believed these occur prenatally. According to Judy Mesch, former associate director of the University of Miami-Nova Southeastern University Center for Autism and Related Disability (CARD), research has found autism is genetic and neurological. Regardless of when in infancy to the first three years of life autism presents itself, the evidence to date counters theories of environmental exposures during childhood, like watching television or bad parenting, but points to neurological changes that begin before birth.

“At this time, there is no evidence that specific toxins in the environment, immunization practices, dietary differences, or immunologic differences cause autism,” according to the Association for Science in Autism Treatment. But what is known, is that theories of faulty parenting practices is not a cause of autism. “The theory of faulty parenting as causative of autism has not disappeared entirely however, and still crops up in “failure to bond” theories and “attachment disorder” theories about the genesis of autism. Most researchers reject this type of explanation as misguided and harmful.”

* Even before getting into how autism was defined, these figures are “based on people who were receiving services in 2005 at a California Department of Developmental Services regional center.” Based on enrollments in these developmental disability services, they wrote that rates doubled from 1970 among those born in 1980 and again in 1986 and 1992. The reporting system for the Dept. of Developmental Services for the state of California determines how much federal special education funding is available. According to its figures released in 2005, rates of autism had been dropping in California since 2002, which weren’t reflected in the rise shown in the Waldman paper.

[Click on images to view larger.]