The U.C. Davis MIND Institute has an ongoing study to explore autism risk factors due to both genetic and environmental influences: the CHARGE study. The CHARGE study website describes itself as:
CHARGE (Childhood Autism Risks from Genetics and the Environment) was launched in 2003 as a study of 1,000 to 2,000 children with differing patterns of development. The goal is to better understand the causes and contributing factors for autism or developmental delay. Three groups of children are being enrolled in the CHARGE study: children with autism, children with developmental delay who do not have autism and children from the general population. All of them are evaluated for a broad array of exposures and susceptibilities.
A paper from this study was just released. In it, they claim that maternal proximity to freeways might be a risk factor for autism:
Residential Proximity to Freeways and Autism in the CHARGE study.
Volk HE, Hertz-Picciotto I, Delwiche L, Lurmann F, McConnell R.
Abstract
Background: Little is known about environmental causes and contributing factors for autism. Basic science and epidemiological research suggest that oxidative stress and inflammation may play a role in disease development. Traffic-related air pollution, a common exposure with established effects on these pathways, contains substances found to have adverse prenatal effects. Objectives: To examine the association between autism and residence proximity, during pregnancy and near the time of delivery, to freeways and major roadways as a surrogate for air pollution exposure. Methods: Data were from 304 autism cases and 259 typically developing controls enrolled in the Childhood Autism Risks from Genetics and the Environment (CHARGE) Study. The mother’s address recorded on the birth certificate and trimester specific addresses derived from a residential history obtained by questionnaire were geo-coded and measures of distance to freeways and major roads were calculated using ArcGIS software. Logistic regression models compared residential proximity to freeways and major roads for autism cases and typically developing controls. Results: Adjusting for sociodemographic factors and maternal smoking, maternal residence at the time of delivery was more likely be near a freeway (?309 meters) for cases, as compared to controls (odds ratio (OR), 1.86, 95% confidence interval (CI) 1.04-3.45). Autism was also associated with residential proximity to a freeway during the third trimester (OR, 2.22, CI, 1.16-4.42). After adjustment for socio-economic and demographic characteristics, these associations were unchanged. Living near other major roads at birth was not associated with autism. Conclusions: Living near a freeway was associated with autism. Examination of associations with measured air pollutants is needed.
Note: Pubmed lists the affiliation as the University of Southern California. I don’t know if any of these authors are from USC, but I know that CHARGE and Dr. Hertz-Picciotto are with the U.C. Davis MIND Institute.
It is an interesting idea and it will be interesting to see if this result holds up in studies with larger groups.
Left Brain Right Brain 
It _is_ an interesting idea and much more of a likely environmental issue than you-know-what.
Sounds like Texas sharp-shooting to me. Betcha it’s crawling with post-hoc subgroup analyses.
Following questions immediately crawl to mind:
(a) Another example of ‘correlation doesn’t equate causation’?
(b) Why were Freeways ‘bad’ and not other major roads?
(c) What about data from other countries where people live close to busy major roadways equivalent to Freeways?
Kausik Datta,
all good questions. I wouldn’t take this as a definitive study. It poses an interesting question and it is the sort of question that many groups claim they want explored but never seem to care about in the end.
And unfortunately, that’s where pseudoscientific kookery jumps in to fill the void, as you are well aware.
If this latest hypothesis was true, then the purported autism epidemic began around 1956 with the passage of the Interstate Highway Act.
I think that this is way too small a number of controls to look at something like this. You need a number of controls that samples the “freeway proximity space” and weights it for the number of children exposed in utero.
The metric found (less than 309 meters) sounds like a metric dredged up via data mining and not an a priori hypothesis.
Air pollution and particle inhalation does cause oxidative stress and does reduce NO/NOx levels, so there is a possible physiological connection. But if so, then places with higher air pollution levels (i.e. LA, Mexico City, Scranton PA (before they put out the coal fires)) should have massive autism clusters.
In any case, most exposure would be at night, if the mother’s worked, and at night there is less traffic. Also, if it is air pollution, then the direction of the wind matters. Up wind there should be little or no pollution from the freeway.
Oh noes. http://www.latimes.com/health/la-he-autism-20101217,0,2040535.story
Proximity to freeways increases autism risk, study finds
More research is needed, but the report suggests air pollution could be a factor.
Proximity to freeways increases autism risk, study finds
More research is needed, but the report suggests air pollution could be a factor.
After looking at Google Maps & consulting my memory, I’m mentally driving around “freeways” in San Francisco, and I wonder if they controlled for multi-family housing (apartments etc.) & single family housing
Kausik Datta: the LA Times story mentions diesel fuel in passing. From my personal knowledge of roadway vs. freeway traffic in the SF area, higher density of diesel-fuelled big rigs on the freeways than on “major roads”.
Hello friends –
I’m not so sure about the scattermap plotting style of study; I took a pretty hard line against the proximity social network via scatterplot study that came out a while ago, this doesn’t look too different from a methodology standpoint. In my town, which has plenty of kids with autism, most homes are tucked away from major roads such as they are here, and everyone I know with a child with autism lives in a subdivision far, far removed from the Interstate (and each other).
That being said, the proposed mechanism of pollutants interferring with development strikes me as plausible, just likely not a major contributor, in my humble opinion.
Considering the way that the genetic studies have been going; away from the big score, and more towards a confluence of many genes, I think we may need to expand our horizons in the environmental front in a similar manner, towards the concept of low penetrance environmental impacts, subtly affecting change. Within such a context, these findings could find a home.
Thanks for posting this.
– pD
This study is silliness to the nth degree. I expect this kind of reporting on The Onion. Next they’ll be testing for mothers who consume black olives vs. green olives during pregnancy.
I live on an isolated rural peninsula a good 30 min from any freeway, and we exceed the daily dose of ASDs in both children and adults. not because of rural pollutants or shellfish contamination but because it’s a notoriously quirky community and people with like traits have bred here for generations. I work at the middle school and every year I come up with about a 1/25 ASD rate, and that’s a conservative estimate.
This reminds me curiously of the study (I can’t remember the source at this moment) that found incidents of drowning were highly and positively correlated with ice-cream sales in certain tropical countries…
If anyone’s read the paper, what do they mean by “sociodemographic factors” exactly? I would imagine it refers to things like income.
There’s a problem with that. A poor community near downtown Los Angeles will probably not have the same prevalence of diagnosed autism as an equally poor community in, say, rural Alabama. (I mean that awareness in those regions is not the same, and availability of specialists is probably not the same either.)
In addition to wealth, it would make sense to control for population density and other factors.
The other problem I have with the CHARGE study in general is that it appears to be a “data dredging” effort. (The pet shampoo finding is what first gave me this impression.)
I was alarmed to see that my local paper ran this headline for the story: Study finds link between autism, air pollutants
There seems to be a lot of oversimplification and jumping to conclusions here. I’d rally appreciate it if you would read what I wrote about it and let me know if my thinking makes any sense to you:
http://thautcast.com/drupal5/content/do-freeways-cause-autism-next-bad-science
Thanks–
Landon Bryce
Sounds like “proximity to freeways” is a surrogate variable for “urbanicity”, since more people will live “close” to freeways in urban areas than in rural areas. If you look at a map of the US interstate highway system, you’ll see that not only do urban areas have a higher population density, they also have a higher freeway density. As a result, the bulk of people living “close” to a freeway – in any state – will be living in urban settings.
I realise they are desperate to find “environmental” causes/triggers/correlations to autism, but this seems awfully sloppy work. If they were my graduate students, I’d send them back to do it again.
BTW, Sullivan, the link to the paper is “broken” – here’s a link to the free full text:
http://ehp03.niehs.nih.gov/article/info:doi/10.1289/ehp.1002835
Prometheus
Thanks for the link, Prometheus. I’ll answer my own question. Here’s what they controlled for:
In short, they did not control for anything resembling population density of the place of diagnosis, which would’ve warranted taking the study more seriously.
It’s also not ideal to take the variables and just plug them into a logistic regression tool. For example, the relationship between population density and administrative prevalence is quasi-logarithmic, so it’s best to apply a transformation first.
To answer some points here: freeways in the United States are already being studied to see how much they contribute to heart disease, asthma, and complications in pregnancy, especially in California. With so much data being gathered on people for other studies, the public health authorities would be remiss if they didn’t also look at autism diagnoses.
I’d bet on Brian Deer’s answer being right, based on what my thesis advisor said about studies that look at too many variables. Include enough variables, and you’ll get a false positive somewhere. (I’m too tired to look up how the statistics work, unfortunately.)
Prometheus also had a good point that freeways = urban = easier to get diagnosed than folks in the boonies.
I also have a bone to pick with their hypothesis: the whole “oxidative stress is associated with autism” began as a way to tie thimerosal to autism. (I don’t have access to most of the journal articles I found online, but I’ve seen posters at oxidative stress conferences citing the Geiers for background. IIRC they were funded by Autism Speaks.) Is this really a legitimate hypothesis despite its dubious origins? Recently there was a paper from the MIND Institute project discussing minor changes in oxidative status and mitochondrial activity but the authors did not correlate any of this to actual symptoms.
And speaking of oxidative stress research, even if we found a solid cause-and-effect between oxidative stress and autism, the most we could do is try to minimize maternal exposure to stressors. Anti-oxidant treatments have been found largely useless for disorders that clearly involve much higher levels of oxidative stress, and the side effects cause damage too. Which, of course, won’t stop supplement-peddlers from using these studies to pimp megadoses of anti-oxidants to parents of autistic kids just like they did for older folks worried about longevity.
The other annoying thing about autism research that chases dubious hypotheses of causation is that it’s wasting funds that could study interventions, such as the recent study on teaching social attention to preschoolers. (Or to develop methods of identifying adult autistics who need supports and interventions. It would be useful to have an accurate headcount of adult autistics just to show that there is no sudden epidemic, just a diagnostic shift.)
“I realise they are desperate to find “environmental” causes/triggers/correlations to autism, but this seems awfully sloppy work”.
There are six syndromes with high rates of autism that have such a massive enviromental effect that they are entirely preventable.
Prevention measure are in place for Congenital Rubella Syndrome and Thalidomide embryopathy with the introduction of an effective rubella vaccine and the ban on the use of Thalidomide in obstretics.
Fetal Alcohol Syndrome, Valproate Acid Syndrome, Prenatal Cocaine exposure sydrome and Post Institutional Autism Syndrome are all preventable syndromes.
http://www.ncbi.nlm.nih.gov/pubmed/8157157?
http://www.ncbi.nlm.nih.gov/pubmed/16167089?
http://www.ncbi.nlm.nih.gov/pubmed/9344050?
http://www.springerlink.com/content/j25pqu8546115m47/
http://www.ncbi.nlm.nih.gov/pubmed/16108456?
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2637680/?tool=pubmed
This is far from a great or conclusive study–especially when it comes to cause and effect. It is an interesting idea that the CHARGE study has–look for possible causes that should be followed up upon.
One major problem I have with the way the CHARGE study appears to be using is that they are breaking out each “hit” as a separate study. If they had taken the CHARGE study as a single study and stated, “we looked at 1,000” (or whatever) “possible links to autism and we found these 5 possibilities” that would put it into better context. Then we would be in a better place to answer whether these are really statistically significant results or not.
Also, if the idea is discussed as a possibility, it is one thing. When people take the possibility as a causal factor, that is far overstretching the value of the study.
Such was the case with the “Autism Spectrum Disorders in Relation to Distribution of Hazardous Air Pollutants in the San Francisco Bay Area” which was touted for a long time as proof of the mercury hypothesis (not by the authors, but by others).
http://ehp03.niehs.nih.gov/article/fetchArticle.action;jsessionid=707363540A3325CB33280189D4EAF045?articleURI=info%3Adoi%2F10.1289%2Fehp.9120
That study had some major limitations, in my view. A major limitation being the lack of controlling for urbanicity. One of the counties in the study, San Francisco, is also a city. All the other counties have some degree of rural area, and much more suburban communities. San Francisco, likely due to this city/county combination, also has by far the highest level of mercury in the pollutants. An abstract at IMFAR this year looked at the Los Angeles area and failed to find any correlations.
Back to the Freeway study. The study’s lead author was interviewed on Los Angels local TV. Besides making the common mistake of claiming a increased incidence for autism (as opposed to prevalence), she appears to be supporting a causal role for freeways beyond which the study would support.
http://www.myfoxla.com/video/videoplayer.swf?dppversion=6512Story link: MyFoxLA.com
She also missed the chance to teach the TV presenters about the lack of real evidence on the MMR vaccine and autism. She was short on time, but it irks me to hear the presenter throw the idea out.
Well if we assume an alpha of 5% then you’re bound to get a false positive once for every twenty false premises. Pharmaceutical companies take advantage of this fairly often – they do a study with multiple outcomes and only publish the ones that are correlative.
I don’t doubt that’s possible (if not documented.) It’s likely that negative risk factor studies go unreported all the time as well. For example, let’s say Palmer or H-P or Windham do a study on pollution where they properly control for population density as a continuous variable. Suppose the study is negative. Would they publish it? That’s doubtful.
I think this is one of the biggest statistical problems in published science, and I don’t see how it could be solved.
Hello friends –
I was poking around a while ago and ran into this paper, which may be of interest to the freeway paper:
Prenatal polycyclic aromatic hydrocarbon exposure leads to behavioral deficits and downregulation of receptor tyrosine kinase, MET
Gene by environment interactions (G × E) are thought to underlie neurodevelopmental disorder, etiology, neurodegenerative disorders, including the multiple forms of autism spectrum disorder. However, there is limited biological information, indicating an interaction between specific genes and environmental components. The present study focuses on a major component of airborne pollutants, polycyclic aromatic hydrocarbons (PAHs), such as benzo(a)pyrene [B(a)P], which negatively impacts cognitive development in children who have been exposed in utero. In our study, prenatal exposure of Cpr(lox/lox) timed-pregnant dams to B(a)P (0, 150, 300, and 600 ?g/kg body weight via oral gavage) on embryonic day (E14-E17) consistent with our susceptibility-exposure paradigm was combined with the analysis of a replicated autism risk gene, the receptor tyrosine kinase, Met. The results demonstrate a dose-dependent increase in B(a)P metabolite generation in B(a)P-exposed Cpr(lox/lox) offspring. Additionally, a sustained persistence of hydroxy metabolites during the onset of synapse formation was noted, corresponding to the peak of Met expression. Prenatal B(a)P exposure also downregulated Met RNA and protein levels and dysregulated normal temporal patterns of expression during synaptogenesis. Consistent with these data, transcriptional cell-based assays demonstrated that B(a)P exposure directly reduces human MET promoter activity. Furthermore, a functional readout of in utero B(a)P exposure showed a robust reduction in novel object discrimination in B(a)P-exposed Cpr(lox/lox) offspring. These results confirm the notion that common pollutants, such as the PAH B(a)P, can have a direct negative impact on the regulated developmental expression of an autism risk gene with associated negative behavioral learning and memory outcomes.
It is actually a really beautiful illustration of the often talked about, but infrequently detailed, gene environment interaction. Of particular salience here, is that the environmental input, PAH, is found in car exhaust (among other things). The MET data is strong for an association, but it does seem to be of relatively low penetrance, perhaps it is interactions like this that are participating.
– pD