So correlation... Yep I know correlation is not the same as causation. Cause-and-effect is a notoriously difficult relationship to map out when talking about conditions like autism with all their heterogeneity (autisms if you please) and that old genes vs. environment grudge match.
But that doesn't mean correlational studies are all bunk.
|Mind the smog @ Wikipedia (Claude Monet)|
Take for example the relationship between air pollution and risk of autism presented in a paper by Volk and colleagues*. I recently covered these findings from the CHARGE initiative and how, not for the first time, there was speculation on either in-utero or early years exposure to air pollution such as traffic fumes potentially having some modifying effect on a child's risk of receiving a diagnosis of an autism spectrum disorder (ASD). Some interesting research but correlational research nonetheless.
Then the study by Tracy Ann Becerra and colleagues** is published (a pre-print version available here open-access) and suddenly things start to get a little more interesting, as now more than one study based in a similar geographical area (California, USA) start to arrive at similar conclusions.
A few choice factoids from the Becerra paper:
- A large sample were included for analysis - over 7,500 children diagnosed with autism based on Department of Developmental Services (DDS) records - and an even larger control group (10 controls to each autism case) matched for gender and quite a few other variables.
- The paper was a very statistical affair looking at birth addresses and "measured and modeled exposures to prenatal air pollution and autism in children". I can't give you all the statistical details because they are a little complicated and somewhat beyond my very limited statistical capabilities. It basically boiled down to correlation and some statistical modeling based on various factors including something called a land use regression (LUR) model.
- Results: "We estimated an approximately 3 to 9% relative increase in the odds of Autistic Disorder per inter-quartile range increase in entire pregnancy exposure to NO (9.40 ppb) and NO2 (5.41 ppb) as estimated by our two-pollutant LUR models". In other words, a possible relationship was hinted at by the data and its analysis between risk of offspring autism and facets of traffic fume exposure.
- A few other points of note: maternal education (a marker for socio-economic status) was a strong modifier of this pollution-autism risk relationship and ozone (O3) (remember that hole in the ozone layer) also featured very strongly among the statistics, potentially increasing risk for autism by 12 - 15%.
Reiterating the correlational nature of this study and all the problems associated with tracking participant records and examining individual specific environmental exposures as being 'the one', I'm sufficiently intrigued by this study adding as it does the growing literature in this area.
You could perhaps question that autism rates are rising (yes they are) yet generally our pollution levels are dropping (the removal of lead from petrol as one example) so why would this be a particularly strong relationship? Of course you'd be right, assuming that is, that such pollution acts solely and directly on the foetus and doesn't for example start to mess around with things like the germline or possess epigentic effects***. That and the fact that pollution levels tend to vary from area to area, city to city, country to country, so we can't necessarily assume any uniform exposure to be widely distributed across everyone and every place. Take a look at similar research on air pollution and things like cardiovascular disease mortality for example (see here****) and you'll perhaps see that dose might not necessarily be the overriding factor in any relationship.
I am of course open to other explanations for the growing literature looking at the air pollution - autism risk correlations being reported including factors such as the availability of neighbourhood-level diagnostic resources***** (covered here), but I must admit at the moment, I'm a little stumped by any environmental factor which could replace the air pollution variable seemingly so universally. Unless one starts to consider things like water sources, food sources or things like medication and their residues as potentially showing other correlations?
To finish, I'm all Christmas-movied out after watching quite a few classics with the brood. Mary Poppins is always a favourite, so how about a spoonful of sugar with that correlation?
* Volk HE. et al. Traffic-related air pollution, particulate matter, and autism. Arch Gen Psychiatry. November 2012.
** Becerra TA. et al. Ambient air pollution and autism in Los Angeles County, California. Environ Health Perspect. December 2012.
*** Stoccoro A. et al. Epigenetic effects of nano-sized materials. Toxicology. December 2012.
**** Chen H. et al. Long-term exposure to traffic-related air pollution and cardiovascular mortality. Epidemiology. 2013; 24: 35-43.
***** Mazumdar S. et al. Spatial clusters of autism births and diagnoses point to contextual drivers of increased prevalence. Soc Sci Med. December 2012.
Becerra TA, Wilhelm M, Olsen J, Cockburn M, & Ritz B (2012). Ambient Air Pollution and Autism in Los Angeles County, California. Environmental health perspectives PMID: 23249813