This brief report is concerned with only one set of research results (plus reference to some existing studies and to reactions) but it highlights a further aetiological hypothesis …. with implications for that view of autism and ASD as conditions with multiple causal pathways and influences.




The research paper by Windham et al (2006) begins by noting the marked increase during the last 10+ years in the rates by which children have been identified with autistic spectrum disorders. 

They cite the continuing uncertainty whether these current prevalence rates reflect a real increase in the number of children who have ASD, or if they are a matter of increased awareness of the wide range of signs and symptoms indicative of ASD with a corresponding increase in the diagnosis of autism and ASD among many cases which, previously, would have been given a different diagnostic label (such as

“ pervasive developmental disorder ”).


There is still no precise information about causation, although the current authors quote the commonly-shared belief that autism results from some impairment or dysfunction in normal neuro-biological mechanisms largely in the prenatal period, and which has a strong genetic influence probably involving various gene locations.


Meanwhile, non-genetic (environmental) factors appear to be influential, and the emergence of some of these factors may be significant in respect of any actual increase in prevalence.


In their review of studies, Windhan et al make reference to the use of medication

(such as valproic acid) impacting upon the unborn child, and to maternal cigarette-smoking during the pregnancy. 

Foetal alcohol syndrome has also been linked to autism. 


The potentially dangerous effects of exposure to heavy metals such as lead or mercury have been much discussed in respect of underlying features of impaired neurodevelopment, although there have been few studies which have specifically explored a possible link between such exposure and autism.

Compounds that impact upon the endocrine system, especially where there are effects upon the maternal thyroid hormones, have been implicated as potentially damaging to foetal brain development; and pre-natal exposure to certain solvents has been linked to developmental delays among some children. 


Various air pollutants have been identified by the US Environmental Protection Agency as hazardous in respect of possible neurological and developmental damage, although direct monitoring of their impact is reported to have been limited despite evidence that the estimated concentration of some of these pollutants, including a number of metals, exceeds safety limits.


The authors describe efforts to track prevalence rates for autism by setting up monitoring systems in several US states; and their own recent and exploratory analysis in the San Francisco Bay area of autism surveillance data and their relationship with ambient chemical exposure during pregnancy or early infancy.


The target population included children born during 1994 to mothers resident at the time of their birth in one of the six counties within the San Francisco Bay area.

Those children with ASD were identified via monitoring of medical assessments and records by the California Centres for Autism and Developmental Disabilities Research and Epidemiology.


The eventual sample, following stricter diagnostic checks against specific DSM-IV criteria on the part of experienced clinicians, included nearly 300 children (by then around 11 years old) with ASD; and a control group of 600+ matched by sex and month of birth and randomly selected from the 1994 California birth or early infant death certificate file relating to maternal residence in one of the six counties. 


The birth addresses were systematically coded by area in order to gain an indication of the level of exposure to given concentrations of air pollutants identified as neuro-toxicants or developmental toxicants and/or endocrine-function disruptors (as provided by the 1996 data base of hazardous air pollutants compiled by the US Environmental Protection Agency.


This fairly broad approach was taken because of the lack of existing information about which of the 33 compounds listed in the 1996 database might be related to autism.


The results indicated that the target cases, compared to controls, were more likely to be white rather than Hispanic, and to have been born to mothers who were somewhat older and had higher levels of education. 

This pattern also applied to paternal age and education.

Demographic data suggested that non-white, younger, and less educated parents were generally more likely to live in areas with higher exposure concentrations of metals and solvents.

The target group had a male-female ratio of 4:1 (which is in accord with existing and converging prevalence data).


Regression analyses indicated a potential association between autism and higher air concentrations of metals and (possibly) chlorinated solvents in the geographical area of residence at the time of birth.

The individual compounds which contributed most to the association included mercury, cadmium, nickel, trichloroethylene, and vinyl chloride.


However, the authors recognised a number of limitations to these exposure data.


      (particularly for metals) diet.


(The present writer - MJC - was struck by an apparent consistency in that the target sample were said more commonly to be born of older and better educated parents than control cases, and yet controls cases were more likely to have been located in areas with higher exposure to pollutant metals and solvents.

One might speculate that the higher pollutant rate was reflected in the more serious physical/neurological difficulties before any effect of an ASD could have been determined, or that aspects of socio-economic disadvantages, such as poor housing or poor diets or less prompt access to medical care, confounded any pollutant effects. 

It is also, presumably, impossible to say what percentage of the control sample might have gone on to show signs and symptoms of autism.

There is also something anomalous about an environmental variable which selectively impacts upon males rather than females and upon certain families rather than others.

One might further ponder whether the impact of pollutants, if impact indeed there is, is another example of an environmental trigger which acts upon an existing propensity ….a propensity which is based upon genetic influences and not equally distributed among the whole of a birth cohort.)


In any event, it is noted that Windham et al do note this issue of socio-economic status as a potential covariate, as reflected in the information concerning ethnic background and education.

They accept that, given the limited state of knowledge concerning the risks for ASD, uncontrolled effects may be confounding the apparent associations reported. (They accepted, for example, that there was no information available about control mothers’ health or behaviours which could have a bearing upon their own health and that of the child).


Nevertheless, the authors felt able to conclude that living in areas with higher ambient levels of hazardous air pollutants, particularly metals and chlorinated solvents, during prenatal or early infant life may be associated with a moderately increased risk of autism. 

For example, San Francisco County had the greatest mean levels for the main compounds assessed (mercury, cadmium, diesel fuel products, methylene chloride, toluene, and vinyl chloride) and also had a higher ratio of cases to controls than the other counties.

The county with the lowest levels of these chemicals (Marin) had a significantly lower ratio (albeit with the acknowledgement from the authors that this could reflect various other factors including differences in diagnostic procedures or in care-seeking actions).


The authors emphasise that more systematic aetiological studies are required by which to gain more accurate information about individual exposure to multiple compounds via various pathways (air, water, diet) alongside genetic information if there is to be a valid unravelling of the interaction of endogenous and exogenous factors in the development of autism and ASD.


(One would further add the note of caution that the data presented are correlational, and, while it might be plausible and somehow reassuring to make a link between the apparent rise in ASD prevalence rates and the increased levels of air pollutants in many places, there is still a difference between a correlation and a causal attribution … as well as continuing debate whether there has been a real increase in prevalence rates of ASD rather than shifts in diagnostic sensitivity and criteria.)


A commentary by Chew (2006) also stresses the need for more evidence about any link between ASD and air pollutants, and for caution about drawing any causal conclusions.  She regards the particular limitation of this current study as failing to measure the compounds to which the mothers were actually exposed, and the dependence, instead, upon estimates calculated by the US Environmental Protection Agency for ambient pollutants in given areas.  She also echoed the acknowledgement of the authors themselves that there could be alternative explanations for the apparent link … such as, she suggests, a possibility that those mothers living in the most polluted areas eat more contaminated seafood; and life-styles, such as smoking habits, were not part of the analysis.

Further, in California, it is possible to identify regions less subject to pollution than the San Francisco Bay area but which have higher overall prevalence rates for autism.

(Sacramento is cited as one example).  


Meanwhile, a study in progress at the University of Texas Health Science Centre, San Antonio (Palmer et al 2005- on) is also looking at the possible aetiological significance of prenatal exposure to mercury.  They are examining agency data on industrial mercury release - mostly from coal-fired power stations - across counties in Texas, and comparing these figures against ASD prevalence rates available from all the school districts. 

Their study suggests that for every 1000 pounds of mercury released through industry there is an appreciable rise (17%) in ASD rates. 

However, while this is recognised as considerable potential exposure, the mechanisms that underlie this putative link are not clear.  Similarly, Palmer points out that these (correlational) data do not prove that mercury pollution causes autism and ASD.

There is an apparent  relationship … but that is all so far.  Much harder data are needed if there is to be any pressure for governmental action on chemical release controls.


A further study is ongoing at the University of Rochester School of Medicine and Dentistry to assess the effects of eating mercury-polluted fish upon child development. 

In an interview on in December 2006, Philip Davidson (research leader) states that, at this moment, there is no clear answer to the question whether the threshold for autism is raised in the case of a child exposed prenatally to mercury.

He can cite no evidence from any existing research in respect of isolating symptoms of autism among individuals exposed to methyl mercury.  Autism appears to have pre-natal aetiological sources, but there is no means of determining whether mercury exposure can act as an autism trigger.


The ongoing study concerns mercury exposure among pregnant women and their children in the Seychelles where diets include large quantities of fish … but Davidson recognises the complexity involved in unravelling any possible effects of the different forms of mercury, the level of exposure, and the age at which a person is exposed to mercury.


Davidson indicates that, if mercury does have harmful effects, the impact is more likely at a prenatal stage than during childhood …. and it is noted that US governmental guidelines of 2004 recommended, as a precaution, that women who are contemplating pregnancy, who are pregnant, or who have young infants, should limit themselves to no more than two servings of fish each week. 


In other words, the evidence is not yet established, but there are (correlational) indicators that there could be some link between exposure to metals and other hazardous substances, notably mercury, via air pollution (or via water pollution or elements of diet),  particularly at the early stages of  prenatal life. Further and clarifying evidence is awaited.



                             *          *          *          *          *          *


M.J.Connor                                                                                              January 2006        




Chew K.  2006   Autism linked (?) to air pollutants in the San Francisco Bay area.  Autism Vox.  June 2006.


Palmer R.  2005   Health and place.   Published on-line, February 2005 


Windham G., Zhang L., Gunier R., Croen L., and Grether J.  2006   Autism spectrum disorders in relation to distribution of hazardous air pollutants in the San Francisco Bay area.   Environmental Health Perspectives  114(9)  1438-1444   


This article is reproduced by kind permission of the author.

© Mike Connor 2007.

Back to NAS Surrey Branch Welcome Page