Sunday, July 02, 2006

On The Relation To Environmental Pollutants

A couple of studies have been published claiming to document a correlation between autism and environmental pollutants. It is not surprising that such studies would come out as the thimerosal hypothesis of autism is phased out. It is also not surprising considering that it is now well known that the "epidemic" of autism was geographically isolated for the most part. Palmer et al. (2006) did a correlation analysis in Texas, focusing on mercury emissions. Windham et al. (2006) is a more methodologically sound study, not married to mercury, that looked at the relation between autism and "hazardous air pollutants" (HAPs) in the San Francisco Bay Area, California.

Lewandowski commented on Palmer et al. and noted several important considerations that invalidate it:

  • Mercury deposits in the western U.S. (including Texas) come mainly from Asia. Texas emissions are deposited in the eastern U.S.
  • The primary source of mercury exposure in humans is fish, and fish are also exposed to mercury elsewhere.
  • Autism incidence was already known to be linked to degree of urbanization. A correlation with many factors, including other chemicals, is thus expected.

The response by Palmer et al. to Lewandowski was unsatisfactory and essentially conceded that the correlation was inconclusive and should be studied further.

Another point is that a major source of environmental mercury exposure is coal burning, which is not as much an issue as it used to be.

The rough link to degree of urbanization needs to be emphasized. Because of this link, if a researcher were to look for a correlation between, say, consumption of French fries and the administrative incidence of autism, they could very well find a correlation, giving credence to the autism fries hypothesis. So it's not sufficient to document these types of correlations.

Windham et al. admit that there could be "uncontrolled confounding" in their study. This confounding may or may not be environmental in nature.

The California Department of Developmental Services (CDDS) provides sufficient information to determine if regional differences in autism prevalence are real. I will try to use this data to show that one should be skeptical of the correlations found by Windham et al. and any correlations where degree of urbanization acts as a proxy.


I will use data provided to me by CDDS. Readers can obtain this data by emailing datax at and asking for the file named CDER Qrt Data.XLS with autistic client characteristics for each regional center (or post your email address in the comments section). Readers may also find the regional center map useful to follow the analysis.

Lack of Equivalence

In order to make a valid claim that there is correlation between autism and something else, a researcher should make sure they are comparing apples with apples. If what "autism" is in one region differs from what "autism" is in another region, then the correlation analysis has no merit. One way to tell if "autism" is equivalent across regions is to compare the characteristics of those diagnosed with autism. Table 1 lists autistic client characteristics in regional centers across the Bay Area.

Table 1: Q4 2005 Autistic Client Characteristics By Regional Center
Regional CenterAutism-Epilepsy RatioEpilepsyMRSevere BehaviorsIn MR Population
Golden Gate0.568.26%39.11%36.2%6.70%
North Bay0.596.90%31.60%17.84%5.92%
East Bay0.766.30%22.05%13.97%5.99%

The autism-epilepsy ratio turns out to be a good indicator of prevalence for a region. (This could also be autism-cerebral palsy ratio, or the reader could determine the total population served by a regional center if so inclined). The last column of Table 1 represents the recognition of autism in the population with mental retardation. I have found that differences in this rate of recognition throw off the other proportions. So in this case we're fortunate that differences in this proportion in Table 1 are small.

Table 1 shows that there is inequivalence between regional centers. But it is also of interest that in regional centers with higher autism prevalence, the proportion of autistic clients with mental retardation, epilepsy and severe behaviors is lower. A more dramatic example of this phenomenon can be observed in Table 2.

Table 2: Q4 2005 Autistic Client Characteristics By Regional Center
Regional CenterAutism-Epilepsy RatioEpilepsyMRSevere BehaviorsIn MR Population
San Diego0.626.82%54.73%23.42%9.26%
North LA1.506.06%22.56%21.11%9.59%

We can see that North LA has an autism prevalence about 2.4 times that of San Diego, but proportionally, about 2.4 times as many autistics have mental retardation in San Diego as they do in North LA. (I chose these two regional centers because, again, their recognition of autism in the mental retardation population is similar, which allows for a fair comparison).

Evidence of Catch-Up

The differences in autism prevalence that exist today across regional centers were widened considerably during the "autism epidemic" of the 1990s. In fact, the "epidemic" was mostly confined to the Los Angeles area. These observations are illustrated in the following figure [courtesy of CDDS]:

Presumably, something unusual happened in the 1990s in LA that did not happen nearly as much elsewhere, e.g. there was a significant upsurge of environmental pollution. Whatever it was, it's apparently leveling off now. What's strange is that the regional center with the lowest prevalence of autism in the state, Central Valley, currently has an annual autism caseload growth of 23.94%, whereas the regional center with highest prevalence in the state, Westside, has an annual autism caseload growth of 8.7%. (Note that annual caseload growth is 10.5% for the state as a whole). One would have to assume that there has been a very significant shift in emissions of environmental pollution, with Central Valley all of the sudden surpassing Westside in emissions and growing much faster. It is far more likely that Central Valley was behind in its recognition of autism and is now simply catching up.

Why Only the Bay Area?

It is unfortunate that Windham et al. decided to only look at the Bay Area. It would have been interesting to see their take on why the prevalence of autism in the Far Northern regional center (apparently a very sparsely populated area) is about 1.7 times that of the Central Valley regional center (which has a city, Fresno, population 400,000). Or why the prevalence of autism in East LA is slightly lower than that of Westside. Or why the South Central regional center has a considerably lower prevalence than all of the surrounding regional centers in the LA area.


Before we can move forward with regional prevalence differences and whether they correlate to various environmental triggers, it is necessary to determine that these differences exist in reality. Recognition of autism in the population with mental retardation is the key variable, because once this proportion is adjusted, it appears that the rest of the variation can be explained through differences in the definition of "autism". So I would recommend carrying out the following pilot study. Random groups of clients with mental retardation from Central Valley and Westside should be selected (at least 100 each). An evaluator blind to region and to pre-existing diagnoses would evaluate the clients for DSM-IV autism. This would tell us if it is true that Westside has significantly more autism than Central Valley in reality.


  1. Joseph,

    Xie xie, Ni.


    I hope you will copy this to the authors of the Bay Area pollution paper.

  2. Hi Joseph
    I think that your work is well presented. However, I have some comments about. To consider causation for an exposure
    1- A correlation must be found
    2-A biochemically plausible mechanism of generation of imbalances must be, hypothesized, analyzed properly and proven.
    The french fries hypothesis (and whatever you want) could be analyzed in terms of correlation (1) but not in terms of (2) because there are no scientifically sounded hypothesis that can explain the supposed correlation
    You can find here an interesting discussion
    Some quotes from here in " ".
    “The notion that educational researchers believe that evidence about cause and effect will be any stronger in causal comparative research than in correlational research appears to be an overgeneralization.
    Many, if not most, of us hold a viewpoint along the following lines:It is generally accepted that the only way to _establish_ cause and effect relationships is via experimental research; However, either causal-comparative or correlational research may indicate _possible_ causes, which can be tested further through experimentation.”
    “Causal-comparative, as the name implies, are (or should be)
    grounded in causal hypotheses and _designed_ to explore evidence of causal relationships within the constraints of a comparative context.”
    My comment
    My point is that in general epidemiology and this kind of studies are correlational. I do think that we need more Causal-comparative, experimental ( or clinical in this case) research, in ASD, considering all the confounding variables.

    Another quote ”What causal-comparative _attempts_ (or can attempt to do) is sample from real-world domains and identify possible patterns of influence (based on correlational-related techniques)” “The key word is _attempt_ and perhaps it would be more accurate to say that they attempt to find consistencies between observed patterns and hypothesized causal relationships and then argue that this suggests or supports the validity of the hypothesized relationships.”
    "The statement is criterion one in the commonly cited three criteria for establishing causation: 1. a relationship of some form between two events or variables, 2. time ordering, and 3. lack of alternative explanations).
    1. Causation is neither necessary nor sufficient for observed nonzero
    correlation. 2. Observed nonzero correlation is neither necessary norsufficient for causation.
    There is a well-know situation in which a causal effect exists, but the correlation is zero: the suppressor effect. Sufficiently tangled relationships might also result in an controlled coefficient with a near-zero estimate when the true effect is nonzero (i.e., when key independent variables are poorly measured or highly correlated)”
    My comment
    And this is the situation that I do think it happens in the majority of the published epi studies about ASD.Key independient variables are poorly measured or are highly corelated and non-analyzed.

    You said
    “What's strange is that the regional center with the lowest prevalence of autism in the state, Central Valley, currently has an annual autism caseload growth of 23.94%, whereas the regional center with highest prevalence in the state, Westside, has an annual autism caseload growth of 8.7%. (Note that annual caseload growth is 10.5% for the state as a whole). One would have to assume that there has been a very significant shift in emissions of environmental pollution, with Central Valley all of the sudden surpassing Westside in emissions and growing much faster. It is far more likely that Central Valley was behind in its recognition of autism and is now simply catching up.”
    I wonder
    1-what is the exposure to xenobiotics in patients to these different regional center, considering them genetically susceptible to xenobiotics?
    a-kind of antibiotics used in infections by local peditricians. Management of viral/bacterian infections of local peditricians. Management of schedule of vaccinations.Exposure to outbreaks of infectious diseases.
    b-exposure to air contamination, food quality and water contamination.
    c- local exposure to hazardous materials (petrochemicals, PCS, PAHS, others)
    2-What is the composition of the population today and how has changed in time since 1992?
    3-New industries of any kind located last 15 years near the residences of the patients of the different regional centers. Use of chemical compounds in these industries. Management of residuals solids, liquid and gaseous.
    4-Changes in diagnostic criteria for ASD and higher awareness to the condition in younger children. Attitude of health providers and doctors to ASd..
    Until these confounding variables are not going to be properly addresed I do think it will be very difficult to know about the TRUE impact of environmental pollution in ASD and to separate from other xenobiotics - that can impact as such in susceptible children for me.

    María Luján

  3. Thanks Camille, María and NM. I've emailed Dr. Windham and invited her to come and discuss the findings.

    María, I'm not saying correlation studies are useless. But awareness/cultural confounding in autism appears to be special. The fact that there's inequivalence between regions is in itself too significant.

    On your last point, what I noted is that not only is there a difference in prevalence between regional centers, there's a difference in caseload growth in the opposite direction which should also be explained. And an awareness/cultural explanation seems more likely on this as well.

  4. Hi Joseph,
    Thank you for your comment. I read many of your posts and links to articles with great interest. I am open to neurodiversity and respect your position, as far as I can understand it. I also realize there is another side to the mercury connection, I fully accept some do not agree, and as it is with a lot of scientific questions, one can present data to support one or the other side of any some extent. However I wonder why an autistic person wouldn't be concerned about the possibility that mercury/thimerosal could be connected, even if certain studies proved unreliable? What about all the parents of children who noticed the change in their children after vaccines, or the fact that thimerosal isn't really necessary, why is it in there (or was it, if it's truly being removed now)? I am new to this side of the debate - esp. as it comes from autistic people themselves - and am interested to learn more. As I said, I will continue to read more about it; if there is one or two particular posts or articles that would help, please direct me there. And feel free to email me at I am interested to perhaps write an article about this contrasting opinion. Thanks for your work.

  5. Thanks for stopping by A.V.Michaels.

    However I wonder why an autistic person wouldn't be concerned about the possibility that mercury/thimerosal could be connected, even if certain studies proved unreliable?

    That's a good question. I can't speak for all autistics, but my guess is that intuitively it doesn't make sense to most of us. Now, finding out about causes has its place, and the neurodiversity movement favors sound autism research. While disability rights still apply whatever the causes are, it's inevitable that more stigma will be associated with certain types of causes. So if some theories are clearly crap and we feel those theories are detrimental to the interests of autistics, many of us will obviously take on them head on. The other issue is that autism is a field that is attracting a lot of quakery. Many of us in the neurodiverstiy movement happen to be skeptics too.

  6. What about all the parents of children who noticed the change in their children after vaccines, or the fact that thimerosal isn't really necessary, why is it in there (or was it, if it's truly being removed now)?

    I'm not opposed to thimerosal being removed from all vaccines. There might be some technical issues, but I think it's better to conclusively know what happens when all thimerosal is removed (and I'd say we already know what happens: nothing). As to regression, note that it also occurs in Rett syndrome, a clearly genetic condition. Regression occurs in Landau-Kleffner syndrome, except that it occurs later than it usually does in autism. There's a paper that offers some theories regarding anomalous EEGs and causation similar to Landau-Kleffner. Before the vaccine theories, parents would associate regression mostly with stressful events, such as the birth of a sibling.

  7. Regression also occurs in normal development for that matter.

  8. That's true. For example, NTs start to lose their ability to acquire a new language after the age of 6.

  9. Dr. Windham took the time to comment on the post and answer additional questions via email. I thought it was only fair to give her a chance to respond. Her comments appear in bold. Mine in italics.

    I did take a look previously. I appreciate your interest in the work and my comments.

    Just to let you know, we chose the Bay Area for the study because that
    is where we have been conducting multi-source surveillance, e.g. we
    include more sources of ASD cases than just DDS, such as Kaiser, and we conduct chart review to verify a standardized case status.

    I looked a week or so ago, but as I recall you were using DDS data to
    discuss geographic variations, which we don't really think is the most appropriate use of that data, because the data comes from local regional centers that are geographically defined. However, administrative practises may vary by RC, so geographic differences don't necessarily reflect environmental differences.

    I read the study and you do have more granularity than is available in the public data from CDDS (6 counties, right?), and I do not have the specifics on where each of the children came from. But in your sample of autistic children, clearly, counties with higher prevalence will be represented more than those with lower prevalence. So if there's a correlation between environmental pollutants and degree of urbanization in a county (which would be expected), and if there's also a correlation between degree of urbanization in the county and administrative prevalence (which is known to happen), then children in your sample will tend to come from mothers exposed to environmental pollutants. This would be independent of whether pollutants are a risk factor for autism. Now, client characteristics across regional centers in the CDDS suggest that differences in prevalence aren't real, but only administrative. So while there is a rough correlation with degree of urbanization (not everywhere though) this correlation seems to be better explained by differences in the recognition of autism. Seem plausible?

    Another point is that even if a RC serves several counties, there could be inequivalence between the counties. That is, the RC's administrative practices are not necessarily the only confound. Differences could be due to awareness and medical culture in each county.

    Our analysis was not done on a county-level so that focus doesn't matter much. Of course more urbanized areas tend to have higher pollutants and will have
    larger populations. However, we did not just look at number of autism cases per county, or any other geographic unit for the matter, nor did we look at rates of
    autism. If we had done the later, the larger population denominator in urban areas should have somewhat accounted for finding more cases in such areas. Our
    primary analysis was a case-control analysis, e.g. we compared where cases were born vs. non-cases, and then the pollutant levels in those areas. We have almost all cases in the 6-county area from our surveillance system and a population-based comparison of non-cases, so would expect them to distribute similarly if no other factors played a role, e.g. BOTH should come from more
    urban areas where this is greater population.

    There are some limitations with using this type of data, as explained in the paper, but more because they don't take into account individual patterns of
    behavior or exposure, not because of this urbanization issue. It is not clear urbanization and administrative practices are directly correlated and we have
    cases from non-DDS sources. I don't think this one study is sufficient as I also said. It is always possible that results are due to statistical chance, but they
    also can't be ignored. If anything, scientists tend to be pretty cautious...

    This is where I wouldn't agree [i.e. that "BOTH should come from more urban areas where this is greater population"] . Children in the control group would of course be found more often in urban areas, and so would autistic children. But with autistic children, the urban representation should be more pronounced. This is because prevalence of autism correlates with urbanization, and prevalence calculations already account for population density. (To clarify this point, note that prevalence of NT does not depend on population density in any significant way). As an example, suppose region # 1 has a total population of 10,000 children and a prevalence of 30 in 10,000; suppose region # 2 has a total population of 20,000 children and a prevalence of 60 in 10,000. You'd find 30 autistic children in region #1 and 120 autistic children in region #2; that's a 1:4 ratio. Your controls would have a 1:2 ratio. This shows that a prevalence correlation to population density does matter when comparing autistic kids to controls.

    This is a large prevalence difference to claim is attributed to urbanization. I am not familiar with literature that establishes a strong correlation of autism
    prevalence with urbanization per se. If you are attributing that to better medical services for detecting or treating cases, that still seems extreme and
    is very unlikely in our largely metro region, perhaps in other areas. Also, in our study children had to be born in the area, so there are less issues regarding moving to areas for/after diagnosis.

    Lauritsen (2005) had previously found that "the risk of autism was associated with increasing degree of urbanisation of the child's place of birth and with increasing paternal, but not maternal, age." Hoshino (1982) found that the prevalence in urban districts was considerably higher than in rural districts in Japan. The findings of Mandell (2005) in regards to diagnosis age and income, and availability of specialists are probably significant as well.

    Contrast this with Vignatelli (2005), which finds no significant differences in the prevalence of epilepsy in Italian urban vs. rural regions. Incidentally, there has not been an epidemic of epilepsy in California, nor will you find significant differences in the prevalence of epilepsy across regional centers in the CDDS. (One might expect this, considering the association of epilepsy with autism).

    The CDDS data shows, for example, that the prevalence in the Westside (West LA) regional center appears to be about 5 times that of the Central Valley (Fresno) regional center. In fact, LA has a high prevalence of autism overall, whereas the rest of the state does not. So my example is actually pretty tame. San Diego is significant too because they have a pretty good recognition of autism in the population with mental retardation (over 9%) but their prevalence of autism is still relatively low. So while I don't have detailed data on prevalence around the Bay Area, except for 3 or 4 regional centers, it's not difficult to extrapolate and note that this has to be a significant confound; also considering that your finding is only of 50% increased risk, correct?

  10. Thanks, Joseph. It looks like you've given this researcher something to chew on.

  11. You say that coal buring is not so much of a problem now. When you look at the EPA's own database it appears that it is just as much of a problem, however they now measure Mercury "compounds", which are also highly toxic.

    Here is the data. I was wondering why you beleiove it's not so much of a problem now?

    The amount of Mercury emissions by year (In pounds).

    1988 296,299 - 20,328 in Mercury compounds

    1989 157,679 - 60,431 in Mercury compounds

    1990 204,341 - 37,293 in Mercury compounds

    1991 137,098 - 40,018 in Mercury compounds

    1992 61,471 - 195,527 in Mercury compounds

    1993 32,065 - 59,640 in Mercury Compounds

    1994 26,900 - 29,040 in Mercury Compounds

    1995 28,698 - 210,395 in Mercury Compounds

    1996 28,303 - 32,435 in Mercury compounds

    1997 39,916 - 27,879 in Mercury compounds

    1998 290,454 - 9,061,942 in Mercury compounds

    1999 142,952 - 3,110,087 in Mercury Compounds

    2000 99,530 - 3,643,198 in Mercury compounds

    2001 114,561 - 4,701,124 in Mercury Compounds

    2002 92,470 - 5,161,814 in Mercury compounds

    2003 68,531 - 7,332,965 in Mercury compounds

  12. Apparently there's some conflicting data. Those last ranges you listed look extremely broad for some reason too.

    "Significant decreases in total mercury observed in the last 15-20 years of the ice-core record indicate reductions in emissions that correspond to the time period when the Clean Air Act and other emission reduction strategies became prominent." (source)

    "We present stratigraphic data from a suite of Minnesota lakes that indicate mercury deposition peaked in the 1960s and 1970s, at least for the upper Midwest." (source)

    "Taken as a whole it is projected that total annual nationwide MSW incinerator emissions of mercury could decrease from about 97 tonnes (1989 baseline uncontrolled emissions) to less than about 4 tonnes in the year 2000." (source)

    "A comparison of today's total gaseous Hg levels in air and the total Hg concentrations in precipitation with the ones found earlier, shows a clear decrease with time." (source)

    "The inventories of global anthropogenic emissions of mercury for years from 1979/1980 to 1995 suggest a substantial reduction in the 1980s and almost constant emissions afterwards. In contrast to emission inventories, measurements of atmospheric mercury suggest a concentration increase in the 1980s and a decrease in the 1990s" (source)

  13. Joseph,

    Mercury emissions were reduced because since the 80's they now record Mercury and Mercury compounds. So yes pure Mercury emissions did go down, however compounds sky rocketed.

    What are Mercury compounds? Well only Methyl mercury the most nurotoxic version of Mercury. My argument against Ethyl Mercury has always been it Nero toxicity is far lower than Methyl. And we know women have high levels of Mercury at birth due to umbilical chord studies. We also know some people with ASD excrete heavy metals differently (less found in the hair, more in the teeth).

    “Mercury exists in three forms: elemental mercury, inorganic mercury compounds (primarily mercuric chloride), and organic mercury compounds (primarily methyl mercury). All forms of mercury are quite toxic, and each form exhibits different health effects.”

    The data i posted is actual data as to how much is being pumped in to the environment.

    As for your links. I have not looked in to them in much detail, however;

    1) The first link is a local study of ice cores. There could be many reasons why local levels have been reduced. As we can see nation wide more is being pumped out.
    2) Again local. There could be hundreds of reasons for a local decline in ice cores when we know that more was entering the environment nationwide.
    3) “Taken as a whole it is projected that total annual nationwide MSW incinerator emissions of mercury could decrease from about 97 tones (1989 baseline uncontrolled emissions) to less than about 4 tones in the year 2000.” As you can see from the data they were substantially incorrect in their predictions, and that is including just Mercury and conveniently forgetting about the compounds.
    4) Local Scandinavian study.
    5) This report appears to have been based on the EPA data and again does not take in to account compounds.

    Then there ar studies such as this one, which has also been duplicated in other areas of the country.

    "On average, for each 1,000 lb of environmentally released mercury, there was a 43% increase in the rate of special education services and a 61% increase in the rate of autism. The association between environmentally released mercury and special education rates were fully mediated by increased autism rates." uids=16338635&query_hl=1&itool=pubmed_docsum

    My point is that you claim that it is expected, that people will move to enviromental Mercury when the vaccine debate is over. However much of the best research has always been looking in to links with enviromental Mercury. Even if there is no link, we are at a stage where pregnant women are advised not to eat many types of fish while pregnant, and no real way to clean up the Mercury which stays in the enviroment for an extreamly long time.

  14. Back2basics: The study you cite is Palmer et al. which you will note is mentioned in the post, and has been addressed by Lewandowski. That study documents a correlation between autism and degree of urbanization, but fails to consider factors resultung in that correlation which may act as confounds (e.g. availability of specialists). What I explain in the post is that it's easy to demonstrate in California that "autism" in one region is not the same as "autism" in other regions, which invalidates correlations to factors that depend on population density or degree of urbanization.

  15. Joseph,

    Absolutely it's very difficult to draw a conclusive link to mercury; it could well be some other factor of urbanization, including better diagnosis. In the same way the recent (Californian) study that suggested a link to Thermisal, could quiet easily have been caused by outside factors. Taking one study and trying to correlate factors, or to dismiss correlation of factors are both equally as invalid.

    However when you link in the excretion issues, Mercury levels in the hair and teeth are easy to measure, and we find less in the hair in ASD and more (3 times as much) in the teeth, then it is either coincidence or there is more to the link.

    4) Mercury in Baby Hair: A recent study found that children with autism have 1/8 the normal amount of mercury in their baby hair, suggesting an inhibited ability to excrete mercury which is consistent with their high oral antibiotic usage. Also, they found that the severity of autism had a strong inverse correlation with the hair level, such that the children with the lowest levels of mercury in their hair (least excretion) were the most severe. We will also report on our replication study, which we carried out with the help of NIH and researchers at MIT, and which generally supports the Holmes' et al study.

    "5) Baby Teeth: We will present our study with the Un. of Texas, which evaluated the level of mercury, lead, and zinc in the baby teeth of children with autism vs. controls, and found that children with autism have 3x as much mercury in their baby teeth.

    6) Glutathione: We will briefly review the published study by Jill James, which found that children with autism have low levels of glutathione, which is required for excretion of mercury and other toxins, so it strongly suggests that children with autism have a decreased ability to excrete mercury."

    Glutathione could be important, as females have higher levels of Glutathione, which could be why we see more males with ASD than females.

    For the record - i do not beleive ASD has one causation. I personally beleive that there may be some parts to the spectrum that are extirely genetic (aspergers for instance). I also have not made up my mind, anybody who has does not follow the research. There are no firm answers yet.

  16. Ugh - there's more mercury in the teeth, less in the hair. If an autistic has high mercury levels it's a problem, if they have low mercury levels it's an excretion problem...every study is contradictory.

  17. anonymous,

    You misunderstand what that means.

    If excretion is working correctly we would see more in the hair, as that is correct excretion. As there is less, the natural excretion is happening more slowly, which is why we find it in the teeth not the hair.

    It's certainly not a case of too much / too little, its where we find it. We can gain information from that and draw conclusions. And it's certainly not contradictory research. The excretion process is very well understood. heavy metal testing is very accurate. So when we find more in the teeth and less in the hair we can basically be assured that there is an issue and Mercury is staying in the system for longer.

    There is no research at all that contradicts the tests done on hair and teeth. None. Nobody is questioning the test procedure as it's well established.

  18. The Holmes baby hair study is quite flawed and contradicts Fido (2005), Ip (2004) and Wecker (1985).

    Autism diva posted a critique of the Holmes study, and so has Prometheus.

    Science that comes from those trying to prove a connection between autism and thimerosal is notoriously flawed and suspect. I'll leave it a that.

  19. In the same way the recent (Californian) study that suggested a link to Thermisal,

    Ah, this is Geier & Geier (2006) I presume. I've discussed this in detail. Search my blog for Geier. You should also Google the Geiers for additional context.

  20. Hello,
    Joseph cites Hoshino as an example of urbanization relating to high autism in and of itself. If I understand, this was regarding the question of whether diagnostic differences account for the urbanization link. It might be worth noting that the study considered this and discuss this in terms of initial querry and case-recovery rates. The pattern does not support the idea that diffrence is due to different medical care practices or more services, etc. As the author says: "Therefore, it would be difficult to explain simply the difference in prevalence rates of the two areas with respect to the difference of medical or welfare services,” (p. 123). This finding of higher rates associated with urbanization fits with an exposure scenario, not with diagnositic differences.

  21. Anon: Notice that Hoshino et al. is not the only study I mentioned.

    The differences are also not necessarily related to wealth or medical services alone.

    While there's no study that I know of that tells us for sure whether regional diagnostic differences are an artifact or environmental, I've looked at data.

    There's clearly an association between (log of) population density and the administrative prevalence of autism (e.g. in California). Once you adjust prevalence for this association, that is, you determine which counties have prevalence that is higher or lower than expected given their population density, and you try to associate this adjusted prevalence to environmental pollution, the association disappears.

    If you attempt the exercise in the opposite direction, the association remains. It seems obvious to me that population density mediates the apparent association between autism and environmental pollution.