Frank Belantonio said the following on 2/5/2008 5:15 PM:


I appreciate your speedy reply!  I guess I assumed you had seen the two posts and had some thoughts about them already.  As I see it, the main questions raised in the blog posts are these:

1.  Have you been able to substantiate the assertion that there is a process of mercury excretion (as opposed to passive absorption) into hair?
2.  Do you think it was wise to build on the assumptions of the Holmes study, in light of the findings of the Gundacker study and the NHANES mercury level measurements?
2.  Do you disagree with Autism Street's assertion that the Chrysochoou paper fails to support your contention that mercury poisoning has sometimes been confused with autism, and have you found any other instances of mercury poisoning being misdiagnosed as autism?
3.  How do you justify relying on the Medical Hypotheses article when there was other literature that was on point, newer, and had been peer-reviewed?
4.  What is the basis for stating there is a *consensus* in favor of the proposition that there has been a marked increase in the incidence of autism?  (I assume the blog authors are thinking here of a consensus among autism researchers rather than among autism activists or lay journalists.)
5.  Given that your paper acknowledges mercury in hair having being found to be both both higher and slightly lower among autistics, do you still believe a one-tailed test was warranted?

6.  Are you aware of any evidence that would support the contention that autistics are impaired in excretion of mercury via the kidneys, liver, etc.?

I think this is all very interesting.  Thanks,


Hi Frank,
Thanks for your two emails and for clarifying your questions. I was reading the blogs for a while, but have pretty much stopped..... However, you have the distinction of being the ONLY person who has bothered to take me up on my invitation to write and ask me to clarify points of confusion regarding things written about our article. It is a bit of a problem because you have asked six questions, each of which I could write a book about, and then when I read the blogsites you refer to, I have even more to say.

Truly, I am honored that instead of just reading information on an anonymous blog site, you have taken the step of asking the expert on the article directly. Thank you for doing so. Thus, I did take some time and read though some of the recent Autism Street and other blogsite postings to try and understand the basis for some of your questions more fully. As I begin to try and respond, it becomes very time consuming. To respond carefully to all the points of confusion on the blogsites would take several weeks of fulltime work.  I have tried to respond to some. I hope you will consider how much time I am putting into answering your questions and come away believing I really am trying to help clarify --  and do not take future silences as admission of “being had,” but realize I just do not actually read the blog sites on a regular basis.


To start, it is clear even from a quick read that significant misinterpretation remains and that the various statistical analyses that were performed by Hitlan and me are still not grasped by all the authors on the blogosphere.  This is understandable. I tell beginning majors when they embark on familiarizing themselves with peer reviewed research that they will not be able to begin to understand statistics presented in peer review research articles until they have taken the major’s required statistics and research methods sequence on their own (which is 8 credit hours, or about three regular courses).  Even after this, they will only be able to understand the more straightforward statistical analyses. Peer reviewed articles often employ statistics that require a couple of specialized courses in graduate school to fully grasp, and even then sometimes reading within a specialized topic of statistics that might be relevant. They can and should still be read, but beginning students have to trust the author about what the statistics mean. There are statements with in blog sites like “we disagree” in reference to straight up factual statistical information.  This is not a good starting point for any serious academic discussion.  The FAQ website was meant to help those who are really curious but who might be misled by things they are reading on various blogsites which serve to misrepresent our article.  To that end, I really feel that 1 and 3, are covered in either print articles, the FAQ web site or both.  I will direct my attentions to the remainder, here goes:


2.  Do you think it was wise to build on the assumptions of the Holmes study, in light of the findings of the Gundacker study and the NHANES mercury level measurements?

2a. Yes. The data as they are seem inexplicable to me otherwise. I am open and interested in other ideas that would explain the observed pattern—but it has to explain the pattern of the data. Be sure you are grasping what the stats actually say and not relying on third party interpretations. 


2. Do you disagree with Autism Street's assertion that the Chrysochoou paper fails to support your contention that mercury poisoning has sometimes been confused with autism, and have you found any other instances of mercury poisoning being misdiagnosed as autism?

2b. Essentially this is a reasonable point, but to me is just not a very important or interesting one --this is my opinion. I have gathered that this opinion is not shared by all, and I think this itself is not unreasonable.  I see that even within the blogosphere there are those that have read both Chrysochoou and our article who point out that the actual difference between the Chrysochoou’s  and Hitlan and my words might well be seen as slight (that the point is essentially frivolous). I think this issue is debatable. I do have some problems with the way this is being debated, though. As I read the most recent description posted on Autism Street about this “issue,” I have to say I feel the bloggers description of the case as reported is getting a bit far afield. To wit, their most recent post on the matter reads (Jan 27) :

“According to the authors, the diagnosis “remained obscure”. There is no preliminary diagnosis of “autism of unknown etiology” mentioned. In fact, there is no “preliminary diagnosis” mentioned at all. Furthermore, there is absolutely no “autism” (of any kind) “later shown to be due to mercury exposure”. In our opinion, the use of this citation (footnote #4) to support DeSoto & Hitlan’s claim, looks like a misrepresentation of the Chrysochoou et al. paper …”

And they go on to demand that if we be honest scientists we should retract our article as a consequence of our misstating Chrysochoou.

Here is what Chrysochoou et al DID say about the case:


“An 11-month-old Swiss boy was brought to his paediatrician. He had been in good health and had developed normally until then, but his parents mentioned that over the previous 2 weeks, the child no longer laughed or played, was becoming more and more restless”…. It was also noted that the child, “showed stereotypic movements of the hands (kneading)…”

(To be fair, physical examination also showed some skin peeling, brisk reflexes, and that “the child sweated profusely” along with the new inability to sit and crawl, lack of interest in surroundings and the new propensity to repeatedly bite himself.) It was stated that he, “would sit in a crouched position, had a sad and apathetic look, took little interest in his surroundings and would occasionally bite his hands and feet.” It is stated that the diagnosis was obscure and the child, “was referred for further evaluation of severe psychomotor regression with autistic features of unknown aetiology” .   

          Later, it was discovered the symptoms occurred because he had been mercury poisoned.  

We stated this:

A review of what is known about the neurotoxic effects of mercury is beyond the scope of this paper,2 but the observable symptoms of acute mercury poisoning have been reported to match up with many of the problems observed in autism.3 Furthermore, mercury poisoning has sometimes been presumptively diagnosed as autism of unknown etiology until the mercury poisoning has been uncovered.4


(Note that an additional separate source --besides Chrysochoou case study-- was also included for the symptom matching up statement.) I also saw from the blogsite that other authors whom I have never met or seen or previously read have apparently taken a similar meaning regarding the Chrysochoou case report as we have (Rossignol, “Evidence of Metal Toxicity” Autism One, 2007). That said, I really might write this one sentence slightly differently if I had it to do over again ... for example, I might instead write that mercury poisoning has “sometimes been confused as autistic behavior of unknown etiology”, or something like that ... the point would be the same, to most people. I think it would make no difference in the paper whatsoever, which is why I think it is not very important.  To suggest that such a disagreement about one statement within a short lit review would merit a retraction of the article does reflect a momentous misunderstanding of when and why retractions are done.


4.  What is the basis for stating there is a *consensus* in favor of the proposition that there has been a marked increase in the incidence of autism?  (I assume the blog authors are thinking here of a consensus among autism researchers rather than among autism activists or lay journalists.)
4. Ok, I regret my saying there is consensus in my short post on the Autism Street blog site. This is fair criticism of me. This is an interesting question that deserves more research. For example, I can agree I would need to back this up with citations if I wanted to try and persuade the readers of the site. Of course, this has nothing to do with the journal article, only saying that I should have been more careful in my response to the blog authors.  J


5.  Given that your paper acknowledges mercury in hair having being found to be both both (sic) higher and slightly lower among autistics, do you still believe a one-tailed test was warranted?
 5. Your question number 5 is trying. To what statistical test are you referring?  All of our statistical analyses are appropriate and correctly done. Please try to read our article carefully and make sure you understand all the statistical tests employed. I did take the time to look through the blog site for reference to this, and found some things posted in the blogoshere that might have served to confuse.


posted as critique of our article by trrll — 28 January, 2008:

You should only use a one-tailed test if you are absolutely certain that it would not be meaningful in any way if the difference went in the other direction–meaning you would ignore the result.

Does anybody seriously believe that it would not be considered meaningful if levels of mercury in hair were elevated in autistics? After all, elevated levels of heavy metals in hair can constitute evidence of heavy metal poisoning. Given that there is no evidence that hair constitutes a major route of mercury excretion, a strong case could be made that elevated hair mercury would constitute evidence that excretion of mercury by other routes is impaired.

Please note this passage (DeSoto and Hitlan, 2007) for anyone who has not taken the time to read our 2007 article and are drawing conclusions about what our position might be based on passages like the one above:

“Whether to use a one-tailed test or a two-tailed test can be decided based on considering what would happen if the results ended up in the opposite direction of what one suspects. In this case, it would mean that the blood mercury levels were lower in the autistic group. Would this support the original hypothesis? (No!) However, if this were to happen, that is, if the autistic group were significantly lower in their blood mercury levels than the normal group, the researchers would find themselves in the incongruous position of having to accept their hypothesis that autism is related to elevated levels of mercury in the blood”(p. 1311)


The person who posted the above appears to be critiquing the idea of using a one-tailed test on the exploratory statistical analysis related to hair-- he/she writes:  “Does anybody seriously believe that it would not be considered meaningful if levels of mercury in hair were elevated in autistics?”  We can not emphasize enough the importance of not relying on anonymous bloggers as sources of critique of peer reviewed research. We did not use a one-tailed t-test on the hair related exploratory analysis. It would not have been appropriate to use a one-tailed t-test on hair for the reasons we stated in 2007, are reviewed in our article and which are echoed by trrll.  And, of course, the blood and mercury connection was statistically significant in both our one-tailed and two-tailed statistical tests, both of which were both reported (DeSoto & Hitlan, 2007). This is actually representative of many “critiques” posted on these not-refereed free-for-all websites. They sometimes reflect false accounts or misunderstanding of the article. Then people read it and post “good point” – but it is not a good point at all because it is criticizing something that was not even done…. As a person who values honest debate and critical thinking (higher education), it is not fun to read.



Comment by Prometheus — 27 January, 2008 @ 4:38 pm

One small nit that you left unpicked. DeSoto and Hitaln (sic) claim: “Because there has been a several-fold increase in environmental mercury exposure, the hypothesis that the rise in autism could be related to an environmental increase in mercury levels is a reasonable one to pursue.” However, they fail to support their assertion that there has been a “…several-fold increase in environmental mercury exposure…” In fact, several recent studies [e.g. Schuster PF et al, “Atmospheric mercury deposition during the last 270 years: a glacial ice core record of natural and anthropogenic sources.”, Environ Sci Technol, 2002 June 1; 36(11) 2303-2310] indicate that environmental mercury “exposure” has been dropping since the early 1980’s. As a result, the “autism epidemic” got its start at almost the exact moment that atmospheric mercury levels began to decrease.

It’s strange the DeSoto and Hitlan didn’t check on that “fact” beforehand. In fact, it seems that there were a lot of “facts” that DeSoto and Hitlan didn’t bother to check.

“Prometheus”- Please tell us who you are, what your line of work is and what your qualifications are. Also please explain how you know what I have read. The report you refer to above is one I have had in my office for about a year and is one I carry with me when asked to speak as an expert of neural development. It is you who do not check your facts when you confidently assert we are ignorant of this research as you sarcastically write: “DeSoto and Hitlan didn’t check on that “fact” beforehand.”  I am aware that the rise in Autism diagnoses began to appear for children who were born in the mid 1980’s and that this ice core data shows a peak about 1980 due to both industrial pollution and a volcano eruption. I am aware that the time for environmental mercury to pass into the foodchain and into animals up the food chain does not perfectly correspond to the time of atmospheric peak. The lag is a few years. Are you aware of this? There was a study done by USGS and METAALICUS that very recently came out that carefully documented the time period for the mercury to enter and be removed from the foodchain, which documented it is faster than at least some have previously argued, but is of course still not instantaneous. The levels of environmental mercury (even with the drop in the late 1980's that occurred as a result of pollution controls) is certainly unusually high in terms of historical averages, and blood levels among women of child bearing age are dangerously high for about 8% of the population, putting about 60,000 children at risk for neurological problems. This is really the key: that mercury levels are high and potentially putting neurological health of children at risk....Are you saying otherwise? If so, be prepared to take on the CDC, the WHO, the UCS, the AAP, NAS and NIH. One might note that the highest mercury levels appeared just a few years before the spike in autism diagnosis, which might be seen as consistent with the time period it takes for environmental mercury to get into the foodchain and onto dinner plates. Do you understand? To be sure—I would not draw this as direct cause and effect, I have never done so in the MANY times I have referred to this study which you say I have not read. I am doing so now to make this point loud and clear:             That you have not made a good point and you have falsely presumed we would be somehow befuddled by the ice core study you refer to.





Frank – in your original long email of pasted blog postings, we read this:


“That’s right, folks. The “poor excretors” had elevated hair mercury. And Dr. DeSoto cited the study that showed it in her own “FAQ”. If that’s not hoisting yourself on your own petard, I don’t know what is.”

We check and find this also was posted by “Prometheus”.  While we appreciate the polite tone and professional wording (ha ha), we would encourage this person to consider the idea of statistical control and to make sure that he has actually understood our arguments correctly before presuming we have hoisted ourselves. Has this person considered absolute levels of hair mercury level would be high in persons with high blood levels, but also that ability to rid the body of heavy metal toxins could make the levels of mercury in the hair lower when blood level is statistically controlled?  These can both be true, and when they are true – it is quite consistent with the idea of “poor excreters”, if one is so inclined. There may be other explanations we have not thought of and we would be open to hearing them. Still, to involve oneself in this debate, one needs to understand multiple effects on one variable and the idea of statistical control to try and tease them apart, think in terms of partitioning the variance.  Should a person be so inclined to consider the effect of multiple variables on a variable of interest, he or she might begin by reading about multivariate statistics, parsing of variance, MANCOVA’s and ANCOVA’s. Although we appreciate good-natured puns and vibrant writing, a person who has been known to criticize others for assuming they know as much as experts (see The Arrogance of Ignorance at might refrain from such confident appraisals until he or she has mastered the nuances of multivariate inferential statistics...(petard: Middle French, from pet expulsion of intestinal gas). Just a thought.



We see that Prometheus expresses sharp disappointment that our article did not include articles he/she feels are relevant, s/he asserts: “It is traditional to try to cite all the relevant literature (or as much as is practical) in scientific papers.”

Yes…. We wonder how this person know this? Does he hold an editorial position on an academic journal?  On what basis is this person so confident of this? This blanket assertion is actually not true at all, at least in our field. It is true that when an author publishes a “review paper” it is customary to provide readers with a broad and very thorough literature review. On the other hand, some articles are short, and it is customary to just give a few brief references selected to illustrate a few key points and to give readers a starting point into the literature if they are interested. A complete literature review would have made the article thirty or forty pages long, and, would have been asked to be shortened by an editor… many times editors or peer reviewers will say the lit review is too long for the paper and ask that it be shortened. When we were preparing the paper, we at first had a longer lit review by a few paragraphs, but decided even that was not too long for the type of article we prepared….. A through lit review would never be attempted in such a paper.  It isn't a plot to hide the "man behind the curtain" as someone suggested.

            In trying to understand who Prometheus is and what his qualifications are, we looked at his site blog site again. We found hidden within this blogsite an interesting exchange, and we think Prometheus' writings are a bit like the pot calling the kettle black.

Prometheus writes in defense of so his critiquing of experts, “In addition, I took the step of consulting with a Professor of Mathematics - who teaches statistics - on the statistical issues.”

Hmmm. Mathematics has many branches, and statistics has many branches. Note thatt the type of statistics one needs to understand our article at the highest level is called “inferential” and specifically the idea of statistical control of multiple variables. To those who think our stats are poorly done: can you explain what it means to use blood level as a covariate, and why was this particular statistical test employed in this case? Why was the t-test of the residuals performed? Can you explain what a residual is? How is the absolute level of hair mercury different that the hair level of mercury after controlling for blood level?  Honestly, several essential points seem utterly lost in the blogoshere analyses of the article. This conversation – at the level attempted by the bloggers --  can not be meaningfully had with persons who do not have the requisite background to have it.

A compromise some may wish to grasp: 

I am not insisting that those who may have been following the debate longer than I have and have rejected the idea that mercury could have anything to so with the problems of ASD must now change everything they believe. If one believes the results are an anomaly, this is something that could be said without much effort. Such a person who has read much on the subject might say, “Based on our understanding of the body of literature relevant to this subject, we doubt that these results could ever be replicated. There is always the chance that these results are just a fluke, a mere chance occurrence. Time may tell otherwise, but we believe this is what has happened.

(Remember-- stats always include an estimate of the likelihood of the results being a chance occurrence-- it is reflected in the associated p value.) To those who find the results offensive, such would be reasonable sort of response: it notes the small but nonetheless real possibility that these results are a fluke --and it is better than spending hours/days/weeks/months continuing to try and find errors in the stats that are not there or disagreeing about wording in the review of the literature and such  (no two authors would ever right the exact same lit review and there would always be those who would say it different).  In my opinion, which is the view of objective experts in biology, these results do suggest that at least a very slight opening of the mind might possibly be warranted, but you don’t have to do so. You don’t have to be logical at all. But our stats were done right.




Closing comments (which also address question 6)

First, we want to make clear that blogsites can be well done and move science forward. We would even (if forced) note that some sites which badly misrepresent our article have done so on some issues. We think that readers should review the following blogsite post which to us epitomizes how blogsite authors can effectively move science forward.


On the other hand, we have tried to illustrate the following: some of the bloggers/authors who are writing as if they wish to be considered as authorities on the matter-- but do illustrate their lack of understanding about the statistics we used and the nature of the data. We grow weary when we read persons write they have caught us in a contradiction, and refer to two statements that are perfectly congruent—but apparently appear incongruent because a nuance is missed by the writer. We then see others write, “bravo” or “that Dr. Desoto is sure an idiot” or “I wonder how they feel having their their bottoms handed to them?” or something, and we just shake our heads in disbelief.  But we wonder -- how do the authors explain the difference in correlation between the hair and blood for normal versus autistic children? Do they have an opinion on how best to proceed with a date set that appears heteroscedacic? If there is no difference in the way autistics as a group deal with mercury in the body, please propose a theory as to why the relationship between hair and blood levels of autistic children was r = .7, but r = .9 for non-autistics. An alternate theory that would account for the observed data would be of interest. I hope that no one expects me to regularly take the time to go through each misunderstanding as it appears and start from scratch explaining what is wrong. Consider the time it takes for the few I am responding to now…. I am neglecting other work to do so, I am doing so (for Frank) because he was the FIRST person to ask any questions of me directly (and I do think this speaks well of you, Frank).

As far as the final question you posed: I do not stake my career on the idea of poor excreters, but so far, this idea seems the only one that explains the observed data in this data set. I did not propose the theory, it was already there and the pattern of the data matches the theory. That is the way it is. To some extent, I have a responsibility to try and help lay persons accurately understand what is and is not said in our article. The important finding is that mercury blood levels might actually relate to autism severity… they did in this data set (it might not in future data sets). But our responsibility has limits. If a reader is still willing to consider the anonymous blogger’s viewpoints of our article as superior to our own and that of independent experts… sadly, there is probably not much more I can do to help.   I am satisfied with my attempt to clarify and can only ask that persons read the original article and, perhaps submit their own analysis to peer reviewed outlets.


Keep your eyes open for March’s issue of JCN: “About Mercury and Autism: A need to clarify” by DeSoto and Hitlan.
And Frank, as the FIRST and ONLY person to have taken up my invitation to ask a question from the blogsites- I hope that my answers are helpful to some, let me know.