Friday, September 17, 2010

Follow-up to My Critique of Pediatric Article Proclaiming No Link Between Thimerosal and Autism

I have been asked to respond to a critique of my comments regarding the study published in the journal Pediatrics, which was used earlier this week to proclaim (once again) that there is no link between thimerosal and autism. I do not know the identity of the person who responded to my comments, so I do not have a name or title to use to help delineate my responses from his or her responses. The person did self-identify as a one of the "shills," so I will use that term to indicate his/her statements, and MPT (my initials) to indicate mine.

If you haven't read my comments from which this discussion stems, here is a link to the previous blog post:

SHILL: Let's look at Ms. Piper-Terry's complaints:

"1. Lower functioning children were excluded because their problems were so severe that it made it tough to assess them."

She makes this complaint based on an interview that the author did:

She manages to quote the author's explanation as to why they did a subgroup analysis excluding the more serious cases, but, like any good quote miner, she managed to leave out a key quote from Dr. Price:
Therefore, an outcome category for AD with low cognitive functioning excluded was created and its relationship to exposure was estimated. The results for this subgroup were very similar to those for the overall analysis"

The reason the low-functioning children were excluded was because there was the concern that it could mask a true association. She also seemed to miss the point that the results for this subgroup are no different than the overall analysis (i.e. that thimerosal was not associated with development of autism even in this severe group). This data can all be found in the technical reports that are publically available, but I doubt she or T4TN has read them.

It is true that I took the quote from an interview of one of the authors. It is also true that I did not include the entire explanation and omitted the information about the establishment of an "outcome category for AD with low cognitive functioning excluded..." My reason for leaving out this information is simple. I do not know how to quantify or interpret the author's statements that "exposure was estimated" or "results for this subgroup were very similar..." There is no statistical information provided in the article to either back up this statement, or to assist in further analysis of the author's assertion. Apparently, we are just supposed to take his word for it.

As for the exclusion of children with low cognitive functioning "because there was the concern that it could mask a true association," I would like to know more about the specific concerns that led to this exclusion criteria. Since it appears from the article that the researchers employed the ADOS as their sole method of criteria in their direct assessment of case-children, the possibility for over-inclusion could be considered a valid concern for children who were higher-functioning and whose scores may be closer to the neurotypical range of functioning on the domains this instrument adresses. For that reason, Best Practices dictates that assessment of children suspected of having an autism diagnosis must also include review of the developmental history, interview with parents (and teachers, if applicable), and review of previous medical history, including pre-existing diagnoses. The ADOS itself has demonstrated high reliability and validity in the standardization process, including demonstrating very high inter-relator reliability coefficients. At the risk of being accused of "quote-mining" the following may provide further clarification:

"The authors reported good inter-rater reliability estimates on the Communication, Reciprocal Social Interaction, Total, and Stereotyped Behaviors and Restricted Interests domains, with intraclass correlations ranging from .82 to .93 (Lord,
Rutter, DiLavore, & Risi, 2001). Test-retest reliability was also good, with intraclass correlations ranging from .73 to .82 on the Communication and Reciprocal Social Interaction domains, and .59 to .86 on the Stereotyped Behaviors and restricted Interests domain. Published validity studies also suggest good predictive validity, with sensitivities ranging from 90% to 97%, and specificities ranging from
87% to 94% for autism/ASD versus other clinical diagnoses" (Lord, Rutter, DiLavore, & Risi, 2001).
For additional information about the ADOS, follow this link:

I have not read the technical report. When I looked at the study and wrote out my observations, it was not yet available for review. Whether or not I will take the time to read them remains to be seen. I really don't know what they could possible reveal that would change my mind about the validity of the study. If there is something in the technical report that is significant enough to negate the impact of the design flaws in the published article, then the authors should have included that information in the article.

SHILL: The technical reports are located here:­eport_Vol1­_090310.pdf­eport_Vol2­_090310.pdf

They also include some other interesting information as well including:

1. Mother's who took prenatal vitamins with folic acid had an approximately 2-fold higher risk (p value 0.0176) of having an autistic child. So, a possible non-vaccine related environmental factor right there it would seem.

MPT RESPONSE: First of all... (and this is a pet-peeve of mine), it is difficult for me to take seriously the comments of someone who is attempting to debate such a highly technical and complex issue as thimerosal and autism, when that person does not understand the basic rules of punctuation. For future reference, "Mothers" is a plural word, indicating that there is more than one "Mother" involved. "Mother's" is the possessive term, to be properly used when you are referring to an object or possibly a trait that belongs to a single (1) "Mother."

Now that that is out of the way... The finding that mothers who took prenatal vitamins with folic acid had a higher risk of having an autistic child is very interesting. Without knowing anything further about this, I would hypothesize that it is possible that these women may have had deficiencies of other B-vitamins (folic acid is one of the family of B-vitamins, which work together in concert), and if they were given folic acid without assessing the levels of B-12 and B-6, in particular, they may have problems in their methylation pathways. They methylation pathway is one of two major detoxification pathways in the body (the other being the sulfation pathway). The methylation pathway is responsible for detoxifying heavy metals, including thimerosal from the body. If the methlyation pathway is not working, thimerosal will not clear from the body as it is supposed to, and as vaccine-makers insist it does. When it does not clear the body because the pathways are not working properly (for example, because of a B-12 or B-6 deficiency), thimerosal deposits in soft tissues, including the kidneys and brain. Thimerosal has an especially strong affinity for the brain because it is very attracted to lipids, and the brain is the most lipid-dense organ in the human body. With regard to the folic acid, it is possible that by elevating folic acid in a group of women who have unknown deficiencies of the other nutrients necessary for methylation to take place, the tendency of the body to increase retention of thimersosal was facilitated. This is, of course, all hypthetical, as I am sure none of these women had their B-12 or B-6 levels checked during their pregnancies. This is, however a very important area of future research, that should not be overlooked as a possible contributor to the increase in autism, especially since oral contraceptives, antibiotics, and steroids ALL deplete B-6. Perhaps we could start looking at WHICH subgroups of women might be more prone to have infants that are at an increased risk of autism from thimerosal and other toxins in vaccines, rather than assuming that if it doesn't happen to everyone, it couldn't possibly happen to some.

SHILL: 2. There were COMPLETELY unvaccinated children both in the ASD group and the control group. That would seem to blow the "unvaccinated children can't get autism" hypothesis out of the water, now wouldn't it?

MPT RESPONSE: If this is a reference to my critique of the article, it is out of place. I never said unvaccinated children cannot get autism. (This tactic is called a "red herring" and is often used to steer a discussion off-topic.)

SHILL: 3. They examine the ASD prevalence as a function of HMO and year of birth.

What they find is that the rate of autism per 1000 is pretty flat and quite near the current estimated CDC values (~1.1%). That would certainly argue against any massive increase in overall incidence or an "autism epidemic".

MPT RESPONSE: No, that wouldn't argue against anything, since the explosion of the autism epidemic began in the 1980s and increased exponentially with the addition of the hepatitis B vaccine at birth, all of which happened prior to the time frame that included the birth dates of the subjects in the study (1994-1999).

SHILL: Ms. Piper-Terry's other complaint is rather perplexing:

"2. The participants were pre-selected by virtue of mandatory physician's consent before they could be recruited into the study."

Yes, the evils of requiring consent. Only a very small number of potential subjects were removed via this mechanism (3.9%). Hardly suspicious of anything, except to the conspiratorial. I find it interesting though that Ms. Piper-Terry failed to castigate all of the parents who refused participation. This was a far larger proportion (32%) and some survey results are included as to why they refused to participate. Where is the conspiracy suggesting that those parents are involved in covering up the "truth"? But why expect consistency, right?

MPT RESPONSE: My criticism is regarding the mandate that the child's PHYSICIAN had to give consent for the family to be contacted about potentially participating in the study. This is very different from the absolutely necessary practice of obtaining informed consent from study participants, and in the case of children and vulnerable populations, obtaining consent from parents or guardians. There is NO established precident in research supporting the mandate that the physician pick and choose which parents of his patients deserve the right to participate in research. There may well be some reason as to WHY phsyicians were given this power to censor who participates and who doesn't, but the article does not elaborate, and as noted above, the technical report was not available when I wrote out my comments. I would still like to know what the rationale was, and may actually have to invest the time to go through the technical reports after-all. At any rate, the fact that physicians were given the power to pre-select which patients were allowed to consider whether or not to participate presents a HUGE problem with the study design because it means that the cases who were contacted and given the opportunity to participate do not represent a random sample. That fact alone is enough to invalidate the study results.

SHILL: Ms. Piper-Terry also complains about some children not meeting the requirements for the study. The authors are quite clear as to why those children were removed. Eligibility required that a child lived with it's mother since birth, the family spoke fluent English, and that the child did not already have a condition to which autism has been causally linked (in order to reduce the signal to noise ratio and provide a clearly observable signal).

MPT RESPONSE: Actually, the authors are quite clear about the fact that many of the issues raised in the above paragraph were considered before the children were classified as cases. I have no problem with the exclusion criteria listed here. Similarly, I was quite clear in my objections about children who were excluded, either by virtue of low-cognitive functioning or by the denial of access to them at all, by virtue of their physicians' denial of consent to even contact the families. The exclusion of children by either of these means is problematic because in doing so integrity of the study design is compromised, resulting in a lack of validity regarding any results reported.

SHILL: And last, thank goodness

Her final criticism:

"When comparing groups of 49 and 652, there is just a tad of inequality in the number of subjects."

Perhaps Ms. Piper-Terry should stick to being a biomedical consultant since her knowledge of statistical analysis is quite poor. One does not need the number of subjects in two groups to be equal in order to define a statstical difference in the analysis.

MPT RESPONSE: It is true that one does not need the number of subjects in two groups to be equal. However, the closer the groups are to being equal in number, the more power there is in the analysis. Inequality of subject groups influences which statistical analysis can be performed, for example when considering whether to use parametric or non-parametric analysis for comparing differences between groups. This is not a small issue and goes directly to the question of whether or not your results will reveal significance. I'm sure the researchers know this, even if "Shill" does not.

SHILL: Overall, Ms. Piper-Terry's arguments show some good quote mining, poor understanding of statistics, and overall lack of understanding in complex study design. I highly doubt she has read the 400 pages of technical reports detailing the intricacies of the study yet feels confident to comment on all of the details involved. Somehow I think the general acceptance of this study will not be much affected by Ms. Piper-Terry's rather poor commentary.

There you go Time4TruthNow. I guess one of the shills wasn't quite stumped, now was I?

MPT RESPONSE: I don't know this person (who self-identifies as a "shill") and I do not know the extent of his or her education, qualifications, experience, or associations. I am not going to comment on his or her personal strengths or weaknesses. The fact that he or she felt it necessary to attack me personally rather than confining his/her comments to the study design is another tactic often employed as a method of distracting readers from the issue at hand. It's called Ad Hominem and it is the practice of attempting to discredit the message by attacking the messenger. Watch out for this. It's a tactic frequently employed by those who know their arguments cannot withstand the scrutiny of close inspection by unbiased observers.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.