Thimerosal: Update

On Sunday I promised two posts on the thimerosal/autism issue. Here is the science post.

Don't worry, I haven't reneged on my promise to write about the alleged conspiracy between the CDC and the major vaccine manufacturers. It's just that I'm still wading through the 386-page Simpsonwood transcript.

In June 2000, the CDC invited a large number of experts from government, academia, and industry Simpsonwood to solicit their opinion on thimoseral in infant vaccines. Their comments would be used by the Advisory Committee on Immunization Practices (ACIP), the policy arm of the National Immunization Program (NIP).

Apparently, the meeting was convened on very short notice. A team of CDC researchers headed by Thomas Verstraeten had found some interesting preliminary results in their retrospective database analysis using data from the CDC's Vaccine Safety Datalink (VSD).

Here are some key documents to keep you occupied while I finish reading the report:

Verstraeten T, Davis RL, DeStefano F, et al. Safety of Thimerosal-Containing Vaccines: A Two-Phased Study of Computerized Health Maintenance Organization Databases. Pediatrics. 2003 Nov;112(5):1039-48.

Verstraeten T. Letter. 2004;113:932. Letter to the editor defending his methodology and his ties to GlaxoSmithKline.

Robert F. Kennedy's article, Deadly Immunity, originally published in Salon on June 16, but now available through Common Dreams with no ads.

Comments

[suzib]

Lindsay,
The Deadly Immunity link doesn't work for me.
S.

[Lindsay Beyerstein]

It's working for me. Try again.

[Njorl]

Verstraeten's paper doesn't exactly clear thimerisol. A 95% confidence that the relative risk with increasing thimerisol exposure at 1-month is between 0.78 and 1.71 isn't saying much one way or the other.

[Lindsay Beyerstein]

Exactly. Phase I of the Verstraeten trial found statistically significant effects for tics, speech delays, and ADD. The conspiracy mongers are alleging that the CDC and Big Pharma got together in the back woods of Georgia to "huddle" about what to do next. The implication is that the Simpsonwood retreat was a chance for the big players to pressure Verstraeten into changing his study design so as to negate the Phase I results.

So far, I haven't found any evidence of that, but then again, I'm only halfway through the transcripts.

[Bill Gardner]

I have to applaud your willingness to read those transcripts!

[BumperStickerist]

some initial, related blogging on the topic:

http://skeptico.blogs.com/skeptico/2005/06/robert_f_kenned.html#comments

[Orac]

Njorl, yes it does. A relative risk whose 95% confidence interval encompasses 1.0 is considered a negative study, particularly when it encompasses 1.0 by a large margin (not being something like 0.95 to 1.9, or something like that, which could imply a small effect that didn't quite reach statistical significance). Statistically, that means there's only a 5% chance that there was an effect not detected by the study.

[Hedley Lamarr]

If what Kennedy contends is true, this is a big deal.

Look for more bills in the Senate to get the drug companies off the hook.

[Dwight Meredith]

I cant recal if it is in the Smallwood minutes but Figure 5 of the Phase I Verstaeten study shows that kids who were exposed to more than 62.5 micrograms of mercury by age 3 months had relative risk ratio of 2.48 at a CI of 95%.

That is, the kids exposed to more than 62.5 microgramns of mercury by age 3 months had two and a half times the chance of being autistic than those kids exposed to less than 37.5 micrograms. It is my inderatnding that by the final edition of the study, that number was not reported.

IIRC, a relative risk of 2.48 for that sample size, (an n of about 15 if my memory serves) was on the edge of statistical significance but not over it.

It has been a while since i looked up those numbers so please forgive me if they are not exactly corect. I have a clear memory that it was Figure 5 of the study and a clear memory of the 2.48 relative risk number.

I also have a clear memory of my blood pressure rising when I first read the study.

[dwight meredith]

I meant Simpsonwood, of course. I am traveling, do not have my files, and it is late.

[Njorl]

"Njorl, yes it does. "

No Orac, it doesn't. The confidence interval is too broad to be useful. 0.78-1.71 means there is significant likelihood that there is no effect of the chemical or that there is significant effect of the chemical. Thimerisol being innocuous, or causing 1/3 of the cases of unexplained autism would both be reasonable expectations given this study.

No one ever claimed that thimerisol caused every case of unexplained autism. If we were looking for a 1-to-1 relationship, as was practically the case with thalidamide, then yes, this data would be exculpatory, but we're not. Everyone (everyone rational) knew going in that a significant amount of autism was not caused by thimerisol. The convolution of uncertainties makes the problem much more difficult to solve.

[Neo Logic]

Autism has increased for only two reasons IMO: 1) much more standardized measures and reporting and 2) HFA (high function) people with Autisim (i.e. light Aspergers) now possses a disorder with a high $$ value. When they meet and breed, you get more autisim cases of all types because of the extreme genetic link for development of the condition. Simple as you would expect: Autism makes you a mini Bill Gates, and you breed more cause now you are a stud. Mercury shmercury. Its not good, of course, but its not the issue here. Thats why the cases increased in geographic areas likely to have higher earners.

[Orac]

Njorl:

Broad intervals of this type are not uncommon at all in the medical literature and do not necessarily imply anything other than that the study failed to find a statistically significant increase or decrease in relative risk at the 95% confidence level. That's all it means. Perhaps the study was underpowered, or perhaps there's a wide natural variation in the data even if the study was adequately powered, or perhaps both. Also, in most biomedical studies for environmental exposure, a relative risk less than 1.5 is considered quite small indeed, and you have to get to a relative risk of 2.0 or more before people really start to sit up and take notice, particularly for relatively infrequent diseases. You're reading far too much into what was a negative study. Even if this study's 95% confidence interval were small enough to exclude 1.0, you'd be looking at a relative risk of 1.25 (the mean of 0.78 and 1.71) at most.

Second, are you aware that mercury causes autism is exactly what many mercury-autism advocates are saying? Indeed, Generation Rescue states quite bluntly: "Generation Rescue believes that childhood neurological disorders such as autism, Asperger's, ADHD/ADD, speech delay, sensory integration disorder, and many other developmental delays are all misdiagnoses for mercury poisoning." Ditto Boyd Haley and the Geiers, although not quite as bluntly. This study is definitely not consistent with such a contention.

[Flint]

I realize that it isn't pertinent to your extensive evaluation of the risk factors and the data in the transcripts, but I wonder if it isn't a combination of factors involved in the autism spike in the population.

I wonder how many of the mothers of those children were drinking diet sodas and consuming large amounts of artificially sweetened foods. They have made a one to one direct relationship between their use and the acceleration of neurological disorders like ALS and Parkinson's disease.

[Thad]

"They"?

[Njorl]

Orac,

I probably did not make myself clear. I was only stating that thimerisol should be tossed into the bin with all of the other potential causes which have no hard evidence supporting them. Actually, it should be knocked down the list quite a ways, since this study showed that it was not responsible for a large number of autism cases, and possibly not responsible for any.

It may be that the increase in autism rates is mostly due to one cause, but it is not likely. Before the increase, autism was known to have many known causes and other unknown causes. In the last 50 years, there have been many changes in behaviors and exposures to hazards that could exacerbate many of these causes. Studies like this will not be capable of detecting modest correlations. If all we ever look for is the-one-cause-of-autism we will almost certainly fail.

So, yes, roll this study up and hit the Geiers on the nose with it. They deserve it. Thimerisol is obviously not the cause of an epidemic of autism. But don't dismiss thimerisol as a possible contributor to the problem based on this study.

[Flint]

"They"

I stand corrected on this one, at least partially. I came across a reference to it while reseaching a drug called Namenda, a new drug for the treatment of Alzheimers.

An article from Harvard made reference to its actions and discussed excitotoxins that cause neurons in the brain to dymylinate and glutamine was the key culprit and it mentioned aspartame in passing. Namenda supresses glutamine in some fashion.

A few google searches on aspartame came up with torrents of research and stuff dating to about 1996-97 and the FDA hearings. Later that was disproven or at least researchers failed to demonstrate their claims sufficiently to be considered proof.

There have been recent studies though that it does play a role in the neuron damage, but it is hotly contested, so the jury is out on what role if any it may have in the degradation process.

[Flint]

Oops... "demylinate"

[louisa silva]

There are two graphs posted in Rolling Stone that don't show up on the Salon.Com article: More Vaccines, More Autism plots the total micrograms of mercury received in the first two years of vaccinations by year of birth, against the national prevalence rate of autism. It is quite damning. Exactly where did the data come from? I would like to review the primary sources.
I would also like to review the primary source of the second graph: percent cumulative growth in cases of autism vs all other disabilities against year surveyed. What is the source of that data?