Also,
c) This small group of mercury-sensitive children never show the usual symptoms of mercury poisoning.

Dr. Manhattan:

The problem with your criticism is as follows: the hypothesis that some children have a genetic sensitivity to mercury such that mercury exposure might cause autism in those children is just that -- a hypothesis, one that explains why a small subset children in a large population might be affected in an unusual way by mercury exposure. Assuming arguendo that it's true, we don't have any way of testing for it at present, and there is no reason that a study collecting data on thimerosal exposure and autism would allow us to identify that group.

(Consider the BRCA genetic mutations which predispose women who carry them to braest cancer. Lots and lots of epidemiological studies on braest cancer were performed before those mutations were identified -- the carrier subgroup doesn't just jump out of the data.)

markm:

There's no reason a mercury-sensitive person would exhibit classic symptoms of mercury poisoning. Like most people, if I eat too many peanuts, I get heartburn. That doesn't mean someone doesn't have a peanut allergy if the symptoms of his allergic reaction don't include heartburn.

(This is not to say that I buy this hypothesis; just that I don't think these are valid criticisms.)

(Also, there is a word deliberately misspelled above to avoid the comment filter.)

The BRCA comment is true, but only serves to reinforce Dr. Manhattan's point. BRCA carriers are a very small proportion of patients with braest cancer. Only around 2% of braest cancer patients carry them, maybe 5-6% in young patients. They are very important in that small subset, but they are generally not that significant in the overall population. If a similar proportion of the population had some sort of mercury "sensitivity" that led to autism, it would not account for the supposed "epidemic."

(Words intentionally mispelled to avoid comment filter. BTW, as much as I like your post, your comment filter sucks because it blocks a post based solely on that word.)

There may be no reason that a study collecting data on thimerosal exposure and autism would allow us to identify a particular, highly sensitive group, but I don't think that's the point. Holding the size of the sensitive group constant, increased exposures to mercury through thimerosal should be strongly positively correlated with increased levels of autism, if the thimerosal is causing autism for that group.

I thought that the point of the post was this: If the increased levels of autism show up in the data, and if those levels are driven by increased exposure to mercury through vaccinations, then shouldn't the correlation between thimerosal and autism also show up in the data?

Orac: There's no reason to expect that the percentage of the population with the BRAC mutation is comparable to the (hypothetical) percentage of the population that is mercury-sensitive.

Ann: Suppose 99.5% of the population can tolerate the levels of mercury involved in a full course of childhood immunizations with mercury-preserved vaccines, and 0.5% can only tolerate 1/3 of the course with mercury-preserved vaccines. Then you would have many children with no autism and a lot of mercury exposure and a few children with autism and not very much mercury exposure. Any correlation would be very weak, even though the mercury -- together with the mercury-sensitive genetic predisposition -- is actually causing the autism.

(Again, I'm not saying I buy this hypothesis. I'm just explaining why a big, well-performed epidemiological study might not identify this particular mechanism, if it actually existed.)

alkali,

Your example proves the point. The jump is autism has been enormous - much larger than such a small group could account for.

Indeed, alkali seems to have shot down his own argument. A small subset of people with a "sensitivity" to mercury could not account for the "epidemic" that mercury-autism advocates claim.