In the end, it doesn’t matter, because either way you call it, the thimerosal-causes-autism hypothesis is dead.

[1] If the rise in autism prevalence is real, then thimerosal cannot be causing it because thimerosal was removed from vaccines over eight years ago.

[2] If the rise in autism prevalence isn’t real, then thimerosal cannot be the cause of the rise in autism prevalence because the rise doesn’t exist

Is that simple enough for you, Harold?

Prometheus

No, the increase is not necessarily a real increase. What that obviously refers to are increases recorded by passive systems, such as CalDDS or the Danish Psychiatric Registry.

So far, there’s no active surveilance system of autism with perfect case-finding and equivalent criteria year after year.

Now, let’s consider Harold’s proposition: If the increase is not real, then no inference can be drawn with respect to thimerosal removal…

This is logically mistaken. If the increase is not real, then the thimerosal hypothesis is a no-go to begin with. It doesn’t even need to be considered.

The thimerosal hypothesis basically says that at least part of the increase is real, and caused by thimerosal. In order to evaluate the hypothesis, we assume part of the increase is real, and then confirm predictions of the hypothesis.

It’s a fact that proponents of the hypothesis predicted a drop in adminstrative counts, but let’s leave this aside.

There’s probably been something like a 7-fold rise in administrative counts in the last couple decades. To be conservative, let’s say thimerosal is responsible for only a 2-fold rise.

Now, between 1999 and 2002, the thimeorsal dose dropped quickly and well below any levels seen in decades. Therefore, administrative counts should’ve dropped by a factor of 2 between 1999+N and 2002+N, based on our assumption. To offset that drop, there would need to be a 2-fold rise for other reasons between those two years. That would only level it off.

It would be very noticeable once you compare the trend, say, 1996 to 1999 vs. 1999 to 2002 vs. 2002 to 2005.

In statistical terms, it’s just wrong to say that no inference can be drawn. It’s simply a multivariate analysis where thimerosal is one variable and the other variable might be nothing more than ‘year’.

Does that mean that:

(1) the increase in autism diagnoses resulted from REAL increases in autism or:

(2) that if the increase in autism was NOT real, was attributable entirely to diagnostic change, social awareness etc. then no inference should be drawn with respect to thimerosal based on increased autism diagnoses after its removal from vaccines?

As I often do for stories of this type, I’m keeping a running list or index of pro- and con- blog posts. This one’s on the list.

The list is here