Orac at Respectful Insolence (one of the ScienceBlogs) does a thorough but restrained trashing of the Geier and Geier paper and the journal it appears in. Along the way, he answers my question about cohort analysis: in the California data, the rate of diagnosed autism among 3-5-year-olds has continued to rise since thimerosal was taken out of vaccines.
My original interest in the topic came from outrage at the legislative process being used to keep thimerosal cases out of court. But I was also willing to entertain the hypothesis that mercury in vaccines might have had some real effects, simply because the explosion in the rate of autism diagnosis seemed to demand some explanation.
However, a reader who knows the literature tells me otherwise:
Many (perhaps most) epidemiologists think the explosion of autism diagnoses was entirely attributable to changes in diagnostic practices. The increase coincided with significant liberalizing changes in the DSM criteria; there was a massive rise in awareness of and attention to autism; in 1991, the Dept of Education established a new category of special ed services for autism. Dept of Ed stats show that the number of children getting services for mental retardation then fell at the same time services for autism swelled, suggesting “diagnostic substitution.”
Looking at far-fetched explanations when there are no others available is a reasonable thing to do, but only if there’s something — an epidemic of autism, for example — to explain. If there’s nothing to explain, then the sensible thing to do is go back to sleep.
I think we can now officially declare the Geiers’ paper a dead horse, along with the larger claim about thimerosal and autism. It looks to me as if there’s just no there there.