Madsen, KM, A Hviid, M Vestergaard, D Schendel, J Wohlfahrt, P Thorsen, J Olsen and M Melbye 2002. A population-based study of measles, mumps and rubella vaccination and autism. New England Journal of Medicine 347:1477-82.

In the largest study yet conducted examining possible links between autism and the vaccination for measles, mumps and rubella (MMR), Madsen et al. find no hint of an association. A theory linking MMR with autism had grown out of observations that children without hints of autism early in their childhood then developed autism after vaccination.

This paper does not address the question of whether or not vaccines containing a mercury-based preservative, thimerosal, may cause autism, as the MMR vaccine as formulated by Merck does not contain thimerosal (a list of vaccines containing thimerosal). According to the authors of this study, the form of the MMR vaccine used in Denmark is identical to that used in the US. Mercury is a powerful neurotoxicant which alters patterns of brain development. Whether or not it causes autism is the focus of a huge debate.

For an in-depth review of the strengths and weaknesses of this study, see comments at Safe Minds.

What did they do? Madsen et al. examined the health records of all children born in Denmark from 1991 through 1998: 537,303 children of whom 440,655 (82.0 percent) had received the MMR vaccine and 96,648 had not. Records also allowed them to identify children diagnosed with autism and related disorders. Of those diagnosed, they carried out detailed assessments of the records of 13%, and found it necessary to reclassify only 8% of the 13% (or roughly 1% of the total). The diagnosis used in this detailed assessment was consistent with that employed by the US CDC in its survey of Brick Township, New Jersey.

The MMR vaccine was first used in Denmark in 1987. The Danish version is identical to that used in the United States.

What did they find? Among the children in the study, 316 children were diagnosised with autism and 422 with other autistic-spectrum disorders.

In the key comparison, Madsen et al. found no elevation of risk of autism in vaccinated children vs. unvaccinated. Adjusting for potential confounding variables, the relative risk for vaccinated children was 0.92 (with 95% confidence intervals extending from 0.68 to 1.24). They found a similar result for other autistic-spectrum disorders (RR = 0.83; 95% CI from 0.65 - 1.07).

[An RR of 1 would have indicated exactly equivalent risk. An RR less than 1 suggests that there is less likelihood of autism in vaccinated kids, but it only becomes interesting when the range of risk values bracketed by the 95% confidence intervals no longer include 1. As long as the range of RR values bracketed by the lower and higher 95% CI includes 1, then the difference between the calculated RR and 1 is likely to be due to chance.]

The Danish data also allowed Madsen et al. to look at the timing of autism's appearance in relation to when a child was vaccinated. They found no particular pattern, other than the fact that the average age at which autism was diagnosed (4 years, 3 months) was later than the average age at which vaccination occurred (17 months).

What does it mean? Madsen et al.'s findings offer strong evidence against the role of MMR vaccinations in the causation of autism. Some critiques of the study, however, suggest alternative explanations. Unfortunately, US public health records do not allow comparable research in the US, where use of thimerosal in vaccines is widespread.