Friday, January 16, 2009

Why would I question medical statistics related to autism?

I am Ed and the father of David who is autistic. I am an engineer by trade and as an engineer I have had a number of occasions to work with statistics. Many times I have heard medical research expound on statistical measurements and what they mean. While I seldom dig into the methods used to collect the statistics, I have looked at the inferences that were drawn and if there is a weakness in medical research, it is in the conclusions that the researchers draw from their data.

The first statistic that the medical community swore to was that the number of autistics in the general population is 1/2500. My son attended an elementary school with approximately 250 children in it. The odds of the school having no autistic children is about 90.48%. That would make my son part of the 9.52% of schools that had one or more autistic children at the time. No big deal. Then a second autistic child came to the school. That would make the school part of the 0.47% of schools that size that have 2 or more autistic children. Then a third child joined my son's school. That made my son's school one of the 0.015% of schools that size with 3 or more autistic children at the time. Had my son's school been a statistical experiment, the inference would have been virtually conclusive. The 1/2500 statistic had to be wrong. Add to that the autistic girl just up the road whose parents did not send her to the local school and the 1/2500 statistic became absurd. Yet all the way through February 2007 the CDC maintained that the autism rate was 1/2500.

This next paragraph is not a comment on whether or not thimerosal is related to the autism rate. It is a comment on the inferences that the medical community makes with their statistics.

When the Danish study was done the medical community waved it around. The autism rate in the study went up as thimerosal was taken out of the Danish vaccines. The conclusion was that thimerosal and the autism rate are not related. That is not the way it works. The measurement being done was to determine whether there is a relationship between autism and thimerosal. Logically you can infer one of two things. Either a little thimerosal prevents autism or the experiment got contaminated by another variable. The second one was true which made the study invalid. The study had been through peer review. The study was trumpeted by the medical experts as proof that autism and thimerosal were not related. The study was not valid. Invalid means that after the study was done, we do not know whether there is a relationship. The medical community made a mistake in logic to conclude that there is no relationship based on the Danish study. Invalid does not say that there is a relationship.

If you want a more modern one just take a look at the Syracuse University study where thimerosal levels were taken after vaccine injection. The thimerosal disappeared from the blood stream after three days. The conclusion made by the study was that the body got rid of the thimerosal and therefore thimerosal is safe as a preservative. What an inference for a potent neurotoxin. What an inference for a cumulative poison. What a leap of faith. The right conclusion is that the blood levels dropped after three days. There is nothing that can be concluded about what happened to it. But the Syracuse University study supports the idea that autism and vaccines are not related. Wrong or not it passed peer review and got plenty of press,

The proper experiment to determine if there is a relationship between autism and vaccines involves comparison of the vaccinated population against the unvaccinated population. I have been handed MMR studies comparing those who had and those who had not received the MMR vaccine and told that the MMR studies are proof that there is no relationship between vaccines and autism. It is an inference that is not valid.

When I suggest that the study needed is one that compares the vaccinated population against the unvaccinated population, the reactions are not favorable. I do not understand this. When I look for the autism rate of the vaccinated population I know that since 98% of the population is vaccinated the autism rate of the general population is about that of the vaccinated population, that is, 1/160. This works as long as the autism rate for the vaccinated is equal to the autism rate of the unvaccinated. Since this is what the medical community has been advertising, I can't see them objecting to that assumption. I only have one known and valid measurement for the unvaccinated population. That is the time before vaccines when the autism rate is 1/2500. From that I have to make one of two conclusions. Either vaccines made the autism rate rise or there is another factor that occurred concurrently. If I were in the medical community, I would be rushing to measure the autism rate among the unvaccinated population. If I am in the medical community then I know that there is no relationship between autism and vaccines and I would want to conclusively prove it. An autism rate of 1/160 among the unvaccinated population would conclusively show that there is no relationship between autism and vaccines.

Why would I question medical statistics related to autism? The medical community is willing to accept any study that says autism and vaccines are not related whether the study is valid or not. The medical community will not do the experiment that is a valid measurement to determine if there is a relationship.

No comments:

Post a Comment