Wednesday, August 8, 2007

Vaccine, autism and statistical nonsense

Okay, it had to happen some day, and it happened yesterday: I came across a new study on thimerosal and autism. My first intention was to leave it aside because others, like Orac, have followed the controversy for a long time and have posted book chapter size posts about the subject: Read more at Respectful Insolence on Autism.

Just in brief for all to whom this is new: Thimerosal (thiomersal), a mercury containing preservative agent in vaccines until 2001, has been blamed to increase the risk of autism in children but the studies cited by the anti-vaccine activists do not really support such a view: Read more about the thiomersal controversy.

Now to the study I have found. Fifty-three patients with autism spectrum disorder (a broader definition of the disease, leading to more cases than autism in strict sense) have been examined. 28 percent of their mothers are rhesus-negative and have received a vaccine that prevents a serious, life-threatening illness of the baby. In a general population not affected by autism, the amount of rhesus-negative mothers is only 14 percent. In other words: Compared to non-autists, patients with autism spectrum are twice as likely to have a rhesus-negative mother and therefore have twice as likely been exposed to the thiomersal containing vaccine. The study authors conclude:

«The results provide insights into the potential role prenatal mercury exposure may play in some children with ASDs.»
Where the statistical nonsense begins

Nothing can be said against this statistics as such, and the conclusion is at least worth to be discussed. But there are two main sources of nonsense that may be drawn from this result. The first one is to conclude that it is necessarily the mercury and to jump from correlation to cause. The second one is to focus only on the double relative risk of autism without taking into account the absolute risks of (a) autism and (b) life-threatening disease of the newborn.

It can be the mercury, but ...

A fetus of a rhesus-negative woman may be exposed to
  1. Mercury from the vaccine.
  2. Immunoglobulines that are the main content of the vaccine, preserved by the mercury-containing agent.
  3. Substances produced by the pregnant woman as a reaction to the vaccine.
  4. If the fetus is rhesus-positive, the effect of the vaccine may not suppress the maternal antibody reaction by a hundred percent. The fetus will be exposed to some of these antibodies.
Absolute risks must be compared

As said before we should not jump to the conclusion that mercury doubles the risk of autism. But I need this hypothesis for my next calculation:
  1. In the United States, about one in 250 persons suffers from autism, that is 0.4 percent of the population. (Source: Fighting Autism). I have not found any figures about the spectrum, so let me be generous and assume that one percent of the population is affected. Let me also assume that mercury doubles the risk, thus causing one additional case in every hundred babies.
  2. Rhesus factor D (the dangerous part of it) is dominant, therefore a baby from a rhesus-negative mother will most likely be positive. I am not a genetics specialist so let me again be generous and assume that only fifty of every hundred babies will be positive and therefore threatened by the antibody reaction of the mother.
Now let's do some math. If we have vaccinated a hundred pregnant women, we may have caused one case of autism spectrum disorder. If we did not vaccinate, we may have caused fifty babies suffering from a life-threatening disease. The ratio of benefit to harm is fifty to one. And this is a very conservative estimate. Any questions left?

All other vaccines are different

After having settled the case of rhesus vaccine, what about MMR and all the other vaccines? Most of them never contained thiomersal. And against the tiny rest, this rhesus study does not add a scrap of new evidence. Just because it is a major difference between an exposed fetus and an exposed child. While I can imagine mercury doing some harm to the developing brain of a fetus in certain sensible stages, this is very unlikely to happen to the fully functional brain of a child.

The source of statistical nonsense

I do not blame the study authors. Their statement is correct. But such a study is very likely to be wrongly cited by anti-vaccine activists. Just wait until the media take up the story. I already see a science editor sit at his desk, creating a catchy headline such as «vaccine doubles autism risk, study shows». But if you, dear reader, happen to be a lawyer of parents of autistic children in a class action suit, forget it. This is not a corpus delicti.

Photo credit:

Judy of Tiggers don't Jump tells me I have missed an important point: Since 2001 there is no thiomersal at all in the vaccine (Rhogam). Thanks Judy, I forgot to mention that all subjects of the study have been born before 2001 and that all mothers have been vaccinated with the old product.

No comments: