Advice Goddess Blog
« Previous | Home | Next »

Conspiracy Theory-Based Medicine
On Slate, Arthur Allen debunks the woo of the anti-vaccination nutwads who believe thimerosal causes autism:

...Four perfectly good studies comparing large populations of kids have showed that thimerosal did not cause the increased reporting of autism. The best evidence comes from Denmark, which stopped putting thimerosal in vaccines in 1992; the rate of autism in kids born afterward continued to increase. A U.S. study showed slightly higher rates of tics in children who got more thimerosal at earlier ages, but no autism. An as yet unpublished study of about 1,000 children exposed to different levels of thimerosal in the United States also showed more tics in the children heavily exposed to thimerosal, but in addition showed they had fewer than average language delays. Both findings were marginally significant in statistical terms—though in unscientific terms, they are probably meaningless.

Parents who are convinced thimerosal damaged their babies attack the big epidemiological studies as a whitewash by vaccine makers. They're especially concerned about the U.S. study, which in its early drafts showed a link between thimerosal and neurodevelopmental problems—though not autism, despite Kennedy's claims to the contrary. He extols the studies by David and Mark Geier, a father-and-son team who work out of their basement in Silver Spring, Md. The Geiers have done a series of studies published in obscure journals that purport to show a link between autism and mercury, and they spend a lot of their time testifying on behalf of allegedly vaccine-injured kids. In the polite language of the Institute of Medicine report that dismissed the vaccine-autism link, the Geier studies are "uninterpretable." The main Geier approach is to mine data from a CDC reporting system that contains a mishmash of real and garbage vaccine-injury allegations, according to the vast majority of the scientists who work in this area. The Geiers have found a sixfold increase in autism in children who got thimerosal-containing vaccines. But nearly all the reports of autism they tallied came after allegations of the vaccine link had been publicized in the newspapers. In other words, the Geiers report the public's response to a scare as if it were meaningful data.

Probably the most damning epidemiological evidence against the vaccines-cause-autism theory, and another point that Kennedy gets wrong, is contained in the document that got critics started on their claim of a vaccine-provoked epidemic—a 1999 Department of Developmental Services report from California. Like reports from other states in the country, it shows a dramatic increase in autistic children seeking state services, from 2,778 autistics on the rolls in 1987 to 10,360 in 1998. An impressive diagram of this increase was projected on a screen at a Committee for Government Reform hearing chaired by Indiana Republican Dan Burton, who believes that vaccines gave his grandson autism. "Look at that graph," Burton said. "They are having an epidemic out there." But the graph actually vindicated vaccines. MMR vaccination began in children born in 1970, but there was no increase in autism reports in the state until 1980, which also happened to be the first year the psychiatric definition of autism spectrum disorders changed. A 2001 study showed that while MMR vaccination rates increased 14 percent from 1980 to 1994, autism intakes in California's state programs increased 373 percent. The increase also showed no apparent connection to the addition of thimerosal-containing vaccines to state pediatric immunization schedules.

A far more obvious explanation for the increase in autism rates in California was the one that mainstream autism experts expounded: diagnostic changes, new laws that expanded federal payments to care for autistics, and greater parental awareness of these resources. In 1990, Congress made autism one of the disabilities that qualified for federal funding. Thereafter, states were obliged to report all cases of autism. In a Minnesota study, to take one example, admissions of autistic children to developmental programs jumped starting in the 1991 school year and continued to do so for a decade. Often these increases occurred within the same grade. For example, 13 autism cases were reported per 10,000 Minnesota 6-year-olds in the 1995-96 school year—that is, among children born roughly in 1989. Five years later, the prevalence rate for this cohort was reported at 33 per 10,000. These were the same kids. Between the ages of 6 and 11, they'd suddenly "become" nearly three times as autistic—or rather, doctors, parents, and school counselors were enrolling them in programs more aggressively.

Most of the scientists who study autism trends are not ready to rule out entirely some real increase in the disease. But the causes may have nothing to do with industrial toxins like mercury. Interestingly, a 2003 California study found that mothers older than 35 were four times as likely to give birth to autistic children as mothers younger than 20. One of the only known environmental causes of autism is congenital rubella infection (or German measles); during a 1965-66 rubella epidemic in the United States, about 1,500 rubella babies were born with autism in addition to their other handicaps. Other perinatal developments, which increase with maternal age, can't be ruled out.

So much easier to blame chemicals instead of your way-past-the-sell-by-date eggs. Can we please base our policy on data instead of emotion?

Posted by aalkon at June 19, 2007 10:11 AM


Well, so far we haven't confirmed the maternal age hypothesis...whereas we have, IMHO, confirmed the follow-the-money hypothesis. (I am agreeing on the main point, natch.) I think it is somewhat comforting to parents to think that there is something out there that is discrete and identifiable that causes autism...something that they can work to stop. A murky interaction between genes and the environment is a lot trickier. Plus, if you look at the number of kids being diagnosed as "borderline/mildly retarded," you'll see that it has not been keeping pace with the population - in fact, it's been going down. Kids who would have been diagnosed as borderline retarded a generation or two ago are now diagnosed as autistic in at least some cases. I don't object to this in principle - I just object to it serving as the basis for a lot of anti-vaccine wackery, which does, contrary to some people's beliefs, have some very nasty consequences. (I'm thinking of the just-born grandson of a friend who came down with whooping cough a few years ago, for example. He survived, thank goodness, but...)

Posted by: marion at June 19, 2007 5:22 AM

The whole "mercury/autism" and "vaccines/autism" pseudoscience is a perfect storm of crankery. The problem is, the evidentiary standards for the Autism Omnibus are so low that it is actually possible that some of the plaintiffs could well prevail.

Posted by: Orac at June 19, 2007 6:59 AM

Autism has no relation to maternal age.

Posted by: deja pseu at June 19, 2007 7:06 AM

Thanks so much, Orac. He's done great posts on this, over at a blog that's a daily read for me, his "Respectful Insolence."

A search of autism on RI, brings up a slew of entries:

P.S. Orac, I ran into some guy in Santa Monica, e-mail address DuaneLaw, who said he'd posted on your blog (argumentatively, I'd imagine). When I met him, he proudly told me that he's doing work to (I'm paraphrasing from memory), basically, bring (gag!) alternative medicine into greater favor. I quoted Marcia Angell and Jerome Kassirer:

“There cannot be two kinds of medicine - conventional and alternative. There is only medicine that has been adequately tested and medicine that has not, medicine that works, and medicine that may or may not work.”

Hilariously, the guy was shocked that a person in Santa Monica would be in favor of such thinking. (They let us logical, rational types out across city lines these days, imagine that!)

Posted by: Amy Alkon at June 19, 2007 7:32 AM

I agree with Orac, for somewhat different reasons. If I were a vaccine manufacturer I'd be quite worried about these cases. Legal standards in civil cases (a preponderance of the evidence, or p > .5) are much lower than in criminal cases or scientific research, and juries can be quite easily confused by dense research and statistics. Given the widely held and irrational dislike of pharmaceutical companies, I'm betting they lose one of these test cases and the floodgates open. Hope not, cause it will only encourage more fools to 1) not vaccinate and 2) sue big medical companies on the thinnest of evidence.

Autism has no relation to maternal age.

Deja, I don't think you're right. A quick google search finds many studies showing a link between increased maternal (and also paternal) age and autism.

Posted by: justin case at June 19, 2007 8:03 AM

The Grandmother Hypothesis
Ten years ago, Kristen Hawkes, an anthropologist at the University of Utah, was examining the biological basis and evolutionary origin of menopause in humans. Her observations led to the so-called “grandmother hypothesis,” which suggests that grandmas reap an evolutionary boon from ceasing reproduction in their fifties and just helping to raise the grandkids. Hawkes’ idea rested on the fact that as women age, pregnancy-related health problems rise dramatically. For example, the incidence of Down syndrome rises from 1 baby in 1,490 when a mother is 24 to 1 in 160 when she is 40. At 50, the risk is 1 in 11. Older mothers are also at a greater risk for high blood pressure, gestational diabetes and difficult labors – first-time mothers 35 and older are almost seven times as likely to have C-sections or assisted deliveries.

I think this is one of those times, like the link between abortion and breast cancer, where people are going to demand a little extra epidemiology before they sign up with the program....

Posted by: Crid at June 19, 2007 10:30 AM

Perhaps the apparent link between maternal age and autism mirror the other findings. Perhaps older mothers are more likely to suspect/realize their child is autistic and push for the diagnosis. Much as the public and medical awareness of autism leads to more reported cases.

Posted by: Kimberly at June 19, 2007 10:31 AM

The graph after the link was me, all the rest is a cut&paste from the article

Posted by: Crid at June 19, 2007 10:31 AM

That's interesting, because our son's neurologist says latest research still hasn't pinpointed any definitive correlating factors for autism (including maternal age, vacinations, diet, etc.). Just anecdotally, most of the moms I personally know with Autistic/Asperger's kids had them in their 20's/30's. My son isn't autistic; he's developmentally disabled (mentally retarded for those who aren't sqeamish) but you get to know a lot of parents of autistic kids through Spec Ed, summer camps, etc.

Also just a note, there's nothing new about women having kids into their 40's. What's new is women waiting until later to have their *first* child. Autism is on the increase among children of mothers of all ages.

Posted by: deja pseu at June 19, 2007 10:33 AM

Until someone is willing to fund a study that looks at all factors inculding but not limited to:
medications given to both child and parents(durring pregnacy and breast feeding)
foods eaten by mother durring pregnancy and breast feeding
food given to children and possible effect of articfical preservitives
gene structuring of both biological parents(though this could pose a problem for some married women)
what possible toxins the parents might have been exposed to throught their lives
a detailed scan of the parents sperm and eggs in comparison to a resonble pre-determined "healthy" control group
an enviormental survey of the home and previous homes and workplaces of the parents to scan for airborne toxins
a check of outdated stanndrds to see what particular cases would have been classifed as a different disorder in the past.

And a nch of other things I'm sure doctors wold think of that I havent.

I'm supprised though that no one has thought about sueing the makers of prenatal vitiams as well as vaccine companies - you could makethe same half assed cliams

Posted by: lujlp at June 19, 2007 3:07 PM

In my blog, I reproduce a table with seven alternatives to evidence based medicine:

Posted by: Pedro Morgado at June 19, 2007 8:03 PM

Just a not-so-quick note about all this talk about links to vaccinations and autism:

WARNING: majorly boring statistics crap follows:

Statistics is based on a null hypothesis, which is a statistical inference of the data you got (i.e. something is either true of false). All this statistics crap falls into two steaming piles that describe how the data is interpreted (or misinterpreted): a type I error means you reject the null hypothesis when the data was right (I think the sky is orange when it is really blue) and type II error that means you accepted the null hypothesis when the data was wrong (I think the Iraq war was right and just when all the data is contrary). Statistics have great ways of insuring that you don’t make either type of error by all sorts of boring mathematical formulas. Let’s look at ways (with both types of errors) where vaccinations cannot be related to autism. A type I error basically states ‘if I ran this test 100 times, how many times out of those 100 would I have a positive value just because of plain ol’ luck?’. This is called a p value. Significance usually is cut off at a p value of 0.05 (5% chance of getting the same result by just luck). Flipping coin has a p value of 0.5 (50% chance of getting the same result); getting laid at a bar by using the line ‘how come a girl like you is in a place like this’ has a p value of 0.001 (1 out of a 1000 chance). You get the picture. All the studies on autism have fallen well below the p value of 0.05. The more powerful argument is the type II error that is intimately linked to the size of the population in the study: in short, the larger the population, the larger the power. Now, if you want to find something that is fairly common (i.e. 1 out of 100 or 1000 result) you don’t need much power (in other words, not as many study subjects, just enough to ferret out that 1 out of 100 or 1000; in the range of high 1000s or low 10,000s subjects). However, if you want to find something much more uncommon (1 out of 10,000 or 100,000) you need much more power (in the range of high 100,000s or many millions subjects). Autism has a prevalence of 2-5/10,000 (from mainstream medical epidemiology, not the crap you get online). Now, think of how many children have received vaccinations (quite a few, easily in the many tens of millions in the past few decades)! Taking all those children who received vaccinations and comparing those who have autism with those who did not receive vaccinations and who have autism you ultimately have a statistical stew that has a sh*tload of power. In other words, if there was any reason to think that autism was not randomly related to vaccination (a type I error) or that vaccination caused an increase in autism (a type II error) you are not actually looking at the data! However, data is like a Guantanamo bay detainee in that if you torture it enough it will say about anything; but when you have the huge, friggin’ vast majority of data against your tenuous, feeble accusations it is hard to say ‘don’t’ look at the facts because I am right’.

Posted by: Doc Jensen at June 19, 2007 8:48 PM

OMG Pedro, that is the BEST table for evidence-based medicine ever!

Posted by: Doc Jensen at June 19, 2007 8:50 PM

Bravo! I nearly snorted ice cream up my nose!

Posted by: Amy Alkon at June 19, 2007 10:47 PM

Alternative medicine ... my wife went for an allergy test the other day. I was expecting that they would expose her to a range of potential allergens and measure her response to them, but no. Instead, she had something that from her description of it sounds very like the scientology gadget - she held two metal handles connected to a computer (wow!) that did something. At the end of this she was presented with a huge list of allergies. (This was private, not NHS.)

Anyone know anything about these devices?

It also occurs to me that if this is a quack scheme, then it pays to find as many allergies as possible. First, it reduces the risk of missing a genuine allergy and getting sued. Second, it seems like value for money ("Gosh, I must be really special to have so many allergies!"). Third, it helps create a market for lots of specialist products. Oooh, I hate being skeptical and cynical.

Posted by: Norman at June 20, 2007 2:07 AM

Posted by: Amy Alkon at June 20, 2007 4:58 AM

Posted by: Amy Alkon at June 20, 2007 4:59 AM

Interesting - thanks!

Posted by: Norman at June 20, 2007 11:43 PM

Leave a comment