Non-stick pans, water-proof clothing, stain-proof carpet, pizza boxes, microwave popcorn bags, the air outdoors, the air indoors– if youve ever been exposed to these things, you have been exposed to ‘Perfluorinated Compounds’ (PFCs).

Because the most common PFCs have a relatively long half-life (~4 years!), and because weve basically all been exposed, you can find PFCs in the blood of about 90% of us here in the US. Its ubiquitous. And, like with a lot of things, ‘some’ isnt that big of a deal, but ‘more’ can cause problems. The questions are where that ‘some’ to ‘more’ threshold is, and what kinds of problems can emerge.

A rather unexpected correlation emerged today in JAMA:

Serum Vaccine Antibody Concentrations in Children Exposed to Perfluorinated Compounds

Previous research indicated that PFCs had a negative impact on the immune systems of mice. How can we find out if they have a negative impact on humans? We cant inject humans with PFCs and cut out their spleens to study their white blood cells. But mice arent humans– maybe what they saw was an artifact of ‘mice’ and PFCs have no impact on humans.

‘Maybe’ isnt good enough. So, this group of scientists went to a place where the folks eat a lot of fish (a common source of PFCs, so, if youve eaten fish, welcome to the 90%). Then they did two things:
1– Looked at the immune response of kids (5 years old) to tetanus and diphtheria vaccines, and correlated it with the blood PFC concentration of their moms in pregnancy.
2– Looked at the memory antibody response (aka how much the kids were protected from the pathogen) two years later (7 years old), and correlated it with their blood PFC concentration.

The higher the PFC concentration, the worse the kids reacted to the vaccines.

Enter media circus, stage left. And some of the circus was fuels by the lead author of this study himself, Dr. Philippe Grandjean.

“When the PFC concentration increases in the body, the immune system gets more sluggish and is less capable of maintaining a defense mechanism against microorganism”

You dont have data to support that statement in your paper. And wtf is ‘immune system gets more sluggish’? That is a meaningless phrase. Talk like a grown up, people are not stupid.

Although the findings don’t prove the chemicals themselves are harming the immune system, Grandjean said he thought that is “very likely” to be the case.

And Judy Mikovits thought XMRV was ‘very likely’ to cause CFS. We dont say things to the public without evidence to back it up, Grandjean.

“I don’t feel comfortable with the compounds for myself and my family and would rather eliminate them,” he told Reuters Health.


He added that parents might want to avoid microwave popcorn, and treatment of furniture, carpets and clothing with stain repellants to reduce their family’s exposure to PFCs.


Say it with me everyone: Correlation does not equal causation.

What Grandjean has is a very interesting observation. The biochemical and physiological ins-and-outs of that observation are still unknown. He did not do studies with microorganisms– he did studies with vaccines. Maybe there is some genetic quirk that has a negative effect on processing PFCs that also disturbs immune function. Maybe exposure to PFCs interferes with vaccine adjuvents (it seems PFCs and alum are in the same size range). Maybe there is something else about eating more fish that is negatively effecting vaccine responses (MERCURY!!!!!WARBLEGARBLE!!!), and PFCs are just a proxy measure that have no direct effect on the phenotype. An innocent correlation without causation, while the real culprit goes unrecognized.

Absolutely, if scientists can turn this observation into a causation, we would then know to keep an eye on PFC exposure and PFC waste in our environment. Of course. But screaming ‘WOLF!!!!’ when you find a dead goat that may or may not have died from a wolf attack helps no one, and can in fact be a snipe we waste time and money chasing if it turns out the wolf had nothing to do with it. I know youre excited, but be more careful with your statements, Grandjean.


  1. #1 Justicar
    January 25, 2012

    I suspect XMRV is the culprit for suppressed immune responses in lab mice. I’ll call this my special theory of XMRV suppression of immune response in laboratory mice. I’ll go cook up some data to prove it in a few minutes – then I’ll work on my general theory, which will incorporate the known threat to immune response that comes from hoodlum slug; hence, the sluggish component.

    Shit, I just read that you mentioned XMRV. *grumbles*

    I’ll forgive you for introducing me to a funny youtube channel . . . this time.

  2. #2 Prometheus
    January 25, 2012

    Ha ha ha. Too late. I gave my lazy cheap self obsessed sister good wine and a cashmere scarf for Christmas.

    She gave me a little stack of crappy plastic bowls from Ikea and when I gave her my patented WTF stare she announced with great pomp, “They are bpa and pfc free!”

    They are still under the back seat of my car, if I don’t get around to putting them in the dog’s big box of chew toys I’ll give em back to her next Christmas and buy myself some wine and cashmere.

  3. #3 DocOsc
    January 25, 2012

    In the Gee-Whiz catagory, these same compounds have been shown to affect development (in mice, anyway) when exposed prenatally. Nonstick chemicals upset behavior
    While not a claim of the study authors, the description of these pups sounds a lot like some of my autistic patients.

  4. #4 Ben
    January 26, 2012

    ERV, I’m a little curious about how much we know about the effects in mice. If you can experimentally establish genuine causation in mice, but don’t have that luxury in humans, although you can demonstrate that the correlation exists, doesn’t that look rather good for the causation hypothesis in humans? Sure it isn’t watertight, but it is *much* more likely to be causal in humans if it is causal in mice.

  5. #5 Jack
    January 26, 2012

    Microwave popcorn – no chance I’m cutting that out me diet ergo this research is crap 🙂

    Then again perhaps microwave popcorn addiction has caused my CFS? Nah. Am still not cutting it out of me diet 🙂

    Would probably be willing to spend a fortune for some nasty ‘medicine’ though – ya think he’s gonna sell me a test kit and some cure-aid?

    Maybe y’all could let me know? Thanks 🙂

    ‘Science Journalism’ Jees!

  6. #6 Ian Musgrave
    January 26, 2012

    I made similar observations myself (shamless plug)

    Also, the claim that there was double-to quadruple the risk of having antibody concentrations below 0.1 IU/ml for a doubling of PFC concentration didn’t seem to be bourne out by either the raw data or the modelling data

  7. #7 Ian Musgrave
    January 26, 2012

    Ben@4 The animal results are contradictory (some find no effect, some find an effect at concentrations modestly higher than the human study found). This could be strain differences or differences in administration. The best that can be said is that the animal studies are suggestive.

  8. #8 Ben
    January 27, 2012

    Thanks Ian. That sounds pretty weak. If the animal results for causation were conclusive and predicted exactly the correlation we see in humans, then I’d probably be a believer. As is, I guess I’ll wait.

  9. #9 harold
    January 29, 2012

    “Correlation does not equal causation”

    I agree that the author of this (actually rather decent) study publicly exaggerated the conclusions. So basically, I don’t have any technical disagreement with the post here. I do have a point about language.

    “Correlation does not equal causation” is, while technically true, a cliche so frequently used by weasels that it should be retired from the language. I recommend saying something more precise, like “correlation alone does not directly demonstrate causation, but is often a trigger for further study or a valid piece of a body of supporting evidence”.

    “Correlation does not equal causation” is frequently used by some weasel to dismiss vast amounts of convincing evidence. The most famous instance of its use was by the otherwise great statistician R. A. Fischer when he argued against a causal relationship between smoking and lung cancer. “Fisher was opposed to the conclusions of Richard Doll and A.B. Hill that smoking caused lung cancer. He compared the correlations in their papers to a correlation between the import of apples and the rise of divorce in order to show that correlation does not imply causation.[28]” From

    Fischer was being a hypocrite in this case, because, for one thing, there is intuitively valid reason to suspect that repeated exposure to cigarette smoke might damage lung epithelial cells in a way predisposing to cancer. There is no similar intuitively credible mechanistic hypothesis relating apple imports to divorce rates.

    A hypothesis of causation predicts correlation. Lack of any correlation (in the broad sense, or even in the sense of a linear relationship for many things) usually rules out causation to any reasonable degree. Starting with a hypothesis and seeking supporting correlation data is a valid way to support a hypothesis. Obviously, work that demonstrates a causative mechanism directly is even better, but may not always be immediately possible.

    Furthermore, when there are two competing hypotheses, and one is supported by correlation data, and the other by nothing at all, the former is generally stronger, even if neither can be ruled in or out. The onus is on supporters of the latter to get some data, rather than merely dismissing the former with “correlation does not equal causation” and assuming victory by default. Creationists and other science denialists sometimes use this tack – dismissing scientific evidence as “correlation” (which in some cases it is), then dismissing “correlation” as meaningless, and then assuming victory by default despite no evidence of any sort for their own hypothesis.

    Correlation should be viewed skeptically, and nonsensically meaningless correlation relationships are easy to find or create.

    However, from a Bayesian perspective, correlated things are far more likely to have an important relationship (not necessarily one of direct causality, but identifying a third variable that drives them both can sometimes be of great interest, too) than non-correlated things.

    Correlation does not equal causation, but it is often an important and valuable thing to note.

  10. #10 Moody
    April 7, 2012

    Excellent comments Harold. You understand science and what it takes to be good at it!

New comments have been disabled.