Autism and obesity, a confounder

The journal Pediatrics released a study recently claiming an association between autism and maternal obesity during pregnancy. As soon as I heard this study being splashed all over the media, I winced. As much as the journalists point out “correlation, not causation”, they also throw out vague warnings about obesity. On NPR this morning, the story was covered as being another casualty of the “rather striking epidemic of obesity”, and, in this article in the Washington Post, despite the lack of causation, the author warns “[s]ince more than one-third of U.S. women of child-bearing age are obese, the results are potentially worrisome and add yet another incentive for maintaining a normal weight, said researcher Paula Krakowiak, a study co-author and scientist at the University of California, Davis.”

Because, it’s all just our fault because of our behavior, right moms?

I don’t think so.

I was already pondering autism and causation recently. It has been an interest of mine for quite some time. Before I knew that I was going to go to medical school, when I got my first job at the first health food store, I was intrigued by the parents of children on the autism spectrum who would come into the store, desperately seeking anything to help. Many were trying gluten-free, casein-free diets, long before the recent gluten-free craze. Many were buying supplements. I was surprised at how many children on the autism spectrum there were. This was in the mid 90’s, and my first glimpse at the burgeoning numbers of children living with this diagnosis. I decided I wanted to work in the field, and help unravel this mystery for these parents.

Since then, my focus has obviously shifted. However, I have two cousin Susans with two sons each on the autism spectrum or with related developmental delays, which were also included along with autism in the study. One is closely related to my younger son’s father, and one is closely related to me. They are not related to each other at all. I know it is just anecdote, but I was already trying to look for a pattern – something they had in common. Why were their children affected, and mine not? I was overweight, possibly obese according to BMI, during both pregnancies. One of my cousins was, and one wasn’t.

Well, this study made things click in my head. There is a confounder strongly associated with obesity that was not looked at in the study. It is also associated with high circulating androgen levels, which have a known association with autism spectrum disorders. And, interestingly enough, both of my cousin Susans have this condition, and I don’t. It’s polycystic ovarian syndrome (PCOS). It is a lot less common than obesity, and would probably make more sense as a causation, both prevalence-wise, and physiology wise. I did a quick literature search, and couldn’t find anything on it.

I am not sure how difficult it would be to do a case-control study on this. It would be easy if I was at Kaiser. It was probably easier to look at weight and height at delivery than delve into gynecologic histories to find if there was any diagnosis of PCOS in the subjects of the study.

Well, trying to do some research on this is a definite possibility. Hopefully in the near future.


Filed under Uncategorized

11 responses to “Autism and obesity, a confounder

  1. irishup

    Longtime lurker here. So very glad to see you’ve tackled this! When I heard it on the radio driving in this morning I made sure to hunt around my favorite med blogs to see what was what. The UGH! factor with the media plan on this is really high. And so glad (where glad=eyes crossed with frustration and anger) to see we’re back to blaming Mommies for ASD! So very refreshing after all these years.

    “It was probably easier to look at weight and height at delivery than delve into gynecologic histories to find if there was any diagnosis of PCOS in the subjects of the study.”

    (Disclosure: clinical researcher here with 20yrs experience using EMR data for epidemiological / outcome studies). Well, IDK; if you’ve got that kind of health information at your disposal, plus birth outcomes and pediatric daignoses, how far a stretch would it have been to get ICD-9 codes? The methods in the abstract is sparse, so I’d have to see the whole to judge whether in fact such information *was* outside of the scope of this study. But I find it vanishingly unlikely that PCOS would NOT be in the OB/GYN notes of pregnant women, or in the billing codes. If they had EMR access, you’d like to think they thought about grabbing as much covariate data as they could.

    The full text is still behind the subscription wall; but judging from the abbreviation list, the covariates looked at are sparse. Far more likely to me is that the study question was framed poorly. It’s super easy to get “obesity linked to … ” papers published these days; editors and reviewers lower the scrutiny hurdles considerably when your findings gel nicely with their current confirmation biases.

    • MomTFH

      Thanks for delurking and for your thoughtful reply!

      Yes, it would depend on EMR, and in the United States, it would depend on how the information is shared. I am assuming you are in Ireland. We don’t have any national health system here. Health information is fragmented, and people change from one of hundreds of private medical insurances to another year after year, if they are lucky enough to have private insurance. People also tend to hide preexisting conditions here if they don’t need current treatment for them, to keep from being penalized in the future, when applying for private insurance.

      Even if there is an EMR, in our hospital, for example, it has just been implemented. The only information in the EMR would be data from hospital admission. That would include a height and weight, and from that one could calculate a BMI.There is no outpatient information there. Most of the time, the outpatient medical history is incomplete or empty. That was why I was saying it would be a good study if I was at a Kaiser institution, or possibly the Veteran’s Administration, two health systems that would be more likely to have comprehensive EMRs with outpatient information. Yes, a good medical history should include PCOS. But, many patients may not think to include that when they show up in labor. It isn’t on our checklist for preexisting conditions on the labor floor. I have never seen it is a problem list for any patient, much less on an obstetric patient.

    • binney'smom

      Just wanted to say that my daughter is on the autism spectrum and I was not overweight during my pregnancy, gained 30 lbs., but I had recently been diagnosed with hypothyroidism, and my doctor had mentioned that I had high androgen levels which was odd to her. Anyway, when I aksed about PCOS, two docs said no, you’re not overweight, blah,blah blah, you don’t fit the profile, but they did not listen to me about my other symptoms, issues with hormone imbalance, hair growth on my face, tired, irritable and very irregular with my periods despite the thyroid medicine. So you may be on to something. I hope someone finds the cause(s) and hopefully some real ways to help our children because life can be exhausting and we need help.

      • MomTFH

        Many clinicians can’t see beyond obesity, either as a cause or a definition of an associated condition. I had the opposite issue – I had a physician recently try to argue with me that I must have PCOS because I’m obese. No matter that I’ve never been diagnosed with it despite obstetric ultrasounds, clockwork menstrual cycles and ridiculous fertility. But, I’m obese, so…hey.

  2. irishup

    LOL! I can see my moniker being misleading, but I am USian, and have been working with US EMR.

    I did make a few assumptions, which will need to be verified or discarded
    when I get access to the full paper. I am used to the academic-tertiary care hospital setting, and practice that I see most definitely does not translate to community and for-profit hospital settings. So I do need to be mindful of my own biases! For instance, it is a helpful reminder to me that you’ve never seen PCOS on an OB problem list. We have a big high risk pregancy and REALLY big IVF program here; the reproductive hx documentation tends to be exhaustive because it’s easier & quicker to get one form through all the approval committees than to get 8 different forms through, and then train staff to remember which one goes for what!

    Back OT, a little lunchtime research reveals that this study is a spinoff of the CHARGE Study, (abstract & protocol from PubMed Central: ). This is data from a longitudinal case-control study being done from academic teaching hospitals in the UC system, AFAICT. The data is not being collected in the heat of the moment @ some community hospital L&D; it’s getting collected over several contacts with study staff as well as by a slew of clinicians. Also, as *study* information, it is off limits to insurers. Sure, some people will still hide stuff – for the very good reasons you cite, or they will be poor historians, but again, these are self-selected study participants who are probably well supported and well coached in disclosing their medical histories.

    Further, there are blood draws associated with this study, so there are ways to elicit conditions the person may have no knowledge of or have not reported. Lastly, it looks like study participation involves consent waivers for release of information from a variety of clinicians.

    From a study design POV, it looks like, were the primary research question being asked “Does obesity CAUSE $X_developmental outcome”, there would be ample opportunity to get PCOS data, and an IRB review section, not to mention NIH reveiw committee, would certainly require you to have your uterus-related conditions covered!

    The problem is that, while the CHARGE study is designed to identify genetic and environmental differences between Index ASD subjects (and their 1st degree relatives) and “normal” (*ahem*) Control subjects(& etc), this is not the same thing as DETERMINING whether any differences they identify are in fact causal. It’s only a very important step in identifing a “short list” of likely causal factors OR markers for risk. FURTHER, as far as I can tell from the study protocol, CHARGE was NOT designed with obesity as a specific factor of interest in mind.

    Not that I expect MSM reporting to address these issues. But I’d like to think that journal editors and reviewers WOULD. Alas, it seems that this level of rigor is not currently expected when it comes to The Obesity Epidemic Booga Booga Booga(TM). If your research shows something that goes against the current thinking, the level of scrutiny your methods and results recieve becomes positively MOLECULAR. When you’re going with the flow, even egregious mistakes can wind up being overlooked.

    (I’ll save my rant for why “obesity” is a poorly framed variable for another time. I’ve already written a tome! My apologies!).

  3. Ethel

    You want a cause then look to genetics and dad’s sperm (as in:, I suspect the fat moms had older partners as well. Hell, who’s to say that men on the autistic spectrum are chubby chasers because they love the tactile sensation and are less affected by visual status as a motivator (and since it is genetic their offspring have increased rates of autism as well).

    There are a million reasons why the study was weak sauce as my mildly affected and tactilely motivated husband would postulate.

  4. sara r.

    You really should read the book Gut and psychology syndrome, if you haven’t already. The author says that almost all of the children that she treats have a parent (mother generally ) with compromised immune system and an overgrowth of pathogenic gut bacteria, which is strongly tied to obesity and insulin resistance issues. So the results of the study aren’t surprising, but obesity might be a red herring ;it’s the gut disbyosis causing the obesity that may set the child up for issues later, since vaginally- born children inherit their gut bacteria from the mother, and babies born by c-section get theirs from the hospital, which is even worse.

  5. Very interesting take on it. I took wanted to scream hearing who the media changed what the study actually said, it’s so frustrating. I can’t wait to hear your research.

  6. Alice, online community manager, MANA

    Anytime someone can call into question how the media jumps on a newly released study without necessarily having a handle on what actually makes a good study, I am happy. So, thanks! Now I am compelled to look more into the possible connection…

  7. Hi, MomTFH!

    I don’t know how I missed this one, I’m usually all over the latest random-factor-dubiously-linked-to-autism story!

    Anyway, your treatment of it is good. I like your PCOS suggestion, as it actually makes sense — what with the higher androgen levels, which there is a fair amount of research showing a relationship between high levels of androgen exposure in utero and later development of autistic traits, and also because PCOS seems to be associated with a higher risk of pregnancy complications, which would be another thing it would share with autism — makes a lot more sense than “uhhhh, obesity did it!”, anyway.

    (I found you via Shakesville — I am autistic and have a degree in biochemistry. I don’t have PCOS, that I know of, though I do have some stuff going on that would be consistent with high androgen levels.)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s