Can feeding your child canned soup and vegetables make her fat? According to study published in the Journal of the American Medical Association (JAMA), it just might -- but only if your child is white. That's the latest junk science "finding" from yet another study designed to condemn the chemical bisphenol A (BPA). Despite obvious flaws with the study and the implausibility of its findings, newspapers around the nation, news websites, blogs and others continue to declare that there is “more evidence” that BPA poses a health problem.
You would think that reputable medical journals like JAMA would focus on science that truly adds insight and value about medical research and public health. After all, the group that publishes it -- the American Medical Association -- states on its website that its mission is "[t]o promote the art and science of medicine and the betterment of public health." But this study really doesn't do that. It's more akin to political science than it is to medical research and may harm public health if it leads to bans on food packaging that uses BPA resins. BPA bans are used to prevent the development of deadly pathogens in our food, and there are no good substitutes.
Surely, all publications have a bias toward studies that come up with positive associations even if dozens of others fail to find associations. Positive findings are more interesting news, even though this bias can generate false impressions as to the state of research. But that alone does not explain why JAMA published this study, which, much like a recent study on BPA and heart disease, is so flawed that no one -- let alone a serious medical journal -- should take it seriously.
The JAMA obesity study used data from the National Health and Nutrition Examination Survey (NHANES), a Centers for Disease Control and Prevention (CDC) program to assess national health trends. Each year, CDC collects data from a different group of volunteers (rather than follow the same group), conducting physical exams and interviews. From this database, the JAMA study pulled data related to BPA levels found in urine and body fat measurements for 2,838 subjects aged 6- through 19 years old. It found that for some participants high BPA levels were associated with higher body fat, which it concluded suggests that BPA contributes to obesity.
Like the heart disease study, the JAMA obesity study is seriously flawed for two main reasons. First, it relies on “spot sampling” of BPA -- measuring exposure via one urine sample per subject. The fact is BPA levels in urine change practically hourly, which means that one-time measurements tell us nothing! Second, it largely ignores the fact that the human body quickly metabolizes BPA before it can have any health impacts. For more background on these two issues, see my blog post on BPA and coronary heart disease.
Those two facts alone are reason enough for JAMA to reject publication of this study, but there are many other reasons why JAMA should have refused publication. First, the association they found -- if it means anything at all -- may simply suggest that children who eat processed foods (much of which is packaged in containers that use BPA resins) tend to be overweight. In fact, processed foods contain greater calories than do fresh produce. The study notes:
Obese children may drink more canned or bottled beverages, or eat more canned food, and thus have higher urinary BPA levels. Similarly, although we adjusted for excessive caloric intake and television watching—lifestyle-associated risks for childhood obesity—it may be those sedentary children consume foods high in BPA.
Moreover, the association they found was not even particularly compelling or consistent. Specifically, it found that BPA is associated with higher body fat among non-Hispanic white children, but not for Hispanic or black children. Unless there is some compelling and plausible body of research to suggest why this is the case, such disparate results suggest that the association was little more than a statistical accident.
The study authors themselves admit up front in the "abstract" that they have not really discovered anything new. They note:
BPA exposure is plausibly linked to childhood obesity, but evidence is lacking to date.
In the "discussion" section, they note the study’s key flaw:
If BPA is rapidly and completely excreted, as suggested by the few adult pharmacokinetic studies to date, then a single measurement of urinary BPA concentration would be a poor proxy for long-term exposure.
However, they then enter into an unscientific rationalization to justify drawing conclusions from what is admittedly a “poor proxy”:
Recent data from adults in NHANES 2003-2004, however, suggest that urinary BPA concentration does not decrease rapidly with fasting time, suggesting that it is stored in fat or other physiologic compartments.
In other words, the JAMA authors rest the validity of their methodology -- and ultimately their conclusions -- on an existing outside data source that simply “suggests” the human body stores BPA, which is hardly a scientific finding. In fact, the citation they provide for "recent" data goes to a single study from 2004 that also uses NHANES data and that is not very compelling for drawing the conclusions that JAMA authors use it to do.
In the 2004 study, the participants were asked to fast for various intervals ranging from less than 4.5 hours, between 4.5 and 8 hours, and between 8 and 24 hours. For those participants asked to fast 8.5 to 24 hours, 7.5 percent of the participants still had “relatively high” BPA in their urine samples when it was expected that they would have eliminated BPA from their body by then. The authors conclude that this shows that BPA either remains in the body longer than expected or there were other non-food exposures. It is worth noting that the study authors explain that some of the study participants may have cheated, possibly eating something or drinking something from a BPA container since they participants self-reported fasting times. Even drinking water from a water cooler could have increased BPA levels since the five-gallon jugs are made with BPA. The authors tried to control for this, but admit it is a weakness of their study.
In other words, the JAMA obesity study amounts to nothing more than a rationalization based on another study that also has conclusions that are nothing more than hypothesis generating. This is pretty thin because, after all, the mere existence of inconclusive study cannot fix a flawed study design.
Meanwhile, the JAMA obesity study did not cite or mention a much more recent and robust study conducted for U.S. EPA, perhaps because it contradicts these conclusions. This study measured BPA levels hourly in study subjects who consumed food from containers made with BPA levels. It found that even highly exposed individuals pass BPA quickly and little if any enters the blood stream. They concluded:
[R]eported BPA concentrations in human blood …are highly unlikely in the general population exposed orally to amounts as much as ~4 times greater than the 95th upper percentile of aggregate exposure in the general U.S population.
Nonetheless, JAMA sure did get lots of headlines and attention. But JAMA members should ask how this serves their association’s mission.