It's been fascinating to watch journalists wrestle with the latest risk study involving bisphenol-A (BPA) the controversial ingredient in plastics that, as Jon Hamilton at NPR, notes "environmental groups have blamed for everything from ADHD to prostate disease."
And that note of skepticism is not meant to suggest that BPA - which is identified by multiple studies as a notable endocrine disruptor - is without human and environmental risks. It is meant instead to point out a different kind of risk - the challenge for science writers covering a a well-promoted study about a high-profile compound in which even the researchers acknowledge that they aren't sure what their findings actually mean.
The study (paywall) was published this week in the Journal of the American Medical Association, under the title "An Association Between Urinary Bisphenol A Concentration and Obesity Prevelance in Children and Adolescents." The word "association", of course, tell us that this is not a study that elucidated a mechanism by which BPA might cause obesity but rather a data analysis that found an association between higher levels of BPA in urine and higher rates of obesity in the American children. In fact, the study authors based their findings on data from the U.S. Centers for Disease Control and Prevention (CDC).
Specificially, they looked at a CDC survey of some 2,838 youth, between the ages of 6 and 19, which included information on body weight and uranalysis of a series of compounds including BPA. (As background, CDC surveys suggest that thanks to the wide commercial use of BPA over past decades, traces of the compound can be found in about 93 percent of U.S. citizens). Anyway, to simplify here, these numbers found that obesity rates in those at the low end of BPA-testing stood at about 10 percent. At the high end, the obesity rate was 22 percent.
Oh, you say, does this mean that BPA exposure contributes to making you fat? Let's not forget that last year, a different study found an association between the compound and "abdominal obesity" in adults. Again, a legitimate question. But - to return to Jon Hamilton and his excellent story for NPR - "BPA could be making kids fat. Or not" - this study does not answer that question.
Hamilton quotes the study's lead author, Leonardo Trasande, an associate professor of pediatrics at New York University, on that issue: "When we find an association like this, it often raises more questions than it answers." For instance, Trasande points out, the data don't tell whether BPA promotes obesity or whether overweight children simply have more of it stored in their fat, which raises the levels in urine. It may also be that chubbier young adults eat more plastic-packaged snack foods and that it's their diets, not the compound, that is responsible for packing on the weight.
And none of that explains a significant anomoly in the results - that the obesity association was found for those of white/Caucasian descent but not with any statistical significance for either African-Americans or Latinos. That gap was so notable that the Los Angeles Times headline for the study was a wonderfully precise: Study Links chemical BPA to obesity in white children. As Melissa Healy wrote in that story: "Deepening the mystery surrounding the health effects of bisphenol A, a large new study has linked high levels of childhood and adolescent exposure to the industrial chemical to higher rates of obesity — in white children only."
Such complications also became the focus of this nicely done story from Mike Stobbe at AP which begins: "A provocative new study suggests a connection between the BPA chemical used in food packaging and childhood obesity, but the researchers say their findings don't prove it's the cause." And they came up in CNN health writer Jacquie Wilson's story on The Chart which included an entire "caveat" section, featuring a quote from JAMA editor-in-chief Dr. Howard Bauchner: “This paper is speculative.”
These four stories exemplified (at least to me) the best of the coverage (more than 400 stories as catalogued by Google). I won't tell you I loved the headlines which were mostly of the BPA= Fat Kids variety. So that even though Brian Bienkowski's story in Scientific American has a lead that focuses on the white children angle, the headline is a more generic: Chemical BPA Linked to Obesity in Children. (UPDATE: For clarity regarding Bienkowski's story, I should emphasize that my complaint is with the Scientific American headline on his piece, which was picked up from Environmental Health News. And at EHN, the headline was: BPA linked to obesity in white children.)
Of course, there were definitely stories without - let's say - nuance at all, such as this one by Christine Conville at The Boston Herald which begins:
"The more packaged food your children eat, the more likely they’ll be obese, an astounding new Journal of American Medical Association report finds.
Dr. Leonardo Trasande of the NYU School of Medicine established the staggering link between childhood obesity and plastic food packaging after studying the urine of nearly 3,000 American children."
And at the other embarrassing extreme, The Sacramento Bee merely reposted a press release from the American Chemistry Council (the umbrella group for industrial chemical manufacturers) headlined: ACC: New Study Provides No Evidence that BPA Causes Obesity. The Bee also picked up the Mike Stobbe story from AP. I would assume that the combination should definitely confuse a few of its readers.
But - without at all endorsing the Bee's rely-on-the work-of-others approach - perhaps confusion is the right message anyway. This is, as its authors acknowledge, a study that emphasizes an uncertain science and a complicated finding. And if we're really trying to do our job right, that's exactly the point that needs to be made. Probably more than once.
Comments
Good points about what the finer details of the study and what happens when you drill down for much more specific information about the BPA and obesity link. The problem - as with a lot of these association studies - is that the intent of the CDC survey was not to measure for this particular link, rather the authors have reanalyzed the government data for information related to their specific question. To get at the kind of smart questions being asked here, we'd need, I think, a much more targeted analysis. My own sense is that there is a kind of journal zeitgeist in accepting hot topics like BPA that raise the profile of the publication. Otherwise, why give so much attention to a study that the JAMA editor himself describes as speculative? And that's a point, of course, that we ought to think more about as journalists as well.
Hey, there.
I've been reading the article for the past few hours. I'm still trying to pick it apart. One of the things that continually gets me is the fact that, yes, there is a significant difference in the number of obese and non-obese individuals in the study. However, I don't really see how they're linking this to the BPA chemical as the exposure variable affects the outcome of obesity.
I'm looking at Table 1, and it surely shows that, from the study, 82.2% of people are not obese and 17.8% are obese. And then it appears that they grab that 17.8% and further re-analyze them (in table 2) with the socio-economic relationships. However, I don't feel that this really understand how this means "uniary BPA concentration was significantly associated with obesity."
If anything, it looks like the data says, "If someone is obese, they can have this range of BPA. And if someone is not obese, they can fall into this range of BPA." Given that, from table 1, I'm interpreting that people are more likely to be non-obese and have a high level of BPA. At least, there was a much larger number of individuals in the study population (from the total original study population).
To me, it just looks like there was a number of obese people in the study. Then again, I'm still having trouble picking apart what kind of statistical methods they've used. I know they have it presented as chi-square, but I think it's multiple regression. Anyone knowledable, feel free to point it out, please.
I have not read the study, but the key question that strikes me is whether time from eating to excretion was measured. As we know from the Teeguarden et al. feeding study, the level of urinary BPA will vary considerably depending on the time food was consumed due to rapid metabolism. Did the study do one spot measurement or multiple measurements? Given that the pharmacokinetics of BPA are now very well described, any failure to take this into account raises questions about the utility of the study design. Once again, we have amazing studies like Teeguarden et al. which actually tell us a huge amount about what BPA does and doesn't do (unconjugated, i.e. "estrogenic" BPA was below the level of detection in their 24-hour high-BPA feeding study), and we have the media (with the exception of Jon Hamilton) ignoring them for sensational yet largely meaningless headlines.
Bienkowski is a staff writer for Environmental Health News (as am I), where his story originally appeared with a more direct hed: http://www.environmentalhealthnews.org/ehs/news/2012/bpa-linked-to-obesi...