BPA causes emotional, behavioral problems in girl toddlers?

Is canned food your girl toddler’s problem?

A new study in Pediatrics reports a statistical association between bisphenol A (BPA) levels in a group of 244 pregnant women and increased anxiety, hyperactivity and depression in girls (but not in boys) at ages 2 and 3 as measured by parent-reported assessment on two behavioral scoring tests. BPA is used in food and beverage container linings and other consumer and medical/dental products.

The authors note that “the clinical relevance of the findings is unclear.” Yet they nonethless state that clinicians can advise concerned patients to reduce their exposures to BPA.

Here’s what you need to know about this study:

As with any epidemiological study, the results only show a statistical association or correlation — not cause and effect. As noted by the authors, the clinical relevance of the findings is unclear, meaning that there is no claim that the children exhibited abnormal levels of anxiety, depression or hyperactivity behavior. This point is especially relevant after examining Figure 2 and Tables 3 and 4, which show that the boys at age 3 had reduced levels of anxiety, depression and hyperactivity behavior and that these changes were of a greater magnitude than in the girls.

In short, if the authors wish to claim that girls are harmed by BPA exposure, they must admit that boys are helped, and to a greater degree. The fact of the matter is that both the statistical associations (increased anxiety, depression and hyperactivity in girls, reduced anxiety, depression and hyperactivity in boys) are likely due to chance and of no biological relevance at all.

The statistical associations or correlations are actually quite weak. So much so, that the researchers abandoned the customary 95% statistical testing for their correlations.

As the authors noted, they made many comparisons and never applied the standard mathematical correction process for multiple comparisons, likely because none of their comparisons would have been significant if they had done so. In short, it is uncertain if any of the claimed statistical associations would have been significant if proper statistical procedures had been applied. Note that in Tables 3 and 4, the only statistically significant associations claimed are for boys.

The authors state that they have based their conclusions on the “patterns, magnitudes and consistency” of their results as well as the consistency with the findings in their previous study (which also found increased hyperactivity and aggressive scores in 2-year-old girls) and with animal studies.

But this ignores the most relevant human study, that of Teeguarden et al. (Toxicological Sciences, vol. 123, pages 48-57, 2011), which found no free (biologically active) BPA in the blood of 20 volunteers who consumed a diet that contained BPA at a level greater than the 95 percentile level of the US population as measured by the CDC NHANES biomonitoring survey.

If there is no free-BPA in the blood, then there is no free-BPA to cross the placental barrier and interact with neuro-behavior receptors in the developing human embryo. So BPA cannot be the cause of the changes in behavior — positively or negatively — in 3-year-olds. Based on the results of Teeguarden et al., the statistical associations observed by Braun et al. are not biologically plausible.

Finally, we must note that the anchor author of this study is Bruce Lanphear, a sort of jack-of-all-junk-science.

JunkScience.com readers were introduced to Lanphear in Steve Milloy’s March 2001 FoxNews.com column “Get the Lead Hysteria Out.”

More recently, Lanphear produced this gem of a study with a size of N = 1.

The only thing science has to fear is Lanphear himself.

3 thoughts on “BPA causes emotional, behavioral problems in girl toddlers?”

  1. From the PR sheet: “Virtually all persons in industrialized countries are exposed to bisphenol A (BPA)…” – i.e. this is an uncontrolled ‘study.’
    Ditto: “BPA was detected in >97% of the … urine samples.” – i.e. we have nothing to compare this to.
    Again: “With adjustment for confounders…” – we diddled the data to conform to all our pre-conceived notions.
    After noting that there were several undefineable and unquantifiable but significant uncertainties, we are left with nothing more than a marketing breif.

  2. Do these people not understand the concept of a null study?

    If you have multiple, conflicting results of an effect, all of them weak, then the proper conclusion to draw, proper statistics or not, is that there isn’t an effect. You don’t need a heavy statistical background to understand that if girls are very mildly more hyper, and boys are mildly less hyper, that it probably means nothing. Did I just get good science teachers in grade school? I expect better from people who have gotten doctoral degrees.

Comments are closed.