MNN - Mother Nature Network

made from a cross-sectional non-placebo controlled study. While it can be a common mistake, be careful not to confuse correlation with causation. One cannot say, based on this study, that BPA causes anything. At best, it can indicate areas for further examination. As the researchers were, no doubt, sensitive to the potential audience that might find their work interesting, they specifically note that "obesity develops over time, and causation cannot be inferred from a cross-sectional association of urinary BPA concentration with increased body mass, even when consistent with increasing laboratory evidence" (Trasande, Attina, Blustein, 2012). The "links" you might be referring to were calculated using Pearson's chi-squared tests or multivariable regression analysis (all tests were 2-sided with an alpha of .05), neither of which can be used to lay any claim on causation. Touching briefly on statistical significance, your concurrent statements that black children were "1.25 times more likely to be obese than those with lower levels" and that "scientists said [this was] not statistically significant" suggests to me that the meaning of statistical significance is unfamiliar to you. While there's not really the time or space to go into a lengthy discussion about this (but give it a quick Google search, plenty of schools put up excellent starter guides), for the purposes of this conversation if something isn't significant we cannot make any claim about it--in other words, if there's no significant difference between two groups, there is no difference between them. (I write this with a full appreciation for the arbitrary limitations of alpha levels, that non-significance can be suggestive, etc. etc.; but one must first know the rules before one can debate them.) Even in the case of white children, where there was significance, one cannot say it's "five times" more likely. Again, the study only informs us about correlation. Greater significance only increases our confidence that the difference is not the result of random chance (urinary BPA concentration quartiles for non-hispanic white children were p < .001 in all quartiles, all other race/ethnicity, p = ns). Finally, while the primary study outcome was obesity--using BMI z scores as secondary outcomes--it's important to note that obesity was defined by Trasande et al. using BMI z scores (obesity as 1.64 or greater [95th percentile]). Obesity, in other words, was determined by BMI z scores. While reading the study's results section is the best way to get into the data, the researchers nicely summarize the issue with their finding in their discussion section: "Our finding of the specificity of association in white participants is only confirmed in interaction tests for obesity as an outcome and not for BMI z score" (Trasande, et al., 2012). Typically, we might expect to see a dose-response curve for BPA. Instead, we see a significant difference for obese participants (p < .001) but no difference between those who are overweight (p = 0.08). Maybe there's a critical dose threshold one needs to get over to see an effect, but again, we do not yet know. So, as studies often are, the results of this one are far less racy than we might imagine. It helps add color to our understanding of this health issue, and that's good. It does not say that BPA causes obesity. It's more interesting that we haven't seen this effect in adults--considering metabolic systems are much closer to adults in the studied patient population than one might realize--because this finding was not found in two previous articles on the subject (Lang et al., 2008; vom Saal & Myers, 2008). There's certainly more on this subject that can be discussed at great length, but I do hope you find this post helpful.