Treehugger.com – a widely read environmental activist site – says in a comment reposted on the Huffington Post:
“At some point, even the people at Stats.org are going to have to acknowledge the growing pile of studies from all over the world adding to the case against Bisphenol A (BPA). The latest, From the University of Exeter, looked at the CDC (American Center for Disease Control) data and found that 60 year old men with the highest levels of BPA have about a 45% greater risk of heart disease than those with lower levels.”
The study doesn’t, in fact, say this. It is a cross-sectional analysis that expresses risk in the form of odds ratios, and odds ratios are not the same as percentage changes in risk. More importantly, cross-sectional studies cannot determine causality. They are snapshots in time of two factors and the relationship between the two may be arbitrary. As the Exeter researchers themselves note in the actual study, “The cross sectional nature of the associations reported need to be treated with caution, as it is theoretically possible, for example, that those with cardiovascular disease change their diets in such a way as to increase BPA exposure.”
In other words, people who eat more fatty food are at greater risk from heart disease – and they may also be more likely to ingest more BPA by virtue of eating more packaged or canned food. This is not a trivial objection to the study – in fact, it’s why an identical, earlier study by the same researchers was rejected by the European Union’s agency responsible for evaluating BPA. The Exeter researchers claim they have replicated their key finding, thus giving their claim for an association more weight, but many of the endpoints they measured lost statistical significance the second time around. In research that mines data for associations, this is called an alarm bell. They also concede that biologically plausible causes for disease based on such minute changes in BPA are speculative.
When the researchers are so open about not finding a causal link to heart disease and the need to interpret their results with caution, why should STATS charge ahead, like Treehugger, and pronounce the link and the risk certain? (To read our review of the Exeter study, click here).
More to the point, whether there is a “growing pile of studies from all over the world” indicating a risk from BPA all depends on what and how you count. So far, not a single risk assessment has pronounced BPA a threat anywhere in the world. Here’s a flavor of the counter evidence:
Since the EU risk assessment in 2006, there has been a review by Japan’s National Institute of Advanced Industrial Science and Technology (2007); an examination of claims of neurotoxicity by the Norwegian Scientific Committee for Food Safety (2008); an update to the European Union’s risk assessment (2008); an evaluation by the French Food Safety Agency (2008); a risk assessment by NSF International, a World Health Organization collaborative center (2008); a review of new data by the German Federal Institute for Risk Assessment (2008); a joint regulatory review for manufacturers by the FDA and Health Canada; a survey by Health Canada (2009); a risk assessment by Food Standards Australia/New Zealand (2009); two more surveys by Health Canada, one on canned powdered infant formula, the second on bottled water products (2009); a hazard assessment by California’s Environmental Protection Agency (2009); and a modeling study of BPA in humans by the German Federal Institute for Risk Assessment (2009).
Risk assessments take research, like that of the Exeter study, and see whether it is statistically robust and methodologically rigorous enough to be used for the purposes of risk assessment. Not all peer-reviewed research is equal, and every scientist knows this. Unfortunately, with BPA, the activist groups and the media recognize no such quality control. (And for those who don’t think quality control in statistics is important, read Richard A. Friedman M.D.’s analysis in the New York Times of a recent study on antidepressants which claimed that they were ineffective.)
For example, the latest study from the EPA failed to prove the low-dose hypothesis (the second from the EPA that failed to replicate the original theory of BPA’s risk to humans). And both went ignored. One of the EPA’s and the world’s leading experts on endocrine disruption dismissed claims made by Consumer Reports about the research as “an ad hominem attack… without scientific merit.” And he was ignored. The lead author of the European Union’s 2006 risk assessement said the Consumer Report’s investigation into BPA in cans was “highly biased” and hard to believe, but who listened to him (except STATS)?
It has become an article of faith in the environmental movement and on the left that BPA is lethal, and the U.S. government irresponsible for not banning it. But even the National Institute of Environmental Health Sciences has conceded that many of the studies it funded and which claimed a risk from BPA, were insufficiently rigorous for risk assessment. It has now toughened up its criteria for funding BPA research – demanding of its grantees the same experimental methodologies used in those studies which failed to find a risk.
Until risk assessments around the world find evidence of a risk from BPA, STATS is bound to give greater weight to their judgment. This is also why each new study needs to be read critically – and why studies that don’t find a risk need to taken seriously. We apologize for the inconvenience.