Health Canada is in a fine mess at the moment, and there's only one way to unravel it. Trouble is, it involves opening up a controversial study about bisphenol-A in "BPA-free" plastics to scrutiny it may not be ready for.

Last month Health Canada published a study on bisphenol-A in non-polycarbonate plastics that demands both attention and a high level of skepticism, and merits replication before any significant conclusions are drawn from it. Unfortunately, while Health Canada is declaring their agency to have "no concerns with respect to the safety of baby bottles from [non-polycarbonate] plastics," they have yet to respond to the study's critics, and will not release detailed information that would match test results to the specific bottles they tested. The end result has been an outcry not only from journalists, but from some of the very companies the study implicates by its failure to disclose details.

Naturally, people like us want to know. Dozens of bottles were tested, including items made of polypropylene, polyethersulfone (PES) and unidentified "BPA FREE" plastics, in addition to a large number of polycarbonate bottles and even the liners used in drop-in style bottles. If there's BPA in products that are supposed to be free of the endocrine-disrupting chemical, everyone should know. But there's more to it than that.

And before we get wonkish, let me say that we do not believe that consumers should be concerned about the possible presence of BPA in non-polycarbonate baby bottles at this time. The levels detected by this study are significantly lower than those in polycarbonate, and we have reason to suspect that the data in this study is not as good as it should be to be actionable. We do think everyone should be concerned with the way Health Canada has handled this situation so far, and everyone should expect more from them.

The backstory

Health Canada quietly published "Migration of bisphenol A from plastic baby bottles, baby bottle liners and reusable polycarbonate drinking bottles" in the June issue of Food Additives and Contaminants, and noted that "trace" amounts of the chemical had been found in non-polycarbonate plastic bottles when tested in two solutions - one of clean water, with findings corrected to remove trace background BPA levels present in water samples, and one in a solution that mimics fatty liquids like breastmilk. Because the levels of BPA detected were very low, they explained them as possibly coming from background environmental variables outside of the control of most manufacturing facilities - we're talking about BPA in dust here - that was probably contaminating plastic that was not being made from BPA. (Unlike other plastics, bisphenol-A is the basic building block of polycarbonate plastic; it basically consists of chains of the stuff.)

Only after publishing the study results did Health Canada turn its eye to a request for public information (filed last November!) under Canada's Access to Information Act. When they did, however, the data released to CanWest News Service's Sarah Schmidt included at least one very hot piece of paper: Correspondence between the study's lead scientist and a department official stating that two non-polycarbonate baby bottles that were self-identified as BPA-free had shown "high readings." The brands of these bottles, which had clearly been discussed by name, were redacted with black censor boxes.

Our concerns about this situation span several domains - scientific, political, and public health - and we'll address each of them briefly. We're less concerned about the findings of this study, given some red flags we'll address below, than we are about the way Health Canada has been handling this situation. Canada is at the leading edge of BPA regulation, as the first country to ban it as a "toxic substance," and the way its public health regulators are handling this controversy is disquieting.

In investigating this issue ourselves, we have examined the full published study; discussed the situation with the reporter in CanWest's own investigation; and interviewed the head of a U.S. baby bottle company. We have also worked with a Canadian citizen to file an Access to Information Request for information that has not yet been released. For us, the situation boils down to three points of analysis we think can help consumers understand what's at stake here, and why Canada should disclose the full facts of the study.

1. The study was weird, and the writeup is downright suspicious.

Five polypropylene bottles, three PES bottles, and one whose plastic type was identified as "n.a." were tested for BPA migration into hot liquids, in addition to 14 polycarbonate bottles. The published study does not name brands, but since the bottles were purchased at retail outlets by Health Canada for testing, it is plausible to assume that the products were selected in order to survey the current market and bottles that are widely available for purchase.

Aligning the study's apparent aim - identifying BPA levels in non-polycarbonate bottles - with its design and reporting of its results is an exercise in futility. For example:

The results tables combine all non-polycarbonate plastics into a single category and report the findings only as an average. Doesn't this seem odd for a study designed to examine the potential BPA content of several plastics made from different ingredients and in different manufacturing facilities?

The internal correspondence revealing "high readings" for two bottles means that the average is skewed higher by these outliers. This makes providing readings only as a single average even more questionable.

Although researchers "corrected" findings to account for BPA in the water supply, the study lacks a control group that showed no BPA leaching even with this adjustment. This basic failure of study design means that the collected data cannot demonstrate that the BPA detected was actually present in the items being tested. That's what control groups are for - to show that the evidence collected has not been tainted by the process of collecting it.

I can think of at least two reasons a study like this could be compromised, off the top of my head:

1. If dust in the manufacturing facility could add BPA to the plastic, could dust in the lab add BPA to the testing environment, beyond that measured initially in the water?


2. Bottles were tested by dishwashing as well as by letting them rest with hot water in them; it is unclear (at least to me) whether these were the same bottles, but if they were, a dishwasher with a plastic interior could itself leach BPA.

But even thinking of and correcting for such possibilities isn't good science. Good science assumes that we may not be able to think of all the potential ways our data could be corrupted, and uses a control group to test our assumptions about what is happening in a study and provide evidence that we are doing things right.

Glass bottles were tested for lead and cadmium levels, and reported on in the study, but were not tested for BPA migration. Two thoughts:

1. If plastics that are not made with BPA are being challenged as potentially not BPA-free, why is glass assumed to be BPA-free? It too is produced in manufacturing facilities that are potentially subject to contamination, and it is also sometimes produced using additives to add strength to the glass.


2. If glass truly is BPA-free, as the scientists were willing to assume, see above: Instant control group.

2. Lack of disclosure harms all by damning none.

The brands involved in the study are obscured, according to CanWest's reporting, "on the grounds that the information could result in financial loss or prejudice the competitive advantage of a company" [Link]. But by failing to identify the levels of BPA found in each product, Health Canada has simply distributed the reputational damage equally among all companies claiming to sell BPA-free bottles.

The disclosure in correspondence that two bottles tested "high" has roiled the sub-industry composed of BPA-free startup bottle companies, in part because these companies all do their own testing and have the results to prove it. Green to Grow and Born Free have both called publicly for the full data to be released, and the head of another company, who spoke with ZRecs on condition of anonymity, has also requested that Health Canada release the detailed findings of the study.

Put yourself in the position of one of these companies. They have worked hard to provide BPA-free products to consumers, and have made their names developing safer options for babies. Even as major holdouts began abandoning polycarbonate plastic, ending with a watershed agreement by the big six U.S. baby bottle manufacturers agreeing to phase out polycarbonate entirely, consumers turned in record numbers to fully BPA-free companies, confident they were getting a safer product. The result was an age of unparalleled competition in a market usually locked up by established major players, and a massive shift of market share towards startups whose founders ushered in new ways of thinking about infant products. The case Health Canada's study essentially makes - although it clearly doesn't wish to - is that even using "BPA-free" products can be a crap shoot.

Left unchallenged, nothing could be more damaging to the fortunes of leading BPA-free companies like Born Free, Green to Grow, Medela, Thinkbaby, Adiri, and others, and as each is armed with its own testing data, they have multiple reasons to call for the information's release - especially the two bottles the lead scientist believed contained "high" levels of BPA. First, they are committed in principle to offering BPA-free products, and if they are selling contaminated product, they want to know it and correct the problem. Second, these companies have invested in more testing on their own than most bottle companies ever bothered to do before, some even branching out into biologic testing, which is far more sensitive - and more expensive - than tests to measure BPA, as they measure estrogenic activity in exposed cells and have a much higher sensitivity. Third, if the high readings come from a competitor's product, they are in an even better position to sell their own.

So with public-minded, scientific, and competitive reasons to contest Health Canada's suppression of the study's findings, why is the agency refusing to do so?

3. Health Canada's real goal is to protect itself from criticism.

The Access to Information Act is Canada's rough equivalent to the Freedom of Information Act. It allows government agencies like Health Canada to restrict access to information under specific circumstances. The argument against releasing this information is located in Section 20, part (1)(c).

(1) Subject to this section, the head of a government institution shall refuse to disclose any record requested under this Act that contains

(a) trade secrets of a third party;

(b) financial, commercial, scientific or technical information that is confidential information supplied to a government institution by a third party and is treated consistently in a confidential manner by the third party;

(c) information the disclosure of which could reasonably be expected to result in material financial loss or gain to, or could reasonably be expected to prejudice the competitive position of, a third party; or

(d) information the disclosure of which could reasonably be expected to interfere with contractual or other negotiations of a third party.

Item (c) above describes the grounds on which the agency has declined to provide data that matches specific bottles with the BPA levels their study found in them. If, for example, one "BPA-free" bottle by Avent and another by Playtex had been shown to have high levels of BPA, sales of those brands would likely plummet. But the Access to Information Act includes an additional clause in a different subsection of the same chapter, which provides the counterweight to excessive protection of companies at the expense of any agency's ability to protect public health, public safety, and the environment:

Disclosure authorized if in public interest

(6) The head of a government institution may disclose any record requested under this Act, or any part thereof, that contains information described in paragraph (1)(b), (c) or (d) if that disclosure would be in the public interest as it relates to public health, public safety or protection of the environment and, if the public interest in disclosure clearly outweighs in importance any financial loss or gain to, prejudice to the competitive position of or interference with contractual or other negotiations of a third party.

Considering that the Act was written for every government agency from the Canada Council for the Arts to the Taxpayers' Ombudsman, it seems highly ironic that an agency suppressing information about the presence of a chemical its chief minister declared a banned hazardous substance, and thus ignoring the mandate to override 20(1)(c) in the event a public health threat, would be the agency specifically charged with the protection of public health.

So what we have here is a major government study that disavows the significance of its own findings but fails in its published scientific study to report on outliers that are far more disturbing; that protects some specific companies to the detriment of everyone who claims to be dealing fairly and responsibly with consumers; and that ignores the spirit and letter of the law regarding public disclosure.

Failure to establish a control group, and failure to test a major class of relevant products. Suspect pooling of data points despite testing multiple types of materials. Private suggestions that the study should be redone to "verify" its findings. Refusal to release data to the very constituency that might be able to contest it.

Some might say it doesn't add up, but to us, it adds up quite neatly. Health Canada is afraid of fallout from and embarrassing challenges to its own questionable study. And we think it's wrong.

Health Canada has an obligation - to Canadians, by law, and to world citizens, by virtue of their leadership position regarding BPA regulation and, we would hope, research - to reveal the complete test results obtained from their 2008 study of the baby bottle market. If there are bottles which contain BPA that are not supposed to contain it, everyone - consumers, manufacturers, regulators, and scientists - need to know. What could follow would then be a frank discussion of advances or shortcomings in testing methods, about what peer review and reproducible findings mean when dealing with a highly charged public health issue with high levels of media attention, about potential BPA contamination during the manufacturing process, and about educating the public about what "free" really means when we talk about "BPA-free." But none of those discussions can occur until the data is released.

We don't like the position the agency has been put in any more than they do. But we do think it's their mess, and they now need to clean it up in the only way it can be cleaned up. Consider it a lesson in the intersection between science and public health policy: If you have a bad study that needs tweaking, redo it before you publish. Don't assume you can publish and then maintain a dispassionate distance from your own research, or release only selective portions of it. Once you publish, you are exiting the lab and entering the public realm, and you had better be damned sure your science is going to weather whatever comes next. The clear likelihood that Health Canada did not do so in this case boggles the mind.

On Friday, we engaged the services of a willing Canadian citizen to file an Access to Information Act request. (You don't have to be a citizen, but you do have to be in Canada to file a legal request.) The request read, in part:

I am seeking (1) the precise quantities of Bisphenol-A discovered in 2008 tests of non-polycarbonate infant bottles, (2) the precise testing method used, and (3) the bottle brand and model which tested positive for BPA. I believe that the information regarding potential public exposure to this toxic, banned substance must legally be released based on the spirit and provisions of the Access to Information Act, namely section 20 (6), which provides for disclosure of information that might cause financial harm to a company when such disclosure is "in the public interest as it relates to public health, public safety or protection of the environment."

CanWest's request sat on the agency's desk for eight months, so our own Canadian contact won't be checking the mailbox just yet. But we do believe this is a story that will evolve over the course of the next couple of weeks. Like we said, this report has nothing but enemies.

Like what we're doing? Our research, consumer advocacy, and independent product reviewing is supported primarily by readers' online shopping through ZRecs links to Amazon.com or Amazon's Canadian portal, Amazon.ca.