Tuesday, November 09, 2010

Sixth Circuit invalidates milk labeling rule

International Dairy Foods Association v. Boggs, Nos. 09-3515, 09-3526 (6th Cir. Sept. 30, 2010)

In the mirror image of IDFA v. Amestoy, 92 F.3d 67 (2d Cir. 1996), here the plaintiffs invalidated on First Amendment grounds most of a regulation designed to discourage milk producers from identifying that their milk came from cows not treated with recombinant bovine growth hormone (rBGH, aka rBST, sometimes capitalized differently). The decision contains some troubling signs of skepticism about consumer protection regulation more generally.

The Ohio Department of Agriculture (ODA), following a line of reasoning pushed by Monsanto, adopted a regulation prohibiting dairy processors from making claims about the absence of artificial hormones in their milk products (composition claims), and requiring them to include a disclaimer when making such claims about their production processes. IDFA and the Organic Trade Association sued for violation of the First Amendment and the dormant Commerce Clause.

rBST is a genetically engineered hormone used to increase milk production. The FDA approved its use in cows in 1993, concluding that milk from treated cows was safe for human consumption. The FDA "found that there was no significant difference between milk from treated and untreated cows," Interim Guidance on the Voluntary Labeling of Milk and Milk Products from Cows that Have Not Been Treated with Recombinant Bovine Somatotropin, 59 Fed. Reg. 6279 (Feb. 10, 1994). The FDA therefore concluded that composition claims such as “rBST-free” were likely to be misleading, because the FDA found no compositional difference between milk from treated and untreated cows. Production claims such as "from cows not treated with rbST” were less problematic, but could be misleading if they implied that milk from untreated cows was safer or of higher quality than milk from treated cows. The FDA suggested that production claims should be paired with a statement such as “[n]o significant difference has been shown between milk derived from rbST-treated and non-rbST-treated cows," or with a statement of "the firm's reasons (other than safety or quality) for choosing not to use milk from cows treated with rbST."

However, because the states traditionally regulate milk production, the FDA stated that its Guidance was nonbinding. It also recommended that states require companies to maintain records substantiating no-rBST claims and to make those records available for inspection.

Since the Guidance, consumers have increased their demand for milk from non-rBST-treated cows. Many dairy processors no longer accept milk from rBST-treated cows, and of course organic milk can’t come from cows treated with rBST.

When the ODA solicited comments about its proposed regulation defining what constitutes false and misleading milk labels, it received under 70 (out of 2700) emails and letters in favor of the proposed rule. Nonetheless, Director Boggs adopted the final rule:
(A) dairy products will be deemed to be misbranded if they contain a statement which is false or misleading.
(B) A dairy label which contains a production claim that "this milk is from cows not supplemented with rbST" (or a substantially equivalent claim) may be considered misleading on the basis of such language, unless:
(1) The labeling entity has verified that the claim is accurate, and proper documents, including, but not limited to, producer signed affidavits, farm weight tickets and plant audit trails, to support the claim, are made readily available to ODA for inspection; and
(2) The label contains, in the same label panel, in exactly the same font, style, case, and color and at least half the size (but no smaller than seven point font) as the foregoing representation, the following contiguous additional statement (or a substantially equivalent statement): "The FDA has determined that no significant difference has been shown between milk derived from rbST-supplemented and non-rbST-supplemented cows."
(C) Making claims regarding the composition of milk with respect to hormones, such as "No Hormones", "Hormone Free", "rbST Free", "rbGH Free", "No Artificial Hormones" and "bST Free", is false and misleading. ...
(D) Statements may be considered to be false or misleading if they indicate the absence of a compound not permitted by the United States [F]ood and [D]rug [A]dministration to be present in any dairy product, including, but not limited to antibiotics or pesticides. Except as otherwise provided in this rule, accurate production claims will not be deemed false or misleading.

The relevant standard is provided by Central Hudson. False and misleading commercial speech may be banned; other speech regulations have to pass a more stringent test. However, Supreme Court doctrine has developed a not-very-well-theorized distinction unknown to false advertising law generally: “inherently” misleading speech may simply be banned. “Potentially” misleading speech, however, should be cured with more disclosure rather than with a ban. The Lanham Act jurisprudence has some conflicting cases about whether disclosures should be presumed effective or ineffective; First Amendment cases, however, have not attempted to engage in empirical analysis of effectiveness, and this case won’t be the happy exception.

There were two sets of claims at issue: composition (the milk doesn’t contain rBST) and production (the milk comes from cows not treated with rBST). The district court found that composition claims were inherently misleading because they implied a composition difference between milk from the two sets of cows, conflicting with the FDA’s finding that there is no measurable difference.

The court of appeals held, however, that the record showed that there was a compositional difference between milk from untreated cows and conventional milk:
the use of rbST in milk production has been shown to elevate the levels of insulin-like growth factor 1 (IGF-1), a naturally-occurring hormone that in high levels is linked to several types of cancers, among other things. The amici also point to certain studies indicating that rbST use induces an unnatural period of milk production during a cow's "negative energy phase." According to these studies, milk produced during this stage is considered to be low quality due to its increased fat content and its decreased level of proteins. The amici further note that milk from treated cows contains higher somatic cell counts, which makes the milk turn sour more quickly and is another indicator of poor milk quality.

I find this an extremely interesting set of statements. (1) There is no discussion of whether the FDA also considered this evidence, or what level of deference would be given to the FDA’s rejection if it had done so. (2) Assuming this is post-1994 evidence, is the FDA required by the First Amendment to revisit its conclusions about appropriate labeling every time there is a new study? (3) Are these studies any good? That is, a government entity has found a certain fact: there is no compositional difference. What evidentiary standard is being applied to invalidate that factfinding?

Regardless, the court of appeals found that there was a compositional difference. But, even more sweepingly, “the failure to discover rbST in conventional milk is not necessarily because the artificial hormone is absent in such milk, but rather because scientists have been unable to perfect a test to detect it.” So it is possible that conventional milk contains rBST; we just can’t tell yet, but we might be able to tell someday.

Query: given this, can conventional milk producers advertise that “our milk contains no rBST”? If it’s constitutional to impose a substantiation requirement, then let’s concede that the FTC could go after this claim—though compare what supplement makers get to say based on cases such as Pearson. But imagine an organic producer suing a conventional producer under the Lanham Act for making a no-rBST claim: it seems to me that, given this evidence, the organic producer could not meet its burden of showing falsity.  Suppose the ODA responds by allowing conventional producers to advertise "our milk contains no rBST"?  Careful what you wish for.

The court of appeals concluded that there were “two distinct types of milk”: untreated cows cannot produce milk containing rBST, but conventional milk might contain rBST. If factual uncertainty is constitutionally sufficient to invalidate the regulation (and the court of appeals punted on this point, emphasizing the IGF-1 as well), then the court may have kicked the supports out of fifty or so years of FDA regulation, plus knocked some serious holes in the general substantiation requirement as well. That is, if the First Amendment requires allowing advertisers to make claims that might or might not be benefits—claims whose truth we simply do not know—then what’s left of substantiation?

The court of appeals found that the only uncertainty was in the extent of the difference. Thus, “rBST free” “at best informs consumers of a meaningful distinction between conventional and other types of milk and at worst potentially misleads them into believing that a compositionally distinct milk adversely affects their health.” Thus, the composition claim was not inherently misleading and needed to be tested under Central Hudson’s remaining three factors.

(Note the sleight of hand on what the compositional difference is. That is, we might accept that there is an IGF-1 difference, but that’s not what the producers want to advertise. They want to advertise “rBST-free,” which says nothing about IGF-1 and may be inherently misleading.)

Separately, the processors challenged the ban on composition claims about the absence of antibiotics and pesticides. The state argued that antibiotics and pesticides are "largely detectable in milk" and that "all milk is routinely tested for antibiotics, and the presence of any antibiotic in any amount renders the milk unacceptable for consumption," though cost considerations prevent the routine testing of every batch of milk. It did not submit evidence about the testing procedures used. The court of appeals concluded that evidence about testing “might well influence our determination” as to whether "antibiotic free" and "pesticide free" are inherently misleading:

If the State's testing can detect these substances and prevent any amount of them from being present in conventional milk, then such claims would be inherently misleading because they falsely imply that conventional milk contains antibiotics and pesticides when in fact the State tests to ensure that it does not. But there is no evidence in the record to verify the State's contention.

So this issue was remanded.

For a claim that is only potentially misleading, a complete ban will be upheld if the state shows that (1) its interest is substantial, (2) the ban directly advances that interest, and (3) a ban is no more extensive than necessary to serve that interest. Preventing deception is a substantial interest, but the state has the burden to demonstrate that “the harms it recites are real and that its restriction will in fact alleviate them to a material degree.” Here, the record of deception was weak at best. The FDA’s guidance merely said that the “rBST-free” claim “may imply a compositional difference,” but that doesn’t establish that the claim is necessarily misleading in every context. And the FDA cited no evidence or studies about consumer confusion. Thus it’s not evidence that the harm is real.

This is conflating a couple of issues. The rBST-free claim necessarily implies a compositional difference, and, as the court seemed to recognize above, no one could seriously dispute that implication. The real question is whether the compositional difference is meaningful to consumers for health or nutrition reasons, or whether this is like advertising “OUR milk is lead-free,” falsely implying the inferiority of other milk.

The court of appeals also dismissed the relevance of the consumer comments received by the ODA. Some comments did demonstrate confusion, such as one consumer who stated that she needed “to know that the milk I drink has no added hormones.” But “few if any” commenters indicated that their confusion stemmed from product labels, rather than other sources.

Anyway, the rule didn’t directly advance the state’s interest and was more extensive than necessary. Any confusion could be alleviated with a disclaimer informing consumers that rbST has yet to be detected in conventional milk. “Examples of possible disclaimers include a statement regarding the lack of evidence that conventional milk contains rbST, or even the disclaimer already required by the Rule to accompany production claims: ‘The FDA has determined that no significant difference has been shown between milk derived from rbST-supplemented and non-rbST-supplemented cows.’”

It’s really too bad that the court didn’t consider (and possibly wasn’t presented with; I haven’t read the briefs) the really compelling evidence that long, involved disclaimers of this sort—specifically, disclaimers about the FDA’s conclusions—have precisely zero effect on consumers’ understanding of the main claims. Consideration of this evidence would bear strongly on Central Hudson factors (2) and (3). See, e.g., Karen Russo France & Paula Fitzgerald Bone, Policy makers’ paradigms and evidence from consumer interpretations of dietary supplement labels, 39 Journal of Consumer Affairs 1 (2005); research presented here.

Anyway, the district court reasoned that a disclaimer “would only serve to confuse a consumer” by contradicting itself—“it would say a product is ‘free’ of rbST, but at the same time state that there is no rbST in other products, which defeats the purpose of making the claim in the first place." The court of appeals, however, thought that this argument assumed that conventional milk had conclusively been shown to be free of rBST. An appropriate disclaimer “could assure consumers that the substance is definitively not in milk so labeled while also advising them that it has yet to be detected in conventional milk.” Thus, the prophylactic ban on composition claims was more extensive than necessary.

The Rule also regulates production claims, requiring them to be accompanied by a disclosure stating that "[t]he FDA has determined that no significant difference has been shown between milk derived from rbST-supplemented and non-rbST-supplemented cows." This disclosure must be on the same label panel, "in exactly the same font, style, case, and color and at least half the size (but no smaller than seven point font)" as the production claim.

The district court granted partial summary judgment to the state, first finding that required disclosures are tested under a reasonableness standard rather than under Central Hudson. It found that production claims imply that conventional milk is inferior or unsafe and that the state has an interest in correcting this potentially misleading information. The court did find a factual dispute over whether the formatting requirements were unduly burdensome for small containers.

On appeal, the processors argued that more stringent review should apply to disclosure requirements. Zauderer v. Office of Disciplinary Counsel of the Supreme Court of Ohio, 471 U.S. 626 (1985), explained that, "because disclosure requirements trench much more narrowly on an advertiser's interests than do flat prohibitions on speech, warnings or disclaimers might be appropriately required in order to dissipate the possibility of consumer confusion or deception” (alterations, citation, and ellipsis omitted). Disclosure requirements are allowed where they’re "reasonably related to the State's interest in preventing deception of consumers," but not when they’re "unjustified or unduly burdensome."

The court of appeals here concluded that Zauderer applied to misleading commercial speech, whether it’s inherently or potentially misleading. This is consistent with the rule that commercial speech is protected because of consumers’ interests in receiving truthful information, not because of commercial speakers’ autonomy interests. Commercial speakers thus have minimal interests in avoiding the provision of factual information. (This rule is also consistent with the idea that the preferred remedy for potentially misleading speech is disclosure; or it could be, anyway, if we had any real idea what the difference between potentially and inherently misleading speech was.)

The disclosure requirement was reasonably related to the state’s interest in preventing consumers from being deceived by production claims. Like composition claims, production claims are potentially misleading “because they imply that conventional milk is inferior or unsafe in some way. But neither the FDA nor any study has conclusively shown that to be the case.” So, I guess, if conventional milk had been conclusively shown inferior, the disclaimer requirement would be unreasonable.

The processors argued that the state failed to show that production claims are misleading. But the state’s burden is more relaxed where the remedy is disclosure and where the possibility of deception is self-evident. Zauderer specifically stated that no survey was necessary to require disclosure. Although the FDA’s Interim Guidance and the consumer comments here were “weak” evidence of deception, they were enough to show that the risk was not speculative. The disclosure requirement is “reasonably related to thwarting that risk.”

However—shockingly!—the court of appeals found that there was no rational basis between the state’s concern and the requirement that the disclosure be “contiguous.” As a reminder, the Rule stipulates that disclosures must be "in the same label panel, in exactly the same font, style, case, and color and at least half the size (but no smaller than seven point font)" as the production claim, and also must be contiguous to the production claim. The font, style, case, and color requirements for the disclosure's text had “a self-evident rational basis” to prevent marketers from making the disclosure unreadable.

But the producers wanted to be able to use an asterisk instead of making the disclosure contiguous. Director Boggs testified that the ODA decided against this based on his "anecdotal experience" of talking to consumers in grocery stores, who informed him that "oftentimes it's hard to understand labels, especially when the print is so small." The court of appeals observed that this showed nothing about whether using an asterisk was effective, and there was no other basis in the record for Boggs’ belief. In light of the paucity of evidence in the record, the contiguity requirement had no demonstrable connection to preventing deception and lacked a rational basis. (This decision is somewhat worrisome for the FTC’s position on disclosures and disclaimers, except that I strongly hope that the FTC would be able to introduce consumer research and testimony from marketing experts who would convey the conventional understanding that, in fact, asterisks don’t work very well if at all for important claims.)

Finally, the processors argued that the disclosure requirement was unduly burdensome, but that concern stemmed largely from the anti-asterisk rule. There was no evidence that the font, size, and color requirements impaired their ability to communicate effectively.

They also alleged a dormant Commerce Clause violation from the disclaimer regulation, but given the First Amendment holdings, there was no significant burden on their ability to participate in interstate commerce. Without the prophylactic ban on composition claims and the asterisk prohibition, the Rule would be “largely indistinguishable from similar regulations in other states.”

Ohio’s rule also didn’t govern conduct occurring entirely out of state, which would have violated the dormant Commerce Clause. The processors argued that, given the complex national distribution channels for milk and the costs of changing labels, the Rule effectively required them to create a nationwide label corresponding to Ohio’s requirements. But the Rule had no direct effect on out-of-state labeling conduct, nor did complying with Ohio law risk conflict with the law of another state. The Rule didn’t impede or control the flow of milk products across the country, and the FDA has explicitly recognized the state role in labeling milk from untreated cows.

The processors argued that the Rule was purposefully and in practical effect protectionist, designed to help in-state interests at the expense of out-of-state interests. Unlike the situation in Vermont, Ohio’s milk producers are more likely to produce conventional milk, and thus the Rule stripped out-of-state processors of a competitive advantage from advertising their nonuse of rBST. But the purpose of the regulation was to prevent deception, which doesn’t indicate protectionism. Moreover, the discriminatory purpose argument was undermined by the processors’ own story, which was that traditional Ohio dairy farmers and Monsanto lobbied for the Rule to keep other Ohio dairy processors from converting to nontreated milk. As for practical effect, the rule’s burdens and benefits don’t correlate with being in or outside of Ohio.

Without an impermissible extraterritorial effect or a finding of protectionism, the court of appeals weighed the burdens and benefits of the Rule. State regulations are to be upheld unless the burden imposed on interstate commerce is clearly excessive in relation to the claimed local benefits. Here, without the absolute ban on composition claims and the asterisk ban, the processors’ argument that economic realities would require them to either stop selling in Ohio or conform nationwide labels to Ohio law was no longer persuasive. (I don’t think this is a dormant Commerce Clause violation, but I’m not sure why it follows from the invalidation of some of the Rule that the producers won’t be forced to print Ohio-conforming labels nationwide as a matter of economic reality. I presume that the idea is, once they can make composition claims and move the disclaimer around, putting those labels on nationwide wouldn’t be unduly burdensome given the lack of conflict with other states’ laws.) Anyway, the intended benefit of consumer protection is significant, and states have always had a particular interest in truthful advertising of food.

No comments: