claims is large, the science remains in substantial doubt, and the costs of

careful assessment are high. The key, however, is to base all decisions on at least a rough cost/benefit analysis.

[ page x ]

VI. CONCLUSION

We support the FDA's move away from a ban on health information in food labeling. Even without such a ban, the market is likely to produce and disseminate less than the optimal amount of consumer information on diet and health. The ban only exacerbates this problem. It is not surprising that in this restrictive information environment consumers are often thought to be uninformed about nutrition and health. Lifting the ban on health claims is likely to improve the nutrition IQ of consumers and encourage the production of more healthful food products.

This paper does not question whether it is necessary and cost effective for the government to regulate health claims after the ban is lifted. Rather, given the widely perceived need for government regulation of health claims, we consider which of two likely regulatory strategies is likely to enhance consumer welfare the most. We conclude that a fixed consensus rule is less desirable than an expected value (or flexible consensus) rule.

This conclusion is based upon an application of the standard cost/benefit framework under conditions of uncertainty. The framework shows that if the likely benefits of allowing a claim that turns out to be true are equal to the likely costs of allowing claim that turns out to be false, there only has to be slightly more than a 50:50 chance that a claim will turn out to be true to make the expected value of allowing the claim positive. In contrast, fixed consensus rules generally require more than a 95:5 chance that a claim is true before it is allowed. A consensus rule therefore prematurely emphasizes harm from Type I regulatory error

[ page 74 ]

(allowing claims about relationships that eventually prove to be false) over harm from Type II regulatory error (prohibiting claims that eventually prove to be true).

It does not appear that this premature emphasis on Type I error avoidance is appropriate for health claims about mainstream food products. A case history of prohibitions on health claims about saturated fat and serum cholesterol illustrates that Type II error is also important. Without evidence that. Type I errors consistently impose greater costs than Type II errors, it is hard to see how consumers benefit from a ban or a rigid consensus standard.

Several possible objections to the use of an expected value rule merit consideration. One might argue that the expected value rule is too hard to implement. We find that the feasibility of an expected value approach has been demonstrated by the FTC's advertising substantiation standard, which is an application of expected value principles. Moreover, the FDA, as an agency with risk assessment expertise, is well equipped to employ an expected value rule. Finally, although the likely costs and benefits of allowing (or prohibiting) a claim cannot be measured very precisely, we think that regulators should be face the questions posed by the expected value framework before they prohibit information on the grounds that it has not been "proven" to be true. More specifically, it is important to focus not only on the probability that a message linking diet and health will turn out to be false, but to focus also on the likely harm that would occur if information about a relationship is prohibited, and the message turns out to be true.

[ page 75 ]

One might also argue that resource constraints make the application of an expected value rule impossible. We believe that such an important policy should not be dictated by short-run resource constraints. Bans and rigid consensus rules bear hidden long-run resource costs which are likely to exceed the long-run costs of implementing an expected value rule.

Another argument against our proposal may be that it is old news, so it should be ignored. More specifically, one might counter that the FDA already bases its decisions on this standard cost/benefit framework. Although this is quite possible, it does not seem likely because the public debate has not focused on trade-offs between Type I and Type II error. Moreover, if this were the case, then it should be easy for the agency to follow our recommendation to make clear in its policy statement that the "weight of scientific evidence" necessary to support the use of health research findings in marketing messages will be determined by an expected value analysis. Otherwise, policy makers might find it too enticing to avoid controversy by maintaining the status quo through the use of a fixed consensus rule.

Explicitly requiring regulators to consider harm from Type I and Type II regulatory errors would not prevent the agency from taking a compromise approach. The FDA could use a flexible substantiation standard in most situations, while reserving the right simply to prohibit claims when a preliminary cost/benefit analysis indicates that the potential danger from a subset of claims is large, the science remains in substantial doubt, and the costs of careful assessment are high. This approach would still be a considerable improvement over a policy that always requires a high degree of consensus regardless of the consequences.

[ page 76 ]

APPENDIX 2.—ADDITIONal StatemenTS AND MATERIALS FOR THE

RECORD

STATEMENT OF SENATOR ORRIN G. HATCH

HUMAN RESOURCES AND INTERGOVERNMENTAL RELATIONS SUBCOMMITTEE
COMMITTEE ON GOVERNMENT OPERATIONS
JULY 20, 1993

THE FDA'S REGULATION OF DIETARY SUPPLEMENTS

I appreciate the opportunity to provide you with my viewpoint on the Food and Drug Administration's regulation of dietary supplements.

At the outset, I would like especially to thank you, Mr. Chairman, for your cosponsorship of Rep. Richardson's "Dietary Supplement Health and Education Act" (H.R. 1709), the companion my Senate legislation (S. 784).

Our legislation will bring much needed sanity and order to the regulation of the dietary supplement industry. I believe your hearing will amply demonstrate the need for regulatory clarification in this area.

Both pieces of legislation have the twin goals of providing a new framework for rational regulation of supplements, and the freedom for consumers to have the benefits of truthful information about these products. These bills establish sensible regulation for safe and properly labeled products.

The support for this legislation is clear. More than 100 million Americans regularly use supplements that include vitamins, minerals, herbal products, amino acids, and other nutritional substances to supplement their diets and improve their health. They know what the cosponsors know: supplements can prevent disease; improve healthy lifestyles; and help hold down health costs.

Our reforms should complete an effort initiated last year when Energy and Commerce Chairman John Dingell and Health Subcommittee Chairman Henry Waxman joined with me to author a new law -- the Dietary Supplement Act of 1992. That Act imposed a moratorium on FDA action to implement a health claims regulation for dietary supplements. And, it required the FDA to look at this issue all over again, and make some recommendations on how claims for supplements should be regulated.

On June 18, 1993, FDA published an Advanced Notice of Proposed Rulemaking, which, regrettably, reaffirms the need for a legislative solution because FDA wants to:

Restrict health claims for supplements, even though the legislative history makes clear FDA has ample authority to take efforts to promote the availability of information to consumers;

(457)