The Nutrition Labeling and Education Act of 1990 (NLEA) gives the FDA authority to regulate information displayed on food products and describes how a lack of certain types of information would be considered misbranding.1 Two rules, sharing one preliminary regulatory impact analysis (PRIA),2 have been proposed based on the authority granted by NLEA.3
The first rule covered by the PRIA, titled “Food Labeling: Revision of the Nutrition and Supplement Fact Labels” (NPRM1), includes a laundry list of potential changes that are designed to “assist consumers in maintaining healthy dietary practices.” 4 The changes required are numerous and involve changes to labels, increases in recordkeeping, and new analytic requirements.
The second rule covered by the PRIA, titled, in part, “Food Labeling: Serving Sizes of Foods” (NPRM2), focuses on labeling changes affecting food packages that hold a small number of servings.5 Specifically, NPRM2 requires that foods in packages that contain less than 200% of reference amounts customarily consumed (RACC) must be labeled as single-serving containers, while food packages with 200–400% of RACC must employ a dual labeling format that gives nutrition information for both amount per serving and amount per package. Additionally, the rule defines new RACC for a number of products and gives a new serving size for breath mints, among other smaller changes.
The major provisions associated with NPRM1 and NPRM2 are listed in table 1. Despite the large number of requirements and changes associated with the two rules, the FDA chose to take the unusual step of preparing one PRIA for both rules. Although this can be an acceptable strategy, when using it the FDA must be careful to analyze the costs and benefits of each sufficiently different provision separately. Instead, the PRIA only addressed a limited number of the proposed changes and only evaluated costs and benefits for a much smaller portion of these changes. This approach has contributed to an extremely weak justification for the rule.
Furthermore, the PRIA analyzed an insufficient set of alternatives and did not adequately incorporate uncertainty into the analysis. Most disturbingly, the FDA utilizes a single unpublished article that has not gone through peer review as the basis for benefits estimates for all provisions. This approach makes it impossible to assess benefits for individual provisions, and indeed, to ascertain whether the rules generate any benefits whatsoever. Though information provision has the potential to improve social welfare, the agency has failed to demonstrate that these rules will do so.
Executive Order (EO) 12866 states,
In deciding whether and how to regulate, agencies should assess all costs and benefits of available regulatory alternatives, including the alternative of not regulating.6
This essentially compels agencies to use regulatory impact analysis as a decision-making tool. In this case, the PRIA is used only as a justification for the rules. I believe the FDA should use the science cited in NPRM1 and NPRM2 as a basis for the benefits estimates in the PRIA and assess the efficacy of each provision separately.7 Quantitative risk analysis appears to be an appropriate method of analysis in this case. If the science is not sufficient to support this level of assessment, the FDA should abandon the provisions, make them voluntary (as seems appropriate for the “added sugars” requirement), or conduct original research to close the gaps in the science and assess the value of each provision separately. Doing so will place the agency in compliance with the letter and spirit of EO 12866.