Requirements for Additional Traceability Records for Certain Foods

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Food and Drug Administration

21 CFR Part 1

[Docket No. FDA-2014-N-0053]

RIN 0910-AI44

Requirements for Additional Traceability Records for Certain Foods

AGENCY:

Food and Drug Administration, HHS.

ACTION:

Final rule.

SUMMARY:

The Food and Drug Administration (FDA, the Agency, or we) is issuing a final rule establishing additional recordkeeping requirements for persons who manufacture, process, pack, or hold foods the Agency has designated for inclusion on the Food Traceability List (FTL). The final rule adopts provisions requiring these entities to maintain records containing information on critical tracking events in the supply chain for these designated foods, such as initially packing, shipping, receiving, and transforming these foods. The requirements established in the final rule will help the Agency rapidly and effectively identify recipients of foods to prevent or mitigate foodborne illness outbreaks and address credible threats of serious adverse health consequences or death resulting from foods being adulterated or misbranded. We are issuing this regulation in accordance with the FDA Food Safety Modernization Act (FSMA).

DATES:

This rule is effective January 20, 2023. For the applicable compliance dates, see section VI “Effective and Compliance Dates” in the

SUPPLEMENTARY INFORMATION

section of this document.

ADDRESSES:

For access to the docket to read background documents or comments received, go to

https://www.regulations.gov

and insert the docket number, found in brackets in the heading of this final rule, into the “Search” box and follow the prompts, and/or go to the Dockets Management Staff (HFA-305), 5630 Fishers Lane, Rm. 1061, Rockville, MD 20852, 240-402-7500.

FOR FURTHER INFORMATION CONTACT:

With regard to the final rule:

Katherine Vierk, Office of Analytics and Outreach, Center for Food Safety and Applied Nutrition, Food and Drug Administration, 5001 Campus Dr., College Park, MD 20740, 240-402-2122,

Katherine.Vierk@fda.hhs.gov.

With regard to the information collection:

Domini Bean, Office of Operations, Food and Drug Administration, Three White Flint North, 10A-12M, 11601 Landsdown St., North Bethesda, MD 20852, 301-796-5733,

PRAStaff@fda.hhs.gov.

SUPPLEMENTARY INFORMATION:

Table of Contents

I. Executive Summary

A. Purpose and Coverage of the Rule

B. Summary of the Major Provisions of the Final Rule

C. Legal Authority

D. Costs and Benefits

II. Table of Abbreviations/Commonly Used Acronyms in This Document

III. Background

A. Need for the Regulation/History of This Rulemaking

B. Summary of Comments on the Proposed Rule

C. General Overview of the Final Rule

IV. Legal Authority

V. Comments on the Proposed Rule and FDA Response

A. Introduction

B. Food Traceability List

C. General Comments on the Proposal

D. Scope (§ 1.1300)

E. Exemptions (§ 1.1305)

F. Definitions (§ 1.1310)

G. Traceability Plan (§ 1.1315)

H. Assignment of Traceability Lot Codes (§ 1.1320)

I. Critical Tracking Events Framework

J. Records of Harvesting and Cooling (§ 1.1325)

K. Records of Initial Packing (§ 1.1330)

L. Records of First Land-Based Receiving of Food Obtained From a Fishing Vessel (§ 1.1335)

M. Records of Shipping (§ 1.1340)

N. Records of Receiving (§ 1.1345)

O. Records of Transformation (§ 1.1350)

P. Procedures for Modified Requirements and Exemptions (§§ 1.1360 to 1.1400)

Q. Waiver Procedures (§§ 1.1405 to 1.1450)

R. Records Maintenance and Availability (§ 1.1455)

S. Consequences of Failure To Comply (§ 1.1460)

T. Updating the FTL (§ 1.1465)

U. Other Issues

VI. Effective and Compliance Dates

VII. Economic Analysis of Impacts

VIII. Analysis of Environmental Impact

IX. Paperwork Reduction Act of 1995

X. Federalism

XI. Consultation and Coordination With Indian Tribal Governments

XII. References

I. Executive Summary

A. Purpose and Coverage of the Rule

This final rule, which is part of FDA's implementation of FSMA (Pub. L. 111-353), establishes additional traceability recordkeeping requirements for persons who manufacture, process, pack, or hold foods for which the Agency has determined these additional requirements are appropriate and necessary to protect the public health in accordance with FSMA. These traceability recordkeeping requirements will help FDA rapidly and effectively identify recipients of such foods to prevent or mitigate a foodborne illness outbreak and address threats of serious adverse health consequences or death as a result of such foods being adulterated or misbranded (with respect to allergen labeling) under the Federal Food, Drug, and Cosmetic Act (FD&C Act). The requirements will reduce the harm to public health caused by foodborne illness outbreaks and limit adverse impacts on industry sectors affected by these outbreaks by improving the ability to quickly and efficiently trace the movement through the supply chain of foods identified as causing illness, identify and remove contaminated foods from the marketplace, and develop mitigation strategies to prevent future contamination.

We are issuing this rule because Congress directed us, in FSMA, to establish recordkeeping requirements for foods we designate that would be additional to the existing traceability recordkeeping requirements in the FD&C Act and FDA regulations. The existing regulations are designed to enable FDA to identify the immediate previous sources and immediate subsequent recipients of foods to address credible threats of serious adverse health consequences or death to humans or animals. This final rule adopts additional recordkeeping requirements for foods we have designated as high-risk foods in accordance with factors specified by Congress in FSMA. We are listing these foods on an FTL, which is included as a reference for the final rule. In accordance with FSMA, we also are publishing the FTL on our website concurrently with the issuance of the final rule. (See section V.B of this document for more information on the FTL.)

B. Summary of the Major Provisions of the Final Rule

The requirements of the final rule are focused on having persons who manufacture, process, pack, or hold FTL foods maintain and provide to their supply chain partners specific information (key data elements) for certain critical tracking events (CTEs) in the handling of the food, consistent with the developing industry consensus approach to food tracing. The information that firms must keep and send forward under the rule varies depending on the type of supply chain activities they perform with respect to an FTL food, from harvesting or production of the food through

processing, distribution, and receipt at retail or other point of service. Central to the proposed requirements is the assignment, recording, and sharing of traceability lot codes for FTL foods, as well as linking these lot codes to other information identifying the foods as they move through the supply chain.

The final rule requires persons who manufacture, process, pack, or hold an FTL food to establish and maintain a traceability plan that, among other things, describes their procedures for maintenance of records under the new requirements, identification of FTL foods handled, and assignment of traceability lot codes to FTL foods. Entities that grow or raise an FTL food (other than eggs) will also need to keep (as part of their traceability plan) a farm map showing the area in which the food is grown or raised, including geographic coordinates for the growing/raising area. Harvesters and coolers of raw agricultural commodities (RACs) (not obtained from a fishing vessel) that are on the FTL must keep records of their activities and provide information on them to the initial packers of these RACs. These initial packers, along with the first land-based receivers of FTL foods obtained from a fishing vessel, as well as entities that transform an FTL food (by manufacturing/processing a food or by changing the food or its packaging or labeling), must assign a traceability lot code to the food to help ensure accurate identification of the food as it moves through the supply chain, as well as maintain other records relating to their activities. Shippers and receivers of FTL foods must keep records of these actions, and shippers must provide the traceability lot code and other information identifying the food to the recipients of the food, including information relating to the traceability lot code source (

i.e.,

the entity that assigned the traceability lot code to the food). To avoid disclosing confidential information about their suppliers, instead of directly identifying the traceability lot code source of an FTL food, the shipper may instead choose to provide a traceability lot code source “reference,” such as an FDA Food Facility Registration number or a web address (which could be configured to require authentication for access), that provides an alternative means for FDA to identify and contact the traceability lot code source for the food. Taken together, these core subpart S requirements establish a structure for maintaining and providing traceability information that will enable FDA to more rapidly and effectively identify the source of contamination when investigating a foodborne illness outbreak than is possible under existing traceability recordkeeping requirements.

The final rule exempts certain small producers (including small produce farms, shell egg producers, and other producers of RACs) and, at the other end of the supply chain, certain small retail food establishments (RFEs) and restaurants. The rule also provides several other exemptions, including, but not limited to, those for the following: farms when food is sold or donated directly to consumers; food produced and packaged on a farm whose packaging maintains product integrity and prevents subsequent contamination; foods that receive certain types of processing, including produce that receives commercial processing that adequately reduces the presence of microorganisms of public health significance, shell eggs that receive a certain treatment, foods that are subjected to a kill step, and foods changed such that they are no longer on the FTL; produce rarely consumed raw; certain raw bivalve molluscan shellfish; persons who manufacture, process, pack, or hold FTL foods during or after the time when the food is within the exclusive jurisdiction of the U.S. Department of Agriculture (USDA); commingled RACs (not including fruits and vegetables subject to the produce safety regulation); RFEs and restaurants purchasing directly from a farm; certain ad hoc purchases by RFEs and restaurants from other such entities; farm to school and farm to institution programs; fishing vessels; transporters; nonprofit food establishments; and food for research or evaluation. (See section V.E of this document for more information on exemptions provided in the final rule.)

In addition to the exemptions codified in the final rule, the rule establishes procedures under which persons may request modified requirements or an exemption from the new traceability recordkeeping requirements for a specific food or a type of entity on the grounds that application of the requirements to that food or type of entity is not necessary to protect the public health. The rule also establishes procedures for requesting a waiver of one or more of the requirements for an individual entity or a type of entity on the grounds that having to meet the requirements would result in an economic hardship due to the unique circumstances of that entity or type of entity.

The rule specifies that persons subject to subpart S may have another entity establish and maintain required records on their behalf, although the person remains responsible for ensuring the records can be provided onsite to FDA within 24 hours of our request for official review. In addition, when necessary to help prevent or mitigate a foodborne illness outbreak, assist in the implementation of a recall, or otherwise address a threat to public health, firms must provide an electronic sortable spreadsheet containing information FDA requests on CTEs involving particular FTL foods for the date ranges or traceability lot codes specified in our request. Certain smaller entities are exempt from the requirement to provide this information in an electronic sortable spreadsheet, though they must still provide the information in other electronic or paper form. To help speed our access to information in such exigent circumstances, we may request the information remotely (

e.g.,

by phone) instead of onsite at the entity's place of business.

In response to many comments expressing concern about the ability of some entities to come into compliance within 2 years after the rule's effective date (as proposed), the final rule extends the compliance date for all persons subject to the rule to 3 years after the effective date. In this interim period, we intend to provide outreach and training, as well as guidance and other materials, to help all sectors of the food industry come into compliance with the new traceability recordkeeping requirements applicable to them under the new regulation.

C. Legal Authority

FSMA directs FDA to publish a notice of proposed rulemaking to establish recordkeeping requirements, in addition to the requirements under the FD&C Act and existing regulations, for facilities that manufacture, process, pack, or hold foods FDA designates. FSMA also directs FDA to designate the foods for which such additional recordkeeping requirements are appropriate and necessary to protect the public health.

D. Costs and Benefits

This final rule will impose compliance costs on covered entities by increasing the number of records that are required for covered foods. Entities that manufacture, process, pack, or hold covered foods will incur costs to establish and maintain a traceability plan and traceability records and one-time costs of reading and understanding the rule. Some firms may also incur initial and recurring capital investment and training costs for systems that will enable them to keep, maintain, and make available to other supply chain entities (and to us upon our request)

their traceability records. We estimate that the present value of costs of the rule over 20 years ranges from about $0.7 billion to $24.6 billion, with a primary estimate of about $6 billion in 2020 dollars at a 7 percent discount rate, and from $0.8 billion to $33.7 billion, with a primary estimate of $8.2 billion at a 3 percent discount rate. At a 7 percent discount rate, annualized costs range from about $63 million to $2.3 billion, with a primary estimate of $570 million per year. At a 3 percent discount rate, annualized costs range from about $53 million to $2.3 billion, with a primary estimate of $551 million per year.

By allowing faster identification of contaminated foods and increasing rates of successful tracing completions, the rule will result in public health benefits if foodborne illnesses directly related to those outbreaks are averted. This might also lead to more efficient use of FDA and industry resources needed for outbreak investigations by potentially resulting in more precise recalls and avoidance of overly broad market withdrawals and advisories for covered foods. We estimate public health benefits using several case studies of outbreak tracebacks for four pathogens associated with illnesses caused by covered foods. We calculate these benefits based on an estimated 83 percent reduction of traceback time resulting from the requirements of this rule. These benefits have a tendency toward underestimation of the total public health benefits because these four pathogens do not represent the total burden of all illnesses associated with foods on the FTL. However, adjustments made for undiagnosed and unattributed illnesses may have the opposite tendency of overstating both illnesses and benefits associated with listed foods. The present value of health benefits over 20 years ranges from about $0.6 billion to $23.7 billion, with a primary estimate of $8.3 billion at a 7 percent discount rate, and from about $0.9 billion to $34.5 billion, with a primary estimate of $12.0 billion at a 3 percent discount rate. The annualized monetized health benefits range from $59 million to $2.2 billion, with a primary estimate of $780 million at a 7 percent discount rate, and from $61 million to $2.3 billion, with a primary estimate of $810 million at a 3 percent discount rate.

The present value of (non-health) benefits from avoiding overly broad recalls and market withdrawals and advisories over 20 years ranges from about $2.5 billion to $18.8 billion, with a primary estimate of $6.1 billion at a 7 percent discount rate, and from about $3.6 billion to $27.3 billion, with a primary estimate of $8.9 billion at a 3 percent discount rate. At a 7 percent discount rate over 20 years, these benefits range from $233 million to $1.8 billion, with a primary estimate of $575 million. At a 3 percent discount rate over 20 years, these benefits range from $242 million to $1.8 billion, with a primary estimate of $596 million. Additional benefits of the rule may include increased food supply system efficiencies, such as improvements in supply chain management and inventory control; more expedient initiation and completion of recalls; avoidance of costs due to unnecessary preventive actions by consumers; reduction of food waste; and other food supply system efficiencies due to a standardized approach to traceability, including an increase in transparency and trust and potential deterrence of fraud.

II. Table of Abbreviations/Commonly Used Acronyms in This Document

Abbreviation

or acronym

What it means

ASN

Advance shipping notice.

BOL

Bill of lading.

CSA

Community supported agriculture.

CTE

Critical tracking event.

FDA

Food and Drug Administration.

FD&C Act

Federal Food, Drug, and Cosmetic Act.

FOIA

Freedom of Information Act.

FSIS

Food Safety and Inspection Service.

FSMA

FDA Food Safety Modernization Act.

FTL

Food Traceability List.

FTE

Full-time equivalent employee.

GPS

Global positioning system.

HACCP

Hazard analysis and critical control point.

KDE

Key data element.

LACF

Low-acid canned food.

NSSP

National Shellfish Sanitation Program.

OMB

Office of Management and Budget.

PTI

Produce Traceability Initiative.

RCR

Rarely consumed raw.

RAC

Raw agricultural commodity.

RTE

Ready-to-eat.

RFR

Reportable Foods Registry.

SECG

Small entity compliance guide.

SOI

Standards of identity.

SME

Subject matter expert.

USDA

U.S. Department of Agriculture.

WGS

Whole genome sequencing.

III. Background

A. Need for the Regulation/History of This Rulemaking

On January 4, 2011, President Obama signed FSMA (Pub. L. 111-353) into law. As a component of FSMA's overhaul of U.S. food safety law to ensure the safety and security of the nation's food supply, section 204 of FSMA requires FDA to establish recordkeeping requirements for facilities that manufacture, process, pack, or hold foods the Agency designates as high risk to facilitate the rapid and effective traceability of such foods. These recordkeeping requirements are additional to the food traceability requirements under section 414 of the FD&C Act (21 U.S.C. 350c) (added to the FD&C Act in title III, subtitle A, section 306, of the Public Health Security and Bioterrorism Preparedness and Response Act of 2002 (the Bioterrorism Act) (Pub. L. 107-188)) and the implementing regulation in subpart J of part 1 of title 21 of the Code of Federal Regulations (CFR) (§§ 1.326 to 1.368) (the subpart J regulation).

Congress directed FDA to adopt the subpart J recordkeeping requirements to allow the Agency to identify the immediate previous sources and immediate subsequent recipients of foods (commonly referred to as “one-up, one-back” recordkeeping) to address credible threats of serious adverse health consequences or death to humans or animals. We issued a final rule promulgating the subpart J regulation in 2004 (69 FR 71562, December 9, 2004).

In the case of a foodborne illness outbreak or evidence of contaminated food, the ability to follow the movement of foods through the supply chain—called product tracing or traceability—helps government agencies identify the points in the food supply chain, including the source of the product, where contamination may have occurred and, working with industry, remove the food from the marketplace. Efficient traceability enables the government and the food industry to take action more quickly to prevent illnesses and reduce economic harm.

In the years following the adoption of the subpart J regulation, FDA has learned that the one-up, one-back recordkeeping requirements in those regulations do not capture all the data elements necessary to effectively and rapidly link shipments of food through each point in the supply chain. Among the significant gaps in the subpart J requirements are the following:

• The lack of coverage of all sectors involved in food production,

distribution, and sale (

e.g.,

farms and restaurants are exempt);

• The lack of uniform data collection (

e.g.,

regarding the source of food ingredients used in each lot of finished product; no requirement to record a lot code or other identifier for all foods); and

• An inability to link incoming product with outgoing product within a firm and from one point in the supply chain to the next (see 85 FR 59984 at 59990, September 23, 2020).

These shortcomings of the subpart J regulation have hindered FDA outbreak investigations in many ways, including by making it more difficult to obtain tracing information from point-of-service firms that are exempt from the regulations. Even when such information is available, the records required under subpart J often are inadequate to facilitate swift and accurate traceback through the distribution chain to the producer of a contaminated food.

Recognizing the need for improvement in food traceability, in section 204(d)(1) of FSMA, Congress directed the Agency to adopt additional recordkeeping requirements to prevent or mitigate foodborne illness outbreaks and address credible threats of serious adverse health consequences or death to humans or animals resulting from foods being adulterated under section 402 of the FD&C Act (21 U.S.C. 342) or misbranded with respect to allergen labeling under section 403(w) of the FD&C Act (21 U.S.C. 343(w)). The additional recordkeeping requirements set forth in this final rule, which will be codified in 21 CFR part 1, subpart S (the subpart S regulation), will help FDA more effectively follow the movement of food products and ingredients on the FTL (“FTL foods”) both backward and forward throughout the supply chain.

Even before the enactment of FSMA, FDA had been considering ways to improve food product traceability and increase the speed and accuracy of our traceback and traceforward investigations, including holding public meetings and engaging in a pilot tracing project. Following the enactment of FSMA, FDA continued its work to improve food product traceability and to lay the groundwork for this rulemaking. Section 204(a) of FSMA directed FDA to establish pilot projects in coordination with the food industry to explore and evaluate methods to rapidly and effectively identify recipients of food. At FDA's request, and in accordance with that provision, the Institute of Food Technologists (IFT) conducted two product tracing pilots and issued a 2012 final report to FDA regarding those pilot studies (Ref. 1). In 2016, in accordance with section 204(a)(3) of FSMA, FDA submitted a Report to Congress that discussed the findings of the pilot projects and included recommendations for improving the tracking and tracing of food (Ref. 2).

In addition, on February 4, 2014, we issued a notice in the

Federal Register

(79 FR 6596) seeking public comment, scientific data, and other information to inform our draft approach to identifying high-risk foods. Section 204(d)(2)(A) of FSMA requires that the designation of high-risk foods be based on the following factors:

• The known safety risks of a particular food, including the history and severity of foodborne illness outbreaks attributed to such food, taking into consideration foodborne illness data collected by the Centers for Disease Control and Prevention (CDC);

• the likelihood that a particular food has a high potential risk for microbiological or chemical contamination or would support the growth of pathogenic microorganisms due to the nature of the food or the processes used to produce such food;

• the point in the manufacturing process of the food where contamination is most likely to occur;

• the likelihood of contamination and steps taken during the manufacturing process to reduce the possibility of contamination;

• the likelihood that consuming a particular food will result in a foodborne illness due to contamination of the food; and

• the likely or known severity, including health and economic impacts, of a foodborne illness attributed to a particular food.

On September 23, 2020, FDA published a proposed rule entitled “Requirements for Additional Traceability Records for Certain Foods” (85 FR 59984), to establish additional recordkeeping requirements for foods on the FTL, a proposed version of which was made available in the public docket for the rulemaking as well as on our website (Ref. 3). At the same time, we made available our “Methodological Approach to Developing a Risk-Ranking Model for Food Tracing FSMA Section 204 (21 U.S.C. 2223)” (RRM-FT Methodological Approach Report) (Ref. 4), which described how we generated the results from the risk-ranking model for food tracing (“RRM-FT” or “the Model”) that we used to help develop the FTL. The Model, which was peer reviewed, used a semiquantitative, multicriteria decision analysis risk-ranking approach, consistent with the factors set forth in section 204(d)(2) of FSMA, and it was operationalized with data relevant to those factors to generate results for foods we regulate (85 FR 59984 at 59991). We also made available a memorandum entitled “Designation of the Food Traceability List Using the Risk-Ranking Model for Food Tracing” (Ref. 5), explaining how we designated the foods on the FTL using the results of the RRM-FT.

As stated in the preamble to the proposed rule, the proposed traceability requirements were focused on having persons who manufacture, process, pack, or hold FTL foods maintain and share specific key data elements (KDEs) for certain CTEs in a food's supply chain, consistent with the developing industry consensus approach to food tracing. The information that firms would need to keep and send to their supply chain partners would vary depending on the type of supply chain activity they were performing with respect to an FTL food, from production of the food through processing, distribution, and receipt at retail or other point of service. Central to the proposed requirements is the assignment, recording, and sharing of traceability lot codes and traceability lot code sources (

i.e.,

the entity that assigned the traceability lot code) for FTL foods, as well as linking the traceability lot codes to other information identifying the foods as they move through the supply chain.

Since the publication of the proposed rule, there is still a need for improved traceability. Foodborne illness continues to have serious public health impacts. In the United States, there are approximately 800 foodborne illness outbreaks reported every year from all foods according to CDC outbreak surveillance reports, including about 200 outbreaks caused by foods covered by this rule (Refs. 6, 16). We estimate that nearly 770,000 illnesses annually in the United States are associated with foods covered by the rule (Ref. 16). Further, many Americans, besides those who become ill, are impacted by supply chain disruptions and temporary shortages due to overly broad recalls and less than fully efficient traceback investigations. A lack of consistent recordkeeping continues to hinder FDA's traceback investigations (Ref. 7). As described in the proposed rule, we have sometimes been unable to determine links between illnesses and specific product distribution due to inconsistent, unstandardized recordkeeping, lack of a deliberate method to connect records, and the frequent lack of lot tracing regarding distribution to specific retail locations. A lack of effective traceability

throughout the food supply has led to delays in product recalls and notification to the public, allowing potentially contaminated foods to remain on the market longer. While this rulemaking does not prevent the occurrence of outbreaks, these recordkeeping requirements can help identify the source of the contaminated food more quickly, potentially reducing the severity of the outbreak.

While parts of the industry have made progress in implementing traceability systems, the success has been confined to a subset of firms and product types, primarily in large firms where there is vertical integration in the supply chain or across the production of relatively homogenous products. Coordination through the supply chain across a wide range of firms varying in size, product mix, and production systems remains burdensome for many firms, especially those not vertically integrated. It is unlikely that without regulation the industry will ever achieve the level of systematic uniformity, accuracy, and efficiency needed to protect public health. The final rule—which applies only to covered foods and maintains the CTE/KDE structure of the proposed rule, but with modifications to address concerns raised in comments—provides a uniform set of requirements and expectations for traceability, reducing the challenges of coordination through the supply chain. The rule will greatly improve the efficiency and accuracy of FDA's traceback and traceforward operations, which should have a direct impact on the public health by allowing us to more quickly identify the source of contaminated food and remove it from the market.

B. Summary of Comments to the Proposed Rule

Although many comments express support for the proposed rule and its purposes, a number of comments request changes to simplify the traceability recordkeeping and record-sending requirements and reduce the burden of the rule on entities throughout the supply chain. Several comments ask that we reduce and simplify the CTEs for which records must be kept and the KDEs that firms must maintain for each event. While many comments acknowledge the importance of documenting the traceability lot code as an FTL food moves through the supply chain, several question how much information on the product and its producer is necessary or appropriate to share with downstream supply chain members.

Some comments ask that we broaden the circumstances under which a traceability lot code may be assigned. Several comments express concern about the feasibility of establishing requirements applicable to the “first receiver” of an FTL food, suggesting that others in the supply chain would be better suited to having and maintaining the required KDEs. Several comments request that we streamline the KDEs to be documented for shipping, receiving, and transformation events, and revise the information that shippers would be required to send to the recipients of the FTL foods, including the requirements applicable to farms.

Several comments ask that we clarify the scope of proposed exemptions from the FTL recordkeeping requirements, with some requesting that we broaden those exemptions to cover additional foods and/or firms. In particular, many comments maintain that having to comply with the rule would impose an undue burden on small farms and small RFEs, as well as other small supply chain firms. In addition, some comments request that we establish additional exemptions (different from those we proposed) for certain foods and supply chain entities.

Many comments object to the proposed requirement to make available to FDA, when necessary to help prevent or mitigate a foodborne illness outbreak, assist in the implementation of a recall, or otherwise address a threat to public health, an electronic sortable spreadsheet containing information in required traceability records for specified FTL foods and date ranges. In addition, although the proposed rule would permit firms to use existing records to meet the proposed recordkeeping requirements, several comments assert that the proposed rule would require unnecessary creation of duplicative records.

The comments generally express support for the proposed RRM-FT we used to determine the foods on the FTL, although some comments take issue with certain aspects of the Model as well as how we used it to generate the FTL. In addition, many comments request clarification as to whether particular foods or food products are on the FTL, and several comments ask that the final FTL not include several foods that were on the proposed FTL.

C. General Overview of the Final Rule

In response to comments we received, we have made several changes to the proposed traceability recordkeeping requirements for FTL foods that will make the final rule easier for supply chain entities to understand and comply with, while still ensuring that the rule substantially improves FDA's ability to respond quickly and effectively to foodborne illness outbreaks involving foods on the FTL. We believe the final rule more closely aligns the FTL recordkeeping requirements with developing industry best practices and effectively addresses stakeholder concerns about the complexity of the requirements and the need to protect the confidentiality of commercial information regarding suppliers.

The final rule includes changes to the requirements for a traceability plan (referred to in the proposed rule as “traceability program records”), including more streamlined requirements for what must be included in the plan and deletion of the proposed requirement to maintain a list of FTL foods shipped. In addition, for those who grow or raise an FTL food, the final rule requires the retention of a relevant farm map containing geographic coordinates instead of the proposed records documenting the growing area coordinates for individual traceability lots of the food.

The final rule also includes changes to certain of the CTEs for which persons subject to the rule must maintain KDEs. Instead of requiring the “first receiver” of an FTL food (which the proposed rule had defined as the first person other than a farm who purchases and takes physical possession of an FTL food that has been grown, raised, caught, or (in the case of a non-produce commodity) harvested) to maintain information on the origination, harvesting, cooling, and packing of food, the final rule places similar responsibility on the initial packer of a RAC (other than a food obtained from a fishing vessel) or the first land-based receiver of a food obtained from a fishing vessel. The KDEs required for shipping and receiving FTL foods have been streamlined and the shipping KDEs no longer apply to shipments that occur before a RAC is initially packed. A new CTE has been added to explain the requirements specific to harvesting and cooling of RACs before they are initially packed, and the CTEs for “transformation” and “creation” of an FTL food have been combined and clarified under a single transformation CTE.

The final rule includes changes to protect the privacy of individuals employed by supply chain entities and the confidentiality of business information concerning suppliers. To address the former, the final rule only requires firms to identify a point of contact within their traceability plan and the point of contact can be identified as a job title (along with a phone number), instead of the person's

name; all of the proposed requirements to provide a point of contact as part of the records sent to other supply chain entities have been deleted. In response to concerns about having to pass forward information on the traceability lot code generator for an FTL food, which could reveal information about a firm's suppliers, the final rule permits firms to provide a traceability lot code source reference, which is an alternative method through which information on the traceability lot code source could be made available to FDA, such as through a web address that provides the location description for the traceability lot code source. If the firm uses a web address as the traceability lot code source reference, the associated website may employ reasonable security measures, such as only being accessible to a government email address, provided the Agency has access to the information at no cost and without delay.

The final rule includes revisions to several of the proposed exemptions from the rule (generally broadening or clarifying the exemptions). We revised exemptions for certain small producers, and we expanded the exemption for farms when food is sold directly to consumers, such that it now covers donations as well as sales. We expanded the exemptions for foods that are subjected to a kill step and commingled RACs to extend these partial exemptions to include certain situations where it is known that the food will be subjected to a kill step (by an entity other than an RFE, restaurant, or consumer) or be commingled in the future, and to include foods that will be changed such that they are no longer on the FTL. Regarding the co-proposal for the exemption of small RFEs (full exemption vs. exemption from the requirement to make available, in certain circumstances, an electronic sortable spreadsheet containing requested tracing information), we have elected to fully exempt certain small RFEs and restaurants but also exempt from the requirement to provide a sortable spreadsheet somewhat larger but still relatively small RFEs and restaurants (along with certain farms and other entities that are relatively small). In addition, in response to comments we have added other partial or full exemptions from the regulations, including for the following: raw bivalve molluscan shellfish; persons who manufacture, process, pack, or hold certain foods subject to regulation by the USDA; certain ad hoc purchases by RFEs and restaurants from other such entities; and food for research or evaluation.

We have not made any changes to the risk-ranking model that we developed, consistent with the factors set forth in section 204(d)(2)(A) of FSMA, to determine which foods should be placed on the FTL. With respect to the FTL itself, on January 11, 2021, we provided additional clarity on the foods on the proposed FTL in response to stakeholder input following the release of the proposed rule (Ref. 8). With the publication of the final rule, we are providing additional description and clarification of FTL foods, including examples of foods that are and are not considered part of certain commodity designations on the FTL.

Finally, in response to the many comments expressing concern about the ability of farms, manufacturers, distributors, retail food establishments, and others to come into compliance with the new traceability recordkeeping requirements within 2 years after the effective date of the final rule, as we had proposed, we are extending the compliance date for all persons subject to the rule to 3 years after its effective date (which is 60 days after the date of publication of the final rule in the

Federal Register

).

IV. Legal Authority

Under section 204(d) of FSMA, in order to rapidly and effectively identify recipients of a food to prevent or mitigate a foodborne illness outbreak and to address credible threats of serious adverse health consequences or death to humans or animals as a result of such food being adulterated under section 402 of the FD&C Act or misbranded under section 403(w) of the FD&C Act, FDA was required to publish a proposed rule to establish recordkeeping requirements, in addition to the requirements under section 414 of the FD&C Act and the subpart J regulation, for facilities that manufacture, process, pack, or hold foods that FDA designates under section 204(d)(2) of FSMA as high-risk foods. We published the required proposed rule on September 23, 2020, and we are completing the rulemaking process with this final rule by establishing the subpart S regulation. We are promulgating this regulation under the following authorities:

• Section 204 of FSMA, the specific provisions of which are discussed throughout this document;

• Section 701(a) of the FD&C Act (21 U.S.C. 371(a)), which provides FDA with the authority to promulgate regulations for the efficient enforcement of the FD&C Act; and

• Sections 311, 361, and 368 of the Public Health Service Act (PHS Act) (42 U.S.C. 243, 264, and 271), which relate to communicable disease, including by providing FDA with authority to make and enforce such regulations as in FDA's judgment are necessary to prevent the introduction, transmission, or spread of communicable diseases from foreign countries into the States or possessions, or from one State or possession into any other State or possession (see section 361(a) of the PHS Act).

The legal authority for this rulemaking is discussed further in the preamble to the proposed rule (see 85 FR 59984 at 59993 and 59994).

V. Comments on the Proposed Rule and FDA Response

A. Introduction

We received approximately 1,100 comment submissions on the proposed rule to establish traceability recordkeeping requirements for persons who handle FTL foods (including comments on the FTL itself and the risk-ranking model used to develop it) by the close of the comment period, each containing one or more comments on one or more issues. We received comments from consumers, consumer groups, trade organizations, farmers, industry (

e.g.,

food manufacturers, processors, distributors), public health organizations, State and local governments, foreign governments and organizations, and others.

We describe and respond to the comments in Sections V.B through V.U of this document, as well as certain comments in Sections VI through IX. We have numbered each comment to help distinguish between different comments. We have grouped similar comments together under the same number, and, in some cases, we have separated different issues discussed in the same comment and designated them as distinct comments for purposes of our responses. The number assigned to each comment or comment topic is purely for organizational purposes and does not signify the comment's value or importance or the order in which comments were received.

B. Food Traceability List

Included as a reference to this final rule (and as seen in table 1) is the FTL, which sets forth the foods that will be subject to the subpart S requirements. In accordance with section 204(d)(2)(B) of FSMA and § 1.1300 of the final rule, we are publishing the FTL on our website concurrently with the issuance of this final rule. We included as a reference to the proposed rule the RRM-FT Methodological Approach Report (Ref.

4), which discusses the risk-ranking model for food tracing we used to determine the foods on the FTL. As stated in the proposed rule, the RRM-FT uses a semiquantitative, multicriteria decision analysis risk-ranking approach that is consistent with the factors specified in section 204(d)(2) of FSMA for use in designating the foods that will be subject to the additional traceability recordkeeping requirements and is operationalized with data relevant to those factors.

Using the results of the RRM-FT, we tentatively identified foods for which additional traceability records will be required, as we discussed in the Designation of the FTL Memorandum (Ref. 5). Based on that analysis, we developed a tentative list of FTL foods (Ref. 3). In response to questions and comments we received regarding the tentative FTL, in January 2021 we updated the table on our website showing the tentative FTL (Ref. 8). The updated table did not reflect a change in which foods were on the tentative FTL, but it included text to clarify the food products that are included in certain categories of foods on the tentative FTL.

Table 1 shows the current FTL that we are publishing with this final rule. The FTL being published with the final rule has not changed from the tentative list issued with the proposed rule. However, we have provided additional revisions to the descriptions of the commodities on the FTL to address some of the comments we received and provide greater clarity. The process for changing the FTL, which includes advance notice and an opportunity for the public to provide comment, is discussed in Section V.T of this document. We intend to update the FTL approximately every 5 years, subject to available resources. For the initial update to the FTL following publication of the final rule, we will take into consideration the compliance date for the final rule when deciding when to begin the process.

Table 1—Food Traceability List

Food traceability list

Description

Cheeses, other than hard cheeses, specifically:

• Cheese (made from pasteurized milk), fresh soft or soft unripened

Includes soft unripened/fresh soft cheeses. Examples include, but are not limited to, cottage, chevre, cream cheese, mascarpone, ricotta, queso blanco, queso fresco, queso de crema, and queso de puna. Does not include cheeses that are frozen, shelf stable at ambient temperature, or aseptically processed and packaged.

• Cheese (made from pasteurized milk), soft ripened or semi-soft

Includes soft ripened/semi-soft cheeses. Examples include, but are not limited to, brie, camembert, feta, mozzarella, taleggio, blue, brick, fontina, monterey jack, and muenster. Does not include cheeses that are frozen, shelf stable at ambient temperature, or aseptically processed and packaged.

• Cheese (made from unpasteurized milk), other than hard cheese

1

Includes all cheeses made with unpasteurized milk, other than hard cheeses. Does not include cheeses that are frozen, shelf stable at ambient temperature, or aseptically processed and packaged.

Shell eggs

Shell egg means the egg of the domesticated chicken.

Nut butters

Includes all types of tree nut and peanut butters. Examples include, but are not limited to, almond, cashew, chestnut, coconut, hazelnut, peanut, pistachio, and walnut butters. Does not include soy or seed butters.

Cucumbers (fresh)

Includes all varieties of fresh cucumbers.

Herbs (fresh)

Includes all types of fresh herbs. Examples include, but are not limited to, parsley, cilantro, and basil. Herbs listed in 21 CFR 112.2(a)(1), such as dill, are exempt from the requirements of the rule under 21 CFR 1.1305(e).

Leafy greens (fresh)

Includes all types of fresh leafy greens. Examples include, but are not limited to, arugula, baby leaf, butter lettuce, chard, chicory, endive, escarole, green leaf, iceberg lettuce, kale, red leaf, pak choi, Romaine, sorrel, spinach, and watercress. Does not include whole head cabbages such as green cabbage, red cabbage, or savoy cabbage. Does not include banana leaf, grape leaf, and leaves that are grown on trees. Leafy greens listed in § 112.2(a)(1), such as collards, are exempt from the requirements of the rule under § 1.1305(e).

Leafy greens (fresh-cut)

Includes all types of fresh-cut leafy greens, including single and mixed greens.

Melons (fresh)

Includes all types of fresh melons. Examples include, but are not limited to, cantaloupe, honeydew, muskmelon, and watermelon.

Peppers (fresh)

Includes all varieties of fresh peppers.

Sprouts (fresh)

Includes all varieties of fresh sprouts (irrespective of seed source), including single and mixed sprouts. Examples include, but are not limited to, alfalfa sprouts, allium sprouts, bean sprouts, broccoli sprouts, clover sprouts, radish sprouts, alfalfa & radish sprouts, and other fresh sprouted grains, nuts, and seeds.

Tomatoes (fresh)

Includes all varieties of fresh tomatoes.

Tropical tree fruits (fresh)

Includes all types of fresh tropical tree fruit. Examples include, but are not limited to, mango, papaya, mamey, guava, lychee, jackfruit, and starfruit. Does not include non-tree fruits such as bananas, pineapple, dates, soursop, jujube, passionfruit, Loquat, pomegranate, sapodilla, and figs. Does not include tree nuts such as coconut. Does not include pit fruits such as avocado. Does not include citrus, such as orange, clementine, tangerine, mandarins, lemon, lime, citron, grapefruit, kumquat, and pomelo.

Fruits (fresh-cut)

Includes all types of fresh-cut fruits. Fruits listed in § 112.2(a)(1) are exempt from the requirements of the rule under § 1.1305(e).

Vegetables other than leafy greens (fresh-cut)

Includes all types of fresh-cut vegetables other than leafy greens. Vegetables listed in § 112.2(a)(1) are exempt from the requirements of the rule under § 1.1305(e).

Finfish (fresh and frozen), specifically:

• Finfish, histamine-producing species

Includes all histamine-producing species of finfish. Examples include, but are not limited to, tuna, mahi mahi, mackerel, amberjack, jack, swordfish, and yellowtail.

• Finfish, species potentially contaminated with ciguatoxin

Includes all finfish species potentially contaminated with ciguatoxin. Examples include, but are not limited to, grouper, barracuda, and snapper.

• Finfish, species not associated with histamine or ciguatoxin

Includes all species of finfish not associated with histamine or ciguatoxin. Examples include, but are not limited to, cod, haddock, Alaska pollock, salmon, tilapia, and trout.

2

Siluriformes fish, such as catfish, are not included.

3

Smoked finfish (refrigerated and frozen)

Includes all types of smoked finfish, including cold smoked finfish and hot smoked finfish.

4

Crustaceans (fresh and frozen)

Includes all crustacean species. Examples include but are not limited to shrimp, crab, lobster, and crayfish.

Molluscan shellfish, bivalves (fresh and frozen)

5

Includes all species of bivalve mollusks. Examples include, but are not limited to, oysters, clams, and mussels. Does not include scallop adductor muscle. Raw bivalve molluscan shellfish that are (1) covered by the requirements of the National Shellfish Sanitation Program; (2) subject to the requirements of 21 CFR part 123, subpart C, and 21 CFR 1240.60; or (3) covered by a final equivalence determination by FDA for raw bivalve molluscan shellfish are exempt from the requirements of the rule under § 1.1305(f).

Ready-to-eat deli salads (refrigerated)

Includes all types of refrigerated ready-to-eat deli salads. Examples include, but are not limited to, egg salad, potato salad, pasta salad, and seafood salad. Does not include meat salads.

1

“Hard cheese” includes hard cheeses as defined in 21 CFR 133.150, colby cheese as defined in 21 CFR 133.118 and caciocavallo siciliano as defined in 21 CFR 133.111. Examples of hard cheese include, but are not limited to, cheddar, romano, and parmesan.

2

For a more comprehensive list, see Chapter 3 of the Fish and Fishery Products Hazards and Controls Guidance at

https://www.fda.gov/media/80637/download.

3

Data for catfish were excluded from the Risk-Ranking Model because Siluriformes fish (such as catfish) are primarily regulated by the U.S. Department of Agriculture.

4

“Smoked finfish” refers to a finfish product that meets the definition of a smoked or smoke-flavored fishery product in 21 CFR 123.3(s).

5

Under 21 CFR 123.3(h),

molluscan shellfish

means any edible species of fresh or frozen oysters, clams, mussels, or scallops, or edible portions of such species, except when the product consists entirely of the shucked adductor muscle.

We received several comments on the RRM-FT, the designation of foods on the FTL, and whether certain foods should or should not be included on the FTL. We respond to these comments in the following paragraphs.

1. Risk-Ranking Model for Food Tracing

(Comment 1) Several comments express general support for the RRM-FT methodology and the process FDA used to develop the FTL, as well as for our solicitation of stakeholder input. The comments maintain that the methodology is grounded in science and the process (including peer reviews) was rigorous, resulting in a targeted list of foods on the FTL. Conversely, other comments assert that the FTL fails to include key FSMA requirements and that the RRM-FT approach is not consistent with the goal or the statutory factors in section 204(d)(2)(A) of FSMA. These comments assert that the RRM-FT differs significantly from some of the FSMA requirements by adding criteria not in the statute and inappropriately merging multiple statutory factors into one Model criterion.

(Response 1) We appreciate the support for the RRM-FT and disagree with the assertions that it does not align with the statutory factors or that it differs from the FSMA requirements. As discussed in the Response to External Peer Review—Model Review (Ref. 9), subject matter experts (SMEs) reviewed the types of concerns raised in the comments when developing the draft RRM-FT, and peer reviewers generally agreed that the seven criteria we adopted were appropriately within the bounds of the FSMA-mandated factors.

(Comment 2) One comment claims that the RRM-FT methodology and the weighting used were not developed according to best practices for a multicriteria model, and the necessary expertise was not available to develop the Model appropriately. The comment maintains that the RRM-FT uses “an additive weighted approach” that is not appropriate when the model criteria are not preferentially independent because it would likely lead to some double counting of information.

(Response 2) We disagree with this comment. The results of the RRM-FT are founded on well-constructed criteria and the best available data. FDA addressed the issues raised by the comment during the peer review process (Ref. 9). As described in the final version of the RRM-FT Methodological Approach Report (Ref. 10), we recognize that mutual independence of criteria is desirable in a multicriteria-based model such as the RRM-FT. Within the constraints of the FSMA-mandated factors, we acknowledge that there are some correlations among the seven criteria or overlaps of data and information used in scoring, but we have taken steps to minimize potential overlaps. Most importantly, in cases where criteria are correlated, the RRM-FT defines them to represent separate aspects of value (of the data and information) to help ensure that the criteria represent independent preferences in ranking (see Methodological Approach Report, section 5.5 (Ref. 10)). The RRM-FT Methodological Approach Report and the peer review-model review report provide further explanation on how the RRM-FT operationalizes the seven criteria to minimize potential overlaps. FDA relied on the expertise of SMEs both within and outside of the Agency to develop the RRM-FT.

In developing the RRM-FT, we reviewed a number of available risk tools, including some developed by FDA and others from the published literature, including qualitative, semi-quantitative, and quantitative methods. We directly addressed the criteria independence issue by consulting with the project advisory group and multiple external expert panels and by considering comments and suggestions provided by peer reviewers.

(Comment 3) Many comments suggest that data used in the RRM-FT should be timely and reflect current food safety practices adopted by the industry. A few comments express support for using a 20-year timeframe (with appropriate weighting based on the year) for data for outbreaks and recalls and suggest that data older than 20 years not be used. Some comments express concern that the 20-year timeframe used in the RRM-FT is too long and suggest use of a shorter timeframe, such as 10 years, to reflect current industry practices. Whether comments prefer the use of 10 or 20 years, their concerns about older data are that it may not represent the current state of the industry because of advancements in science and food safety management, including the implementation of the produce safety regulation and the regulation on preventive controls for human food promulgated under FSMA. Furthermore, the comments assert that because industry usually attempts to address food safety problems and adopt enhanced food safety practices and mitigations to prevent recurrence of outbreaks, the use of older data may misrepresent risk. A few comments express support for the data weighting method in the RRM-FT, in which a weight of 0.4, 0.7, or 1 is applied

depending on the age of the data, but they request clarification as to whether we will always use the most recent 20 years of data and whether we will continue to use the same data weighting method in future updates of the Model.

(Response 3) We concur that data used in the RRM-FT should be timely and agree with the comment suggesting that a 20-year timeframe for outbreak and recall data is appropriate, while giving lower weight to (down-weighting) the older data. The RRM-FT incorporates a rolling data window in which the most recent 20-year data is used for scoring Criterion 1 (Frequency of Outbreaks and Occurrence of Illnesses), Criterion 7 (Cost of Illness), and Criterion 3 (Likelihood of Contamination), and within the 20-year timeframe, we down-weight older data. We believe a 20-year timeframe with down-weighting for older data provides an appropriate time window and scoring method to accurately capture the history of outbreaks and contamination associated with a commodity.

Criterion 5 (Manufacturing Process Contamination Probability and Industry-Wide Intervention) in the RRM-FT considers the current state of industry-wide interventions applied to each commodity-hazard pair. We acknowledge that industry may make concerted efforts to address food safety problems such as in response to outbreaks, and that food safety management practices may improve because of the implementation of regulations such as those for produce safety or preventive controls for human food, and these efforts are accounted for in the RRM-FT through the scoring of Criterion 5. Furthermore, to the extent that industry-wide preventive controls and interventions reduce food safety risk, the reduction in risk would also be reflected in the scoring, such as when the number of recent outbreaks (not down-weighted) is declining compared to older outbreaks, which would be down-weighted.

(Comment 4) Many comments state the RRM-FT criteria should be weighted differently, with more emphasis given to foods with validated preventive controls and less to epidemiological data. Specifically, some comments claim that the RRM-FT does not give sufficient weight to the three factors specified by Congress in FSMA section 204(d)(2)(A) that are related to contamination and production and processing activities,

i.e.,

factors (ii) (the likelihood that a particular food has a high potential risk for microbiological or chemical contamination or would support the growth of pathogenic microorganisms due to the nature of the food or the processes used to produce such food), (iii) (the point in the manufacturing process of the food where contamination is most likely to occur), and (iv) (the likelihood of contamination and steps taken during the manufacturing process to reduce the possibility of contamination). According to the comments, the RRM-FT gives too much weight to the other three FSMA factors, which are related to outbreaks or are epidemiological in nature. The comments assert that because the RRM-FT has five criteria to represent the three factors that are epidemiological in nature, this places too much emphasis on those factors in comparison to the two criteria that represent the factors related to the nature of food and manufacturing activities. The comments maintain that the over-emphasis of epidemiology in the Model contradicts Congressional intent and results in certain RACs such as leafy greens, herbs, tomatoes, cucumbers, peppers, and melons being deemed risky when, in the view of the comments, industry and the scientific community have greater food safety concerns about further processing of fresh produce such as fresh-cut fruits and vegetables (

e.g.,

because of a greater potential for contamination and for pathogen growth).

Conversely, other comments maintain that the Model puts too much weight on poor processing conditions rather than on inherent risk. The comments recommend that we weight criteria so that when a food goes through a validated kill step or other preventive control (including hurdle technology), the food is not on the FTL. Similarly, some comments ask FDA to weight Criterion 5 most heavily and not give too much weight to Criterion 6 (Consumption), maintaining that if there are strong industry interventions, the amount consumed is less relevant. Finally, some comments claim the sensitivity analysis in the RRM-FT is very limited and that we have not provided sufficient information to justify equal weighting of the criteria in the Model or the impact of such equal weighting on the ranking.

(Response 4) We do not agree with these comments concerning the appropriate weighting of the statutory risk factors, and the comments have not provided data to support their recommendations. As indicated in the RRM-FT Methodological Approach Report (Ref. 10), the RRM-FT uses the FSMA statutory factors to define the seven criteria used in the Model, and FDA considered different criteria weighting schemes in the approach that was peer reviewed. Peer reviewers generally agreed the Model's seven criteria were appropriate, and there was no general consensus for use of a different weighting scheme other than equal weighting of the criteria (Ref. 9). Therefore, we decided to weight the seven criteria equally in the RRM-FT. With regard to the comments requesting acknowledgment of the importance of a kill step in risk reduction, we agree and, as discussed in Section V.E.5 of this document, § 1.1305(d) of the final rule sets forth exemptions and partial exemptions for FTL foods that receive or will receive a kill step.

(Comment 5) Several comments suggest that FDA consider relevant data representative of the inherent food safety risk, including data relevant to intrinsic characteristics of the food (

e.g.,

pH, application of a validated kill step) and outbreak data from credible sources (both State and Federal Agencies). The comments assert that it is not appropriate to use outbreak data and other information from isolated events or problems specific to a particular facility or consumer misuse of the food, such as data from the Reportable Food Registry (RFR), because this information concerns facility-specific incidents that do not reflect overall risks to public health. The comments also suggest that FDA should have a scientific basis for including any food on the FTL.

(Response 5) The RRM-FT provides the scientific basis for the designation of the foods on the FTL. As described in the RRM-FT Methodological Approach Report (Ref. 10), the RRM-FT uses data and information on the intrinsic characteristics of the food and considers information on validated control measures in risk scoring. The RRM-FT uses the FDA Coordinated Outbreak Response and Evaluation (CORE) outbreak dataset (Ref. 11) that includes the CDC outbreak data for outbreaks in which the outbreak investigation demonstrated an association with FDA-regulated products. In addition, for outbreaks involving

Vibrio

spp. and marine and plant biotoxins, the Model uses data from CDC's National Outbreak Reporting System (NORS). To the extent that State agencies and other health departments report their foodborne illness outbreaks involving microbial and chemical hazards to the NORS, outbreaks relevant to FDA-regulated human foods have been considered in the RRM-FT. To apply the factors specified in FSMA section 204(d)(2)(A), it is necessary to consider both the characteristics of foods and hazards. In the RRM-FT, we classify FDA-regulated human foods into 47 commodity categories. Within each commodity

category, we identify food commodities and associated known or reasonably foreseeable hazards,

i.e.,

commodity-hazard pairs, using outbreak data, contamination data, and other information from multiple sources (Ref. 10). The RRM-FT uses RFR data as a source for scoring Criterion 3 only when sampling data are not available. When RFR data are used in the RRM-FT, these data are aggregated,

e.g.,

RFR reports from 2009 to 2019 are attributed to a commodity-hazard pair (a specific hazard in a specific food such as Shiga toxin-producing

Escherichia coli

O157 (STEC O157) in leafy greens), which minimizes the potential issue raised in the comments about overemphasis of facility-specific problems.

(Comment 6) Several comments state that the FTL should exclude foods that, according to the comments, are “not inherently dangerous.” Many comments maintain that fresh produce commodities have varying degrees of food safety risk; furthermore, the comments assert that fresh produce itself is not inherently risky and that risks are introduced by food production conditions and processing activities. These comments maintain that the risk of contamination is much greater with fresh-cut produce than intact RACs and that covering unprocessed produce under the food traceability rule will not improve public health. Several comments suggest that we factor production methods (

e.g.,

controlled environment vs. field environments for growing produce) and growing conditions for RACs into the RRM-FT, or that the designation of foods on the list be specific to where the food was produced. One comment states that the likelihood of contamination for fresh produce varies greatly because growing conditions vary greatly across farms and regions. The comment provides contrasting examples of fresh produce sourced from protected high tunnels irrigated with well water vs. from open fields irrigated with water from a canal near concentrated animal feeding operations. According to the comment, the risk of a fresh produce commodity (

e.g.,

leafy greens) is related to the latter type of growing environment and conditions. Therefore, the comment maintains that FDA should not require all leafy greens to meet the same traceability requirements because this would not be science-based or consistent with requirements in FSMA. Another comment asserts that, compared to field-grown leafy greens, those produced under controlled environments have a significantly lower risk of causing foodborne illness because of different risk factors (including minimal exposure to animals, potable water irrigation through root systems, minimal impacts from weather events, and other control measures). The comment suggests that such “controlled environment-produced leafy greens” should be given different consideration in the RRM-FT than other leafy greens.

(Response 6) We disagree with these comments, and the comments do not provide scientific data to support their assertions. As previously stated, the RRM-FT scores commodity-hazard pairs according to data and information relevant for seven criteria that account for the factors specified in FSMA section 204(d)(2)(A). As discussed in the RRM-FT Methodological Approach Report (Ref. 10), the RRM-FT criteria are related not only to the characteristics of the food but also to the production and manufacturing processes at the commodity level. For example, we evaluate the impact of fresh-cut processing by first identifying a variety of commodities under the Produce—RAC commodity category, and a variety of commodities under the Produce—Fresh Cut commodity category; for each of the commodities, we then identify known or reasonably foreseeable hazards,

i.e.,

commodity-hazard pairs for the commodities of Leafy Greens and Leafy Greens (Fresh-cut). Thus, the methodology accommodates on-farm production practices by identifying and evaluating hazards introduced on-farm (

e.g.,

STEC O157 in Leafy Greens), and it accommodates processing activities by identifying and evaluating hazards introduced in a processing facility (

e.g., Listeria monocytogenes

(

L. monocytogenes

) in Leafy Greens (Fresh-cut)). The Model then scores each commodity-hazard pair using data and information relevant to the seven RRM-FT criteria. For example, the impacts of production conditions and processing activities are reflected, on an industry-wide basis, in the data used to score Criterion 3 (Likelihood of Contamination) and the expert judgment used to score Criterion 5 (Manufacturing Process Contamination Probability and Industry-Wide Intervention). As such, the Model does consider production and manufacturing risks, as well as other aspects of risks such as the potential for the food to support growth of a pathogen (if present).

We agree with the comments that not all fresh produce is the same. Therefore, the Model identifies approximately two dozen fresh produce commodities based on the nature of the food and evaluates each of them separately,

e.g.,

Leafy Greens, Melons, Tomatoes, Stem Vegetables (see Ref. 10, Table A-2). In the Model, the identification of commodity-hazard pairs is based on available data and information,

e.g.,

foods and hazards associated with outbreaks and illnesses and detection of hazards in foods. The Model does not rank fresh produce at a more granular level than at the commodity level. Regardless of production practices (

e.g.,

field-grown vs. controlled environment), fresh produce within the same commodity group typically share similar characteristics in the potential for the food to support pathogen growth, and many contamination risk factors in controlled environments are similar to those found in traditional agriculture (Ref. 12). Moreover, we are not aware of data that warrant a separate evaluation based on production practices, and data are not available to evaluate commodity-hazard pairs at that level of granularity for the various criteria in the Model.

(Comment 7) Several comments maintain that the RRM-FT inappropriately grouped foods of different natures. According to the comments, FDA's approach to risk ranking is problematic because it groups different types of commodities together without consideration of the variety in each commodity, and, the comment claims, the risk of the commodity (

e.g.,

melons, leafy greens) varies depending on the variety (

e.g.,

watermelon vs. cantaloupe, spinach vs. lettuce). Several comments state that there are no data to suggest certain fresh herbs (

e.g.,

fresh bay leaf, makrut lime leaf, curry leaf, rosemary leaf) present any significant risk to human health or to support identification of many tropical fruits and leafy greens as high-risk foods. One comment asserts that while foods within a category may share similar characteristics in production and processing, the RRM-FT's analysis of a broad food category cannot adequately consider all the criteria because some criteria are specific to varieties, not commodities (

e.g.,

food safety technologies and innovations are usually developed for particular foods, not commodity groups). The comments suggest that we conduct individual analyses for particular foods and revise the FTL accordingly.

(Response 7) The RRM-FT considers the nature of the food through a categorization scheme that classifies FDA-regulated foods into 47 commodity categories. Furthermore, within each commodity category, the RRM-FT identifies individual commodities. In total, the RRM-FT identifies more than 200 commodities (see Ref. 10, Table A-2).

The Model does not rank commodities such as fresh produce at a more granular level than at the commodity level. We are not aware of scientific evidence that warrants a separate evaluation based on the varieties within a fresh produce commodity. Moreover, data on individual foods, such as specific varietals, are sparse and inconsistent across the variety of foods in the Model and on the FTL. For the purposes of the FTL, we determined that the appropriate level of granularity is at the level of “commodity,”

e.g.,

“tomatoes (fresh)” rather than “Roma tomatoes” or “cherry tomatoes.” Food items within the same “commodity” designation generally have similar characteristics, associated hazards, and production and supply chain practices and conditions, and peer review for the RRM-FT supported this approach (Ref. 13). Further, data used to assess components of the Model (

e.g.,

outbreak and illness data, likelihood of contamination, degree to which product supports growth, consumption, annual cost of illness) are available and adequate at the “commodity” level of granularity.

(Comment 8) A few comments assert that the RRM-FT does not adequately represent FSMA section 204(d)(2)(A) factors (iii) and (iv) (

i.e.,

“the point in the manufacturing process of the food where contamination is most likely to occur” and “the likelihood of contamination and steps taken during the manufacturing process to reduce the possibility of contamination”) and that the Model does not appropriately reflect differences in production systems and practices. According to the comments, the RRM-FT uses one criterion (Criterion 5: Manufacturing Process Contamination Probability and Industry-wide Intervention) to represent the two FSMA factors, which minimizes their impact on risk ranking, especially if there is a validated kill step for pathogens in the manufacturing process. The comments suggest that we consider more broadly the point in the overall supply chain where contamination is most likely to occur and include data to represent differences in potential contamination associated with different production, manufacturing, and handling processes and practices. The comments request that we revise the RRM-FT and the FTL to address their concerns and provide the public with an opportunity to comment on the revisions.

(Response 8) We decline to revise the RRM-FT and to solicit additional public comment before issuing the final rule. Regarding FSMA section 204(d)(2)(A) factors (iii) and (iv), these are incorporated into Criterion 5 of the RRM-FT (Manufacturing Process Contamination Probability and Industry-wide Intervention) as well as through the identification of commodity-hazard pairs under the broad range of commodity categories of FDA-regulated human foods. The commodities and the commodity categories (see Table A-1 in the RRM-FT Methodological Approach Report (Ref. 10)) represent a broad range of foods at different points in the supply chain with differences in production, manufacturing, and handling processes and practices. As discussed in the Response to External Peer Review—Model Review (Ref. 9), subject matter experts reviewed and addressed the types of concerns raised in the comments during the development of the draft RRM-FT, and peer reviewers generally agreed that the seven criteria we adopted were appropriately within the bounds of the FSMA-mandated factors, including the representation of FSMA factors (iii) and (iv) in the Model.

(Comment 9) Many comments assert that fresh produce from smaller-scale farms with relatively short supply chains (sometimes just a few miles) have lower risk than produce grown on larger farms, shipped long distance, or transformed without a kill step and shipped long distance. The comments maintain that locally grown commodities on the FTL, such as tomatoes, leafy greens, peppers, and cucumbers, do not have a greater risk than fresh crops not on the FTL. Some comments also assert that it is not scientifically sound to group locally grown and non-locally grown produce into one commodity in the RRM-FT because supply chain conditions and complexity vary between the two, so the food safety risk varies. The comments express concerns that such broad grouping will hurt the local food system, drive up the price of food, and limit the availability of fresh produce without reducing the risk of foodborne illness. Similarly, several comments claim the scoring of Criterion 5 in the RRM-FT is subjective, subject to change over time, and might not adequately represent small farms or local and regional food systems (LRFS). According to the comments, the scoring of Criterion 5, which is based on expert elicitations with several expert panels, reflects outcomes rather than root causes. One comment maintains that the size and type of production system and the length of supply chain are among the root causes of foodborne illness from fresh produce, but these factors are not adequately considered in the Model. Comments also note that the Criterion 5 score could change when industry improves production and manufacturing processes to better manage risk, which could affect both large and small operations. The comments suggest FDA obtain and use qualitative data that represent the scale and diversity of small, local farms and food businesses serving LRFS supply chains for scoring Criterion 5 and for use otherwise in the Model.

(Response 9) We do not agree that locally produced foods are inherently less risky than non-locally produced foods, and the comments do not provide scientific data to support their assertions. The Model does not differentiate locally grown fresh produce because how near to the point of sale the produce was grown does not change the characteristics of the food (

e.g.,

the potential for supporting pathogen growth) or the potential for on-farm contamination. The RRM-FT considers customary shelf life of fresh produce in scoring the potential for growth at a temperature at which the commodity (locally grown or not) is intended to be held and stored. While locally grown produce might be purchased and consumed within a time period shorter than that for non-locally grown produce, data are not available to show the potential for pathogen growth is sufficiently different between the two to result in a different score in Criterion 4 (Growth Potential, with Consideration of Shelf Life). Fresh produce commodities on the FTL, including locally grown produce, score higher than fresh produce commodities not on the FTL based on data relevant to the seven criteria in the RRM-FT. While we do not agree that locally grown FTL food is less risky than non-locally grown food, we understand that small operations may be particularly burdened by the provisions of the rule. We also understand that full traceability records may not be necessary when a consumer or RFE purchases food directly from a farm. Therefore, the final rule provides exemptions from some or all of the provisions of subpart S for certain smaller operations and in certain short supply chain situations, as discussed in sections V.E.2 and V.E.3, respectively, of this document.

With regard to the scoring of Criterion 5, FDA scores the seven criteria in the Model based on available data, both quantitative and qualitative. If quantitative data are not available for a certain criterion, the criterion is scored based on qualitative data. The RRM-FT relies on qualitative information from consultations with SMEs, including external expert panels, to score Criterion 5. The scoring of Criterion 5 is based on

the SMEs' assessments of each of the commodity-hazard pairs based on the status of industry-wide interventions as of 2019 (Ref. 10). The SMEs' assessment is based on the entire industry sector, including consideration of farms and operations of all sizes and scale collectively. It is not feasible to assess a commodity specific to the scale of a farm or LRFS supply chain because data for the seven criteria are unavailable at that level of granularity. In the peer review process, we specifically inquired about the adequacy of the expert elicitation process used to obtain qualitative data and address data gaps in the RRM-FT (Ref. 13), and there was general consensus among the peer reviewers that the process was adequate for the purpose. Changes in industry-wide interventions over time will be assessed as the data in the Model are updated in the future (see Response 488 about updating the Model).

(Comment 10) Several comments state that certain ingredients (

e.g.,

peanut butter) could be considered low risk but, because of their incorporation into many diverse foods, the magnitude of the impact if a contamination issue arises becomes greater, especially if no kill step is applied.

(Response 10) We agree that ingredients that are incorporated into many different foods have the potential to introduce widespread contamination. In the Model, we consider this possibility by including multi-ingredient foods, identifying and evaluating multi-ingredient commodity-hazard pairs based on data (

e.g.,

from outbreaks, recalls, and surveillance studies) and expert knowledge.

(Comment 11) One comment maintains that the RRM-FT does not provide justification for the criteria scores of 1, 3, and 9. According to the comment, these values can inappropriately inflate risk scores, and it is unusual to have the same value for a high, medium, and low score for all criteria when the ranges of values in each of the criteria are different. The comment also maintains that a multi-criteria model should include the elicitation of the value function, but the RRM-FT does not show that such an elicitation was done. The comment asserts that the RRM-FT uses arbitrary scoring bins of 0, 1, 3, and 9, leading to the top bin score of 9 being 9 times as bad as the bin score of 1, and FDA does not justify this difference. Another comment suggests that FDA use more evenly distributed scoring bins, claiming the 0-1-3-9 binning approach could over-inflate the criterion score, especially for Criterion 1 (Frequency of Outbreaks and Occurrence of Illnesses), Criterion 4 (Growth Potential, with Consideration of Shelf Life), and Criterion 5 (Manufacturing Process Contamination Probability and Industry-wide Intervention).

(Response 11) In developing the RRM-FT, we evaluated multiple value functions, including using an evenly distributed scale (1-2-3-4) and essentially a logarithmic scale (0-1-3-9) for scoring Model criteria. The scoring and binning methodology chosen was based on extensive consultations with external and internal SMEs as well as peer review. Given the intended use of the Model, an essentially logarithmic scale was recommended by multiple external panels in the expert elicitation process and the peer reviewers in the Model review panel. A justification of the chosen methodology is provided in the RRM-FT Methodological Approach Report (Ref. 10). The rationale behind using the scoring scale of 0-1-3-9 is that risk is not necessarily operating on a linear scale. Furthermore, using the 0-1-3-9 scale facilitates a greater degree of differentiation between higher- and lower-ranked food-hazard pairs, which is useful for informing the designation of the FTL. The RRM-FT methodology does not consider a criterion score of 9 to be 9 times “as bad as” a score of 1. Rather, as is the case with all multi-criteria decision analysis models, results from the RRM-FT provide a risk ranking of alternatives but do not directly quantify risk to the consumer (

e.g.,

the probability of illnesses), which requires a different methodology such as a quantitative risk assessment. The RRM-FT methodology appropriately gives the same criterion score to a range of data points that fall into the same scoring bin because, for its intended purpose, the RRM-FT does not attempt to quantify risk on a continuous risk basis, as would be done in a quantitative risk assessment.

(Comment 12) One comment claims the RRM-FT uses a method to determine the contribution of multiple hazards in which the total risk score for a food is determined by summing the risk scores of the food-hazard pairs associated with the food. According to the comment, this method makes a food associated with multiple hazards more likely to be designated high-risk because it would have a higher score. Furthermore, the comment suggests that FDA consider other factors (such as processing controls) so that a food is not more likely to be designated high-risk simply because it is associated with multiple hazards.

(Response 12) The RRM-FT does not use the summing method stated by the comment; instead, the Model uses an aggregation method that involves exponential transformation, summing, and log transformation taking into consideration the risk scores for all food-hazard pairs under the food. This aggregation method is not sensitive to the number of hazards associated with the commodity, but rather the risk score for the commodity is driven by the highest-scored commodity-hazard pair(s). With regard to considering processing controls, the RRM-FT considers processing controls when scoring Criterion 5, which accounts for steps taken to reduce contamination and industry-wide interventions.

(Comment 13) Several comments claim that Criterion 6 (Consumption) in the RRM-FT does not align with FSMA section 204(d)(2)(A)(v), which directs FDA to consider the “likelihood that consuming a particular food will result in a foodborne illness due to contamination of the food. . . .” The comments maintain that section 204(d)(2)(A)(v) was intended to be more about consumer handling of the food, such as whether there is temperature abuse, whether the food is cooked properly, and amount consumed. The comments maintain that the consumption criterion in the RRM-FT (which focuses on frequency and amount of consumption) may skew risk ranking, especially for popular foods. One comment acknowledges that higher consumption of a food could cause an outbreak with greater public health consequences but argues that is not what Congress directed FDA to evaluate.

(Response 13) We disagree with the comments and believe that Criterion 6 in the Model appropriately reflects FSMA factor (v) because consumption patterns affect the likelihood that consuming a particular food will result in a foodborne illness when the food is contaminated. Inclusion of the consumption criterion in the RRM-FT is based on extensive consultation with SMEs including external expert panels, and it has been subject to peer review (Refs. 9 and 13). Additionally, consumption is a standard component of a risk assessment, as described in the Food and Agriculture Organization (FAO)/World Health Organization (WHO) microbiological risk assessment guidance for food (Ref. 14). FDA defines Criterion 6 by using two data indicators, consumption rate and amount consumed (Ref. 10). When contaminated, products that are consumed frequently, in large amount, or both are more likely to cause widespread outbreaks. We think that FSMA factor (ii) (“the likelihood that a particular food has a high potential risk for microbiological or chemical

contamination or would support the growth of pathogenic microorganisms due to the nature of the food or the processes used to produce such food”) is the factor that relates more directly to the consequence from the potential for temperature abuse during the customary shelf life of the food, and we therefore considered that issue in the scoring of Criterion 4 (Growth Potential, with Consideration of Shelf Life) for the commodity-hazard pair. The RRM-FT does not consider consumer cooking because the commodities in the RRM-FT are defined as foods available for purchase by consumers.

(Comment 14) One comment asserts that the Model does not identify or explain a “cut-off” risk score above which foods are on the FTL, which makes it impossible to evaluate the impacts of the Model.

(Response 14) The RRM-FT methodology is designed to evaluate what the risk score is, not what risk score is used to designate a line above which foods are on the FTL. The final version of the Designation of the FTL Memorandum (Ref. 15) describes this cut-off score and explains how FDA uses results from the Model to determine whether a food is on the FTL.

(Comment 15) One comment asserts that the Model attributes fresh-cut leafy green outbreaks to both fresh-cut and RAC leafy green commodities. According to the comment, this inappropriately inflates the risk scores for both categories, particularly in the case of RAC products where it is often unknown if the contamination occurred after processing, and results in the RRM-FT scoring RAC leafy greens as higher risk than fresh-cut leafy greens. The comment asserts that this contradicts industry understanding and well-known science that fresh-cut produce by its very nature presents a higher risk than the same produce in RAC form.

(Response 15) The RRM-FT does not attribute outbreaks associated with fresh-cut leafy greens to both fresh-cut and RAC leafy green commodities. The Model does not “double count” outbreaks; each outbreak is attributed to a single commodity-hazard pair,

e.g.,

either the RAC or the fresh-cut product, depending on the source of the outbreak. FDA scores Criterion 1 (Frequency of Outbreaks and Occurrence of Illnesses) in the RRM-FT based on the Agency's determination of the source implicated in an outbreak,

i.e.,

whether it was determined to be a food vehicle (such as fresh salsa) or a contaminated ingredient used in the vehicle (such as contaminated tomatoes used in the fresh salsa) (Ref. 10). We attribute the number of illnesses and outbreaks to a commodity-hazard pair according to information on the contaminated ingredient (

i.e.,

the source of the contamination), not to the food vehicle implicated (if it is different from the contaminated ingredient), when both the contaminated ingredient and the food vehicle were identified in the outbreak investigation. For example, if fresh salsa was implicated in a foodborne illness outbreak but tomatoes were identified as the contaminated ingredient, the outbreak would be attributed to tomatoes and not fresh salsa.

We disagree with the comment's assertion that the RRM-FT methodology contradicts the current scientific understanding of the route of pathogen contamination in fresh produce. We considered public comments on the 2014 draft methodological approach in the development of the RRM-FT (Ref. 4), and we had the methodological approach peer reviewed in 2016 (Refs. 9 and 13). Based on the peer-reviewed approach, we updated the underlying data, where major data sources for scoring in the Model were updated to 2019 or the latest available data (Ref. 10). Consequently, our approach to outbreak attribution is based on the best available information on the source of contamination, which remains consistent with current scientific understanding. For example, the fact that the commodity-hazard pair risk score is higher for the pair “Leafy greens—STEC O157” than for the pair “Leafy greens (fresh-cut)—STEC O157” (risk score of 430 vs. 310) (Ref. 10) reflects the fact that STEC O157 is more likely to originate in RAC leafy greens (but can sometimes remain in fresh-cut leafy greens after processing). However, for a hazard associated with leafy greens for which the processing environment is a typical route of contamination (such as

L. monocytogenes

), the risk score is higher for “Leafy greens (fresh-cut)—

L. monocytogenes

” than “Leafy greens—

L. monocytogenes

” (risk score of 370 vs. 330). The RRM-FT systematically scores relevant commodity-hazard pairs for RAC leafy greens and fresh-cut leafy greens. The Model then calculates a risk score for each commodity using an appropriate aggregation method (Ref. 10), where the risk score for the commodity is driven by the risk score for the highest-scored commodity-hazard pair(s); this results in a commodity risk score that is higher for RAC leafy greens than fresh-cut leafy greens.

(Comment 16) One comment suggests that we consider the wide variations in shelf life and pathogen growth potential among dairy products. As an example, the comment compares a pathogen like

L. monocytogenes

in a soft Hispanic-style cheese, which has strong growth potential, to any pathogen in ice cream, which has effectively zero growth potential. The comment maintains that having two indicators for scoring Criterion 4 (

i.e.,

using a scoring matrix of Growth Potential and Shelf Life) is problematic and may skew the criterion score for a commodity as a whole compared to the scores for individual foods. For example, the comment maintains that it does not seem accurate to have the same Criterion 4 score for a dairy product with a short shelf life/strong growth potential as for a dairy product with a moderate shelf life/moderate growth potential.

(Response 16) We agree that it is important to consider the variations in pathogen growth potential. Consistent with the comment's suggestion, results from the Model show a wide range of Criterion 4 scores among commodity-hazard pairs for dairy commodities. To determine the score for Criterion 4, we use a single indicator based on the potential that a food would support the growth of pathogenic microorganisms due to the nature of the food, and the extent of growth as affected by the customary shelf life of the food and the temperature at which the food is intended to be held and stored. This reflects a revision that we made to the draft approach, taking into consideration comments we had received from the public and from peer reviews of the RRM-FT (Refs. 9, 13). The commenter incorrectly stated that Criterion 4 in the 2020 RRM-FT Methodological Approach Report (Ref. 4) used for the proposed rule included two indicators. We changed the Criterion 4 scoring definition to one indicator in the revised Model (2020) in response to comments peer reviewers and stakeholders had made on the 2014 draft. As a result, the revised Model uses only one indicator to score Criterion 4, which is “Growth potential, with consideration of shelf life,” instead of using “Growth potential/shelf life,” which was evaluated as two separate indicators in the draft approach. The scoring definition for Criterion 4 includes the amount of growth (log

10

increase) given customary shelf life. As described in the RRM-FT Methodological Approach Report (Ref. 10), the revised definition allows us to appropriately apply data from growth studies and predictive microbiology databases, as well as avoid potentially skewing the criterion score if two indicators were used.

(Comment 17) One comment expresses concern about treating “Dairy” as one group in the RRM-FT and asserts that foods selected in the RRM-FT are not representative of the wide diversity of the dairy industry. The comment states that the dairy industry makes a wide variety of products, including ice cream, yogurt and cultured dairy products, butter, hard cheeses, soft cheeses, sour cream, cottage cheese, dips, canned sweetened condensed and evaporated milks, pasteurized flavored and unflavored fluid milks, dried milk, whey powders, raw milk, and raw milk products. The comment asserts that each of these products has unique intrinsic characteristics and that the manufacturing process of each product may involve a unique combination of processing steps. The comment further maintains that it is not appropriate to combine pasteurized and unpasteurized dairy products into a single category because some dairy products are virtually risk-free, while raw milk and raw milk products are inherently risky. For support, the comment cites CDC data indicating that over 70 percent of outbreaks associated with dairy products are attributed to raw milk and raw milk cheeses. Therefore, the comment suggests that we revise the dairy food classification considering intrinsic properties (

e.g.,

pH and a

w

) and potential for pathogen growth in the product, choose representative dairy foods that reflect the diversity of the industry, and ensure that risks from raw milk and raw milk products do not affect the risk scores of other dairy products. The comment specifically recommends that we separate dairy products into three categories—cheese, ice cream, and milk—and further divide the cheese category into four subcategories: soft ripened cheese, semi-soft cheese, hard cheese, and other cheese. The comment also suggests that we amend the food facility registration classification scheme by adding a new category for yogurt and other fermented milks and cultured dairy products because of their unique intrinsic properties. Finally, the comment urges us to put raw milk and raw milk products in a stand-alone category named “Raw Milk for Consumption and Raw Milk Products.”

(Response 17) We do not believe it is necessary to make the revisions suggested by the comment. We agree that each of the dairy commodities has its unique food characteristics and manufacturing processes. In fact, the RRM-FT considers such unique characteristics and processes, as well as most of the dairy products suggested by the comment, in scoring each of the dairy commodities and associated commodity-hazard pairs.

The RRM-FT does not treat “Dairy” as one group but instead includes six separate commodity categories for dairy products (see Ref. 10, Table A-1), several of which contain multiple specific commodities (see Ref. 10, Table A-2). The Model identifies as separate commodities different types of cheeses (fresh cheese, soft-ripened cheese, and hard cheese) made from pasteurized milk. Furthermore, cheeses made from raw milk are classified into their own commodities separate from cheeses made from pasteurized milk. Ultimately the RRM-FT identifies and evaluates 21 individual dairy commodities (see Ref. 10, Table A-2).

The concerns expressed in the comment do not reflect the handling of the dairy commodity categories in the Model (Ref. 10). The RRM-FT uses data relevant to seven criteria for each commodity and associated commodity-hazard pairs to generate risk scores, taking into consideration the intrinsic characteristics of the food (such as the low pH of yogurt) in scoring Criterion 4 (Growth Potential, with Consideration of Shelf Life), among other data. The RRM-FT does consider “Dairy—Fermented dairy products other than cheese” as a stand-alone commodity category that includes two separate commodities (Yogurt and Cultured Products (excluding yogurt)) and associated commodity-hazard pairs. Amending the food facility registration scheme to add a new category for yogurt as the comment suggests is beyond the scope of this rulemaking. Additionally, while the RRM-FT does not include a raw milk commodity because FDA prohibits the sale of raw milk in interstate commerce, the RRM-FT evaluates raw milk in two separate commodities, one for hard cheeses made from unpasteurized milk and one for cheeses other than hard made from unpasteurized milk.

(Comment 18) One comment asserts that FDA did not include or consider costs of complying with the FTL traceability rule in Criterion 7 (Cost of Illness) of the RRM-FT and recommends that we include these costs.

(Response 18) The RRM-FT includes public health risk criteria as specified by FSMA section 204(d)(2)(A). Criterion 7 of the RRM-FT is defined as the cost of illness for the commodity-hazard pair; therefore, it is not appropriate to include in this criterion non-public health economic impacts such as the cost of complying with the rule. FDA considers the costs and benefits associated with the rule in the Final Regulatory Impact Analysis (FRIA) (Ref. 16).

(Comment 19) One comment requests clarification on how FDA will address changes in consumer habits. Specifically, for a food that is not on the FTL because FDA has determined that the food is rarely consumed raw, the comment requests clarification on whether covered entities are responsible for knowing that consumer habits have changed such that the product is no longer rarely consumed raw or if the FTL remains the same until FDA changes it. The comment also asks if we will indicate that we are planning to update the FTL due to changes in consumer habits.

(Response 19) The FTL will remain the same until we change it. The process for changing the FTL, which includes advance notice and an opportunity for the public to provide comment, is discussed in Section V.S of this document.

It is possible for a food to be part of a commodity that is on the FTL but to nonetheless be exempt under § 1.1305(e) of the final rule because it is listed as rarely consumed raw in § 112.2(a)(1) (21 CFR 112.2(a)(1)). For example, collards fall within the commodity “Leafy Greens,” but they are exempt from the subpart S requirements because they are listed as rarely consumed raw in § 112.2(a)(1). Because any changes to the rarely consumed raw list in § 112.2(a)(1) would have to be made through notice and comment rulemaking, firms would receive notice that the rarely consumed raw list might change and would have an opportunity to provide comments on the potential change.

(Comment 20) Some comments ask FDA to clarify the growing and production processes that were evaluated and used to place foods on the FTL. The comments also request that we clarify, if processes and practices change, how that type of information will be used to support inclusion or removal of foods from the FTL.

(Response 20) The growing and production processes that we evaluated and used to place foods on the FTL are described in the RRM-FT Methodological Approach Report (Ref. 10), specifically in section 3 of the report (“Identification of Food-Hazard Pairs”), where we describe the food classification scheme, and in the description of Criterion 5 (Manufacturing Process Contamination Probability and Industry-wide Intervention), which evaluates the possibility of hazard introduction

during manufacturing and the ability to control contamination with interventions through growing and production practices and processes throughout the supply chain. We will consider changes in industry processes and practices when we update the Model (see Response 488).

(Comment 21) Several comments ask that we make an interactive model tool available for stakeholders to test hypothetical changes to the scores for each criterion in the RRM-FT. Additionally, the comments ask that we make the data inputs and risk scores for all foods evaluated (not just those on the FTL) available to the public to increase transparency and help stakeholders with future business decisions. Comments also request that we provide the commodity category level analyses as well as the analyses for individual commodities in the commodity category. One comment that requests revisions to the RRM-FT further suggests that we conduct a pilot test with an interactive version of the revised RRM-FT to demonstrate to stakeholders how the scores are determined for the criteria and how that results in food being placed on the FTL. This comment suggests that stakeholders be given an opportunity to comment on the revised Model and the demonstration, which the comment maintains would give credibility to the Model and promote public acceptance.

(Response 21) We have already made public a substantial amount of information that allows stakeholders to analyze and interact with information relating to the RRM-FT, including testing hypothetical changes to the Model scores. For example, we provided a web-based tool (Ref. 17), the RRM-FT Methodological Approach Report (Ref. 10), and a full list of references for the data and information used in the Model (see link to references in Ref. 17). These materials provide the details of the methods on which the analyses are based (including examples) with all the information stakeholders need to reproduce such analyses. The tool also provides the total score for each of the commodities on the FTL as well as the criteria scores for the commodity-hazard pairs that make up each commodity on the FTL. In response to comments, we are considering making public the scores for all the foods evaluated in the Model, including those food/hazard pairs not included on the FTL. The Designation of the FTL Memorandum (Ref. 15) describes key aspects of how FDA uses the RRM-FT to designate the FTL.

With regard to the suggested pilot of the Model and additional opportunities for stakeholder comment, we have provided stakeholders with opportunities to comment throughout the development of the FTL. As previously stated, we published our draft approach for developing a risk-ranking model for public comment in 2014. We then refined the approach, taking into consideration the public comments received. Two separate external peer-review panels reviewed a draft model and the data used to generate risk scores with the Model, respectively. Concurrently with issuance of the proposed rule, we made available a revised model and updated the data, taking into consideration comments from the peer reviews. Additionally, we provided opportunities for stakeholders to obtain clarity on how the scores are determined for the criteria and which foods would be placed on the FTL during three public meetings. When we develop a new FTL in the future, we intend to publish a proposed updated FTL in the

Federal Register

for public input, review comments from the public, and publish a final updated FTL in the

Federal Register

. We believe this will provide stakeholders sufficient opportunity to provide input on any potential changes to the FTL.

(Comment 22) Several comments suggest that FDA use the RRM-FT to evaluate the risk of any new food, such as a multi-ingredient food that contains an ingredient on the FTL (FTL ingredient). The comments maintain that the dose-response curve should be considered in each instance and the risk of a multi-ingredient food that contains an FTL food may change depending on the ability of the relevant microbial pathogen(s) to survive and grow in the new food. The comments acknowledge practical challenges in a potentially enormous number of new foods that contain FTL ingredients that would each need to be evaluated. The comments suggest that, if FDA does not have the resources to evaluate all the new foods, it should apply a threshold to the amount of an FTL food that needs to be in a multi-ingredient food for the new food to be on the FTL, or help industry use the RRM-FT methodology to self-assess the risk of a new food to determine whether subpart S would apply.

(Response 22) We decline to use the RRM-FT to make individual evaluations of each multi-ingredient food that contains an FTL food. This would not be practical, nor is it necessary. Elsewhere in the final rule, we are providing additional clarity on which foods containing FTL foods as ingredients are on the FTL (see Response 27). For example, for a food that is specified on the FTL as being fresh or fresh-cut, if the nature of the FTL food has not changed in the new multi-ingredient food containing the FTL food as an ingredient (

e.g.,

bagged salad mix containing lettuce, smoothie containing fresh cantaloupe, sandwich containing fresh-cut tomato), the risk of the FTL food used as an ingredient in the new food is not expected to decrease. In fact, in some cases, the ability of bacterial pathogens to grow could be greater in the fresh FTL food when it is cut or sliced and included in the new multi-ingredient food.

With respect to the dose-response curve, we acknowledge there might be different levels of risk of illness when a different amount of an FTL food is consumed. However, there is no generalizable evidence with regard to risk of illness from a specific amount of the FTL foods that would enable us to set a threshold amount for FTL foods used as ingredients in other foods, as suggested by the comments.

(Comment 23) One comment maintains that in developing the RRM-FT, FDA should have ensured that risk managers agreed the Model criteria were relevant to the decision for designating the FTL. The comment maintains that FDA did not report work done in this area.

(Response 23) We disagree with the comment. FSMA section 204(d)(2)(A) establishes six factors for assessing risk of foods and designating the FTL that are represented by the criteria in the RRM-FT. The RRM-FT Methodological Approach Report (Ref. 10) describes the iterative process for developing the RRM-FT. This process included extensive and iterative consultations with an FDA Project Advisory Group, consisting of members from FDA's Center for Food Safety and Applied Nutrition, Office of Foods and Veterinary Medicine, Office of Food Policy and Response, Office of Policy, Legislation and International Affairs, Center for Veterinary Medicine, and Office of Regulatory Affairs, as well as the CDC (Ref. 10). The Project Advisory Group provided both technical and policy perspectives in the development of the Model. Furthermore, as discussed above in Response 2, during the development of the Model we consulted multiple external expert panels and considered comments and suggestions provided by peer reviewers.

(Comment 24) Several comments oppose using customer reviews as data for scoring in the RRM-FT. The comments voice concern with FDA's expressed interest in using artificial intelligence to mine non-traditional data

sources, specifically customer online reviews, as part of our efforts to gather additional data to support risk modeling and inspection prioritization. These comments do not believe customer online reviews will meaningfully contribute to data gathering.

(Response 24) The RRM-FT does not use customer reviews in scoring because the Model only includes data relevant to seven criteria based on the factors specified in section 204(d)(2)(A) of FSMA (Ref. 10), including the number of reported outbreaks and illnesses for commodity-hazard pairs. However, under FDA's New Era of Smarter Food Safety initiative, we will continue to explore ways to utilize non-traditional data sources and the use of artificial intelligence to protect the U.S. food supply. Additional information on this effort can be found in FDA's Blueprint for New Era of Smarter Food Safety (Ref. 18).

(Comment 25) Several comments assert that FDA does not appear to have considered comments they submitted on FDA's draft methodological approach in 2014. Specifically, the comments maintain that some issues they had submitted in 2014 remain not adequately addressed in the RRM-FT (2020 version), including the following claims: (1) the RRM-FT is not aligned with FSMA section 204(d)(2)(A) because it combines factors (iii) and (iv) into one criterion (Criterion 5—Manufacturing Process Contamination Probability and Industry-wide Intervention) and the Model's consumption criterion does not align with FSMA; (2) foods selected are not representative of the diversity of the dairy industry; (3) having two indicators for Criterion 4 (

i.e.,

using a scoring matrix of Growth Potential and Shelf Life) is problematic; (4) use of summing as an aggregation method (

i.e.,

summing risk scores for commodity-hazard pairs to calculate a risk score for the commodity) is not appropriate; and (5) the RRM-FT does not provide a cut-off score for foods on the FTL.

(Response 25) We considered each of these issues that were submitted in comments on the draft methodological approach in 2014 in the iterative process we used to develop and refine the RRM-FT. As previously stated, the iterative approach involved consulting with the RRM-FT Project Advisory Group and multiple external expert panels, and considering comments and suggestions provided by peer reviewers. As previously discussed, we have responded to these issues in this final rule (see Response 26 for discussion of the RRM-FT alignment with statutory factors in FSMA section 204(d)(2)(A); Response 17 for discussion of foods selected in the Dairy group; Response 16 for discussion of the indicators for Criterion 4; Response 12 for discussion of the aggregation method used for risk scores in the RRM-FT; and Response 14 for discussion of the cut-off score for foods on the FTL).

2. Designation of Foods on the FTL

a. General

(Comment 26) Some comments are supportive of the designation of the foods on the FTL. Conversely, other comments raise concerns with how we determine which foods are on the FTL and suggest our approach was not what Congress intended.

(Response 26) We appreciate the comments that are supportive of the FTL. In section 204(d)(1) of FSMA, Congress directed us to establish recordkeeping requirements for certain designated foods that would be additional to the traceability recordkeeping requirements in section 414 of the FD&C Act and the subpart J regulations. In section 204(d)(2) of FSMA, Congress directed us to consider specific factors in determining for which foods additional traceability recordkeeping requirements are needed. To determine which foods should be included on the FTL, we developed the RRM-FT based on the factors Congress identified in section 204(d)(2)(A) of FSMA. The Model considers FDA-regulated human foods, identifies commodities available for purchase at retail, and for each commodity identifies associated known or reasonably foreseeable hazards. The Model scores commodity-hazard pairs according to data and information relevant to the seven criteria described in the RRM-FT Methodological Approach Report (Ref. 10), which are based on the factors Congress identified in section 204(d)(2)(A) of FSMA. A commodity was included on the FTL if its risk score, aggregated across all associated hazards, was 330 or higher in the Model or if the evidence of outbreaks and illnesses and cost of illness scores for one or more associated commodity hazard pairs was “strong” (Ref. 15). This approach is science-based and reflects the intent of Congress in identifying the foods for which additional traceability records are necessary.

b. FTL Foods as Ingredients

(Comment 27) Some comments support our proposal to include on the FTL both foods specifically listed as well as foods that contain a listed food as an ingredient. However, many comments oppose this approach. Some comments claim that FDA exceeded its statutory authority by expanding the FTL beyond “particular” foods (as specified in section 204(d)(2)(A)(i), (ii), (v), and (vi) of FSMA). Some comments assert that the proposed approach would impose a burden on industry to identify every food that contains an FTL food as an ingredient without a corresponding public health benefit. Other comments maintain that this approach would lead to confusion and a lack of clarity for the food industry and increase the burden, particularly on retailers and distributors. One comment asserts that this approach would reduce consumption of produce because multi-ingredient foods would be formulated to avoid including foods on the FTL, such as certain produce items. Some comments provide examples of products for which we should not require additional recordkeeping for traceability, such as frozen pizza with cheese, granola bars with dried fruit, herbed bread, and quiches that use different types of peppers. Many comments ask that we exempt foods containing FTL foods as ingredients unless they are otherwise a listed food, such as a deli salad containing tomatoes, or to specifically list on the FTL certain multi-ingredient foods that should be covered under the final rule, such as bagged salads. Some comments recommend that the final rule apply only to foods on the FTL and foods containing listed foods as ingredients that will be consumed without a kill step.

(Response 27) We are clarifying our approach to the FTL in response to the comments. For several of the commodities on the FTL, we have clarified which version of the commodity is on the FTL and therefore covered by the final rule. For example, if a commodity is specified as “fresh” on the FTL, then only the fresh version of the commodity is covered by the final rule. If such a commodity is used in its fresh form as part of a multi-ingredient food, then the multi-ingredient food would be covered under the final rule. For example, fresh lettuce used in a bagged salad mix, fresh cantaloupe in a commercially prepared smoothie, or a sandwich containing a fresh tomato would be covered, but a frozen pizza with a spinach topping or trail mix with dried papaya would not be covered. We believe this approach is appropriate because the risk of the fresh FTL food would not be diminished just because it is used as an ingredient in a multi-ingredient food, if no kill step is applied or the FTL food is not otherwise changed, for example by drying or

freezing, such that it is no longer on the FTL. Further, the multi-ingredient food may be a key signal in an outbreak investigation that ultimately leads to identification of the contaminated ingredient. For example, we may receive a signal of fresh salsa in an outbreak investigation, and after further investigation be able to attribute the outbreak to the fresh tomatoes in the salsa. This example demonstrates not only why it is important to have the multi-ingredient food covered by the rule (because it is causing illness and serves as a key signal), but also why a commodity such as fresh salsa might not independently appear on the list if it is associated with outbreaks that are not attributed to it in our outbreak database because they are found to have been caused by an ingredient such as fresh tomatoes (see Response 15). Therefore, we believe it is appropriately protective of public health for the subpart S requirements to apply to multi-ingredient foods with FTL foods as ingredients, provided the FTL food remains in the same form (

e.g.,

“fresh”) that is specified on the FTL. We do not think Congress's use of the word “particular” in section 204(d)(2)(A)(i), (ii), (v), and (vi) of FSMA precludes this approach.

For foods on the FTL that are not designated as “fresh,” if those FTL foods are used as ingredients in a multi-ingredient food and no kill step is applied or the FTL food is not otherwise changed such that it is no longer on the FTL, then the multi-ingredient food would be covered by the final rule. For example, peanut butter in a sandwich cracker for which no kill step is applied (to either the peanut butter or the peanut butter sandwich cracker) will be covered by the rule. As discussed in Response 75, the commodities on the FTL related to finfish and seafood include both the fresh and frozen forms of those products. As such, freezing finfish or seafood would not be considered a change such that the food is no longer on the FTL, so frozen finfish or seafood would not be exempt from the subpart S requirements.

(Comment 28) One comment asserts that additional recordkeeping requirements are unnecessary for foods containing FTL foods as ingredients because processors already keep records under the preventive controls for human food regulation and the FSVP regulation, which require documentation of application of a kill step and verification of suppliers. In addition, the comment maintains that food companies still have to keep records for the immediate previous source and immediate subsequent recipient of the food under subpart J.

(Response 28) While many food companies are required to keep records under subpart J documenting the immediate previous source and immediate subsequent recipient of their food, FSMA directed FDA to develop a regulation requiring additional traceability records for certain foods beyond what FDA already requires under subpart J. We recognize that food processors also must keep records under other regulations, but many of those records are for purposes other than traceability. For records required under subpart S, § 1.1455(f) specifies that firms may use records kept for other purposes and do not have to duplicate records (see Section V.R.3 of this document). For example, we anticipate that many manufacturers/processors would be able to use records required under existing regulations, such as those requiring documentation of monitoring of a preventive control (see 21 CFR 117.190(a)(2)) or documentation of thermal processing of low-acid canned foods (LACF) (see 21 CFR 113.100), to meet the requirement in § 1.1305(d)(3)(ii) to document application of the kill step to a food.

(Comment 29) One comment requests that we exclude foods from the final rule for which the Harmonized Commodity Description and Coding System does not provide sufficient classification of the food because it would be too confusing, particularly for trading partners, to clearly identify the food on the FTL if there is not a corresponding code in that system. Another comment suggests that we use the Harmonized Commodity Description and Coding System to provide additional clarity on the foods on the FTL.

(Response 29) We decline the comment's suggestion to exempt from the final rule foods that are insufficiently classified under the Harmonized Commodity Description and Coding System. We believe the FTL issued with the final rule (Ref. 19) provides sufficient information for firms to know whether a particular food is on the FTL. While Harmonized Commodity Description and Coding System codes are typically used for tariff and not food safety purposes, we recognize that in some cases providing additional information on FTL foods using classification systems used by importers could be useful. We will explore ways to provide additional guidance for importers as needed regarding identification of foods on the FTL.

c. Changing the Form of an FTL Food

(Comment 30) Many comments request clarification on the version of the food that is covered by the proposed rule and whether a fresh version of an FTL food would be considered an ingredient in a dried or frozen version of the food and be covered, or if the dried or frozen version of the food would not be considered an FTL food. The comments note that the Model contains separate commodity designations for some frozen foods such as frozen fruits and frozen vegetables. If the dried or frozen version is covered by the rule, the comments ask for clarification on which KDEs would apply to the food. The comments maintain that including on the FTL these foods that have changed their form would result in coverage of numerous foods that do not present the same public health risk as listed foods and would increase the rule's economic and resource burden on covered entities.

(Response 30) We have clarified the FTL in response to the comments. For foods that are designated as “fresh” on the FTL, if the form of the food is no longer fresh and has been changed (

i.e.,

through freezing, drying, or another change in the form of the food), then the food would no longer be an FTL food. For example, frozen spinach, frozen cut mangoes, dried peppers, or dried herbs would not be covered by the rule if only the fresh form is listed on the FTL. The person changing the FTL food such that it is no longer on the FTL would need to maintain receiving records of the FTL food but would not be required to maintain subpart S records for its subsequent handling of the food (

e.g.,

transformation and shipping), and subsequent recipients of the food would not have to maintain records under the rule.

However, as discussed in Response 75, the commodities on the FTL related to finfish and seafood include both the fresh and frozen forms of those products. As such, freezing finfish or seafood would not be considered a change such that the food is no longer on the FTL, and frozen finfish and seafood are therefore covered by the final rule.

We believe our approach to this issue is appropriate because of how foods are categorized within the Model. For example, the Model includes several commodity designations that could include peppers (

e.g.,

peppers (fresh), frozen vegetables, dried vegetables), but it is the fresh peppers that had a risk score high enough to be included on the FTL. Frozen vegetables and dried vegetables did not have a risk score that

placed them on the FTL (see Response 26 for a description of the method by which foods on the FTL were determined).

d. Clarify Foods on the FTL

(Comment 31) Several comments express appreciation for the additional clarification FDA provided on the FTL on January 11, 2021, and request that we include those clarifications in the final rule. Many comments ask that we provide additional clarity and specificity in describing the foods on the FTL, maintaining that this would reduce confusion for the food industry and regulators.

(Response 31) As the comments note, we provided additional clarity regarding the foods on the FTL on January 11, 2021, in response to stakeholder input following the publication of the proposed rule. The FTL we are issuing with the publication of the final rule maintains those clarifications and provides additional clarifications and descriptions for the commodities on the FTL (Ref. 19). For some commodities, we have added examples of foods that are and are not considered part of that commodity designation on the FTL.

(Comment 32) Multiple comments request that we provide exhaustive lists of the foods for each commodity on the FTL and for commodities not on the FTL.

(Response 32) Considering the variety and range of food products for each commodity, it would be very challenging to provide an exhaustive list of foods for each commodity. As stated in Response 31, we have provided additional clarifications and descriptions for the commodities on the FTL, and for some commodities we have added examples of foods that are and are not considered part of that commodity designation on the FTL. We believe these clarifications and examples will help stakeholders better understand the foods under each commodity on the FTL.

(Comment 33) One comment asks where they can find the commodity risk scores mentioned in the proposed rule.

(Response 33) The risk scores for the commodities on the FTL are available in the RRM-FT Methodological Approach Report (Ref. 10).

(Comment 34) A few comments support the use of the term “Food Traceability List” to identify the list of foods that are covered by the rule. The comments note that the term is preferable to use of the term “high-risk list,” which could result in consumers avoiding certain foods such as fruits and vegetables due to public perception of the term “high-risk.” One comment argues that FDA must use the term “high-risk list” in the food traceability regulation to be consistent with the language and intent of FSMA.

(Response 34) While we acknowledge that section 204(d) of FSMA uses the phrase “high-risk foods,” we believe the term “Food Traceability List” is appropriate for the purposes of this rule. We agree with the concerns raised about potential negative consumer perceptions of a “high-risk list” and resulting efforts to avoid foods on the list. Furthermore, the FTL is based on specific concerns related to traceability and is not meant to encompass all possible risk factors associated with foods. To determine which foods should be included on the FTL, we developed the RRM-FT based on the factors that Congress identified in section 204(d)(2)(A) of FSMA. Those factors are specific to what Congress required under FSMA and may not reflect other approaches to assessing risk. Furthermore, in identifying foods for inclusion on the FTL, we focused on hazards for which improved traceability records would help protect the public health. For example, as discussed below (see Response 86), we concluded that enhanced traceability recordkeeping requirements would not greatly improve our ability to identify and respond to undeclared allergens in food. Therefore, although undeclared allergens pose a significant risk, we did not incorporate this risk into our decision of which foods to designate for the FTL. Consequently, to avoid unnecessary consumer concerns and confusion with other risk determinations, we conclude that it is appropriate to use the term “Food Traceability List” rather than “High-Risk Foods List.”

e. Foods vs. Commodities

(Comment 35) Several comments claim that FSMA required FDA to designate “particular foods” for the FTL rather than commodities. The comments maintain that some foods within certain commodities, if scored separately, would not have sufficient risk scores to be listed on the FTL. One comment argues that grouping foods into commodities does not accurately capture the risk of individual foods. Some comments assert that the boundaries of the commodities on the FTL are not clearly defined, which could result in confusion and ambiguity for some parts of the industry. These comments maintain that submitting questions through the FDA Technical Assistance Network (TAN) to inquire about coverage of specific foods is complicated and not timely.

(Response 35) We interpret the term “particular food” in section 204(d)(2)(A)(i), (ii), (v), and (vi) of FSMA in a way that is reasonable and consistent with section 204(d), and that accurately reflects the specificity of data available to us in developing the FTL. As discussed in Response 7, data on individual foods, such as specific varietals, is sparse and inconsistent across the variety of foods in the Model and on the FTL. For the purposes of the FTL, we determined that the appropriate level of granularity is at the level of “commodity,”

e.g.,

“tomatoes (fresh)” rather than “Roma tomatoes” or “cherry tomatoes.” Food items within the same “commodity” designation generally have similar characteristics, associated hazards, and production and supply chain practices and conditions. Further, data used to assess components of the Model (

e.g.,

outbreak and illness data, likelihood of contamination, degree to which product supports growth, consumption, annual cost of illness) are available and adequate at the “commodity” level of granularity. See also Response 68 for a discussion on the scope of the seafood commodity categories.

As stated in Response 31, we have provided additional clarifications and descriptions for the commodities on the FTL, and for some commodities we have added examples of foods that are or are not considered part of that commodity designation on the FTL. We believe these clarifications and examples will help stakeholders better understand the foods under each commodity on the FTL. As part of our outreach to stakeholders regarding the final rule (see Section V.U.4 of this document), we will continue to use the TAN to provide timely responses to questions about the FTL and the subpart S requirements, recognizing that some answers may take longer depending on the nature of the question.

(Comment 36) One comment argues that listing commodities would make it more difficult to remove foods from the FTL because new food safety technologies are typically applied to individual foods rather than commodities as a group.

(Response 36) As discussed in Section V.T.1 of this document, we plan to periodically conduct a review to determine whether it is appropriate to revise the FTL in accordance with the procedures set forth in § 1.1465 of the final rule. While there are several factors that we must consider in determining which foods are on the FTL, changes in industry practice, such as the use of new food safety technologies, may result in a sufficient change in the risk score of a commodity such that it would no longer be on the FTL.

We encourage the development and adoption of new food safety technologies to improve the safety of specific foods. If a company develops a new food safety technology which they believe provides an additional level of food safety for the food they produce, that company might consider submitting a citizen petition requesting modified requirements or an exemption from subpart S for certain products based on use of that technology, using the procedure set forth in § 1.1370 (see Section V.P of this document). We note that if new technologies provide a “kill step” to FTL foods, the food might be exempt from subpart S under § 1.1305(d) of the final rule.

f. Add Foods to the FTL

(Comment 37) Several comments suggest additions to the FTL. A few comments suggest the FTL should be expanded to include all foods or all foods that have caused foodborne illness. A few comments suggest expanding the FTL to include all produce and all seafood. One comment suggests expanding the FTL to include additional foods associated with outbreaks, such as dried and frozen fruits, tahini, pistachios, hazelnuts, and flour.

(Response 37) We decline to make these changes to the FTL. Congress explicitly directed us to establish additional recordkeeping requirements for traceability for foods that meet certain risk-based criteria. To determine which foods should be included on the FTL, we developed the RRM-FT based on the factors that Congress identified in section 204(d)(2)(A) of FSMA. The Model scores commodity-hazard pairs according to data and information relevant to seven criteria described in the RRM-FT Methodological Approach Report (Ref. 10). A commodity was included on the FTL if its risk score, aggregated across all associated hazards, was 330 or higher in the Model or if the evidence of outbreaks and illnesses and cost of illness scores for one or more associated commodity hazard pairs was “strong” (Ref. 15). If the foods suggested by the comments are not on the FTL, it is because their risk scores were not high enough to warrant inclusion on the FTL. As noted elsewhere, we intend to revise the FTL on a regular basis based on updates of the data in the Model. If the risk scores for foods (including those specified in the comments) change, those foods could be added to the FTL in a subsequent update to the list.

We recognize that there are foods that have been linked to past outbreaks but that are not on the FTL. Future outbreaks might also occur among foods not on the FTL. No food is completely risk-free, and we encourage all supply chain members to have systems and procedures in place to enable them to rapidly and effectively engage in traceback and traceforward activities for all of their foods, including those not on the FTL. However, Congress made clear that the additional recordkeeping requirements established by this rulemaking should only apply to foods that FDA designated for inclusion on the FTL, and that these requirements should have no effect on foods that are not so designated (see section 204(d)(7) of FSMA).

g. The FTL and the High-Risk Designation

(Comment 38) One comment requests that we not use the FTL for purposes other than the traceability recordkeeping requirements, such as establishing inspection frequencies or setting performance standards. The comment asserts that “high-risk” is defined differently depending on its context or use.

(Response 38) We agree that “high-risk” is defined differently depending on its context or use. Congress directed us to consider specific factors in determining which foods should have additional recordkeeping requirements for traceability. Those factors were specific to section 204(d) of FSMA. Section 201 of FSMA, which is codified as section 421 of the FD&C Act (21 U.S.C. 350j), directs FDA to consider a different set of factors to identify high-risk facilities for the purpose of determining the frequency of domestic inspections. Performance standards can be used in a wide range of settings, and any risk determination used for a performance standard would have to be appropriate to that context.

h. Description of Foods on the FTL

(Comment 39) One comment requests that we provide the scientific name of plants and animals on the FTL. Another comment requests that we use the naming conventions of the Codex Alimentarius or the Code of Federal Regulations in identifying foods on the FTL.

(Response 39) We decline these requests. The foods identified on the FTL were based, in part, on data from FDA's RFR and facility registration systems, which have existing naming conventions within FDA systems. Further, FDA typically uses the common name of plants and animals in its documents to help ensure that all stakeholders have an understanding of the foods to which regulations or guidance apply. Regarding requests to use other naming conventions, such as those in the Codex Alimentarius or the Code of Federal Regulations, those naming conventions were not developed for traceability, nor do they necessarily conform to FDA's typical naming conventions.

i. Produce

(Comment 40) Several comments ask for clarifications on the types of melons that would be covered in the “melon” category and how melons were deemed to be high-risk foods. The comments also request that whole fresh watermelon be excluded from the FTL.

(Response 40) In the melon category, the FTL includes all types of fresh melons. Examples include, but are not limited to, cantaloupe, honeydew, muskmelon, winter melon, bitter melon, and watermelon. As previously stated, a commodity was included on the FTL if its risk score, aggregated across all associated hazards, was 330 or higher in the Model, or if the evidence of outbreaks and illnesses and cost of illness scores for one or more associated commodity hazard pairs was “strong.” Based on the seven criteria used in the Model and the data we have for melons, this commodity has a risk score that warrants its inclusion on the FTL. Response 26 provides a description of the method by which foods, including melons, on the FTL were determined, while Response 6 discusses why the list uses commodity groupings (such as melons) rather than individual foods (such as watermelons).

(Comment 41) Several comments ask for clarification on how tropical fruits were determined to be in the tropical tree fruit category and whether certain fruits like bananas, avocado, and citrus are in that category.

(Response 41) The RRM-FT Methodological Approach Report (Ref. 10) describes the classification of food commodities, including tropical tree fruits. The tropical tree fruit designation allows for a grouping of similar tree fruits, not other tropical fruit, that are typical to locations that are hot and humid and whose longer day lengths allow for fruit maturity. Examples of tropical tree fruits include (but are not limited to) mango, papaya, mamey, guava, lychee, jackfruit, and starfruit. Tropical tree fruits do not include non-tree fruits (such as bananas, pineapple, dates, soursop, jujube, passionfruit, loquat, pomegranate, sapodilla, and figs); tree nuts (such as coconut); pit fruit (such as avocado); or citrus (such as orange, clementine, tangerine, mandarins, lemon, lime, citron, grapefruit, kumquat, and pomelo). However, derivatives or components of

some of the fruits that are not considered tropical tree fruits may be on the FTL in other commodity categories, such as coconut butter in the nut butter category, as discussed in this document.

(Comment 42) Several comments ask whether the “Tropical Tree Fruits (fresh)” category is limited to high-risk tree fruits and includes other tropical tree fruit products that have undergone processing but not a validated kill step, such as guava paste.

(Response 42) The “Tropical Tree Fruits (fresh)” commodity is one of two dozen commodities we identify in the commodity category “Produce—RAC (raw agricultural commodity)” based on the consideration of the characteristics of the foods and production and supply chain practices and conditions. The RRM-FT evaluates several commodities for fresh fruits, including Tropical Tree Fruits (

e.g.,

papaya), Tropical Fruits NEC. (

e.g.,

banana), Citrus (

e.g.,

orange), Pome Fruits (

e.g.,

apple), and Pit Fruits (

e.g.,

avocado), and finds that only the Tropical Tree Fruits commodity has a high enough risk score to meet the threshold for inclusion on the FTL. Therefore, the FTL includes fresh tropical tree fruits but does not include other fresh tropical fruits. Fresh guava is covered under the “Tropical Tree Fruits (fresh)” commodity. If fresh guava is used as an ingredient in guava paste, the guava paste would also be included on the FLT. However, if the guava paste is subjected to a kill step, the exemption language in § 1.1305(d) would apply.

(Comment 43) Several comments request that we clarify the scope and definition of leafy greens that are on the FTL. Some comments also suggest that the FTL align with the Leafy Greens Marketing Association (LGMA) definition of leafy greens.

(Response 43) We have provided additional clarification to the description of the commodity “Leafy Greens (fresh)” on the FTL, specifying that it includes all types of fresh leafy greens (Ref. 19). Examples include, but are not limited to, arugula, baby leaf, butter lettuce, chard, chicory, endive, escarole, green leaf, iceberg lettuce, kale, red leaf, pak choi, Romaine, sorrel, spinach, and watercress. The “Leafy Greens (fresh)” category does not include whole head cabbages such as green cabbage, red cabbage, and savoy cabbage, nor does it include banana leaf, grape leaf, and leaves that grow on trees. Also note that fresh leafy greens listed as rarely consumed raw in § 112.2(a)(1), such as collards, are exempt from the requirements of subpart S under § 1.1305(e) of the final rule.

We believe the description of “Leafy Greens (fresh)” that is on the FTL is generally aligned with the LGMA list of leafy greens. However, we acknowledge that there are some differences. The LGMA list includes whole head cabbages, which are not on the FTL, and spring mix, which is not part of the “Leafy Greens (fresh)” category on the FTL (but which is nonetheless on the FTL as part of the commodity “Leafy Greens (fresh-cut)”). The FTL description of “Leafy Greens (fresh)” includes some leafy greens that are not on the LGMA list, such as chicory, watercress, pak choi, and sorrel.

(Comment 44) A few comments request that collards be removed from the proposed FTL as they are listed in the produce safety regulation (in § 112.2(a)(1)) as rarely consumed raw.

(Response 44) Collards are exempt from the subpart S requirements under § 1.1305(e) of the final rule because they are currently listed as rarely consumed raw in § 112.2(a)(1). Otherwise, collards would be subject to subpart S because they are part of the leafy greens commodity category. To avoid confusion, we have removed collards from the list of examples of leafy greens on the FTL.

(Comment 45) One comment requests that we individually list, with the applicable plant part(s), every fruit, vegetable, and culinary herb that is subject to the rule, or expand the language in each category to fully describe the intended subjects, including information such as the species name(s), the plant part(s), the botanical characteristics (

e.g.,

whether the plant grows on the ground vs. a tree or a climbing vine) and other information as appropriate to provide clear and accurate descriptions.

(Response 45) We do not agree that this level of detail is necessary. Furthermore, adding botanical names could inadvertently include or exclude commodities not intended to be on or off the FTL. However, the revised FTL (Ref. 19) points out differences when necessary, such as between beet root and beet greens, as well as dill leaves and dill seed. The revised FTL also includes additional examples of foods on the FTL.

(Comment 46) Some comments ask that we confirm that “frozen” and “fresh-frozen” vegetables are not included on the FTL.

(Response 46) Vegetables that are sold as “frozen” or “fresh-frozen” are not included on the FTL because this product category was analyzed separately from vegetables that are sold in other forms (

e.g.,

fresh, dried), and frozen/fresh-frozen vegetables did not meet the scoring criteria for inclusion on the FTL.

(Comment 47) One comment agrees with FDA that whole apples, pears, cherries, and fresh berries should not be on the FTL.

(Response 47) Whole apples, pears, cherries, and fresh berries did not have risk scores high enough to be included on the FTL and therefore are not covered by the final rule.

(Comment 48) Several comments request that we limit the FTL to sprouts, fresh produce, and/or high-risk herbs like cilantro with risk scores above the cutoff threshold of 330, and then phase in other foods as part of subsequent FTL updates. The comments maintain that this would allow FDA to “test” its traceability approach in the final rule, especially since some sectors of the produce industry have experience with traceability via participation in private traceability initiatives.

(Response 48) We decline to adopt the phased-in approach suggested by the comments. Congress directed FDA to identify foods for which additional recordkeeping requirements for traceability are necessary to protect the public health. Limiting the foods on the FTL to a subset of the commodities that had risk scores that merited inclusion on the list would not be based in science and would reduce the public health protections anticipated for the food traceability regulation.

(Comment 49) A comment suggests that we clarify whether fresh-cut produce that is “rarely consumed raw” under the produce safety regulation falls under the subpart S requirements for fresh-cut produce. One comment suggests that we provide more clarity about which fresh-cut produce is included on the FTL, and additional clarity on the methodology used to reach these conclusions.

(Response 49) Produce that is “rarely consumed raw” according to the produce safety regulation (§ 112.2(a)(1)) is exempt from the subpart S regulations under § 1.1305(e) for the entirety of the supply chain, regardless of whether it is fresh-cut. For example, although all fresh-cut fruits and vegetables are on the FTL, a fresh-cut “rarely consumed raw” vegetable such as fresh diced butternut squash would be exempt under § 1.1305(e) because the fact that the butternut squash is fresh-cut does not change its status as “rarely consumed raw.”

(Comment 50) Some comments suggest that we reevaluate coverage of mung bean sprouts under the FTL. These comments maintain that mung bean sprouts should be considered rarely consumed raw and assert that few food safety issues have been linked to mung bean sprouts and mung beans.

The comments also ask us to reevaluate mung bean sprout consumption data using more recent datasets.

(Response 50) Fresh mung bean sprouts, as well as other types of fresh sprouts, are covered by the produce safety regulation and are not considered to be “rarely consumed raw” under § 112.2(a)(1). Section 112.2(a)(1) codifies an exhaustive list of all produce that is considered “rarely consumed raw,” and revising that list is outside the scope of this rulemaking. The commodity risk scores for fresh sprouts, including mung bean sprouts, qualified this commodity for inclusion on the FTL, as it has associated commodity-hazard pairs with criteria scores in the moderate to strong range (Ref. 15, Table 1 and Appendix I). We further note that, according to the FDA CORE Outbreak Dataset (Ref. 11), between 1999-2019 there were eight documented outbreaks related to consumption of mung bean sprouts, resulting in 319 illnesses and at least 2 deaths.

j. Herbs and Spices

(Comment 51) One comment asks that we clarify that it is the fresh version of herbs that are on the FTL and not the dried form (

i.e.,

spices). The comment further maintains that tomatoes and peppers that are dried or will be dried for spices or seasonings should not be included on the FTL. The comment also asks for clarification on whether capsicum annum pepper, if grown to become a spice, would be covered by the rule. Another comment asserts that herbs that are destined to be dried should not be covered by the rule because those herbs are grown, processed, and consumed differently than fresh herbs. Another comment recommends that spices, seasonings, and flavorings not be included on the FTL. Another comment states that it understands that dried herbs and spices are not covered by the rule because they are a separate commodity in the Model and are not on the FTL.

(Response 51) In the additional information on the FTL that we provided on January 11, 2021, we noted that the form of herbs on the FTL is the fresh form. Spices, seasonings, and flavorings are not included on the FTL and therefore are not covered by the final rule. In Response 30, we provide additional clarity regarding foods on the FTL that are designated as “fresh.” Section 1.1305(d)(4) and (d)(5) of the final rule (see Section V.E.5 of this document) provide further clarification that if a food is changed such that it is no longer on the FTL, then the food would not be covered. Therefore, dried herbs, dried tomatoes, and dried peppers would not be covered by the final rule because the FTL only includes the fresh versions of those foods.

In addition, under § 1.1305(d)(6), if an FTL food is destined to be changed (

e.g.,

through freezing, drying, or another change in form of the food) such that it is no longer on the FTL, then that food would not be covered from the point at which it is known that the FTL food is destined to be changed, provided that the entities have a written agreement as described in Response 196.

Regarding the capsicum annum pepper, if the peppers are destined to be dried for spices and the pepper shipper has a written agreement with the receiver that the peppers will be dried, then, as noted above, the shipper and receiver of the pepper would not be required to keep subpart S records for the food. However, if the pepper shipper does not have a written agreement, the shipper would need to maintain the relevant subpart S records.

(Comment 52) Comments request that we provide more clarity regarding the specific part of the herb plant that is covered under the FTL.

(Response 52) For fresh herbs, any part of the herb that is fresh and sold for human consumption would be covered under the FTL.

(Comment 53) One comment asks that we limit the FTL to fresh culinary herbs rather than all herbs.

(Response 53) As discussed in Response 51, we have clarified that the form of herbs on the FTL is the fresh form. We believe that further clarification and distinction as “culinary” herbs is not necessary. The “Herbs (fresh)” commodity is one of two dozen commodities we identify in the commodity category “Produce—RAC” based on the consideration of the characteristics of the foods and production and supply chain practices and conditions. The Model scores the commodity-hazard pairs at the commodity level

(e.g.,

all fresh herbs) regardless of the purpose of use because we are not aware of scientific evidence that fresh produce within the same commodity does not share a similarity in the characteristics of the food and in how they are produced. Furthermore, we are not sure how the phrase “culinary herbs” would be defined. In the Model, the “Herbs (fresh)” commodity has criteria scores high enough to meet the threshold for inclusion on the FTL.

k. Deli Salads

(Comment 54) Several comments assert that “deli salad” is a vague term that has different meanings in some sectors of the food industry, and other comments request that we clarify how we interpret the deli salad category for the RRM-FT. Some comments ask that we specify whether an “antipasti” salad would be considered a deli salad.

(Response 54) The ready-to-eat (RTE) deli salads commodity in the RRM-FT includes prepared refrigerated and RTE deli salads (

e.g.,

potato salad, egg salad, pasta salad, seafood salad). While the term “deli salad” appears to be a broad term, it is intended to capture multiple types of RTE deli salads, including the aforementioned examples as well as a prepared antipasti salad. However, a prepared, RTE antipasti salad could include meat as an ingredient, which may place it under the jurisdiction of USDA and therefore make it exempt from the requirements of subpart S under § 1.1305(g).

(Comment 55) Several comments request exemption of deli salads from the subpart S requirements. Some comments assert that RTE deli salads like pasta and potato salad that are processed and prepared using hurdle technology or other controls to minimize pathogen growth should not be included on the FTL. Similarly, other comments assert that these types of RTE salads that are processed and prepared using controls such as pH and preservatives (

e.g.,

antimicrobials and

Listeria

inhibitors) do not pose the same risk as RTE salads that do not use the hurdle approach.

(Response 55) While we acknowledge that the use of preservatives and antimicrobials in deli salads helps to minimize bacterial growth, the data provided in the comments do not change how we score deli salads in the RRM-FT. The hurdle approach, as opposed to a kill step, can vary widely in terms of procedure and is not consistently applied throughout industry.

Therefore, based on the available data, we conclude it is not appropriate to grant a blanket exemption for deli salads processed using hurdle technology or related procedures.

l. Nut Butters

(Comment 56) Some comments ask us to include all butters (nut, soy, and seed) on the FTL that are considered allergenic. Other comments question why soy and seed butters in general were not included on the FTL. These comments assert that soy and seed butters have similar manufacturing processes and supply chain standards, and thus pose the same risk as nut butters. Additionally, some comments assert that consumption patterns might be shifting from peanut butter to seed butter due to allergies.

(Response 56) We decline to include all butters considered allergenic or all soy and seed butters on the FTL. As previously stated, we developed a risk-ranking model for food tracing based on the factors in section 204(d)(2)(A) of FSMA. A commodity was included on the FTL if its risk score, aggregated across all associated hazards, was 330 or higher in the Model, or if the evidence of outbreaks and illnesses and cost of illness scores for one or more associated commodity hazard pairs was “strong.” Using the RRM-FT, we evaluated nut butters (

e.g.,

made from tree nuts and peanuts) and soy and seed butters (

e.g.,

made from edible seeds) as separate commodities and found that only the nut butters had a risk score high enough to meet the threshold for inclusion on the FTL. Therefore, only nut butters are covered by the rule. As previously stated, we will periodically review data and information relevant to the RRM-FT criteria for commodity-hazard pairs, including the consideration of consumption patterns and food safety improvements across commodities.

The inclusion of nut butters on the FTL does not relate to the fact that nut butters can be allergenic. See Response 86 for a discussion of how we assessed the risks that are related to allergens.

(Comment 57) Several comments request clarification on whether nut butters made with raw nuts pose the same level of risk as nuts that are roasted, even when applying a process control during the roasting process that results in a 4- to 5-log reduction of the pertinent pathogen.

(Response 57) We acknowledge that adequate process controls resulting in a 4- to 5-log reduction in the pertinent pathogen should minimize the risk associated with nuts. However, it is the nut butter, not the nuts, that is on the FTL and covered by the final rule. The nut butters commodity, regardless of whether the ingredient nuts were raw or roasted, ranked high in the RRM-FT, which is why nut butters are included on the FTL. While applying a validated roasting process control for peanuts may mitigate the associated hazard, we continue to see multiple outbreaks associated with recontamination of peanuts and peanut butter after the roasting step. We also know from previous FDA investigations that there are sources of environmental pathogens (

e.g., Salmonella

spp.,

L. monocytogenes

) in facilities, and routes of contamination for these pathogens into the nut butters have been associated with employee practices, insanitary conditions, and inadequate sanitation practices. Using roasted nuts that have undergone a properly designed and implemented process control should mitigate the hazard associated with this ingredient; however, it does not reduce the risk of the potentially significant hazards posed by the exposed nut butters in the post-processing environment.

(Comment 58) Several comments ask whether nut meals and powders, nut flours, nut flavoring extracts, and similar commodities are on the FTL. Some comments request that we clarify whether peanut butter chips fall under the nut butter category on the FTL. Some comments assert that peanut butter chips should not be considered nut butters but should be a separate commodity that is exempt from the rule.

(Response 58) “Nut meals and powders,” “Flours (wheat, rice or soy),” and “Flavorings” are all separate commodity designations from the “nut butters” designation. These commodities were assessed separately in the RRM-FT and did not have risk scores that would include them on the FTL.

Peanut butter chips are not in the “nut butters” commodity. However, if peanut butter chips are produced using peanut butter as an ingredient, they are covered by the rule because they contain an ingredient on the FTL (peanut butter). However, if a kill step is applied to the peanut butter chips, the exemption in § 1.1305(d) would apply.

(Comment 59) Some comments request that we clarify whether “coconut butter” and “Chinese chestnut butter” are covered by the rule under the nut butter category. The comments maintain that “coconut” qualifies as a “tree nut” for purposes of the Food Allergen Labeling and Consumer Protection Act of 2004, but that in many countries it is not considered a “tree nut” because it does not meet common definitions of “nut,” nor does it grow on “trees.” The comments suggest that if we intend “nut butter” to include coconut butter, we should say so explicitly in the FTL and have data appropriate to deem coconut nut butter a “high-risk food.”

(Response 59) As discussed in Response 39, we use data from FDA's RFR and facility registration systems to help determine commodity designations for the FTL. Based on those classification systems, we consider coconut to be a nut; therefore, coconut butter is included on the FTL as a nut butter. This is consistent with 21 CFR 170.3, which also classifies coconut as a nut. We consider Chinese chestnut to be a tree nut and, therefore, Chinese chestnut butter also is an FTL food subject to the subpart S requirements. We have added both coconut butter and chestnut butter to the FTL as examples of “nut butters” to clarify that they are included in this category. See the RRM-FT results tool (Ref. 17) for information about risks associated with nut butters.

(Comment 60) One comment expresses support for the fact that almonds/tree nuts are not

This text is long and has been trimmed here. Open the source document for the complete record.