SEARCH

SEARCH BY CITATION

Keywords:

  • FSMA;
  • product tracing;
  • public policy;
  • traceability pilots

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodology
  5. Results and Discussion
  6. Recommendations
  7. Conclusion
  8. References

In September 2011, the U.S. Food and Drug Administration (FDA) asked the Institute of Food Technologists (IFT) to execute product tracing pilot projects as described in Section 204 of the FDA Food Safety Modernization Act (FSMA). IFT collaborated with representatives from more than 100 organizations—including the U.S. Dept. of Agriculture, state departments of agriculture and public health, industry, and consumer groups, as well as not-for-profit organizations—to implement the pilots. The objectives of the pilot projects were 1) to identify and gather information on methods to improve product tracing of foods in the supply chain and 2) to explore and evaluate methods to rapidly and effectively identify the recipient of 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 or misbranded. IFT conducted evaluations to determine the impact of currently available technologies, types of data and formats, and the data acquisition process, as well as the use of technology on the ability to follow product movement through the supply chain. Results from the pilots found inconsistencies in the terminology, numbering systems, formatting, legibility, and occasionally the language that sometimes required IFT to contact the submitting firm to gain clarity, thus increasing the time required to capture data before any meaningful analysis could begin. However, the pilot participants appeared to have many of the tools and processes in place which are required to allow the capture and communication of critical track and trace information (such as, key data elements) at critical points of product transfer and transformation (such as, critical tracking events). IFT determined that costs associated with implementing a product tracing system can vary widely as determined by numerous factors: the size of the firm/facility, the method of product tracing already in use (manual or electronic), and the range of each firm's capabilities to implement or improve its product tracing system, to name a few. IFT found that there are several areas (such as uniformity and standardization, improved recordkeeping, enhanced planning and preparedness, better coordination and communication, and the use of technology) in which industry improvements and enhancements to FDA's processes would enable tracebacks and traceforwards to occur more rapidly. IFT developed 10 recommendations for FDA to consider for improving the state of system-wide food product tracing. The recommendations outlined in the report will enable FDA to conduct more rapid and effective investigations during foodborne illness outbreaks and other product tracing investigations, thus significantly enhancing protection of public health.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodology
  5. Results and Discussion
  6. Recommendations
  7. Conclusion
  8. References

In September 2011, the U.S. Food and Drug Administration (FDA) charged the Institute of Food Technologists (IFT) to coordinate and conduct the product tracing pilots required by Section 204 of the FDA Food Safety Modernization Act (FSMA), including an evaluation of costs and benefits to industry and consumers.

Representatives from more than 100 organizations, including state departments of agriculture and public health, the U.S. Dept. of Agriculture (USDA) Agricultural Marketing Service (AMS) and Food Safety Inspection Service (FSIS), industry trade associations, not-for-profit organizations, consumer groups, technology solution providers, and a diverse cross section of the food industry including supply chain partners from farm to point of sale/service as well as large and small firms, collaborated with IFT to execute product tracing pilots for 3 ingredients (chicken, peanuts, and crushed red pepper) used in the production of 4 multi-ingredient processed food products (2 dry and 1 frozen Kung Pao chicken products, and peanut butter) as well as tomatoes (both whole and sliced).

In the continuum of an outbreak—from the time a person becomes ill to the time that the product has been removed from the distribution system—there are several points in the product tracing and recall processes where improvements can have positive and meaningful impacts on public health. This task primarily focused on traceback investigations. Tracebacks can occur when one or more foods (including ingredients) are suspected of being a potential health risk and there is a need to determine the path of a product through the supply chain. A traceback investigation generally involves documenting the distribution paths of products from several locations to determine if there is a common point of convergence in the supply chain, for example, a common date and location of harvest or place of manufacture. Determination of a convergence point is critical to the next step in conducting a source investigation to determine how the contamination occurred in order to prevent future illnesses. A traceforward investigation, explored in this task, but to a lesser extent than traceback investigations, follows the distribution path of a product from the point of convergence toward its point of consumption, including through manufacturing, distribution, retail, and foodservice. During a traceback investigation, the key question is “What do these products have in common: a lot number, common date at the same location, and so on?” When that information is known, the key question in a traceforward is “Where did these specific products (defined by lot numbers, production dates, and so on) go?” These investigations often occur after some or all of the products have exited the supply chain, thus the investigations are heavily dependent on residual records.

Tracebacks and traceforwards rely primarily on record-keeping. Current record-keeping requirements stem, in part, from the Public Health Security and Bioterrorism Preparedness and Response Act of 2002 (the BT Act; U.S. Congress 2002). Requirements based on the BT Act include having firms know from whom they received products and to whom they were sent, commonly referred to as one up—one down tracing; however, some supply chain members, such as restaurants and farms, are exempt. The specific types of information required to be kept is dependent on the role of the firm in the supply chain. When a product is transformed, the regulations resulting from the BT Act state that lot numbers, if available, be used to link incoming ingredients to outgoing products (U.S. Food and Drug Administration 2004).

Methodology

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodology
  5. Results and Discussion
  6. Recommendations
  7. Conclusion
  8. References

IFT solicited industry, government, consumer advocates, and other stakeholders for input on the selection of food products and participation within the 2 pilots. IFT disseminated a request for formal input and publicized the 3 stakeholder input sessions through a variety of outreach mechanisms, including posting the request on the IFT website, emailing all individuals who had previously expressed interest in IFT's product tracing work (approximately 700 contacts), and using social media outlets. Nearly 70 organizations, including 3rd-party technology providers, food industry representatives, trade associations, consumer groups, academicians, and others responded, either in writing or at 1 of 3 public stakeholder input sessions. Later, IFT advertised the opportunity for technology providers to serve as “collaboration platforms.” Recognizing the multitude of technologies available to assist firms or regulators in tracing products, IFT also solicited input regarding additional technologies in existence or in development that enhance the ability to track and trace foods.

In order to identify and quantify product tracing improvement opportunities, it was necessary to develop a clear sense of the current processes used during traceback investigations and document any obstacles (and enablers) to effective product tracing. IFT spoke to numerous state-level traceback investigators, epidemiologists, and representatives from USDA-FSIS and FDA. Each individual shared thoughts and experiences and several themes emerged. It was clear that the amount of epidemiological information, and the confidence in that information, played a notable role in the distinction between “easy” and “difficult” tracebacks. To meet the requirements of FSMA, IFT did not implement any dramatic changes within firms participating in the pilots (such as, installation of new technologies) but instead evaluated what the current capabilities are within the firms and which technologies are being used.

IFT considered stakeholder input and the requirements of FSMA when presenting FDA with recommendations for the types of foods that would be good candidates for the pilot projects. Ultimately, FDA determined that IFT should evaluate the tomato supply chain in the produce pilot. Ingredients were a key focus for the processed food pilot, and given the range of recent outbreaks and recalls associated with nuts (including peanuts) and spices, FDA tasked IFT with conducting the pilot with products that contained these ingredients. Further, FSMA requires FDA to collaborate with USDA-FSIS. FDA asked IFT to determine the feasibility of working with several food ingredients, including chicken, in the pilots. Frozen or dried Kung Pao chicken (containing one or more of the following ingredients: peanuts, spices, and chicken) was identified by FDA as the best candidate for the processed food ingredients pilot. While efforts were underway to identify participants for the pilot studies, IFT was approached by a peanut butter manufacturer who sells both private-label and branded peanut butter. FDA agreed that this product was also a suitable candidate for the processed food ingredient pilot.

Two teams of individuals, including industry experts and state-level traceback investigators, and USDA-FSIS in the case of the processed food ingredient pilot, worked together to conceptualize the types of situations that would prompt a traceback or traceforward inquiry. Given the potential for a brand and label to be associated with processed food products at retail, the team working on that pilot determined that the 4 pilot scenarios should be constructed to vary the nature of the information provided at the beginning of the mock traceback. Accordingly, the timeframes for which records were requested also varied (ranging from asking for information on a specific lot to products produced during the course of a 10-mo timeframe).

Given the multitude of participants in the tomato pilot, and the known diversity in product tracing practices and processes within similar portions of the supply chain, it was determined that each of the 12 scenarios executed through the mock tracebacks should be based upon a similar “story.” Depending on the exact scenario, participants at the retail and foodservice points in the supply chain (where most of the requests began) were generally asked for records covering a 1- to 5-wk timeframe. Eight of the scenarios were conducted as multistep tracebacks, beginning with restaurants or retail outlets, and following the paths backwards through the supply chain, based on the pre-existing relationships between trading partners. However, there were 4 participants in the tomato pilots who were not linked to other pilot participants. These firms—2 re-packers and 2 wholesalers—were each asked to trace one shipment of tomatoes forwards and backwards within their own operations. While these 4 scenarios were not like typical tracebacks, they did allow IFT to assess the technologies and processes used by these firms, of which some were small businesses.

IFT developed summary response templates based on previous work, which expanded on the concepts for critical tracking events (CTEs) and key data elements (KDEs) that IFT developed in 2009 (McEntire and others 2010). The use of the summary templates was optional. Firms were asked to provide the information they deemed necessary to respond to the IFT request. All participants supply chains were evaluated on the basis of a number of factors:

  • Breadth and precision: the amount, nature, quality, and accuracy of information provided
  • Access: a combination of the following factors:
    • ○ Total time: cumulative supply chain and individual firmresponse times
    • ○ Minimum time: time before convergence was found (or thetrace was otherwise ended)
    • ○ Analysis time: time needed by IFT to understand and analyzeparticipant-provided data
  • Depth: a firm's ability to readily provide information for more than one supplier back in their supply chain (whether they themselves had the information or could readily acquire it)
  • System ranking: the sum of a firm's self-reported abilities, including the technologies currently in use by the firm that enable them to link incoming and outgoing product and their reported ability to meet 9 options for improved product tracing (described in the next section)
  • Quantity: total number of pages of documents provided
  • Format: use of IFT-supplied or company-generated summary document

Key to this task was the exploration of how technology can be used by investigators to enhance the speed, effectiveness, and accuracy of the product tracing process. Additionally, IFT also conducted a qualitative study of industry's use of technology to improve product tracing capabilities.

IFT was tasked with using a “collaboration platform” for the mock tracebacks involving ingredients and processed foods, and IFT opted to use similar platforms for select mock tomato tracebacks. Because the term “collaboration platform” was not defined in the task, this was one area around which IFT solicited stakeholder input. Ultimately, the “collaboration platform” functioned as more of a data analysis system, which could be used by FDA (or other regulators) to share and analyze data collected from industry. Collaboration platforms were not used in this task by food industry members to submit data. Instead, industry data were collected by IFT through these pilots. IFT in turn blinded and supplied these data to the collaboration platform providers. These collaboration platforms were then used to query the data to look for convergence and conduct tracebacks.

A transparent process was used to broadly solicit input on how the collaboration platforms should be selected. Ultimately, 9 firms that currently offer commercially available track-and-trace solutions participated in the evaluation process. During the evaluation, roughly half the firms received identical data sets for the tomato pilot and the other half received data for the ingredients and processed foods pilot. The names of the supply chain participants were blinded before being shared with the collaboration platform providers (identified only generically as “Distributor 7,” for example). The data were provided to the collaboration platform providers in the same format that IFT received them from the pilot supply chain participants (for example, in PDFs, spreadsheets).

After uploading the data, all 9 collaboration platform providers explained their approach and demonstrated their systems, using the provided pilot data, for a broad panel that included FDA, pilot participants, and other Subject Matter Experts. The goal of the demonstrations was not to select one provider; rather, it was to observe capabilities that seemed to improve the speed and accuracy of traceback investigations.

To conduct the in-depth assessment of the costs associated with product tracing, IFT conducted a literature review and sought information from pilot participants and others (such as, technology providers, companies, and organizations) that generally is not published.

Results and Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodology
  5. Results and Discussion
  6. Recommendations
  7. Conclusion
  8. References

IFT was successful in conducting 14 mock tracebacks, ranging from simple (tracing 1 shipment of tomatoes or 1 lot code of peanut butter) to complex (for example, finding convergence when tomatoes were thought to be sourced from 2 different growers; finding a common lot of ingredient between different processed food products manufactured in different locations). Traceforwards were also explored as elements of some of the scenarios; however, when firms provided information on the recipients of one or more lots of product, the lists were often lengthy and the majority of recipients were not pilot participants which caused the traceforward to end.

The process of conducting a step-wise traceback (one supply chain node at a time) was complicated and oftentimes confusing. Most firms provided information in the form of PDF documents. While information in this format can be transmitted electronically via email, the information is image-based and cannot be manipulated electronically, which makes analysis of data slow and potentially error-prone as data must be re-entered or extracted via optical character recognition for software analysis. Additionally, inconsistencies in the terminology, numbering systems, formatting, legibility, and occasionally the language sometimes required IFT to contact the submitting firm to gain clarity, thereby increasing the time required to capture data before any meaningful analysis could begin. In many instances, firms provided a document to explain how the numerous documents and reports (in some cases, scores of pages) were linked together to demonstrate how the product moved through the facility. This was extremely helpful, as was the use of summary documents. While there were occasional errors in the summary documents, they provided an “at a glance” means to better understand the information provided in the detailed source documentation.

Most notably, IFT found that some participating firms were surprised by the process used, and expected an experience more like a mock recall in which they would be provided with a lot number and asked to identify where the product was sent. Many had never considered how their records would need to be pieced together with those of their supply chain partners to facilitate an effective traceback.

Challenges aside, the pilot participants appeared to have many of the tools and processes in place which are required to allow the capture and communication of critical track-and-trace data (such as, KDEs) at critical points of product transfer and transformation (CTEs). Many of the collaboration platforms were able to demonstrate the flow of specific lots of product through the supply chain with minimal effort, and some were able to identify convergence. However, while querying occurred within seconds, the collaboration platform providers reported spending between 3 and 7 d uploading the data into their systems due to the lack of a standard structure or format and the need to re-enter data.

Based on the discussions with the pilot participants and other industry stakeholders, IFT observed that firms provided track-and-trace data in several ways. Ultimately, the way in which data were readily accessed and transmitted to IFT was dependent on the systems and processes in place within a firm to capture, store, and report this information.

IFT identified 9 specific processes firms could use to improve product tracing. The first 4 options revolve around data capture. IFT believes that capture of the right data, regardless of format, is a prerequisite to any substantial improvements in product tracing. Thus, the first 4 options explore different ways that the same data could be captured to account for what is practicable for facilities of varying sizes, including small businesses. For reasons described below, the KDEs included in the options presented to pilot participants did not include lot/batch number.

The first 4 options (for which questions were asked around current capabilities and costs) were:

  • Capture KDEs (Supplier ID, Product ID, Purchase Order Number, Quantity-pack size, Receipt Date) by writing on paper
  • Capture KDEs (Supplier ID, Product ID, Purchase Order Number, Quantity-pack size, Receipt Date) by writing on paper and later entering into a database/spreadsheet
  • Capture KDEs (Supplier ID, Product ID, Purchase Order Number, Quantity-pack size, Receipt Date) by scanning labels
  • Capture KDEs (Supplier ID, Product ID, Purchase Order Number, Quantity-pack size, Receipt Date) by electronic message

IFT observed that some segments of the distribution chain did not generally record the grower/producer-assigned lot number. Distributors, for example, are not required to record this information and those who manufacture, process, or pack food are required to record lot numbers only if the information exists (U.S. Food and Drug Administration 2004). Therefore, as noted above, lot/batch number was not included as a data element in the 4 options above, but was treated as its own question. The remaining 5 options related to the use of standards, communicating data forward to customers, and the use of a summary data sheet. They included a firm's ability to:

  • capture incoming quantity by received lot number, assuming a lot number is provided
  • link incoming and outgoing product, whether there is transformation (such as, ingredients into a finished product) or not (for example, relating lot numbers received to lot numbers shipped)
  • use nonproprietary standards (for example, Global Trade Item Number [GTIN], Global Location Number [GLN], state-issued plant/registration number)
  • send KDEs electronically to customers
  • provide a data summary sheet (or template such as that provided by IFT) that highlights the links between KDEs for the products of interest

A literature review was conducted to analyze previously published studies on the costs and benefits of improving record-keeping and product-tracing capabilities. However, there were very few studies that published quantitative costs or benefits. Instead, they described more qualitative characteristics in their observations and analysis. For example, the costs associated with improvements include fixed and variable costs, like capital equipment, software, consulting, design and implementation, training, labor, materials, and impact on speed of business operations. The qualitative benefits associated with improvements include protection of public health, improved trade, sustainability tracking, limited recall scope, increased market access, quality assurance, and supply chain efficiencies. Due to the limited availability of published studies, IFT collected additional data through the use of nonpeer-reviewed case studies and white papers including data from technology solution providers and standards organizations. Several nonpeer-reviewed studies, some of which are tied to implementing bar code systems in produce, show that there are a range of capabilities and associated costs and benefits to the firm by implementing or improving a firm's product tracing system.

In addition, Deloitte Consulting worked with the pilot participants to determine costs associated with the 9 identified options, as well as the types of benefits that firms had realized from their investments. Pilot participants were asked to indicate whether they had systems and processes in place to perform any of the 9 activities listed above (as options), and for any activity not already in place to provide an estimate of the resources needed and cost required to attain the goal.

In terms of the costs needed to implement the 9 options identified above, the 22 firms that provided data reported the ability for some form of data capture. For those capturing data by hand or who had invested to convert manually captured data to spreadsheets, the cost of this capability ranged between $40 and $350K. In contrast, capturing the same types of data, but doing it by scanning (as by bar code) was reported to be roughly an order of magnitude more expensive, ranging between $125K and $4.5M. This is consistent with the experience reported by firms implementing the produce traceability initiative (which requires the use of GS1 128 bar codes); the reported range of costs was generally from several hundred thousand to a few million dollars. Further, these ranges reflect all business sizes and supply chain segments.

Many firms reported the ability to capture incoming lot numbers (assuming they were provided), however, the pilot demonstrated that even if this capability exists, it is more likely to be used by processors, especially of multi-ingredient products, compared to others in the supply chain. Therefore, while the estimate to reach this capability ranged from $0 to $150K, IFT expects that implementation of this practice would be more costly, although a focused effort would be required to quantify these costs.

Of all the options presented, the development of a data summary, whereby industry would present the KDEs in a logical fashion that illustrates the internal and external links, was deemed the easiest to achieve in terms of expenditures. Firms generally reported this capability, and where resources were required, they were never reported to be in excess of $10K annually.

To assess the benefits associated with improved product tracing, IFT conducted a literature review, evaluated 8 case studies of previous outbreaks and sought information from pilot and nonpilot participants. The benefits associated with improved recordkeeping and, therefore, improved product tracing fall into 3 general categories:

  • Benefits to the FDA: FDA expends resources during an investigation that can presumably be decreased if investigations could be conducted more rapidly and with less manual manipulation and analysis of trace data. IFT did not quantify the resources used by FDA in investigations to ascertain the extent of the benefit.
  • Benefits to public health: Protecting public health is the key goal of an improved product tracing system. To quantify the benefits to public health, IFT examined 8 previous outbreak investigations. The duration of the traceback investigation and the illnesses that occurred during this timeframe were determined. Working with Deloitte Consulting, IFT translated the number of illnesses into costs using existing government figures. The cost savings (driven by reductions in illness) resulting from reducing traceback duration by 25%, 50%, and 75% were calculated. The range of the public health benefit per outbreak spanned $18K to $14M depending on the characteristics of the outbreak.
  • Benefits to the industry: Quantifying the benefits to a particular firm is completely dependent on the way a firm chooses to meet the required track and trace objectives and was therefore difficult to calculate. The literature and nonpeer-reviewed information was either qualitative or demonstrated the benefits of a very specific system. The types of benefits described by pilot participants were consistent with those suggested in the published literature. Table 1 identifies areas of benefit which were reported by the pilot participants and illustrates how these benefits varied depending on a firm's location in the supply chain.
Table 1. Percent of pilot participants identifying with recordkeeping benefits
Recordkeeping benefitsGrowers (n = 2)Processor (n = 6)Distributors (n = 8)Retailers (n = 4)
  1. Whenever the response to an individual benefit was left blank, it was treated as a “does not identify this benefit” answer in the calculations above. (Percent of Pilot Participants Identifying the Recordkeeping Benefit)

Improved brand reputation100%33%62%50%
Increased consumer confidence0%67%75%25%
Expanded markets50%33%50%25%
Improved supply chain management50%67%62%100%
Insurance cost reduction50%33%12%0%
Supply chain confidence0%83%75%25%
Decreased spoilage50%67%75%25%
Process improvement100%33%100%100%

It was noted that many of the tangible benefits to industry of recordkeeping could potentially be enabled through process and technology improvements that may or may not also enable product tracing. It is unclear if the identified tangible benefits can be fully captured by all industry participants and whether these benefits will be sufficient to cover the investment required for improving product tracing. Therefore, recognition of public health benefits is critical.

Recommendations

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodology
  5. Results and Discussion
  6. Recommendations
  7. Conclusion
  8. References

The pilots demonstrated some of the challenges that FDA has in achieving its goal of being both quick and accurate when conducting traceback investigations. However, the pilots also demonstrated areas in which improvements can be implemented to reduce traceback time and ensure the accuracy of information.

IFT has 1 overarching recommendation to improve product tracing, 2 recommendations for FDA to consider during the rule making process, and 7 additional recommendations. While these recommendations are actions FDA can take, those in the food supply chain should view these recommendations in the context of the nature of improvements that may be expected of them.

  1. From an overarching perspective, IFT recommends that FDA establish a uniform set of recordkeeping requirements for all FDA-regulated foods and not permit exemptions to recordkeeping requirements based on risk classification. Further, FDA should issue guidance documents defining these requirements.
  2. With regard to future rulemaking, IFT recommends that FDA require firms that manufacture, process, pack, transport, distribute, receive, hold, or import food to identify and maintain CTE and corresponding KDE-related records as defined by FDA based on input from the food industry.
  3. Also in regards to rulemaking, IFT recommends that FDA require each member of the food supply chain to develop, document, and exercise a product tracing plan.
  4. FDA should encourage and support existing industry-led initiatives for the development of implementation guidelines and should seek stakeholder input by issuing an Advance Notice of Proposed Rulemaking (ANPR) or using other input mechanisms.
  5. FDA should clearly and more consistently articulate and communicate to industry the information needed during a product tracing investigation.
  6. FDA should develop standardized, structured, and electronic mechanisms for industry to provide the Agency CTE and KDE product tracing data when requested during a specific food safety investigation.
  7. FDA should accept CTE and KDE data sent in summary form through standardized and structured reporting mechanisms and initiate investigations based on such data.
  8. If available, FDA should request CTE and KDE data for more than one up–one back in the supply chain.
  9. FDA should pursue the adoption of a technology platform to allow the Agency to efficiently aggregate and analyze data reported in response to a specific request from regulatory officials. The technology platform should also be available to regulatory counterparts.
  10. FDA should coordinate traceback investigations and develop response protocols between and among state and local health and regulatory agencies using existing commissioning and credentialing processes. Further, FDA should formalize the use of industry Subject Matter Experts (SMEs) to address FDA's general questions about the characteristics of a particular supply chain at the outset of an investigation.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodology
  5. Results and Discussion
  6. Recommendations
  7. Conclusion
  8. References

IFT recognizes that there are several barriers to implementing the recommendations presented above. Barriers include issues related to: current availability of KDEs and other prerequisites to efficient data capture and sharing; availability and accessibility of technology, particularly to small businesses and firms in developing countries; and need for continued education and recognition of cultural differences.

With FDA positioned to commence the rulemaking process requiring additional records for high-risk foods, the food industry is anxiously awaiting direction from the Agency regarding the expectations of a product tracing system. Many of the industry-led initiatives have met a level of resistance owing to the concern that FDA might require something at odds with the initiatives and implementation guidance. Numerous individuals contributing to the pilot studies expressed their hope that the results of this work would be used to enhance industry best practices and drive change. There are many documents discussing the challenges associated with tracing food products, and some that offer a path forward. Recognizing that change takes time, our hope is that with this report, and FDA's subsequent report to Congress, change starts now.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methodology
  5. Results and Discussion
  6. Recommendations
  7. Conclusion
  8. References