Will the Produce Traceability Initiative expedite the ability to

Document Sample
Will the Produce Traceability Initiative expedite the ability to Powered By Docstoc


    Will the Produce Traceability Initiative expedite the ability to isolate the source of 
                           foodborne illness in fresh produce? 






                                        Justin Porter 









                                  Submitted in fulfillment of 

                                   the requirements for the  

                                       Master’s Thesis 

                                  SANTA CLARA UNIVERSITY 


                                       March 21, 2011                                 

                                  Table of Contents 
i.        Thesis Statement                                                         5 
ii.       Procedures                                                               7 
iii.      Implications of Research                                                 11 
Part I.   Introduction:  Foodborne Illness & The Impetus for Rapid Traceability     
1)        Foodborne Illness and the Call for Enhanced Traceability                 13 
2)        Traceability: Comparing Preventative and Reactive Food Safety  
          Approaches                                                               17 
Part II. From Farm to Fork:  Understanding the Fruit & Vegetable Supply Chain 
1)       Points of Origin – Growing, Harvesting & Packing                          20 
          a)    Growers                                                            21 
          b)    Harvesters                                                         22 
          c)    Shippers                                                           23 
          d)    Summary                                                            24 
2)       Buyers                                                                    27 
          a)    Retail                                                             27 
          b)    Foodservice Distribution                                           27 
          c)    Brokers                                                            27 
          d)    Processors/Re‐packaging & Consolidation                            28 
          e)    Terminal Markets                                                   28 
          f)    Direct to Consumers                                                29 
          g)    Summary                                                            29 
3)       Transportation                                                            30 

Part III.       Current State of Traceability in the U.S. Fresh Produce Supply Chain 
1)       Perishable Agricultural Commodities Act                                  32 
2)       Country of Origin Labeling                                               32 
3)       The Bioterrorism Act of 2002                                             33 
4)       Reportable Food Registry                                                 34 
5)       Future Directions: The Food Safety Enhancement Act of 2009               35 

Part IV.        The Traceability Investigation: Agencies & Roles, Recent Outbreaks & 
        Lessons to Learn From  
1)      The Current State of Traceability and Outbreak Investigations           36 
2)      Systems Utilized for Outbreak Monitoring                                36 
3)      State, County and Local Health Agencies                                 37 
4)      CDC – Centers for Disease Control                                       38 

5)      FDA – Food and Drug Administration                                   38 
6)      The Trace‐back Procedure                                             39 
7)      Recent Major Outbreaks 
         a)    2006 E. coli O157:H7 Outbreak in Bagged Spinach               44 
         b)    2008 Salmonella Outbreak in Peanut Butter                     46 
8)      Supply Chain Challenges in the Trace‐back Process                    47 
         a)    Farm to Fork Revisited:  Where are we Today?                  47 
               1.     Original Growers/Shippers                              48 
               2.     Processing/Reconfiguring                               50 
               3.     Brokers                                                50 
               4.     Distribution                                           51 
               5.     Retail                                                 52 

Part V.  Future Goals of Traceability:  A Foundation for the Future 

1)      Analysis Methodology                                                 54 
2)      Future Traceability:  The Consumer and Food                          54 
3)      Standardization of Datasets 
         a)       Uniqueness of Identifiers                                  56 
                i.       Common Nomenclature                                 57 
                         1.     Company Identification                       58 
                         2.     Discernable Product Characteristics          59 
                         3.     Lot or Batch Information                     60 
             ii.        Common Level of Detail                               61 
4)      Linking Data Elements                                                63 
5)      Standardization of Reporting Methodology                             64 
6)      Equal Requirements for Recordkeeping Across Supply Chain             66 
7)      Visibility                                                           69 

Part VI.       The Produce Traceability Initiative 

1)      Case Level Identification vs. Item Level Identification              72 
2)      Standardization with PTI and GS1                                     73 
3)      PTI – Milestones of Implementation                                   74 
        a.       PTI Milestone #1:  Obtain a Company Prefix                  75 
        b.       PTI Milestone #2:  Assign GTINs to Cases                    77 
               i.       Reference Numbers for Unique Case Configurations 
              ii.       A Final GTIN                                         80 
        c.       PTI Milestone #3:  Provide GTIN Information to Buyers       80 
        d.      PTI Milestone #4 & #5:  Show Human‐Readable Information           81 
                On Cases and Encode Information in a Barcode 
                i.     Human Readable Information                                 82 
               ii.     Barcode – Computer Readable Information                    83 
              iii.     Additional Case Label Requirements and    
                       Recommendations                                            84 
        e.      PTI Milestone #6:  Read and Store Information on Inbound Cases    85 
        f.      PTI Milestone #7:  Read and Store Information on Outbound  
                Cases                                                             85 

Part VII:      Analysis:  PTI and the Future Goals 

1)      Standardization of Datasets                                               88 
        a.       Uniqueness of Identifiers                                        88 
               i.          Common Nomenclature                                    88 
                              1. Company Identification                           88 
                              2. Discernable Product Characteristics              92 
                              3. Lot or Batch Information                         95 
            ii.            Common Level of Detail                                 96 
2)      Linking Data Elements                                                     98 
3)      Standardization of Reporting Methodology                                  99 
4)      Equal Requirements for Recordkeeping Across Supply Chain                  100 
5)      Visibility                                                                101 

Part VIII:    Discussion, Conclusion & Recommendations:  The Future of Fresh 
        Produce Traceability in the U.S.	

1)      Discussion                                                                102 
2)      Conclusion                                                                105 
3)      Recommendations                                                           106 
        a.     Stick to Uniformity Principles                                     106 
        b.     Standardize Product Reference Number Elements                      107 
        c.     Create a Reporting Mechanism for Investigative Purposes            108 
        d.     Create an Environment of Open & Transparent Communication          108 

Bibliography                                                                      110 

Appendix A:  Base Open‐Ended Questions for Supply Chain Interviews                114 



i.	           Thesis	Statement	

              As the American population moves further away from any direct link to our

domestic food supply, consumer confidence in the safety of that food supply is critical.

Despite an increase in government oversight and significant advances in the detection and

investigation of foodborne illnesses and outbreaks, the number and severity of incidents

has not decreased in recent history. Accordingly, in 2009 the Produce Traceability

Initiative (PTI) was proposed by a coalition of fresh produce industry leaders as a

voluntary industry effort1 to “enhance total supply chain traceability to better serve our

customers, to expedite trace-backs and recalls, and more narrowly isolate potential

recalls or other problems when they do occur.” The PTI was developed to help establish

a common system of traceability across commodities in the produce industry.

              Although it was proposed by industry leaders2, industry support for the PTI is

mixed at best. The cost of implementation has been a major hurdle for both suppliers and

distributors of fresh produce due to the additional technological and labor expenses

involved in adding a lot-specific barcode to each box and then reading and storing that

data. Additionally, the use of a highly standardized approach to labeling a wide variety of

commodities with different growing, packing, and handling processes creates additional

burdens for users. For those firms that are already complying with one-up, one-down

regulations under the U.S. Bioterrorism Act, doubt exists that the addition of a barcode

on a case will be helpful to a traceback investigation by the FDA. Finally, critics are
   The PTI was not developed to be a mandatory process nor is it a legal requirement.  Adherence to the 
PTI is voluntary across the supply chain. 
   PTI is sponsored by the Produce Marketing Association, United Fresh, and the Canadian Produce 
Marketing Association and was developed by a steering committee of representatives from major 
grower/shippers, foodservice distributors and retail outlets. 

worried that the PTI makes it easier for regulators to use the grower as a scapegoat by

providing a simpler mechanism to point straight back to point of origin instead of the

path through which food traveled on its way to consumers.  This worry is compounded by

recent research that suggests that contamination of fresh produce outbreaks occurs less

commonly in the field and in original handling than in subsequent steps through the

supply chain.  

       Proponents of the PTI cite the differences between process flow (the physical

flow of the product through the supply chain) and transactional flow (the flow of money

and paperwork in the sales and distribution through the supply chain) as a major benefit

of the PTI. A more robust tracing system that identifies each physical location (the

process flow) through which produce is handled instead of only transactional data (such

as invoices) could shorten the time-consuming process required to trace the physical path

of produce through the supply chain. The electronic footprint left by the PTI can provide

a standardized link between each point in the supply chain through which the product

moves and, subsequently, might be contaminated. As the flow of money and sales is very

often different than the flow of physical produce, authorities will be able narrow the

number of commodities, brands, and companies implicated in an outbreak by matching

outbreak locations to specific lots of product sold through those locations. Accordingly,

labels identifying the brand owner, commodity, product identification, and lot would not

be used simply to identify the original supplier but rather to identify all relevant handlers

along the supply chain.

       This paper examines current regulations, adherence to those regulations by supply

chain members and analysis of trace-back successes and failures to establish future goals

for traceability in fresh produce. Using this approach, PTI will be analyzed to establish

whether or not it accomplishes its goal of enhancing the supply chain by expediting trace-

backs and recalls, and more narrowly isolating potential recalls or other problems when

they occur. More specifically, this paper examines whether or not PTI will enable

investigators to trace consumed product through the entire supply chain more effectively

than the current status quo. Finally, in addition to the conceptual result of PTI, elements

of implementation, uniformity and economics will be examined to provide as realistic as

possible an outcome and ultimate impact upon firms in the fresh produce industry and

consumers purchasing their products.

ii.	   Procedures	

       The first objective was to develop a set of future goals for an effective traceability

system. To that end, research started with a thorough review of existing local, state and

federal regulatory requirements to develop a complete picture of the current chain-wide

traceability system as it applies to produce, as well as its strengths and shortcomings.

Primary research was then conducted in the form of open-ended interviews with growers,

shippers, brokers, retailers, foodservice providers, and distributors in the fresh produce

supply chain. The objective of the interviews was to determine the strengths and

shortcomings of the current system in order to formulate a set of future goals for a

successful traceability system. This primary research was supplemented with publicly

available comments and information from presentations by major firms in the produce

supply chain. A total of 17 interviews were conducted with representatives of

grower/shippers, brokers, processors, retail chains, technology vendors, and industry

associations using an interview guide. The majority of these interviews were carried out
in-person; a few interviews were conducted by telephone. Secondary research was then

analyzed, focusing on the strengths and shortcomings of the existing system. The

analysis was based on previous academic work in the field, government documents,

public testimony of supply chain participants before Congress, the FDA, the CDC, the

USDA, legislation, and documentation from various electronic sources regarding specific

initiative data. Two case studies were also analyzed for a deeper understanding of the

food safety system, the 2006 E. coli contamination of

fresh spinach and the 2008 salmonella contamination of peanut butter.

Type of Firm                                              Public
                                                               Interviews       Total
Growers/Shippers                         10               0                     10
Retail/Distribution/Foodservice          3                2                     5
Processors                               4                1                     5
Technology Vendors                       2                2                     4
Industry Associations                    1                8                     9
Brokers                                  8                0                     8
       Second, a set of future goals that would improve upon the effectiveness and

timeliness of the current traceback system was developed. A four stage iterative process

was employed such that these goals encompassed, as accurately as possible, data

accumulated through interviews as well as the analysis of secondary research. The first

stage was to lay out each “shortcoming” in the current system as illustrated by primary

and secondary research (interviews with supply chain members, FDA testimony and

public releases, and an analysis of recent major outbreak and recall shortcomings during

open ended interviews). The second stage was to examine the shortcomings identified
   Total interviews were 17; however, the sum of interviews in this column is above 17 as some interviews 
represented multiple points of expertise in the supply chain 
   Six brokers were also considered Growers/Shippers that had a defined business for brokering fruit.  Two 
interviews were engaged primarily in the business of brokerage 

through various methods, looking for similarities such that they could be grouped into

more cohesive categories. The third stage was to turn each category of “shortcomings”

on its head and to develop “goals” for an effective traceability system. Finally, the future

goals were compared to the original shortcomings from stage one to ensure that the

synthesized final goals did not lose their original substance. This process was then

repeated until the substance of the original shortcomings and the substance of the

synthesized goals were one and the same.

        The third step was to assess the PTI relative to the future goals. The PTI

implementation is laid out in detail and compared the system that the PTI would create in

the supply chain to that of the future goals. This analysis allowed us to highlight what the

PTI accomplishes conceptually and where it falls short. It is important to note that the

PTI is described as a “living document” which is regularly updated. In order to maintain

a consistent analysis, a single version of the PTI documentation as of May, 2010 was


        Finally, based on the above analysis of the future traceability goals and the

assessment of the PTI, recommendations or principles were developed that should be

adopted in order to help ensure that future versions of the PTI will result in an effective,

timely and accurate traceability system for the produce industry.

        Specifically, analysis of the supply chain (Part II) relies heavily upon data

obtained via interviews and public comments from supply chain members and serves to

lay a foundation for the differing functions of each point in the supply chain. The process

of an outbreak investigation (Part IV) relies much more heavily upon cited research and

public commentary by regulatory bodies (also cited) but also relies heavily upon supply

chain interviews for those parts of an outbreak investigation specific to supply-chain

members. The creation of future goals for traceability in fresh produce (Part V) is a

synthesis of interview data, research, public commentary and conclusions drawn therein

by the author. Additionally, analysis of the PTI in terms of its ability to achieve the

future goals as specified in Part V and recommendations for moving forward (Parts VII

and VIII) is based largely on a synthesis of data already presented and conclusions drawn

by the author. Analysis of goals was an iterative process of qualitative hypothesis testing

through which each goal was built through the use of each interview as well as public

testimony such that, while the summarization and definition of the goals is that of the

author, the goals themselves are a direct reflection of the accumulation of interviews


iii.	   Implications	of	Research	

        With a great deal of academic research supporting traceability methodologies

from a health and economic standpoint (as well as the enhancement of current

methodologies) in the food industry and an increasingly loud call from both consumers

and government, the fresh produce industry will inevitably move forward with

traceability standards. In this manner, the industry sits at the crossroads: provide results

through self-regulation or prepare to be more heavily regulated. PTI has been proposed

and endorsed by a significant number of major retailers, foodservice providers and

dominant shippers (suppliers) in the industry as a method for enhanced rapid traceability

through the fresh produce supply chain. Given current efforts of governmental bodies

involved in traceability investigations and established traceability practices already in

existence in the fresh produce industry, a thorough and realistic analysis of PTI as the

answer to enhanced rapid traceability has yet to be accomplished. This research and

accompanying findings aims to provide a “down to earth” view of traceability as it exists

currently and traceability as it would exist utilizing PTI as an enhanced methodology.

        This research and findings are not intended, nor should it be used or endorsed as,

legal guidance for any member of the supply chain. Should this research be effective, it

will cause a necessary discussion among members and across levels in the supply chain

as current roadblocks to traceability often come back to the act of pointing the finger.

Naturally, no member of the supply chain wants the blame for a major outbreak or the

financial fallout related to an outbreak. The truth, though, is that even as food safety

programs are enhanced across the supply chain, widespread foodborne illness outbreaks

continue to occur. The goal is to enact a program that will minimize the trace-back

investigation whereby societal and industry costs as the result of outbreaks is minimized.

This cannot occur without thorough and open discussion at all levels of the supply chain.


                                           Part	I:	

    Introduction:		Foodborne	Illness	&	The	Impetus	for	Rapid	Traceability	

Foodborne	Illness	and	the	Call	for	Enhanced	Traceability	

         In the U.S., foodborne illness account for approximately 76 million illnesses,

325,000 hospitalizations and 5,000 deaths annually (Mead, et al., 1999). Based upon

USDA Economic Research Service (ERS) and Centers for Disease Control (CDC)

estimates of incidence, outcome severity, as well as level of medical, productivity and

disutility costs, outbreaks of Salmonella and E coli O157:h7 alone are estimated to cost

$3,124,811,739 annually (USDA, 2009). While there are greater than 200 known

diseases transmitted through food, a primary complicating factor in detecting foodborne

illness are pathogens or agents not identified or diagnosable. CDC estimates that more

than 75% of illnesses, hospitalizations and deaths are caused by unknown agents and

pathogens. Some of the most recognizable pathogens today (i.e. E coli O157:H7) were

not recognized as the cause of foodborne illness until the last twenty years (Mead, et al.,


         Advances in technology have made the detection of foodborne illness and

foodborne illness outbreaks more rapid. A foodborne illness outbreak is defined as two

or more cases of a similar illness resulting from ingestion of common food (Mead, et al.,

1999). Hailed by CDC because of its standardization of protocols, software and

nomenclature, PulseNet was created after a major E. coli O157:H7 outbreak in 1993 that

caused approximately 800 illnesses. In the E. coli outbreak, scientists used DNA

“fingerprinting” to match the strain of E. coli found in sick patients with a strain found in

hamburger patties that were served by a fast food chain. They realized that had they been

able to more rapidly identify the DNA match, they likely could have prevented many of

the illnesses that occurred. PulseNet was created as an on-demand electronic database of

DNA “fingerprints” of disease-causing bacteria taken in a standardized manner from both

sick humans and food suspected of infection. The CDC conducts searches of the

database in order to identify clusters of illnesses with the same DNA footprint report

results. This rapid electronic detection of indistinguishable DNA “fingerprint” clusters

across the U.S. enables the CDC to distinguish and react to outbreaks far more rapidly

than was previously possible—outbreaks that once took days to discover can now be

detected in hours (National Center for Infectious Diseases, 2009).

        While technological and communicative advances have aided in the detection of

outbreaks, there is still considerable difficulty investigating and tracing back to the

source of the problem. Recent studies have shown that, of those investigations in which

there was a successful source implication, only 2% of outbreaks and 6% of illnesses were

confirmed as problems associated with growing, packing, shipping or processing fresh

produce. This data is based upon confirmed cases, however, and does not account for

those outbreaks and illnesses in which a source was never implicated (Alliance for Food

& Farming, 2010). Once the source of contamination and the scope of products affected

are identified, the FDA attempts to trace-forward to each location in which the product

was shipped to recall. While the specific problems encountered in the traceability

process will be addressed later in this analysis, there are three central tenets to the reality

that advances in traceability protocol in the produce supply chain will likely be impacted

over the foreseeable future:

    1) Whether the result of better reporting and detection or publicity, the

       number and economic cost of foodborne illnesses and outbreaks is on the

       rise in the eyes of regulators and consumers as seen through research and

       media reports

           Between 1990 and 2008, over 700 foodborne illness outbreaks were linked

       back to produce (Center for Science in the Public Interest, 2008). The number

       of annual incidences is also on the rise—40 outbreaks were linked to produce

       in 1999 and increased to 86 by 2004. This is caused by a variety of factors,

       including the increased consumption of fresh produce, wider distribution,

       enhanced reporting of outbreaks and illnesses and an aging population more

       susceptible to foodborne illness (Shin, 2006).

           A recent study estimated that foodborne illness outbreaks cost the U.S.

       approximately $152 billion annually in healthcare and lost productivity costs.

       $39 billion of this figure is attributed to produce related outbreaks. From an

       industry point of view, Kellogg Co. alone estimates that it lost approximately

       $70 million from recalled products during the recent peanut recall and the

       spinach industry has sustained between $50 and $100 million in losses in the

       year following a major 2006 E. coli outbreak (Van Cassell, 2007).

    2) Current Research Supports the Need for Enhanced Traceability

           The Office of Inspector General under the U.S. Department of Health and

       Human services recently released the results of a traceability study it

       embarked upon in which forty food products were purchased randomly at

       retailers and then traced back through the supply chain. Of the forty food

       products, investigators were only able to fully trace five of the products back

       to their sources. They were unable to trace through the supply chain because

       lot specific information was not always recorded at the processor, packer and

       manufacturer level, lot information was often not recorded by other facilities

       because it was not mandatory, products were not always labeled with lot

       specificity and some products were the result of comingling products from

       many different farms.    Additionally, the OIG found that 59% of the food

       facilities did not meet the recordkeeping requirements set forth by the FDA

       while 25% of food facilities were unaware of the requirements (Levinson,


    3) The U.S. Government Is Working Toward New Regulation and Guidance

       for Traceability

       Tom Vilsack, the U.S. Secretary of Agriculture, recently indicated that the

    prevention of foodborne illness is a top priority of President Obama as well as the

    USDA. He emphasized the goals of the Food Safety Working Group (FSWG)

    organized by the President (Bottemiller, 2010). A top FSWG goal is the creation

    of a National Trace-back and Response System that will “improve our national

    capacity for detecting the origins of foodborne illness” (The President's Food

    Safety Working Group).

       Congress has also acted to push forward with a broader and more defined

    traceability system for all food located in or imported in to the United States. In

    the Food Safety Enhancement Act of 2009, Congress calls for the establishment

    of a tracing system for all food in or imported into the United States such that

              each person who “grows, produces, manufactures, processes, packs, transports,

              holds, or sells food” can be quickly identified (H.R. 2749--111th Congress: Food

              Safety Enhancement Act of 2009, 2009).

                      As the result of recent major foodborne illness outbreaks and calls from

              Congress to act, the FDA has held public meetings on traceability efforts and

              requested public testimony regarding traceability efforts and methods that can

              improve the rapidity and accuracy of trace-back efforts. While outbreaks are on

              the rise, the ability of the FDA to police the industry is on the decline—the budget

              for their Center for Food Safety and Applied Nutrition is on the decline and each

              year the number of facilities they are able to inspect decreases (Shin, 2006). The

              onus is on industry to provide more efficient means of both food safety and

              traceability investigations.

Traceability	‐	Comparing	Preventative	and	Reactive	Food	Safety	


              There is an important distinction between the concept of preventative and reactive

food safety paradigms—food safety and traceability are not the same thing. Food safety

is defined by the University of Rhode Island Cooperative Extension as: “Protecting the

food supply from microbial, chemical (i.e. rancidity, browning) and physical (i.e. drying

out, infestation) hazards or contamination that may occur during all stages of food

production and handling-growing, harvesting, processing, transporting, preparing,

distributing and storing.5” In other words, the concept of food safety is a preventative

methodology through which efforts to safeguard food from foodborne illness causing

     (University of Rhode Island, 2000) 

bacteria are employed.     Traceability, on the other hand, is the “ability to trace the

history, application or location of that which is under consideration” (Golan, Krissoff,

Barry, Kuchler, Calvin, Nelson, & Price, 2004). More specific to food, the definition of

traceability can be further refined as the ability to “follow the movement of a food

product through the stages of production, processing and distribution.” Traceability, in

the context of a foodborne illness outbreak, is a reactive methodology as opposed to a

preventative methodology as it is a means to perform both a trace-back to the source of

the food product through the supply chain as well as a trace-forward from the source to

the retail shelf. The traceability process is what allows investigators to locate the source

of contamination in the event of an outbreak and subsequently locate any other outlets

through which the same contaminated product was delivered for a recall (Levinson,

2009). Additionally, it is equally important that an effective trace-back allows

investigators to learn and identify holes in the U.S. food system such that more effective

preventative controls can be utilized (USDA, 2009, p. 61).

       It is important to note that traceability is incredibly broad in definition—that is to

say, there is no standard means by which traceability is utilized and/or measured (i.e.

standard location, processes that should be identified, technology to be utilized, etc.)

There is no single approach to traceability that works for every application. In order for

effective implementations of traceability to occur, the objective and breadth, depth and

precision must first be identified in order to establish what details are worth tracking

(Golan, Krissoff, Barry, Kuchler, Calvin, Nelson, & Price, 2004). Traceability is an

abstract in which the ends of a traceability implementation dictate the means by which it

is applied and utilized.

       PTI and this analysis are focused upon the trace-back and trace-forward

procedures, the speed and accuracy with which they are currently carried out as well as

the changes to the speed and accuracy through implementation of PTI.


                                 Part	II	
    From	Farm	to	Fork:		Understanding	the	Fresh	Produce	Supply	Chain	

              The fresh produce supply chain is complex and must function rapidly and

efficiently in order to maintain freshness and safety for the consumer. Fresh produce

commonly passes through multiple points in the supply chain before reaching the end

consumer. The following section examines the fresh produce supply chain and maps the

typical means by which fresh produce moves from the farm to the final consumer6. This

section will provide an overview of the supply chain and graphical representations of

common arrangements through which product flows and transactions flow—the flow of

product and the flow of transactions (monetary transactions) do not always follow the

same path. In the produce supply chain, it is often the case that the authority, control and

liability of the product are in different hands at different times.

Point	of	Origin	‐	Growing,	Harvesting,	Packing		

              While a farming operation is the first point in the supply chain, farming

operations come in many “shapes and styles”—some farms perform only the growing

function while others perform additional value-added functions. There are typically three

key value-added points in the supply chain that occur between the growing stage and the

point of first shipment: growers, harvesters and shippers. Depending upon the

characteristics of a given product and farming company, organizations in the fresh

produce supply chain may perform any or all of the above functions.

  This section relies heavily upon industry interviews conducted by the author in order to represent the 
fresh produce supply chain as accurately as possible.  Unless specifically cited, data in this section was 
obtained through interviews. 


       Growers plant seed (or purchase transplants) for the purpose of growing a crop.

These crops are grown in many different fashions depending on the commodity. Some

crops are planted, tended and harvested within months while other crops require years of

growth and investment before the crop is mature enough to harvest. Crops are soil

dependent and climate dependent—that is to say, each commodity has a type of climate

and soil in which it grows most effectively or at all. Commodities require differing

quantities of water, specialized mechanization and vary in both the quantity and type of

input applications. Some crops are grown in rows, some in orchards. To summarize,

each of the hundreds of major produce commodities has distinctly different prime

growing conditions and requires in depth knowledge to successfully grow and yield fresh

produce. It is the responsibility of the grower to plant the crop, tend the crop and

successfully ready it for harvest.

       Depending on the size of the growing operation, growers may perform all of the

growing functions internally (fully vertical) or through the use of multiple service and/or

material providers to assist or augment the growing process. Thus, the growing process

often includes an entire subset of the supply chain related to seed acquisition, planting,

thinning or weeding, irrigating, pesticide application, fertilizer, etc. The nature of the

crops and crop rotations also dictate the type of service providers utilized throughout the

growing process. Farms have many crop components; permanent crops like nuts and

citrus trees last decades, while other row crops like cantaloupes and strawberries are

annual row crops and, depending upon the crop schedule, are rotated (i.e. the same crop

is not necessarily grown in the same field each year). In the case of many annual row

crops, individual growers seldom plant the same commodity in the same field each year.

This is because of soil conditioning, “compatible” neighboring crops and maximizing the

productive and effective use of the land. Growers don’t often run “melon farms” or

“cotton farms”—farms are distinctly different in the composition and makeup of a variety

of commodities each year to suit the best use of soil in a given climate. Growers must

consider neighboring farms and crops, application of inputs to those crops, and differing

qualities and quantities of water available to them (including distance from the source) as

each can have distinct impacts upon crops. Because of the nature of the business,

traceability back to an individual field will entail the recordkeeping of a myriad of

potential food safety hazards—each field is different, each crop brings with it its own

challenges (i.e. pest infestation, crop disease), and thus each year a single field will have

differing commodities with different inputs to produce a harvestable crop.


       Harvesting is the process of collecting mature produce for packaging and

shipping. The process used to harvest a crop differs from commodity to commodity.

Some commodities require hand picking or cutting from the host plant. Others are

harvested using machinery or a combination of human labor and machinery. In a field-

pack harvesting operation, the produce is gathered and then packed into boxes while still

in the field of origin. Harvesters who pack in the field also commonly perform functions

such as applying labeling to individual products, additional packaging for smaller

quantities than a full case (i.e. plastic clamshells in which berries are commonly placed

before being loaded into a box) or cleaning or brushing product and grading. In a shed-

pack operation, the produce is gathered in bulk and transported to an initial packing

station or location for processing. Harvesting has five key elements: severing the crop

from the plant/tree/vine, initial collection into bins, boxes, bulk trailers, etc. (prepacking,

which can involve transporting), final packaging in consumer, retail or food service

containers and palletizing with similar products and precooling (eliminating field heat)

prior to shipment.


              The initial location to which fresh produce is transported is most commonly

referred to as a shipping point or shipper. Depending on the commodity, a shipper may

cleanse or brush fresh produce, package bulk produce into cases (if the commodity is

shed-packed as opposed to field-packed) and repackage product. The shipper cools (if

required) the product to lengthen freshness and, thus, shelf life. Different commodities

require different methods of cooling and/or cleansing. For example, cantaloupes are

maintained at a temperature of 36-38 degrees Fahrenheit in order to most effectively

maintain freshness and extend shelf life while a broad spectrum of lettuce varieties are

maintained at 34-36 degrees Fahrenheit for the same reasons. Similarly, differing

technologies and methods are utilized for cooling—cantaloupes utilize forced air cooling

technology while the spectrum of lettuce varieties are top-iced. Some shippers are single

commodity focused while others specialize in a wide variety of commodities and must

provide infrastructure to deal with different cooling and refrigeration requirements. After

preparing the product for sales, the product is marketed and shipped to the shippers’


  Some shippers also perform additional processing (discussed further in following section) before selling
the product.

Summary	of	Growing,	Harvesting	&	Shipping	

       While each of the three functions adds its own distinct value to the fresh produce

supply chain, some companies perform more than a single function and others perform all

three. Some shippers conduct receiving, cooling, processing and marketing functions

with fruit procured from multiple growers (sometimes including themselves). Shippers

sometimes utilize the distribution and cooling facilities of another organization for their

product. It is also feasible that a single grower is distributing fresh produce to multiple

shippers. The following charts illustrate the common links between these three members

of the supply chain in the various forms described.











                                1                  2

  Grower also                                                 
harvests produce

                          Grower/Harvester       Shipper











                                  1                    2                          3

     Multiple Growers           Grower #1
    to Single Shipper,                                                         Shipper          
    Cooled at 3rd Party
     Storage Facility
    Owned by another                                                                            
                                Grower #2

                                                                           3rd Party Storage
                                Grower #3                                       Facility




       Across the broad spectrum of businesses growing, harvesting and shipping fresh

produce, a wide variety of these combinations is the reality of the supply chain. A more

realistic supply chain view, considering many growers, many harvesters and many

shippers is as follows.


           After the initial handling of the product by an initial grower, harvester and

shipper, the product is sold to the next member in the supply chain. A brief description

of each of the next potential supply chain members follows.

           Retail stores such as grocery chains are often the immediate purchasers of fresh

produce. Product is most commonly shipped to a distribution center which can either be

owned by the retail chain or by a 3rd party. Fresh produce is also shipped directly to retail


Foodservice	Providers	

           Foodservice distribution organizations buy fresh produce (among many other

items) and add value in the supply chain through consolidation and highly efficient

shipping operations to restaurants, cafeterias and other preparers of food.

           Depending on the size and strategy of a foodservice organization, fresh produce

may be purchased directly from suppliers and shipped to a distribution center for

consolidation and shipment to individual outlets with other items (i.e. restaurants).


           Brokers facilitate transactions between shippers and other members of the supply

chain by buying fresh produce from shippers and re-selling the produce to other members

of the supply chain. Brokers work with a diverse collection of shippers in order to

provide consistent supply to other members of the supply chain. In other situations,

shippers purchase product from competitors by way of a broker in order to fill an order

for a specific product in the event of a shortage.

Processors/Re‐packaging	and	Consolidation	

       Further processing may be performed on a wide variety of fresh produce before it

reaches the consumer. Many fresh vegetables and fruits are cut and repackaged in “ready

to eat” packages (i.e. bags of mixed lettuce, fresh cut produce trays). This processing is

sometimes vertically integrated into the original shipper while other organizations exist

solely for processing. Processors commonly utilize product from multiple shippers,

growers and fields to produce a bulk output.

       Re-packaging and consolidation of many fresh produce items is also common.

Organizations buy diverse commodities of product and provide a broad range of value-

added services such as repackaging in a private label, quality assurance and consolidation

of multiple commodities such their customers can go to a single supplier for all or many

of their fresh produce needs. Organizations that perform this function also commonly

utilize product from multiple shippers, growers and fields to produce the consolidated

supply for their customers.

Terminal	Markets	

       Terminal markets exist in major cities and provide a single stop for many smaller

retailers, processors, brokers, foodservice businesses and consolidators to purchase all of

their fresh produce needs at a single location. While many buyers of fresh produce that

visit terminal markets are small, major national organizations (retail, foodservice, etc.)

are also buyers at terminal markets. Many shippers and brokers of fresh produce have

open markets at the terminals and allow customers to inspect and purchase produce on

the spot. Produce is delivered daily to the market and buyers come to negotiate prices,

make purchases and take delivery. Terminal markets are a means for consumer outlets

(whether that be foodservice, retail or brokerage) to purchase wholesale produce for

resale to consumers.

Direct	to	Consumers	

              Farms also sell product directly to consumers in outlets such as farmers markets,

fruit stands and pick-yourself operations. Though this is a small portion of overall

produce sales in the U.S., it is important to note as a method of “farm-to-fork.”

Summary	of	Buyers	

              It is not uncommon for fresh produce to travel through multiple members of the

aforementioned list of potential buyers before actually reaching the consumer.

Consumers typically purchase produce through retail outlets, foodservice and farmers

markets (See Figure 2.1). The chart on the following page illustrates the common

complexity in this portion of the supply chain.

                     Where Consumers Purchase 
                          Fresh Produce

                        32%                                          Retail/Grocery
                                                                     Farmers' Markets

Figure 2.1: Where Do People Buy Produce8

     (The Great Atlantic & Pacific Tea Company, Inc., 2010) 


       It should be noted that an integral part of the fresh produce supply chain is the

transportation between each member of the supply chain. Transportation options vary

based upon need, strategy, space, distance and perishability. Trucks and trains are the

most common methods of transport between members of the supply chain for domestic

product. For imports and exports, air and ship travel are common methods of transport

(depending on the geographic distance between the import and export countries).

The Supply Chain: Visually

       In order to better describe the flow of product through the supply chain, the

following charts are intended to provide a visual idea of the flow of product from

shippers through the rest of the supply chain.

                                                                                                 Direct to




    Shipper                                                                                Foodservice

              Processor / Reconfigurer





                                                  Distributor                 Retailer



                                   Part	III	
    Current	State	of	Traceability	in	the	U.S.	Fresh	Produce	Supply	Chain	

Perishable	Agricultural	Commodities	Act	

        The regulation of traceability requirements for fresh produce dates as far back as

the Perishable Agricultural Commodities Act (PACA), which was enacted in 1930. The

intent of the act was the promotion of fair trading practices in the fresh produce and

vegetable industry. PACA mandates significant recordkeeping requirements regarding

who buyers and sellers are, types of products sold, quantities that are sold and the method

in which the transaction takes place. Records must be maintained for a minimum of two

years and link the original source of the produce (the grower) to the transaction in which

it was sold by any intermediary (as a first handler of that produce) (Golan, Krissoff,

Barry, Kuchler, Calvin, Nelson, & Price, 2004).

Country	of	Origin	Labeling	

        First incorporated into the Farm Security and Rural Investment Act of 2002,

Country of Origin Labeling (COOL) requirements were designed to provide consumers

with more information regarding the origin of the commodity. COOL covers all fresh

fruit and vegetable products, with the exception of those products used as ingredients in a

processed food item or as a combination with other food products (i.e. orange juice,

bagged salad, mixed nuts). The requirements stipulate that a label, placard, mark, stamp,

etc. is clear and visible on the commodity (whether on the packaging, display, bin, etc.) at

the final point of sale. Records must be kept that identify both the immediate previous

source and immediate recipient of the food product in order to provide accountability for

the transfer of country of origin through links in the supply chain. There is flexibility in

the types of records companies can maintain to provide the link through the supply chain

(i.e. there are a range of documents in which the transfer of country of origin labeling can

be accomplished) to accommodate firms with differentiated procedures as well as those

already maintaining records. Records must be kept for a minimum of two years (Golan,

Krissoff, Barry, Kuchler, Calvin, Nelson, & Price, 2004).

       Food service businesses including restaurants, food stands those similar are

exempt from the labeling requirements. Additionally, any grocery store whose total

invoice value of fresh produce and vegetable sales is under $230,000 is also exempt from

the requirements (Golan, Krissoff, Barry, Kuchler, Calvin, Nelson, & Price, 2004).

The	Bioterrorism	Act	of	2002	

       The body of regulation currently in place, most often cited and the most stringent

for traceability requirements in the fresh produce supply chain is the U.S. Bioterrorism

Act of 2002. The core tenet of the Bioterrorism Act in terms of traceability is the

requirement that members of the supply chain maintain a “one-up, one-down” record of

fresh food products transported domestically and imported into the United States. That is

to say, each member of the supply chain must record and maintain records of the

immediate source of the product and the immediate recipient of the product. The FDA

has regulatory power over food safety through the act and, since the legislation was

passed and put into effect, the FDA has released guidance for the industry to describe

who must keep records, what records they must keep and when they must report records

to authorities (Levinson, 2009).

              Any “persons” (where “persons” includes individuals, partnerships, corporations

and associations9) who “manufacture, process, pack, transport, distribute, receive, hold,

or import food” in the United States, as well as foreign food transporters must maintain

the records required by the Bioterrorism Act (McEntire, et al., 2010).

              There are a few notable exclusions from transaction recordkeeping requirements,

notably farms, distributors who sell directly to consumers, restaurants, transfers of food

within multiple facilities in a vertically integrated company, outer food packaging and

food processed for personal consumption10. Additionally, there are some establishments

that, while exempt from recordkeeping requirements, must adhere to record access

requirements. These businesses include food retailers with 10 or fewer employees, inner

food packagers and nonprofit food establishments (McEntire, et al., 2010).

              In general, records must be available for examination within 24 hours of request

and must be kept for between 6 months and two years, depending on the type of food.

(McEntire, et al., 2010).

Reportable	Food	Registry	

              The Reportable Food Registry was established as an electronic means of reporting

FDA-regulated food that “may cause serious adverse consequences or death of humans or

animals.” Companies are required to submit, by means of an electronic portal, a

company identifier, quantity of food, information on the packaging (i.e. UPC, brand

names) and lot-specific information if available. Immediate source(s) and recipient(s)

(based upon those companies that are covered under the record keeping requirements of

the Bioterrorism Act) as well as contact information may also be required. Each case
     (U.S. FDA, 2005) 
      (U.S. FDA, 2005), (McEntire, et al., 2010) 

reported to the registry is given a unique identifier which is then used by the contacts

provided to input the same data. This information is not public to others but is subject to

the Freedom of Information Act (McEntire, et al., 2010).

Future	Directions:		The	Food	Safety	Enhancement	Act	of	2009	

        The Food Safety Enhancement Act of 2009 covers a broad range of issues in the

area of food safety and traceability. As cited in the introduction to this paper, the act

calls for a national tracking system for the food industry that would allow for the ability

to more rapidly trace a food product back to its source. With senate acting on broad

legislation surrounding food safety and fresh food traceability in the wake of major

outbreaks, it is increasingly clear that either the industry will act to sufficiently trace

produce for the minimization of outbreak spread or the federal government will

implement its own policy.


                                           Part	IV	

    The	Traceability	Investigation:	Agencies	&	Roles,	Recent	Outbreaks,	
            Lessons	Learned	and	a	Foundation	for	the	Future 

The	Current	State	of	Traceability	and	Outbreak	Investigations	

         The traceability process starts with the reporting of illnesses from local and state

agencies. When a major foodborne illness outbreak occurs, a number of major

governmental bodies become involved to perform an investigation and, potentially, a

recall. In order to assess the current state of traceability in the fresh produce supply

chain, we first look to understand the process by which these bodies work with industry

and consumers to trace a product as well as the systems they utilize. We then turn to

several of the most recent major outbreaks in fresh produce and introduce an analysis of

what worked, what didn’t work and what the agencies have stated is lacking. Finally, we

will come to a set of future goals that any next generation traceability system would

require in order to be effective.

Systems	Utilized	for	Outbreak	Monitoring	

         There are a multitude of systems and organizations of professionals that play a

role in the outbreak detection, investigation and resolution processes.

System             Function
PulseNet           An electronic monitoring system that utilizes results from DNA
                   “fingerprinting” to match specific strains of illnesses across the United
                   States. The Center for Disease Control (CDC) utilizes PulseNet to
                   identify clusters of illnesses and outbreaks (National Center for
                   Infectious Diseases, 2009).
OutbreakNet        A group of health officials at the state level who investigate food borne
                   illness and share information throughout an outbreak (Porter & Lister,
                   2006, p. 3).

CDC’s Health A nationwide program that specializes in the communication,
Alert Network dissemination of information and education and organizational
              infrastructure with which to defend against health threats. Includes
              warning system to rapidly broadcast warnings or threats to local, state
              and federal governments as well as media (CDC, 2001)
Epidemic      The CDC’s secure, web-based network that facilitates communication
Information   between CDC, state and local health departments, public health
Exchange      professionals and poison control centers (Porter & Lister, 2006, p. 3)
Reportable    An electronic reporting mechanism for industry to report an article of
Food Registry food when there is reasonable certainty it will cause serious adverse
              health consequences (FDA, 2010).

State,	County	and	Local	Health	Agencies	

       County and local health departments are the places in which patients are first

interviewed regarding their illness. It is imperative that the information garnered during

these interviews gives investigators enough insight to initiate or continue an investigation

into a foodborne illness. If the information from these interviews cannot help

investigators narrow down how that patient may have gotten sick, there is no need for

traceability as there is no way to understand what product to trace. Competing needs for

resources and any other major health epidemics (i.e. H1N1) can cause foodborne illness

investigations to be prioritized at a lower level. As a result, there is not always

consistency from the county and local health departments in their ability to perform

comprehensive interviews at all times. Additionally, patients often simply don’t

remember what they ate. In reference to fresh produce, if they do remember what they

ate, they don’t remember what bin they picked a tomato from or exactly which day they

bought a head of lettuce. This lack of reference from consumers in regard to what they

ate makes transmission of useful information to national authorities difficult and/or

impossible (USDA, 2009, pp. 49-58).

CDC	–	Centers	for	Disease	Control	

              The primary role of the CDC is problem identification and source implication. To

this end, the CDC performs disease surveillance, outbreak detection and investigation and

education of public health staff. Additionally, the CDC provides support for states in

terms of epidemiologists, coordination and lab support. States do not report or answer to

the CDC—it serves only a supporting role (USDA, 2009, pp. 58-69).

FDA	–	Food	and	Drug	Administration	

              With the exception of meat, poultry, dairy and egg products (which are regulated

by the USDA11), the FDA is responsible for the safety of all food products in the U.S.,

including seafood12. The FDA has the authority to regulate food safety through the

Federal Food Drug and Cosmetic Act as well as amendments to the act granted through

the U.S. Bioterrorism Act of 2002. The FDA has the authority (and has issued guidance

regarding) recordkeeping requirements for those who “manufacture, process, pack,

transport, distribute, receive, hold, or import food.” In the event that the FDA has

reasonable belief a food product presents a serious health threat, it has the authority to

inspect those records. Records officially requested by the FDA must be presented in no

more than 24 hours (Levinson, 2009).

              It is important to note the inherent conflict in the regulatory powers created by the

granted authority of the FDA to regulate food safety through the Federal Food, Drug and

Cosmetic Act as well as the U.S. Bioterrorism Act. While the acts encompass all “food”

products, the FDA does not have the responsibility of food safety for all food products.

  www.fda.gov; The National Oceanic and Atmospheric Administration also manage a seafood inspection 
program, commonly referred to as the Department of Commerce Seafood Inspection program.  See 

This conflict creates an environment in which one federal agency has regulatory

authority over a body of foods it does not have the responsibility to ensure the safety of.

The	Trace‐back	Procedure	

       Typically, the time it takes from exposure to confirmation that an individual

illness is a part of a specific outbreak is two to three weeks. DNA fingerprinting and

reporting through a system such as PulseNet is employed to match a specific illness to a

specific outbreak. The process of confirming an illness as part of an outbreak involves a

number of steps, contributing to the lengthy timeframe.

    1. Eat contaminated food to first symptoms:                                      2-4


    2. First symptom to seek medical care & sample taken:                            1-5


    3. Initial lab diagnosis                                                         1-3


    4. Shipment of sample from lab to state authorities for DNA fingerprinting       up to 1


    5. DNA Fingerprinting Analysis and Match                                         1-4


    6. Reporting to CDC (during outbreak)                                            1 day

    Assuming a full week for shipment to state authorities for DNA fingerprinting, the

best case scenario from the time of consumption to confirmation at the CDC level is 13

days (Figure 4-1) but can take up to 24 days (Figure 4-2)


Figure 4‐1:  Best Case Confirmation 


Figure 4‐2:  Worst Case Confirmation

         The CDC performs surveillance on an ongoing basis to detect illness outbreaks.

They do this by monitoring for any increase in the number of infections above a baseline

for a period of time. This increase is referred to as a cluster. Approximately 25% of

clusters are detected through reporting from state and local health departments. The

remaining 75% of clusters are detected through Serotype results as well as PulseNet.

When a cluster is positively identified, OutbreakNet is contacted to initiate an

investigation (USDA, 2009, pp. 58-69).

         At this point, an epidemiological investigation is undertaken in which the first

step is to gather data through interviews with ill patients. Some interviews are done

through standard questionnaires while others are in-person interviews with open ended

interviews. The goal is to gather as much history of what the individuals have eaten as is

possible. After data is gathered, the process moves to hypothesis forming (what product)

and testing (how statistically feasible) to determine whether or not a particular type of

food should be implicated or cleared (USDA, 2009, pp. 58-69). In many cases, food has

already been thrown out by the consumer or is no longer on retail store shelves (USDA,

2009, p. 77). To this end, those retailers who keep consumer information through

consumer cards have helped to narrow down the specific products that were purchased

(USDA, 2009, p. 101).

       While these tests are being performed, a trace-back investigation is often going on

to test sources of food potentially implicated. None of the processes happen in a vacuum;

the trace-back and epidemiological investigation normally overlap and changes in one

investigation can assist in the other (USDA, 2009, pp. 58-69). The FDA gets involved in

the epidemiological investigation as well as the trace-back investigation.

       The FDA’s involvement begins when a cluster of illnesses believed to be caused

by food is reported from the CDC or state partners. The FDA aids in the epidemiology

process by helping with hypothesis generation and providing information that may assist

to this end including historic sampling, seasonality of products and industry practices. A

specific problem often encountered in the investigation is the lack of laboratory testing

available for contaminated food—most often the suspected products have been consumed

or disposed of. At the point in which FDA can statistically implicate a specific product,

the recall process is initiated in an effort to remove as much of the product from store

shelves as is possible. For a trace-back to be successful, the epidemiologic investigation

must be productive; without an effective epidemiological investigation in which certain

foods can be implicated, there is no need to trace-back. Dates of illness onset help FDA

investigators determine the dates of records necessary to perform a trace-back (i.e.

production records, shipment records) (USDA, 2009, pp. 77-84). To that point, the

existence of records isn’t always the issue; sometimes the issue is that the quality of the

records that exist is insufficient to reach a justified implication (USDA, 2009, p. 73).

       The FDA cites a few challenges to their procedures. First, a widespread outbreak

with small numbers of cases across multiple states can make the implication of a specific

food difficult or impossible. In this scenario, extra coordination must be established to do

interviews and questionnaires for food implication. Another difficulty lies within

identifying suspect ingredients—patients don’t often know what ingredients were used in

the food they ate (USDA, 2009, pp. 77-84).

       Implication of a single source is ultimately a challenging task. In an outbreak,

time is of the essence—the longer a source implication remains a mystery, the greater the

possibility of tainted product in commerce remains a threat. Considering that it is

generally impossible to predict an outbreak (i.e. knowledge of an outbreak before until

either an illness or tainted sample is reported), tracing the problem back to a source as

quickly and accurately as possible is the key to removing other potentially tainted product

from commerce. As such, the FDA also works backwards to assist the epidemiological

investigation with what they call epi investigations—they perform a trace-back to limit

the type of food that could be implicated to help epidemiologists with their questioning.

For example, the implication of “melons” over the summer can be misleading if there is

not more specific detail. The FDA works to establish the types of melons (i.e.

cantaloupe, honeydew, watermelon) that are in harvest as well as the organization from

which they are sourced (USDA, 2009, pp. 77-84).

       In its challenges, the FDA points to a few areas that would improve the ability to

implicate a specific food. First, all potential parties should be engaged early in the

process with the same information. Second, the process of implicating a specific product

more rapidly needs to be identified. Third, the evidence required to implicate a particular

product should be examined to ensure that the method in which “implication” is

measured is the most effective. If it isn’t, new methods of associating an outbreak to an

implicated food should be employed (USDA, 2009, pp. 77-84).

       In the end, an effective trace-back relies upon three primary things: the epi

information, the exposure and the distribution. The epi information provides the link

between different pieces of the investigation (i.e. lab results, PulseNet matches, etc.).

The exposure is the information that proves the patient was, indeed, exposed to a source.

The distribution is the linking of the food product through its dispersed route to ensure

that it is plausible that the patient could have been exposed to the source. As such,

industry is heavily relied upon to provide this data as accurately and quickly as possible

to limit the distribution of contaminated product by quickly implicating a lot and

removing it from commerce. A key point to the trace-back investigation is the uniformity

and standardization of data and common records throughout the supply chain. Uniform

data makes following a specific product through the supply chain (each party that handles

the product) much quicker to verify. This data includes the date and time the product was

produced, exact name and type of product, quantity of product, sell by/use by/production

code, other information used to identify product, and lot specific coding from suppliers.

Knowing accurately that a specific product was at a specific spot at a specific time is key

to determining whether that product is implicated or not. Inaccurate data causes the

process to take longer and keeps contaminated product in commerce longer (USDA,

2009, pp. 91-97). In some cases, comingling and differing in-house recordkeeping

practices and reports cause the actual product being traced to be lost through the

distribution chain13. This procedure of gathering the data and building a history to

confirm an implicated product can take weeks and is complicated further when the

original illness stems from something like a fruit salad in which there are multiple

possible sources of contamination (USDA, 2009, pp. 99-102).

2006	E.	coli	O157:H7	Outbreak	in	Bagged	Spinach	

              Recent major outbreaks have served as the impetus for enhanced traceability

across the supply chain. As such, examination of the events leading to and throughout

investigations of major outbreaks and shortfalls throughout the investigation of these

outbreaks is an important point of analysis in determining the correct path forward with

traceability. One of the most extensive recent outbreaks was the 2006 E. coli O157:H7

outbreak in bagged spinach.

       The E. coli O157:H7 outbreak in bagged spinach in the summer and fall of 2006 was

extensive in its geographical distribution of infection, severity of illnesses and deaths and

financial fallout to the leafy greens industry. In the end, the outbreak spanned 26 states,

causing 199 illnesses, 102 hospitalizations, 31 cases of HUS (a condition that can lead to

kidney damage and death) and 3 deaths. Financial impact upon the industry was

estimated at $100 million by November of 2006 (Porter & Lister, 2006).

   This is to say, each company’s reports and documents have a different form and differing levels of detail 
such that following the route of a specific product from company to company becomes a difficult task. 

       In the 2006 outbreak, the average time for confirmation was 15 days (Porter & Lister,

2006). Conversely, the average shelf life of a bagged spinach product is 14-16 days14.

The FDA and CDC were first alerted on September 8, 2006 of four illnesses in Wisconsin

and immediately began an investigation. By September 12, 2006, PulseNet was used to

confirm that each of the cases had identical E. coli 0157:H7 strains in addition to multiple

patients from other states. Five days after first notification (September 13), fresh spinach

was first suspected as the source of infection based upon the investigation in progress.

On September 14, based solely on the epidemiological investigation and not traces of E.

coli found in fresh bagged spinach products, a consumer alert was first issued by the FDA

and CDC in which consumers were advised to avoid consuming bagged fresh spinach. A

first voluntary recall was initiated by Natural Selection Foods LLC of San Juan Bautista,

CA on September 15 and a second voluntary recall was initiated by River Ranch of

Salinas, CA on September 17. River Ranch recalled packages of spring mix obtained in

bulk from Natural Selection Foods. A third voluntary recall came from RLB Food

Distributors, L.P. of West Caldwell, New Jersey on September 20. On September 23,

two more voluntary recalls were announced, both with spinach products originating from

Natural Selection Foods. It wasn’t until September 29, 21 days after first notification,

that the FDA confirmed it had implicated the source of the contamination from Natural

Selection Foods (Porter & Lister, 2006), a processor, packer and shipper (Chadwell,


       The major breakthrough that led investigators to know the details of the product

implicated were codes on the side of the bags, indicating (among other things) the date it

was bagged, worker shift and packing line. The implicated product had been bagged on
      Obtained through interview data 

August 15, which led investigators to sample fields in the nine different farms in which

bagged spinach was being packed on that date. Ultimately, the trace-back was effective

only because of a mix of epidemiologic and laboratory investigation, as well as multi-

agency cooperation (Porter & Lister, 2006). 

2008	Salmonella	Outbreak	in	Peanut	Butter	

        On November 25, 2008, an investigation began as the result of a growing cluster

of salmonella infections. By January 28, 2009, 529 people in 43 states were reported

infected with the same strain of salmonella, 116 had been hospitalized and the infection

may have contributed to 8 deaths. The CDC began its investigation with open-ended

patient interviews conducted by both the CDC and state and local health departments.

Two products were initially reported as likely eaten by patients in the seven days prior to

consumption: chicken (86% reported eating) and peanut butter (77% reported eating).

While the reported results for eating chicken were in line with historical averages (85%),

only 59% of respondents historically reported eating peanut butter. An investigation with

patients in Minnesota revealed that of the three institutions some of the patients had eaten

at (two long term care facilities and one cafeteria), there was a single common distributor

and the only common item purchased between the three was a single brand of King Nut

creamy peanut butter. After obtaining an open container of the King Nut creamy peanut

butter from one of the facilities, the Minnesota Department of Health confirmed a

positive match for Salmonella on January 12, 2009. By January 28, 2009, there were 16

clusters of illnesses; each cluster was in an institutional facility and each facility

exclusively purchased King Nut brand creamy peanut butter from Peanut Corporation of

America in Georgia (CDC, 2009).

        While it was initially believed that the outbreak was limited to peanut butter

distributed to foodservice institutions, further investigation uncovered that Peanut Butter

Corp had shipped the contaminated product to approximately 2,100 accounts in the form

of dry and oil-roasted peanuts, granulated peanuts and peanut meal. As of January 28,

2009, 431 peanut butter containing products were recalled (CDC, 2009).

        The outbreak of Salmonella in peanut butter and peanut butter-containing

products provides two valuable lessons in traceability. First is the need for rapid

association of foods purchased or consumed at a large variety of grocery stores,

restaurants and institutions for the purpose of statistical analysis; the ability to associate

commonalities between multiple foods in a myriad of illnesses is important to do rapidly.

Second is the issue of ingredients in food; the rapid ability to associate source

contaminated products with other foods that may use the product as an ingredient is


Supply	Chain	Challenges	in	the	Trace‐back	Process	

        In addition to the cases presented above, further testimony and releases from

governmental bodies charged with tracing back contaminated food products to their

sources provide grounding for the establishment of a set of Future Goals of Traceability.

These goals shall serve as a body of comparison against proposed traceability initiatives

intended to enhance traceability throughout the supply chain.

Farm	to	Fork	Revisited:		Where	are	we	today?	

        Given the difficulties that the FDA faces in traceability, it is important to

understand the current methodologies employed by various members of the supply chain

for traceability15. It should be noted that the following section is organized by function

and is not necessarily indicative of an individual group of the supply chain—i.e. some

organizations perform more than one of the following functions in the supply chain.

Original	Growers/Shippers	

              In the interviews conducted, it was the original shipper’s responsibility to apply

any marking or encoding of barcodes for traceability purposes. All shippers had a

traceability system in place that allowed them to trace product back to the original field,

date of harvest and specific product characteristics on a pallet level. The quantity of each

pallet (also tracked) differs by shipper and is based upon commodity characteristics,

marketing requirements and weight requirements. Some cartons are filled with raw

product while other cartons are filled with additional smaller packages such as clamshells

for individual sale at the retail level. In each case, shippers encode the information into a

barcode and track the pallet through the warehouse until shipment, at which point the

barcode is electronically linked to the bill of lading such that each pallet shipped on an

order can be traced back to its original source. In many cases, additional marking (with

or without a barcode) is placed on the individual carton identifying the date and lot of the

product inside.

              Across shippers, access to existing traceability data was rapid. Reported times of

access varied between five minutes and one hour. The shippers interviewed used a wide

variety of backoffice software to run their operations and, as such, formatting and

presentation differed though base data elements including shipment information, lot-

   This section is based largely upon interview data with growers, shippers, processors, brokers, 
distributors and retailers as well as FDA presentations from other members of the supply chain (available 
publicly as cited) 

specific data, dates of harvest, commodities and other identifying product characteristics

were widely reported. In all cases, shippers who sell product raw or in bulk (i.e. no

additional processing, cutting, or reconfiguration, etc.) maintain internal records

electronically of product characteristics and lot information.

        A common function among those shippers interviewed is the purchase and resale

of another shipper’s product in the event of a shortage. In all cases, product that was

purchased was received and inventoried as such and traceability data tracked with the

identity of the original source (immediate prior source) as well as identifying product

characteristics (commodity, brand, quantity, etc.) This data was traceable via the same

mechanisms used for tracing any other produce shipped through the facility. Almost

without exception, this information recorded and/or retained did not include a lot or batch

number with bulk produce.

        Preparation for traceability events was common across all shippers. Each shipper

conducted a minimum mock recall each year while other shippers perform multiple mock

recalls each year.

        An issue shippers often deal with is the wide variation in the products that they

provide for sale; many commodities of produce come in multiple varieties, styles of pack,

grades and brands. Differing commodities most often have incredibly different styles of

growing and harvesting. This creates the need for a system of traceability that is able to

work in a variety of settings (i.e. field and facility application), climates (i.e. humidity,

temperature, exposure to moisture) and worker skill/literacy. As the supply chain

becomes increasingly complex and demands for customization increase, the ability to

accurately trace each “flavor” of product becomes more difficult and cost intensive. As

explained by one shipper, the “nature of the product” limits how fresh produce can be

traced (as well as the level of detail).


        Additional processing is not uncommon for fresh produce before it reaches

consumers. Additional processing includes cutting, comingling, repackaging and

rebranding. While all processors interviewed had the ability to trace product from

source(s) to finished lots, a mix of electronic and paper records exist to varying degrees.

In one case, almost the entire process was stored electronically, save a few specific

documents that hadn’t been integrated into the data records. On the opposite end of the

spectrum was an almost fully paper based approach that had full visibility from beginning

to end in paper records. In the case of those processors who were engaged in the business

of cutting and/or comingling of products, it was commonplace that coding on the

individual packaging delivered to end customers was traceable back to the original source

lot(s) in which product was sourced.


        Brokers maintain records as received from shippers and customers. As brokers

typically don’t physically handle the product, they are dependent upon the immediate

prior source and immediate recipient for verification of lots and product characteristics.

In interviews, brokers had experienced and participated in recalls on a number of

occasions (both voluntary and mandatory). In those events, the brokers used a mix of

electronic and paper based documentation to link the source produce being recalled with

the immediate subsequent recipient(s) of that product. However, in most cases the

reliance in identifying product to recall lies with the immediate prior source as lot

information is not commonly identifiable at the broker level16.


              Distributors (both foodservice and wholesale) have differing abilities in terms of

tracing product back. Based upon interviews, the level of data retained was similar to

that of brokers. Product is tracked by some unique reference that identifies the original

order for that particular product (normally the PO#) dates associated with barcoded pallet

tags. In order to maintain traceability when pallets are broken down and shipped to

individual customers, the quantities of each individual product as well as the originating

PO# (from the original pallet) are recorded for the new pallet(s), which consist of a

combination of product commodities and varieties. In recall situations, this method has

proved (through use) effective in quickly determining and contacting any immediate

recipients of individual product specified by the immediate prior source as contaminated.

This method requires that the immediate prior source is able to effectively locate the

immediate recipient of individual contaminated lots of product as well.

              As distributors’ primary focus is on providing a singular point of contact and

purchase for a wide variety product needs, there is no additional information regarding

the product generated at the distribution level; distributors are highly reliant upon data

received from prior sources as to the origin and specific product attributes. In that

context, information that is received must be readily available on the exterior of the

packaging—this includes any information relevant or necessary to perform a trace-back

or recall procedure. An effective recall or trace-back additionally requires that the

   This was true with shippers who bought and sold fruit (brokerage) to fill their own orders in the event of 
a shortage 

information provided originally on the packaging is uniform and consistently provided by

suppliers to the information communicated by prior sources or subsequent recipients of

the product(s) implicated such that an accurate match can be made (IFDA, 2010).

              In order to be effective and profitable, distributors must be incredibly efficient in

massive17 storage facilities that receive, on average, approximately 500,000 cases of food

products each week. These cases are received and stored by pallet and then re-distributed

to individual stores based on complex orders (at the case level). This ability to correctly

distribute, at the case level, complex orders of a wide variety of foods to each individual

store is the value that is added through the distribution process (FMI, 2010).


              As with distributors, retail outlets must efficiently handle a massive number of

products—a single grocery store may have 75,000 SKU’s at any given time (FMI, 2010,

p. 2). The sheer volume of food products makes traceability in any detail a daunting

concept. Based upon interviews, grocery stores largely do not have any internal

recordkeeping as to individual lots received through distribution channels (including their

own). Considering the difficulty inherent in tracing lot-level product through a DC,

many of the largest grocery stores (who are vertically integrated) are not required to trace

data from beyond the point of receipt at the initial DC. Similar in many ways to brokers

and foodservice distributors, retail outlets rely upon the immediate prior source of the

product for lot-specific traceability purposes.

              Interviews shed some light on the retail distribution process as well. As product

is received, the pallets are associated with the original order on which they were placed

(i.e. a PO#). By tracking that product through the facility and then linking it to the
      Median size of a United States DC in 2007 was 583,000 square feet (FMI, 2010, p. 3) 

outbound store, there is some indication that whatever lots were on the original shipment

are the outer bounds for product that could have been contaminated. This is, however,

complicated when product from multiple shipments, each potentially consisting of

multiple lots are shipped on a single order to an individual store from the distribution



                                         Part	V	

                           Future	Goals	of	Traceability 

Analysis	Methodology	

       There are a number of future goals that are key to expediting the trace-back and

recall process in the fresh produce industry. After analyzing current regulations, analysis

of investigations, analysis of current recordkeeping and supply chain practices and

reviews of regulator releases on trace-backs and recalls, these goals were created as a

baseline comparison for future traceability efforts in fresh produce. There is one

overarching theme and five further goals that are more elemental to the theme of future

traceability systems. The goals are neither the “specific” output of individual interviews

nor directly pulled from research. Rather, they are an accumulation and synthesis of data

presented in previous sections. Initial hypotheses were created based upon a holistic

view of the material (including interviews and secondary research materials as cited) and

then tested against the material on a granular basis. That is to say, the goal hypotheses

were tested against interview and research data in an iterative fashion such that they met

the needs for enhanced traceability as presented in the material from previous sections.

When research and publicly available information has been utilized in this analysis, it is

cited as such. Other elements of data or implied “fact” are based upon interview data.

Analysis of those data elements and facts for each goal are based upon conclusions drawn

by the author therein.

Future	Traceability:		The	Consumer	and	Food	

       Any future traceability system must, first and foremost, address the issue of

determining what food a consumer has eaten and further which of those foods is the

cause of illness. While this is not to say that the interview process currently employed is

ineffective, it is time consuming and incredibly subjective. A more refined approach to

connecting what consumers ate to what made them sick is the first step in accomplishing

more rapid trace-backs and recalls. To this end, a number of methodologies have been

employed with mixed success. Consumer purchasing or rewards cards can be an

effective means of limiting the scope of a recall. In the event of a widespread outbreak,

knowing where consumers shop as opposed to relying on recall of what they ate (be it at

home, in restaurants, cafeterias, etc.) and utilizing available purchase data can statistically

limit the scope of the investigation of good records are maintained. To cite a comment

from one interviewee, an effective traceability system will aid investigators by providing

them with the right questions to ask rather than chasing every possibility in a time

sensitive investigation.

       Additionally, a system in which specific products and/or lots of fresh food

products can be limited by consumer outlet (i.e. which lot-specific products were shipped

to each individual outlet) can also provide a rapid mechanism for correlating

commonalities between outbreak cases. Armed with this data and the proper ability to

trace each of the suspected products through the supply chain to common spots (whether

a distribution center, trucking service or grower/shipper) could rapidly yield statistically

valid results for use in validating the origin of the illness and recalling any further

affected product.

       The use of data modeling to map commonalities between products has been tested

and proven to be both successful and rapid, though the actual process of gathering and

standardizing data (a necessary process) caused huge delays in processing. Researchers

working with Harvard University spent weeks “cleansing” the data such that the reported

information was fed to a computer modeler in a standard fashion. Once standardized, a

visual trace-back and implication was possible in an incredibly rapid manner. Thus, the

more data available in a pre-standardized format (such that investigators need not

“cleanse” the data), the more rapidly one can expect statistical analysis to yield results for

both implicating an individual source product while eliminating many others (McEntire J.

C., 2009).

       To this end, the following set of goals are set forth to establish of a traceability

system that allows investigators to utilize data analysis and statistics as an augmentation

for interviews with consumers regarding the food they have consumed.

I.     Standardization	of	Datasets	

Uniqueness	of	Identifiers	

       In truth, the FDA itself has indicated that a lack of uniqueness (or existence) of lot

or code information makes any trace-back much wider in scope and, subsequently, more

complicated and timely (USDA, 2009, p. 35). The level of detail that should be captured

is the primary issue. As evidenced through grower/shipper interviews discussing a wide

variety of commodities, the growing, harvesting, handling, processing and distribution of

each unique commodity can vary so dramatically, there is no single level of detail (i.e. by

field or by day) that most accurately and efficiently fits each process and accomplishes

the goal at hand. To this end, interview data were used in this to examine the level of

specificity that would allow the narrowest possible recall in which all possible

contaminated product could be effectively identified. That is to say, the level of detail in

which the least financial damage is incurred by the supply chain that would

simultaneously provide the breadth necessary to remove all potential contaminated

product from commerce. The level of uniqueness in identification comes down to two

primary categories of information: common nomenclature and common level of detail.

    1. Common	Nomenclature		

    The application and use of a common language for fresh produce is necessary. The

Harvard Business Tomato study cites that lack of consistency in the data presented causes

major delays in rapidly following product through the supply chain (McEntire J. C.,

2009). Use of the same nomenclature to describe individual products is essential to

performing a rapid trace-back if technology is to be employed for more rapid data mining

techniques. However, a technology vendor put it most succinctly that when dealing with

this nomenclature, the specific means of delivery are irrelevant. That is to say, utilization

of whatever technology or method of transfer that is most readily adopted and

implemented across the supply chain is more important than picking that technology or


    As an example, this nomenclature is analogous in many ways to a language: if the

first member in the supply chain only speaks English, the next only speaks Spanish and

the third speaks only Chinese, rapid communication is virtually impossible; a traceability

nomenclature should be the same across the supply chain such that each member of the

supply chain is able to read and understand the exact same meaning as the next. This

nomenclature should be universal and allow for uniqueness on an international scale

(considering the high degree of fresh produce imports into the U.S. food supply). This

allows the FDA to skip the step of “interpretation” from multiple languages and instead

focus upon commonalities in source implication.

    Interviews revealed that perhaps the most important part of a traceability system in

terms of speed and accuracy is providing regulators with enough information such that

they know both the right questions to ask as well as the right people to ask. To this end,

through analysis of interviews and investigation of the current traceability process in Part

IV, there seem to exist three variables of key importance that need to exist within this

nomenclature: unique company identification, discernable product characteristics and lot

or batch information.

               a. Company	Identification	

               The ability to identify companies in a uniform and, most importantly,

       unique, fashion is paramount in any traceability nomenclature. Issues in

       identifying which company handled the product or grew the product drastically

       limit the scope of the investigation. As one interviewee pointed out, following the

       paper trail sometimes means following the financial flow of products rather than

       the physical flow.

               By way of interpretation, this means that knowing the unique physical

       locations that the product travels through allows analysis of commonalities among

       multiple implicated products. The outcome of a limited investigation means that

       the speed and ability to implicate a specific product (and, simultaneously exclude

       others from the implication) is limited only by the specificity of the products

       identified through consumer outlets. In an industry where private labels are

       packed by multiple companies, identification of the specific company from which

       the fruit originated as well as those facilities through which the product passes is


           Additionally, this interpretation suggests that the use of unique

    identification for each company is an effective method of following the physical

    product through the supply chain (i.e. the physical path the product takes as

    opposed to the monetary path in which dollars flow through the supply chain).

    Standardizing datasets throughout the supply chain requires that each member of

    the supply chain who handles product be identified uniquely and tracked

    uniformly through the process. In this way, the immediate previous source of the

    product (the prior firm who physically handled the product) can be recorded as

    well as the immediate subsequent recipient (the subsequent firm who physically

    handles the product).

           b. Discernable	Product	Characteristics:	        	

           Discernable characteristics allow each level of the supply chain, including

    the end consumer, to identify and distinguish the product from any other. This

    detail can include commodity, variety, brand, packaging, and post-harvest

    processing (i.e. cut or bagged)—the important point is not the level of detail as

    much as the commonality of detail across the supply chain and to the consumer.

    This level of detail may or may not be referred to by each member of the supply

    chain or on individual paperwork used throughout the supply chain (i.e. Bills of

    Lading, Invoices). While not every member of the supply chain has access to a

    visual of products still available for consumption (i.e. product that is still on

    grocery shelves but is no longer in pipeline inventory), it is also the case that not

    every member of the supply chain has access to existing electronic data (i.e.

    consumers at home, grocery clerks in remote areas without constant access to

    electronic means of communication). By matching the electronic identifier with

    the visual identification that makes the product discernable from another,

    implicated fresh produce can be communicated throughout the supply chain as

    well as with consumers and regulatory bodies in a uniform manner.

           c.   Lot	or	Batch	Information	

           Lot or batch information should be unique to the degree that it separates

    one group of product to the maximum scope in which cross contamination could

    occur. For example, in some commodities this would constitute identifiers at the

    level of an individual field where contamination could be caused in a small

    portion of the field (i.e. droppings from wild pigs) or could be the cause of a

    contaminated water source. The level of uniqueness for the lot should also

    represent the lowest common denominator of potential contamination (as in the

    case of water contamination). In a great deal of situations, the cause of

    contamination could be a single field or could be multiple fields. However, the

    lowest common denominator is that which provides the narrowest scope. In other

    contexts, the lot or batch identification would be defined by a totally different set

    of parameters (for instance, cut and bagged mixed lettuce).

           The key point in utilization of lot or batch # is to minimize the impact of a

    recall by narrowing the scope of product that is implicated. Additionally, lot or

    batch # becomes incredibly important when products are processed (i.e. cut or

    comingled) such that a recall from a single distributor does not cause a much

    broader recall of processed products. If data on lots and batches isn’t consistent

    and doesn’t follow the product, FDA’s efforts to trace the product back through

          the supply chain can cause them to lose the trail of the product—at the end of the

          day they may or may not even know they’re still tracing the right product (USDA,

          2009, p. 102)

                 Additionally, it can be argued that simply following a product by its

          “product” characteristics can lead to a much wider recall than with a lot or batch

          attached to it—many trucks carry multiple commodities from multiple companies.

          Many of the products are from multiple original lots. Thus, a trace-back for an

          apple that may have only had a problem on a particular truck where temperature

          controls were not monitored could lead to a much broader recall of apples from a

          particular shipper or even from the entire industry (as was experienced with

          spinach products in 2006).

    2. Common	Level	of	Detail	

    In many ways, the level of detail is an integral part of the interpretation of traceability

nomenclature. Depending on the commodity, post-harvest processes, and location in the

supply chain (i.e. shipper’s warehouse, processor’s cutting line or on the shelf at a

grocery store), fresh produce has many levels of detail. In some cases, product is in

shipped and/or stored in bulk bins. In others, the product is stored in cases and palletized.

Many commodities are placed on grocery shelves in bulk with no packaging (i.e. tree


    Interviewee’s were relatively consistent in their thinking that some method of

implementing precise labeling and tracking at the unit level (i.e. on each individual potato

or carrot) would allow investigators to minimize a recall if the consumer still had a

sample available of a suspected food product. However, that idea is reliant upon

consumers both possessing whatever portion of the product is remaining and a positive

test implicating that product sample. Additionally, the reality is that labeling each

individual type of produce by unit is neither feasible nor effective for a number of

reasons—shippers and distributors alike pointed this out. First, it is physically impossible

to affix individual labels to many items such as bulk carrots or green beans. Second,

traceability would not be improved through the supply chain by labeling individual items

as they are shipped and moved through the supply chain in cases. Third, considering that

a major problem for investigators is that consumers either don’t remember what they ate

in detail or have disposed of the product by the time regulators perform interviews,

individual item labels don’t do much to help narrow an investigation (USDA, 2009, pp.


    At the end of the day, a trace-back that accurately identifies a source product that has

been contaminated after the product has been consumed or disposed of does nothing to

mitigate the spread of an outbreak. Trace-backs must not only occur with accuracy but

also with speed. In the world of perishable fresh produce, traceability yields little gain to

consumers if product has already expired or been consumed by the point in which it is

implicated. In order to implicate a product, regulatory bodies are looking for a statistical

link between clusters of the same strain of illness products consumed by those patients.

In order to establish this link, they must first have the ability to narrow the options. As

customers often don’t remember specifics of the products they consumed, they are more

likely to remember in general what they consumed and where they shop (i.e. a bag of

lettuce as opposed to a specific variety of lettuce) (USDA, 2009, pp. 100-101). In this

case, the knowledge of what fresh produce (i.e. which specific types of romaine lettuce as

well as which specific lots each of those bags of romaine lettuce) they have a heightened

ability to statistically implicate a very narrow line of products and lots. Knowledge of

what products could have been consumed at a narrow level of detail gives investigators

an opportunity to not only exclude other products, but also follow each product’s path

back through the supply chain for source commonality.

    This being said, the level of detail in which the application of nomenclature is that

which is capable of both:

    1) Most narrowly identifying a specific company’s lot-specific product

          - and -

    2) Applied on a level of detail such that it is both identifiable and visible to the

          largest number of supply chain members as any other form (i.e. pallet level of

          detail vs. case level of detail)

    II.       Linking	Data	Elements	

          The FDA has also noted that it is not only important that there be standardized

data elements, but also linkages between those data elements from farm to fork. More

specifically, the FDA has made the point that while standardized data elements are

incredibly important to any traceability system moving forward, they “can’t stand alone”

simply because the existence of a data element does not create a link for the product

across the supply chain (USDA, 2009, p. 34).

          Additionally, shippers with FDA investigation experience pointed out that it is

imperative that fresh produce be traceable first by the physical path it traversed through

the supply chain. That is to say, a trace-back motivated by existing documents such as

invoices and bills of lading are suited to trace product back at the transactional and

monetary level. As noted in a recent study18, contamination of food product can happen

at any point in the supply chain in which the product it handled or moved (for example,

cooling failure or exposure to other harmful substances, etc.) Establishing a statistical

link between individual lots of product to implicate as precisely as possible, links

between physical paths of multiple products/lots must be available and accurate.

              As pointed out by the International Foodservice Distribution Association in their

public response to the FDA’s request for comments on traceability, the need for each link

in the supply chain to maintain any list of ingredients for each product received is neither

feasible nor helpful in the trace-back process (IFDA, 2010, p. 4). Rather, the

organization in the supply chain that reconfigures or remanufactures the product should

be responsible for establishing the link between each source product and the final

lot/batch produced and shipped. As such, for an accurate and rapid recognition of each

individual product’s movement through the supply chain, there need be an electronic link

between individual products used as ingredients and the lot/batch produced such that a

trace-back can effectively locate product still in raw, bulk form or as part of a comingled

or remanufactured product.

       III.           Standardization	of	Reporting	Methodology	

              In a major outbreak, regulators face multiple dilemmas in terms of information

gathering for trace-backs. Among those are the window of time in which each

organization has to report traceability data as well as the wide variety of not only formats

of information but also formats of reporting information. Thousands of pages received in

      (Alliance for Food & Farming, 2010) 

differing reporting formats and mediums (i.e. fax, mail, photocopies, etc.) means more

time disseminating what they’re looking at. It is not only the access to and availability of

uniform product information that is important, but also the uniformity in which the data is

received. The Harvard Tomato study reinforced the need to establish commonality both

in data and in reporting (McEntire J. C., 2009). Common nomenclature on its own

doesn’t mean that the data reported is uniform (i.e. each individual backoffice system

produces reports that have unique layouts, characteristics, descriptions, etc.)

Additionally, recent outbreaks have proven that even a single day saved in the discovery

and implication can make a difference in the outcome of an outbreak. Regulators must

have rapid access to data that is not only collected in a uniform nomenclature but

reported in a uniform manner as rapidly as possible. The fastest possible way for

regulators to obtain the necessary information is electronically if it is reported uniformly

across the supply chain.

              A major concern to this end is the issue of privacy19. Across the supply chain is

concern over the privacy of trade data and the availability of trade data to the public (and

suppliers) if the information is readily available to regulators in a cloud-based system.

Shipments between supply-chain members constitute trade secrets and exposure of these

“secrets” to competitors presents a major challenge to maintaining competitiveness. If all

traceability data is accumulated in a single location or available “on-demand” there is a

distinct and widespread fear that it will be used for purposes other than traceability by

regulators and that shipment and product data will fall into the hands of competitors. In

an era of enhanced scrutiny and availability of data from regulators through means such

as the Freedom of Information Act, organizations are rightly concerned that widely
      As evidenced often in the interview process 

available data will not stay private and be used fairly. While the use of electronic

reporting in a standardized format for the traceability nomenclature is ultimately an

important facet of traceability, the issue of information privacy must be addressed. Based

upon information from interviewees, a method for reporting in which regulators can

receive rapid responses to standardized requests while simultaneously minimizing the

chance of potential privacy violations for supply chain members through exposure of

non-necessary data is ideal. This is to say, the more legally protected the data, the more

cooperative industry is likely to be in providing it.

    IV.      Equal	Requirements	for	Recordkeeping	Across	Supply	Chain	

          The comingling of fresh food products at any level of the supply chain can be

cause for worry when not properly documented. A 2008 outbreak of E. coli O157:H7 in

ground beef was traced back to a common retailer who didn’t keep records (called grind

logs) of which lots of beef were comingled into packages of ground beef. As a result, the

FSIS was unable to trace the contaminated product back to its source to ensure no

additional contaminated beef from that lot was in still being sold in other locations

(USDA, 2009, p. 16).

          Specifically conveyed through grower/shipper interviews is a resounding

discomfort that all organizations will not be expected to follow the same guidelines;

small companies need to be held to the same standard as large companies. The concern

lies not in a principle of fairness but with an outlook towards effectiveness. In a supply

chain with as many links between the source and the ultimate consumer, a single broken

link can cause the trail to be lost in a trace-back. If small farms, distributors and retailers

are allowed to bypass any requirements to track product and product source data that

would make a trace-back and subsequent trace-forward more effective, the

implementation across other members of the supply chain becomes more self-defeating.

This is to say, a partial trace-back through only members of the supply chain deemed as

“large” can leave holes in the investigation and potentially cause inconclusive implication

and more widespread outbreaks and recalls.

       Pressure from not only industry partners but also from buyers may also serve to

“raise the minimum bar.” Shippers who do not comply with existing regulations may be

forced to, at a minimum, increase the effectiveness of current traceability programs.

However, it was noted by multiple supply chain members that large upstream supply

chain members are the key to positing appropriate pressure to ensure widespread

adoption of voluntary new data standards. While small buyers may not have the ability to

influence a large distributor or shipper’s operations, a much larger buyer can. In this

way, consistency on the part of upstream members is also imperative if standardized data

is expected to be broadly adopted. Almost without exception, it was noted in interviews

that prior initiatives first endorsed by major downstream purchasers (such as the use of

databar technology) were often cast aside in the instance of cheaper alternatives not

meeting the requirements.

       Following the idea that each time the product is introduced to a new environment,

data requirements should also be applied in a uniform manner. Whether the product is

packed, processed, distributed or simply handled and not reconfigured, each level of

touch presents an opportunity for the introduction of contamination. In devising full-

chain traceability for perishable produce items, it must be possible to track produce at any

location where there is an opportunity for contamination.

              Different members of the supply chain suffer in more or less dramatic ways

depending on two factors: the quantity and quality of traceability data maintained as well

as level in the supply chain. In defining appropriate quantity and quality, it can be noted

that an effective trace-back procedure relies on a quantity of data sufficient to follow

individual product lots from entry (source) to exit (recipient) in a given location while the

quality of data would allow specific identification for minimal impact in a recall

situation. Additionally, the level in the supply chain can dictate the stakes involved in a

trace-back. To illustrate, take two fictitious (but commonly referred to20) examples in

which those members of the supply chain who dutifully record and maintain trace-back

information are harmed by those who do not.

              (1) Many growers/shippers comply with recordkeeping and labeling by

                      application of and maintenance of relevant traceability data by lot/batch. In

                      the event of a recall, product is received from grower/shippers who have

                      provided sufficient information as well as those who have not. Because there

                      is only a partial dataset available for trace-back, a widespread recall on a

                      commodity is initiated, harming all grower/shippers equally.

              (2) Trace-back information is provided and accurate at a grower/shipper level but

                      maintained only partially at a distribution or retail level. In the event of an

                      outbreak, there is not sufficient data to implicate a single source due to lack of

                      traceability records between DC’s and individual retail outlets such that all

                      product must be recalled for a given commodity or commodities. While all

                      members of the supply chain bear some cost in the recall, the ultimate burden

      Commonly referred to in interviews with Grower/Shipper organizations 

            is upon the grower/shipper who must refund the cost of the product through

            the supply chain in addition to performing a recall.

         While these two examples are not intended to provide the full breadth of potential

issues in which some players “follow the rules” while others do not, they are provided to

support the idea that traceability, in order to be effective, must be implemented (without

exception) chain-wide.

    V.      Visibility	

         A final and, perhaps most difficult, part of traceability is the concept of visibility

to and the perception of consumers and regulators. If consumers and regulators do not

trust the accuracy of a new traceability process, widespread recalls will continue to occur

and consumers will continue to avoid the purchase of food not affected by a recall merely

because it is the same commodity. Consumers must ultimately have the perception that

the traceability system in place across the supply chain will help ensure that less

contaminated product has a chance to stay in commerce in the event of an outbreak, thus

minimizing reputational and financial damage to industry. Additionally, consumers must

have the means to know what food is affected in as specific detail as is possible. If

consumers only know that bagged spinach is affected but have no knowledge of brands,

‘best-if-used-by’ dates or other identification, they will likely cease to purchase bagged

spinach products and the entire industry will suffer. As the industry is all-to-aware

through recent major outbreaks (such as the Spinach recall in 2006), wariness by

consumers and regulators in terms of trusting traceability records has continued to build

mistrust—if consumers don’t trust the supply chain to accurately perform a recall such

that contaminated food is pulled from commerce effectively, many will cease to purchase

entire commodity groups for fear that they can still get sick or die. This concept of

visibility and understanding of food safety on the part of the consumer was touched upon,

in some capacity, across almost every interview conducted.


                                    Part	VI	
                       The	Produce	Traceability	Initiative	

       The central tenet of PTI is the creation of a uniform method of labeling and

identifying fresh produce by the case that is recognizable outside the walls of a single

facility. That is to say, the same barcode symbology applied at the source of product

packaging can be scanned and stored electronically by each link in the supply chain to

maintain a physical product flow. Barcodes were picked as the technology of choice

because of their prevalent and long history of use throughout the supply chain—people

understand what a barcode is and most companies (both large and small) are using

barcodes to some degree in their operations. This labeling process is largely an addition

to existing infrastructure such that it need not compromise or complicate existing internal

traceability systems (i.e. within the confines of an individual facility) and is proposed on

a fully voluntary basis—there is no legal requirement to implement the PTI for any

member of the supply chain. Companies would be able to retain and maintain their

original software and hardware investments in internal traceability (for inventory control

and management capabilities) while simultaneously applying a common identifier to each

package. PTI was designed to make use of existing technologies (barcodes), standards

(GS1), information (lot #, batch #), legislation (Bioterrorism Act and “one-up one-down”

methodology) and systems to augment a fully transparent view of product flow through

the supply chain (Produce Marketing Association, 2009).

       It is important to note that the PTI is self-described as a “living” document and,

thus, has changed and will likely continue to change over time to reflect lessons learned

in implementation. Thus, this paper’s examination of the PTI is accomplished by

utilizing a frozen frame version of the documentation from March, 2010.

Case	Level	Identification	vs.	Item	Level	Identification	

              There are three common levels of product identification in the fresh produce

supply chain: pallet level, case level and item level. As an example, a pallet may contain

60 cases of product (which may or may not be of the same origin, depending on the prior

source(s)), in which there may be 20 clamshells of a specific item (which, again, may or

may not be of the same origin). Alternatively, the case may hold bulk items such as

melons, tomatoes or lettuce in which there is no additional grouping of product inside the


              PTI specifically seeks to standardize products at the case level. Reference

material available for PTI specifies that while the initial focus is on case-level

traceability, there is a strong feeling that the long-term plan should be to move to the item

level. The steering committee for PTI felt that it was better to focus on case level

tracking initially for several reasons21:

1)       Touch:                All companies in the distribution chain (both grocery and foodservice)

         handle products at the case level (i.e. “touch” the case) but in most cases the case is

         never opened and the item is never handled while the product is in transit between

         locations. The rationale is that case level marking is applicable to every company

         handling the product while item level identification would create unnecessary and

         unrealistic difficulty in implementation.

2)       Space:                Cases have additional room to accommodate the necessary data elements

         for traceability while there is not space on most bulk produce items. Even a much

         smaller and simpler PLU (Price Look-Up) sticker will only fit on some bulk produce

      (PTI Steering Committee, 2009e) 

     while others (such as loose green beans, loose mushrooms, etc.) cannot

     accommodate even a very small sticker.

3)   Function: Adding data that encodes an individual lot # and/or batch # for each item

     would not only add enormous cost but also cause a “severe slowdown in


     The PTI Steering Committee maintains that work required to address traceability at

the case level is a “substantial” first step but that the exploration of an efficient means to

track at the item level will continue.

Standardization	with	PTI	and	GS1	

       The barcode selected by PTI was developed and is maintained by GS1, a not-for

profit 501c6 organization that was originally formed for the purpose of U.P.C. (Universal

Product Code) barcodes in the grocery system. GS1 and GS1 standards are now globally

recognized by 2 million companies in 150 countries. Their standards are used to identify

things, places and organizations through highly structured numbering systems. GS1

standardized numbers have been used to identify a broad range of hierarchical levels of

products; a GS1 standard can be used to identify categories of products such as 16-ounce

boxes of XYZ-brand beef patties as well as serialized (each number or barcode is unique)

numbers in order to uniquely identify each box of those patties (Carpenter, 2009).

       The most rigid portion of the numbering system employed by PTI for

standardization is through the use of a GTIN (Global Trade Item Number), also

administered by GS1. The GTIN specifies the specific product characteristics and unique

company code as a portion of the GS1-128 barcode to be affixed to each case. The GTIN

is specified for use because (among other things) it is well established, has a consistent

format and structure, is used globally, has applicability across a broad range of

technology platforms (i.e. the GTIN can be used with both barcode and RFID

technology), can be re-used by brand owners for a series of identical products and is

creates uniqueness in identifying every variation of a product or service in a globally

unique manner (GS1 US, 2006). The GTIN is defined by PTI documentation as

identification at the case level analogous to the U.P.C. at the item level (Produce

Marketing Association, 2009). While the specific and rigid numbering requirements of

the GTIN in reference to PTI will be discussed further in later portions of this analysis,

the key point is that the numbering system selected by PTI was picked because it is:

              Well established, globally recognized and globally utilized

              Electronically Recordable (via barcode scan)

              Standardized across industries and members of the supply chain

              Unique to the company and item level detail desired

PTI	–	Milestones	of	Implementation	

       PTI was established as a set of milestones for various members across the supply

chain to implement. The Initiative was designed to be implemented in sequential order—

that is, Milestone #1 is a prerequisite for Milestone #2 and so on (PTI Steering

Committee, 2009d). The milestones will be described in this research similarly, followed

by a thorough analysis of the “outputs” gained after the entire initiative is implemented as

compared to the future goals specified in Part V of this document.

       Specifically, the milestones and their deadline for recommended completion are

listed in Figure 6-1.

Milestone       Description                                             Deadline
#1              Obtain a Company Prefix                                 Q1 2009
#2              Assign GTINs to Cases                                   Q1 2009
#3              Provide GTIN Information to Buyers                      Q3 2009
#4              Show Human-Readable Information on Cases                Q3 2010
#5              Encode Information in a Barcode                         Q3 2010
#6              Read and Store Information on Inbound Cases             2011
#7              Read and Store Information on Outbound Cases            2012
Figure 6‐1:  Milestones for PTI Implementation

PTI	Milestone	#1:		Obtain	a	Company	Prefix	

         At the heart of the GS1-128 barcode that will be applied to each case is the GTIN

(Global Trade Item Number). PTI specifies that the GTIN is exactly 14 digits in length

and consists of the following components (Figure 6-2):

                  Global Trade Item Numbers
 PACKAGING  Packaging indicator to differentiate between different 
  INDICATOR  methods of packaging for a given unique case configuration.  
               For example, a size 9 cantaloupe melon may be packaged in a 
               regular corrugated carton, a DRC or an RPC.  The packaging 
               indicator is 9 digits long and can accommodate up to 8                 1 
               specific packaging styles.  A ‘1’ in this spot is recommended by 
               the PTI Steering Committee and is non‐specific to style of 
         GS1  A value of ‘0’ is used in this spot to indicate that the company 
    Indicator  prefix was issued by GS1 US or GS1 Canada                              1 
   COMPANY  Company prefix identifies the organization uniquely from all 
      PREFIX  other organizations in the world.  The length of the company 
               prefix is dependent upon the number of unique case 
               configurations required (see below). 
REFERENCE #  Reference number identifies a given case configuration                  11 
               uniquely.  This number is assigned specifically by the 
               organization to fit their requirements.  The length of the 
               reference number is dependent upon the length of the 
               company prefix. 

    CHECK DIGIT  The check digit is derived from an algorithm.  This number 
                 serves to ensure integrity of the data (i.e. that the 
                 combination of the indicator, company prefix and reference                      1 
                 number was entered correctly).  A tool to calculate this digit is 
                 available at GS1us.org 
                                                                     Total Digits                14
Figure 6‐2:  Meaning of data elements within a GS1 GTIN # 

       The first milestone is meant to establish a unique number (identifier) for each

company that will apply labels to a case. This unique identifier, the ‘company prefix,’ is

issued by GS1. Companies that will need to acquire a company prefix are based on their

position and value-add in the supply chain. Specifically, the companies that require a

GS1 company prefix are:

           Growers or Shippers who wish to maintain their own brand

           Packers that change the composition of the original case or product and re-brand

           Buyers who maintain their own private-label brand

       While the first milestone specified in PTI

documentation only applies to the creation of a               # of Case              Length of 
                                                              Configurations         Company Prefix 
company prefix, it is important to understand the             99,999                 6 digits 
                                                              9,999                  7 digits 
implications of company prefix selected. As                   999                    8 digits 
                                                              99                     9 digits 
specified in Figure 6-2 above, the combined                  Figure 6‐3:  Length of Company Prefix

length of the company prefix and the reference number is 11 digits. This is because the

length of the company prefix is dependent upon the number of unique case configurations

that an organization will need to specify. The larger the number of unique case

configurations required, the shorter the company prefix will be. For example, a nine-

digit-long company prefix only leaves 2 digits for the reference number; thus, only a

maximum of 99 (01-99) reference numbers can be specified. On the opposite end of the

spectrum, a six-digit-long company prefix leaves five digits for the reference number so

there are 99,999 possible unique reference numbers. The specific length of a company

code based on the maximum number of unique case configurations is listed in Figure 6-3.   

      GS1 charges a different fee for company prefixes based upon the length. As the

company prefix must be established before the actual creation and registration of unique

case configurations (via reference numbers), it is imperative to understand the approach

that will be used to uniquely identify cases before the company prefix is established. The

level of specificity required and/or desired will dictate the number of GTINs required for

each company. For example, a shipper that packs four varieties of four commodities each

with four sizes may require four reference numbers (commodity level specificity), sixteen

reference numbers (commodity and variety), or 48 reference numbers (commodity,

variety and size). In this example, a nine digit company prefix is sufficient. However, if

that same example were extended to ten commodities in ten different varieties each with

four difference sizes (and maximum specificity is a requirement), the nine digit company

prefix would not allow for enough unique case configurations—an eight digit company

prefix would be necessary to accommodate the 400 unique case configurations (10 x 10 x


PTI	Milestone	#2:		Assign	GTINs	to	Cases	

Reference	Numbers	for	Unique	Case	Configurations	

         After the creation of a unique company prefix in Milestone #1, an organization

can begin to establish reference #’s for each unique case configuration. As specified

above, reference numbers are an important part of the GTIN to be applied to each case as

they distinctly identify each unique case configuration. While the PTI plan does not call

for this step to be implemented until after a company prefix is established, the assignment

of GTINs to cases is inherently limited by the length of the company prefix an

organization selects for Milestone #1. GTINs (and, thus, reference numbers) are required


           Growers who maintain their own brand

           Packers that change the composition of the original

            case and re-brand it                                    Primary Attributes 
           Shippers who maintain their own brand (GTIN’s           Variety 
            must be provided to party that packs the product        Grade 
            into boxes for them)                                    Count 
                                                                    Shipping Container 
           Buyers with private labeled products                    Inner Pack Style 
                                                                    Inner Pack Quantity 
       The recommendation for creating reference numbers in         Inner Pack Size 
                                                                    Inner Pack UOM 
PTI is based upon the concept of Primary and Secondary              Growing Method 

product attributes. Primary attributes are those that are most likely specified by a buyer

when purchasing the product whereas secondary attributes are product characteristics that

the shipper maintains internally and are not generally of any consequence to the buyer.

Additionally, primary attributes are inclusive of characteristics that change the

composition of a case. For example, the physical configuration of an size 80 versus a

size 90 apple is different—the size is analogous to the count in a given case in this

example and, thus, a difference in size is a difference in composition. For this example,

size would be a primary attribute. However, if the same supplier packs a given case

configuration with multiple labels and label is of no consequence to a buyer, label would

be considered a secondary attribute. Unique reference numbers should be assigned to

every unique case configuration of primary attributes. To apply this to our example,

suppose both sizes (80 and 90) of apples are packed in four distinct labels and that there

are no other primary attributes (i.e. there are multiple labels available but buyers only

typically order based upon size and don’t care about label differences). Only two unique

reference numbers and two unique GTINs would be necessary in this example (one for

size 80 and one for size 90). However, if buyers ordered based upon both size and label

specifically, there would be eight unique reference numbers and eight unique resulting


       The rationale for primary attributes (i.e. based upon the attributes in which a

buyer uses to order the product) is that buyers order product based on significant

attributes that make one case configuration inherently different from another. Further,

this can be seen through price point differentiation in product attributes—differing price

points is indicative of a level of significance inherent in the case configuration. PTI

recommended primary attributes are listed in the figure on the previous page.

       If any organization in the supply chain changes the original composition of the

case in any way, (i.e. through re-packing, processing for fresh cut, etc.), a new GTIN

must be affixed to the new case. Any organization that changes the original composition

of the package must maintain the original GTINs and lot numbers for all product in the

new case. The original GTIN’s and associated lot/batch numbers must be linked by that

organization to the new GTIN and batch/lot# assigned. (Lot and batch numbers will be

described in more detail later in this document).

A	Final	GTIN	

       The final GTIN is the combination of:

    [Packaging Indicator] + [GS1 Indicator ‘0’] + [Company Prefix] + [Reference #] +

                                       [Check Digit]

       As an example, imagine Shipper XYZ registers a nine digit company prefix to

allow for up to 99 unique reference numbers (unique case configurations). The company

prefix assigned to Shipper XYZ is ‘123456789’. The first unique case configuration

Shipper XYZ establishes is ‘01’. Additionally, Shipper XYZ has decided to maintain a

generic packing style of ‘1’ for all products for simplicity’s sake (the recommendation of

the PTI Steering Committee). Using the tool available on the GS1 U.S. website

(www.gs1us.org) , Shipper XYZ can calculate the check digit required for the GTIN.

The GTIN that would be affixed to all cases of unique configuration 1 is as follows

(where the ‘+’ sign is indicative of concatenation, not addition):

[Packaging Indicator] + [GS1 Indicator ‘0’] + [Company Prefix] + [Reference #] + [Check Digit]

          1        +       0           +       123456789     +       01         +      9

                                    = 10123456789019

       This number, 10123456789019, uniquely represents Shipper XYZ’s unique case

configuration ‘01’ and should be affixed to each case with that unique configuration.

PTI	Milestone	#3:		Provide	GTIN	Information	to	Buyers	

       The ultimate goal of affixing an identifying barcode to every case is the ability to

uniformly trace the product through the supply chain. As such, buyers of produce (i.e.

one step downstream in the supply chain), need to be aware what GTINs each of their

suppliers have created so that they can create a methodology for reading the barcodes and

understanding what the barcode signifies. For example, when our fictitious Shipper XYZ

sends a case of product ‘01’ and the buyer scans the barcode on that case, the buyer’s

system will recognize that the product is a case of Shipper XYZ’s product ‘01’.

       The PTI plan and best practices provide a spreadsheet with which a supplier can

create a standardized list of GTINs and associated product characteristics in order to e-

mail buyers with a list of GTINs. This would need to take place in advance of shipping

cases to a buyer. Alternatively, a data synchronization tool in which suppliers may

register their GTINs (including unique reference #’s and descriptions) with GS1 U.S.

through their website (www.gs1us.org) and allow buyers to automatically sync registered

GTINs with their respective internal systems.

       Products received and sold as “external substitutions” (products from another

organization are not materially different in nature) in the event of a shortage of product

do not need new GTIN’s assigned. PTI best practices documentation specify that product

received from external sources should be recorded at the case level (i.e. store existing

inbound GTIN information) and recorded at the case level on outbound shipments by

milestones #6 and #7. As shippers buying and selling fruit for substitution are not

engaged in modifying the case configuration or repackaging/processing, these products

do not need new GTINs or case labels (PTI Steering Committee, 2009c).

PTI	Milestone	#4	&	#5:		Show	Human‐Readable	Information	on	Cases	and	Encode	

Information	in	a	Barcode	

       While PTI Milestones #4 and #5 are listed separately in the PTI guidance

documentation, they are prescribed to be completed at the same time and are two parts of

a single output: a case-level label that includes certain ‘human readable’ information as

well as a barcode signifying this human readable information. As such, this analysis will

combine the two steps for the simplicity of understanding the combined requirement and


       Case level labels will need to be affixed in either of the two following


    1. At the time of original packaging (i.e. in the field or the packing shed when

       product is first placed into a case)

    2. When a case of produce is reconfigured, processed (i.e. co-mingled and/or cut) or

       in any way the original composition of the case is changed

    If a case of product is simply “passing-through” an organization in the supply chain,

no further label needs to be affixed to the case—the original tag is sufficient. When a

new label is affixed to the case (after reconfiguration in some way), the original GTINs

and lot/batch numbers provided on the original label must be recorded and linked to the

new GTIN and lot/batch number assigned to the case.

Human	Readable	Information	

    Human readable information (i.e. actual text on the label not intended to be read by a

computer) on the tag must include, at a minimum:

          The case configuration’s GTIN number

          Lot or Batch #

       If the lot or batch number of the product is not unique by pack or harvest date, the

barcode must have additional information that specifies the pack or harvest date (see

below). Additionally, the following data elements are specified as necessary on the case

label if not already printed on the same side of the case in which the label will be applied:

              Commodity and “additional” description (Example: Variety, Variety

               Group, Color)

              Pack Configuration (Example: 10/4-lb bags)

Data Elements
                                                           Barcode	–	Computer	Readable	
Buyer/Receiver Contact Information
Product Description                                        Information	
UOM (e.g. cases, boxes, etc.)                                      So that all linear
Shipment ID
Transporter that brought load to your company              scanners in use in warehouses
Vendor/Supplier/Sender contact info
Date of Shipment                                           today can read it, the GS1-128
Ship from location (shipping information)
Ship to location (receiving information)                   barcode has been selected by the
Date of receipt
                                                           PTI steering committee as the
Transporter that shipped the load from your company
standard barcode symbology of choice for the human readable portion of the label. This

particular barcode can utilize up to 48 characters of information. There are four pieces of

data that must be included in a PTI compliant case label barcode:

       Application Identifier: GTIN ‘(01)’

       GTIN

       Application Identifier: Lot/Batch # ‘(10)’

       Lot/Batch #

    An application identifier is a code that precedes a specific element of data and tells

the computer what the piece of data means. For example, the application identifier ‘(01)’

tells a computer that the next fourteen digits are the GTIN (i.e. the company prefix and

reference number for that case). Similarly, the application identifier ‘(10)’ tells a

computer that the following eleven characters signify the lot or batch number for the

case. Back to our fictitious Shipper XYZ, a full barcode for reference ‘01’ that came

from lot

    ‘97102910A10’ would be (‘+’ signs indicate concatenation, not addition):

    [‘(01)’]   +               [GTIN]          +      [‘(10)’]        +       [Lot/Batch #]

    (01)       +       10123456789019          +      (10)            +       97102910A10

                          = (10)10123456789019(10)97102910A10

    When a subsequent member of the supply chain scans this barcode, their system will

record that they received a case of ‘01’ from Shipper XYZ originating in lot/batch

number 97102910A10.

Additional	Case	Label	Requirements	and	Recommendations	

        From a functional standpoint, the PTI committee specifies minimum and

recommended requirements for the size of some elements of the label. The barcode itself

requires 3.75” width and 1.25” height on the label (including excess space requirements

on either horizontal side of the barcode. Additionally, the human readable version of the

barcode (the GTIN and lot/batch number) is required to be printed at an 18 point font

size. In order to achieve this barcode size and allow enough space for any other printed

information required (see next section), the recommended minimum case label is 4” x 2”.

For RPC containers (re-usable product containers – made of plastic), a card should be

substituted for the label.

PTI	Milestone	#6:		Read	and	Store	Information	on	Inbound	Cases	

              Reading and storing the information contained on each case is required by the

same organizations who would be required to store “one-up, one-down” information in

the U.S. Bioterrorism Act of 2002. The PTI steering committee defines those excluded

from these requirements as individual restaurants and individual stores (based upon those

organizations excluded from the requirements of the Bioterrorism Act)22. Additionally,

those businesses who are engaged in contractual “cross-docking” (whereby the service

provided is receipt of the product and then subsequent loading of the product as a

convenience to customers), are exempt from scanning inbound packages23. Each scan

(which records the GTIN and Lot/Batch #) should then be tied to the data elements listed

on the previous page (defined as already available to the organization). The contents of

the barcode and accompanying data must be stored in electronic format and retained for

the period required by the U.S. Bioterrorism Act of 2002 (two years) (Produce Marketing

Association, 2009).

PTI	Milestone	#7:		Read	and	Store	Information	on	Outbound	Cases	

              Milestone #7 requires that all organizations specified in Milestone #6 scan all

barcodes and store the same required data with the barcode at the time of shipment

      See Analysis of PTI section for a discussion of these recommendations 
      See “Best Practices for Cross Dock and Load Only Service” (PTI Steering Committee, 2009b) 

(outbound cases) in addition to inbound shipments. Those exempt in Milestone #6 are

similarly exempt from Milestone #7.


                                   Part	VII	
       Analysis	&	Discussion:		PTI	and	the	Future	Goals	of	Traceability	

              In order to establish the potential effectiveness of PTI, future goals to enhance

fresh produce traceability systems were established based upon supply chain interviews,

FDA testimony and public releases and analysis of recent major outbreak and recall

shortcomings in Part V of this document. The following analysis uses the future goals in

Part V as a means of comparison for PTI’s potential in increasing the speed and accuracy

in trace-back and recall events. Baseline comparisons of meeting the future goals as well

as implementation and usage details from members of the supply chain are included in

order to establish not only the ability of PTI to meet each future goal but also the relative

feasibility of industry to implement. In order to accurately analyze PTI, it is important to

note that the assumption herein is full compliance across the supply chain.

              The primary question in this analysis is whether or not the PTI assists in linking

the foods that consumers have eaten with the path through which those products travelled

through the supply chain. This analysis will consider each of the goals presented in Part

V and then revisit this question in the following section to determine whether or not PTI

appears to ultimately assist in this goal of expedited traceability. If good information is

available to trace product back through the supply chain starting from the consumer, then

it almost certainly is reasonable to expect that the same data will be effective at tracing

the product forward for a recall24.

  This analysis and discussion, prefaced and based upon the research and interview data collected and 
analyzed above, is that of the author’s. 

    I.      Standardization	of	Datasets	

         PTI directly addresses the concept of unique identifiers utilizing the GS1 standard

and utilizes the GS1 coding methodology as a means of creating common nomenclature

that is both universally recognized and unique in nature.

    On a macro level, GS1 is an internationally recognized organization of common

standards that provides a platform for unique codes. Standards for the individual codes

that make up PTI data scheme are all recommended based upon standards that are already

in place and being used for other commerce purposes internationally. As a selection for

nomenclature, the use of existing GS1 standards enforces the concept that each member

of the supply chain is speaking the same language to the degree that individual data

elements are presented and utilized in a uniform manner across the supply chain. That is

to say, selection of a uniform body of codes ensures that both visually (i.e. looking at the

codes) and systematically (i.e. storing and accessing the codes), the individual elements

are simultaneously universally understood as the same data elements and designed for

uniqueness in identification.

    a. Company	Identification	

                The use of the GS1 Company Prefix numbers as the recommended code

         methodology within which to store and transmit the common nomenclature allows

         for the creation and use of datasets that ensure uniqueness of the source firm from

         which the product originated. Furthermore, both shippers and growers who

         maintain their own brand and processors who reconfigure or change the product

         in some fashion are required to apply labels with their company prefix.

           While the PTI adequately handles the issue of determining the immediate

    prior firm that modified the product, it does not call for standardization of

    company codes at other levels of the supply chain, including some levels of

    distribution, retail, etc. This presents a situation in which regulators are presented

    with uniform company information at the application level (i.e. the medium

    selected for application of the dataset) that indicates the immediate prior modifier

    of the information but does not provide them with consistent information with

    which to trace the product fully through the supply chain. In other words, they

    will receive uniformly applied company codes that specify an original

    grower/shipper or brand owner in some parts of the investigation while needing to

    use other forms of identification for distribution centers, storage facilities and

    other locations that the product may pass through in other parts of the supply

    chain. This unevenly applied uniformity in the application of traceability presents

    a situation in which data reported are inconsistent and do not fully meet the goal

    of uniform company identification throughout the supply chain. By not utilizing a

    standard for each “stop” along the path of produce through the supply chain,

    investigations must rely upon multiple forms of information gathering in addition

    to the standardized data provided through the PTI. Utilization of standardized

    company codes across every member of the supply chain ensures that a dataset

    from one firm with both a source and destination company identified may be

    matched electronically with the datasets provided by the source company and

    destination company.

           Additionally, the requirement that buyers with private labels utilize

    company prefixes as a standard part of the dataset (which indicates only the prior

    source of the product in which the product was modified or changed) does not

    accurately allow for the rapid traceability of physical product flow throughout the

    supply chain. Instead, regulators must first contact the buyer in order to even

    begin to identify the product. The PTI documentation on private labeling explains

    that the company prefix should be based upon the brand identified on the exterior

    of the box. Only if the exterior of the case is branded by the original grower or

    shipper (with private label packages inside the case) should the company prefix

    reflect the original source (PTI Steering Committee, 2009). Again, it is important

    that company prefixes be used uniformly throughout the supply chain such that

    each firm that handles the product may be identified and that the combination of

    company and other unique data elements (product characteristics and lot/batch

    number) are able to be used uniformly nationwide. As many shippers who

    provide private label services for a buyer also either have a brand of their own

    (often sourced from the same lot) or provide private label services for multiple

    companies, the application of company codes is unnecessarily complicated in that

    it implicates multiple unique products, with different company or brand names,

    that all point back to the same original field.

           As an example, following the goals established above, the dataset at each

    firm between a shipper, distributor, and retailer would include unique identifiers

    that show the flow of product from physical location to physical location. The

    shipper, distributor, and retailer’s unique company codes are referenced,

    respectively, as the immediate prior and subsequent recipient of the product. The

    company identifier on each individual case identifies the last member of the

    supply chain to pack, reconfigure or process the product. Once reported correctly

    to investigators, the product flow can easily be determined because each unique

    location with each unique product creates a flow of product from location to

    location. As currently laid out in the PTI, this data flow is interrupted; the label

    indicates the brand owner (which, in the case of private labels is often a retailer

    and not a grower or processor) meaning further investigative work is required in

    many cases to trace back to the ultimate source. Additionally, because only brand

    owners are required to establish company identifiers, many links in the physical

    path of the product flow will not be identified without significant work tracing the

    product manually via documentation such as purchase orders, bills of lading, and

    invoices to determine each physical stop. As contamination may occur at

    locations in which the product physically passes through as opposed to locations

    in which it may simply be marketed (i.e. retail stores), identifying the brand

    owner doesn’t necessarily assist in identifying possible contamination sources.

           While the PTI does utilize a standardized company code, it does not use

    the company code uniformly across the supply chain. Company codes either

    represent the last organization to modify a package (as in a fresh-cut processor) or

    they do not (a private label owner that contracts the growing/packing/processing

    of products). Utilization of the same exact company code to indicate multiple

    physical points in the supply chain does not create a standardized data set. A

    standardized data set, when assembled with others across the supply chain in an

       investigation, should easily match the immediate prior handler with the immediate

       recipient. In other words, the dataset for the second location in a product’s

       journey should indicate, in a uniform use of coding, that the first location was the

       immediate prior handler and vice versa. This would minimize or eliminate the

       need to fill in gaps for analysis of traceability data thus minimizing unnecessary

       and time-consuming extra steps in an investigation.

    b. Discernable	Product	Characteristics	

           The PTI uses reference numbers to identify each of a company’s products.

    The reference number utilizes GS1’s product identification scheme for use in its

    standardized GTIN numbers. A GTIN, or Global Trade Item Number, is a

    standardized code devised by GS1 to globally identify any item that could be sold or

    appear on a price list (GS1, 2009). In defining the methodology through which

    product reference numbers should be selected and identified, the PTI plan calls for

    product identification based upon the characteristics primarily used for selling. While

    this method makes sense from an e-commerce perspective, it is unclear how the

    standardization of primary product attributes based upon selling characteristics will

    directly enable trace-back investigations and recalls to occur in a more timely fashion.

    As shippers do have different selling processes (even with the same commodities,

    processes, and volumes), the identification of a product in a recall situation involves

    sifting through many product identification codes to interpret the range of codes that

    essentially mean the exact same thing. That is to say, regulators would again find

    themselves trying to connect the dots between unique points of data (disparate

    reference numbers) that all potentially mean the exact same thing in the context of a

    recall. While advanced analysis using analytics software can aid in this process,

    investigators must still ultimately take the time to determine whether or not one code

    is the equivalent to another. This added overhead and work required to tie things

    together is unnecessary for an accurate recall. For example, while two separate

    companies both ship bags of chopped romaine lettuce, each will have a unique

    methodology for creating reference numbers and, thus, will have different reference

    numbers. While this differentiation is, to some degree, due to the nature of the

    differing methods by which individual companies buy and sell, some concrete

    methodology should be established such that multiple reference numbers don’t exist

    to identify the same identical product (as with a bag of chopped romaine lettuce). If

    the differentiation between the two reference numbers is, indeed, brand, this should

    be spelled out. The characteristics necessary for an effective trace-back to link to

    what a consumer actually eats are the characteristics that need to be identified in a

    reference number.

           The sheer quantity of potential unique reference numbers has caused a major

    problem. There is great disparity in the understanding and application of reference

    numbers among shippers. Shippers with a similar number of commodities, processes

    and relative volumes had differences in the number of GTIN’s utilized for their

    products as large as 1,000 reference numbers. One shipper specifies that the lowest

    level of detail required is the commodity while others engaged in the same business

    identify GTIN’s based upon every possible combination of variables in the selling

    process. The level of detail specified for reference numbers as laid out in the PTI is

    clearly ambiguous to those members of the supply chain charged with the creation of

    the reference numbers. In our interviews, not a single set of shippers had the same

    philosophy for assigning GTIN’s. While the voluntary nature of the PTI means that

    some growers and shippers may not follow its guidance, sufficient guidance has

    clearly not been provided to assist those charged with the responsibility for

    implementing the PTI. As evidenced in interviews, spoken nomenclature used to

    refer to products tended to be very similar while the assignment of reference codes

    applied to these same products tended to vary greatly.

               From an implementation perspective, the uneven use of reference numbers

    will create an added cost and burden to each member of the supply chain. Any time

    any single member of the supply chain adds or modifies a reference number, each

    other member that may potentially touch that product would potentially need to be

    contacted with updated GTIN’s so that their method of receiving and recording the

    dataset remains effective. The tools suggested by the PTI for synchronization of

    GTIN’s between supply chain members will largely minimize the difficulty in the

    data exchange portion of implementation. However, the data synchronization tools

    provided cannot match product received with an individual shipper’s internal

    inventory system. That is to say, linking of GTIN numbers to an inventory system

    will, at a minimum, require either separate systems of identification (using internal

    product ID numbers) or a manual process of matching each synchronized GTIN and

    company prefix to a back-office system by hand. For some members of the supply

    chain (such as distributors), this will entail manually linking thousands upon

    thousands of unique codes to internal item numbers. The PTI documentation

    regarding external product substitutions for the PTI refers to “substitutes” as a

    product that can be substituted for another insomuch that it does not “compromise the

    specification or quality of the product” (PTI Steering Committee, 2009c).

    Considering that many of these thousands of items will fall into the category of

    substitutes, linking GTIN’s to existing systems will be a tedious, expensive, and

    arduous task.

               While the PTI has provided a common nomenclature and schema for

    product characteristics, it simply does not fully establish an approach for uniquely

    identifying products throughout the supply chain that will improve the speed of a

    recall. Utilizing a common nomenclature to establish standardization and then

    creating unique reference numbers for every type of product shipped from every

    company that ships the product is analogous to differing interpretations of language

    and slang terms across countries that speak the same language but with different

    dialects. If the goal of future product traceability is a more rapid and limited recall,

    simplicity and uniformity in differentiating individual products should be the goal of

    assigning reference numbers within a given commodity group. Allowing each

    company to individually determine the level of detail to be used in assigning

    reference numbers for each product or product introduces a great deal of ambiguity

    and confusion into the system.

    c. Lot	or	Batch	Information	

               Interviews with shippers showed a wide variety of philosophies in

       lot/batch #. Some shippers have lot/batch #’s schemas that will include very

       detailed information such as grower, field, date of harvest, crew and machine

       while others more broadly referred to lots as windows of time (i.e. a single day of

       harvest) or a single ranch. The approach each shipper took in deciding how lot or

       batch # would be devised was based largely upon their own processes and needs

       and not on any form of uniformity. This is not to say that lot or batch # will and

       must entail differing characteristics in differing firms and processes—it is only to

       say that the approach itself should have uniformity. While the lot or batch

       identification is a key element to traceability, the PTI action plan has provided no

       guidance in establishing uniformity in lot/batch #’s. The resulting variety of

       detail and lack of consistency is the current result.

    1. Common	Level	of	Detail	

       The PTI posits that, while item level traceability may occur in the future, the case

    level of identification makes the most sense. The case level 1) may be implemented in

    a way that is not prohibitively costly or time consuming; 2) may be applied to all

    produce, including bulk produce that is too small or in some way cannot

    accommodate a unit-level sticker; and 3) is the lowest level of detail that is both

    feasible and handled by each member of the supply chain.

       The debate over level of detail is one with fiercely differing views across the

    supply chain. Many individuals in the supply chain question the reliability of case-

    level traceability because produce that arrives at a retail store in a case is seldom

    purchased by consumers in that case. That is to say, there is still a degree of

    variability in the implication of a single product and/or lot because products from

    different companies and lots are comingled outside of the box for consumers to

    purchase. The argument is that case-level traceability does not allow regulators to

    implicate individual products in any narrower scope than is currently available.

       Based upon the methodology that is used in the epidemiological investigations

    that precede and accompany trace-back investigations, a broader outbreak of a

    specific illness provides many more points of data. For example, if (at a given time)

    there are twenty company prefixes in commerce for a given product and each of the

    twenty company prefixes are accompanied by five different lots that were received

    and placed into commerce at the store level, regulators are provided a great deal of

    data in which to build a confidence interval to statistically determine the likely culprit

    of contamination.

       To illustrate, consider a nationwide outbreak in which several commodities are

    suspected. If data is readily available that allows investigators to view the potential

    products and associated lots for each of these commodities, then they can begin to

    establish commonalities. For instance, an instant illumination of each product flow

    provides the possibility of finding a common point of contamination (a particular

    warehouse, truckload or origin lot). This allows investigators to limit the scope of an

    investigation by quickly focusing on the commonality and pulling only that product

    from commerce rather than causing a recall for an entire commodity or multiple

    commodities (as has happened in recent history).

       Considering that the greatest damage to a commodity supply chain is the result of

    a massive outbreak (such as the 2006 E. coli outbreak in spinach), the case level as

    proposed in the PTI seems to adequately provide regulators with a great deal of

    uniform, readily available data that identifies product in 1) the narrowest feasible

    scope (i.e. each individual company’s product by lot/batch) and 2) at the lowest level

    of detail utilized by each member of the supply chain (and which is subsequently


    II.      Linking	Data	Elements	

             The PTI provides recommendations for linking data elements (such as

          reference #’s) across the supply chain through the use of tools provided by GS1.

          GS1 can, through the use of standardized company codes and nomenclature,

          provide an electronic means to link identifiers to actual description of the

          company and product. It is unclear at this stage to what extent this tool is being

          utilized through the supply chain and unclear that it is the methodology of choice

          for organizations across the supply chain. While it is reasonable to assume that

          the PTI itself should prescribe a specific means of linking data elements to the

          description of those data elements (i.e. a reference # linked to the description of

          that product), it must be the goal of any traceability initiative to work across

          levels of the supply chain to establish commonality. A manual system (i.e. e-

          mailing spreadsheets with new reference #’s) of exchanging data elements that are

          intended to link together between systems is cumbersome at best and, as

          evidenced through interviews, not currently working. In most cases, there is

          confusion as to where reference #’s are to be sent and who needs them.

             The concept of linking data elements is also complicated by the utilization of

          “brand owner” as the preferred methodology of company prefix. While company

          codes that clearly indicate the physical path (i.e. each warehouse through which

          the product flows) allows investigators to rapidly trace multiple products to

          establish commonality, the use of brand owner creates a situation in which some

              product will show the physical path by linking the company prefix of each

              location through which the product travels to the company prefix of the owner of

              that location. As currently implemented and recommended, the use of company

              prefix simply provides a link back to the grower or prior processor of the

              produce—the company in which recent analysis has indicated is only responsible

              for 2% of all outbreaks25. Additionally, the use of “brand owner” in the case of

              private labeling will not even go so far as to provide that link—as currently

              written, the labeling will only lead to the organization who contracted some other

              member of the supply chain to grow and ship the product (a whole new path

              through which to trace product).

                      In order to effectively trace product through the supply chain, it is imperative

              that the issue of tracing physical product flow versus the monetary flow of

              transactions be addressed.

       III.           Standardization	of	Reporting	Methodology	

                      PTI does not, in any way, suggest or speak to a reporting methodology. The

              absence of standardized electronic reporting of data is an incredibly important

              method with which to ensure that data is “cleansed” as regulators utilize it. While

              PTI suggests that data be recorded electronically, it provides no guidance

              whatsoever for what further information the dataset should be connected to and no

              method or recommendation for the provision of data to regulators. Lack of

              universal company prefixes for each firm in the supply chain that handles the

      (Alliance for Food & Farming, 2010) 

          product adds additional ambiguity in the linkage of products from one supply

          chain member to the next.

             The spinach recall of 2006 made clear that a huge amount of hours were

          wasted simply sorting through the wide variety of data formats received (i.e. fax,

          mail) as well as differing formats and descriptions of data elements on internal

          documentation (i.e. Bills of Lading, Invoices). These hours spent meant huge

          slowdowns on the investigation. The storage of standardized data as described by

          the PTI plan is only a first step toward improving the speed and accuracy in which

          recalls occur—the presentation of that data in relation to the data stored by other

          members in the supply chain is paramount in speeding the process. It does

          investigators little good to know the data exists—the key to standardization is the

          reporting of that data.

    IV.      Equal	Requirements	for	Recordkeeping	Across	Supply	Chain	

             The future goals call for a full trace of product from one end of the supply

          chain to the next such that any link in the supply chain in which contamination is

          possible can be identified. However, PTI allows certain supply chain members to

          forgo recordkeeping for the purpose of cross-docking. While product is handled

          by a third party, there is no requirement for recordkeeping. This physical “stop”

          along a fresh product’s trip to the consumer presents an opportunity for

          contamination—to ignore any stop is to ignore the possibility of an invisible and

          untraceable contamination if the data provided by the PTI is to be utilized in


             Additionally, PTI utilizes the Bioterrorism Act’s definitions of exemptions

         from recordkeeping requirements; this means that small grocery stores, small

         farms and direct-to-consumer restaurants have limited or no recordkeeping

         requirements. These consumer-facing companies are an important link between

         consumer illness and the source of food from which the contamination occurred.

    V.       Visibility	

             The concept of providing visibility is a difficult one—probably the most

         difficult of the future goals. In this regard, it is also difficult to expect a plan such

         as PTI to have integrated visibility into its outline. Visibility, in the case of PTI,

         may be more in the form of recalls becoming less visible with its implementation.

         This is dependent upon regulators utilizing the system, narrowing trace-backs

         more rapidly and outbreaks being minimized as quickly as product can be pulled

         from commerce. At the end of the day, this goal will be met so far as PTI is

         successful at reducing the publicity and damage from foodborne illness outbreaks.



                                                 Part	VIII	
                               Discussion,	Conclusion	&	Recommendations:	

                      	The	Future	of	Fresh	Produce	Traceability	in	the	U.S.	


              Traceability systems in perishable produce commodities must provide a conduit

through which investigators and industry can trace the source of a foodborne illness as

quickly as is possible. If traceability investigations can reach a scientific conclusion

more rapidly, regulators and industry can respond more rapidly and pull any potentially

contaminated produce product from commerce before the outbreak spreads. This would

also mean that those products not implicated will not need to feel the drastic impact of a

recall that should not, in any way, affect the sales of their produce (as the spinach recall

in 2006 affected, and still affects, the entire industry26). If a traceability system cannot

allow regulators to more quickly respond and allow industry to more effectively recall

affected produce products from commerce, it is a failed system. PTI seeks to accomplish

this goal through the tracing of product through the supply chain in a common

nomenclature provided by a universally recognized standards organization—a success by

all comparisons and analysis of the needs of investigators.

              Where the PTI falls short is in its lack of direction and specificity in certain

elements of that nomenclature (such as supply-chain wide company codes) which are of

prime importance to the success of the initiative for the industry. In the opinions of many

of the interviewees for this paper, it is clear that part of the problem with PTI is in the

lack of leadership displayed in its application. One interviewee in particular put it

succinctly when he stated that the success of an enhanced traceability system in produce
      As evidenced through interviews 

is reliant upon industry leaders who can not only harness added value through

traceability systems but also communicate that value. Communication in terms of value

and education is lacking. As a result, confusion on such basic elements of the PTI as

reference #’s and lot/batch #’s has put a halt on the initiative for many organizations in

the supply chain.

       Even halfway through implementation, PTI is a hotly contested proposal in terms

of its ability to expedite trace-backs and have a positive impact on narrowing the scope of

implication. In the words of one interviewee, PTI is a ‘step backwards’ for many

industry members who not only have systems in compliance with existing regulations but

also capable of very granular identification of products by lot. The real momentum of

progress in traceability is dependent upon the right incentives. Those who do fail to play

by existing regulations are likely to hold back progress in initiatives such as PTI. For

those members of the supply chain, the incentive to not comply (or perhaps lack of

disincentive therein) is stronger than the incentive to comply. The success or failure of

any industry-led initiative is reliant upon those with the power to grant or withhold

incentive exercising that ability. In an industry where some past initiatives (such as

databar) were led by downstream partners who first expressed mandatory supplier

compliance and then purchased product from those who didn’t there is decreased

incentive to comply by those who did not reap benefits from past compliance. The truth

is, PTI has not been fully implemented and the industry is not yet in a position to test how

well it works—but it is only a matter of time before it will be. The occurrence of another

major foodborne illness outbreak in fresh produce is not so much an ‘if’ as a ‘when.’

       As it stands today, PTI provides a basis for partial traceability; that being said,

partial traceability doesn’t necessarily indicate any improvement in trace-back

investigations. PTI fails to establish uniformity and clarity throughout—an issue that has

become a resounding problem for those currently attempting to implement PTI in

industry. Lack of consistent direction in terms of company identification throughout the

supply chain, product identification that is not redundant in nature and lot specificity that

establishes a framework for suppliers would take the common nomenclature a long way.

Additionally, it is commonly cited that with PTI comes great organizational efficiencies

(such as payroll tracking and enhanced marketing visibility)—a point that is neither

described nor expanded upon by the steering committee that formed PTI.

       Finally, a lack of coherent direction on reporting is the most troubling shortfall.

The Harvard Business School study on tomato traceability (McEntire J. C., 2009) (as well

as the interviews conducted in the course of this paper) show that, in most cases, the data

needed for more rapid trace-back is already in existence due to existing regulations and

requirements. To that end, records in a database do not provide accurate or meaningful

answers to investigators simply through existence—it is through their combined use that

they create a powerful mechanism for tracing product from the farm to the fork in search

of common origins in which a contamination could occur. A successful traceability

system should equip investigators with enough answers to know the further questions

they need to ask. PTI, in its current state, leaves unanswered queries that must first be

matched and cleansed by investigators before proceeding to a comprehensive trace-

back—a problem that PTI was developed to fix.

       To this end, it is important to note again that the specific technology or

application used to create uniformity in fresh produce through the supply chain is not the

important issue at stake—the important issue at stake is getting that standardized data to

the right people in a timely manner. Before the industry embarks upon a time consuming

implementation of the data elements, the issue of transferring that data to investigators

such that rapid trace-backs can occur without the added need to disseminate differing

reporting formats for each and every member of the supply chain.


       The Produce Traceability Initiative does establish a necessary “first step” towards

creating a common nomenclature and methodology for transferring that nomenclature

throughout the supply chain. In that way, it cannot and, in the opinion of this author,

should not, be discarded or cast aside because of its shortcomings. It is evident that the

committee who first created the PTI realized that it would not fully answer the call of a

future traceability system from its first inception simply by its status as a “living”

document—it was created as a starting point.

       There are gaps in the PTI that need to be filled before it can effectively fulfill its

lofty goals—goals that don’t seem to be met by its current implementation. The problem

the industry has run into is that implementation is happening in silos rather than as a

group—until better communication is established between members of the supply chain,

the implementation, understanding and use of PTI will be difficult. If this

communication does not occur, gaps may never be filled and this may be yet another

initiative paid for but not demanded. It cannot be stressed enough that the utilization of

data mining techniques to narrow the scope of recalls depends of a quantity of data—if

the industry is not compelled to store and produce this data rapidly, these techniques fall


         It also remains to be seen whether or not the information that is provided by PTI

will be communicated and/or used by the FDA for investigative purposes. Even if the

data is created and stored industry wide, a methodology for reporting by industry and use

by investigators must be established or it becomes unclear how much standardization

actually helps.

         In the end, PTI isn’t and shouldn’t be regarded as “the solution,” or the “end-all”

to solving the problem of tracing perishable produce through the supply chain. Rather, it

is very much the end of the beginning.


    In the previous sections of this paper, we develop future goals for an ideal trace-back

system for the fresh produce industry and assess to what extent the PTI has the potential

to achieve these goals. In this section, we establish a set of principles as

recommendations for moving forward. These principles are not intended to be individual

steps towards meeting the future goals identified in this research. Rather, they are meant

to be high level guidelines that will help meet the future goals and fill the gaps identified

in the PTI.

    1. Stick to Uniformity Principles

         In order to effectively utilize data analysis techniques, it important that

    investigators not waste valuable time “connecting the dots.” If a principle is applied

    to one sector of the industry for recordkeeping, it should be uniformly applied to

    other sectors such that records are utilized in a consistent fashion. The point of

    traceability is to implicate the narrowest line of contaminated product(s) possible

    such that public health as well as financial damage may be limited throughout the

    industry to the maximum extent possible. A standard should be chosen, applied across

    the board, and used consistently. This includes identification of locations and

    companies that handle the product, identification of one product from another, and

    identification of source or origin of lots and batches. This uniformity needs to exist,

    without exception, between growing and ultimate consumption. Codes representing

    these elements should be universally recognizable and their meaning must be

    commonly understood across the supply chain.

    2. Standardize Product Reference Number Elements

       Related to uniformity but specific to one part of the nomenclature is the issue of

    assigning unique reference numbers. Unique reference numbers should identify the

    product uniformly. Otherwise the implementation of tracking systems will become

    monumental in scope and archaic in interpretation. A system that utilizes ten to

    twenty individual product codes for the same essential product for tens of thousands

    of products will undoubtedly lead to confusion on the part of regulators. The industry

    must settle on a common nomenclature for the sake of efficiency and the good of the

    industry. By reducing the ambiguity in defining products, the overall complexity of

    the system is reduced (both in implementation and ultimate use). Minimization of

    redundancy and interpretation should be a major goal in traceability data. Failure to

    clarify the interpretation of a product description simply because a standard level of

    detail was never agreed upon will exact a high price in terms of the speed and

    accuracy of an investigation.

    3. Create a Reporting Mechanism for Investigative Purposes

        Perhaps the most troubling aspect of the PTI is that there is not a single mention

    of reporting or recommendations or guidance for doing so. In reality, the FDA has

    recently made strides toward standardized technological reporting with the

    Reportable Food Registry. If the industry is working towards creating the dataset for

    the government, they should know that it will in fact be used by the government. If

    the industry implements the PTI and proceeds to print out volumes of reports (each in

    their own company-specific format) to mail and fax to the FDA, the industry has

    gained next to nothing. In the case of a food-borne illness outbreak, time is of the

    essence. It is critical that a method for rapidly communicating information to the FDA

    be developed so as to speed up future investigations, and reduce human illnesses (and

    deaths) and industry losses. There must be a uniform method of rapid data delivery

    to the FDA from each point in the supply chain in order to successfully minimize

    traceability investigation timeframes.

    4. Create an Environment of Open and Transparent Communication

        It will be critical that industry members within the supply chain and across

    commodities find an effective means of communication. It is clear, as discussed

    above, that issues of standardization and uniformity not directly met through the PTI

    must be addressed. However, it is also clear that it is not realistic for the

    representatives that form the steering committee to universally address these issues

    across all commodity groups. Industry members both within levels and across levels

    of the supply chain need to be in communication regarding the needs and uses of the

    datasets. Understanding the expectations and realities of the data that must be

    captured will go a long way towards collecting and reporting it. In this way,

    commodity specific associations should be involved in formulating a traceability plan

    for the future. If the common nomenclature to be utilized is that provided through

    GS1 (the standards organization), leaders and representatives across commodity

    groups will be prepared to formulate methodologies for such items as reference

    numbers and universally applied levels of detail for each commodity. Without proper

    communication and cooperation, the industry will simply create another layer of

    complexity using a new nomenclature.




H.R. 2749‐‐111th Congress: Food Safety Enhancement Act of 2009. (2009). Retrieved April 07, 
        2010, from GovTrack.us (database of federal legislation): 

Alliance for Food & Farming. (2010, January). Analysis of Produce Related Foodborne Illness 
        Outbreaks. Retrieved May 25, 2010, from Alliance for Food and Farming: 

Bottemiller, H. (2010, March 25). Vilsack: Food Safety Top Priority for Obama. Retrieved April 9, 
       2010, from Food Safety News: http://www.foodsafetynews.com/2010/03/vilsack‐food‐

Carpenter, R. W. (2009, December 9‐10). Public Comment of GS1 US before the U.S. Food and 
       Drug Adminsitration and USDA Food Safety and Inspection Services. Retrieved March 25, 
       2010, from Regulations.Gov: 

CDC. (2001, October 4). The Health Alert Network. Retrieved May 20, 2010, from Centers for 
        Disease Control: http://www.bt.cdc.gov/documentsapp/han/han.asp 

CDC. (2009, February 6). Multistate Outbreak of Salmonella Infections Associated with Peanut 
        Butter and Peanut Butter‐Containing Products ‐‐‐ United States, 2008‐‐2009. Retrieved 
        April 13, 2010, from Morbidity and Mortality Weekly Report: 

Center for Science in the Public Interest. (2008, July 03). Emergency Regs Needed for Tracking 
        Produce, Food Groups Say. Retrieved April 07, 2010, from 

Chadwell, J. (2007, September 2007). Natural Selection Foods Learns From Experience. Retrieved 
      May 18, 2010, from The Packer: http://thepacker.com/Natural‐Selection‐Foods‐learns‐

FDA. (2010, March). Reportable Food Registry for Industry. Retrieved May 22, 2010, from Food 
        and Drug Administration: 

FMI. (2010). Product Tracing Systmes for Food; Public Meeting. Retrieved from 

Golan, E., Krissoff, Barry, Kuchler, F., Calvin, L., Nelson, K., & Price, G. (2004). Traceability in the 
        U.S. Food Supply: Economic Theory and Industry Studies. Washington, DC: United States 
        Department of Agriculture. 

GS1 US. (2006, December). An Introduction to the Global Trade Item Number (GTIN). Retrieved 
       April 11, 2010, from GS1 US: 

IFDA. (2010). Product Tracing Systems for Food; Request for Comments. Retrieved from 

Levinson, D. R. (2009, March). Traceability in the Food Supply Chain. Retrieved April 22, 2010, 
       from Department of Health and Human Services: http://oig.hhs.gov/oei/reports/oei‐02‐

McEntire, J. C. (2009, December 1). Task Order No.7 Final Report: Revised Tracing Systems: An 
       Exercise Exploring Data Needs and Design. Retrieved May 15, 2010, from IFT: 

McEntire, J. C., Arens, S., Bernstein, M., Bugusu, B., Busta, F. F., Cole, M., et al. (2010). 
       Traceability (Product Tracing) in Food Systems: An IFT Report Submitted to the FDA, 
       Volume 1: Technical Aspects and Recommendations. Comprehensive Reviews in Food 
       Science and Food Safety, 9, 93‐158. 

Mead, P. S., Slutsker, L., Dietz, V., McCaig, L. F., Bresee, J. S., Shapiro, C., et al. (1999). Food‐
       Related Illness and Death in the United States. Emerging Infectious Diseases, 607‐625. 

National Center for Infectious Diseases. (2009, March 9). What is PulseNet? Retrieved April 6, 
       2010, from CDC PulseNet: http://www.cdc.gov/pulsenet/whatis.htm 

Porter, D. V., & Lister, S. A. (2006). Food Safety: Federal and State Response to the Spinach E. coli 
        Outbreak. Washington D.C.: Congressional Research Service; U.S. Library of Congress. 

Produce Marketing Association. (2009, March). About PTI. Retrieved February 17, 2010, from 
       Produce Traceability Initiative: 

Produce Marketing Association. (2009). PTI Action Plan. Retrieved February 18, 2010, from 
       Produce Traceability Initiative: 

Produce Traceability Initiative. (2009, May 22). Action Plan. Retrieved April 11, 2010, from The 
       Produce Traceability Initiative: 

PTI Steering Committee. (2009a, October). Best Practices for Private Label/Brand. Retrieved 
        April 11, 2010, from Produce Traceability Initiative: 

PTI Steering Committee. (2009b, September). Best Practices for Cross Docking and Load Only 
        Service. Retrieved April 2, 2010, from Produce Traceability Initiative: 

PTI Steering Committee. (2009c, June). Best Practices for External Product Substitutions. 
        Retrieved April 10, 2010, from Produce Traceability Initiative: 

PTI Steering Committee. (2009d, March). PTI Toolkit Introduction. Retrieved April 10, 2010, from 
        Produce Traceability Initiative: 

PTI Steering Committee. (2009e, March). Traceability at the Item Level. Retrieved April 11, 2010, 
        from Produce Traceability Initiative: 

Shin, A. (2006, December 11). Outbreaks Reveal Food Safety Net’s Holes. Retrieved April 07, 
         2010, from Washington Post: http://www.washingtonpost.com/wp‐

The Great Atlantic & Pacific Tea Company, Inc. (2010). Produce FAQ's. Retrieved June 5, 2010, 
       from SuperFresh. 

The President's Food Safety Working Group. (n.d.). Food Safety Working Group: Learn. Retrieved 
       April 10, 2010, from Food Safety Working Group: 

U.S. FDA. (2005, January). Presentation: Overview of Bioterrorism Act Establishment and 
        Maintenance of Records Final Rule. Retrieved April 18, 2010, from U.S. Food and Drug 

University of Rhode Island. (2000, May). Rhode Island Food Safety Education: Glossary. 
        Retrieved May 14, 2010, from University of Rhode Island Cooperative Extension: 

USDA. (2009). Day 1: Overview of Outbreak Response and Product tracing in both FDA and FSIS 
       Regulated Foods. Transcript: Improving Product Tracing of Foods: More Rapid Response. 
       . Washington, D.C.: Free State Reporting, Inc. 

USDA. (2009, May 22). Foodborne Illness Cost Calculator. Retrieved February 17, 2010, from 
       USDA Economic Research Service: http://www.ers.usda.gov/Data/FoodborneIllness/ 

Van Cassell, M. (2007, September 12). Spinach Industry Struggles. Retrieved April 10, 2010, from 
       The Gilroy Dispatch: http://www.gilroydispatch.com/news/225031‐spinach‐industry‐



Appendix	A:		Base	Open‐Ended	Questions	for	Supply	Chain	

Background Questions

1.    Commodities shipped/handled/distributed

2.    Number and physical distance between locations

3.    Quantity of cases processed each year

4.    Primary lines of business (e.g. Harvesting, Marketing, Distribution, Cooling, etc.)

5.    Acres harvested (if applicable)

6.    Current food safety measures taken and existence of mock recall program

7.    Has the firm been directly or indirectly involved with an outbreak investigation?

Internal Traceability Systems Questions

1.    Current layout of traceability system

2.    Current time from notification to identification of source lot in recall events

3.    Current method of recordkeeping (i.e. paper and electronic)

4.    Current data retention policies

5.    Technologies engaged to accomplish recordkeeping

6.    Current level of detail retained for traceability purposes

External Traceability Systems Questions

1.    How well equipped are other trading partners at providing recall data?

2.    How is traceability data reported to external bodies (if applicable)?

3.    What shortcomings exist in current industry traceability practices?

4.    How can current shortcomings be remedied?

5.    What level of detail makes the most sense for traceback? Why?

6.    What recordkeeping requirements is the firm subject to?

7.    Does the firm keep all recordkeeping requirements up to date?

PTI Specific Questions

1.    Does PTI enhance current internal traceability practices? How?

2.    Does PTI enhance industry wide traceability practices? How?

3.    Has the firm begun PTI implementation?

4.    What obstacles remain for your firm’s implementation of PTI?

5.    If applicable, what would you change about PTI and why?


Shared By: