GROWTH PROBABILITY OF ESCHERICHIA COLI O157:H7 AND LISTERIA
MONOCYTOGENES IN PACKAGED FRESH-CUT ROMAINE MIX AT
FLUCTUATING TEMPERATURES DURING SIMULATED TRANSPORT, RETAIL
STORAGE AND DISPLAY
By
Wenting Zeng

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Food Science
2012

ABSTRACT
GROWTH PROBABILITY OF ESCHERICHIA COLI O157:H7 AND LISTERIA
MONOCYTOGENES IN PACKAGED FRESH-CUT ROMAINE MIX AT FLUCTUATING
TEMPERATURES DURING SIMULATED TRANSPORT, RETAIL STORAGE AND
DISPLAY
By
Wenting Zeng
Temperature abuse during commercial transport and retail sale of leafy greens negatively
impacts both microbial safety and product quality. The effect of fluctuating temperatures on
Escherichia coli O157:H7 and Listeria monocytogenes growth in commercially bagged salad
greens was simulated during transport, retail storage, and display. Time/temperature profiles for
bagged salads were obtained from 16 transportation routes (432 profiles) and during retail
storage (4867 profiles) and display (3799 profiles) at 9 supermarkets. Five different
time/temperature profiles collected during transport, retail storage, and display were then
duplicated in a programmable incubator to assess pathogen growth in bags of romaine mix.
Microbial growth predictions using the Baranyi/Ratkowsky and McKellar models were validated
by comparing the root mean square error (RMSEs) and biases between the growth data and
model predictions. Both models yielded acceptable RMSEs and biases. Monte Carlo simulations
were then performed to calculate the probability distribution of microbial growth from
8,122,127,472 scenarios. E. coli O157:H7 and L. monocytogenes populations increased a
maximum of 3.1 and 3.0 log CFU/g at retail storage. Based on the simulation results, both
pathogens generally increased < 2 log CFU/g. However, retail storage duration can significantly
impact the extent of E. coli O157:H7 and L. monocytogenes growth. This first large-scale US
study should be useful in filling some of the data gaps in current risk assessments for leafy
greens.

To my mom Changbin Sun, my dad Jinxiang Zeng, and my husband Yuanteng Pei
I dedicate this thesis
谨以此献给我的母亲, 父亲和我的爱人

iii

ACKNOWLEDGEMENTS
I would like to sincerely thank my major advisor Dr. Elliot Ryser for bringing me in the
food safety area. It is my great pleasure to work for and learn from him. His brightness, patience,
dedication, support, and encouragement have inspired me for all of my accomplishments. I am
grateful to Drs. Bradley Marks and John Linz for graciously serving on my advisory committee.
Their guidance and support keep me on the right track of research directions. I am very thankful
to Dr. Sanghyup Jeong and Isabel Tenorio, without their help on the predictive models I might
not have completed my research. I also want to express my thanks to Drs. Keith Vorst and Wyatt
Brown from California Polytechnic State University for sharing millions of their temperature
data. I would like to specially thank Dr. Phil Elliott, without his suggestion I might not have used
the Monte Carlo simulation. I would like to thank Drs. Fernando Pérez-Rodríguez from Spain
and Donald Schaffner for the guidance on the risk assessment and probability distributions.
I want to acknowledge all of my labmates, past and present, for their vital help, friendship,
and support. I really had a great time here. Amy and Rick Oldejans, from MSU OISS friendship
family program, have moved me hundreds of times for their warm care. My husband Yuanteng
Pei is always supportive. I am dedicated to my mom and dad, for their understanding,
encouragement, and their tolerance for the long time of missing me.
Wenting Zeng

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES .......................................................................................................... vii
CHAPTER 1: Introduction and literature review.. ............................................................... 1
Introduction.. ......................................................................................................................... 1
Literature review ................................................................................................................... 5
1.1 Foodborne pathogen............................................................................................ 5
1.1.1 Escherichia coli ....................................................................................... 5
1.1.2 Listeria monocytogenes ........................................................................... 6
1.2 Foodborne illness ................................................................................................ 7
1.3 Contamination ................................................................................................... 13
1.3.1 Pre-harvest contamination ..................................................................... 13
1.3.2 Postharvest contamination ..................................................................... 14
1.4 Temperature ...................................................................................................... 17
1.4.1 Growth of E. coli O157:H7 and L. monocytogenes at constant
temperatures ............................................................................................ …17
1.4.2 U. S.and international guidelines on temperature control .................... 19
1.4.3 Temperature fluctuations ...................................................................... 19
1.4.4 Temperature monitoring in industry..................................................... 27
1.5 Modeling bacterial growth .............................................................................. 29
1.6 Risk assessment .............................................................................................. 33
CHAPTER 2: Growth of Escherichia coli O157:H7 and Listeria monocytogenes
in packaged fresh-cut romaine mix at fluctuating temperatures during commercial
transport, retail storage and display .................................................................................. 36
2.1 Abstract ........................................................................................................... 36
2.2 Introduction ..................................................................................................... 37
2.3 Materials and methods .................................................................................... 40
2.3.1 Temperature monitoring during commercial transport and retail ........ 40
2.3.2 Culture preparation ............................................................................... 41
2.3.3 Inoculation ............................................................................................ 42
2.3.4 Incubation ............................................................................................. 43
2.3.5 Microbial analyses ................................................................................ 43
2.3.6 Model validation .................................................................................. 44
2.3.7 Monte Carlo simulation ........................................................................ 48
2.3.8 Statistical analyses ................................................................................ 48
2.4 Results ............................................................................................................. 48
2.4.1 Temperature selection for lab validation .............................................. 48
2.4.2 Microbial growth .................................................................................. 53
2.4.3 Model validation................................................................................... 64
2.4.4 Monte Carlo simulation ........................................................................ 64
2.5 Discussion ....................................................................................................... 78

v

APPENDICES .................................................................................................................. ..84
Appendix A Matlab code for Escherichia coli O157:H7………………………....84
a Baranyi dynamic model……………………………….………………....84
b Bacterial growth predicted from Baranyi/Ratkowsky model……………88
c Monte Carlo simulation for E. coli O157:H7………………..………......89
Appendix B Matlab code for Listeria monocytogenes…………............................94
REFERENCES ................................................................................................................... 95

vi

LIST OF TABLES

Table 1.1 Lettuce- and leafy green-associated outbreaks…………………………………9
Table 1.2 Outbreaks of listeriosis associated with produce (Hoelzer and others 2012)….12
Table 2.1 Baranyi and Ratlowsky model parameters for growth of E. coli O157:H7 and L.
monocytogenes on Iceberg lettuce (Koseki and Isobe 2005a) …………………………..46
Table 2.2 Root Mean Square Errors and biases between selected sensor and duplicated lab
temperatures during transport, retail storage and display ................................................ .62
Table 2.3 Growth of mesophiles and psychrotrophs in bagged romaine mix during transport,
retail storage, and diplay…………………………………………………………………63
Table 2.4 Root Mean Square Errors and biases between observed pathogen growth and the
Baranyi and McKellar model predictions ……………………………………………….66

vii

LIST OF FIGURES

Figure 1.1 Production and distribution chain for leafy greens (Carrasco and others 2010)… 16
Figure 1.2 Temperature history for lettuce from the field to retail display in Japan (Koseki and
Isobe 2005a). ............................................................................................................................22
Figure 1.3 Temperature profiles for different locations (top, middle and bottom) of salad bags
during retail display. Dashed horizontal lines correspond to the recommended temperature range
for fresh-cut vegetable salads (Nunes and others 2009)…………………………………….. 26
Figure 2.1 Average temperatures for all profiles collected during transport, retail storage, and
display with the 5 selected temperatures indicated (A, B, C) and the time/temperature profiles for
each of the average temperatures selected (D, E, F)……………………………………….. ..50
Figure 2.2 Growth of E. coli O157:H7 and L. monocytogenes under selected temperatures during
transport (A1 – A5), retail storage (B1 - B5) and retail display (C1 – C5).. .......................... 54
Figure 2.3 Probability distributions for E. coli O157:H7 (A, B) and L. monocytogenes (C, D)
after 24 and 72 h of retail storage, respectively. ......................................................................67
Figure 2.4 Probability distributions for E. coli O157:H7 (A, B, C, D) and L. monocytogenes (E,
F, G, H) growth in romaine mix transported during spring, summer, fall and winter……….69
Figure 2.5 Probability distributions for E. coli O157:H7 (A, B, C) and L. monocytogenes (D, E,
F) growth in romaine mix at the top, middle, and bottom of the truck during transport…….73
Figure 2.6 Probability distributions for E. coli O157:H7 (A, B) and L. monocytogenes (C, D)
growth in romaine mix at the front and back of the retail display ...........................................76

viii

CHAPTER 1
INTRODUCTION
Due to the abundant health benefits and year-round availability of fresh produce,
consumption of fresh and fresh-cut leafy greens in the United States has increased over the past
few decades (Calvin and others 2003; ERS 2003; Meng and Doyle 2002; Nicola and others
2006), with per capita consumption of romaine lettuce increasing from 1.5 to 5.1 kg from 1982
to 2003 (ERS 2003). Salad greens, which are consumed raw, are highly prone to microbial
contamination in the field. Although relatively small quantities of leafy greens are presumed to
be contaminated with bacterial pathogens, outbreaks of illness continue to be reported.
Since 1995, at least 19 outbreaks of foodborne illness were caused by Escherichia coli
O157:H7 linked to fresh or fresh-cut lettuce as the vehicle of infection (Falkensten, 2010). These
outbreaks can result from low levels of contamination since as few as 10 cells of E. coli
O157:H7 are able to cause human illness (FDA 1998). Although traceability to growers was not
completed in all outbreak investigations, eight of the outbreaks associated with lettuce and
spinach were traced back to Salinas, California.
In addition to E. coli O157:H7, Listeria monocytogenes can also grow on fresh produce
when present (De Simon and others 1992; Francis and others 1999; Szabo and others 2000). L.
monocytogenes has thus far been linked to 16 produce-related outbreaks internationally, with a
recent large-scale cantaloupe outbreak involving 139 cases, including 29 deaths and 1
miscarriage (CDC 2011b). Several recent recalls also have been issued for Listeria-contaminated
produce, including bagged salads, baby spinach, Napa kimchi, chopped or shredded romaine,
and diced red onions in the U. S. and Canada.

1

These outbreaks have greatly shaken consumer confidence concerning the consumption of
ready-to-eat fresh produce. As a result, the produce industry is being driven to make major
changes in how leafy greens are grown, harvested, and processed. The industry is reviewing
current operation guidelines for minimizing microbial food safety hazards as well as other
available information regarding pathogen reduction strategies for fresh produce. This thesis
focuses on the effect of temperature abuse during postharvest transportation and distribution,
with this research expected to provide practical solutions that can be used to enhance safety and
keeping quality of leafy greens during transportation, retail storage, and display.
Produce can be easily contaminated at any point in the farm-to-fork continuum. The high
moisture and nutrient content, together with the large surface area of fresh-cut produce, make
these products highly susceptible to microbial contamination and subsequent growth. The many
benefits of low temperature storage on product quality and microbial growth during transport and
distribution have been well documented (Jacxsens and others 2002; Koseki and Isobe 2005a;
Luo and others 2010). Therefore, the U. S. along with many other countries has developed
temperature guidelines for most perishable foods during transport and retail sale. The U. S. Food
and Drug Administration 2009 Food Code requires that ready-to-eat fruits and vegetables be
refrigerated and received at or below 5°C (41°F) to limit growth of organisms of public health
concern (FDA 2009). The Canadian Food Inspection Agency code of practice for minimally
processed ready-to-eat vegetables requires that such products be maintained at 4°C during
transportation and storage (Canadian Food Inspection Agency 2009). The EU food hygiene
legislation (2006) requires foods which are likely to support the growth of pathogenic microorganisms to be held at or below 8°C. Nunes and others (2008) summarized the recommended
storage temperatures of 1 to 3°C for various fruits, vegetables, and bagged salads. While these

2

temperature ranges are considered ideal, temperatures of 5 or 8°C are more realistic in the
distribution chain.
Temperature monitoring during distribution, retail storage, and display provides a major
advantage in the ability to manage temperature sensitive products in the cold chain with most
food and pharmaceutical suppliers of perishable goods already tracking time/temperature
histories. Many large-scale retailers have now mandated the use of temperature sensors and data
loggers for their suppliers, with others using various temperature monitoring systems to track
specific cases and pallets of product in the supply chain. In this study, temperature sensors were
used to monitor temperature fluctuations and collect time/temperature profiles during transport,
at retail storage room, and retail display cases.
However, maintaining a consistently low temperature throughout the distribution chain is
challenging. Opening of the truck doors during loading and unloading, outside temperature
extremes, and retail storage/display conditions can all lead to temperature fluctuations. Thus far,
only three studies from Canada, Japan, and Belgium have assessed the growth of E. coli
O157:H7 and L. monocytogenes at real-time temperatures recorded during preharvest,
transportation, and retail sale of fresh-cut bagged salad greens (McKellar and others 2012;
Koseki and Isobe 2005a; Rediers and others 2008). Moreover, the time/temperature profiles from
these three studies may not sufficiently reflect that segment of the U. S. cold chain during
transport and distribution.
Consequently, the objectives of this study were to:
1. Assess the impact of various real-time time/temperature histories on the growth of E.
coli O157:H7 and L. monocytogenes as well as psychrotrophic and mesophilic bacteria in
commercial romaine mix during transport, retail cold storage, and display;

3

2. Validate the Baranyi/Ratlowsky and McKellar models using laboratory data;
3. Determine the probability distributions for E. coli O157:H7 and L. monocytogenes
growth during transport, retail storage, and display.

4

LITERATURE REVIEW
1.1 Foodborne pathogens
Foodborne pathogens cause widespread and growing public health problems over the
world. The global incidence of foodborne disease leads to approximately 1.8 million deaths
annually, with most cases from less developed countries (WHO 2007). In developed countries
foodborne pathogens are responsible for millions of cases of gastrointestinal diseases each year,
resulting in significant medical cost as well as productivity loss. Emerging foodborne pathogens
and foodborne diseases are also likely to occur in recent years, driven by factors such as genetic
evolution, changes in agricultural and manufacturing practices, and the diversity of host status.
There are also growing concerns for terrorism reasons to contaminate food and water supplies in
attempts to incapacitate people and disrupt economic growth. Fuelled by these concerns, 25
major microbial pathogens are monitored by various surveillance programs in the U. S., such as
the Foodborne Diseases Active Surveillance network (FoodNet), the National Notifiable
Diseases Surveillance System (NNDSS), the Cholera and other Vibrio Illness Surveillance
(COVIS) system, the National Tuberculosis Surveillance System (NTSS), and the Foodborne
Disease Outbreak Surveillance System (FDOSS).
1.1.1 Escherichia coli
Escherichia coli is a facultative anaerobic, Gram-negative, non-sporulating bacterium
predominantly found in the colonic flora of warm-blooded animals (Todar 2008a). Although
most strains are harmless, some pathogenic serotypes such as E. coli O157:H7 are responsible
for serious food poisoning and product recalls. Serotyping of pathogenic E. coli is based the
presence of different O (lipopolysaccharide, LPS), H (flagellar), and K (capsular) antigens

5

(Nataro and Kaper 1998). Among the estimated 50,000 to 80,000 serotypes, six categories of
pathogenic E. coli are recognized based on serological characteristics and virulence properties enterotoxigenic E. coli (ETEC), enteropathogenic E. coli (EPEC), enteroinvasive E. coli (EIEC),
enterohemorrhagic E. coli (EHEC), enteroaggregative (EAEC), and diffuse-adhering E. coli
(DAEC) (Guerrant 2005; Nguyen 2006; Scaletsky 2002). E. coli O157:H7 belongs to EHEC,
causing human diseases including non-bloody diarrhea, hemorrhagic colitis, hemolytic uremic
syndrome (HUS), acute kidney failure, and thrombotic thrombocytopenic purpura.
1.1.2 Listeria monocytogenes
The genus Listeria comprises a group of facultative anaerobic, Gram-positive, non-spore
forming, short rod-shaped bacterium and contains the following species: L. monocytogenes, L.
innocua, L. invanovii, L. seeligeri, L. welshimeri, L. grayi, L. marthi and L. rocoyrtiae. Listeria
monocytogenes is of greatest concern as a foodborne pathogen of the aforementioned species due
to its ability to grow at refrigeration temperatures (Rocourt 1999). Its resistance to acid, salt, and
heat makes this pathogen a major concern to food manufacturers. Thus, while typically found in
the natural environment, including soil and water, as well as wild and domesticated animals, L.
monocytogenes can be commonly found in many food-processing environments and typically
enters food as a post-processing contaminant.
L. monocytogenes is the causative agent of listeriosis, which has a fatality rate of about
20% (Scallan and others 2011). Thirteen serotypes are recognized, with serotypes 1/2a, 1/2b, and
4b responsible for most foodborne illnesses (Farber and others 1991). Listeriosis usually lasts 710 days and occurs predominantly in newborns, the elderly, and immunocompromised
individuals (Hof 1996). Symptoms of listeriosis include fever, muscle aches, vomiting, headache,
nausea and diarrhea. After crossing the blood-brain barrier, L. monocytogenes can

6

cause meningitis, an infection of the covering of the brain and spinal cord, characterized by
headache, stiff neck, confusion, loss of balance, and convulsions (Todar 2008b).
1.2 Foodborne illness
Foodborne illnesses have become a considerable public health burden in the United States.
An estimated 48 million foodborne illnesses, 128,000 hospitalizations, and 3,000 associated
deaths occur annually, with 55% of outbreaks traced to bacterial pathogens with known etiology
(CDC 2006b; Scallan and others 2011). A total of 63,153 annual cases of shiga toxin-producing
E. coli O157 (STEC O157), 112,752 cases of non-O157 shiga toxin-producing E. coli (nonSTEC O157), and 1,591 cases of L. monocytogenes infection were estimated by Scallan and
others (2011). The hospitalization rate for laboratory-confirmed cases was highest for L.
monocytogenes (94%), followed by Vibrio vulnificus (91%) and Clostridium botulinum (83%),
whereas the death rate was highest for C. botulinum (83%), followed by V. vulnificus (35%) and
L. monocytogenes (16%) (Scallan and others 2011). The estimated annual cost of medical care
and premature death for the major bacterial pathogens exceeds $6.9 billion (ERS 2010). The cost
of E. coli O157 alone is at least $478 million (ERS 2010).
Numerous foodborne outbreaks involving E. coli O157:H7 have shaken consumer
confidence. An estimated 63,153 illnesses are caused by E. coli O157:H7 in the United States
annually, with 2,138 hospitalizations and 20 deaths (Scallan and others 2011). E. coli O157:H7
was first identified as a foodborne pathogen in 1982 during an investigation of hemorrhagic
colitis (Riley and others 1983). It has remained a threat to the food industry since 1993 when a
large multistate outbreak was linked to ground beef. Ground beef, produce, and dairy products
are the most common vehicles associated with E. coli O157:H7 outbreaks.

7

During 1999-2008, contaminated produce accounted for 21% of all foodborne outbreaks
and 31% of all foodborne illnesses in the United States, with leafy greens frequently implicated
(DeWaal and others 2012). The term leafy greens includes iceberg lettuce, romaine lettuce, leaf
lettuce, butter lettuce, baby leafy lettuce, escarole, endive, spring mix, spinach, cabbage, kale,
arugula and chard. Since 1993, at least 800 cases of illness and 8 deaths have been traced to leafy
greens (FDA 2005a), particularly California-grown lettuce (iceberg, romaine, red leaf, and
mesclun) and spinach (Table 1.1). In one 2005 outbreak, 250,000 bags of pre-washed and pre-cut
salad greens were recalled. In late summer of 2006, E. coli O157:H7 contaminated Californiagrown baby flat-leaf spinach was responsible for 199 cases of illness in 28 states, including 31
cases of hemolytic uremic syndrome, 141 hospitalizations, and 3 deaths (FDA, 2006). In
December 2006, commercially-shredded lettuce used by Taco Bell sickened over 71 people with
53 hospitalizations and 8 cases of kidney failure traced to E. coli O157:H7 (CDC 2006c).
Produce outbreaks usually peak in summer and fall. A total of 34% of produce outbreaks were
linked to lettuce, 18% apple cider or apple juice, 16% to salad, 11% to coleslaw, 11% to melons,
8% to sprouts and 3% to grapes (Rangel and others 2005). With increasing frequency in recent
years, the median number of cases in produce-associated outbreaks was significantly larger than
that of ground beef outbreaks.
Other serogroups of shiga toxin-producing E. coli are also becoming a major concern
incuding serotypes O26, O45, O103, O111, O121 and O145 – now known as ‘the big six’. In
June 2011, E. coli O104:H4 contaminated sprouts were responsible for 2229 reported infections,
852 cases of hemolytic uremic syndrome and 32 deaths in Germany as well as 6 confirmed cases
and 1 death in the US (CDC 2011a). In May 2010, a multistate outbreak of human E. coli O145

8

Table 1.1 Lettuce- and leafy green-associated outbreaks (Anonymous, 2011; Falkensten).
Date
May 2011
May 2010
Spring 2010
Jul.-Aug. 2009
Aug. 2009
Feb. 2009
Nov. 2008

Causative Agent
Salmonella Typhimirum
E. coli O145
Salmonella spp
Salmonella spp
Salmonella spp
Salmonella spp
E. coli O157:H7

Oct. 2008
Sep. 2008
Aug. 2008
Jun. 2008
May 2008
Jul. 2007
Jul. 2007
Feb. 2007
Jan. 2007
Nov. 2006

E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
Shigella sonnei
E. coli O157:H7
Norovirus
Norovirus
E. coli O157:H7

Oct. 2006
Sep. 2006
Sep. 2006
Sep. 2005
Nov. 2004
Nov. 2003
Sep. 2003
Dec. 2002
Nov. 2002
Jul. 2002
Nov. 2001
Oct. 2000
May 2000
May 2000
Feb. 2000
Oct. 1999
Oct. 1999
Oct. 1999
Oct. 1999
Sep. 1999
Sep. 1999
Sep. 1999
Aug. 1999

E. coli O157:H7
Norovirus
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
Campylobacter jejuni
Norovirus
Norovirus
E. coli O157:H7
E. coli O157:H7
E. coli O157:H7
Norovirus
E. coli O157:H11
Norovirus
E. coli O111:H8
Norovirus

Illnesses
Reported
15
23
?
> 100
124
5
130

Source
Salad?
Lettuce (Romaine)
Lettuce
Lettuce
Lettuce?
Lettuce?
Lettuce

Location

Lettuce
Lettuce
Spinach
Lettuce
Lettuce
Salad
Lettuce
Lettuce
Salad
Lettuce
Spinach
Salad
Lettuce
Lettuce (Romaine)
Lettuce
Spinach
Lettuce (Mixed salad)
Lettuce
Lettuce
Lettuce (Romaine)
Lettuce
Salad
Salad
Salad
Salad
Lettuce
Lettuce
Salad
Salad
Lettuce
Lettuce
Salad
Salad

59
34
5
9
9
72
26
8
9
77
199 (3
deaths)
9
30
11
6
16
51
3
13
55
20
6
13
3
7
40
47
5
16
6
115
58
25

9

IL
MI, OH, NY, TN
Upper Midwest
5 States
?
MS
Ontario, Canada
Multistate, US;
Ontario, Can
MI
OR
WA
CA
CA
AL
TN
ID
Multistate, U. S.
Multistate, U. S.
VA
Ontario, Canada
?
NJ
CA
CA
MN
IL
WA
TX
IN
CT
OH
OH
PA
OH
OR
WA
WA
WA
TX
MN

Table 1.1 (cont’d)
Date
May 1999
Feb. 1999
May 1998
Jun. 1996
May 1996
Oct. 1995
Sep. 1995
Sep. 1995
Jul. 1995

Causative Agent
Norovirus
E. coli O157:H9
E. coli O157:H7
E. coli O153:H49
E. coli O157:H10
E. coli O153:H46
E. coli O153:H47
E. coli O157:H7
E. coli O153:H48

Illnesses
Reported
28
65
2
7
61
11
30
20
74

10

Source
Salad
Lettuce (Iceberg)
Salad
Lettuce
Lettuce (Mesclun)
Lettuce (Iceberg)
Lettuce (Iceberg)
Salad (Romaine)
Lettuce (Romaine)

Location
FL
NE
?
NY
?
OH
ME
ID
MT

was linked to shredded romaine lettuce, with 26 confirmed and 7 probable cases in Michigan,
New York, Ohio, Pennsylvania, and Tennessee (CDC, 2010).
In addition to E. coli O157:H7, L. monocytogenes can also grow on fresh produce when
present (De Simon and others 1992; Francis and others 1999; Szabo and others 2000). L.
monocytogenes was first recognized as a foodborne pathogen in 1981 when contaminated
coleslaw sickened at least 41 people and caused 18 deaths (Schelech and others 1983). L.
monocytogenes has this far been linked to 16 produce-related outbreaks (Table 1.2).

In

September of 2011, cantaloupes grown in Colorado were implicated in a multistate outbreak that
included 139 cases of illness, 29 deaths and 1 miscarriage (CDC 2011b). From January to
October 2010, L. monocytogenes- contaminated celery from Sangar Fresh Cut Produce in Texas
was responsible for 10 listeriosis cases and 5 deaths (FDA 2010a). Several recent recalls have
also been issued for Listeria-contaminated produce. These include the 2010 recall of Tuv Taam
salads brand Nova Lox packaged salad (FDA 2010b), the 2011 recalls of various bagged salad
products from River Ranch Fresh Foods, chopped or shredded romaine from True Leaf Farms,
and organic baby spinach and cut Napa Kimchi, and the 2012 recall of romaine lettuce, bagged
salads and diced red onions in the U. S. and Canada.

11

Table 1.2 Outbreaks of listeriosis associated with produce (Hoelzer and others 2012).
Illnesses
Year
Source
Reported
Location
1979
Tuna fish and chicken salads with
20 (5 death)
U. S.
celery, lettuce, and tomatoes; cheese
1981
Coleslaw
41 (18 death)
Canada
1986
Vegetables or unpasteurized milk
28 (5 death)
Australia
United
1988
Vegetable rennet
1 miscarriage
Kingdom
United
1989
Salad containing lettuce (hypothesized) 1
Kingdom
1989
Salted mushrooms
1
Finland
-Frozen broccoli, cauliflower
7
U. S.
1993
Rice salad (cheese, pickled vegetables,
18
Italy
frozen vegetables, hard-boiled eggs)
1997
Corn and tuna salad
1566
Italy
1998-1999 Fruit salad
6 (5 death)
Australia
-Melon and watermelon
-U. S.
2001
Potato salad
56
U. S.
2006
Taco/nacho salad
2
U. S.
2010
Alfalfa sprouts
20
U. S.
2010
Celery
10 (5 death)
U. S.
2011
Cantaloupe melon
146 (30 death)
U. S.

12

Produce outbreaks have been most commonly associated with fresh-cut bagged products.
The increasing number of outbreaks traced to fresh and fresh-cut produce points to the need to
develop farm-to-table strategies to better unsure their safety.
1.3 Contamination
1.3.1 Pre-harvest contamination
The potential routes for the spread of E. coli O157:H7 and L. monocytogenes in the natural
environment are numerous and include the following: irrigation water, airborne dust, cattle feedlot run-off, wild animals (e.g., feral pigs, deer, birds), manure particulates, improperly treated
compost, field processing (e.g., cores/trimmings, farm equipment), and employee hygiene (DiezGonzalez and Mukherjee 2009). Once present in the soil, survival of E. coli O157:H7 and L.
monocytogenes is further influenced by variations in soil type, background microflora and
climate (e.g., UV light, temperature, relative humidity). In addition, cattle, feral pigs, and other
animals may also become asymptomatic carriers of E. coli O157:H7 and L. monocytogenes
(Mandrell 2009). When leafy greens become contaminated, pathogen survival is based on
relative humidity, temperature, overall health, nutrients, and maturity of the plant. The high
moisture and nutrient content, together with the large surface area, make fresh produce highly
susceptible to microbial growth.
Increased outbreaks of E. coli O157:H7 and L. monocytogenes in recent years have
prompted significant efforts towards the development and implementation of improved Good
Agricultural Practices (GAPs) to minimize pathogen contamination in the field. One such
outcome was issuance of the California Leafy Green Products Handler Marketing Agreement in
2007, together with modified Good Agricultural Practices specifically for the production and

13

harvest of leafy greens. While adherence to GAPs will aide in pathogen reduction, current leafy
green harvesting, post-harvest processing, and distribution practices cannot guarantee that these
products will be free of E. coli O157:H7, L. monocytogenes, Salmonella, norovirus or other
foodborne pathogens of public health concern.
1.3.2 Postharvest contamination
Leafy greens are prone to contamination during washing and sorting in the packinghouse,
distribution in retail stores, food-service facilities, and at the home (Fig. 1.1). Contamination can
result from contact with wash water, cooling ice, manure and compost, equipment in the field,
transportation vehicles, cross-contamination from other foods, and improper storage, packaging,
and preparation (FDA, USDA, and CDC 1998). Leafy greens destined for the fresh-cut market
are typically vacuum-cooled, shredded, washed in disinfectant water, dewatered by shaking and
centrifugation, and packaged in a modified atmosphere. Key contamination points observed by
Kaneko and others (1999) at two fresh-cut produce facilities included the trimming knives, wash
water, the blades located in the interior surfaces of a mechanical slicer, and the interior of a
dewatering centrifuge, with most samples taken during production yielding aerobic plate counts
above 5 log CFU/cm2.
Washing is an essential step for removing soil and debris, decreasing the microbial load,
and increasing product shelf life. While washing has abundant benefits, the wash water may
become a vehicle for pathogen cross-contamination since the water is recirculated in the wash
system (Buchholz and others 2009; FDA 2008). Therefore, addition of sanitizers to the water
during leafy green processing is a widely used in the industry to reduce microbial populations
and cross contamination. Chlorine-based sanitizers, which are most widely used in industry,

14

function by oxidizing intracellular enzymes (Cho and others 2010) and typically reduce
microbial populations by 1 - 2.5 log CFU/g (Beuchat and others 2004).
In some cases, these sanitizers may be no more effective than water, particularly when the
organic load increases in the wash water during processing, with the ongoing recalls reinforcing
the unreliability of commercial sanitizers (Parish 2003; Beuchat and others 2004). Furthermore,
those cells that enter the stomata of leafy greens or those that reside in hydrophobic regions,
cavities, and on rough surfaces are likely to escape sanitizer exposure (Takeuchi and Frank 2000;
Zhou and others 2009). Biofilm production by E. coli O157:H7 and L. monocytogenes, which
has been observed on stainless steel surfaces in the presence of lettuce juice (Ryu and others
2004), can also protect pathogens from sanitizers and increase the extent of cross contamination.
Moreover, internalization of pathogens from any processing step that cools, tears, cuts, or shreds
leafy greens can withstand the exposure to sanitizers.

15

Figure 1.1 Production and distribution chain for leafy greens (Carrasco and others 2010).

16

After processing, bagged salads are shipped through the cold chain transport networks with
no effective strategies availble to reduce microbial contamination after processing. Therefore, the
safety of fresh-cut leafy greens is primarily based on temperature control. Temperature abuse
leading to growth/survival of E. coli O157:H7 and L. monocytogenes during storage, transport,
retail sale, and home/restaurant refrigeration is one of the major concerns in the current risk
assessments for leafy greens.
1.4 Temperature
1.4.1 Growth of E. coli O157:H7 and L. monocytogenes at constant temperatures
E. coli O157:H7 will grow in a temperature range from 8 to 48°C, with the optimal growth
at 37°C (Buchanan and Klawitter 1992). A comprehensive study on the growth kinetics of E. coli
O157:H7 as impacted by temperature, pH, and salt concentration was conducted by Buchanan
and Klawitter (1992), then calculated the lag phase durations, generation times, and maximum
population densities. When grown in a laboratory medium at 37°C (pH 6.5) containing 5 g/L
NaCl, E. coli O157:H7 exhibited a fastest generation time of 0.3 h, lag phase duration of 1.4 h,
maximum population density of 9.6 log CFU/ml, and exponential growth rate of 1.0 log
CFU/ml/h.
Growth and survival of E. coli O157:H7 in fresh produce, meat products, and juices under
various temperatures have been extensively investigated under laboratory conditions (AbdulRaouf and others 1993a; Abdul-Raouf and others 1993b; Del-Rosario and Beuchat 1995).
Studies monitoring the growth of E. coli O157:H7 in shredded lettuce, sliced cucumber,
cantaloupe, watermelon, and ground beef under various storage temperatures demonstrated no
growth or even slightly reductions in population at 5°C, and rapid growth at 10°C or higher

17

(Abdul-Raouf and others 1993a; Abdul-Raouf and others 1993b; Del-Rosario and Beuchat 1995;
Kauppi 1996). Significant growth of E. coli O157:H7 can occur on lettuce at temperatures of
12oC or above (Koseki and Isobe 2005a; Luo and others 2010; McEvoy and others 2009;
O’Beirne 2007). McKellar and others (2012) and Jones and others (2005) showed slight die-off
of E. coli O157:H7 when incubated for 4 - 12 days at temperatures below 6°C. E. coli cells will
elongate without dividing, with all cells regardless of length losing viability at similar rates
(Visvalingam and others 2012).
The impact of temperature fluctuations on safety of commercially packaged produce
remains poorly understood. Most growth studies have used laboratory rather than commercially
prepared leafy greens, with the levels of background spoilage microorganisms likely differing
from commercially packaged products. Additionally, Luo and others (2009) and Lee and Baek
(2008) demonstrated that commercially packaged baby spinach leaves inoculated with E. coli
O157:H7 stored at 7°C or above would support pathogen growth. However, these baby spinach
leaves were packaged using micro perforated films. Hence, more studies on commercially
bagged salads fresh-cut lettuce under modified atmosphere are needed.
Unlike E. coli O157:H7, L. monocytogenes is psychrotrophic and will grow under
refrigeration conditions. Hoelzer and others (2012) synthesized and compared the available
information regarding the growth of L. monocytogenes on produce with the listeriosis outbreak
data. They found that growth rates and maximum population densities for L. monocytogenes
differed markedly among produce commodities at 10°C. Post-harvest processing also had a
major impact on the growth dynamics for certain types of fresh produce. In general, the
minimum growth temperatures were 1.0°C for L. monocytogenes serovars 1/2a, 1/2b, 1/2c and
1.3°C for serovar 4b (Junttila and others 1988). The maximum growth temperature was 48°C.

18

Populations of L. monocytogenes on lettuce increased after eight days at 5°C (Beuchat and
Brackett 1990). Growth of L. monocytogenes on lettuce at 5°C, 10°C, 15°C, 20°C, and 25°C was
also demonstrated by Koseki and Isobe (2005b), with the fastest growth rate and shortest lag
phase seen at 20°C.
1.4.2 U. S. and international guidelines on temperature control
The U. S. Food and Drug Administration 2009 Food Code requires that ready-to-eat fruits
and vegetables be refrigerated and received at or below 5°C (41°F) to limit growth of organisms
of public health concern (FDA 2009). The Canadian Food Inspection Agency code of practice
for minimally processed ready-to-eat vegetables requires that such products be maintained at 4°C
during transportation and storage (Canadian Food Inspection Agency 2009). The EU food
hygiene legislation (2006) requires foods which are likely to support the growth of pathogenic
micro-organisms to be held at or below 8°C. In Ireland, the core food temperature must be
maintained at 5°C or below during transport (Food Safety Authority of Ireland 2006). The Food
and Agriculture Organization (FAO) recommends optimum refrigeration temperatures of 2 to
5°C for fruits and vegetables (FAO 2003). Nunes and others (2008) summarized the
recommended storage temperatures of 1 to 3°C for various fruits, vegetables, and bagged salads.
While these temperature ranges are considered ideal, temperatures of 5 or 8°C are more realistic
in the distribution chain.
1.4.3 Temperature fluctuations
Storage, processing, and distribution temperatures have a major impact on microbial
quality and shelf-life of fresh-cut produce. The beneficial effects of low temperature storage (1 to
3°C) on product quality and microbial growth during transportation have been well documented
(Jacxsens and others 2002; Koseki and Isobe 2005a; Luo and others 2010). Low temperature is

19

not only able to decrease the respiration rate and enzyme activity of fresh-cut leafy greens, but
also reduces microbial growth and survival. Several studies have demonstrated the survival of E.
coli and L. monocytogenes in salad vegetables at 4 to 5.5º C (Abdul-Raouf and others 1993a;
Kauppi 1996; Koseki and Isobe 2005a,b). This temperature range is assumed to be frequently
encountered during commercial transport of leafy greens in the U. S. Hence, under such ideal
conditions, survival rather than growth of E. coli O157:H7 and limited growth of L.
monocytogenes would be expected during cold chain distribution and retail handling of leafy
greens.
Maintaining a consistent low temperature during transport from the manufacturer to the
retail store is challenging due to both opening of the truck during loading and unloading of
products as well as the extreme outside temperatures. The effect of these temperature
fluctuations on growth and survival of pathogens is not yet fully understood. Thus far, only three
published studies – one each from Canada, Japan and Belgium, have monitored the growth of E.
coli O157:H7 or L. monocytogenes in fresh-cut bagged salad greens during actual fluctuating
temperatures that were recorded during commercial harvest, transport, and retail sale (McKellar
and others 2012; Koseki and Isobe 2005a; Rediers and others 2008).
Koseki and Isobe (2005a) monitored the growth of E. coli O157:H7 and L. monocytogenes
for two time/temperature scenarios that were documented in Japan during commercial harvest,
transport and retail sale of iceberg lettuce. The lettuce was harvested at a temperature of 16 to
17°C and then pre-cooled to below 5°C in the field. The temperature ranged from 4 to 8°C, 2 to
15°C, and 8 to 12°C, during storage, transport and display, respectively (Fig. 1.2). Overall, from
the farm to retail sale, populations of E. coli O157:H7 and L. monocytogenes increased 1.3 and 1
log CFU/g, respectively. E. coli O157:H7 was sensitive to temperature variations on inoculated

20

Iceberg lettuce, with increased growth seen immediately after temperature abuse. Longer periods
of temperature abuse at higher temperatures further increased the growth of E. coli O157:H7. In
contrast, L. monocytogenes was less sensitive to temperature abuse with the growth rate being
more consistent over time.

21

Commercial temperature history in Japan

Figure 1.2 Temperature history for lettuce from the field to retail display in Japan (Koseki and
Isobe 2005a).

22

Working in Canada, McKellar and others (2012) obtained 27 cold chain temperature
profiles for fresh-cut lettuce during storage at the processing facility, commercial transport,
storage at the distribution center, delivery, and retail storage. The distribution times and
temperatures significantly varied for the different stores, routes, and the segments of the cold
chain. In the worst case scenario, the temperature fluctuated from 2 to 4°C, 2 to 6.8°C, 1.6 to
6°C, and 1.6 to 5.4°C, and 3.5 to 6.3°C over 12.5 days during storage at the processor, transport,
storage at the distribution center, delivery, and retail storage, respectively. These temperatures
remained below the minimal growth temperature for E. coli O157:H7. Therefore, the mean
estmated populations of E. coli O157:H7 declined 0.98 log CFU/g overall, indicating slight cell
die-off. However, since this study was only conducted in winter, microbial growth could be
higher during warmer months.
Rediers and others (2009) obtained three temperature profiles for endive throughout the
supply chain from producer, via processor and distributor to a restaurant in Belgium. Similar to
the above study, the time/temperature profiles indicated that the cold chain was generally
properly maintained with only small temperature fluctuations observed. Total coliforms and
Enterobacteriaceae populations were significantly higher in endive samples subjected to
temperature fluctuations in the supply chain, compared to samples stored under optimal
laboratory conditions. However, none of endive samples were inoculated with pathogens.
Nunes and others (2009) evaluated produce temperatures during retail storage, retail
display, and household storage. The results showed that heat-sensitive fruits and vegetables were
transported at higher temperatures than those recommended (Fig. 1.3). Upon arrival at three
retail stores, the bagged salad temperatures were 6.8, 8.1, and 7.7°C, all of which were higher

23

than the recommended 1 to 3°C during cold chain transport. These recorded temperatures were
higher than the above studies where sensors were placed in the refrigerated trucks. However, this
study was conducted under warmer ambient conditions. At retail display, the temperature
profiles for bagged salads varied widely from 1.5 to 16°C, with significant temperature
differences seen in top, middle, and bottom layers of the display cases (Fig. 1.3).
Time/temperature profiles for chilled ready-to eat foods including salads from preparation
and packaging in a centralized kitchen to refrigerated storage in a school canteen showed ample
opportunity for growth of L. monocytogenes, especially during extended storage over the
weekend (Rosset and others 2004). Time/temperature profiles for ready-to-eat foods stored in U.
S. household refrigerators also indicate many opportunities for temperature abuse (Pouillot and
others 2010). However, limited support was found for a correlation between home storage time
and temperature.
Other studies monitoring temperatures during transport and retail display highlight the need
to adopt temperature-monitoring technologies for perishable foods during transport. When
Whyte and others (2006) evaluated meat temperatures during transport and retail display in New
Zealand, more than 8% of the samples exceeded the 7ºC industry standard during summer. The
retail storage temperature averaged 2ºC higher than the transport temperature, leading to
potential quality and safety concerns. Another study monitoring transport of bulk beef showed
broad fluctuations in product temperature during summer storage and transport (Gill and others
1997), again suggesting ample opportunity for growth of both pathogenic and spoilage
organisms. A similar study in Australia monitoring chilled foods for temperature abuse during
transport found that 60% of the products were above the maximum acceptable storage
temperature of 5ºC (Estrada-Flores and others 2006), which resulted in abundant growth of

24

Pseudomonas. These findings reinforce the importance of strict temperature control during
transport and retail display.

25

Retail display temperature in the US

Figure 1.3 Temperature profiles for different locations (top, middle and bottom) of salad bags
during retail display. Dashed horizontal lines correspond to the recommended temperature range
for fresh-cut vegetable salads (Nunes and others 2009).

26

The effect of sudden temperature change on the intermediate microbial lag phase has been
studied (Swinnen and others 2005). The intermediate lag phase is the phenomenon that a
microbial population needs some time to adapt to changes in the environment when subjected to
dynamic conditions, such as temperature fluctuations in the cold chain. Temperature fluctuations
within the temperature range from 20 to 37°C do not result in a lag phase for E. coli, while
temperature changes below this range lead to a period of adaptation. Therefore, at low
temperatures during transport and retail sale, rapid and slight temperature changes caused by
normal refrigeration cycles may have little effect on the growth of E. coli O157:H7.
Similarly, listeriosis outbreaks traced to fresh produce have been most often associated
with fruits or vegetables that have been contaminated during processing and/or subjected to
temperature abuse during storage (Hoelzer and others 2012). However, data are currently
insufficient to reliably estimate growth behavior for many commodities.
In summary, the temperatures during large-scale commercial cold chain transport are
generally properly maintained with only small temperature fluctuations. Temperature abuse is
most likely to occur at retail storage and display. However, few temperature profiles with relative
low temperatures are presently available, which may be not representative. Consequently, the
impact of temperature fluctuations on both the microbial safety and quality of fresh produce in
the U. S. is not yet fully understood.
1.4.4 Temperature monitoring in industry
Application of accurate temperature monitoring systems during perishable food transport
and distribution can help maintain food quality and improve food safety, particularly during
warm months. This information is also needed for HACCP validation and supplier compliance

27

records. One study used a Safety Monitoring and Assurance System to track temperatures during
shipment of hams, and these time-temperature histories then were used to simulate and identify
critical control points in the supply chain (Koutsoumanis and others 2005). The results showed
that the percentage of unacceptable hams decreased from 16 to 4%. Surprisingly, over 12.5% of
ham samples were beyond acceptable quality at the time of consumption, with the potential
growth of L. monocytogenes and Lactobacillus sakei up to 107 CFU/g using growth models.
These results reaffirm the need for monitoring transport temperatures to accurately assess current
transport conditions.
Currently, various sensors with cold chain logistics are available for temperature
®

monitoring. One of the latest innovations in temperature monitoring, TimestripPlus , is a new
temperature indicator used for temperature sensitive goods (Hafen 2012). The major advantage
of this indicator is that it can be modified for each application to reflect different temperature
requirements (freeze, refrigerated, and control room temperature) or different lengths of abuse
time. The alert can also be color-scaled for different levels (high/moderate/low) of temperature
abuse. This user-friendly design has the potential to significantly help truck drivers, retail
personnel and FDA inspectors.
Implementation of Active Radio Frequency Identification (RFID) wireless sensors for
temperature monitoring of perishable foods from harvest to retail sale has allowed both food
processors and retailers to improve backroom operations for quality control (Songini 2006).
Some large-scale retailers have even mandated the use of RFID sensors by their more than 1000
suppliers (Songini 2006). Using these RFID sensors resulted in a significant improvement in
replenishment rate and stock-out reduction, and an estimated cost savings of over US $1.7 billion
(Hardgrave and others 2005). Some grocery chains have also integrated RFID tracking for cases

28

and pallets of produce throughout its supply chain, improving customer demand chain
management (Gonsalves 2004). Recent use of advanced sensor-based RFID systems has helped
to minimize the growth of E. coli in meat (Gessner and others 2007). These sensor-based
technologies are able to notify retailers of potential temperature abuse that could lead to
unacceptable microbial growth at the unit level (Wessel 2006).
1.5 Modeling bacterial growth
Since the 1980s, various models have been developed to predict the growth of foodborne
pathogens in foods (Zwietering and others 1990). When using isothermal growth models, curves
are fitted to the experimental growth data using one of the sigmoid functions. Several sigmoid
functions have been developed with part of the resulting curve defined as the lag phase (Gill
1984; Fu and others 1991). The lag phase is then modeled, together with the maximum specific
growth rate, as a function of environmental factors such as temperature, pH, and water activity.
In this approach, lag phase was defined empirically without mechanistic principles. Other models
including logistic, vitalistic, Gompertz, Richards, Schnute, and Stannard can also be used to
predict microbial growth (Cole and others 1990; Membre and others 1997; Sorrells and others
1989; Zwietering and others 1990).
Additional models have been developed to predict bacterial growth during non-isothermal
temperature abuse. The U. S. Department of Agriculture – Agricultural Research Service (USDA
– ARS) Pathogen Modeling Program 7.0 (PMP) has been widely applied in the food industry to
estimate the growth, survival, and inactivation of Listeria and other pathogens based on pH,
storage temperature, and salt concentration (Buchanan and others 1990; Houstsma and others
1996; Le Marc and others 2002). Under non-isothermal temperature conditions, pathogen growth
can be predicted by dividing the time-temperature history into multiple small time/temperature

29

intervals and then applying the USDA PMP to predict the pathogen growth in each interval. The
overall estimated pathogen growth is the sum of the predicted growth for each interval. However,
this model was developed from pure-culture and broth-based studies. Because pure-culture
systems contain high levels of nutrients without competitive microflora, the PMP model is
generally assumed to provide conservative estimates of pathogen growth. Koseki and Isobe
(2005a) also indicated that the PMP model tended to overestimate the pathogen growth.
Therefore, USDA regulators have not generally accepted this approach as the sole means of
scientific validation (USDA FSIS 2005).
Baranyi and Roberts (1994; 1995) developed a state-of-the-art approach to predict
microbial growth based on the physiological state of the cell. This dynamic model has been
successfully implemented for continuously changing growth conditions such as temperature, pH
and water activity (Baranyi and others 1994; 1995). Combined with Baranyi’s primary model
and Ratkowsky’s square-root model, the growth of Brochothrix thermosphacta was accurately
predicted under gradually changing temperatures and sudden changes in temperature between 5
and 25°C. Accuracy of the model prediction depends on optimizing an appropriate value for
model parameter α0, the actual specific growth rate to the potential specific growth rate at the
time of inoculation. However, the extent of growth was overestimated when the temperature
profile contained sudden changes between 2.8 and 25°C (Baranyi and others 1995). This might
be due to changes in the physiological state of the microorganism as a result of sudden cold
shock.
Koseki and Isobe (2005a) applied the Baranyi model to predict the growth of E. coli
O157:H7, Salmonella, and L. monocytogenes on lettuce at fluctuating temperatures encountered
from farm to retail store. The growth kinetic parameters (lag time, maximum growth rate, and

30

maximum population density) were obtained from pathogen growth studies on lettuce at
isothermal temperatures (5 to 25°C). These model parameters were then used to predict pathogen
growth under fluctuating temperatures in the Baranyi primary microbial growth model along
with the Ratkowsky secondary model. Their results demonstrated that the Baranyi growth model
was able to accurately predict pathogen growth under fluctuating temperatures in most cases
when fresh produce was properly cooled after harvest. Most predictions agreed with the observed
viable counts based on the small root mean square error (RMSEs) and biases. In comparison, the
PMP model, the Gamma concept, and the model of Farber and others (1996) overestimated the
growth rate with larger RMSEs (Sant’Ana and others 2012; Carrasco and others 2008; Farber
and others 1996; Koseki and Isobe 2005a).
However, Baranyi’s model has limitations. Overestimation of E. coli O157:H7 and
Salmonella growth using Baranyi model occurred when the temperature history started high
(25°C) for 5 hours and decreased later (Koseki and Isobe 2005a), while the prediction of L.
monocytogenes under the same condition fitted the observed growth data. Danyluk and Schaffner
(2011), McKellar and Delaquis (2012) also showed that Baranyi’s model overestimated the
growth rate of E. coli O157:H7. However, most data are from constant temperature studies
(Chang and Fang 2007; Delaquis and others 2002; Lee and Baek 2008; McEvoy and Luo 2009;
Theofel and Harris 2009). One explanation was that the estimated growth kinetics calculated
from the isothermal environments were inadequately used for determining coefficients of
differential equations for predicting growth in dynamic environments. In addition, the growth
kinetic parameters used by Koseki and Isobe (2005a) were obtained under ambient conditions.
The optimizing kinetic parameters for predicting microbial growth in commercial fresh-cut
bagged salad greens under Modified Atmosphere Packaging (MAP) remain unknown which also

31

increases uncertainty. When Carrasco and others (2007) used the Baranyi model to predict the
growth of L. monocytogenes on ready-to-eat MAP-packaged iceberg lettuce, Listeria growth was
significantly inhibited.
Juneja and others (2008) recently modified the Baranyi model by adding a memory effect
so that the kinetic parameter predictions depended on prior cell history. This modification
allowed for a less restrictive assumption and might lead to models that provide more precise
estimates of growth. Although limited to lag and exponential growth phase, this model can
characterize the growth kinetic parameters relative to prior history in dynamic environments.
Future studies are needed to validate this modified Baranyi model for predicting the growth of E.
coli O157:H7, L. monocytogenes, and Salmonella in bagged fresh-cut salads.
To improve the accuracy of model prediction, McKellar and others (2009) developed a
dynamic growth-death model for E. coli O157:H7 in minimally processed leafy greens.
McKellar’s model contains a primary three-phase linear model for growth, a secondary square
root model for growth, and a tertiary log-linear model for death. Therefore, McKellar’s model
has two advantages. Firstly, bacterial cell death and survival over time have been considered.
Secondly, the model can accurately predict microbial growth at low temperatures. McKellar and
others (2009) successfully used this model to predict pathogen behavior under both isothermal
and non-isothermal conditions when fitted to the data from 13 published studies, with better
predictions compared to Baranyi’s model.
McKellar and others (2012) applied this dynamic growth-death model to simulate the
behavior of E. coli O157:H7 in fresh-cut lettuce under the dynamic temperatures encountered
during distribution from processing to retail. Since this study was conducted in Canada during
winter, the 27 temperature profiles reported in this study showed only slight temperature abuse.

32

At these low temperatures, the model predicted an overall mean decline in cell numbers and
slight growth for a few cases when the temperature rose above 5°C. However, no experimental
growth data were available to validate the model predictions, making accuracy of the model
prediction difficult to determine. In addition, since this model includes a linear death model, it
has the potential to underestimate microbial growth if the parameters are not optimized.
Therefore, model validation during periods of more severe temperature such as during summer is
necessary since underestimation of microbial growth is usually not acceptable for model
predictions.
1.6 Risk assessment
Quantitative microbial risk assessment (QMRA) has been an increasingly used approach to
integrate and evaluate information from diverse sources concerning the origin and fate of
pathogens in the food chain and to determine the magnitude of public health risks. The Sanitary
and Phytosanitary Agreement (SPS Agreement) of the World Trade Organization (WTO)
recognized the necessity of scientific verification and validation when determining food safety
risks. The principles and guidelines for food safety risk analysis were defined by the Codex
Committee on Food Hygiene (FDAP/WHO 1995). Risk is defined as the probability and the
consequence of a hazard to occur. A traditional QMRA consists of four components: hazard
identification, dose-response assessment, exposure assessment, and risk characterization, which
when considered together provide an expression of public health risk.
Risk assessment has its roots in the concerns for toxic chemicals in food. While these
assessments are mostly based on toxicology and carcinogenity studies, their application to
microbial foodborne pathogens poses some significant challenges. One difficulty relates to the
fact that unlike chemical, toxicological, or environmental contaminants, populations of bacterial

33

pathogens can change dynamically since conditions vary widely from farm to consumption.
Fortunately, researchers are making progress in developing the predictive models, simulations,
and other tools that will meet the technical requirements for quantifying estimates of risk. In
addition to this technical challenge, there are many data gaps that limit the precision necessary
for quantitative risk assessments. For instance, little information is available to precisely estimate
the relationship between the quantity of a biological agent and the frequency and magnitude of
adverse human health effects, particularly as this might relate to susceptible sub-populations.
There is also limited information on exposure assessment - the amounts of foods consumed by
populations and their probable contamination.
QMRA usually involves various mathematical models, probability distributions, and Monte
Carlo simulation. Mathematical models are used to describe the introduction of pathogens into
food, multiplication of microbes in food over time, microbial survival following various
treatments, consumption of microbes in food, and subsequent illness. The inherent variability
and uncertainty in the estimation of microbial risk can be described by probability distributions.
A probability distribution is a mathematical representation of the relative likelihood of a random
variable taking on a specific value. Monte Carlo simulation can be used to estimate the level of
human illness and uncertainty associated with the model. Control strategies can be calculated in
the same way and a cost/benefit can be conducted for prevention of illness.
The preliminary QMRA framework for the risks associated with leafy greens from farm to
table began a few years ago. Up to now, three research groups have assessed the risk of E. coli
O157:H7, Salmonella, and L. monocytogenes in leafy green vegetables from farm to
consumption in the U. S., Spain, and the Netherlands (Carrasco and others 2010; Franz and
others 2010; Danyluk and Schaffner 2011). Danyluk and Schaffner (2011) presented a

34

preliminary framework that identified currently available data and provided initial risk estimates
for E. coli O157:H7 in leafy greens in the U. S. The risk model predicted a starting level of -1
log CFU/g and a prevalence of 0.1% which could have resulted in an outbreak approximately the
size of the 2006 E. coli O157:H7 spinach outbreak. Temperature abuse will support growth of E.
coli O157:H7, with populations increasing as much as 1 log CFU/day. Carrasco and others
(2010) assessed the risk of L. monocytogenes in ready-to-eat lettuce salads in Spain. The
estimated number of listeriosis cases was 102 and 105 for low- and high-risk subpopulations, with
similar numbers of listeriosis cases reported in Spain. They also found that MAP was the most
effective means to reduce the number of cases. Franz and others (2009) estimated the average
number of cases per year associated with the consumption of leafy greens at salad bars in the
Netherlands at 166, 187, and 0.3 for E. coli O157:H7, Salmonella, and L. monocytogenes,
respectively. The temperatures in that study were maintained at < 5°C, with considerably greater
growth of L. monocytogenes (194%) and minimal growth of E. coli O157:H7 (19%). The
mathematical models and data used in these studies provide a preliminary database for future
leafy green risk assessments. However, many data gaps still remain to be filled, including
transport/distribution/retail sale/home storage temperature fluctuations, the correlation between
storage time and temperature, the importance of lag time in microbial growth models, and
validation of the importance of cross-contamination in wash water during processing.
Quantitative microbial risk assessment has become the framework for the scientificallybased decisions used in formulating sound risk management strategies. The work presented in
this thesis on temperature abuse of leafy greens in the cold chain will help fill one of the
important data gaps in the current risk assessment.

35

CHAPTER 2
MODELING THE GROWTH OF ESCHERICHIA COLI O157:H7 AND LISTERIA
MONOCYTOGENES IN PACKAGED FRESH-CUT ROMAINE MIX AT FLUCTUATING
TEMPERATURES DURING SIMULATED TRANSPORT, RETAIL STORAGE AND
DISPLAY

2.1 Abstract
Temperature abuse during commercial transport and retail sale of leafy greens negatively
impacts both microbial safety and product quality. Consequently, the effect of fluctuating
temperatures on Escherichia coli O157:H7 and Listeria monocytogenes growth in commercially
bagged salad greens was assessed during transport, retail storage, and display. Over a 16-month
period, a series of time/temperature profiles for bagged salad greens were obtained from 16
transportation routes covering four geographic regions (432 profiles), as well as during retail
storage (4867 profiles) and display (3799 profiles) at nine supermarkets. Five different
time/temperature profiles collected during 2 to 3 days of transport, 1 and 3 days of cold room
storage and 3 days of retail display (including the best and worst scenarios) were then duplicated
in a programmable incubator to assess E. coli O157:H7 and L. monocytogenes growth in
commercial bags of romaine lettuce mix. Microbial growth predictions using the
Baranyi/Ratkowsky and McKellar models were validated by comparing the root mean square
error (RMSE) and bias between the laboratory growth data and model predictions. Monte Carlo
simulations (~100,000 iterations) were then performed to calculate the probability distribution of
microbial growth from a total of 8,122,127,472 scenarios during transport, cold room storage,
and retail display. Using inoculated bags of retail salad, E. coli O157:H7 and L. monocytogenes
populations increased a maximum of 3.1 and 3.0 log CFU/g at retail storage. Both models

36

yielded < 1 log CFU/g RMSEs and biases with the Baranyi model fitting slightly better to the
laboratory data. Based on the simulation results, both pathogens generally increased < 2 log
CFU/g. However, retail storage duration can significantly impact the extent of E. coli O157:H7
and L. monocytogenes growth. Greatest growth of E. coli O157:H7 would be expected in product
located at the bottom of the pallet during summer transport. This large-scale U. S. study – the
first using commercial time/temperature profiles to assess the microbial risk of leafy greens
during transport, retail storage, and display, should be useful in filling some of the data gaps in
current risk assessments for leafy greens.
2.2 Introduction
Consumption of fresh and fresh-cut leafy greens in the United States has increased over the
past few decades due to the abundant health benefits and year-round availability (Calvin and
others 2003; Meng and Doyle, 2002; Nicola and others, 2006).

Leafy green vegetables

consumed raw are highly susceptible to microbial contamination. Since 1993, numerous
foodborne outbreaks involving Escherichia coli O157:H7 have shaken consumer confidence
with at least 800 cases of illness and 8 deaths having been traced to leafy greens (FDA 2005 a),
particularly California-grown lettuce (Iceberg, Romaine, red leaf, and mesclun) and spinach. In
late summer of 2006, consumption of contaminated California-grown baby flat-leaf spinach was
responsible for 205 cases of E. coli O157:H7 infection in 28 states, including 31 cases of
hemolytic uremic syndrome and 3 fatalities (FDA 2006). In December 2006, commerciallyshredded lettuce used in tacos that sickened over 150 people was again traced to E. coli O157:H7
(CDC 2006a).
In addition to E. coli O157:H7, Listeria monocytogenes is also of particular concern
because of its wide distribution in the environment and its ability to grow during refrigeration. In

37

2011, 139 cases of listeriosis, including 29 deaths and 1 miscarriage, were traced to consumption
of cantaloupe (CDC 2011b) with a smaller outbreak linked to celery one year earlier (FDA
2010a). While cases of listeriosis involving lettuce are few (FDA 2012), 8 recalls have been
issued since 2010 for L. monocytogenes-contaminated leafy greens, thus legitimizing concern for
the pathogen in lettuce.
Produce can become contaminated at any point in the farm-to-fork continuum. The high
moisture and nutrient content, together with the large surface area of fresh-cut produce, make
these highly susceptible microbial contamination and subsequent growth. Although washing and
sanitizing can decrease microbial numbers, bacterial cells internalized in stomata or cut surfaces
and those found in hydrophobic regions, cavities, and rough surfaces of leafy greens are likely to
survive sanitizer exposure (Takeuchi and Frank 2000; Zhou and others 2009).
Temperature monitoring during distribution provides a major advantage in the ability to
manage temperature sensitive products in the cold chain with most food and pharmaceutical
suppliers of perishable goods already tracking time/temperature histories. Many large-scale
retailers have now mandated the use temperature sensors and data loggers for their suppliers with
others using various temperature monitoring systems to track specific cases and pallets of
product in the supply chain. In this study, temperature sensors were placed in trucks during
transport, at retail storage room, and retail display cases in order to monitor temperature
fluctuations and collect time/temperature profiles.
The many benefits of low temperature storage on product quality and microbial growth
during transport and distribution have been well documented (Jacxsens and others 2002; Koseki
and Isobe 2005a; Luo and others 2010). The U. S. Food and Drug Administration Food Code
requires that ready-to-eat fruits and vegetables be refrigerated at < 5°C (41°F) to minimize

38

growth of foodborne pathogens (FDA 2009). However, maintaining a consistently low
temperature throughout the distribution chain is challenging. Opening of the truck doors during
loading and unloading, outside temperature extremes and retail storage/display conditions can all
lead to temperature fluctuations. Thus far, only studies from Canada, Japan and Belgium have
assessed the growth of E. coli O157:H7 and L. monocytogenes at real-time temperatures
collected during pre-harvest, transportation and retail sale of fresh-cut bagged salad greens
(McKellar and others 2012; Koseki and Isobe 2005a; Rediers and others 2008). However,
limited time/temperature profiles in the above studies may not represent the U. S. cold chain
transport and distribution.
Several mathematical models have been developed over the last 25 years to predict the
likelihood for growth of E. coli O157:H7, L. monocytogenes and other pathogens in foods,
primarily under constant temperature conditions (Buchanan and others 1990; Cole and others
1990; Membre and others 1997; Sorrells and others 1989; Zwietering and others 1990). The
Baranyi/Ratlowsky model has been successfully used to predict microbial growth under
continuously changing conditions (Baranyi and others 1994; 1995). Koseki and Isobe (2005a)
applied this same model to predict the growth of E. coli O157:H7, Salmonella, and L.
monocytogenes on lettuce at fluctuating temperatures. While some studies showed a small root
mean square error between observed bacterial growth and the model prediction (Koseki and
Isobe 2005a; Carrasco and others, 2008), others found that the Baranyi/Ratlowsky model tended
to overestimate E. coli O157:H7 growth (Danyluk and Schaffner 2011; McKellar and Delaquis
2011). A dynamic growth – death model was recently developed by McKellar and Delaquis
(2011) to quantitatively predict the growth/survival of E. coli O157:H7 at variable temperatures
during transport and retail sales.

39

A recent quantitative microbial risk assessment of leafy greens from farm to fork by
Danyluk and Schaffner (2011) supported the growth of E. coli O157:H7 during periods of
temperature abuse. However, the retail temperature profiles used were obtained from meat,
cheese and yogurt rather than leafy greens, with temperature abuse during commercial transport
from the manufacturer to retail stores also lacking.
Consequently, the objectives of this study were to: 1) assess the impact of various real-time
time/temperature histories on the growth of E. coli O157:H7 and L. monocytogenes as well as
psychrotrophic and mesophilic bacteria in commercial romaine mix during transport, retail cold
storage and display, 2) validate the Baranyi/Ratkowsky and McKellar models using laboratory
data, and 3) determine the probability distributions for E. coli O157:H7 and L. monocytogenes
growth during transport, retail storage and display. To our knowledge, this is the first large-scale
study to report the impact of fluctuating temperatures on microbial growth for fresh-cut leafy
greens distributed in the United States.
2.3 Materials and methods
2.3.1 Temperature monitoring during commercial transport and retail
Temperature fluctuations were continuously recorded in typical refrigerated truck trailers
@

using TempTale 4 sensors (Sensitec Inc., Beverly, MA). During cross-country transport of
bagged salads, five different commercial routes were assessed during four seasons – California to
North Carolina (5 trials), Arizona to North Carolina (1 trial), California to Texas (3 trials),
California to Ohio (5 trials) and California to Arizona (2 trials). Within each refrigerated cargo
trailer, 20-30 sensors were placed in three key zones - front, midway and rear, midway up the
sidewall of the truck and at three pallet levels - top, middle, and bottom (unpublished data).

40

Time/temperature readings were recorded at 5-minute intervals during 2 to 3 days of transport
until delivery with a total of 432 time/temperature profiles generated.
Temperature fluctuations during retail storage and display were monitored at 9
supermarkets. Depending on the size of the store, 2 to 6 and 6 to 24 sensors were respectively
placed in the storage room and display case for 1 to 2 months with two replications. The bagged
salads were typically stored for 1 to 3 days and displayed for a maximum of 3 days. Therefore,
each original temperature profile during storage and was generated at 1- and 3-day intervals. In
total, 4867 retail storage and 3863 retail display temperature profiles were obtained.
Five temperature profiles, including the best and worst case scenarios as well as three inbetween temperatures, were selected from those collected during transport, retail storage, and
retail display for the microbial growth studies. Criteria for selection were based on the average,
maximum and minimum temperatures, standard deviations, and model predictions (see model
validation).
2.3.2 Culture preparation
The four avirulent, GFP-labeled strains of Escherichia coli O157:H7 chosen for study were
ATCC 43888 (pGFPuv, stx1-, stx2-, ampicillin resistant), CV 2b7 (pGFPuv, stx1-, stx2-,
ampicillin resistant), 6980-2 (pGFP, stx1-, stx2-, ampicillin resistant), and 6982-2 (pGFPuv, stx1-,
stx2-, ampicillin resistant), all of which were obtained from Dr. Michael Doyle (University of
Georgia, Griffin, GA). The four unlabeled strains of Listeria monocytogenes were N1-227
(serotype 4b, food outbreak), C1-056 (serotype 1/2a, human sporadic case), J1-108 (serotype 4b,
human outbreak), and N3-031 (serotype 1/2a, hot dog sporadic case). All stock cultures were
maintained at –80°C in Trypticase Soy Broth containing 0.6% (w/v) Yeast Extract (TSB-YE)

41

(Becton, Dickinson & Co., Franklin Lakes, NJ)] and 10% glycerol (Mallinckrodt Baker, Inc.,
Phillipsburg, NJ).
Prior to inoculation, the four strains of E. coli O157:H7 and L. monocytogenes were
respectively streaked to plates of Trypticase Soy Agar containing 0.6% (w/v) Yeast Extract
(TSA-YE) and 100 ppm ampicillin or TSA-YE which were incubated overnight at 37°C. Single
colonies of each E. coli O157:H7 and L. monocytogenes strain were then subjected to two
consecutive 24 h/37°C transfers in TSB-YE with or without 100 ppm ampicillin, respectively.
The cultures were then combined in equal volumes to obtain two separate 4-strain cocktails
containing ~108 CFU/ml as determined by optical absorbance and then diluted in sterile
phosphate buffer solution (PBS) to a level of 2.6×106 CFU/ml.
2.3.3 Inoculation
Retail bags of pre-washed classic romaine mix (255 g/bag) were obtained from a local
supermarket (Meijer, Okemos, MI) at least 15 days before the “best sell by” date and
immediately placed in a walk-in cold room at 4°C with a maximum 2-day storage time before
use. Each bag was surface-sterilized using 70% ethanol prior to inoculation. Working in the coldroom, each bag of romaine mix was inoculated with 2 ml of the E. coli O157:H7 or L.
monocytogenes cocktail using a BD PrecisionGlide™ general use syringe (Becton, Dickinson &
3

Co.) so as to contain ~10 CFU/g, immediately sealed with tape to maintain the same package
atmosphere and then vigorously shaken for 1 min to evenly distribute the inoculum. The package
atmosphere was measured using Gas Chromatography at the Department of Packaging, Michigan
State University. The average oxygen and carbon dioxide concentration based on three bags of
salads were 4.3% and 3.2%, respectively. Three bagged salads were similarly syringe-inoculated
with Glow Germ (0.5 g in 2 ml of distilled water) and placed under UV light. Glow Germ
42

fluorescence under UV light indicated that the inoculum was evenly distributed inside the bags
after the 1-min shake. In addition, three 50 g sub-samples from three bags of salads inoculated
with E. coli O157:H7 were separatedly examined for E. coli O157:H7. Statistical results also
showed no significant difference for E. coli O157:H7 populations in each sub-samples of salad
bags to support the conclusion that the inoculum was evenly distribution in salad bags.
2.3.4 Incubation
The inoculated bags were immediately placed in a Thermo Forma Environmental Chamber
(Model 3851, Thermo Fischer Scientific, Inc., Waltham, MA) that was programmed to duplicate
®

the same time/temperature profiles recorded by the TempTale 4 sensors during commercial
transport and retail storage/display. All of the programmed time/temperature profiles were
®

monitored and recorded at 5-min intervals using a HOBO data logger (Onset , Bourne, MA).
These profiles were then validated based on the root mean squared errors (RMSEs) and biases
between the transport/storage/display and environmental chamber time/temperature data. For
each individual temperature selected:

and

Each experiment was repeated three times with two replicates per trial.
2.3.5 Microbial analyses
Bags of romaine mix were collected at pre-determined times (average of three to four times
per day for transport and retail storage, and once per day for retail display) and quantitatively

43

analyzed for E. coli O157:H7 and L. monocytogenes. The salad bags were weighed and
aseptically opened after surface decontamination with 70% ethanol. Duplicate 50-g lettuce
samples from each bag were transferred to sterile Whirl-Pak

®

filter bags (1.7 L, Nasco, Fort
®

Atkinson, WI) and macerated in 200 ml of sterile PBS using a Stomacher

400 Circulator

(Seward, London, U.K.) at 260 rpm for 1 min. After appropriate serial dilution in PBS, 100 µl
aliquots were plated on TSA-YE containing 100 ppm ampicillin or Modified Oxford Agar
(Neogen, Lansing, MI) to respectively quantify E. coli O157:H7 and L. monocytogenes after 48 h
of incubation at 37°C and 35°C. Similarly diluted samples were surface-plated in duplicate on
TSA-YE with these plates then incubated at 37°C for 48 h or 4°C for 10 days to quantify
mesophilic and psychrotrophic bacteria, respectively.
2.3.6 Model validation
The Baranyi primary model, a square-root Ratlowsky secondary model for maximum
growth rate, and a linear model for maximum population density (MPD) were applied to the
time/temperature profiles, with the dynamic solution based on the following coupled differential
equations (Baranyi et al., 1995):

dq
= μ max q
dt

and

x
q
dx
)x
(1 = μ max
x max
1+ q
dt

where x is the natural log of the bacterial concentration, q is the physiological state of the
population, μmax is the maximum specific growth rate, and xmax is MPD. The initial conditions for
the above equations are fixed as q(0) = q0 and x(0) = x0. The secondary models are:

µ max = b(T - Tmin )
44

where the model parameters used were those initially reported by Koseki and Isobe (2005a) for
growth of E. coli O157:H7 and L. monocytogenes on iceberg lettuce (Table 2.1).
®

MATLAB (R2010b, MathWorks, Natick, MA) codes were developed for the Baranyi/
Ratlowsky model so that microbial growth could be predicted by inputting the time/temperature
profiles. Initial predictions were used to select the time/temperature profiles used in the
previously described laboratory experiments.

45

Table 2.1 Baranyi and Ratlowsky model parameters for growth of E. coli O157:H7 and L.
monocytogenes on Iceberg lettuce (Koseki and Isobe 2005a).

Pathogen
E. coli O157:H7
L. monocytogenes

α0
0.056
0.072

b
0.033
0.016

T min (°C)
4.54
-4.26

46

A dynamic growth-death model developed by McKellar and Delaquis (2011) was also
applied to simulate the growth of E. coli O157:H7:

;
;
;
;

;
;
;

The validity of the above two growth models was quantified based on the root mean
squared error (RMSEs) and biases between the predicted and observed experimental growth
data. RMSEs and biases were evaluated as follows for each individual time/temperature profile
selected:

and

47

The model with the smaller RMSE and bias was then used to predict the increase in growth
rate for E. coli O157:H7 and L. monocytogenes at every temperature profile collected during
transport, retail storage and display.
2.3.7 Monte Carlo simulation
Monte Carlo simulations (~100,000 iterations) were performed in MATLAB to calculate
the probability distribution of microbial growth for 432 (during transport) × 4867 (24 h at retail
storage) × 3863 (72 h at retail display) in a total of 8,122,127,472 time/temperature scenarios.
Seasonal effect during transport, location of the bag (top, middle, bottom layer) within the
pallets, retail storage time (24 h vs 72 h), and location in the retail display (front vs back of the
display case) were examined to further simulate the probability distributions for microbial
growth.
2.3.8 Statistical analyses
Data were log transformed to satisfy the assumption of homogeneity of variance. The
standard errors for bacterial populations over time were calculated by ANOVA using the PROC
MIXED procedure in the SAS package (v.9.2, SAS Institute, Cary, NC). Fisher’s Least
Significant Difference was used to compare the difference between the initial and maximum
populations of mesophiles and psychrotrophs (P < 0.05).
2.4 Results
2.4.1 Temperature selection for lab validation
Over a 16-month period, a series of time/temperature profiles were obtained for bagged
salad greens from 16 transport routes covering four geographic regions (432 profiles), as well as
during retail storage (4867 profiles) and display (3863 profiles) at nine supermarkets. Overall,

48

the average temperature profiles ranged from -0.3 to 7.7°C, 0.6 to 15.4°C, and -1.1 to 9.7°C
during transport (Fig. 2.1A), retail storage (Fig. 2.1B), and retail display (Fig. 2.1C),

49

Figure 2.1 Average temperatures for all profiles collected during transport, retail storage, and
display with the 5 selected temperatures indicated (A transport, B retail storage, C retail display)
and the time/temperature profiles for each of the temperatures selected (D transport, E retail
storage, F retail display).

50

Figure 2.1 (cont’d)

51

Figure 2.1 (cont’d)

52

respectively. Five transport (Fig. 2.1D), retail storage (Fig. 2.1E) and retail display (Fig. 2.1F)
time/temperature profiles having minimum and maximum temperature peaks ranging from 2.1 to
9.7°C, 1.8 to 18.2°C, and 1.0 to 14.1°C, respectively, were selected based on average, minimum
and maximam temperatures, and preliminary model predictions for the microbial studies.
Root mean square errors (RMSEs) and biases between sensor and laboratory-based
temperatures duplicated in the programmable incubator were calculated for the five transport,
storage and display time/temperature profiles (Table 2.2). All biases were < 1°C and all RMSEs
except those for C1 and C2 were < 1°C.
2.4.2 Microbial growth
In bags of inoculated romaine mix, E. coli O157:H7 populations remained at ~ 3 log CFU/g
during 48 h to 52 h of incubation at simulated transport temperatures (Fig. 2.2 A1 – A5).
Simulated retail storage led to E. coli O157:H7 population increases of 0.1 to 3.1 log CFU/g
(Fig. 2.2 B1 – B5) with no significant growth seen during 72 h at retail display (Fig. 2.2 C1 –
C5). L. monocytogenes populations increased < 0.6 log CFU/g at simulated transport
temperatures (Fig. 2.2 A1 – A5). Greater growth of Listeria was seen during 72 h at retail storage
and display temperatures with populations increasing up to 3.0 (Fig 2.2 B1 – B5) and 1.1 log
CFU/g (Fig 2.2 C1 – C5), respectively. Using the same time/temperature profiles as above,
mesophilic and psychrotrophic populations increased significantly (P < 0.05) in all samples
except A3 (Table 2.3). Similar to E. coli O157:H7 and L. monocytogenes, mesophiles and
psychrotrophs exhibited greater growth – up to 3.1 and 3.3 log

53

Figure 2.2 Growth of E. coli O157:H7 and L. monocytogenes under selected temperatures during
transport (A1 – A5), retail storage (B1 - B5) and retail display (C1 – C5).

54

Figure 2.2 (cont’d)

55

Figure 2.2 (cont’d)

56

Figure 2.2 (cont’d)

57

Figure 2.2 (cont’d)

58

Figure 2.2 (cont’d)

59

Figure 2.2 (cont’d)

60

Figure 2.2 (cont’d)

61

Table 2.2 Root Mean Square Errors and biases between selected sensor and duplicated lab
temperatures during transport, retail storage and display.
Transport RMSE
(°C)
A1
0.57
A2
0.88
A3
0.88
A4
0.47
A5
0.14

Bias
(°C)
0.30
0.57
0.57
0.21
-0.02

Retail
Storage
B1
B2
B3
B4
B5

RMSE
(°C)
0.88
0.62
0.93
0.89
0.69

62

Bias
(°C)
0.11
-0.06
0.45
0.69
-0.19

Retail
Display
C1
C2
C3
C4
C5

RMSE
(°C)
1.32
2.11
0.81
0.78
1.00

Bias
(°C)
0.71
0.98
0.21
0.28
-0.08

Table 2.3 Growth of mesophiles and psychrotrophs in bagged romaine mix during transport,
retail storage, and display

Transport
A1
A2
A3
A4
A5
Retail Storage
B1
B2
B3
B4
B5
Retail Display
C1
C2
C3
C4
C5

Mesophiles
Initial (log
Final (log
CFU/g)
CFU/g)

Psychrotrophs
Initial (log
Final (log
CFU/g)
CFU/g)
a

5.19
6.28
5.86
5.06
6.61

6.62
6.83
5.91
6.29
6.82

*
*

5.50
4.79
4.52
5.10
5.78
5.99
4.96
6.03
6.20
5.26

*
*

5.79
6.61
6.10
5.54
6.90

7.06
7.54
6.71
6.70
7.06

*
*
*
*
*

7.36
7.87
6.30
8.02
6.86

*
*
*
*
*

5.48
6.08
4.83
5.35
5.89

8.37
7.78
6.69
8.65
7.21

*
*
*
*
*

6.16
7.31
6.73
7.12
6.65

*
*
*
*
*

5.94
5.13
6.35
6.01
6.06

7.14
7.72
7.39
7.45
7.19

*
*
*
*
*

a

*-Significant growth during incubation (P < 0.05).

63

CFU/g, respectively, during retail storage where the average temperatures were higher as
compared to transport and retail display. During transport, retail storage, and display,
psychrotrophic populations were generally higher than the mesophiles.
2.4.3 Model validation
All RMSEs and biases between the observed microbial growth and the model predictions
were < 1 log CFU/g (Table 2.4). These small RMSEs and biases indicated that both the Baranyi
model and McKellar models were acceptable. Based on our growth data, Baranyi’s model
yielded smaller RMSEs and biases for most of the transport retail storage and display scenarios.
Only one RMSE (Table 2.4 B4) and two biases (Table 2.4 B3 and C4) were slightly higher using
Baranyi’s in comparison to McKellar’s model. Therefore, Baranyi’s model was used for the
Monte Carlo simulations.
2.4.4 Monte Carlo simulation
In most cases, the simulated populations of E. coli O157:H7 and L. monocytogenes
increased < 1 and < 2 log CFU/g, respectively (Fig. 2.3, 2.4, 2.5, and 2.6). Retail storage duration
had the greatest impact on E. coli O157:H7 and L. monocytogenes growth (Fig. 2.3). Storing the
bagged salads for 24 and 72 h at retail storage yielded overall growth (> 1 log CFU/g)
probabilities of 2 and 34%, respectively, for Listeria. Similarly, 24 and 72 h of storage produced
overall E. coli O157:H7 growth (> 1 log CFU/g) probabilities of 0.5% and 3.6%.
Season, bag location in the pallet (top, middle, or bottom), and location in the retail display
case (front or back) did not significantly impact (P > 0.05) microbial growth (Fig. 2.4, 2.5, and
2.6). However, the probability of E. coli O157:H7 exceeding 3 log CFU/g was 29.6 times greater
during transport in summer as compared to fall, winter, and spring. The probability of E. coli

64

O157:H7 exceeding 3 logs CFU/g was also 36.1 times greater when the bag was located near the
bottom of the pallets.

65

Table 2.4 Root Mean Square Errors and biases between observed pathogen growth and the
Baranyi and McKellar model predictions.
Temp
profiles

Transport

Retail
storage

Retail
display

Pathogen
Model
Errors (log
CFU/g)
A1
A2
A3
A4
A5
B1
B2
B3
B4
B5
C1
C2
C3
C4
C5

E. coli O157:H7
Baranyi
McKellar
RMSE
0.20
0.44
a
N/A
N/A
N/A
0.38
0.35
0.26
0.54
0.16
0.11
0.18
0.13
0.14
0.13

L. monocytogenes
Baranyi

Bias
-0.14
0.04

RMSE
0.44
0.49

Bias
0.37
0.18

RMSE
0.33
0.17

Bias
0.16
0.04

N/A
N/A
N/A
0.24
0.10
0.14
0.16
-0.09
0.01
0.11
-0.08
0.05
0.00

N/A
N/A
N/A
0.57
0.72
0.26
0.43
0.35
0.25
0.34
0.15
0.15
0.45

N/A
N/A
N/A
0.47
0.58
0.00
0.20
-0.28
-0.18
0.25
-0.10
0.00
0.37

0.18
0.12
0.14
0.94
0.86
0.38
0.44
0.09
0.17
0.93
0.23
0.25
0.12

0.11
-0.04
0.05
-0.03
-0.20
-0.27
-0.29
0.05
0.11
-0.54
0.15
0.12
-0.02

a

N/A - Growth of E. coli O157:H7 was not assessed.

66

(Log CFU/g)

(Log CFU/g)
Figure 2.3 Probability distributions for E. coli O157:H7 (A, B) and L. monocytogenes (C, D)
from transport to retail storage with 24 and 72 h of retail storage, respectively.

67

Figure 2.3 (cont’d)

(Log CFU/g)

(Log CFU/g)

68

(Log CFU/g)

(Log CFU/g)
Figure 2.4 Probability distributions for E. coli O157:H7 (A, B, C, D) and L. monocytogenes (E,
F, G, H) growth in romaine mix transported during spring, summer, fall and winter.

69

Figure 2.4 (cont’d)

(Log CFU/g)

(Log CFU/g)

70

Figure 2.4 (cont’d)

(Log CFU/g)

(Log CFU/g)

71

Figure 2.4 (cont’d)

(Log CFU/g)

(Log CFU/g)

72

(Log CFU/g)

(Log CFU/g)
Figure 2.5 Probability distributions for E. coli O157:H7 (A, B, C) and L. monocytogenes (D, E,
F) growth in romaine mix at the top, middle, and bottom of the truck during transport.

73

Figure 2.5 (cont’d)

(Log CFU/g)

(Log CFU/g)

74

Figure 2.5 (cont’d)

(Log CFU/g)

(Log CFU/g)

75

(Log CFU/g)

(Log CFU/g)

Figure 2.6 Probability distributions for E. coli O157:H7 (A, B) and L. monocytogenes (C, D)
growth in romaine mix at the front and back of the retail display.

76

Figure 2.6 (cont’d)

(Log CFU/g)

(Log CFU/g)

77

2.5 Discussion
Temperature abuse has the potential to increase the chance for E. coli O157:H7 and L.
monocytogenes growth in bagged salad greens during transport and subsequent handling. To our
knowledge, this is the first large-scale study in the United States using commercial
time/temperature profiles to assess the microbial risk of leafy greens during transport, retail
storage, and display. The impact of season on temperature abuse during transport and product
location in the trucks and retail display cases was also considered. In general, the commercial
cold chain distribution system in the U. S. is well-controlled with most temperatures under 6°C the minimum growth temperature for E. coli O157:H7 (Delaquis and others 2007). Properly
maintained time/temperature profiles also have been reported during cold chain transport in
Canada (McKellar and others 2012) and Belgium (Rediers and others 2009). However,
temperature abuse during transport did exist. Working in Florida, Nunes and others (2009)
measured the mean surface temperatures for bagged salads of 6.8, 7.7 and 8.1°C upon delivery to
three retail stores. These temperatures were higher than our truck temperatures and those from
other studies, probably due to the warm ambient environment. Moreover, a small proportion of
non-commercial trucks have also been reported to carry unrefrigerated perishable foods when
headed to grocery stores and restaurants (Segall 2011).
Our findings and those of others (McKellar and others 2012; Rediers and others 2009)
demonstrate that temperature abuse of commercially produced fresh-cut salad greens is most
likely to occur during retail storage, such as when the refrigeration system is temporarily shut
down during cleaning and sanitizing of the cold room. Storage duration typically varies between
1 and 3 days, depending on the size of the supermarket, supply chain distribution pattern, and

78

consumer sales activity (Danyluk and Schaffner 2011). When combined with varying retail
storage times, these short-term periods of temperature abuse can lead to significant microbial
growth.
Storing bags of romaine mix for 72 h significantly increased the growth probabilities for
both E. coli O157:H7 and L. monocytogenes. Our storage temperatures were similar to the retail
display temperatures recorded by Nunes and others (2009), with temperatures fluctuating from
-1.2 to 19.2°C. Although temperature variations due to seasonal differences and sensor location
did not generally impact microbial populations, increased growth of E. coli O157:H7 was
observed for summer temperatures when products were located near the floor of the truck during
transport. In our study, the 432, 4867, and 3799 time/temperature profiles collected during
transport, retail storage and display, respectively, were handled separately due to logistics. While
impractical, continuous monitoring from transport through home storage would be most
beneficial for risk assessments.
In our study, E. coli O157:H7 and L. monocytogenes populations increased a maximum of
3 logs, which is consistent with the results from Koseki and Isobe (2005a) who used similar
temperatures. However, more growth of E. coli O157:H7 was observed throughout cold chain
distribution compared to McKellar and others (2012) who used colder temperatures during
winter monitoring in Canada. In most cases however, E. coli O157:H7 populations remained
relatively stable during 3 days of simulated transport at well-controlled refrigeration
temperatures. Several researchers also reported slight decreases in E. coli O157:H7 numbers
during 4 to 12 days of incubation at temperatures below those needed for sustained growth
(McKellar and others 2012; Jones and others 2005). Visvalingam and others (2012) showed that

79

E. coli cells in culture at 6°C became elongated but did not divide with cells losing viability at
similar rates.
The effect of sudden temperature changes on the microbial lag phase was previously
assessed by Swinnen and others (2005). Lag phase changes for E. coli are typically not seen
when the temperature fluctuates from 20 to 37°C. However, temperature changes outside this
range will lead to a period of adaptation. Therefore, at the low temperatures encountered in the
cold chain, the temperature spikes caused by normal refrigeration cycles should not impact E.
coli O157:H7 growth.
Using the time/temperature profiles for storage at 9 to 16°C (B1, B2, and B4), L.
monocytogenes, which is psychrotrophic, generally grew better than E. coli O157:H7. However,
E. coli O157:H7 and L. monocytogenes exhibited similar growth rates (P > 0.05) at temperatures
below 9°C or above 16°C with growth of the background microflora similar to the two
pathogens. Greater growth of the mesophiles and psychrotrophs was seen during retail storage
with these average temperatures higher than those observed during transport and display. Koseki
and Isobe (2005a) also indicated that the background bacterial flora did not affect the growth rate
of E. coli O157:H7 or L. monocytogenes.
In this study, laboratory-generated growth data were also used to validate two currently
available predictive models by Baranyi and McKellar for growth of E. coli O157:H7 or L.
monocytogenes in romaine mix at fluctuating temperatures. While both models were able to
predict microbial growth, Baranyi’s model yielded slightly smaller RMSEs and biases. Therefore
Baranyi’s model was used for Monte Carlo simulation. When combined with various secondary
models, Baranyi’s model can also reliably predict E. coli O157:H7 and L. monocytogenes growth
in bagged salads as a function of temperature and relative humidity or carbon dioxide

80

concentration (Carrasco and others 2008; Ding and others 2012). However, Danyluk and
Schaffner (2011) and McKellar and Delaquis (2012) both showed that Baranyi’s model
overestimated the rate of E. coli O157:H7 growth. But most of these growth studies have been
conducted at constant temperatures (Chang and Fang 2007; Delaquis and others 2002; Lee and
Baek 2008; McEvoy and Luo 2009; Theofel and Harris 2009). McKellar and others (2011)
recently developed a growth-death model based on previously published data with lab validation
of their prediction errors needed for future comparison. In one study modified atmosphere
packaging (MAP) was shown to inhibit the growth of L. monocytogenes in fresh-cut lettuce
(Carrasco and others 2008). These lower experimental counts could be due to MAP since Koseki
and Isobe (2005a) did not consider MAP for parameter determination of Baranyi’s model. More
studies and model parameter optimizations are needed to better predict the behavior of pathogens
under MAP conditions at fluctuating temperatures.
Retail storage duration had the greatest impact on E. coli O157:H7 and L. monocytogenes
growth. Our probability distributions demonstrate that the simulated populations of E. coli
O157:H7 and L. monocytogenes increased < 2 log CFU/g in most cases. However, slight growth
does not necessarily equate to a safe product. Low levels of contamination in a large number of
servings are presumed to be responsible for most cases of foodborne illnesses (Danyluk and
Schaffner 2011). In addition, profiles of long term temperature abuse and high average
temperature over time rather than short time temperature abuse such as the defrost cycle of the
refrigerated unit lead to greater probability distributions of E. coli O157:H7 and L.
monocytogenes growth.
Tracing distribution temperatures during commercial transport and retail sale of leafy
greens and mathematically predicting the potential for growth and/or survival of E. coli O157:H7

81

and L. monocytogenes have both been identified as major data gaps in current risk assessments
(Carrasco and others 2010; Danyluk and Schaffner 2011). In a recent quantitative microbial risk
assessment of leafy greens from farm-to-table, the effect of temperature abuse during transport
was not considered (Danyluk and Schaffner 2011). The findings from this study should help fill a
vital data gap in the current E. coli O157:H7 and L. monocytogenes risk assessments being
developed to better define the risks associated with consumption of ready-to-eat bagged salad
products in the U. S.

82

APPENDICES

83

Appendix A

Matlab code for Escherichia coli O157:H7
a) Baranyi dynamic model
function out = Baranyi_dyn(Tempvec, secmod, primod)
% time - time vector in MINUTES
% temp - temperature vector in degrees Celsius
% Secondary model: mu = a(T-Tmin)^2. (1-exp(b(T-Tmax)))
% a, b, Tmin, and Tmax are parameters of the secondary model
% Secondary model: mu = a(T-Tmin)^2
% a, Tmin, are parameters of the secondary model
% a = b^2 from Koseki
% xo - initial innoculum value in log(CFU/g)
% xmax - maximum microbial population in log (CFU/g): equation function of
% T was adopted from Koseki
% h - paremeter in the Baranyi model
close all
clc;
TIME_min = Tempvec(:, 1);
Temp_C = Tempvec(:, 2);
Tmin_C = secmod(1);
a = secmod(2);
x0 = primod(1);
xmax_const1 = primod(2);

84

xmax_const2 = primod(3);
q0 = primod(4);
% clear the screen
clc;
% define time step
dt_min = 5;
dth = dt_min/60;
% Create time vector - user input (will be in hours now)
% it is assumed in uniform time steps.
t = TIME_min/60;
nt = length(t);
tt = [t(1):dth:t(nt)];
csT = spline(t,Temp_C);
TT = ppval(csT,tt);
n = length(TT);
tfinal = tt(n);
% SOLVE DYNAMIC BARANYI'S MODEL USING RUNGE-KUTTA 4th ORDER
% Spline interpolation of temperature to obtain midpoints of each time interval
% Oversampling the time and temperature vectors by a factor of two for
% calculation purposes
csT = spline(tt,TT);
t1 = [t(1):dth/2:tfinal];
T1 = ppval(csT,t1);

% new time vector containing midpoint values
% new temperature vector containing midpoint values

nn = length(t1);
% Discretized Secondary Model

85

% secondary model predicts growth rate as a function of temperature
% determine the growth rate for the oversampled temperature vector.
% initialize growth rate vector
mumax = [];
xmax = [];
for i = 1:length(T1)
xmax(i)=xmax_const1*T1(i)+xmax_const2; % dynamic xmax
if T1(i) >= Tmin_C
mumax(i) = a*(T1(i) - Tmin_C)^2;
elseif T1(i) < Tmin_C
mumax(i) = 0;
end
end
% Numerical Solution of dQ/dt using RK4
% Define Initial Condition
% the "work to be done", constant for a given microorganism at a given
% substrate
%

%h0 = 1.75;

%

alpha = exp(-h0);

%

q0 = 1/(exp(h0)-1);

% initial physiological state of cells
% initial bottleneck substance

q = [];
q(1) = q0;

% initial condition for dq/dt

% Runge-Kutta algorithm
i = 1;
for j = 2:n

86

% PREDICTION
k1 = dth*(mumax(i))*q(j-1);
k2 = dth*(mumax(i+1))*q(j-1);
k3 = dth*(mumax(i+1))*q(j-1);
k4 = dth*(mumax(i+2))*q(j-1);
i = i+2;
% CORRECTION
q(j) = q(j-1) + ((1/6)*(k1 + (2*k2) + (2*k3) + k4));
end
% Numerical Solution of dy/dt using RK4
% Define Initial Condition
x = [];
x(1) = x0;

% initial condition for dy/dt - user input

% Define Maximum Cell Population - constant for a given microorganism at a given
% substrate
% xmax = xmax;
% Spline interpolation of q to obtain midpoints of each time interval
% Oversample or increase the length of q vector by a factor of two
csq = spline(tt,q);
t1 = [t(1):dth/2:tfinal];
qq = ppval(csq,t1);

% new time vector containing midpoint values
% new q vector containing midpoint values

% Runge-Kutta algorithm
i = 1;
for j = 2:n
% PREDICTION

87

k1 = dth*mumax(i)*(qq(i)/(1+qq(i)))*(1-(x(j-1)/(xmax(i))))*x(j-1);
k2 = dth*mumax(i+1)*(qq(i+1)/(1+qq(i+1)))*(1-((x(j1)+(k1/2))/(xmax(i+1))))*(x(j-1)+(k1/2));
k3 = dth*mumax(i+1)*(qq(i+1)/(1+qq(i+1)))*(1-((x(j1)+(k2/2))/(xmax(i+1))))*(x(j-1)+(k2/2));
k4 = dth*mumax(i+2)*(qq(i+2)/(1+qq(i+2)))*(1-((x(j1)+(k3))/(xmax(i+2))))*(x(j-1)+(k3));
i = i+2;
% CORRECTION
x(j) = x(j-1) + ((1/6)*(k1 + (2*k2) + (2*k3) + k4));
end
xpred = x;
out = [tt' xpred'];
% figure, plot (t', y, 'b');
% figure, plot(t', T);
figure, plotyy(tt,xpred,tt,TT);
[AX,H1,H2] = plotyy(tt,xpred,tt,TT,'plot');
grid on
xlabel('Time (h)')
set(get(AX(1),'Ylabel'),'String','log(CFU)/mL')
set(get(AX(2),'Ylabel'),'String',' Temp (C)')
title ('Time (h), Temp (C), log (CFU)/mL')

b) Bacterial growth predicted from Baranyi/Ratkowsky model
% read Excel file and corresponding tab.
% CHANGE DEPEDING ON WHAT FILE YOU WANT TO READ!

88

data = xlsread('sept 15.xls','Sheet1');
t_hr = data(:,1);
%transform t to minutes for model solving
t = t_hr*60;
T = data(:,2);
% define t/T vector
Tempvec = [t,T];
% primary model parameters
primod = [6.91, 0.218, 10.320, 0.059322];%2.63 8.35, 1.7788 % E. coli
% primod = [7.53, 0.037, 12.434, 0.077586];%2.63 8.35, 1.7788 % LM
% x0, xmax_const1, xmax_const2, q0
% secondary model parameters
secmod = [4.54,0.001089]; %2.8988, 0.00164 % E. coli
% secmod = [-4.26,0.000256]; %2.8988, 0.00164 % LM
% Tmin_C, a
% solve/run program!
out = baranyi_dyn(Tempvec, secmod, primod);
maxgrowthlog = out(length(out),2);

c) Monte Carlo simulation for E. coli O157:H7
clear;
% how many random points to collect
randomSelectLength=100000;
% use random seed by time: for better random
% rng('shuffle');
89

format long g
[dates,routes,departs,arrvs,tags,locTrucks,locPallets,seasons,incrEC, incrLM,tmp] =
textread('input\phase1.csv','%s %n %s %s %s %s%s%s%f %f%s','delimiter',',');
% first 1D array to save the 1st phase end value; to serve as the init value
% of the 2nd phase;
phase1Res=incrEC;
% phase1Res=incrLM;
phase1ResLen=length(phase1Res);
% secondary model parameters: const
secmod = [4.54,0.001089]; %2.8988, 0.00164 % E. coli
% secmod = [-4.26,0.000256]; %2.8988, 0.00164 % LM
% Tmin_C, a
% Generate file lists for type B & compute file numbers
phase2_list = dir('phase2_data_output_all_3days/*.csv');
phase2ResLen = length(phase2_list);
% Generate file lists for type C & compute file numbers
phase3_list = dir('phase3_data_output_all_3days_front/*.csv');
phase3ResLen = length(phase3_list);
% Flattened: for drawing chart: Pre-allocation result space: rows and 1 column
allCaseLength = phase1ResLen * phase2ResLen * phase3ResLen;
% save all random selected result: for the final value.
randomSelectAllPhaseCases = nan(randomSelectLength,1);
randomCaseIndex=1;
for rand_iter = 1 : randomSelectLength
randomPhase1Index= randi(phase1ResLen);

90

% Load phase 1 result: note that this value needs to be changed to base n
% (log) from base 10
phase2init= ( phase1Res(randomPhase1Index) +3.0)* log(10);
% phase2init =log( 10.^( phase1Res(iter)+3.0 ));
% Load phase 2
randomPhase2Index=randi(phase2ResLen);
phase2CSVFileName = ['phase2_data_output_all_3days/'
phase2_list(randomPhase2Index).name];
[phase2Time,phase2Temp,dateTimePhase2] =
textread(phase2CSVFileName,'%n%f%s','delimiter',',');
phase2Vec = [phase2Time,phase2Temp];
% now to put the variables into the model to compute the increased
% b number in log
% primary model parameters: need to use the 1st phase result to
% serve as the initial value of the 2nd phase
phase2Primod = [phase2init, 0.218, 10.320, 0.059322];%2.63 8.35, 1.7788 % E. coli
%phase2Primod = [phase2init, 0.037, 12.434, 0.077586];%2.63 8.35, 1.7788 % LM
phase2IncrResultMatrix = Baranyi_dyn(phase2Vec, secmod, phase2Primod);
phase2IncrResult = phase2IncrResultMatrix(end,2);
% just use this log base n to be the init value of the next phase:
% phase 3
phase3init = phase2IncrResult;
% enter phase 3
randomPhase3Index=randi(phase3ResLen);
% Load phase 2 result: note that this value must be changed to base n
% (log) from base 10
91

phase3CSVFileName = ['phase3_data_output_all_3days_front/'
phase3_list(randomPhase3Index).name];
[phase3Time,phase3Temp,dateTimePhase3] =
textread(phase3CSVFileName,'%n%f%s','delimiter',',');
phase3Vec = [phase3Time,phase3Temp];
% now to put the variables into the model to compute the increased
% b number in log
% primary model parameters: need to use the 2nd phase result to
% serve as the initial value of the 3rd phase
phase3Primod = [phase3init, 0.218, 10.320, 0.059322];%2.63 8.35, 1.7788 % E. coli
%phase3Primod = [phase3init, 0.037, 12.434, 0.077586];%2.63 8.35, 1.7788 % LM
phase3IncrResultMatrix = baranyi_dyn(phase3Vec, secmod, phase3Primod);
phase3IncrResult = phase3IncrResultMatrix(end,2);
% change to base 10 for final result
phase3IncrResult=phase3IncrResult / log(10) - 3.0;
randomSelectAllPhaseCases(rand_iter) = phase3IncrResult;
if( mod(rand_iter,10)==0)
disp( ['Reaching case No. ', num2str(rand_iter), '.'])
end
end
% Distribution plot for "result"
clf, hist(randomSelectAllPhaseCases,20);
[n,myHist]=hist(randomSelectAllPhaseCases,20);
bar(myHist,n/randomSelectLength,1);
myFont=30;
xlabel('Growth of \it{E. coli} \rm{O157:H7}','fontsize', myFont);
92

ylabel('Probability Density','fontsize', myFont);
ylim([0 1.0]); xlim([0 5.0]);
%x = normplot(randomSelectAllPhaseCases);
%y = histfit(randomSelectAllPhaseCases);
%data = (randomSelectAllPhaseCases);
%[mu,sigma,muci,sigmaci] = normfit(data);

93

Appendix B

Matlab code for Listeria monocytogenes
Most codes for L. monocytogenes are the same as the codes for E. coli O157:H7 except
parameters xmax_const1=0.037, xmax_const2=12.434, and q0=0.077586.

94

REFERENCES

95

REFERENCES

Abdul-Raouf, U. M., L. R. Beuchat, and M. S. Ammar. 1993a. Survival and growth of
Escherichia coli O157:H7 on salad vegetables. Appl. Environ. Microbiol. 59:1999-2006.
Abdul-Raouf, U. M., L. R. Beuchat, and M. S. Ammar. 1993b. Survival and growth of
Escherichia coli O157:H7 in ground, roasted beef as affected by pH, acidulants, and
temperature. Appl. Environ. Microbiol. 59:2364-2368.
Anonymous. 2011. Outbreaks related to leafy greens. Available at: http://bites.ksu.edu/ leafygreens-related-outbreaks. Accessed 06 November 2011.
Baranyi, J., and T. A. Roberts. 1994. A dynamic approach to predicting bacterial growth in
food. Int. J. Food Microbiol. 23:277-294.
Baranyi, J., and T. A. Roberts. 1995. Mathematics of predictive food microbiology. Int. J. Food
Microbiol. 26:199-218.
Beuchat, L. R., B. B. Adler, and M. M. Lang. 2004. Efficacy of chlorine and peroxyacetic acid
sanitizer in killing Listeria monocytogens on iceberg and romaine lettuce using simulated
commercial processing conditions. J. Food Prot. 67:1238-1242.
Beuchat, L. R., and R. Brackett. 1990. Survival and growth of Listeria monocytogenes on lettuce
as influenced by shredding, chlorine treatment, modified atmosphere packaging and
temperature. J. Food Sci. 55:755-758.
Buchanan, R. L., and L. A. Klawitter. 1992. The effect of incubation temperature, initial pH, and
sodium chloride on the growth kinetics of Escherichia coli O157:H7. Food Microbiol.
9:185-196.
Buchanan, R. L., and J. Phillips. 1990. Response surface model for predicting the effects of
temperature, pH, sodium chloride content, sodium nitrate concentration and atmosphere on
the growth of Listeria monocytogenes. J. Food Prot. 53:370-376.
Buchholz, A. L., G. R. Davidson, B. P. Marks, E. C. D. Todd, and E. T. Ryser. 2009. Transfer of
Escherichia coli O157:H7 from equipment surfaces to iceberg and romaine lettuce during
simulated commercial processing. Abst. Ann. Mtg. Int. Assoc. Food Prot. Grapevine, TX
July 12-15.
Calvin, L., R. Cook, M. Denbaly, C. Dimitri, L. Glaser, C. Handy, M. Jekanowski, P. Kaufman,
B. Krissoff, G. Thompson, and S. Thornsbury. 2003. Changes in fresh fruit and vegetable
marketing. Market and Trade Economics Division, Economic Research Service, U.S.
Department of Agriculture. Agricultural Economic Report No. 795.

96

Cantwell, M. 2002. Optimal handling conditions for fresh produc. In: Kader, A.A. (Ed.),
Postharvest Technology of Horticultural Crops, 3311. University of California, Division of
Agricultural and Natural Resources (Special publication), pp. 511–518.
Carrasco, E., F. Perez-Rodriguez, A. Valero, R. M. Garcia-Gimeno, and G. Zurera. 2008.
Growth of Listeria monocytogenes on shredded, ready-to-eat iceberg lettuce. Food Control.
19:487-494.
Carrasco, E., F. Pérez-Rodríguez, A. Valero, R. M. García-Gimeno, and G. Zurera. 2010. Risk
assessment and management of Listeria monocytogenes in ready to eat lettuce salads.
Compr. Rev. Food Sci. F. 9:498-521.
Canadian Food Inspection Agency. 2009. Code of practice for minimally processed ready-to-eat
vegetables. Available at: http://www.inspection.gc.ca/english/fssa/frefra/safsal/minproe.
shtml#s6. Accessed 4 April 2012.
Centers for Infectious Disease Research and Policy. 2006. Diarrheagenic Escherichia coli.
University of Miniesota. Available at: http://www.cidrap.umn.edu/cidrap/ content/ fs/fooddisease/causes/ecolioview.html. Accessed 5 January 2011.
Centers for Disease Control and Prevention. 2006a. Surveillance for foodborne disease outbreaks
United
states,
1998-2002.
Available
at:
http://www.cdc.gov.
proxy2.cl.msu.edu/mmwr/preview/mmwrhtml/ss5510a1.htm. Accessed 22 December 2010.
Centers for Disease Control and Prevention. 2006b. Multistate outbreak of E. coli O157
Infections, Dec 2006. Available at: http://www.cdc.gov/ecoli/2006/december/ 121406. htm.
Accessed 23 December 2010.
Centers for Disease Control and Prevention. 2010. Investigation update: multistate outbreak of
human E. coli O145 infections linked to shredded romaine lettuce from a single processing
facility. Available at: http://www.cdc.gov/ecoli/2010/ecoli_o145/ index.html. Accessed 03
November 2011.
Centers for Disease Control and Prevention. 2011a. Investigation update: outbreak of shiga
toxin-producting E. coli O104 (STEC O104:H4) infections associated with travel to
Germany. Available at: http://www.cdc.gov/ecoli/2011/ecoliO104/ index.html. Accessed
03 November 2011.
Centers for Disease Control and Prevention. 2011b. Multistate outbreak of listeriosis linked to
whole
cantaloupes
from
Jensen
farms,
Colorado.
Available
at:
http://www.cdc.gov/listeria/outbreaks/cantaloupes-jensen-farms/index.html. Accessed 04
November 2011.

97

Chang, J. M., and T. J. Fang. 2007. Survival of Escherichia coli O157:H7 and Salmonella
enterica serovars Typhimurium in iceberg lettuce and the antimicrobial effect of rice
vinegar against E. coli O157:H7. Food Microbiol. 24:745–751.
Chapman, B. 2012. North Carolina’s adoption of the FDA Model Food Code and requiring cut
leafy
greens
to
be
received
at
41F.
Available
at
http://barfblog.foodsafety.ksu.edu/blog/154529/12/04/26/north-carolinas-adoption-fdamodel-food-code-and-requiring-cut-leafy-greens-be- Accessed 22 May 2012.
Cho, M., J. Kim, J. Y. Kim, J. Yoon, and J. H. Kim. 2010. Mechanisms of Escherichia coli
inactivation by several disinfectants. Water Res. 44:3410-3418.
Cole, M. B., M. V. Jones, and C. Holyoak. 1990. The effect of pH, salt concentration and
temperature on the survival and growth of Listeria monocytogenes. J. App. Bacteriol. 69:
63-72.
Danyluk, M. D., and D. W. Schaffner. 2011. Quantitative assessment of the microbial risk of
leafy greens from farm to consumption: preliminary framework, data, and risk estimates. J.
Food Prot. 74:700-708.
Delaquis, P., S. Bach, and L. D. Dinu. Behavior of Escherichia coli O157:H7 in leafy
vegetables. J. Food Prot. 70:1966-1974.
Delaquis, S., S. Stewart, S. Cazaux, and P. Toivonen. 2002. Survival and growth of Listeria
monocytogenes and Escherichia coli O157:H7 in ready-to-eat iceberg lettuce washed in
warm chlorinated water. J. Food Prot. 65:459–464.
Del-Rosario, A. A., and L. R. Beuchat. 1995. Survival and growth of enterohemorrhagic
Escherichia coli O157:H7 in cantaloupe and watermelon. J. Food Prot. 58:105-107.
De Simon, M., C. Tarrago, and M. D. Ferrer. 1992. Incidence of Listeria monocytogenes in fresh
foods in Barcelona (Spain). Int. J. Food Microbiol. 16:153–156.
DeWaal, C. S., C. Roberts, C. Catella. 2012. Outbreak alert! 1999-2008. Available at:
http://www.cspinet.org/foodsafety/PDFs/Outbreak_Alert_1999-2008.pdf. Accessed 31
May 2012.
Diez-Gonzalez, F., and A. Mukherjee. 2009. Produce safety in organic vs. conventional crops. In
Microbial safety of fresh produce edited by Fan, X., B. A. Niemira, C. J. Doona, F. E.
Feeherry, and R. B. Gravani. p 83-99. IFT Press, Wiley-Blackwell.
Ding, T., J. Wang, F. Forghani, S.-D. Ha, M.-S. Chung, G.-J. Bahk, I.-G. Hwang, E. Abdallah,
and D.-H. Oh. 2012. Development of predictive models for the growth of Escherichia
coli O157:H7 on cabbage in Korea. J. Food Sci. doi: 10.1111/j.1750-3841.2012.02660.

98

Estrada-Flores, S., N. Smale, and D. J. Tanner. 2006. Temperature regulations during transport
of perishables in Australia. Proc. Int. Cong. Refrig. ICR. 0704:87-95.
EU food hygiene legislation. 2006. The Food Hygine (England) Regulations 2006 No.14.
Farber, J. M., and P. I. Peterkin. 1991. Listeria monocytogenes, a food-borne pathogen.
Microbiol. Rev. 55:476-511.
Food and Agriculture Organization. 2003. Handling and preservation of fruits and vegetables by
combined methods for rural areas. FAO agricultural services bulletin 149.
Food and Agriculture Organization/World Health Organization. 1995. Application of risk
analysis to food standards issues. Report of the Joint FAO/WHO Expert Consulation.
Available at: http://www.who.int/foodsafety/publications/micro/en/march19 95.pdf
Accessed 17 April 2012.
Food Safety Authority of Ireland. 2006. Guidance note No.16: food stalls (revision 1). Available
http://www.fsai.ie/assets/0/86/204/38b489aa-c7c5-4526-b035-f136860593f4.pdf.
at:
Accessed 11 April 2012.
Francis, G. A., C. Thomas, and D. O’Beirne. 1999. The microbiological safety of minimally
processed vegetables. Int. J. Food Sci. Technol. 34:1–22.
Fu, B., P. S. Taoukis, and T. Labuza. 1991. Predictive microbiology for spoilage of dairy
products with time-temperature integrators. J. Food Sci. 56:1209-1215.
Gessener, G. H., L. Volonino, and L. A. Fish. 2007. One-up, one-back ERM in the food supply
chain. Inf. Syst. Manag. 24:213-222.
Gill, C. O. 1984. Prevention of early spoilage of livers. In: Proceedings 30th European Meeting
of Meat Research Workers, pp. 240–241.
Gill, C. O., J. C. McGinnis, K. Rahn, and A. Houde. 1996. Control of product temperatures
during the storage and transport of bulk containers of manufacturing beef. Food Res. Int.
29:647-651.
Gonsalves, A.
2004.
Albertsons launches RFID initiative.
TechWeb Network.
http://www.networkcomputing.com/data-protection/albertsons-launches-rfid-initiative. php.
Accessed 12 January 2011.
Guerrant, R. L., T. S. Steiner. 2005. Principles and syndromes of enteric infection, p. 1215-1231.
In G. L. Mandell, J. E. Bennett, and R. Dolin (ed.), Mandell: Douglas and Bennett's
principles and practice of infectious diseases, Ed 6. Churchill Livingstone, Philadelphia,
PA.

99

Hafen, D. 2012. Temperature lane mapping
http://www.danhafen.com/ Accessed: 22 May 2012.

with

indicators.

Available

at:

Hardgrave, B., M. Waller, and R. Miller. 2005. Does RFID reduce out of stocks? A Preliminary
Analysis. Available at: http://itri.uark.edu/104.asp?code=rfid&article= ITRI-WP058-1105.
Accessed 12 January 2011.
Hardenburg, R.E., A.E. Watada and C.I., Wand. 1986. The Commercial Storage of Fruits,
Vegetables, and Florist and Nursery Stocks. United States Department of Agriculture,
Agriculture Research Service, Agriculture Handbook Number 66, 130 pp.
Hoelzer, K., R. Pouillot, and S. Dennis. 2012. Listeria monocytogenes growth dynamics on
produce: a review of the available data for predictive modeling. Foodborne Pathog. Dis.
doi:10.1089/fpd.2011.1087.
Hof, H. 1996. Listeria monocytogenes, Ch. 16. In S. Baron (ed.), Baron's Medical Microbiology,
Ed 4. Univ of Texas Medical Branch. NCBI Bookshelf, ISBN 0-9631172-1-1.
Houtsma, P., M. Kant-Muermans, F. Rombouts, and M. Zwietering. 1996. Model for the
combined effects of temperature, pH, and sodium lactate on growth rates of Listeria
innocua in broth and bologna-type sausages. Appl. Environ. Microbiol. 62:616-1622.
Jacxsens, L., F. Develieghere, and J. Debevere. 2002. Temperature dependence of shelf-life as
affected by microbial proliferation and sensory quality of equilibrium modified atmosphere
packaged fresh produce. Postharvest Biol. Technol. 26:59-73.
Johannes, L., and W. Romer. 2010. Shiga toxins-from cell biology to biomedical applications.
Nat. Rev. Microbiol. 8:105-116.
Jones, T., C. O. Gill, and L. McMullen. 2002. The behavior of log phase Escherichia coli at
temperatures below the minimum for sustained growth. Food Microbiol. 19:83-90.
Junttila, J. R., S. I. Niemela, and J. Hirn. 1988. Minimum growth temperatures of Listeria
monocytogenes and non-haemolytic listeria. J. Appl. Bacteriol. 65: 321-327.
Juneja, V. K., H. Marks, and H. Thippareddi. 2008. Predictive model for growth of Clostridium
perfringens during cooling of cooked uncured beef. Food Microbiol. 25:42-55.
Kauppi, K. L. 1996. Influence of substrate and low temperature growth and survival of
verotoxigenic Escherichia coli. Food Microbiol. 13:397-405.
Koseki, S., and S. Isobe. 2005a. Prediction of pathogen growth on iceberg lettuce under real
temperature history during distribution from farm to table. Int. J. Food Microbiol. 104:239248.
Koseki, S., and S. Isobe. 2005b. Growth of Listeria monocytogenes on iceberg lettuce and solid
media. Int. J. Food Microbiol. 101:217-225.

100

Koutsoumanis, K., P. S. Taoukis, and G. J. E. Nychas. 2005. Development of a safety
monitoring and assurance system for chilled food products. Int. J. Food Microbiol.
100:253-260.
Lee, S. Y., and S. Baek. 2008. Effect of chemical sanitizer combined with modified atmosphere
packaging on inhibiting Escherichia coli O157:H7 in commercial spinach. Food Microbiol.
25:582–587.
Le Marc, Y., V. Huchet, C. Bourgeois, J. Guyonnet, P. Mafart, and D. Thuault. 2002. Modeling
the growth kinetics of Listeria as a function of temperature, pH, and organic acid
concentration. Int. J. Food Microbiol. 73:219-237.
Luo, Y., Q. He, J. L. McEvoy, and W. S. Conway. 2009. Fate of Escherichia coli O157:H7 in the
presence of indigenous microorganisms on commercially packaged baby spinach as
impacted by storage temperature and time. J. Food Prot. 72:2038–2045.
Luo, Y., Q. He, and J. L. McEvoy. 2010. Effect of storage temperature and duration on the
behavior of Escherichia coli O157:H7 on packaged fresh-cut salad containing romaine and
iceberg lettuce. J. Food Sci. 75: 390-397.
Mandrell, R. E. 2009. Enteric human pathogens associated with fresh produce: sources,
transport, and ecology, p. 5-41. In X. Fan, B. A. Niemira, C. J. Doona, F. E. Feeherry, and
R. B. Gravani (ed.), Microbial Safety of Fresh Produce. IFT Press, Wiley-Blackwell,
Hoboken, NJ.
McEvoy, J. L., Y. Luo, B. Zhou, H. Feng, and W. S. Conway. 2009. Potential of Escherichia coli
O157:H7 to grow on field-cored lettuce as impacted by postharvest storage time and
temperature. Int. J. Food Microbiol. 128:506–509.
McKellar, R. C., D. L. LeBlanc, J. Lu, and P. Delaquis. 2012. Simulation of Escherichia coli
O157:H7 behavior in fresh-cut lettuce under dynamic temperature conditions during
distribution from processing to retail. Foodborne Pathog. Dis. 9:239-244.
McKellar, R. C., and P. Delaquis. 2011. Development of a dynamic growth-death model for
Escherichia coli O157:H7 in minimally processed leafy green vegetables. Int. J. Food
Microbiol. 151:7-14.
Membre, J. M., J. Thurette, and M. Catteau. 1997. Modeling the growth, survival and death of
Listeria monocytogenes. J. Appl. Microbiol. 82:345-350.
Meng, J., and M. P. Doyle. 2002. Introduction. Microbiological food safety. Microbes Infect.
4:395–397.
Nataro, J. P., and J. B. Kaper. 1998. Diarrheagenic Escherichia coli. Clin. Microbiol. Rev.
11:142-201.

101

Nguyen, R. N., L. S. Taylor, M. Tauschek, and R. M. Robins-Browne. 2006. Atypical
enteropathogenic Escherichia coli infection and prolonged diarrhea in children. Emerg.
Infect. Dis. 12:597-603.
Nicola, S., E. Fontana, C. Torassa, and J. Hoeberechts. 2006. Fresh-cut produce: postharvest
critical issues. Acta Hortic. (ISHS) 712:223–230.
Nunes, M. C. N., J. P. Emond, M. Rauth, S. Dea, and K. V. Chau. 2009. Environmental
conditions encountered during typical consumer retail display affect fruit and vegetable
quality and waste. Postharvest Biol. Technol. 51:232-241.
O’Beirne, D. 2007. Microbial safety of fresh-cut vegetables. Acta Hortic. 746:159–172.
Pizarro-Cerda, J., S. Sousa, and P. Cossart. 2004. Exploitation of host cell cytoskeleton and
signaling during Listeria monocytogenes entry into mammalian cells. C. R. Biol. 327:523–
531.
Pouillot, R., M. B. Lubran, S. C. Cates, and S. Dennis. 2010. Estimating parametric distributions
of storage time and temperature of ready-to-eat foods for U.S. households. J. Food Prot.
73:312–321.
Rangel, J.M., P. H. Sparling, C. Crowe, P. M. Griffin, and D. L. Swerdlow. 2005. Epidemiology
of Escherichia coli O157:H7 outbreaks, United States, 1982-2002. Emerg. Infect Dis.
11:603-609.
Rediers, H., M. Claes, L. Peeters, and K. A. Willems. 2008. Evaluation of the cold chain of
fresh-cut endive from farmer to plate. Postharvest Biol. Technol. 51:257-262.
Riley, L. W., R. S. Remis, S. D. Helgerson, H. B. McGee, J. G. Wells, B. R. Davis, R. J. Hebert,
E. S. Olcott, L. M. Johnson, N. T. Hargrett, P. A. Blake, and M. L. Cohen. 1983.
Hemorrhagic colitis associated with a rare Escherichia coli serotype. N. Engl. J. Med.
308:681–5.
Rosset, P., M. Cornu, V. Noel, E, Morelli, and G. Poumeyrol. 2004. Time-temperature profiles
of chilled ready-to-eat foods in school catering and probabilistic analysis of L.
monocytogenes growth. Int. J. Food Microbiol. 96:49-59.
Ryu, J. H., H. Kim, and L. R. Beuchat. 2004. Attachment and biofilm formation by Escherichia
coli O157:H7 on stainless steel as influenced by exopolysaccharide production, nutrient
availability, and temperature. J. Food Prot. 67:2123-2131.
Sant’Ana, A. S., B. D. G. M. Franco, and D. W. Schaffner. 2012. Modeling the growth rate and
lag time of different strains of Salmonella enteric and Listeria monocytogenes in ready-toeat lettuce. Food Microbiol. 30:267-273.
Segall, B. 2011. 13 Investigates “hot trucks”. Available at: http://www.wthr.com/story/
15144020/13-investigates-hot-trucks. Accessed 31 May 2012.

102

Scaletsky, I. C. A., S. H. Fabbricotti, R. L. B. Carvalho, C. R. Nunes, H. S. Maranhão, M. B.
Morais, and U. Fagundes-Neto. 2002. Diffusely adherent Escherichia coli as a cause of
acute diarrhea in young children in northeast Brazil: a case-control study. J Clin
Microbiol. 40:645–648.
Scallan, E., R. M. Hoekstra, F. J. Angulo, R. V. Tauxe, M. Widdowson, S. L. Roy, J. L. Jones,
and P. M. Griffin. 2011. Foodborne illness acquired in the United States – major pathogens.
Emerg. Infect. Dis. 17: 7-15.
Schlech, W. F., P. M. Lavigne, R. A. Bortolussi, A. C. Allen, E. V. Haldane, A. J. Worth, A. W.
Hightower, S. E. Hohnsone, S. H. King, E. S. Nicholls, and C. V. Broome. 1983. Epidemic
listeriosis-evidence for transmission by food. N. Engl. J. Med. 308: 203–206.
Songini, M. L. 2006. Wal-Mart details its RFID journey. Available at: http://www.
computerworld.com/s/article/109132/Wal_Mart_details_its_RFID_journey. Accessed 11
January 2011.
Sorrells, K. M., D. C. Enigl, and J. R. Hatfield. 1989. Effect of pH, acidulant, time, and
temperature on the growth and survival of Listeria monocytogenes. J. Food Prot. 52:571573.
Swinnen, I. A. M., K. Bernaerts, K.Gysemans, and J. F. Van Impe. 2005. Quantifying microbial
lag phenomena due to a sudden rise in temperature: a systematic macroscopic study. Int. J.
Food Microbiol. 100:85-96.
Szabo, E. A., Scurrah, K. J., & Burrows, J. M. 2000. Survey for psychrotrophic bacterial
pathogens in minimally processed lettuce. Lett. Appl. Microbiol. 30: 456–460.
Takeuchi, K., and J. Frank. 2000. Penetration of Escherichia coli O157:H7 into lettuce tissues as
affected by inoculum size and temperature and the effect of chlorine treatment on cell
viability. J. Food Prot. 63:434–440.
Tam, P. J., and C. A. Lingwood. 2007. Membrane cytosolic translocation of verotoxin A1
subunit in target cells. Microbiology 153:2700-2710.
Theofel, C. G., and L. J. Harris. 2009. Impact of preinoculation culture conditions on the
behavior of Escherichia coli O157:H7 inoculated onto romaine lettuce (Lactuca sativa)
plants and cut leaf surfaces. J. Food Prot. 72:1553–1559.
Todar, K. 2008a. Pathogenic E. coli. Available at: http://www.Textbookofbacterio logy.net/e.coli
2.html. Accessed 30 December 2010.
Todar, K. 2008b. Listeria monocytogenes. Available
bacteriology.net/Listeria.html. Accessed 5 January 2011.

103

at:

http://www.Textbookof

U. S. Department of Agriculture Economic Research Service. 2003. Changes in fresh fruit and
vegetable marketing.
U. S. Department of Agriculture Economic Research Service. 2010. Foodborne illness cost
calculator. Available at: http://www.ers.usda.gov/Data/FoodborneIllness. Accessed 22
December 2010.
U. S. Department of Agriculture, Food Safety and Inspection Service. 2005. Use of microbial
pathogen computer modeling in HACCP plans, FSIS notice 25-05. Available at:
http://www.fsis.usda.gov/regulations/notice25-05/default.asp. Accessed 21 December
2010.
U. S. Food and Drug Administration. 1998. Guide to minimize microbial food safety hazards for
fresh fruits and vegetables. Center for Food Safety and Applied Nutrition, Washington, D.
C.
U. S. Food and Drug Administration. 2005a. FDA letter to California firms that grow, pack,
process, or ship fresh and fresh-cut lettuce, 4 Nov 2005. Available at:
http://www.cfsan.fda.gov/~dms/prodltr2.htm. Accessed 23 December 2010.
U. S. Food and Drug Administration. 2005 b. Listeria monocytogenes outbreaks on the safety of
raw milk, Nov 3 2005. Available at: http://www.fda.gov/Food/FoodSafety/
ProductSpecificInformation/MilkSafety/ConsumerInformationAboutMilkSafety/ucm16546
8.html. Accessed 25 November 2010.
U. S. Food and Drug Administration. 2006. FDA statement on food-borne E. coli O157:H7
outbreak
in
spinach.
Available
at:
http://www.fda.gov/bbs/topics/NEWS/
2006/NEW01486.html. Accessed 23 December 2010.
U. S. Food and Drug Administration. 2007. FDA Statement on food-borne E. coli O157:H7
recall
in
packaged
salads.
Available
at:
http://www.fda.gov/oc/po/
firmrecalls/dole0907.html Accessed 27 December 2010.
U. S. Food and Drug Administration. 2008. Guide to minimize microbial food safety hazard for
fresh fruits and vegetables.
U. S. Food and Drug Administration. 2009. FDA Food Code 2009. Available at:
http://www.fda.gov/Food/FoodSafety/RetailFoodProtection/FoodCode/FoodCode2009/def
ault.htm. Accessed 7 January 2011.
U. S. Food and Drug Administration. 2010a. DSHS Orders Sangar Produce to close, recall
products. Available at: http://www.fda.gov/Safety/Recalls/ucm230709.html. Accessed 26
November 2010.
U. S. Food and Drug Administration. 2010b. Kosher First LLC. Issues Alert on Listeria in Tuv
Taam salads nova lox salad. Available at: http://www.fda.gov/Safety/Recalls/
ucm231736.html. Accessed 21 December 2010.

104

U.S. Food and Drug Administration. 2012. Search result for lettuce in FDA-recalls. Available at:
http://google2.fda.gov/search?q=lettuce&client=FDAgov&proxystyle
sheet=FDAgov&output=xml_no_dtd&sort=date%253AD%253AL%253Ad1&x=38&y=12
&entqr=3&entqrm=0&oe=UTF-8&ie=UTF-8&ud=1&site=FDAgov-recalls&filter
=0&btnG=Search
U. S. Food and Drug Administration. U. S. Department of Agriculture, Center for Disease
Control and prevention. 1998. Guidance for industry: guide to minimize microbial food
safety hazard for fresh fruits and vegetables.
Visvalingam, J., C. O. Gill, and R. A. Holley. 2012. The viabilities of cells in cultures of
Escherichia coli growth with formation of filaments at 6°C. Int. J. Food Microbiol.
153:129-134.
Wessel, R. 2007. German meat-tracking project focuses on lasers and RFID. Available at:
http://www.rfidjournal.com/article/articleview/2941/. Accessed January 12, 2010.
Whyte, R., N. King, G. Bayne, P. van der Logt, and A. Hudson. 2006. Temperature control of
meat during transport and retail display. Abstract for the International Association for Food
Protection Annual Meeting, Calgary, Alberta, Canada, Aug 13-16, (Abstract number P524).
Zhou, B., H. Feng, and Y. Luo. 2009. Ultrasound enhanced sanitizer efficacy in reduction of
Escherichia coli O157:H7 population on spinach leaves. J. Food Sci. 74:308–313.
Zwitering, M. H., I. Jongenburgen, F. M. Rombouts, and K. Van’t Rier. 1990. Modeling of the
bacterial growth curve. Appl. Environ. Microbiol. 56:1875-1881.
Zwietering, M. H., De Wit, J. C., & Notermans, S. 1996. Application of predictive microbiology
to estimate the number of Bacillus cereus in pasteurized milk at the point of consumption.
Int. J. Food Microbiol. 30:55-70.

105