Predictive food microbiology as a tool in risk ... - ProSafeBeef

Predictive food microbiology as a tool in
risk assessment
Kostas Koutsoumanis
Assistant Professor
Lab of Food Microbiology and Hygiene Dpt. Of Food
Science and Technology
Aristotle University of Thessaloniki,

Presentation Outline
1.Food Safety Management
1.1 The Risk Analysis process
1.2 Risk Assessment
1.3 Risk Management
2.Predictive Microbiology
2.1 Kinetic Models
2.1.1 Primary
2.1.2 Secondary
2.1.3 Tertiary (software)
2.2 Growth/no growth boundary models
2.3 Stochastic Models
3.Case Studies
3.1 Exposure assessment of L monocytogenes in
pasteurized milk
3.2 Risk-based approach for evaluating compliance of
RTE food to the safety criteria for L. monocytogenes

Food Safety Management
Traditional Food Safety Management approach was based
on end-product testing
End Product Sampling
Is the product
safe?
Yes
No
Food Safety
Qualitative (Discrete)
variable

Food Safety Management
Traditional Food Safety Management approach was based
on end-product testing
Food Safety Crisis of 90’s
Antibiotics
Listeria
BSE
SARS
Dioxins
E. coli
Campylobacter
Acrylamide
Salmonella
Need for change

Food Safety Management
Development of new tools
New Tools
Predictive Microbiology
Risk Assessment

Risk Analysis
Risk Assessment is a component of Risk Analysis
(WHO/FAO, 1995):
Risk
Assessment
Risk
Communication
Risk
Management

Risk Analysis
Risk Assessment is a component of Risk Analysis

Risk Analysis
The Risk Analysis Process - CAC

Risk Assessment
Risk Assessment Stages
Hazard Identification: what biological, chemical and
physical agents are we dealing with and with which foods is
it associated?
Hazard Characterization: what illness can be caused,
associated in relation to dose and population?
Exposure Assessment: how likely it is that an individual or
a population will be exposed to a microbial hazard and what
numbers of organisms are likely to be ingested?
Risk Characterization: the integration of the above
resulting in the probabilities of illness

Risk Assessment
Exposure Assessment
Exposure assessment provides an estimate of the
occurrence and level of the pathogen in a specified portion
of a certain food at the time of consumption in a specified
population by taking into account uncertainty and variability
Exposure = f(Contamination ; Consumption)
Prevalence (proportion of
contaminated units)
and levels of contamination
(in contaminated units)
at the time of consumption
Consumption rates
and serving sizes

Exposure Assessment
Data and tools required for Exposure Assessment
‣Food product flows
-Farm-to-fork
-Import and export
‣Pathogen dynamics at starting point
-Prevalence
-Numbers
‣Processing
-Conditions
-Effects (models)
‣Distribution, retail and domestic storage
-Conditions (Time-Temperature)
-Effects (models
‣Consumption patterns
-Intake
-Preparation

Exposure Assessment
Variability and Uncertainty (Definitions)
Variability represents a true heterogeneity of the
population that is a consequence of the physical system and
irreducible (but better characterized) by further
measurements.
‣Variability between sub-populations
Examples: differences in serving sizes between
infants/children/teenagers/adults, male versus female…
‣Variability within a (sub-) population
Examples: variability of serving sizes from one person to
another,from one serving (cocktail) to another (main meal)…

Exposure Assessment
Variability and Uncertainty (Definitions)
Uncertainty represents our lack of knowledge and may
include:
‣scenario uncertainty
‣model uncertainty
‣parameter uncertainty

Exposure Assessment
Variability (Example)
We all want to move to the 5 th floor using the elevator in
groups of 5 (randomly selected) people
The weight limit of the elevator is 480 kg
Estimate the chance of exceeding the weight limit
Deterministic method (variability is not taken into account)
Average individual weight=70 kg
5 persons x 70 =350 kg

Exposure Assessment
Variability (Example)
Stochastic method (variability is taken into account)
25
Normal (70,8 kg)
20
%Frequency
15
10
5
0
50 55 60 65 70 75 80 85 90 95 100
Individuals average Weight

Variability (Example)
Exposure Assessment
Stochastic method (variability is taken into account)
%Frequency
25
20
15
10
5
0
50 55 60 65 70 75 80 85 90 95 100
Individuals average Weight
Random selection of
5 values
Repeat 100000
(iterations)
Sum the 5 values
0,1
Monte Carlo
Simulation
Probability
0,08
0,06
0,04
0,02
Limit
P(>480)=3x10 -4
0
240
260
280
300
320
340
360
380
400
420
440
460
480
500
Total Weight of 5 persons

Exposure Assessment
Uncertainty (Example)
Stochastic method (variability is taken into account)
Uncertainty: We don’t know the weight limit of the
elevator
Expert Opinion: Min:450, Max:550 Most likely:500
Triang(450; 500; 550)
2,5
X

0
0
2
1
4
2
3
6
8
0
2
4
0
6
8
1
2
3
0
1
2
3
0,40
0,35
0,30
0,25
Expon(2,5) Shift=-2,5
Exposure Assessment
0,40
Triang(-2,5; 0; 2,5)
0,20
0,35
0,15
0,30
0,10
0,25
0,05
0,20
0,00
-4
-2
90,0%
-2,37 4,99
5,0%
Contamination
at t=0
10
>
0,15
0,10
0,05
0,00
-3
-2
-1
5,0% 90,0%
5,0%
-1,709 1,709
Consumption
data
MODELS (growthinactivation-survival
Contamination at
consumption time
Normal(0; 1)
0,40
Triang(-2,5; 0; 2,5)
Monte
Carlo
0,45
0,40
0,35
0,35
0,30
0,30
0,25
0,20
0,15
0,25
0,20
0,15
Exposure
0,10
0,10
0,05
0,00
-3
-2
-1
0,05
0,00
Lognorm(2,5; 2,5) Shift=-2,5
< 5,0% 90,0%
5,0% >
-1,645 1,645
-3
-2
-1
5,0% 90,0%
5,0%
-1,709 1,709
0,40
0,35
Processing,
distribution,
storage
parameters
0,40
0,35
0,30
0,25
0,20
0,15
0,10
0,05
0,00
-4
-2
Lognorm(2,5; 2,5) Shift=-2,5
90,0%
-2,05 4,45
5,0%
10
12
>
0,30
0,25
0,20
0,15
0,10
0,05
0,00
-4
-2
0
2
4
90,0%
-2,05 4,45
6
8
5,0%
10
12
>

Risk Characterization
< 5,0% 90,0%
5,0% >
-1,645 1,645
Exposure
Dose-Response
Lognorm(2,5; 2,5) Shift=-2,5
100
0,40
0,35
0,30
0,25
0,20
0,15
0,10
0,05
0,00
-4
-2
0
2
4
90,0%
-2,05 4,45
6
8
5,0%
10
12
>
Frequency of Infection, %
90
80
70
60
50
40
30
20
10
0
0 1 2 3 4 5
Dose (Log10 cfu)
6
7
Data
Gompertz-Log
Probability of illness
Normal(0; 1)
0,45
0,40
0,35
0,30
0,25
0,20
0,15
0,10
0,05
0,00
-3
-2
-1
0
1
2
3

Food Safety Management
Microbiological Risk Assessment estimates
Number of cases of (a certain) illness per year per
(e.g.) 100.000 persons in a given population caused
by a certain micro-organism or group of
microorganisms in a particular food or food type
The output of Risk Assessment
is a probability (Risk)
e.g 10 -6

Food Safety Management
Food Safety Management approach
Food Safety
Is the product safe? Yes/No
Qualitative (Discrete)
variable
Risk Assessment
Quantitative (Continues)
variable
What is the degree of product safety? Risk

Food Safety Management
Food Safety Management approach
ACCEPTABLE RISK?
100% Safety (zero risk) does not
exists and should not be expected

Food Safety Management
Food Safety Management approach
1996: WTO/SPS agreement
Appropriate Level of Protection (ALOP)
Level of protection deemed appropriate by a
member (country) establishing a sanitary or
phytosanitary measure to protect human, animal or
plant life or health within its territory.

Food Safety Management
Food Safety Management approach
Appropriate Level of Protection (ALOP)
Degree of risk that a society is willing to
tolerate/accept
The “costs” that society is willing to bear to
achieve a specific degree of control over a
Hazard
“Costs” includes: human, quality, nutritional,
economic, ethical, medical, legal, etc
ALOP example: less than 0.25 cases of
Listeriosis per 100,000 people per year in USA

Food Safety Management
Food Safety Management approach
Need for conversion of ALOP into actions. To
implement actions, an objective must be defined,
expressed in terms of a level of a hazard in a
food
Food Safety Objective (FSO)
The maximum frequency and/or concentration of
a microbial hazard in a food at the moment of
consumption that provides the ALOP

Food Safety Management
Food Safety Management approach
Establishment FSO based on ALOP
Dose-Response relationship

Food Safety Management
Food Safety Management approach
ICMSF Safety Equation
H 0 - ΣR+ ΣG+ ΣC < FSO.
H 0 : Initial contamination
ΣR: Total Reduction (log CFU/g)
ΣG: Total Growth (log CFU/g)
ΣC: Total Contamination (log CFU/g)

Food Safety Management
Conversion of FSO to practical criteria for the
food industry
Performance Objective: The maximum frequency and/or
concentration of a (microbial) hazard in a food at a
specified step in the food chain before time of consumption
that still provides or contributes to the achievement of an
FSO.
Performance Criterion: The effect of one or more control
measure(s) needed to meet or contribute to meeting a PO.
Process Criterion: The conditions pf a process that lead to
a PO.

Food Safety Management: Summary
Gorris, 2004

Predictive Microbiology
THE CONCEPT
A detailed knowledge of microbial responses to
environmental conditions, synthesized in a
mathematical model, enables objective evaluation
of processing, distribution and storage operations
on the microbiological safety and quality of foods,
by monitoring the environment without
recourse to further microbiological analysis

Predictive Microbiology
THE CONCEPT
‣Growth, survival and inactivation of microorganisms in
foods are reproducible responses
‣A limited number of environmental parameters in foods
determine the kinetic responses of microorganisms
(Temperature, Water activity/water phase salt, pH, Food
preservatives
‣A mathematical model that quantitatively describes the
combined effect of the environmental parameters can be
used to predict growth, survival or inactivation of a
microorganism and thereby contribute important information
about product safety and shelf-life
Roberts and Jarvis (1983)

Predictive Microbiology
APPLICATIONS
‣Predict the effect of product characteristics and storage
conditions on microbial responses (safety and shelf-life)
‣Predict effect of changes in parameters (product
development)
‣HACCP plans – establish limits for CCP
‣Food safety objectives – equivalence of processes
‣Education – easy access to information
‣Quantitative microbiological risk assessment (QMRA)
(The concentration of microbial hazards in foods may
increase or decrease substantially during processing and
distribution)

Predictive Microbiology
TYPES OF MODELS
‣Primary models: describing the microbial evolution
(growth, inactivation, survival) as a function of
time. Estimate kinetic parameters
‣Secondary models: describing kinetic parameters
as a function of influencing factors like pH,
temperature, water activity, concentration of
preservatives, …
‣Tertiary models: integrate primary and
secondary models in a software tool

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Experimental design- data collection
‣Estimation of kinetic parameters (primary
models)
‣Mathematical description of the effect of the
environmental factors to the kinetic parameters
(secondary models)
‣Validation of the model
‣Integration in a software tool (tertiary models)

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣data collection
11
9
7
Log cfu/g
4
2
0
0 5 10 15 20 25 30
TIME (h)

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Fitting data to primary model
11
9
Log cfu/g
7
4
2
0
0 5 10 15 20 25 30
TIME (h)

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Estimation of kinetic parameters
9
11
Log N max
Log cfu/g
7
4
Log N max Maximum rate
0 5 10 15 20 25 30
Log N 0
2
0
0 5 10 15 20 25 30
Lag
TIME (h)

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Primary models

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Primary models

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Mathematical description of the effect of the
environmental factors to the kinetic parameters
(secondary models)

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Secondary models
‣Kinetic models
• Polynomial and constrained linear polynomial
models
• Square-root-type models
• Arrhenius type models
• Cardinal parameter models
• Artificial neural networks
‣Growth/no growth interface models
(probabilistic models)

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Secondary models
Cardinal parameter model

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Secondary models
Square root type model

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Growth no growth boundary models

Predictive Microbiology
STEPS IN MODEL DEVELOPMENT
‣Validation of models
Most models are developed in laboratory media. There can
be no guarantee that predicted values will match those
that would occur in any specific food system. Before the
models could be used in such a manner, the user would
have to validate the models for each specific food of
interest.
Internal validation: Comparison between predicted and
observed values for data used for model development
External validation: Comparison between predicted and
observed values for independent data

Predictive Microbiology
Tertiary models-Software
Pathogen Modeling Program (on-line)
(http://pmp.arserrc.gov/PMPOnline.aspx)
ComBase (database) (http://www.combase.cc)
ComBase Predictor (models) (http://www.combase.cc)
Seafood Spoilage Predictor
(http://www.dfu.min.dk/micro/sssp/Home/Home.aspx)
Refrigeration Index
(ijenson@mla.com.au; http://www.mla.com.au)

Predictive Microbiology
Tertiary models-Software
www.combase.cc
‣large, searchable, database of microbiological
raw data
‣still growing, users can add data
‣web-based, free access
‣integrates “Food Micromodel” and “Pathogen
Modeling Program” data, and many more
‣includes new models in “ComBase Predictor”

Predictive Microbiology
Tertiary models-Software
www.combase.cc
35,000 records on growth and survival of
pathogens and spoilage organisms
– ~28,000 records on pathogens
– ~4,000 on spoilage organisms, including
– ‘total spoilage bacteria’ (346)
– ‘bacillus spoilage bacteria’ (65)
– Brocothrix thermosphacta (741)
– enterobacteriaceae (338)
– lactic acid bacteria (701)
– Shewenella putrefasciens (57)
– “spoilage yeast” (44)

Predictive Microbiology
Tertiary models-Software

Predictive Microbiology
Tertiary models-Softwares

Predictive Microbiology
Tertiary models-Softwares

Predictive Microbiology
Tertiary models-Softwares

Predictive Microbiology
Tertiary models-Softwares

Predictive Microbiology
Tertiary models-Software
seafood spoilage and safety predictor-SSSP
www.dfu.min.dk/micro/sssp/
predicts growth of bacteria in different fresh and
lightly preserved seafoods
allows prediction of:
– rates of spoilage of seafood
– shelf life of various seafoods
– effect of fluctuating conditions
– simultaneous growth of Listeria monocytogenes (a
pathogen) and spoilage bacteria in cold-smoked
salmon

Predictive Microbiology
Tertiary models-Softwares
seafood spoilage and safety predictor-SSSP
www.dfu.min.dk/micro/sssp/
predicts growth of bacteria in different fresh and
lightly preserved seafoods
allows prediction of:
– rates of spoilage of seafood
– shelf life of various seafoods
– effect of fluctuating conditions
– simultaneous growth of Listeria monocytogenes (a
pathogen) and spoilage bacteria in cold-smoked
salmon

Predictive Microbiology
Tertiary models-Software
seafood spoilage and safety predictor-SSSP
www.dfu.min.dk/micro/sssp/

Predictive Microbiology
Tertiary models-Software
“refrigeration index”
www.mla.com.au
Australian product with regulatory approval for use
under Australian Export Meat Orders
• predicts growth of E. coli (as an indicator of
safe temperature control) from continuous
temperature history using the idea of timetemperature
functionintegration

Predictive Microbiology
Tertiary models-Softwares
“refrigeration index”
www.mla.com.au
Australian product with regulatory approval for use
under Australian Export Meat Orders
• predicts growth of E. coli (as an indicator of
safe temperature control) from continuous
temperature history using the idea of timetemperature
functionintegration

Predictive Microbiology
Tertiary models-Softwares
“refrigeration index”
www.mla.com.au
Australian product with regulatory approval for use
under Australian Export Meat Orders
• predicts growth of E. coli (as an indicator of
safe temperature control) from continuous
temperature history using the idea of timetemperature
functionintegration

Predictive Microbiology
Risk Assessment Software
“Risk Ranger”
www.foodsafetycentre.com.au/riskranger.htm
‣semi-quantitative risk assessment tool
‣simple, spreadsheet-based
‣deterministic, but similar logic to more complex
stochastic models
‣good for beginning to understand microbial food
safety risk assessment, contributions/interplay of
factors,differentiation of levels of risk

Predictive Microbiology
Risk Assessment Software
“Risk Ranger”
www.foodsafetycentre.com.au/riskranger.htm
‣semi-quantitative risk assessment tool
‣simple, spreadsheet-based
‣deterministic, but similar logic to more complex
stochastic models
‣good for beginning to understand microbial food
safety risk assessment, contributions/interplay of
factors,differentiation of levels of risk

Predictive Microbiology
Stochastic modeling-single cell behavior
‣For many years predictive microbiology deals with
the development of deterministic models based on
studies with large microbial populations
‣In “real life” however, contamination of foods occurs
at very low levels and deterministic models are not
suitable for such situations
‣Recently, the need for stochastic models which are
able to predict the effects of more “realistic”
contamination events (low microbial numbers) in food
safety was stressed (quantitative risk assessment)

Predictive Microbiology
Stochastic modeling-single cell behavior

Predictive Microbiology
Stochastic modeling-single cell behavior

WHAT IS LIFE?
ERWIN SCHRODINGER
First published 1944
What is life? The Physical Aspect of the
Living Cell.

Predictive Microbiology
Stochastic modeling-single cell behavior

Predictive Microbiology
Stochastic modeling-single cell behavior

Predictive Microbiology
Stochastic modeling-single cell behavior

Predictive Microbiology
Stochastic modeling-single cell behavior
Distribution of NaCl growth limits for Salmonella single
cells at different pH values
0.5
0.4
pH:5.0
pH:5.5
pH:7.3
Probability
0.3
0.2
0.1
0
4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0
%NaCl

pH:7.3, NaCl:4.5%, time=0

pH:7.3, NaCl:4.5%, time=2.5 days

pH:7.3, NaCl:6.5%, time=5.5 days

Predictive Microbiology
Stochastic modeling-single cell behavior
Simulation of Salmonella growth at 30 o C, NaCl=6.0%,
pH=5.5 (with lag)
5
4
Growing Fraction
Non Growing Fraction
Population
lag
Lag:
Wn
μ
N G
pseudolag:
NNG
3
Total Population
Ln cfu/cm 2
2
1
Population lag
0
lag
pseudolag
-1
0 10 20 30 40 50 60 70 80
Time (h)

Predictive Microbiology
Stochastic modeling-single cell behavior
Results from 20 simulations of Salmonella growth at 30 o C,
NaCl=6.0%, pH=5.5 (no lag)
12
9
Ln cfu
6
3
0
0 5 10 15
Time (h)

Predictive Microbiology-Risk Assessment
Case Study 1
Exposure assessment of L monocytogenes in
pasteurized milk
Production
Transportation
To Retail
Retail Storage
Domestic Storage
Consumption

Predictive Microbiology-Risk Assessment
Case Study 1
Exposure assessment of L monocytogenes in
pasteurized milk
Growth
model
Production
Transportation
To Retail
Retail Storage
Domestic Storage
Consumption
Prevalenceconcentration
Timetemperature
data
Handling
information

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Development of a model for Listeria monocytogenes
growth in pasteurized milk

Data collection
10
9
8
1.5 o C
4.0 o C
8.0 o C
12 o C
16 o C
Log10CFU ml -1
7
6
5
4
3
0 500 1000 1500
Time (h)

Fitting data to primary model (Baranyi and Roberts
1994)-Estimation of kinetic parameters
9
8
9
8
9
8
9
8
7
7
7
7
6
6
6
6
5
5
5
5
4
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
0
0
0
0 500 1000 0 500 1000 0 100 200 300 0 100 200 300
9
9
9
9
8
8
8
8
7
7
7
7
6
6
6
6
5
5
5
5
4
4
4
4
3
3
3
3
2
2
2
2
1
1
1
1
0
0 100 200 300 400
0
0 100 200 300 400
0
0 50 100 150 200
0
0 50 100 150 200
9
10
10
10
8
7
6
5
4
3
2
1
9
8
7
6
5
4
3
2
1
9
8
7
6
5
4
3
2
1
9
8
7
6
5
4
3
2
1
0
0 50 100 150 200
0
0 50 100 150 200
0
0 50 100 150
0
0 50 100 150

Fitting data to primary model (Baranyi and Roberts
1994)-Estimation of kinetic parameters
curve rate lag y0 yEnd se(fit) R^2_stat
1A 0.0046 502.4 3.70 0.225 0.973
1B 0.0035 460.9 3.80 0.133 0.985
4A 0.0121 199.3 3.79 7.74 0.172 0.991
4B 0.0110 164.0 3.60 7.77 0.164 0.992
4C 0.0111 181.2 4.03 8.07 0.202 0.987
4D 0.0083 164.9 4.11 7.68 0.146 0.991
8A 0.0267 16.4 3.50 8.26 0.146 0.992
8B 0.0252 25.2 3.66 8.29 0.106 0.995
8C 0.0250 17.1 3.59 8.46 0.161 0.992
8D 0.0257 19.0 3.80 8.37 0.129 0.995
12A 0.0476 13.3 3.68 8.35 0.126 0.994
12B 0.0476 14.3 3.74 8.25 0.113 0.995
12C 0.0502 9.3 3.60 8.31 0.214 0.984
12D 0.0478 12.8 3.89 8.83 0.160 0.992
16A 0.0773 4.0 3.57 8.51 0.176 0.991
16B 0.0804 5.9 3.71 8.56 0.161 0.992
16C 0.0829 8.8 3.44 8.23 0.173 0.992
16D 0.0832 6.0 3.30 8.15 0.178 0.992
16E 0.0967 10.2 3.60 8.45 0.063 0.999
16F 0.0973 7.8 3.63 8.62 0.100 0.998

Fitting kinetic parameters to a secondary model
0.5
0.4
μ
max
= b −
( T Tmin
)
(μmax) 0.5
0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18
Temperature ( o C)
Estimated value Lower 95% CI Upper 95% CI r 2
Parameter
b 0.024 0.023 0.025 0.988
T min ( o C) -2.32 -3.02 -1.61

Validation at dynamic temperature conditions
prediction based on the square root model for the
estimation of the “momentary” rate and the differential
equations of Baranyi and Roberts model (eq. 2 and 3),
which were numerically integrated with respect to time:
d
dt
x
=
⎛
q
⎞
q
⎜
⎝ + 1
⎠
⎝
x
x
[ b( T
t
T )] ⎟⎜
⎛ 2
( ) − ⎜
−
⎟
x
min 1
max
⎞
⎟
⎠
m
d q
=
2
min
dt
[ b
( T
( t
)
−
T
)
] q

Validation at dynamic temperature conditions
10
20
9
18
Log10 cfu/ml
8
7
6
5
4
16
14
12
10
8
6
4
Temperature ( o C)
3
2
2
0
0 30 60 90 120 150 180
Time (hours)

Validation at dynamic temperature conditions
10
20
9
18
8
16
Log10 cfu/ml
7
6
5
4
3
14
12
10
8
6
Temperature ( o C)
2
4
1
2
0
0
0 30 60 90 120 150 180 210 240
Time (hours)

Validation at dynamic temperature conditions
10
20
Log10 cfu/g
9
8
7
6
5
4
3
2
1
0
4
2
0
0 30 60 90 120 150
Time (hours)
18
16
14
12
10
8
6
Temperature ( o C)

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Collection of data on pasteurized milk chill chain
Wholesaler
Packer/Processor
Pro
duc
er
Consumer
Retailer
Hospital

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Collection of data on pasteurized milk chill chain
Collection of Time-Temperature data from
Transportation trucks
retail refrigerators
domestic refrigerators
Pro
duc
er

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Data from trucks during transportation to
retail
Transportation
time
Transportation
temperature
20
18
30
%Products
16
14
12
10
8
6
%Trucks
25
20
15
10
4
2
5
0
0 1 2 3 4 5 6 7 8 9 10 11
0
3 4 5 6 7 8 9 10 11 12
Transportation time (hours)
Temperature in truck ( o C)

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Representative temperature profiles of retail
cabinets
Θερμοκρασία o C
15,0
13,0
11,0
9,0
7,0
5,0
3,0
1,0
-1,0
-3,0
-5,0
15,0
13,0
0 50 100 150 200 250 300 350
Χρόνος (ώρες)
Θερμοκρασία o C
15,0
13,0
11,0
9,0
7,0
5,0
3,0
1,0
-1,0
-3,0
-5,0
15,0
13,0
0 50 100 150 200 250 300 350
Χρόνος (ώρες)
11,0
11,0
Θερμοκρασία o C
9,0
7,0
5,0
3,0
1,0
Θερμοκρασία o C
9,0
7,0
5,0
3,0
1,0
-1,0
-3,0
0 50 100 150 200 250
-1,0
-3,0
0 50 100 150 200 250 300 350
-5,0
-5,0
Χρόνος (ώρες)
Χρόνος (ώρες)

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Temperature distribution of retail cabinets
35
30
%Refrigerators
25
20
15
10
5
0
-2 0 2 4 6 8 10 12 14
Temperature o C

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Storage time at retail cabinets
%Products
45
40
35
30
25
20
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100
Storage time at retail (hours)

Temperature o C
Temperature o C
15
13
11
9
7
5
3
1
-1
-3
-5
Representative temperature profiles of
domestic refrigerators
Upper shelf
Middle shelf
Lower shelf
Door shelf
0 4 8 12 16 20 24
15,0
13,0
11,0
9,0
7,0
5,0
3,0
1,0
Upper shelf
Middle shelf
Lower shelf
Door shelf
Time (hours)
Temperature o C
Temperature o C
15,0
13,0
11,0
9,0
7,0
5,0
3,0
1,0
-1,0
-3,0
-5,0
15,0
13,0
11,0
9,0
7,0
5,0
3,0
1,0
Upper shelf
Middle shelf
Lower shelf
Door shelf
0 4 8 12 16 20 24
Upper shelf
Middle shelf
Lower shelf
Door shelf
Time (hours)
-1,0
-3,0
0 4 8 12 16 20 24
-1,0
-3,0
0 4 8 12 16 20 24
-5,0
-5,0
Time (hours)
Time (hours)

Temperature distribution of domestic
refrigerators
%Refrigerators
35
30
25
20
15
10
Upper shelf
Middle shelf
Lower shelf
Door shelf
5
0
0 2 4 6 8 10 12 14 16 18
Average Temperature o C

Consumer Handling information

Consumer Handling information
60
Storage time in domestic
refrigerator
50
% Consumers
40
30
20
10
0
1 2 3 4 5
Storage time of pasteurized milk in domestic refrigerator (days)

Consumer Handling information
60
50
Position of milk in domestic
refrigerator
% Consumers
40
30
20
10
0
Upper shelf Middle shelf Lower shelf Door shelf
Position of pasteurized milk in domestic refrigerator

0
0
2
1
4
2
3
6
8
0
2
4
0
6
8
1
2
3
0
1
2
3
0,40
0,35
0,30
0,25
Expon(2,5) Shift=-2,5
Exposure Assessment
0,40
Triang(-2,5; 0; 2,5)
0,20
0,35
0,15
0,30
0,10
0,25
0,05
0,20
0,00
-4
-2
90,0%
-2,37 4,99
5,0%
Contamination
at t=0
10
>
0,15
0,10
0,05
0,00
-3
-2
-1
5,0% 90,0%
5,0%
-1,709 1,709
Consumption
data
MODELS (growthinactivation-survival
Contamination at
consumption time
Normal(0; 1)
0,40
Triang(-2,5; 0; 2,5)
Monte
Carlo
0,45
0,40
0,35
0,35
0,30
0,30
0,25
0,20
0,15
0,25
0,20
0,15
Exposure
0,10
0,10
0,05
0,00
-3
-2
-1
0,05
0,00
Lognorm(2,5; 2,5) Shift=-2,5
< 5,0% 90,0%
5,0% >
-1,645 1,645
-3
-2
-1
5,0% 90,0%
5,0%
-1,709 1,709
0,40
0,35
Processing,
distribution,
storage
parameters
0,40
0,35
0,30
0,25
0,20
0,15
0,10
0,05
0,00
-4
-2
Lognorm(2,5; 2,5) Shift=-2,5
90,0%
-2,05 4,45
5,0%
10
12
>
0,30
0,25
0,20
0,15
0,10
0,05
0,00
-4
-2
0
2
4
90,0%
-2,05 4,45
6
8
5,0%
10
12
>

Concentration of L. monocytogenes in contaminated
pasteurized milk at the tine of consumption
60
56.57
50
40
%Products
30
20
18.50
10
11.55
6.18
0
3.66 1.77 0.87 0.44 0.27 0.09 0.04 0.03 0.02 0.01
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
L. monocytogenes concentration at consumption time (Log CFU/ml)

Sensitivity Analysis
Domestic Storage Time
0.515
Domestic Temperature-Door Shelf
0.307
Retail Storage Time
Retail Storage Temperature
Domestic Temperature-Upper Shelf
Position in Domestic Refrigerator (Shelf)
Domestic Temperature-Middle Shelf
Transportation Temperature
Domestic Temperature-Lower Shelf
Transportation Time
0.209
0.188
0.149
0.114
0.097
0.046
0.038
0.032
0 0.1 0.2 0.3 0.4 0.5 0.6
Corelation Coefficient

Case Study 1: Exposure assessment of Listeria
monocytogenes in pasteurized milk
Evaluation of “what if” scenarios
Int. I: better control of retail temperature,
Int. II: exclude door shelf from domestic storage,
Int. III: Int. II+ decrease domestic mean
temperature by 2 oC

Evaluation of “what if” scenarios
20
18
Int. III
81.9
16
Int. II
67.6
14
Int. I
61.1
%Products
12
10
8
6
4
2
0
Current situation
Current situation
Int. I
Int. II
Int. III
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Total growth of L. monocytogenes (Log CFU/ml) during distribution,
retail and domestic storage
56.6
0 20 40 60 80 100
Percent of products in which L. monocytogenes will not grow

Evaluation of “what if” scenarios
Increase shelf life of pasteurized milk from 5 to
10 days
40
35
30
%Products
25
20
15
10
Current situation
Int. III
5
0
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
L. monocytogenes concentration at consumption time (Log CFU/ml)

Predictive Microbiology-Risk Assessment
Case Study 2
Risk-based approach for evaluating compliance of
RTE food to the safety criteria for L.
monocytogenes

The new EC regulation 2073/2005 on
microbiological criteria for foodstuffs
(safety criteria for L. monocytogenes)
Before January 1 st 2006 the Microbiological criterion
for L. monocytogenes was
Zero Tolerance
(EC Directive 92/46)
The problem………
Zero Tolerance was not feasible in practice

Source:Draft Assessment of the Relative Risk to Public Health from Foodborne
Listeria monocytogenes Among Selected Categories of Ready-to-Eat
Foodshttp://www.foodsafety.gov/~dms/lmrisk.html

The new EC regulation 2073/2005 on
microbiological criteria for foodstuffs
(safety criteria for L. monocytogenes)

Lm in sliced RTE meat products of the Hellenic Market
Mean Prevalence 8,1% (n=209)
Product Product Product
Tested Positive
type name category a Manufacturing company
samples samples
1 Bresaola F F 2 0
2
Turkey
breast
HT H,I,K,S,U,X,ZA 21 0
3
Smoked
tongue
HT T 1 0
4
Ham
(cooked)
HT E,H,J,K,P,S,X,ZA 27 1
5
Ham
(fermented)
F E,I,N 6 1
6 Copa F E,F 3 0
7
Chicken
breast
HT Q 4 0
8 Mortadella HT C,D,E,H,I,S,X,ZA 13 0
9 Bacon HT B,C,D,E,H,I,J,M,O,S,U,W,X,Y,Z,ZA 49 12
10 Pork loin HT C,D,H,J,S,ZA 10 0
11 Pariza HT A,C,D,H,X,ZA 14 0
12 Pastirma F T 2 0
13 Prosiuto F F,M,R 6 0
14 Salami F D,E,F,G,I,M,P,R,S,U,V,X,ZA 30 3
15
Salami
(cooked)
HT K,L,Q 6 0
16
Pork
shoulder
HT D,H,M,P,S,U,X,ZA 15 0
Total 209 17

The new EC regulation 2073/2005 on
microbiological criteria for foodstuffs
(safety criteria for L. monocytogenes)

The new EC regulation 2073/2005 on
microbiological criteria for foodstuffs
Safety Criteria for L. monocytogenes
Food category
Samplingplan
Limits
n c m M
Stage where the criterion
applies
1.2 Ready-to-eat
foods able to support
the growth of L.
monocytogenes
5 0 100 cfu/g 4 placed on the market and
Products ready to be
during their shelf-life
5 0 Absence
in 25 g 5
Before the food has left
the immediate control of
the food business operator,
who has produced it
1.3 Ready-to-eat
foods unable to
support the growth
of L. monocytogenes
5 0 100 cfu/g
Products ready to be
placed on the market and
during their shelf-life

The new EC regulation 2073/2005 on
microbiological criteria for foodstuffs
(emphasis on L. monocytogenes)
the traditional approach based on end-product
sampling and examination is no longer sufficient to
assess the compliance of RTE foods with the new
criteria
Use of alternative “tools”
Predictive Microbiology

Requirements (technical)
for the Food Industry for compliance with the new
safety criteria for LM in RTE foods
1. Prove that the products support or do not support
growth of L monocytogenes
Probabilistic (Growth/no Growth) models
If the product supports growth
2. Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Kinetic models

Requirements (technical)
1.Prove that the products support or do not support
growth of L monocytogenes
Note in the regulation
Products are automatically considered to belong to the
category of not supporting growth of Lm if
pH

Requirements (technical)
1.Prove that the products support or do not support
growth of L monocytogenes
7,00
pH-a w combinations of RTE meat
products in the Hellenic market
6,50
6,00
pH
5,50
5,00
4,50
4,00
0,890 0,910 0,930 0,950 0,970 0,990
Only 6.1% of the products are automatically considered to belong
to the category of not supporting growth of Lm
a w

Requirements (technical)
1.Prove that the products support or do not support
growth of L monocytogenes
Probabilistic (Growth/no Growth) models
7,00
4 o C
6,50
Support Growth
6,00
pH
5,50
5,00
4,50
Do not Support Growth
Growth boundary (P=0.5)
4,00
0,920 0,930 0,940 0,950 0,960 0,970 0,980 0,990
a w
Model:Koutsoumanis et al., 2004, Int. J. Food Micro.

Requirements (technical)
1.Prove that the products support or do not support
growth of L monocytogenes
Probabilistic (Growth/no Growth) models
7,00
15 o C
10 o C
4 o C
6,50
6,00
pH
5,50
5,00
4,50
Growth boundary (P=0.5)
4,00
0,920 0,930 0,940 0,950 0,960 0,970 0,980 0,990
Model:Koutsoumanis et al., 2004
a w

Requirements (technical)
1.Prove that the products support or do not support
growth of L monocytogenes
At which temperature ?

Requirements (technical)
2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Before the food has left
the immediate control of the
food business operator, who
has produced it
Products ready to be placed
on the market and during
their shelf-life
Absence in 25 g or

2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Use of available predictive modeling software (i.e PMP)

2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Use of available predictive modeling software (i.e PMP)
11.0
Listeria monocytogenes
11.0
Listeria monocytogenes
10.0
9.0
4 o C
10.0
9.0
10 o C
8.0
8.0
7.0
7.0
log(CFU/ml)
6.0
5.0
4.0
3.0
2.0
1.0
25 days for 3 logs
increase
log(CFU/ml)
6.0
5.0
4.0
3.0
2.0
1.0
8 days for 3 logs
increase
0.0
0 5 10 15 20 25 30 35 40 45 50
Time (Days)
0.0
0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25
Time (Days)

2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
At which temperature ?
General Requirements in Reg 2073/2005
the food safety criteria applicable
throughout the shelf-life of the products
can be met under reasonably,foreseeable
conditions of distribution, storage and use.

Chill Chain Conditions
Temperature distribution in retail refrigerators
(survey in Greece)
35
30
% Refrigerators
25
20
15
10
5
0
0 2 4 6 8 10 12 14
Temperature o C

Predictive Microbiology tools at Production level
A probabilistic modelling approach for evaluating
the compliance of RTE foods with the new safety
criteria for Listeria monocytogenes
The approach is based on the combined use of: a)
growth/no growth boundary models, b) kinetic growth
models, c) data on product characteristics (pH, aw, shelflife)
and d) storage temperature data
A probabilistic analysis of the above components using
Monte Carlo simulation, can lead to a more realistic
estimation of the behavior of L. monocytogenes which can
be further used as a decision-making tool regarding the
design or modification of a product’s formulation or its
“use-by-date” in order to ensure its compliance

Requirements (technical)
1.Prove that the products support or do not support
growth of L monocytogenes
Probabilistic approach
G/NG
model
Product
Characteristics
(pH, a w …)
Monte Carlo
simulation
Probability of growth
Chill chain
characteristics
1
0,9
Cumulative probability
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
0 0,2 0,4 0,6 0,8 1 1,2
Probability of growth for L. monocytogenes
% Refrigerators
35
30
25
20
15
10
5
0
0 2 4 6 8 10 12 14
Temperature o C

Requirements (technical)
1.Prove that the product supports or do not support
growth of L monocytogenes
Case study
Product: cooked ham (pH=5.49, a w =0.943)
1
1
Cumulative probability
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
Binomial (1, Pg)
probability
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
No Growth
0,6143
Growth
0,3857
0,1
0,1
0
0 0,2 0,4 0,6 0,8 1 1,2
Probability of growth for L. monocytogenes
0
0 1
Growth/No growth for L. monocytogenes

X

Model used
Combination of G/NG and kinetic models
0,1
0,08
G/NG boundary model
Koutsoumanis et al., 2004
Kinetic model
Buchanan., 2001
0,06
mmax
0,04
0,02
Probability of growth
1,2
1
0,8
0,6
0,4
0,2
0
4 6 8
mmax=
Normal (m, sePred)
0
0 2 4 6 8 10 12 14 16 18 20
μ
max
Pg =
=
Temperature ( o C)
fT ( ),Ross et al., 2003
Buchanan, 2001
g( T), Presser et al., 1998
Primary model: Baranyi, μ
a: Binomial (1, Pg)
Koutsoumanis et al., 2004
μ: Normal ( ˆ μ , sepred)
max
max
= a×
μ
max

Requirements (technical)
2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Case study
Product: cooked ham (pH=5.49, a w =0.943, N=50ppm, SL=60 days)
Distribution of L. monocytogenes
at the end of shelf life in retail
0,25
0,2
In 28% of products Lm will
exceed 100 cfu/g
Probability
0,15
0,1
0,05
0
-2 -1 0 1 2 3 4 5 6 7 8 9 10
L. monocytogenes at the end of retail storage (Log cfu/g)

Requirements (technical)
2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Potential actions for achieving the desired level of
compliance to the Safety criteria
‣ Adjust the shelf life of the product
‣ Modify the formulation of the product

Requirements (technical)
2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Potential actions for meeting the Safety criteria
Adjust the shelf life of the product
Probability
0,6
0,5
0,4
0,3
0,2
Shelf Life 60 Days Compliance in 72th percentile
Shelf Life 43 days:Compliance in 90th percentile
Shelf Life 35 Days:Compliance in 95th percentile
Shelf Life 25 Days:Compliance in 99th percentile
0,1
0
-2 -1 0 1 2 3 4 5 6 7 8 9 10
L. monocytogenes at the end of retail storage (Log cfu/g)

Requirements (technical)
2.Prove that the concentration of the pathogen will
not exceed 100 cfu/g at the end of shelf life in retail
Cooked ham
Potential actions for meeting the Safety criteria
Modify the formulation of the product
a w 0.943 0.935
N 50ppm 80ppm
0,45
0,4
0,35
0,3
Compliance in 72th percentile
Compliance in 95th percentile
a w
-0.848
Tornado graph
Regression Sensitivity for Self-life
Probability
0,25
0,2
0,15
pH
-0.458
0,1
Nitrites
0.269
0,05
-1 -0.5 0 0.5 1
Std b Coefficients
0
-2 -1 0 1 2 3 4 5 6 7 8 9 10
L. monocytogenes at the end of retail storage (Log cfu/g)

Using the probabilistic approach for both safety
and spoilage
X

Using the probabilistic approach for both safety
and spoilage
Case study
Product: cooked ham
pH=5.49, a w =0.943, N=50ppm, SL=60 days
SSO: L. sake
Spoilage level 10 7 cfu/g

Using the probabilistic approach for both safety
and spoilage
Case study
Product: cooked ham SL 60 days
poilage
10
compliant/spoiled
non-compliant/spoiled
Level
LAB at end of SL (log cfu/g)
9
8
7
6
5
4
compliant/unspoiled
3
non-compliant/unspoiled
2
-2 -1 0 1 2 3 4 5 6 7
L. monocytogenes at the end of SL (log cfu/g)
Safety criterion

Using the probabilistic approach for both safety
and spoilage
Case study
SL 60 days
Product: cooked ham
SL 25 days
10
compliant/spoiled
non-compliant/spoiled
10
compliant/spoiled
non-compliant/spoiled
9
9
LAB at end of SL (log cfu/g)
8
7
6
5
4
compliant/unspoiled
LAB at end of SL (log cfu/g)
8
7
6
5
4
compliant/unspoiled
3
2
-2 -1 0 1 2 3 4 5 6 7
L. monocytogenes at the end of SL (log cfu/g)
72% compliance
non-compliant/unspoiled
17% spoiled before end
of Shelf life
3
2
-2 -1 0 1 2 3 4 5 6 7
L. monocytogenes at the end of SL (log cfu/g)
99% compliance
non-compliant/unspoiled
0.1% spoiled before end
of Shelf life

Acknowledgement
EU Framework VI programme on Food Quality and Safety,
‘ProSafeBeef’ Food-CT-2006-36241
EU Framework V programme on “Quality of Life and
Management of Living Resources”, Key Action 1-Health
Food and Environment,”SMAS” QLK1-CT2002-02545
“Development and Application of Microbial Time
Temperature Indicators for Monitoring Food Quality &
Safety” Greek Government & Commission of the European
Community, PENED 2003

Predictive food microbiology as a tool in
risk assessment
Kostas Koutsoumanis
Lab of Food Microbiology and Hygiene
Dpt. Of Food Science and Technology
Aristotle University of Thessaloniki,
Contact: kkoutsou@agro.auth.gr