University of Nebraska - Lincoln

DigitalCommons@University of Nebraska - Lincoln

Dissertations, Theses, & Student Research in Food
Food Science and Technology Department
Science and Technology

12-2016

Risk Assessment and Research Synthesis

methodologies in food safety: two effective tools to
provide scientific evidence into the Decision
Making Process.
Juan E. Ortuzar
University of Nebraska-Lincoln, [email protected]

Follow this and additional works at: http://digitalcommons.unl.edu/foodscidiss

Part of the Food Microbiology Commons

Ortuzar, Juan E., "Risk Assessment and Research Synthesis methodologies in food safety: two effective tools to provide scientific
evidence into the Decision Making Process." (2016). Dissertations, Theses, & Student Research in Food Science and Technology. 79.
http://digitalcommons.unl.edu/foodscidiss/79

This Article is brought to you for free and open access by the Food Science and Technology Department at DigitalCommons@University of Nebraska -
Lincoln. It has been accepted for inclusion in Dissertations, Theses, & Student Research in Food Science and Technology by an authorized
administrator of DigitalCommons@University of Nebraska - Lincoln.
RISK ASSESSMENT AND RESEARCH SYNTHESIS METHODOLOGIES IN FOOD

SAFETY: TWO EFFECTIVE TOOLS TO PROVIDE SCIENTIFIC EVIDENCE INTO

THE DECISION MAKING PROCESS

by

Juan E. Ortuzar

A THESIS

Presented to the Faculty of

The Graduate College at the University of Nebraska

In Partial Fulfillment of Requirements

For the Degree of Master of Science

Major: Food Science and Technology

Under the Supervision of Professors Bing Wang and John Rupnow

Lincoln, Nebraska

December, 2016
 RISK ASSESSMENT AND RESEARCH SYNTHESIS METHODOLOGIES IN

FOOD SAFETY: TWO EFFECTIVE TOOLS TO PROVIDE SCIENTIFIC

EVIDENCE INTO THE DECISION MAKING PROCESS

Juan Eduardo Ortúzar, MSc.

University of Nebraska, 2016

Advisors: Bing Wang and John Rupnow

The food supply chain is a complex and diverse system. Some food products need

minimum processing to reach the consumers, while others involve several different

processes, countries and suppliers, can take several months to be on the table of the end

consumer. Regarding food safety, the public health of consumers is at stake and the

consequences of outbreaks could prove disastrous. This has been recognized as a matter

of global importance for the food industry and authorities around the world since several

efforts to improve quality, safety and trade of food have arisen since the early 1960s. The

birth of the Codex Alimentarius Commission, a joint organism lead by the World Health

Organization and the Food and Agriculture Organization, marks a milestone, creating the

first organization dedicated to proposing international food safety standards and to foster

fair global food trade.

All these organizations agree that the use of solid scientific evidence in the decision

making process is the cornerstone in creating a safe global food supply chain. Although

there is widespread consensus about this, developing countries usually encounter heavy

difficulties in accomplishing these objectives due to obstacles such as low funding to

sample their food products, a weak regulatory system, insufficient technology and

scientific capabilities. Therefore, addressing the question “how can we provide tools for

these countries to strengthen their capacities to create scientific evidence based

regulations with the consideration of these limitations?” is in great need. In this project

two case studies were used to show that risk assessment, in conjunction with the use of

research synthesis methodologies, are two approaches that can be used by the food

industry and governments to provide effective scientific insights into their respective

decision making processes. The focus of this research project is food safety in Chile, thus

the analysis, results and overall direction will be narrowed to the perspective of this

developing country.
 iii

ACKNOWLEDGEMENTS

In first place, this experience would not have been possible if Dr. Bing Wang and John

Rupnow have not believed in me. For that I am really grateful. It was an honor to have

worked together. Thanks to my Committee Members, Dr. Jennifer Clarke, George Gray

and Rolando Flores; they were wonderful mentors always eager to answer my questions

and guide me throughout this project.

My family were always supporting me in every possible aspect of this adventure in the

United States. Having such a loving, happy and united family has created a mark in my

personality and soul, which I carry with great pride anywhere I go. Thank you, for you

are the reason I am happy in life and able to reflect that on others.

Living abroad can be sometimes hard for many reasons. But I want to acknowledge that

the people I met during this two years living in Lincoln have made this journey incredibly

enjoyable and rewarding. The American friends I made here, John, Michaela, Michael,

John Jimma, Josh, Brian, Becca, Kim and Emika, among many, gave me the best

impression of this beautiful country and its people. Keep being so unique, smart and

loving. I hope our paths cross again and I will call you my friends forever!

Thanks to all my coworkers, Anand, Ece, Yangyunna, Sapna and Bin. All the best in your

academic and personal lives. I am sure every single one of you will turn into incredible

scientists!

I can’t close this dedication without naming the most amazing group of individuals that

gathered in Lincoln, Nebraska. The almighty Cabroncos! With an unquenchable thirst for

music (country and heavy metal), adventures and paellas, this group were my first close
 iv

friends. All the trips together, the fun times, the culinary meetings and the

scientific/philosophical conversations we had made my stay here so awesome. Carolina,

Denis, Ivo, Felipe, Isabel, Michael, K, Hugo and of course, our beloved leader, Isidro

Campos, thank you all for being such wonderful people and being my friends.

To everyone that somehow affected my life in a positive, which I am proud to say it has

been every single person here in the U.S., thanks.

And always remember;

“Don’t worry about a thing, ‘cause every little thing gonna be alright”

Bob Marley
 v

TABLE OF CONTENTS

ABSTRACT ....................................................................................................................... i

ACKNOWLEDGEMENTS ............................................................................................. iii

TABLE OF CONTENTS .................................................................................................. v

LIST OF TABLES ........................................................................................................... ix

LIST OF FIGURES ......................................................................................................... xi

CHAPTER 1: FOOD SAFETY, TRADE, AND THE

NEED FOR SCIENCE IN POLICY ........................................................................... 13

I. INTRODUCTION ....................................................................................................... 13

A. A brief history of food safety ..................................................................................... 13

B. About this study ......................................................................................................... 13

II. FOOD SAFETY – SELECTED NATIONS COMPARISON ................................... 16

A. United States .............................................................................................................. 16

B. European Union.......................................................................................................... 18

C. Chile ........................................................................................................................... 20

D. International Organizations ........................................................................................ 25

E. Conclusion .................................................................................................................. 26
 vi

III. SCIENCE INTO THE DECISION MAKING PROCESS A GLOBAL REVIEW . 27

A. United States .............................................................................................................. 27

B. European Union.......................................................................................................... 30

C. Chile .......................................................................................................................... .30

D. International Organizations ........................................................................................ 31

IV. TECHNOLOGIES AND TOOLS IN FOOD SAFETY ........................................... 32

A. Food Safety Risk Assessment .................................................................................... 32

B. Research Synthesis Methodologies ............................................................................ 34

C. Meta-analysis ............................................................................................................. 35

D. Others ......................................................................................................................... 36

V. CONCLUSIONS ........................................................................................................ 36

VI. REFERENCES ......................................................................................................... 37

CHAPTER 2: RISK ASSESSMENT COLLABORATION PROJECT .................. 40

I. ABSTRACT ................................................................................................................ 40

II. INTRODUCTION ...................................................................................................... 41

A. Risk Assessment Collaborative Project ..................................................................... 42

B. Raspberries Official Control Program (ROCP) ......................................................... 42

C. Need of risk assessment in ROCP .............................................................................. 45

D. Specific Objectives of the Risk Assessment project .................................................. 45

III. MATERIALS AND METHODS .............................................................................. 46

A. On-farm, collection center and packaging practices survey ...................................... 48

B. Build raspberry supply chain flow chart .................................................................... 49

 vii

C. Justification to exclude data points ............................................................................ 50

D. Collect data from literature and survey ...................................................................... 51

E. Fill data gaps with expert elicitation .......................................................................... 53

F. Build quantitative risk assessment model ................................................................... 53

G. Pre-harvest module (Farm) ........................................................................................ 59

H. Cross contamination module – Harvest (Farm) ......................................................... 59

I. Growth model (Farm, Collection Center and Packing Plant) ...................................... 60

J. Cross contamination module – Handling (Packing Plant) .......................................... 62

K. Run Monte Carlo simulation ...................................................................................... 63

L. Scenario analysis ........................................................................................................ 63

IV. RESULTS AND DISCUSSION ............................................................................... 67

A. Risk estimates of current practices............................................................................. 67

B. Expert elicitation ........................................................................................................ 69

C. Sensitivity analysis ..................................................................................................... 69

D. Scenario analysis and interventions evaluation ......................................................... 75

E. Data gaps identification .............................................................................................. 77

F. Significance for regulators and evidence-based policies ............................................ 78

V. CONCLUSIONS ........................................................................................................ 79

VI. REFERENCES ......................................................................................................... 80

CHAPTER 3: SYSTEMATIC REVIEW.................................................................... 83

I. ABSTRACT ................................................................................................................ 83
 viii

II. INTRODUCTION ...................................................................................................... 84

III. MATERIALS AND METHODS .............................................................................. 87

A. Search Strategy .......................................................................................................... 87

B. Relevance screening ................................................................................................... 89

i. Title screening.............................................................................................................. 90

ii. Abstract screening ...................................................................................................... 90

iii. Full-text screening ..................................................................................................... 90

C. Data extraction ........................................................................................................... 91

D. Standardization of milk supply steps ......................................................................... 91

E. Definitions .................................................................................................................. 92

F. Data analysis ............................................................................................................... 93

IV. RESULTS AND DISCUSSION ............................................................................... 94

A. Systematic review process ......................................................................................... 94

B. Characteristics of the relevant citations and extracted data ....................................... 98

C. Concentration of spore-forming bacteria along the milk supply chain .................... 103

D. Prevalence of spore-forming bacteria along the milk supply chain ......................... 105

E. Meta-analysis for prevalence .................................................................................... 111

F. Meta-regression analysis .......................................................................................... 114

G. Significance for regulators and evidence-based policies ......................................... 115

V. CONCLUSIONS ...................................................................................................... 116

VI. REFERENCES ....................................................................................................... 118

SUMMARY AND FUTURE DIRECTION .............................................................. 124

ANNEXES ................................................................................................................... 126

 ix

LIST OF TABLES
CHAPTER 1
Table 1.1 Largest economic activities and its contribution to the GDP in
Chile for 2011……….…………………………………………...21
Table 1.2 Foodborne illnesses comparison chart from selected countries.....23

CHAPTER 2
Table 2.1 Activity list agreed on the project profile………………………..47
Table 2.2 List of parameters, values and distributions used in the farm module
for both fresh and frozen raspberries……………………………..54
Table 2.3 List of parameters, values and distributions used in the collection
center module for fresh raspberries……………………………….56
Table 2.4. List of parameters, values and distributions used in the collection
center module for frozen raspberries……………………………..56
Table 2.5 List of parameters, values and distributions used in the packing
plant module for fresh
raspberries………………………………………………………..57
Table 2.6 List of parameters, values and distributions used in the packing
plant module for frozen
raspberries……………………………..…………………………58
Table 2.7 Bacterial growth model parameters for temperatures over
8°C………………………………………………………………..61
Table 2.8 Growth model parameters for temperatures between 0°C and
8°C………………………………………………………………..61
Table 2.9 Growth model parameters for temperatures below
0°C……………………………………………………………………...62
Table 2.10 Water uses scenario analysis for bacterial
contamination…………………………………………………….65
Table 2.11 Transportation reduction time scenarios for bacterial
contamination…………………………………………………….66
Table 2.12 Temperature reduction at Collection Center scenario
results…………………………………………………………….66
Table 2.13 Pesticide applications time scenarios…………………………….66
 x

Table 2.14 Summary of the scenario analysis results………………………..76

CHAPTER 3
Table 3.1 Summary of the search strategies for the electronic
databases…………………………………………………….…..89
Table 3.2 Summary of the main characteristics of the citations that were
included in the data extraction process………………………….96
Table 3.3 Summary statistics of concentration data by Standardized Supply
Chain............................................................................................100
Table 3.4 Summary statistics of prevalence data by Standardized Supply
Chain………….………………………………………………...101
Table 3.5 Concentration pooled estimates for each processing
step……………………………………………………………...102
Table 3.6 Prevalence estimates pooled by random effect meta-analysis model
for each supply chain step…………………….………………..111
 xi

LIST OF FIGURES

CHAPTER 1

Figure 1.1 Important milestones in food safety in the United

States……………………………………………….……….…..15
Figure 1.2 The U.S. agricultural exports evolution
from 2000-2015.………………………………………………..17
Figure 1.3 The U.S. agricultural imports evolution
from 2000-2015……………………………….……….……….17
Figure 1.4 Evolution of agri-food related imports and exports in the
EU………………………………………………………..……..19
Figure 1.5 Interactive graph showing Chile’s largest trading partners in 2014,
in terms of exports……………….………………………….….21
Figure 1.6 Interactive graph showing Chile’s exports to the US in
2014……………………………………………………….……22
Figure 1.7 Imports of the EU from Chile in 2014…………………………22
Figure 1.8 Diagram of the food safety management
system in Chile…………………………………….….....……..25
Figure 1.9 Comparison chart between Systematic Review and Literature
Reviews………………………………………………….……..35

CHAPTER 2

Figure 2.1 Political map of Chile showing the region numbers……………44

Figure 2.2 Flow chart of the processing steps for raspberries, potential hazards
and reduction steps at the packing………………………..…….52
Figure 2.3 Bacterial Log CFU/gr contamination distribution of 10,000
iterations simulation for the fresh raspberry model………….....68
Figure 2.4 Viral log PDU/gr contamination distribution of 10.000 iterations
simulation for the fresh raspberry model……………………….68
 xii

Figure 2.5 Bacterial log CFU/gr contamination distribution of 10.000

iterations simulation for the frozen raspberry model....………69
Figure 2.6 Tornado plot for the final bacterial concentration for the fresh
raspberry model……………………………………………….72
Figure 2.7 Tornado plot for the final bacterial concentration for the frozen

raspberry model………………………………………………..73

Figure 2.8 Tornado plot for the final viral concentration for the fresh

raspberry model………………………………………………..74

CHAPTER 3

Figure 3.1 Flow chart of the standardized milk supply steps, with their
coverage of samples described in the retrieved articles…..…..92
Figure 3.2 Model selection procedure……………………..……………..94
Figure 3.3 Process flow of studies being retrieved, screened, appraised,
selected, data-extracted in this systematic review and meta-
analysis………………………………………………….…….99
Figure 3.4 Stacked box plot for the concentration of spore-forming bacteria
throughout the milk processing chain……...…………………103
Figure 3.5 Stacked box plot for prevalence of spore-forming bacteria
throughout the milk processing chain………………………..105
Figure 3.6a Plot of the trends for individual studies reporting prevalence in
cells…………………………………………………………..107
Figure 3.6b Plot of the trends for individual studies reporting concentrations in
cells………………..…………………………………………107
Figure 3.6c Plot of the trends for individual studies reporting spore
prevalence…………………………………………………….108
Figure 3.6d Plot of the trends for individual studies reporting spore
concentration………………………………………………….109
Figure 3.7 Forest plot of reported cell prevalence………………………..112
Figure 3.8 Forest plot of reported spore prevalence……………………..113
 13

CHAPTER 1: FOOD SAFETY, TRADE, AND THE NEED FOR SCIENCE IN

POLICY

I. INTRODUCTION

A. A brief history of food safety

Food has played a pivotal role in the development of mankind, in both the nutritional and

cultural dimensions. Food safety practices can be tracked to prehistoric times, starting

with the Chinese that developed the first preservation methods for vegetables in 4000 BC

(Uemura and Bari, 2016), which provided them means to attain higher levels of food

safety. As eating patterns and foods changed and evolved over time, food safety laws

started to appear (Uemura and Bari, 2016).

The first food laws can be seen in the book of Leviticus around 2000 BC and in the

Quran by 570 AC (Hutt and Hutt, 1984). Although these were targeting food adulteration,

as with food preservation, the population indirectly received the first benefits of food

safety practices. In spite of food safety being a very old subject which almost every early

civilization was addressing to some extent, it was not until the 19th century that

comprehensive food legislations were adopted (Uemura and Bari, 2016). Figure 1.1 uses

the United States (U.S.) as an example to show important milestones in the history related

to food safety, starting from the late 1800’s.

B. About this study

The purpose of this study is to demonstrate how the two systematic approaches, risk

assessment and research synthesis methodologies, can be utilized by food industry and

regulatory authorities to provide effective scientific insights to inform the process of

 14

designing intervention strategies, regulations, policies or laws. In this chapter, the

foundation of why we need science in the decision making process is going to be

explained, from the perspective of domestic food safety protection and international

trade. The whole work will revolve around how Chile – a developing country in terms of

the International Monetary Fund (IMF, 2016) – can harness the potential of these two

tools and include them in their decision making process for food safety matters. This will

be achieved by comparing the epidemiological status, current trade situation and use of

science by the two most powerful actors in food trade, the United States of America and

the European Union (EU), compared to the Chilean reality. These nations were selected

since they are the leaders in food safety sciences and technologies, employ more

advanced regulatory frameworks and, as we will see later, are the most important trading

partners to Chile. The epidemiological status was surveyed to have a broad understanding

of the range of deaths and illnesses caused by food in each nation. The integration of

science into the decision making process is something that in developing countries is hard

to achieve. Thus, it is important to have a look in developed countries and understand

how they are achieving this. Finally, a description is given of the tools that this thesis is

proposing should be used in order to achieve the food safety protection objective.
 15

Figure 1.1 Important milestones in food safety in the United States (adopted from Reneé
Johnson, 2014)
 16

II. FOOD SAFETY – SELECTED NATIONS COMPARISON

A. United States

i. Economics and food trade

According to 2016 estimates by the IMF, the United States (US) Gross Domestic Product

(GDP) is US$ 18,562 billion, making the U.S. the second largest economy in the world

after China. Its GDP per capita is US$ 56,084 ranking 11 worldwide (IMF, 2016).

The United States Department of Agriculture (USDA) has predicted for 2017 an export

forecast for agricultural trade of US$ 133.0 billion and imports of US$ 113.5 billion,

worth 1.327% of the total GDP. As shown in Figure 1.2 and Figure 1.3, a considerable

portion of trade corresponds to human food.

The total local retail and food services sales for 2015 were US$ 1,511 billion, which is

worth 8.14% of the total GDP. (Economic Research Service, 2016)

ii. Food safety epidemiology situation

The Center for Disease Control and Prevention (CDC) has estimated that yearly, 48

million people get sick, 128,000 are hospitalized and 3,000 die due to foodborne illnesses

(CDC, 2016). However, these numbers are underestimated due to the surveillance

methods used, under-diagnosis because of variations in medical care seeking, specimen

submission, laboratory testing and sensitivity (CDC, 2016).

 17

Figure 1.2 The U.S. agricultural exports evolution from 2000-2015. Data retrieved from
the Economic Research Service.

Figure 1.3. The U.S. agricultural imports evolution from 2000-2015. Data retrieved from
the Economic Research Service.
 18

iii. Regulatory framework

Food safety responsibilities are divided among several different agencies in the U.S. The

USDA and Food and Drug Administration (FDA) have direct enforcing and regulation

power over different sets of foods, while the CDC is the supporting agency that collects

data on foodborne illnesses and supports foodborne disease surveillance and response

(Foodsafety.gov, 2016)

Meat, poultry and egg products are under the jurisdiction of USDA, through its Food

Safety and Inspection Service (FSIS). Any other types of food are regulated by FDA.

B. European Union

i. Economics and trade

The European Union is a political and economic union of 28 member states. If treated as

a single country, according to the IMF for 2016, its GDP is US$ 16.673 billion, ranking

the third largest economy of the world.

According to the Agricultural and Rural Development Department of the European

Commission, for 2015, the agri-food exports ascended to US$ 129 billion, while imports

were worth US$ 113 billion. This is equal to 1.45% of the total GDP. As shown in Figure

1.4, an important part of the exports and imports correspond to human food.

ii. Food safety epidemiology situation

The World Health Organization has estimated that the number of foodborne illnesses in

the EU is approximately 2,431 cases per 100,000 persons and the number of deaths is 0.4

in every 100,000 (WHO, 2010). Taking into consideration the EU population of about
 19

508 million (European Union, 2016), the cases calculated are lower in foodborne adverse

outcomes compared to the U.S.: 12 million illnesses and 2,000 deaths.

Figure 1.4. Evolution of agri-food related imports and exports in the EU. It is important
to note that the “commodities” class includes live livestock and some other non-edible
items. (Agricultural and Rural Development, 2016)

iii. Regulatory framework

Each member state is allowed to have its own food safety agencies, research and outreach

efforts. There is, nonetheless, a general guideline called “The General Food Law”. Under

Regulation (EC) No 178/2002, the General Food Law is defined as “the foundation of

food and feed law. It sets outs an overarching and coherent framework for the

development of food and feed legislation both at Union and national levels. To this end, it

lays down general principles, requirements and procedures that underpin decision making

in matters of food and feed safety, covering all stages of food and feed production and

distribution.” (European Commission, 2016)

 20

This regulation also creates the European Food Safety Authority (EFSA), which is an

independent agency that provides scientific advice and support to member states

regarding food safety and public health matters. It is important to highlight that EFSA

does not enforce food safety, which is still a responsibility of each member state.

C. Chile

i. Economics and trade

According to the IMF, for 2016 the GDP for Chile is of US$ 422,422 billion. With a

population of about 18 million, the GDP per capita is about US$ 23,507.

Table 1.1 shows the main economic activities of Chile and its corresponding share of the

GDP. With a total of US$ 5.749 million, Agriculture and forestry exports make 6.43% of

the exports. In particular, US$ 4.738 million correspond to fruit exports and the rest to

other agri-food related items (Chilean Central Bank data for 2011). Figures 1.5, 1.6 and

1.7 shows worldwide trade, to the U.S., and to the EU, respectively.

ii. Food safety epidemiology situation

The latest epidemiology report from the Health Ministry in Chile indicated that in 2015,

there were 5,901 diagnosed foodborne illnesses and 119 hospitalizations (Chilean

Ministry of Health, 2015). The number of deaths attributable to foodborne illnesses is not

available. The estimated number of illnesses is also not available.

 21

Table 1.1. Largest economic activities and its contribution to the GDP in Chile for 2011.
Agriculture is showed as it is the class that includes agri-food related items.
Economic Activity Percentage of the GDP

Mining 15.2%

Business Services 13%

Manufacturing industry 10.9%

Personal services (health, education, others) 10.6%

Retail 7.9%

Agriculture and forestry 2.8%

Remaining activities 39.6%

Figure 1.5. Interactive graph showing Chile’s largest trading partners in 2014, in terms of
exports. Data taken from the Observatory of Economic Complexity from the
Massachusetts Institute of Technology.
 22

Figure 1.6. Interactive graph showing Chile’s exports to the US in 2014. Data taken from
the Observatory of Economic Complexity from the Massachusetts Institute of
Technology.

Figure 1.7. Imports of the EU from Chile in 2014. Data taken from Eurostat webpage.
 23

Proposal of an estimated number of illnesses in Chile, based on diagnosed cases.

As a foodborne illness estimate is missing in Chile’s statistics, data from CDC was

extracted and adjusted to Chile’s numbers with the purpose of doing a comparison. Table

1.2 shows the comparisons between the nations of interest based on the population.

Table 1.2. Foodborne illnesses comparison chart from selected countries.

Total Yearly estimated illnesses Yearly estimated deaths (% of total

Country
population (% of total population) population)

United States 324,099,593a 48,000,000 (14%) 3,000 (0.0009%)

European Union 510,056,011b 12,000,000 (2.35%) 2,000 (0.0004%)

Chile 18,006,407c 1,513,800 (11.89%)d No data

a. United States Census Bureau. Retrieved on October 13, 2016

b. Eurostat – Population on 1 January 2016". European Commission. Retrieved on October 13, 2016.

c. Chilean National Statistics Institute. Retrieved on October 13, 2016.

d. Derived in this study based on CDC’s adjustment factor.

The latest CDC report on foodborne illnesses indicated that the number of diagnosed

foodborne illnesses for 2006 was 142,481. Scallan et al 2011 proposed an estimate of

37,220,098 foodborne illness cases, based on the number of diagnosed cases. Therefore,

with the latest technology and science available, there is a 261.2 factor difference

between the foodborne illnesses estimate and the number of actual diagnosed cases. This

factor was used to estimate Chilean foodborne illnesses based on the number of

diagnosed cases. Caution should be used when using this number as there are several

differences in laboratory technology, scientific capacities, pathogen prevalence and

 24

dietary differences between the U.S. and Chile that makes this estimate only valid when

looking at the data from a general perspective.

iii. Regulatory framework

The only agency that enforces food safety in Chile is the Ministry of Health, through the

SEREMIS (Regional Health Secretariat), which are regional independent secretariats

with full legal power. Nonetheless, other agencies have compliance authority - but they

cannot recall a food product. Figure 1.8 indicates the organization of this multi-sectorial

management of food safety in Chile.

The Chilean Food Safety and Quality Agency (ACHIPIA) is a scientific advice and

support agency, created with the model of EFSA in mind. The main difference is that

ACHIPIA gives integral scientific support to the three agencies involved in food safety:

the Service for Livestock and Agriculture (SAG) and the National Service of Fisheries

(SERNAPESCA) and SEREMIS, instead of to member states.

 25

Figure 1.8. Diagram of the food safety management system in Chile. Courtesy of the
Chilean Agency for Food Safety and Quality.

D. International Organizations

There are some international organizations that are worth mentioning mainly because of

their significant impact on the development of standardized food safety standards,

epidemiologic data generation, education and scientific integration into regulatory issues.

i. WHO and FAO

The World Health Organization (WHO) and the Food and Agriculture Organization

(FAO) are both entities from the United Nations. Although their missions are different,

they share a common goal in terms of food safety. That is the reason why, even though

the WHO and the FAO have their own food safety capacity building, outreach and

support teams, they co-manage the Codex Alimentarius Commission (CAC), which is a

food standards creation program. In the CAC sessions, all member nations participate and
 26

scientific evidence is taken with high regard, to promote fair international trade and safe

food.

ii. ILSI

The International Life Sciences Institute (ILSI) is a nonprofit scientific organization

whose mission is: “to provide science that improves human health and well-being and

safeguards the environment” (ILSI, 2016). ILSI advocates for better and transparent

scientific advice in topics such as food and environment. It has stable funding sources,

which are mainly agri-food related industries.

E. Conclusion

This section introduced the three actors in this Chapter from a trade, food safety and

regulatory perspective. The US and the EU are the most important trade partners along

with China for Chilean agri-food exports. It is essential to understand how they manage

their food safety issues and what their current epidemiological situation is.

“As previously noted, not everything is run by the government. Instead, key international

actors such as the FAO, WHO, and ILSI contribute to the harmonization of food safety

standards, placing great efforts on ensuring a safe food supply while simultaneously

promoting fair global food trade.

This section is fundamental to understand the key players in food safety around the world

and to understand the structure of this thesis. The next section explains how these

recently introduced countries and organization take into consideration the scientific

support in their decision making process and regulation design.

 27

III. SCIENCE INTO THE DECISION MAKING PROCESS: A GLOBAL

REVIEW

A. United States

i. National Academy of Sciences

The National Academy of Sciences (NAS) is a private, non-profit society, founded in

1863 with its mission of “providing independent, objective advice to the nation on

matters related to science and technology” (NAS, 2016). Any governmental departments

can call upon the NAS for scientific advice. More than 6,000 experts have served in

different policy studies and reports, on matters of critical importance to the society.

The NAS is constantly collaborating with the Government in order to provide the best

independent scientific advice that would ultimately be used in the design of public

policies. An example of such is the request of the US Congress on November 22, 2015 to

the NAS to create a Forensic Sciences Committee, with the objectives of, among others:

(National Criminal Justice Reference Service, 2015)

(1) Assess the present and future resource needs of the forensic science community, to

include State and local crime labs, medical examiners, and coroners;

(2) Make recommendations for maximizing the use of forensic technologies and

techniques to solve crimes, investigate deaths, and protect the public;

(3) Make recommendations for programs that will increase the number of qualified

forensic scientists and medical examiners available to work in public crime laboratories;

This kind of collaborations explains how important the link is with the scientists in the

U.S. and how evidence is taken strongly into account when dealing with public policies.
 28

ii. FDA

The FDA’s mission is to: “protect and advance public health by helping to speed

innovations that provide our nation with safe and effective medical products and that

keep our food safe. The Agency achieves this by applying the latest technology and

science-based standards to the regulatory challenges presented by drugs, biologics

(vaccines, blood products, cell and gene therapy products, and tissues), medical devices,

food additives, and, since 2009, tobacco.” (FDA, 2016)

Science is fundamental in the creation of regulations for the FDA, as there is recognition

that science-based standards are essential to providing effective public health. There are

several examples on how the FDA does that, but in the food safety area, the most

important is the Food Safety Modernization Act (FSMA). The main objective of FSMA

is to shift the food production system from being reactive to being preventative, with a

risk-focus.

FSMA was born from several scientific risk assessments of the potential contamination

routes and recent foodborne outbreaks. For example, Section 105 of FSMA, which

contains the rule “Standards for the Growing, Harvesting, Packing, and Holding of

Produce for Human Consumption” was initially created after the findings of the ‘‘Draft

Qualitative Assessment of Risk to Public Health from On-Farm Contamination of

Produce”. (FDA, 2016)

iii. Joint Organisms and Homeland Security Centers of Excellence

The need to create better science and to extend the scientific knowledge to the public is

taken in high regard by the US agencies. For food safety issues, it is of paramount
 29

importance to leverage the resources given by Academia and to create synergies using

Government-Academia alliances.

One of the successful experiences is the FDA’s Joint Institute for Food Safety and

Applied Nutrition (JIFSAN), which is a collaborative project with the University of

Maryland. Its mission is to “be a premier source of scientific information and education

programs on food safety and applied nutrition that enables the development of sound

public health policy and reduces the incidence of food-related illness.” (JIFSAN, 2016).

Established in 1996, one of its numerous achievement is to have delivered in-country

international training programs over 70 times in 24 different countries. These training

programs range from Good Agricultural Practices to seafood HACCP trainings. (JIFSAN,

2016)

The second successful collaborative program worth mentioning is the Homeland Security

Centers of Excellence. The “DHS S&T Centers of Excellence (COEs) develop

multidisciplinary, customer-driven, homeland security science and technology solutions

and help train the next generation of homeland security experts.” (DHS, 2016). There are

eight centers for excellence that focus on protecting the US from external and internal

attacks on any critical supply chain or infrastructure. Regarding food safety, the “Food

Protection and Defense Institute (FPDI), led by the University of Minnesota, defends the

safety and security of the food system by conducting research to protect vulnerabilities in

the food supply chain”. (DHS, 2016)

 30

B. European Union

i. EFSA

The work of EFSA is mainly focused on answering member states, European

Commission and Parliament. The scientific advice comes from the Scientific Panels and

Scientific Committee, organisms that adhere to several working principles such as

transparency, cooperation and independence. There is a structured process on how EFSA

conducts science and a quality assurance system that “continually monitors and

strengthens the quality of EFSA’s scientific work” (EFSA, 2016).

Among the myriad number of activities that EFSA conducts, there is a multi-annual

project called: “Promoting Methods for Evidence Use in Science” that defines principles,

processes and methods for the use of evidence in scientific assessment. (EFSA, 2015).

Moreover, the project: “Application of systematic review methodology to food and feed

safety assessments to support decision making” was performed in 2010. (EFSA, 2010).

These kinds of activities indicate the high regard which the EU holds for scientific

evidence in the decision making process.

C. Chile

1. ACHIPIA: Scientist Network

ACHIPIA has set the Risk Analysis Process as the prime resource to integrate science

into its advisory responsibilities. The Scientist Network has become one of the main

sources for local data and expert elicitations.

The Food Safety Scientist Network was created in 2014 to establish an effective link

between ACHIPIA and the scientific community. Its activities range from local data
 31

collection, expert elicitation panels and an Advisory Scientific Committee that manages

all the collaboration between the agency and the scientific community and sets the

priorities for the Network. (ACHIPIA, 2016)

During 2016, five expert elicitations have been conducted and more than ten Scientific

Opinions had been submitted to international fora such as the Codex Alimentarius

Commission and EFSA.

D. International Organizations

i. WHO, FAO and the Codex Alimentarius Commission

The World Health Organization and the Food and Agriculture Organization were

pioneers in integrating science into their decision making process through the Joint

FAO/WHO Expert Committees. These committees provide independent scientific advice

upon request to WHO and FAO. The oldest is the JECFA, which stands for Joint

FAO/WHO Expert Committee on Food Additives and was founded in 1956 (WHO,

2016). There are two other committees, the JEMRA - Joint FAO/WHO Expert Meeting

on Microbiological Risk Assessment – and the JMPR - Joint FAO/WHO Meetings on

Pesticide Residues – that are currently working and collaborating with the FAO and

WHO. Later, in 1963, when the Codex Alimentarius Commission (CAC) was

established, these committees found an improved meaning and mission, turning into the

prime resource of scientific advice and priority setting for the CAC.

With the creation of the World Trade Organization in 1995, the major multilateral food

agreement was signed: the Sanitary and Phytosanitary (SPS) agreement, which: “sets out

the basic rules for food safety and animal and plant health standards.” (WTO, 2016)
 32

For food safety, the key success of these negotiations was the acknowledgment of the

Codex Alimentarius Commission as the definitive resource of scientific information for

food international standard setting and the harmonization of food laws. Specifically:

“Harmonization with international food safety standards means basing national

requirements on the standards developed by the FAO/WHO Joint Codex Alimentarius

Commission. Codex standards are not "lowest common denominator" standards. They are

based on the input of leading scientists in the field and national experts on food safety.”

(WTO, 2016)

IV. TECHNOLOGIES AND TOOLS IN FOOD SAFETY

A. Food Safety Risk Assessment

Risk Assessment is the “scientifically based process consisting of the following steps: (i)

hazard identification, (ii) hazard characterization, iii) exposure assessment, and (iv) risk

characterization”. (Codex Alimentarius Procedural Manual, 24th edition, 2016). The

World Health Organization defines it more specifically as “the scientific evaluation of

known or potential adverse health effects resulting from human exposure to foodborne

hazards” (WHO, 2016). It is embedded in a broad food safety framework called risk

analysis, which is the “process consisting of three components: risk assessment, risk

management and risk communication”. (Codex Alimentarius Procedural Manual, 24th

edition, 2016). Risk Analysis is the modern focus that Governments are undertaking to

manage Food Safety issues.

The first mentions of risk assessments on public health in the scientific literature start

around the late 1960’s. It is not until 1983 that the National Research Council (NRC), by
 33

request of the United States Congress, wrote the book: “Risk Assessment in the Federal

Government: Managing the Process”. This book contains the first guidelines and

scientific opinions on how to use risk assessment and its related tools to “strengthen the

reliability and objectivity of scientific assessment that forms the basis for federal

regulatory policies applicable to carcinogens and other public health hazards”. (Risk

Assessment in the Federal Government: Managing the Process, NRC, 1983). This book is

the cornerstone of all the subsequent work on how scientific advice can be useful to the

Regulatory Agencies, with the objective of creating science-based regulations and

guidelines.

The WTO recognizes the value of Risk Assessment and considers it nowadays as an

essential source of evidence for managing food safety issues, not only at a national level

but international as well. The SPS agreement, for example, ensures that all international

standards are science based, which is reflected in the first paragraph of Article 5 of the

SPS agreement text: “Members shall ensure that their sanitary or phytosanitary measures

are based on an assessment, as appropriate to the circumstances, of the risks to human,

animal or plant life or health, taking into account risk assessment techniques developed

by the relevant international organizations.”

 34

B. Research Synthesis Methodologies

i. Literature Review

Harvard University describes a literature review as an: “assessment of a body of research

that addresses a research question” Its purpose is to “1) Identify questions a body of

research does not answer, and 2) Make a case for why further study of research questions

is important to a field”. (Harvard Graduate School of Education, 2016)

The Cochrane Collaboration explains that literature reviews are usually characterized by

the use of informal, unsystematic and subjective methods to collect and interpret

information. Thus, they are subject to the author’s bias, statistically, incomplete or

incorrect analysis and potentially inconsistent conclusions that may suit the author’s

experiences or overall direction of the review.

ii. Systematic Review

The Systematic Review, on the other hand, is a literature review that collects and

critically appraises several different research studies or papers, following a pre-specified

procedure and criteria. The Cochrane Collaboration defines it as: “a high-level overview

of primary research on a particular research question that tries to identify, select,

synthesize and appraise all high quality research evidence relevant to that question in

order to answer it”.

Figure 1.9 shows the main differences between literature review and systematic review.

There are a number of successful experiences of Systematic Reviews informing the

decision making process, most of them in the Health Care management area (Lavis et al,

2015, Mays et al, 2005 and Keown et al, 2008).

 35

Figure 1.9. Comparison chart between Systematic Review and Literature Reviews.
Adopted from Lynn Kish, MLIS. University of Southern California.

iii. Meta-Analysis

The purpose of a meta-analysis is to provide an estimate of an effect or observation

across two or more studies. George Washington University defines it as: “A subset of

systematic reviews; a method for systematically combining pertinent qualitative and

quantitative study data from several selected studies to develop a single conclusion that

has greater statistical power” (Himmelfarb Health Sciences Library, 2016)

 36

Usually meta-analyses are conducted within the Systematic Review framework. It is a

widely used tool in epidemiology but it has been lately used very frequently in the agri-

food public health sector. (Sargeant et al, 2006).

iv. Others

Young and colleagues (2013), defined other two research synthesis methodologies: 1)

The scoping reviews and 2) Structured rapid reviews. Scoping reviews are usually

performed to summarize the state of knowledge in a certain area, to identify data gaps

and to prioritize questions in a systematic review (Young et al, 2013). They are usually

policy-driven so they are aimed to answer specific questions. On the other hand,

structured rapid reviews are short, accelerated systematic reviews aiming to quickly

inform decision-making officers for policy and practice (Gannan et al, 2010)

V. CONCLUSIONS

During this chapter, the current economical and food safety and science situation was

described for the United States, European Union and Chile. These concepts set the

foundation to understand why it is important to develop tools and resources for

developing countries such as Chile, when evidence-based policies are needed.

Systematic Review and Risk Assessment are two tools widely used in the agri-food

public-health sector. The outputs are several and they can be used for many purposes.

Throughout this thesis, it will be shown that these two processes can be effectively

conducted by developing countries and that the outputs are easily interpretable and ready

to be integrated as a source of valuable information for decision makers or politicians.

 37

VI. REFERENCES

ACHIPIA. (2016). Vinculación con el mundo Científico. Only available in Spanish.

Retrieved from http://www.achipia.cl/vinculacion-con-el-mundo-cientifico/
Agriculture and Rural Development (2016). AGRI-FOOD TRADE STATISTICAL
FACTSHEET: European Union - Chile. Retrieved from
http://ec.europa.eu/agriculture/trade-analysis/statistics/outside-eu/countries/agrifood-
chile_en.pdf
Agriculture and Rural Development (2016). Trade statistics. Retrieved from
http://ec.europa.eu/agriculture/trade-analysis/statistics/index_en.htm
Bari, T. U. et al. (2016). History and Safety of Food: Past, Present and Future: Taylor and
Francis Group, LLC.
Centers for Disease Control and Prevention (2016). Estimates of Foodborne Illness in the
United States. Retrieved from https://www.cdc.gov/foodborneburden/
Chilean Health Ministry (2015). Situación de las Enfermedades Transmitidas por los
Alimentos en Chile. Only available in spanish. Retrieved from
http://www.achipia.cl/wp-content/uploads/2016/03/Silvia-Baeza-Minsal-Situacion-de-
las-Enfermedades-de-Transmision-Alimentaria-en-Chile-1.pdf
Committee on Identifying the Needs of the Forensic Sciences Community. (2015).
Strengthening Forensic Science in the United States: A Path Forward. Retrieved from
https://www.ncjrs.gov/pdffiles1/nij/grants/228091.pdf
European Union (2016). Living in the EU. Retrieved from https://europa.eu/european-
union/about-eu/figures/living_en
European Food Safety Authority (2016). How we work. Retrieved from
http://www.efsa.europa.eu/en/about/howwework
European Commission (2016). General Food Law. Retrieved from
http://ec.europa.eu/food/safety/general_food_law_en

Food and Drug Administration (2016). About Science & Research at FDA. Retrieved
from http://www.fda.gov/ScienceResearch/AboutScienceResearchatFDA/default.htm
Food and Drug Administration (2016). Sec. 105. Standards for Produce Safety. Retrieved
from http://www.fda.gov/Food/GuidanceRegulation/FSMA/ucm247548.htm#SEC105
Food and Drug Administration (2016). Application of systematic review methodology to
food and feed safety assessments to support decision making. EFSA Journal, 6(8).
 38

Food and Drug Administration (2015). Consolidated Annual Activity Report. Retrieved
from http://www.efsa.europa.eu/sites/default/files/corporate_publications/files/ar2015.pdf
Foodsafety.gov. (2016). Selected Federal Agencies with a Role in Food Safety. Retrieved
from https://www.foodsafety.gov/about/federal/
Ganann, R., Ciliska, D., & Thomas, H. (2010). Expediting systematic reviews: methods
and implications of rapid reviews. Implementation Science, 5, 10-19. doi:10.1186/1748-
5908-5-56
Harvard Graduate School of Education (2016). The Literature Review: A Research
Journey. Retrieved from http://guides.library.harvard.edu/literaturereview
Himmelfarb Health Sciences Library (2016). Meta-analysis. Retrieved from
https://himmelfarb.gwu.edu/tutorials/studydesign101/metaanalyses.html
Homeland Security (2016). Current Centers of Excellence. Retrieved from
https://www.dhs.gov/science-and-technology/centers-excellence
Homeland Security (2016). Food Protection and Defense Institute (FPDI). Retrieved from
https://www.dhs.gov/science-and-technology/centers-excellence
Hutt, P. B. (1984). A history of government regulation and misbranding of food. Food,
Drug Cosmet. Law L., 3(39).
International Monetary Fund (2012). Chile and the IMF. Retrieved from
http://www.imf.org/external/country/CHL/index.htm
International Life Sciences Institute (2016). Mission & Operating Principles. Retrieved
from http://ilsi.org/about/mission/
Johnson, R. (2014). The Federal Food Safety System: A Primer. Congressional Research
Service. Retrieved from www.crs.gov
Joint Institute for Food Safety and Applied Nutrition (2016). History. Retrieved from
http://jifsan.umd.edu/history/
Joint Institute for Food Safety and Applied Nutrition (2016). What is JIFSAN? Retrieved
from http://jifsan.umd.edu/about/
Keown, K., Van Eerd, D., & Irvin, E. (2008). Stakeholder engagement opportunities in
systematic reviews: Knowledge transfer for policy and practice. Journal of Continuing
Education in the Health Professions, 28(2), 67-72. doi:10.1002/chp.159
Lavis, J. N., Davies, H., Oxman, A., Denis, J.L., Golden-Biddle, K., Ferlie, E. (2005).
Towards systematic reviews that inform health care management and policy-making.
Journal of Health Serv. REs. Policy, 10(1), 35-48.
 39

Massachusetts Institute of Technology (2014). Where does Chile export to? (2014).
Retrieved from
http://atlas.media.mit.edu/en/visualize/tree_map/hs92/export/chl/show/all/2014/
Mays, N., Pope, C., & Popay, J. (2005). Systematically reviewing qualitative and
quantitative evidence to inform management and policy-making in the health field.
Journal of health services research & policy, 10 Suppl 1, 6-20.
doi:10.1258/1355819054308576
National Academy of Sciences (2016). Mission. Retrieved from
http://www.nasonline.org/about-nas/mission/
Sargeant, J. M., Rajic, A., Read, S., & Ohlsson, A. (2006). The process of systematic
review and its application in agri-food public-health. Preventive Veterinary Medicine,
75(3-4), 141-151. doi:10.1016/j.prevetmed.2006.03.002
Scallan et al (2011). Estimated annual number of episodes of illnesses caused by 31
pathogens, United States. Retrieved from
https://www.cdc.gov/foodborneburden/pdfs/scallan-estimated-illnesses-foodborne-
pathogens.pdf
World Health Organization (2016). Risk Assessment. Retrieved from
http://www.who.int/foodsafety/risk-analysis/riskassessment/en/
World Health Organization (2010). WHO estimates of the global burden of foodborne
diseases. Retrieved from
https://extranet.who.int/sree/Reports?op=vs&path=/WHO_HQ_Reports/G36/PROD/EXT
/FoodborneDiseaseBurden
World Health Organization (2016). Fact Sheet - What is JECFA? Retrieved from
http://www.who.int/foodsafety/areas_work/chemical-risks/FactSheet-
whatisJECFA.pdf?ua=1
World Trade Organization (2016). Understanding the WTO Agreement on Sanitary and
Phytosanitary Measures. Retrieved from
https://www.wto.org/english/tratop_e/sps_e/spsund_e.htm
World Trade Organization (2016). The WTO Agreement on the Application of Sanitary
and Phytosanitary Measures (SPS Agreement). Retrieved from
https://www.wto.org/english/tratop_e/sps_e/spsagr_e.htm

Young, I., Gropp, K., Pintar, K., Waddell, L., Marshall, B., Thomas, K., . . . Rajic, A.
(2014). Experiences and Attitudes towards Evidence-Informed Policy-Making Among
Research and Policy Stakeholders in the Canadian Agri-Food Public Health Sector.
Zoonoses and Public Health, 61(8), 581-589. doi:10.1111/zph.12108
 40

CHAPTER 2: RISK ASSESSMENT COLLABORATION PROJECT

I. ABSTRACT

Risk Assessment is a widely used tool for many fields. It is especially important for food

safety as it has been recognized by numerous governments and international

organizations as the main scientific evidence provider to the risk managers or decision

making bodies. Risk Assessment has reached an unprecedented relevance for food trade,

as the World Trade Organization recognizes it as the main dispute resolution system

when two nations differ in the setting of a certain food safety standard. Thus, it is very

important for all nations to be able to conduct Risk Assessments and create regulations

and policies that are based on these results. It is, however, complicated for developing

nations to achieve this. A number of factors such as a fragmented regulatory system and

insufficient scientific capabilities and technology, among others, make this process hard

to perform. In this project, we demonstrate that collaborations between the Academia and

Government are essential to narrow these gaps. Specifically, the Chilean Food Quality

and Safety Agency (ACHIPIA) engaged in a collaborative project with the University of

Nebraska-Lincoln to assess the risk on the production of raspberries destined to export to

the United States. The results indicate that the most important factors contributing to the

bacterial and viral concentration are the water used for pesticide applications and that a

considerable effort must be done to improve the data quantity and quality. This Risk

Assessment project provides simple and straightforward recommendations to the Chilean

policy makers to effectively focus their financial and human resources to solve issues that

are significantly affecting the contamination of raspberries. This collaboration was a pilot

experience and a number of lessons were learned during the process, such as the need to
 41

improve the Food Safety Scientist Network from ACHIPIA and to further bolster

Government-Academia alliances, since they are very effective in narrowing the gap

between science and policy.

II. INTRODUCTION

One of the main roles of the Chilean Food Quality and Safety Agency, ACHIPIA, is to

support the incorporation of a risk analysis framework in the context of a National Food

Quality and Safety System (SINCA). ACHIPIA is currently undergoing a design phase of

the structure and operation of a risk analysis process in its internal procedures. To

achieve this, it has been coordinating the development of several pilot programs in

collaboration with food safety scientists throughout the world (ACHIPIA, 2016). The

long-term goal is to build the capacity to implement a risk analysis framework to provide

evidence-based decisions in the agri-food sector in Chile. The results of these studies will

provide essential and new scientific information to the public services to advance SINCA

and enforce food safety for both domestic consumption and international trade.

To achieve its goal, ACHIPIA signed a cooperation agreement with the University of

Nebraska-Lincoln, Department of Food Science and Technology (UNL-FDST), with the

specific objective to support and strengthen ACHIPIA’s capacities to conduct research

projects in a variety of issues related to food quality and safety, especially within the food

safety risk analysis scope.

 42

Risk Assessment Collaborative Project

Risk Assessment is one of the three components of the Risk Analysis process. The other

two are Risk Management and Risk Communication. It is the main tool that provides

scientific evidence to the Risk Managers.

The first activity under this cooperation agreement was to conduct a risk assessment

project of the Raspberries Official Control Program (ROCP), which is enforced by the

Livestock and Agriculture Service of Chile (SAG). SAG, through ACHIPIA, reached out

to UNL-FDST to advance the current ROCP through a risk-based project for the

raspberry safety protection. Three parties, including SAG, ACHIPIA and UNL-FDST,

were involved in this collaborative risk assessment project, with the agreement that the

research group at UNL-FDST will conduct the specific risk assessment project under the

risk management objectives discussed among the three parties, based on the information

shared by SAG. The results of this assessment will be taken by ACHIPIA and SAG to

evaluate and improve the effectiveness of the ROCP.

Raspberries Official Control Program (ROCP)

The ROCP was designed to verify the fitness for human consumption and complete

traceability of the raspberries produced in Chile destined for export to the United States

of America, by establishing the auditable requirements to guarantee the safety of the

raspberries.

Two outbreaks related to raspberries set the first alarm to Chile’s producers, though they

had been systematically increasing their exports. The first one was the detection of
 43

Cyclospora on raspberries from Guatemala in 1995 (Ho et al, 2002) and later a

Calicivirus outbreak in Canada in 1997 (Berger, 2016). Though the two outbreaks were

not linked to Chilean raspberry exports, in consultation with different stakeholders

Resolution N°3410 was enacted in 2002 by the Chilean Ministry of Agriculture, which

created the ROCP. The ROCP was designed under a public consultation meeting, where

many different stakeholders had the chance to comment and work together with

government agencies. The ROCP has two main objectives: 1) verify the traceability of

the raspberries and 2) guarantee the safety for human consumption. These two objectives

are accomplished using on-site audits of the participants of the ROCP. The ROCP covers

participants in the administrative regions VI-X (Figure 2.1), which are located in the mid-

south part of Chile and covers the majority of raspberry producers in the country.
 44

Figure 2.1. Political map of Chile showing the region numbers. Taken from Icarito

encyclopedia.

Most of the participants of ROCP are small family oriented farmers. Every owner of a

raspberry farm who wishes to export their raspberries has to be accredited by SAG,

otherwise their exports will be halted by Chile’s custom before leaving the country. This

accreditation consists in the completion of a small, farmer-tailored Good Agricultural

Practices Program (GAP). This limited GAP focuses on the most common issues for

small farmers, such as water quality, hygiene measures for harvesters and animal controls

on the farm. (SAG auditor Manual, 2008). With the compliance of the GAP program, the
 45

farmers will be accredited and automatically included in the registry held by SAG, which

enables the export of their raspberries. The accreditation is active for one year and is

required to be renewed annually to stay in the registry.

Need for risk assessment in ROCP

Though ROCP has been running almost for 15 years, there is limited knowledge about

the real hazards and risk factors, since these were not formally evaluated, based on the

information collected through the auditing program conducted by SAG (SAG's personal

indication). Consequently, there is no chance to propose improvements to the program or

to the raspberry production process.

Risk Assessment is a tool that allows this kind of evaluation and furthermore, the

progression to a risk-based program where they can propose improvements in controlling

hazards that are significantly affecting the contamination. This will allow the SAG to

better allocate their human and financial resources as well as to improve the exports

amounts and raspberry safety.

Specific Objectives of the Risk Assessment project (Project Profile between ACHIPIA

and UNL, 2015)

1) Assess the risks of E. coli and Hepatitis A in the frozen and fresh raspberry

production chain;

2) Identify risk-based interventions to control microbial contamination in raspberry

end products;

3) Develop a collaborative model between academia and a regulatory agency for

food safety protection.

 46

III. MATERIALS AND METHODS

The project started from the development of a project profile, which consisted of the

problem formulation, project scope and outline, and the role and responsibilities of

involved parties (Project Profile, 2015) in detail. Briefly, a list with all the activities,

expected outcomes and responsibilities is shown in Table 2.1. The list was agreed by all

parties serving as the roadmap of this project.

 47

Table 2.1. Activity list agreed upon the project profile.

Expected result Activity Responsible

1. Description of i. Visit to a farm and packaging facility. i. SAG-
the production, ii. Create a process flow using the ACHIPIA
storage and information of the visit. ii. ACHIPIA
packing stages iii. Complement the process flow with iii. ACHIPIA
of the frozen the activities done by SAG under the
raspberries ROCP program.
process.

2. Microbiological i. Detailed description of the current i. SAG

risk assessment actions taken by SAG in the ROCP.
of the process. ii. ROCP results evaluation with current ii. SAG
available information.
iii. Collect the data generated by ROCP iii. SAG-
during the last and current season. ACHIPIA
iv. Identification and prioritization of iv. SAG-
hazards. ACHIPIA
v. Data analysis regarding ROCP v. UNL
management and water quality tests.
vi. Define the risk assessment model to vi. UNL
be used and the information needed.
vii. Expert identification for expert vii. ACHIPIA
panel/elicitation. SAG
viii. Mitigation measures identification. viii. UNL
ix. Development of the risk assessment. ix. UNL
x. Preliminary report of the risk x. ACHIPIA-
assessment. SAG
xi. Comments session on the preliminary xi. UNL
report.
xii. Final report of the risk assessment. xii. UNL
xiii. Translation of the final report. xiii. ACHIPIA
xiv. Proposal of scientific publications. xiv. SAG-
xv. Validation of the publications. ACHIPIA-
xvi. Workshop UNL
 48

The following section summarizes the process of conducting the raspberry risk

assessment project, including the main steps as follows:

1. Create and administer on-farm, collection center and packaging practices survey;

2. Build raspberry supply chain flow chart;

3. Collect data from literature and survey;

4. Fill data gaps with expert elicitation;

5. Build quantitative risk assessment model;

6. Run Monte Carlo simulation;

7. Scenario analysis; and

8. Result inference.

III.1) On-farm, collection center and packaging practices survey

A non-scheduled data collection activity additional to the activities planned in the project

profile was conducted in early 2016 (February-March), which is usually the time for

raspberry harvest and SAG audits conducted more intensively. During December 2015,

before the harvest season of 2016, the UNL-FDST group provided a list of data needed

for the development of the quantitative risk assessment model, and drafted three surveys

in English to collect data regarding the practices on the farm, at collection center and

packing plants. The draft surveys were discussed and finalized between UNL-FDST and

ACHIPIA experts, translated into Spanish by ACHIPIA and distributed by SAG to the

raspberry farmers registered in ROPC. The objectives of the surveys were to 1) obtain a

real picture of the current practices of raspberry supply in Chile, 2) collect data that can

be incorporated in the quantitative risk assessment to simulate how the practices can
 49

influence the introduction and transmission of the microbial loads towards the end

products. Therefore, these surveys provided the fundamental to narrow the data gap and

significant insights on the process from a local perspective. The surveys are provided in

English in Annex I, II, and III focusing on practices on farm, at collection center and

packing plant, respectively.

III.2) Build raspberry supply chain flow chart

A three step module process was established based on the preliminary data: Farm,

Collection Center and Packing Plant (shown in Figure 2.2). A general overview of the

process is as follows: at the farm, raspberries are planted, irrigated, applied pesticides and

fertilizers and finally harvested during summer (January-March). The Collection Center

is a place where raspberries from different farmers are gathered and sold as one package

to a Packaging Plant. The Packaging Plant is the place where raspberries are visually

inspected and selected for export (best quality), sent to juice and other processed fruits

(lower quality) or discarded.

The end products of interest include both fresh and frozen raspberries. In discussion with

SAG and ACHIPIA, the contamination of Escherichia coli and Hepatitis A virus was

studied as they had previous border detections (SAG’s personal indication). To

understand their behavior and to identify potential contamination and reduction stages,

we used the information contained in the surveys to model each event. The data collected

through the survey were vast and sometimes too complicated to be integrated in the risk

assessment model, especially because there are no mathematical models available in the

literature to relate the data. Therefore, data that were determined as not significantly

impacting the microbial contamination in raspberries were excluded.

 50

III.2.1) Justification to exclude data points

a. Irrigation practices

The Expert Elicitation indicated that the possibility of contamination with the irrigation

water is insignificant. Raspberries are extremely sensitive to the contamination with the

fungal species Botryotinia fuckeliana, which causes a gray mold disease almost always

when the fruits are exposed to high humidity situations. In the situations where the fruits

are touched by irrigation water, they would be spoiled immediately due to this fungi and

would not be harvested.

b. Frequency of pesticide application and type of application system

Water used to dilute the pesticide is considered as a potential risk factor to introduce

microbial contamination during the growth of the fruits through the pesticide application.

No data were found on the cumulative impact of multiple pesticide applications on the

microbial loads in fresh produce at the pre-harvest stage. The only similar information

found was in Petterson et al (2001), which showed that the last irrigation is the most

significant in terms of contamination. So, the last pesticide application was used in the

model. The transfer mechanisms or transfer rates were not found.

d. Hygiene of harvest trays

Cannon et al (2014) evaluated the persistence and transfer of enteric viruses in food-

contact surfaces and in foods. However, contamination data for viruses in the harvesting

trays as well as transfer rates for bacteria could not be found. Quadros Rodrigues et al

(2014) investigated the bacterial contamination on the harvesting tray, however, the

transfer rate from harvesting tray to the fruits were not found.
 51

e. Food contact surface hygiene at packing plant

Butot et al (2008) found the bacterial contamination reduction due to the use of chlorine

in food contact surfaces. No data was found on the distribution of food contact surfaces

contamination so it was impossible to model this step.

III.3) Collect data from literature and survey

As mentioned earlier, the Chilean farmers were surveyed and information production

practices was collected. The surveys were received in Spanish, translated and answers

collected in an Excel spreadsheet. For the farm module, 226 surveys were received, 23

for the collection center and 36 for the processing plant.

Literature searches were conducted using UNL’s library resources, mainly the Web of

Science database. Data was fitted by @risk (Palisade Corporation, 2016) and integrated

in the risk assessment model with the proposed distribution. Tables 2.2-2.9 summarize

the information collected and the sources.

 52

Figure 2.2. Flow chart of the processing steps for raspberries including potential hazards and reduction steps at the packing

plant.

52
 53

III.4) Fill data gaps with expert elicitation

A spreadsheet was designed to collect missing data and was sent to the Food Scientists

Network, managed by ACHIPIA. The spreadsheet is shown in Annex IV.

III.5) Build quantitative risk assessment model

Tables 2.2-2.9 list the inputs used in the risk assessment model. Based on the information

collected, different equations were constructed to model each one of the steps in the risk

assessment model.
 54

Table 2.2. List of parameters, values and distributions used in the farm module for both
fresh and frozen raspberries.
Parameter
Description Distribution/Unit Reference
(information
source)
w_t_pest Type of water used for pesticide Discrete @ risk fit from survey
(Survey) applications
1 – Groundwater 1 – 71%
2 – Surface 2 – 15%
3 – Potable 3 – 14%
Cw_1 Bacterial groundwater Uniform (0,1000) GDWQ, 3rd Edition
(Lit. search)
contamination CFU/L

Cw_2 Bacterial Surface water Pareto (1.31,2900) @risk fit from de Roda Husman et al.,
(Lit. search)
contamination CFU/L 2006

Cw_3 Bacterial Potable water Uniform (0.01,0.1) Chilean potable water regulation “Nch
(Lit. search)
contamination CFU/L 409”

Cw_4 Viral groundwater contamination Uniform (0,2) GDWQ, 3rd Edition

(Lit. search)
PDU/L

Cw_5 Viral Surface water contamination Uniform (0.01,10) GDWQ, 3rd Edition
(Lit. search)
PDU/L

Cw_6 Viral Potable water contamination Uniform (0.006-4) Borchard et al, 2012
(Lit. search)
PDU/L

Tap How much times goes by between Laplace (30,21.88) @ risk fit from surveys
(Survey) the last application and the
harvest? Days

D Bacterial and viral decay rate Triangular Danyluk et al, 2011

(Lit. search)
(0.008,0.019,0.039)

Log CFU/day

Log PDU/day

Bac_transf Percentage of bacterial transfer Uniform (0.000081, Gerba et al, 2005 and 2011
(Lit. search) per 0.5gr
0.00011)

Vir_transf Percentage of virus transfer per Uniform (0.021, Gerba et al, 2005 and 2011
(Lit. search) 0.5gr
0.031)

Prev_hands Bacterial prevalence in harvesters Beta (7,41) Aceituno et al, 2016

(Lit. search) hands
 55

Table 2.2 (Continuation) List of parameters, values and distributions used in the farm
module for both fresh and frozen raspberries.
Parameter
Description Distribution/Unit Reference
(information
source)
F_prod Transferred proportion per touch Beta(15.64,41.94) Verhaelen et al, 2013
(Lit. search) from produce to hand

W_harv Surface area of hands that touch 2.1 cm2 Verhaelen et al, 2013
(Lit. search) the produce

W_hand Total surface area of one side of 245 cm2 USEPA, 2011
(Lit. search) one hand

W_prod Surface area of produce Normal (1064,167) Bouwknegt et al, 2015

(Lit. search)
mm2

F_hand Transferred proportion per touch Lognormal(-8.34,0.58) Verhaelen et al, 2013

(Lit. search) from hand to produce

C_harv_vir Virus number on harvester's hand Gamma(0.14,54.6) Bouwknegt et al, 2015

(Lit. search)
PDU/hand

C_harv_bac Bacterial number in harvester's Uniform(1,1.9) Quadros Rodrigues et al, 2014

(Lit. search) hands
CFU/cm2

transp_time How long does it take from the Loglogistic(0.0014937 @risk fit from survey
(Survey) Farm to the Collection Center
,0.044281,1.7081)

Days

transp_temp At which temperature are the Triangular(11.256,28, @risk from survey

(Survey) raspberries usually transported?
28) °C
 56

Table 2.3. List of parameters, values and distributions used in the collection center
module for fresh raspberries.
Parameter
Description Distribution/Unit Reference
(information
source)
Time_cc Average time raspberries stay in Triangular @ risk fit from survey
(Survey) the Collection Center (0.041667,0.041667,0.33716)
Days

Temp_cc What is the average temperature Extreme Value(24.3522,5.1304) @ risk fit from survey
(Survey) of the Collection Center?
°C

transp_temp What it the temperature in the Triangular (-7.6691,27,27) °C @ risk fit from survey
(Survey) transport?

transp_time Time taken from the Collection Exponential (0.060343) Days @ risk fit from survey
(Survey) Center to the Packing Facility

Table 2.4. List of parameters, values and distributions used in the collection center
module for frozen raspberries.
Parameter
Description Distribution/Unit Reference
(information
source)
Time_cc_frz Average time raspberries stay Uniform (30,40) Days Survey
(Survey) in the Collection Center

Temp_cc_frz What is the average Uniform (-22.5,-18) °C Survey

(Survey) temperature of the Collection
Center?

transp_temp_frz What it the temperature in the Uniform (-22.5,-20) °C Survey

(Survey) transport?
transp_time_frz Time taken from the Collection Uniform (0.0007,0.0834) Survey
(Survey) Center to the Packing Facility
Days
 57

Table 2.5. List of parameters, values and distributions used in the packing plant module
for fresh raspberries.
Parameter
Description Distribution/Unit Reference
(information
source)
wait_time_rec Waiting time when receiving the Exponential (0.010305) Days @ risk fit from survey
(Survey) raspberries

wait_temp_rec Average temperature in the Triangular (0.050215,27,27) °C @ risk fit from survey
(Survey) receiving

cold_time Time that the fruits stays at the Cold Triangular (- @ risk fit from survey
(Survey) Chamber
0.0093158,0.083333,0.56261) Days

cold_temp Target temperature in the Cold Exponential (0.79688) °C @ risk fit from survey
(Survey) Chamber
C_food_vir Virus number on handler's hand Gamma(0.67,1.62) Bouwknegt et al, 2015
(Lit. search)
PDU/hand

Proportion of the food handler’s

πfood hand touching the produce
Uniform (0,1) Bouwknegt et al, 2015
(Lit.Search)
C_food_bac Bacterial number in handler's hands Uniform(1,1.9) Quadros Rodrigues et
(Lit. search) al, 2014
CFU/cm2

pack_time Time taken from selection to Loglogistic @ risk fit from survey
(Survey) transport
(0.0043615,0.0080573,1.7482) Days

pack_temp What is the temperature inside the Logistic (7.6448,1.4959) °C @ risk fit from survey
(Survey) Packing area
time_transp Time taken to destination. Pareto(0.77518,0.083333) days @ risk fit from survey
(Survey)
temp_transp Temperature of the cooling truck Loglogistic(-23.0679,4.6603,4.4384) °C @ risk fit from survey
(Survey) during transport
 58

Table 2.6. List of parameters, values and distributions used in the packing plant module
for frozen raspberries.
Parameter Description Distribution/Unit Reference

wait_time_rec The wait time in the receiving Laplace (0.021,0.0164) Days @ risk fit from survey
(Survey)

wait_temp_rec Average temperature in the Triangular (0.050215,27,27) °C @ risk fit from survey
(Survey) receiving

cold_time Time that the fruits stays at the Cold Triangular (- @ risk fit from survey
(Survey) Chamber
0.0093158,0.083333,0.56261) Days

cold_temp Target temperature in the Cold Exponential (0.79688) °C @ risk fit from survey
(Survey) Chamber
C_food_vir Virus number on handler's hand Gamma(0.67,1.62) Bouwknegt et al, 2015
(Lit. search)
PDU/hand

C_food_bac Bacterial number in handler's hands Uniform(1,1.9) Quadros Rodrigues et

(Lit. search) al, 2014
CFU/cm2

pack_time Time taken from selection to freeze Loglogistic @ risk fit from survey
(Survey) chamber
(0.0043615,0.0080573,1.7482) Days

Proportion of the food handler’s

πfood hand touching the produce
Uniform (0,1) Bouwknegt et al, 2015
(Lit.Search)
pack_temp What is the temperature inside the Logistic (7.6448,1.4959) °C @ risk fit from survey
(Survey) Packing area
Frz_temp The target temperature is Uniform (-35,-25) °C @ risk fit from survey
(Survey)
Frz_time Time at freezing chamber Inverse Gaussian (16.348,1.3124) days @ risk fit from survey
(Survey)
time_transp Time taken to destination. Pareto(0.77518,0.083333) days @ risk fit from survey
(Survey)

temp_transp Temperature of the cooling truck Loglogistic(-23.0679,4.6603,4.4384) °C @ risk fit from survey
(Survey) during transport
 59

Pre-harvest contamination (Farm module)

The objective of this module is to understand how the contamination from the water is

being transferred to the crops during the pesticide application. The concentration on the

raspberry during the pre-harvest stage (Cph) were calculated as a function of the

concentration in the raspberry after the last pesticide application (Cap), the time between

the last application and harvest (Tap) and the decay rate (D) using the following

calculations proposed by Danyluk et al (2011):

𝐶𝑝ℎ = 𝐶𝑎𝑝 − 𝑇𝑎𝑝 ∗ 𝐷 (1)

Gerba and collegues calculated the transfer rate of bacteria and viruses during a pesticide

application (Gerba et al, 2005). This information was used to calculate Cap, which is the

product of concentration of the water used (Cw) and the bacterial or viral transfer rate

(Bac_transf and Vir_transf).

Cross-contamination at harvest (Farm module)

To assess the potential contamination contribution due to harvesting practices of

raspberries, the Bouwknegt et al (2015) model was used. The number of bacteria or

viruses per gram (Nharv) of raspberry during harvest was calculated as

𝑊ℎ𝑎𝑟𝑣 𝑊ℎ𝑎𝑟𝑣
𝑁ℎ𝑎𝑟𝑣 = 𝐶𝑝ℎ − 𝑓𝑝𝑟𝑜𝑑 𝐶𝑝ℎ + 𝑓ℎ𝑎𝑛𝑑 𝐶ℎ𝑎𝑟𝑣 (2)
𝑊𝑝𝑟𝑜𝑑 𝑊ℎ𝑎𝑛𝑑

with Fhand being the proportion of viruses transferred from hand to raspberries. The size

of a hand (Whand) corresponds to the total surface area of a harvesters’ hand. (USEPA,
 60

2011) Wharv is the area of the hand that actually touches the raspberries. Finally, Charv is

the concentration of bacteria or viruses in the hand.

Growth model (Farm, Collection Center and Packing Plant modules)

One of the main effects on the bacterial populations is the growth due to temperature

abuse and the reduction due to freezing and cooling practices. Danyluk and colleagues

(2011) studied the growth parameters of E. coli O157:H7 in leafy greens and proposed a

growth model. Survival of E. coli O157:H7 was studied as well in strawberries during

cooling and freezing temperatures (Harris et al (2011)). Based on data extracted from

these two publications that were found the most similar to this research, three models

were created based on the temperature of the process under modelling: over 8°C, between

0°C and 8°C, and under 0°C . Tables 2.7, 2.8 and 2.9 indicate the summarized

parameters and values.

For the cooling and freezing temperatures, a maximum reduction (rmax) was proposed

based on the data from Harris et al (2001). Additionally, the first days of freezing have a

stronger reduction in bacterial populations, so two different reduction rates (r1 and r2)

were proposed based upon the freezing times. For less than 8 days, r1 was used and for

more than 8 days, r2 was used.

 61

Table 2.7. Bacterial growth model parameters for temperatures over 8°C.
Parameter Parameter Equation Value/Distribution/ Unit
ID Description Calculation
µ Growth rate (b*(T-T0))^2 - Log CFU
T Temperature of - See Table 5,6,7,8 °C
modelled step and 9
T01 Temperature constant - 2.628 sqrt(log
1 cfu/day/°C)
b1 Temperature constant - 0.0616 °C
2
t Time of the modelled - See Table 5,6,7,8 Days
step and 9
Ci initial concentration - From previous step Log CFU/gr
- Final concentration Ci+ µ*t - Log CFU/gr
1
Equations and constants are adopted from Danyluk et al., 2011.

Table 2.8. Growth model parameters for temperatures between 0°C and 8°C.
Parameter Parameter Description Equation Value/Distribution/ Unit
ID Calculation
r1 Reduction per day - 0.18l Logs/day
rmax1 Maximum log reduction - 1.225 Logs
t Time of the modelled - See Table 5,6,7,8 and Days
step 9
Ci Initial concentration - From previous step Log CFU/gr
- Final concentration Ci-r*t - Log CFU/gr
or
Ci- rmax
1Parameters derived from Danyluk et al, 2011 data.
 62

Table 2.9. Growth model parameters for temperatures below 0°C.

Parameter Parameter Description Equation Value/Distribution/ Unit
ID Calculation
r 1a Reduction per day, less than - 0.18l Logs/day
8 days
r 2a Reduction per day, more - 1.225 Logs
than 8 days
rmaxa Maximum reduction - 1.6 Logs
t Time of the modelled step - See Table 5,6,7,8 Days
and 9
Ci Initial concentration - From previous step Log CFU/gr
- Final concentration Ci-r1*t, if t<8 - Log CFU/gr
or
Ci-r2*t, if t>8
or
Ci- rmax
a
Parameters derived from Danyluk et al, 2011 data.

Cross-contamination due to handling (Packing Plant module)

Similar to the harvesting module, the Bouwknegt et al (2015) model was used for the

handling of raspberries during selection in the packaging plant. The selection process

consists of workers manually handling raspberries to assess their visual quality. The

number of bacteria or viruses per gram (ntouch) of raspberry during the selection process

was calculated as

𝑊𝑓𝑜𝑜𝑑
𝑛𝑡𝑜𝑢𝑐ℎ = 𝐶𝑐𝑐 − 𝑓𝑝𝑟𝑜𝑑 𝜋𝑓𝑜𝑜𝑑 𝐶𝑐𝑐 + 𝑓ℎ𝑎𝑛𝑑 𝐶 (3)
𝑊ℎ𝑎𝑛𝑑 𝑓𝑜𝑜𝑑

with Ccc being the concentration in the raspberry after the Collection Center, which is the

previous step to the Packaging Plant where raspberries are stored and selected. Cfood is the

concentration of viruses or bacteria in the handler’s hands, πfood is the proportion of the

food handler’s hand touching the produce and Wfood is the touching surface of a handler’s

hand which is the same as Wharv at the Farm.

 63

III.6) Run Monte Carlo simulation

Once the model was developed, the Monte Carlo simulation using Latin Hypercube

sampling for 10,000 iterations was performed to obtain stochastic estimates of the output

variables, namely, bacterial and viral contamination loads in both fresh and frozen

raspberry products, using Microsoft Excel add-on package @Risk (version 7.0, Palisade

Corporation, New York, USA). Sensitivity analysis was conducted to evaluate the

importance of input variables on the changes in contamination risks, represented in

tornado charts.

III.7) Scenario analysis

The efficacy of microbial control interventions that can be potentially adopted at different

points along the raspberry supply chain were evaluated through a scenario analysis. A

total of 13 scenarios were run in the model, including a baseline scenario for comparative

purposes using the data mentioned above for the estimate of “no intervention” scenario

and 10 other alternative scenarios to predict the food safety protection in end raspberry

products if a specific intervention technology or regulation would be adopted. For each

scenario, the model was run for 10,000 iterations to generate the mean risk estimates. All

the scenario analysis were conducted on fresh raspberries. The list of scenarios evaluated

is shown in Table 2.10 for water interventions and in Table 2.11 for the reduction of time

when raspberries are stored at the collection centers.

Previous studies show that water is one of the prime sources of contamination for berries

and leafy greens (Bern et al, 1999 and Ashbolt et al, 2001). As shown in the on-farm

practice survey, raspberry farms in Chile mainly rely on three types of water sources with
 64

microbial safety level in the order of portable water as the cleanest source, followed by

ground water and surface water. Therefore, one of the water intervention actions

evaluated in this study is changing the use of potable water and/or ground water instead

of surface water with the improvement of public water treatment and supply

infrastructure in Chile. The changes in water sources were modeled by increasing the

proportions of raspberry farms using potable and/or ground water in the model. To

control the microbial loads in the water sources, the introduction of ultraviolet light is the

other intervention actions evaluated in this thesis, because it has been shown that

ultraviolet lamps are easy to install and operate in less expensive costs and do not create

harmful byproducts (Pariseau et al, 2010). Ultraviolet light has been demonstrated to

reduce bacterial and viral contamination in water by 2-4 logs (Chang et al, 1985 and

Pariseau et al, 2010). Combinations of the two water intervention actions were also

evaluated. Relative changes in mean risk estimates of each alternative scenario were

calculated, compared to the baseline scenario.

 65

Table 2.10. Water uses scenario analysis for bacterial contamination.

Water type
Occurrence
Water Occurrence Occurrence
Scenario of
contamination of surface of potable
groundwat Water type change
water use water use
er use
(SW) (PW)
(GW)
Contamination as Current occurrence
Baseline 71% 15% 14%
current

A No intervention 86% 0% 14% 100% SWGW

B No intervention 42% 8% 50% 50% SW  GW & 50%

GW  PW
C No intervention 5% 5% 90% GW&SWPW

UV light Current occurrence

D 71% 15% 14%
intervention

A+D UV light 86% 0% 14% 100% SWGW

intervention

B+D UV light 42% 8% 50% 50% SW  GW & 50%

intervention GW  PW
 66

Table 2.11. Transportation reduction time scenarios for bacterial contamination.

Transport time from farm to Transport time from collection
Scenario
collection center center to packing plant
Baseline 0-9 hours 0.5-8 hours

E 1 hour Baseline

F Baseline 1 hour

E+F 1 hour 1 hour

Table 2.12 Temperature reduction at Collection Center scenarios

Scenario Temperature at Collection Center

Baseline 0.5-30 °C

G 50 % reduction

H 4-8 °C
(fully implemented refrigeration system)

Table 2.13 Pesticide applications time scenarios

Scenario Harvest time after last application

Baseline 0-120 days

I 25% increase

J 50% increase
 67

IV. RESULTS AND DISCUSSION

Risk estimates of current practices

Figures 2.3 and 2.4 show the contamination distribution at the end of the process for fresh

raspberries for E.coli and Hepatitis A, respectively. The bacterial contamination for

frozen raspberries is shown in Figure 2.5. Data for viral contamination in the frozen chain

is not shown because the only parameters changed are the freezing practices, which does

not result in difference from fresh fruits. Note that the baseline scenario was not an

accurate representation of the current risk estimate of contamination in raspberry

products, since some initial input parameters were populated with data extracted from

studies conducted in countries other than Chile.

For the fresh raspberries, bacterial contamination mean was -1.89 log CFU/gr. The

majority of the results (95% probability interval) for 10,000 iterations ranged between -

5.48 and 0.13 log CFU/gr with the maximum value over 8 logs. The contamination mean

for the frozen raspberries was -4.44 log CFU/gr.

The viral contamination mean for fresh raspberries was -2.07 log PDU/gr. The majority

of the results (95% probability interval) for 10,000 iterations ranged between -3.67 and -

0.93 log PDU/gr with a maximum value of 0.03 log PDU/gr.

 68

Density

Log CFU/gr
Figure 2.3. Bacterial Log CFU/gr contamination distribution of 10,000 iterations
simulation for the fresh raspberry model. The 95% probability interval of the results are
highlighted in the top portion of the plot.
Density

Log PDU/gr
Figure 2.4. Viral log PDU/gr contamination distribution of 10.000 iterations simulation
for the fresh raspberry model. The 95% proportion of the results are highlighted in the top
portion of the plot.
 69

Density

Log CFU/gr
Figure 2.5. Bacterial log CFU/gr contamination distribution of 10.000 iterations
simulation for the frozen raspberry model. The 95% proportion of the results are
highlighted in the portion of the plot.

Expert elicitation

This project demonstrated that the Food Scientist Network is at its early development

stage and that risk assessment procedures are still widely unknown, even to scientists. A

number of questions were received indicating that the scientist were not understanding

what was being asked, although examples were given. No data was received directly

from the spreadsheet, but useful information was delivered, for example, that irrigation

water should not be considered because the soft rot caused by Botryotinia fuckeliana.

Sensitivity analysis

The tornado plots shown in Figures 2.6, 2.7 and 2.8, indicate the inputs that have the

largest impact in the simulations. For the bacterial contamination in the fresh chain, the

three largest inputs that changes the results are the type of water used, times of transport
 70

time from the Packing Plant and time after the last pesticide application. For the viral

contamination, the three largest inputs that changes the results are the time after the last

application, the groundwater contamination and the decay rate. Finally, for the frozen

raspberry supply chain, the most important parameters are the type of water used, the

freezing times and time of transport from the Packing Plant.

In all Monte Carlo simulations for every data set, one of the recurring significant

parameters is the water used for pesticides applications. This is intuitive as several reports

had indicated that water is one of the main vehicles for contamination of fresh produce

(Herwaldt et al. 1997), especially in the case of water used for pesticide applications

(Gerba et al. 2011, Caceres et al.1998; Herwaldt and Beach 1999). Initially irrigation

water was also considered but later discarded due to the impossibility of harvesting a

raspberry due to fungal spoilage associated with this event (Expert Elicitation, ACHIPIA

2016).

Freezing practices in the freezing chamber and the transport from the Packing Plant are

also significant in the outputs since very low temperatures and extended periods of time

reduces the bacterial load significantly (Harris et al, 2001).

As seen in Figure 2.8, time after the last application, groundwater contamination and the

decay rate – all data related to pesticides applications – have the largest impact in viral

concentrations. This is largely due to the fact that this stage is the only source of entry for

viral contamination in this model.

 71

Manipulation by harvester and handler hands does not show in the simulation as a

significant factor. Due to the lack of the data, the prevalence data was not considered

while is the most important parameter to study when assessing the impact of cross-

contamination. The latter is especially important because with the model and data

collected from different authors (Aceituno 2016, Quadros Rodriguez 2015 and

Bouwknegt 2015) the net effect of touching a raspberry is an actual transfer of

contamination from the raspberry to the hand, rather than the opposite direction.

All the results are within a low range, the fresh raspberry chain is the one with the highest

counts of E. coli. The reason is that during the fresh raspberry chain, there are more

waiting periods with higher temperatures. Nevertheless, the latter is not seen in the

tornado plot in Figure 2.6, where one would expect these times and temperatures to have

larger effects in the estimates. This is very likely due to the uncertainties linked to the

water contamination data and transfers ratio to the fruit due to the pesticides applications.

There are significant uncertainties in the model that are classified in two categories: 1)

non-local data and 2) non-optimized data. The first relates to the need to use data that has

not been created from Chilean sources, such as the water contamination and the handler’s

hand contamination. The second class refers to data that was collected from other models

and uses. Among others, the transfer rates proposed by Gerba et al (2005) were intended

for lettuce not for raspberries, thus, this is an important limitation of the model.

Collection of data in terms of reducing uncertainty and in terms of having appropriate

parameters closer to our research food matrix are invaluable to improving the quality of

the risk estimates.

 72

Log CFU/gr
Figure 2.6. Tornado plot for the final bacterial concentration for the fresh raspberry model.

72
 73

Log CFU/gr
Figure 2.7. Tornado plot for the final bacterial concentration for the frozen raspberry model.

73
 74

Log PDU/gr

Figure 2.8. Tornado plot for the final viral concentration for the fresh raspberry model.

74
 75

Scenario analysis and interventions evaluation

Table 2.14 summarizes the results from the different scenario analyses.

For the Scenarios A-C, changing the frequency of the type of water in use had a strong

impact on the bacterial populations but not in the virus populations. Increasing the use of

potable water reduced the bacterial populations by 66.35% and 136.96% for scenarios B

and C, respectively. Viral populations were slightly affected by the changes in the

frequency of use of the water sources.

Using UV-lamps had a much more marked effect, reducing bacterial populations to a

similar level than when using mainly potable water (Scenario C). All scenarios with the

UV lamp had at least 100% log reduction in bacterial populations and 50% viral.

Scenarios E and F had little effect on the simulations, resulting in reductions less than 6%

in every case.

The scenario cases provides interesting insight on the production chain. As seen in Table

2.14, increasing the frequency of the use of potable water (Scenario C) is very effective

for bacteria populations, but not for viruses. The minor increase in viruses may be due to

the lack of data for potable water; the few data points collected from Borchard et al

(2012), describes slightly higher concentration numbers compared to the global estimates

of the WHO for groundwater. (GDWQ, 3rd Edition)

 76

Table 2.14. Summary of the scenario analysis results.

Mean contamination
Scenario % change compared to baseline
(log CFU/gr or log PDU/gr)
Bacterial
-1.84
Viral -
-2.07
(Baseline)
9.24% reduction
-2.01
A 5.31% reduction
-2.18
66.30% reduction
-3.06
B 2.89% increase
-2.01
136.96% reduction
-4.36
C 7.73% increase
-1.91
133.15% reduction
-4.29
D 158.94% reduction
-3.29
194.02% reduction
-5.41
A+D 55.07% reduction
-3.21
133.7% reduction
-4.30
B+D 59.9% reduction
-3.31
E
-1.90 3.26% reduction
(Only bacterial)
F
-1.88 2.17% reduction
(Only bacterial)
E+F
-1.94 5.43% reduction
(Only bacterial)
G
-1.99 8.15% reduction
(Only bacterial)
H
-2.03 10.33% reduction
(Only bacterial)
8.15% reduction
-1.99
I 9.18% reduction
-2.26
-2.15 16.85% reduction
J
-2.44 17.87% reduction

On the other hand, the proposed ultraviolet lights intervention indicates significant

reduction in both bacterial and viral populations. As shown in Table 2.14, the log
 77

reduction achieved by this technology (Scenario D) for bacteria and virus are up to 133%

and 159%, respectively. The effect on viruses is larger probably because there are no

further grow stages as with bacteria.

The combination of both UV lamps and increasing potable water use (Scenarios A+D and

B+D) does not seem to provide considerable further reduction, especially considering that

the groundwater is increasing the virus counts (Scenarios B and C)

This technology is currently being applied in small farms in Chile (Expert Elicitation,

ACHIPIA, 2016) so it arises as an interesting potential intervention.

As the receiving in the Collection Center is currently unrefrigerated, two scenarios were

simulated (Scenarios G and H). The reduction achieved for a 50% decrease in

temperature is 8.15%. Even implementing a refrigeration system in this step, which can

be very costly, only reduces the contamination by 10.33%. The waiting time in this stage

is very short (Table 2.3) so any temperature intervention would affect the final

contamination considerably.

Although the time of application before the harvest appears to be an important input in

the simulations (Figures 2.6 and 2.8), the reduction achieved for scenarios I and J is

considerably smaller than previous scenarios. The practices associated with these last

scenarios can be very resource consuming so it does not seem a practical intervention.

Data gaps identification

Several contamination routes were dropped due to the lack of models available to connect

the Chilean data – mostly about frequency of use – or inexistent prevalence and

concentration data for the selected microorganisms in raspberries. Animal contamination

 78

on farm, harvester tray contamination, food contact surfaces and others were some

datasets that had to be discarded due to these reasons. There is need to increase the data

available, not only from an experimental perspective but from an observational point of

view.

Nevertheless, there is much uncertainty as Chilean specific water contamination data was

not obtained. Another uncertainty factor is the decay rate, Danyluk et al (2011) was the

only author that proposed a usable estimate, although on spinach for an Escherichia coli

surrogate.

The transfer rate used was estimated on lettuce, due to the lack of studies conducted in

raspberries; data from experimental research was taken and applied. (Gerba et al, 2005)

Even considering these limitations, a comprehensive estimate was given for the behavior

of the bacterial and viral populations in the fresh and frozen raspberry production chain.

There is a need for open access information and the creation of continuous surveillance

systems that provide this kind of data to researchers. Academia-Government

collaborations are useful to accomplish this objective, as shown in this study for some

datasets.

Significance for regulators and evidence-based policies.

This collaborative project is the first in its kind in the realm of food safety in Chile.

ACHIPIA and SAG were effective collaborators and the outputs of this study are ready to

be evaluated by risk managers or policy makers. The results are displayed in a simple

way and very visual. There is no need to have specific expertise to critically analyze these
 79

results and scientific evidence has been effectively provided to take well informed

decisions.

V. CONCLUSIONS

Risk Assessment is a tool that has been used since the decade of the 1980’s. It is a very

well structured process that takes into consideration the data limitations and provides

easy to understand information to risk managers. Although the process itself requires

scientific expertise, this is when strategic alliances such as collaborations between

Academia and Government are most useful.

In this particular Risk Assessment project the key findings from the perspective of

Chile’s government are:

1. Water quality needs to be improved as it is the main effector of contamination in

raspberries.

2. Frozen raspberries are much safer in terms of bacterial contamination. Virus

contamination is similar as in fresh raspberries.

3. Relatively cheap and easy to use technologies, such as ultraviolet light application,

provide important contamination reduction. These interventions could be applied while a

more definitive solution is developed, such as stronger regulation on water quality.

4. The use of Risk Assessment provides critical insight on the information gaps. There is

a need for more research into water sources, raspberry-specific contamination transfer

due to animal waste, and the prevalence of bacteria and viruses in the food operation

premises, among others.

 80

5. As stated previously, one of the objectives was to try the collaborative experience

between Academia and ACHIPIA. Although no data was collected directly from the

Expert Elicitation spreadsheet, very useful guidance and general comments were

received. These kind of tools proved to be key in narrowing the gap between developing

and developed countries when trying to integrate science into their decision making

process.

VI. REFERENCES

ACHIPIA. (2016). Area de soporte al análisis de riesgos. Retrieved from

http://www.achipia.cl/inicio/areas-de-coordinacion/area-de-soporte-al-analisis-de-
riesgo

Berger, S. (2016). Gastroenteritis - Viral Infectious Diseases of Canada: 2016 Edition

(pp. 187).

Bern, C., Hernandez, B., Lopez, M. B., Arrowood, M. J., de Mejia, M. A., de Merida, A.
M., . . . Klein, R. E. (1999). Epidemiologic studies of Cyclospora cayetanensis in
Guatemala. Emerging Infectious Diseases, 5(6), 766-774.

Borchardt, M. A., Spencer, S. K., Kieke, B. A., Lambertini, E., & Loge, F. J. (2012).
Viruses in Nondisinfected Drinking Water from Municipal Wells and Community
Incidence of Acute Gastrointestinal Illness. Environmental Health Perspectives,
120(9), 1272-1279. doi:10.1289/ehp.1104499

Bouwknegt, M., Verhaelen, K., Rzetutka, A., Kozyra, I., Maunula, L., von Bonsdorff, C.
H., . . . Husman, A. M. D. (2015). Quantitative farm-to-fork risk assessment
model for norovirus and hepatitis A virus in European leafy green vegetable and
berry fruit supply chains. International Journal of Food Microbiology, 198, 50-
58. doi:10.1016/j.ijfoodmicro.2014.12.013

Butot, S., Putallaz, T., & Sanchez, G. (2008). Effects of sanitation, freezing and frozen
storage on enteric viruses in berries and herbs. International Journal of Food
Microbiology, 126(1-2), 30-35. doi:10.1016/j.ijfoodmicro.2008.04.033
Caceres, V. M., Ball, R. T., Somerfeldt, S. A., Mackey, R. L., Nichols, S. E., MacKenzie,
W. R., & Herwaldt, B. L. (1998). A foodborne outbreak of cyclosporiasis caused
by imported raspberries. Journal of Family Practice, 47(3), 231-234.
 81

Chang, J. C. H., Ossoff, S. F., Lobe, D. C., Dorfman, M. H., Dumais, C. M., Qualls, R.
G., & Johnson, J. D. (1985). UV INACTIVATION OF PATHOGENIC AND
INDICATOR MICROORGANISMS. Applied and Environmental Microbiology,
49(6), 1361-1365.

Danyluk, M. D., & Schaffner, D. W. (2011). Quantitative Assessment of the Microbial

Risk of Leafy Greens from Farm to Consumption: Preliminary Framework, Data,
and Risk Estimates. Journal of Food Protection, 74(5), 700-708.
doi:10.4315/0362-028x.jfp-10-373

de Aceituno, A. F., Heredia, N., Stern, A., Bartz, F. E., Venegas, F., Solis-Soto, L., . . .
Garcia, S. (2016). Efficacy of two hygiene methods to reduce soil and microbial
contamination on farmworker hands during harvest. Food Control, 59, 787-792.
doi:10.1016/j.foodcont.2015.06.059

Herwaldt, B. L., Ackers, M. L., Farrar, J., Richardson, S., Nelson, R., Fletcher, M., . . .
Messonnier, M. (1997). An outbreak in 1996 of cyclosporiasis associated with
imported raspberries. New England Journal of Medicine, 336(22), 1548-1556.
doi:10.1056/nejm199705293362202

Herwaldt, B. L., Beach, M. J., & Cyclospora Working, G. (1999). The return of
Cyclospora in 1997: Another outbreak of cyclosporiasis in North America
associated with imported raspberries. Annals of Internal Medicine, 130(3), 210-+.

Ho, A. Y., Lopez, A. S., Eberhart, M. G., Levenson, R., Finkel, B. S., da Silva, A. J., . . .
Herwaldt, B. L. (2002). Outbreak of cyclosporiasis associated with imported
raspberries, Philadelphia, Pennsylvania, 2000. Emerging Infectious Diseases,
8(8), 783-788.

Knudsen, D. M., Yamamoto, S. A., & Harris, L. J. (2001). Survival of Salmonella spp.
and Escherichia coli O157:H7 on fresh and frozen strawberries. Journal of Food
Protection, 64(10), 1483-1488.

Kotwal, G., & Cannon, J. L. (2014). Environmental persistence and transfer of enteric
viruses. Current Opinion in Virology, 4, 37-43. doi:10.1016/j.coviro.2013.12.003

Masse, D. I., Masse, L., Topp, E., Seguin, G., Ortega, L. M., Scott, A., & Pariseau, E.
(2011). Maintenance strategies for on-site water disinfection by ultraviolet lamps
on dairy farms. Water Quality Research Journal of Canada, 46(1), 2-12.
doi:10.2166/wqrjc.2011.014
Petterson, S. R., & Ashbolt, N. J. (2001). Viral risks associated with wastewater reuse:
modeling virus persistence on wastewater irrigated salad crops. Water Science
and Technology, 43(12), 23-26.
 82

Petterson, S. R., Teunis, P. F. M., & Ashbolt, N. J. (2001). Modeling virus inactivation on
salad crops using microbial count data. Risk Analysis, 21(6), 1097-1108.
doi:10.1111/0272-4332.216178

Rodrigues, R. D., Loiko, M. R., de Paula, C. M. D., Hessel, C. T., Jacxsens, L.,
Uyttendaele, M., . . . Tondo, E. C. (2014). Microbiological contamination linked
to implementation of good agricultural practices in the production of organic
lettuce in Southern Brazil. Food Control, 42, 152-164.
doi:10.1016/j.foodcont.2014.01.043

Stine, S. W., Song, I., Choi, C. Y., & Gerba, C. P. (2005). Effect of relative humidity on
preharvest survival of bacterial and viral pathogens on the surface of cantaloupe,
lettuce, and bell peppers. Journal of Food Protection, 68(7), 1352-1358.

Stine, S. W., Song, I., Choi, C. Y., & Gerba, C. P. (2011). Application of Pesticide
Sprays to Fresh Produce: A Risk Assessment for Hepatitis A and Salmonella.
Food and Environmental Virology, 3(2), 86-91. doi:10.1007/s12560-011-9061-x
USEPA., 2011. Exposure Factors Handbook. United States Environmental Protection
Agency, Washington.
 83

CHAPTER 3: SYSTEMATIC REVIEW OF SPORE FORMING BACTERIA IN

MILK

I. ABSTRACT

Approximately one third of the produced fluid milk in the United States is lost annually.

One important factor contributing to the loss is the contamination with spore-forming

bacteria, which can not only survive the pasteurization process, but also grow under

refrigeration conditions resulting in subsequent spoilage. The objective of this study is to

describe the population dynamics of spore-forming bacteria and spores in milk from farm

to packing plant through a systematic review approach. A database search was conducted

to identify, appraise, and summarize primary research studies that describe the prevalence

and/or concentration of spore-forming bacteria and spores at more than one

production/processing point in the same study. Literature searches retrieved 9,778

citations, among which data were extracted from 31 relevant citations for meta-analysis.

Due to variant milk sampling points recorded in citations, we standardized the sampling

points by clustering similar ones as follows: Milking machine, Raw milk, Bulk tank,

Transportation, Silo, Pasteurized milk and Packaged milk. Bacillus cereus was the most

reported organism. Concentration data were more abundant with 582 data points for both

vegetative cells and spores, compared to prevalence data with 68 points. In general, great

heterogeneity was observed among studies in the contamination of milk samples. Spore

concentrations remain stable until pasteurization, in a range of 0-2.5 log spores/ml. After

pasteurization, spore concentrations decrease in accordance with an increase in vegetative

cells. Although considerable research has been conducted on this topic, there are limited

studies to holistically describe the population dynamics of spore-forming bacteria under

 84

the current milk production system. Meta-regression analysis indicates that moderators

Steps (in the milk chain), Season and Year of Publication explains 65.71% of

heterogeneity for cells and 35.11% for spores. Findings of this study can provide insights

regarding steps where spore-forming bacteria could be introduced for potential effective

management, as well as further research needs to increase the quality and shelf life of

milk products in the United States. This project demonstrated that the outputs of

Systematic Review can feed the decision making process, through simple and clear

recommendations to the risk managers using a high-level evidence synthesis analysis

procedure.

II. INTRODUCTION

The previous chapter shows the useful application of risk assessment in food safety

protection by a collaborative project of assessing microbial risks of raspberry products in

Chile. In this chapter, the approach of systematic review is demonstrated via a case study

of evaluating the changes in spore-forming bacteria along milk supply chain.

In the milk production process, contamination with microorganisms is the most important

hurdle to overcome to provide safe milk products with long shelf life. Microorganisms

that create spores, referred to as spore-forming bacteria throughout this paper, can persist

along the downstream processing. This is due to its capacity of spores to resist

pasteurization temperatures; leading to microbial growth and premature spoilage. Cotter

and colleagues classified spore-forming bacteria in two groups. The first are the aerobic

psychrotrophic thermophilic spore formers such as B. cereus, Paenibacillus sp. and

Geobacillus stearothermophilus. The second ones are the anaerobic psychrotrophic

 85

thermophilic spore-forming bacteria, such as C. botulinum and C. perfringens. (Cotter et

al, 2015)

Spore-forming bacteria have been declared by the USDA and FDA to be the greatest

threat to dairy products in terms of spoilage (Hull et al., 1992). The spores of these

organisms, under the heat treatment of milk (e.g UHT), trigger the growth of its

vegetative form. The subsequent growth of these microbes will generate the secretion of

different thermostable lipolytic and hydrolytic enzymes that will breakdown the major

constituents of milk (Samaržija et al, 2012). Under these circumstances, milk spoilage

results and follows economic losses to farmers and processors. On the other hand, Gram-

positive spore-forming bacteria such as Bacillus cereus, produce enterotoxins which can

cause diarrhea and emetic disease due to food poisoning (Lindabk and Granum 2006).

Therefore, the potential contamination of spore-forming bacteria is a very important issue

that the dairy industry is aware of and constantly tries to address using exhaustive

hygiene and preventive control programs, such as HACCP and Good Manufacturing

Practices. Of particular interest to the milk industry are the spore-forming psychrotrophic

bacteria, which are able to grow at 7˚C or less, regardless of their optimal temperature of

growth (International Dairy Association, 1976) and synthetize thermoresistant spores.

Spore-forming bacteria can be introduced through multiple points along the liquid milk

production chain. The initial contamination starts in the milking facilities. Teat skin is

considered one of the major sources of spores in raw milk (McKinnon and Pettipher,

1983, Samaržija et al, 2012). It has also been documented that the number of spores

present in milk is significantly correlated to the degree of soil contamination on teats

(Christiansson et al, 1999), which indicates the significance of soil and dust attached to
 86

the teat skin contributing to the spore-forming bacteria contamination in raw milk. The

bulk milk storage tanks, pipelines and filling machines during processing procedures are

also key contamination sources, via the formation of biofilms on the food-contact

surfaces. Most of the spore-forming bacteria are able to create biofilms, which are very

resistant to temperature and sanitation, therefore generating an additional hurdle to the

industry.

Significant research has been conducted to develop the modern interventions to prevent

microbiological contamination, which are contained at the farm and processing level. The

application of Good Farm Managing Practices is critical to achieve low spore

contamination of raw milk. While the dairy industry relies on pasteurization to achieve a

reduction in the number of pathogenic and spoilage microorganism, pasteurization is

ineffective against spores (Cotter et al., 2015, Gleeson et al., 2013). Usually the research

focuses on specific points, but a limited number of studies have reported the cumulative

impact of control efforts over the entire system. In addition, research papers quantifying

the contamination of spore-forming bacteria in milk are available, but data with great

heterogeneity may be reported depending on study design, size and quality. Holistic and

systematic understanding of the dynamics of populations of spore-forming bacteria

throughout the whole milk supply chain is a very valued information set that no research

group has addressed, as most of the efforts are put in one or few steps.

In both situations, systematic review (SR) can facilitate the data collection conducted in a

structured and comprehensive process to identify data gaps and to fully capture the

naturally occurring variations among studies. Differing from narrative review, SR uses a

structured research protocol to minimize selection bias and evaluate data quality. Data
 87

extracted from independent studies selected in SR are commonly synthesized by meta-

analysis (MA), which is a subset of SR to use statistical approaches to combine the

results from multiple studies to develop a single conclusion with greater statistical power

over individual studies. SR, together with MA, can independently address research

questions by synthesizing relevant scientific evidence and also result in quantified

estimates that are suitable for quantitative microbial risk assessment (QMRA) model

parameterization to inform sound food safety risk management decision makings. The

use of results from SR and MA will increase the confidence in the QMRA model input

estimates and subsequent risk predictions, compared to using the “author-picked” data.

The present study was aimed at answering two research questions aided by SR: i) What

are the magnitudes of the changes in prevalence and/or concentration of spore-forming

bacteria and spores across steps along the pasteurized milk supply chain, and, ii) what are

the factors that could explain the variability of prevalence and/concentration of spore-

forming bacteria and spores in the intermediate and end milk products. Since the

information to resolve these questions was collected in the farm-to-processing

continuum, findings of this study will indicate the cumulative efficacy of the agricultural

and manufacturing practices employed in the current milk supply system in controlling

spore-forming bacteria. In this study, we report our first findings focused on spore-

forming bacteria dynamics along the pasteurized liquid milk supply chain.

III. MATERIALS AND METHODS

2.1 Search strategy

In consultation with the University of Nebraska – Lincoln subject specialist for Food

Science and Technology, Veterinary and Biomedical Sciences, a search strategy was
 88

developed using different key words and syntax. The databases used were: Food Science

and Technology Abstracts, Centre for Agriculture and Bioscience International database

(CABI), MEDLINE®, BIOSIS Previews, Biological Abstracts and the Web of Science.

The initial searches were narrow and specific, containing keywords that made reference

to food products, spore-forming bacteria related terms and specific bacterial species. An

initial screening of those results revealed that potential relevant manuscripts were being

discarded. After testing several search strategies, a search strategy utilizing more general

terms was determined appropriate to prevent losing relevant studies. A summary of the

search strategy for each database is shown in Table 3.1. Proceedings of conferences were

included when the full text was available. This study started on March 2015 and was

finished in June 2016.

 89

Table 3.1. Summary of the search strategies for the electronic databases.

Database name Search strategy

spore* OR "bacterial spores" OR sporeformer* OR "spore
former" OR spore-former* OR sporeforming OR spore-forming
CABI, Web of Science and OR "spore forming" OR endospore*
Biological Abstracts AND "milk products" OR milk OR "ice cream" OR cheese* OR
cream OR butter OR yogurt OR yoghurt OR dairy.

milk OR milks OR "ice cream" OR "ice creams" OR cheese OR

cheeses OR butter OR yogurt OR yoghurt OR cream OR dairy
OR dairy products [MeSH]
AND
PubMed
spore OR spores OR "spore forming" OR sporeform* OR spore-
form* OR "spore former" OR "spore formers" OR endospore OR
endospores OR spores, bacterial [MeSH]

milk OR milks OR "ice cream" OR "ice creams" OR cheese OR

cheeses OR butter OR yogurt OR yoghurt OR cream OR dairy
Biosis Citation AND spore* OR “spore forming” OR sporeform* OR spore-
form* OR "spore former" OR “spore formers” OR endospore*

spore* OR "bacterial spores" OR sporeformer* OR "spore

former" OR spore-former* OR sporeforming OR spore-forming
Food Science and
OR "spore forming" OR endospore* AND "dairy products" OR
Technology Abstracts
milk OR "ice cream" OR cheese* OR cream OR butter OR
yogurt OR yoghurt OR dairy

2.2 Relevance screening

Two graduate-level students conducted independent relevance assessment of the initially

retrieved publications in three steps: 1) title screening, 2) abstract screening, 3)full-text

screening. The software EndNote X7® (Thomson Reuters, Toronto, Canada) was used to

manage the references.

 90

2.2.1 Title screening

Due to a large number of articles obtained and the broad search strategy selected, the title

screening was first conducted to remove retrieval noise and evident non-relevant articles,

such as “analysis of spore-forming bacteria in canned vegetables”.

2.2.2 Abstract screening

Primary research was included at this stage if the following information was covered,

including 1) English language; 2) data from countries with similar milk production

systems as the United States of America. (We consider all European countries, Australia,

New Zealand and Canada as having close characteristics as the United States); 3)

prevalence and/or concentration of; 4) cells and/or spores in milk samples on; 5) any step

in the milk chain supply system. Reviews were collected to be used later as a quality

check of our retrieved literature.

2.2.3 Full-text screening

The full-texts for the selected articles at the previous stage were collected for the final

screening. Using the online resources, subscriptions and interlibrary load service

available at the University of Nebraska-Lincoln, the full texts were downloaded and

stored in the Endnote reference library. Any article whose corresponding manuscript was

not retrievable was discarded at this stage.

Articles with available full-texts were further screened for data extraction and analysis, if

the following information were reported, including 1) data of nationally-occurring

contamination on, 2) at least one data point in the defined milk supply chain, 3)

concentration and/or prevalence of spoilage sporeforming organisms in either raw or

 91

pasteurized liquid milk with their respective sample sizes and 4) arithmetic mean

concentration and/or prevalence with sample sizes reported. The variance and sample

sizes are fundamental data needed to propose pooled estimates using MA tools.

(Cochrane Collaboration webpage, 2016)

Articles were excluded if they pertained solely to detection or challenge studies to

evaluate the efficacies of specific spore-forming bacteria/spore reduction. Our main focus

was on observational research that studied the populations of spore-forming bacteria

along the milk supply chain.

2.3 Data extraction

Relevant data were manually extracted, organized and stored in a spreadsheet. The

following information from each selected articles was extracted: first author, year of

publication, country where the study was conducted, study duration, study season,

bacterial species, sample size (volume), sample number, production step involved,

concentration/prevalence, detection method and its corresponding detection limit (when

available) and statistical descriptors (when available) such as median, range, standard

deviation, standard error and confidence intervals.

2.4 Standardization of milk supply steps

Due to the great heterogeneity of the studies, especially regarding sampling plans, the

data extraction and grouping process yielded several different datasets within the milk

supply chain. Different names among the manuscripts were combined into the same

processing step, thus, developing a standardized process for the milk production chain

was essential to group representative data and analyze it in a logical structure. Figure 3.1
 92

shows the standardized steps and an explanation of the milk supply chain proposed in this

study with their coverage.

Milking Includes all raw milk samples taken from the milking equipment
Machine during milk extraction.

FARM Raw Milk Includes all raw milk samples after the milking until filling the
bulk tank, such as milk samples from pipelines just before
reaching the raw milk bulk tank.

Includes all raw milk samples from the raw milk bulk tank at the
Bulk Tank farm before the transport from farm to the processing facility.

Includes all raw milk samples taken during the transport before
filling the silo at the processing facility.
Transport

Includes all raw milk samples taken between the arrival of raw
Silo milk from the transport, storage silo at the processing plant and
immediately before entering the pasteurizer.

Includes all pasteurized milk samples taken from the

PROCESSING Pasteurized pasteurizer, pipelines, storage tanks and fillers until
PLANT Milk immediately before packaging.

Includes all pasteurized milk samples from within any

Packaged package at the facility or destiny market that it’s associated
Milk with a certain dairy.

Figure 3.1. Flow chart of the standardized milk supply steps, with their coverage of
samples described in the retrieved articles

2.5 Definitions

For the purpose of delivering straightforward and consistent discussion and conclusion,

we propose the following definitions. A citation refers to a unique publication in which

data from the primary research was collected, analyzed, and reported by the article
 93

authors. Within a citation, data from multiple trials can be reported, which is referred to

as a study. Multiple studies can be present in a single citation. Stating these differences is

critical for the following descriptive and meta-analyses, which are based on the synthesis

of studies within the same and also from different citations.

2.6 Data analysis

In spite of the large number of results and research in this topic, few studies were

considered relevant to answer the research questions. The scarcity of statistical

descriptors further limited qualification of selected articles for meta-analyses. Therefore,

a descriptive approach was mainly used to analyze the data and informative plots were

developed to describe the observed trends and data gaps. Dot plots, lattice plots and

statistical descriptors such as minimum, maximum and quantiles were also obtained using

the R statistical software package (Vienna, Austria).

A pooled estimate in each step is fundamental for data synthetizing studies. Due to the

lack of statistical descriptors, specifically variance, we can’t provide a pooled estimate of

the concentrations. Nonetheless, we provided a weighted mean based on the sample size.

Random effects Meta-analyses were conducted, when possible, for prevalence data to

establish a proper combined estimate in each step. Random effects analysis, model

selection and meta-regression analysis were performed in R 3.1.3 using the “Meta” and

“Metafor” packages”.

The Cochrane Collaboration defines the chi-squared test for heterogeneity (Q) as: “it

assesses whether observed differences in results are compatible with chance alone”. To

quantify heterogeneity we used the I2 statistic which is calculated as Higgins et al, 2003

proposes:
 94

𝐼 2 = 100% ∗ (𝑄 − 𝑑𝑓)/𝑄 (1)

The purpose of meta-regression is to assess the impact of selected variables on the study

effect size, in this case, prevalence and concentration. Figure 3.2 shows the model

selection procedure. The model selection process and meta-regression analysis were

conducted using a modified version of the method proposed by Islam, (Islam et al, 2014).

Figure 3.2. Model selection procedure

IV. RESULTS AND DISCUSSION

3.1 Systematic review process

Figure 3.3 summarizes the systematic review process conducted for this study. The

search strategies retrieved 16,193 articles from six electronic databases. After

deduplication, 8,553 unique articles remained for relevance screening. Of the 8,553

citations, 7,930 were excluded during the title and abstract screening because the articles
 95

did not describe the primary research or were not deemed to be relevant based on the

inclusion criteria that was pre-determined. Of the 623 articles that passed the title and

abstract screening, another 503 articles were excluded either during or after full-text

collection process. The articles were excluded because the full text was unavailable (89

articles) or did not pass the inclusion criteria (414 articles). Finally, 31 articles were

deemed relevant and data was successfully extracted. Table 3.2 describes the data

collected from each selected citation.

 96

Table 3.2. Summary of the main characteristics of the citations that were included in the
data extraction process.
Spore-forming
Production Sample Cell Analytical Concentration
Reference Country bacteria
steps covered number stage method or prevalence
class/species
Buehner et al United States Raw milk – Bulk 738 Spores Thermophilic, Spore count and Concentration
(2014) tank and cells Mesophilic and Thermoduric
Total Spores. bacteria count
Thermophilic and
thermoduric
bacteria.
McAuley et al Australia Raw milk 15 Cells Bacillus cereus AS 5013.2- Prevalence
(2014) 2007; Standards
Australia 2007

Tabit et al South Africa Silo – Pasteurized Not Spores Bacillus BHI agar plates Concentration
(2011) milk – Packaged available sporothermodurans
milk
Bartoszewicz et Poland Silo – Pasteurized 44 Spores Bacillus cereus Egg yolk Concentration
al (2008) milk – Packaged precipitation on
milk MYP medium
Vissers et al Netherlands Bulk tank 137 Spores Bacillus cereus Dutch standard Concentration
(2007a) 6875 (NEN-
ISO, 1994)
Vissers et al Netherlands Raw milk 110 Spores Mesophilic spores Plate count milk Concentration
(2007b) agar
Vissers et al Netherlands Bulk tank 327 Spores Butyric acid bacteria Dutch Standard Concentration
(2007c) spores (NEN-ISO-
6877, 1994)

Magnusson et al Sweden Bulk tank 81 Spores Bacillus cereus Phase-contrast Concentration and
(2007) microscopy and Prevalence
plating on MYP
agar

Scheldeman et Belgium Raw milk 18 Spores Total spores Milk plate count Concentration
al (2005) agar (Oxoid)

Moussa- Algeria Milking machine 530 Spores Bacillus cereus AFNOR Prevalence
Boudjemaa et al – Raw milk – procedure
(2004) Bulk tank
Hanus et al Czech Republic Bulk tank 70 Cells Bacillus Standard ČSN Concentration
(2004) licheniformis, ISO 7932
Bacillus cereus,
Other bacilli and
Total bacilli.
Giffel et al Netherlands Bulk tank 25 Spores Aerobic spores PCMA Concentration
(2002)

Lukasova et al Czech Republic Raw milk – Bulk 576 Cells Bacillus cereus and MYP agar Concentration and
(2001) Tank Total Bacilli Prevalence

Eneroth et al Sweden Pasteurized milk – 168 Cells Bacillus cereus Blood agar plate Concentration
(2001) Packaged milk

Svensson et al Norway/Sweden Silo – Pasteurized 44 Cells Bacillus cereus Blood agar plate Concentration and
(2000) milk Prevalence

Svensson et al Norway/Sweden Silo – Pasteurized 98 Cells Bacillus cereus MYP and blood Concentration
(1999) milk – Packaged and agar
milk Spores
Mayr et al Germany Packaged milk Not Cells Psychrotrophic API50CHB Concentration
(1999) available Bacillus sp. and system
Mesophilic Bacillus
sp.
Lin et al (1998) Canada Silo – Pasteurized 232 Spores Bacillus cereus BHI plates Concentration and
milk – Packaged and Prevalence
milk Cells
 97

Table 3.2. (Continuation) Summary of the main characteristics of the citations that were included
in the data extraction process.
Spore-forming
Production Sample Cell Analytical Concentration
Reference Country bacteria
steps covered number stage method or prevalence
class/species
Boor et al United States Raw milk 855 Spores Mesophilic aerobic BHI plates Concentration
(1998) spores

Slaghuis et al Netherlands Raw milk – Bulk 1318 Spores Aerobic spores and Aerobic Spore Concentration and
(1997) tank Bacillus cereus Count Prevalence
spores and
Voges-
Proskauer on
Tryptic Soy
Agar (TSA)
Larsen et al Denmark Silo – Pasteurized 830 Spores Bacillus cereus Tryptose blood Concentration and
(1997) milk and agar Prevalence
Cells
Giffel et al Netherlands Transport – Silo – 388 Cells Bacillus cereus Voges- Prevalence
(1996) Pasteurized milk – Proskauer on
Packaged milk TSA

Christiansson et Sweden Raw milk 144 Spores Bacillus cereus Blood agar plate Concentration
al (1996)

Giffel et al Netherlands Raw milk Not Cells Bacillus cereus Voges- Prevalence
(1995) available Proskauer on
TSA

Sutherland, A. Milking machine - 951 Spores Aerobic Na+MnSO4 Concentration

D (1994) Scotland Bulk tank – psychrotrophic
Transport – Silo - spores,
Pasteurized milk Aerobic mesophilic
spores,

Griffiths et al Scotland Bulk tank, Silo, 113 Spores Psychrotrophic Psychrotrophic Concentration and
(1990) Pasteurized milk spores and spore colony Prevalence
Bacillus spp spores count (PSC)

Dasgupta, A Australia Bulk tank Not Spores Anaerobic spores RCM and Concentration
(1989) available and RCM-lactate +
C. tyrobutyricum LATA
McKinnon et al United Kingdom Bulk tank – 126 Spores Psychrotrophic Total spore Concentration
(1983) Transport – Silo – spores and count (TSC) and
Pasteurized milk Total spores PSC

Oterholm, B Norway Bulk tank 15480 Cells Anaerobic Weinzirl Prevalence

(1981) sporeformers method

Falkowski et al Poland Bulk tank – 300 Spores Thermophilic Kosmachev Concentration

(1978) Pasteurized milk streptomyces spores media

Saywell et al New Zealand Raw milk – Bulk 60 Spores C. tyrobutyricum RCM-L Prevalence
(1977) tank spores

3.2 Characteristics of the relevant citations and extracted data

Research described in the 31 citations were conducted worldwide, with the majority in

Europe (23), North America (3) and Australia and New Zealand (3). The citations were

published in a year range of 1977 to 2015. Samples sizes were very variable, from sizes

down to 15 samples and up to 15480. The sample size depended largely on the duration
 98

of the studies, which ranged from one week up to two years. In terms of data coverage,

the number of citations covering each processing steps were: 2 (7%) on the Milking

Machine, 13 (42%) on the Raw Milk, 17 (55%) on the Bulk Tank, 3 (10%) on the

transport, 6 (19%) on the Silo, 11 (35%) on the Pasteurized Milk and 7 (23%) on the

Packaged Milk.

Overall, concentration data are more abundant compared to prevalence data. As shown in

Table 3.3 and Table 3.4, spores concentrations at the standardized steps were reported

and synthesized, ranging from 11 to 161. For both prevalence and concentration data, the

results vary considerably within and between the processing steps (Tables 3.3 and 3.4).

The more extreme cases are spores for concentration data, especially in the Silo,

Pasteurized Milk and Packaged Milk.

 99

FSTA CABI MEDLINE BIOSIS Biological Abstracts Web of Science Core

1,927 3,858 1,097 4,251 3,754 1,306

Total
16,193

After automatic de-duplication through EndNote X7 Duplicates

9,778 6,415

After manual de-duplication Duplicates

8,553 1,225

Excluded After peer 1 title screening After peer 2 title screening Excluded
6,731 1,822 1,621 6,932

Merge results Duplicates

2,566 877

After peer 1 abstract screening After peer 2 abstract screening

806 1124

Merge results Duplicates and

623 non-relevant
1307

Not found using UNL’s Full text search Data extraction

available tools 534 31
89

Did not pass inclusion criteria

414

Figure 3.3. Process flow of studies being retrieved, screened, appraised, selected, data-
extracted in this systematic review and meta-analysis
 100

Table 3.3 Summary statistics of concentration data by Standardized Supply Chain (log CFU/ml)

Supply Chain Number of

Minimum 1st Quantile Median Mean 3rd Quantile Maximum
Step data points

Spores Cells Spores Cells Spores Cells Spores Cells Spores Cells Spores Cells Spores Cells

Milking
64 NA -2.3 NA -0.1 NA 0.55 NA 0.58 NA 1.58 NA 6.11 NA
machine

Raw milk 26 3 -1.39 1.4 1.49 1.82 1.73 2.24 1.59 2.16 1.9 2.54 3.74 2.84

Bulk tank 161 10 -2.3 0.91 -0.1 0.99 0.39 1.21 0.53 1.54 1.23 2.06 6.23 2.76

Transport 65 NA -2.3 NA 1.03 NA 2.01 NA 1.88 NA 2.45 NA 7 NA

Silo 92 6 -2.3 -2 -0.16 -1.93 1.38 -1.58 1.5 -0.66 2.38 0.74 7 1.7

Pasteurized milk 89 60 -2.3 1 0.67 1 1.98 1 1.86 1.97 2.38 2.62 7 5.7

Packaged milk 11 64 -1.4 -1.3 -1.35 1 -1.22 1.5 0.41 1.92 -0.04 2.54 6.74 6.7

100
 101

Table 3.4. Summary statistics of prevalence data by Standardized Supply Chain (% of positives)

Supply Chain Number of

Minimum 1st Quantile Median Mean 3rd Quantile Maximum
Step data points

Spores Cells Spores Cells Spores Cells Spores Cells Spores Cells Spores Cells Spores Cells

Milking machine 1 NA 26.32 NA 26.32 NA 26.32 NA 26.32 NA 26.32 NA 26.32 NA

Raw milk 13 8 0 0 10 18.25 15 23.5 23.13 23.54 40 31.25 53 40

Bulk tank 11 13 3 12 12.09 25 20 34 33.43 36.36 59 50 100 57

Transport NA 1 NA 35 NA 35 NA 35 NA 35 NA 35 NA 35

Silo 4 5 80 7 81.5 10 83.5 25.22 85 22.44 87 35 93 35

Pasteurized milk 4 7 76 55 82.75 56 89.5 61 87.25 61.86 94 67 94 71

Packaged milk 2 1 90 71 91.5 71 93 71 93 71 94.5 71 96 71

101
 102
3.3 Concentration of spore-forming bacteria along the milk supply chain

Concentration data was the most abundant in the selected studies with 582 data points

extracted from the publications. Figure 3.4 shows the distribution of concentration for

both the vegetative cell stage and spores. Table 3.5 shows a summary of the pooled

concentration estimates in each step.

Table 3.5. Concentration pooled estimates for each processing step, ND = No data
available
Step Cells (log cfu/ml) Spores (log cfu/ml)
Milking machine ND 0.58
Raw milk 2.34 1.34
Bulk Tank 2.35 0.43
Transport ND 1.67
Silo 0.06 1.59
Pasteurized milk 2.00 2.44
Packaged milk 2.65 3.30

As shown in Figure 3.4, the overall trend of weighted average keeps relatively stable for

concentration of both cells and spores. The concentration of spores remains stable

between 0-2.5 logs until milk is packaged, where we can see an increase in dispersed

data. The great heterogeneity of concentration data of spore-forming bacteria at the step

of Packaged Milk can be due to the fact that the studies that reported these data points are

very different in the study design, season, location and methods of estimating the

concentrations. For example, in the study from Lin and colleagues (Lin et al, 1998),

enrichment at 80˚C for 14 days was conducted before counting, whereas Bartoszewicz
 103
and colleagues (Bartoszewicz et al, 2008) enriched the sample only for 48 hours at 25˚C.

Differences in methodologies are one of the major issues to overcome when pooling the

data together and providing meaningful critical review of the results.

Concentration (log CFU/ml)

Standardized supply chain steps

Figure 3.4. Stacked box plot for the concentration of spore-forming bacteria throughout
the milk processing chain. The top chart gives information for vegetative cells and the
chart below for spores. The red line represents the weighted mean. The dot size
corresponds to the sample size associated to a particular dataset. The spread of the dots
corresponds to “jittering” to avoid excessive overlapping and improve visualization.

After pasteurization, spores stay somewhat stable but cells increase dramatically. This is

intuitive as it is commonly known that the vegetative cells do not survive a pasteurization

process, but spores will germinate as a result of a thermal shock. Nonetheless, there are

only eleven data points contributing to the Silo stage in cell concentration, as opposed to
 104
60 and 64 for pasteurized and packaged milk (Table 3.3), which in turn have more

consistent datasets.

Raw milk and Bulk tank counts of cells are within the same range of 1.5-2.5 logs but with

no data available in the Milking machine and the Transport which are the previous and

following steps, respectively. These data fit well with the previously described

concentration ranges in Pasteurized and Packaged milk.

As aforementioned, concentration across steps remain stable, which is either because

contamination entry points are limited to the farm mostly, such as teat contamination

(McKinnon, 1982), or because modern control procedures are moderately effective or

both. More data is needed in the packaged milk step particularly to study the fate of these

spores.

3.4 Prevalence of spore-forming bacteria along the milk supply chain

Prevalence data were scarce compared to concentration, with 70 data points, especially in

certain processing steps such as Milking Machine and Transport, where one data point or

less was available. It is noteworthy that a significant portion of studies were focused in

the Bulk Tank, both for spores and cells, with pooled sample sizes of 15492 and 848. As

shown in Figure 3.5, prevalence data have more data gaps which makes the analysis more

difficult to conduct, but it is shown that prevalence of spore-forming bacteria is

increasing as milk moves from the farm to the processing Plant.

Prevalence (%)
105

Standardized supply chain steps

Figure 3.5. Stacked box plot for prevalence of spore-forming bacteria throughout the
milk processing chain. The top chart gives information for vegetative cells and the chart
below for spores. Red lines show the pooled estimates from random effects analyses. The
spread of the dots correspond to “jittering” to avoid excessive overlapping and improve
visualization. When there are no red lines but data points available, there is no sample
size available to provide an estimate.
 106

Figures 3.6a-3.6d show individual study trends for both concentration and prevalence.

While trying to detect individual trends that would be otherwise hidden in the summary

charts on Figure 3.4 and Figure 3.5, we found out that before Transport, the individual

study trend indicates a stable prevalence and concentration, and in some cases a slight

reduction in prevalence. After the Transport, the individual study trends seems to be

stable but with a moderate increase. Although the trend is not dramatically increasing, it

is certainly shedding light on where the industry should focus their efforts to control the

growth and proliferation of spore-forming bacteria. As seen in Figure 3.5, within the

Processing Plant (after Transport) there are significant chances that spores and cells may

eventually rise, so even if concentration and prevalence might seem to be under control,

the results of the present Systematic Review suggest that the focus should be set before

and after pasteurization.

 107

10 = Giffel et al, 1996

11 = Larsen et al, 1997

19 = Lukasova et al, 2001

Standardized Supply Chain Step

Figure 3.6a. Plot of the trends for individual studies reporting prevalence in cells.

16 = Svensson et al, 1999

17 = Svensson et al, 2000

18 = Eneroth et al, 2001

19 = Lukasova et al, 2001

Standardized Supply Chain Step

Figure 3.6b. Plot of the trends for individual studies reporting concentrations in cells.
 108

1 = Saywell et al, 1977

12 = Slaghuis et al, 1997

14 = Lin et al, 1998

22 = Moussa-Boudjemaa et
al, 2004

Standardized Supply Chain Step

Figure 3.6c. Plot of the trends for individual studies reporting spore prevalence.
 109

2 = Falkowski et al, 1978

4 = McKinnon et al, 1983
6 = Griffiths et al, 1990
7 = Sutherland, A.D et al, 1997
12 = Slaghuis et al, 1997
14 = Lin et al, 1998
28 = Bartoszewicz et al, 2008
29 = Tabit et al, 2011

Standardized Supply Chain Step

Figure 3.6d. Plot of the trends for individual studies reporting spore concentration.

In Figure 3.6d, the data reported from the Silo-Pasteurized Milk-Packaged Milk steps is

variable and shows different trends. Lin et al (1998) results indicate a high spore

concentration of about 6 logs cfu/ml and continuously increasing along the supply chain.

On the other hand, Falkowski et al (1978), Tabit et (2011), Bartoszewicz et al (2008) and

Griffiths et al (1990) indicate a considerable lower concentration, of about 0.5 logs and

that is decreasing. The variability of this data has multiple reasons: detection method,

season and location of the study among others.

 110
3.4.1 Meta-analysis for prevalence

Random effects meta-analysis was conducted to estimate pooled prevalence through the

use of the meta() package in R. Figure 3.7 and Figure 3.8 shows several forests plots for

the prevalence data. For meta-analysis, data points without sample size reported were

discarded. Between study variance (tau squared) was always significant (P-value<0.1) so

random effects estimates where used, except in the cell prevalence in the Silo. To

estimate pooled prevalence estimates, sample size is needed and very often it was not

provided in the studies. Nonetheless, Table 3.6 shows the estimated prevalence when

sample size is available. Modern meta-analyses procedures takes into account within and

between study variability, so these estimates are much more powerful than normal

average estimates. For the last three steps, although there are enough data to provide a

pooled estimate, sample sizes are missing. In all cases but the Silo prevalence,

Heterogeneity was estimated to be extremely high, so we conducted meta-regression

analysis to look for the sources and propose a regression model that accounts for the most

heterogeneity possible.
 111

Table 3.6. Prevalence estimates pooled by random effect meta-analysis model for each
supply chain step, ND = No data available

95% Confidence Spores 95% Confidence

Step Cells (%)
Interval (%) Interval
Milking machine ND ND ND ND
Raw milk 14 2-63 23 16-32
Bulk Tank 36 28-45 23 11-42
Transport ND ND ND ND
Silo 33 21-49 ND ND
Pasteurized milk 58 54-62 ND ND
Packaged milk ND ND ND ND
 112

Figure 3.7 Forest plot of reported cell prevalence. Study refers to “Study” definition in
section 2.5. Studies can come from the same Citation or different. The vertical dashed
lines represent the estimates for the Fixed and Random effects models.
 113

Figure 3.8. Forest plot of reported spore prevalence. Study refers to “Study” definition
in section 2.5. Studies can come from the same Citation or different. The vertical dashed
lines represent the estimates for the Fixed and Random effects models.

3.4.2 Meta-regression analysis

Sources of heterogeneity can be detected through the use of this approach. The variables

identified in this study were: Location of the study, year of publication (clustered in a 10
 114
year range), season when the study was conducted, type of bacteria detected, the step in

the processing chain where the sample was taken and the detection method used.

For both cells and spores, the final model was Season, Step and Year of publication. In

cells, the meta-regression model explained 65.71% of heterogeneity, while in the spores

was 35.11%.

Seasonality has been reported as a critical factor in the variation of spore-forming

bacteria populations (Sutherland et al, 1994), whether it is increasing or decreasing, the

general consensus is that the season is a major force driving spore-forming bacteria

population along the milk chain. The step where the sample was taken is also relevant as

very different characteristics are present in different steps. Finally, Year of publication is

also critical as sampling plans and detection methods are being updated and perfected

along the years, generating significantly different results.

All these three moderators were expected to be relevant in the meta-regression analysis

but the detection method was a variable that would not be deemed as explaining

heterogeneity. This could be based on the fact that it is closely linked to the publication

year.

3.4.3 Significance for regulators and evidence-based policies.

Systematic Review is readily usable by Governments as it is a structured process and it is

recognized world-wide as a powerful tool to synthetize data. Although it requires some

statistical expertise and is time-consuming, it can be done successfully by looking at

 115
online resources and seizing the opportunities of creating strategic collaborations with the

Academia.

The outputs of the Systematic Review act as a source of evidence for policy makers and

also feed the Risk Assessment data gaps.

V. CONCLUSIONS

This study is the first systematic review in this field to our knowledge. Holistic

understanding of food processing systems is fundamental to provide bias free conclusions

and when proposing more focus on certain interventions or processing steps. Although

the outputs of a systematic review of this type is not investigating specific interventions

or practices in the dairy industry, it does give useful insight for researchers, policy

makers and the industry itself about where are the potential issues for controlling spore-

forming bacteria and evidently where the current system seems to be working well, in

order to refocus resources where needed.

In this particular Systematic Review project, the conclusions in relation to this thesis are:

1. There is a critical need for more research in this topic, especially in the steps where no

or very scarce data are available, such as Milking Machine, Raw Milk, Bulk tank milk

and Transport for cell concentration and Milking Machine, Raw milk, Transport and

Packaged milk for prevalence in both cells and spores. Not only are more data needed,

but also data with quantified variability.

2. Prevalence meta-regression analysis indicates that Year of Publication, Season and

Step are the moderators explaining 65.71% of heterogeneity in cells and 35.11% in
 116
spores. There is still a significant amount of heterogeneity yet to be explained. We

believe that in the first place, more data is needed in the steps where little information is

available and also to explore new variables such as Detection Limits, and Sampling

Plans. Regarding concentrations, more statistical descriptors are needed in the

publications retrieved to provide a pooled estimate for each step.

3. These results are very useful for establishing performance objectives, which provide the

dairy industry solid and easy to establish metrics to add another layer of assurance of

quality to their products. Performance objective is a term borrowed from food safety

sciences, which refers to a specific level that must be met in earlier steps in the food chain

to comply with a Food Safety Objective, which in turn consists of the “maximum frequency

and/or concentration of a hazard in a food at the time of consumption” (IMCSF, 2006).

These metrics can be easily converted to food quality levels that must be met, for example,

not to surpass a certain threshold, which was proposed using data from this present study.

4. To fully harness the potential of data synthesis technologies such as SR, it is highly

recommended for developing countries to form Government-Academia collaborations.

Academics usually have the resources and expertise but lack the data, which

Governments can provide by consulting their surveillance or regulation compliance

control systems. Governments in turn benefit from acquiring evidence to support their

decision making process that was created using high quality, robust and non-biased

methods to synthetize data.

 117
VI. REFERENCES

Bartoszewiez, M., B. M. Hansen, and I. Swiecicka. "The Members of the Bacillus Cereus

Group Are Commonly Present Contaminants of Fresh and Heat-Treated Milk."

Food Microbiology (London) 25, no. 4 (Jun 2008): 588-96.

Boor, K. J., D. P. Brown, S. C. Murphy, S. M. Kozlowski, and D. K. Bandler.

"Microbiological and Chemical Quality of Raw Milk in New York State." Journal

of Dairy Science 81, no. 6 (June 1998): 1743-48.

Buehner, K. P., Anand Sanjeev, and A. Garcia. "Prevalence of Thermoduric Bacteria and

Spores on 10 Midwest Dairy Farms." Journal of Dairy Science 97, no. 11 (2014):

6777-84.

Christiansson, A., J. Nilsson, J. Bertilsson, and Federat Int Dairy. Bacillus Cereus Spores

in Raw Milk. Factors Affecting the Contamination of Milk During the Grazing

Period. Symposium on Bacteriological Quality of Raw Milk, Proceedings. 1996.

Cochrane Colaboration,

http://handbook.cochrane.org/chapter_9/9_analysing_data_and_undertaking_meta

_analyses.htm

Conor J. Doyle, David Gleesonc, Kieran Jordan, Tom P. Beresford, R. Paul Ross, Gerald

F. Fitzgerald, Paul D. Cotter. "Anaerobic sporeformers and their significance with

respect to milk and dairy products". International Journal of Food Microbiology

197 (2015) 77–87

 118
Dasgupta, A. P., and R. R. Hull. "Late Blowing of Swiss Cheese Incidence of

Clostridium-Tyrobutyricum in Manufacturing Milk." Australian Journal of Dairy

Technology 44, no. 2 (1989): 82-87.

Elmoslemany, A. M., G. P. Keefe, I. R. Dohoo, and B. M. Jayarao. "Risk Factors for

Bacteriological Quality of Bulk Tank Milk in Prince Edward Island Dairy Herds.

Part 2: Bacteria Count-Specific Risk Factors." [In English]. Journal of Dairy

Science 92, no. 6 (Jun 2009): 2644-52.

Eneroth, Asa, Birgitta Svensson, Goran Molin, and Anders Christiansson.

"Contamination of Pasteurized Milk by Bacillus Cereus in the Filling Machine."

Journal of Dairy Research 68, no. 2 (May 2001): 189-96.

Falkowski, J. "Occurrence and Thermo Resistance of Thermophilic Streptomycetes

Spores in Milk." Zentralblatt fuer Bakteriologie Parasitenkunde

Infektionskrankheiten und Hygiene Erste Abteilung Originale Reihe B Hygiene

Krankenhaushygiene Betriebshygiene Praeventive Medizin 167, no. 1-2 (1978):

171-76.

Giffel, M. C. te, A. Wagendorp, A. Herrewegh, and F. Driehuis. "Bacterial Spores in

Silage and Raw Milk." Antonie van Leeuwenhoek 81, no. 1/4 (2002): 625-30.

Gleeson, D., O'Connell, A., Jordan, K., 2013. "Review of potential sources and control of

thermoduric bacteria in bulk-tank milk". Irish J. Agric. Food Res. 52 (2), 217–

227.
 119
Griffiths, M. W., and J. D. Phillips. "Incidence Source and Some Properties of

Psychrotrophic Bacillus-Spp Found in Raw and Pasteurized Milk." Journal of the

Society of Dairy Technology 43, no. 3 (1990): 62-66.

Hanus, O., J. Frelich, M. Vyletelova, P. Roubal, Z. Vorlicek, and R. Jedelska.

"Technologically Difficult, Pathogenic and Food Risky Bacterial Contamination

of Raw Milk and Other Materials from Dairy Cow Herds." Czech Journal of

Animal Science 49, no. 11 (2004): 489-99.

Higgins, J, Thompson, S.G, Deeks, J.J and Altman D.G. Measuring inconsistency in

meta-analyses. BMJ. 2003 Sep 6; 327(7414): 557–560.

Hull, R.R., Toyne, S., Haynes, I.N., Lehmann, F.L. Thermoduric bacteria: A re-emerging

problem in cheesemaking. Aust. J. Dairy Technol. 1992;47:91–94.

Larsen, H. D., and K. Jorgensen. "The Occurrence of Bacillus Cereus in Danish

Pasteurized Milk." International Journal of Food Microbiology 34, no. 2 (1997):

179-86.

Lin, S., H. Schraft, J. A. Odumeru, and M. W. Griffiths. "Identification of Contamination

Sources of Bacillus Cereus in Pasteurized Milk." International Journal of Food

Microbiology 43, no. 3 (Sept. 8 1998): 159-71.

Lukasova, J., J. Vyhnalkova, and Z. Pacova. "Bacillus Species in Raw Milk and in the

Farm Environment." Milchwissenschaft 56, no. 11 (2001): 609-11.

 120
Magnusson, M., A. Christiansson, and B. Svensson. "Bacillus Cereus Spores During

Housing of Dairy Cows: Factors Affecting Contamination of Raw Milk." Journal

of Dairy Science 90, no. 6 (Jun 2007): 2745-54.

Mayr, R., I. Eppert, and S. Scherer. "Incidence and Identification of Psychrotrophic (7°C-

Tolerant) Bacillus Spp. In German Htst Pasteurized Milk." Milchwissenschaft 54,

no. 1 (1999): 26-29.

McAuley, Catherine M., Kate McMillan, Sean C. Moore, Narelle Fegan, and Edward M.

Fox. "Prevalence and Characterization of Food Borne Pathogens from Australian

Dairy Farm Environments." Journal of Dairy Science 97, no. 12 (Dec 2014):

7402-12.

McKinnon, C. H., and G. L. Pettipher. "A Survey of Sources of Heat-Resistant Bacteria

in Milk with Particular Reference to Psychrotrophic Spore-Forming Bacteria."

Journal of Dairy Research 50, no. 2 (1983): 163-70.

Moussa-Boudjemaa, B., M. Kihal, M. Lopez, and J. Gonzalez. "The Incidence of Bacillus

Cereus Spores in Algerian Raw Milk: A Study of the Chief Sources of

Contamination." Archiv Fur Lebensmittelhygiene 55, no. 4 (Jul-Aug 2004): 94-96.

Oterholm, B., and I. Engan-Skei. "Anaerobic Sporeformers and the Milking

Environment." Kieler Milchwirtschaftliche Forschungsberichte 33, no. 4 (1981):

337-42.

R Core Team (2015). R: A language and environment for statistical

computing. R Foundation for Statistical Computing, Vienna, Austria.

 121
URL https://www.R-project.org/.

Saywell, B. A., M. A. McGee, and C. M. Veitch. "The Incidence of Clostridium-

Tyrobutyricum in a Factory Milk Supply." New Zealand Journal of Dairy Science and

Technology 12, no. 4 (1977): 274-75.

Scheldeman, Patsy, Annelies Pil, Lieve Herman, Paul De Vos, and Marc Heyndrickx.

"Incidence and Diversity of Potentially Highly Heat-Resistant Spores Isolated at

Dairy Farms." Applied and Environmental Microbiology 71, no. 3 (Mar 2005):

1480-94.

Slaghuis, Betsie A., Meike C. Te Giffel, Rijkelt R. Beumer, and Geert Andre. "Effect of

Pasturing on the Incidence of Bacillus Cereus Spores in Raw Milk." International

Dairy Journal 7, no. 4 (April 1997): 201-05.

Sutherland, A. D., and R. Murdoch. "Seasonal Occurrence of Psychrotrophic Bacillus

Species in Raw Milk, and Studies on the Interactions with Mesophilic Bacillus

Sp." International Journal of Food Microbiology 21, no. 4 (1994): 279-92.

Svensson, Birgitta, Asa Eneroth, Johanne Brendehaug, and Anders Christiansson.

"Investigation of Bacillus Cereus Contamination Sites in a Dairy Plant with Rapd-

Pcr." International Dairy Journal 9, no. 12 (1999): 903-12.

Svensson, Birgitta, Asa Eneroth, Johanne Brendehaug, Goran Molin, and Anders

Christiansson. "Involvement of a Pasteurizer in the Contamination of Milk by

Bacillus Cereus in a Commercial Dairy Plant." Journal of Dairy Research 67, no.

3 (August 2000): 455-60.

 122
Tabit, F. T., and E. M. Buys. "Incidence and Survival of Bacillus Sporothermodurans

During Processing of Uht Milk." British Food Journal 113, no. 4 (2011):

1906684.

Te Giffel, M. C., R. R. Beumer, M. H. Bonestroo, and F. M. Rombouts. "Incidence and

Characterization of Bacillus Cereus in Two Dairy Processing Plants." Netherlands

Milk and Dairy Journal 50, no. 4 (1996): 479-92.

Tegiffel, M. C., R. R. Beumer, B. A. Slaghuis, and F. M. Rombouts. "Occurrence and

Characterization of (Psychrotrophic) Bacillus-Cereus on Farms in the

Netherlands." Netherlands Milk and Dairy Journal 49, no. 2-3 (1995): 125-38.

Vissers, M. M. M., F. Driehuis, M. C. Te Giffel, P. De Jong, and J. M. G. Lankveld.

"Minimizing the Level of Butyric Acid Bacteria Spores in Farm Tank Milk."

Journal of Dairy Science 90, no. 7 (Jul 2007): 3278-85.

Vissers, M. M. M., F. Driehuis, M. C. te Giffel, P. de Jong, and J. M. G. Lankveld.

"Quantification of the Transmission of Microorganisms to Milk Via Dirt Attached

to the Exterior of Teats." Journal of Dairy Science 90, no. 8 (2007): 3579-82.

Vissers, M. M. M., M. C. Te Giffel, F. Driehuis, P. de Jong, and J. M. G. Lankveld.

"Minimizing the Level of Bacillus Cereus Spores in Farm Tank Milk." Journal of

Dairy Science 90, no. 7 (2007): 3286-93.

 123
SUMMARY AND FUTURE DIRECTION

In this thesis, two cases studies were performed to understand how can two commonly

used research tools such as Risk Assessment and Systematic Review in food safety can

feed the decision making process, in developing countries were technology and research

itself is as not as developed compared to the United States or the European Union.

To show the application of risk assessment in food safety regulatory decision making

procedure, a collaborative project with the Chilean Food Quality and Safety Agency to

assess the risks of raspberry production of Chilean farmers was conducted. Regarding

the Systematic Review application in agri-food field, it was demonstrated through a case

study of evaluating the contamination of spore-forming bacteria along the milk supply

chain, which can be extended to address food safety questions of other hazard-food pairs.

For example, the systematic review approach can be used to fill up the data gaps and

further improve the risk assessment model of the microbial contamination in Chilean

raspberry products by reducing the parameter uncertainty involved. On the other hand,

Risk Assessment can tell the Agencies which are the production steps that needs

improvements and focused allocation of resources or new regulations. It also indicates in

a visual and simple way which are the main factors who are driving the risk along a

certain process flow. The ability to evaluate scenarios and interventions in-silico gives

Governments unprecedented opportunities to have a wide arrange of scientifically

assessed recommendations and potential interventions to improve whatever process is

being assessed, without the need of experimentation, field trials or further data collection.
 124
The focus of this thesis was set in the method investigation and demonstration, and how a

non-scientific stakeholder can benefit from the results of these high-end scientific

procedures. Both activities delivered simple to understand and sound evidence, although

they were performed under strict scientific procedures and state-of-art knowledge. The

collaboration between Academia and Government was fundamental in achieving these

accomplishments, since it harness the comparative advantages of each party, creating

synergies and successfully delivering evidence that is ready to be used for regulation and

policy making.

This thesis can serve as the basis of several different projects, for example, turning Risk

Analysis and Systematic Reviews procedures presented in this thesis into guidelines for

developing countries on how to conduct these processes. For this purpose, it is very

important to design it in collaboration with the Government Agencies as they know their

limitations and the best way to convey these topics to their target audiences.

This successful experience can be replicated in other developing countries, specifically

making Chile a strategical center of training in creating these collaborations and how to

bridge the gap between science and policy in developing countries.

 125
ANNEX I
STUDY SURVEY
FARMS

Region: ______________________________________

Municipality: ______________________________________

Location: ______________________________________

Geographical Coordinates (WGS 84) X: ______________ Y: ____________

0.a) Farm size: __________ ha

0.b) Average production: __________ kg/season
Mark with an x the way you trade your raspberries:

0.c) Collection Center ☐ Sells to intermediary ☐ Direct sale to packing ☐ Local sells
☐

Please answer this questions in the simplest way possible. If you don’t have detailed
information, please provide a simple estimate.

1. IRRIGATION PRACTICES
1.a) What irrigation type you use?

Drip ☐ Surface ☐ Furrow ☐ Other ☐

Other: _________________________________________________________
1.b) During the growth of the fruits, how often you irrigate?

Daily ☐ _____ per week ☐ Other ☐

Other:_______________________________
1.c) How many times a day?
____ times a day
Other:_______________________________
 126

1.d) How much water you use per irrigation event? (Approximate flow).

_______ per hectare ☐ farm total ☐

1.e) What is the source of the irrigation water?

Well ☐ Dike ☐ Ferris ☐ Deep well ☐ Other ☐

Other: _______________________________________

2. PESTICIDE APPLICATIONS

How many times and how often does pesticides had been applied during the flowering
and fruit formation during the present season? (If possible, provide a simple
description on pesticide application)
2.a.1) Number of applications: _____
2.a.2) Time between applications: ____ days
2.b) What type of water you use for pesticide applications?

Well ☐ Dike ☐ Ferris ☐ Deep well ☐ Potable ☐ Other ☐

Other: ______________________________________________
2.c) What type of pesticide application system you use?

Pulverize ☐ Knapsack sprayer ☐ Nebulizer ☐ Dredger ☐ Other ☐

Other: ______________________________________________
Please indicate the type of pesticide and the amount used (pesticide + water) in the
farm per application.

Pesticide name Active Ingredient Liter/Application

2.d.1) 2.d.2) 2.d.3)
 127

2.e) How much times goes by between the last application and the harvest?
(withholding period)
_____ days.

3. SOIL AMENDMENTS PRACTICES

Do you apply any soil amendment procedure?

3.a.1)Yes ☐ No ☐

If yes, what type?

3.a.2) Compost ☐ Sludge ☐ Manure ☐ Other ☐

Other: _____________________________________________________________
3.b) When and how often are these procedures applied?

_______________________________________________________________

3.c) How much do you apply? (kg per hectare, per farm o any information available)

____________________________________________________________________

3.d) How many days goes by between application and harvest?

_____ days

4. HARVEST PRACTICES
4.a) What kind of personal security/hygiene equipment are used in the harvest?
Safety footwear ☐ Gloves ☐ Apron ☐ Mask ☐

4.b) During the current season, have any worker been absent for diseases?

Yes ☐ No ☐
 128
4.c) If yes, for how long? (average)

____ days

4.d) Does any of these diseases had been food poisoning, diarrhea or vomit?

Yes ☐ No ☐

Before the harvest, are the trays meant for the harvest:

Washed?
4.e.1) Yes ☐ 4.e.2) With potable water ☐ Non- potable water ☐
No ☐

Disinfected?
4.e.3) Yes ☐ 4.e.4) Indicate chemical:_______________________
No ☐

5. ANIMAL PRESENCE IN THE FARM

5.a) Have you detected the presence of animals (mammals or birds) on the farm?
Yes ☐ No ☐
5.b) What type of animals?

Domestic mammals or birds ☐ Wild mammals or birds ☐

5.c) Do these animals come in direct contact with the fruits?

Yes ☐ No ☐

5.d) How often does the latter happen?

Daily ☐ Weekly ☐ Monthly ☐

5.e) Have you seen animal waste in direct contact with the fruits of harvest
equipment?
Yes ☐ No ☐

5.f) How often does the latter happen?

 129

Daily ☐ Weekly ☐ Monthly ☐

6. FARM TO PACKING TRANSPORT

6.a) How long does it take from the Collecting Center or Farm to the Processing
Plant?

________ hours ☐ minutes ☐

6.b) At which temperature are the raspberries usually transported?

_____ °C No refrigeration ☐

6.c) The shipment is carried:

Closed wagon ☐ Covered with loom ☐ Covered with raschel mesh ☐

Just tied without mesh or loom ☐ Other ☐

Other: _______________________________________

ANNEX II
STUDY SURVEY
1. COLLECTION CENTERS
 130
Region: ______________________________________

Municipality: ______________________________________

Location: ______________________________________

Geographical Coordinates (WGS 84) X: ______________ Y: ____________

Infrastructure:
Type of ceiling ______________________________
Type of floor _______________________________
Type of walls ______________________________

Closed ☐ Open ☐ (with or without access doors?)

0.a) Is it located alongside a raspberry farm Yes ☐ No ☐

0.b) Is it located in a location with no raspberry farm Yes ☐ No ☐

0.c) Average number of farmers that collects here by season _____
1.a) For the raspberries that come from a farm, how much time in average stays in the
Collecting Center?
______ days ☐ hours ☐

1.b) Is the same tray used in the harvest used in the Collection Center?

Yes ☐ No ☐

1.c) What is the average temperature of the Collection Center?

_____ °C

1.d) What is the storage capacity of the Collection Center? (Indicate the number of trays
and average weight of the tray with raspberries)

_____ trays _______ grams ☐ kilograms ☐

1.e) Is there any ventilation system? If yes, which type?

Yes ☐ No ☐
 131
Type: ________________________________

1.f) Have you ever detected the presence of animals (mammals or birds)?
Yes ☐ No ☐

1.g) What kind of animals?

Domestic mammals or birds ☐ Wild mammals or birds ☐ Pests ☐

1.h) Does this animals take direct contact with the fruits?
Yes ☐ No ☐

1.i) How often does the previous happen?

Daily ☐ Weekly ☐ Monthly ☐

1.j) Only answer this if the collected fruit comes from different farmers:

The fruit from different farmers is stored in different places?

1.j.1) Yes ☐ No ☐
1.j.2) In pallets ☐ Directly on the ground ☐ Other ☐

1.j.b) Is there a label that identifies the farm source on the trays?

Yes ☐ No ☐

1.j.c) Is there a label that identifies the farm source on the pallets?

Yes ☐ No ☐

2. TRANSPORT FROM COLLECTION CENTER TO PACKING OR

PROCESSING CENTER
 132
2.a) How long does it takes the transport from the Collection Center to the Processing
Plant or Packing Facility?

______ minutes ☐ hours ☐

2.b) What it the temperature in this process?

_____ °C No refrigeration ☐

2.c) Describe the transportation process:

Closed wagon ☐ Covered with canvas (or similar) ☐ Covered with raschel mesh ☐
Only tied and no cover ☐ Other ☐

Other: ________________________________

ANNEX III
 133
STUDY SURVEY
EXPORT PACKING

Region: ______________________________________

Municipality: ______________________________________

Location: ______________________________________

Geographical Coordinates (WGS 84) X: ______________ Y: ____________

1. RASPBERRIES RECEIVING

1.a) The place is:

Open ☐ Closed ☐ Under a ceiling ☐

1.b) The wait time is around (in minutes):

1.b.1) Max ____ 1.b.2) Min _____ 1.b.3) Average ___

1.c) Average temperature in unloading place: ______°C

2. FIRST COLD CHAMBER

2.a) Target temperature is:

______ °C

2.b) Time needed to reach target temperature:

______ minutes ☐ hours ☐

2.c) Time that the fruits stays here?

______ hours ☐ days ☐

3. OPERATIONS

3.a) How many shifts? (even if they work with different fruits)
 134

_____ shifts

3.b) How long does the shifts lasts?

_____ hours

3.c) What is the temperature inside the Packing area? (Temperature records)

3.c.1) Max ____ 3.c.2) Min _____ 3.c.3) Average _____

3.d) Generally, how long does it takes since the fruit exits the cold chamber and goes
through the first selection and goes into the freeze chamber?

_____ minutes

4. SANITATION

4.a) Do you conduct any Sanitation procedure?

Yes ☐ No ☐

4.b) How often you conduct these procedures?

Infrastructure/Equipment Routine cleaning Deep cleaning

Steel tabletops 4.1) 4.15)
Conveyor belt 4.2) 4.16)
Calibrators 4.3) 4.17)
Bins 4.4) 4.18)
Boxes transporter truck 4.5) 4.19)
Metal detector 4.6) 4.20)
Hands washing station 4.7) 4.21)
Precooling tunnel 4.8) 4.22)
Static tunnel 4.9) 4.23)
IQF Frost tunnel 4.10) 4.24)
Tray washer 4.11) 4.25)
Wash tub 4.12) 4.26)
Walls and ceiling 4.13) 4.27)
Trash bins 4.14) 4.28)
 135
4.c) Washing
4.c.1) Type of soap used: ______________________________
4.c.2) Dilution used: _______________

4.d) Disinfection
4.d.1) Name of chemical used: ______________________________
4.d.2) Concentration used: ________

4.e) Type of personal protection and/or hygiene that workers use.

Safety footwear ☐ Gloves ☐ Apron ☐ Mask ☐ Hat ☐ PVC apron ☐

4.f) How often are the work clothes changed?

______________________________

4.g) Do workers change clothes when the shift starts/end?

______________________________

4.h) During the current season: How many workers had shown symptoms related to a
possible foodborne illness, such as diarrhea?

___________ workers

4.i) Do you conduct a hands sampling procedure to look for fecal coliforms and
pathogens?
(If yes, please describe shortly the procedure, if it is done to all the personnel or only
some. Please describe the criteria that selects who is going to be sampled)

Yes ☐ No ☐

4.i.1) If yes, please describe as requested:

_____________________________________________________________________
_____________________________________________________________________
_____________________________________________________________________
_____________________________________________________________________

5. FREEZING PRACTICES

5.a) Target freezing temperature?

 136
______ °C

5.b) How much time is needed to reach the target temperature?

______ minutes ☐ hours ☐

5.c) How long does the fruits stay here?

______ hours ☐ days ☐

6. ENVIRONMENT

6.a) Is there any other sampling plan in the Process or Packing plants? (surface
contact materials and other surfaces for example)

Yes ☐ No ☐

6.a.1) If yes, please describe briefly:

_____________________________________________________________________

_____________________________________________________________________

7. TRANSPORT

7.a) Is there any disinfection and/or cleaning procedure applied to the trucks or cold
chambers, before loading?

7.a.1) Cleaning Yes ☐ No ☐

7.a.2) Disinfection Yes ☐ No ☐

7.b) Temperature of the loading room.

______ °C

7.c) Temperature of the cooling truck during transport

______ °C

7.d) Time taken to destination.

 137

7.d.1) Minimum
______ hours ☐ days ☐

7.d.2) Average
______ hours ☐ days ☐

7.d.3) Maximum
______ hours ☐ days ☐
 138

ANNEX IV: Expert Elicitation data spreadsheet.

Data Type of PREVALENCE Parameter

Explanation Microorganism Distribution Value
requested data Positives Total 1

Example 1 Distribution Hepatitis A distribution in water used for raspberry irrigation Hepatitis A - - Gamma Alfa 0.084
Example 2 Occurrence Time between last application and harvest - - - Laplace Mean 120
Example 3 Distribution E. coli concentration for groundwater in Chile E. coli - - Pert - -
Example 4 Prevalence E. coli prevalence in harvester's hands E. coli 6 40 - - -
E. coli or coliforms
Prevalence Microbiological prevalence in manure
Hepatitis A or norovirus
Contamination E. coli or coliforms
due to soil Distribution Microbiological distribution in manure
amendments Hepatitis A or norovirus
E. coli or coliforms
Occurrence Frequency that manure touches the fruits
Hepatitis A or norovirus
E. coli or coliforms
Contamination Prevalence Prevalence in trays
due to harvest Hepatitis A or norovirus
tray E. coli or coliforms
contamination Distribution Distribution in trays
Hepatitis A or norovirus

E. coli or coliforms
Prevalence Microbiological prevalence in animal waste
Hepatitis A or norovirus
Contamination E. coli or coliforms
due to animal Distribution Microbiological distribution in animal waste
contact Hepatitis A or norovirus
E. coli or coliforms
Occurrence Frequency that animal waste touches the fruits
Hepatitis A or norovirus

136
 139

ANNEX IV: Expert Elicitation data spreadsheet. (continuation)

Parameter Min Max

Value Mode Units Reference
2 value value

PCR-detectable
Beta 0.039 - - - units/Liter M. Bouwknegt et al 2015
SD 61.29 - - - Days ACHIPIA survey
- - 10^2 10^7 10^3 cfu/ml Expert elicitation

- - - - - - Aceituno et al, 2016