Food safety behavior in primary cook and health outcomes of household in ho chi minh city

This research’s objective is to find out the factors that modify the food safety practice of household primary food preparer and the effect of these behaviors on food poisoning risk.. Th

MINISTRY OF EDUCATION AND TRAINING UNIVERSITY OF ECONOMICS HO CHI MINH CITY

NGO HOANG TUAN HAI

FOOD SAFETY BEHAVIOR IN PRIMARY COOK AND HEALTH OUTCOMES OF HOUSEHOLD IN HO CHI MINH CITY

MASTER OF ECONOMICS THESIS

Ho Chi Minh City - Year 2016

UNIVERSITY OF ECONOMICS HO CHI MINH CITY

NGO HOANG TUAN HAI

FOOD SAFETY BEHAVIOR IN PRIMARY COOK AND HEALTH OUTCOMES OF HOUSEHOLD IN HO CHI MINH CITY

Major: Development Economics ID: 60310105

MASTER OF ECONOMICS THESIS

SUPERVISOR: DR PHAM KHANH NAM

Ho Chi Minh City - Year 2016

COMMITMENT

I commit the thesis “Food safety behavior of primary cook and health outcomes

of household in Ho Chi Minh city” is my own research

Except the references which are extracted in this thesis, there is no any others research or documents which is used in the thesis against regulatory

I would bear the full responsibility of my research The data, conclusion in this thesis is fidelity and not published in any research yet

Ho Chi Minh City, October 31st, 2016

Ngo Hoang Tuan Hai

TABLE OF CONTENT

COMMITMENT

TABLE OF CONTENT

ACKNOWLEDGEMENT

LIST OF ABBREVIATIONS

LIST OF FIGURES

LIST OF TABLES

CHAPTER 1 : INTRODUCTION 1

1.1 PROBLEM STATEMENTS 1

1.2 RESEARCH OBJECTIVES AND RESEARCH QUESTIONS 6

1.3 SCOPE OF RESEARCH 8

1.4 THESIS STRUCTURE 9

CHAPTER 2 : LITERATURE REVIEW 10

2.1 FOOD SAFETY AND FOOD-BORNE DISEASES 10

2.2 THE HEALTH BELIEF MODEL: 11

2.3 EMPIRICAL REVIEWS ON DRIVERS OF FOOD SAFETY PRACTICES: 13

CHAPTER 3 : RESEARCH METHODOLOGY 18

3.1 ANALYTIC FRAMEWORK 18

3.2 ECONOMETRIC MODELS 19

3.3 DATA 25

CHAPTER 4 : RESEARCH RESULTS 28

4.1 FOOD SAFETY PROBLEMS IN VIETNAM 28

4.2 DESCRIPTIVE STATISTICS 31

4.3 RESULTS FROM MULTIVARIATE PROBIT MODELS 39

4.4 RESULTS FROM PROPENSITY SCORE MATCHING MODEL 45

CHAPTER 5 : DISCUSSION AND IMPLIED POLICY 49

5.1 DISCUSSIONS AND CONCLUSIONS 49

5.2 POLICY IMPLICATION 50

5.3 LIMITATION AND IMPLICATIONS FOR FURTHER RESEARCH 51

Appendix 1: The correlation matrix of perception’s factors 53

Appendix 2: PCA result 54

Appendix 3: MVP regression (reduced form) 55

Appendix 4: MVP regression (original form) 56

Appendix 5: Poisson regression 57

Appendix 6: Questionaire form 58

References 67

ACKNOWLEDGEMENT

Firstly, I would like to express my gratitude to Dr Pham Khanh Nam, my supervisor, for all the suggestions, recommendations, knowledge and guidance that he did to support me to finish the thesis

Secondly, I am very grateful to doctor, MPH Nguyen Thi Huynh Mai, vice director of Safety Hygiene Food Branch of Ho Chi Minh city, for the permission as well as the advice to use the food safety data

Thirdly, I would like to thankful to my colleagues, my friend for all the encouragement and support they gave to me during the thesis processing

Lastly, my sincere thanks are all to the member of School of Economics – University of Economic Ho Chi Minh city for their effort to create the best environment for studying and researching for me as well as other students during the course

Ho Chi Minh city, October 31st, 2016

Ngo Hoang Tuan Hai

LIST OF ABBREVIATIONS

FBD: Food-borne disease

WHO: World Health Organization

FAO: Food and Agriculture Organization

HBM: Heal Belief Model

MVP: Multivariate Probit

KAP: Knowledge, Attitude and Practice

PSM: Propensity Score Matching

CDC: Center for Disease Control and Prevention

LIST OF FIGURES

Figure 1.1: The number of food-borne cases annually (WHO, 2015) 2

Figure 1.2: The number of death caused by FBD annually (WHO, 2015) 3

Figure 1.3: The burden of FBD (WHO, 2015) 4

Figure 2.1: Health belief Model Components and Linkages (Glanz et al, 2008) 13

Figure 3.1: The Health Belief Model application in food safety 19

Figure 4.1: The number of food poisoning cases in Vietnam (MOH, 2016) 28

Figure 4.2: The number of food poisoning outbreaks and death in Vietnam (MOH, 2015) 29

Figure 4.3: The number of food poisoning cases in HCM city (FSBDH, 2016) 30

Figure 4.4: The nonparametric relationship between food safety practice and knowledge, perception 38

LIST OF TABLES

Table 2.1: The concepts of Health Belief Model 11

Table 3.1: Variables’ description 23

Table 4.1: Demographic characteristics of participants (category variables) 31

Table 4.2: Demographic characteristics of participants (continuous variables) 32

Table 4.3: Factor analysis result 35

Table 4.4: Food safety practices 37

Table 4.5: MVP regression reduced form 39

Table 4.6: MVP regression original form 41

Table 4.7: Marginal effect after MVP regression 42

Table 4.8: Poisson regression 44

Table 4.9: Probit regression result 46

Table 4.10: Differences of continuous variables 47

Table 4.11: Correlations between binary variables and FBD 48

ABSTRACT

Nowadays food safety issue appeals a lot of attention from the global

organization to local authorities Each year, the food-borne diseases cause an enormous

burden on people heath as well as national’s economy As climate change suffered

country with the developing economy, Vietnam has to face to many food safety

challenges In recent years, the Vietnam government has invested much effort in order

to maintain the rate of economic growth while trying to improve the people’s health by

a lot of new law and institutions However, the food safety policy does not focus on

adjusting the consumer’s behavior As a result, the impact of individual’s practice on

their health was not determined clearly in Vietnamese community

This research’s objective is to find out the factors that modify the food safety

practice of household primary food preparer and the effect of these behaviors on food

poisoning risk The results show that the food safety knowledge and perception have

significant effect on household cook’s practices while their behaviors do not have the

explicit impact on the food poisoning’s risk

Keywords: food safety, food knowledge, perception and behavior, food poisoning

CHAPTER 1 : INTRODUCTION 1.1 PROBLEM STATEMENTS

The agricultural revolution has brought a large amount of food, rations for human and improved laborer’s health, a fundamental factor of human capital Therefore, food safety1 problem would give negative impact to the sustainable development of developing countries as well as the national’s security The fact that many diseases related to food demonstrated recently has driven many countries’ attention from food quantity to food safety

The usage of pesticides, chemical fertilizer and feedstuff improve agriculture’s capacity However, the overuse and misuse of them affect the quality of food In addition, the preservation and processing method as well as food additive exploitation both make food become less safety According to WHO (2015), the main factors cause food-borne disease are bacteria, virus, parasites, chemicals and toxins The most dedicated people of food-borne illness are children, pregnant women and the elder Due to the development of transportation and international trade, food safety issue is not only the problem of any countries but also a global problem For examples: Chinese milk scandal, New Zealand material milk crisis caused an enormous loss for the manufacturer and impinge on other countries as well as consumer’s health

1 World Health Organization (WHO) and Food and Agriculture Organization (FAO) defined:

“Food safety is the assurance that food will not cause harm to the consumer when it is prepared and eaten according to its intended use” (WHO and FAO, 2009, p 6).

Figure 1.1: The number of food-borne cases annually (WHO, 2015)

FBD occur in all countries in the world However, the developing regions, such as South East Asia and Africa, distributed the majority of food-borne cases while the developed areas, such as Europe and America, had the least number of food-borne cases An unexpected result is that although African countries are lack of food source and food safety control, this region had less amount of food-borne illness cases than the South East Asian region The reasons of this phenomenon may be due to the diversity

of high nutrition food and the tropical weather in the Asian area These two factors are the ideal condition for bacteria and other food risky factors to affect the human health

WHO (2015) estimated that every year food-borne disease cause almost 10% global populations to fall ill and responsible for 420,000 deaths (one third are children) Among all FBD, diarrheal diseases are the most common illnesses resulting from

0 100,000,000

unsafe food, accounted for half of global burden of FBD and made 550 million people

falling ill (including 220 million children), cause 230,000 deaths (96,000 children’s)

Figure 1.2: The number of death caused by FBD annually (WHO, 2015)

Similar with the number of food-borne cases, the number of death caused by FBD showed the same trend South East Asian and African regions continued to distribute the largest number of death cases while the European and American had the least amount of death by FBD The African region also had less number of death case than the South East Asian This consequence may be the effect of many international medical supports to the Africa than the Asian In addition, the distinction of physical strength of local population between these two regions is another fundamental reason The FBD caused the burden about 33 million DALYs2 Diarrheal diseases agents were the largest contributors, accounted for 18 million DALYs, 54% of total All three

2 DALYs: Disability-adjusted life year, a health gap measure that combines the years of life lost due to premature death (YLL) and the years lived with disability (YLD) from a disease or condition, for

figures 1.1, 1.2, 1.3 both indicated that South East Asian and African region’s food safety issue is severe and these areas suffered an enormous burden from FBD

However, despite of the number of death cases in Africa was less than Asia, the burden of FBD in this area was almost twice than the South East Asian and much more than other areas The European, Western Pacific and American total burden was almost equal to the South East Asia’s and half than the Africa’s burden These figures exposed the distance of the health care facilities as well as the food safety controlled policy between each areas and their impact of population’s health outcomes

Figure 1.3: The burden of FBD (WHO, 2015)

varying degrees of severity, making time itself the common metric for death and disability One DALY equates to one year of healthy life lost (WHO, 2015)

0 500

According to the region category of WHO, Vietnam is categorized in the Western Pacific areas In the general view of this organization, food-borne illness figure’s of this area is at the middle level of the world The similarity result repeated in Vietnam case However, this region includes various kinds of countries, from developed countries such as Australia, Japan, Republic of Korea to the developing ones such as Cambodia, Philippines, Vietnam Thus, the contribution of each country has a huge distinction The specific information of Vietnam case showed in Chapter 4

Although the developing countries suffer the most from FBD, the developed countries also have difficulty with food-borne illness According to Center for Disease Control and Prevention (CDC), The USA had 864 food-borne disease outbreaks, resulting in 13,246 illnesses, 712 hospitalizations, caused 21 deaths, and 21 food recalls in 2014 In detail, the majority of the food poisoning cases occurred in restaurant (485 cases, accounted for 65%), followed by private home (86 cases, accounted for 12%) The most popular causes of food-borne illness in 2014 was bacteria (149 cases, confirmed and suspected) distributed 22% of total cases Those figures proved that even the country with well-organized healthcare system and policy has to struggle with food poisoning

CDC also predicted that food safety issue would continue emerge in the future due to:

- Changes in our food production and supply, including more imported foods

- Changes in the environment leading to food contamination

- Better detection of multistate outbreaks

- New and emerging bacteria, toxins, and antibiotic resistance

- Changes in consumer preferences and habits

- Changes in the tests that diagnose foodborne illness

Despite of the wave of immigrants, the increase of trade exchange, the globalization problem, the food-borne disease is not the attention of any single country but it is the issue of the modern world

1.2 RESEARCH OBJECTIVES AND RESEARCH QUESTIONS

1.2.1 Benefits of the research

The diversity of food and food market in Vietnam create the convenience for household lady to purchase food Therefore, most of the Vietnamese’s households often cook and eating at home at least one meal per day For that reason, as well as the outside factor’s effects, the people’s behaviors also play an important role in their food poison possibility Those behaviors established from their socio-economic statement, individual characteristics and living condition In addition, the habit of cooking in Vietnam community transmitted from generations, thus the Asian tradition and agricultural culture caused a deep impact on the household cooking behaviors The role

of tradition and culture expose through the knowledge of food processing, preserving

as well as their practice in the kitchen

Some research about the Knowledge, Attitude and Practice (KAP) of the people

in Thua Thien Hue province (Duong, 2013), Ho Chi Minh city (Nguyen, 2010) both indicate the correlation between knowledge, attitude toward food safety and the food safety practice of individual at food factory, restaurant as well as household According

to the “10 golden principles in food processing” (MOH, 2005), the food safety behaviors include:

- Clean, tidy kitchen and the cooker surface

- Using waste basket with cover

- Toilet outside the kitchen

- Use clean water to handle food

- Separate well-done food and raw food in processing and preserving

- Use clean tool to prepare and divide food

- Clean vegetable precisely

- Not use forbidden food additives or out of date food

- Washing hand before cooking and after toileting

- Not smoke, spit out or nail polished while cooking

These golden principles were used frequently in many researches in Vietnam and were the outcome behavior in this thesis However, most of these researches were medical perspective so the socio-economic statement and individual characteristics are not well considered As a result, the effect of these factors on food safety behaviors has not assessed explicitly

On the other hand, the finding of those research merely indicated the current statement of food safety issue in the community without the relative analyzing of multi factors to confirm their impact Moreover, the participants of the above research did not participate in the annual survey In consequence, the effect of government policy and activities did not mention in the research

In order to review and confirm the factors which have impact on food safety behavior, this thesis’s objective is determining these factors as well as estimating their effect on individual behavior and predict their food-borne disease possibility These potential factors include the socio-economic status, the knowledge about food safety, the perception and the information source of each individual Whether the personal causes are determined, the government could develop the compatible strategy to adjust the people’s behavior, prevent those risky ones, minimized the individual food-borne diseases possibility in the public and its burden

1.2.2 General objectives

To analyze food safety behavior and health outcomes of household primary cook

Although the survey in this theme held annually, the kind of participants is distinct in every year, vary from the householder (in 2010 and 2013) to the restaurant worker (in 2012, 2014, 2015, 2016) In addition, the specific responders in the same kind is different from years, thus the latest data for the household is in 2013 and is not connected with the 2010 survey so thesis is not able to create the panel data to analyze Basing on this secondary data, the scope of research is the behavior and acute food poisoning statement of the community in Ho Chi Minh City in the period from March to April of 2013 to evaluate the impact of the factors Besides the descriptive statistics, econometric tools are the main methods used in the thesis, such as: factor analysis, multivariate probit, propensity score matching

1.4 THESIS STRUCTURE

Due to the available of the data, thesis is composed as the structure below:

- Chapter 1: Introduction about the research problem, the benefit and the scope of research This chapter present the general view of FBD and its burden in the world as well as the outline and objective of the thesis

- Chapter 2: Literature review This chapter review the concept’s definition and the previous research about the factors and models used in research which are the base

to create the analytic framework and method to analyze the effect of each component

- Chapter 3: Research methodology This chapter provide the framework and econometric tools which used in the research In addition, the data source and its collecting method are presented in this chapter as well as the variables’ description

- Chapter 4: Research result This chapter analyzes the data as well as indicate the result finding and compare it with other results The descriptive statistic of variables are also presented in this chapter

- Chapter 5: Conclusion and policy implications This chapter concludes the research finding, provides implication, further suggestion as well as the research limitations

CHAPTER 2 : LITERATURE REVIEW 2.1 FOOD SAFETY AND FOOD-BORNE DISEASES

According to WHO (2007, p 11), “Food-borne diseases (FBD) can be defined as those conditions that are commonly transmitted through ingested food FBD comprise

a broad group of illnesses caused by enteric pathogens, parasites, chemical contaminants and biotoxins Two methodological approaches for food-borne disease burden estimation exist Firstly, the etiologic agent (or risk assessment) approach which commences with the exposure and identifies the exposure levels of agents commonly transmitted though food, which is followed by determining the proportion that is food-borne Secondly, the syndromic (or epidemiological) approach, which commences with the outcome and estimates incidence of disease syndromes (e.g gastroenteritis or chemically induced anaphylaxis), followed by attributing a proportion

to food-borne agents A comprehensive burden of disease assessment will require a combination of both approaches”

WHO (2015, p X) also estimated there were about 600 million food-borne illnesses and 420,000 deaths in 2015 due to 31 food-borne hazards, and 40% of the food-borne disease burden was among children under 5 years of age However, the data that is used in the thesis, was collected by interviewing the participants through a questionnaire and not including any food testing Therefore, the FBD diagnose was depended on the responders’ own estimation combine with the assessment of the medical staff through individual description symptom

On the other hand, Food safety is “the assurance that food will not cause harm to human’s health or life” (Vietnam Ministry of Health, 2010) This definition is not as specific as WHO’s, however it covers the whole procedure of growing, harvesting, preserving, processing of food instead of only preparing and eating Due to the wide

coverage of Vietnam Ministry of Health and the spreading of Food Safety Regulatory

in Vietnamese population, the thesis used this definition as the concept of Food Safety

2.2 THE HEALTH BELIEF MODEL:

The HBM was initially derived from the theory of psychology and behavior of individual (Maiman and Becker, 1974) making decision in uncertainty condition, where their behavior were predicted by evaluating the “value – expectancy” of possible outcome Adapting this theory in health area, the model assumes that individual highly assesses illness prevention and health status improvement As a result, they expect their specific action would prevent disease, improve their health That expectation is affected by individual’s calculation about their sensibility to illness, seriousness of disease as well as the possibility of getting sick by their behavior According to Glanz

et al (2008, p 47 - 48), the main components of HBM include:

Table 2.1: The concepts of Health Belief Model

Perceived susceptibility Belief about the chances of

experiencing a risk or getting a condition or disease

Define population(s) at risk, risk levels

Personalize risk based on a person’s characteristics or behavior

Make perceived susceptibility more consistent with individual’s actual risk

Perceived severity Belief about how serious a

condition and its sequelae are

Specify consequences of risks and conditions

Perceived benefits Belief in efficacy of the

advised action to reduce risk or seriousness of impact

Define action to take: how, where, when; clarify the positive effects to be expected

Perceived barriers Belief about the tangible

and psychological costs of the advised action

Identify and reduce perceived barriers through reassurance, correction of misinformation, incentives, assistance

Cues to action Strategies to activate

“readiness”

Provide how-to information, promote awareness, use appropriate reminder systems

Self-efficacy Confidence in one’s ability

to take action

Provide training and guidance in performing recommended action Use progressive goal setting

Give verbal reinforcement Demonstrate desired behaviors

Reduce anxiety

These components combine with other individual characteristics then divide into three groups of factors included: modifying factors, individual belief and action The relation, components and impact of each groups showed in the figure below:

Figure 2.1: Health Belief Model Components and Linkages (Glanz et al, 2008)

Many research implemented HBM in analyzing food safety behavior among various kind of responders, for instance: restaurant workers (Cho et al, 2010), primary food preparers in family with young children (Lum, 2013; Meysenburg et al, 2013), older adults (Hanson and Benedict, 2002) All these papers show that the components

in HBM interact with other, affect on individual’s behavior, especially the food safety knowledge had strong impact to their perception about food safety

2.3 EMPIRICAL REVIEWS ON DRIVERS OF FOOD SAFETY PRACTICES: 2.3.1 Socio-economic characteristics

The participants who had college degree or higher degree got a better score in food safety knowledge and food safety behavior than others (Meysenburg et al, 2013) The author group used the Health Belief Model with the mixed method analysis to analyze the sample of 72 participants by script interview and group discussion Another finding

to and severity

of disease

Perceived benefits

Perceived barriers Perceived self-efficacy

Perceived threat

Individual behaviors

Cues to action

of Unusan (2005) determined a positive effect of education level to the confidence in food safety practice Furthermore, higher education level groups get less risk behavior than the lower ones However, this research found that socio-economic status does not correlate with individual food safety practice In this research, Unusan collected the data from Turkish households and analyzed using MANOVAs

On the other hand, Unusan’s research indicated the impact of gender and education level to the food safety knowledge The reason for this finding is quite acceptable due

to most of the primary food preparers in household are women, and the high educated ones likely pay more attention on information Byrd-Bredbenner et al (2007) and Mullan et al (2014) found the similar result after reviewing many researches in food safety issue Their research also showed that age affect to food knowledge, for instance the older tend to get higher score in food knowledge In addition, women would be more responsible in food safety issue than men (Jevsnik et al, 2006) This consequence concluded from the investigation of participants by analyzing using ANOVA

Another research leading by Langiano et al (2012) figured out that the married participants had more precise food behavior than the singles Moreover, the more members the family had, the more accurate practice the primary cook did

2.3.2 Food safety knowledge

Food preparers mainly study food process knowledge from family’s members and relatives (Meysenburg et al, 2013) Other research also indicated family as a food safety knowledge resource which affects individual behavior (Kwon et al, 2008; Trepka et al, 2006) Kwon investigated participants of the Special Supplemental Nutrition Program for Woman, Infants, and Children with a questionnaire about food knowledge and behavior to conclude the result by ANOVA analyzing Furthermore, respondents with excellent food knowledge would behave precisely in food practicing

(T H Vo et al, 2015) These authors group investigated in the canteens’ and restaurant’s workers by using logistic regression model to get this finding.

However, the finding of Cho et al (2010) show that food safety knowledge of participants do not influence their food practice This research focused on restaurant’s worker with the multiple regression and maximum likelihood estimation In addition, Roberts et al (2008) prove that there is only limited distortion in food workers’ behavior even though they have just trained, educated in food safety Those consolidate the assumption that knowledge had little impact on individual’s behavior change

2.3.3 Perception on food safety

The high self-efficiency is confident that they can prevent health threat as well as FBD when handling food was in their control (Meysenburg et al, 2013) This confidence would decrease whether the food were prepared by others In addition, ones had suffer food-borne illness or caused the illness for family member due to their improper handling food also were less confident in their food prepare

Many studies have shown mixed results on food safety perception and behavior Nesbitt et al (2013) found that many consumers felt that food contamination occurs before food reaches their kitchen and majority of those who experienced FBD felt that their illness was cause by food prepared outside the home However, Unusan’s research (2007) indicated that consumer did not recognize food poisoning/FBD as a health problem, they even acknowledged it as a normal issue This misleading awareness leads to the fact that they rarely tend to adjust food safety behavior or pay attention on food safety issue

Jevsnik et al (2006) found the fact that the household cook preferred farmer’s product to the industrial factories They believed that the food produced by farmer would be safer However, the consumers said that they were not responsible for food safety but claim this is food handler (farmer, food factory, retailing, catering) and the

government The research also figured out that the under 30 group regularly evaluated their ability in handling food safely highly, though the fact was not (Byrd-Bredbenner et al, 2007)

self-A research by T H Vo et al (2015) proved the correlation between the food safety knowledge and individual attitude about food safety problem while the relation of attitude and food practice was insignificant Moreover, Cho et al (2010) found that responders with good knowledge about food safety would perceive well about the severity and probability of food poisoning In addition, precise food knowledge consumers seldom struggled with barriers, difficulties when handling food safely However, this paper did not figure out any direct impact of knowledge to perception of FBD prevention as well as individual’s behaving food practice safely, though it found that the benefit perception affect to individual safety food behavior The responders who acknowledged the advantages of not suffering food poisoning would behave food safety practice better and more frequently

On the other hand, Hanson and Benedict (2002) demonstrated the good awareness about FBD severity would improve individual behaviors while the correlation between the perception of FBD hazard and food safety practice was not strong The result was calculated through the nonparametric statistics with Spearman rank correlation coefficients

2.3.4 Cues to actions

Cho et al (2010) found the strong correlation of cues to individual food safety practice Ones who used to suffer FBD were more likely to perform safe food handling practice (Lum, 2010) However, Lum indicated that experiencing symptoms of an illness does not always lead to favorable behavior

A similar result from Hanson and Benedict (2002) showed that the cue, content of communication, education material had impact to safety food practice of responders

This paper also found that males are less affect from education than females while the older age got stronger impact from education This influence varied according to the food handling frequency of individual

Another research of Byrd-Bredbenner et al (2013) indicated that the messages of risky food or oriented practices, which are printed on the food labels, had positive effect to individual practice Furthermore, this research also figured out the consumers

in various age both concerned about food safety knowledge However, each age was only susceptible and concerned when the information was specifically communicated for their group

On the other hand, Mullan et al (2014) showed that past behavior or habits is an important predictor of current behavior Habits are formed through the repetition of a behavior in a consistent context or in response to a cue Individuals may not be practicing food safety behavior in their homes due to a lack of cue to action that remind them to do so

CHAPTER 3 : RESEARCH METHODOLOGY 3.1 ANALYTIC FRAMEWORK

Consulting from other relating research, thesis implemented the HBM framework with the components evaluated through specific variables for food safety behaviors The modifying factors include individual and demographic characteristic of participants and their family whilst the knowledge focuses about the food safety issue only In addition, the individual belief measured by the attitude, awareness of responders about food safety relating issue On the other hand, individual’s actions evaluated through a range of food safety practices while the cues for these behaviors are the information source

Due to the limitation of the secondary data, the components from HBM measured

- Individual behavior: hygiene, process, preserve practice

- Cues to action: the food safety information source

The relatives and interactions of those components illustrated in the figure below:

Figure 3.1: The Health Belief Model application in food safety

3.2 ECONOMETRIC MODELS

3.2.1 Multivariate probit model

Thesis applied the multivariate probit model (MVP) in order to analysis the influence of independent variables to each behavior group The data include 3 aspects

of food safety behavior such as hygiene kitchen practice, process and preserve practice, hygiene individual practice, so this paper would use the MVP with 3 equations to

+ TV, newspaper + Local food safety

communicator

Individual behavior

- Hygiene kitchen practice

- Process, preserve practice

- Hygiene individual practice

Action

ắ

Perceived susceptibility to and severity of disease

- Attention about food safety problem

- Reason of food poisoning

- Risky group

Perceived benefits

- Attitude about food selection

Perceived barriers

- Food source

Perceived efficacy

self Attitude about food practice

Perceived threat

Individual belief

predict the dependent variables According to Cappellari and Jenkins (2003), the trivariate probit model is:

yim∗ = βmXim+ ϵim, m = 1,2, 3

yim = 1 if yim* > 0 and yim=0 otherwise

ϵ im, m = 1,… 3 are error terms distributed as multivariate normal each with a mean of zero, and variance–covariance matrix V, where V has values of 1 on the leading diagonal and correlations ρjk = ρkj as off-diagonal elements

The log-likelihood function for the sample of N independent observation is given by:

= Pr(ϵ 3≤β3 ’X3 ∣ ϵ 2≤β1 ’X1, ϵ 1≤β1 ’X1) x Pr(ϵ 2≤β1 ’X1, ϵ 1≤β1 ’X1) x Pr(ϵ 1≤β1 ’X1)

 Variables in the model:

The dependent variable is nine food safety practices of individual, divided into three behavior groups with varied behaviors for each group Every dependent variable receives the binary value: 1 (for precise practice) or 0 (for wrong practice) The thesis defined the right practice is that all behaviors in the group were precise, otherwise is wrong practice, and the three behavior groups are symbolized as “kprac” (for hygiene kitchen practices), “pprac” (for process and preserve practices) and “iprac” (for hygiene individual practices) The behaviors are “The 10 golden principles in food processing” but the “Clean, tidy kitchen and the cooker surface” and “separate well-done food and raw food in processing and preserving” divided into 4 behaviors in order for the interviewer easy to evaluate

The independent variables:

- “sex” is the dummy variable indicate the sexuality of participant, 0 for male and

1 for female The expected regression coefficient of this variable is predicted insignificant, due to the fact that most of the responders were females;

- “loc” is dummy variable indicate the location of responder, 0 for suburban participant (include 12 districts: Binh Tan, Binh Chanh, Thu Duc, Go Vap, 9, 12, 6, 8, Nha Be, Can Gio, Cu Chi, Hoc Mon) and 1 for urban participant (include 12 districts:

1, 2, 3, 4, 5, 7, 10, 11, Phu Nhuan, Tan Binh, Tan Phu, Binh Thanh) The expected regression coefficient of this variable is predicted to get the positive value, that mean the urban participant get the higher probability to behave precisely than the suburban ones;

- “age”, “exp”, “f_member” are the variables indicate the age (years old), the amount of money for food expenditure (hundred thousand VND), the number of family’s member, respectively The expected regression coefficients of these variables

are predicted getting the positive values However, the thesis used the natural logarithm

of age (lnage) and expenditure (lnexp) to estimate the coefficients so as to adjust these variables to get normal distribution

- “job” and “edu” are category variables show the professional (office clerk, retirement, household lady, physical labor, farmer) and the education level (primary, junior high, high school, college – university, below primary) of the responders In order to reduce the number of dummy variable as well as easier to interpret, the thesis use the transformed variable of “edu” (the schooling years) and job (only householder - hholder, common labor – com_labor and other)

- “know” is the examining result about food safety knowledge of participant This variable is recorded through the questionnaire followed WHO’s Food safety knowledge, divide into 2 group: knowledge about safety food selection, knowledge about food processing and preserving The value of this variable is the score that the participants get through the survey questions That score was estimated by the difficulty index method (Collen, 2006, p 98 – 100 ):

𝜌 = 𝑛𝑐

𝑁 , where:

ρ: difficulty index

nc: the number of right answer

N: the total number of responders

- “per” is the indicator of participant’s perception, estimating by the questionnaire

of attitude about food safety issues, include 4 groups: perception about susceptibility

to and severity of FBD, perception about benefits, perception about barrier and perception about self-efficacy However, in spite of the limited of the data, the thesis applied Factor Analysis method to figure out the perception factor from 3 of 4 groups, except the perception about benefits

- “cue” is the category variable show the information source about food safety of responders (TV, radio, newspaper, local medical staff, food documentary) In order to analysis, the thesis use the dummy variable of these cues (TV, radio, news, local_staff, food_doc respectively) All variables which are used in the model are described in the table below:

Table 3.1: Variables’ description

Expenditure Amount of money for food consumption Hundred thousand VND

years

labor, householder, other

point Perception Awareness about food safety issue 3 point Likert’s scale

Cue to action Food safety information source Category variable: TV,

radio, newspaper, local medical staff, food documentary

3.2.2 Multicollinearity problems

Multicollinearity is the situation that the explanatory variables in the regression model have the linear relationship There are two types of multicollinearity: the perfect

multicollinearity and imperfect multicollinearity If multicollinearity is perfect, the regression coefficients of the dependent variables are indeterminate and their standard errors are infinite If multicollinearity is less than perfect, the regression coefficients, although determinate, possess large standard errors (in relation to the coefficients themselves), which means the coefficients can not be estimated with great precision or accuracy (Gujarati, 2004)

The paper of Cho et al (2010) and T H Vo et al (2015) found out the relation between knowledge and perception about food safety The perception variable was measured by Likert scale, that mean its coefficient does not show the extent but only the trend of perception’s impact to behavior Due to this reason and the large number

of observation (above 1000), the effect of multicollinearity in the regression model would decrease However, the thesis also used the reduced form of MVP as well as the original form to estimate the regression coefficient

3.2.3 Propensity Score Matching (PSM) Method

According to Chow and Mullan (2009), the past behavior was a significant predictor of food safety behavior so they suggest providing a cue to carry out food-safety behaviors and made those become individual habits in order to change primary cook behaviors Moreover, the consumers regularly are not aware about their role in food safety chain (Jevsnik et al, 2007) and preventing FBD (Byrd-Bredbenner et al 2007) Due to these reason, the consumers could hardly change their behavior in a short time period after suffering food poisoning

On the other hand, individuals use rationality when they are aware of and have some knowledge about the cause-effect relationship between the correct behavior and the health benefits (Mari et al 2008) But it might be difficult for the household cook to figure out the wrong practice which lead them to the FBD In addition, the food poisoning information in the data only covered the health statement for 2 weeks, thus

this paper assume that the food poisoning incident did not affect to individual behavior and using individual behavior to estimate the food poisoning probability by PSM method

Khandker et al (2009) “Propensity score matching (PSM) constructs a statistical comparison group that is based on a model of the probability of participating in the treatment, using observed characteristics Participants are then matched on the basis of this probability, or propensity score, to nonparticipants” In order to compare the effect

of food practice on food poisoning risk, this thesis used the PSM in 4 steps:

- Step 1: establish the logit regression model with the dependent variable receive value as “0” if the participant had not suffer FBD within 2 week at the survey time, and

“1” otherwise The explanatory variables are individual food safety behaviors

- Step 2: using the probit regression model to predict the FBD possibilities of each responder in the survey data

- Step 3: remove all observations with too high or too low prediction among the sample

- Step 4: compare the food safety practices between 2 groups “suffer FBD” or

“non suffer FBD” in order to evaluate the impact of behavior to FBD probability of individual

The relative of continuous variables would check by the t-test while the bivariate variables’ tests by the Chi-square test

3.3 DATA

Thesis used the data from investigation about individual food poisoning in Ho Chi Minh City survey (2013) and Knowledge, Attitude, Practice (KAP) in food safety

of Ho Chi Minh City’s Household survey (2013)

- Data source: Safety Hygiene Food Branch of Ho Chi Minh city

- Data description: the data had two parts:

+ Part 1 is the investigation in individual food poisoning and the relevant symptoms, included: individual characteristic, demographic information, food poisoning statement and clinical symptoms

+ Part 2 is the KAP survey of primary cook of household, included: individual characteristic, food safety knowledge, food safety attitude and examining the food safety practicing of primary cook

- Size of samples: 1,174 households and 4,593 individual participate in the survey The households were chosen by Probability Proportional to Size sampling technique (PPS):

+ Among 319 wards of Ho Chi Minh City, the researchers chose randomly 30 wards on the list In each ward, the surveyor would investigate randomly the first household, then interview the next 39 households on the right side with 4 questionnaires include: acute food poisoning investigating, food knowledge examination, attitude investigating and food practice evaluation The food practice checklist would evaluate by the surveyor whilst the others would answer by the responders Therefore, the investigators in the research were the local medical staffs who take in charge of food safety issue at each chosen ward

+ The household who took part in the survey had to satisfy the standards: all the member of the household had lived at the location at least 6 months before the investigation The household had to approve to participate in the survey, the household’s member are ones who live in the same address, having at least one meal with other members and share the similar relation and household chores The household that is unapproachable for three times will be replaced by other

+ The individual participate in the survey do not have mental illness, deafness or dumb The children who joined in the survey must from 6 months old at least because the infant’s food is mostly lac feminnum The answer of the children under 10 was confirmed by their mother or the primary custodian

+ The symptoms to diagnose food poisoning case: after having meal, the patient

had the stomach-intestine symptoms (colic, vomit, diarrhea…), nerve symptoms

(stiffen tongue, illusion, less visible, delirium, convulsion…) or other symptoms

depend on the poisoning pathogen However, the food poisoning case only counted if

the responders had used the meal at home before the first symptom occurred A

household had suffered food poisoning when the primary cook or any of the family’s

members had the symptoms after having meal at home

Thesis combined two parts of the data to evaluate the relevant of knowledge,

perception about food safety and individual behavior as well as estimating the effect of

food safety behavior of individual to their food poisoning probability

CHAPTER 4 : RESEARCH RESULTS 4.1 FOOD SAFETY PROBLEMS IN VIETNAM

4.1.1 Food poisoning statement

WHO (2016) estimated the food-borne diseases burden in Vietnam are about 1 billion USD per year (2% of GDP), including the total costs of lost work time, lost productivity due to illness and related market losses On the other hand, according to Vietnam Food Administration’s (VFA) statistics from 2007 to 2015, there are 150 to

250 mass food poisoning outbreaks in Vietnam each year, impact to above 5,000 people annually Although the National Strategy on Food safety has already been implemented from 2006 and the strategy in the period 2011 – 2020 has validated, the number of food poisoning cases in Vietnam still fluctuate around 5,000 cases per year

Figure 4.1: The number of food poisoning cases in Vietnam (MOH, 2016)

The reason for this stagnant may be the slackness of legacy institution The Law

of Food Safety was promulgated in 2011, effective since 2011 but the other regulatory documents related to, are in adequate to meet the actual needs, therefore the Food Safety Department do not have the institution to administrate while the manufacturers and consumers do not have enough information and instructions to apply food safety practice In addition, the Food Safety Administration network is complex with the involving of many Ministries, Departments Furthermore, the tropical climate and climate change problem create opportunities for food poisoning occurred The diversity

of Vietnamese food combine with the limited knowledge of consumer, enhance the possibility of FBD Despite of the stability of food poisoning cases, the number of death dropped down slowly and half of that is due to natural toxic (Nguyen, 2016)

Figure 4.2: The number of food poisoning outbreaks and death in Vietnam

Death

In Ho Chi Minh City, the food poisoning incidents trend to decrease annually with total 20 incidents in the period 2012 - 2016 and no death case Nineteen of twenty incidents were caused by bacterium, and the other is non-identified cause

Figure 4.3: The number of food poisoning cases in HCM city (FSBDH, 2016)

The figure of food poisoning cases in Vietnam and Ho Chi Minh City do not include the individual incidents due to the lack of attention from the government and researchers to this issue As a result, there are few programs to educate and evaluate the accurate of household food preparing and cooking Most of the resources of the government are spending on managing the manufacturers and merchandising However, the finding from the survey in 2013 show that the individual food poisoning ratio is 2.18% That figure implies the high possibility of population with food-borne diseases

4.1.2 Problems with household’s cooking behavior

Due to the lack attention of the Vietnamese government on the food safety in household, the primary cook‘s practice is considered badly According to the investigation of Safety Hygiene Food Branch of Ho Chi Minh city (2010), there’s only