spiral.imperial.ac.uk · 2 abstract opisthorchiasis is a zoonotic parasitic infection caused by...
TRANSCRIPT
A Comprehensive Epidemiological
Study of Opisthorchiasis in a Rural
Community of Thailand
Picha Suwannahitatorn
Imperial College London
Department of Infectious Disease Epidemiology
Thesis submitted for PhD examination
2015
2
Abstract
Opisthorchiasis is a zoonotic parasitic infection caused by human liver flukes. Opisthorchis viverrini
(OV) is endemic in Southeast Asia along Mekong basin which estimated 9 million people are
infected. Transmission to humans occurs through the consumption of uncooked fish; cyprinoid or
white-scale freshwater fish containing infective stage metacercariae. The infection is mainly
asymptomatic. Adult flukes could live up in the bile duct in the absence of treatment. Chronic
infection is strongly related with bile duct cancer or cholangiocarcinoma. The International Agency
for Research on Cancer has declared that Opisthorchis viverrini is a member of group 1 agent,
carcinogenic to humans. Thailand has the highest incidence of cholangiocarcinoma in the world.
However, opisthorchiasis is acknowledged as a neglected and underestimated disease on the global
scale.
In the present day, the epidemiological data on opisthorchiasis from OV infection in Thailand are
considered outdated. An extensive cross-sectional study was undertaken from 1984 to 2001 under
the National Control Program, but data on incidence is very limited. Moreover, infection dynamics
are still poorly understood.
This project aimed to comprehensively study the infection dynamics of OV infection using multiple
tools in order to explore the infection in various aspects. Statistical models were developed to
explore epidemiological data; prevalence, incidence and infection intensity, with risk factors for
acquiring the infection. Infection dynamics will be described using mathematical modelling. The
qualitative technique, by interviewing and group discussions, will be used to explain the cause of
uncontrolled infection rate in bio-psycho-social aspect. Integrated study results will be used to
develop community intervention strategies under the framework of the public health planning
model. The overall outcome will be valuable for Thailand National Health Policy and epidemiological
data will provide the basis for further rigorous academic research.
3
Declaration of originality
I hereby declare that this project is entirely my own work and produced under the supervision of
Prof Christl Donnelly, Prof Joanne Webster and Prof Steven Riley.
Co-researchers and collaborators who participated in this work are acknowledged as stated below.
1. Prof Saovaanee Leelayoova, PhD Research investigator
2. Prof Mathirut Mungthin, MD PhD Research investigator
3. Assist Prof Ram Rangsin, MD DrPH Research investigator
4. Assoc Prof Paanjit Taamasri, MSc Research investigator
5. Dr Phunlerd Piyaraj, MD PhD Research investigator
6. Assoc Prof Wirote Areekul, MD Research collaborator
7. Assist Prof Pote Aimpun, MD PhD Research collaborator
8. Assist Prof Suthee Panichkupl, MD Research collaborator
9. Dr Tanongsan Tienthavorn, MD PhD Research collaborator
10. Mr Tawee Naaglor, MSc Laboratory manager
11. Mrs Worarachanee Imjaijitt, MSc Data unit manager
Any forms of additional source of information including published works, figures or quotes have
been fully referenced in the thesis.
Picha Suwannahitatorn, MD MSc
October 2015
4
Copyright declaration
The copyright of this thesis rests with the author and is made available under a Creative Commons
Attribution Non-Commercial No Derivatives licence. Researchers are free to copy, distribute or
transmit the thesis on the condition that they attribute it, that they do not use it for commercial
purposes and that they do not alter, transform or build upon it. For any reuse or redistribution,
researchers must make clear to others the licence terms of this work.
5
Contents
Abstract ....................................................................................................................................... 2
Declaration of originality .............................................................................................................. 3
Copyright declaration ................................................................................................................... 4
List of figures ............................................................................................................................. 11
List of tables .............................................................................................................................. 13
List of abbreviations and acronyms ............................................................................................ 15
Acknowledgments ..................................................................................................................... 17
Chapter 1 Introduction ............................................................................................................... 19
1.1 Overview ............................................................................................................................... 19
1.2 Biology of Opisthorchis viverrini .......................................................................................... 20
1.2.1 Morphology and life cycle ............................................................................................. 20
1.2.2 Opisthorchiasis and cholangiocarcinoma ..................................................................... 23
1.3 Epidemiology of OV infection .............................................................................................. 24
1.3.1 Current situation in the Mekong region ....................................................................... 24
1.3.2 Epidemiology of OV infection in Thailand ..................................................................... 27
1.3.2.1 Prevalence of OV infection ....................................................................................... 27
1.3.2.2 Infection intensity of OV infection ............................................................................ 35
1.3.2.3 Incidence of OV infection .......................................................................................... 36
1.4 Control of OV infection in Thailand ..................................................................................... 39
1.4.1 Timeline for OV infection control in Thailand ............................................................... 39
1.4.1.1 Survey phase ............................................................................................................. 39
1.4.1.2 National Control Program phase .............................................................................. 40
1.4.1.3 Passive surveillance strategy phase .......................................................................... 44
1.5 Main conceptual framework................................................................................................ 45
1.6 Study aims and thesis outline .............................................................................................. 47
1.7 Project outline ...................................................................................................................... 48
1.7.1 Background of study community .................................................................................. 48
1.7.2 Background of community healthcare system ............................................................. 51
1.7.3 Overview of study methods .......................................................................................... 52
6
Chapter 2 Epidemiology of Opisthorchis viverrini infection: distribution and risk factors .............. 53
2.1 Introduction .......................................................................................................................... 55
2.2 Methodology ........................................................................................................................ 57
2.2.1 Study areas and population .......................................................................................... 57
2.2.3 Questionnaires .............................................................................................................. 59
2.2.4 Collection of stool specimen ......................................................................................... 59
2.2.5 Data analysis ................................................................................................................. 60
2.2.6 Model for prevalence and incidence data .................................................................... 60
2.2.7 Model for infection intensity ........................................................................................ 61
2.2.8 Model analysis and fitting ............................................................................................. 66
2.2.9 Ethical consideration ..................................................................................................... 66
2.3 Results ................................................................................................................................... 67
2.3.1 Population characteristics ............................................................................................. 67
2.3.1.1 Response from the study .......................................................................................... 67
2.3.1.2 Demographic data ..................................................................................................... 71
2.3.2 Uncooked fish consumption behaviors......................................................................... 73
2.3.3 Distribution and risk factors of OV infection ................................................................ 76
2.3.3.1 Distribution of prevalence ........................................................................................ 76
2.3.3.2 Distribution of incidence ........................................................................................... 77
2.3.4 Evaluation of risk factors for acquiring OV infection .................................................... 79
2.3.4.1 Univariable analysis of risk factors for acquiring OV infection ................................. 79
2.3.4.2 Model fitting ............................................................................................................. 80
2.3.4.3 Multivariable analysis of risk factors for acquiring OV infection .............................. 81
2.3.5 Study of infection intensity ........................................................................................... 85
2.3.5.1 Prevalence and intensity profiles .............................................................................. 86
2.3.5.2 The prevalence-age structure ................................................................................... 92
2.3.5.3 The age-prevalence-intensity structure .................................................................... 93
2.3.5.4 Estimation of parasite aggregation ........................................................................... 95
2.3.5.5 Infection dynamics of infection intensity.................................................................. 96
2.3.5.6 Relationship of infection intensity and risk factors using count model .................... 98
2.4 Discussion ........................................................................................................................... 103
2.4.1 Study design and response ......................................................................................... 103
2.4.2 Distribution of OV infection ........................................................................................ 106
2.4.2.1 Prevalence of OV infection ..................................................................................... 106
7
2.4.2.2 Incidence of OV infection ........................................................................................ 107
2.4.2.3 Intensity of the infection ......................................................................................... 107
2.4.3 Risk factors for acquiring the infection ....................................................................... 109
2.5 Conclusion .......................................................................................................................... 112
Chapter 3 Infection dynamics of OV infection...................................................................... 113
3.1 Introduction ........................................................................................................................ 115
3.1.1 The relationship of prevalence and incidence ............................................................ 117
3.2 Methodology ...................................................................................................................... 119
3.2.1 Model definition ......................................................................................................... 119
3.2.2 Parameters of the model ............................................................................................ 122
3.2.3 Fitting the model ......................................................................................................... 125
3.2.4 Basic reproductive number ......................................................................................... 127
3.2.5 Estimation of treatment effectiveness using individual-level data ............................ 128
3.3 Results ................................................................................................................................. 130
3.3.1 Infection dynamics of OV infection ............................................................................. 130
3.3.2 Estimation of treatment effectiveness using population data and mathematical
model ……………………………………………………………………………………………………………………………….132
3.3.3 Estimation of treatment effectiveness using individual-level data ............................ 133
3.3.4 Basic reproductive number; R0 ................................................................................... 134
3.3.5 Estimating the effectiveness of the National Control Program using the model ....... 135
3.4 Discussion ........................................................................................................................... 136
3.4.1 Infection dynamics and study limitation ..................................................................... 136
3.4.2 Basic reproductive number ......................................................................................... 137
3.4.3 Treatment effectiveness ............................................................................................. 138
3.4.4 Treatment failure ........................................................................................................ 138
3.5 Conclusion .......................................................................................................................... 140
Chapter 4 Evaluation of risk behaviours for OV infection as determined by a qualitative approach
................................................................................................................................................ 141
4.1 Introduction ........................................................................................................................ 143
4.2 Methodology ...................................................................................................................... 146
4.2.1 Group interview .......................................................................................................... 146
4.2.2 In-depth interview ...................................................................................................... 146
4.2.3 Study population ......................................................................................................... 147
4.2.4 Study process .............................................................................................................. 148
8
4.2.4.1 Focus group discussion ........................................................................................... 148
4.2.4.2 In-depth interview .................................................................................................. 149
4.2.5 Data processing and analysis ...................................................................................... 149
4.2.5.1 Thematic content analysis ...................................................................................... 150
4.2.5.2 Framework analysis................................................................................................. 150
4.2.5.3 Grounded theory ..................................................................................................... 150
4.3 Results ................................................................................................................................. 153
4.3.1 Population characteristics ........................................................................................... 153
4.3.2 Thematic contents ...................................................................................................... 156
4.3.2.1 Situation of the OV infection in the community ..................................................... 156
4.3.2.2 Current situation of consumption behaviours ........................................................ 157
4.3.2.3 Knowledge of the infection ..................................................................................... 159
4.3.2.4 Perception of OV infection and its consequences .................................................. 161
4.3.2.5 Health concern of risk factors and the infection .................................................... 162
4.3.2.6 Diagnosis and treatment ......................................................................................... 164
4.3.2.7 Role of National Control Program ........................................................................... 165
4.4 Discussion ........................................................................................................................... 167
4.4.1 Social component integrated in the main conceptual framework ............................. 168
4.4.2 Knowledge, attitude and perception of OV infection contributed by age structure.. 169
4.4.3 Attitude and perception towards risk behaviours ...................................................... 171
4.4.4 Impact of OV infection on health, treatment and prevention .................................... 174
4.5 Conclusion .......................................................................................................................... 176
Chapter 5 Community-based intervention: PRECEDE-PROCEED model framework for controlling OV
infection .................................................................................................................................. 178
5.1 Introduction ........................................................................................................................ 180
5.1.1 Overview of PRECEDE-PROCEED model ...................................................................... 181
5.2 Methodology ...................................................................................................................... 185
5.2.1 Overview of study methods ........................................................................................ 185
5.2.2 PRECEDE – PROCEED model establishment ................................................................ 186
5.2.2.1 Step 1 Social diagnosis ............................................................................................ 186
5.2.2.2 Step 2 Epidemiological diagnosis ............................................................................ 188
5.2.2.3 Step 3 Behavioural and environmental diagnosis ................................................... 188
5.2.2.4 Step 4 Educational and ecological diagnosis........................................................... 188
5.2.2.5 Step 5 Administrative and policy assessment ......................................................... 193
9
5.2.2.6 Step 6 Implementation ........................................................................................... 194
5.2.2.7 Fieldwork methodology .......................................................................................... 195
5.2.2.8 Step 7 Process evaluation ....................................................................................... 203
5.2.2.9 Step 8 Impact evaluation ........................................................................................ 203
5.2.2.10 Step 9 Outcome evaluation .................................................................................... 204
5.3 Results ................................................................................................................................. 205
5.3.1 Step 1 Social diagnosis ................................................................................................ 205
5.3.1.1 Demographic data ................................................................................................... 205
5.3.1.2 Characteristics of the study populations from qualitative approach ..................... 206
5.3.1.3 Precipitating factors for promoting uncooked fish consumption ........................... 207
5.3.1.4 Precipitating factors to discontinue Koi pla consumption ...................................... 209
5.3.2 Step 2 Epidemiological diagnosis ................................................................................ 209
5.3.3 Step 3 Behavioural and environmental diagnosis ....................................................... 209
5.3.4 Step 4 Educational and ecological diagnosis............................................................... 210
5.3.5 Step 5 Administrative and policy assessment ............................................................. 211
5.3.6 Step 6 Implementation ............................................................................................... 213
5.3.7 Step 7 Process evaluation ........................................................................................... 214
5.3.7.1 Summary of study flow and response rate ............................................................. 214
5.3.7.2 Uncooked fish consumption from distribution and risk factors study ................... 219
5.3.7.3 Incidence of OV infection ........................................................................................ 221
5.3.8 Step 8 Impact evaluation ............................................................................................ 222
5.3.8.1 Evaluation of risk factors for acquiring OV infection .............................................. 222
5.3.8.2 Evaluation of impact of community intervention on Koi pla consumption ............ 223
5.3.9 Step 9 Outcome evaluation ........................................................................................ 223
5.3.9.1 Evaluation of impact of community intervention on incidence of OV infection .... 223
5.3.9.2 Qualitative approach for evaluation of community intervention ........................... 224
5.4 Discussion ........................................................................................................................... 225
5.4.1 Study design and response ......................................................................................... 225
5.4.2 Situation of OV infection ............................................................................................. 228
5.4.2.1 Situation of Koi pla consumption ............................................................................ 229
5.4.2.2 Incidence of OV infection ........................................................................................ 229
5.4.3 Community intervention ............................................................................................. 230
5.4.3.1 Impact of intervention on Koi pla consumption ..................................................... 230
5.4.3.2 Impact of intervention on incidence of OV infection ............................................. 232
10
5.4.4 Implication of PRECEDE – PROCEED model framework .............................................. 233
5.5 Conclusion .......................................................................................................................... 236
Chapter 6 Discussion ................................................................................................................ 237
6.1 Overview ............................................................................................................................. 237
6.1.1 Host and agent relationship ........................................................................................ 239
6.1.2 Host and environment relationship ............................................................................ 239
6.1.3 Agent and environment relationship .......................................................................... 240
6.1.4 Development of planning model ................................................................................ 240
6.2 Summary of study findings ................................................................................................ 242
6.2.1 Epidemiology of Opisthorchis viverrini infection: distribution and risk factors .......... 242
6.2.2 Infection dynamics of OV infection ............................................................................. 243
6.2.3 Evaluation of OV infection and risk behaviours using qualitative approach .............. 244
6.2.4 Community-based intervention: PRECEDE-PROCEED model framework for controlling
OV infection ................................................................................................................................ 246
6.3 Public health implication ................................................................................................... 247
6.3.1 Primary prevention ..................................................................................................... 247
6.3.2 Secondary prevention ................................................................................................. 248
6.4 Future work ........................................................................................................................ 249
6.5 Conclusion .......................................................................................................................... 252
References ............................................................................................................................... 253
Appendices .............................................................................................................................. 261
11
List of figures
Figure 1.1 Morphology of Opisthorchis viverrini egg and adult ........................................................... 20
Figure 1.2 Life cycle of Opisthorchis viverrini ....................................................................................... 21
Figure 1.3 Bithynia snail and cyprinoid fish .......................................................................................... 22
Figure 1.4 Prevalence of opisthorchiasis in Southeast Asian and Southern China .............................. 24
Figure 1.5 Strategic approach for liver fluke control ............................................................................ 40
Figure 1.6 Prevalence of opisthorchiasis under National Control Program Implementation 1987 -
2001 ...................................................................................................................................................... 43
Figure 1.7 Main conceptual framework of study project ..................................................................... 45
Figure 1.8 Overview of study area ........................................................................................................ 48
Figure 2.1 Conceptual framework for distribution and risk factors of opisthorchiasis ........................ 56
Figure 2.2 Flowchart of fieldwork ......................................................................................................... 58
Figure 2.4 Distribution of count data in hurdle model ......................................................................... 63
Figure 2.3 Distribution of count data in zero-inflated model ............................................................... 63
Figure 2.5 Summarized flowchart for cross-sectional study ................................................................ 67
Figure 2.6 Summarized flowchart for follow-up study ......................................................................... 69
Figure 2.7 Population pyramids ............................................................................................................ 71
Figure 2.8 Prevalence of OV infection distributed by age-group ......................................................... 76
Figure 2.9 Incidence of OV infection distributed by age-group ............................................................ 77
Figure 2.10 Prevalence and intensity of Na-yao 2002 cross-sectional study ....................................... 86
Figure 2.11 Frequency distribution of EPG from Na-yao 2002 cross-sectional study .......................... 86
Figure 2.12 Prevalence and intensity of Na-yao 2004 follow-up study ................................................ 88
Figure 2.13 Frequency distribution of EPG from Na-yao 2004 follow-up study................................... 88
Figure 2.14 Prevalence and intensity of Na-ngam and Na-isarn cross-sectional study ....................... 90
Figure 2.15 Frequency distribution of EPG from Na-ngam and Na-isarn cross-sectional study .......... 90
Figure 2.16 Fitting –ln(s) with age ........................................................................................................ 92
Figure 2.17 k estimates from study areas ............................................................................................. 95
Figure 2.18 Predicted prevalence from varying k and M for Na-yao cohort data ................................ 96
Figure 2.19 Predicted incidence from varying k and M for Na-yao prevalence data ........................... 96
Figure 2.20 Predicted prevalence from varying k and M for Na-ngam and Na-isarn prevalence data 97
Figure 3.1 Conceptual framework for infection dynamics. ................................................................ 116
Figure 3.2 Fitting prevalence with age-structure incidence and fieldwork incidence ....................... 118
Figure 3.3 Model diagram for opisthorchiasis infection dynamics .................................................... 119
Figure 3.4 Differential equations of S1IS2R model.............................................................................. 122
Figure 3.5 Model diagram considered the R0 ..................................................................................... 127
Figure 3.6 Diagram for estimation of treatment effectiveness using individual-level data ............... 128
Figure 3.7 Infection dynamics of prevalence in study area ................................................................ 131
Figure 3.8 Estimation of treatment effectiveness from the model .................................................... 132
Figure 3.9 Basic reproductive number with various duration of infectiousness ................................ 134
Figure 3.10 Fitting model with prevalence from National Control Program. ..................................... 135
Figure 4.1 Conceptual framework for qualitative approach .............................................................. 144
Figure 4.2 Population pyramid of Na-ngam 2012 prevalence data .................................................... 153
Figure 4.3 Conceptual framework of opisthorchiasis ......................................................................... 167
12
Figure 4.4 Social component integrated to conceptual framework ................................................... 167
Figure 5.1 Diagram of PRECEDE – PROCEED model ............................................................................ 181
Figure 5.2 Conceptual framework for community intervention study ............................................... 183
Figure 5.3 Summary of study flowchart ............................................................................................. 185
Figure 5.4 Summary of fieldwork flowchart ....................................................................................... 194
Figure 5.5 Community intervention flowchart ................................................................................... 197
Figure 5.6 Distribution and risk factors study flowchart .................................................................... 198
Figure 5.7 Population pyramid of Tung-heang 2012 data .................................................................. 205
Figure 5.8 Implementation of fieldwork ............................................................................................. 213
Figure 5.9 Summarized flowchart for community trial study ............................................................. 214
Figure 5.10 Summarized flowchart for distribution and risk factors study ........................................ 217
Figure 5.11 Incidence of opisthorchiasis distributed by age-group ................................................... 221
Figure 5.12 PRECEDE – PROCEED model integrated to conceptual framework ................................. 233
Figure 6.1 Integrated conceptual framework from comprehensive study ........................................ 237
13
List of tables
Table 1.1 Summary of prevalence and intensity of OV infection ......................................................... 29
Table 1.2 Prevalence of OV infection in overseas Thai population ...................................................... 32
Table 1.3 Summary of incidence of OV infection ................................................................................. 33
Table 1.4 Summary of annual prevalence during National Control Program implementation............ 42
Table 1.5 Summary of fieldwork data collection .................................................................................. 52
Table 2.1 Study areas for risk and distribution analysis ....................................................................... 57
Table 2.2 Response for cross-sectional study ....................................................................................... 68
Table 2.3 Response for follow-up study ............................................................................................... 70
Table 2.4 Population characteristics of cross-sectional study .............................................................. 72
Table 2.5 Population characteristics of follow-up study ...................................................................... 72
Table 2.6 Multivariable analysis of population characteristics and uncooked fish consumption
behaviours in cross-sectional study ...................................................................................................... 74
Table 2.7 Multivariable analysis of population characteristics and uncooked fish consumption
behaviours in follow-up study .............................................................................................................. 75
Table 2.8 Proportional distribution of OV infection diagnosis categorised by stool examination
methods ................................................................................................................................................ 78
Table 2.9 Model fitting for prevalence and follow-up study ................................................................ 80
Table 2.10 Multivariable analysis of risk factors for acquiring OV infection of prevalence data from
Na-isarn 2013 study .............................................................................................................................. 81
Table 2.11 Multivariable analysis of risk factors for acquiring OV infection of prevalence data from
Na-ngam 2012 study ............................................................................................................................. 82
Table 2.12 Multivariable analysis of risk factors for acquiring OV infection of follow-up data from Na-
yao 2012 study ...................................................................................................................................... 83
Table 2.13 Multivariable analysis of risk factors for acquiring OV infection of follow-up data from Na-
ngam 2014 study ................................................................................................................................... 84
Table 2.14 Summary of positive egg count diagnosed by Kato-Katz and FECT methods ..................... 85
Table 2.15 Summary of Na-yao 2002 intensity data ............................................................................ 86
Table 2.16 Summary of Na-yao 2004 intensity data ............................................................................ 88
Table 2.17 Summary of Na-isarn and Na-ngam 2012 - 13 intensity data ............................................. 90
Table 2.18 Summary of prevalence-age structure ............................................................................... 92
Table 2.19 k estimates from study areas .............................................................................................. 95
Table 2.20 Summary of count model for infection intensity analysis .................................................. 98
Table 2.21 Multivariable analysis of risk factors and infection intensity of Na-yao 2004 follow-up data
.............................................................................................................................................................. 99
Table 2.22 Multivariable analysis of risk factors and infection intensity of Na-isarn and Na-ngam
2012-13 prevalence data .................................................................................................................... 101
Table 2.23 Model fitting for Na-yao follow-up study ......................................................................... 102
Table 2.24 Model fitting for Na-isarn and Na-ngam cross-sectional study ........................................ 102
Table 3.1 Obtained incidence from fieldwork and estimated incidence from prevalence-age structure
............................................................................................................................................................ 117
Table 3.2 Summary of fieldwork data for Na-yao area ...................................................................... 120
Table 3.3 Summary of fieldwork data for Na-ngam area ................................................................... 121
14
Table 3.4 Model parameters .............................................................................................................. 123
Table 3.5 Estimation of duration of infectiousness and recovery rate .............................................. 130
Table 3.6 Estimation of treatment effectiveness from Na-yao 2002-04 data .................................... 133
Table 3.7 Estimation of treatment effectiveness from Na-ngam 2012-14 data ................................ 133
Table 3.8 Basic reproductive number ................................................................................................. 134
Table 4.1 Selection criteria for focus group discussion ...................................................................... 147
Table 4.2 Analysis approach to qualitative data ................................................................................. 149
Table 4.3 Age structure classified by generations of Na-ngam 2012 prevalence data ...................... 153
Table 4.4 Characteristics of focus group discussion participants ....................................................... 154
Table 4.5 Characteristic of in-depth interview participants ............................................................... 155
Table 4.6 Situation of the OV infection in the community ................................................................. 156
Table 4.7 Current situation of consumption behaviours .................................................................... 157
Table 4.8 Knowledge of the infection ................................................................................................. 159
Table 4.9 Perception of OV infection and its consequences .............................................................. 161
Table 4.10 Health concern of risk factors and the infection ............................................................. 162
Table 4.11 Diagnosis and treatment ................................................................................................... 164
Table 4.12 Role of National Control Program ..................................................................................... 165
Table 4.13 Knowledge, attitude and perception of OV infection contributed by age structure ....... 169
Table 4.14 Common characteristics regarding knowledge, attitude and perception across
generations ......................................................................................................................................... 172
Table 4.15 Impact of OV infection on health, treatment and prevention ......................................... 174
Table 5.1 Community mutual agreement........................................................................................... 192
Table 5.2 Selection criteria for focus group discussion ...................................................................... 201
Table 5.3 Review of study process and method of assessment ......................................................... 203
Table 5.4 Population characteristics of Tung-heang area .................................................................. 206
Table 5.5 Response for community trial study ................................................................................... 215
Table 5.6 Population characteristics of community trial study .......................................................... 216
Table 5.7 Response for distribution and risk factors study ................................................................ 218
Table 5.8 Population characteristics of distribution and risk factors study ....................................... 219
Table 5.9 Distribution of uncooked fish consumption ...................................................................... 219
Table 5.10 Univariable and multivariable analysis of population characteristics and uncooked fish
consumption behaviours .................................................................................................................... 220
Table 5.11 Univariable and multivariable analysis of risk factors for acquiring OV infection ............ 222
15
List of abbreviations and acronyms
AIC Alkaline information criterion
BPP Border Patrol Police
CAB Community advisory board
CCA Cholangiocarcinoma
CI Confidence interval
CIP Community intervention package
DALY Disability-adjusted life year
EPG Egg per gram
FECT Formalin-ether concentration technique
FGD Focus group discussion
HBM Health belief model
IARC International Agency for Research on Cancer
IRR Incidence rate ratio
IQR Interquartile range
k parasite aggregation parameter
LC Latent class
M Mean infection intensity
MIF Minute intestinal fluke
MCMC Markov chain Monte Carlo
NBLH Negative binomial logit hurdle model
NBRM Negative binomial regression model
NTDs Neglected tropical diseases
16
OR Odds ratio
OV Opisthorchis viverrini
PCR Polymerase chain reaction
PDR People's Democratic Republic
PLH Poisson logit hurdle model
PPM PRECEDE-PROCEED model
PRM Poisson regression model
Q1 1st quartile
Q2 2nd quartile
Q3 3rd quartile
Q4 4th quartile
R0 Basic reproductive number
S.D. Standard deviation
SCT Social cognitive theory
ZINB Zero-inflated negative binomial model
ZIP Zero-inflated Poisson model
17
Acknowledgments
I would like to gratefully thank my supervisors; Prof Christl Donnelly and co-supervisors Prof Joanne
Webster and Prof Steven Riley for all their support. Prof Christl Donnelly has used her expertise in
statistics to guide me through this project for technical support in rigorous data analysis. Prof Joanne
Webster provided her knowledge supporting me in the aspect of parasitology and Prof Steven Riley
has guided me to use fieldwork data to formulate mathematical models. Moreover, the entire
project is carefully supervised through my PhD. They have enabled me to use all my potential to
achieve this challenging task.
Before I came to Imperial College London, I believed this University to be one of the top universities,
and from my earliest experiences during my MSc course in 2011, this was quickly confirmed by the
wealth of inspiration, the perfect atmosphere conducive to learning and abundance of friendly,
warm and resourceful colleagues supporting my growth and development. I knew right away, that
studying at ICL will stand to be one of my best decisions throughout life. I consider every moment
here as precious and I wish to express my gratitude to the entire college.
I acknowledge and gratefully thank the research team and members of staff from Thailand for their
collaboration. I would also like to thank staff members from the Department of Parasitology and
Department of Military and Community Medicine, Phramongkutklao College of Medicine for
cooperation and Phramongkutklao Research Fund for financial support to my research.
I would like to thank former head of Department of Parasitology; Prof Saovaanee Leelayoova and my
academic mentor Prof Mathirut Mungthin for motivation and inspiration to commence work on the
neglected opisthorchiasis disease. I also thank my current head of department; Assoc Prof Paanjit
Taamasri and my colleague Dr Phunlerd Piyaraj for all the support since the beginning.
The fieldwork could not have been done without the initiation of a community project started in
2002 under the visionary researchers from Department of Military and Community Medicine; Assoc
Prof Wirote Areekul (former head of department) and Assist Prof Ram Rangsin (current head of
department). Research colleague; Assist Pote Aimpun and Assist Suthee Panichkupl provided full
support for my research.
I would like to address Mr Tawee Naaglor and Mrs Worarachanee Imjaijitt as potential collaborators.
Mr Tawee’s skill in laboratory work and Mrs Worarachanee’s expertise in biostatistics are
outstanding. I am very grateful to have them in the research team.
18
On behalf of the research team, I acknowledge the collaboration of the authorities of The
Chachoengsao Provincial Health Office, The Sanamchaikhet District Health Office, Local health
volunteers and all participants from Na-yao, Na-ngam, Na-isarn and Tung-heang village for their
great contributions.
I would like to thank my parents and my younger sister. It is a tough decision for choosing academic
pathway from my medical doctor background. From the moment I decided to focus on research,
rather than medical practice, the way was not smooth and there were many challenging tasks.
Without them, I would not have succeeded to accomplish my academic life. Their love and care are
unconditional and infinite. The overwhelming gratitude also extends to the support from all of my
family members.
I sincerely want to thank those who give me true love and care. Every element of your support helps
me to develop this work along the way. For those who sometimes make my life difficult, I forgive
them.
Lastly, I wish to grant this contribution to the Royal Thai Government, Ministry of Science and Royal
Thai Army Medical Department for their sponsorship and making this opportunity come true.
Picha Suwannahitatorn, MD MSc
October 2015
19
Chapter 1 Introduction
1.1 Overview
Opisthorchiasis is a zoonotic parasitic infection caused by human liver flukes. Three major species
are responsible for the infection (1-4): Opisthorchis viverrini, Opisthorchis felineus and Clonorchis
sinensis (disease is also called clonorchiasis). An estimated 700 million people are at risk worldwide
(5). Opisthorchis viverrini is endemic in Southeast Asian along the Mekong Basin (6-9) including
Thailand, Lao PDR and Cambodia where an estimated 9 million people are infected (5, 9-11).
OV is a pathologically important food-borne trematode (12, 13), leading to infection within the
hepato-biliary system. Transmission to humans occurs through the consumption of uncooked fish;
cyprinoid or white-scale freshwater fish containing infective stage metacercariae (2, 14). After
infection, OV can live in the bile duct in the absence of treatment (15-17). Many studies have
indicated that chronic infection is strongly related to a bile duct cancer, cholangiocarcinoma; CCA
(17-21). The International Agency for Research on Cancer (IARC) has declared that OV is a member
of group 1 agents, carcinogenic to humans (5, 18). Thailand is the highest ranked in the world for
CCA incidence, with ranges from 93.8 to 317.6 /100000 person-years (15-17, 20, 22-24). However,
opisthorchiasis is acknowledged as a neglected and underestimated disease on the global scale (10,
11, 25).
In Thailand, the prevalence of OV infection was extensively studied in the past especially during 1984
– 2001 under the National Control Program implementation (26-28). The latest national survey was
performed in 2001 (28). The record shows the geographical variation in infection prevalence is not
homogenously distributed through the country. The endemic areas are confined to locations where
intermediate hosts and uncooked fish consumption behaviours are present (1, 5, 15, 29).
In the present day, the data on disease epidemiology are considered outdated and the infection
dynamics are still poorly understood (10, 11, 25, 30).
20
1.2 Biology of Opisthorchis viverrini
1.2.1 Morphology and life cycle
Human liver flukes are digenetic trematodes belonging to the family Opisthorchiidae (2, 3, 13, 31).
The 3 majors species found to be infectious to humans are Opisthorchis viverrini, Opisthorchis
felineus and Clonorchis sinensis. Others species are rarely reported (13).
Figure 1.1 Morphology of Opisthorchis viverrini egg and adult
Image of egg and adult OV as shown in Figure 1.1 were obtained by myself at Department of
Parasitology, Phramongkutklao College of Medicine (October, 2015).
The adult worm (Figure 1.1, right) of OV is 7.0 (5.4-10.2) X 1.5(0.8-1.9) mm on average. The fluke
morphology is dorso-ventrally flattened, thin and transparent with reddish-blue colour. Eggs are oval
(Figure 1.1, left), 17 X 15 microns, with Ill-developed operculum and an aboperculum knob at the
bottom. The life cycle is complex (12, 32-36) and comprises multiple intermediate hosts involving
several stages of parasitic growth development as shown in Figure 1.2.
21
Figure 1.2 Life cycle of Opisthorchis viverrini
Figure 1.2 is obtained from free-access database from Laboratory Identification of Parasites of Public
Health Concern, US CDC, http://www.dpd.cdc.gov/dpdx/html/opisthorchiasis.htm.
The life cycle of Opisthorchis viverrini was fully described in 1965 (2). Embryonated eggs are released
with the faeces (1) in natural water bodies such as ponds, rivers or flooded rice fields. In the water,
eggs are ingested by Bithynia snails (2) serving as the first intermediate host. Miracidia (2a) hatch
from the eggs and grow in many stages as seen in the Figure, miracidia (2a) sporocysts radiae
(2c) cercariae (2d). The growth and development from miracidia to cercariae takes approximately
2 months after snail infection. Free-swimming cercariae are released from the snails (3) and become
encysted under freshwater fish muscle layer or scales becoming infective metacercariae (4) in 21
days. Cyprinoid freshwater fish serve as the second intermediate host.
Uncooked freshwater fish containing encysted metacercariae are consumed by humans. Cysts are
digested and become excysted in the duodenal part of the small bowl (5). Metacercariae will travel
to the biliary duct through the ampulla of Vater (the opening of common bile duct to the intestine)
where they grow to fully developed adults. The adult flukes are ready to reproduce by laying eggs in
3 – 4 weeks. In total, it would take 4 months to complete the full life cycle.
22
There are two medically important stages involving human explained below.
The infective stage (i): encysted metacercariae in fish muscles or scales serve as the infective stage
when humans consume raw fish.
Diagnostic stage (d): opisthorchiasis can be diagnosed by identifying fluke eggs in faecal materials or
bile content. More importantly, eggs release parasite offspring into the environment.
Figure 1.3 Bithynia snail and cyprinoid fish
Image of Bithynia snail (left) and cyprinoid fish (right) as shown in Figure 1.3 were obtained by
myself at Department of Parasitology, Phramongkutklao College of Medicine (October, 2015). Figure
1.3 shows the general appearance of the Bithynia snail and cyprinoid fish. The prevalence of
cercariae in snails is typically low; 0.05 - 0.07% (32). The diversity and abundance of snails are
strongly related to seasonal variation mainly depending on rainfall. The population of snails are
abundant during the rainy season, which is also the time of rice planting and dramatically decrease
during the dry season. Therefore, controlling snails is impractical due to the seasonal variation in the
snail population (32, 37).
Unlike snails, the prevalence of infection in cyprinoid fish is high (38). Up to 95% of cyprinoid fish
species harbour the metacercariae. In Thailand, more than 15 species are found in natural water
bodies and serve as a good source of food.
The reservoir hosts are also reported in various species of animals including dogs, cats, rats and pigs.
These reservoir hosts are responsible for the transmission when they excrete contaminated stool in
the environment. Studies have shown that reservoir hosts might have some important role as an
area with minimal transmission from humans that still maintains re-infection rates (9, 39).
23
1.2.2 Opisthorchiasis and cholangiocarcinoma
Once the flukes become fully grown, they reside in the biliary duct. The host’s immune system will
extensively response to the pathogen producing inflammatory mediators also damaging the host’s
own tissues (12, 21). The parasites can evade immune mechanisms. However, the recitative host
response from chronic inflammation causes abnormal tissue growth: biliary epithelial cell
metaplasia. Long-term infection also induces accumulation of tissue fibrosis; a scar to the epithelial
wall. Many cytokines (IL-6, TNF-α) are responsible for altering the regular cell cycle to abnormal cell
proliferation including the mucus glands and fibrous tissue proliferation ending in advanced
periductal fibrosis (12, 15, 16). Carcinogenesis of OV is explained by secretion of growth factor from
the parasite. Many studies and experiments have demonstrated an ES product (excretory/secretory)
called OV-GRN-1 can stimulate abnormal cell proliferation and create a suitable environment for
tumour development (17, 22).
Opisthorchiasis is typically asymptomatic (15, 20). Abdominal discomfort is also seen but the
symptom is nonspecific. Even though bile duct thickening is always found, it does not cause bile
passage blockage so jaundice rarely occurs from chronic infection. The simple way to diagnose
infection is through stool examination. Although adult flukes can be detected by bile duct aspiration,
the procedure is invasive and is impractical in clinical use. Opisthorchiasis can be treated with a
single dose of praziquantel. Drug side effects are typically minimal including nausea or drowsiness. A
field study in Thailand shows a high drug efficacy of 90 – 95% (28, 40).
Cholangiocarcinoma is a primary malignancy of bile duct (16, 17, 19, 23). It is an extensively invasive
tumour and often causes metastasis. No definitive treatment is currently available result in death
within a short period of time. Early stage is always asymptomatic or causes nonspecific
manifestations, which are difficult to detect in general health practice. Late stage tumour grow and
invade the liver and major blood vessels causing extensive metastasis, which is always non-
resectable (15, 17, 20, 22). Bile duct obstruction results in jaundice and liver failure. Prognosis is very
poor (41), median 5-year rate is approximately 30 % for non-resectable tumour. The IARC has
declared that Opisthorchis viverrini is carcinogenic to humans (2, 11, 42). The burden from OV
infection and subsequent CCA costs 120 million USD annually for medical care and lost productivity
(11) and results in 74,070 DALYs (43).
24
1.3 Epidemiology of OV infection
1.3.1 Current situation in the Mekong region
Liver flukes are geographically distributed with regard to intermediate hosts including snails and
cyprinoid fish (33, 34, 37, 44, 45). Opisthorchis viverrini and Clonorchis sinensis are prevalent along
the Mekong region including Southeast Asian and Southern China (1, 5)
Figure 1.4 is adapted from Sithithaworn et al. (5) depicting the prevalence of Opisthorchis viverrini
and Clonorchis sinensis, which is geographically-related to the Mekong River. The Mekong River
origin is located in southern China with an estimated length of 4000 km ranking it the 12th – longest
river in the world. From the Tibetan Plateau, the river runs through China's Yunnan Province and
flows through Southeast Asian forming a natural border along Thailand, Lao PDR and Myanmar, then
running eastward along the Thai and Laos PDR border and through the country of Cambodia then
continuing to the South Chinese Sea. The river serves as an important water reservoir along the
Mekong Basin.
B
A
Figure 1.4 Prevalence of opisthorchiasis in Southeast Asian and Southern China
25
Opisthorchis viverrini is abundant in area A below the grey dotted line (A) including Thailand, Lao
PDR, Cambodia and Vietnam while Clonorchis sinensis is prevalent in area B within Southern China.
Clonorchiasis is not common in Southeast Asian except as reported in northern Vietnam bordering
China. However, sporadic cases are reported from central Thailand from PCR technique and within
the study area by Traub et al. (46).
Opisthorchis viverrini has been reported in Thailand for almost 100 years. In the past, multiple
surveys have shown a nearly 100% prevalence in northern and northeastern provinces (27, 28). The
prevalence is relatively low in central and southern Thailand. Biologically, natural intermediate hosts
can be found across the country (11, 47). However, further studies indicate the relationship of
uncooked fish consumption behaviors and high prevalence of OV infection (4, 29) related to
geographical population characteristics.
With the aid of foreign support and governmental concern (26-28, 48), the National Control Program
was implemented from 1984 – 2001 mainly focusing on case diagnosis and treatment. The national
prevalence was dramatically decreased to 10%. However, no on-going surveillance was conducted
due to the national policy and economic burden.
Accuracy regarding the field diagnostic methods, mainly performed with simple smear and Kato-Katz
technique, is a concerned issue. Concentration and molecular techniques indicate that simple smear
and Kato-Katz are less sensitive, which could lead to gross underestimation of the infection
prevalence (49, 50).
For Lao PDR, the prevalence is particularly higher in lowland rather than highland areas. The Laotian
cultural context is considered similar to Thai especially in the northeastern and also northern
provinces along the Mekong Basin. Thai and Lao languages are partially similar; therefore, cultural-
transfers across the border are common. The practice of uncooked fish consumption is similar and
prevalence of OV infection along the basin is also high (29, 51, 52). However, a recent study has
evaluated the efficacy of several drug choices against OV infection in Lao PDR showing promise for
future implementation of the National Control Program (53).
In Cambodia, data on OV infection are considered limited. As shown in Figure 1.4, the data are
available only along the Mekong Basin in limited part of Cambodia. Therefore, no accurate estimate
exists on the situation of the infection (5). Additionally, numbers of Cambodian immigrants are
currently settled in Thailand and also travel freely across the border. The given population dynamics
might impact transmission dynamics especially in the future when Southeast Asia will become an
economic community with free population movement.
26
Vietnam, has reported both Opisthorchis viverrini and Clonorchis sinensis transmission. Clonorchiasis
is reported in the northern region (54) bordering southern China. Consumption of uncooked fish is
also reported in Vietnamese populations (5). However, no official data is available on national
prevalence.
Moreover, sporadic cases of OV infection have been occasionally reported in other native Southeast
Asian populations such as Malaysia and Singapore (55). Moreover, OV infection has also been
reported among overseas Thais who work abroad (discussed in Section 1.3.2.1 and Table 1.4).
27
1.3.2 Epidemiology of OV infection in Thailand
1.3.2.1 Prevalence of OV infection
As seen in Table 1.1, prevalence and intensity have been estimated across the country. With regard
to Figure 1.1, the prevalence is high in the north and northeastern of Thailand. Most of the reports
emphasized these areas considered endemic. Previous reports from 1980 – 1990 also focus on the
assessment to support the National Control Program.
The first report of OV infection in Thailand was described by Leiper and Kerr (56) in 1911 where they
found that 17% of 230 adult male prisoners in Chiang Mai, Northern Thailand were infected with
Opisthorchis felineus. It was later confirmed by Sadun (39) in 1965 that reported cases in Thailand
were from Opisthorchis viverrini not Opisthorchis felineus.
Regional surveys aimed to assess the burden of opisthorchiasis from 1950 – 1987. The surveys
supported by the US government from 1950 – 1958 promoted the intestinal helminth and liver fluke
control program in the National policy. After a field trial of treatment was conducted the National
Control Program of OV infection was then finally initiated in 1984 (discussed in Section 1.4). The data
collected from 1984 – 1987 in 4 provinces indicated that the prevalence of OV infection was 63.6%
and the nationwide treatment was started in 1987 (26) under the 6th National Public Health
Development Plan (1987 – 1991) which reported the prevalence before the treatment was 33.9%.
National surveys were reported twice by the 7th (1992 – 1996) and 8th National Public Health
Development Plans (1997 – 2001) (26-28, 48).
For the 7th National Public Health Development Plan 3,007,125 subjects were examined with Kato-
Katz method reporting a national prevalence of opisthorchiasis of 21.5%. The prevalence for each
region was 32.5 % in the North, 15.3% in the Northeast, 16.7% in Central region and 0.1% in the
South.
The National Control Program finished in 2001 at the end of the 8th National Public Health
Development Plan, and the reported national prevalence was 9.7%. Regional prevalence was 19.3%
in the North, 12.4% in the Northeast, 3.8% in the Central region and 0% in the South.
Additionally, national surveys were based on Kato-Katz technique. A report based on the FECT
method was noted from 1982 – 89 in a study conducted on human autopsy cases from fatal car
accidents where stool specimens were directly collected from the large intestine (26).
From 1987 – 2001, prevalence was reported mostly in the northern and northeastern regions. In
1985, the reported prevalence in Khon Kaen was 38.1% in the study of self-pay scheme for diagnosis
28
and treatment of OV infection (57). A report from Ramathibodi Hospital, Bangkok investigated
patients of northeast origin, living in rural and urban areas, who visited or were admitted. The result
should that the prevalence of OV infection among rural dwellers was significantly higher than among
those living in urban areas (79.4% and 54.8%, respectively) (58). A study in Loei and Nong Khai
Provinces in Northeastern Thailand reported that the people living far from the river had a higher
prevalence of the infection than those living on the bank despite the result that bank dwellers
consumed more uncooked fish (59).
29
Table 1.1 Summary of prevalence and intensity of OV infection
Author Study area Study
period
Study
population
Diagnostic
method
Prevalence
(%)
Intensity
(EPG)
Details
Jongsuksuntikul (26-28,
48)
Northeast;
Khon Kaen, Roi-Et,
Sakol Nakorn and
Ubon Ratchathani
province
1984-87 629522 Kato-Katz 63.6 National survey initiated by Department of
Communicable Disease Control
Northeast;
19 provinces
1994 1912 Kato-Katz 18.6 1371.0 Thai-German Community Health Development
through Parasitic Control Project
Consumption of uncooked fish 7.0-42.0 %
Northeast 1987-91 5238062 Kato-Katz 33.9 6th National Public Health Development
Plan (1987–1991)
Nationwide 1992-96 3007125 Kato-Katz 21.5 876.3 7th National Public Health Development
Plan (1992–1996) North 32.6 1226.4
Northeast 15.3 268.5
Central 16.7 1034.6
South 0.1 N/A
Nationwide 1997-
2001
1062725 Kato-Katz 9.7 8th National Public Health Development
Plan (1997–2001)
Consumption of uncooked fish 15.0-59.8 %
North 19.3
Northeast 12.4
Central 3.8
South 0.0
Sithithaworn (60) Northeast;
Khon Kaen province
1982-89 181 Stoll’s dilution
technique
FECT
66.9 53.3 Study conducted in human autopsy cases from
fatal accidents
Feces collected from rectum or terminal colon
30
Author Study area Study
period
Study
population
Diagnostic
method
Prevalence
(%)
Intensity
(EPG)
Details
Mongkolintra (57) Northeast 1985 1417 N/A 38.1
Kurathong (58) Northeast;
Urban area
1987 126 Kato-Katz 54.8
Northeast;
Rural area
433 79.4
Tesana (59) Northeast;
Loei and Nong Khai
province
1991 N/A N/A 41.3 Prevalence of infection among the people
residing far from the rivers was higher than those
residing on the banks
Maleewong (61) Northeast;
Nakhon Phanom
1991 2412 N/A 66.4 uninfected: 33%
light: 59%
moderate: 7%
heavy: 1%
Radomyos (62, 63) Northeastern;
16 provinces
1994 681 Stool
sedimentation
technique
92.4
North;
16 provinces
1998 431 11.6
Sriamporn (64) Northeast;
Khon Kaen province
1990-
2001
18393 Stoll’s dilution
technique
FECT
24.5 Study aimed to determine the association of
prevalence of OV infection and incidence of
cholangiocarcinoma
Waree (65) Northern;
Phitsanulok province
2000 584 FECT 10.8
Wiwanitkit (66) Northeast;
Udonthani province
2001 183 FECT 8.1
Waikagul (67) Northern; 2001 1010 N/A 1.7
31
Author Study area Study
period
Study
population
Diagnostic
method
Prevalence
(%)
Intensity
(EPG)
Details
Nan province
Saksirisampant (68) Various regions 2001 2213 FECT 28.9 Health screening program in Thai workers for
oversea employment conducted in The King
Chulalongkorn Memorial Hospital in Bangkok
Rangsin (69) Central;
Chachoengsao
province
2002 668 Simple-smear
Kato-Katz
FECT
21.3 Study served as baseline population for cohort
study
Suwannahitatorn (70) Central;
Chachoengsao
province
2007 1024 Simple-smear
Kato-Katz
FECT
18.6 Study served as baseline population for cohort
study
2011 1038 Simple-smear
Kato-Katz
FECT
6.2 Unpublished data
Study served as baseline population for PhD
research project
Tungtrongchitr (71) Northeast;
Khon Kaen province
2007 479 Kato-Katz 14.8 Study compared sensitivity of direct simple smear
to Kato-Katz technique
Northeast;
Ubon Ratchathani
province
1124 Simple smear
Kato-Katz
32.0
Kaewpitoon (72, 73) Northeast;
Surin province
2011 333
Kato-Katz 9.9 Study conducted in age ≥ 60 years population
Northeast;
Nakhon Ratchasima
province
2011 1168 Kato-Katz 2.5
Saengsawang (8) Northeast;
Yasothon province
2012 1569 Kato-Katz 38.7
32
Table 1.2 Prevalence of OV infection in overseas Thai population
Author Study area Study
period
Study
population
Diagnostic
method
Prevalence (%)
Intensity
(EPG)
Details
Hira (74) Kuwait 1982-86 N/A FECT 10 cases were
reported
Case series from annual report of in-patient and
out-patient screening program
Estimated screening specimen 18000 per annum
from 4 hospitals
Peng (75) Taiwan 1991-92 1364 FECT 7.0 Thai laborers joining the Six-Year National
Construction Project in Taiwan
Cheng (76) Taiwan 1992-93 302 FECT 43.0 Among infected cases, 77.1 % known to have
infection and 67.4 % received treatment before
arrival
Lo (77) Taiwan 1993-94 N/A FECT 16.6
Greenberg (78) Israel 1994 98 N/A 56.1
Wang (79, 80) Taiwan 1992-96 7670 FECT 7.0
Taiwan 1998-99 702 FECT 2.1
Maneeboonyang (81) Thai-Myanmar
border
2004 286 FECT 19.2 Thai male army personnel operating along Thai-
Myanmar border
33
Table 1.3 Summary of incidence of OV infection
Author Study area Study
period
Study
population
Diagnostic
method
Incidence (%)
Intensity
(EPG)
Details
Upatham (82)
Brockelman (83)
Northeast; Khon
Kaen province
1980 -
1982
1651 Stoll’s
dilution
technique
46.4 % during
1980 – 1981
19.4 % during
1981 - 1982
Incidence is significantly higher in male than
female
Brockelman used catalytic infection model for fit
age-prevalence profile
Sornmani (84) Northeast; Khon
Kaen province
1984 942 Trial
group
442
Control
group
Stoll’s
dilution
technique
2.0 /month in
trail group
5.0 /month in
control group
A pilot trial to assess praziquantel efficacy against
OV infection for community implementation
Upatham (40) Northeast; Khon
Kaen province
1987 -
1988
704 Stoll’s
dilution
technique
87.7 % of re-
infection rate
after 1 year
follow-up
Re-infection rate was assessed 1 year after
treatment with praziquantel
Rangsin (69) Central;
Chachoengsao
province
2002 -
2004
526 Simple-smear
Kato-Katz
FECT
21.6/100
person-years
Suwannahitatorn (70) Central;
Chachoengsao
province
2007 -
2009
980 Simple-smear
Kato-Katz
FECT
21.4/100
person-years
34
Concentration techniques were used from 1994 – 2001 studies. Radomyos el al. (62, 63) reported
prevalence from stool sedimentation technique: 92.4% from 16 northeastern provinces and 11.6%
from 16 northern provinces. A cohort study was conducted among 18,393 subjects from 1990 – 2001
to measure the incidence of CCA (64); the study aimed to determine the association of OV infection
prevalence with incidence of CCA. The reported prevalence was 24.5% using FECT. More reports
with FECT technique were in Phitsanuloke (65) and Nan (67) Provinces in Northern Thailand with
prevalence of 10.8% and 1.7%, respectively. Another survey in Udon Thani, northeastern Thailand
(66) reported a prevalence of 8.1%.
In 2002, Rangsin et al. (69) initiated a health study in a rural area of Chachoengsao, central Thailand.
The multiple fieldwork projects reported on the prevalence of OV infection. Published data (69, 70)
indicated that prevalence was 21.3% in 2002 and 18.6% in 2007. Fieldwork prevalence will further
discussed.
A study in 2007 (71) compared the sensitivity of direct simple smear and Kato-Katz technique
reporting a prevalence of 14.8% in Khon Kaen and 32.0% in Ubon Ratchathani Provinces. Prevalence
surveys in northeastern provinces reported various geographic prevalence of 9.9 – 38.7 % from 2011
– 2012 (72, 73).
Additional prevalence was reported among overseas Thais (Table 1.2) who worked in Taiwan, Israel
and Kuwait. All of them were examined in their working country during employment. The population
movement caused impact on population dynamics and parasite transmission dynamics. In 2001, a
health screening program was conducted for Thai workers before overseas employment (68)
reporting that 28.9% of 2213 screened candidates were infected with OV infection.
The study of Thai workers in Taiwan was under the Six-Year National Construction Project from 1991
– 1997 (75, 77, 79) indicating the prevalence ranged from 7.0 to 43.0%. Additional health screening
from 1998 – 1999 reported a prevalence of 2.1% (76). The study suggested the prominent
characteristics of particular local Thai uncooked fish dishes were associated with OV infection.
The infection was also reported from Middle East countries, which were nonendemic area, and
where uncooked fish consumption was not a regular habit. One case report was described in Kuwait
(74) for 10 Thai migrants infected with Opisthorchis viverrini. The screening test for Thai workers in
Israel (78) found a prevalence of 56.1%.
Maneeboonyang (81) reported the prevalence among Thai male army personnel operating along the
Thai-Myanmar border in Ratchaburi Province was 19.2%. The area was endemic but almost all
personnel were northeastern Thais.
35
1.3.2.2 Infection intensity of OV infection
As seen in Table 1.1, available data for infection intensity were considered limited during the
implementation period of the National Control Program from 1992 – 1996 (26, 27). A 1994 survey
supported by the Thai-German Community Health Development though Parasitic Control Project
reported a mean EPG of 1371.0 among 1912 subjects. From the 7th National Public Health
Development Plan from 1992 – 1996, the national mean EPG was 876.3 and the highest mean EPG
was in the northern region.
A study conducted among human autopsies reported an EPG of 53.3 (60) from stool samples. A
study conducted in 14 villages in Nakhon Phanom Province near the Mekong Basin on the Thai-Lao
PDR border (61) in 1994 reported a prevalence of 66.4%. Considering the infected cases, 59% of
cases were light intensity, only 1% had heavy infection compared with the 1996 National Survey
reporting a heavy infection of 1.9% (27).
A notable study of infection intensity was conducted in 1988 in the evaluation of praziquantel
treatment where the intensity of infection was measured between pre- and posttreatment. The
result stated that high pre-treatment intensity subjects were more likely to have a heavier intensity
of re-infection indicating the predisposing factors contributing to heavy infection (40).
Ramsey (85) and Haswell-Elkins (86) conducted stool expulsion with chemotherapy from 1989 - 1991
and found that heavy intensity was aggregated in a small proportion of the infected population and
over-dispersion distribution was suggested. However, no recent data regarding infection intensity
and its distribution characteristics of OV infection have been reported in Thailand. This study
provides up-to-date data regarding the infection intensity.
36
1.3.2.3 Incidence of OV infection
Although data on prevalence are available, estimates of the incidence of OV infection are extremely
limited. From Table 1.3, notable studies of incidence were conducted from 1980 – 1982 in a rural
area of Khon Kaen Province, northeastern Thailand. Upatham et al. (82) conducted a fieldwork
measuring incidence within 2 periods: 1980 – 81 and 1981 – 82. The incidence was 46.4% and 19.4%,
respectively. Brockelman et al. (83) later used a catalytic model for fitting age-prevalence data from
the fieldwork to estimate incidence from mathematical modeling compared with directly-estimated
incidence (discussed in Chapter 2).
Sornmani et al. (84) conducted a pilot clinical trial to assess the efficacy of praziquantel in 1984 to
implement the treatment in the National Control Program. The result indicated that incidence in the
control group was 5.0% monthly compared with 2.0% monthly in the trial group. Upatham (40)
assessed the re-infection rate after praziquantel treatment. One-year follow up in a cohort
population revealed an average re-infection rate of 87.7% yearly.
Rangsin (69) reported the incidence of OV infection in northeastern-originated community in
Chachoengsao Province, central Thailand was 21.6/100 person-years from 2002 – 2004.
Suwannahitatorn (70) later re-evaluated the incidence in the same study area and reported the
incidence of 21.4/100 person-years from 2007 – 2009. Moreover, both Rangsin and
Suwannahitatorn cohort studies were conducted with similar methodology. Only negative cases for
OV infection from baseline cross-sectional study were re-examined in follow-up study which could
resulted in bias when the incidence of OV might be underestimated.
1.3.3 Source of infection
From Section 1.2.1, mode of infection is consumption of uncooked fish containing metacercariae.
Therefore, the primary mode of infection will be intentionally consumed uncooked fish (2). Utensils
and hands contaminated with fish scraps are also responsible for the infection.
OV infection is endemic in the north and northeastern parts of Thailand where eating habits closely
follow local traditions. Uncooked fish is considered popular resulting in various dishes based on this
method of preparation. In Thai including the local dialect, the word “Pla” means fish;
1) Instantly prepared uncooked fish: freshwater fish is chopped and mixed with spicy ingredients.
Raw fresh meat is usually denatured by lime juice due to acidity which dramatically changes its
colour to a cooked-like texture. One Ill-known popular dish is Koi pla (chopped raw fish salad). Koi
pla is always consumed instantly. The variation of Koi pla is Larb pla, fish meat that is thoroughly
chopped and mixed with particular ingredients.
37
2) Moderately fermented fish: fish meat is preserved in highly-concentrated salt for a few days to
one week. The period of preservation varies and might affect the viability of metacercariae. Studies
have reported that fish preserved less than 7 days were likely to be infective (47, 87) The viability of
metacercariae depends on the concentration of preserving solution and the period of preservation
(88), which can be extended to minute intestinal flukes as Ill. Popular dishes for moderately
fermented fish are Pla som and Pla jom.
3) Extensively fermented fish: freshwater fish are preserved in highly concentrated salt solution for 3
- 6 months. The dish is widely known as Pla ra. Chunks of fish could be consumed uncooked and the
remaining fish sauce is often used as a daily ingredient for local dishes (29, 87, 89).
Various choices of uncooked dishes result in a variety of eating practices. The pattern of
consumption behavior differs regarding the particular uncooked fish dish. Koi pla is consumed as a
main dish and limited to special occasions such as social drinking, religious ceremonies or festive
celebrations. However, the dish is popular with respect to a high proportion of individuals who have
experienced this dish. Pla ra itself has different patterns of consumption; the fish sauce is the main
ingredient for local dishes, for example, local Thai papaya salad or well-known Som tum is dressed
with Pla ra sauce for local consumption. The consumption behaviors are considered related to social
factors. However, the underlying social influence dynamics have not yet been extensively studied
(9). Despite the importance of epidemiological data, a study sociocultural dynamics relating to
Opisthorchis viverrini in the Mekong Basin (29) by Grundy-Warr et al. suggested the importance of
social dynamics, which should be incorporated with parasite infection dynamics, human behaviors
and public health awareness.
Consumption of Koi pla and Pla ra were firstly described by Sadun in 1955 (39) for the method of
preparation. Association of Koi pla consumption with risk of OV infection are mostly addressed by
cross-sectional studies (25, 58, 90, 91). The prevalence of Koi pla consumption was high in the north
and northeastern Thailand, up to 80%. The pattern was still observed even among overseas Thai
workers, who were from endemic areas of OV infection (68, 79, 80). Based on the cross-sectional
design, the causal relationship of uncooked fish consumption and OV infection could not be fully
explained due to the exposure and outcome is measured at the same time. However, the National
Control Program has implemented interventions on avoiding uncooked fish based on given biological
properties consistent with the mode of infection. Without a clear underlying explanation regarding
the epidemiology of OV infection and its risk factors, the intervention package could have issues with
efficacy especially with primary prevention, which could prevent newly-infected individuals.
38
A cohort design can provide better understanding in terms of causal-relationship and data with
which to estimate the incidence of infection, which is important considering the role of prevention
and control campaigns. As described in Section 1.3.2.3, the data on incidence are scant and available
data are considered outdated. The incidence from follow-up studies conducted in Na-yao area rom
2002 – 2009 (69, 70) provided up-to-date data and highlighted the relationship of Koi pla
consumption and OV infection. My study will analyze more data on risk factors related to infection
intensity, which could provide more understanding of infection dynamics.
39
1.4 Control of OV infection in Thailand
1.4.1 Timeline for OV infection control in Thailand
The history of OV infection dates back to 1950 when the field survey was initiated by the
government. The timeline to manage OV infection in national level can be divided in 3 main phases
described below.
1.4.1.1 Survey phase
This phase included the assessment of parasitic infection burden, field study of prevention and field
trial of chemotherapy.
First, a wide-scale survey was conducted from 1950 – 1958 by the Ministry of Health with technical
and financial support from the US government. The Helminthiasis Control Unit was organized to
diagnose and treat intestinal parasitic infections including OV infection. The service was provided to
5 provinces, 4 provinces in the northeastern region (Nakhon Ratchasima, Udon Thani, Sakon Nakhon
and Ubon Ratchathani) and Songkhla Province in the Southern region.
The support from the US government was terminated in 1958 resulting in the termination of The
Helminthiasis Control Unit. However, the service scheme was still active and was later integrated
with the rural health development project of the Health Development Division. The main strategy
still targeted diagnosis and treatment.
The implementation of prevention campaigns was conducted from 1967 – 1974 under the Liver
Fluke Control Unit. The trial targeted primary prevention in discontinuing uncooked fish
consumption and preparation. The pilot study was conducted in Sakon Nakhon Province
emphasizing health education including introducing methods for cooked fish preparation and also
distributed cooking pots to support cooking equipment.
So far, the implementation of OV infection control from 1950 – 1974 included health education as
primary prevention and diagnosis and treatment as secondary prevention.
From 1980 - 1983, praziquantel efficacy was firstly assessed as a widely used chemotherapy. The
Faculty of Tropical Medicine, Mahidol University initially reported the efficacy of a single dose of 40
mg /1 kg body weight for praziquantel treatment against OV infection that provided a cure rate of
91%.
The collaboration was established between Mahidol University and the Department of
Communicable Disease, Ministry of Health to extend the study to field trials which reported a cure
rate with praziquantel of 95.5%.
40
Furthermore, the Helminthiasis Section of the Department of Communicable Disease set up liver
fluke control units in 4 northeastern provinces; Khon Kaen, Roi Et, Sakon Nakhon and Ubon
Ratchathani. The result should that treatment was provided to 400,452 infected cases from 629,522
infected cases (approximately 63.6%, Table 1.4) from 1984 – 1987.
1.4.1.2 National Control Program phase
This phase included the implementation of the National Control Strategy following 3 main
approaches. This phase emphasized the control program and evaluated the reduction in prevalence.
The scheme was extended to national scale where the activity was prioritized and incorporated in
the National Public Health Development Plan from 1987 – 2001. The National Control Program was
integrated from the 6th to 8th National Public Health Development Plans following strategic activity.
Figure 1.5 is adapted from Jongsuksuntigul et al. (28) summarizing 3 main strategic concepts for
controlling liver flukes. The primary endpoint was to reduce prevalence of OV infection. The
campaign included primary and secondary prevention.
Primary prevention: the prevention scheme focused on providing health education to promote
cooked fish consumption.
Figure 1.5 Strategic approach for liver fluke control
41
Secondary prevention: the mobile stool examination unit was organized to provide service once a
year including diagnosis and treatment for OV infection. A small fee was collected from the
participants for the diagnosis (10 baht – 20p) and additional 10 baht for positive result which later
treated with praziquantel. Hygienic defecation was promoted through health education to reduce
the transmission of fluke egg in natural water bodies.
Community participation was engaged by the mobile stool examination in which the strategy
prioritized case seeking and treatment. The activity continued for consecutive years. If prevalence
decreased below 10% in any community, the program switched to passive surveillance strategy.
Therefore, the outcome from primary prevention was not directly determined as the incidence of
infection was not obtained. Moreover, outcome from providing health education was not yet
quantified. Re-infection rate and infection intensity were not measured as well; therefore, the
infection dynamics regarding the program efficacy solely relied on the report of prevalence of
infection.
42
Table 1.4 Summary of annual prevalence during National Control Program implementation
National Public Health
Development Plan Year No. positive/total N Prevalence (%) 95% CI*
Pre-implementation 1984-87 400452 / 629522 63.6 63.5 – 63.7
6th
1988 160308 / 450677 35.6 35.4 – 35.7
1989 407309 / 1343110 30.3 30.2 – 30.4
1990 533147 / 2193275 24.3 24.2 – 24.4
1991 437241 / 2747011 15.9 15.8 – 16.0
7th
1992 363689 / 2876423 12.6 12.6 – 12.7
1993 311809 / 2834725 11.0 10.9 – 11.0
1994 300720 / 2766837 10.9 10.8 – 10.9
1995 330780 / 3130563 10.6 10.5 – 10.6
1996 365010 / 3007125 12.1 12.1 – 12.2
8th
1997 125805 / 1532678 8.2 8.1 – 8.3
1998 128037 / 1375915 9.3 9.2 – 9.4
1999 110920 / 1234408 9.0 8.9 – 9.0
2000 97419 / 1448877 6.7 6.6 – 6.8
2001 100059 / 1062725 9.5 9.2 – 9.6
* revised from original paper
The prevalence from Table 1.4 was modified from Jongsuksuntigul P et al. (28). The prevalence in
1995 was corrected to 10.6% from direct calculation; 330780/3130563 as it reported 11.0% from the
original Table. The 95% CI was calculated by binomial distribution regarding the reported results.
A large-scale field trial of praziquantel efficacy was conducted from 1984 – 87. The successful cure
rate led to expanding the treatment coverage to the National Control Program implementation,
initiated in the 6th National Public Health Development Plan from 1987 – 1991. The program was
applied to all health facilities in the northeastern region of Thailand under the supervision of the
Department of Communicable Disease Control. Stool examination was conducted among 5,238,062
participants of whom 1,774,929 were diagnosed with OV infection and treated with praziquantel;
33.89% (95% CI: 33.84 – 33.93).
Moreover, the Promotion of Community Health through Parasite Control Project was conducted by
the support of Germany to provide stool examination and treatment in 7 northeastern provinces. A
total of 1,839,813 cases were examined while 531,174 cases were positive and treated; 28.87% (95%
CI: 28.81 – 28.94).
43
As shown in Table 1.4, the success in reducing the prevalence from 63.6 to 15.9% by the end of the
6th plan resulted in continuing the programs in the following 7th (1992 – 1996) and 8th (1997 – 2001)
National Public Health Development Plans. The implementation covered all provinces in the
northeast and some provinces in the central region resulting in 42 provinces receiving coverage
(55.3% from 76 provinces). From1992 – 1994, 5,238,062 cases were examined and 632,869 infected
cases were treated; 12.08% (95% CI: 12.05 – 12.11).
Figure 1.6 shows the prevalence trend from the data in Table 1.4. From 1997, the prevalence fell
under 10% and continued to remain below 10% for 5 consecutive years until the end of the 8th
National Public Health Development Plan as shown in Table 1.6. Therefore, the program was
considered to have achieved the goal and then switched to passive surveillance strategy. Figure 1.6
shows the prevalence trend from the data in Table 1.4. From 1997, the prevalence fell under 10 %
and continued to remain below 10 % for 5 consecutive years until the end of 8th National Public
Health Development Plan as shown in Table 1.6. Therefore, the program was considered archiving
goal and then switched to passive surveillance strategy.
Figure 1.6 Prevalence of opisthorchiasis under National Control Program Implementation 1987 - 2001
0.64
0.36
0.30
0.24
0.16
0.130.11 0.11 0.11
0.12
0.080.09 0.09
0.07
0.09
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Pre
vale
nce
(as
pro
po
rtio
n)
Year1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Years
44
1.4.1.3 Passive surveillance strategy phase
From the 9th National Public Health Development Plan to the current 11th plan (2001 up to the
present), the control strategy was merged with rural health services. The scheme focuses on
targeted areas where the infection is still a burden. Therefore, the priorities of the control activity
depend on the situation and policy in each particular province.
From Table 1.1, the cross-sectional study conducted after 2001 focused on the northeastern and
northern provinces which were identified as endemic areas. This study was conducted from 2002 –
2007 in Chachoengsao Province, central Thailand, which was considered as a non-endemic area
showing that prevalence and incidence were high (69, 70). This issue raised a question regarding the
infection dynamics of OV infection for the distribution and the factors contributing to the infection.
45
1.5 Main conceptual framework
In this project, opisthorchiasis will be defined as OV infection which do not examine the clinical
disease, but rather emphasize on epidemiological parameter in infection dynamics.
From reviewed literature and available data, gaps were presented with regard to infection dynamics
of OV infection. A comprehensive study was required to establish up-to-date knowledge in various
aspects.
The study aim was to explore the infection dynamics of OV infection to fill the gap with regard to
various aspects of the infection. Based on the main conceptual framework in Figure 1.7, the infection
dynamics were considered a complex interaction shaped by multiple factors as described below.
OV infection risk is affected by risk behaviors and also shaped by risk factors. The aim was to
explore the epidemiology of the infection using fieldwork study with rigorous methods. The
Risk behaviors
Social influence
Opisthorchiasis
Intervention
Prevention Diagnosis and Treatment
Figure 1.7 Main conceptual framework of study project
46
results included the up-to-date prevalence and incidence of infection. The risk factors were
identified with a better explanation of the causal relationship through the cohort study design.
The infection intensity was examined. Other studies addressed the over-dispersion and density-
dependent property (85, 86), which will be further explored in this thesis. Statistical modeling
was formulated to analyze the relationship between risk factors and OV infection with respect to
its intensity distribution.
The relationship between prevalence and incidence was described with mathematical modeling.
The model was fitted to fieldwork data to explore the infection dynamics.
Socio-culture dynamics have been important (29) in determining the prevalence of risk
behaviors. The qualitative approach was introduced to obtain social-related data. The qualitative
method allowed me to collect and analyze underlying bio-psycho-social aspects of the infection
and explore responses and interactions from individual to community levels to and with the
infection.
Quantitative and qualitative methods were combined as comprehensive data to design a
planning model for public health implications to conduct a pilot community trial study to explore
the effectiveness and feasibility of the intervention.
Effectiveness of primary prevention was assessed through community trials and secondary
prevention was assessed by statistical and mathematical modeling fitted to fieldwork data to
determine the cure rate from praziquantel treatment.
Comprehensive data provided more insights about infection dynamics. The introduction of
qualitative methods, novel statistical and mathematical modeling completed the gap of research
questions addressed in previous studies, which was crucial for the future implementation of OV
infection studies in Thailand.
47
1.6 Study aims and thesis outline
The project aims to comprehensively study the infection dynamics of Opisthorchiasis from OV
infection by using multiple approaches with regard to statistical analysis, mathematical modelling,
social science and public health planning.
The thesis comprises 6 chapters. Sub-objectives are outlined below.
Chapter 1: The first chapter aims to explain the current situation of OV infection in Thailand and the
role of the National Control Policy.
Chapter 2: This chapter explores the distribution of OV infection included prevalence, incidence and
infection intensity. The age-prevalence-intensity models were formulated allowing for over-
dispersion (for example, using zero-inflated distributions), a common characteristic of helminthic
density distributions.
Chapter 3: Chapter three constructs mathematical modelling to explore the infection dynamics and
the effectiveness of praziquantel treatment using obtained fieldwork data.
Chapter 4: This chapter explores the complexity of bio-psycho-social aspects to explain risk-
behaviour relationships using qualitative methods.
Chapter 5: This chapter introduces the principle and theory behind the community-based approach,
the relationship between disease biology, human factors and environmental context.
Chapter 6: The final chapter summarizes the overall picture of disease burden, what the current
distribution and risk factors are and how intervention would impact on disease and community.
Research data would extend knowledge of disease epidemiology and reflect the current situation of
OV infection to initiate a community intervention package (CIP) and expand this work on the
national scale.
48
1.7 Project outline
The study area was located in Tha-kradan Subdistrict, Sanamchaikhet District, Chacheongsao
Province in central Thailand as shown in Figure 1.8. Chacheongsao Province is about 100 km east of
Bangkok. However, the research site was isolated and remotely located from the centre of the
province.
Figure 1.8 Overview of study area
The fieldwork was conducted in 4 areas; Na-yao, Na-ngam, Na-isarn and Tung-heang Villages. The
total population consisted of approximately 5000 residents.
1.7.1 Background of study community
Before the settlement of Tha-kradan and neighboring villages, the geographical area was covered
with jungle comprising natural ponds and small rivers making it rich in natural resources. Therefore,
a private sawmill was later opened and operated under governmental authorization for the wood
industry. The “Sukapoom” was the first operational sawmill located in a nearby area along the
riverside to produce raw wooden materials for the industrial sector.
Most of the workers were employed from the northeastern region of Thailand. Some of them were
working in the wood factory as a part-time job during off-harvest season and returned to their
hometown when the rice farming started in the early rainy season.
49
The northeastern region of Thailand was originally arid and droughty and sometimes made the locals
find it difficult to make a living and the Tha-kradan area was also more resourceful and more suitable
for agriculture. Therefore, some native villagers chose to emigrate and settle permanently in the
Tha-kradan area.
The very first settlement comprised small groups of people from various provinces in the
northeastern moving to the Tha-kradan area in 1975; they claimed and possessed large areas of
empty lands. The following immigrants purchased lands from them at a very cheap price; 40 pence
for 1 rai (local Thai unit, approximately 1600 m2 or 0.4 acres). Each household owned more than 100
rai because title deeds for land ownership were not yet available.
The first settled community was peaceful but not uniformly organized. Each household lived on their
own creating a problem when the community grew bigger. In 1983, the Border Patrol Police (BPP)
managed to take care of the community safety. Because the area was close to the country’s border,
incidences of drug smugglers and conflicts were encountered from minority tribes, so the BPP setup
a community center and relocated the outer households to be gathered together as a whole
community.
In 1984, the community leader cooperated with the BPP to organize the landscape planning; areas
were designated for schools and a temple later built by community members. The landscape was
divided in 4 rai squares (800x800 m). Each block was surrounded by road and filled with 4
households. The school was under BPP supervision; their staff served as teachers and also
maintained border security.
From 1989 - 91, the community grew rapidly; current households attracted their relatives to move
in. The lands were purchased at much higher prices. Compared with 1975, the pricing was 20 times
higher in 1983. Within 15 years, the land price was 100 times higher in 1990.
The rapidly growing community resulted in resource management. Even though the landscape was
Ill-planned, the community setting was rural. They still lacked fundamental infrastructure.
In 1994, HRH Princess Maha Chakri Sirindhorn visited the BPP School and noticed that poverty still
affected the villagers’ quality of life. Agriculture lacked adequate technology resulting in
underproduction and irrigation was poorly managed. HRH the princess took the BPP School under
her patronage and also setup an office for community development. The scheme was under her
major royal projects covering all BPP schools located along the national border. The office for
community development has since been continuously operated and she has makes regular visits to
Na-yao BPP School until the present.
50
Agriculture has remained a major occupation since the community foundation. In the early period,
most agricultural works involved the timber industry as they were primarily sawmill workers. After
the forest was reclaimed in the very first period, the landscape was transformed into lowlands
suitable for rice field. As they kept moving to higher ground, the upland became suitable for farm
plants such as corn and sesame.
The forest area dramatically decreased raising environmental concerns. The authorities later
declared the remaining forest was a conservation area, which ended the timber industry in the area.
As a consequence, agricultural areas expanded. The villagers started to grow various kinds of plants
including cassava, soybeans, jute, green beans and sunflowers. However, corn was still a major
agricultural product. The lowland was exclusively suitable for rice fields because it needed large
amounts of water. The authorities supported the villagers by promoting rice seed and proper
irrigation.
As the community grew, the commercial sector also expanded. The central market was set up to
exchange local products. Local groceries imported goods such as clothes and household appliances
from external suppliers so the villagers did not need to travel downtown. The local cooperative was
formed in 1995 to facilitate the distribution of agricultural products.
Up to the present, the villages are still considered a rural environment in context where the majority
of residents still preserve Northeastern culture. The inflow of modernization has resulted in the
development of infrastructure and thus started the transitional period to urbanization. However, the
local traditions including the local dialect, local folk customs, and local calendar ceremonies and
eating behaviours are still widely maintained.
51
1.7.2 Background of community healthcare system
In the beginning, no official health care facilities existed in the community either modern medicine
physician or traditional doctor. Villagers relied on self-care or herbal medicine. In 1984, a major
outbreak of malaria resulted in more than 200 cases with over 20 fatalities. The fatal outbreak
triggered health concerns within the community.
In 1987, local clinics were opened by two medical doctors providing the first local healthcare. One of
them was obstetrics, so modern delivery was also available. Morbidity and mortality from malaria
were still reported annually but continued to decrease. In 1990, the malarial unit was setup to
provide immediate diagnosis and treatment to the villagers and local clinics also grew.
The local healthcare center was established in 1995 located within the temple area in the middle of
the community. Three staff were stationed at the facility providing primary health care and some
secondary healthcare services. One of the staff was a medical doctor who also opened his own
private clinic. However, the facility was not always equipped with medical doctors. The majority of
staff was health officers and nurses.
The local healthcare center was under district hospital supervision. Medical doctors paid regular
visits especially for chronic disease appointments. Any cases beyond the facility’s scope were
transferred to higher level healthcare. The closest facility was the district hospital, a 1-hr drive away.
The road was the only exclusive method of transportation for referrals.
52
1.7.3 Overview of study methods
Table 1.5 Summary of fieldwork data collection
Area Year Source of data Study type
Study outcome Availability of study outcome
Outcome Egg count Risk factor
assessment
Na-yao
2002 Secondary Cross-sectional Prevalence Yes No
2004 Secondary Follow-up Incidence Yes Yes
2007 Secondary Cross-sectional Prevalence No No
2011 Secondary Cross-sectional Prevalence No Yes
2013 Primary Follow-up Incidence No Yes
Na-ngam 2012 Primary
Cross-sectional
Qualitative
Prevalence
Qualitative
Yes
No
Yes
Yes
2014 Primary Follow-up Incidence No Yes
Na-isarn 2013 Primary Cross-sectional Prevalence Yes Yes
Tung-heang
2010 Secondary Cross-sectional Prevalence No Yes
2012 Primary Community trial
Qualitative
Incidence
Qualitative
No
No
Yes
Yes
As seen in Table 1.5, the method of data collection was based on fieldwork study since 2007. Data
obtained from fieldwork came from 2 sources; primary data was directly obtained by the research
team. Secondary data was obtained from earlier fieldwork. The method was a mixed-method design
comprising quantitative and qualitative methods.
The quantitative method consisted of two approaches: cross-sectional and cohort design. Enrolled
participants were required to provide stool examination and undergo questionnaire assessment.
A cross-sectional study was conducted in Na-ngam and Na-isarn Areas to assess disease burden. A
follow-up study was conducted in Na-yao and Na-ngam Areas for which the corresponding cross-
sectional study served as baseline. Each participant was required to complete a standardized
questionnaire to assess basic demographic data and risk behaviours for acquiring OV infection.
The qualitative techniques used in this study were focus group discussions and in-depth interviews
conducted in Na-yao and Tung-heang Area.
For community trials, a planning model was applied to the community to systematically engage OV
infection. The community intervention was developed with an effort to build community
participation concerning local and environmental context under health behavioural theory to archive
goals to reduce OV infection through behavioural modification by reducing risk behaviours.
53
Chapter 2 Epidemiology of Opisthorchis viverrini infection:
distribution and risk factors
Abstract
Opisthorchis viverrini infection or OV infection is a pathologically important food-borne trematode,
where transmission to humans occurs through the consumption of uncooked fish containing
infective stage metacercariae leading to infection within the hepato-biliary system. Chronic infection
is strongly associated with to a bile duct cancer or cholangiocarcinoma, where Thailand holds the
highest ranked incidence in the world. Cholangiocarcinoma can be partially prevented through
prevention and control measures against the parasite itself. However, OV infection is acknowledged
as a neglected and underestimated disease on a global scale and the available epidemiological data
are considered outdated.
This study aimed to provide the most up-to–date and comprehensive study of the epidemiology of
OV infection in Thailand. Prevalence and incidence were directly measured through both cross-
sectional and follow-up studies in three study areas: Na-yao, Na-isarn and Na-ngam Areas during
2012 – 2014 in population 1 – 87 years of age. Risk factors for acquiring the infection were assessed
by age-group and uncooked fish consumption and identified through statistical modelling. Infection
intensity was determined to reflect the infection burden through eggs per gram. Regression models
for count outcome were used to examine patterns of intensity distribution including Poisson and
negative binomial distributions. Compound model, zero-inflated and hurdle models were
incorporated to incorporate over-dispersed and zero excess data.
For age structure, quartiles were used for equally-proportioned age groups. The prevalence of OV
infection was 7.85% (95% CI: 4.96 – 10.75) in Na-isarn and 9.29% (95% CI: 6.64 – 11.93) in Na-ngam
area. The incidence rate was 7.98/100 person-years (95% CI: 5.49 – 11.20) in Na-yao and 6.80/100
person-years (95% CI: 4.68 – 9.54). The infection intensity data were over-dispersed, suggesting that
the use of negative binomial distribution was more appropriate than Poisson distribution. A negative
binomial hurdle model was the model of choice for zero excess and over-dispersed count data in this
study. Final models included logistic regression model for prevalence data, Poisson model for cohort
data and negative binomial logit hurdle model (NBLH) for intensity data with two covariates: age
group and consumption of Koi pla (local instantly-prepared uncooked fish). The results
demonstrated that consumption of Koi pla was a potential risk factor for acquiring the infection as it
showed a significant association from 5 of 6 models. Koi pla consumption was also shown to
increases the risk of OV infection and infection intensity when incorporated with the NBLH model.
54
Increasing age-groups linearly increased infection risk when compared with the 1st quartile age-
group in the NBLH model but not for infection intensity. Prevalence linearly increased with age.
Incidence was calculated from prevalence through the regression model. The age-prevalence-
intensity relationship indicated an age-structure component within the intensity and level of parasite
aggregation. Assuming incidence is uniform in all ages, the intensity tended to increase and the
parasite spread more equally.
Up-to-date information on disease dynamics has a potential impact on public health and further
ongoing extensive research in this field is essential to provide effective public health management.
55
2.1 Introduction
OV infection is a parasitic infection caused by human liver flukes. Three major species are
responsible for the infection (1-4): Opisthorchis viverrini (OV), Opisthorchis felineus and Clonorchis
sinensis (infection is also called clonorchiasis). As described in Section 1.3, OV infection is a
pathologically important food-borne trematode, where transmission to humans occurs through the
consumption of uncooked fish containing infective stage metacercariae (12, 13), leading to infection
within the hepato-biliary system (15-17). Many studies have indicated that chronic infection is
strongly related to a bile duct cancer, cholangiocarcinoma CCA (17, 19-21, 92). In Thailand,
knowledge of the current incidence of infection is extremely limited and the available prevalence
data are considered outdated in many areas (5, 9, 30, 40, 42, 69, 83). The record showed
geographical variation in infection prevalence. The prevalence was higher in the North (19.3%) and
the Northeast (15.7%), where people regularly consume dishes containing uncooked freshwater fish
(27, 28). Consumption patterns are strongly influenced by culture and traditional beliefs and a
similar pattern was also observed throughout the Mekong basin region. Endemic areas were thus
revealed to be confined to where intermediate hosts and uncooked fish consumption behaviours are
present.
In late 2002, Rangsin et al. (69) conducted a health survey in Na-yao Area, a rural community located
in central Thailand, and found an infection prevalence of 21.3% (95% CI: 18.2 – 24.6). The negative
cases were enrolled in a follow-up study in February 2004. The incidence of OV infection and risk
factors were evaluated by a community-based cohort design showing the incidence was 21.6/100
person-years. In 2007 - 2009, Suwannahitatorn et al. (70) re-evaluated the situation in the same area
found prevalence was 18.6% and incidence was at 21.4/100 person-years. This indicated a high
prevalence area outside endemic areas where the national control strategy is no longer targeted.
The study of the prevalence and incidence can be extended to infection intensity. Studies of
helminths such as Schistosoma haematobium, S. mansoni, Ascaris lumbricoidis, Trichuris trichiura
and hookworm, for example (93-95), all tend to be characterized by a pattern of infection intensity
with over-dispersed distributions, where few individuals are infected with a large burden of worm
load while the majority are infected with light or no infection. Such a typical over-dispersion
characteristic of helminth infection raise technical issues to analyses, such as in relation to excess
zero count outcomes and parasite aggregation. Conventional statistical models such as the Poisson
model may be technically problematic to properly explain such a distribution. Extended models can
therefore be introduced to count model families to explore the issues such as negative binomial
distribution or compound model.
56
My hypothesis is that Opisthorchis viverrini infection will be characterized by an over-dispersed
distribution pattern. Therefore, compound models including zero-inflated and hurdle models were
introduced in this study to characterise infection intensity. The model framework was useful for
excess zero data (96, 97) and are applicable in many areas, such as economics, ecology and
agriculture. Zero-inflated models are also becoming more popular in medicine (97). For helminth
epidemiology, however, the application of the compound model remains limited (95).
As illustrated in Figure 2.1, I will explore the distribution of OV and evaluate risk factors contributing
to the infection though statistical modelling. The study aimed to:
Explore the prevalence and incidence of OV and evaluate risk factors for acquiring the
infection, and
Explore the infection intensity; distribution pattern, population dynamics of age-prevalence-
intensity data and apply models for over-dispersion and excess zeros in count outcomes.
Risk factors
Social influence
Opisthorchiasis
Intervention
Prevention Diagnosis and Treatment
Figure 2.1 Conceptual framework for distribution and risk factors of opisthorchiasis
57
2.2 Methodology
2.2.1 Study areas and population
The study area was located in Na-yao, Na-isarn and Na-isarn area located in Tha-kradan Subdistrict,
Sanamchaikhet District, Chacheongsao Province in central Thailand. Population characteristic can be
referred to Section 1.7.1 for details.
Table 2.1 Study areas for risk and distribution analysis
Area Year Source of data Study type
Data collection
Infection
outcome
(yes/no)
Intensity
(egg count) Risk behaviours
Na-yao
2002 Secondary Cross-sectional Yes Yes No
2004 Secondary Follow-up Yes Yes Yes
2011 Secondary Cross-sectional Yes No Yes
2013 Primary Follow-up Yes No Yes
Na-ngam 2012 Primary Cross-sectional Yes Yes Yes
2014 Primary Follow-up Yes No Yes
Na-isarn 2013 Primary Cross-sectional Yes Yes Yes
This chapter comprises two sources of data. Primary data were obtained from the fieldwork from
2012 - 2014. Secondary data were obtained from the Na-yao area, which was surveyed before and
conducted as a continuing study as summarized in Table 2.1.
The Na-yao 2002 - 04 published study (69) by Rangsin and colleagues described incidence and risk
factors associated with OV.
A cross-sectional study was conducted in Na-ngam and Na-isarn areas to assess disease burden by
comprehensive survey.
Follow-up population was enrolled from corresponding baseline cross-sectional study. A follow-up
study was conducted using a community-based retrospective cohort design in Na-yao and Na-ngam
areas. From baseline cross-sectional study, negative cases for OV infection were eligible for a current
follow-up to assess the incidence of OV. Enrolled participants were also required to report their risk
behaviours retrospectively from the time they joined the survey
58
2.2.2 Fieldwork procedure
Figure 2.2 Flowchart of fieldwork
The fieldwork protocol for cross-sectional and follow-up studies was similar, as illustrated in Figure
2.2. Fieldwork was organized from the community health centre in the middle of the research area
where a field laboratory and data processing unit were set up. The team worked together with
village heath volunteers to co-operate with the research participants. Eligible participants were
invited in the study. The enrolment process included providing written informed consent forms and
some participants were excluded from the study by exclusion criteria.
After the enrolment process, each participant received a survey package which included a
questionnaire and a sealable container for collecting a stool specimen. Stool examination was
performed after specimen collection. The stool examination result and questionnaire data were
recorded and matched for subsequent data analysis. Participants received their personal stool
examination result; reporting intestinal pathogenic parasite infection such as helminth and protozoa
infection, and health education based on National Health Guidelines. Participants with OV received
proper medical treatment depending on their identified pathogen.
Eligible subject
Informed consent
Enrolled subject Receive survey package
Stool container
Questionnaire
Stool examination
Positive for parasitic infection
Health education
Receive medical treatment
Negative for parasitic infection
Health education
Excluded from exclusion criteria
Receive stool examination result
Data management
Questionnaire data entry
Stool examination data entry
Matching result
Subject excluded for incomplete data
Data analysis
59
2.2.3 Questionnaires
Each participant was required to complete a standardized questionnaire to assess basic demographic
data and risk behaviours for acquiring the infection. The questionnaire had been validated and used
earlier in the same study area. Basic demographic data included sex, age and occupation. Risk
behaviours were determined by assessing uncooked-fish eating behaviour patterns (69, 70, 98).
Uncooked fish preparations are classified in two major groups as described below.
1) Instantly prepared uncooked fish
A well-known popular dish is Koi pla (chopped raw fish salad). Freshwater fish is chopped and mixed
with spicy ingredients. Raw fresh meat is usually denatured by lime juice due to acidity, which
dramatically changes its colour to a cooked-like texture. Koi pla is always consumed instantly.
2) Extensively fermented fish
Pla ra is a freshwater fish preserved in a highly concentrated salt solution for 3-6 months. Chunks of
fish could be consumed uncooked and remaining fish sauce is often used as daily ingredient for local
dishes (29, 87, 89).
2.2.4 Collection of stool specimen
After receiving a sealable container, participants collected stool samples with a scoop provided.
Then the container was sealed and transferred to a village health volunteer and delivered to the
field lab. Each participant provided 1 stool sample, processed by 3 examination methods. Two
methods were performed in the field: 1) Wet preparation and 2) Kato-Katz technique. Both
processes were conducted within six hours after collection. The specimen was examined under a
light microscope to identify intestinal parasite eggs, focusing on the presence of OV eggs by
experienced examiners. Then specimens were kept in a refrigerator and transported to the
Department of Parasitology, Phramongkutklao College of Medicine for processing for the third
examination method with formalin-ethyl acetate concentration technique and also examined under
light microscopy. At least one positive result from any method was interpreted as positive for the
infection (69, 70). Additionally, OV eggs were counted under Kato-Katz technique; the specimen
were examined throughout the slide and reported as total egg per slide.
As described in Section 1.2.1, the morphology of OV eggs is distinct from other intestinal helminths,
but could be physically similar to MIF under the microscope. Evidence indicated that no MIF was
presented in the study area (46), but no further confirmation was made from the molecular study in
the research project. All OV-like eggs were assumed to be Opisthorchis viverrini eggs.
60
2.2.5 Data analysis
Each participant was matched for questionnaire and specimen using a code-embedded identification
system. Stool examiners were blinded to the individual who provided the specimen. The data
management unit matched all corresponding data (questionnaire and stool examination results) for
data analysis. Incomplete questionnaires with missing data on relevant exposures, i.e., sex, age,
occupation and consumption behaviours, were excluded to perform risk factor analysis.
Descriptive statistics were used to describe data; outcome was reported as frequency and
percentage. Statistical analysis was reported with significance level of 0.05 and 95% confidence
intervals.
Outcomes of infection were considered in two ways: binary outcome as infected or not and egg
count as number of eggs per slide for infection intensity. Distribution of OV was reported as
prevalence (%) in the cross-sectional study and incidence (/100 person-years) for the follow-up
study. Egg intensity was reported as eggs per gram (EPG). The number of eggs per slide was
multiplied by 24 to obtain EPG. The distribution was explored by histogram to visualize the pattern
of intensity.
2.2.6 Model for prevalence and incidence data
Univariable analysis was initially performed to examine the association of OV of each exposure of
interest and providing crude association. Exposures included sex, age, occupation and uncooked fish
consumption behaviours. The association between univariable risk factors and the infection was
quantified using Pearson’s chi-square.
Multivariable analysis was performed using regression modelling for adjusting risk factors. Risk
factors were adjusted for age based on previous works (69, 70). Factors with p value <0.2 from
univariable analysis were also considered for inclusion in the multivariable model.
Instant fish preparation (Koi pla) was recognised as a risk factor from previous works in this project
(69, 70). However, extensively fermented fish (Pla ra) was considered to be included into the model
to control for confounder based on the data that they were also consumed on a regular basis.
For the cross-sectional study, outcomes were reported as categorical data (infected/uninfected). A
logistic regression model was used for model fitting. Risk factor associations were described using
odds ratios.
61
For the follow-up study, each infected case was counted as a singular event. All participants were
followed for the specified follow-up time (14-15 months). Poisson regression was developed to study
incidence outcomes with risk factors and reported as incidence rate ratio.
2.2.7 Model for infection intensity
Biologically, adult OV worms reside in the human host bile duct (2). Therefore, direct worm counts
for OV infection are not available for practical fieldwork (58, 86). Therefore, egg per gram
measurements of intensity are performed, where it is assumed, for logistical reasons that the
amount of eggs excreted in the stool is associated with the intensity of worm load presented in the
host bile duct (95).
Data were analyzed using a count model approach. Studies across many species of helminthes have
addressed the problem of over-dispersion data (93-95, 99-101),which results in large amounts of
zero count eggs for which classical Poisson distribution may not be appropriate. Therefore, extensive
models for count data were used here to explore the characteristics of over-dispersion and evaluate
the association of risk factors and outcomes, and also the age-prevalence-intensity relationship.
The regression models for count data and related equations were based on model framework by
Zeileis et al. (102) published for count data analysis in R.
The count data model was based on the concept of the generalized linear model (GLM) framework
(102, 103) which explains the dependence of a scalar variable yi (when i = 1, 2,…, n) on a vector of
regressors xi with the probability density function as shown in Equation 2.1 below.
𝑓(𝑦; 𝜆, 𝜙) = exp (𝑦∙𝜆−𝑏(𝜆)
𝜙) + 𝑐(𝑦, 𝜙), where (2.1)
The conditional distribution of yi|xi follows a linear exponential family.
λ is a canonical parameter.
𝜙 is a dispersion parameter.
The function b() and c() refer to a particular family which will be used; normal, binomial
or Poisson distribution.
Mean was defined as 𝐸[𝑦𝑖|𝑥𝑖] = 𝑏′(𝜆𝑖) or μi and variance could be defined as 𝑉𝐴𝑅[𝑦𝑖|𝑥𝑖] = 𝜙 ⋅
𝑏′′(𝜆𝑖); therefore, the distribution of yi is determined by μi.
The dependence of the conditional mean 𝐸[𝑦𝑖|𝑥𝑖] = 𝜇𝑖 on regressors xi was given by 𝑔(𝜇𝑖) = 𝑥𝑖⊺𝛽
where g() was a link function and 𝛽 was a vector of regression coefficient.
62
The simple model for count data within the GLM framework was the Poisson distribution with
probability density function as described in Equation 2.2 below.
𝑓(𝑦; 𝜇) = exp(−𝜇)∙𝜇𝑦
𝑦! (2.2)
Poisson distribution refers to the number of eggs excreted in the stool specimen during the
fieldwork period. As the mean egg count was low, most data consisted of low egg counts and few
data held highly skewed values. When the mean egg count increased, the distribution approached a
normal distribution pattern which meant equal to variance; V(μ) = μ, so the dispersion (φ) was fixed
at 1. The canonical link between mean and linear predictor was defined as g(μ) = log(μ) resulting in a
log-linear relationship.
In particular, the data included a larger amount of extreme values resulting in variance spreading
greater than mean, called over-dispersion and is explained below.
1. Some individuals carried extremely high egg count. Many helminth infections exhibit a
particular pattern when only few cases carry a heavy burden of infection while the
remaining carry a few or zero egg counts, resulting in a visibly longer tail in distribution.
2. A large number of zero egg counts resulted in a heavily skewed distribution.
One way to deal with this problem was to assume a negative binomial distribution as shown in
Equation 2.3 below.
𝑓(𝑦; 𝜇, 𝜃) = Γ(𝑦+ 𝜃)
Γ(𝜃)⋅𝑦! ⋅
𝜇𝑦⋅𝜃𝜃
(𝜇+𝜃)𝑦+𝜃 , where (2.3)
𝜃 is a shape parameter.
Γ() is a gamma function.
The 𝜙 is 1 but the variance could refer to 𝑉(𝜇) = 𝜇 + 𝜇2
𝜃
A negative binomial distribution has been proposed for a better choice for over-dispersion data
regarding some parasite intensity data (94, 95) . Thus, the preliminary analysis will deal with the
property of egg count outcomes from fieldwork data with classical Poisson and negative binomial
distributions.
63
Compound model
Zero egg count raised an interesting issue as the zero result tended to be a major proportion of stool
outcomes. Zero result could be derived from true or false origin whether the examination tools
could truly capture the zero count with high sensitivity, or the light infection could be missed due to
false negatives (95).
The compound model has been addressed to treat the issue of zero egg excretor from various
assumptions (95).
Figure 2.4 Distribution of count data in hurdle
model
As illustrated in Figures 2.3 and 2.4, the zero component was based on two assumptions: whether
the zero components were sampling or structural. Structural zero is derived from true negative cases
referring to uninfected individuals. The difference of the zero-inflated and a hurdle model depends
on how the zero will be sampled. Basically, the hurdle model assumes that all zero egg outputs are
true negative; therefore, a non-zero output should belong to infected cases with sampling
distribution (Figure 2.4). However, the zero-inflated model assumes that a zero egg count could
come from either true negative or false negative due to diagnostic sensitivity. A true negative result
will refer to uninfected individuals. For infected cases, when the test fails to detect fluke eggs in
stool, then the outcome will be negative as a result of a false negative. Therefore, the sampling of
egg counts will include zero from the assumption.
Zero-inflated model
As shown in Figure 2.3, the zero-inflated model assumes that zero outcomes stem from different
origins. The projection is a joint model consisting of two parts: one for the predicting zero and the
other for the count model.
0
20
40
60
80
100
120
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Freq
uen
cy
Number of egg count
Zero-inflated model
Structural
Sampling
0
20
40
60
80
100
120
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Freq
uen
cy
Number of egg count
Hurdle model
Structural
Sampling
Figure 2.3 Distribution of count data in zero-
inflated model
64
Some infected cases produced zero egg counts, i.e., zero excretor. This portion of zero excretor will
be a sampling zero since it is assumed to happen by chance.
As a result, the zero-inflated model will comprise two components when one part produces only
zero, a point mass at zero 𝐼{0}(𝑦), and other part generates both zero and non-zero outcomes, a
count component. A binary model is used to model the unobserved probability with 𝜋 = 𝑔−1(𝑧⊺𝛾).
Therefore, the density of the inflated-model will be incorporated with point mass at zero and a
count distribution, as shown below in Equation 2.4.
𝑓(𝑦; 𝑥, 𝑧, 𝛽, 𝛾) = 𝑓𝑧𝑒𝑟𝑜(0; 𝑧, 𝛾) ⋅ 𝐼{0}(𝑦) + (1 − 𝑓𝑧𝑒𝑟𝑜(0; 𝑧, 𝛾)) ⋅ 𝑓𝑐𝑜𝑢𝑛𝑡(𝑦; 𝑥, 𝛽), (2.4)
Where the probability of observing zero count is inflated with probability 𝜋 = 𝑓𝑧𝑒𝑟𝑜(0; 𝑧, 𝛾) and a
count model assumed to follow Poisson or negative binomial distributions depends on the property
of count outcome, as referred to dispersion parameter. Regarding the mean, the regression function
will be;
𝜇𝑖 = 𝜋𝑖 ⋅ 0 + (1 − 𝜋𝑖) ⋅ exp(𝑥𝑖⊺𝛽),
Where the default link function; 𝑔(𝜋), based on the binomial, is logit link.
Hurdle model
The hurdle model is a compound model to deal with excess zero data with a different assumption
from the zero-inflated model. The model comprises two components:
The first hurdle will determine the zero part of the result. The model assumes that all zero outcomes
are structural. Therefore, zero egg excretor are truly uninfected cases. Thus, this component will
assume the probability whether the individual will be infected or not following binary outcome (0/1)
under the logit regression model, defined as zero hurdle model 𝑓𝑧𝑒𝑟𝑜(𝑦; 𝑧, 𝛾), which generally
follows binomial distribution.
The second hurdle will deal with count outcome as infection intensity. From the first hurdle, I
assume that zero egg excretor is uninfected so the nonzero egg excretor will be truly infected.
Therefore, the infected cases must produce a nonzero result since they cannot have zero egg count
by assumption. The count outcome from this component will follow the truncated count model
either Poisson or negative binomial distribution,; 𝑓𝑐𝑜𝑢𝑛𝑡(𝑦; 𝑥, 𝛽) in Equation 2.5.
𝑓ℎ𝑢𝑟𝑑𝑙𝑒(𝑦; 𝑥,𝑧,𝛽,𝛾) = {𝑓𝑧𝑒𝑟𝑜(0; 𝑧, 𝛾) 𝑖𝑓 𝑦 = 0
(1 − 𝑓𝑧𝑒𝑟𝑜(0; 𝑧, 𝛾)) ⋅ 𝑓𝑐𝑜𝑢𝑛𝑡(𝑦;𝑥,𝛽)
1− 𝑓𝑐𝑜𝑢𝑛𝑡(0;𝑥,𝛽) 𝑖𝑓 𝑦 > 0
(2.5)
65
The likelihood for zero hurdle and count model estimation can be maximized separately. The mean
and regressors can be described as shown below in Equation 2.6.
log(𝜇) = 𝑥𝑖⊺𝛽 + log (1 − 𝑓𝑧𝑒𝑟𝑜(0; 𝑧, 𝛾))) − log (1 − 𝑓𝑐𝑜𝑢𝑛𝑡(0; 𝑥, 𝛽)). (2.6)
Based on the assumption regarding the zero component, the zero-inflated model will predict the
probability for obtaining sampling zero when the hurdle model will predict the probability for
obtaining nonzero outcome.
The results from model fitting are interpreted in light of the characteristics of the infection and
method of stool examination.
66
2.2.8 Model analysis and fitting
Statistical modeling was performed using STATA/IC 14 and R 3.1.2. The zero-inflated and hurdle
models were fitted using the functions countfit, hplogit, hnblogit in STATA and PSCL package in R
(Political Science Computational Laboratory).
For prevalence and incidence data, covariates comprised five variables: sex, age, and occupation,
consumption of Koi pla and consumption of Pla ra. All model building began with a full model and
compared with a reduced model. The method of covariate selection was based on the following.
All covariates were preliminarily explored with univariable analysis to obtain statistical
value.
Full models were performed with corresponding analysis with five covariates.
Reduced model excluded non-significant covariates from full models.
The full model was compared with the reduced model by log-likelihood and AIC to obtain
the best fitted model.
For intensity data, the compound models incorporated covariates in both components and
covariates were selected from the best fitted model.
Over-dispersion was explored and tested. To test over-dispersion, the likelihood ratio test between
Poisson and negative binomial model was performed. The over-dispersion parameter (α) was used.
When α was zero, the negative binomial distribution was equivalent to standard Poisson
distribution.
The Poisson model was assigned as a baseline model. The ∆AIC; difference of AIC to baseline model
≥10 indicated that the smallest AIC model was best fitted. The observed zero counts were compared
to expected zero count to determine the performance of handling excess zero outcomes.
2.2.9 Ethical consideration
The research protocol of this study was approved by the Ethics Committee of the Royal Thai Army
Medical Department (approval code S045h/54). Study participants or parents of participants under
18 years of age agreed to join the study after reading the data sheet provided in Thai and provided
their written informed consent. Appropriate treatment was provided for those who were found to
be positive for any intestinal parasitic infection.
67
2.3 Results
2.3.1 Population characteristics
2.3.1.1 Response from the study
As illustrated in Figure 2.5, Consenting participants returned the survey package including stool
specimens and questionnaires for 86.1% and 66.1%, respectively. Test on the equality of proportions
between two study areas were reported with significant p-value.
Figure 2.5 Summarized flowchart for cross-sectional study
Baseline population
Na-isarn and Na-ngam village
2012 - 2013
1514 Participants
1514 eligible subjects Informed for the consent
Dispatched of survey package Consent form document Stool specimen container Questionnaire Did not return survey
package
Did not consented
Unable to contact
Processing of survey package
Stool examination
Questionnaire data entry
Stool examination data entry
Matching for individual record
Completed data records for analysis 794 records
Collection of survey package
Stool container (1303)
Questionnaire (1000)
Excluded from the analysis Unmatched or incomplete questionnaires (509)
A
C
D
B
68
Table 2.2 Response for cross-sectional study
Na-isarn 2013 Na-gnam 2012 p Total
N % N % N %
A. Survey package dispatched 583 931 1514
B. Consented and returned
survey package
Returned stool specimen
Returned questionnaire
431
354
73.9
60.7
872
646
93.7
69.4
<0.001
<0.001
1303
1000
86.1
66.1
C. Matching and excluded for
incomplete questionnaire 100 17.1 409 44.0 <0.001 509 33.7
D. Completed questionnaire
and specimen 331 56.8 463 49.7 0.01 794 52.4
As illustrated in Table 2.2, the response for returned stool specimens was significantly higher in Na-
ngam Area (93.7% and 73.9 %, respectively, p <0.001). After pairing stool examination results with
matched questionnaires and excluding incomplete questionnaires, eligible records from Na-ngam
were more numerous than Na-isarn Area (44.0 % and 17.1 %, respectively, p <0.001).
For the whole study area, completed questionnaires matched with specimens eligible for data
analysis totalled 52.4 %. Only 13.9% of participants did not return the packages during the survey
period and were unable to contact. Some of them could not provide the stool specimen at the time
of collection. Returned questionnaire rate was also lower than stool specimens from both areas.
69
Figure 2.6 Summarized flowchart for follow-up study
Eligible participants comprised negative cases from the baseline survey conducted from 2013-14
with a total of 1803 subjects: 974 cases from Na-yao Area and 829 cases from Na-ngam Area.
As illustrated in Figure 2.6, the enrolment process was conducted at the time of fieldwork. Cases
were enrolled retrospectively; therefore, deceased subjects or those with confirmed permanently
moved-out status were excluded from the study. Test results on the equality of proportions between
the two study areas were reported with significant p-value.
Baseline cross-sectional study
Na-yao and Na-ngam village
2011 - 2012
1910 Participants
Negative for OV infection 1803 subjects
1803 eligible subjects for a follow-up study
Invited to participate in retrospective cohort study in 2013-2014 14-15 months follow-up period Informed consent
Excluded from the study
Did not consented (8)
Deceased (15)
Permanent moved-out (82)
Collection of survey package 884 packages
Processing of survey package
Stool examination
Questionnaire data entry
Stool examination data entry
Matching for individual record
Completed data records for analysis 746 records
Positive for OV infection 107 subjects
Enrolled to the study 1698 subjects received survey package
Stool container
Questionnaire
Excluded from the analysis Unmatched or incomplete questionnaires (138)
Loss to follow-up
Temporary moved-out (361)
Unable to contact/return specimen (453)
A
C
E D
B
G
F
70
Table 2.3 Response for follow-up study
Na-yao 2013 Na-gnam 2014 Total
N % N % p N %
A. Eligible participants
Consented
974
916
94.0
829
782
94.3
0.80
1803
1698
94.2
B. Excluded participants
Did not consent
Died during follow-up period
Permanently moved-out
5
3
50
0.5
0.3
5.1
3
12
32
0.4
1.4
3.9
0.63
0.01
0.20
8
15
82
0.4
0.8
4.5
C. Enrolled participants 916 782 1698
D. Loss to follow-up participants
Temporarily moved-out during
specimen collection
Unable to contact/return specimen
187
352
20.4
38.4
174
101
22.3
12.9
0.36
<0.001
361
453
21.2
26.7
E. Survey package collected 377 41.2 507 64.3 <0.001 884 52.1
F. Excluded for unmatched or incomplete
questionnaire 46 5.1 92 11.2 0.02 138 9.9
G. Completed questionnaire and
specimen 331 36.1 415 53.1 <0.001 746 42.2
As shown Table 2.3, 94.2% of eligible cases consented to join the follow-up study. No significance
was observed concerning consented participants for both areas (p = 0.80). The percentage of those
who did not consent or confirmed as permanently moved-out was insignificant. Subjects who died
during the follow-up period were more numerous in Na-gnam Area (p = 0.01). After the enrolment
process, 916 participants for Na-yao and 782 participants for Na-ngam joined the follow-up study.
The follow-up time for Na-yao area was 15 months and Na-ngam area, 14 months.
The survey package was dispatched to participants. Completed packages were returned with stool
specimens and completed questionnaires to perform data analysis. The overall response for
returned survey packages was 52.1 % and loss to follow-up participants comprised 47.9 %. The
difference in response rate was significant (41.2 % and 64.3 %, respectively, p <0.001). The
difference was mainly derived from the subjects whom I were unable to contact or did not return
specimens. Overall excluded incomplete questionnaires totalled 9.9%, resulting in a lower response
rate of 42.2%. The final response rates in Na-yao and Na-gnam were 36.1% and 53.1%, respectively.
71
2.3.1.2 Demographic data
Figure 2.7 Population pyramids
As shown in Figure 2.7, the population pyramids showed an irregular pattern of age structure where
the near-bottom class was relatively limited in size in all four study areas.
The pattern indicated that the majority of subjects were aged between 30 - 59 years. Considering
the working age, which usually starts at early adolescence for Thai rural population, the age period
20 - 39 was relatively small in proportion compared with age ≥40 years.
For the Na-isarn 2013 cross-sectional study, no females were included in the 0 - 9 years age class.
More females were in the middle to old age in all areas.
However, the age structure of the follow-up population needed to be compared with the baseline
population because the individuals identified as defined at baseline were excluded.
Na-isarn 2013 cross-sectional study
Na-yao 2013 follow-up study Na-ngam 2014 follow-up study
Na-ngam 2012 cross-sectional study
72
Table 2.4 Population characteristics of cross-sectional study
Characteristic Na-isarn
N (%)
Na-ngam
N (%)
Total
N (%) p
Sex
Female
Male
179 (54.1)
152 (45.9)
255 (55.1)
208 (44.9)
434 (55.1)
360 (45.3)
0.78
Age profile
Mean ± S.D.
Median ± IQR
Min – max
44.6 ± 15.9
44.0 ± 20.0
7 – 84
44.7 ± 18.6
46.0 ± 23.0
2 - 86
44.7 ± 15.9
45.0 ± 21.0
2 - 86
0.46
Age group (in years, as quartiles)
0 – 35
36 – 46
47 – 58
≥ 59
117 (25.3)
119 (25.7)
116 (25.1)
111 (24.0)
0 – 35
36 – 45
46 – 57
≥ 58
200 (26.2)
186 (24.4)
202 (26.5)
175 (22.9)
Q1
Q2
Q3
Q4
0 – 36
37 – 44
45 – 56
≥ 57
88 (26.6)
78 (23.6)
88 (26.6)
77 (23.3)
Occupation
Unemployed
Non-agriculture
Agriculture
30 (9.1)
81 (24.5)
220 (66.5)
45 (9.7)
147 (31.7)
271 (58.5)
75 (9.4)
228 (28.7)
491 (61.8)
0.06
Table 2.5 Population characteristics of follow-up study
Characteristic Na-yao
N (%)
Na-ngam
N (%)
Total
N (%) p
Sex
Female
Male
176 (53.2)
155 (46.8)
228 (54.9)
187 (45.1)
404 (54.2)
342 (45.8)
0.63
Age profile
Mean ± S.D.
Median ± IQR
Min – max
47.2 ± 19.1
50.0 ± 22
1 – 85
43.3 ± 19.9
47.0 ± 23
2 – 87
45.0 ± 19.6
48.0 ± 23
1 - 87
0.88
Age group (in years, as quartiles)
0 – 34
35 – 47
48 – 57
≥ 57
104 (25.1)
117 (28.2)
91 (21.9)
103 (24.8)
0 – 36
37 – 48
49 – 59
≥ 60
189 (25.3)
190 (25.5)
191 (25.6)
176 (23.6)
Q1
Q2
Q3
Q4
0 – 38
39 – 50
51 – 60
≥ 60
83 (25.1)
86 (26.0)
87 (23.6)
75 (22.7)
Occupation
Unemployed
Non-agriculture
Agriculture
17 (5.1)
100 (30.2)
214 (64.7)
36 (8.7)
146 (35.2)
233 (56.1)
53 (7.1)
246 (33.0)
447 (59.9)
0.33
73
The majority of population worked in the agricultural-related sector. As shown in Table 2.4, the
overall mean age in the cross-sectional study was 44.7 ± 15.9 years with a range of 2 - 86 years and
median age of 45.0 ± 21.0 years. As shown in Table 2.5, the follow-up study showed the overall
mean age was 45.0 ± 19.6 years with a range of 1 - 87 years and median age was 48.0 ± 23.0 years.
Due to the irregularity of age structure, age was categorized in four age groups by quartile intervals.
2.3.2 Uncooked fish consumption behaviors
From the uncooked fish consumption behaviors from the 2 prevalence and 2 follow-up studies; all
four areas showed similar patterns of uncooked fish consumption behaviors. The range of
consumption was 39.8 - 44.1 %. (Supplement Table S-1 and S-2)
However, Pla ra consumption was more popular: 82.1% from the cross-sectional study and 77.5%
from the follow-up study. Moreover, consumption of Koi pla did not significantly differ for both
areas within the same study method. The proportion of Pla ra consumption was significantly higher
in Na-isarn in the cross-sectional study and in Na-ngam in the follow-up study.
74
Table 2.6 Multivariable analysis of population characteristics and uncooked fish consumption
behaviours in cross-sectional study
Characteristic
Na-isarn 2013 Na-ngam 2012
Chopped raw fish salad
(Koi pla)
Extensively
fermented fish
(Pla ra)
Chopped raw fish
salad
(Koi pla)
Extensively
fermented fish
(Pla ra)
Adjusted
OR
(95% CI)
P
Adjusted
OR
(95% CI)
P
Adjusted
OR
(95% CI)
P Adjusted OR
(95% CI) P
Sex
Female
Male
1
3.46
(2.13-5.63)
<0.01
1
1.91
(0.95-3.85)
0.07
1
4.20
(2.81-6.27)
<0.01
1
1.30
(0.81-2.08)
0.28
Age group
(as quartiles)
Q1
Q2
Q3
Q4
1
2.18
(1.05-4.54)
3.18
(1.53-6.61)
2.64
(1.23-5.68)
0.04
<0.01
0.01
1
0.91
(0.36-2.28)
3.40
(0.89-12.98)
0.47
(0.19-1.14)
0.84
0.07
0.10
1
2.18
(1.21-3.93)
1.86
(1.02-3.81)
2.54
(1.36-4.73)
0.01
0.04
<0.01
1
2.47
(1.26-4.86)
2.45
(1.22-4.91)
1.12
(0.59-2.12)
0.01
0.01
0.74
Occupation
Unemployed
Non -
agriculture
Agriculture
1
1.33
(0.45-3.93)
2.13
(0.83-5.49)
0.60
0.12
1
1.27
(0.43-3.76)
2.72
(1.03-7.17)
0.67
0.04
1
1.02
(0.47-2.23)
1.26
(0.61-2.59)
0.96
0.53
1
1.00
(0.45-2.20)
2.13
(1.01-4.50)
0.99
0.04
Table 2.6 shows the multivariable analysis of population characteristics and uncooked fish
consumption behaviours. Sex, age group and occupation were considered potential demographic
factors based on previous studies contributing risk behaviours (69, 70). Age-group is referred to in
quartiles as shown in Table 2.4. Both study areas show similar Koi pla consumption patterns, male
sex and age-group were statistically significant factors. For age-group, the adjusted OR was highest
in Q3 for Na-isarn and Q4 for Na-ngam area indicating that the OR could be higher in older age. No
statistical significance for occupation and Koi pla consumption was found from both study areas.
For Pla ra, age was only significant for Q2 and Q3 in Na-ngam area. Agricultural-related work
significantly increased risk of eating Pla ra in both areas compared with unemployment.
75
Table 2.7 Multivariable analysis of population characteristics and uncooked fish consumption
behaviours in follow-up study
Characteristic
Na-yao 2013 Na-ngam 2014
Chopped raw fish salad
(Koi pla)
Extensively
fermented fish
(Pla ra)
Chopped raw fish
salad
(Koi pla)
Extensively
fermented fish
(Pla ra)
Adjusted RR
(95% CI) P
Adjusted RR
(95% CI) P
Adjusted RR
(95% CI) P
Adjusted RR
(95% CI) P
Sex
Female
Male
1
1.20
(0.86-1.67)
0.27
1
0.93
(0.71-1.21)
0.57
1
1.56
(1.14-2.13)
<0.01
1
0.97
(0.79-1.20)
0.81
Age group
(as quartiles)
Q1
Q2
Q3
Q4
1
2.25
(1.22-4.17)
2.53
(1.38-4.64)
2.94
(1.58-5.45)
0.01
<0.01
<0.01
1
1.24
(0.83-1.86)
1.17
(0.77-1.76)
1.11
(0.72-1.71)
0.29
0.46
0.64
1
1.90
(1.11-3.26)
1.84
(1.04-3.26)
1.64
(0.92-2.93)
0.02
0.04
0.09
1
1.15
(0.84-1.59)
1.08
(0.76-1.53)
1.17
(0.84-1.65)
0.39
0.68
0.36
Occupation
Unemployed
Non-
agriculture
Agriculture
1
1.32
(0.53-3.30)
1.52
(0.65-3.55)
0.55
0.34
1
1.07
(0.52-2.18)
1.21
(0.62-2.36)
0.86
0.57
1
1.37
(0.59-3.21)
2.25
(1.02-4.94)
0.47
0.04
1
1.00
(0.95-1.54)
1.10
(0.73-0.64)
0.99
0.66
As illustrated in Table 2.7, age-groups showed that increasing age was associated with Koi pla
consumption in both study areas when compared with referenced age-group except the Q4 age-
group in Na-ngam study. Male gender and agricultural-related work were also statistically significant
in Na-ngam area for Koi pla consumption. Age-group was referred to in quartiles as shown in Table
2.5. No significant association was observed for population characteristics and Pla ra consumption.
76
2.3.3 Distribution and risk factors of OV infection
2.3.3.1 Distribution of prevalence
Figure 2.8 Prevalence of OV infection distributed by age-group
Prevalence of OV in Na-isarn was 7.85 % (95% CI: 4.96 - 10.75) and Na-ngam is 9.29 % (95% CI: 6.64 -
11.93), respectively. The overall prevalence was 8.69 % (95% CI: 6.82 - 10.87) and both prevalence
did not vary significantly between the two study areas. The prevalence increased by age where the
prevalence was higher in older age groups as shown in Figure 2.8. (Supplement Table S-3)
From the study, 1303 survey packages were returned for which stool samples were processed.
However, only 794 specimens were matched to completed questionnaires to perform data analysis.
Table 2.11 compared the prevalence of OV between two groups indicating that no significant
difference for the completed and incomplete questionnaire groups for both areas. However, the
prevalence was statistically higher among those completing questionnaires in the Na-isarn area.
(Supplement Table S-4)
0
5
10
15
20
25
1 2 3 4
Pre
vale
nce
(%
wir
h 9
5% C
I)
Age group
Na-isarn 2013 cross-sectional study
0
5
10
15
20
25
1 2 3 4
Pre
vale
nce
(%
wir
h 9
5% C
I)
Age group
Na-ngam 2014 cross-sectional study
0-36 37-44 45-56 ≥57
Age-group
0-34 37-48 49-59 ≥60
Age-group
77
2.3.3.2 Distribution of incidence
The incidence of OV infection was 7.98/100 person-years (95% CI: 5.49 – 11.20) in the Na-yao study
and 6.80/100 person-years (95% CI: 4.68 – 9.54) in the Na-ngam study. The incidences from both
areas did not significantly differ. (Supplement Table S-5)
Figure 2.9 shows the distribution of incidence by age-group. For Na-yao area, incidence increased in
the younger age and remained relatively stable in the older age groups. Na-ngam study showed that
the incidence was highest in the 35 - 47 age group.
The incidence rate of OV infection between the two groups indicating no significance for the
completed and incomplete questionnaires groups. (Supplement Table S-6)
0
5
10
15
20
1 2 3 4
Inci
den
ce (
/100
per
son
-yea
r w
irh
95%
CI)
Age group
Na-yao 2013 follow-up study
0
5
10
15
20
1 2 3 4
Inci
den
ce (
/100
per
son
-yea
r w
irh
95%
CI)
Age group
Na-ngam2014 follow-up study
Figure 2.9 Incidence of OV infection distributed by age-group
0-38 39-50 51-60 ≥61
Age-group
0-36 37-48 49-59 ≥60
Age-group
78
Table 2.8 Proportional distribution of OV infection diagnosis categorised by stool examination
methods
Test Cross-sectional study Follow-up study Total
N % N % N %
N of cases 69 66 135
1 positive test
Simple smear 0 0.0 0 0.0 0 0.0
Kato-Katz 0 0.0 0 0.0 0 0.0
FECT 47 68.1 43 65.2 90 66.7
2 positive tests
Simple smear
+ Kato-Katz 0 0.0 0 0.0 0 0.0
Simple smear
+ FECT 1 1.4 2 3.0 3 2.2
Kato-Katz
+ FECT 17 24.6 16 24.2 33 24.4
3 positive tests
Simple smear
+ Kato-Katz
+ FECT
4 5.8 5 7.6 9 6.7
Table 2.8 reports positive stool examination results diagnosed by each method. From both study
designs conducted in 4 areas; Na-yao, Na-isarn and Na-ngam from 2011 - 2014, 135 cases were
diagnosed with OV; 69 cases from the cross-sectional study and 66 cases from the follow-up study.
In all, 66.7% of diagnoses were from FECT only while the other 33.3% of cases were positive with a
combination of two of three tests.
FECT was responsible for all positive cases either single or combination results. Regarding the
combination of results, simple smear test results were positive in the diagnosis of 12 cases (8.9%)
and Kato-Katz for 42 cases (31.1%). A total of 9 (6.7%) cases all tested positive. No case was
diagnosed with simple smear or Kato-Katz alone as well as simple smear and Kato-Katz method.
79
2.3.4 Evaluation of risk factors for acquiring OV infection
2.3.4.1 Univariable analysis of risk factors for acquiring OV infection
Univariable analysis assesses the association between each factor to OV infection. The analysis
indicates that sex, occupation and Pla ra consumption were not significantly associated with OV in
all 4 areas. (Supplement Table S-7 and S-8)
From the cross-sectional study, the oldest age group, (Q4) was statistically significant for having the
infection compared with the Q1 age group in both areas. However, increasing age was associated
with increasing risk in Na-isarn area. Koi pla was significantly associated with the infection in both
areas. (Supplement Table S-7)
The follow-up study showed a different pattern of risk factors from the cross-sectional study. The 35
- 47 age group was only observed for significant factors for OV in the Na-ngam 2014 study. Koi pla
consumption was statistically significant in the Na-yao 2013 study. (Supplement Table S-8)
80
2.3.4.2 Model fitting
Table 2.9 Model fitting for prevalence and follow-up study
Cross-sectional study
Na-isarn 2013 Na-ngam 2012
Covariates 5 4 2 5 4 2
Log L -79.89 -80.03 -80.31 -133.87 -135.58 -136.38
AIC 177.78 174.06 170.61 285.75 285.16 282.76
Follow-up study
Na-yao 2013 Na-ngam 2014
Covariates 5 4 2 5 4 2
Log L -105.07 -105.13 -105.46 -112.00 -112.07 -112.38
AIC 228.14 224.25 220.92 242.01 238.15 234.77
Table 2.9 shows the model fitting with regard to covariate selection, age group and consumption of
Koi pla was significant from univariable models. Therefore, they were selected in the final model. Sex
and occupation revealed a potential risk obtained from the qualitative approach detailed in the
Chapter 5.
Moreover, Pla ra is a very popular uncooked fish dish for which the consumption is approximately
80%. The qualitative approach also revealed that Pla ra was a main ingredient for local dishes.
Additional analysis revealed that Pla ra held a significantly positive correlation with Koi pla in all
study areas (correlation coefficient = 0.5 - 0.6, p <0.01).
The full multivariable model comprised five covariates as a baseline model; sex, age, occupation, Koi
pla and Pla ra consumption.
The four-covariate model excluded the occupation covariate from the model. Sex remained in the
model since previous studies showed the potential risk (70) besides the qualitative result and Pla ra
was still included in the model as the role of main uncooked fish selection.
The final model selected only those two statistically significant covariates obtained from current
fieldwork study: age and Koi pla consumption.
The results indicated no significant difference for log-likelihood ratio between models while the AIC
was smallest in the two-covariate model suggesting that besides age and Koi pla consumption, the
other three covariates may not contribute much influence in the model.
81
2.3.4.3 Multivariable analysis of risk factors for acquiring OV infection
Table 2.10 Multivariable analysis of risk factors for acquiring OV infection of prevalence data from
Na-isarn 2013 study
Characteristic
5 covariates 4 covariates 2 covariates
Adjusted OR
(95% CI) p
Adjusted OR
(95% CI) p
Adjusted OR
(95% CI) p
Sex (female as reference)
Male
0.74 (0.31 – 0.18)
0.51
0.75 (0.31 – 1.80)
0.52
Age group (as quartiles, age 0 – 36 as reference)
1.45
(0.23 – 9.04)
4.21
(0.87 – 20.35)
5.58
(1.16 – 26.95)
0.69
0.07
0.032
1.44
(0.23 – 8.97)
4.31
(0.89 – 20.80)
5.43
(1.13 – 26.13)
0.70
0.07
0.034
37 – 44
45 – 56
≥ 57
1.37
(0.21 – 8.69)
3.86
(7.60 – 19.63)
5.33
(1.03 – 27.50)
0.74
0.10
0.045
Occupation (unemployed as reference)
Non-agriculture
Agriculture
1.09 (0.15 – 7.89)
1.39 (0.28 – 6.94)
0.94
0.67
Fish menus
Koi pla
Yes
2.54 (1.01 – 6.39)
0.048
2.57 (1.02 – 6.47)
0.045
2.43 (1.01 – 5.80)
0.046
Pla ra
Yes
1.31 (0.27 – 6.36)
0.74
1.36 (0.28 – 6.53)
0.70
82
Table 2.11 Multivariable analysis of risk factors for acquiring OV infection of prevalence data from
Na-ngam 2012 study
Characteristic
5 covariates 4 covariates 2 covariates
Adjusted OR
(95% CI) p
Adjusted OR
(95% CI) p
Adjusted OR
(95% CI) p
Sex (female as reference)
Male
1.31 (0.65 – 2.61)
0.45
1.38 (0.70 – 2.73)
0.36
Age group (as quartiles, age 0 - 35 as reference)
1.53
(0.53 – 4.45)
1.76
(0.62 – 5.04)
2.66
(0.98 – 7.19)
0.44
0.29
0.054
1.45
(0.50 – 4.18)
1.68
(0.59 – 4.76)
2.61
(0.97 – 7.04)
0.49
0.33
0.057
36 - 46
47 - 58
≥ 59
1.24
(0.42 – 3.69)
1.39
(0.47 – 4.08)
2.10
(0.74 – 6.00)
0.70
0.55
0.17
Occupation (unemployed as reference)
Non-agriculture
Agriculture
0.73 (0.18 – 2.85)
1.57 (1.16 – 5.15)
0.65
0.45
Fish menus
Koi pla
Yes
2.44 (1.16 – 5.15)
0.019
2.43 (1.16 – 5.12)
0.019
2.48 (1.28 – 4.82)
0.007
Pla ra
Yes
0.63 (0.27 – 1.49)
0.29
0.69 (0.29 – 1.60)
0.39
Tables 2.10 and 2.11 show the multivariable analysis for prevalence data. From the Na-isarn study in
Table 2.10, age ≥57 years old and consumption of Koi pla remained significant risk factors in the
three models indicating their consistency when compared with the reference age-group.
For the Na-ngam study presented in Table 2.11, the consistency of Koi pla association was still
observed in all three models. When adjusted for other covariates, age was not statistically
significant.
83
Table 2.12 Multivariable analysis of risk factors for acquiring OV infection of follow-up data from Na-
yao 2012 study
Characteristic
5 covariates 4 covariates 2 covariates
Adjusted IRR
(95% CI) p
Adjusted IRR
(95% CI) p
Adjusted IRR
(95% CI) p
Sex (female as reference)
Male
1.23 (0.61 -2.48)
0.56
1.21 (0.61 – 2.42)
0.58
Age group (as quartiles, age 0 – 38 as reference)
1.38
(0.45 – 4.27)
1.40
(0.46 – 4.27)
1.40
(0.45 – 4.39)
0.58
0.55
0.56
1.35
(0.40 – 4.12)
1.40
(0.46 – 4.25)
1.40
(0.45 – 4.37)
0.60
0.55
0.56
39 – 50
51 – 60
≥ 61
11.36
(0.41 – 4.47)
1.37
(0.42 – 4.47)
1.33
(0.38 – 4.65)
0.61
0.60
0.66
Occupation (unemployed as reference)
Non-agriculture
Agriculture
0.75 (0.14 – 3.98)
0.77 (0.17 – 3.51)
0.73
0.74
Fish menus
Koi pla
Yes
2.37 (1.10 – 5.12)
0.028
2.35 (1.09 – 5.06)
0.029
2.34 (1.10 – 4.99)
0.027
Pla ra
Yes
0.81 (0.39 – 1.70)
0.58
0.81 (0.39 – 1.69)
0.57
84
Table 2.13 Multivariable analysis of risk factors for acquiring OV infection of follow-up data from Na-
ngam 2014 study
Characteristic
5 covariates 4 covariates 2 covariates
Adjusted IRR
(95% CI) p
Adjusted IRR
(95% CI) p
Adjusted IRR
(95% CI) p
Sex (female as reference)
Male
1.23 (0.61 – 2.47)
0.56
1.22 (0.61 – 2.45)
0.58
Age group (as quartiles age 0 – 34 as reference)
1.38
(0.45 – 4.27)
1.40
(0.46 – 4.27)
1.40
(0.45 – 4.39)
0.58
0.55
0.56
3.14
(1.02 – 9.65)
1.29
(0.34 – 4.91)
1.83
(0.54 – 6.16)
0.046
0.71
0.33
35 – 47
48 – 57
≥ 58
3.08
(0.95 - 10.03)
1.26
(0.31 – 5.16)
1.70
(0.46 – 6.32)
0.06
0.75
0.43
Occupation (unemployed as reference)
Non-agriculture
Agriculture
0.76 (0.18 – 3.11)
0.80 (0.22 – 2.93)
0.70
0.70
Fish menus
Koi pla
Yes
1.31 (0.63 – 2.72)
0.47
1.30 (0.63 – 2.66)
0.48
1.39 (0.69 – 2.80)
0.35
Pla ra
Yes
1.40 (0.42 – 4.67)
0.59
1.40 (0.42 – 4.67)
0.59
Table 2.12 shows the Na-yao follow-up data, and Koi-pla consumption was significantly associated
with OV. Age-group was not associated with the infection in any models.
The Na-ngam follow-up data from Table 2.13 revealed different patterns from other areas.
Consumption of Koi pla and Pla ra were not statistically significant. The five-covariate model showed
no association with OV. When reduced to 4 and 2 covariates, the age group of 35 -47 was
significantly associated with the infection when compared the reference age-group.
85
2.3.5 Study of infection intensity
From Table 2.1, four study areas with EPG results were eligible for infection intensity analysis.
Table 2.14 Summary of positive egg count diagnosed by Kato-Katz and FECT methods
Na-yao Na-isarn and Na-ngam
Year 2002 2004 2013-14
Study design Prevalence Follow-up Prevalence
Positive by Kato-Katz
(%, 95% CI))
17.14
(14.02-20.64)
20.77
(16.25-25.29)
4.19
(2.77-5.62)
Positive by FECT
(%, 95% CI)
22.86
(19.33-27.31)
26.52
(21.60-31.43)
7.99
(6.07-9.92)
Sensitivity of Kato-Katz
compared to FECT as gold
standard (%, 95% CI)
74.98
(72.53-75.58)
78.32
(75.2-80.48)
52.44
(45.63-56.65)
% of excess positive from FECT
(% of excess proportion)
5.72
(25.02)
5.75
(21.68)
3.80
(47.56)
Table 2.14 reports the percentage of positive cases diagnosed by Kato-Katz or FECT. Both methods
are microscopic-based where fluke eggs need to be identified. However, FECT is a more sensitive
technique than Kato-Katz (p <0.05).
Considering FECT as a gold standard and responsible for all positive results (Table 2.8), sensitivity for
Kato-Katz in Na-yao from 2002 - 04 study was 74.98 - 78.32%. Sensitivity for Na-isarn and Na-gnam
was 52.44% or approximately half the proportion of FECT result.
To perform the infection intensity analysis, EPGs were calculated through the egg output per slide
from Kato-Katz technique. Table 2.14 shows the prevalence and intensity from Na-ngam and Na-
isarn data were also less than Na-yao 2002-24 data. As a result, I combined Na-isarn and Na-ngam
prevalence data to gain sufficient positive results for intensity analysis.
86
2.3.5.1 Prevalence and intensity profiles
Table 2.15 Summary of Na-yao 2002 intensity data
Age group
(years) N
Prevalence (%)
(95% CI)
EPG, S.D.
(95% CI) Variance Range
0 – 11 99 6.06
(2.26 – 12.73)
7.52, 61.87
(6.98 – 8.08) 3828.01 0 - 600
12 – 32 157 20.38
(14.38 – 27.54
12.09, 42.59
(11.54 – 12.63) 1813.84 0 - 312
33 – 46 132 25.00
(17.88 – 33.28)
53.45, 263.00
(52.21 – 54.72) 69169.17 0 - 2376
≥ 47 137 35.77
(27.77 – 44.40)
153.46, 460.01
(151.39 – 155.55) 211606.80 0 - 3048
Total 525 22.86
(19.33 – 27.31)
58.51, 277.43
(57.86 – 59.17) 76969.91 0 - 3048
0
50
100
150
200
250
300
350
400
450
Freq
uen
cy
Number of egg count
Na-yao 2002 cross-secitonal study
0
5
10
15
20
25
30
35
40
45
50
0 5 10 15 20 25 30 35 40 45 50 55 60
Pre
vale
nce
(%
)
Age group
Na-yao cross-sectional study
0
20
40
60
80
100
120
140
160
180
0 5 10 15 20 25 30 35 40 45 50 55 60
EPG
Age group
Na-yao 2002 cross-sectional study
Figure 2.11 Frequency distribution of EPG from Na-yao 2002 cross-sectional study
Figure 2.10 Prevalence and intensity of Na-yao 2002 cross-sectional study
87
Table 2.15 shows that EPG ranged between 0 and 3048, but the majority of individuals harboured
few eggs or zero eggs as shown in Figure 2.10. Only few cases carried large amounts of eggs;
therefore, the distribution resulted in a high variance exceeding the mean egg count, i.e., EPG of
58.51 with S.D. of 277.43 and variance of 76969.91.
Figure 2.11 shows the prevalence of the infection rapidly increased in the very first years from 0 - 15
years of age and then slowly increased thereafter. The mean intensity showed a different pattern in
that the egg count gradually increased by age at younger ages and raise quickly in older years. The
egg count ≥100 was observed in the age group ≥ 47years.
88
Table 2.16 Summary of Na-yao 2004 intensity data
Age group N Incidence (%)
(95% CI)
EPG, S.D.
(95% CI) Variance Range
0-14 91 8.23
(3.64 – 15.58)
5.54, 21.78
(5.07 – 6.04) 474.58 0 - 120
15-40 75 25.11
(16.58 – 35.46)
52.8, 212.84
(51.17 – 54.47) 45301.62 0 - 1800
41-53 72 23.18
(14.68 – 33.75)
41.33, 149.51
(42.81 – 45.90) 22353.01 0 - 1080
≥54 75 36.42
(26.93 – 46.90)
188.48, 532.44
(185.39 – 191.61) 283496.20 0 - 3048
Total 313 23.08
(18.87 – 27.77)
69.62, 296.68
(68.70 – 70.55) 88017.10 0 - 3048
0
50
100
150
200
250
Freq
uen
cy
Number of egg count
Na-yao 2004 follow-up study
0
5
10
15
20
25
30
35
40
45
50
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Inci
den
ce (
%)
Age group
Na-yao 2004 follow-up study
0
50
100
150
200
250
0 5 10 15 20 25 30 35 40 45 50 55 60 65
EPG
Age group
Na-yao 2004 follow-up study
Figure 2.13 Frequency distribution of EPG from Na-yao 2004 follow-up study
Figure 2.12 Prevalence and intensity of Na-yao 2004 follow-up study
89
Figure 2.12 reveals the pattern of egg count distribution was similar to the cross-sectional study.
Table 2.16 indicates the range of egg count was 0 - 3048 and mean intensity was 69.62 with S.D. of
296.68 and variance of 88017.10. The oldest age group still exhibited the highest egg intensity.
Figure 2.13 demonstrates the incidence increased from age 0 to age 35 and then decreased before
continuing to reach the highest incidence at age group ≥54 years. The mean intensity showed the
same pattern where intensity rose abruptly from the 41 - 53 age group to the ≥54 age group.
90
Table 2.17 Summary of Na-isarn and Na-ngam 2012 - 13 intensity data
Age group N Prevalence
(95% CI)
EPG, S.D.
(95% CI) Variance Range
0 – 35 200 3.00
(0.62-5.37)
1.08, 6.93
(0.94 – 1.23) 48.03 0 - 72
36 – 45 186 9.14
(4.98-13.30)
5.16, 28.75
(4.84 – 5.50) 826.32 0 - 312
46 – 57 202 9.41
(5.36-13.45)
4.87, 30.22
(4.57 – 5.19) 913.23 0 - 408
≥ 58 175 10.86
(6.23-15.49)
6.17, 29.86
5.81 – 6.55) 891.90 0 - 336
Total 763 7.99
(6.07-9.92)
4.25, 25.72
(4.10 – 4.40) 661.50 0 - 408
0
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
Freq
uen
cy
Number of egg count
Na-ngam and Na-isarn 2012-13 cross-sectional study
0
2
4
6
8
10
12
14
16
18
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Pre
vale
nce
(%
)
Age group
Na-ngam and Na-isarn 2012-13 cross-sectional study
0
1
2
3
4
5
6
7
0 5 10 15 20 25 30 35 40 45 50 55 60 65
EPG
Age group
Na-ngam and Na-isarn 2012-13 cross-sectional study
Figure 2.15 Frequency distribution of EPG from Na-ngam and Na-isarn cross-sectional study
Figure 2.14 Prevalence and intensity of Na-ngam and Na-isarn cross-sectional study
91
The infection intensity in Na-isarn and Na-ngam from 2012 – 13 study was significantly lighter (p =
0.03) than the Na-yao 2002 study. Table 2.17 indicates EPG was 4.25 with S.D. of 25.72 and variance
of 661.50. However, the intensity distribution was similar to Na-yao data where the majority of cases
had zero egg count.
Prevalence and intensity showed similar patterns where they increased from zero to 40 years and
slightly increased afterward as shown in Figure 2.15.
92
2.3.5.2 The prevalence-age structure
Assuming that incidence (λ) is constant over time and for all age groups;
p(a) proportion of infected individuals at age a
s(a) proportion of uninfected individuals at age a
Therefore;
𝑠(𝑎) = 1 − 𝑝(𝑎) (2.7)
From Equation 2.7, s(a), can be described using following expression;
𝑠(𝑎) = 𝑒−𝜆𝑎 (2.8)
Taking the natural logarithm on both sides of Equation 2.8;
− 𝑙𝑛(𝑠) = 𝜆𝑎 (2.9)
Refers to general linear function: Y = mX + c, the incidence (λ) from Equation 2.9 can be estimated as
the slope of –ln (s) against age;
Table 2.18 Summary of prevalence-age structure
Area R2 𝝀
(/person-years) 95% CI p - value
Na-yao 2002 0.902 0.0070 0.0028 – 0.0113 0.013
Na-isarn and Na-ngam 2012 - 13 0.985 0.0019 0.0014 – 0.0023 0.001
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
-ln
(s)
Age
Na-isarn and Na-ngam: fitting -ln(s) with age
Figure 2.16 Fitting –ln(s) with age
0
0.1
0.2
0.3
0.4
0.5
0.6
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
-ln
(s)
Age
Na-yao 2002: fitting -ln(s) with age
93
Fitting a regression model for –ln(s) as a function of age indicates that a linear relationship exists in
both cross-sectional study areas as shown in Figure 2.16. The regression coefficient could be used as
the incidence rate. Table 2.18 reveals the estimated incidence for Na-yao was 0.0070/person-years
(95% CI: 0.0028 - 0.0113) and Na-isarn and Na-ngam was 0.0019/person-years (95% CI: 0.0014 -
0.0023). By assuming that incidence is constant for all ages, the prevalence provides a linear
relationship for age-structure from both study areas.
2.3.5.3 The age-prevalence-intensity structure
Figures 2.10, 2.12 and 2.14, which refer to distribution of EPG, indicates that a large proportion of
egg count was zero and only few individuals carried large amounts of eggs. The standard Poisson
model assumes equal mean and variance for the distribution. From three datasets from Tables 2.15 -
2.17, the variance of egg counts was much greater than the mean, indicating the over-dispersion
distribution.
From Section 2.2.8, the likelihood ratio test of α = 0 showed a χ2 = 2385.77 with p-value <0.001 in
Na-yao 2002 prevalence data. For Na-isarn and Na-ngam prevalence data, the likelihood ratio test
gave χ2 = 325.46 with p-value <0.001. The α was significantly different from zero result in over-
dispersion; therefore, Poisson distribution was inappropriate. The results confirmed the evidence of
over-dispersion in the study for both prevalence and incidence data.
Over-dispersion has been observed in many helminth infections. The studies on hookworm,
ascariasis and schistosomiasis (93-95, 101) indicate the practical use of negative binomial
distribution to explore density-dependent patterns of infection dynamics.
The relationship between prevalence and infection intensity can be described by the following
Equation 2.10 below.
𝑝(𝑎) = 1 − (1 +𝑀(𝑎)
𝑘)−𝑘 (2.10)
Where the parameters are defined as
p(a) proportion of infected individuals at age a
M(a) infection intensity at age a which referred to EPG
k parasite aggregation parameter
94
As seen in Equation 2.10, I assume that the parasite distribution in the population is in negative
binomial manner where k estimates refers to level of aggregation. Worm burden, defined as
infection intensity (M), is estimated through the mean EPG for each age-group.
95
2.3.5.4 Estimation of parasite aggregation
Figure 2.17 k estimates from study areas
Table 2.19 k estimates from study areas
Na-yao 2002 cross-sectional
study Na-yao 2004 follow-up study
Na-isarn/Na-ngam
cross-sectional study
Quartile
Age group
(as years,
median)
k
(95% CI)
Age group
(as years,
median)
k
(95% CI)
Age group
(as years,
median)
k
(95% CI)
Q1 0 – 11
(7)
0.009
(0.002 – 0.023)
0 – 14
(11)
0.014
(0.005 – 0.032)
0 – 35
(17)
0.006
(0.002 – 0.015)
Q2 12 – 32
(13)
0.040
(0.025 – 0.060)
15 – 40
(32)
0.036
(0.020 – 0.061)
36 – 45
(41)
0.017
(0.009 – 0.030)
Q3 33 – 46
(39.5)
0.040
(0.026 – 0.059)
41 – 53
(45)
0.041
(0.024 – 0.067)
46 – 57
(51)
0.018
(0.010 – 0.030)
Q4 ≥ 47
(57)
0.056
(0.039 – 0.077)
≥ 54
(61)
0.056
(0.037 – 0.082)
≥ 58
(64)
0.020
(0.011 – 0.034)
Median age
(as years) 33
0.035
(0.028 – 0.043) 38
0.034
(0.026 – 0.044) 45
0.015
(0.011 – 0.020)
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0 5 10 15 20 25 30 35 40 45 50 55 60 65
K e
sim
ates
Age (year)
Na-yao prevalence data Na-yao cohort data Na-isarn/Na-ngam prevalence data
K for median age K for median age K for median age
96
Table 2.19 shows the k estimates from 3 datasets. The pattern from Figure 2.17 shows a similar
trend in that infection is highly aggregated in the younger age-group. As k increased, the infection
became progressively less clustered for older age groups. For Na-yao prevalence data, k estimates
became relatively stable from age group 12 - 32 and 33 - 46 years. When k is assumed to be constant
at median age, k estimates ranged between 0.015 - 0.035.
2.3.5.5 Infection dynamics of infection intensity
Table 2.18, confirms that uniform incidence for all ages was fitted for linear prevalence-age
structure. Figures 2.18 and 2.19 explore the infection dynamics whether infection intensity or
parasite aggregation, which denoted by M and k, are show to be age-dependent or not. Constant k is
derived using mean-age k and constant M is defined by mean-age EPG.
To explore the age-dependent structure with respect to infection dynamics of infection intensity, the
assumptions are explained below.
1. When the degree of parasite aggregation; k, was assumed to be constant with age-varying
intensity; predicted incidence with variable M was linearly fitted for observed incidence in Na-
yao follow-up data with p-value = 0.032 (Figure 2.19). For prevalence data in Figure 2.18, the p-
value = 0.058, which is not statistically significant.
2. When infection intensity; M, was assumed to be constant with age-varying parasite aggregation;
predicted prevalence with variable k was linearly fitted for both datasets in from Figures 2.18
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Pre
vale
nce
Age (year)
Na-yao 2002 prevalence data
variable M and constant k constant M and variable k
observed
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Infe
cted
pro
po
rtio
n
Age (year)
Na-yao 2004 incidence data
variable M and constant k constant M and variable k
observed
Figure 2.19 Predicted incidence from varying k
and M for Na-yao prevalence data
Figure 2.18 Predicted prevalence from varying
k and M for Na-yao cohort data
97
and 2.19 with p = 0.004 for cross-sectional study and p = 0.002 for follow-up study. Parasite
aggregation was age-dependent when infection intensity was assumed to be constant.
Constant k overestimated the prevalence in younger ages, and predicted prevalence became higher
than observed prevalence for the Q2 age-group. Predicted prevalence from constant M was
overestimated from young age until the Q3 age group and became lower than observed prevalence
in Q4.
Figure 2.20 Predicted prevalence from varying k and M for Na-ngam and Na-isarn prevalence data
Figure 2.20, Na-isarn and Na-ngam plots, show similar patterns to Na-yao follow-up data. The data
was linearly fitted for both constant k; p = 0.033 and constant M; p = <0.001.
Parasite aggregation was age-dependent from all three datasets and infection intensity was age-
dependent from 2 of 3 datasets under the assumption that incidence was constant for all age-
groups.
Therefore, the prevalence and intensity tended to increase with age and the distribution of infection
became less aggregated.
0
0.02
0.04
0.06
0.08
0.1
0.12
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Infe
cted
pro
po
rtio
n
Age (year)
Na-isarn and Na-ngam prevalence data
variable M and constant k
constant M and variable k
observed
98
2.3.5.6 Relationship of infection intensity and risk factors using count model
Table 2.20 Summary of count model for infection intensity analysis
PRM NBRM ZIP ZINB PLH NBLH
Model
Poisson
regression
Negative
binomial
regression
Zero-inflated
Poisson
Zero-inflated
negative
binomial
Poisson logit
hurdle
Negative
binomial logit
hurdle
Count model
Poisson Negative
binomial
Poisson Negative
binomial
Truncated
Poisson
Truncated
negative
binomial
Zero model Logit Logit Logit Logit
Table 2.20 summarizes 6 count models for infection intensity analysis; Poisson and Negative
binomial model (NB), and compound model; zero-inflated Poisson (ZIP), zero-inflated negative
binomial (ZINB), Poisson logit hurdle (PLH) and negative binomial logit hurdle (NBLH) model. The
covariates were comprised of age-group categorized by quartiles and consumption of Koi pla, while
the compound model incorporated the same set of covariates in both components of the model.
All models are based on the count model, reflecting the EPG as infection intensity. The compound
model incorporated with zero modeling will reflect the infection probability.
Based on the concept of modeling zero component; the zero-inflated model describes the
probability of observing zero value whereas the hurdle model predicts the probability of obtaining
positive count value 1 in the logit part. For better interpretation and comparison, the results were
transformed to the probability of obtaining 1 to provide more understanding with respect to risk
factors.
99
Table 2.21 Multivariable analysis of risk factors and infection intensity of Na-yao 2004 follow-up data
Model
PRM NBRM ZIP ZINB PLH NBLH
Count model
Age (years)
0-14 as reference
15-40 6.47
(5.89-7.10)
5.21
(1.32-20.59)
2.30
(2.09-2.52)
1.97
(0.79-4.92)
2.30
(2.09-2.52)
1.97
(0.79-4.92)
41-53 4.39
(3.99-4.82)
4.49
(1.13-17.90)
2.13
(1.94-2.34)
1.97
(0.74-4.74)
2.13
(1.94-2.34)
1.97
(0.74-4.73)
≥54 19.20
(17.56-21.00)
16.65
(3.99-66.94)
5.42
(4.96-5.92)
4.26
(1.76-10.33)
5.41
(4.96-5.92)
4.26
(1.75-10.33)
Koi pla consumption
Yes 5.56
(5.34-5.80)
4.77
(1.70-13.40)
3.32
(3.18-3.46)
2.76
(1.62-4.70)
3.32
(3.18-3.46)
2.76
(1.62-4.70)
Zero model
Age (years)
0-14 as reference
15-40 3.54
(1.44-8.68)
3.51
(1.43-8.63)
3.63
(1.48-8.90)
3.63
(1.48-8.90)
41-53 2.66
(1.06-6.71)
2.63
(1.04-6.66)
2.68
(1.06-6.77)
2.68
(1.06-6.77)
≥54 6.45
(2.68-15.53)
6.34
(2.63-15.33)
6.01
(2.74-15.92)
6.01
(2.74-15.92)
Koi pla consumption
Yes 2.54
(1.44-4.46)
2.51
(1.42-4.42)
2.54
(1.45-4.47)
2.54
(1.45-4.47)
As shown in Table 2.21, the probability of infection in zero modeling was significantly associated
with Koi pla consumption in all 4 models; the OR was 2.54 (95% CI: 1.45 - 4.46) in the PLH and NBLH
models. The OR in the ZIP and ZINB models was 2.54 (95% CI: 1.44 - 4.46) and 2.51 (95% CI: 1.42 -
4.42), respectively. All models reported similar results for age-groups. Older age-groups; Q2, Q3 and
Q4, showed significantly increased risk of acquiring OV when compared with the Q1 age-group. The
age group ≥54 years had highest risk for acquiring the infection; OR = 3.48 (95% CI: 1.34 - 9.01) from
all 4 count models.
For the count model component, Koi pla was statistically significant in all models; infection intensity
was higher among individuals who consumed Koi pla. Age groups showed similar patterns of effect
100
where older age increased intensity. Zero-inflated and hurdle models showed similar results with
respect to the same distribution.
Older age groups; Q2, Q3 and Q4, were significantly related with intensity when compared with the
Q1 age group in Poisson, NB and Poisson-based compound models; ZIP and PLH models. Age ≥54
years significantly increased infection intensity in all 6 models.
101
Table 2.22 Multivariable analysis of risk factors and infection intensity of Na-isarn and Na-ngam
2012-13 prevalence data
Model
PRM NBRM ZIP ZINB PLH NBLH
Count model
Age (years)
0-35 as reference
36-45 3.31
(2.85-3.84)
2.32
(0.41-13.28)
1.19
(1.02-1.39)
0.99
(0.44-2.25)
1.19
(1.02-1.39)
0.99
(0.44-2.25)
46-57 3.15
(2.71-3.66)
2.26
(0.41-12.31)
1.07
(0.92-1.24)
0.93
(0.42-2.07)
1.07
(0.92-1.24)
0.93
(0.42-2.07)
≥58 4.35
(3.75-5.04)
3.61
(0.69-18.90)
1.21
(1.04-1.40)
1.17
(0.54-2.52)
1.21
(1.04-1.40)
1.17
(0.54-2.52)
Koi pla consumption
Yes 3.09
(2.87-3.34)
3.01
(3.85-4.40)
1.86
(1.72-2.01)
1.95
(1.25-3.06)
1.86
(1.72-2.01)
1.95
(1.25-3.06)
Zero model
Age (years)
0-35 as reference
36-45 2.73
(1.03-7.20)
2.73
(1.03-7.20)
2.73
(1.03-7.20)
2.73
(1.03-7.20)
46-57 2.83
(1.09-7.37)
2.83
(1.09-7.37)
2.83
(1.09-7.37)
2.83
(1.09-7.37)
≥58 3.48
(1.34-9.01)
3.48
(1.34-9.01)
3.48
(1.34-9.01)
3.48
(1.34-9.01)
Koi pla consumption
Yes 1.76
(1.02-3.02)
1.75
(1.02-3.00)
1.76
(1.02-3.02)
1.76
(1.02-3.02)
As shown in Table 2.22, the infection probability model within the compound model indicated that
Koi pla consumption increased risk of infection by OR = 1.76 (95% CI: 1.02 - 3.02) in all 4 compound
models. Moreover, increasing age group significantly increased risk of infection when compared with
the Q1 age group. Risk was highest in age ≥58 years with OR = 3.48 (95% CI: 1.34 - 9.01). All models
reported similar results in age group for infection probability.
For the count model component, Koi pla significantly increased infection intensity in all models
except the NB model. Older age groups and Q2, Q3 and Q4, were observed to increase infection
intensity in the Poisson-based model. All NB-based models did not provide any significance for age
group association.
102
Table 2.23 Model fitting for Na-yao follow-up study
PRM NBRM ZIP ZINB PLH NBLH
log L -35531.44 -730.24 -17191.51 -685.23 -16830.46 -685.88
AIC 71073 1472 34395 1392 33681 1394
∆AIC 69600 36678 69680 37392 69680
Zero count
capturing (%) 2.2 97.1 100 100 100 100
Table 2.24 Model fitting for Na-isarn and Na-ngam cross-sectional study
PRM NBRM ZIP ZINB PLH NBLH
log L -9193.62 -543.15 -1955.30 -501.79 -1955.30 -501.79
AIC 18397 1098 3931 1026 3931 1026
∆AIC 17299 14467 17372 14467 17372
Zero count
capturing (%) 14.1 99.8 100 100 100 100
Model fitting for 2 datasets are described in Tables 2.23 and 2.24 revealing similar patterns.
Evidence of over-dispersion from Section 2.3.5.3 indicated that the negative binomial distribution
should be more appropriate than Poisson distribution. AIC of NB-based models was lower than
Poisson-based models: 1472.48 for NB, 1392.46 for ZINB, 1393.76 for NBLH in the Na-yao study and
1098.3 for NB and 1025.59 for ZINB and HLNB in the Na-isarn and Na-ngam study. The ∆AIC shows
the difference of AIC from each model to PRM used as the reference model. The largest ∆AIC means
the particular model provides minimal value of AIC. From both datasets, ZINB and NBLH models
provided the largest ∆AIC. Modeling the zero counts was conducted to compare the expected
number of zero counts with the observed zeros by % of zero count capturing. The results indicated
that Poisson model was not suitable when comparing with other models. Poisson model only
achieved 2.2% of zero capturing in Table 2.23 and 14.1% in Table 2.24. NB models were much better
in modeling zero counts; 97.1% and 99.8% from both datasets. However, all compound models
performed better with 100% matching of expected zero to observed zero count.
The zero-inflated and hurdle models provided a better result than Poisson and negative binomial
models. Based on fit statistics, ZINB and NBLH provided the best results from both datasets.
However, NBLH might be slightly more preferable when considering the process of how the stool
egg count is produced when the first process determining individuals are infected with OV or not,
and the second process determines how many eggs are produced from infected individuals.
103
2.4 Discussion
From this project, extensive studies were conducted in three study areas with seven fieldwork
projects highlighting various epidemiological data including prevalence, incidence, infection intensity
and risk factors.
2.4.1 Study design and response
The cross-sectional study provided prevalence data from the Mekong Basin area where OV is
endemic. However, many studies showed that prevalence of OV varies for different study areas and
tended to be localized in subpopulations depending on consumption patterns and cultural influences
(5, 10, 11, 29, 30, 69, 70). The results may not fully represent the whole picture of the Thai rural
population. Many studies indicate a strong association between uncooked freshwater fish
consumption and OV from cross-sectional designs. Causal relationship should be cautiously
interpreted with respect to study method.
Cohort designs provide better evidence of a causal relationship which strengthens the evidence for
potential risk factors. However, only negative cases from baseline survey were enrolled in the
follow-up study while positive cases were excluded. Thus, the true population-level incidence rate
might be underestimated because the high risk group was disregarded.
The cross-sectional study was conducted in two areas; Na-isarn and Na-ngam Areas. Basic
demographic profile including sex, age and occupation showed no significant differences between
the two areas. From the population pyramid, the constrictive pattern appears in late teenage to
early adult stage, considered as the working age. Informal conversations with community leaders
revealed that some villagers in this generation regularly moved out temporarily for job seeking in
urban areas especially during off-season for rice planting. Additionally, the efficiency of the local
database system was still limited for providing accurate population data. The population dynamics
are later explored and discussed with the qualitative approach in Chapter 4.
Regarding the study response in Table 2.1, 1514 participants received the survey package including
the research questionnaire and a container for stool collection. The return of the survey package
was 86.1%. Although the response rate was significantly lower than in Na-isarn area, the rate was
still over 70%. However, excluded questionnaires were approximately twice as high as in Na-ngam
Area (44.0% and 17.1%, p <0.001) resulted in lower final response rate for Na-ngam Area. After
exclusions for incomplete questionnaires, the effective response was 52.4%. The impact from
incomplete questionnaires resulted in a loss response of 33.7%. The completeness of questionnaires
104
significantly affected the prevalence in the Na-isarn study (Supplement Table S-4), where the
prevalence of OV of completed questionnaires group was significantly higher (p = 0.03) than the
incomplete questionnaire group. However, study participants provided impressive cooperation for
the high return rate of stool specimens while some limitations may have affected the completeness
of the survey packages, for example, illiteracy.
The follow-up study in Na-yao and Na-ngam Areas required a more complicated enrolment process
(Figure 2.6). Eligible participants comprised the population from baseline survey for each study area.
The follow-up study was conducted retrospectively where the eligible participants were enrolled
during the fieldwork period. The follow-up time was 15-17 months. From Table 2.3, eligible
participants totalled 1803 subjects from baseline survey.
After enrolment, only 5.8% of baseline participants were excluded due to lack of consent, death and
permanently moved-out accounting for 4.5 %. The returned survey packages were received at a
52.1% response rate. Loss to follow-up was 47.9%. Permanently moved-out participants did not
significantly differ in the two areas. However, lost participants due to unable to contact/return
specimen was statistically higher in Na-yao Area which might have resulted from less cooperation
from the villagers.
Overall the percentage of questionnaires excluded was 33.7% for the cross-sectional and 9.9% for
the follow-up study. Major obstacles were due to literacy and translation of the standard-Thai
questionnaire to the Northeastern-Thai dialect spoken by locals. The Northeastern dialect posed
some communication challenges. Local health volunteers and educated participants contributed
greatly by helping to translate local words to the local Thai dialect. In addition, any information
provided to the villagers needed to be translated back as well. From the loss of incomplete
questionnaires statistically affected the results in the Na-isarn cross-sectional study when the three
other study areas showed no significant difference from the prevalence or incidence outcome.
(Supplement Table S-4 and S-6)
As I collected uncooked fish consumption data retrospectively, potential recall bias was possible. I
collected exposure data and processed stool examination at the same time, so the outcome should
be less affected by bias (69, 70).
Considering the stool examination, the diagnostic methods depend on detecting Opisthorchis
viverrini eggs in the specimen. However, the study also aimed to provide diagnosis for other
intestinal parasites including helminthic and protozoa infection for public health benefits for the
study participants.
105
Among three diagnostic methods: simple smear, Kato-Katz and FECT, the Kato-Katz technique is
considered a standard technique used in ambulatory settings such as healthcare facility and research
fieldwork (49, 104-106). However, single Kato-Katz smear has low sensitivity for Opisthorchis
viverrini diagnosis especially from light infections. Therefore, an increased number of smears or
specimens would gain higher sensitivity. A study on praziquantel efficacy against Schistosoma
mekongi and Opisthorchis viverrini revealed that conducting one specimen with triplicate Kato-Katz
smears (1X3 scheme) or single Kato-Katz smear with 3 stool specimens (3X1 scheme) provided the
same estimate of OV prevalence. However, collecting 3 stool specimens consecutively from one
individual is likely to demand more time and workload, which seems unfeasible for such fieldwork
(49).
I used similar diagnostic methods in both prevalence and follow-up studies. A single stool specimen
was tested with three diagnostic methods instead of collecting multiple specimens assuming that I
would achieve a higher response rate using the FECT technique that could improve sensitivity for
detecting Opisthorchis viverrini eggs (107). Multiple studies have reported various effectiveness
levels of FECT for diagnosing different species of trematode including schistosomiasis and
clonorchiasis (50, 108) but information on Opisthorchis viverrini is still lacking. However, my results
revealed that FECT was exclusively responsible for all diagnoses in this study while Kato-Katz was
positive in 31.1% of OV diagnoses from combined results (Tables 2.14 and 2.22).
Many issues involve difficulty in microscopically distinguishing Opisthorchis viverrini eggs from other
parasite eggs with similar structure such as a minute intestinal fluke (MIF) and co-infection of MIF
and Opisthorchis viverrini is common in the Mekong Basin area (109, 110). However, MIF was not
found in this area in the study of Traub et al. (46).
106
2.4.2 Distribution of OV infection
2.4.2.1 Prevalence of OV infection
The overall prevalence of OV was 8.69 % (95% CI: 6.82 - 10.87) which indicated that a disease burden
still exists in the studied communities. The prevalence was slightly lower than the latest national
survey in 2001 reporting a nationwide average of 9.4%. However, the prevalence was higher
compared with regional average for the central region of Thailand (3.8%) (28). Regarding the well-
preserved Northeastern environment, cultural background could play a major role in continuing risk
behaviours, which could potentially maintain a higher prevalence in non-endemic area.
The youngest OV case was 2 years old. The prevalence for Q1 age group 0 - 36 years in Na-isarn 2013
was 3.41% (95% CI: 3.41 - 9.64) and 0 - 35 years in Na-ngam 2012 was 5.13% (95% CI: 1.90 - 10.83). A
study from 1982 - 84 also revealed that the youngest age group had the lowest prevalence (90).
Children were assumed less likely to acquire uncooked fish for consumption by themselves as their
parents probably fed them intentionally or accidentally contaminated them with metacercariae from
fish during food preparation (90, 111).
As seen in Figure 2.8, prevalence seems to increase with age. The prevalence remains relatively low
at younger age and continues to increase by age. To maintain high prevalence in older age, the
accumulation of parasites could play a major role for contributing such high prevalence. The
infection period needs to be long from either long parasite life expectancy or sustained re-infection
(99).
Prevalence-age structure was analysed in both study areas based on the assumption that the
incidence was constant over time and for all age groups. Brockelman et al. (83) has explored the age-
structure of OV prevalence in Northeastern Thailand and also estimated the incidence from the
prevalence-age relationship. The Brockelman et al. (83) result stated that prevalence would increase
accordingly with age from birth until the age of 15 years and remain relatively stable thereafter. My
results showed that prevalence would continue to increase until older age.
I assume that the difference of infection pattern depended on the prevalence itself; the Brockelman
study was conducted in an area with very high prevalence which became saturated at nearly 100 %
before 15 years of age. My study revealed a relatively lower prevalence; therefore, age influence can
be observed through older ages. Sustained prevalence has raised a question whether the infection
maintains itself for a long period. Unlike other cross-sectional studies, this study revealed a relatively
high prevalence of OV outside the traditional endemic area exhibiting some unique characteristics
for infection.
107
2.4.2.2 Incidence of OV infection
Unlike cross-sectional studies, most published incidence data could be considered outdated. Surveys
on the incidence of infection were not incorporated in the Thai national survey (27, 28). Moreover,
the national survey scheme was suspended in 2001 and the program was exclusively active in high
endemic area (28).
The incidence showed an increasing trend with age in the Na-yao 2013 study. However, a
nonspecific pattern was observed in the Na-ngam 2014 study, where the incidence peaked at 35 - 47
years of age (Figure 2.9). The youngest case detected was 2 years old. Considering the cohort
method, the 2-year-old case acquired the infection before age of 1 year. Infection in very young
children provides an interesting issue since they are newly infected during a short follow-up period.
Na-ngam Area was the area conducted with both study designs during the project timeframe. The
study began in 2012 where a cross-sectional study served as baseline for follow-up study in 2014.
Na-ngam Area received a significantly higher response rate than Na-yao Area (64.3 % and 41.2 %,
respectively, p <0.001).
As illustrated in Table 2.18, the prevalence-age structure allowed the incidence to be estimated from
the prevalence trend (Equation 2.9). Interestingly, the estimated incidence differed from that
estimated from the cohort data; 0.19/100 person-years (95% CI: 0.14 - 0.23) compared with
6.80/100 person-years (95% CI: 4.68 - 9.54) to be further explored in Chapter 3.
2.4.2.3 Intensity of the infection
The mean worm burden was indirectly assessed by the intensity of egg counts in stool specimens as
EPG. From the stool examination methods, the Kato-Katz technique was less sensitive than FECT.
Therefore, zero egg count was classified as negative even though showing a positive result from
FECT result in less infected populations than all 3 methods combined (Table 2.8).
The standard method for stool egg count for OV infection is the Kato-Katz technique, also used to
diagnose other helminth infections (71, 104, 108). However, diagnosis is made under microscope so
detection of parasite egg relies on the expertise of the examiner. False negatives mainly come from
two reasons: the examined portion of infected specimen does not contain eggs or the examiner
missed the presence of eggs in stool. The examination is considered more difficult when the
intensity is very light resulting in a low production of Opisthorchis viverrini eggs. From the three
studies for which egg count results were available, i.e., Na-yao 2002 follow-up, Na-isarn and Na-
ngam prevalence, the Kato-Katz test diagnostic sensitivity was 52.44 - 78.32 % (Table 2.14)
108
compared with FECT meaning about 21.68 - 47.56% of infected individuals were missed in the
intensity analysis. The sensitivity for Na-isarn 2013 and Na-ngam 2012 cross-sectional study was also
significantly lower than Na-yao 2002 follow-up due to the lighter intensity.
Many studies have indicated that parasitic infection is based on host heterogeneity regarding risk
exposure and transmission dynamics (95, 112, 113). Few individuals harbour a large burden of
worms, while remaining individuals carry minimal or no infections including zero excretor.
In general, standard Poisson distribution was used to analyse count data assuming equal mean and
variance. Regarding egg count data, distribution was heavily skewed due to over-dispersion resulting
in extremely large variances. The analysis indicated that negative binomial distribution was suitable
to handle over-dispersed data. Besides the prevalence-age structure, intensity and parasite
aggregation have played a major role in infection dynamics of helminth infection.
I explored the age-prevalence-intensity following the relationship as described in Section 2.3.5.3
with Equation 2.10. Assuming that incidence is uniform in all ages, infection intensity tended to
increase with age while the parasite became less aggregated.
In this study, the level of parasite aggregation was evaluated as k estimates. The data indicated that
k estimates increased with age as well, meaning that the flukes were highly aggregated at younger
age (Figure 2.17). The parasite became less clustered as k increased. The estimates of k at median
age from all three dataset ranged from 0.015 - 0.035 (Table 2.27). A study of S. mansoni reported an
estimated k value of 0.03 - 0.5 (99).
Assuming that the infection was equally distributed in all ages (constant k), the infection intensity
was statistically associated with linear age-structure from 2 of 3 datasets (Section 2.3.5.5).
Furthermore, assuming constant M provides the result that the level of parasite aggregation was
progressively less clustered with older age as well. The result highlighted that age-structure was
involved in prevalence, intensity and parasite aggregation. At very first years of age, the parasite
locally aggregated in particular hosts with relatively low intensity and the prevalence was still low as
the infection occurred in few individuals. The infection became less aggregated and distributed more
evenly with higher intensity among older hosts.
Bundy et al. (93, 94) reported in the study of intensity in three helminths (Ascariasis, hookworm and
Trichuriasis) that despite the prevalence of all helminths becoming stable after age 5 years, the
parasitic aggregation was also stable while the intensity seemed to fall at older ages. I assumed that
the different pattern came from the reproductive biology of the parasite. As the individual host is a
closed system, the flukes would need adequate resources to support their life and contribute to
109
reproduction. When the number of flukes reaches the critical level, the lack of resources and highly
competitive environment will affect their ability to reproduce, resulting in a decline in egg output
per fluke. The parasite becomes more equally-distributed because almost everyone is infected.
My fieldwork reflected a picture of lighter infection when the age-structure can be observed through
the ages. Despite the mode of infection, children can acquire infection from a very young age. The k
estimate ranged from 0.015 - 0.035. The infection seemed to progressively accumulate and be
distributed across the population.
2.4.3 Risk factors for acquiring the infection
Infection status was assessed by logistic and Poisson regression models according to study methods.
The model shows that using two covariates, age group and consumption of Koi pla, were appropriate
for the analysis, which could extend to infection intensity outcome where the models were more
complicated. The results summarize that consumption of Koi pla is a potential risk factor for
acquiring the infection as it shows a significant association from 5 of 6 models (Section 2.3.4.3 and
2.3.5.6). Koi pla consumption also increases the risk of OV and infection intensity when incorporated
with the NBLH model.
Overall picture reveals that Q4 age-group or oldest age is significantly related with increasing risk of
infection from 3 of 6 studies; age ≥57 years in Na-isarn 2013 prevalence data, age ≥54 years in Na-
yao 2004 cohort data and age ≥58 years in Na-isarn and Na-ngam prevalence data from the NBLH
model. A trend was observed among age-groups in the NBLH model where risk of acquiring infection
increased with age when compared with the reference age group. However, the same trend was not
observed for infection intensity.
OV is biologically acquired by uncooked fish consumption where the infective stage; metacercariae,
is found in raw fish meat. Consumption of these dishes might directly lead to increasing risk of the
infection regarding method of preparation.
Two main local dishes from the study also contained raw fish meat for which various methods of
preparation depend on how the fish is cooked and preserved. Koi pla is consumed immediately after
mixing with ingredients. A study on survival of metacercariae by preserving fish meat under sodium
chloride and acetic acid resembling the preparation method reveals that metacercariae were
completely viable in the very first hours and completely inactive or presumed dead after two days of
preservation. For Pla ra, it takes at least 3 - 6 months with extremely concentrated salt preservation,
a conclusively unsuitable environment for the parasite to survive (87, 88).
110
Regarding the distribution of uncooked fish consumption (Supplement Table S-1 and S-2), only 12.5%
from the cross-sectional study and 14.3% from the follow-up study did not report consumption of
any uncooked fish dish within the study period, resulting in 85.7 - 87.5% of the population
experiencing at least one dish. For Pla ra only, the consumption could reach as high as almost 90%.
Therefore, a health campaign carrying a broad message to avoid all kinds of uncooked fish might
result in confusion and be impractical for daily lifestyle of the local villagers.
Population characteristics play an important role for the uncooked fish consumption. Male sex and
older age-group was significantly associated with consumption of Koi pla when compared with Q1,
the youngest age-group. Agricultural-related work was related with consumption of Pla ra in some
areas. In terms of consumption behaviours, the quantitative approach may not fully explain the
pattern. Method of consumption was a matter of behaviour relating to attitude, social value and
more importantly, traditional culture. While OV is more likely to be a global burden, consumption
patterns seemed to be localized and environment-related (29, 30). A qualitative approach was able
to explain and describe its epidemiology in terms of bio-psycho-social aspects (114). Concerning
uncooked fish consumption in this community, a strong attachment was found to tradition and
culture including main popular dishes as described in the results.
National Control Program suggests avoiding all uncooked fish to prevent the infection and provide
treatment for infected cases (27, 28). Prevention campaigns might be impractical to villagers
because they have to change most of their daily habit. Pla ra was the main ingredient in most all
local Northeastern dishes while Koi pla was more likely a main course. In addition, the chemical
properties of Pla ra, contain highly concentrated salt, and did not allow metacercariae to survive (47,
87, 89). As a result, implementation of controlling specific risks seemed to be more practical and
convenient (69).
Compound model
Over-dispersion poses a challenge for analysis of egg intensity data. The negative binomial
distribution provides a better fit than the classical Poisson model for the over-dispersed data
observed in many helminth intensity studies (93, 94, 115).
The NBLH model separated the infection probability component of the model from the intensity
component. In the logit model, where the probability is determined, risk for acquiring OV increases
with older age-groups when compared with the reference age-group. For logistic and Poisson
models, only particular age-group was associated while the trend was not clear.
111
If I assume that incidence is uniform at all ages, infection intensity tended to linearly increase with
age when the parasite aggregation was assumed to be constant. The level of parasite aggregation
also showed a linear trend with age when the infection became less clumped as age increased.
Therefore, both intensity and parasite aggregation would interact accordingly to age and provide a
different pattern from risk factor analysis. Moreover, it would be more realistic to think about
incidence as a dynamic parameter rather than in constant terms (94).
Zero excretor may be the result of light infection in which small amounts of eggs are produced
increasing the chance to miss the diagnosis under the microscope. The egg count method relies on
Kato-Katz technique with single stool sample. Because only 42 mg of stool was used the chance was
increased that the tested stool sample would not contain parasite egg as well.
The zero-inflated model included explicit consideration of the zero outcomes with sampling (from
lightly infected individuals) and structural (from uninfected individuals) origin. Thus, the count part
of the model will contain zero count intensity. Regarding the outcome, sampling zero could be the
result of zero excretor where I assume that infected cases produce zero egg.
The zero excretor can be seen by comparing FECT to Kato-Katz results; the additional infected cases
diagnosed by FECT and report zero count by Kato-Katz were 5.75% for Na-yao 2004 data and 3.80%
for Na-isarn and Na-ngam 2012 – 13 data. However, based on the examination method, zero count
was regarded as an uninfected case in the first place. Therefore, the infection intensity could be
measured by Kato-Katz only.
The results point out that Kato-Katz underestimates the results when compared with FECT. This is
considered a limitation to my results of intensity analysis.
The model only needs to handle zero excess as a cause of over-dispersion where few individuals
carry a large burden of intensity while the remaining report zero or minimal intensity. The NBLH
model with negative binomial in the count module showed a better fit to the data than PLH. Zero
excretor may not interfere with Kato-Katz result. However, a false negative will definitely affect the
outcome in underestimating prevalence and incidence in the compound model.
Consumption of Koi pla is statistically associated with infection intensity but age-groups do not
provide any consistent pattern. Considering the method of data collection, the question on
consumption behaviour was more likely a qualitative assessment whether the participants ever
consumed uncooked fish dishes or not within the study period and frequency of consumption was
not recorded. More data may be required on exposure in terms of consumption frequency for a
better explanation for its association with infection intensity.
112
2.5 Conclusion
Extensive epidemiological study and statistical analyses were performed in a rural area of Thailand
where OV infection is still a health burden. Prevalence and incidence are relatively high considering
that the studied areas were considered non-endemic for the infection. Population characteristics
played an important role in contributing to risk of infection, especially the consumption of particular
traditional uncooked fish menus.
In summary, age-structure was observed in prevalence of OV with regard to intensity and level of
parasite aggregation; population characteristics and risk factors were identified contributing to OV.
Infection intensity showed the pattern of over-dispersion and compound models were introduced to
handle the distribution issues and excess zeros outcomes.
Consumption of instant uncooked fish dishes such as Koi pla has been identified as linked with OV (1,
5, 15, 27-29). My results with two cohorts and additional supplement data from another 2 follow-up
studies (69, 70) provided rational supporting evidence for the casual relationship of Koi pla, which is
clearly identified as a potential risk factor for acquiring OV and should be prioritized as a major
concern for prevention efforts.
Prevalence and consumption of uncooked fish are both age-related. Because the infection can
apparently reside and remain for long periods, I could assume that flukes may have a long life
expectancy or the host is constantly re-infected. Infected cases could accumulate in the long run and
could effectively transmit the parasite in completing its life cycle. Moreover, newly-emerging cases
are also capable of parasite transmission. Especially in the event of no treatment or intervention, the
infection remained stable and increased the risk of cholangiocarcinoma, a bile duct cancer.
The study highlights the epidemiological view of OV in the rural community setting the majority area
of Thailand. Data obtained from the field study could be useful for extensive epidemiological tools to
gain more insights into the complexity of the biology and disease risk relationship providing greater
understanding of the infection dynamics and parasite-host relationship. Disease surveillance might
be useful for areas with a high burden of disease where the high risk population could be primarily
identified with a history of uncooked fish consumption. Community-based intervention should be
extensively considered with respect to National Control Programs, but the strategy needs to be
adapted to the social/cultural settings, which change over time. Up-to-date information on disease
dynamics has a potential impact on public health and further ongoing extensive research in this field
is essential to provide effective public health management.
113
Chapter 3 Infection dynamics of OV infection
Abstract
Mathematical models were applied to explore the infection dynamics of OV infection in a rural area
in Thailand. They were calibrated using epidemiological data obtained from fieldwork to gain more
insights regarding the infection.
The model was based on the classic susceptible – infectious – recovered (SIR) model. The main
extension was adding a second susceptible class to create primary susceptible individuals S1 who
were infected for the first time. Infected individuals will naturally recover and become secondary
susceptible (S2) again for re-infection with reduced infectivity. The model assumed that infected
individuals recover because the Praziquantel treatment will gain longer immunity before potentially
becoming re-infected. The force of infection was assumed to be uniform and the recovery rate
depended on the parasite life expectancy.
The model was based on key data from the fieldwork. The fieldwork data included 2 study areas; Na-
yao Area comprising 4 consecutive prevalence studies with an incidence rate of 22.1/100 person-
years (95% CI: 17.6 – 27.3), Na-ngam Area comprised 2 prevalence studies with an incidence rate of
6.8/100 person-years (95% CI: 4.7 – 9.5). The model assumed that the infection was in the endemic
phase where Na-yao prevalence was 22.9% (95% CI: 19.3 – 26.5) and Na-ngam was 9.3 % (95% CI:
6.9 – 11.6). The effectiveness of praziquantel treatment was estimated in Na-yao area where
multiple surveys had been performed. Basic reproductive numbers (R0) were estimated from the
model. Finally, the model was used to assess the effectiveness of the National Control Program.
Using model, the average infectious time was estimated in the range 1.51 – 1.81 years and that
treatment effectiveness, serving as the cure rate of praziquantel was 80.1 – 92.5%. An alternative
calculation from individual-level fieldwork data gave a similar estimate of 88.7 – 92.2 % of treatment
effectiveness. R0 for Na-yao was 2.74 (95% CI: 2.61 – 2.89) and for Na-ngam was 2.30 (95% CI: 2.23 –
2.40). Compared with other trematodes; S. mansoni had an infectious period of 3 – 4 years and R0 1
– 2. The provided treatment with praziquantel appeared to yield the same effectiveness as that
observed in previous studies at 80 – 95%. When the model was applied to the National Control
Program data; uniform incidence was estimated to be 9.8 – 47.3/100 person-years. Re-infection
played an important role for the chronic infection picture and shaped a sustained prevalence.
The model provided a good insight for infection dynamics of OV infection. However, parameter
values using a likelihood approach were unable to be formally estimated and therefore the
114
numerical findings could have been more robust. The density-dependent pattern could have played
an important role as it was a common characteristic of parasitic infection. Further studies with more
data can put forward the understanding of model development with prospective up-to-date
information.
115
3.1 Introduction
Mathematical modelling has been applied to almost all scientific branches. Medical science is one
discipline where the application of mathematical modelling is constantly increasing including the
field of infectious disease (116). Transmission is a natural process of inter-organism interaction
when the infectious agent invades another organism; the host. Sometimes the process is pathogenic
causing illness or disease (112, 117, 118). The challenging issues are that rates of transmission are
influenced in a complicated way by the epidemiological triangle; host, pathogen and environment.
None of them interact with transmission in a straightforward way such that it is easy to describe the
system accurately with a mathematical formula.
Helminths are naturally complex organisms that mostly require multiple stages in their life cycle to
grow and develop (116). Human infection with helminths usually occurs in tropical areas where
intermediate hosts are abundant. Helminth infections are still considered to be neglected tropical
diseases (NTDs). Infection is typically prevalent in underdeveloped and developing countries where
resources are considerably limited (14, 43).
Some NTDs have contributed a huge impact on global scale. Schistosomiasis, or blood fluke, affects
over 100 million people mostly in underdeveloped countries raising concerns from international
authorities (119). Studies of its infection dynamics provide good insights on the disease (99, 112,
113, 117). Human and parasite factors have been incorporated in complex models capable of
describing the infection in almost every aspect leading to massive intervention on prevention and
control programs (120-123).
OV infection is regarded as an NTD because of the lack of resources and has been prioritized for
control (10, 11, 30). Infection posts some serious complications and threatens the population’s
health in endemic areas. OV infection is caused by Opisthorchis viverrini prevalent in Southeast Asia
especially along the Mekong River Basin including Thailand, Lao PDR and Cambodia. The prevalence
is very high in some areas and can reach up to 90% (29, 51).
Opisthorchis viverrini is a major pathological human liver fluke (along with other 2 species), endemic
in other areas of the world. Almost 10 million people are affected by the infection. More
importantly, the parasite is classified as a carcinogen causing cholangiocarcinoma; a lethal bile duct
cancer. The incidence of cholangiocarcinoma in Thailand is the highest in the world (15-17, 20, 22-
24). In addition to the benefits from prevention and control, resulting in better hygiene and quality
116
of life, it is promising that cholangiocarcinoma can be prevented from Opisthorchis viverrini
infection.
This chapter aims to explore the infection dynamics of OV infection. Based on the conceptual
framework shown in Figure 3.1, the relationship of risk behaviours and OV infection can be reflected
as uninfected and infected individuals for the infection dynamics situation. Data obtained from
fieldwork including prevalence and incidence of infection can be used to formulate a mathematical
model to explore its interaction [A] as shown in Figure 3.1. Additionally, the effectiveness of
treatment on OV infection can be evaluated [B] through follow-up study with regard to study design.
The relationship of prevalence and incidence has been discussed previously in Section 2.3.5.2 where
a linear regression model was fitted with prevalence-age structure. The following section will discuss
the interesting issues between directly-obtained incidence from the fieldwork and incidence from
prevalence-age structure leading to the formulation of more robust mathematical modelling.
A and B are referred in the main text
Uninfected
(Risk behaviors)
Social influence
Infected
(Opisthorchiasis)
Intervention
Prevention Diagnosis and Treatment
A
B
Figure 3.1 Conceptual framework for infection dynamics.
117
3.1.1 The relationship of prevalence and incidence
In general, disease burden can be measured by prevalence and incidence (124). In this study,
prevalence is defined as the proportion of individuals being diagnosed for OV infection by stool
examination over total studied population at a given point time. Incidence refers to the number of
new cases during a follow-up period of time. Therefore, incidence can be thought of as the rate of
emerging cases at a given time.
Prevalence is usually considered a less challenging way to measure disease burden. Incidence
requires more effort because direct measurement needs a follow-up study to estimate newly-
emerged cases from baseline data. Considering that and population measurement consumes
considerable time and resources, and for field investigators to conduct at least two surveys to obtain
a proper incidence is often difficult in some circumstances (83, 124).
Indirect methods for measuring the incidence of OV infection have been proposed by Brockelman et
al. using age-prevalence data (40, 83). Their results demonstrated a good age-prevalence profile
using a catalytic infection model following the regression method. Therefore, the single cross-
sectional study could provide the incidence.
According to Brockelman’s work, the regression method should consider the following assumptions
stated below;
● The infection should be in the endemic state where prevalence shows stable pattern.
● The incidence should remain stable as long as the oldest age population have been alive.
From Section 2.3.5.2 in Chapter 2, data from Na-yao 2002 – 04 and Na-ngam 2012 – 14 study shows
that incidence can be fitted from prevalence-age structure.
Table 3.1 Obtained incidence from fieldwork and estimated incidence from prevalence-age structure
Study area
Incidence from prevalence-age structure
(/100 person-years)
(95% CI)
Incidence from fieldwork
(/100 person-years)
(95% CI)
Na-yao 2002 - 2004 0.70
(0.28 – 1.13)
22.05
(17.57 – 27.34)
Na-ngam 2012 - 2014 0.21
(0.12 – 0.30)
6.80
(4.68 – 9.54)
Table 3.1 shows that the incidence from prevalence-age structure is significantly lower than direct
incidence obtained from fieldwork (p < 0.01).
118
As shown in Figure 3.2, fitting prevalence with provided incidence with Equations 2.7 and 2.8 in
Chapter 2 showed that incidence from prevalence-age structure provided a linear trend for
underling prevalence; p – value = 0.012 and 0.005 for Na-yao and Na-ngam areas, respectively.
Therefore, the possibility of using a mathematical model to explore the relationship between
prevalence and incidence from the fieldwork was considered to better understand infection
dynamics.
The intentions for this section of the thesis are stated below.
● To explore the infection dynamics of OV infection with epidemiological data from the fieldwork
emphasizing human population dynamics
● To explore the impact of treatment on infection dynamics and apply the findings to the National
Control Program data.
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20 25 30 35 40 45 50 55
Pre
vale
nce
(%
)
Age
Na-yao 2002: fitting prevalence
Observed
Predicted from age-structure incidence
Predicted from fieldwork incidence
0
10
20
30
40
50
60
70
80
90
100
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Pre
vale
nce
(%
)
Age
Na-ngam 2012: fitting prevalence
Observed
Predicted from age-structure incidence
Predicted from fieldwork incidence
Figure 3.2 Fitting prevalence with age-structure incidence and fieldwork incidence
Figure STYLEREF 1 \s 3. SEQ Figure \* ARABIC \s 1 2 Fitting prevalence with age-structure incidence and fieldwork incidence
119
3.2 Methodology
3.2.1 Model definition
The infection dynamics are explored with a compartmental deterministic model with some
parameters obtained from the literature and other obtained by comparing model output with
observed field data. The individuals in the population were formed in subgroups as compartmental
models and the infection dynamics were observed among these individuals uniformly. A differential
equation was applied to describe the rate of change in the number of individuals over time.
The model structure was based on the SIR framework. That infectious individuals tended to be re-
infected once they temporarily recovered was assumed (99).
Therefore the model is extended to S1IS2R framework; Primary susceptible – Infectious – Secondary
susceptible – Recovered.
Figure 3.2 show the model diagram with S1IS2R framework. The model assumes that the infection
process starts with primary susceptible individuals (S1) where they have never been infected with OV
infection before.
Then they become infectious (I); the infectious time depends on parasite life expectancy in the
human host body. The incubation period of OV infection is relatively short; approximately 2 – 4
weeks (3, 9), compared with the entire infection process, so the pre-infectious compartment is not
incorporated in the model.
μβI/N
Births (b) Primary
Susceptible
S1(t)
Infectious
I(t)
Secondary Susceptible
S2(t)
βI/N
(1-ε)·r
ε·r Deaths (m)
Deaths (m)
Deaths (m)
Recovered
R(t)
Deaths (m)
γ
Figure STYLEREF 1 \s 3. SEQ Figure \* ARABIC \s 1 3 Model diagram for Opisthorchiasis infection dynamics Figure 3.3 Model diagram for opisthorchiasis infection dynamics
120
Infectious individuals naturally recovered with rate r. To maintain its endemicity, recovered
individuals are considered as re-infected, so the compartment was defined as secondary susceptible
(S2). That infectious individuals would have some immunological response to the infection was
assumed; therefore, the secondary susceptible individuals would be re-infected with reduced
infectivity (μ) as a result of partial immunity.
In case treatment was provided, a fraction (ε) of infectious individuals would recover and be immune
for some period while the remaining fraction (1 - ε) of infectious individuals would proceed to the S2
compartment as a natural process of the infection. However, the immunity against infection was
considered temporary as well; protective immunity will prevent the recovered individuals from being
re-infected for some period. Waning of the immunity will move them to the S2 compartment again
with the rate γ.
In addition, the model also aimed to describe the long-term transmission of the infection. Population
demographic parameters such as birth rate and death rate were also incorporated in the model.
Table 3.2 Summary of fieldwork data for Na-yao area
Year Outcome Value
(95% CI) Source
2002 Prevalence 22.9 %
(19.3-26.5)
Revised data from (69)
2004
Prevalence 20.7 %
(17.6-23.9)
Incidence 22.1 /100 person-years
(17.6-27.3)
2007 Prevalence 18.6 %
(16.4-20.9) Revised data from (70)
2011 Prevalence 10.7 %
(8.7-12.7) Project fieldwork
The data used in this chapter comprised 2 sets of fieldwork data: Na-yao and Na-ngam area. The Na-
yao area has been extensively studied since 2002. The results from studies in 2002-04 and 2007-09
were published with incidence and risk factors of OV infection (69, 70) and the current 2011 - 13
study was included in this PhD project. The 4 point prevalence data available was used to observe
infection dynamics. The incidence rate from the 2004 (69) data served as a force of infection of the
study as shown in Table 3.4.
121
Table 3.3 Summary of fieldwork data for Na-ngam area
Year Outcome Value
(95% CI) Source
2012 Prevalence 9.3 %
(6.9 – 11.6)
Project fieldwork
2014
Prevalence 8.1 %
(6.2 - 10.0)
Incidence 6.8 /100 person-years
(4.7 – 9.5)
As shown in Table 3.5, the summary cross-sectional study was conducted in Na-ngam Area in 2012
and a follow-up study in 2014 with 14 months follow-up period. Both studies used comprehensive
surveys, so prevalence and incidence were applicable for the study.
122
3.2.2 Parameters of the model
Diagram of compartmental model in Figure 3.3 could refer to following differential equations in
Figure 3.4 from Equation 3.1 to 3.4:
𝑑𝑆1(𝑡)
𝑑𝑡= 𝑏𝑁(𝑡) −
𝛽𝐼(𝑡)
𝑁(𝑡)𝑆1(𝑡) − 𝑚𝑆1(𝑡) (3.1)
𝑑𝐼(𝑡)
𝑑𝑡=
𝛽𝐼(𝑡)
𝑁(𝑡)𝑆1(𝑡) − 𝜀𝑟𝐼(𝑡) − (1 − 𝜀)𝑟𝐼(𝑡) + 𝜇
𝛽𝐼(𝑡)
𝑁(𝑡)𝑆2(𝑡) − 𝑚𝐼(𝑡) (3.2)
𝑑𝑆2(𝑡)
𝑑𝑡= (1 − 𝜀)𝑟𝐼(𝑡) − 𝜇
𝛽𝐼(𝑡)
𝑁(𝑡)𝑆2(𝑡) + 𝛾𝑅(𝑡) − 𝑚𝑆2(𝑡) (3.3)
𝑑𝑅(𝑡)
𝑑𝑡= 𝜀𝑟𝐼(𝑡) − 𝛾𝑅(𝑡) − 𝑚𝑅(𝑡) (3.4)
Figure 3.4 Differential equations of S1IS2R model
The definitions of population compartments are as follow:
𝑑𝑆1(𝑡)
𝑑𝑡 The rate of change in the number of primary susceptible individuals at time t
𝑑𝐼(𝑡)
𝑑𝑡 The rate of change in the number of infectious individuals at time t
𝑑𝑆2(𝑡)
𝑑𝑡 The rate of change in the number of secondary susceptible individuals at time t
𝑑𝑅(𝑡)
𝑑𝑡 The rate of change in the number of recovered individuals time t
N(t) Total population size at time t which equals to S1(t) + I(t) + S2(t) + R(t).
123
Table 3.4 Model parameters
Parameters Definition Value
βI/N
Force of infection; the rate at which primary
susceptible individuals becoming infected
per unit time at time t
● Na-yao study: 0.221/person-years (95% CI:
0.176 – 0.273)
● Na-ngam study: 0.068/person-years (95%
CI: 0.047 – 0.095)
r
Recovery rate; the rate at which infectious
individuals recover from the infection per
unit time
● Fitted from the model
ε Fraction of infectious individuals being
recovered from treatment
● ε = 0 for fitting recovery rate
● Fitting for the treatment effectiveness
1 - ε
Fraction of infectious individuals being
naturally recovered and become secondary
susceptible
μ Fraction of reduced infectivity due to
immune modulation 0.9 (40)
μβI/N
Force of re-infection; the rate at which
secondary susceptible individuals becoming
re-infected per unit time at time t
Force of infection reduced by fraction μ
γ The rate of waning immunity 0.1/person-years (125-127)
b Per capita birth rate 0.0129*
m Per capita mortality rate 0.0072*
* From World Bank database http://data.worldbank.org/country/thailand
The force of infection is determined from the rate at which susceptible individuals become infected
per unit time at time t. The rate is obtained from the fieldwork where incidence rate (λ) was directly
measured from the follow-up study from both Na-yao and Na-ngam studies.
The model framework assumed that force of infection was uniform. Individuals were mixed
homogeneously regardless of demographic characteristics.
The incidence from both published follow-ups study (69, 70) in Na-yao area indicate that the
incidence rate is fairly constant with value of 22.1/100 person-years (95% CI; 17.6 – 27.3) and
21.4/100 person-years (95% CI; 18.5-24.7), respectively. Therefore the model uses incidence rate
from first follow-up from both 2004 Na-yao and 2014 Na-ngam study as a force of infection.
124
Force of infection is following frequency-dependent assumption where the risk of infection remains
unchanged over increased population referring to the Equation 3.5;
𝜆(𝑡) = 𝛽𝐼(𝑡)
𝑁(𝑡) (3.5)
𝛽 is the number of individuals effectively contacted by each person per unit time. Since I assume the
incidence is constant, 𝛽 at time t can be indirectly calculated by Equation 3.6;
𝛽(𝑡) = 𝜆𝑁(𝑡)
𝐼(𝑡) (3.6)
So, number of susceptible individuals getting infected is calculated by Equation 3.7;
𝛽𝐼(𝑡)
𝑁(𝑡)𝑆(𝑡) = 𝜆
𝑁(𝑡)
𝐼(𝑡)
𝐼(𝑡)
𝑁(𝑡)𝑆(𝑡) = 𝜆𝑆(𝑡) (3.7)
In this setting, 𝛽 will be vary in order to maintain constant incidence rate in growing population.
The force of infection was assumed to refer to infection of the primary susceptible individuals. Re-
infection occurring in naturally-recovered individuals resulted in lower rates due to the partial
influence from immune modulation.
The data on re-infection rate was very limited. The rate was estimated from the evaluation of re-
infection rates after praziquantel treatment from previous studies (40) indicating that the force
infection could be reduced to 90% from primary infection.
125
3.2.3 Fitting the model
From mathematical model, I fit the parameters by finding a point estimated value from the
differential equation.
Little is known about OV life expectancy which reflects the duration of infectiousness. Therefore, the
model was used to fit the duration of infectiousness considering the fieldwork data. The indirect
evidence obtained by the immunological response from the infected host could be used to estimate
a life expectancy at 1 – 10 years (9, 13, 45, 70, 128). Compared with other helminths, life expectancy
of S. mansoni is 3 – 4 years and other soil-transmitted helminths are 1 – 4 years (99, 113).
Assuming that re-infection is a main feature of the infection, where no infected individuals could
develop protective immunity (ε = 0), infected individuals should recover due to the parasite life
expectancy solely and tend to be re-infected again.
The duration of infectiousness was fitted from the model and the rate of recovery; defined as r,
duration of infectiousness was estimated from Equation 3.8;
𝐷𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑖𝑛𝑓𝑒𝑐𝑡𝑖𝑜𝑢𝑠𝑛𝑒𝑠𝑠 = 1
𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦 𝑟𝑎𝑡𝑒 (3.8)
When treatment was provided with praziquantel, the fraction ε of infectious individuals recovered
and remained immune for some time. Waning immunity will remove them with the rate γ to be re-
infected again in compartment S2. The fraction ε estimated from the model reflected the
effectiveness of the praziquantel treatment.
The period of immunity was assumed to be longer than the S2 period. However, the data on
protective immunity was limited. Based on available data, the immunity was assumed to last 10
years (125-127).
Birth rate; b and mortality rate; m, are also incorporated to the model in order to explore the long-
term infection dynamics.
The study emphasized the epidemiological view of the infection dynamics regarding human factors,
so the result mainly accounted for disease distribution. The study outcome was reported as
prevalence using Equation 3.9;
Prevalence = 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑒𝑥𝑖𝑠𝑡𝑖𝑛𝑔 𝑐𝑎𝑠𝑒𝑠 𝑜𝑛 𝑎 𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝑑𝑎𝑡𝑒
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑒𝑜𝑝𝑙𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑜𝑛 𝑡ℎ𝑖𝑠 𝑑𝑎𝑡𝑒 (3.9)
126
According to the study, the point prevalence constituted the proportion of infected individuals at
time t determined from Equation 3.10;
Prevalence(t) = 𝐼(𝑡)
𝑁(𝑡) (3.10)
The model was used to calculate the impact of the National Control Program on annual prevalence
from 1984 – 2001 (27, 28) based on estimates of treatment effectiveness from the mathematical
model.
127
3.2.4 Basic reproductive number
Conventionally, the basic reproductive number or R0 was defined as the average number of
secondary infectious individuals resulting from infectious persons and introduced in a totally
susceptible population (112, 113, 116, 117).
In the model incorporated with population demographic data, assuming that the population will
increase over time, the R0 was defined as demonstrated by Equation 3.11;
𝑅0 = 𝛽𝐷 (3.11)
Where β is contact rate and D is duration of infectiousness. Considering my model;
The contact rate is incorporated with 2 components as shown in Figure 3.5;
● Primary infection contact rate (β): obtained from force of infection of primary
susceptible individuals.
● Re-infection contact rate (μβ): force of infection with reduced infectivity (μ)
The duration of infectiousness depends on
● Recovery rate (r)
● Death rate (m)
The R0 for the infection would be as Equation 3.12;
𝑅0 =𝛽 + 𝜇𝛽
𝑟 + 𝑚 (3.12)
Figure STYLEREF 1 \s 3. SEQ Figure \* ARABIC \s 1 5 Model diagram considered the R0
μβI/N
Births (b) Primary
Susceptible
S1(t)
Infectious
I(t)
Secondary Susceptible
S2(t)
βI/N
(1-ε)·r
ε·r Deaths (m)
Deaths (m)
Deaths (m)
Recovered
R(t)
Deaths (m)
γ
Figure 3.5 Model diagram considered the R0
128
3.2.5 Estimation of treatment effectiveness using individual-level data
Figure 3.6 Diagram for estimation of treatment effectiveness using individual-level data
Treatment effectiveness could be conventionally estimated from the fieldwork to give an overview
of the infection dynamics. Because post-treatment follow-up was not immediately assessed after the
baseline study, the treatment effectiveness was calculated from the proportion of positive and
negative cases reported in the follow-up study by assuming the conditions below.
● All positive cases received praziquantel treatment.
● Newly negative cases are cured as a result from treatment.
● Still positive cases come from treatment failure.
● Newly positive cases acquired the infection during follow-up period.
● Still negative cases did not acquire infection during follow-up period.
The eligible studies for assessing treatment effectiveness included the Na-yao study in 2002-04 and
the Na-ngam study in 2012 – 14, where both baseline and follow-up studies were comprehensive
surveys. The assumptions are based on all fieldworks being conducted with the same method and
response rate being fairly equal (70 – 80 %).
As illustrated in Figure 3.6, the epidemiological outcome could be calculated as;
● Total survey was conducted in the follow-up study. However, the study design emphasized
on the follow-up population (C + D). The portion of A + B are the additional population
included in the total survey. I assume that the proportional population structure would
remain the same. Therefore the proportion of A and B are indirectly estimated from
prevalence and incidence measured from fieldworks.
Positive
Still positive (B)
Newly negative (A)
Newly positive (C)
Negative
Still negative (D)
rfgf
Treatment
Follow-up study Baseline study
129
● Incidence of infection is 𝑁𝑒𝑤𝑙𝑦 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 (𝐶)
𝐵𝑎𝑠𝑒𝑙𝑖𝑛𝑒 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 (𝐶+𝐷) , the negative cases from baseline study
will be followed-up. There would be a portion of baseline negative that becomes newly
positive.
● Prevalence of follow-up study will be the sum of positive cases divided by total study
population, 𝐴𝑙𝑙 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 (𝐵+𝐶)
𝑇𝑜𝑡𝑎𝑙 𝑠𝑡𝑢𝑑𝑦 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 (𝐴+𝐵+𝐶+𝐷)
● After baseline survey, positive cases receive treatment and result in follow-up study. Newly
negative cases are assumed to the cured from the treatment,𝑁𝑒𝑤𝑙𝑦 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 (𝐴)
𝐵𝑎𝑠𝑒𝑙𝑖𝑛𝑒 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 (𝐴+𝐵) ,
Treatment effectiveness will be the proportion of newly negative cases and baseline
positive.
● I assume that still positive cases are the result of treatment failure, so it could be roughly
estimated with indirect calculation by, 𝐹𝑜𝑙𝑙𝑜𝑤−𝑢𝑝 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 (𝐵)
𝐵𝑎𝑠𝑒𝑙𝑖𝑛𝑒 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 (𝐴+𝐵), However, the follow-up
positive individuals were based on the total value; the original method was not intended to
include the participants in this portion to be matched for treatment failure rate, and
immediate posttreatment evaluation was not conducted. The calculation was a crude
estimation.
130
3.3 Results
3.3.1 Infection dynamics of OV infection
Table 3.5 Estimation of duration of infectiousness and recovery rate
Study area
Fieldwork data Model estimation
Incidence rate
(/person-years)
(95% CI)
Duration of infectiousness
(years)
(95% CI)
Recovery rate
(/person-years)
(95% CI)
Na-yao 0.221
(0.176 – 0.273)
1.81
(1.21 - 1.84)
0.662
(0.543 – 0.825)
Na-ngam 0.068
(0.047 – 0.095)
1.51
(1.15 - 2.22)
0.595
(0.449 – 0.869)
As seen in Table 3.5, the model shows that the estimated recovery rate is 0.662 /100 person-years
(95% CI: 0.543 – 0.825) for Na-yao data and 0.595/person-years (95% CI: 0.449 – 0.869) for Na-ngam
data, respectively. Therefore, the duration of infectiousness will be 1.81 years (95% CI: 1.21 - 1.84)
for Na-yao and 1.51 years (95% CI: 1.15 - 2.22) for Na-ngam area.
131
As shown in Table 3.5, the result is that duration of infectiousness allows the model to fit the data in
the two different study populations as shown in Table 3.7.
For the two study sites, the values of the duration of infectiousness were consistent with both the
prevalence of infection and the transmission model. The parameters in the transmission model
(other than the duration of infection) were well constrained by other evidence available in the
literature (Table 3.2). Based on the assumption that the force of infection was uniform and the
prevalence was in a stable phase, the duration of infectiousness from both studies was in the range
1.51 – 1.81 years (Table 3.3). The model solution found was consistent with these observations
suggesting that prevalence increased within the first 10 years and then remained stable thereafter
for both studies.
Figure 3.7 Infection dynamics of prevalence in study area
132
3.3.2 Estimation of treatment effectiveness using population data and mathematical model
The serial prevalence data from the Na-yao study (Table 3.4) is appropriate to estimate treatment
effectiveness. The trend of prevalence can be observed through the 9-year study period with initial
prevalence starting at 22.9% (95% CI: 19.3 – 26.5) and assumed to be in a stable phase. Subsequent
prevalence with 95% CI was fitted in the model to estimate treatment effectiveness as a result of the
ε fraction. Incidence was assumed to be uniform at 22.1/100 person-years (95% CI: 17.6 – 27.3).
Using the model solutions found above, the proportion of treated infectious individuals (ε) who
entered a recovered stage was identified. Values between 80.1 – 92.5% for the effectiveness of
treatment gave prevalence patterns that were consistent with the observed data (Figure 3.8).
Figure STYLEREF 1 \s 3. SEQ Figure \* ARABIC \s 1 8 Estimation of treatment efficacy from the model Figure 3.8 Estimation of treatment effectiveness from the model
1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022
Year
133
3.3.3 Estimation of treatment effectiveness using individual-level data
Under the assumption that the overall force of infection was substantially reduced immediately after
community-wide treatment, effectiveness of treatment was calculated for individuals as a
comparison for the calculation using population data and the model.
Table 3.6 Estimation of treatment effectiveness from Na-yao 2002-04 data
Baseline survey Follow-up study
Results
Baseline Prevalence (%, 95% CI)
Follow-up time (years)
Incidence rate (/100 person-
years)
Results
Cumulative Incidence
(%, 95% CI)
Cumulative Proportion (%, 95% CI)
Follow-up Prevalence (%, 95% CI)
Positive 22.9
(19.3-26.5) 1.19
New negative 20.0
(16.5-23.4)
Still positive 2.9
(1.4-4.3) 20.7 (17.3-24.2)
Negative 77.1
(73.5-80.7) 1.19
22.1 (17.6-27.3)
New positive 23.1
(19.1-27.7) 17.8
(14.6-21.4)
Still negative 76.9
(72.7-80.9) 59.3
(55.0-63.4)
Table 3.7 Estimation of treatment effectiveness from Na-ngam 2012-14 data
Baseline survey Follow-up study
Results
Baseline Prevalence (%, 95% CI)
Follow-up time (years)
Incidence rate (/100 person-
years)
Results
Cumulative Incidence
(%, 95% CI)
Cumulative Proportion (%, 95% CI)
Follow-up Prevalence (%, 95% CI)
Positive 8.9
(6.8-10.8) 1.17
New negative 8.2
(6.3-10.1)
Still positive 0.7
(0.1-1.2) 8.1 (6.2-10.0)
Negative 91.1
(89.2-93.1) 1.17
7.2 (5.0-10.0)
New positive 8.1
(6.1-10.2) 7.4
(5.6-9.3)
Still negative 91.9
(89.8-93.9) 83.7
(81.1-86.3)
As reported in Table 3.6 and 3.7, the result estimated treatment effectiveness of praziquantel
obtained from indirect calculation from fieldwork data. Considering the 2 areas by merging their
results, the treatment effectiveness should have ranges between 88.7 – 92.2%. Additionally, the
estimated treatment failure would be 12.7% (95% CI: 7.8 – 15.5) in Na-yao Area and 7.9% (95% CI:
1.6 – 10.6) Na-gnam Area. Tables 3.6 and 3.7 shows the calculated treatment effectiveness for Na-
yao Area equalled 87.3% (95% CI: 84.5 - 92.2) and for Na-ngam Area equalled 92.1 % (95% CI: 88.7 –
98.4).
134
3.3.4 Basic reproductive number; R0
Table 3.8 Basic reproductive number
Study area R0 95% CI
Na-yao 2002 2.74 (2.61 – 2.89)
Na-ngam 2012 2.30 (2.23 – 2.40)
The relationship was investigated between the basic reproductive number and the duration of
infectiousness, within the content of previous results. R0 was estimated to be 2.74 (95% CI: 2.61 –
2.89) for Na-yao area and 2.30 (95% CI: 2.23 – 2.40) for Na-ngam area, assuming that the duration of
infectiousness was a fixed ranged between 1.5 and 1.8.
This range was much lower than the range of values of R0 consistent with the duration of
infectiousness from the reviewed literature (9, 13, 45, 70, 128). Due to various durations of
infectiousness from 1 to 10 years, the R0 slightly increased from the shorter duration of
infectiousness and then rapidly rose after 8 years especially in the Na-yao study area. For Na-yao
Area, the R0 ranged from 2.52 (95% CI: 2.44 – 2.62) to 6.10 (95% CI: 5.30 – 7.30) and in Na-ngam
Area ranged from 2.21 (95% CI: 2.16 – 2.27) to 3.37 (95% CI: 2.98 – 3.86). The functional form that
showed this dependency can be seen in Figure 3.9.
Figure STYLEREF 1 \s 3. SEQ Figure \* ARABIC \s 1 9 Basic reproductive number with various duration of infectiousness
0
1
2
3
4
5
6
7
8
1 2 3 4 5 6 7 8 9 10
R0
Infectious time (years)
R0 with infection dynamics
Na-yao incidence 95% CI of incidence Na-ngam incidence 95% CI of incidence
Figure 3.9 Basic reproductive number with various duration of infectiousness Figure STYLEREF 1 \s 3. SEQ Figure \* ARABIC \s 1 9 Basic reproductive number with various duration of infectiousness
135
3.3.5 Estimating the effectiveness of the National Control Program using the model
The prevalence of OV infection was plotted against study years from 1987 to 2001 during the
National Control Program implementation presented in Figure 3.10. The blue shaded area
represents the calculated prevalence from the approximate range of effectiveness.
The prevalence was 63.6% at the beginning of the program. After praziquantel treatment was
provided, prevalence decreased to approximately 10% within 5 years and remained stable after
1992. Assuming the parameters setting was similar to current fieldwork and incidence was uniform
through time, the incidence was estimated from the model fitted to the prevalence data.
The estimate of treatment effectiveness was used to calculate the likely impact of the National
Control Program in reducing prevalence. The model could fit the data to estimate incidence rate.
From Figure 3.10, model fitting indicated that the incidence rate should range from 9.8 – 47.3/100
person-years considering 80 – 100% treatment effectiveness. Estimation of treatment effectiveness
was assessed though the fitted model with fieldwork data (Section 3.3.2) and indirect calculation
from individual-level data (Section 3.3.3). The approximate range of effectiveness was 80 – 100%
(maximum and minimum of shaded area shown in Figure 3.10).
Figure STYLEREF 1 \s 3. SEQ Figure \* ARABIC \s 1 10 Fitting model with prevalence from National Control Program.
Figure 3.10 Fitting model with prevalence from National Control Program.
136
3.4 Discussion
In this chapter, the duration of infectiousness was determined from model fitting. The point
estimates of the duration of infectiousness were 1.51 years (95% CI: 1.15 - 2.22) to 1.81 years (95%
CI: 1.21 - 1.84). Two methods were used to calculate the effectiveness of Praziquantel treatment.
From the mathematical model, the effectiveness was estimated to be 80.1 – 92.5%. The
effectiveness for individual-level obtained from indirect calculation from fieldwork data ranged from
88.7 – 92.2%. R0 was estimated to be 2.30 (95% CI: 2.23 – 2.40) to 2.74 (95% CI: 2.61 – 2.89)
assuming that the duration of infectiousness was a fixed ranged between 1.51 and 1.81. The
approximate range of treatment effectiveness was used to calculate the likely impact of the National
Control Program in reducing prevalence. The model could fit the data to estimate incidence rate
ranging from 9.8 – 47.3/100 person-years.
3.4.1 Infection dynamics and study limitation
The major concerns regarding model analysis are presented and discussed here. The method of
model fitting was the fine tuning of one free parameter to visually find the optimal value of another.
The analysis could have been more robust by using the maximum likelihood allowing the robust
confidence intervals to be defined. For further work, future models will likely to incorporate more
unknown parameters. However, because only one free parameter was tuned, the difference in point
estimates between the results presented here and more statistically robust results should be small.
The infection dynamics of this system are more complicated than presented here. The actual
biological life cycle of OV is complicated involving multiple intermediate hosts including snails and
cyprinoid fish. The fish themselves contain more than 10 species and all are edible. Besides that,
many animal species also serve as a reservoir hosts (2, 3, 34, 44, 45).
The force of infection was assumed to be constant. From the model, the force of infection value was
fixed to the incidence rate obtained from the fieldwork. Therefore, the contact rate (β) varied over
time. The contact rate itself posed some interesting issues. By definition, the contact rate is the
interaction between infectious and susceptible individuals. OV infection is not transmitted directly
from person to person. The infection is transmitted through the parasite life cycle within the host
body and external environment. Therefore, susceptible individuals indirectly transmit the infection
from infectious individuals by infective larvae; metacercariae, found in uncooked cyprinoid fish
undergoing multiple steps of growth and development. Little is known about the contact rate for OV
infection. When compared with other trematodes such as Schistosomiasis, the mode of infection still
differs much (94, 121, 123).
137
Concerning the relationship within force of infection components, the interaction between each
compartment, the contact rate and the population dynamics, it would be more realistic to think
about the force of infection as a dynamic quantity. The input parameters were assumed to be
constant over time including birth and death rates. The death rate should vary according to the host
characteristics for each compartment. For example, chronic infection among individuals might suffer
from pathological changes in their hepatobiliary system resulting in higher mortality rates than those
of uninfected individuals (113).
Migration may have affected the transmission dynamics. The population pyramid structure shows
the constricted pattern referring to the active population movement in the early to middle-age
classes. The main reason obtained from the qualitative study revealed that temporary immigration
occurs during off season. This setting could potentially have affected the infection dynamics as a
result from the remaining proportion of the current population and might not truly reflect the
population structure. As mentioned earlier, this would also result in an irregular age structure from
statistical analysis.
The incidence obtained from the fieldwork was derived from the follow-up population, so only
negative cases were followed up for incidence. The true incidence might have been underestimated
because the high-risk population with infection were excluded from the follow-up population. The
re-infection rate could be calculated using the indirect method from the data but would not be
accurate as direct incidence from the study. As result, fieldwork incidence was used as a reference
for the summary of force of infection.
3.4.2 Basic reproductive number
Parasite Study location R0
Onchocerca volvulus Sudan 5 – 35
Ascaris lumbricoides Iran 4 – 5
Necator americanus India 2 – 3
Schistosoma mansoni Brazil 1 – 2
Opisthorchis viverrini Thailand 2.21 – 2.52
The table is adapted from Roy M Anderson et al. (99)
From epidemiological data, the R0 from Na-yao was 2.74 (95% CI; 2.61 – 2.89) and for Na-ngam was
2.30 (95% CI; 2.23 – 2.40), compared with R0 from other flukes; S. mansoni was 1 - 2 and O. volvulus
was 5 - 35.
138
3.4.3 Treatment effectiveness
This was the very first model attempting to assess the treatment effectiveness regarding the
infection dynamics of OV infection. As stated earlier, the National Control Program has been applied
to reduce the infection from 1984 – 2001 covering up to 40 provinces in Thailand with praziquantel
regimen (26-28, 48). The prevalence decreased from 63.6% from 1984-87 to 9.4% in 2001. Then the
program was switched to a passive strategy where the control program was no longer a national
scale.
The model agreed with various treatment efficacies based on the incidence of the infection. The
fitting incidence was assumed to be uniform and ranged from 9.8 – 47.3/100 person-years. As study
areas dated after the National Control scheme, the result might reflect some similarity of
epidemiological pattern of the infection between my fieldwork and the National data.
However, the incidence should be higher considering the initial prevalence started at 63.6% in 1987
and became endemic. In other surveys, the prevalence and incidence could reach up to 100% (40,
83). Hence, the estimated incidence was rather an average rate to fit the overall prevalence trend.
Clearly, the incidence would vary according to the transmission factors as well. The model still needs
a more flexible approach on parameters such as demographic factors or updated data on
transmission dynamics.
Either pre-national deworming field trials or National Control Program focused on the cure from
infection. The reduction of worm intensity was not taken into account because it did not reflect the
achievement of infection control. Moreover, that the infection intensity decreased while diagnosis
was still positive is not practically understandable; which might be reasonable for public perceptions.
3.4.4 Treatment failure
The estimated treatment failure rate of Na-yao Area was 12.7% (7.8 – 15.5, 95% CI) and Na-ngam
Area was 7.9% (1.6 – 10.6, 95% CI) probably consistent with estimated treatment effectiveness
(Table 3.9).
Originally, the treatment failure rate was not directly evaluated because only negative cases from
the baseline survey were followed in the subsequent follow-up study. The remaining participants in
the follow-up comprehensive survey did not completely match the original population from the
baseline survey. However, the treatment failure rate could be calculated by indirect method from
the fieldwork data assuming that all baseline positive cases received Praziquantel treatment and
treatment failure cases remained infected until the follow-up study. Therefore, the obtained
139
treatment failure rate stated here is a crude estimation by indirect method and still far less accurate.
Further fieldwork studies should include immediate posttreatment evaluation when treatment
failure needs to be assessed.
Helminths are multicellular and complex-structure organisms. Their size is relatively much larger
than microorganisms such as bacteria or viruses. As a result, they require effective defence
mechanisms against immune systems to evade and remain viable within their host body. Assuming
this, helminth infection is more likely a constant state where the host cannot produce adequate
immunity against re-infection or further infection. Thus, the infection could be accumulated within a
host producing a temporal relationship with respect to time-dependence.
The phenomenon does occur where the prevalence increases in the aging population as seen in
many helminth infections. In terms of formulating a mathematical model, the age pattern could
shape a more precise structure to a model considering differing levels of risk population.
However, this model did not consider transmission dynamics within parasite biology. Greater insight
regarding transmission dynamics could potentially contribute various control strategy choices such
as the role of intermediate host control or vaccination.
Much fieldwork on chemotherapy and mathematically-statistical analysis favour treatment
strategies directed at target populations (28, 40, 83, 105). Potential risk behaviours and high-risk
populations could be identified. Therefore, the population was less likely a random mix. Infection
intensity is more likely to be associated with the intensity of uncooked fish consumption practice
particularly local dishes such as Koi pla (70, 83). Moreover, qualitative aspects of OV infection
indicate that the consumption pattern is strongly related to demographic influence such as
agriculture-related jobs or alcohol-related social events (70, 114). Therefore, the contact pattern
regarding demographic heterogeneity should be concerned.
140
3.5 Conclusion
Infection is a complex phenomenon involving multiple factors including infectious agents, hosts and
the environment the so-called epidemiologic triangle. Understanding infection dynamics is crucial as
it provides the strategy to battle the infectious disease; both treatment and prevention.
Mathematical modelling has become an effective tool to explore these complicated relationships. It
could visualize the flow of infection dynamics. Some infectious diseases have been well-investigated
such as schistosomiasis and malaria, leading to contributions of ongoing interventions with huge
efforts.
OV infection caused by Opisthorchis viverrini is considered an NTD, and prevalent within Southeast
Asian, where most countries are still developing. The limited resources result in difficulties and a
constant struggle to sustain prevention and control. Ultimately, OV infection can cause a lethal
malignancy: cholangiocarcinoma. Hence, the prominence of this infection should drive the health
community to expand treatment and prevention measures while ensuring resources are used under
effective strategies.
Mathematical modelling is among the many useful tools to extend the knowledge of infection
dynamics. The data from the study project with additional data from previous works are able to
formulate effective models and provide greater insight about how each compartment can be
related.
141
Chapter 4 Evaluation of risk behaviours for OV infection as
determined by a qualitative approach
Abstract
Central Thailand was considered a non-endemic area for OV infection. However, my fieldwork
revealed that the prevalence and incidence were relatively high compared with the regional average
data. I have hypothesized that the behavioural-psycho-social background of the study population
played an important role in the high burden of the infection which quantitative approach may not
fully describe such characteristics. As a result, the qualitative approach was used to highlight the
potential of social components affecting the infection dynamics to gain more understanding of the
risk behaviours and their contexts.
The qualitative approach including focus group discussion and in-depth interview was employed in
Na-ngam area in 2012. Study participants were allocated to each group based on their stool
examination result from baseline and follow-up study reflecting the current infection status as
never-infected, previously-infected, newly-infected and re-infected within 14 months of study
period. Local heath volunteers were also assessed as to their role in the community regarding the
infection. Discussion themes were directed by predefined topics regarding the current quantitative
data and reviewed literature to develop greater understanding of OV infection and the associated
health risks. Framework analysis was mainly used to explore the association between factors from
thematic content extracted from gathered data. Additionally, grounded theory was used when new
ideas were generated from the analysis.
The results showed that Koi pla was still a popular dish in the community as the dish itself
represented Northeastern culture. The cultural norm had been transferred from the ancestors to
their descendants; the evidence could be visually observed that they still preserved their lifestyle
despite the fact that they had already moved out from the northeast around 40 years ago.
Population dynamics also influenced the social aspects of OV infection. Some elders complained that
discontinuing the consumption of Koi pla went against old traditions with respect to cultural norms
and socialization. Higher education teaches about hygienic living conditions including OV infection
transmission, so teenagers and young adults tended to modify their lifestyle including eating habits.
Therefore, health education only might not be practical to motivate the behavioural change. More
efforts are needed to support the transformation. Children are a potential key to pass knowledge to
142
their parents and school-based education programs can serve as a practical hub for knowledge
distribution.
For rural settings, community leaders function as an important key in the social structure. As the
local administration is managed by the central and regional authority, governmental staff might not
be aware of local cultural norms. Community leaders could bridge this gap and help facilitate the
governmental affairs. Local health volunteers play an important role for the communication and
interaction among villagers and health authorities. They could act as community leaders and as role
models for health interventions. Because they are locals, they could more easily gain trust and
cooperation from the villagers that are a crucial part in maintaining the sustainability of health
campaigns in the community.
Social influence showed a strong impact on infection dynamics of OV infection. Within the social
component, the interaction of knowledge and culture transfer across generations provides more
insight to bio-psycho-social aspects.
143
4.1 Introduction
Health is becoming a subject of interest for social science disciplines (129). From the quantitative
research perspective, the questions for social aspect concerning health have become more
important to understand the underlying process of human behaviours associated with disease.
In terms of infectious disease, 3 main factors contribute to the infection: host, agent and
environment (130). From Figure 4.1, Chapters 2 and 3 described the infection dynamics regarding
the quantitative approach including statistical and mathematical modelling; the results quantified
the size of the problem such as prevalence and incidence of the infection or what was the risk of the
infection. For my study, age and Koi pla consumption were significant risk factors for acquiring OV
infection.
144
Consumption of Koi pla is a direct mode of transmission of OV infection (2). Therefore, it is directly
involved with human behaviour. However, understanding the behaviour may require more tools to
more completely understand the underlying causes.
A study in a Northeastern population (4) revealed that knowledge toward OV infection was at a good
level (80 %). However, good knowledge did not reflect safe behaviours regarding consumption of
uncooked fish. The pattern of consumption was also observed (70, 114, 131) for uncooked fish
consumption; male was the predominant sex for Koi pla consumption and it was associated with
alcoholic drinking. The change from active to passive strategies to prevent OV infection created
some issues such as the role of primary prevention (28, 132). It was challenging to fill the missing gap
with qualitative tools for a more comprehensive result.
My fieldwork was conducted in Central Thailand, which was considered a non-endemic area (11, 69).
However, the prevalence and incidence found from the study were relatively high compared with
regional average data (Sections 2.3.3.1 and 2.3.3.2 from Chapter 2). I have hypothesized that the
Risk behaviours
Social influence
Opisthorchiasis
Intervention
Prevention Diagnosis and Treatment
Figure 4.1 Conceptual framework for qualitative approach
145
behavioural-psycho-social background of the study population played an important role involving a
high burden of infection.
The quantitative approach may not fully describe such characteristics regarding the behavioural-
psycho-social background. Therefore, the qualitative approach highlighted the potential of the social
component, which has played a major role in infection dynamics to gain more understanding with
regard to the “how” and “why” (129) of the concerning issues. The contribution of the social
component could shape the risk behaviours and also influence the underlying process of
intervention including prevention and control.
This Chapter aims to describe the social influence on infection dynamics with regard to the stated
questions below.
Why OV infection is still a burden in the area, and how does the community react to the
problem?
How do knowledge, attitudes and perceptions interact with OV infection and related risk
behaviours?
What is the interaction between OV infection and social components from the conceptual
framework of OV infection (Figure 4.1)?
What is the impact of the OV infection from individual to community perspective?
146
4.2 Methodology
The study design comprised 2 approaches including group and in-depth interviews.
4.2.1 Group interview
In the qualitative study, group interviews allowed researchers to investigate the study topic from
more than one perspective. In the social sciences and urban planning, group interview allows
researchers to study people in a more natural conversation pattern than typically occurs in a one-to-
one interview. In combination with participant observation, they can be used for learning about
groups and their patterns of interaction. An advantage is their fairly low cost compared with surveys,
as one can get results relatively quickly and increase the sample size of a report by talking with
several people at once.
Group interviews could be classified in subtypes depending on particular features. However, the
fundamental concepts are based on gathering data from groups of people. Characteristics of
participants and the aims of data implication would define its subtypes including consensus panels,
focus group discussions, natural group discussion and community interviews (129).
Focus group discussion was the selected technique used in this study. The pattern of the focus group
discussion was typically more formal than the general meeting with a more structural pattern. The
participants were enrolled to meet the sampling criteria. They would discuss and provide opinions
under the facilitator’s direction with predefined topics. However, the participants were allowed to
express their views freely to seek a broad range of ideas on open-ended topics. Meeting places and
times were usually scheduled. The conversation was stored in a recording device for later decoding
and analysis.
Focus group discussions usually contained 8 – 12 participants to ensure the coverage of opinion
expression. When fewer members attended, the meeting format would be replaced with small
group interviews.
4.2.2 In-depth interview
In-depth interviews are face-to-face conversations between the researcher and participants.
Generally, interview is the most common form of data collection in both quantitative and qualitative
design. In this study, the in-depth interview was used to assess some interesting points that had
emerged during focus group discussions or deviant cases reflecting alternative ideas from the
discussed issues.
147
4.2.3 Study population
Qualitative study was conducted in the area of Na-ngam Area from 2012 – 14. Study participants
were enrolled from the quantitative study where demographic data, questionnaire data and stool
examination results were collected (Chapter 2).
Research participants were invited to a group conference or interview in a focus group discussion.
The selection method was stratified purposive sampling. The enrolled participants were categorized
in 5 groups based on stool examination results for OV infection. Demographic characteristics were
considered the selection criteria based on homogeneity of the study population to avoid influential
members leading or dominating the conference. However, the enrolment process was flexible and
preliminary interview was given to demonstrate attitudes and opinions of the invited members.
Table 4.1 Selection criteria for focus group discussion
1 2 3 4 5
Never-infected Previously-infected Newly-infected Re-infected Local heath
volunteers Baseline study 2012 Negative Positive Negative Positive
Follow-up study 2014 Negative Negative Positive Positive
As shown in Table 4.1, the main criteria was based on stool examination results showing infection
dynamics during the follow-up period reflecting the corresponding factors contributing to the
infection. For the re-infected group, infected cases from baseline study were treated with
praziquantel and assumed to be cured. If their follow-up stool examination result was positive, they
were assumed to be re-infected during the follow-up period.
The local health volunteers who acted as fieldwork collaborators were categorized in separate
groups to avoid them dominating the discussions.
For some unique cases, in-depth interviews were arranged for a specific case with interesting issues.
148
4.2.4 Study process
4.2.4.1 Focus group discussion
1. The stool examination result was analysed for classification criteria with respect to study results
in 2012 and 2014.
2. The method of selection was purposive sampling. Joining the study was voluntary and written
informed consent was given by the enrolled participants. The consent also included the
permission for voice and image recording during the conversation.
3. Each participant was informed about their group, time and venue for the group discussion. They
were notified again shortly before the appointment by any means of communication such as
mobile phone or personal reminder.
4. On the conference day, an ice-breaking exercise was performed at the beginning of the meeting.
The moderator would introduce him/herself and let each of the participants speak as an
introduction to get to know each other and establish a friendly environment. Then participants
were informed about the objectives and the process of the discussion.
5. The moderator managed the conversation by asking focused questions to meet the study
objectives and maintain the flow of opinions expressed. The moderator maintained the structure
of the conference while continuing the session in a relaxed atmosphere; the participants
received equal opportunity to freely express their opinions.
6. Discussions focused on various aspects relating to OV infection and its risk factors. The flow of
discussion followed the main topic themes listed below.
Knowledge
o Basic knowledge of the infection; liver fluke life cycle, mode of transmission and
infection, risk factors, diagnosis and treatment
o Health-related consequences from the infection; role of carcinogens,
cholangiocarcinoma
Attitude
o Perception of OV infection and its consequences
o Health concerns of the risk factors and the infection
Uncooked fish consumption behaviours
o Situation of uncooked fish consumption in the community
o Current pattern of consumption behaviours regarding social aspects of the
community
Impact of the infection on the community
o Health and social impacts from the individual to community levels
149
Solutions
o Roles of the National Control Program from the community perspective
o Concerns regarding treatment and control
o Roles of primary prevention
o Accessibility to healthcare services
7. The moderator used open-end questions to contribute various expressions from the members.
Closed-end questions were also used to clarify some interesting points.
8. The conversations were recorded using a voice recorder. The data collection proceeded
continuously until the discussions were saturated when nothing new was being generated from
the meeting. It usually took 30 – 45 minutes for each session.
4.2.4.2 In-depth interview
Some members were invited to face-to-face interviews. The conversations focused on specific points
of interest raised during the focus group discussions.
4.2.5 Data processing and analysis
Data processing and analysis included multiple approaches to interpret physical conversations in
systematic content, exploring and explaining the situation of OV infection in qualitative aspects.
Table 4.2 Analysis approach to qualitative data
Analysis approach Aim
Thematic content analysis To identify and organize themes from data
Framework analysis To explore content within data framework
Grounded theory To develop theory from data
The methods used included direct quotations and selected words to consider actual local words used
by the participants. Table 4.2 summarizes the analysis approaches to the collected and processed
data as described below.
150
4.2.5.1 Thematic content analysis
The recorded conversations from each session were transcribed to text using a word processing
program. Then text-based data were sorted and coded according to categories or common themes.
This first step helped rearranging unorganized conversation data in systemic form of thematic data.
The guidelines from the discussed topics are indicated below.
Knowledge of and attitude towards the infection
Risk behaviours
Impact on the community
Solutions
In addition, relevant themes emerging from the analysis were taken into account.
4.2.5.2 Framework analysis
The main framework followed the OV infection model from Figure 4.1 highlighting the broad
interaction of social influence to other compartments.
Thematic contents were categorized in a systematic chart and used to analyse of the thematic
framework. The thematic framework was elaborated and expanded from the social influence
compartment. The analysis included exploring and comparing data within this framework to explain
the interaction between each factor regarding thematic contents.
4.2.5.3 Grounded theory
This method was used when new ideas were encountered from the analysis. In addition to analysing
contents within the framework in Section 4.2.3.2, new theories could be obtained from the data.
The qualitative analysis aimed to achieve the outcomes listed below.
1. Contributing conceptual definition from discussed topics.
2. Developing typologies and classifications.
3. Explaining the situation of the infection with consideration of bio-psycho-social aspects.
4. Exploring the association between attitudes, health behaviours and social influence of OV
infection and its risk factors.
5. Generating new theories or ideas regarding the infection.
151
Since the discussion theme was directed by predefined topics regarding the current quantitative
data and reviewed literature to develop an understanding of OV infection and the health risks,
framework analysis was mainly used to explore the associations between factors. From the thematic
content extracted from the gathered data, the analysing steps included those listed below.
Data familiarization: data was carefully read and transcript to computer-based text.
Thematic analysis: the thematic content analysis was preliminarily used to identify main
topics; repetitive quotes, expression or statements were also summarized by frequency.
Saturation was observed when little or no change was found in the codebook indicating
that the thematic topics were extracted and well demonstrated.
Indexing: data was revised, organized and summarized for analysis. Each theme was
extracted from the content then coded and labelled by category.
Charting: Summarized thematic data was rearranged to visualize the whole picture of the
contents. The chart summary allowed the researcher to look through the range of data to
sort, compare and organize the content systematically.
Mapping and interpretation: diagrams and tables were used to explore the relationship
between concepts and typologies developed from the contents.
Additionally, grounded theory was a helping tool to explore new ideas and theories emerging from
the contents. The method employed the deductive approach where data were moved back and forth
between theory and data content.
Firstly, data was broken into smaller pieces of information or data fractures. Each entry was marked
as open coding and intensively analysed line-by-line to explore the insight meaning to develop initial
concepts of what was occurring from the situation.
Initial concepts provided further steps of analysis. Open coding can produce a list of concepts and
then be reorganized into similar phenomena to explore properties and dimensions.
Properties comprised the characteristics or attributes of the phenomena and dimensions were the
continua along which properties could be arranged. The method attempted to explain the current
situation or reasoning the phenomena based on the generated ideas.
152
The next step included axial coding; fracture data were classified in categories to develop a coding
paradigm entailing a set of questions about each code as described below.
Condition: What conditions give rise to the categories?
Context: What was the context of the issue?
Interactional strategies: How to manage to interact or handle the phenomena?
Consequences: What were the consequences from the strategy?
Finally, selective coding was used to analyse and compare to explore the most relevant concepts or
core categories to explain most of what was occurring.
153
4.3 Results
4.3.1 Population characteristics
The qualitative study was conducted in Na-ngam. Details of the study population and community
background are described in Sections 1.6.1 and 1.6.2.
Table 4.3 Age structure classified by generations of Na-ngam 2012 prevalence data
Age group N %
Generation Years
Children 0 – 12 47 10.2
Teenagers 13 – 17 16 3.5
Adults 18 - 60 305 65.9
Elderly ≥ 60 95 20.5
According to the Institute for Population and Social Research (IPSR), Mahidol University and the
National Statistical Office of Thailand, age-class can be classified in 4 major generations based on
bio-psycho-social aspects. From Figure 4.2, the population pyramid showed a constrictive pattern
where age-class 10 – 29 years was relatively low compared with other classes. Table 4.3 indicates
that majority of the population comprised adults. It was observed that the elderly population was
about 20% of the total proportion. Referring to Table 2.5 from Chapter 2, the proportion of females
and males was nearly equal (54.9% and 45.1%, respectively). Ages ranged from 2 – 87 years. Mean
age was 43.3 ± 19.9 and median age was 47.0 ± 23.
Figure 4.2 Population pyramid of Na-ngam 2012 prevalence data
154
Table 4.4 Characteristics of focus group discussion participants
Characteristic Discussion group
Local health volunteer Never-infected Newly-infected Previously-infected
No. of members 12 9 7 7
Gender
Male
Female
1 (8.3 %)
11 (91.7%)
1 (11.1 %)
8 (88.9 %)
5 (71.4 %)
2 (28.6 % )
1 (14.3 %)
6 (85.7 %)
Mean age (years) 39.8 49.8 62.8 56.0
Occupation
Agricultural work
Agricultural work
Self-employed
Student
Agricultural work
Self-employed
Agricultural work
Self-employed
Educational level Primary school Primary school Primary school Primary school
Average income
(baht/month) 12000 5000 5000 5000
The sessions were categorized in 4 groups; local health volunteers, never-infected, newly-infected
and previously-infected groups (Table 4.4). Participants in the re-infected group were too few to
form a group discussion, so they were allocated to in-depth interviews only. From a total of 35
members; 8 were male (22.9%) and 27 were female (77.1%). Females dominated in all groups except
the newly-infected group. Overall mean age of participants was 51.1 years. Most of them were
agriculture-related workers. The most common educational level was primary school and average
income was approximately 6750 baht/month (135 GBP).
155
Table 4.5 Characteristic of in-depth interview participants
Characteristic In-depth interview group
Re-infected Previously-infected Newly-infected
No. of participants 3 2 1
Gender
Male
Female
1 (33.3 %)
2 (66.7 %)
2 (100 %)
0
0
1 (100 %)
Mean age (years) 59.7 38.5 57.0
Occupation
Agricultural work
Self-employed
Unemployed
Unemployed
Student
Self-employed
Educational level Primary school Primary school Primary school
Average income
(baht/month) 8300 2500 4000
Some members who made interesting contributions to the group discussions were invited to join in-
depth interviews to provide more details of the unique issues. Re-infected cases were also invited to
be interviewed to provide information about the recurrence of the disease. As shown in Table 4.5, 6
members enrolled for the interview reorganized in 3 groups, i.e., Re-infected, control and newly-
infected groups. Males and females were equal in proportion. However, previously-infected and
newly-infected groups comprised only one sex. Total mean age was 51.7 years. The major
occupation was self-employed. The main educational level was primary school and average income
was 4933 baht/month (100 GBP).
156
4.3.2 Thematic contents
4.3.2.1 Situation of the OV infection in the community
Table 4.6 Situation of the OV infection in the community
Themes Concerning topics Emerging issues
Acknowledgement of the
infection
Uncooked fish consumption
was identified as health risk
Cultural value and social
influence of uncooked fish
consumption
Misunderstanding of
uncooked fish menus
Local health volunteers were
regarded as respected
members of community
Health campaign based on the National Control Program was provided to the community including
health education, diagnosis and treatment by the collaboration between research team and local
health volunteers.
Local health volunteers played an important role to communicate and interact among villagers and
local health authorities. They gained respect from the villagers as important key members of the
community.
As shown in Table 4.6, most participants acknowledged that OV infection was still a health burden to
the community. They perceived that the infection was strongly associated with consumption
behaviours and chronic infection could result in serious outcome. However, misunderstanding and
misconception still remained in some villagers. The potential of social and cultural influence also
impacted on attitudes and perceptions of the infection. High-risk populations and risk factors
continued to be identified. Moreover, the prevalence and incidence of OV infection were still
relatively high compared with the baseline population.
Most villagers knew that some of their popular fish dishes were uncooked. Some were still confused
that all uncooked dished could cause OV infection while some of them knew that only fresh fish
material could lead to infection.
Koi pla (chopped raw fish salad) is a popular dish that has been consistently identified as a major risk
factor of the infection. Although the community learned that it could lead to infection, the dish was
still valued as a Northeastern culture norm. Daily consumption might have decreased; Koi pla
consumption was still popular during festive events, friend or family meeting and male social
drinking.
157
Older age showed strong attachment to the traditional value of Northeastern culture; “Most of the
elders still eat raw fish; they came from the Northeastern region” said one 48-year-old woman from
the newly-infected group mentioning the strong attachment between the elders and their ancestors
who originally lived in Northeastern Thailand and later migrated to the current area.
4.3.2.2 Current situation of consumption behaviours
Table 4.7 Current situation of consumption behaviours
Themes Concerning topics Emerging issues
Popularity of uncooked fish
consumption
Eating pattern among
generation
Subpopulations were
identified for risk behaviours
Burden in behavioural
modification
Some uncooked fish menus
were the way of life
Eating habit as family
tradition
The current situation of consumption behaviours is shown in Table 4.7; adult participants
acknowledged that instantly-prepared uncooked fish could cause the infection. Some of them knew
that Koi pla dish was a risk factor, so they avoided eating Koi pla but sought a substitute dish. Some
of them perceived that all kinds of uncooked fish dishes led to infection, so they chose to continue
their eating behaviours unchanged because modifying a familiar lifestyle was too complicated.
Moreover, some dishes represented a traditional value, especially Koi pla, which they always
consumed during festive events. Some elderly members stated that,
“I’ve been living with this food for a long time; it’s a part of my life. It’s valuable. But children don’t
eat it anymore”
“When I have family meetings, if they serve you (Koi pla) you should eat it, or else you are insulting
them” said a 51-year-old woman from the newly-infected group.
Although uncooked fish dishes varied in this community, Koi pla and Pla ra were the most popular
dishes villagers frequently consumed in both sexes and all age groups.
“Everybody eats raw fish; it’s very rare to see someone who doesn’t” said one member of the re-
infected group. However, consumption of Koi pla decreased in females and at younger ages, “As I
recognize, it’s once a year, twice as a maximum.” said one 40-year-old woman describing having Koi
pla.
158
Koi pla was shown to be strongly related to multiple aspects as their way of life. It was considered a
food frequently eaten with alcoholic drinks especially in male social drinking.
“Koi pla was served in a drinking party among male friends. Koi pla was the best when paired with
alcohol”, one 70-year-old man from the newly-infected group mentioned about the relationship
between alcohol drinking and uncooked fish eating habit. “It’s a socialization process,” said one 61-
year-old from the newly-infected group.
“If I have to quit Koi pla, I have to quit alcohol, which is a highly unlikely thing for me to want to do,”
said one 48-year-old man from the re-infected group.
Some of them stated that eating Koi pla was related to agriculture-related occupations: “Farmers
always eat Koi pla, fish is easily caught from the river when they go to a paddy field” said one 40-
year-old woman from the never-infected group.
On the contrary, Pla ra consumption was popular at all ages and both sexes. Pla ra was used as a
seasoning, food ingredient or even consumed as a main dish. The roles of Pla ra are various. When
mentioning uncooked fish dishes, Pla ra would be the very first dish they recognized. “I can’t live
without Pla ra,” said one 48-year-old woman from the previously-infected group.
Some families whose parents still consumed Koi pla kept their children away from it unless cooked.
Moreover, if children and teenagers consumed Koi pla, they were always from families with
uncooked fish eating habits. “I want to eat it,” one 7-year-old boy mentioned about Koi pla, who had
parents who regularly consumed Koi pla and he was a newly-infected case.
Some villagers stated that uncooked fish consumption was an old tradition. It might be difficult to
avoid or stop eating due to the strong cultural attachment. However, it began to gradually fade out
from the younger generation as one 64-year-old women from the never-infected group said, “They
don’t even recognize Pla ra; they just enjoy hotdogs.”
The teenage group said that they did not like Koi pla. They thought the preparation method and food
appearance looked disgusting and unacceptable. Moreover, the fishy taste did not match their more
urbanized lifestyle. Younger generations loved to eat instantly prepared foods such as instant
noodles or ready-to-eat meals. Some thought it was fashionable to eat what was advertised on the
television.
“Teenagers do not have much uncooked food. They enjoy instant noodles to anything easy to prepare
such as fried eggs”
159
“They like cooked food, whatever kind of preparation; fried, grilled or roasted”
Health education provided more insights about the threat of uncooked fish consumption. The roles
of local health volunteers and health campaigns influenced the behavioural patterns of
consumption.
“The newer generation never eats raw fish; they were educated from their school,” said one 64-year-
old woman from the never-infected group.
4.3.2.3 Knowledge of the infection
Table 4.8 Knowledge of the infection
Themes Concerning topics Emerging issues
Causes of parasitic infection
come from many routes
Individual parasite needs its
own mode of infection
Inaccurate knowledge leads
to risk behaviours
Lack of visible evidence of
parasite affect awareness
Misinterpretation
Information overload
Healthy lifestyle does not
guarantee healthy life
Table 4.8 shows that participants knew that uncooked Cyprinoid fish contained metacercariae,
which could lead to infection. However, some of them were confused about the life cycle of other
parasites;
“When I go out barefooted, the worms penetrate my skin and go up to infect the liver”
“They go through your nail once you chop fish and leave the utensil unwashed”
Some of them perceived that the mode of infection was through the faecal-oral route, the common
intestinal helminth life cycle.
“It’s not about eating raw fish; it’s about fresh vegetables. Don’t you see eating raw fish was
obviously reduced nowadays but people still become infected? I’m sure it was contaminated with
vegetables,” said one 51-year-old woman from the newly-infected group.
Some villagers perceived that fish from clean water sources was free of metacercariae.
“The water body was clean without aquatic weeds or chemical substances, so the fish was also clean
from parasites,” said one 70-year-old from the newly-infected group.
160
They also knew that Koi pla was a risk factor for OV infection. Most of them understood that
dressing uncooked fish meat with acidic agents such as lime juice or spirits only changed its colour, it
was not completely cooked and metacercariae were still viable.
However, one member from the re-infected group had never heard about OV infection.
“I knew nothing until the doctors came. It’s very usual around here; fish are chopped and mixed with
ingredients, tasty,” one 48-year-old man admitted.
Teenagers showed better knowledge of OV infection. They knew the important part of the life cycle
and were able to explain the mode of infection and transmission. However, most of their knowledge
was acquired from textbooks or internet. They stated that they only knew of the parasite without
experiencing it which might have affected their awareness.
The adult and elder groups recognized that OV infection could lead to bile duct cancer but some
believed it was a result of agricultural insecticide contaminated with fish meat or
immunocompromised host as evidenced below.
“I don’t see any association between the infection and cancer. In fact, chemical toxins do. I see many
people who never smoke, drink or eat raw fish but still die of cancer. I think it depends on each
individual. Of course these stuffs have toxins, but if you were strong enough you’ll be fine. It doesn’t
matter what you take into your body. If your immunity was weak you’ll get cancer even you eat
vegetables,” One 70-year-old-man discussed about the cause of cancer. A 62-year-old man from
newly-infected group also supported that; “Some people have a very healthy lifestyle. Sadly they died
at 40 or 50 years from cancer. It doesn’t make any sense and it’s proven that they’re related”
The villagers acknowledged the transmission process of the parasite. They knew that once stool
contaminated with parasite eggs were released in a natural water source, the eggs hatched and
continued its life cycle.
Some of them mentioned that public excretion could be avoided and in fact most of them used
toilets on a regular basis, but some factors could not prevent disease transmission.
“I saw the toilet pumping service car dumped stool in the river,” one 70-year-old man mentioned
about how the authorities managed the public toilet pumping service.
161
4.3.2.4 Perception of OV infection and its consequences
Table 4.9 Perception of OV infection and its consequences
Themes Concerning topics Emerging issues
Cooked food was considered
hygienic
Doctors were responsible for
peoples’ health
The infection was easy to
threat
Primary prevention was not a
primary concern
Stigmatization from being
infected
School was an information
hub
The villagers perceived that OV infection was the major problem to the community (Table 4.9). They
showed better attitudes towards OV infection. Although some of them might have misunderstood
about the life cycle, they agreed that cooked food was more hygienic and could prevent not only OV
infection but other intestinal parasites as well. Their consumption behaviours might be difficult to
observe in practice, but parents were trying not to let their children consume Koi pla; “Cooked food
was good, vegetables need to be washed as well”
After receiving stool examination results, some villagers decreased their uncooked fish consumption
as they mentioned, “Though I’ve never seen the worms, the result makes me stop eating raw fish”.
Even when the result was negative, some were still concerned about other’s results or they were
afraid that if they chose to maintain risk behaviours. One day the result might turn to be positive, “I
see doctors collect stool and give medicine; I think it’s time to stop eating raw fish. I don’t want to be
one of the infected”
Some were concerned about possible side effects of taking praziquantel. They were afraid that if
they were infected, they would have to take the medication again, “It didn’t feel good after taking
the medicine,” said one 48-year-old from the re-infected group.
The younger generation had better accessibility to information. The school taught them about basic
hygiene. Most households could not afford the internet service and IT infrastructure was still under
development. However, internet was available at school and internet cafés in community centres.
Some of them spent time during break or after class looking for more information about OV
infection, “I saw the pictures of worms on the internet,” said one school-age participant.
As the villagers perceived that OV infection was a disease, they preferred to focus on treatment
rather than prevention. In addition, some of them learned that treatment for the infection was
simple and feasible so they were willing to wait for the healthcare worker as they thought it should
162
be the physician’s responsibility to take care of them; “Only the doctors know the disease, so it was
better to leave it in their hands” said one 70-year-old man.
4.3.2.5 Health concern of risk factors and the infection
Table 4.10 Health concern of risk factors and the infection
Themes Concerning topics Emerging issues
Infection threats the
community
Each generations approach to
disease threat in different
ways
Asymptomatic manifestation
affects health awareness
Health concerns do not
associate with eating habit
The term “liver fluke” was a
misnomer since the liver was
not directly involved
Concern was primarily raised
by the doctors
Table 4.10 reveals the health concerns of risk factors and the infection, high prevalence and
incidence of OV infection triggered concerns about the health consequences to the community.
However, they suggested that the main clinical manifestation was asymptomatic and presented
symptoms were also nonspecific. They understood that OV infection caused abdominal distension,
ascites and jaundice and were actually the symptoms of cholangiocarcinoma or liver diseases. They
also knew that OV infection could be cured by anthelminthic medication. Most of them
acknowledged the health threat of the infection. Because they were physically healthy, health
awareness seemed to decrease due to the disease’s natural history.
All groups expressed that they were asked for stool examination because the doctors (research
team) thought it was important. Practically, nobody ever asked for a stool examination at the local
healthcare facility, “I consider what the doctor concerns”
However, some thought to look after the health was the doctor’s responsibility.
“I’ll do whatever they say. They said I have the parasite, so they gave me a drug.” said one 70-year-
old man.
Most villagers were not sure about the relationship between OV infection and cholangiocarcinoma.
It was noticeable that the villagers also expressed various opinions toward the parasite and infection
with respect to their age; teenage, adult or elderly.
Teenagers and adults knew that chronic infection could lead to cancer, but they did not know exactly
what kind of cancer resulted from OV infection. Moreover, the name “liver fluke” (also called in Thai)
confused the villagers and that the pathology occurred in liver tissue. Some believed that consuming
163
Koi pla with alcohol could accelerate the process of liver cirrhosis. Elderly members thought OV
infection was not related to cholangiocarcinoma. They observed that many were still healthy despite
the fact that they regularly consumed uncooked fish.
Although all age groups realized the medical importance of OV infection, younger aged members
were concerned about live worms living in their body while they were less aware of being sick from
the infection. Male youths started to consume uncooked fish in adolescence to imitate their friends.
Small children might be fed uncooked fish by their parents.
Adults and the elderly paid more attention to infection from the parasite. Lastly, the elderly
members were concerned about both parasite and the infection. However, some admitted that they
could not avoid eating uncooked fish despite knowing the unhealthy consequences.
The villagers stated that uncooked fish consumption needed to be slowly reduced. Abruptly
discontinuing consumption seemed impractical. For Koi pla consumption, quantity and frequency
were decreased over time except in the elder group that stated it was difficult to give up their way of
life. Moreover, some of them mentioned that contributing to a major change was not worthwhile
because they only had a few years left in their life. “I’ve been living so far and I all have to die
anyway, why not just enjoy life while it lasts,” said one 83-year-old woman from the re-infected
group.
164
4.3.2.6 Diagnosis and treatment
Table 4.11 Diagnosis and treatment
Themes Concerning topics Emerging issues
OV infection was curable
Praziquantel was not
available in the local health
facilities
Misunderstanding of
anthelmintic drug
Treatment causes more
symptoms than infection
From Table 4.11, standard treatment with praziquantel was provided by the research team under
the project after the stool was examined. OV infection cases received their diagnosis with
praziquantel and other parasitic infections also received corresponding medication.
When they were aware of being infected, two approaches allowed them to receive medication.
Firstly, they waited for the stool examination result from the research team who regularly visited
them. They would definitely receive praziquantel with this method, “Doctors come very often. If I got
infected, then they’ll give me drugs,” said one 47-year-old woman in the local volunteer group.
Secondly, they also perceived that anthelminthic medication was available at the community
healthcare centre or they could directly purchase over-the-counter medicine from the pharmacy for
their convenience. Unfortunately, those anthelminthic drugs were for intestinal helminths such as
albendazole or mebendazole. Praziquantel was not available in the community pharmacies unless
villagers went to the district hospital which was 1 hour away.
Additionally, some thought that a single drug could cure all parasitic infections including OV
infection. Especially for those who currently consumed uncooked fish, some purchased medicine
from a drugstore and misunderstood that it could cure liver flukes so they returned to resume their
eating behaviours, “It’s simple, just go to the drug store,” Said one 61-year-old from the newly-
infected group. After receiving praziquantel, some felt that the infection was truly eradicated from
their body, “After I took your drug, all worms were killed. So I feel I’m stronger,” said one 60-year-old
woman from the previously-infected group.
However, some who used to take medicine still recognized praziquantel from its side effects, “The
drug was so strong, I vomited until nothing was left in my stomach,” said one 46-year-old from the
local volunteer group.
165
Some stated that a home remedy was available to cure parasitic infection, “From the old days, I
mixed salt with coconut cream to treat parasites”
4.3.2.7 Role of National Control Program
Table 4.12 Role of National Control Program
Themes Concerning topics Emerging issues
National control program
generally improves heath
Only fresh-prepared
uncooked fish can cause the
infection
Avoidance of all uncooked
fish menus was not practical
Primary prevention needs
more specific strategy
As shown in Table 4.12, the participants stated that they had better knowledge after they received
health education. Consumption of Koi pla decreased, even some could not definitely quit but could
reduce in terms of frequency and amount of consumption.
“The doctors come and talk about it every year. I see the health campaign posters every day, it keeps
reminding me not to eat Koi pla,” said one 54-year-old from the previously-infected group.
Because the development of cholangiocarcinoma is a long process and mainly asymptomatic, so the
cancer was not a major concern to the villagers after being infected with OV infection. In fact, the
prevention was simple. Avoidance of uncooked fish did not require sophisticated medical
intervention. Moreover, the treatment was also effective and affordable. At this point, some of them
admitted that they did not pay much attention to cholangiocarcinoma, “I can seek healthcare
whenever I think I’m getting the infection,” said one member of the adult group.
The National Control Program has focused on the control of uncooked fish consumption, which
would interrupt the infection process; therefore, it could potentially prevent the occurrence of
cholangiocarcinoma from OV infection. However, the consumption behaviours were strongly
attached to the local culture, “There were many raw fish dishes, I can’t avoid all of them,” some
mentioned about uncooked fish dishes. The control program suggested that all uncooked fish dishes
should be avoided.
The group discussion agreed that when a particular dish was identified as a risk factor, it would be
easier and more promising for behavioural modification.
166
Children had a better understanding about the knowledge of infection. Some adults and elderly felt
more comfortable and hesitated less to be educated from their offspring. Children served as an
effective medium for transferring knowledge. Additionally, they did not want their children to
practice the same habit.
“It was a good strategy. They’re ashamed if they do not behave as role models,” said one 38-year-old
female from the adult group.
Most villagers had toilets in their house. Some of them installed toilets on their farm. Otherwise they
would excrete in the field and cover with soil. Just a few people excreted directly in the natural
water resource.
“Toilets are everywhere, nowadays. Those who can’t find one use a spade to cover it with soil,” one
55-year-old local health volunteer mentioned about how to excrete in the paddy field without toilet.
“I would hold it until I get home,” said one 18-year-old from the never-infected group.
167
4.4 Discussion
Risk behaviours
Social influence
Opisthorchiasis
Intervention
Prevention Diagnosis and Treatment
B
D
A
E
C
F G
Children
Teenagers
Adults
Elderly
Modern education
Traditional-belief
Knowledge
transfer
Information-based
Traditional-based
Culture transfer
J I
H
Figure 4.3 Conceptual framework of opisthorchiasis
Figure 4.4 Social component integrated to conceptual framework
168
4.4.1 Social component integrated in the main conceptual framework
Figure 4.3 illustrates the comprehensive conceptual framework with regard to social components.
The infection dynamics (A) have been discussed from previous Chapters (2 and 3) with rigorous
statistical analysis and mathematical modelling to describe the epidemiology of OV infection and its
risk factors with regard to the interventions (B and C).
As illustrated in Figure 4.4, the social component has played a major role in influencing the infection
dynamics. Thematic contents provide various outcomes regarding bio-psycho-social aspects; the
impact of social components to risk factors (D) and also the roles of intervention (F and G). OV
infection itself impacts the population as well (E).
I have elaborated on the social context from the baseline framework (Figure 4.3) to explore the
behavioural-psycho-social aspects of the infection. The main theme is based on age structure (H)
where I have observed that the process of knowledge (I) and culture (J) transfer shows some
interesting and unique characteristics.
Regarding the overall picture, basic knowledge of OV infection requires modern education usually
provided by formal resources such as schools. Some traditional beliefs still play important roles
concerning knowledge of infection. Knowledge and attitudes towards OV infection leads to the
choice of practicing risk behaviours and also the perception of prevention and control. The pattern
of knowledge transfer is likely to start with children and teenagers when their parents agree to let
them have modern education from school (I). Younger generations can access more informative
sources such as the internet, and therefore, can pass modern knowledge to the older generation,
i.e., their parents and the elderly.
However, the culture transfer seems to show a confronting direction (J) where the traditional-beliefs
of the local Northeastern culture still have a strong influence on attitudes and perceptions toward
risk behaviours passed on from previous generations. This complex relationship shows that younger
generations with modern education tend to have multiple sources of information and start to justify
the risks and benefits from practicing risk behaviours.
The interaction between modern information and traditional beliefs can be observed during
transitional periods among teenagers and early adults when self-esteem and independence starts to
develop.
169
4.4.2 Knowledge, attitude and perception of OV infection contributed by age structure
Table 4.13 Knowledge, attitude and perception of OV infection contributed by age structure
Young
children
Knowledge was school-based
o Health education and hygiene lesson were given at the school.
o Some of the younger generation tended to have higher educational
levels than their parents; therefore, school was merely a knowledge
hub for children.
Knowledge from school was also transferred to parents by children.
Teenagers
Role models were important
o Family
o School-based: teachers and friends
o Social trend: internet and social media
Sought behaviours for information and tended to believe by evidence-based
data.
Adults
Direct evidence triggers concerns such as fatal cases from
Cholangiocarcinoma.
Adults hesitated to listen to their children less than non-family members.
Elders Age was a natural barrier to learning new knowledge.
Trust was given doctors and medical personnel.
As shown in Figure 4.4, the interaction between the process of knowledge and culture transfer could
describe the method of how knowledge and attitudes could contribute to the practice of
consumption.
Knowledge of OV infection comprised several parts: liver fluke life cycle, mode of infection and
transmission, health consequences and diagnosis and treatment. Results indicated that participants
perceived information and understood the infection concerning different aspects (Table 4.13).
170
It would be impractical to gain full insights in the infection considering that other diseases need to of
equal concern. However, some vital issues were important to raise concerns leading to prevent risk
behaviours as described below.
Mode of infection was consumption of uncooked fish. The specific dishes were instantly
prepared such as Koi pla.
The parasites grow and live in the body for many years.
An infected person may not be aware of the condition because of the asymptomatic features,
but chronic infection leads to cancer of the bile duct (Cholangiocarcinoma) was and is highly
fatal.
Risk of infection maybe higher when practicing risk behaviours, so avoiding uncooked fish
consumption was preferred as a primary prevention.
All age groups acknowledged that the term “parasite” referred to a kind of pathogen and caused
illness. Children and teenagers are formally educated about parasitic infections in school on a
regular basis as core content in hygiene lessons including liver fluke infection. Additionally, OV
infection was also introduced to the community by the research team. As a part of the community
medicine course, medical students provide health education emphasizing OV infection in the
community including community meetings, public relation campaigns and also local radio
broadcasts.
School plays a major role as an information hub equipped with academic resources such as teachers,
books and internet access. Therefore, the majority of health knowledge among young children and
teenagers is school-based.
Teenager behaviours are also motivated by social influence; they are naturally sensitive to
information with respect to physical and psychological development during their puberty. Therefore,
they are more likely to follow social trends including eating habits. The modern lifestyle leads them
to choose more urbanized food. However, teenagers have come to the point that they are starting
to choose independently. In terms of eating habits, they might choose to consume uncooked fish
when they perceived that their previous generation was a role-model. From their information
seeking behaviours, they might have to adjust between knowledge they had about OV infection and
how they imitated others or developed life-long eating habits.
Teenagers and adults also required evidence-based data. For instance, they asked the research team
as they could be shown actual liver flukes or eggs. Data presentation from health professionals also
focused their attention on the infection.
171
The elderly had barriers for learning new knowledge, especially when new knowledge conflicted
with their beliefs. This issue also occurred among adult as they did not pay much attention to the
messages from youth. However, data analysis showed that age barriers could be minimized.
Knowledge transfer could be performed more smoothly by a family member. Adults showed less
hesitation when the information was provided by their offspring. This connection provided the
mechanism for the transfer of school-based knowledge to the community by school children as a
medium creating a practical strategy for knowledge distribution.
4.4.3 Attitude and perception towards risk behaviours
An earlier qualitative study in Na-yao area from 2007 – 09 (70) indicated that the National Control
Program strategy may have some issues regarding the prevention of uncooked fish consumption
campaign (26-28). The participants found that avoiding all uncooked fish dishes was impractical as
indicated by the guidelines provided, so they were more likely to continue the same eating habits
(70) resulting in continued Pla ra consumption. Pla ra was widely used as a main ingredient for local
dishes, so the indirect impact from unspecified uncooked fish avoidance campaign resulted in
continued Koi pla consumption.
For liver fluke infection, the mode of infection was simple and straightforward. However, eating
practices could result in different approaches with respect to attitudes, perceptions and socio-
economic aspects. From the study, patterns of uncooked fish consumption varied across many
factors. Age group had a distinctive pattern observed from the analysis.
Children’s food was mainly prepared by their parents Therefore, the attitudes towards health risk
was difficult to determine because children’s eating behaviours were strongly dependent on the
parents’.
Teenagers were the first to start deciding what to eat. They were sensitive to information such as
social media or TV and they were more likely to become a trend follower. Urbanization has
suggested more modern food choices but eating behaviours could proceed in two ways as described
below.
They may decide to have more modern food while rejecting local dishes as outdated.
They still consumed uncooked fish dishes.
Practical strategies suggest that teenagers could try both approaches. Trial and error could be a
good strategy during physical/mental development in puberty.
172
Adult represented an independent generation; therefore, they had confidence to decide
independently. The underlying knowledge incorporated with personal experience could motivate the
choices of what to eat or not eat. This study revealed an important practical point where
participants perceived that heath risk was controllable and an occasional exception was acceptable.
Strong evidence may be needed to modify existing beliefs often misunderstandings.
The lifestyle was an important factor to be more greatly exposed to health risks. Fish were caught
and prepared instantly during agricultural fieldwork. Men also consumed alcohol at social events.
A relationship was established between states of dependence and independence referring to adult-
and childhood; while they contributed their own decisions, most of them forbade their children from
consuming uncooked fish. However, in high-risk families, all family members consumed Koi pla on a
regular basis.
The elderly exhibited strong attachment to local tradition and culture. Long life expectancy has
allowed some excuses seem logical and rational to them. They have been maintaining a long-term
lifestyle without any health problems relating to parasitic infection, so they did not wish to change
or modify anything. They wished to enjoy their remaining years with a peaceful life.
Knowledge, attitudes and perceptions across generations raised some interesting issues. Common
characteristics of a particular age group indicated points for further intervention.
Table 4.14 Common characteristics regarding knowledge, attitude and perception across
generations
Common characteristics
Children Teenager Adult Elder
Family tradition Transitional period Knowledge misunderstanding Personal belief
Although children were mainly educated in school, they may be fed uncooked fish by their parents.
Food and utensils accidentally contaminated with raw fish might also cause the infection.
Additionally, some families have shown a strong preference for uncooked fish consumption affecting
children’s attitudes in a long run.
As seen in Table 4.14, teenagers were in a transitional period and became more independent when
selecting food. Therefore, they tended to absorb and process a great deal of information.
Information overload may interfere with how they would comprehend the data. Imitation among
173
friends occasionally occurred due to biologically impulsive behaviours. Concern could arise when
they chose to pursue agricultural careers contributing a higher chance to contact risk behaviours and
at risk environments.
Adults and the elderly revealed a misunderstanding through knowledge and personal beliefs. Solid
evidence could provide a stronger impact on their perception such as evidence of toxic waste,
polluted natural water bodies and causes of illness. Moreover, some elderly believed that they have
been living long enough without harmful diseases confirmed by their healthy lifestyle. Regarding
cultural norms, some of them believed that serious illness was a matter of destiny or Karma.
174
4.4.4 Impact of OV infection on health, treatment and prevention
Table 4.15 Impact of OV infection on health, treatment and prevention
Young
children
Health was under the parent’s supervision.
OV infection was demonstrated at school.
Teenagers
Healthy lifestyle also incorporates social influence; peer pressure, social
media and urbanization.
OV infection was intimidating, the parasite was disgusting. Local uncooked
fish dishes were not trendy.
Adults
Physical health must be maintained to work every day.
Some risk behaviours were considered a reward or treat after hard work.
o Alcohol consumption
o Uncooked fish consumption
OV infection was asymptomatic; others diseases were considered more
harmful. Moreover, it did not obviously affect physical health.
Elders
Healthcare was more likely a defensive strategy.
o Doctors were responsible for their health.
Their long life has proved that they have a healthy lifestyle.
OV infection was intimidating, but they couldn’t change their way of life.
They felt that some medical treatments were more harmful than the disease.
Table 4.15 describes the impact of OV infection on health, treatment and prevention across
generations.
Young children and teenagers are in a growing phase of life. They were generally considered at a
healthy age with energetic activity. Therefore, physical health was not a primary concern. Young
children’s health is under parental supervision. Most of them were in education institutions; they
were taught hygiene lessons beginning in school including basic hygienic practices generally
indicating that uncooked food was unhealthy.
Teenagers are just starting to learn that pathogens can cause disease. Moreover, uncooked food was
not considered trendy. Teenagers tended to pay more attention to their physical body; they became
more sensitive to their image and also body transformation. After seeing actual specimens, then
175
knowing that raw fish meat contained live metacercariae that actually resided in their bodies made
them feel revolted.
Adult showed more concern to physical function than general appearance. They had to maintain
physical fitness to perform work almost every day. Moreover, some risk behaviours were considered
rewards after a long and hard day at work such as social drinking group or consuming Koi pla
considered a tasty dish along with alcohol.
Many chronic diseases begin at adult age. Some conditions such as hypertension or diabetes have
asymptomatic manifestations at the early phase. OV infection shares this feature and affects health
awareness. In fact, they were more likely to be concerned about what would immediately
compromise their fitness such as myalgia or trauma.
The elderly thought that healthcare should be under a doctors’ supervision. Their ideas represented
a passive defensive strategy where they only sought medical care after contracting illness. They also
justified the efforts compared with the benefits for behavioural modification. The benefits from
primary prevention may not be visible because the disease is prevented rather than when an illness
is cured.
However, the issue involves self-conflicts. They also perceived that taking praziquantel as a
treatment for OV infection caused more illness than being infected and remaining asymptomatic.
Feeling nausea and dizziness were unfavourable outcomes.
The reason may lead to the point where they think that medical care should only involve medical
equipment and healthcare workers such as seeing doctors in the hospital or receiving medicine. As a
result, primary prevention represented a separate activity rather than involving a part of healthcare.
Because prevention was less prioritized, treatment became a crucial part for health issues.
Lack of motivation was an important issue for both prevention and treatment. Moreover, even when
they understood fatalities resulted from Cholangiocarcinoma, it could be a result from chronic
infection occurring many years ago. Considering their remaining years of age, they calculated having
cancer now was unlikely.
176
4.5 Conclusion
The qualitative approach including focus group discussions and in-depth interviews revealed that
uncooked fish consumption was decreased in terms of quantity and frequency. However, some
people still consumed uncooked fish regularly. Uncooked fish dishes such as Koi pla are still popular.
From a psychosocial aspect, the dish itself represents Northeastern culture.
Population dynamics also influenced the social aspects of OV infection. Regarding the quantitative
study results, the infection increased among older ages; the elderly generation recently migrated
from Northeastern Thailand. Moreover, traditional beliefs were transferred from their ancestors to
their descendants. The evidence could be visually observed while they still preserved their lifestyle
despite the fact that they already left their origins.
Additionally, Koi pla was considered a tasty dish, which many people considered an important factor
to continue consumption. However, the generation gap also revealed that the younger generation
was likely to absorb modernized culture and become more urbanized.
Some elderly complained that refusing to eat local fish dishes might go against old traditions with
respect to cultural norms and socialization. Higher education teaches about hygienic conditions
including OV infection, so teenagers and young adults tended to modify their lifestyle including
eating habits.
Therefore, health education alone would not be sufficient to motivate the behavioural changes
required. More efforts are needed to support the transformation. Children represent a potential key
to pass knowledge to their parents and schools are practical hubs for knowledge distribution.
From rural settings, community leaders are important keys in the social structure. As the local
administration is managed by the central and regional authority, governmental staff might not be
aware of local cultural norms. Community leaders could bridge this gap and help facilitate the
government’s strategies.
Medical doctors and healthcare personnel are also considered community leaders looking after
villagers’ health. People respect and trust doctors when they are instructed and advised regarding
their health. However, doctors are not permanently stationed at local health centres. They regularly
visit the community once a week mostly for chronic disease appointments. In Thailand, the shortage
of human resources is still a major problem in rural areas.
177
Local health volunteers played important roles for communicating and interacting among villagers
and health authorities. They could act as community leaders and as role models for health
interventions. Because they are locals, they could more easily gain trust and cooperation from the
villagers, a crucial part to maintain the sustainability of health campaigns in the community (70, 114,
132).
Social influence plays an important role for shaping the infection dynamics of OV infection. Within
the social components (Figure 4.3), the interaction of knowledge and culture transferred across
generations provided more insight to bio-psycho-social dynamics (Figure 4.4).
In Chapter 5, community intervention employing a mixed-method study as a comprehensive tool
revealed the potential use of the application of quantitative and qualitative approaches with regard
to the host-agent-environment aspect of the infection dynamics.
178
Chapter 5 Community-based intervention: PRECEDE-PROCEED
model framework for controlling OV infection
Abstract
OV infection is a liver fluke infection caused by Opisthorchis viverrini. The important mode of
infection is consumption of uncooked freshwater fish containing infective metacercariae. OV
infection mainly features asymptomatic manifestations but chronic infection could lead to bile duct
cancer or Cholangiocarcinoma. Over decades, the National Control Program has targeted endemic
areas aiming to suppress prevalence down to 10%.
Current fieldwork conducted in 4 areas of Ta-Kadarn District located in rural central Thailand
indicated that the prevalence was higher than the regional average and incidence was also high
considering the studied areas were outside the endemic Northeastern Thailand. Follow-up studies
have report significant association of Koi pla consumption with increased risk of OV infection and a
qualitative approach has revealed the potential influence of bio-psycho-social factors contributing to
the practice of Koi pla consumption related to Northeastern culture where the community founders
originally moved from.
The PRECEDE-PROCEED model framework was applied to the community of Tung-heang to
systematically engage the problem with the aim of reducing the incidence of OV infection through
behavioural modification by decreasing Koi pla consumption. The community-based intervention
aimed to reduce risk behaviours was developed using a mixed-method study design. Community
trials were conducted to assess the effectiveness of the intervention on incidence of OV infection
and consumption of Koi pla. The distribution and risk factors of OV infection were obtained by the
cohort design and qualitative approach using up-to-date information.
Community trial was followed by an open-label, nonrandomized community trial design. The
intervention group participated in the community by mutual agreement to discontinue Koi pla
consumption. After the 17-month study period, the intervention group and control group were
diagnosed for OV infection by stool examination and assessed for risk behaviour by questionnaire.
The results indicated that community intervention could significantly reduce incidence of OV
infection by 63% (95% CI: 7 – 85), p = 0.03 and reduce Koi pla consumption by 46% (95% CI: 1 - 71), p
= 0.04. For the cohort analysis, Koi pla consumption could increase risk of OV infection with IRR =
2.52 (95% CI: 1.07 – 5.92) when adjusted for age group. Additionally, social and cultural attributes
were potential factors leading to Koi pla consumption.
179
Consumption behaviours were complex involving traditional beliefs, attitudes and diverse cultural
backgrounds. Following the PRECEDE-PROCEED framework, community participatory action
including community-derived intervention was considered adaptive and practically suited to
villagers’ lifestyles. Local health volunteers were originally community members and could play an
important role as a reinforcing factor.
From the fieldwork study, the mixed-method design provided comprehensive approaches to assess
OV infection in various aspects. The PRECEDE-PROCEED model framework also provided a
comprehensive approach for intervention/evaluation planning. The project initiative provided more
insight of community management which was expanded to a larger scale in which the community
intervention package could be developed and applied for public health implementation on a national
scale.
180
5.1 Introduction
The prevalence of OV has been estimated across Thailand. The record from the 2001 national survey
shows various distributions of infection. The prevalence was high in the North (19.3%) and the
Northeast (15.7%) while the prevalence was 3.8% in the central Thailand and 0% in the South.
Overall, the national prevalence was roughly 10%. The endemic areas were limited to local areas
with local habitats for intermediate hosts and traditional eating habits (27, 28).
The first national survey in the 1980s revealed the prevalence of OV infection was 63.6%. Therefore,
an OV control program was started from the 6th 5-year National Public Health Development Plan
(1987 – 1991) aimed to control the infection as described in Section 1.4.1.2. The national prevalence
declined to 9.4% in 2001. Recently, the control activity has focused on areas in the North and
Northeast where the infection has remained endemic (28).
Infection dynamics of OV infection involve multiple factors. Consumption behaviours were complex
activities, which could be influenced by many factors such as traditions or cultural norms (1, 5, 15,
29). The qualitative approach showed that bio-psycho-social aspects played an important role for
the consumption pattern. Apart from the qualitative components, OV infection itself shows a
dynamic picture of prevalence and incidence (Chapters 2 and 3). I have observed that the prevalence
and incidence were not linearly associated where the incidence obtained from fieldwork was
significantly higher than calculated incidence from prevalence-age linear function (Section 3.1.2 from
Chapter 3). Moreover, re-infection could play an important role for a chronic infection picture and
shape a sustained prevalence (Section 3.4.1 from Chapter 3).
Based on the previous situation when the national prevalence was high from 1984 – 1987; 63.6%
(28), the National Control Program was designed to mainly focus on secondary prevention along
with treatment of infected cases (26, 28) from1987 - 2001. In the present day, this strategy should
be re-evaluated to deal with emerging cases or re-infection.
A Planning model has been proposed to systematically engage health burden (133-135). The
PRECEDE-PROCEED model (PPM) framework (133, 136) is a planning model which organizes the
method of examining root cause of the problem and sets up the goals for reducing the burden with
the planned intervention. The PPM has been used in some parasitic infection cases such as the plan
for controlling Taenia solium in Nepal (137) and school-based health promotion for helminth control
in KwaZulu-Natal (138). Both studies used PPM to assess and develop community intervention with
regard to social and environmental factors with consideration of feasibility and compatibility with
current policy.
181
In the macroscopic level, the PPM framework was used to approach the challenges posed by OV
infection transmission at the community level while reflecting the Thailand rural community
perspective.
5.1.1 Overview of PRECEDE-PROCEED model
The
PRECEDE-PROCEED planning model was a framework for the public health planning aim to reduce
disease burdens at the macroscopic level (133, 136, 139). To establish an effective intervention,
many factors were considered from individual to health policy levels. Thus, the PPM was very much
an ecological approach. The relationship between individual characteristics and environmental and
social contexts were incorporated in the framework.
Figure 5.1 was adapted from Crosby R. (133) based on a primary framework developed by Green
L.W and Kreuter, M.W (136). Basically, heath behaviors were considered complex and dynamic
(140). It was challenging for public health practitioners to engage heath issues in the community. To
approach complicated community-based problems, local assessment was needed to identify specific
problems fitting in a particular setting. From Figure 5.1, the PRECEDE-PROCEED diagram was used to
analyze factors while intervention could be planned and applied at the same time using appropriate
theory for practical guidelines. Two key aspects to form a framework were planning and evaluation.
From the diagram, PPM involves 9 stages of health planning and evaluation. The framework begins
with largest goal, ultimately judged by the outcome evaluation after the relevant factors were
assessed and the intervention was implemented. The process was performed step by step.
Figure 5.1 Diagram of PRECEDE – PROCEED model
182
PRECEDE stands for Predisposing, Reinforcing, and Enabling Constructs in
Educational/environmental Diagnosis and Evaluation (steps 1 to 4). In general, public health issues
normally start with a disease or health threat. To generate a health plan, the root cause was traced
back to the source of the problems where relevant factors were assessed by a multidisciplinary
approach. Health-related objectives were created to set the ultimate goals for the intervention. The
objectives were usually measurable and time-limited, which was practical, to be later evaluated for
the outcome.
PROCEED stands for Policy, Regulatory, and Organizational Constructs in Educational and
Environmental Development (steps 5 to 9). After the planning stage was carefully assessed, an
intervention based on health theory was used to modify health-related risk factor in this stage. The
intervention package was implemented in the community and the evaluation process finally
assessed the outcome of the intervention such as the reduction of disease burden. The vital key was
the evaluation process, which monitored the intervention program to ensure the fidelity of the plan.
183
Figure 5.2 Conceptual framework for community intervention study
Based on the OV infection conceptual framework from Figure 5.2, the planning process involved
assessing data regarding the infection dynamics described below.
Social diagnosis assessed the burden of disease and the strengths and weaknesses of the
community setting to engage the problem.
Epidemiological study assessed the burden the disease and determined risk factors contributing
to the infection.
Administrative and policy assessment evaluated the roles involved in the current management
of OV infection, which will help in guiding the design of intervention with regard to feasibility
and resources.
Collected data will guide the specification of objectives to reduce disease burden and its associated
risk factors. The planning process will establish community intervention for achieving the objectives.
Risk behaviours
Social influence
Opisthorchiasis
Intervention
Prevention Diagnosis and Treatment
Planning phase
Evaluation
phase
184
The evaluation process involves the implementation of the intervention, which requires an
appropriate study design. Intervention aimed to achieve the objectives involving risk factor
modification to eventually reduce OV infection.
185
5.2 Methodology
5.2.1 Overview of study methods
[A] Cross-sectional conducted in 2010
[B] Study outcome was explained to the community
[C] Initiation of community intervention to reduce opisthorchiasis
[D] Planning process: PRECEDE
[Step 1] Evaluate burden of the opisthorchiasis
[Step 2] Set key objective to reduce opisthorchiasis
[Step 3] Set sub-objective to reduce risk behaviour contributing to opisthorchiasis
[Step 4] Design of community intervention to achieve objective and sub-objective
[E] Evaluation process: PROCEED
[Step 5] Evaluate opisthorchiasis-related policy
[Step 6] Implement community intervention
[Step 7] Evaluate the flow of study
[Step 8] Evaluate the outcome of sub-objective
[Step 9] Evaluate the final outcome of objective
[F] Inform and discuss the study result to the community
Figure 5.3 Summary of study flowchart
186
Figure 5.3 summarizes the entire process of this Chapter. The study was conducted in Tung-heang
Area. In 2010, the baseline survey was performed for OV infection for which prevalence was 11.42%
(95% CI: 8.69 – 14.63) [A].
After the baseline survey, the fieldwork team provided information to the community [B]. The
community perceived the burden of the problem and realized this was a priority issue for the
community, which initiated the idea of reducing OV infection by community-based solutions.
Members of Tung-heang agreed to initiate the community-based intervention [C]. The whole Tung-
heang Area also agreed to join a follow-up study for evaluating the incidence of OV infection.
Tung-heang community members brainstormed for a method of controlling OV infection based on
behavioural modification by mutual agreement. The fieldwork team agreed to participate in this
project by providing diagnosis at the end of follow-up time and provide data management to
evaluate the effectiveness of the intervention [E]. In the final process, the research team analyse
data and transfer information to the community [F].
The PRECEDE-PROCEED model was used to holistically approach the problem, the process of
intervention and study outcome. A planning model was created following the key steps. The pilot
community-based intervention was then initiated.
5.2.2 PRECEDE – PROCEED model establishment
The study flow followed the 9 steps in the PPM framework as presented in Figures 5.3 [D] and [E].
PRECEDE (planning phase)
5.2.2.1 Step 1 Social diagnosis
This step investigated the burden of OV infection, strengths and weaknesses of the community
setting and community acknowledgement of the health issues.
Initially, the baseline prevalence from the study survey in 2010 was provided to the community.
Additionally, risk factors obtained from adjacent study areas; published data (69, 70) and fieldwork
data (Chapter 2), indicated that Koi pla consumption was a major risk factor for acquiring the
infection.
The community perceived the burden of the problem and realized this was a priority issue. Social
diagnosis was a description of community characteristics and a relationship with the targeted
disease; OV infection. Social contexts were evaluated by qualitative approach including focus group
187
discussion and in-depth interview. Therefore, the qualitative approach was used to gain social-
related data (129).
These key questions were provided and discussed to assess the impact of the infection.
1. How do community members acknowledge OV infection?
Basic knowledge of OV infection; risks factors, modes of transmission, diagnosis and
treatment
Consequences of chronic infection; role of carcinogens, Cholangiocarcinoma
2. Attitudes toward the infection and its subsequent outcome
Perceptions of becoming infected
Burden of problem in psycho-social aspect.
3. Uncooked fish eating behaviours
Current situation of eating habit
Precipitating factors to promote uncooked fish consumption
Precipitating factors to discontinue uncooked fish consumption
4. How does the community react to the problem?
Community concern of the problem
How will they prioritize the situation?
How will they manage a solution?
In this circumstance, community leaders played an important role as a respectful key person and
also acted as the coordinators between the research team and community members.
Community leaders comprised of administrative and spiritual leaders such as head of the village,
local intellectuals and monks. Local governmental officers who work in the area also served as key
persons organizing governmental-related affairs in the community. For healthcare aspects,
community leaders worked as healthcare personnel at the local health facility and as local heath
volunteers.
The Community Advisory Board (CAB) was a key part of the entire PPM framework. The members of
the CAB were chosen from community leaders and community members. As a part of community
participation, the CAB coordinated communication with the community members following the PPM
framework: gathering and exchanging information. Therefore, the CAB was an integral part of the
entire PPM and the maintenance of the CAB was required to coordinate communication between
community members and the research team.
188
5.2.2.2 Step 2 Epidemiological diagnosis
This step emphasized creating the objective for final outcomes of the program. In general, the aim
was the reduction of disease burden or increased rate of health promotion campaign. The objective
was usually measurable and time-limited and was later evaluated. Additionally, available field data
aided the planner to set the ultimate goal for the intervention program.
The primary objective was to reduce incidence of OV infection by the end of study with a 17-month
follow-up period.
5.2.2.3 Step 3 Behavioural and environmental diagnosis
This step identified potential behavioural and environmental factors influencing the burden of
disease. These factors may increase risks for acquiring disease or prevent effective prevention and
control.
Then the identified factors were applied as sub-objectives directly leading to planning for the
intervention. The sub-objectives were also set as measurable goals for example, reducing the
prevalence of risk behaviours contributing to the infection or modifying some environmental factors
which naturally supported the pathogen. From the framework, the behavioural and environmental
modification ultimately reduced the disease burden.
The sub-objective was to reduce risk factors for acquiring OV infection, i.e., consumption of Koi pla,
by the end of study with 17-month follow-up period.
5.2.2.4 Step 4 Educational and ecological diagnosis
In this step, an intervention plan was developed to achieve the sub-objectives proposed in step 3.
This was a vital step to manage either behavioural or environmental modification through a health
intervention.
The PPM framework also provided a systemic strategy to assist the planner to develop an
intervention plan. Each sub-objective required a separate plan to fulfil each goal. At this point, three
subsequent factors were considered (133, 136).
Predisposing factors
Predisposing factors were originated at the cognitive level through attitudes and beliefs. From
individual to community levels, perception, knowledge and health beliefs toward the disease were
important factors shaping attitudes for health concerns. They could be derived from either internal
189
or external sources, for example, passing traditional beliefs from the previous generation or health
education provided by health authorities. In general, modification of predisposing factors was
achieved by educational campaigns.
Reinforcing factors
According to the framework, the planned intervention should contribute to behavioural or
environmental modification. After intervention was applied, reinforcing factors assured the
sustainability of the program effectiveness. This was a crucial step as normally public health
interventions were only performed in a point or period of time. Repeating desired behaviours as
routine is a challenge for health planners to maintain the sustainability of the intervention program
in the community.
Enabling factors
Enabling factors make it easier or possible for the behaviour to occur such as required skills to
perform the intervention, and an affordable price for the medical campaign or even convenient
transportation to the health facility. More support will provide more cooperation from the
community. Skill-related enabling factors and environmental factors should be concerned to develop
suitable community-based intervention.
Behavioural science theory was widely used to support the intervention planning. The PPM
framework was not attached to any specific theory. In fact, each designated intervention had
different objectives. Therefore, health theory was selected based on the proposed intervention
developed in this step.
190
The role of behavioural science theory in the developing of community intervention
In general, health intervention was perceived as an action aimed to reduce health burden (141).
Many studies suggest that health intervention will be more effective if the design is based on
theoretical foundation. Many theories have been used to construct health interventions. Therefore,
the development and implementation of the behavioural science theory should be based on the
conceptual framework and objectives of the health program. The core fundamental concepts
regarding the success of intervention was to comprehensively understand the risk behaviour and its
context (128, 139, 140, 142).
From steps 2 and 3, objectives and sub-objective, goals were set to reduce incidence of OV infection
by modifying one of the potential risk factors; consumption of Koi pla. Two theories served as an
underlying framework for the community intervention development.
Health belief model (HBM)
The HBM is widely used in public health programs (143, 144) as a core principle of how an individual
reacts to a health problem and justifies the action to reduce the risk of a given problem (140). The
HBM explains that the individuals will justify the risk behaviour once they perceive appropriate
information on the severity of the health problem and the feasibility and obstacles to modify risk
behaviour (128). The theory emphasizes self-efficacy where the individuals will recognize and modify
their risk behaviours by themselves.
The HBM was effective against diseases with long asymptomatic periods requiring the accumulation
of risk behaviour such as the study of reducing risk factors for cardiovascular disease (145). The
intervention developed under the HMB framework aims to reduce the health risk to reduce or
prevent the subsequent disease as a primary prevention.
Therefore, the concept was consistent with the OV infection framework where the risk behaviour
was a major key contributing to the infection. Moreover, OV infection presents mainly an
asymptomatic period with silent development of subsequent Cholangiocarcinoma. Information
about OV infection including prevalence and incidence, consequence of chronic infection and results
from fieldwork study indicated the risk from consuming uncooked fish were transferred to the
community.
Self-efficacy comes from the individual to community levels. As each member perceives the threat of
OV infection, they might still have the problem regarding the method of behavioural modification.
191
Then community perception and awareness plays an important role as it ensures the community
members that the goal for handling OV infection problem will be carried on in the same direction.
Social cognitive theory (SCT)
From the framework, SCT expanded the behavioural modification regarding self-efficacy factors by
proposing that personal factors also interacted with environmental factors (128). Health behaviour
was not only influenced by personal experience but also by observing other actions, which was
called the reciprocal model (146).
Community intervention developed by the members of community themselves can play a major role
as the activity derives from the collective opinions and are easier and more accessible to the local
norm. Self-efficacy was still a major concept for SCT with additional concerns of social perspective to
push the intervention forward as a whole community.
More efforts were provided to the intervention planning to increase self-efficacy by three strategies
described below.
Setting precise and achievable objectives; clear targets can be more easily approached. Avoiding
confusion with simple objectives was more appropriate to engage the problem in a larger scale
such as the community level. Problems cannot be solved in every respect at once. Therefore, the
objectives from steps 2 and 3 were clear for reducing the incidence of OV infection by reducing
Koi pla consumption.
Applying specific instructions on behavioural modification, which was more practical to conduct
within the same direction. Health education was also provided to the community to build up
self-efficacy to engage in the intervention.
Reinforcement: the factor contributing the continuity and sustainability of behavioural
modification. In general, monitoring system were an effective method for sustaining individual
awareness to the ongoing intervention.
192
Initiation of community intervention
From the PPM framework, the intervention was initiated on the basis of community participation
with the SCT strategy. Following the HBM, community was concerned about the problem and agreed
to take action to lower OV infection incidence emphasizing Koi pla consumption. The intervention
was developed based on community mutual agreement (Table 5.1). The research team agreed to
evaluate the intervention outcome by providing diagnosis of OV infection at the end of study as a
result of behavioural modification and observed the intervention process along with the CAB.
Table 5.1 Community mutual agreement
Community mutual agreement
• Joining agreement was voluntary. After being informed, participants agreed to sign in consent form
to confirm their intention. The participants must be 18 years of age and over.
• Participants should follow guidelines to avoid risk of acquiring OV infection:
o Targeted strategy emphasize on Koi pla consumption (risk was identified through statistical
modelling and qualitative approach).
o Avoid eating Koi pla (chopped raw fish salad)
o Maintain good hygiene of the kitchen to prevent food contaminated with raw fish scraps.
• Defecate in hygienic toilets to prevent transmission.
• Following the programme schedule, participants should send their stool to fieldwork team for OV
diagnosis at the end of study.
The intervention group was occasionally reminded by CAB as a monitoring system to promote the
sustainability of intervention.
Those who did not participate in the intervention group were invited to join the study to provide a
stool specimen to evaluate the incidence as a control group and risk factor analysis. Once a stool
examination diagnostic result was available, it was delivered to participants individually. At this step,
it was their decision whether to provide their own result to other community members. The
fieldwork team only took part in overall result analysis and did not participate in any subsequent
activity from the agreement. Incidence of OV infection in the intervention group was compared with
the control group to assess effectiveness.
193
PROCEED (evaluation phase)
5.2.2.5 Step 5 Administrative and policy assessment
This was the most challenging step in the entire process, which assessed the available resources to
implement the intervention in the community. Careful and informative assessment will aid the
successfulness of the intervention establishment.
The objectives of the assessment can be divided into two categories explained below.
Health education
Health education serves as a long-standing scheme in public health campaigns. Basically, health
education efforts will focus on predisposing and reinforcing factors. Knowledge could be widely
provided to the community through various sources. Social norms could be reshaped to serve as a
method of reinforcement for healthy practices. For enabling factors, better health knowledge also
enables the community to improve the performance for performing the intervention.
Policy, regulation and organisation structures
The health policy involves all aspects of healthcare serving as an infrastructure to the community.
The medical facility was the primary source for the community members to seek for healthcare.
Thus, the role of existing health policy needed to be assessed to achieve the research goal.
An intervention plan might be already compatible with the current health policy. Otherwise, the
planner can consider altering the plan or modifying the policy to finally suit the feasibility and fidelity
of the program. An assessment can also aid the planner to seek political support to aid the policy
establishment.
The focus group discussions and interviews with participants provided information about the health
policy from the local healthcare administrators in the community.
Qualitative data could provide two-way opinions from the healthcare worker viewpoint who manage
to follow National guidelines for OV infection and the community’s expression to the healthcare
roles. It could also assess the feasibility of the policy to the intervention. Furthermore, the health
policy should support when the intervention needs to be self-sustained and solely managed by the
community in the future.
194
5.2.2.6 Step 6 Implementation
This step was one of the crucial steps for the planning model. To achieve the goal of the
intervention, planners should consider the role of program evaluation. While the health program
was running through timeframe in the population, the evaluation plan should be also assessed from
the beginning to evaluate all important aspects of the program comprising three parts: process
evaluation, impact evaluation and outcome evaluation.
419 Baseline population from 2010 cross-sectional study with negative result for opisthorchiasis
357 Age ≥ 18 years 62 Age < 18 years
[A] Enrolled to community intervention
study
Not consented to
intervention study
[B] Enrolled for study of risk factors
Intervention group Control group
Figure 5.5
Figure 5.6
Subject allocation
Figure 5.4 Summary of fieldwork flowchart
195
As shown in Figure 5.4, negative cases from the baseline survey were invited to join the follow-up
study in 2012 to measure the incidence of OV infection. The eligible subjects were enrolled i the
community-based intervention program. Based on 2010 baseline survey, total eligible subjects for a
follow-up study totaled 419. The studies were divided in community intervention [A] and risk factors
studies [B].
The whole study was conducted in Tung-heang Area. Individuals with 18 years of age and over were
enrolled in the community intervention study and then allocated to intervention and control groups.
The remaining subjects from the intervention group including ≥18 years-old control group, those
who did not consent to community intervention and the remaining <18 years-old individuals were
enrolled to provide stool specimens and questionnaires for the risk factors study.
The community intervention followed the protocols shown in Figure 5.5 and risk factors study was
undertaken as described in Figure 5.6.
Both community intervention and risk factors studies had the same end-point at 17 months of
follow-up period. The demographic data and risk behaviours were assessed through questionnaires.
The additional qualitative study was used to provide in-depth information regarding bio-psycho-
social aspects.
5.2.2.7 Fieldwork methodology
The study aimed to evaluate the community intervention based on fieldwork data and explore
disease epidemiology using multimodality approaches. The study design was a mixed-method design
described below.
The quantitative study aimed to use statistical analysis to capture epidemiological data.
o Open-label, nonrandomized community trial for community intervention study
o Retrospective cohort study for distribution and risk factors of OV infection
The qualitative study aimed to explain the cause of risk factors and evaluate the impact of
community-based intervention in bio-psycho-social aspects.
196
Mixed-method approach
Eligible subjects were invited to participate in the study. The inclusion criteria are stated below.
Resided in Tung-heang Village
Participated in the epidemiological survey in 2010 where OV infection was diagnosed by
stool examination and negative for OV infection
Consented to participate in the study project
Participants in intervention group were ≥18 years of age. Excluded participants from this
criterion were still eligible for the risk factors study.
Able to communicate with the fieldwork team in answering the questionnaire or joining
the discussion groups.
From October 2010, participants were divided in two study groups: intervention and control. The
intervention group formed a community-based intervention called a community mutual
agreement; they volunteered to avoid risk behaviours aimed to reduce the incidence of OV
infection.
The follow-up time was 17 months and the study finished in March 2012. The fieldwork team
evaluated the intervention outcome by measuring incidence of infection compared with
intervention with the control group.
Epidemiological data were assessed to explore the situation of OV infection along with updating
disease distribution and risk factors.
Social contexts were evaluated by qualitative approaches, i.e., focus group discussions and in-
depth interviews.
197
As shown in Figure 5.5, the study design was an open-label, nonrandomized community trial.
Enrolled subjects were at least 18 years of age. Enrolled subjects were allocated in the intervention
and control groups. Based on community mutual agreement protocols, the participants in the
intervention group avoided eating Koi pla. Therefore, the allocation process was not randomized
because consenting participants volunteered for the behavioural modification.
The intervention group followed protocols from the community mutual agreement (Table 5.1). The
intervention group were reminded about the intervention protocols individually every 1 – 2 months
by the CAB. No records were made with regard to their consumption behaviours in both study
419 subjects with negative result for opisthorchiasis from 2010 cross-sectional study
Excluded subjects for
Age < 18 years
Not consented
Eligible subjects allocated to the study group
Subjects allocated to intervention group Subjects allocated to control group
Follow-up at 17 months
Completed study
Excluded for incomplete
questionnaire
Final analysis
Figure 5.5 Community intervention flowchart
198
groups. The study outcome was assessed through stool examination and questionnaires for the
incidence of OV infection and risk factors after 17 months of follow-up period.
As shown in Figure 5.6, the study of distribution and risk factors of OV infection followed the
retrospective cohort design. From 419 baseline subjects, participants enrolled in the intervention
group in community trial were excluded.
Enrolled subjects for the study included subjects age <18 years. Subjects with age ≥18 years were
allocated to the control group from community trial, and those who were excluded from the
community trial study.
419 subjects with negative result for opisthorchiasis from 2010 cross-sectional study
Excluded subjects for
Age ≥ 18 years and allocated to intervention group from intervention study
Not consented to risk factor study
Enrolled subjects for risk factor study
Age < 18 years
Age ≥ 18 years and allocated to control group from intervention study
Excluded from intervention study
Follow-up at 17 months
Completed study
Excluded for incomplete
questionnaire
Final analysis
Figure 5.6 Distribution and risk factors study flowchart
199
After 17 months of follow-up, study participants provided stool specimen to diagnose OV infection
and complete questionnaires to assess risk behaviours regarding the infection within the follow-up
period.
Outcome evaluation
Based on standard fieldwork protocols explained in Section 2.2.2 from Chapter 2, incidence of OV
infection was estimated from stool examination data. Demographic data and risk factors were
collected by standardized questionnaires.
Stool collection and processing
Based on Section 2.2.4 protocols, each participant received a survey package including questionnaire
and a sealable container for collecting a stool specimen at the end of follow-up time
Each participant received their personal stool examination result; reporting any detected intestinal
pathogenic parasite infections such as helminth and protozoa infection, and health education based
on National Health guidelines. Infected participants received proper medical treatment according to
their pathogen(s).
Questionnaires
Each participant was required to complete a standardized questionnaire to assess basic demographic
data and risk behaviours for acquiring the infection. The questionnaire had been validated and used
in earlier studies in the same study area. Basic demographic data included sex, age and occupation.
Risk behaviours were evaluated by assessing uncooked-fish eating behaviour patterns (69, 70, 98).
Uncooked fish preparations were classified in two major groups; Koi pla and Pla ra as described in
Section 2.2.3
Data analysis
Each participant was matched to questionnaire and specimen using a code-embedded identification
system. Stool examiners were blinded to the individual who provided the specimens. The data
management unit matched all corresponding data (questionnaire and stool examination results) for
data analysis. Incomplete questionnaires with missing data on relevant exposures; sex, age,
occupation and consumption behaviours were excluded to perform risk factor analysis.
200
Descriptive statistics were used to describe data; outcome was reported as frequency and
percentage. Statistical analysis was reported with a significance level of 0.05 and 95% confidence
interval.
For the community trial study, the results were based on two objectives illustrated below.
Primary objective: to reduce the incidence of OV infection by the end of study with 17 months
follow-up. Incidence rate of OV infection between intervention and control group were
compared using incidence rate ratio. p – value <0.05 indicating a significant reduction of OV
infection from the intervention group.
Sub-objective: to reduce of Koi pla consumption by the end of study with 17 months of follow-
up. Incidence of Koi pla consumption between intervention and control groups was compared
using relative risk. p – value <0.05 indicating a significant reduction of Koi pla consumption from
intervention group.
For the distribution and risk factors study, distribution of OV infection was reported as incidence
rate (/100 person-years). Univariable analysis was initially performed to examine the association of
OV infection of each exposure of interest and providing crude association. Exposures included: sex,
age, occupation and uncooked fish consumption behaviours. The association between univariable
risk factors and the infection was quantified using Pearson’s chi-square. Multivariable analysis was
performed using regression modelling to adjust risk factors. Risk factors were adjusted for age based
on previous works (69, 70). Factors with p value <0.2 from univariable analysis were also considered
for inclusion in the multivariable model. Instant fish preparation (Koi pla) consumption was
recognized as risk factor from previous works in this project (69, 70). However, extensively
fermented fish (Pla ra) were considered to be included in the model to control for confounders
based on the data that they were also consumed on a regular basis. Poisson’s regression models
were developed to study incidence outcome with risk factors and with results reported as incidence
rate ratios.
201
Qualitative study
Research participants were invited to a group conference or interview in a focus group discussion,
and in-depth interviews. The qualitative study design was based on the study protocols from Section
4.2 in Chapter 4 comprising group (Section 4.2.1) and in-depth interviews (Section 4.2.2). The
qualitative approach aimed to achieve the following objectives.
Evaluate the impact of community-based intervention
Evaluate the risk-behaviour relationship of OV infection with respect to bio-psycho-social
aspects
The impact of community intervention was assessed through the intervention group. The risk-
behaviours’ relationships were evaluated from participants in the distribution and risk factors study.
The interview sessions were flexible according to the number of participants, schedule and
environment. The method of participant selection was purposive sampling.
Table 5.2 Selection criteria for focus group discussion
1 2 3 4 5
Koi pla consumption
CAB Baseline study 2010 + + - -
Follow-up study 2012 - + - +
As shown in Table 5.2, the selection criteria were based on patterns of Koi pla consumption. The
pattern was assessed from the 2010 baseline study whether the subjects reported consuming Koi pla
or not, and then the 2012 follow-up study when they still reported consuming Koi pla or not. Group
5 involved the CAB which was evaluated for additional administrative and policy assessment from
Section 5.2.2.5.
The moderator organized the meeting and managed the questions about the relationship of
uncooked fish consumption behaviour and the community intervention program. The themes of
questions are described in Section 5.2.2.1 (step 1 for social evaluation).
202
Data processing
The recorded conversations from each session were transcribed to text. Then text-based data were
sorted and coded according to categories. Repetitive quotes, expression or statements were also
summarized in frequency. Data were revised, organized and summarized for analysis. The methods
included content analysis, direct quotations and selected words to consider actual local words used
by the participants.
Analysis approach included thematic content analysis, framework analysis and grounded theory. The
analysis process is described in Section 4.2.3 from Chapter 4.
Ethical considerations
The project was approved by the Ethics Committee of the Medical Department of the Royal Thai
Army. Informed consent was obtained from enrolled participants or parents of young participants
following standard protocols. Those who were positive for intestinal parasitic infection received
appropriate anthelmintic treatment.
203
5.2.2.8 Step 7 Process evaluation
This step was to monitor the flow of the intervention. During the course of the health program, the
protocols might not have been accurately followed. Therefore, the process evaluation was acquired
to ensure the fidelity of the program blueprints. The process included continuous monitoring and
correction which was conducted when the program was nearly completed.
Table 5.3 Review of study process and method of assessment
Step Task Method of assessment
PRECEDE
(Planning phase)
1 Social diagnosis Data from baseline cross-sectional study
Qualitative approach
2 Epidemiological diagnosis Data from baseline cross-sectional study
3 Behavioural and
environmental diagnosis
Data from baseline cross-sectional study
4 Educational and ecological
diagnosis
Qualitative approach
Community intervention establishment
PROCEED
(Evaluation phase)
5 Administrative and policy
assessment
Interview with community advisory board
Review of literatures
6 Implementation Summary of study process
7 Process evaluation Summary of study response
8 Impact evaluation Follow-up fieldwork data
9 Outcome evaluation Follow-up fieldwork data
As shown in Table 5.3, the process was monitored by the CAB. The research team observed the
process evaluation by occasionally attending their monthly meetings. At the end of study, the
research team provided stool examinations for the study participants and conducted analysis. The
process of public relations and specimen collection require the cooperation from the CAB.
5.2.2.9 Step 8 Impact evaluation
This step assessed goals from the behavioural and environmental sub-objectives developed in step
3. Impact evaluation determined immediate outcome or risk factors of the disease burden. For
example, the planner could rapidly measure the reduction of health-risk behaviours along the
implementation of the program or notice that some environmental factors had been modified to
decrease health burden. While the planner could instantly measure the success of these sub-
objectives outcome, the disease burden might need a longer time to show the effect from the
204
intervention. From the framework, risk factors were controlled, in other words, the behavioural and
environmental modifications theoretically reduced the disease burden.
From social diagnosis in step 1 (Section 5.2.2.1), Koi pla was identified as a potential risk factor for
OV infection. Therefore reducing consumption was assumed to reduce the incidence of the
infection.
5.2.2.10 Step 9 Outcome evaluation
This step measured the ultimate goal of the program by the objectives created in step 2 (Section
5.2.2.2). The final outcome was usually a certain health burden.
However, the success of impact evaluation in step 8 does not definitely guarantee a successful
outcome, or at least show immediate outcome in the program timeframe. In fact, it was more likely
to reduce the disease burden according to theory. Planners should also consider other factors which
could unavoidably interfere with the study’s proposed risk-outcome relationship such as genetic
variability within the population, socioeconomic factors, pre-existing underlying health conditions or
spatial factors. Moreover, some disease burdens need long-term intervention. Therefore, the
sustainability of the program was another key factor considered.
Thus, the success of outcome evaluation was the ultimate achievement for the health program.
However, the planner could evaluate its performance in various aspects. For example, a decline in
risk behaviours or health-related risk environments was an immediate and measurable outcome
while health burden reduction could be a long-term endpoint.
The final outcome was measuring the incidence of OV infection in 2012 at the end of intervention
program and comparing with the intervention group with the control group.
Additionally, the intervention was also evaluated by a qualitative approach. The stakeholders were
the subjects who participated in the intervention, which directly impacted their lifestyle and the
community leaders who were influential to community policy and villager’s mentality.
205
5.3 Results
Results from PRECEDE-PROCEED model process
The community intervention program was processed through the PPM framework which followed
key steps.
5.3.1 Step 1 Social diagnosis
Epidemiological data were successfully communicated to the community; previous baseline
prevalence of OV infection and its risk factors, leading to the establishment of community-based
intervention. Additional data were obtained by qualitative approach indicating social context of
infection in the community.
5.3.1.1 Demographic data
As illustrated Figure 5.7, the population pyramid revealed the structure of Tung-heang population
based on age-class and sex. The 2012 data refer to follow-up population from the 2010 baseline
study; therefore, the pyramid reflected only the population with negative results for OV infection
from the previous prevalence survey. The structure shows a similar pattern to other study areas
from Figure 2.6 in Chapter 2. Age-class 20 – 29 years was relatively constricted compared with other
age-classes.
Figure 5.7 Population pyramid of Tung-heang 2012 data
206
Additionally, the age structure of this follow-up population needed to be compared with the
baseline population because the individuals identified as infected at baseline were excluded.
The population from the 2012 data was allocated to the two studies; community intervention and
risk factor. The subpopulation structure will be presented in the final section.
Table 5.4 Population characteristics of Tung-heang area
Characteristic N %
Sex
Female
Male
180
149
54.7
45.3
Age profile
Mean ± S.D.
Median ± IQR
Min – max
39.3 ± 21.3
42.0 ± 37.0
2 – 88
Age group
(in years, as quartiles)
0 – 19
20 – 42
43 – 56
≥ 57
77
73
77
72
25.8
24.4
25.8
24.1
Occupation
Unemployed
Others
Agriculture
79
33
193
24.0
10.0
58.7
As shown in Table 5.4 shows that the study population comprised more females than males. The
mean age was 39.3 ± 21.3 years and median age was 42.0 ± 37.0 years. The most common
occupation was agriculture-related work.
5.3.1.2 Characteristics of the study populations from qualitative approach
The mean age of the focus group discussion participants was 28 years for females (range: 22 - 65
years) and 38 years for males (range: 30-58 years). All were literate. Most had primary education
and the highest education they obtained was junior high school. The majority of discussion
participants’ occupations were agriculture.
207
From the designated groups according to Table 5.2, no subjects were found to meet criterion 4; new
Koi pla consumption subjects. The results were categorized in two main domains shown below.
1. Precipitating factors to promote Koi pla consumption
2. Precipitating factors to discontinue Koi pla consumption
5.3.1.3 Precipitating factors for promoting uncooked fish consumption
Knowledge of OV infection
At first, a misunderstanding was observed about the threat of eating uncooked fish. Koi pla has been
recognized as a popular dish for a long time. They were still confused about the link between eating
raw fish and becoming infected with OV infection. In fact, the majority of infected cases were
asymptomatic. Their ancestors consumed Koi pla for long time and they looked healthy. Therefore,
they had no reason to go against what their ancestors did as evidenced by the following statements.
“They didn’t teach about it in school, I just followed the elders. They said it’s safe”
“They saw them (the elders) as a model; they seemed alright, so should I”
Some of them were still confused about the source of infection. They were not sure which of the
foods they ate caused liver fluke infection. One of the villagers unconvincingly stated, “It (liver fluke
eggs) may be in shrimps and I ate chopped raw shrimps salad so I got infected. Water vegetables in
the pond can also have liver fluke eggs.”
Some of them were not sure about the route of transmission. They thought that eating any raw
food, skin penetration or drinking water can cause the infection. The following quotes serve as
examples.
“The liver flukes penetrate my skin by not wearing shoes.”
“I can get infected by drinking water.”
“Any type of raw food has liver flukes.”
208
Cultural context
They insisted that the Northeastern people like to eat raw food. It was their lifestyle to eat raw fish,
which was very hard to change. Some of them insisted that this tradition was acceptable.
“I think environment forces us to eat this way. I eat whatever my parents do. It’s inevitable.”
“It’s more than traditional; it’s a part of my lives,”
“It’s my living style that I eat simple dishes in the rice field. Doctors wake up and cook something to
eat. Farmers wake up and go to the field and find something to eat.”
Fish flavour
In the harvest season (dry season), Koi pla was consumed more than other seasons due to the fact
that cyprinoid fish in this season were more delicious. One of infected male insisted, “In the harvest
season, fish was delicious.”
Even though they knew about the threat of eating Koi pla, they just could not deny that they enjoyed
eating it. Some of them voiced the following.
“It was delicious, just simple like that.”
“Better than cooked (laughing)”
“I knew what could happen, but if I die I won’t have a chance to eat it, right?”
Socialisation
Children and adult females chose to consume Koi pla less often than adult males. Usually, adult
females cooked food at home and families had their meals together. Although females did not
prepare uncooked food at home, males who wanted to eat uncooked fish would prepare it
themselves. Koi pla was not consumed daily. However, it was consumed more frequently when
alcohol was consumed, as evidenced by the following quotes.
“When you drink, it would be a great side dish”
“When I got drunk, I just let myself go.”
Some of them believed that mixing spirit beverage (40% alcohol) and lime in the raw fish dish could
kill the liver flukes but some of them, especially females, disagreed.
209
5.3.1.4 Precipitating factors to discontinue Koi pla consumption
Health education campaign
They were asked where they could obtain the information about liver flukes. Some of them
thankfully expressed the following quotes.
“I just knew about liver flukes when the doctor (fieldwork team) came in 2010.”
“Doctors teach us; they showed us scary pictures (of cholangiocarcinoma patient). What would
happen to me if I get that?”
Impact of OV infection to the community
After the community was educated concerning OV infection, some of them learned that several
patients in the village suffered from cholangiocarcinoma. They expressed the following.
“My father passed away with this disease. He asked me for the last time, eat cooked fish.”
“I saw him suffering from cancer, it was really painful”
Emotional effect
Fear of death seemed to be an important factor for them to initiate health concerns and the mutual
agreement as they mentioned below.
“It’s so scary”
“OV was scaring me, and the agreement keeps reminding me what I should do.”
“Nobody forced me to do it (agreement), but it’s about time to change.”
5.3.2 Step 2 Epidemiological diagnosis
Based on primary objective, the study period was 17 months starting in October 2011 and finished in
March 2013. The outcome will be reported in step 9 (Section 5.3.9) as outcome evaluation.
5.3.3 Step 3 Behavioural and environmental diagnosis
Up-to-date incidence and risk factors were estimated from the distribution and risk factors study.
The impact of intervention on Koi pla consumption were reported in step 8 (Section 5.3.8) as impact
evaluation.
210
5.3.4 Step 4 Educational and ecological diagnosis
To achieve behavioural modification and reduce OV infection, three potential factors were explored
to enhance strategic plans.
Predisposing factors:
The current National Control Program, which has been underway for almost 30 years, focused on
prevalence reduction. The main strategies for the control program included stool examination and
treatment of infected cases, a health education campaign to promote cooked fish consumption and
improved hygienic defecation through environmental sanitation development. The strategy already
emphasized knowledge to control the infection.
Enabling factors:
When the Nation Control Program was still active (before 2001), the strategy was active case
detection and treatment. After reaching the goal (prevalence <10%), the service was switched to a
passive strategy.
However, the disease is mainly asymptomatic. Some of cases would suffer from bile duct cancer
from chronic infection. An active strategy was still needed for communities with endemic infection,
but a program similar to the National Control Program required too much human resources and
financial support. However, preventing OV infection would prevent bile duct cancer. Therefore,
intervention in prevention and/or treatment form was mandatory.
Reinforcing factors:
The study aimed to identify reinforcing factors which encouraged infected communities to maintain
safe eating habits on the basis that knowledge about OV infection was available (predisposing
factors), and also health facilities were accessible (enabling factors).
Along with community perceptions, the fieldwork team urged Tung-heang villagers to think over the
burden of the problem. They agreed to initiate the participatory agreement control risk behaviours,
which would result in decreased incidence of OV infection.
The community-derived solution might be an important reinforcing factor for the consistency of the
program in the community.
211
5.3.5 Step 5 Administrative and policy assessment
The national OV infection control program began in 1950 with small groups of high risk areas. After
many studies and trials, large-scale control began from 1984-1987 with nationwide prevalence as
high as 63.6% (28). The control activities involved mobile stool examination teams, promotion of
cooked fish consumption and hygienic defecation, and finally, mass anthelminthic chemotherapy. In
2001, the prevalence was 9.6% (27, 28). Recently, control programs were targeted rather than
national in scale. The activities were prioritized by the degree of the problem in each area. Cross
sectional studies could not identify the newly infected cases so the effectiveness of cooked-fish
consumption promotion campaigns could not be clearly evaluated.
As mentioned before, programs similar to the National Control Program required too much human
resources and financial support for the current passive strategy. Although health expenses were fully
covered by the National Universal Coverage welfare, the local health authority was not equipped
with laboratory and medication to handle OV infection. The regional secondary health care, the
district hospital, equipped to provide full support, was 60 km away.
Multiple studies conducted in Tarkradarn Subdistrict including Na-yao and Tung-heang Areas
contributed field data to gain insights into community-based situations. From understanding the
relationship between host, agent and environment of OV infection, community-based intervention
was designed and initiated. Relevant studies in this area indicated the role of primary prevention.
Focus group discussions also gained information directly from the community members.
Access to health facility
The local health centre could not provide any diagnostic services for the villagers. The community
members also were not concerned with OV infection because they presented no symptoms.
Praziquantel, the drug of choice, was not available in the local health centres and local pharmacies.
Seeking for treatment was difficult. They expressed the following opinions.
“There was no liver fluke drug in the market; there was only general anthelmintic drug.”
212
Attitude toward diagnosis and treatment
The community approach of the fieldwork team provided them more accessibility to diagnosis and
treatment of OV infection. Some of them realized the threat and urged others to avoid contracting
the infection, as evidenced below.
“It must be really important, that’s why doctors come and do lots of work for helping us, and I should
do something about this.”
“Kids love doctors (medical students), they keep reminding us what doctors have said”
They also knew that praziquantel was effective against the infection so they did not question taking
the medicine.
Role of community leaders
In the rural setting, a community leader is a vital key to drive and influence the community’s activity.
It could be either an administrative or spiritual leader; head of village, monk or spiritual leader. The
discussion sessions approached community leaders as important stakeholders to drive the OV
control strategy forward; as discussed below.
“The reduction of OV infection (idea of community intervention) Yes, I were talking about it every
monthly meeting”
“This was a very important problem, it needs immediate response”
“I always told them (the villagers), if I cannot make it, how could I say anything (to the doctors;
fieldwork team)”
213
5.3.6 Step 6 Implementation
Figure 5.8 summarizes the fieldwork under the PPM framework. From 419 eligible subjects with
negative results of OV infection from the 2010 study, 357 subjects were ≥18 years of age and eligible
to enrol in the community trial study. Enrolled subjects were 295 which 62 subjects did not consent
to participate in the community trial. After allocation, 115 subjects allocated in the intervention
group and 180 subjects belonged to the control group. The disproportion between intervention and
control groups was the result of the enrolment process when joining the intervention was voluntary.
419 Baseline population from 2010 cross-sectional study with negative result for opisthorchiasis
357 Age ≥ 18 years 62 Age < 18 years
295 Enrolled to community intervention
study 62 Not consented to
intervention study
304 Enrolled for study of risk factors
115 Intervention group 180 Control group
Figure 5.9
Figure 5.10
Subject allocation
Figure 5.8 Implementation of fieldwork
214
A total of 62 subjects with age <18 were enrolled in the distribution and risk factors study. The
control group from the community trial; that did not receive any intervention, also enrolled in the
risk factors study. Additionally, subjects who did not consent to the community trial were still invited
to join the risk factors study again as if they were willing to participate in the follow-up study. The
enrolled subjects for distribution and risk factors study totalled 304.
5.3.7 Step 7 Process evaluation
5.3.7.1 Summary of study flow and response rate
419 subjects with negative result for opisthorchiasis from 2010 cross-sectional study
124 Excluded
62 Age < 18 years
62 Not consented
295 Eligible and allocated to the study group
115 Allocated to intervention group 180 Allocated to control group
Follow-up at 17 months
1 Withdrew: died or severe illness
29 Loss to follow-up/unable to contact
Follow-up at 17 months
6 Withdrew: died or severe illness
66 Loss to follow-up/unable to contact
93 Completed study
8 Excluded for incomplete
questionnaire
85 Included in final analysis
131 Completed study
23 Excluded for incomplete
questionnaire
108 Included in final analysis
Figure 5.9 Summarized flowchart for community trial study
215
Table 5.5 Response for community trial study
Subjects from 2010 baseline
survey 419
Excluded from the study
Age < 18 years
Not consented
62 (14.8 %)
62 (14.8 %)
Eligible subjects 295 (70.4 %)
Intervention group
N (%)
Control group
N (%) p
Total
N (%)
Subjects allocated to study group 115 180 295
Withdrew: died or severe illness 1 (0.9) 6 (3.3) 0.18 7 (2.4)
Loss to follow-up/unable to
contact 21 (25.2) 43 (36.7) 0.25 87 (29.5)
Completed study
Excluded for incomplete
questionnaire
Eligible for final analysis
93 (80.9)
8 (7.0)
85 (73.9)
131 (72.8)
23 (12.8)
108 (60.0)
0.11
0.11
0.01
224 (75.9)
31 (10.5)
193 (65.4)
Figure 5.9 provides details of study flow for the community trial study. According to Table 5.5,
eligible subjects totaled 295 or 70.4% of the baseline population. Withdrew and lost to follow-up
subjects did not significantly differ between the two study groups. However, after excluding
incomplete questionnaires, which were not eligible for data analysis, the final subjects significantly
differed between both groups. The percentage of response rate in the intervention group was
significantly higher than in the control group (73.9% and 60.0%, respectively, p = 0.01)
216
Table 5.6 Population characteristics of community trial study
Characteristic Intervention
N (%)
Control
N (%)
Total
N (%) p
Sex
Female
Male
44 (51.8)
41 (48.2)
61 (56.5)
47 (43.5)
105 (54.4)
88 (45.6)
0.51
Age profile
Mean ± S.D.
Median ± IQR
Min – max
46.4 ± 15.9
49.0 ± 26.0
18 - 88
51.3 ± 13.0
50.0 ± 19
19 - 84
49.1 ± 14.5
49.0 ± 22.0
18 - 88
0.02
Age group
(in years, as quartiles)
18 – 38
39 – 49
50 – 60
≥ 61
30 (35.3)
14 (16.5)
25 (29.4)
16 (18.8)
20 (18.5)
33 (30.6)
29 (26.9)
26 (24.1)
50 (25.9)
47 (24.4)
54 (28.0)
42 (21.8)
0.02
Occupation
Unemployed
Others
Agriculture
7 (8.5)
12 (14.6)
63(76.8)
4 (4.0)
13 (13.1)
82 (82.8)
11 (6.1)
25 (13.8)
145 (80.1)
0.41
As shown in Table 5.6, the proportion of male and female subjects was not significant for the
intervention and control groups. Mean age and age-group significantly differed between the two
study groups which meant age was significantly higher in the control group compared with the
intervention group. Occupation showed no difference in both study groups.
217
419 subjects with negative result for opisthorchiasis from 2010 cross-sectional study
Excluded 115 subjects ≥ 18 years and allocated to intervention group in intervention study
304 Eligible subjects for risk factor study
62 Age < 18 years
62 Not consented to intervention study
180 Age ≥ 18 years and allocated to control group from intervention study
Follow-up at 17 months
6 Withdrew: died or severe illness
43 Loss to follow-up/unable to contact
248 Enrolled to risk factor study
56 Not consented
199 Completed study
25 Excluded for incomplete questionnaire
174 Included in final analysis
Figure 5.10 Summarized flowchart for distribution and risk factors study
218
Table 5.7 Response for distribution and risk factors study
N %
Subjects from 2010 baseline survey
Excluded for subjected allocated to intervention group
419
115
27.4
Eligible subjects for risk factor study 304 72.6
Excluded to risk factor study
Not consented
56
13.4
Enrolled subjects 248
Withdrew: died or severe illness 6 2.4
Loss to follow-up/unable to contact 56 22.6
Completed study
Excluded for incomplete questionnaire
Eligible for final analysis
199
25
174
80.2
10.1
70.2
As described in Table 5.7, 115 subjects (27.4%) were initially excluded for allocating to the
intervention group in the community trial study resulting in 304 subjects being eligible for this study.
As illustrated in Figure 5.10, 62 subjects with age <18 years, 62 subjects who did not consent to the
community trial and 180 subjects with age ≥18 years and allocated to the control group from
community trial, were enrolled in the study.
After enrolment, 56 subjects or 13.4% of eligible subjects did not consent to participate. Therefore,
248 subjects joined the study. At the end of study, the number of subjects eligible for the risk factors
analysis totalled 174 (or 70.2%).
219
Table 5.8 Population characteristics of distribution and risk factors study
Characteristic N %
Sex
Female
Male
98
76
56.3
43.7
Age profile
Mean ± S.D.
Median ± IQR
Min – max
41.4 ± 21.2
45.0 ± 30.0
2 - 84
Age group
(in years, as quartiles)
0 – 27
28 – 45
46 – 57
≥ 58
44
43
44
43
25.3
24.7
25.3
24.7
Occupation
Unemployed
Others
Agriculture
38
15
109
21.8
8.6
62.6
Table 5.8 shows the population characteristics from risk factor study. The age-group was categorized
into quartiles. There were more females than males. Agriculture-related work was a common
occupation. The mean age was 41.4 ± 21.2 years, median age was 45.0 ± 30.0 years and age ranges
from 2 to 84 years old.
5.3.7.2 Uncooked fish consumption from distribution and risk factors study
Table 5.9 Distribution of uncooked fish consumption
Fish menus N %
Chopped raw fish salad (Koi pla)
No
Yes
143
31
82.2
17.8
Extensively fermented fish (Pla ra)
No
Yes
90
84
51.7
48.3
As shown in Table 5.9, the consumption of uncooked fish showed that Pla ra was consumed in
approximately half of the study population and Koi pla was consumed by 17.8% of the study
participants.
220
Table 5.10 Univariable and multivariable analysis of population characteristics and uncooked fish
consumption behaviours
Characteristic
Koi pla Pla ra
Crude RR
(95% CI)
Adjusted RR
(95% CI)
Crude RR
(95% CI)
Adjusted RR
(95% CI)
Sex
Female
Male
1
1.21 (0.60 – 2.45)
1
1.09 (0.51 – 2.34)
1
1.06 (0.69 – 1.64)
1
0.95 (0.61 – 1.50)
Age group
(years, as quartiles)
0 - 27
28 - 45
46 - 57
≥ 58
1
5.12 (1.12 – 23.35)
5.50 (1.22 – 24.81)
4.09 (0.87 – 19.27)
1
8.97 (0.89 – 90.85)
8.58 (0.86 – 86.02)
5.42 (0.66 – 44.60)
1
2.47 (1.26 – 4.85)
2.00 (1.00 – 4.00)
1.62 (0.79 – 3.34)
1
3.37 (1.03 – 11.11)
2.57 (0.77 – 8.54)
2.05 (0.68 – 6.24)
Occupation
Unemployed
Others
Agriculture
1
0.84 (0.09 – 8.12)
2.67 (0.80 – 8.90)
1
0.17 (0.01 – 2.66)
0.52 (0.78 – 3.43)
1
1.06 (0.37 – 3.00)
1.78 (0.95 – 3.29)
1
0.48 (0.12 – 1.84)
0.76 (0.25 – 2.29)
Table 5.10 reports the univariable and multivariable analysis of population characteristics and
uncooked fish consumption behaviours. The rationale for selection of factors was discussed from
Section 2.3.2 in Chapter 2 and shown in Tables 2.8 and 2.9.
From univariable analysis, age-groups 28 – 45 and 46 – 57 years significantly increased the risk of
consumption of both Koi pla and Pla ra compared with the reference age-group (0-27 years).
However, none of population demographic characteristics; sex, age-group and occupation, showed
any significant association in the multivariable analysis.
221
5.3.7.3 Incidence of OV infection
Incidence rate of OV infection was 9.31 /100 person-years (95% CI: 5.90 – 13.97) from the
distribution and risk factor study (Figure 5.10).
From Figure 5.11, the quartiles age-group indicated that incidence was highest in the 3rd quartile;
age-group 46 – 57 years with incidence of 16.01 /100 person-years (95% CI: 7.68 – 29.43).
(Supplement Table 5-9)
0
5
10
15
20
25
30
35
0 - 27 28 - 45 46 - 57 ≥ 58
Inci
den
ce (
/100
per
son
-yea
r w
irh
95%
CI)
Age group
Tung-heang 2012 cohort study
Figure 5.11 Incidence of opisthorchiasis distributed by age-group
222
5.3.8 Step 8 Impact evaluation
5.3.8.1 Evaluation of risk factors for acquiring OV infection
Table 5.11 Univariable and multivariable analysis of risk factors for acquiring OV infection
Characteristic Crude IRR
(95% CI) P
Adjusted IRR
(95% CI) p
Sex (female as reference)
Male
1.41 (0.62 – 3.19)
0.41
Age group
(years, as quartiles)
0 - 27
28 - 45
46 - 57
≥ 58
1
1.53 (0.26 – 9.19)
5.00 (1.10 – 22.82)
4.09 (0.87 – 19.27)
0.64
0.04
0.08
1
1.21 (0.20 – 7.41)
3.87 (0.82 – 18.20)
3.41 (0.71 – 16.32)
0.84
0.09
0.12
Occupation
Unemployed
Others
Agriculture
1
5.07 (0.93 – 27.66)
2.27 (0.51 – 10.04)
0.1
0.28
Fish menus
Koi pla
Yes
2.97 (1.28 – 6.85)
0.01
2.52 (1.07 – 5.92)
0.03
Pla ra
Yes
1.67 (0.72 – 3.85)
0.23
Table 5.11 shows univariable and multivariable analysis of risk factors for acquiring OV infection
from the follow-up study. From univariable analysis, the relative risk was not significant for males,
occupation and Pla ra consumption compared with the reference. Age-group showed that age 28 –
45 years had a significantly higher risk of infection (IRR = 5.00, 95% CI: 1.10 – 22.82) compared with
reference age-group (0 – 27 years).
Based on Section 2.3.4.2 from Chapter 2, the final multivariable model selected two significant
covariates obtained from current fieldwork study: age and Koi pla consumption. Adjusted IRR for Koi
pla was 2.52 (95% CI: 1.07 – 5.92) and age-group showed no significance for increasing risk
compared with the reference.
.
223
5.3.8.2 Evaluation of impact of community intervention on Koi pla consumption
Result showed that community intervention significantly reduced the incidence of consumption of
Koi pla within the study period; RR = 0.54 (95% CI: 0.29 – 0.99). The intervention could reduce the
consumption by 46% (95% CI: 1 – 71). (Supplement Table S-10)
The qualitative approach from social diagnosis in Section 5.3.1 also provided additional evidence of
how psycho-social factors could influence Koi pla consumption patterns
5.3.9 Step 9 Outcome evaluation
5.3.9.1 Evaluation of impact of community intervention on incidence of OV infection
At the end of study, the incidence of OV infection in the intervention group was 5.13/100 person-
years (95% CI: 1.88 – 11.17) compared with the control group 13.69/100 person-years (95% CI: 8.48
– 20.93). The IRR was 0.37 (95% CI: 0.15 – 0.93) and statistically significant. Thus, the intervention
yielded a 63% (95% CI: 7 – 85) reduction in the incidence of infections. (Supplement Table S-11)
224
5.3.9.2 Qualitative approach for evaluation of community intervention
The impact of intervention was also evaluated by the qualitative approach. Most of them expressed
that the agreement was a good policy. They were keeping in mind that they needed to avoid those
risks of infection; consuming raw fish especially Koi pla, adequate sanitation or how to manage
kitchen waste. They perceived that the aim of agreement was for their own health. Some expressed
the opinions below.
“It keeps us alert, to be aware of eating raw fish. It’s good for you, not anyone else. I think it’s a good
project. Everyone recognizes it and so do I. That’s the way I’ve tried to control myself. ”
“I mean yes, I volunteered. Of course sometime it makes me feel uncomfortable. But in the end it’s
good for me at all.”
“I think it’s about the process. You commit to do something, something very important. Not just for
yourself, also community. Look at your family, your neighbors, everyone concerned about it
(agreement). It’s about how to change or modify your life. It’s challenging.”
“It’s like you make a promise and you want to keep it. If you don’t want to keep it at the first place,
you wouldn’t commit anything like this (laugh)”
225
5.4 Discussion
5.4.1 Study design and response
The Tung-heang fieldwork comprised two studies: a community trial and follow-up study. The
community trial assessed the effectiveness of community-based intervention using the PRECEDE-
PROCEED framework and the follow-up study assessed the relationship of risk factors and OV
infection.
Considering the study design, the method resembled that in the other two areas; Na-ngam and Na-
yao Villages. After exclusions for unmatched and incomplete questionnaires, the final response rate
was 70.2%, significantly higher (p <0.001) than the Na-ngam (53.1 %) and Na-yao study (36.1 %). It
was assumed that incorporating the control group from the community trial would result in a higher
response rate. However, the study population was considered low: 248 enrolled subjects and 174
subjects at the end of study due to the size of Tung-heang community.
The Tung-heang cohort design highlighted the same issues as found in the other two areas (Section
2.4.1 from Chapter 2). Since the consumption of Koi pla was collected retrospectively by
questionnaires, potential bias included recall bias and social desirability bias (147, 148).
Recall bias can occur when the study subjects have to recall a prior exposure differently depending
on the disease status (148, 149). Recall bias is a concerned issue for case-control design when the
disease outcome is addressed first and then the corresponding exposures are assessed
retrospectively from past memory (124). Reporting of exposures might be overestimated or
underestimated depend on participants’ concern, for example, cases might pay more attention on
assessing risk factors they believe would be the cause of their illness. The impact of recall bias can
result in differential misclassification.
The study design used in this chapter included a community trial and follow-up study. The outcome
was prospectively determined when stool specimens were examined at the end of the study for OV
infection diagnosis. However, the uncooked fish consumption behaviors regarded as disease
exposure were determined retrospectively. The study participants provided the details of their
consumption habits by answering the self-administered questionnaire at the same time as providing
stool specimens. From the study method, the data collection method was designed concerning the
recall bias issue as described below.
Stool specimen and questionnaires were collected at the same time at the end of study period
when fieldwork was conducted. Therefore, the stool examination process was ready for
226
reporting the result to the participants after collecting the questionnaire so they were not aware
of their infection status during the questionnaire survey. Moreover, the study participants
comprised a follow-up population and were preliminary negative for OV infection from the
previous baseline cross-sectional study. Therefore, they were less likely to be influenced by the
stool examination result for giving details of consumption behaviors.
Based on the questionnaires, the study participants needed to provide details whether they ever
consumed Koi pla or Pla ra during the study period. The quantity of food and frequency of
consumption were not determined, which could reduce the confusion to provide recalled
details.
However, survey questions about sensitive issues could lead to misreporting of the variables of
interest resulting in social desirability bias.
Generally, individuals tended to preserve their image within the social norm (150). They would try to
avoid the negative impact from social interaction such as negative criticism, feeling of shame or
embarrassment resulting in a state of insecurity and loss of self-confidence. Psychologically, the
defense mechanisms will respond to maintain social desirability including lying or manipulating data
in any way to avoid such negative feedback (151). In the research field, social desirability bias will be
likely to appear when the study issues are considered sensitive such as sexual, racial or religious-
related issues (150). Therefore, misreporting the variable of interest could impact the result of the
study. Psychological studies indicate that misreporting on sensitive questions in a social desirable
way is a result from voluntary process under responder’s consciousness (152, 153). The responders
are more likely to answer the question with a self-controlled process than a subconsciously
autonomic mental response. Therefore, data collection method could be designed to reduce this
bias.
In this study, the association of Koi pla consumption and OV infection was communicated to the
community when Koi pla was identified as a potential health threat. The data collecting method
aimed to reduce recall bias by reporting study results after collecting questionnaires. However, the
undesired outcome of Koi pla consumption might still affect the answers provided in the
questionnaire for consumption behavior history.
From qualitative data from Chapter 4, Koi pla consumption was still a common practice in the
community with regard to cultural background. The study participants acknowledged Koi pla as a
health risk from the research team. Therefore, some of them might provide the answer in a
desirable way to the research team as they perceived that the team comprised healthcare
personnel.
227
Considering social desirability bias, the community trial protocols were designed to reduce the bias
by the methods stated below.
The confidentiality of the data of study participants was preserved through the entire process.
The questionnaire and stool specimen were processed under a code-embedded system. The
research protocol administrator, specimen examiner and data analyzer were different persons.
The information was used with regard to the whole group. Names and personal data were not
individually identified in the research. Stool result was reported to each participant in person
with a sealed envelope.
For the qualitative approach, each group in the focus-group discussion was categorized based on
same Koi pla consumption pattern to observe the direction of opinions from each particular
eating practice. The homogeneity in each group would let the participants express their opinions
more freely. For in-depth interview, a special session was conducted in a private setting with the
responder’s consent. The interviewer always introduced the session with the academic benefits
from the participant’s cooperation. The provided answers would contribute scientific benefit
and would not interfere with regular healthcare. Moreover, they would not lose any benefit
when they were unwilling to join the study.
The data of Koi pla consumption was collected at the end of the study. Study participants were
informed that their identity remained anonymous during data analysis. The self-administered
questionnaire would decrease concern about admitting to Koi pla consumption (154, 155).
The given principle was also applied to all part of the fieldwork involving data collection on
uncooked fish consumption.
The issues of recall bias and social desirability bias could be improved in future studies. The
prospective data collection on variables of interest could reduce the effort to recall memory. To
avoid answering sensitive questions about uncooked fish consumption, a food diary could be
proposed to collect daily food intake in a more general way, which could reduce the frustration of
answering any specific food item (154). The method could also collect the amount and frequency of
a particular food type as well.
The stool examination comprised three methods to improve sensitivity. However, the
microscopically-based methods might reduce specificity because minute intestinal fluke (MIF) eggs
were very similar to liver fluke eggs. Moreover, co-infection of MIF and OV (156) have been reported
in Thailand. However, MIF was not found in this region by PCR technique (46). Based on cohort
methods, only negative cases from the baseline study were enrolled in the follow-up study.
228
Therefore, the incidence might have been underestimated due to the high-risk group; previously
infected cases were not enrolled in the study.
The community trial was conducted to assess the community-based intervention based on open-
label, nonrandomized method due to the design of the intervention. The intervention was a
community mutual agreement where the study participants agreed to follow instructions
emphasizing behavioral modification with respect to risk factors. Therefore, the participants
consented and volunteered to join the intervention. Thus, the allocation method was not
randomized. The CAB coordinated between study participants and research team. The control group
only agreed that they would provide a stool specimen and answer a questionnaire at the end of the
follow-up period. Therefore, they were eligible for the study of risk factors.
Considering the intervention protocol and method for study allocation, the design was open-label
because the study group and investigator were not blinded. However, the specimen was blinded to
examiners during stool examination process to avoid bias toward reporting results. Age-structure
significantly differed between study groups (Table 5.6) which meant age for the control group was
higher than the intervention group. Subjects that withdrew or were lost to follow-up and incomplete
questionnaire subjects were not significant in the two groups. However, the final response rate was
significantly higher in the intervention group. Probably, the lower loss to follow-up and incomplete
questionnaire proportion in the intervention group resulted because the study participants
volunteered; therefore, they might be more likely to provide cooperation to the study.
An interesting issue arose for allocating the study population in the qualitative approach with regard
to patterns of Koi pla consumption. Within the 17 months of study period, no individuals, who never
consumed Koi pla before, choose to have it during study period. Hence, the factor to promote Koi
pla consumption was obtained through individuals who sustained their Koi pla consumption
behaviors.
5.4.2 Situation of OV infection
Two surveys were conducted on OV infection in Tung-heang area in 2008 and 2010. However, no risk
factors were assessed. From the fieldwork database, the prevalence of OV infection was 10.38%
(95% CI: 7.85 – 13.39) in 2008 and 11.42% (95% CI: 8.69 – 14.63) in 2010. However, the data were
not yet published. Risk factors were evaluated in this follow-up study. Koi pla consumption was
consistently identified as a risk factor for contributing OV infection in adjacent areas including Na-
ngam, Na-yao and Na-isarn (Table 2.10 – 2.12 from Chapter 2).
229
5.4.2.1 Situation of Koi pla consumption
As shown in Table 5.9, Koi pla was consumed by 17.8% of the follow-up population. Based on the
questionnaire, individuals were asked whether they consumed Koi pla at least once during the
follow-up period to evaluate the relationship of Koi pla consumption and OV infection. It was
observed that the rate of consumption was lower than the other 4 areas (Section 2.3.2 from Chapter
2); 39.8 – 44.1 %. However, Koi pla consumption was 25.9% in the control group from the
community trial.
Univariable analysis showed that age-group significantly increased the risk of Koi pla consumption
compared with the reference age (Table 5.10). However, multivariable analysis did not show any
significant population characteristics on Koi pla consumption. However, the confidence interval for
each age-group was wide, which could have resulted from the small sample size (174 subjects from
Table 5.7).
The qualitative approach could provide more details about Koi pla consumption patterns. Classifying
consumption patterns in 5 different groups (Table 5.2) provided more insights of how the behavior
would change over time with respect to underlying causes.
Precipitating factors for continuing Koi pla consumption emphasized the cultural context of the
behavior. Consumption of Koi pla itself did not provide any health benefit and the dish was not a
regular dish for daily living. The potential of Koi pla was an underlying core of local norms and
traditions. Male always consumed Koi pla in terms of socialization, while females accepted it as a
norm and children learned that was pathway to growth in the community.
However, eating habits were dynamic and could be potentially influenced by many factors. Many
individuals stopped eating Koi pla during the study period. The interesting issue was that initiation of
community-based intervention derived from the members of the community. The mutual
agreement might have acted as a new type of social norm for the study participants, to be discussed
in Section 5.4.3.1.
5.4.2.2 Incidence of OV infection
From the follow-up study, the incidence of OV infection was 9.31/100 person-years (95% CI: 5.90 –
13.97), higher than both national and regional incidence (11, 30, 69, 70). Compared with other
follow-up studies in this area, the incidence of OV infection was 7.98/100 person-years (95% CI: 5.49
– 11.20) in the Na-yao 2013 study and 6.80/100 person-years (95% CI: 4.68 – 9.54) in the Na-ngam
2014 study. Tung-heang incidence was significantly higher than other two study areas (p <0.05).
230
Multivariable analysis indicated that Koi pla consumption significantly increased the risk of infection
with IRR = 2.52 (95% CI: 1.28 – 6.85) when adjusted for age-group (Table 5.11). Age-group showed
no significant association with the infection. Considering the age-group pattern in the Na-yao 2013
and Na-ngam 2014 follow-up studies, no age-pattern was observed as well.
From the qualitative approach, newly-infected cases were likely derived from individuals with
constant Koi pla consumption. From the study method, the questionnaire aimed to determine Koi
pla consumption during the study period. Therefore, long-term consumption was not assessed using
this method. Based on assumptions from Chapter 3, OV infection needed sustained re-infection to
maintain its endemicity. According to the mode of infection, uncooked fish should be constantly
consumed as well. Study on factors shaping consumption patterns could play an important role for
the understanding of infection dynamic with regard to risk behaviors.
5.4.3 Community intervention
5.4.3.1 Impact of intervention on Koi pla consumption
From the National Control Program (28), the strategy to interrupt transmission of the infection was
avoiding uncooked fish consumption. The problem was that the villagers could not avoid all types of
dishes, especially extensively fermented fish (Pla ra), which was the main ingredient for I-sarn
(northeastern) dishes. Pla ra was regarded as a cultural symbol for Thailand (69). Individuals might
interpret the campaign to reduce the consumption of unspecific uncooked fish as saying they have
to stop eating Pla ra as well as Koi pla. Quantitative data revealed that Pla ra was more popular than
Koi pla in terms of incidence of consumption (48.3% and 17.8%, respectively). Many studies showed
that metacercariae, the infective larva, could not survive in some particular environments. It
definitely survives in fresh fish meat, so consuming Koi pla would biologically increase risk infection
with metacercariae, while metacercariae degenerate after two days of preservation. For extensive
Pla ra, the fish is preserved in a highly concentrated salt solution for 3 – 6 months, which was not
suitable for the metacercariae to survive (87, 88) . The results show that of the fish dishes
considered, only Koi pla was an independent risk factor for acquiring OV infection. This could lead to
the design of a focused control strategy, especially combined with the bio-psycho-social data from
the qualitative analysis.
To deal with the situation, the available quantitative data may not be sufficient to establish an
effective health policy to control OV infection. The problems involve every aspect not only the
incidence rate and the risk factors obtained from the study, but also important bio-psycho-social
aspects.
231
Based on the PPM framework, the reasons why Koi pla was still popular; attitudes, knowledge, or
even ideas of conducting a solution were then obtained from the qualitative approach based on the
community’s own thinking. The framework allowed us to acknowledge the problems in various
aspects. Results showed that both quantitative and qualitative data were consistent; Koi pla
consumption was a main risk factor for contributing to OV infection. Study outcomes highlighted
that particular uncooked fish dishes were traditional and regarded as tasty. Changing or modifying
eating habits probably pose a large burden for their lives. Some perceived that despite consuming
uncooked fish dishes for long periods, they were still healthy and sometimes stool examination came
with negative results even when they were classified as a high-risk person for acquiring the infection.
Thus, little incentive existed to change their eating habits.
Poor knowledge or insufficient information of disease risk and routes of transmission were obstacles
for behavior change among the study participants. They required adequate knowledge from easily
understandable messages to clarify the risk of acquiring OV infection. Social and traditional cultural
factors were important to explain Koi pla consumption. The risk and benefit from eating Koi pla
could be demonstrated as they should eventually perceive problems themselves and if successful, a
sustainable solution would come from the community itself.
The intervention program should have had a substantial impact on reducing Koi pla consumption,
because it did not target uncooked fish consumption generally but used the precise term “Koi pla” in
the community (Table 5.1). Therefore, they would not be confused with other foods, such as Pla ra.
This made modification of their eating behaviours much easier.
The impact evaluation from step 8 (Section 5.3.8.2) indicated that the intervention could significantly
reduce Koi pla consumption by 46% (p = 0.04) within the study period. The outcome highlights the
potential effectiveness of the primary prevention strategy. However, the non-random allocation
process should be concerned as it could potentially produce selection bias.
Educational and ecological diagnosis from step 4 (Section 5.3.4) indicated that reinforcing factors
were a key point for the community to maintain consistency of the intervention. A community-
derived solution could play an important role to reinforce the community agreement to reduce OV
infection. The CAB, along with the role of local health volunteers, acted as a key role for
implementing the intervention at the community level with minimal assistance from outsiders such
as the research team. However, outcomes of intervention still need evaluation. Microscopic-based
stool examination requires technical skill, but can be conducted in an ambulatory setting or local
health facility.
232
5.4.3.2 Impact of intervention on incidence of OV infection
The outcome evaluation from Section 5.3.9.1 reported that the intervention could significantly
reduce incidence of OV infection by 63% (95 % CI: 7 – 85, p = 0.03) within the study period.
Moreover, the administrative and policy assessment in step 5 (Section 5.3.5) identified some
interesting issues regarding the control strategy. Improving diagnosis and treatment at the local
health centre should be implemented under the national strategy because so many people
originating from Northeastern Thailand have dispersed around the country like in this community.
Stocking praziquantel at the local pharmacy might encourage the community members to buy it
themselves and take it whenever they consumed uncooked fish.
The intervention group from community trials showed positive attitudes toward the intervention.
Some of them mentioned that even though they could not abruptly stop eating Koi pla, the
consumption eventually became less frequent. Further studies may need to assess frequency of
consumption. Moreover, the burden of OV infection could be evaluated by infection intensity as well
to monitor the decline of severity in addition to curing the infection (as described in Section 2.4.2.3
from Chapter 2).
233
5.4.4 Implication of PRECEDE – PROCEED model framework
Figure 5.12 shows how the PPM framework integrated the conceptual framework of the study. With
the aid of mixed-method study designs, quantitative and qualitative fieldwork provided results in
various aspects highlighting that complex structure regarding infection dynamic could be holistically
approached by systematic framework as described below.
Step 1: evaluate the burden of OV infection through the prevalence and incidence of the
infection. Qualitative approach visualized the social influence affecting infection dynamics. This
was a vital step to identify the root cause of the problem.
Step 2: setup the objective to reduce OV infection. From the framework, reducing the
prevalence of OV infection was directly involved with diagnosis and treatment, which is
considered a secondary prevention. Considering the reduction of incidence of the infection, the
intervention should be applied through risk behaviours to interrupt the mode of infection. By
emphasizing primary prevention, the objective was to reduce incidence of infection.
Opisthorchiasis Risk behaviours
Social influence
Intervention
Prevention Diagnosis and Treatment
6
4
1
5
1
2 3
9
7
8
Figure 5.12 PRECEDE – PROCEED model integrated to conceptual framework
234
Step 3: setup the sub-objective to reduce risk behaviours. From studies conducted in this area,
Koi pla consumption was identified as potential risk behaviour (Chapter 2). The qualitative
approach (Chapter 3) also suggests that the intervention should specify precise action such as
“reduced Koi pla consumption” (Table 5.1) to better modify behaviour.
Step 4: setup community intervention (Table 5.1) focusing on reduction of Koi pla consumption,
which could eventually reduce the incidence of OV infection. The crucial step of this process was
that the intervention was developed and initiated by the community. Moreover, the community
members also hold responsibility to manage the flow of intervention with the support of the
research team for the study design and outcome evaluation. The concept of community
intervention was based on three important factors: predisposing, reinforcing and enabling
factors.
Step 5: current strategy and resource with respect to National policy were assessed to evaluate
the feasibility and compatibility of community intervention. The designated intervention should
be compatible with existing policy and also adapted to community characteristics. This was a
challenging step as the intervention could be applied to the community. From step 4, the aim of
reducing Koi pla consumption was based on the National Control guidelines (28), but the details
have been modified to achieve more specific goals; precise intervention was also supported by
the evidence-based mixed-method study conducted in multiple areas.
Step 6: the fieldwork was designed and conducted to achieve objectives and sub-objectives with
the mixed-method design. The community trial was used to evaluate the community
intervention for reducing incidence of OV infection and Koi pla consumption. The follow-up
study served as an additional study to evaluate the distribution and risk factor for up-to-date
information. The qualitative approach could also access data in many steps to describe the
details of given issues.
Step 7: the flow of the whole study was summarized (Table 5.3) to ensure the completeness of
fieldwork. The CAB was a vital key for managing the study flow. The initiation of intervention and
cooperation from community members reflected the success of community-based action against
the health problem.
Step 8: the impact of intervention on Koi pla consumption was evaluated from the community
trial. Up-to-date results of risk factors contributing the infection were obtained from the follow-
up study.
Step 9: the impact of intervention on the incidence of OV infection was evaluated from the
community trial.
235
Following the PPM framework, community participatory action could play an important role as a
reinforcing factor as the intervention was developed within the community environment which was
considered adaptive and practically suited to the villagers’ lifestyles when consumption behaviours
were complex involving traditional beliefs, attitudes and cultural backgrounds. To initiate
behavioural modification, strict law enforcement may not be available and probably not suitable for
such a public health campaign.
It appears that the National Control Program effectively reduced OV infection prevalence (26-28),
but failed to suppress the incidence of disease sustainably (69, 70). Seeking cases and providing
treatment were more likely an effective secondary prevention. The PPM framework allowed us to
perceive and understand the weak points of the public health campaign. Reinforcing factors were
crucial to sustain health campaign continuity.
236
5.5 Conclusion
From the fieldwork study, the mixed-method design provided comprehensive approaches to assess
OV infection in various aspects. The PRECEDE-PROCEED model framework also provided
comprehensive approaches for intervention/evaluation planning.
Although this project was a pilot study to assess the community in small scale, the study process and
results provided insights into community management to support expansion into a larger scale in
which CIP (Community Intervention Package) could be developed and applied on a national scale.
Providing adequate knowledge of OV infection and improving health care facilities would offer the
best alternatives for disease control at the community level. Community participatory action was
considered important for self-sustained activity. In a resource-limited setting, primary prevention
was promising for a public health campaign.
237
Chapter 6 Discussion
6.1 Overview
This project aimed to comprehensively study the infection dynamics of OV infection using multiple
tools including statistical model, mathematical model and qualitative approach to explore the
infection in various aspects.
Study results then could be used to develop community interventions using the framework of the
planning model. The planning model was based on integrated outcomes from the comprehensive
study with respect to a rural community setting which reflects the common characteristics of rural
Thailand.
As shown in Figure 6.1, the conceptual framework was integrated with comprehensive results from
the study with respect to the epidemiological triangle relationships of host, agent and environment.
[Host]
Risk factors
[Environment]
Social influence
[Agent]
Opisthorchiasis
Intervention
Prevention Diagnosis and Treatment
Planning
Evaluation
Figure 6.1 Integrated conceptual framework from comprehensive study
238
Study outcomes would aid the developing of intervention planning and lead to design and
implementation of interventions in the evaluation phase. The project explored the fundamental
knowledge and background of OV infection and built up consolidated information to formulate a
community intervention package.
From the planning phase, the relationship between risk factors, OV infection and social influences
were explored reflecting the host-agent-environment interaction.
The study of host referred to the population characteristics and factors determining the risk
behaviours, which could lead to acquiring the infection. The population structure was visualized by a
population pyramid to examine the pattern of age with respect to age-class.
The results from Chapter 2 (Figure 2.7) showed that the population pyramid from 4 study areas was
considered a constricted pattern with the relatively small proportion of age-class 20 – 29 years. The
qualitative approach supported the underlying reasons that many individuals in their 20s
temporarily move out to seek jobs in an urban setting during the off-harvest season. This is a
common characteristic of Thai rural villages where the primary jobs are agriculture-related with
part-time jobs during off-season (70).
The study of agent referred to OV infection caused by Opisthorchis viverrini. This project assessed
the epidemiology of the infection including prevalence and incidence by extensive fieldwork. The
infection dynamics were explored through a mathematical model highlighting the complicated
relationship between prevalence and incidence.
Environmental factors were considered to have important roles in shaping the infection dynamics.
OV infection was known to be locally specific to each area (43). This project emphasized the study of
the social factors which influenced the risk behaviours. The study examined the advantages and
limitations of the National Health Policy with regard to controlling OV infection through health
campaigns proposed by health authorities for the community. The Long-term National Control
Program has successfully reduced the prevalence of OV infection as national goal. A new challenging
issue has emerged from this project when the study site selected was a non-endemic area with high
incidence of OV infection.
239
The interaction between factors was assessed by particular study methods described below.
6.1.1 Host and agent relationship
In Chapter 2, prevalence and incidence of OV infection were extensively examined through cross-
sectional and follow-up studies within 3 study areas. Regarding risk behaviour; Koi pla consumption
was determined to increase the risk of acquiring infection and also increase the intensity of the
infection. For infection intensity, the age-prevalence-intensity relationship indicated that the age-
structure component within the intensity and the level of parasite aggregation interacted to where
the intensity tended to increase and the parasites were more equally spread based on the
assumption that incidence was uniform at all ages. Over-dispersion was observed for the infection
intensity, and a two-part model was introduced to handle over-dispersed and zero-excess data.
From Chapter 3, mathematical modelling was also formulated to explore the relationship between
prevalence and incidence of the infection. The model was fitted to serial prevalence obtained from
the Na-yao area to project the longitudinal trend of infection dynamics. The treatment effectiveness
of praziquantel was also estimated from both mathematical models; 80.1 – 92.5% and indirect
calculation from individual-level fieldwork data; 88.7 – 92.2 %. Additionally, the model was fitted to
the prevalence trend of OV infection under the National Control Program.
From Chapter 4, the qualitative approach was conducted to explore the social dynamics which
provided important underlying reasons behind the practice of risk behaviours with regard to
uncooked fish consumption. The social component was considered an important part for infection
dynamics with the influence of knowledge, attitudes and perceptions.
6.1.2 Host and environment relationship
The qualitative approach from Chapter 4 described the relationship between host and environment
in terms of risk behaviours and social influence with group discussion conducted in Chapters 4 and 5
based on the age generation and pattern of Koi pla consumption.
Age structure, which was classified in age generations, played an important role for the underlying
attitudes and thinking processes toward OV infection and its risk behaviours. The qualitative study
investigated how the knowledge transfer and cultural transfer interacted within particular
generations.
From Chapter 2, the impact of sex, age-group and occupation on uncooked fish consumption; Koi pla
and Pla ra were evaluated by statistical modelling. Male sex and increasing age-group significantly
240
increased the risk for Koi pla consumption and agriculture-related work also increased the risk of Pla
ra consumption.
The underlying social-influence details were additionally explained by focus group discussions and
in-depth interviews. Male-associated environments might have influenced the practice of Koi pla
consumption; fish were usually caught and instantly prepared during their agricultural fieldwork.
Male social drinking was also associated with Koi pla consumption by qualitative approach. Age was
considered a potential factor; older age population showed a picture of cultural attachment in which
Koi pla represented the original Northeastern tradition.
Pla ra provided different perspectives of uncooked fish consumption. Biologically, metacercariae
could not survive in Pla ra because of its high salt concentration but the pattern of consumption
practice could influence other uncooked fish dishes. Pla ra was considered the main ingredient in
most local dishes; therefore, any campaigns to avoid all uncooked fish consumption might be too
broad to target the actual risk behaviour, and therefore, it might be problematic for behavioural
modification.
6.1.3 Agent and environment relationship
OV infection itself had an impact on the community which can be evaluated from Chapter 4. The
study results were transferred to the community which triggered some concerns and lead to
subsequent action as described in Chapter 6 for the community intervention.
Community background and social structure precipitate and limit the burden of OV infection as
described in Chapter 4.
6.1.4 Development of planning model
Figure 6.1 briefly describes the interaction of host-agent-environment discussed in Section 6.1.1 to
6.1.3. A planning model integrated the study results from Chapters 2 – 4 to engage the disease
burden at the community level under PPM.
The community trial initiated the community intervention package based on the fieldwork data
obtained from both quantitative and qualitative studies. The comprehensive data supported the
development of a planning model with community participation and considered the importance of
background especially population characteristics and social components.
The intervention was built based on community perspectives reflecting on the current situation of
OV infection. As the community members perceived and acknowledged that OV infection was a
241
burden, they set up a goal to reduce the incidence of OV infection using behavioural modification to
reduce Koi pla consumption. The planning phase was followed by the evaluation phase where the
intervention was implemented and the outcomes were evaluated.
The PPM also assessed the advantages and disadvantages of the current health policy to justify and
adapt the proposed intervention to be more feasible and practical in the community setting.
Therefore, Chapter 5 provides an integrated method to use comprehensive data obtained from
fieldwork and analysed with rigorous statistical methods, then applied the community intervention
package to the actual public health setting.
242
6.2 Summary of study findings
6.2.1 Epidemiology of Opisthorchis viverrini infection: distribution and risk factors
Chapter 2 highlighted the potential of OV infection epidemiology concerning distribution and risk
factors, within the rural community setting regarded as a majority part of Thailand. Regarding the
well-preserved Northeastern environment, cultural background could play a major role in continuing
risk behaviours, which could potentially maintain a higher prevalence in this non-endemic area.
This Chapter aimed to explore the epidemiology of OV infection for better understanding of the
infection in various aspects. The main findings showed that prevalence and incidence were relatively
high considering the study area was non-endemic. The prevalence of OV infection was 7.85% (95%
CI: 4.96 – 10.75) in Na-isarn 2013 and 9.29% (95% CI: 6.64 – 11.93) in the Na-ngam 2012 study. The
incidence rate was 7.98/100 person-years (95% CI: 5.49 – 11.20) in Na-yao 2013 and 6.80/100
person-years (95% CI: 4.68 – 9.54) in the Na-ngam 2014 study. Compared with the regional average,
the prevalence of OV infection in central Thailand was (3.8%) (28) and the incidence has never been
officially published before in this region (9). The population demography is linked to the
Northeastern background, their ancestors’ origin.
Consumption of Koi pla was determined to present a potential risk factor for acquiring the infection.
Data from the two follow-up studies and additional supplemental data from another two follow-up
studies (69, 70) provided rational supporting evidence for the causal relationship between Koi pla
and OV infection.
Infection intensity showed an over-dispersed distribution of egg count outputs suggesting the use of
negative binomial distribution for count outcomes. The age-prevalence-intensity relationship
indicated that an age-structure component within the intensity and level of parasite aggregation.
Assuming incidence was uniform at all ages, the intensity tended to increase and the parasites were
more equally distributed with increasing age within the population.
The compound model including zero-inflated-based models and hurdle-based models was
introduced to handle the zero excess and over-dispersed data. The model provided both infection
probability and infection intensity, which was applicable for the data interpretation. The model
result reported the risk of being infected from identified factors and also the level of infection
intensity. From the study, the negative binomial logit hurdle model (NBLH) was the model of choice
indicating increasing risk of OV infection from Koi pla consumption as observed in prevalence and
cohort data analysis. A trend was observed among age groups in NBLH model where the risk of
243
acquiring infection increased with age when compared with the reference age group. Moreover,
consumption of Koi pla and the 4th quartile age-group also significantly increased infection intensity.
6.2.2 Infection dynamics of OV infection
Chapter 3 showed that re-infection played an important role for sustaining the chronic infection
picture and shaping a sustained prevalence. Praziquantel was estimated to show the same
effectiveness as previous fieldworks at 80 – 95 % (26, 40, 49).
From Chapter 2, the incidence of OV infection can be estimated from catalytic function of prevalence
data using linear regression modelling. However, the incidence directly estimated from the cohort
fieldwork was significantly higher than the fitted incidence obtained from the regression model.
Therefore, the infection dynamics would need a more complex framework with respect to host –
agent – environment.
The inconsistency of the given relationship was further explored in Chapter 3. My aim was to explore
the infection dynamics with regard to complicated relationships between prevalence and incidence
from the fieldwork using a more complex framework which emphasizing the human population
dynamics. Chapter 3 also evaluated the effectiveness of praziquantel treatment on infection
dynamics and applying the findings to the National Control Program data.
The model was based on the classic susceptible – infectious – recovered (SIR) model. The main
extension was to add a second susceptible class so there are primary susceptible individuals (S1) who
will be infected for the first time. Infected individuals will be naturally recovered for some period
and become secondary susceptible (S2) again for the re-infection with reduced infectivity. The model
assumed that infected individuals who recover because of praziquantel treatment will gain longer
immunity before potentially becoming re-infected. The force of infection is assumed to be uniform
and the recovery rate depends on the parasite life expectancy.
The model showed some key data from the fieldwork. The model assumes that the infection is in the
endemic phase where Na-yao 2002 prevalence is 22.9 % (95% CI: 19.3 – 26.5) and Na-ngam 2012 is
9.3 % (95% CI: 6.9 – 11.6). I estimated that the average duration of infectiousness was in the range
1.51 – 1.81 years. Basic reproductive numbers (R0) were estimated from the model. Finally, the
model is used to assess the effectiveness of the National Control Program. R0 for Na-yao is 2.74
(95% CI: 2.61 – 2.89) and for Na-ngam is 2.30 (95% CI: 2.23 – 2.40).
244
The effectiveness of praziquantel treatment is estimated from Na-yao study where multiple surveys
have been conducted. The treatment effectiveness, as cure rate of praziquantel is 80.1 – 92.5 %
from the mathematical model. An alternative calculation of from individual-level fieldwork data
provided a similar estimate of 88.7 – 92.2 % of treatment effectiveness.
The model, fitted to the prevalence data from the National Control Program from 1984 – 2001,
revealed that the incidence was estimated to be 9.8 – 47.3/100 person-years by assuming that
incidence was uniform with praziquantel effectiveness of 80 – 100%.
6.2.3 Evaluation of OV infection and risk behaviours using qualitative approach
Chapter 4 showed that social influence was a potential component for infection dynamics of OV
infection. The interaction of knowledge and culture transfer across generations provided more
insights to the bio-psycho-social dynamics.
The infection dynamics of OV infection were assessed through the qualitative approach using focus
group discussions and in-depth interviews. Discussion themes were directed by pre-defined topics
regarding the current quantitative data and reviewed literature to develop the understanding of OV
infection and related health risks. Framework analysis was mainly used to explore the association
between factors from thematic content extracted from gathered data. Additionally, grounded theory
(129) was used when new ideas were generated from the analysis.
The prevention campaign, based on the National Control Program, targets uncooked fish
consumption broadly. The locals have found some difficulty in avoiding all uncooked fish dishes and
as a result often maintained the same eating habits. The outcome suggested that the campaign
should emphasize the risks associated with specific items and should be supported by evidence-
based data.
Koi pla remained a popular dish in the community as the dish itself represents Northeastern culture.
The cultural norm had been transferred from their ancestors who originally moved from the
Northeastern region around 4 decades ago. Consumption of Koi pla, which was not a regular eating
habit for central Thais, reflected a potential cultural attachment.
From Chapter 2, the NBLH model indicated the potential risk of the increasing age-group for
acquiring OV infection. The qualitative approach revealed that age-generation also influenced the
process of knowledge transfer and cultural transfer in the matter related with OV infection and risk
behaviours. Knowledge of OV infection can be easily distributed to younger generations by formal
education and social networks when they can pass this knowledge to older generations through
245
their family relationship. However, the knowledge-transfer scheme will confront the cultural
transfer, which was conversely passed from older generations. This interaction has an important
impact on the risk behaviours; therefore, it was crucial for the intervention strategy.
For rural settings, the community leader is an important key in the social structure. As centralized
authority might not be aware of local cultural norms, community leaders can help in filling this gap in
terms of communication. Local health volunteers played an important role for the interaction
between villagers and health authorities. They could act as community leaders and as facilitators for
the health intervention. Because they were locals, they could more easily gain trust and cooperation
from the villagers, which were a crucial part to maintain the sustainability of health campaigns in the
community.
246
6.2.4 Community-based intervention: PRECEDE-PROCEED model framework for controlling
OV infection
Chapter 5 showed that the mixed-method study design provided comprehensive approaches to
assess OV infection in various aspects. The PRECEDE-PROCEED model framework also provided a
comprehensive approach for the intervention/evaluation planning. The project initiative provided
more insights of community management, which aimed to expand to a larger scale in which the
community intervention package could be developed and applied for public health implementation
on a national scale.
From Chapters 2 – 4, extensive infection dynamics data were obtained from statistical modelling,
mathematical modelling and also the qualitative approach. In Chapter 5, the aim was to construct an
intervention in the community using a public health framework supported by study data.
The PPM framework was applied in the community to systematically engage the problem. The
framework consisted of two major phases; planning and evaluation. The planning phase aimed to
setup goals to reduce the incidence of OV infection through behaviour modification by decreasing
Koi pla consumption. The community intervention was developed through efforts of community
participation concerning local and environmental contexts. From the framework, the intervention
was initiated on the basis of the health belief model and social cognitive theory. Members of the
community agreed to take action to lower OV infection incidence emphasizing Koi pla consumption.
The evaluation phase involved a process to monitor the intervention and evaluate the study
outcome. Study outcome was measured by a mixed-method study design. An open-label, non-
randomized community trial was conducted to assess the effectiveness of intervention on incidence
of OV infection and consumption of Koi pla.
The study flow was maintained by the community advisory board (CAB) formed by community
leaders with the support of the research team. The participatory action revealed a potential self-
maintenance capability for the community to sustain a community-derived health campaign.
The results indicated that community intervention could significantly reduce the incidence of OV
infection by 63% (95% CI: 7 – 85), p = 0.03 and reduce Koi pla consumption by 46% (95% CI: 1 - 71), p
= 0.04. Additional cohort analysis showed that Koi pla consumption could increase risk of OV
infection with IRR = 2.52 (95% CI: 1.07 – 5.92) when adjusted for age-group. Additionally, social and
cultural attributes were potential factors leading to Koi pla consumption.
247
6.3 Public health implication
The comprehensive epidemiological study provided up-to-date data with the aim that study results
will be useful for public health policy.
6.3.1 Primary prevention
Considering primary prevention, study result provided support for public health campaign planning.
Following the National Control Program guideline, health education was integrated as a strategic
plan for community approach (5, 15). Providing health education was practical and did not require
complex tools to conduct. However, a further study might be needed in order to explore the
effectiveness of health education.
As I can observe from qualitative study, the method of knowledge distribution should be modified.
The younger generation might gain benefit from social network campaigns. School-age children and
young adults were very effective targets; they were capable of learning new things, transfer to their
family and feasible for behavioural modification. Elders were more likely to trust medical doctors.
However, local health care put more effort into providing diagnosis and treatment. I suggested that
local health volunteers could play an important role to provide health education especially for older
age groups when they gain more trust from being a member of healthcare personnel.
School acts as a community centre and perceived as a knowledge hub. A school-based prevention of
schistosomiasis had been initiated in Tanzania and Nigeria using participatory action research (157,
158) using school as a centre for distributing intervention. Moreover, PPM framework had been
used to assess the feasibility of school as an entry point for implementing the intervention of
helminth control to the community in Kwazulu-Natal (138).
In this study, the role of community centre extended to the formation of CAB where the community
members have shown a potential to initiate and sustain health campaign. In OV infection context,
the infection was regarded as a locally-specific where the social and environmental could play a
major role shaping the infection dynamics. The intervention designed for each area might be
different in details, but the fundamental concept can be consolidated. The roles of predisposing,
enabling and reinforcing factors from PPM were major concerns for developing particular health
campaigns. The process of planning and evaluation within PPM indicated that health burden can be
systematically approached with community participation.
248
6.3.2 Secondary prevention
Based on national policy, the management for OV infection switched to a passive strategy where the
activity mainly focused on diagnosis and treatment (28) requiring a hospital-based setting.
However, providing routine stool examination and treatment may encounter both administrative
and technical problems. The process needs a hospital-based setting was and is not yet available in
the community healthcare system. Although the equipment for microscopic stool examination was
convenient for an ambulatory setting, the examination method needs technically skilled health
personnel to identify OV eggs and distinguish them from other parasites. The most reliable method
needs more laboratory work such as formalin-ethyl acetate concentration technique or molecular
technique was and is not feasible for rural settings (106). The follow-up (70) study in 2007 - 09
suggested that diagnostic methods and equipment should be improved in this community. This
recommendation was not yet followed up and the local health centre was still equipped with same
instruments and stool examination was under the research team support, completely. Skill training
for local health staff was potentially possible, but the process would require time and government
support while short-term resolutions may be considered. A nationwide survey may be too costly for
a developing country. Thus, to effectively deploy resources, a screening program is preferable in high
risk areas. The strategy should be capable of identifying high-risk populations especially outside the
endemic areas where OV infection might be less concerned.
Mass chemotherapy was also considered a strategy to decrease the disease burden as a secondary
prevention. Praziquantel was the drug of choice and provides high efficacy and fewer side effects
than other anthelmintic drugs. A recent study in an endemic area in Thailand reported that previous
use of praziquantel may increase risk of re-infection if the participants lacked adequate knowledge
of prevention as only treatment will not prevent subsequent infection as they still continue
consuming uncooked fish (6). The side effects would come from the drug itself and also a result of
the immune system’s reaction to dying parasites (159) including major side effects such as
drowsiness, dizziness or mild abdominal distress. The drug usage should be justified and provided for
confirmed cases to avoid unnecessary collateral health risks. Moreover, high incidence would
require more than effective secondary prevention for a control strategy.
249
6.4 Future work
Future work can be extended to various aspects. Within 5 years, the project will be expanded in the
surrounding areas where the fieldwork has not yet been conducted. Therefore, future research
could collect epidemiological data in untreated populations and also conduct a clinical trial to assess
the effectiveness of treatment and intervention. Effectiveness was indirectly evaluated through
mathematical modelling in this study, but clinical trials can provide more accurate data. Moreover,
side effects and compliance can be assessed.
Clinical trials have been conducted in the Mekong region especially in Lao PDR where the efficacy of
praziquantel and other drugs against OV infection has been evaluated (49, 53). In Thailand, the
efficacy of praziquantel against OV infection was directly studied from 1980 – 1990 in the period of
the National Control Program implementation (27, 28, 105). Clinical trials could be conducted in this
study area to re-evaluate the effectiveness of praziquantel. Moreover, newer drugs such as
tribendimidine could be assessed.
Household and public sanitation have been essentially improved where the villagers use their toilets
when they were home on a regular basis. Defecation at the paddy field might remain a major
problem for transmitting eggs to the environment. Considering the OV biological life cycle involving
multiple intermediate hosts, interruption of uncooked fish consumption seemed a more practical
intervention compared with biological control of OV (9). Many studies have indicated that reservoir
hosts such as pets and livestock might have played an important role in sustained OV transmissions
(160, 161).
The completeness of fieldwork data could provide more insight about the infection dynamics. The
epidemiology of the direct OV life cycle including the intermediate and reservoir hosts can provide
more insights of transmission dynamics. Further models will consider population dynamics such as
non-random mixing with regard to age pattern and contact pattern and density dependence should
be further incorporated in the model.
Bayesian geostatistical modelling was introduced to explore the epidemiology of Schistosomiasis
(100) and OV infection (14). The Bayesian approach provides an effort for handling uncertainty. The
spatial geostatistical model under the Bayesian formulation was used to build spatial correlation on
the distribution of random effects. The model helps in creating risk maps and is capable of predicting
the infection in non-survey areas, which was useful for targeting the intervention or health
campaign.
250
OV infection was known to be locally-specific (14). Within adjacent areas, the infection dynamics
(and thus the prevalence) can vary considerably. The spatial geostatistical model was a tool to
explore the potential of spatial property, which influences the infection and its risk factors. A study
in southern Lao PDR indicated that strong spatial heterogeneity was found among study areas. The
environmental factors played an important role shaping the infection dynamics (51) by incorporating
spatial factors in the disease model to provide more accuracy to the results and also improve cost-
effectiveness in targeting areas with interventions.
Compound models were important for handling over-dispersed data. The hurdle model provided a
rational interpretation of infection data with regard to diagnostic method. A zero-inflated model had
been developed with the Bayesian framework (100) for mapping Schistosomiasis transmission. Along
with geostatistical modelling, the Bayesian geostatistical zero-inflated regression model showed a
better fit than the spatial negative binomial model. Future research will involve collecting spatial
data from fieldwork to create OV infection risk maps based on the Bayesian framework. The
compound models (zero-inflated and hurdle models) will be incorporated in the mapping model. The
result will aid the health authorities to design and revise public health campaigns.
Diagnostic methods used in fieldwork can result in uncertainties and inaccurate reporting of
infection. The latent class (LC) model was used to compare the accuracy of 5 diagnostic measures for
Schistosoma haematobium infection in Ghana (162) providing more accurate sensitivity and
specificity. The comparison results in an ability to identify the most accurate test among multiple
methods when the gold standard test was not yet identified. Furthermore, more accurate diagnostic
results can be further applied to more accurate reporting of infection prevalence, which was
eventually substantial for epidemiological studies.
For OV infection, the considered “gold standard” can refer to direct identification of Opisthorchis
viverrini eggs in stool specimen or bile. The morphology of these eggs was distinct from other
helminths. However, the eggs were very similar to minute intestinal flukes (MIF). Reports of co-
infections of OV infection and MIF were made along the Mekong basin (156). Therefore, prevalence
can be overestimated due to the morphological similarity.
The serological techniques were constantly improved for accuracy over time (15). However, a cross-
reaction problem remains for other flukes such as Clonorchis sinensis. Moreover, the differentiation
of previous and current infections can be problematic due to the long half-life of the immune
response (163). PCR, a molecular technique, provides nearly 100% specificity and ability to
distinguish Opisthorchis viverrini, Clonorchis sinensis and Opisthorchis felineus (164, 165). However,
251
both serological and molecular technique may not be practical for ambulatory settings for rural
fieldwork in Thailand. The microscopic-based diagnosis was still feasible for this research setting.
Therefore, LC models could be used to validate the diagnostic test in the absence of a gold standard
method in the field. Future research will use the LC model for a test validation and re-assess the
prevalence of OV infection.
Finally, the qualitative study showed a potential result with respect to infection dynamics. The
contribution of the planning model also encourages community participation to engage the health
problem. Further fieldwork will then be incorporated with the qualitative approach so the problem
can be carefully assessed in all aspects. The community intervention package will be introduced to
health policy as a product of rigorous research.
252
6.5 Conclusion
The overall outcome was valuable for Thailand National Health Policy and epidemiological data will
provide the basis for further rigorous academic research. The study highlights the epidemiological
view of OV infection in the rural community setting, the majority part of Thailand. Primary data from
the field study could be useful for extensive epidemiological tools such as mathematical and
statistical modelling to gain more insights into the complexity of OV biology and disease risk
relationship providing greater understanding of the infection dynamics and parasite-host
relationship. Disease surveillance might be useful for the high endemic area of OV infection where
the high risk population could be primarily identified by the history of Koi pla consumption. The
qualitative approach provides more understanding of the disease dynamics in bio-psycho-social
aspects. The National Control Program should be maintained as its core principle is fundamental but
the strategy needs to be adapted for social dynamics, which have changed over time especially
among different generations. Koi pla consumption was clearly identified as a potential risk factor for
acquiring OV infection, which must be interpreted as a primary prevention and should be a
prioritized major concern. For secondary prevention, selective treatment for infected cases was
shown to be effective where the stool examination proved positive. The role of mass chemotherapy
should be re-evaluated and justified for effectiveness. Re-infection should be evaluated as well to
determine the magnitude of the OV infection burden. Up-to-date information on disease dynamics
contributes a potential impact on public health and further ongoing extensive research in this field is
essential to provide effective public health management.
253
References
1. Conlan JV, Sripa B, Attwood S, Newton PN. A review of parasitic zoonoses in a changing Southeast Asia. Vet Parasitol. 2011;182(1):22-40. 2. Keiser J, Utzinger J. Food-borne trematodiases. Clin Microbiol Rev. 2009;22(3):466-83. 3. Chai JY, Darwin Murrell K, Lymbery AJ. Fish-borne parasitic zoonoses: status and issues. Int J Parasitol. 2005;35(11-12):1233-54. 4. Macpherson CN. Human behaviour and the epidemiology of parasitic zoonoses. Int J Parasitol. 2005;35(11-12):1319-31. 5. Sithithaworn P, Andrews RH, Nguyen VD, Wongsaroj T, Sinuon M, Odermatt P, et al. The current status of opisthorchiasis and clonorchiasis in the Mekong Basin. Parasitol Int. 2012;61(1):10-6. 6. Saengsawang P, Promthet S, Bradshaw P. Infection with Opisthorchis viverrini and use of praziquantel among a working-age population in northeast Thailand. Asian Pac J Cancer Prev. 2013;14(5):2963-6. 7. Yong TS, Shin EH, Chai JY, Sohn WM, Eom KS, Lee DM, et al. High prevalence of Opisthorchis viverrini infection in a riparian population in Takeo Province, Cambodia. Korean J Parasitol. 2012;50(2):173-6. 8. Saengsawang P, Promthet S, Bradshaw P. Prevalence of OV infection in Yasothon Province, Northeast Thailand. Asian Pac J Cancer Prev. 2012;13(7):3399-402. 9. Sithithaworn P, Haswell-Elkins M. Epidemiology of Opisthorchis viverrini. Acta Trop. 2003;88(3):187-94. 10. Kaewpitoon N, Kaewpitoon SJ, Pengsaa P. Opisthorchiasis in Thailand: review and current status. World J Gastroenterol. 2008;14(15):2297-302. 11. Andrews RH, Sithithaworn P, Petney TN. Opisthorchis viverrini: an underestimated parasite in world health. Trends Parasitol. 2008;24(11):497-501. 12. Sripa B. Pathobiology of opisthorchiasis: an update. Acta Trop. 2003;88(3):209-20. 13. Kaewkes S. Taxonomy and biology of liver flukes. Acta Trop. 2003;88(3):177-86. 14. Furst T, Duthaler U, Sripa B, Utzinger J, Keiser J. Trematode infections: liver and lung flukes. Infectious disease clinics of North America. 2012;26(2):399-419. 15. Sripa B, Bethony JM, Sithithaworn P, Kaewkes S, Mairiang E, Loukas A, et al. Opisthorchiasis and Opisthorchis-associated cholangiocarcinoma in Thailand and Laos. Acta Trop. 2011;120 Suppl 1:S158-68. 16. Kaewpitoon N, Kaewpitoon SJ, Pengsaa P, Sripa B. Opisthorchis viverrini: the carcinogenic human liver fluke. World J Gastroenterol. 2008;14(5):666-74. 17. Sripa B, Kaewkes S, Sithithaworn P, Mairiang E, Laha T, Smout M, et al. Liver fluke induces cholangiocarcinoma. PLoS Med. 2007;4(7):e201. 18. Songserm N, Promthet S, Sithithaworn P, Pientong C, Ekalaksananan T, Chopjitt P, et al. Risk factors for cholangiocarcinoma in high-risk area of Thailand: Role of lifestyle, diet and methylenetetrahydrofolate reductase polymorphisms. Cancer Epidemiol. 2011. 19. Vennervald BJ, Polman K. Helminths and malignancy. Parasite Immunol. 2009;31(11):686-96. 20. Sripa B, Pairojkul C. Cholangiocarcinoma: lessons from Thailand. Curr Opin Gastroenterol. 2008;24(3):349-56. 21. Wongratanacheewin S, Sermswan RW, Sirisinha S. Immunology and molecular biology of Opisthorchis viverrini infection. Acta Trop. 2003;88(3):195-207. 22. Smout MJ, Sripa B, Laha T, Mulvenna J, Gasser RB, Young ND, et al. Infection with the carcinogenic human liver fluke, Opisthorchis viverrini. Mol Biosyst. 2011;7(5):1367-75. 23. Fried B, Reddy A, Mayer D. Helminths in human carcinogenesis. Cancer Lett. 2011;305(2):239-49.
254
24. Poomphakwaen K, Promthet S, Kamsa-Ard S, Vatanasapt P, Chaveepojnkamjorn W, Klaewkla J, et al. Risk factors for cholangiocarcinoma in Khon Kaen, Thailand: a nested case-control study. Asian Pac J Cancer Prev. 2009;10(2):251-8. 25. Lim JH. Liver flukes: the malady neglected. Korean J Radiol. 2011;12(3):269-79. 26. Jongsuksuntigul P, Imsomboon T. The impact of a decade long opisthorchiasis control program in northeastern Thailand. Southeast Asian J Trop Med Public Health. 1997;28(3):551-7. 27. Jongsuksuntigul P, Imsomboon T. Epidemiology of opisthorchiasis and national control program in Thailand. Southeast Asian J Trop Med Public Health. 1998;29(2):327-32. 28. Jongsuksuntigul P, Imsomboon T. Opisthorchiasis control in Thailand. Acta Trop. 2003;88(3):229-32. 29. Grundy-Warr C, Andrews RH, Sithithaworn P, Petney TN, Sripa B, Laithavewat L, et al. Raw attitudes, wetland cultures, life-cycles: socio-cultural dynamics relating to Opisthorchis viverrini in the Mekong Basin. Parasitol Int. 2012;61(1):65-70. 30. Upatham ES, Viyanant V. Opisthorchis viverrini and opisthorchiasis: a historical review and future perspective. Acta Trop. 2003;88(3):171-6. 31. Schuster RK. Opisthorchiidosis--a review. Infect Disord Drug Targets. 2010;10(5):402-15. 32. Brockelman WY, Upatham ES, Viyanant V, Ardsungnoen S, Chantanawat R. Field studies on the transmission of the human liver fluke, Opisthorchis viverrini, in northeast Thailand: population changes of the snail intermediate host. Int J Parasitol. 1986;16(5):545-52. 33. Haas W, Granzer M, Brockelman CR. Opisthorchis viverrini: finding and recognition of the fish host by the cercariae. Exp Parasitol. 1990;71(4):422-31. 34. Petney T, Sithithaworn P, Andrews R, Kiatsopit N, Tesana S, Grundy-Warr C, et al. The ecology of the Bithynia first intermediate hosts of Opisthorchis viverrini. Parasitol Int. 2012;61(1):38-45. 35. Saijuntha W, Sithithaworn P, Wongkham S, Laha T, Pipitgool V, Tesana S, et al. Evidence of a species complex within the food-borne trematode Opisthorchis viverrini and possible co-evolution with their first intermediate hosts. Int J Parasitol. 2007;37(6):695-703. 36. Touch S, Komalamisra C, Radomyos P, Waikagul J. Discovery of Opisthorchis viverrini metacercariae in freshwater fish in southern Cambodia. Acta Trop. 2009;111(2):108-13. 37. Ngern-klun R, Sukontason KL, Tesana S, Sripakdee D, Irvine KN, Sukontason K. Field investigation of Bithynia funiculata, intermediate host of Opisthorchis viverrini in northern Thailand. Southeast Asian J Trop Med Public Health. 2006;37(4):662-72. 38. Vichasri S, Viyanant V, Upatham ES. Opisthorchis viverrini : intensity and rates of infection in cyprinoid fish from an endemic focus in Northeast Thailand. Southeast Asian J Trop Med Public Health. 1982;13(1):138-41. 39. Sadun EH. Studies on Opisthorchis viverrini in Thailand. American journal of hygiene. 1955;62(2):81-115. 40. Upatham ES, Viyanant V, Brockelman WY, Kurathong S, Lee P, Kraengraeng R. Rate of re-infection by Opisthorchis viverrini in an endemic northeast Thai community after chemotherapy. Int J Parasitol. 1988;18(5):643-9. 41. Ghouri YA, Mian I, Blechacz B. Cancer review: Cholangiocarcinoma. Journal of carcinogenesis. 2015;14:1. 42. Torgerson PR, Macpherson CN. The socioeconomic burden of parasitic zoonoses: global trends. Vet Parasitol. 2011;182(1):79-95. 43. Furst T, Keiser J, Utzinger J. Global burden of human food-borne trematodiasis: a systematic review and meta-analysis. Lancet Infect Dis. 2012;12(3):210-21. 44. Rim HJ, Sohn WM, Yong TS, Eom KS, Chai JY, Min DY, et al. Fishborne trematode metacercariae in Luang Prabang, Khammouane, and Saravane Province, Lao PDR. Korean J Parasitol. 2013;51(1):107-14. 45. Harinasuta C, Harinasuta T. Opisthorchis viverrini: life cycle, intermediate hosts, transmission to man and geographical distribution in Thailand. Arzneimittel-Forschung. 1984;34(9B):1164-7.
255
46. Traub RJ, Macaranas J, Mungthin M, Leelayoova S, Cribb T, Murrell KD, et al. A new PCR-based approach indicates the range of Clonorchis sinensis now extends to Central Thailand. PLoS Negl Trop Dis. 2009;3(1):e367. 47. Prasongwatana J, Laummaunwai P, Boonmars T, Pinlaor S. Viable metacercariae of Opisthorchis viverrini in northeastern Thai cyprinid fish dishes--as part of a rational program for control of O. viverrini-associated cholangiocarcinoma. Parasitol Res. 2013;112(3):1323-7. 48. Jongsuksuntigul P, Manatrakul D, Wongsaroj T, Krisanamara P, Sawatdimongkol S, Wongsaroj S. Evaluation of the Helminthiasis Control Program in Thailand at the End of the 8th Health Development Plan, 2001. J Trop Med Parasitol. 2003;26:18-45. 49. Lovis L, Mak TK, Phongluxa K, Aye Soukhathammavong P, Vonghachack Y, Keiser J, et al. Efficacy of praziquantel against Schistosoma mekongi and Opisthorchis viverrini: a randomized, single-blinded dose-comparison trial. PLoS Negl Trop Dis. 2012;6(7):e1726. 50. Glinz D, Silue KD, Knopp S, Lohourignon LK, Yao KP, Steinmann P, et al. Comparing diagnostic accuracy of Kato-Katz, Koga agar plate, ether-concentration, and FLOTAC for Schistosoma mansoni and soil-transmitted helminths. PLoS Negl Trop Dis. 2010;4(7):e754. 51. Forrer A, Sayasone S, Vounatsou P, Vonghachack Y, Bouakhasith D, Vogt S, et al. Spatial distribution of, and risk factors for, Opisthorchis viverrini infection in southern Lao PDR. PLoS Negl Trop Dis. 2012;6(2):e1481. 52. Kobayashi J, Vannachone B, Sato Y, Manivong K, Nambanya S, Inthakone S. An epidemiological study on Opisthorchis viverrini infection in Lao villages. Southeast Asian J Trop Med Public Health. 2000;31(1):128-32. 53. Soukhathammavong P, Odermatt P, Sayasone S, Vonghachack Y, Vounatsou P, Hatz C, et al. Efficacy and safety of mefloquine, artesunate, mefloquine-artesunate, tribendimidine, and praziquantel in patients with Opisthorchis viverrini: a randomised, exploratory, open-label, phase 2 trial. Lancet Infect Dis. 2011;11(2):110-8. 54. De NV, Murrell KD, Cong le D, Cam PD, Chau le V, Toan ND, et al. The food-borne trematode zoonoses of Vietnam. Southeast Asian J Trop Med Public Health. 2003;34 Suppl 1:12-34. 55. Sripa B, Kaewkes S, Intapan PM, Maleewong W, Brindley PJ. Food-borne trematodiases in Southeast Asia epidemiology, pathology, clinical manifestation and control. Adv Parasitol. 2010;72:305-50. 56. Kerr A. Intestinal parasites in northern Siam. Trans Soc Trop Med. 1916;9:82-9. 57. Mongkolintra S, Leelapanmetha P, Sornmani S, Fungladda W, Siripornpitak S, Siriwichai P, et al. Encouraging self-reliance in the fight against liver fluke. World health forum. 1988;9(3):341-4. 58. Kurathong S, Lerdverasirikul P, Wongpaitoon V, Pramoolsinsap C, Upatham ES. Opisthorchis viverrini infection in rural and urban communities in northeast Thailand. Trans R Soc Trop Med Hyg. 1987;81(3):411-4. 59. Tesana S, Sithithaworn P, Prasongwatana J, Kaewkes S, Pipitgool V, Pientong C. Influence of water current on the distribution of Opisthorchis viverrini infection in northeastern villages of Thailand. Southeast Asian J Trop Med Public Health. 1991;22(1):93-8. 60. Sithithaworn P, Tesana S, Pipitgool V, Kaewkes S, Pairojkul C, Sripa B, et al. Relationship between faecal egg count and worm burden of Opisthorchis viverrini in human autopsy cases. Parasitology. 1991;102 Pt 2:277-81. 61. Maleewong W, Intapan P, Wongwajana S, Sitthithaworn P, Pipitgool V, Wongkham C, et al. Prevalence and intensity of Opisthorchis viverrini in rural community near the Mekong River on the Thai-Laos border in northeast Thailand. J Med Assoc Thai. 1992;75(4):231-5. 62. Radomyos B, Wongsaroj T, Wilairatana P, Radomyos P, Praevanich R, Meesomboon V, et al. Opisthorchiasis and intestinal fluke infections in northern Thailand. Southeast Asian J Trop Med Public Health. 1998;29(1):123-7. 63. Radomyos P, Radomyos B, Tungtrongchitr A. Multi-infection with helminths in adults from northeast Thailand as determined by post-treatment fecal examination of adult worms. Trop Med Parasitol. 1994;45(2):133-5.
256
64. Sriamporn S, Pisani P, Pipitgool V, Suwanrungruang K, Kamsa-ard S, Parkin DM. Prevalence of Opisthorchis viverrini infection and incidence of cholangiocarcinoma in Khon Kaen, Northeast Thailand. Trop Med Int Health. 2004;9(5):588-94. 65. Waree P, Polseela P, Pannarunothai S, Pipitgool V. The present situation of paragonimiasis in endemic area in Phitsanulok Province. Southeast Asian J Trop Med Public Health. 2001;32 Suppl 2:51-4. 66. Wiwanitkit V, Suwansaksri J, Chaiyakhun Y. High prevalence of Fasciolopsis buski in an endemic area of liver fluke infection in Thailand. MedGenMed : Medscape general medicine. 2002;4(3):6. 67. Waikagul J, Dekumyoy P, Chaichana K, Thairungroje Anantapruti M, Komalamisra C, Kitikoon V. Serodiagnosis of human opisthorchiasis using cocktail and electroeluted Bithynia snail antigens. Parasitol Int. 2002;51(3):237-47. 68. Saksirisampant W, Wiwanitkit V, Akrabovorn P, Nuchprayoon S. Parasitic infections in Thai workers that pursue overseas employment: the need for a screening program. Southeast Asian J Trop Med Public Health. 2002;33 Suppl 3:110-2. 69. Rangsin R, Mungthin M, Taamasri P, Mongklon S, Aimpun P, Naaglor T, et al. Incidence and risk factors of Opisthorchis viverrini infections in a rural community in Thailand. Am J Trop Med Hyg. 2009;81(1):152-5. 70. Suwannahitatorn P, Klomjit S, Naaglor T, Taamasri P, Rangsin R, Leelayoova S, et al. A follow-up study of Opisthorchis viverrini infection after the implementation of control program in a rural community, central Thailand. Parasit Vectors. 2013;6:188. 71. Tungtrongchitr A, Chiworaporn C, Praewanich R, Radomyos P, Boitano JJ. The potential usefulness of the modified Kato thick smear technique in the detection of intestinal sarcocystosis during field surveys. Southeast Asian J Trop Med Public Health. 2007;38(2):232-8. 72. Kaewpitoon SJ, Rujirakul R, Ueng-Arporn N, Matrakool L, Namwichaisiriku N, Churproong S, et al. Community-based cross-sectional study of carcinogenic human liver fluke in elderly from Surin province, Thailand. Asian Pac J Cancer Prev. 2012;13(9):4285-8. 73. Kaewpitoon SJ, Rujirakul R, Kaewpitoon N. Prevalence of Opisthorchis viverrini infection in Nakhon Ratchasima province, Northeast Thailand. Asian Pac J Cancer Prev. 2012;13(10):5245-9. 74. Hira PR, Al-Enizi AA, Al-Kandari S, Behbehani K. Opisthorchiasis in Kuwait: first report of infections in Thai migrant workers in the Arabian Gulf. Annales de la Societe belge de medecine tropicale. 1987;67(4):363-8. 75. Peng HW, Chao HL, Fan PC. Imported Opisthorchis viverrini and parasite infections from Thai labourers in Taiwan. J Helminthol. 1993;67(2):102-6. 76. Cheng HS, Shieh YH. Investigation on subclinical aspects related to intestinal parasitic infections among Thai laborers in Taipei. Journal of travel medicine. 2000;7(6):319-24. 77. Lo CT, Lee KM. Intestinal parasites among the Southeast Asian laborers in Taiwan during 1993-1994. Zhonghua yi xue za zhi = Chinese medical journal; Free China ed. 1996;57(6):401-4. 78. Greenberg Z, Giladi L, Bashary A, Zahavi H. [Prevalence of intestinal parasites among Thais in Israel]. Harefuah. 1994;126(9):507-9, 63. 79. Wang LC. Parasitic infections among Southeast Asian labourers in Taiwan: a long-term study. Epidemiol Infect. 1998;120(1):81-6. 80. Wang LC. Changing patterns in intestinal parasitic infections among Southeast Asian laborers in Taiwan. Parasitol Res. 2004;92(1):18-21. 81. Maneeboonyang W, Charusabha C, Prommongkol S, Puangsa-art S, Thanyavanich N, Buchachart K. Status of intestinal helminth infections among thai troops working along Thai-Myanmar border, Suanphung, Ratchaburi Thailand. J Trop Med Parasitol. 2004;27:59-64. 82. Upatham ES, Brockelman WY, Viyanant V, Lee P, Kaengraeng R, Prayoonwiwat B. Incidence of endemic Opisthorchis viverrini infection in a village in northeast Thailand. Am J Trop Med Hyg. 1985;34(5):903-6.
257
83. Brockelman WY, Upatham ES, Viyanant V, Hirunraks A. Measurement of incidence of the human liver fluke, Opisthorchis viverrini, in northeast Thailand. Trans R Soc Trop Med Hyg. 1987;81(2):327-35. 84. Sornmani S, Schelp FP, Vivatanasesth P, Patihatakorn W, Impand P, Sitabutra P, et al. A pilot project for controlling O. viverrini infection in Nong Wai, Northeast Thailand, by applying praziquantel and other measures. Arzneimittel-Forschung. 1984;34(9B):1231-4. 85. Ramsay RJ, Sithithaworn P, Prociv P, Moorhouse DE, Methaphat C. Density-dependent fecundity of Opisthorchis viverrini in humans, based on faecal recovery of flukes. Trans R Soc Trop Med Hyg. 1989;83(2):241-2. 86. Haswell-Elkins MR, Elkins DB, Sithithaworn P, Treesarawat P, Kaewkes S. Distribution patterns of Opisthorchis viverrini within a human community. Parasitology. 1991;103 Pt 1:97-101. 87. Sukontason K, Methanitikorn R, Piangjai S, Choochote W. Viability of metacercariae in northern Thai traditional foods. Southeast Asian J Trop Med Public Health. 1998;29(4):714-6. 88. Wiwanitkit V, Nithiuthai S, Suwansaksri J. Motility of minute intestinal fluke, Haplorchinae spp, metacercariae in fish dishes prepared by different uncooked methods. MedGenMed : Medscape general medicine. 2002;4(1):8. 89. Kruatrachue M, Chitramvong YP, Upatham ES, Vichasri S, Viyanant V. Effects of physico-chemical factors on the infection of hamsters by metacercariae of Opisthorchis viverrini. Southeast Asian J Trop Med Public Health. 1982;13(4):614-7. 90. Upatham ES, Viyanant V, Kurathong S, Rojborwonwitaya J, Brockelman WY, Ardsungnoen S, et al. Relationship between prevalence and intensity of Opisthorchis viverrini infection, and clinical symptoms and signs in a rural community in north-east Thailand. Bull World Health Organ. 1984;62(3):451-61. 91. Haswell-Elkins MR, Mairiang E, Mairiang P, Chaiyakum J, Chamadol N, Loapaiboon V, et al. Cross-sectional study of Opisthorchis viverrini infection and cholangiocarcinoma in communities within a high-risk area in northeast Thailand. Int J Cancer. 1994;59(4):505-9. 92. Songserm N, Promthet S, Sithithaworn P, Pientong C, Ekalaksananan T, Chopjitt P, et al. Risk factors for cholangiocarcinoma in high-risk area of Thailand: role of lifestyle, diet and methylenetetrahydrofolate reductase polymorphisms. Cancer Epidemiol. 2012;36(2):e89-94. 93. Bundy DA, Cooper ES, Thompson DE, Anderson RM, Didier JM. Age-related prevalence and intensity of Trichuris trichiura infection in a St. Lucian community. Trans R Soc Trop Med Hyg. 1987;81(1):85-94. 94. Bundy DA, Kan SP, Rose R. Age-related prevalence, intensity and frequency distribution of gastrointestinal helminth infection in urban slum children from Kuala Lumpur, Malaysia. Trans R Soc Trop Med Hyg. 1988;82(2):289-94. 95. Chipeta MG, Ngwira B, Kazembe LN. Analysis of Schistosomiasis haematobium infection prevalence and intensity in Chikhwawa, Malawi: an application of a two part model. PLoS Negl Trop Dis. 2013;7(3):e2131. 96. Saffari SE, Adnan R, Greene W. Hurdle negative binomial regression model with right censored count data. SORT 2012, vol 36, núm 2. 2012. 97. Hu MC, Pavlicova M, Nunes EV. Zero-inflated and hurdle models of count data with extra zeros: examples from an HIV-risk reduction intervention trial. The American journal of drug and alcohol abuse. 2011;37(5):367-75. 98. Dang TC, Yajima A, Nguyen VK, Montresor A. Prevalence, intensity and risk factors for clonorchiasis and possible use of questionnaires to detect individuals at risk in northern Vietnam. Trans R Soc Trop Med Hyg. 2008;102(12):1263-8. 99. Anderson RM, May RM. Population dynamics of human helminth infections: control by chemotherapy. Nature. 1982;297(5867):557-63. 100. Vounatsou P, Raso G, Tanner M, N'Goran E K, Utzinger J. Bayesian geostatistical modelling for mapping schistosomiasis transmission. Parasitology. 2009;136(13):1695-705.
258
101. Walker M, Hall A, Anderson RM, Basanez MG. Density-dependent effects on the weight of female Ascaris lumbricoides infections of humans and its impact on patterns of egg production. Parasit Vectors. 2009;2(1):11. 102. Zeileis A, Kleiber C, Jackman S. Regression Models for Count Data in R. 2008. 2008;27(8):25. 103. McCullagh P, Nelder JA. Generalized Linear Models, Second Edition: Taylor & Francis; 1989. 104. Bergquist R, Johansen MV, Utzinger J. Diagnostic dilemmas in helminthology: what tools to use and when? Trends Parasitol. 2009;25(4):151-6. 105. Saowakontha S, Pipitgool V, Pariyanonda S, Tesana S, Rojsathaporn K, Intarakhao C. Field trials in the control of Opisthorchis viverrini with an integrated programme in endemic areas of northeast Thailand. Parasitology. 1993;106 ( Pt 3):283-8. 106. Keiser J, Duthaler U, Utzinger J. Update on the diagnosis and treatment of food-borne trematode infections. Curr Opin Infect Dis. 2010;23(5):513-20. 107. Truant AL, Elliott SH, Kelly MT, Smith JH. Comparison of formalin-ethyl ether sedimentation, formalin-ethyl acetate sedimentation, and zinc sulfate flotation techniques for detection of intestinal parasites. J Clin Microbiol. 1981;13(5):882-4. 108. Qian MB, Yap P, Yang YC, Liang H, Jiang ZH, Li W, et al. Accuracy of the Kato-Katz method and formalin-ether concentration technique for the diagnosis of Clonorchis sinensis, and implication for assessing drug efficacy. Parasit Vectors. 2013;6(1):314. 109. Tesana S, Srisawangwonk T, Kaewkes S, Sithithaworn P, Kanla P, Arunyanart C. Eggshell morphology of the small eggs of human trematodes in Thailand. Southeast Asian J Trop Med Public Health. 1991;22(4):631-6. 110. Lovis L, Mak TK, Phongluxa K, Soukhathammavong P, Sayasone S, Akkhavong K, et al. PCR Diagnosis of Opisthorchis viverrini and Haplorchis taichui Infections in a Lao Community in an area of endemicity and comparison of diagnostic methods for parasitological field surveys. J Clin Microbiol. 2009;47(5):1517-23. 111. Upatham ES, Viyanant V, Kurathong S, Brockelman WY, Menaruchi A, Saowakontha S, et al. Morbidity in relation to intensity of infection in Opisthorchiasis viverrini: study of a community in Khon Kaen, Thailand. Am J Trop Med Hyg. 1982;31(6):1156-63. 112. Anderson RM, May RM. Population biology of infectious diseases: Part I. Nature. 1979;280(5721):361-7. 113. May RM, Anderson RM. Population biology of infectious diseases: Part II. Nature. 1979;280(5722):455-61. 114. Strandgaard H, Johansen MV, Aagaard-Hansen J, Petlueng P, Ornbjerg N. Local perceptions and practices in regard to opisthorchiasis in two villages in Lao PDR. Southeast Asian J Trop Med Public Health. 2008;39(1):19-26. 115. Brooker S, Alexander N, Geiger S, Moyeed RA, Stander J, Fleming F, et al. Contrasting patterns in the small-scale heterogeneity of human helminth infections in urban and rural environments in Brazil. Int J Parasitol. 2006;36(10-11):1143-51. 116. Anderson RM, May RM. Infectious Diseases of Humans: Dynamics and Control: OUP Oxford; 1992. 117. Grassly NC, Fraser C. Mathematical models of infectious disease transmission. Nat Rev Microbiol. 2008;6(6):477-87. 118. Dobson AP. The population biology of parasite-induced changes in host behavior. Q Rev Biol. 1988;63(2):139-65. 119. Furst T, Silue KD, Ouattara M, N'Goran DN, Adiossan LG, N'Guessan Y, et al. Schistosomiasis, soil-transmitted helminthiasis, and sociodemographic factors influence quality of life of adults in Cote d'Ivoire. PLoS Negl Trop Dis. 2012;6(10):e1855. 120. Feng Z, Eppert A, Milner FA, Minchella DJ. Estimation of parameters governing the transmission dynamics of schistosomes. Applied Mathematics Letters. 2004;17(10):1105-12. 121. Barbour AD. Modeling the transmission of schistosomiasis: an introductory view. Am J Trop Med Hyg. 1996;55(5 Suppl):135-43.
259
122. Chiyaka ET, Magombedze G, Mutimbu L. Modelling within host parasite dynamics of schistosomiasis. Comput Math Methods Med. 2010;11(3):255-80. 123. Woolhouse ME. On the application of mathematical models of schistosome transmission dynamics. I. Natural transmission. Acta Trop. 1991;49(4):241-70. 124. Gordis L. Epidemiology: Elsevier Health Sciences; 2013. 125. Akai PS, Pungpak S, Kitikoon V, Bunnag D, Befus AD. Possible protective immunity in human opisthorchiasis. Parasite Immunol. 1994;16(6):279-88. 126. Haswell-Elkins MR, Sithithaworn P, Mairiang E, Elkins DB, Wongratanacheewin S, Kaewkes S, et al. Immune responsiveness and parasite-specific antibody levels in human hepatobiliary disease associated with Opisthorchis viverrini infection. Clin Exp Immunol. 1991;84(2):213-8. 127. Wongratanacheewin S, Charupatana C, Bunnag D, Sirisinha S. Effect of praziquantel treatment on antibody levels and lymphoproliferative responses in patients with opisthorchiasis. Southeast Asian J Trop Med Public Health. 1988;19(1):109-16. 128. Glanz K, Rimer BK, Viswanath K. Health Behavior and Health Education: Theory, Research, and Practice: Wiley; 2008. 129. Green J, Thorogood N. Qualitative Methods for Health Research: SAGE Publications; 2004. 130. Giesecke J. Modern Infectious Disease Epidemiology, Second Edition: Taylor & Francis; 2001. 131. Petney TN. Environmental, cultural and social changes and their influence on parasite infections. Int J Parasitol. 2001;31(9):919-32. 132. Kaewpitoon N, Kaewpitoon SJ, Pengsaa P, Pilasri C. Knowledge, attitude and practice related to liver fluke infection in northeast Thailand. World J Gastroenterol. 2007;13(12):1837-40. 133. Crosby R, Noar SM. What is a planning model? An introduction to PRECEDE-PROCEED. Journal of public health dentistry. 2011;71 Suppl 1:S7-15. 134. Li Y, Cao J, Lin H, Li D, Wang Y, He J. Community health needs assessment with precede-proceed model: a mixed methods study. BMC health services research. 2009;9:181. 135. Mazur J, Szumska-Olczak A. [Health promotion indicators - selected model approaches]. Medycyna wieku rozwojowego. 2000;4(4 Suppl 1):53-66. 136. Green LW, Kreuter MW. Health Program Planning: An Educational and Ecological Approach: McGraw-Hill; 2004. 137. Joshi DD, Poudyal PM, Jimba M, Mishra PN, Neave LA, Maharjan M. Controlling Taenia solium in Nepal using the PRECEDE-PROCEED model. Southeast Asian J Trop Med Public Health. 2001;32 Suppl 2:94-7. 138. Taylor M, Coovadia HM, Kvalsvig JD, Jinabhai CC, Reddy P. Helminth control as an entry point for health-promoting schools in KwaZulu-Natal. South African medical journal = Suid-Afrikaanse tydskrif vir geneeskunde. 1999;89(3):273-9. 139. Goodson P. Theory in Health Promotion Research and Practice: Thinking Outside the Box: Jones & Bartlett Learning; 2010. 140. Conner M, Norman P. Predicting Health Behaviour: McGraw-Hill Education; 2005. 141. Glanz K, Bishop DB. The role of behavioral science theory in development and implementation of public health interventions. Annual review of public health. 2010;31:399-418. 142. Glanz K, Rimer BK, Institute NC. Theory at a glance: a guide for health promotion practice: U.S. Dept. of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute; 1997. 143. Hochbaum GM. Public Participation in Medical Screening Programs: A Socio-psychological Study: U.S. Department of Health, Education, and Welfare, Public Health Service, Bureau of State Services, Division of Special Health Services, Tuberculosis Program; 1958. 144. Rosenstock IM. The Health Belief Model and Preventive Health Behavior. Health Education & Behavior. 1974;2(4):354-86. 145. Will JC, Farris RP, Sanders CG, Stockmyer CK, Finkelstein EA. Health promotion interventions for disadvantaged women: overview of the WISEWOMAN projects. J Womens Health (Larchmt). 2004;13(5):484-502.
260
146. Bandura A. Social foundations of thought and action: a social cognitive theory: Prentice-Hall; 1986. 147. Hebert JR, Clemow L, Pbert L, Ockene IS, Ockene JK. Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Int J Epidemiol. 1995;24(2):389-98. 148. Holmberg L, Ohlander EM, Byers T, Zack M, Wolk A, Bruce A, et al. A search for recall bias in a case-control study of diet and breast cancer. Int J Epidemiol. 1996;25(2):235-44. 149. Friedenreich CM, Howe GR, Miller AB. An investigation of recall bias in the reporting of past food intake among breast cancer cases and controls. Annals of epidemiology. 1991;1(5):439-53. 150. Krumpal I. Determinants of social desirability bias in sensitive surveys: a literature review. Qual Quant. 2013;47(4):2025-47. 151. DePaulo BM, Kashy DA, Kirkendol SE, Wyer MM, Epstein JA. Lying in everyday life. Journal of personality and social psychology. 1996;70(5):979-95. 152. Holtgraves T. Social desirability and self-reports: testing models of socially desirable responding. Personality & social psychology bulletin. 2004;30(2):161-72. 153. Holtgraves T, Eck J, Lasky B. Face Management, Question Wording, and Social Desirability1. Journal of Applied Social Psychology. 1997;27(18):1650-71. 154. Bradburn NM, Sudman S, Blair E. Improving interview method and questionnaire design: Jossey-Bass; 1979. 155. Marsden PV, Wright JD. Handbook of Survey Research: Emerald; 2010. 156. Wongsawad C, Phalee A, Noikong W, Chuboon S, Nithikathkul C. Co-infection with Opisthorchis viverrini and Haplorchis taichui detected by human fecal examination in Chomtong district, Chiang Mai Province, Thailand. Parasitol Int. 2012;61(1):56-9. 157. Olaseha IO, Sridhar MK. Participatory action research: community diagnosis and intervention in controlling urinary schistosomiasis in an urban community in Ibadan, Nigeria. International quarterly of community health education. 2005;24(2):153-60. 158. Freudenthal S, Ahlberg BM, Mtweve S, Nyindo P, Poggensee G, Krantz I. School-based prevention of schistosomiasis: initiating a participatory action research project in northern Tanzania. Acta Trop. 2006;100(1-2):79-87. 159. Chai JY. Praziquantel treatment in trematode and cestode infections: an update. Infection & chemotherapy. 2013;45(1):32-43. 160. Aunpromma S, Tangkawattana P, Papirom P, Kanjampa P, Tesana S, Sripa B, et al. High prevalence of Opisthorchis viverrini infection in reservoir hosts in four districts of Khon Kaen Province, an opisthorchiasis endemic area of Thailand. Parasitol Int. 2012;61(1):60-4. 161. Enes JE, Wages AJ, Malone JB, Tesana S. Prevalence of Opisthorchis viverrini infection in the canine and feline hosts in three villages, Khon Kaen Province, northeastern Thailand. Southeast Asian J Trop Med Public Health. 2010;41(1):36-42. 162. Koukounari A, Webster JP, Donnelly CA, Bray BC, Naples J, Bosompem K, et al. Sensitivities and specificities of diagnostic tests and infection prevalence of Schistosoma haematobium estimated from data on adults in villages northwest of Accra, Ghana. Am J Trop Med Hyg. 2009;80(3):435-41. 163. Thammapalerd N, Tharavanij S, Nacapunchai D, Bunnag D, Radomyos P, Prasertsiriroj V. Detection of antibodies against Opisthorchis viverrini in patients before and after treatment with praziquantel. Southeast Asian J Trop Med Public Health. 1988;19(1):101-8. 164. Le TH, Van De N, Blair D, Sithithaworn P, McManus DP. Clonorchis sinensis and Opisthorchis viverrini: development of a mitochondrial-based multiplex PCR for their identification and discrimination. Exp Parasitol. 2006;112(2):109-14. 165. Pauly A, Schuster R, Steuber S. Molecular characterization and differentiation of opisthorchiid trematodes of the species Opisthorchis felineus (Rivolta, 1884) and Metorchis bilis (Braun, 1790) using polymerase chain reaction. Parasitol Res. 2003;90(5):409-14.
261
Appendices
Appendix 1 Supplement tables
Chapter 2
Table S-1 Uncooked fish consumption behaviors of cross-sectional study
Characteristic Na-isarn
N (%)
Na-ngam
N (%)
Total
N (%) P
Fish menus
Chopped raw fish salad
(Koi pla)
No
Yes
196 (59.2)
135 (40.8)
263 (56.8)
200 (43.2)
459 (57.8)
335 (42.2)
0.49
Extensively fermented fish
(Pla ra)
No
Yes
44 (13.3)
287 (86.7)
98 (21.2)
365 (78.8)
142 (17.9)
365 (82.1)
0.004
Table S-2 Uncooked fish consumption behaviors of follow-up study
Characteristic Na-yao
N (%)
Na-ngam
N (%)
Total
N (%) P
Fish menus
Chopped raw fish salad
(Koi pla)
No
Yes
185 (55.9)
146 (44.1)
250 (60.2)
165 (39.8)
435 (58.3)
311 (41.7)
0.23
Extensively fermented fish
(Pla ra)
No
Yes
105 (31.7)
226 (68.3)
63 (15.2)
352 (84.8)
168 (22.5)
578 (77.5)
<0.001
262
Table S-3 Prevalence of OV infection
Prevalence (%)
(95%, CI)
Na-isarn Na-ngam Total p
Age group
(as quartiles, median age)
Q1
0 – 36
(26)
3.41
(0.71 – 9.64)
0 – 35
(16)
5.13
(1.90 – 10.83)
0 – 35
(17)
4.04
(1.76 – 7.81)
Q2
37 – 44
(41)
3.85
(0.08 – 10.83)
36 – 46
(41)
8.40
(4.10 – 14.91)
36 – 45
(41)
5.91
(3.09 – 10.10)
Q3 45 – 56
(49)
11.36
(5.59 – 19.91)
47 – 58
(52)
9.49
(4.83 – 16.34)
46 – 57
(51)
11.17
(7.21 – 16.28)
Q4 ≥ 57
(64)
12.99
(6.40 – 22.59)
≥ 59
(65)
14.41
(8.47 – 22.35)
≥ 58
(64)
13.90
(9.29 – 19.70)
Median age 44 7.85
(4.96 – 10.75) 46
9.29
(6.64 – 11.93) 45
8.69
(6.82 – 10.87) 0.48
Table S-4 Prevalence between completed and incomplete questionnaires group
Prevalence (%)
(95% CI)
Na-isarn p Na-ngam p Total p
Completed questionnaire
No
Yes
3.57
(1.28 – 5.87)
7.85
(4.96 – 10.75)
0.03
8.56
(5.85 – 11.27)
9.29
(6.64 – 11.93)
0.71
7.48
(5.19 – 9.77)
8.69
(6.82 – 10.87)
0.44
263
Table S-5 Incidence of OV infection
Incidence (/100 person-years)
(95%, CI)
Na-yao Na-ngam Total p
Age group
(as quartiles, median age)
Q1
0 – 38
(15)
4.82
(1.56 – 11.25)
0 – 34
(12)
3.29
(0.90 – 8.41)
0 – 36
(13)
4.85
(2.42 – 8.67)
Q2 39 – 50
(45)
8.37
(3.83 – 15.59)
35 – 47
(42)
11.69
(6.68 – 18.98)
37 – 48
(43)
8.75
(5.34 – 13.51)
Q3 51 – 60
(55)
9.20
(4.41 – 16.91)
48 – 57
(52)
4.70
(1.52 – 10.96)
49 – 59
(54)
7.80
(4.62 – 12.32
Q4 ≥ 61
(69)
9.60
(4.39 – 18.22)
≥ 58
(64)
6.64
(2.87 – 130.8)
≥ 60
(66)
7.99
(4.65 – 12.79)
Median age 50 7.98
(5.49 – 11.20) 47
6.80
(4.68 – 9.54) 48
7.34
(5.68 – 9.34) 0.52
Table S-6 Incidence between completed and incomplete questionnaires group
Incidence (/100 person-years)
(95% CI)
Na-yao p Na-ngam p Total p
Completed questionnaire
No
Yes
13.91
(6.01 – 27.41)
7.98
(5.49 – 11.20)
0.18
10.22
(5.10 – 18.29)
6.80
(4.68 – 9.54)
0.25
11.51
(6.93 – 18.00)
7.34
(5.68 – 9.34)
0.095
264
Table S-7 Univariable analysis of risk factors for acquiring OV infection of cross-sectional study
Characteristic Na-isarn Na-ngam
Crude OR (95% CI) p Crude OR (95% CI) p
Sex (female as reference)
Male
1.01 (0.45 – 2.26)
0.98
1.80 (0.95 – 3.40)
0.07
Age group (as quartiles)
Q1 as reference
Q2
Q3
Q4
0 – 36
37 – 44
45 – 56
≥ 57
1
1.13 (0.22 – 5.79)
3.63 (0.96 – 13.68)
4.23 (1.12 – 15.98)
0.88
0.057
0.034
0 – 35
36 – 46
47 – 58
≥ 59
1
1.70 (0.60 – 4.83)
1.94 (0.69 – 5.42)
3.12 (1.72 – 8.28)
0.32
0.21
0.023
Occupation (unemployed as reference)
0.51
0.61
0.52 (0.14 – 1.84)
1.37 (0.46 – 4.09)
0.31
0.57
Others
Agriculture
0.54 (0.09 – 3.39)
1.48 (0.33 – 6.64)
Koi pla consumption
Yes
2.95 (1.26 – 6.90)
0.013
2.69 (1.40 – 5.19)
0.003
Pla ra consumption
Yes
1.92 (0.44 – 8.41)
0.39
1.02 (0.47 – 2.20)
0.97
Table S-8 Univariable analysis of risk factors for acquiring OV infection of follow-up study
Characteristic Na-yao Na-ngam
Crude IRR (95% CI) P Crude IRR (95% CI) p
Sex (female as reference)
Male
1.36 (0.69 – 2.70)
0.38
1.30 (0.65 – 2.56)
0.46
Age group (as quartiles)
Q1 as reference
Q2
Q3
Q4
0 – 38
39 – 50
51 – 60
≥ 60
1
1.74 (0.58 – 5.18)
1.91 (0.65 – 5.58)
1.99 (0.67 – 5.94)
0.32
0.24
0.22
0 – 34
35 – 47
48 – 57
≥ 57
1
3.56 (1.19 – 10.66)
1.43 (0.83 – 5.32)
2.02 (0.61 – 6.71)
0.023
0.60
0.25
Occupation (unemployed as reference)
0.63
0.90
0.74 (0.20 – 2.73)
1.08 (0.32 – 3.63)
0.65
0.90
Others
Agriculture
0.68 (0.14 – 3.20)
0.91 (0.22 – 3.87)
Koi pla consumption
Yes
2.53 (1.23 – 5.23)
0.012
1.61 (0.81 – 3.18)
0.17
Pla ra consumption
Yes
0.93 (0.45 – 1.92)
0.84
1.79 (0.55 – 5.86)
0.34
265
Chapter 5
Table S-9 Incidence of OV infection
Incidence rate (/100 person-years) 95% CI
Age group (as quartiles, median age)
0 – 27
(9)
3.20
0.39 – 11.56
28 – 45
(38) 4.91 1.01 – 14.36
46 – 57
(50) 16.01 7.68 – 29.43
46 – 57
(64) 13.10 5.66 – 25.82
Median age
(45) 9.31 5.90 – 13.97
Table S-10 Impact of community intervention on Koi pla consumption
Study group
Incidence (%) of
Koi pla consumption
(95% CI)
RR (95% CI) p
Intervention 14.12
(7.51 – 23.36) 0.54 (0.29 – 0.99) 0.04
Control 25.93
(17.97 – 35.24)
Table S-11 Impact of community intervention on incidence of OV infection
Study group
Incidence rate
(/100 person-years, 95%
CI)
IRR (95% CI) p
Intervention 5.13
(1.88 – 11.17) 0.37 (0.15 – 0.93) 0.03
Control 13.69
(8.48 – 20.93)
266
Appendix 2 Fieldwork questionnaire
The following questionnaire was mainly used for collecting individual data for parasitic infection
assessment. The Thai version was translated and adapted from this original questionnaire.
QUESTIONNAIRE RURAL PART ONE – INDIVIDUAL DATA
(Circle appropriate answer, tick a box or fill in the blanks)
(CODE)
Name individual: _________________________
Address: _________________________
Date: _________________________
Name of interviewer: _________________________
Part 1 Characterization
If individual is a child less than 12 years of age, please ask parents to answer questions on behalf of the
child where questions are applicable
1. Code
2. Gender 1. Male 2. Female
3. Age of individual: ________Years
4. Religion 1. Buddhist 2. Christian 3. Islamic 4. Other
5. Level of education (If child under the age of 12 years, then maternal level of education:
1. Less than primary
2. Primary
3. Secondary 1
4. Secondary 2
5. Tertiary or College
6. Occupation:
1. Farmer
2. Business
3. Laborer
4. Civil servant
5. Others = 7
8. Per month household income (Baht)
1. < 5000
2. 5,000-10,000
3. 10,000-15,000
267
4. 15,000-20,000
5. > 20,000
9. Were you diagnosed with liver fluke infection within the past year?
1. No 2. Yes
10. If yes, were you treated with anthelmintic?
1. No 2. Yes
11. If yes, did you complete the medication?
1. No 2. Yes
Part II Health behavior
To feed your children, have you been done these behaviors?
2. Yes 1. No
12. Feeding raw fish
13. Use the spoon contaminated with raw fish
14. Feeding with contaminated hands
15. Using contaminated utensils
16. Have you ever eaten Koi pla ?
1. No (go to No. 11) 2. Yes
17. If yes, have you been eaten in the past year?
1. No 2. Yes
18. If yes, how often have you been eaten? ………………..times/month
What kind of fish did you use for Koi pla? No. 19-24 = type of fish
Where do you acquire your fish from?
25. Pond in the village
26. Reservoir in the village
27. Regular market in the village
28. Weekend market
29. Other __________
When you acquire fish, what did you do?
30. Eat straight away
31. Eat at home
32. Preserve it
33. Store it, how = 34……………
35. When do you usually have Koi Pla
268
1. all year
2. Feb-Apr
3. Apr-Oct
4. Oct-Feb
5. other,…………………………..
What is the usual occasion for eating Koi Pla?
36. At home (for a normal meal)
37. Party / Celebrations
38. Drinking with friends
39. During working in the field
40. Other, specify________________
Why do you have Koi pla?
41. Habit
42. Tasty
43. Easy to prepare
44. Just to be like others
What other raw, undercooked / fermented fish dish do you consume? _45, 47, 49, 51, 53, 55, 57, 59, 61 = raw
food dish, 46, 48, 50, 52, 54, 56, 58, 60, 62 = monthly consumption
63. Do you wash your hands after preparing raw fish?
1. Never
2. Sometimes
3. At all times
64. After preparing raw fish, do you wash utensils before preparing other dishes?
1. Never
2. Sometimes
3. At all times
65. When cleaning raw fish, do you feed the remains to dogs and cats or leave them in a place where dogs and
cats can find them?
1. No 2. Yes
66. What lavatory (toilet) facilities do you utilize?
1. No toilet, defecate outside
2. using toilet sometimes
269
3. always using toilet
67. If you go to work in the field, where do you defecate?
1. In the field
2. Near water reservoir
3. Nearest Toilet
4. Other = 68, ………………………..
69. Do you drink bottled or tap water?
1. Tap
2. Rain
3. Well
4. Bottle
5. Pond
6. Other,…70…………………..
71. How do you clean the water?
1. Never
2. Boil
3. Filter
4. Other,…= 72……………
73. How often do you wear footwear when you are outdoors?
1. Never
2. Rarely
3. Sometimes
4. All the time
74. Do you bathe or swim in nearby ponds or reservoirs? 1. No 2. Yes
75. How would you categorize your general health status?
1. Very Healthy
2. Healthy
3. Sick
4. Very sick
Have you suffered from any of the following symptoms in the past one week?
76. Upper abdominal pain or discomfort
77. Gurgling abdomen
270
78. Diarrhoea (3 bouts of watery stool / day or one
bout of dysentery )
79. Nausea
80. Indigestion
81. Malaise or general weakness
82. Anorexia
83. Weight loss (10% within past month)
Do you consume raw, undercooked or fermented 84. frog or 85. snake meat or 86. grasshoppers?
1. No 2. Yes
87. How often do you take de-worming medication?
Frequency Anthelmintic name, if known
1. At least every 6 months
2. At least once a year
3. Occasionally
4. Rarely
0. Never
6. Unsure
88. Do you own a dog or regularly feed a dog?
1. Yes 0. No
89. Do you slaughter animals at home? 1. Yes 0. No
90. Type of animal that you slaughter
91. How often do you come in close contact with dogs (petting, playing or feeding)?
1. Daily
2. 2-3 times a week
3. Once a week
4. Occasionally (1-3 times a month)
5. Rarely
6. Never
92. Are you aware that internal parasites can be transmitted between dogs and humans?
1. No 2. Yes
271
Appendix 3 Ethical approval document
272