ECONOMIC DECISIONS ON THE LIVESTOCK PRODUCTION OF HIGHLANDERS TO CONTROL ZOONOSES RISK USING BAYESIAN NETWORK ANALYSIS CHALISA KALLAYANAMITRA MASTER OF ECONOMICS THE GRADUATE SCHOOL CHIANG MAI UNIVERSITY FEBRUARY 2013 ECONOMIC DECISIONS ON THE LIVESTOCK PRODUCTION OF HIGHLANDERS TO CONTROL ZOONOSES RISK USING BAYESIAN NETWORK ANALYSIS CHALISA KALLAYANAMITRA A THESIS SUBMITTED TO THE GRADUATE SCHOOL IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER IN ECONOMICS THE GRADUATE SCHOOL CHIANG MAI UNIVERSITY FEBRUARY 2013 ECONOMIC DECISIONS ON THE LIVESTOCK PRODUCTION OF HIGHLANDERS TO CONTROL ZOONOSES RISK USING BAYESIAN NETWORK ANALYSIS CHALISA KALLAYANAMITRA THIS THESIS HAS BEEN APPROVED TO BE A PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ECONOMICS EXAMINING COMMITTEE THESIS ADVISORY COMMITTEE ....................................................... CHAIRPERSON ....................................................... ADVISOR Prof. Bruce A. Wilcox ....................................................... MEMBER Dr. Pisit Leeahtam Dr. Pisit Leeahtam .......................................................CO-ADVISOR Dr. Manoj Potapohn ....................................................... MEMBER Dr. Manoj Potapohn ....................................................... MEMBER Dr. Veerasak Punyapornwithaya ....................................................... MEMBER Mr. Chalermpol Samranpong 27 February 2013 © Copyright by Chiang Mai University iii Acknowledgments I would like to express my very great appreciation to my thesis supervisors, including Dr. Pisit Leeahtam, Dr. Manoj Potapohn (Faculty of Economics, Chiang Mai University), Dr. Veerasak Punyapornwithaya (Faculty of Veterinary Medicine, Chiang Mai University) and Mr. Chalermpol Samranpong (Center for Agricultural Resource System Research, Faculty of Agriculture, Chiang Mai University), and my honorable external advisory committee chairman, Dr. Bruce A. Wilcox (Integrative Research & Education Program, Faculty of Public Health, Mahidol University and Tropical Disease Research Laboratory, Khon Kaen University), for their valuable and constructive suggestions during the planning and development of this thesis paper and the assistance with transdisciplinary approach, disease risk modeling and the statistical methods used in this research. I would like to express my gratitude to Dr. Fred Unger and Dr. Jeffrey Gilbert (International Livestock Research Institute (ILRI)) for their continued support to the project in particular on the study design and tools. . I would like to offer my special thanks to Assoc. Prof. Dr. Pichart Uparanukraw, Assoc. Prof. Dr. Nimit Morakote (Department of Parasitology, Faculty of Medicine, Chiang Mai University), Dr. Adulsak Wijit (Office of Disease Prevention and Control 10, Ministry of Public Health), Assist. Prof. Panuwat Yamsakul (Faculty of Veterinary Medicine, Chiang Mai University), Ms. Pornpen Tablerk (Department of Livestock Development, Nan Province), Dr. Wandee Kongkaew (Veterinary Research and Development Centre (South Region), iv Department of Livestock Development) Dr. Montakarn Vongpakorn (Parasitology section, National Institute of Animal Health), Dr. Chirasak Khamboonruang (Department of Parasitology, Faculty of Medicine, Chiang Mai University), Dr. Mongkol Srijun, Dr. Wichak Tidchai (Department of Livestock Development, Chiang Mai Province), Dr. Jennifer Steele (Tufts University), Dr. Karin Hamilton (University of Minnesota) for the human health and veterinary technical advice and Prof. Dr. Songsak Sriboonjit (Faculty of Economics, Chiang Mai University) for the assistance with statistics used in this research. I would like to thank Mr. Naret Puntasrivichai (Highland Research and Development Institute (Public organization)), Mr. Sathian Pattamawat (Pua Crown Prince Hospital, Nan Province), Ms. Jitrat Pongtong (Office of Disease Prevention and Control 10, Ministry of Public Health) for their cooperation with the collection of the data, and Mr. Kongchak Jaidee (Global Health Asia, Faculty of Public Health, Mahidol University) for GIS and mapping support. Advice given by Dr. Warangkhana Chaisowwong (Faculty of Veterinary Medicine, Chiang Mai University), Dr. Jan Hinrichs (Animal Health Economist, Food and Agriculture Organization of the United Nations (FAO)), Dr. Prani Rodtian (Department of Livestock Development, Chiang Mai Province), Dr. Parichart Saenna (Tropical Disease Research Laboratory, Faculty of Medicine, Khon Kaen University), has been most helpful in developing the questionnaires for this research I am particularly grateful to Assoc. Prof. Dr. Khwanchai Kruesukhon (Faculty of Veterinary Medicine, Chiang Mai University), Dr. Peter Kunstadter (Program for HIV Prevention and Treatment, Thailand), Mrs. Isaree Khreusirikul, Ms. Weerawan v Komutdang (Country director’s assistant, Heifer International (Thailand)), Mr. Leesor Jalor (Headman Huai Chan Si Village), Mr. Peerawas Paloeng (Ban Luang sub district officer), Mr. Apinun Taotao (Mae Na Wang sub district officer) Ms. Supansa Lorpu and Mr. Pravin Tee-Ngoo (villagers), for their assistance to contact local people and organizations. I gratefully acknowledge the administrative and logistic assistance of the Faculty of Economics, Faculty of Veterinary Medicine and Faculty of Medicine, Chiang Mai University. This piece of valuable research was affiliated to the EcoHealth-One Health Resource Centre, Chiang Mai University, the provided administrative support is appreciated. The research funds were supported by the EcoZD program of the International Livestock Research Institute (ILRI) made possible by Government of Canada’s International Development Research Centre (IDRC) and the Graduate School, Chiang Mai University. My special thanks are extended to the enumerators including Mr. Xin Dang Xuan, Mr. Faron Xu, Ms. Vu Thi Thu Tra, Ms.Varinda Somrit, Ms. Chanakarn Khampilai, Ms. Pornwimon Pata, Ms. Supassorn Chatsiriyingyong, Ms. Pimchanok Mueangchaimoon, Ms. Hathaichanok Wasasiri, Ms. Sarocha Sukrinprom, Ms. Bhurichaya Palasot, Ms. Phiangkwaun Padeang, Mr. Phuttipong Pookjohn and Mr. Nonprapa Bhuranawut, for their efforts in collecting the data in the fields. Finally, and most of all, I wish to thank my parents for their encouragement and support throughout my study. Chalisa Kallayanamitra vi ชือเรืองวทิยานิพนธ์ การตัดสินใจทางเศรษฐศาสตร์ที เ กี ยวกับการควบคุม โรคติดต่อจากสัตว์มาสู่คนจากการทาํปศุสัตว์ในกลุ่มคน บนทีสูงด้วยการวิเคราะห์โครงข่ายของเบยส์ ผู้เขียน นางสาวชลิสา กลัยาณมิตร ปริญญา เศรษฐศาสตรมหาบณัฑิต คณะกรรมการทีปรึกษาวทิยานิพนธ์ อ.ดร.พิสิฐ ลี,อาธรรม อาจารยที์ปรึกษาหลกั อ.ดร.มาโนช โพธาภรณ์ อาจารยที์ปรึกษาร่วม บทคดัย่อ การเลี, ยงสุกรมีความสําคญัอย่างยิงในกลุ่มคนบนทีสูง นอกจากสุกรจะถูกนาํมาปรุงเป็น อาหารแลว้ ยงัถูกนาํมาใชเ้พือประกอบพิธีกรรมสําคญัต่างๆ จากการศึกษาคน้ควา้จากรายงานโรค ระบาดในช่วง 10 ปีทีผ่านมา (พ.ศ.2546-2555) พบว่า ร้อยละ 90 ของการระบาดของ ทริคิเนลโลซิส (Trichinellosis) นั,น เกิดขึ,นในกลุ่มคนบนทีสูงทีอาศยัตามเขตรอยต่อกบัประเทศ เพือนบา้น ด้วยเหตุนี, เพืออธิบายการแพร่ระบาดของทริคิเนลล่า (Trichinella) งานวิจยันี, ไดน้ํา แนวคิดสุขภาพแบบองคร์วม (EcoHealth-One Health Approach) มาประยุกตใ์ชเ้พือพฒันาแผนผงั บูรณาการขา้มสาขาวิชา (Transdisciplinary Framework) โดยคาํนึงถึงปฎิสัมพนัธ์ของกลุ่มคนบนที สูง สุกร และสิงแวดลอ้ม อยา่งเป็นระบบ ในการศึกษาความเสียงต่อการแพร่ระบาดของทริคิเนลล่า งานวิจยันี, ไดจ้าํแนกประเด็นศึกษาออกเป็น 4 ระบบ ไดแ้ก่ ระบบการเลี,ยงหมู ระบบห่วงโซ่อาหาร ระบบสิงแวดล้อม และระบบเศรษฐกิจ และได้พฒันาแบบจาํลองโครงข่ายของเบยส์เพืออธิบาย ความเสียงของการแพร่ระบาดของทริคิเนลล่า ประกอบกบัการศึกษาเชิงลึกในหมู่บา้นของคนบนที สูงสองแห่ง หนึ งในหมู่บ้านดังกล่าวเคยพบการระบาดของทริคิเนลโลซิส ผลการวิจัยพบว่า แบบจาํลองโครงข่ายของเบยส์สามารถใช้เป็นเครืองมือประกอบการตดัสินใจเกียวกบัการจดัการ ดา้นการเลี, ยงสัตวแ์ละสุขอนามยัของมนุษย ์นอกจากนี, ยงัพบวา่ ระบบการเลี,ยงสุกร ระบบห่วงโซ่ vii อาหาร ระบบสิ งแวดล้อม และระบบเศรษฐกิจ มีความสัมพนัธ์กนัอย่างเด่นชัด ดังนั,นผูมี้ส่วน เกียวขอ้งทั,งภาครัฐ เอกชน และองคก์รต่างๆ สามารถแกปั้ญหาการระบาดของโรคสัตวม์าสู่คนได้ หากเขา้ใจแนวคิดสุขภาพแบบองคร์วม viii Thesis Title Economic Decisions on the Livestock Production of Highlanders to Control Zoonoses Risk Using Bayesian Network Analysis Author Ms. Chalisa Kallayanamitra Degree Master of Economics Thesis Advisory Committee Lect. Dr. Pisit Leeahtam Advisor Lect. Dr. Manoj Potapohn Co-advisor ABSTRACT Pig rearing continues to be an important source of food and serves for ritual use among highlanders in Northern Thailand. The review of Trichinellosis outbreak reports from the past ten years (2003-2012) suggests that more than 90 percent of the outbreaks have occurred in the highlands with several major foci scattered throughout the borderland provinces. To help us understand the transmission of the disease, the research applied an EcoHealth-One Health approach to develop a trandisciplinary framework considering the interaction of highlanders with the pigs they grow and their environment as a single system. The research identified four subsystems to investigate Trichinellosis risk, including, animal husbandry, food chain, environment, ix and economic conditions. The research reported the results of a trandisciplinary process involving the development of a Bayesian Belief Network model of Trichinellosis risk and in-depth study of two highlander villages, including one that experienced an outbreak. The models provided a better understanding of the transmission of Trichinella and solve the decision problems in management systems related with pig production and public health concern to reduce Trichinellosis. The models and the survey results suggested that the above subsystems are entirely interdependent, and thus must be considered as an integrated whole when devising disease interventions. x TABLE OF CONTENTS Page Acknowledgments iii Abstract (Thai) vi Abstract (English) viii Table of Contents x List of Figures xiv List of Appendix Tables xv List of Appendix Figures xvii Chapter 1 Introduction 1.1 Rationale/Problem Statement 1 1.2 Research Objectives 4 1.3 Expected Outcomes 5 1.4 Scope and Limitation 6 1.5 Hypothesis 7 1.6 Definition of Terms 7 Chapter 2 Theory and Literature Review 2.1 Introduction of Animal Health Economics 11 2.1.1 What is Economics? 11 2.1.2 What is Animal Health Economics? 12 2.1.2 How is Economics Useful in Controlling Zoonoses Associated with Livestock Production? 13 xi 2.2 Main Contributions of Veterinary Economists to Animal Health Economics Development 14 2.3 Understanding Zoonoses Emergence through EcoHealth-One Health Approach 18 2.4 Understanding Risk Assessment and Probabilistic Risk Assessment 20 2.4.1 Risk Assessment 20 2.4.2 Probabilistic Risk Assessment 21 2.5 Bayesian Belief Network Analysis 22 2.5.1 Introduction 22 2.5.2 Bayesian Statistics 23 2.5.3 Decision Theory 24 2.5.4 Bayesian Influence 25 2.5.5 Building Networks 26 Chapter 3 Research Strategy and Methodology 3.1 Population and Sampling Design 28 3.1.1 Population 28 3.1.2 Sampling Design 28 3.2 Data Collection 29 3.2.1 Institution Survey 29 3.2.2 Household Survey 29 3.2.3 Environmental Survey 30 3.2.4 Focus Groups 30 3.3 Data Analysis 31 3.3.1 Descriptive Statistics 31 xii 3.3.2 Modeling 31 Chapter 4 Research Results 4.1 Trandisciplinary Trichinellosis Risk Framework 46 4.1.1 General Information 46 4.1.2 Animal Husbandry-relevant Trichinellosis Risk Factors 47 4.1.3 Food Chain-relevant Trichinellosis Risk Factors 50 4.1.4 Environment-relevant Trichinellosis Risk Factors 53 4.1.5 Economy-relevant Trichinellosis Risk Factors 55 4.2 Bayesian Belief Network Model of Trichinellosis Risk 58 4.2.1 Institution’s Decision to Support Money for Pig Pen Construction 59 4.2.2 Institution’s Decision to Encourage People to Stop Consuming Raw or Undercooked Meat 60 Chapter 5 Discussion and Conclusion 5.1 Research Summary 62 5.2 Discussion and Interpretation of Findings 63 5.2.1 Trandisciplinary Trichinellosis Risk Framework 63 5.2.2 Bayesian Belief Network Model of Trichinellosis Risk 66 References 69 Appendices Appendix A Highland Population Information 79 Appendix B Trichinellosis Risk Framework 85 Appendix C Pig Husbandry-relevant Information 88 Appendix D Human Health-relevant Information 98 xiii Appendix E Abbreviations of Variables in Bayesian Belief Network 109 Appendix F Descriptive Statistics from the Field Study 122 Appendix G Cost Structures and Revenue Streams of Pig Production in Highlands 140 Curriculum Vitae 146 xiv LIST OF FIGURES Figure Page 1-1 Life cycle of Trichinellosis 3 2-1 Concept of EcoHealth-One Health Approach 18 3-1 Decision tree representing the institution’s decision to encourage people to switch to keep pig in pen 33 3-2 Decision tree representing the institution’s decision to encourage people to stop consuming raw or undercooked meat 35 xv LIST OF APPENDIX TABLES Table Page A-1 The estimated of highland population in Thailand 80 A-2 The estimated of highland population in Chiang Mai, Chiang Rai, and Mae Hong Son 80 C-1 Gains and losses from switching to keep pigs in pen 89 C-2 Utility table of animal health perspective 90 C-3 Scoring rule results of the Trichinella infection in pig 94 C-4 Probability table of the Trichinellosis risk in animal 95 D-1 Economic losses of illness and death per capita from Trichinellosis in human 99 D-2 Utility table of human health perspective 100 D-3 Scoring rule results of the Trichinella risk in human 105 D-4 Probability table of the Trichinellosis risk in human 106 E-1 Abbreviations of variables used in Bayesian Belief Network 110 F-1 General information of the respondents 123 F-2 General information of the pig growers 124 F-3 Knowledge and attitude of the pig growers 125 F-4 Pig production practices 126 F-5 Source of pigs 127 F-6 Pig production modes 128 F-7 General information of food-preparing persons 129 xvi F-8 Knowledge and attitude of the pig growers 131 F-9 Meat preparation and frequency of meat consumption 132 F-10 Source of meat 133 F-11 Environment-related Trichinellosis risk factors 134 F-12 Financial status of the respondents 136 F-13 Access to medical service of the respondents 138 G-1 Cost structures and revenue streams of pig production in highlands 141 xvii LIST OF APPENDIX FIGURES Figure Page A-1 Location of the highland villages in Chiang Mai, Chiang Rai, and Mae Hong Son 81 A-2 Reported cases of Trichinellosis by sub district (Tambol) during 2003-2012 82 A-3 Visited households in Huai Ma Fueang Village 83 A-4 Visited households in Huai Chan Si Village 84 B-1 Trichinellosis risk framework 86 B-2 Trichinellosis risk framework with belief bars 87 C-1 Trichinella infection risk framework (animal perspective) – Model 1 92 C-2 Trichinella infection risk framework (animal perspective) – Model 2 93 C-3 Trichinella infection risk framework (animal health perspective with decision and utility nodes) – Model 2 96 C-4 Trichinella infection risk framework (animal health perspective with decision and utility nodes) – Model 1, when there is information about the level of risk that pigs will be infected by Trichinella 97 D-1 Trichinellosis risk framework (human perspective) – Model 1 103 D-2 Trichinellosis risk framework (human perspective) – Model 2 104 D-3 Trichinellosis risk framework (human health perspective with decision and utility nodes) – Model 1 107 xviii D-4 Trichinellosis risk framework (human health perspective with decision and utility nodes) – Model 1, when there is information about the level of risk that people will be Trichinellosis 108 Chapter 1 Introduction 1.1 Rationale/Problem Statement In the upland areas in Northern Thailand culturally distinct ethnic minority groups reside scattered throughout the mountainous region along Thailand-Burma borders. They consist of mostly Akha, Hmong, Karen, Lahu, Lisu, and Yao ethnicities (Crooker, 2007). Overall, these ethnic minorities in 20 provinces of Thailand make up a population of about 750,000 persons; thereby they represent slightly more than 1% of the overall population (ADB, 2001: 5 cited in Krahl, 2011) (For more information about highlander population, see Appendix A). Throughout this paper the term “highlanders” will be used as it is the generally recognized term for this population. The notion of highlander historically has not only served to improve livelihoods but also aimed to control and incorporate the highland population into the nation, to secure the national boundary, to prevent the production of opium and later on to protect forests and watersheds (Gillogly, 2008: 119 cited in Krahl, 2011). Opium cultivation historically was a major source of income for many of the highlanders. The government worked hard to eradicate their cultivation through crop substitution and livestock farming by the end of 1980s (Crooker, 1988; Dirksen, 1997; Renard, 2001 cited in Crooker, 2007). However, the cash crop-substitution policy led to the highlanders being accused of destroying the country’s forests with their “slash- 2 and-burn” agriculture, as their traditional swidden farming practice was labeled during this period. The government subsequently adopted a policy of relocation forcing highlanders to live on less land and restricting their land use rights. Unintended consequences of these policies have included intensification of land and resource use and permanent settlements. This has in turn contributed to deteriorated soil fertility, diminished crop yields and indigenous crops, increased water contamination from the fertilizer and pesticides use, and significant changes in livestock rearing practices; all of which have raised concerns about food security and health risks (Crooker, 2007). Recently, increased attention has been paid to addressing various issues concerning highlanders (Fujioka, 2002). Numerous institutions are concerned with highlander development. This includes 31 departments and 168 agencies within eleven government ministries involved in hill area development namely the ministries of Interior, Defense, Public Health, Education, University Affairs, Agriculture, Cooperatives, Science, Technology and Energy, Finance, Communication, Industry, and the Prime Minister’s Office (Fujioka, 2002). In spite of this the livelihoods of many villagers are still in question (Hau, 2000). The highlanders are seen as the most disadvantaged and vulnerable groups among Thailand’s rural population. They largely depend on agriculture for income and employment (Fujioka, 2002). Besides crop production, livestock production in highlanders’ farming systems is very important and its role varies widely. It provides draught power for crop production used for subsistence needs or market sale (McDermott et al., 1999). Pig rearing continues to be an important source of subsistence food and serves for sacramental purposes (Rattanaronchart, 1994; Tancho, 3 1997), with almost every family keeping pigs (Cheva-Isarakul, 1998). Thus pig health and production can have a substantial impact on their livelihoods. Source: Centers for Disease Control and Prvention (2011) Figure 1-1: Life cycle of Trichinellosis Even though there is no regularly updated data recording incidences of parasitic infections in the highland villages, interview of experts and review of literature confirms that Trichinellosis is endemic in these borderland highlands. The nature of occurrence, transmission, and circulation of Trichinellosis (Ramasoota, 1991) in these areas can be explained by the interaction of highlanders with the pigs they grow and of both with their environment. Epidemiological investigations reveal that outbreaks of this disease have taken place mostly in rural areas where pigs are 4 most commonly raised on a free range basis and some are kept in low standard pens as a result of limited investment in husbandry beyond minimal needs (Khamboonruang, 1991; Ramasoota, 1991; Rattanaronchart, 1994; Willingham, 2003). Pigs will eat nearly anything, including garbage and wild animal carcasses often accessible to pigs. This fosters the transmission of the parasite to pigs and, in turn, to humans, as pork is popularly eaten raw or undercooked during rituals associated with local and traditional festivals (Ramasoota, 1991). Besides the distinct settings and practices and weak public health infrastructure placing them at higher risk of zoonosis, these populations are confronted with difficulties resulting from the lack of citizenship, language barriers, market pressure, social exclusion, and globalization. As these external factors continue to evolve, often resulting in increasing economic vulnerability of highland populations, concerns are being raised about the potential of the borderland area as an epicenter of outbreaks of zoonoses. 1.2 Research Objectives 1.2.1 Consistent with the EcoHealth-One Health approach, this study attempts to understand Trichinellosis risk in borderland highlanders as a basis for prevention and control measures by developing a transdisciplinary framework. This framework considers the interaction of highlanders with the pigs they grow and their environment as a single system. 1.2.2 Both quantitative and qualitative methods employed in Bayesian Belief Network (BBN) are applied in this study to explain Trichinellosis risk in borderland highlanders and effects of uncertainty in the management system. Also, the concept 5 of optimizing expected utility is applied in solving the BBN in order to indicate the options available and choices made as a basis for decision-making to reduce Trichinellosis risk. 1.2.3 This study does not intend to suggest to that interventions aimed at highlander’s livelihood systems should be launched in the form of a policy forcing villagers to change their modes of pig production to reduce disease risk. Rather, it encourages policy makers to better understand the underlying mechanism by which the livelihoods system affects disease risk in order to find the appropriate policy to control the emergence and spread of diseases. 1.3 Expected Outcomes 1.3.1 Most importantly this study will reveal a paradigm shift taking place in Economics research toward understanding a complex, real-world problem, such as emerging zoonoses, using the EcoHealth-One Health approach. 1.3.2 This study will address the core concept of economics; that of the idea of utilizing scarce resources to satisfy unlimited wants or needs of humans. In this case, Economics can help policy makers or planners make more effective and efficient decisions in terms of their potential to affect pig-rearing strategies and associated development to improve biosecurity and mitigate zoonoses risk. 1.3.3 This study will be considered innovative research for its novel utilization of mixed methods including quantitative and qualitative methods in decision making. 6 1.4 Scope and Limitation The study concentrates on the ethnic minority groups that reside scattered throughout the mountainous region along Thai-Myanmar border of Thailand’s Chiang Mai, Chiang Rai, and Mae Hong Son Provinces, including Akha, Hmong, Karen, Lahu, Lisu, and Yao ethnicities, using mixed methods including qualitative and quantitative methods to assess the risk of being infected with Trichinellosis in human associated with pig production modes. Its data gathering and analysis further concentrated on two villages in a single District, Mae Ae, Chiang Mai Province, of mainly Lahu ethnicity. The highlanders of the Northern Thailand borderland are quite heterogeneous group with substantial intra- as well as inter-ethnicity differences in terms of economic and environmental circumstances, cultural practices and so on. The expectation was not that the study could extrapolate findings based on two villages. Rather it was to use the two villages to further develop and test a novel approach and methods, as well as a disease risk model based on a framework constructed using existing information and data. This included theory from multiple disciplines as well as published and informant-provided data on the borderland highlanders in general, including that on their pig husbandry and eating practices, The village-level field study provided a basis for testing and further refining the model, including the BBN method, as well as documenting the people-pig- environment interactions in their economic aspects for at least the two villages. 7 1.5 Hypotheses 1.5.1 EcoHealth-One Health approach can be applied to develop transdisciplinary Trichinellosis risk frame work in borderland highlanders. 1.5.2 Highlanders make decisions on the allocation of scarce resources in livestock production to optimize their utility 1.5.3 Bayesian Belief Network Analysis can solve the decision-making problems based on the interconnection of human and animal health. 1.5.4 Government institutions or non-governmental organizations can make effective and efficient decisions to reduce zoonoses risk if they understand the concept of EcoHealth-One Health approach 1.6 Definition of Terms Bayesian Belief Network (BBN): A statistical method invented in the 1940’s and 1950’s to take into account the effects of uncertainty in management systems in decision making processes (Henrion et al., 1991 cited in Dambacher et al., 2007). It is a graphical description of the conceptual model that captures the analyst beliefs in the causal relationships of significant variables in the system of interest (Dambacher et al., 2007). EcoHealth-One Health Approach: A systematic and participatory approach to understanding and promoting sustainable health and well-being of humans, animals and the environment thought of as all part of one ecosystem; as well as making decisions, taking action, and evaluating outcomes (Waltner-Toews, 2009). 8 EcoHealth-One Health approach is an emerging field of study and practice that examines the biological, social and economic dynamics of an ecoystem and relates these changes to human and animal health, holistically. It brings together people from various disciplines such as veterinarians, ecologists, economists, social scientists, policy makers, and others to explore and understand how the above dynamics affect human and animal health (UNBC, n.d.). Highlander: The term used within Thailand for all of the various tribal peoples who migrated from China and Tibet over the past few centuries (Srisoontorn, n.d.). They reside scattered throughout the mountainous region along Thailand, Laos, and Myanmar borders (Crooker, 2007). Some people also use the term ‘Ethnic minority’, but this must include Chinese, Laotians, Indians or Malaysians. Some people use the term ‘Highlanders’ to include Chinese people who live in the mountains as well. There are six major highlander groups within Thailand: Akha, Lahu, Karen, Hmong, Mien and Lisu reside in 20 provinces (Srisoontorn, n.d.). Influence diagram: A graphical and mathematical representation of a decision problem (Lumina Decision Systems, 2013) which is an extension of BBN (Watthayu and Peng, 2004) that includes decision making, uncertainties, utility maximization, and how they influence each other. It is also known as decision diagram or decision network (Lumina Decision Systems, 2013). Parasitic Zoonoses: Diseases hosted by animals which are caused by parasites that can be transmitted to humans (Westmount Animal Clinic, 2009). Besides, a number of livestock parasites also cause economic losses from the impact on the quality and quantity of animal products. In addition, the complex life cycles of most 9 parasites, the distinct conditions of animal husbandry in rural areas, slaughtering facilities, and marketing practices can have a severe influence on the transmission and outbreak of the diseases. Probabilistic Risk Assessment (PRA): Application of probability distributions to identify variability or uncertainty in estimations of risk. It is a quantitative explanation of the degree of variability and uncertainty in risk estimates for unwanted events such as the outbreak of diseases (Mitchell, Smith, and Murphy, 2004: 1-10). Risk: Risk is fundamental to any decision making scheme. Risk can be defined as imperfect knowledge for stochastic events where the probabilities of the possible outcomes are known (Hardaker et al., 1997 cited in Kaan, 2000; Siegel and Alwang, 1999 cited in Devereux, 2001). To put it simply, risk is uncertain consequences (Kaan, 2000) resulting in welfare losses (Devereux, 2001). Trichinellosis: Trichinellosis or Trichinosis is a parasitic disease caused by a roundworm (nematode) called Trichinella spiralis. Trichinae can be readily avoided by proper handling and cooking of certain meats, particularly pork products (Medical- dictionary, 2002). The severity of symptoms depends on the quantity of infectious worms consumed. The initial clinical manifestations of the disease are nausea, diarrhea, vomiting, fatigue and fever, following by headaches, fevers, chills, cough, eye swelling, aching joint muscle pains, itchy skin and diarrhea. In case of heavy infection, patients may experience difficulty controlling movements and have cardiovascular and respiratory problems. Severe cases can progress to coma or death (Medterms, 2011). 10 Zoonoses: Zoonoses, also called zoonotic diseases are diseases caused by infectious agents transmitted between animals whether wild or domesticated and humans through a variety of infection routes, including animal bites, vectors, and animal-to-human contact (Olsen, 2004; Koo, 2009). All zoonoses can create a serious health threat if not controlled (Stregowski, 2012). Chapter 2 Theory and Literature Review A review of existing literature was conducted to identify the critical points of current knowledge including findings as well as theoretical and methodological contributions concerning the investigation of economic decisions and the assessment of zoonoses risk associated with livestock production. It focuses on four main themes: (1) introduction of animal health economics, (2) main contributions of veterinary economists to animal health economics development, (3) understanding zoonoses emergence through EcoHealth-One Health approach, (4) understanding risk assessment and probabilistic risk assessment, and (5) novel application of Bayesian Belief Network Analysis. 2.1 Introduction of Animal Health Economics 2.1.1 What is Economics? Economic thinking was first used in the context of agriculture efficiency management between 394 and 365 BC (Backhouse, 2002 cited in Rushton 2009). Lionel Robbins defines economics as “the science which studies human behavior as a relationship between ends and scarce means which have alternative uses” (Backhouse, 2002: 3 cited in Rushton, 2009). Likewise, Black and others mentioned that economics is concerned with decisions about how to allocate and use 12 scarce resources, particularly the production, distribution and consumption of commodities (Perry and Randolph, 1999 cited in Black, 2006). 2.1.2 What is Animal Health Economics? “Animal health economics is a discipline, which does not belong to the core of veterinary science” (Otte and Chilonda, 2000 cited in Ruston, 2009) and is relatively young in relation to other economic disciplines (Rushton, 2009). However, it is becoming more and more important as the assistance for decision making on animal health intervention at all levels (Otte and Chilonda, 2000 cited in Sudan, 2009) attempting to optimize animal health management (Marsh, 1999 cited in Sudan, 2009). In this field economics is not mainly dealing with money but rather with making rational choices in the allocation of scarce resources for achieving competing goals. With the hypothesis that people make decisions in order to optimize their satisfaction, utility or pleasure, some of these decisions have led to unintended consequences such as zoonoses emergence (Black, 2006). When outbreaks occur, scarce resources are used to care for both animals and humans that are sick and to prevent or control the transmission of infection. Productive capacity is constrained and trading relationships are disrupted by infection. Besides, there are likely to be missing markets for infection control caused by many reasons such as externalities, public goods, uncertainty and equity (Roberts, 2006, Ch.1: 12). As a consequence, infection poses a huge economic problem that needs to be addressed. The characteristics of zoonotic infectious diseases raise issues for economists seeking to apply their tools in this area. 13 2.1.3 How is Economics Useful in Controlling Zoonoses Associated with Livestock Production? Evaluating interventions or controls is a central task for economists to contribute to the adoption of efficient policies to control infection (Roberts, 2006, Ch. 1: 8). Furthermore, economic instruments are becoming more and more important as the aids to understand behavior and decision-making processes, especially of small-scale farmers in animal health management (Chilonda and Huylenbroeck, 2001). Economic analysis of the optimal control of zoonoses associated with livestock production is complex as it depends on the nature of occurrence, transmission, and circulation of the diseases. It takes account of the benefits and costs of controlling diseases in monetary terms. Consequently, information from both economists and non-economists is important for this economic analysis (Tisdell, 2006). Economic approaches to infectious disease are embedded in many areas of work nowadays and cannot be ignored. The National Institute for Clinical Excellence in the United Kingdom, for instance now requires economic assessments included in guidelines for interventions. Specifications for grants to assess interventions used to prevent or control infections now often include economic assessments (Roberts, Ch. 2006, Ch.13: 237-240). Even though this was unpopular with the pure economist it is an innovative way of utilizing economics in explaining a complex, real-world problem. 14 2.2 Main Contributions of Veterinary Economists to Animal Health Economics Development (Rushton, 2009) Recent studies show that the emphasis of most of veterinary economists is usually on the economics of production disease (diseases induced by management practices) and in the evaluation of zoonoses intervention and control efforts. Peter Ellis is considered one of the leading veterinary economists (Dijkhuizen and Morris, 1997 cited in Rushton, 2009). In 1970, he was the first to apply cost-benefit analysis techniques to an animal disease, specifically for the analysis of classical swine fever (CSF) eradication in the UK. Furthermore, Roger Morris and Ellis had been working together on the various aspects of veterinary economics with particular emphasis on production disease and the evaluation of zoonoses control. In addition, in 1977, an interdisciplinary team designated as VEERU (Veterinary Epidemiology and Economics Research Unit) was established in the Department of Agriculture at the University of Reading in England. The early contributors to this group included economists, veterinarians, farm management experts, statisticians, and animal production experts. Their major contributions were in the early use in scientific studies of cost-benefit analysis techniques, herd models (CLIPPER and LPEC), herd monitoring systems (DAISY, EVA, MONTY, INTERHERD), promoting the use of economic techniques in planning processes, and examining economic impact across different levels of society. VEERU aimed at developing teams through collaborative projects in various countries and building training schemes for management of veterinary and livestock services. These initiatives have been supported by Office of Development 15 Assistance of OECD, German Aid, Danish Aid, the British Council, FAO, World Organization of Animal Health (OIE) The World Bank and many other agencies (Rushton, 2009). Likewise, Tisdell (1995) and Harrison (1996) also investigated the application of cost-benefit analysis for evaluating animal disease programs. These economists examined how animal health programs can aid sustainable development (Harrison and Tisdell, 1997 cited in Rushton, 2009) in Thailand. Furthermore, Perry at the International livestock Research Institute (ILRI) is one of the world’s most noticeable epidemiologists specialized in a range of diseases. She concentrates on a number of important themes in animal health economics, including farm-level economic evaluations, trade implications of sanitary requirements and veterinary service delivery (Perry, 1999 cited in Rushton, 2009). Additionally, many of veterinary economists such as Ramsay, Tisdell and Harrison (1997) have concentrated on how better information communication for animal health could enhance decision-making. Their findings demonstrate that for endemic diseases there are two options: do nothing or eradication (Harrison et al., 1999; Tisdell, et al., 1999 cited in Rushton, 2009). Another pioneer in the field Richard Bennett initially worked on the advantages of information communication on animal health decisions (Bennett, 1991 cited in Rushton, 2009) and on decision- making for leptospirosis control in cattle (Bennett, 1993 cited in Rushton, 2009). This and other work by him and his colleagues is specifically in the field of animal welfare economics (Bennett, 1995, 1998; Bennett and Larson, 1996; Blaney and Bennett, 1997; Anderson et al., 1999 cited in Rushton, 2009). 16 Some veterinary economists have been working on the development of economic analysis techniques in the study of diseases and their control. Tim Carpenter was the first to examine the use of various economic analysis techniques such as decision tree analysis (Carpenter and Norman, 1983; Carpenter et al., 1987; Ruegg and Carpenter, 1989; Rodrigues et al., 1990 cited in Rushton, 2009), microeconomics analysis of disease (Carpenter, 1983 cited in Rushton, 2009), simulation models to assess animal disease (Carpenter and Thieme, 1980 cited in Rushton, 2009), dynamic programming (Carpenter and Howitt, 1988 cited in Rushton, 2009), dual estimation approach to derive shadow prices for diseases (Vagsholm et al., 1991 cited in Rushton, 2009), estimation of consumer surplus (Mohammed et al., 1987 cited in Rushton, 2009), willingness to pay for vaccination (Thorburn et al., 1987 cited in Rushton, 2009), linear programming (Carpenter, 1978; Carpenter and Howitt, 1980; Chirstiansen and Carpenter, 1983 cited in Rushton, 2009), use of economic analysis to review subsidies to veterinary support institutions (Carpenter and Howitt, 1982 cited in Rushton, 2009), and the use of the cost-benefit analysis approach for selecting veterinary services (Zessin an Carpenter, 1985 cited in Rushton, 2009) He also has been involved in economic assessment using more conventional economic instruments such as financial and cost-benefit analysis (Carpenter et al., 1981,1988; Davidson et al., 1981; Kimsey et al., 1985; Miusing et al., 1988; Vagsholm et al., 1988; Sischo et al., 1990 cited in Rushton, 2009). His work has been based on the very detailed knowledge of a production system and the epidemiology of the disease concerned. 17 Meanwhile Aalt Dijkhuizen at Wageningen Agricultural Univesrity of The Netherlands began researching the use of economic evaluation techniques for animal disease. Dijkhuizen and his team worked on problems including the economics of pig fertility and culling management, cattle problems and diseases, and the economics of Foot and Mouth Disease at a time when Europe was considering changing from a policy of annual vaccination to no vaccination. They also working the problems of a “stamping out policy,” exotic disease risk and the inclusion of risk analysis into economic analysis. Other research conducted by them was on the use of insurance against the outbreak of contagious diseases, and on animal welfare, food safety and animal health economics. Dijkhuizen utilized animal recording systems such as PORKCHOP and decision support systems such as CHESS in his work. His modeling inputs have assisted decision makers at the farm, national and region levels. His experience has \ benefited from examining a wide range of techniques for the economic evaluation of diseases. His contributions to the field of animal health economic analysis have been significant in directing animal health policies in his own country. On the theoretical side some veterinary economists such as McInerney and Howe began researching the economics of livestock disease through the development of conceptual models of farmer behavior towards disease (Howe, 1985; McInerney et al., 1992; Howe and Christiansen, 2004 cited in Rushton, 2009). This group is credited as being the first to apply a conceptual framework for economic analysis of disease and its control. However, their influence on the thinking of animal health economics has largely been limited to concepts and theory. 18 All the above approaches are in the practical field of the economic evaluation of animal disease based on a detailed knowledge of the production system. This research is relatively limited in scope and focused on existing or endemic diseases of interest to the agricultural industry. It is true that they have demonstrated its value in understanding the problem of production disease or the way to maximize utility in this context. However, the field had not yet begun to address the more complex and real-world problem emerging diseases. Here, a more holistic approach such as that advocated by EcoHealth-One Health approaches is required. 2.3 Understanding Zoonoses Emergence through Ecohealth- One Health Approach Source: Department of Environmental and Global Health, University of Florida, 2012 Figure 2-1: Concept of Ecohealth-One Health Approach Veterinary medicine appears to have been a distinct discipline during the Zhou Dynasty in China (11-13th century). This period had one of the earliest organizations of a holistic public health system including human and animal health (Driesch and 19 Peters, 2003 cited in Zinsstag et al., 2010). Later on in the 19th century, based on the discovery of similar disease processes in humans and animals, Rudolf Virchow as a scientist had a strong interest in an interconnection of human and veterinary medicine (Saunders, 2000 cited in Zinsstag et al., 2010). In the 20th century, Calvin Schwabe originated the concept of ‘one medicine’ suggesting that human and veterinary medicine are interconnected and can contribute to the development of each other (Zinsstag et al., 2010). Later on, a broader approach to health and well-being of societies was introduced as ‘one health’. In these years, given the global health thinking, ecosystem approaches to health have emerged. Based on multifaceted thinking that goes beyond humans and animals, these approaches consider as inseparable the interconnection between ecosystems and health. The idea of ecohealth, a term used as a contraction for “ecology and health” and “ecosystem and human health,” was first popularized internationally by the Canadian government’s International Development Research Centre (IDRC) drawing on Lebel’s (2003) “ecosystem approaches to human health”. As described by Wilcox and colleagues (Wilcox et al., 2012), who are among the founders of ecohealth as an academic field, “the ecosystem approach” in general means applying an understanding of the properties of a whole entity of relevance to the health problem of concern, an infectious disease or otherwise. According to these researchers this includes “contextualizing a problem by situating it geographically and identifying the biophysical as well as socio-cultural and economic conditions and forces contributing to a human or veterinary public health issues.” They point out that the objective is to identify the proximal as well as the distal causative factors and how they interact to, for example, understand a zoonoses outbreak. Thus, they point out that in addition to 20 conducting a routine epidemiological investigation, an ecohealth study would consider all the potentially relevant underlying factors as well as the source or origin of the agent(s) responsible, with the aim of targeting the critical variables that will limit the likelihood of emergence events involving existing disease-causing agents or “new” agents, for example, Highly Pathogenic Avian Influenza (H5N1). Accordingly, to understand the complexities of causes behind zoonoses emergence, it is important to call for more holistic and comprehensive approaches to analyze and address this real-world problem (McDermott and Grace, 2011). In pointing this out, EcoHealth-One Health approach has emerged to capture the increasing potential risk of zoonoses at local, regional and global scales (AVMA, 2008). 2.4 Understanding Risk Assessment and Probabilistic Risk Assessment 2.4.1 Risk Assessment Risk is fundamental to any decision making scheme. Risk can be defined as imperfect knowledge for stochastic events where the probabilities of the possible outcomes are known (Hardaker et al., 1997 cited in Kaan, 2000; Siegel and Alwang, 1999 cited in Devereux, 2001). To put it simply, stated in terms of economics, risk is uncertain consequences (Kaan, 2000) resulting in welfare losses (Devereux, 2001). Risk assessment models are important instruments in economic analysis of infectious disease. It has been described as the science of identifying and understanding unwanted events, of estimating the possibility of these events occurring 21 and of the consequences if they do occur (Roberts, 2006: 237). Risk assessments will not be beneficial unless they provide guideline for management. Managing risk is significant for livestock farming. In agriculture, the sources of risk are various such as a fluctuation of price in market for agricultural products, financial viability, or a diversity of hazards related to weather and diseases. Risk management strategies involve decisions on the farm and the household to find out the amount of outputs to be produced, the allocation of land, the use of inputs, etc. Farmers can manage risk through market tools including insurance. However, not all risks are insurable because of a market failure from information asymmetries. Government can empower farmers to take responsibility for risk management by providing a variety of instruments so that they can choose the best that fits their needs (OECD, 2009). 2.4.2 Probabilistic Risk Assessment Probabilistic Risk Assessment (PRA) utilizes probability distributions to identify variability or uncertainty in estimations of risk (Mitchell et al., 2004: 1-10). The method has been proved useful in many fields, including animal health (Roberts, 2006: 246). It is excellent tool for estimating the probability of an unwanted event occurring, such as contamination of food with pathogens (Roberts et al., 1995 cited in Roberts, 2006: 246). The output of a PRA is a range or probability distribution of risks experienced by the receptors. The performance of a PRA is limited by the availability of distributional data that sufficiently describe one or more of the input parameters. PRA can provide a quantitative explanation of the degree of variability and uncertainty in risk estimates for unwanted events such as the outbreak 22 of diseases. This can provide a more comprehensive identification of risk, additional information and potential flexibility that affords the risk manager (Mitchell et al., 2004: 1-10). Additionally, the beauty of PRA model is that it can illustrate the risk of trade-offs associated with various interventions. Once the risk trade-offs have been estimated, economic data can be added to estimate the benefits and costs of alternative options (Narrod et al., 1999 cited in Roberts, 2006: 251). Usually PRA models are tackled by a team composed of decision scientists, economists, modelers and subject matter experts such as veterinarians. The team attempts to capture the scenarios that can lead to significant levels of unwanted events such as contamination in model (Roberts, 2006: 246). However, PRA may not be suitable for every analysis since it generally requires more time, resources, and expertise (Mitchell et al., 2004: 1-10). 2.5 Bayesian Belief Network Analysis 2.5.1 Introduction Bayesian Belief Network (BBN) was invented in the 1940’s and 1950’s for the purpose of incorporating the effects of uncertainty in management systems for decision making (Henrion et al., 1991 cited in Dambacher et al., 2007). It is a graphical conceptual model that captures the components of analyst’s beliefs and probabilistic data in relation to the causal relationships of significant interrelated variables in the system of interest (Dambacher et al., 2007; Wongthanavasu, 2008; Carmona et al., 2011). Both quantitative and qualitative methods are employed in BBN to deliver advanced knowledge-based systems to solve real world problems (Harrison, 1997). The qualitative part represents causality, relevance and relationships between variables, while the quantitative part represents probability 23 distributions that quantify these relationships. Once a complete BBN is constructed it is an efficient instrument for performing inferences (Campos, 2006). In BBN the nodes represent stochastic variables. Each variable is characterized by states that can be indicated as numerical, ordinal, interval or nominal values (Wongthanavasu, 2008; Carmona et al., 2011). The relationships between the variables in a BBN are strictly acyclic (Dambacher et al., 2007) illustrated by the arcs connecting variables (Suermondt, 1992, p.12 cited in Krieg, 2001). For each variable, a conditional probability table (CPT) has to be defined relying on the available information, including Bayesian or physical probabilities. Bayesian probabilities are derived from prior knowledge including elicited judgment of experts and stakeholders in the form of the subjective estimates, whereas physical probabilities are obtained from available data in terms of statistical and empirical frequencies (Heckerman, 1996 cited in Krieg, 2001; Carmona et al., 2011). 2.5.2 Bayesian Statistics Bayesian statistics is the probability language applied to BBNs to determine the probabilities of each variable from the predetermined conditional and prior probabilities (Krieg, 2001). Therefore, Bayesian probability is considered one of the evidential probabilities enabling reasoning under uncertainty (Paulos, 2011). There are three key concepts in Bayesian statistics: a posterior probability, a likelihood function and a priori probability. The a posterior probability of a random event, say parameter θ, is the conditional probability that is assigned after the relevant evidence, say X, is taken into account: pθ|X. It is different from the likelihood function, which is the probability of the evidence given the parameters: 24 pX|θ. On the other hand, a priori probability is the probability distribution of the evidence. It is often the subjective assessment of experienced experts, regardless of any other information: pθ. Although prior probabilities have been criticized as a source of unwanted bias, they are considered as an integral part of human uncertainty reasoning (Jensen, 1996, p.19 cited in Krieg, 2001). The posterior probability is defined as; pθ|X = pX|θpX ∙ pθ The term p(X) is a normalizing factor. Suppose, X = {x1, x2, x3,…, xn} Using the law of total probability, p(X) = p(x1| θ)p(x1)+ (x2| θ)p(x2)+p(x3| θ)p(x3)+…+p(xn| θ)p(xn) For discrete distribution, p(X) = ∑ p|θp For continuous distribution, p(X) = px |θpx di (Krieg, 2001; Watthayu and Peng, 2004; Christopher, 2006) 2.5.3 Decision Theory There are three elements to be considered in decision theory. The first element is actions which are the alternative choices that a decision maker can choose to make. Another element is states which are the uncertainties that the decision maker cannot control. The last element is consequences which are the outcomes of making that particular decision under the uncertainty (Lenk, 2001). 25 2.5.4 Bayesian Influence Bayesian inference is a process of drawing conclusions from random events in which Bayesian interpretation is applied to illustrate how a subjective degree of belief should rationally alter the consideration of additional evidence. The advantage of Bayesian inference is that it always yields an accurate answer even when no data are available (de Finetti, 1974; Dawid, 1982; Ferson, 2005). Decision theory and Bayesian inference provide a consistent theoretical framework for decision making to solve complex and real-world problems. The management objectives are determined as a function, and the expected outcomes of management choices are calculated under the uncertainty (Dorazio and Johnson, 2003). Expected utility: For discrete function, EUD|p = ∑ Uω , Dpω  For continuous function, EUD|p = Uω , Dpω di Where, E[U(D)|p] = Expected utility or expected consequences from a decision making under uncertainty p(ω  = Probability of events that decision maker cannot control We choose D to maximize EUD|p. However, in statistics, we normally use loss function instead of utility function; Lω , D = −Uω , D Expected loss: For discrete function, ρp, D = ∑ Lω , Dpω  = ELω , D 26 For continuous function, ρp, D = Lω , Dpω di = ELω , D The objective is to make a decision (D*) that minimize the expected loss (ρ*) ρp, D∗ = ρ∗p (Lenk, 2001) Bayesian Belief Network can be applied to solve decision problems by extending two additional types of nodes: decision nodes and utility nodes (Watthayu and Peng, 2004). A decision node is a node in an influence diagram that represents action alternatives under the control of the decision maker (Watthayu and Peng, 2004; Norsys Software Corp, 2013). When the net is solved a decision rule that indicates choices in making a certain decision for each possible condition will be found for the node that optimizes the expected utility (Norsys Software Corp, 2013). Instead of holding conditional probability table (CPT) a utility node holds a table of utility values imposed by the decision maker by manual calculation for all value configurations of its parent nodes that meet the optimization objective (Jensen, 1995 cited in Watthayu and Peng, 2004). 2.5.5 Building Networks Designing a BBN involves these following steps (Heckerman, 1996 cited in Krieg, 2001): a) Identify the objectives of the model b) Identify sources of data to achieve these objectives c) Include only the meaningful and worthwhile data in the model d) Transform the data into variables 27 e) Identify thorough states of each variable f) Determine the causal structure between the variables Currently, BBN is becoming increasing popular for policy modeling of livelihoods and natural resource management problems such as water resource management (Cain, 2001; Ames et al., 2005), ecological risk management (Pollino et al., 2007), ecological modeling and conservation (Marcot et al., 2006), and wetland development (Gibbs, 2007). The study of Dambacher et al. (2007) proves that BBN is transparent, repeatable, makes experimental predictions statistically testable, and does not require large amounts of empirical data. However, the most significant drawback of BBN is the time, expertise and data needed to realistically represent complex problems. Chapter 3 Research Strategy and Methodology This part covers the research strategy and methodology; including (1) population and sampling design, (2) data collection, and (3) data analysis. This study has utilized the Bayesian Belief Network (BBN) as quantitative and qualitative instruments for the data analysis. 3.1 Population and Sampling Design 3.1.1 Population The target population of this study includes ethnic minority groups residing scattered throughout the mountainous region along Thailand- Myanmar borders in Chiang Mai, Chiang Rai, and Mae Hong Son, including Akha, Hmong, Karen, Lahu, Lisu, and Yao ethnicities with a total population ofapproximately 378,000 persons or 1,200 villages (for more information, see appendix A). 3.1.2 Sampling Design To determine the appropriate villages; it requires background information of the outbreak of Trichinosis provided by the Office of Disease Prevention and Control 10, the Bureau of Epidemiology, and the interview with knowledgeable individuals including the officers from the Department of Livestock 29 Development in Chiang Mai, Chiang Rai, and Mae Hong Son. The reported cases of Trichinellosis by sub district during 2003-2012 are mapped using Google Earth. Two highlanders’ villages in Mae Ai district, Chiang Mai Province were selected to conduct the in-depth study, including one that experienced an outbreak namely Huai Chan Si village and one that has never experienced an outbreak namely Huai Ma Fueang. There are a total of 84 households in Huai Chan Si village and 118 households in Huai Ma Fueang village. Twenty-six households from Huai Chan Si village and 28 households from Huai Ma Fueang village were selected using simple random selection (See Figure A-3 and Figure A-4). 3.2 Data Collection The survey instruments including questionnaire, environmental survey, in-depth interview, evaluation form and focus group are used in this study. 3.2.1 Institution Survey To understand the roles of institutions, we conducted in-depth interviews with staff working at the Department of Livestock Development, the Tambon Health Promoting Hospital, the Bureau of Epidemiology, and the Office of Disease Prevention and Control 10 in reducing parasitic zoonoses transmission. 3.2.2 Household Survey A questionnaire was developed for the household survey based on the Trichinellosis risk factors deriving from experts’ opinion. Twelve enumerators including 8 students from the faculty of Veterinary Medicine and 4 students from the 30 faculty of Economics, Chiang Mai University were trained on how to conduct the questionnaire in the selected villages and at the same time the questionnaire is tested. 3.2.3 Environmental Survey An environmental survey form was developed by an expert to investigate environmental factors related with Trichinellosis risk. To help the enumerators to understand the transmission of the disease, they were trained by the experts to understand One Health Approach. This form considers the interaction of highlanders with the pigs they grow and their environment as a single system. 3.2.4 Focus Groups After conducting the household survey, we developed a set of data preparing for the experts to evaluate the Trichinellosis risk circumstance in the selected villages using experts’ meeting. Seven experts are invited to join the focus groups, including; a) Animal Health Experts Assist.Prof.Panuwat Yamsakul Faculty of Veterinary Medicine, Chiang Mai University Dr.Veerasak Punyapornwithaya Faculty of Veterinary Medicine, Chiang Mai University Ms. Pornpen Tablerk Department of Livestock Development, Nan Province b) Disease Ecologist Prof. Bruce A. Wilcox Integrative Research & Education Program, Faculty of Public Health, Mahidol University and Tropical Disease Research Laboratory, KhonKaen University 31 c) Human Health Experts Assoc.Prof.Dr.Pichart Uparanukraw Faculty of Medicine, Chiang Mai University Assoc.Prof.Dr.Nimit Morakote Faculty of Medicine, Chiang Mai University Mr. Adulsak Wijit The Office of Diseases Prevention and Control 10 3.3 Data Analysis 3.3.1 Descriptive Statistics Descriptive statistics were used to quantitatively describe the collected data. They are divided into 4 sub-systems, including: animal husbandry, food chain, environment, and economic condition. 3.3.2 Modeling To conduct an in-depth household study within the limited time and financial resources, only 54 households were randomly selected. The complexity of these circumstances has led to model-based approaches for investigating the interconnections and for predicting management outcomes (Jakeman et al., 2006). A probabilistic graphical model for qualitative instrument called BBN is applied for this analysis since it does not need large amounts of empirical data. The conceptual transdisciplinary framework of Trichinellosis risk is developed by experts based on the existing knowledge and the experience from the field study to explain interconnection of the risk factors. It is also applied to solve decision problems related with management of the relevant institutions attempting to reduce the risk. When the net is solved, a decision rule which indicates choices for making a certain 32 decision for each possible condition will be found for the node that optimizes the expected utility. a) Purposes of the Modeling The purposes of this modeling are to, first, gain a better understanding of the transmission of Trichinellosis, second, solve decision problems in management systems related to pig production and the public health situation to reduce Trichinellosis, and, finally, develop a universal Trichinellosis risk model explaining the circumstance in other areas. b) Developing the Models The Trichinellosis risk framework was developed based on the opinions of veterinarians, disease ecologists, medical doctors and public health officers (See Figure B-1). There are a total of 77 variables to be studied categorized into four subsystems to investigate Trichinellosis risk, including: animal husbandry, food chain, environment and economic condition.There are two kinds of variables in this study, including discrete data and continuous data. These variables are associated with probabilistic functions (the states of each variable are explained in the Appendix E). There are two sources of information to feed in the model, including the data from the field study and the data from experts’ opinions. Netica, a powerful and easy- to-use program for working with BBN and influence diagrams are applied to analyze this set of data. 33 c) Specifying Modeling Context We broke the Trichinellosis risk framework into two parts based on the decision problems that we attempted to investigate. There are two main decision problems in management systems related with pig production and public health situation to reduce Trichinellosis, including the decision to switch from the original pig production mode of keeping pigs in pens and the decision to stop consuming raw or undercooked meat. a. Institution’s Decision to Encourage Villagers to Switch to Keep Pig Pen Figure 3-1: Decision tree representing the institution’s decision to encourage people to switch to keep pig in pen 34 U1 = f(X8, D1, X22) Given, U1 = Benefits from switching to keep pigs in pen X8 = Pig production modes D1 = Institution’s decision to construct pig pen X22 = Household’s decision to keep pigs in pens In order to reduce the possibility of getting infected by Trichinella and other parasites in animals, a complex set of issues must be considered, mainly social and economic trade-offs. In considering a campaign to change the original pig production mode to keeping pigs in pens, the benefit of doing this is the reduction in the possibility of getting infected by Trichinella and other parasites in animals which in turn yields a higher productivity and reduces the risk of getting Trichinellosis and other parasitic zoonoses in humans. However, keeping pigs in pens bears a huge cost to the farmer. The cost of construction is seen as a small portion if we take the opportunity costs into consideration (more details of cost structures and revenue streams of different pig production modes can be seen in Appendix G). Those who do crop farming as a primary career need to devote their time in a field whichis located far away from the village. Many of them decide to let their pigs roam freely because they do not have to prepare feed for them which takes hours to prepare. Assuming that an institution, for example, a governmental institution, has unlimited money to construct pig pens for villagers and that will not affect to its utility. The institution’s decision whether to provide money to construct pig pens (3,981.43 Baht each, the average cost derived from the field study) for the pig growers is based on the satisfaction of a household from switching the practices and that we considered only the average gains (or losses) a household 35 will face if it changes the practice. If the institution does not want to support the money to construct pig pens for the pig growers, they have to bear this cost by themselves. We calculated the average gains of each pig production mode and compared those with the average gain from raising pigs in pens (see Table C-1). We ignore the possible benefits from the reduction in the risk that pigs will be infected by Trichinella from keeping pigs in pens. b. Institution’s Decision to Encourage People to Stop Consuming Raw or Undercooked Meat Figure 3-2: Decision tree representing the institution’s decision to encourage people to stop consuming raw or undercooked meat U2 = f(RTH, X60, D2, X64) Given, U2 = Benefits from decision to stop eating raw/undercooked meat RTH = Risk of getting Trichinellosis in human X42 = Meat preparation D2 = Institution’s decision to encourage people to stop consuming 36 raw/undercooked meat X46 = Individual’s decision to stop eating raw/undercooked meat The second decision is to decide whether the government should go to the field to encourage people to stop consuming raw or undercooked meat. In each year, the public health officers in both local offices and provincial offices have put in effort trying to encourage people, especially those who live in the country side, to stop consuming raw or undercooked meat by providing them knowledge about the danger of consuming raw or undercooked meat. Even though these people are educated about the harm of consuming raw or undercooked meat, they still insist on consuming it. This means that no matter how much public health officers put in effort to encourage people to stop consuming raw or undercooked meat, if they are not aware of the danger, they still will not change their behavior. Therefore, if the public health officers understand the behavior and attitude of the villagers very well, they can decide whether they should keep educating them of the danger of consuming raw or undercooked meat or should stop and rather put the effort on other issues instead. On the other hand, if the decision makers know that some villagers are undereducated about the danger of consuming raw or undercooked meat and if they are educated they tend to change behavior, therefore, the effort that public health put will be quite effective and worth the money and time. Contrary to the previous decision, with this decision we assume that an institution, for example, a public health organization, has limited money to encourage people to stop eating raw or undercooked meat and that is its decision to allocate resources wisely. A local institution spends approximately 6,000 Baht each time it visits village providing knowledge about hygiene. 37 The decision of an institution whether to launch a campaign to encourage people to stop eating raw/undercooked meat depends on benefits a household will receive from stopping eating raw/undercooked meat. These benefits can be calculated from the reduction in the burden of illness or the loss from death from Trichinellosis. In so doing, we consider the risk that an individual can get Trichinellosis that is evaluated by human health experts. We also take the severity of the illness in to consideration. However, since we do not have enough information to calculate the severity of getting Trichinellosis, we assume that those who have higher risk may face higher severity. The higher severe case bears higher economic losses. There are three levels of severity including high, medium and low levels. We derived the data on the economic losses from the illness from the case outbreak in Nan Province. Assoc.Prof.Dr.Pichart Uparanukraw, a human health expert determined the levels of severity of the illness from the case outbreak. In addition, we also take the individual’s decision whether to stop eating raw/undercooked meat or not and the costs that the institution bears in visiting a village in order to provide the knowledge to villagers into account. The economic losses of illness and death from Trichinellosis in human can be seen in Table D-1. c) Model Structure and Parameters a. Decision to Switch to Keep Pigs in Pens Posterior Probability Equation of TIP Based on Bayesian statistics, the posterior probability for this model is defined as; pTIPX = pX |TIP pX ∙ pTIP 38 Where, i = Levels of the risk that pigs will be infected by Trichinella = {H,M,L} XA = All risk factors associated Trichinella infection in pigs = {X1, X2,…, X57} pTIPX = Posterior probabilities (or probabilities of the parametersTIP) given evidence X pX |TIP = Likelihood functions (or the probabilities of evidence X ) given the parametersTIP pTIP = Prior probability probabilities (or the probabilities of risk that pigs will be infected by Trichinella based on the subjective assessment of experienced experts) p(X ) = Probability of all evidences in set X , regardless of any other information From the law of total probability, p(X ) = p(X|X1=H)p(X1=H)+p(X|X1=M)p(X1=M)+p(X|X1=L)p(X1=L)+ p(X|X2=H)p(X2=H)+p(X|X2=M)p(X2=M)+p(X|X2=L)p(X2=L)+…+ p(X|X57=H)p(X57=H)+p(X|X57=M)p(X57=M)+p(X|X57=L)p(X57=L) Expected Utility Function of the Decision to Switch to Keep Pig in Pen EUD pX  = U X8j , X22k ,D1l  p X8j  p X22k  pD1l ,, Where, EUD pX  = Expected utility or expected consequences from a decision making of supporting pen construction for pig growers under uncertainty about Trichinellosis risk. j = {P,T,FU,FO,P+T,P+FU,P+FO,P+F,T+FU,T+F,P+T+F,P+FU+F,P+FO+F} k = {N,Y} 39 l = {N,Y} However, X22 = f(TIP), and we are interested to see the effect of TIP on the expected outcome (U1). The posterior probability of X22 is defined as, p!X"" TIP# = pTIP|X"" pTIP ∙ pX"" ∴ p!X"" # = p!X"" TIP# pTIP pTIP|X"" From the posterior probability of TIP, pTIPX = pX |TIP pX ∙ pTIP pTIP = pTIPX pX pX |TIP p!X"" # = p!X"" TIP#pTIPX pX pX |TIP pTIP|X"" ∴ E&U1D1 pX ' =  U1!X( , X"" TIPi , D #p!X( #pD p!X""  TIP#pTIPX pX pX |TIP pTIP|X"" i,j,k,l An institution will make a decision whether to support pen construction for pig growers or not based on expected utility maximization. However, in statistics, we normally use loss function instead of utility function; L!X( , X""  TIP , D # = −UX( , X"" TIP , D Expected loss: 40 ρpTIP|X , D =  L1!X( , X"" TIPi , D1l#p!X( #pD1l p!X""  TIP#pTIPX pX pX |TIP pTIP|X"" i,j,k,l The objective is to make a decision (choose whether to support pen construction for pig growers or not) that minimizes the expected loss (ρ*) based on the posterior probability of TIP. ρpTIP|X , D∗ = ρ∗pTIP|X b. Institution’s Decision to Encourage People to Stop Consuming Raw or Undercooked Meat Posterior Probability Equation of RTH Based on Bayesian statistics, the posterior probability for this model is defined as; pRTH0|X1 = pX1|RTH0 pX1 ∙ pRTH0 Where, m = Levels of the risk that humans will be infected by Trichinellosis = {H,M,L} X1 = All risk factors associated Trichinella infection in pigs = { X1, X2,…, X69} pRTH0|X1 = Posterior probabilities (or probabilities of the parametersRTH0) given evidence X1 pX1|RTH0 = Likelihood functions (or the probabilities of evidence X1) given the parameters RTH0 41 pRTH0 = Prior probability probabilities (or the probabilities of risk that human will be infected by Trichinellosis based on the subjective assessment of experienced experts) p(X1) = Probability of all evidences in set Y, regardless of any other information From the law of total probability, p(X1) = p(X|X1=H)p(X1=H)+p(X|X1=M)p(X1=M)+p(X|X1=L)p(X1=L)+ p(X|X2=H)p(X2=H)+p(X|X2=M)p(X2=M)+p(X|X2=L)p(X2=L)+…+ p(X|X69=H)p(X69=H)+p(X|X69=M)p(X69=M)+p(X|X69=L)p(X69=L) Expected Utility Function of the Decision to Stop Consuming Raw or Undercooked Meat EU"D" pXB  =  U"!RTHm, X456 , X476 , D"8#pRTHm pX456 pX476 p!D"8#0,6,9,8 Where, EU"D" pXB  = Expected utility or expected consequences from decision making to encourage people to stop consuming raw/undercooked meat under uncertainty about Trichinellosis risk. m = {H,M,L} n = {R,C} p = {N,Y} q = {N,Y} However, we are interested to see the effect of RTH on the expected outcome (U2). From the posterior probability of RTH, pRTH0|X1 = pX1|RTH0 pX1 ∙ pRTH0 pRTH0 = pRTH0|X1 pX1 pX1|RTH0 42 ∴ EU"D" pXB  =  U"!RTHm, X456 , X476 , D"8#pX456 pX476 p!D"8#0,6,9,8 p!RTHmXB# p!XB#p!XBRTHm# An institution will make a decision whether to encourage people to stop consuming raw/undercooked meat or not based on expected utility maximization. However, in statistics, we normally use loss function instead of utility function; L"!RTH0, X456 , X476 , D"8# = −U"RTH0, X456 , X476 , D"8 Expected loss: ρ"pRTH0|Y , D" =  L2RTH0, X60n , X64n , D2q pX60n pX64n pD2q m,n,p,q pRTH0|X1 pX1 pX1|RTH0 The objective is to make a decision (choose whether to encourage people to stop consuming raw/undercooked meat or not) that minimizes the expected loss (ρ"*). ρ"pRTH0|X1 , D"∗ = ρ"∗pRTH0|X1  d) Testing the Modeling The objective of this test is to evaluate the quality of the Bayesian Networks (Appendix B) using a set of real cases using Netica. This test will illustrate how well the models match the actual cases by considering the actual belief 43 levels of the states in determining how well they agree with the value of the case file. We first incorporate 60% of the cases into the model. Then, the nodes in which we wish to find their inferences, including, TIP and RTH nodes were selected. We used 40% of the samples to verify the validity of the model. When the Netica was done, it printed a report called scoring rule results of each of the selected nodes (see Table C-3 and Table C-4). The reports included error rate, logarithmic loss score, quadratic (Brier score), and spherical payoff score. Error rate determines how many times the classifier misclassifies a case divided by the number of classifications. It is only with respect to the probability distribution of the test cases. Logarithmic loss values are calculated using the natural log. The values are between zero and infinity. Zero indicates the best performance. Logarithmic loss = MOAC [- log (Pc)] Quadratic loss values or the Brier score are between zero and two. Zero indicates the best performance. Quadratic loss = MOAC [1 – 2(Pc) + ∑ p"6 ] Spherical payoff values are between zero and one. One represents the best performance. Spherical payoff = MOAC [ ABC∑ 9DEFD ] Where, Pc = Probability predicted for the correct state Pi = Probability predicted for state i, n is the number of states MOAC = Mean (average) over all cases (Norsys Software Corp., 2013) 44 Another way to verify the validity of the models is to use Netica to pass through the case file by processing cases one-by-one. For each case, the software reads the case except the nodes that we wish to find their inferences. After that, the software will revise the actual value for those nodes and compare them with the beliefs the model generated. Netica accumulates all the comparisons as illustrated in Table C-4 and Table D-4. The models were selected based on the values of sum square error (SSE). The less SSE, the best the model is. For animal health perspective, the values of logarithmic loss and quadratic loss of the model 2 were slightly smaller than the model 1, while the value of spherical payoff value of the model 2 was slightly larger. Though, the model 2 yielded slightly larger of sum square error (SSE) than the model 1, the model 2 was selected since it yielded a lot less error rate of only 20%. For the human health perspective, the scoring rule results and the error rate yielded no difference values between the model 1 and the model 2. Though, model 2 yielded a slightly larger value of SSE, it was selected since it was much less complicated than model 1. After we derived the models, we incorporated all the data into the selected models. As a result, we would see the learned probability distributions appeared in each node (see Figure C-1, C-2, D-1 and D-2). In order to solve the decision problems, we augmented the decision node and the utility node into the models. For the human health perspective, we augmented the institution’s decision whether to encourage people to stop consuming raw or undercooked meat or not. For animal health perspective, we augmented the institution’s decision whether to support the constructing cost of pens 45 to pig growers or not. Netica would attach a deterministic function which provided a value for the decision node for each possible configuration of parent values. The links into a decision node indicate what the decision maker will know when he is about to make the decision. For the human health perspective, we assume that the institution may know the knowledge of food-preparing persons and their attitudes to change the eating habits. On the other hand, for the animal health perspective, we assume that the institution may know the pig production mode that pig growers apply and their attitudes towards changing the practices. The decision function from the decision node will maximize the expected value of the sum of the utility node (see Figure C-3 and Figure D-3). Chapter 4 Research results This chapter covers the results of a transdisciplinary Trichinellosis risk framework involving the descriptive results of two highlander villagers, including one that experienced an outbreak, and the results from the development of a Bayesian Belief Network model of Trichinellosis risk. 4.1 Transdisciplinary Trichinellosis Risk Framework This section used a One Health approach to develop a transdisciplinary framework considering the interaction of highlanders with the pigs they grow and their environment as a single system. The research identified four subsystems to investigate Trichinellosis risk, including animal husbandry, food chain, environment, and economic conditions. Descriptive statistics were used to quantitatively describe the collected data (see Appendix F). 4.1.1 General Information The in-depth household study was conducted in two highlander villages in Mae Ai District, Chiang Mai Province, including Huai Ma Fueang village and Huai Chan Si village which experienced an outbreak of Trichinellosis in 2004 (BOE, 2004). Fifty-four representative households were randomly selected from these two villages. Out of this number, twenty-three representative households raise 47 pig, and five and three households do not raise pig in Huai Ma Fueang village and Huai Chan Si village, respectively. Most of the respondents (34 persons) are responsible for both pig rearing and food preparation, eleven persons prepare food only and nine persons raise pigs only. On average, they are almost 40 years old. There are approximately 5 persons in the family. 4.1.2 Animal Husbandry-relevant Trichinellosis Risk Factors The data show that females are responsible for pig rearing in 26 households while males are responsible for pig rearing in 20 households. A majority of the pig raisers are Red Lahu (39 persons) and the minority includes Lisu (4 persons), Black Lahu (1 person) and Palong (1 person). Most of them are Buddhists (35 persons) while some of them are Christians (10 persons) and animist (2 persons). In addition, those who are Buddhists or Christians, they also follow their own ethnic culture, including values, beliefs and special celebrations. Most of these households raise pigs for both consumption and commercial purposes (36 households). Some of them raise pigs for self subsistence only (8 households). Among these households, most of them raise pigs for ritual uses (28 households). On average, they have 11 years of experience in pig rearing, but some of them have raised pigs for up to 35 years. It takes approximately 10 months for growing a pig until they can sell for approximately 4,200 Baht for 38 kilograms weight. In considering the pigs these highlanders grow, they are all native pigs (black pigs). Some ritual ceremonies only allow black pigs as the oblation. One villager said that if they do not grow black pigs, when they need one for a ceremony 48 they have to buy it from somewhere else and they have to accept at any price. This is the reason why these people tend to raise pigs by themselves since it saves more money. They will buy more pigs (usually piglets) from neighbors or sometimes from others outside the village when they think there are not enough pigs for their ritual uses or special celebrations such as the New Year, weddings, funerals, etc. Raising pigs can be seen as a kind of investment, and they can invest in a pig with little money. Then a pig grows with time and they can sell it out when they need money for contingency situations or they can wait until it is grown and consume it and save money instead of buying it from another farmer. This is quite a wise decision for them in utilizing their time, land space and other resources available. The profits they will earn from pig rearing depend largely on how they manage these resources and the kinds of pig production modes they apply. There are five kinds of pig production modes, including free range in which farmers allow pigs to wander around finding something to eat by themselves. For this kind of pig production mode, farmers do not need to construct a house for pigs and feed is rarely prepared for them. Usually piglets are allowed to wander around. Tethering pigs with ropes can be seen as another kind of pig production mode that farmers tie a pig with a rope, usually underneath the house for keeping it nearby and they have to prepare feed for it. The other two kinds are the fence located underneath the house and the fence located outdoors. Farmers construct fences for grounding pigs and feed needs to be prepared for them. The last kind of pig production mode is in pens. Farmers construct a pen as a house for pigs. The quality of the pen depends on the budget that farmers have. Most boars and sows are usually kept in pens. Most of these households keep pigs in pens (18 households) 49 following by in fenced areas underneath the house (5 households), tethering (1 household) and outdoor-located fenced area (1 household). Many of them apply a combination of various pig production modes such as a combination of outdoor- located fences and pens (4 households), a combination of fences underneath the house and pens (4 households), a combination of tethering and pens (3 households), etc. The majority of these households claim that they choose the kind of pig production mode depending on convenience. Some of them follow the community regulation of grounding pig in pen or fence. If a pig disturbs any other villagers’ property, the pig owner has to be fined 300 Baht. Surprisingly, many of them are concerned about the hygienic security, and that is why they keep pigs in pens. In addition, they also claim that financial constraint plays a big part on their decision of choosing a pig production mode. A majority of these representative highlanders kill pigs by themselves in the backyard (44 households). Only one household goes to local butcher and one household kills pigs inside the pen. Major feeds that these people feed the pigs are banana trunk and rice chaff (43 households). Some of them use food scrap (23 households) and carcasses left over (3 households) as pig feed. These can be considered as a risk for infection with Trichinellosis in pigs since they have a chance to consume infected animal parts. In considering the pig raisers’ knowledge about Trichinellosis, a majority of them (49 households) do not know anything about this disease. However, twenty-eight households know that keeping pigs in pens is hygienically safe for the pigs. For those who do not keep pigs in pens, most of them (27 households) tend to 50 change the behavior by putting pigs in pens after they know that pens can lead to hygienic security in the pigs. However, only 15 households decide not to change the behavior mainly because of the financial constraint. In considering the health practices, most of the pig raisers (32 households) never check the health situation of their pigs, but the rest of them have checked the health situation of their pigs. They tend to check the health situation of their pigs by themselves (7 households). Furthermore, when their pigs gets sick most of them (18 households) usually treat them themselves. Some of them (12 households) do nothing. Some of them (10 households) ask for someone such as the public health volunteers which are their neighbors to help. A few (5 households) use herbal remedies and only one person (1 household) sells the sick pig. Surprisingly, thirty-eight households apply deworming drugs for their pigs with almost two applications annually. Though, deworming drugs cannot effectively kill parasites located in muscles, the high tendency of applying deworming drugs amongst this population shows a good sign that they are aware of the parasitic diseases in pigs. For those who do not apply the deworming drug, they claim that it is not important to use it (4 households). Some of them claim that their pigs are already healthy (2 households), while one household thought that their pigs are too old, no need to use the deworming drug anymore. In addition, only two households do not even know that they should apply deworming drug for their pigs. 4.1.3 Food Chain-relevant Trichinellosis Risk Factors Most of food-preparing persons (40 persons) in these representative households are female. Most of them are Red Lahu (45 persons) and 51 most of them have already been granted Thai citizenship (52 persons). Thirty-five had never attended school. This probably is the reason why most of them (39 households) do not know that consuming raw or undercooked meat is harmful for their health. In these two villages, in each year, villagers usually kill native pigs on the Lahu New Year festival (or Kin Wor) (41 households), wedding ceremonies (28 households), funerals (27 households), merit making (24 households), sacrifices (17 households), and New Rice Alms’ ceremony (15 households). Additionally, the pork is almost always shared with the neighbors and even with visitors or strangers. Nonetheless, the villagers also eat outside of the house. Most of them (31 households) go to their neighbor’s house. From the interview, we found that they usually have alcoholic drinks with raw or undercooked meat. Twenty-six of them go to a restaurant and a few of them go to other villages, churches and markets. Twenty-nine households go to eat outside occasionally, especially when there is a special occasion. Only a couple households go to eat outside everyday and only 5 households never go to eat outside at all. Fortunately, most of the villagers in these representative households (69.44%) do not like consuming raw or undercooked meat, and most of those who do not like it (72.73%) said it is “nasty”. Around 45.46% of them claim that it is harmful for health. Only a few of them said that their parents do not allow eating raw or undercooked meat (3.03%) and some (3.03%) claimed that the materials for preparing raw or undercooked meat are more expensive. 52 However, for those who like to consume raw or undercooked meat, around 90% of them claimed that it is “delicious”. Some of them (31.71%) believe that it is a tradition to eat it. Surprisingly, many of them (31.71%) believe that consuming rawor undercooked meat can give them strength. Around 12% of them love to eat it with alcohol and a few of them think it is “cool” (2.44%) and they feel used to consuming raw or undercooked (2.44%). After the enumerators educated the danger of consuming raw or undercooked meat to these people, only around 60% of them said they would stop eating raw or undercooked meat. All of them said they are scared of the danger. Surprisingly, around 21% of them insisted on continuing to eat it as usual and the rest claimed to eat it only on special occasions (14.89%). A few of them claimed that they will eat less (2.13%) and some are still unsure (2.13%). For those who insist on continuing to eat raw or undercooked meat, around 62% of them said that it is because of their own preference. A few of them said it is because nothing bad ever happened to them (12.5%). Some of them claimed that they eat the deworming drug after consuming raw or undercooked meat (12.5%). A few of them said it is because of the tradition (6.25%) and some claimed that they have to eat it with their husband (6.25%). From the questionnaire, we found out that there are nine different kinds of animals that the villagers eat, including, white pig, native pig, chicken, wild boar, water monitor, wild cat, snake, dog and rat. These animals can be infected by Trichinella if they consume infectious cysts in meat. Humans become infected when they eat raw or undercooked infected meat. The study found that the villagers do not eat raw or undercooked chicken, water monitor, wild cat, snake, dog and rat while 53 they consume the rest as raw or undercooked. Most of the time they consume white pig (annually around 260 days on average as for cooked and around 86 days for raw or undercooked) that they can buy from other villagers (71.70%), from neighbor (28.30%) and with their own reproduction (3.77%). The villagers often consume native pig (annually around 73 days on average as for cooked and around 33 days as for raw or undercooked) that they can buy from other villagers (9.26%), from neighbors (50%) and from their own production (61.11%). Since the villagers’ houses are located nearby a forest, they often hunt wild boar for consumption (25.93%). However, some of them buy it from other villages (14.81%), from neighbors (24.07%) and from their own production (3.70%). Annually, they consume wild boar around 12 days on average as for cooked and only around 2 days as for raw or undercooked. 4.1.4 Environment-relevant Trichinellosis Risk Factors After slaughtering pigs or other animals, most of the representative households (26 households) use the carcasses left over as pet feed. Some of them (7 households) just sweep them down to the floor. Only a few of them put carcasses in the trash can or bury them. Thirty-eight households bury animals that died of sickness. Unfortunately, a few of these households eat animals that died of sickness within the family, share to neighbors, or sell it out to other neighbors. For the naturally dead animal, forty-one households usually bury it. Amongst these households, a couple of them clean pig areas twice a day, ten households do it once a day, four households do it every other days, 54 four households do it every other two days, five households do it once a week, seven households rarely clean, and nine households never clean pig areas. In considering the pig waste disposal, a couple of these families dispose of it twice a day, eight households dispose of it once a day, seven households dispose of it every other day, a couple of households dispose of it every other two days, seven households dispose of it once a week, a few households rarely dispose of it and fourteen households never dispose of it at all. In considering disposal of the feed left over, twenty-two households dispose of it every time after feeding. Seven households dispose of it sometimes and thirteen households never dispose of it. In considering the cleanliness of surroundings, sixteen households clean surroundings every day, fourteen households do it every other day, fourteen households do it once a week and ten households rarely do it. Twenty-three households tend to use wet garbage as animal feed. Twelve households throw this trash in the forest nearby. Ten households sweep it down to the floor. Twelve households dispose it by burning. Eight households put it in a trash can, while only a couple households bury it. For solid waste, twenty-eight households dispose it by burning. Thirteen households put it in community trash can. Eight households throw it in forest. Seven households sell it out. A couple of households bury it and only one household reuses it. However, a couple households just sweep the trash down to the floor. Most households have seen around 1-5 rats a day. Most of these households do not recognize the danger of rats. In addition, twenty-five households use rat control. Furthermore, ten households have seen wildlife around the villages. 55 4.1.5 Economic-relevant Trichinellosis Risk Factors a) Financial Status As for a primary source of income, these representative households earn money from crop production (50 households), working as laborer (5 households), merchandise (3 households) and animal farming (2 households). As for a secondary source of income, they earn money from working as a laborer (29 households), animal farming (21 households), crop production (2 households) and selling merchandise (1 households). On average, a family earns 58,537.96 Baht annually. Forty households do save some money for different purposes, including, investment in agriculture (30 households), buying products (22 households), preparing for children’s education (12 households), preparing for contingency purposes (7 households), preparing for vacation (1 household). Some of them also provide loans for others (3 households). Thirty-one households use the savings money for pig production investment, including buying deworming drugs (20 households), buying more pigs (15 households), buying pig feed (5 households) and also for improving pig hygiene (4 households). Forty-two of these representative households have fallen into debt. Twenty-one households borrow money from their neighbors. Nineteen households borrow money from village funds. Eleven households are able to access money from the Bank for Agriculture and Agricultural Co-operatives (BAAC). Eight households access sources of funding through informal leasing. A few of them use private leasing or go to Government Saving Bank (GSB) to access sources of funding. 56 Since most of these people are Thai citizens, they can access the universal coverage for health care service (50 households). A few of them can access the health care service for free through other alternative choices such as the free medical service for elderly, public health volunteer, low income people and as a community leader. A couple of them have private insurance or registered in the social security service. Usually, when these people are sick, they will go to a district hospital (40 households), Tambol health promoting hospital (36 households), nearby clinic (23 households), Maharat hospital (7 households) which is a provincial hospital, Prasat neurological hospital (6 households) which is another provincial hospital, or Fang hospital (5 households) which is district hospital in another district. However, some of them decide to buy medicine by themselves at the pharmacy store (34 households). Some of them use a traditional health care (13 households) or spiritual treatment (9 households). In considering the convenience products that these people possess, these households have motorbike (48 households) that facilitates convenient travel from house to the cropping farm or to the town. Only a few of them own a truck (5 households). They also have television (45 households) with satellite dish (43 households) and radio (21 households). Many of them have a cell phone (19 households) and only a couple them have laptops, etc. b) Access to Information Amongst these people, the most important source of information and news is television (47 households). Those who do not have television may watch it with their neighbors who have it. Word of mouth from 57 neighbors is also seen as a channel to pass along information or news to other neighbors (38 households). Besides, some people receive news from the public announcement (38 households). Additionally, they also access information and news by listening to the radio (47 households), and through reading newspapers (5 households). A few of them use internet and read local journals or magazines. Furthermore, they also receive services from persons and institutions, including, headman (on average around 29 days visited in a year), teacher (on average around 13 days visited in a year), public health officers (on average around 3 days visited in a year), animal health volunteer (on average around 1 day visited in a year), public health volunteer (on average around 28 days visited in a year), animal health volunteer (on average around 2 days visited in a year), police (on average once in every other year), heifer officer (on average once in four years), tree bank officer (only once), sub district officer (on average once in four years) and district officer (only once). From the interview, we found out that, on average, most of these people are not satisfied with the services from these institutions. This may be because there are a lot of them do not receive some services from these people or institutions. c) Cost Structures and Revenue Stream of Highlanders’ Pig production modes The cost structure of pig rearing is composed of fixed cost from pig house and variable cost from feed cost, water supply and pig housing maintenance, and other miscellaneous variable cost. Different kinds of pig production modes have different cost structures and yield different revenue stream to the family as illustrated Appendix G.These costs are calculated until a farmer can sell a pig out. 58 As overall, the cost of pig rearing per capita is 926.41 Baht. A pig can be sold for approximately 2,900 Baht. The profit per capita is approximately 2,300 Baht. However, if we consider the opportunity costs including the cost of time spent for raising pigs and the cost saved from not buying pigs, the profit per capita will be approximately 2,200 Baht. From the study, there are two kinds of pig production modes that yield a loss to households. They include pens (3 households) and fences underneath the house (2 households). The combination of raising pigs in pens and tethering yield the highest profit per capita (4,193.24 Baht) while the combination of raising pigs in outdoor-located fences, pens and free range yield the lowest profit per capita (423.13) to the household 4.2 Bayesian Belief Network Model of Trichinellosis Risk The conceptual transdisciplinary framework of Trichinellosis risk is developed (see Appendix B) by experts based on the existing knowledge and the experience from the field study to explain interconnection of the risk factors. It is also applied to solve decision problems associated with management of the relevant institutions attempting to reduce the risk. The decision problems include (1) institution’s decision to support money for pig pen construction to pig growers and (2) institution’s decision to encourage people to stop consuming raw or undercooked meat. 59 4.2.1 Institution’s Decision to Support Money for Pig Pen Construction After we incorporate the data into Model 2 (see Figure C-3), the outcome node illustrates the probability that pigs are at high risk to be infected by Trichinellais 31.48%, at medium risk is 37.04% and at low risk is 31.48%. For the attitudes toward changing the practices of the pig growers, without being educated, the probability that they will not change the practice is 50.27% and the probability that they will change is 49.73%. This means that they are reluctant whether to change the practice or not. With these circumstances, if the institution decides to launch a program to support money for pig pen construction to pig growers in attempting to reduce the risk that pigs will be infected by Trichinella, a household will receive a negative outcome of 3,912.20 Baht per household (in case that the household also agree to change the practice). However, if the institution decides not to launch the program, a household will receive a negative outcome of 5,747.80 Baht per household. Though, without taking possible benefits from the reduction in the risk that pigs will be infected by Trichinella from keeping pigs in pens into consideration and under the assumption that the institution has unlimited resources to construct the pen, we can say that a household will be better off if an institution supports a budget to construct pig pens. Moreover, if we assume that a household receives news from someone that the village has a high prevalence of Trichinella infection in pigs, the model shows that there will be a slight increase in the willingness to change the practice to keep pigs in pens and the decision from the institution to support pen construction costs for pig growers still outperformed the decision of not to support (see Figure C-4). This showed that the information that their pigs are at high risk to 60 be infected with Trichinella has an effect on the decision of pig growers to switch to keep pigs in pens. 4.2.2 Institution’s Decision to Encourage People to Stop Consuming Raw or Undercooked Meat After we incorporate the data into Model 1 (see Figure D-3), the outcome node illustrated the probability that people are at high risk to be infected with Trichinellosis is 16.50%, at medium risk 19.65% and at low risk 63.85%. For the attitudes toward changing the behaviors, without being educated about the danger of consuming raw or undercooked meat, the probability that they will not change the habits is 38.60% and the probability that they will change is 61.40%. This means that they tend to change the habits by themselves easily. With these circumstances, if the institution decides to launch a program to encourage people to stop consuming raw or undercooked meat in order to reduce the risk that people will be Trichinellosis, an individual will receive a negative outcome of 3,912.20 Baht. However, if the institution decides not to launch the program, an individual will receive a negative outcome of 5,747.80 Baht. Since we already include the cost of visiting the village into the model, in this case, we can say that an institution should go to the field and encourage people to stop consuming raw or undercooked meat because it yielded less loss than not to. Moreover, if an institution has heard news that a village is at high risk that people will be infected with Trichinellosis, this information can be updated the model and the decision making process. In this case, if the institution launches a program to encourage people to stop consuming raw or undercooked meat, an 61 individual will receive a loss of 6,000 Baht which is much smaller than the decision of not to (as high as 20,681 Baht) (see Figure D-4). Chapter 5 Discussion and Conclusion This chapter concludes this thesis. It presents a summary of the research. Findings of the study and are discussed and interpreted. Recommendations for further study are also provided at the end of each section. 5.1 Research Summary To help us understand the transmission of the disease the study used a One Health approach to develop a transdisciplinary framework. This framework considers interaction of highlanders with the pigs they grow and their environment as a single system. The study identified four subsystems to investigate Trichinellosis risk: animal husbandry, food chain, environment, and economy. The results of a transdisciplinary process involved the development of a Bayesian Belief Network model of Trichinellosis risk and in-depth study of two highlander villagers, including one that experienced an outbreak. The study developed and tested a novel survey instrument consistent with the model and the One Health approach. The model and our survey results suggested the above subsystems, including pig husbandry, food chain, environment, and highlanders’ economic circumstances are entirely interdependent, and thus must be considered as an integrated whole when devising disease interventions. 63 5.2 Discussion and Interpretation of Findings 5.2.1 Transdisciplinary Trichinellosis Risk Framework a) Animal Husbandry The study found that females have equal opportunity to be part of the pig rearing process. In fact, most of pig growers are females. On average, the pig growers have high experience in pig rearing.The study also presented that the pig growers acquire knowledge about pig rearing from public health officers who are their neighbors. Some households use local wisdom in healing sick pigs such as using herbs. From the interview, we can see that these people have naïve morality. They claimed that they never sell any sick pigs to others. Surprisingly, many households apply deworming drugs to their pigs twice annually. Though, this practice cannot effectively kill parasites located in muscles, it shows a good sign that they are aware of parasitic diseases in pigs. However, most of the pig growers in these villages do not recognize that keeping pigs in pens can prevent their pigs from parasitic zoonoses. After they were educated by the enumerators, most of them tended to change their practices to keep their pigs in pen. However, the main reason that some people decided not to change their practices is because of the financial constraint and time that they needed to devote to pig rearing. b) Food Chain The study found that, each year, villagers kill native pigs on the special occasions related with traditions and beliefs. In addition, the pork is almost always shared with the neighbors and even with visitors or strangers and 64 always served raw. These people may pass along meat that is infected by Trichinella or other parasitic diseases to others. These people have a high tendency to consume a large amount of infected meat in these special occasions. On the other hand, the parasitic zoonoses experts claimed that eating outside of the house such as at arestaurant, market or church haverelatively lower risk since the infected meat may mixed with non-infected meat and shared with several people. The study also found that eating habits and food choice depend largely on personal preference as well as the influence of peers or family members. There is a possibility that children will follow their parents’ behavior and acquire the same eating habits as adults. Knowledge can prevent some people from eating raw meat.Additionally, for those who eat raw meat for pleasure (good taste), being informed about the dangers does not appear to dissuade them from continuing to eat it. Those who consume it mainly for its tonic affect tend to be more easily induced to stop. An area of future research that might provide valuable would be to investigate the impact of raw meat consumption on individual health, productivity in work, level of income, and overall wellbeing including the possibility to falling into a poverty trap. Furthermore, the preliminary results of this study are being used as basis for expanding the research to include a component of participatory prevention and control measures aimed at reducing disease risk in the highlander population. 65 c) Environment Environment is seen as the major risk factor explaining the transmission of Trichinella in highlanders. Since these people allow their pigs to wander around for food in the natural environment, this presents a high risk of infection from wild animals or rodents to their pigs. Though, a few households raise pigs in the forest, many households have seen wildlife around the villages and also lots of rodents. A few of them recognize the danger of these animals that could possibly bring diseases to their pigs or to themselves. This study considered the cleanliness of surroundings and pig areas since we believe they are important factors leading to Trichinella infection in pigs. Keeping pigs in pens can reduce the risk that pigs will be infected by Trichinella if pig growers do not feed their pigs with animal carcasses, cleaning pig areas frequently and removing feed after feeding every time. d) Economic Conditions Economic factors seem to be a driving force for any decision making among these populations. These highlanders use the intuition in decision- making process in pig rearing based on their objectives and constraints such as time, money and knowledge. The study discovered that there are three objectives of pig rearing in highlanders, including self-subsistence, ritual uses and commercial purposes. In terms of the commercial purpose, it is more likely an acquaintanceship selling to neighbors or friends when they are in need. In this case, they can get a fair price considering the quality and weight of the pigs they sell and do not need to compete with each other over price. The objectives of pig rearing determine the kinds of pig production modes these people apply. Those who tend to raise pigs 66 purposively for their own consumption or those who sell pigs to their friends do not pay a lot of attention to the welfare or productivity of pigs compared with those who attempt to sell the pigs out to market for good price. Therefore, the commercial farmers are willing to invest more income in constructing better housing or buying better feed for their pigs. This study also found out that different pig production modes lead to different cost structure and revenue. However, a more in-depth study concerning the mechanism of cost and the revenue stream of pig rearing, agricultural market mechanism and the adaptation of farmers in different situations related with pig rearing should be a focus in the next study. 5.2.2 Bayesian Belief Network model of Trichinellosis risk The conceptual transdisciplinary framework of Trichinellosis risk is developed by the transdisciplinary experts’ team to explain interconnection of the risk factors. Bayesian Belief Networks (BBNs) offered convenient ways to solve the decision problems related with management of the relevant institutions in attempting to reduce the risk including (1) institution’s decision to support money for pig pen construction to pig growers and (2) institution’s decision to encourage people to stop consuming raw or undercooked meat. The accuracy of the models was based on experts’ judgments. In addition, we used scoring rule results, including, logarithmic loss, quadratic loss, spherical payoff, error rate, and sum square errors to select models. We recommend using more advance qualitative statistical tools to measure the accuracy of a model and to select a model in the next study. Regarding the advantages of BBNs, they allowed us to make a decision under an uncertainty such as when we do not know the behaviors of target 67 populations. They also allowed flexibility in the prediction about how the situation will behave which is very useful for policy making. In addition, they also provided an outcome of any decision, and the models are very adaptable. We can start constructing a model with limited knowledge and improve it later as we acquire new understanding. Therefore, we recommend conducting further research in other areas to refine the creditability of the models. In considering the disadvantages of BBNs, for this study we found a so-called curse of dimensionality problem that has often been a difficulty with Bayesian statistics when the posterior distributions often have many parameters. under-determined or under-constrained problem. This problem occurs when there are many more features than data points. This problem we can often find in some real world problems. It can create noise that impedes the learning algorithm from recognizing the features that are distinguishing with respect to the target concept (Pansombut et al., 2011). As a consequence, we recommend the further research to be aware of this problem and try to avoid variables that possess too many features. However, in case that we cannot include this kind of variable in the model, we may need to increase the sample size. In this study, we found this problem in the pig production mode. Indeed, there should be only 5 features. This variable has 13 features since we cannot calculate the cost structure and revenue stream from these 5 different features but we can for the 13 features. Therefore, for the further research, we can avoid the under-determined problem if we can find the cost structure and revenue stream of these 5 different kinds of pig production. The above constraints make these models not the perfect tools to make a decision. However, the research has fulfilled the obligations to explain the 68 transmission of Trichinellosis risk in borderland highlanders and the BBNs have been constructed based on a credible process. This research can be considered as a pilot study to test and refine methods and tools for some improvement in further study. References Ames, D.P., Neilson, B.T., Stevens, D.K. and Lall U. (2005). Using Bayesian Networks to Model Watershed Management Decisions: An East Canyon Creek Case Study. Utah Water Research laboratory, Utah State University, Utah. Available: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.85.5169& rep=rep1&type=pdf [February 5, 2013] AVMA. (2008). One Health Initiative Task Force: Final Report. Available: http://www.avma.org/onehealth/onehealth_final.pdf {May 11, 2012} Black, P.F. (2006). Veterinary Science and Economics: What Next? Available: http://www.sciquest.org.nz/elibrary/download/63836/T6-2.2.5_-_Veterinary_ science_and_economics_%3A_What_next%3F [February 5, 2013] Cain, J. (2001). Planning Improvements in Natural Resources Management. Centre for Ecology and Hydrology, Wallingford, Oxfordshire, UK. Available: http://norsys.com/downloads/BBN%20Guidelines%20-%20Cain.pdf [August 12, 2012] Campos, L.M. (2006). A Scoring Function for Learning Bayesian Betworks based on Mutual Information and Conditional Independence Tests. Available: http://jmlr.csail.mit.edu/papers/volume7/decampos06a/decampos06a.pdf [May 14, 2012] 70 Carmona, G.G., Várela-Ortega C. and Bromley, J. (2011). The Use of Participatory Object-oriented Bayesian Networks and Agro-economic Models for Groundwater Management in Spain. Springerlink, 25 (5), 1509-24. Centers for Disease Control and Prevention. (2011). Parasites and Health: Trichinellosis. Available: http://dpd.cdc.gov/dpdx/html/Trichinellosis.htm [July 6, 2012] Cheva-Isarakul, B. (1998). Livestock Production Development in the Northern Highlands of Thailand. Available:http://aciar.gov.au/files/node/2099/PRO 087%20part%204.pdf [April 23, 2012] Chilonda, P. and Huylenbroeck, G.V. (2001). A Conceptual Framework for the Economic Analysis of Factors Influencing Decision-making of Small-scale Farmers in Animal Health Management. Available: http://citeseerx.ist.psu.edu/ viewdoc/download?doi=10.1.1.110.5455&rep=rep1&type=pdf [February 5, 2013] Christopher, M.B. (2006). Pattern Recognition and Machine Learning. Springer. pp. 21–24. Crooker, R.A. (2007). Ecological Marginalization and Hill Tribe Security in Northern Thailand. Available: http://www.gammathetaupsilon.org/the- geographical-bulletin/2000s/volume48-1/article2.pdf [April 25, 2012] Dambacher, J.M., Shenton, W., Hayes, K.R., Hart, B.T. and Barry, S. (2007). Qualitative Modelling and Bayesian Network Analysis for Risk-based 71 Biosecurity Decision Making in Complex Systems. Available: http://www.acera.unimelb.edu.au/materials/endorsed/0601.pdf [June 10, 2012] Dawid, A.P. (1982). The Well-Calibrated Bayesian. Journal of the American Statistical Association, 379 (77), 605-10. Available: http://fitelson.org/seminar /dawid.pdf [February 5, 2013] de Finetti, B. (1974). Theory of probability. J. Wiley & Sons, Inc., New York. Department of Environmental and Global Health. (2012). Academic Programs New One Health PhD Program Approved for Summer 2012. Available http://egh.phhp.ufl.edu/academic-programs [February 15, 2012] Devereux, S. (2001). Livelihood Insecurity and Social Protection: A Re-emerging Issue in Rural Development. Development Policy Review, 19 (4), 507-19. Dorazio, R.M. and Johnson, F.A. (2003). Bayesian Inference and Decision Theory–A Coherent Framework for Decision Making in Natural Resource Management. Availablie: http://www.au.af.mil/au/awc/awcgate/usgs/bayes_example.pdf [February 5, 2013] Ferson, S. (2005). Bayesian Methods in Risk Assessment. Available: http://www. ramas.com/bayes.pdf [February 5, 2013] Fujioka, R. (2002). Case Study on Education Opportunity for Hill Tribes in Northern Thailand: Implications for Sustainable Rural Development. Available: http://ftp.fao.org/docrep/fao/004/ak216e/ak216e00.pdf [June 12, 2012] 72 Gibbs, M. (2007). Application of a Bayesian Network Model and a Complex Systems Model to Investigate Risks of a Proposed Aquaculture Development on the Carrying Capacity of Shorebirds at the Miranda Ramsar Wetland. Enviroment Waikato Technical Series 2007/04, Hamilton East, New Zealand. Available: http://www.waikatoregion.govt.nz/PageFiles/5148/tr07-04.pdf [February 12, 2013] Harrison, I. (1997). Belief Networks. Artificial Intelligence Applications Institute, University of Edinburgh, Edinburgh, UK. Available : http://www.aiai.ed.ac.uk/ links/bn.html [February 5, 2013] Hau, A.M. (2000). Rural Market Structures and the Impact of Market Access on Agricultural Productivity: A Case Study in DoiInthanon of Northern Thailand. Available: https://www.uni-hohenheim.de/fileadmin/einrichtungen/sfb564/afs- files/protected/e3_protected/hau_diplom.pdf [May 5, 2012] Kaan, D. (2000). Risk and Resilience in Agriculture: Defining Risk and a Framework for Moving Towards Resilience In Agriculture. Available: http://www.uwagec.org/rnrinag/RnR%20Section%201/Defining%20Risk%20an d%20a%20Framework%20for%20Moving%20Towards%20Resilience%20In% 20Agriculture.pdf [May 16, 2012] Kemp-Benedict, E., Bharwani, S., de la Rosa, E., Krittasudthacheewa, C. and Matin, N. (2009). Assessing Water-related Poverty Using the Sustainable Livelihoods Framework. Available: http://www.sei-international.org/mediamanager/ 73 documents/Publications/Sustainable-livelihoods/assessing-water-related- poverty.pdf [May 19, 2012] Khamboonruang, C. (1991). The Present Status of Trichinellosis in Thailand. The National Center for Biotechnology Information, 22: 312-15. Koo, I. (2009). Definition of Zoonotic Disease. Available: http://infectiousdiseases. about.com/od/glossary/g/zoonotic.htm [June 27, 2012] Krahl, T.J. (2011). Empowering Hill Tribe Minorities in Thailand: Moving Towards a People Centred Approach. Master’s thesis. University of Leeds, Leeds, U.K. Krieg, M.L. (2001). A Tutoraila on Bayesian Belief Networks. DSTO Electronics and Surveillance Research Laboratory, Australia. Available: http://www.dsto. defence.gov.au/publications/2424/DSTO-TN-0403.pdf [February 5, 2013] Lebel, J. (2003). Health: An Ecosystem Approach. International Development Research Centre. Ottawa, Canada. Lenk, P. (2001). Bayesian Inference and Markov Chain Monte Carlo. University of Michigan Business School, Michigan, USA. Available: http://webuser.bus.umich.edu/ plenk/Bam2%20Short.pdf [February 6, 2013] Lumina Decision Systems. (2013). Influence Diagrams. Available: http://www.lumina.com/technology/influence-diagrams/ [January 7, 2013] Marcot, B.G., Stevenson, J.D., Sutherland, G.D. and McCann, R.K. (2006). Guidelines for Developing and Updating Bayesian Belief Networks Applied to Ecological Modeling and Conservation. Canadian Journal of Forestry 74 Research 36: 3063-3074. Available: http://www.fs.fed.us/pnw/pubs/journals/ pnw_2006_marcot001.pdf [February 1, 2013] McDermott, J.J. and Grace, D. (2011). Leveraging Agriculture from Improving Nutrition and Health. Available: http://www.ifpri.org/sites/default/files/ publications/2020anhconfbr18.pdf [April 1, 2012] McDermott, J.J., Randolph, T.F. and Staal, S.J. (1999). The Economics of Optimal Health and Productivity in Smallholder Livestock Systems in Developing Countries. U.S. National Library of Medicine National Institute of Health, 18 (2), 399-424. Medical-dictionary. (2002). Trichinellosis. Available: http://medical-dictionary.the freedictionary.com/trichinosis [December 7, 2012] Medterms. (2011). Definition of Trichinellosis. Available: http://www.medterms. com/script/main/art.asp?articlekey=8145 [December 7, 2012] Mitchell, K.L., Smith, R.L. and Murphy, D. (2004). Chapter 31: Probabilistic Risk Assessment. Air Toxics Risk Assessment Reference Library: Technical Resource Manual. pp.1-10. Vol. 1. Virginia: U.S. Environmental Protection Agency. Norsys Software Corp. (2013). Decision-Making Nets. Available: http://www.norsys.com/WebHelp/NETICA/X_Quick_Tour_Decision_Problems .htm [June 21, 2012] 75 OECD. (2009). Managing Risk in Agriculture: A Holistic Approach. Available: http://www.oecd.org/dataoecd/10/35/ 45558582.pdf [May 21, 2012] Olsen, C.W. 2004. Zoonotic Diseases Tutorial. Available: http://www.vetmed. wisc.edu/pbs/zoonoses [June 26, 2012] Pansombut, T., Hendrix, W., Gao, Z., Harrison, B.E. and Samatova, N.F. (2011). Biclustering-Driven Ensemble of Bayesian Belief Network Classifiers for Underdetermined Problems. Available: http://ijcai.org/papers11/Papers/IJCAI 11-243.pdf [February 7, 2013] Paulos, J.A. (2011). The Mathematics of Changing Your Mind. New York Times (US). Available: http://www.nytimes.com/2011/08/07/books/review/the-theory- that-would-not-die-by-sharon-bertsch-mcgrayne-book-review.html?pagewanted =all&_r=0 Pollino, C.A., Woodberry, O., Nicholson, A., Korb, K. and Hart, B.T. (2007). Parameterisation and Evaluation of a Bayesian Network for Use in an Ecological Risk Assessment. Environmental Modelling and Software. 22(8): 1- 13. Available: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.89.49 9&rep=rep1&type=pdf [February 5, 2013] Ramasoota, T. (1991). Current Status of Food-borne Parasitic Zoonoses in Thailand. National Center for Biotechnology Information. 22: 23-6. Rattanaronchart, S. (1994). Present Situation of Thai Native Pigs. Available: http://www.angelfire.com/mi/fafontwin/thaipigf.pdf [February 15, 2012] 76 Roberts, J.A. (2006). Chapter 1: Introduction to the Economics of Infectious Disease. The Economics of Infectious Disease. New York: Oxford University Press INC. Roberts, T. (2006). Chapter 13: Risk Assessment Models, Economic Analysis and Food Safety Policy. The Economics of Infectious Disease. New York: Oxford University Press INC. Rushton, J. (2009). The Economics of Animal Health and Production. Wallingford, UK: CABI. Srisoontorn, C. (n.d.). Prison Ministry with Hill Tribes in Thailand. Available: http://www.ipcaworldwide.org/resources/Articles/Chuleepran-Srisoontorn. pdf [June 27, 2012] Stregowski, J. (2012). Zoonosis. Available: http://dogs.about.com/od/caninediseases/ g/zoonosis_def.htm [June 26, 2012] Sudan, J. (2009). Animal Health Economics: A Review. Available: http://sustech. edu/staff_publications/20111030084740599.pdf [April 30, 2012] Tancho, A. (1997). Food and Fertilizer Technology Center. Available: http://www.agnet.org/library.php?func=view [April 26, 2012] Tisdell, C.A. (2006). Economics of Controlling Livestock Diseases: Basic Theory. Economics, Ecology and the Environment (pp. 1-9), Brisbane: University of Queensland, School of Economics. UNBC. (n.d.). UNBC - Ecohealth 2011. Available: http://www.unbc.ca/qrrc/photo_ albumeco.html [June 29,2012] 77 Waltner-Toews, D. (2009). Eco-Health: A primer for veterinarians. National Center for Biotechnology Information, 50 (5), 519-521. Watthayu, W. and Peng, Y. (2004). A Bayesian Network Based Framework for Multi-criteria Decision Making. Available: http://citeseerx.ist.psu.edu/view doc/download?doi=10.1.1.91.7966&rep=rep1&type=pdf [July 5, 2012] Westmount Animal Clinic. (2009). Parasitic Zoonoses. Available: http://www. westmountanimalclinic.com/parasiticzoonoses.htm [July 6, 2012] Wilcox, B.A., Aguirre, A.A. and Horwitz, P. (2012). Ecohealth: Connecting Ecology, Health and Sustainability. Conservation Medicine. 2nd Edition. Oxford University Press. Pp. 17-32. Willingham, A.L. (2003). New Research Opportunities in Meat-borne and Other Parasitic Zoonoses. Available: http://www.ilri.org/InfoServ/Webpub/fulldocs/ InvestAnim/Book1/media/Appendex/10/append.htm [May 16, 2012] Zinsstag, J., Schelling, E., Waltner-Toews, D. and Tanner, M. (2011). From “one Medicine” to “one Health” and Systemic Approaches to Health and Well- being. Preventive Veterinary Medicine, 101 (3-4), 148-56. Appendices 79 Appendix A Highland population information 80 Table A-1: The estimated of highland population in Thailand Provinces Village (%) Household (%) Population (%) Chiang Mai 571 (27.16) 58,245 (25.21) 244,291 (25.32) Chiang Rai 290 (13.80) 28,160 (12.19) 130,054 (13.48) Mae Hong Son 322 (15.32) 25,670 (11.11) 109,119 (11.31) Tak 205 (9.75) 28,591 (12.38) 130,065 (13.48) Nan 191 (9.08) 18,762 (8.12) 87,253 (9.04) Kanchanaburi 113 (5.38) 17,820 (7.71) 61,816 (6.41) Lamphun 63 (3.00) 8,057 (3.49) 30,825 (3.19) Phitsanulok 61 (2.90) 6,298 (2.73) 25,872 (2.68) Phrae 46 (2.19) 5,095 (2.21) 18,517 (1.92) Lampang 46 (2.19) 4,511 (1.95) 18,432 (1.91) Phayao 41 (1.95) 4,050 (1.75) 18,572 (1.92) Rachburi 26 (1.24) 5,874 (2.54) 20,510 (2.13) Phetchaboon 24 (1.14) 5,176 (2.24) 25,140 (2.61) Phetburi 24 (1.14) 5,176 (2.24) 8,407 (0.87) Kampangphet 23 (1.09) 1,820 (0.79) 8,729 (0.90) Uthaithani 17 (0.81) 1,994 (0.86) 7,511 (0.78) Prachuapkirikhan 14 (0.67) 2,945 (1.27) 9,131 (0.95) Sukhothai 12 (0.57) 1,136 (0.49) 4,413 (0.46) Supanburi 11 (0.52) 1,299 (0.56) 4,783 (0.50) Loei 2 (0.10) 317 (0.14) 1,476 (0.15) Total 2,102 (100) 230,996 (100) 964,916 (100) Source: Highland Research and Development Institute (2007) Table A-2: The estimated of highland population in Chiang Mai, Chiang Rai, and Mae Hong Son Tribes Household (%) Population (%) Akha 9,169 (10.65) 42,617 (11.28) Hmong 7,377 (8.57) 42,791 (11.33) Karen 47,212 (54.84) 199,843 (52.89) Lahu 15,310 (17.79) 63,121 (16.71) Lisu 5,084 (5.91) 21,319 (5.64) Yao 1,931 (2.24) 8,124 (2.15) Total 86,083 (100) 377,815 (100) Source: Highland Research and Development Institute (2007) 81 Source: Highland Research and Development Institute, 2007 Figure A-1: Location of the highland villages in Chiang Mai, Chiang Rai, and Mae Hong Son 82 Figure A-2: Reported cases of Trichinellosis by sub district (Tambol) during 2003- 2012 83 Figure A-3: Visited households in Huai Ma Fueang Village 84 Figure A-4: Visited households in Huai Chan Si Village 85 Appendix B Trichinellosis risk framework 86 Figure B-1: Trichinellosis risk framework Knowledge of food-preparing person Formal education of food preparing person Recognition of the danger of raw/undercooked meat Sick pig handling Frequency of pig health check up Deworming application Health practices Reasons for not deworming People in charge of pig health check up Frequency of having meat Risk of Eating Outside Consumption habits Pig production mode Investment in pig rearing Trichinella infection in pig Source of pig people raise Will keep pig in pen? Raw/undercooked preference Gender of food-preparing person Meat Preparation Source of meat Objectives of raising pigRatio of time spent for pig rearing Ratio of revenue receiving from pig rearing Hygienic security Animal welfare Convenience Financial constraint Social responsibility Regulation Tradition Ethnicity of pig raiser Religion of pig raiser Type of pig Rat control Rat abundance Wildlife presence Frequency of cleaning surrounding Place to slaughter pig Wet garbage handling Solid garbage handling Carcasses handling Dead animal handling Environment cleanliness Sick dead pig handling Environment suitability for Trichinella circulation Waste management Frequency of cleaning pig place Rearing practices Income level Frequency eating outside Place of eating outside Risk of getting Trichinosis Feed handling People clean feed left over everytim Use food scraps as feed Use carcasses left over as feed Knowledge of pig raiser Formal education level of pig raiserPeople know that keeping pig in pen is safeTrichinellosis recognition Experience in raising pig Gender of pig raiser Recognition of rat danger Public announcement Magazine/journal Neighbors/others Village headman Newspaper Internet Radio TV Animal health-relevant factors Environmental factors Food chain-relevant factors Economic factors 86 87 Figure B-2: Trichinellosis risk framework with belief bars Ratio of revenue receiving from pig raisi... Low Medium High 87.8 8.16 4.08 0.207 ± 0.19 Knowledge of pig raiser High Medium Low 33.2 33.5 33.2 Trichinellosis recogniti... No Yes 92.6 7.41 Feed handling Suitable Fair Not suitable 33.3 33.3 33.3 Use carcasses left over as feed No Yes 83.9 16.1 Use food scraps as feed No Yes 57.1 42.9 People clean feed left over everytim No Yes Sometimes 33.3 48.9 17.8 Rearing practices Suitable Fair Not suitable 33.3 33.3 33.3 Waste management Suitable Fair Not suitable 33.3 33.3 33.3 Frequency of cleaning pig place Never Twice a day Once a day Every other day Every other two days Once a week Rarely 22.0 6.00 24.0 10.0 10.0 12.0 16.0 Deworming applicati... No Yes 46.3 53.7 Health practices Suitable Fair Not suitable 33.3 33.3 33.3 Sick pig handling Never been sick Do nothing Consult expert Buy drug yourself Use herbs Sell out 15.2 19.6 32.6 26.1 4.35 2.17 Will keep pig in pe... No Yes 46.7 53.3 Trichinella infection in ... High Medium Low 33.3 33.3 33.3 Source of pig people raise Own reproduction Buy from neighbor Buy from people outside vi... 70.6 23.5 5.88 Meat Preparation Raw/undercooked Cooked 50.0 50.0 Raw/undercooked preference Not like Like 64.0 36.0 Gender of food-preparing person Male Female 27.8 72.2 Frequency of having meat <12 Times 12-30 Times 31-50 Times 51-100 Times 101-300 Times >300 Times 10.2 11.4 9.48 12.2 19.2 37.6 346 ± 380 Frequency eating outside Never Everyday Every other day Once a week Once a month Occasionally 11.9 7.45 11.3 11.4 15.1 42.9 Income level ฿< 12,000 ฿12,000-60,000 ฿60,001-100,000 ฿100,001-180,000 ฿> 180,000 6.78 61.0 22.0 5.08 5.08 60700 ± 59000 Consumption habits Good Moderate Poor 38.5 30.7 30.7 Place of eating outside Neighbor's house Food place Market Church Other village 42.7 29.7 7.09 9.35 11.2 Risk of Eating Outside High Medium Low 30.1 31.0 38.9 Risk of getting Trichinosis High Medium Low 22.9 28.0 49.1 People in charge of pig health check ... Themselves Neighbor Animal health officer Public health volunteer 44.4 16.7 27.8 11.1Investment in pig rearing Buy new pigs Buy deworming drug Promoting animal hygiene Buy high nutrient feed 32.5 40.3 10.1 17.1 Pig production mode Pen Tether Fence (underhouse) Fence (outdoor) Pen+Tether Pen+Fence (underhouse) Pen+Fence (outdoor) Pen+Fence Tether+Fence (underhouse) Tether+Free range Pen+Tether+Free range Pen+Fence (underhouse)... Pen+Fence (outdoor)+Fre... 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 Objectives of raising pig Consumption only Commercial purpose only Both consumption and co... Ritual ceremony 23.2 7.76 56.1 12.9 Ratio of time spent for pig raising per to... Low Medium High 92.5 3.77 3.77 0.19 ± 0.17 Environment suitability for Trichinella ci... Suitable Fair Not suitable 33.3 33.3 33.3 Rat abundance None 1-5 5-10 >10 29.8 53.5 3.71 12.9 Wildlife presence No Yes 80.4 19.6 Rat control No Yes 48.0 52.0 Place to slaughter pig Backyard Local butcher Inside pen 91.8 4.08 4.08 Dead animal handling Bury Burn Never seen one die 80.4 11.8 7.84 Frequency of cleaning surrounding Everyday Every other day Once a week Rarely Never 28.8 25.4 25.4 18.6 1.69 Wet garbage handling Sweep away Use as feed Bury Burn Put in the community trash... Throw in forest 13.7 33.3 3.92 19.6 11.8 17.6 Solid garbage handling Sweep away Bury Burn Put in the community trash... Throw in forest Sell Reuse 5.56 37.0 25.9 5.56 9.26 13.0 3.70 Environment cleanliness Clean Fair Dirty 33.3 33.4 33.3 Source of meat Own reproduction Buy from neighbor Buy outside village Hunt 14.3 14.3 57.1 14.3 Sick dead pig handling Never die Bury Burn Eat within family Eat within family and shar... Sell to neighbor 5.77 75.0 3.85 5.77 3.85 5.77 Hygienic security No Yes 58.3 41.7 Regulation No Yes 50.0 50.0 Tradition No Yes 91.5 8.51 Animal welfare No Yes 91.7 8.33 Social responsibil... No Yes 91.7 8.33 Financial constraint No Yes 79.2 20.8 Convenience No Yes 47.9 52.1 Ethnicity of pig raiser Lisu Black Lahu Red Lahu Palong 8.16 4.08 83.7 4.08 Religion of pig raiser Buddhism Christianism Spiritualism 74.5 23.4 2.13 Type of pig Boar Sow Piglet 16.7 22.2 61.1 Recognition of the danger of consumng ... No Yes 71.4 28.6 Reasons for not deworming Accessibility Finanical constraint Misunderstanding of the i... 35.3 11.8 52.9 Frequency of pig health check up Never Once a week Once a month Once a year 53.7 17.3 14.5 14.4 Formal education level of pig raiser None 10 years 45.1 17.6 37.3 18600 ± 30000 People know that keeping pig in pen is ... No Yes 58.0 42.0 Formal education of food preparing per... None 180,000 6.78 61.0 22.0 5.08 5.08 60700 ± 59000 Newspaper No Yes 89.3 10.7 Village headman No Yes 96.4 3.57 TV No Yes 14.3 85.7 Radio No Yes 66.1 33.9 Internet No Yes 96.4 3.57 Neighbors/others No Yes 30.4 69.6 Magazine/journal No Yes 96.4 3.57 Public announcement No Yes 53.6 46.4 Type of pig Boar Sow Piglet 16.7 22.2 61.1 Rat abundance None 1-5 5-10 >10 29.8 53.5 3.71 12.9 Rat control No Yes 48.0 52.0 Environment cleanliness Clean Fair Dirty 33.3 33.4 33.3 Carcasses left over handling after slaught... No left over Sweep down to the floor Pet feed Pig feed Put in trash can Bury 37.2 11.6 44.2 2.33 2.33 2.33 Place to slaughter pig Backyard Local butcher Ins ide pen 91.8 4.08 4.08 Dead animal handling Bury Burn Never seen one die 80.4 11.8 7.84 Frequency of cleaning surrounding Everyday Every other day Once a week Rarely Never 28.8 25.4 25.4 18.6 1.69 Wet garbage handling Sweep away Use as feed Bury Burn Put in the community trash can Throw in forest 13.7 33.3 3.92 19.6 11.8 17.6 Solid garbage handling Sweep away Bury Burn Put in the community trash can Throw in forest Sell Reuse 5.56 37.0 25.9 5.56 9.26 13.0 3.70 Sick dead pig handling Never die Bury Burn Eat within family Eat within family and share ... Sell to neighbor 5.77 75.0 3.85 5.77 3.85 5.77 Gender of pig raiser Male Female 42.6 57.4 Trichinellosis recognition No Yes 92.6 7.41 Formal education level of pig raiser None 10 years 45.1 17.6 37.3 18600 ± 30000 People know that keeping pig in pen is safe No Yes 58.0 42.0 93 93 Figure C-2: Trichinella infection risk framework (animal perspective) – Model 2 Feed handling Suitable Fair Not suitable 33.3 33.3 33.3 Use carcasses left over as feed No Yes 85.3 14.7 Use food scraps as feed No Yes 55.9 44.1 People clean feed left over everytim No Yes Sometimes 39.3 39.3 21.4 Rearing practices Suitable Fair Not suitable 33.3 33.3 33.3 Waste management Suitable Fair Not suitable 33.3 33.3 33.3 Deworming application No Yes 47.3 52.7 Health practices Suitable Fair Not suitable 33.3 33.3 33.3 Sick pig handling Never been sick Do nothing Consult expert Buy drug yourself Use herbs Sell out 16.1 12.9 32.3 29.0 6.45 3.23 People in charge of pig health check up Themselves Neighbor Animal health officer Public health volunteer 50.0 7.14 35.7 7.14 Reasons for not deworming Accessibility Finanical constraint Misunderstanding of the imp... 33.3 16.7 50.0 Frequency of pig health check up Never Once a week Once a month Once a year 48.7 18.5 14.4 18.5 Trichinella infection in pig High Medium Low 31.5 37.0 31.5 Pig production mode Pen Tether Fence (underhouse) Fence (outdoor) Pen+Tether Pen+Fence (underhouse) Pen+Fence (outdoor) Pen+Fence Tether+Fence (underhouse) Tether+Free range Pen+Tether+Free range Pen+Fence (underhouse)+F... Pen+Fence (outdoor)+Free r... 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 Investment in pig rearing Buy new pigs Buy deworming drug Promoting animal hygiene Buy high nutrient feed 36.4 33.8 13.1 16.7 Source of pig people raise Own reproduction Buy from neighbor Buy from people outside villa... 66.7 25.0 8.33 Income level ฿< 12,000 ฿12,000-60,000 ฿60,001-100,000 ฿100,001-180,000 ฿> 180,000 8.11 56.8 21.6 5.41 8.11 67400 ± 69000 Village headman No Yes 94.1 5.88 TV No Yes 14.7 85.3 Internet No Yes 94.1 5.88 Neighbors/others No Yes 32.4 67.6 Frequency of cleaning pig place Never Twice a day Once a day Every other day Every other two days Once a week Rarely 12.5 9.38 25.0 12.5 6.25 15.6 18.8 Objectives of raising pig Consumption only Commercial purpose only Both consumption and com... Ritual ceremony 23.5 5.88 58.8 11.8 Religion of pig raiser Buddhism Christianism Spiritualism 76.7 20.0 3.33 Type of pig Boar Sow Piglet 15.4 15.4 69.2 Knowledge of pig raiser High Medium Low 33.0 34.1 33.0 Gender of pig raiser Male Female 40.0 60.0 Trichinellosis recognition No Yes 97.0 3.03 Formal education level of pig raiser None 10 years 43.8 12.5 43.8 21900 ± 31000 People know that keeping pig in pen is safe No Yes 62.5 37.5 Ethnicity of pig raiser Lisu Black Lahu Red Lahu Palong 9.68 3.23 80.6 6.45 Hygienic security No Yes 56.7 43.3 Regulation No Yes 53.3 46.7 Tradition No Yes 93.3 6.67 94 Table C-3: Scoring rule results of the Trichinella infection in pig Scoring rule results Values Model 1 Model 2 Logarithmic loss 1.0990 0.9201 Quadratic loss 0.6667 0.5416 Spherical payoff 0.5774 0.6840 Error rate 100% 20% 95 Table C-4: Probability table of the Trichinellosis risk in animal Household P(TIP=High) P(TIP=Medium) P(TIP=Low) Observed Predicted Observed Predicted Observed Predicted Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 1 0 0 0 1.00 1.00 1.00 0 0 0 10 0 0 0 0.75 0.75 0.78 0.25 0.25 0.22 13 0.50 0.50 0.5 0.50 0.50 0.50 0 0 0 15 0 0 0 1.00 1.00 1.00 0 0 0 17 0.50 0.50 0.45 0.50 0.50 0.55 0 0 0 19 0.50 0.50 0.37 0.50 0.50 0.63 0 0 0 20 0.50 0.50 0.46 0.50 0.50 0.54 0 0 0 21 0 0 0 0.75 0.75 0.84 0.25 0.25 0.16 22 * 0.33 0.31 * 0.33 0.37 * 0.33 0.31 23 * 0.33 0.31 * 0.33 0.37 * 0.33 0.31 25 0 0 0 1.00 1.00 1.00 0 0 0 26 0 0 0 0.75 0.75 0.79 0.25 0.25 0.21 27 0.25 0.25 0.22 0.75 0.75 0.78 0 0 0 31 0 0 0 1.00 1.00 1.00 0 0 0 36 * 0.33 0.31 * 0.33 0.37 * 0.33 0.31 37 0.25 0.25 0.19 0.75 0.75 0.81 0 0 0 41 * 0.333 0.31 * 0.33 0.37 * 0.333 0.31 42 0 0 0 1.00 1.00 1.00 0 0 0 44 0.50 0.50 0.50 0.50 0.50 0.5 0 0 0 46 0.50 0.50 0.38 0.50 0.50 0.62 0 0 0 47 0.50 0.50 0.47 0.50 0.50 0.53 0 0 0 51 0.50 0.50 0.50 0.50 0.50 0.50 0 0 0 SSE Model 1 = 0, SSE Model 2 = 0.104 95 96 96 Figure C-3: Trichinella infection risk framework (animal health perspective with decision and utility nodes) – Model 2 Switch to keep pig in pen Feed handling Suitable Fair Not suitable 33.3 33.3 33.3 Use carcasses left over as feed No Yes 85.3 14.7 Use food scraps as feed No Yes 55.9 44.1 People clean feed left over everytim No Yes Sometimes 39.3 39.3 21.4 Rearing practices Suitable Fair Not suitable 33.3 33.3 33.3 Waste management Suitable Fair Not suitable 33.3 33.3 33.3 Deworming application No Yes 47.3 52.7 Health practices Suitable Fair Not suitable 33.3 33.3 33.3 Sick pig handling Never been sick Do nothing Consult expert Buy drug yourself Use herbs Sell out 16.1 12.9 32.3 29.0 6.45 3.23 People in charge of pig health check up Themselves Neighbor Animal health officer Public health volunteer 50.0 7.14 35.7 7.14 Reasons for not deworming Accessibility Finanical constraint Misunderstanding of the imp... 33.3 16.7 50.0 Frequency of pig health check up Never Once a week Once a month Once a year 48.7 18.5 14.4 18.5 Pig production mode Pen Tether Fence (underhouse) Fence (outdoor) Pen+Tether Pen+Fence (underhouse) Pen+Fence (outdoor) Pen+Fence Tether+Fence (underhouse) Tether+Free range Pen+Tether+Free range Pen+Fence (underhouse)+F... Pen+Fence (outdoor)+Free r... 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 Investment in pig rearing Buy new pigs Buy deworming drug Promoting animal hygiene Buy high nutrient feed 36.4 33.8 13.1 16.7 Source of pig people raise Own reproduction Buy from neighbor Buy from people outside villa... 66.7 25.0 8.33 Village headman No Yes 94.1 5.88 TV No Yes 14.7 85.3 Internet No Yes 94.1 5.88 Neighbors/others No Yes 32.4 67.6 Frequency of cleaning pig place Never Twice a day Once a day Every other day Every other two days Once a week Rarely 12.5 9.38 25.0 12.5 6.25 15.6 18.8 Knowledge of pig raiser High Medium Low 33.0 34.1 33.0 Gender of pig raiser Male Female 40.0 60.0 Trichinellosis recognition No Yes 97.0 3.03 Formal education level of pig raiser None 10 years 43.8 12.5 43.8 21900 ± 31000 People know that keeping pig in pen is safe No Yes 62.5 37.5 Income level ฿< 12,000 ฿12,000-60,000 ฿60,001-100,000 ฿100,001-180,000 ฿> 180,000 8.11 56.8 21.6 5.41 8.11 67400 ± 69000 Will keep pig in pen? No Yes 50.3 49.7 Keep pig in pen No Yes -5747.7 -3912.2 Trichinella infection in pig High Medium Low 31.5 37.0 31.5 Religion of pig raiser Buddhism Christianism Spiritualism 76.7 20.0 3.33 Type of pig Boar Sow Piglet 15.4 15.4 69.2 Ethnicity of pig raiser Lisu Black Lahu Red Lahu Palong 9.68 3.23 80.6 6.45 Hygienic security No Yes 56.7 43.3 Regulation No Yes 53.3 46.7 Tradition No Yes 93.3 6.67 Objectives of raising pig Consumption only Commercial purpose only Both consumption and com... Ritual ceremony 23.5 5.88 58.8 11.8 97 97 Figure C-4: Trichinella infection risk framework (animal health perspective with decision and utility nodes) – Model 2, when there is information about the level of risk that pigs will be infected by Trichinella Feed handling Suitable Fair Not suitable 33.3 33.3 33.3 Use carcasses left over as feed No Yes 85.3 14.7 Use food scraps as feed No Yes 55.9 44.1 People clean feed left over everytim No Yes Sometimes 39.3 39.3 21.4 Rearing practices Suitable Fair Not suitable 35.3 29.4 35.3 Waste management Suitable Fair Not suitable 33.3 33.3 33.3 Deworming application No Yes 47.3 52.7 Health practices Suitable Fair Not suitable 33.3 33.3 33.3 Sick pig handling Never been sick Do nothing Consult expert Buy drug yourself Use herbs Sell out 16.1 12.9 32.3 29.0 6.45 3.23 People in charge of pig health check up Themselves Neighbor Animal health officer Public health volunteer 50.0 7.14 35.7 7.14 Reasons for not deworming Accessibility Finanical constraint Misunderstanding of the imp... 33.3 16.7 50.0 Frequency of pig health check up Never Once a week Once a month Once a year 48.7 18.5 14.4 18.5 Pig production mode Pen Tether Fence (underhouse) Fence (outdoor) Pen+Tether Pen+Fence (underhouse) Pen+Fence (outdoor) Pen+Fence Tether+Fence (underhouse) Tether+Free range Pen+Tether+Free range Pen+Fence (underhouse)+F... Pen+Fence (outdoor)+Free r... 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 7.69 Investment in pig rearing Buy new pigs Buy deworming drug Promoting animal hygiene Buy high nutrient feed 36.4 33.8 13.1 16.7 Source of pig people raise Own reproduction Buy from neighbor Buy from people outs ide villa... 64.7 26.5 8.82 Village headman No Yes 94.1 5.88 TV No Yes 14.7 85.3 Internet No Yes 94.1 5.88 Neighbors/others No Yes 32.4 67.6 Frequency of cleaning pig place Never Twice a day Once a day Every other day Every other two days Once a week Rarely 12.5 9.38 25.0 12.5 6.25 15.6 18.8 Knowledge of pig raiser High Medium Low 33.0 34.1 33.0 Gender of pig raiser Male Female 40.0 60.0 Trichinellosis recognition No Yes 97.0 3.03 Formal education level of pig raiser None 10 years 43.8 12.5 43.8 21900 ± 31000 People know that keeping pig in pen is safe No Yes 62.5 37.5 Income level ฿< 12,000 ฿12,000-60,000 ฿60,001-100,000 ฿100,001-180,000 ฿> 180,000 8.11 56.8 21.6 5.41 8.11 67400 ± 69000 Will keep pig in pen? No Yes 50.0 50.0 Switch to keep pig in pen Keep pig in pen No Yes -5732.8 -3886.5 Trichinella infection in pig High Medium Low 100 0 0 Religion of pig raiser Buddhism Christianism Spiritualism 76.7 20.0 3.33 Type of pig Boar Sow Piglet 15.4 15.4 69.2 Ethnicity of pig raiser Lisu Black Lahu Red Lahu Palong 9.68 3.23 80.6 6.45 Hygienic security No Yes 56.7 43.3 Regulation No Yes 53.3 46.7 Tradition No Yes 93.3 6.67 Objectives of raising pig Consumption only Commercial purpose only Both consumption and com... Ritual ceremony 23.5 5.88 58.8 11.8 98 Appendix D Human health-relevant information 99 Table D-1: Economic losses of illness and death per capita from Trichinellosis in human Unit: Baht Economic losses Severity1 Average High Medium Low Direct costs Hospital operating costs OPD costs Pharmacy Lab investigation Service Radiology IPD costs Room and meal Pharmacy Lab investigation Radiology Service Medical supplies Anesthetic service Special test Blood Serodiagnosis Total direct costs 22,773.81 1,778.07 248.07 1,300.00 202.14 195.00 22,745.20 2,853.33 6,506.53 3,040.00 630.00 2,784.00 8,218.33 13,520.00 2,000.00 12,620.00 200.00 22,973.81 115,067.50 - - - - - 115,067.50 7,200.00 41,717.50 9,455.00 1,265.00 8,280.00 21,430.00 38,830.00 - 12,620 200.00 115,267.50 11,696.67 1,636.67 260.11 1,195.56 180.00 220.00 10,060.00 2,700.00 1,313.78 2,299.44 548.57 2,433.33 2,775.00 870.00 2,000 - 200.00 11,896.67 5,795.20 1,956.40 226.40 1,488.00 242.00 1,990.40 3,838.00 820.00 467.80 1,199.00 186.67 660.00 450.00 - 2,000 - 200.00 5,995.20 Indirect costs Transportation2 Absenteeism3 Premature mortality4 Total indirect costs 327.25 1,790.63 2,340.00 2,410.38 102.00 2,325.00 2,340.00 4,767.00 347.56 1,966.67 - 2,314.22 380.80 1,260.00 - 1,640.80 Total economic losses5 25,384.19 120,034.50 14,210.89 7,636.00 Source (raw data): Acting Lt.Sathian Pattamawath, Pua Crown Prince Hospital, Nan Province. 1Levels of the severity is determined by Assoc.Prof.Dr.Pichart Uparanukraw, Department of Parasitology, Faculty of Medicine, Chiang Mai University. 2Transportation cost = Gasoline cost per km.×distance from village to hospital×days visiting hospital Gasoline cost per km. = 4 Baht (Diesel, 29.99Baht/litre), avg.distance from village to hospital = 17 km. 3Absenteeism = Absent days×wage per day 4Premature mortality cost = (Life expectancy at birth of this population - Average age of the villagers)×12 months×wage per day Life expectancy at birth = 40.9 ys, average age of the villagers = 39.6 yrs, wage per day = 150 Baht. 5Total economic losses per capita, calculated by author. 100 Table D-2: Utility table of human health perspective Scenarios RTH F6 D2 F15 U2 1 H R N N -120,034.50 21 H R N Y 60,017.25 3 H R Y N -126,034.50 4 H R Y Y 114,034.50 5 H C N N2 0 5 H C N Y 0 6 H C Y N2 -6,000.00 6 H C Y Y -6,000.00 1 M R N N -14,210.89 21 M R N Y 7,105.44 3 M R Y N -20,210.89 4 M R Y Y 8,210.89 5 M C N N2 0 5 M C N Y 0 6 M C Y N2 -6,000.00 6 M C Y Y -6,000.00 1 L R N N -7,636.00 21 L R N Y 3,818.00 3 L R Y N -13,636.00 4 L R Y Y 1,636.00 5 L C N N2 0 5 L C N Y 0 6 L C Y N2 -6,000.00 6 L C Y Y -6,000.00 1U2 of these scenarios = cost saved from being Trichinellosis×probability to divert the behavior back to consume raw or undercooked meat (we assume to be 0.50). 2These cases mean they continuing consume cooked meat. 101 Possible scenarios 1. Risk loving without intervention This scenario considers those who love to eat raw or undercooked meat. They are seen as the risk lovers. Without any intervention, these people continuing consuming raw or undercooked meat. They face the possibility of getting Trichinellosis and may bear some economic losses from the illness. The size of the losses depends on the severity of the illness. 2. Diverting risk without intervention This scenario considers those who previously love to eat raw or undercooked meat and then divert the behavior to stop consuming it without any intervention or being educated. To stop consuming raw or undercooked meat, we can say that the risk of being Trichinellosis will be 0%. However, we assume that these people can easily divert behaviors back to consume raw or undercooked meat. 3. Risk loving with intervention (ineffective intervention) This scenario considers the situation when there is an intervention from an institution attempting to encourage people to stop consuming raw or undercooked meat. The institution bears the cost from introducing its campaign. Though people are educated about the danger of consuming raw or undercooked meat, they tend to not be aware of. The campaign is ineffective to change their behaviors. These people face the possibility of getting Trichinellosis and may bear some economic losses from the illness. The size of the losses depends on the severity of the illness. 102 4. Diverting risk with intervention (effective intervention) This scenario considers the situation when there is an intervention from an institution attempting to encourage people to stop consuming raw or undercooked meat. The institution bears the cost from introducing its campaign. After people are educated about the danger of consuming raw or undercooked meat, they tend to divert their behaviors. These people can save the economic losses from the illness. The size of the losses depends on the possibility that they could get Trichinellosis if they do not divert the behavior. 5. Risk aversion without intervention This scenario considers those who already aware the danger of consuming raw or undercooked meat and never consuming some. Without the intervention, there are no gains and no losses. 6. Risk aversion with intervention This scenario considers the situation when there is an intervention from an institution attempting to encourage people to stop consuming raw or undercooked meat. The institution bears the cost from introducing its campaign. However, the intervention does not have an effect on those who already aware the danger of consuming raw or undercooked meat because they never consuming some 103 103 Figure D-1: Trichinellosis risk framework (human perspective) – Model 1 Village headman No Yes 94.1 5.88 TV No Yes 14.7 85.3 Internet No Yes 94.1 5.88 Neighbors/others No Yes 32.4 67.6 Knowledge of pig raiser High Medium Low 33.0 34.1 33.0 Feed handling Suitable Fair Not suitable 33.3 33.3 33.3 Use carcasses left over as feed No Yes 85.3 14.7 Use food scraps as feed No Yes 55.9 44.1 People clean feed left over everytim No Yes Sometimes 39.3 39.3 21.4 Formal education of food preparing person None 10 years 43.7 12.5 43.8 21900 ± 31000 Frequency of having meat <12 Times 12-30 Times 31-50 Times 51-100 Times 101-300 Times >300 Times 12.2 13.7 11.2 13.1 18.8 31.0 297 ± 360 Frequency eating outside Never Everyday Every other day Once a week Once a month Occasionally 13.2 8.31 14.2 8.31 14.2 41.8 Income level ฿< 12,000 ฿12,000-60,000 ฿60,001-100,000 ฿100,001-180,000 ฿> 180,000 8.11 56.8 21.6 5.41 8.11 67400 ± 69000 Place of eating outside Neighbor's house Food place Market Church Other village 35.8 29.2 9.01 12.0 14.0 Risk of Eating Outside High Medium Low 30.3 31.4 38.3 Consumption habits Good Moderate Poor 36.5 31.7 31.7 Risk of getting Trichinosis High Medium Low 20.2 32.9 46.8 Religion of pig raiser Buddhism Christianism Spiritualism 76.7 20.0 3.33 Type of pig Boar Sow Piglet 15.4 15.4 69.2 Objectives of raising pig Consumption only Commercial purpose only Both consumption and com... Ritual ceremony 23.5 5.88 58.8 11.8 Regulation No Yes 53.3 46.7 Hygienic security No Yes 56.7 43.3 Tradition No Yes 93.3 6.67 Ethnicity of pig raiser Lisu Black Lahu Red Lahu Palong 9.68 3.23 80.6 6.45 Environment suitability for Trichinella circu... Suitable Fair Not suitable 20.0 20.0 60.0 104 Figure D-2: Trichinellosis risk framework (human perspective) – Model 2 Meat Preparation Raw/undercooked Cooked 49.0 51.0 Frequency of having meat <12 Times 12-30 Times 31-50 Times 51-100 Times 101-300 Times >300 Times 8.83 10.5 6.65 10.5 19.2 44.3 396 ± 390 Knowledge of food-preparing person High Medium Low 21.8 27.6 50.6 Formal education of food preparing person None 180,000 8.11 56.8 21.6 5.41 8.11 67400 ± 69000 Raw/undercooked preference Not like Like 60.0 40.0 Gender of food-preparing person Male Female 33.3 66.7 Recognition of the danger of consumng ra... No Yes 82.4 17.6 Consumption habits Good Moderate Poor 58.0 21.0 21.0 Risk of getting Trichinosis High Medium Low 16.5 24.9 58.7 105 Table D-3: Scoring rule results of the Trichinellosis risk in human Scoring rule results Values Model 1 Model 2 Logarithmic loss 0.6074 0.6107 Quadratic loss 0.3346 0.3403 Spherical payoff 0.7957 0.7882 Error rate 33.33% 33.33% 106 Table D-4: Probability table of the Trichinellosis risk in human Household P(RTH=High) P( RTH =Medium) P( RTH =Low) Observed Predicted Observed Observed Predicted Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 1 0 0 0 1 0 0 0 1 0 10 0 0 0 10 0 0 0 10 0 13 0 0 0 13 0 0 0 13 0 15 0 0 0 15 0 0 0 15 0 17 0 0 0 17 0 0 0 17 0 19 0.33 0.24 0.24 19 0.33 0.24 0.24 19 0.33 20 0.33 0.08 0.05 20 0.33 0.08 0.05 20 0.33 21 0.33 0.17 0.08 21 0.33 0.17 0.08 21 0.33 22 0 0 0 22 0 0 0 22 0 23 0 0 0 23 0 0 0 23 0 25 0.33 0.14 0.14 25 0.33 0.14 0.14 25 0.33 26 0 0 0 26 0 0 0 26 0 27 0 0 0 27 0 0 0 27 0 31 0 0 0 31 0 0 0 31 0 36 0.67 0.49 0.49 36 0.67 0.49 0.49 36 0.67 37 1.00 1.00 1.00 37 1.00 1.00 1.00 37 1.00 41 0 0 0 41 0 0 0 41 0 42 1.00 1.00 1.00 42 1.00 1.00 1.00 42 1.00 44 0.67 0.49 0.48 44 0.67 0.49 0.48 44 0.67 46 0.67 0.40 0.40 46 0.67 0.40 0.40 46 0.67 47 0 0 0 47 0 0 0 47 0 51 0.33 0.16 0.11 51 0.33 0.16 0.11 51 0.33 SSE Model 1 = 7.955, SSE Model 2 = 8.207 106 107 Figure D-3: Trichinellosis risk framework (human health perspective with decision and utility nodes) – Model 2 Formal education of food preparing per... None 180,000 6.... 6... 2... 5.... 5.... 60700 ± 59000 Recognition of the danger of consumng ... No Yes 7... 2... Raw/undercooked prefer... Not like Like 6... 3... Gender of food-preparing per... Male Female 2... 7... Frequency of having meat <12 Times 12-30 Times 31-50 Times 51-100 Times 101-300 Times >300 Times 6.... 7.... 4.... 8.... 1... 5... 461 ± 400 108 108 Figure D-4: Trichinellosis risk framework (human health perspective with decision and utility nodes) – Model 2, when there is information about the level of risk that people will be Trichinellosis Raw/undercooked preference Not like Like 64.0 36.0 Gender of food-preparing person Male Female 27.8 72.2 Source of meat Own reproduction Buy from neighbor Buy outside village Hunt 14.3 14.3 57.1 14.3 Frequency of hav ing meat (Carni... <12 Times 12-30 T imes 31-50 T imes 51-100 T imes 101-300 T imes >300 T imes 10.2 11.4 9.50 12.2 19.1 37.5 346 ± 380 Place of eating outside Neighbor's house Food place Market Church Other village 42.6 29.6 7.13 9.40 11.2 Sick dead pig handling Never die Bury Burn Eat within family Eat within family an... Sell to neighbor 5.77 75.0 3.85 5.77 3.85 5.77 Knowledge of food-preparing per... High Medium Low 19.7 26.2 54.1 Income level ฿< 12,000 ฿12,000-60,000 ฿60,001-100,000 ฿100,001-180,000 ฿> 180,000 6.80 61.0 22.0 5.10 5.11 60700 ± 59000 Formal education of food prepari... None 10 years People have raised pig more than 10 years X6 Source of pig people raise Own reproduction Pigs people raise are from their own reproduction Buy from neighbor Pigs people raise are from their neighbor Buy from people outside village Pigs people raise are from outside village X7 Type of pig Boar People raise boar Sow People raise sow Piglet People raise piglet X8 Pig production mode Free range People allow pig to wandering around the village Tether People tether pig with rope underneath their house Fence (under house) People keep pig in fence underneath their house Fence (outdoor) People keep pig in fence located outdoor Pen People keep pig in pen X9 Convenience No People choose pig production mode not based on convenience Yes People choose pig production mode based on convenience X10 Hygienic security No People choose pig production mode not based on hygienic security Yes People choose pig production mode based on hygienic security X11 Regulation No People choose pig production mode not based on regulation 112 Abbrev. Topic States Descriptions Yes People choose pig production mode based on regulation X12 Tradition No People choose pig production mode not based on tradition Yes People choose pig production mode based on tradition X13 Social responsibility No People choose pig production mode not based on social responsibility Yes People choose pig production mode based on social responsibility X14 Animal welfare No People choose pig production mode not based on animal welfare Yes People choose pig production mode based on animal welfare X15 Financial constraint No People choose pig production mode not based on financial constraint Yes People choose pig production mode based on financial constraint X16 Frequency of pig health check up Never Never check pig health Once a week Check pig health once a week Once a month Check pig health once a month Once a year Check pig health once a year X17 People in charge of pig health check up Themselves People check pig health by themselves Neighbor People ask neighbor to check pig health Animal health officer People ask animal health officer to check pig health Public health volunteer People ask public health volunteer to check pig health X18 Sick pig handling Never been sick People’s pigs have never been sick Do nothing People do nothing when their pigs are sick Consult expert People consult with expert if 113 Abbrev. Topic States Descriptions their pigs are sick Buy drug themselves People buy drug for their pigs when they get sick Use herbs People use herbs for their pigs when they get sick Sell out People sell out their pigs when they are sick X19 Deworming application No Do not use deworming drug Yes Use deworming product X20 Trichinellosis recognition No People do not know anything about Trichinellosis Yes People know what Trichinellosis is X21 People recognition that keeping in pen is safe No People know that keeping pig in pen is hygienically safe Yes People do not know that keeping pig in pen is hygienically safe X22 Household’s decision to keep pigs in pen No If people know that keeping pig in pen is hygienically safe, they will decide to keep pig in pen Yes Though people know that keeping pig in pen is hygienically safe, they will not to keep pig in pen X23 Use of food scraps as feed No Do not use food scraps as feed Yes Use food scraps as feed X24 Use of carcasses left over as feed No Do not use carcasses left over as feed Yes Use carcasses left over as feed X25 Carcasses left over handling after slaughtering a pig No left over There is not carcasses left over after slaughtering a pig Sweep down to the floor After slaughtering a pig, people sweep carcasses left over down to the floor Pet feed After slaughtering a pig, people use carcasses left over as pet feed Pig feed After slaughtering a pig, people use carcasses left over as pig feed Put in trash can After slaughtering a pig, 114 Abbrev. Topic States Descriptions people put carcasses left over in trash can Bury After slaughtering a pig, people bury carcasses left over X26 Frequency of cleaning surrounding Everyday People clean surrounding everyday Every other day People clean surrounding every other day Once a week People clean surrounding once a week Rarely People rarely clean surrounding Never People never clean surrounding X27 Wet garbage handling Sweep away People sweep wet garbage down to the floor Use as feed People use wet garbage as feed Bury People bury wet garbage Burn People burn wet garbage Put in the community trash can People put wet garbage in the community trash can Throw in forest People throw wet garbage in forest nearby X28 Solid garbage handling Sweep away People sweep solid garbage down to the floor Bury People bury solid garbage Burn People burn solid garbage Put in the community trash can People put solid garbage in the community trash can Thrown in forest People throw solid garbage in forest nearby Sell People sell solid garbage Reuse People reuse solid garbage X29 Frequency of cleaning pig place Never People never clean pig place Twice a day People clean pig place twice a day Once a day People clean pig place once a day Every other day People clean pig place every other day Every other two People clean pig place every 115 Abbrev. Topic States Descriptions days other two days Once a week People clean pig place once a week Rarely People rarely clean pig place X30 People clean feed left over every time after feeding No People do not clean feed left over every time after feeding Yes People clean feed left over every time after feeding Sometimes People sometimes clean feed left over every time after feeding X31 Rat abundance None People see none of rat around the house 1-5 People see 1-5 rats around the house 5-10 People see 5-10 rats around the house >10 People see more than 10 rats around the house X32 Recognition of rat danger No People do not know that rat is dangerous Yes People know that rate is dangerous X33 Wildlife presence No People see none of wildlife in the village Yes People see some of wildlife in the village X34 Rat control No People do not use rat control Yes People use rat control X35 Sick dead pig handling Never die There is no pig that is dead with sickness Bury People bury sick dead pig Burn People burn sick dead pig Eat within family People consume sick dead pig within their family Eat within family and share with neighbor People consume sick dead pig within their family and share to neighbor Sell to neighbor People sell sick dead pig to neighbor X36 Dead animal handling Bury People usually bury dead animal Burn People usually burn dead animal Never seen one People never see any animal 116 Abbrev. Topic States Descriptions die die before X37 Place to slaughter pig Back yard People slaughter pig back yard Local butcher People slaughter pig at local butcher Inside pen People slaughter pig inside the pen X38 Environment cleanliness Clean The environment of this house is clean Fair The environment of this house is fair Dirty The environment of this house is dirty X39 Environment suitability of Trichinella circulation Suitable The environment of this household is suitable for Trichinella circulation Fair The environment of this household is fair for Trichinella circulation Not suitable The environment of this household is not suitable for Trichinella circulation X40 Feed handling Suitable The feed handling is suitable Fair The feed handling is fair Not suitable The feed handling is not suitable X41 Health practices Suitable The health practices for their pig is suitable Fair The health practices for their pig is fair Not suitable The health practices for their pig is not suitable X42 Knowledge of pig raiser High Pig raiser has high knowedge about pig rearing Medium Pig raiser has medium knowledge about pig rearing Low Pig raiser has low knowledge about pig rearing X43 Rearing practice Suitable Rearing practice is suitable Fair Rearing practice is fair Not suitable Rearing practice is not suitable X44 Income level <฿12,000 People receive less than 12,000 Baht annually which is considered to be under poverty (earn less than a US 117 Abbrev. Topic States Descriptions dollar a day) ฿12,00-60,000 People receive 12,000- 60,000 Baht annually which is considered to be poor ฿60,001-100,000 People receive 12,000- 60,000 Baht annually which is considered to be normal ฿100,001-180,000 People receive 12,000- 60,000 Baht annually which is considered to be better off >฿180,000 People receive greater than 180,000 Baht annually which is the goal that Ministry of Agricultural and Cooperatives set to achieve in 2013 X45 Investment in pig rearing Buy new pigs People use some of their money to invest in buying new pigs Buy deworming drug People use some of their money to invest in buying deworming drug Promote animal hygiene People use some of their money to invest in promoting animal hygiene Buy high nutrient feed People use some of their money to invest in buying high nutrient feed X46 Objective of pig rearing Consumption only People raise pig for their own consumption only Commercial purpose only People raise pig for commercial purpose only Both consumption and commercial purpose People raise pig for both consumption and commercial purpose Ritual use People raise pig for ritual use only X47 Ratio of time spent for pig rearing per total time spent on career Low Time spent for pig rearing is around 0-30% of the total time spent on career Medium Time spent for pig rearing is around 31-70% of the total time spent on career High Time spent for pig rearing is around 71-100% of the total 118 Abbrev. Topic States Descriptions time spent on career X48 Ratio of revenue receiving from pig rearing per total revenue Low Revenue receiving from pig rearing is around 0-30% of the total revenue Medium Revenue receiving from pig rearing is around 31-70% of the total revenue High Revenue receiving from pig rearing is around 71-100% of the total revenue X49 Newspaper No People do not receive information through newspaper Yes People receive information through newspaper X50 Radio No People do not receive information through radio Yes People receive information through radio X51 TV No People do not receive information through television Yes People receive information through television X52 Public announcement No People do not receive information through public announcement Yes People receive information through public announcement X53 Magazine/journal No People do not receive information through magazine or journal Yes People receive information through magazine or journal X54 Neighbors/others No People do not receive information through neighbors or others Yes People receive information through neighbors or others X55 Internet No People do not receive information through internet Yes People receive information through internet X56 Village headman No People do not receive information through village 119 Abbrev. Topic States Descriptions headman Yes People receive information through village headman X57 Waste management Suitable Waste management is suitable Fair Waste management is fair Not suitable Waste management is not suitable X58 Gender of food- preparing person Male Food-preparing person is a man Female Food-preparing person is a woman X59 Formal education of food-preparing person None Food-preparing person does not go to school 300 times People consume meat more than 300 times in a year X67 Consumption habits Good The consumption habits is good Fair The consumption habits is fair Poor The consumption habits is poor X68 Knowledge of food- preparing person High Food-preparing person has high knowledge about how to prepare good food Medium Food-preparing person has medium knowledge about how to prepare good food Low Food-preparing person has low knowledge about how to prepare good food X69 Risk of eating outside the house High People face high risk of be infected by Trichinellosis when eating outside Medium People face medium risk of be infected by Trichinellosis when eating outside Low People face low risk of be infected by Trichinellosis when eating outside 122 Appendix F Descriptive statistics from the field survey 123 Table F-1: General information of the respondents Unit: Household Number (percent) Name of village Raise pig Do not raise pig Total Huai Ma Fueang Huai Chan Si Total 23 (42.59) 23 (42.59) 46 (85.18) 5 (9.26) 3 (5.55) 8 (14.81) 28 (51.85) 26 (48.15) 54 (100) General information of the respondents Respondent’s main responsibility Prepare food Raise pig Both Average age of respondents(SD.) Average member in family (SD.) Unit: Person 11 (20.37) 9 (16.67) 34 (62.96) 39.59 (15.08) 4.70 (2.01) 124 Table F-2: General information of the pig growers Unit: Household Number (percent) Gender of pig growers Male Female 20 (42) 28 (58) Ethnicity of pig growers Lisu Black Lahu Red Lahu Palong 4 (8.89) 1 (2.22) 39 (86.67) 1 (2.22) Identification of pig growers None ID card Burmese migrant (pink card) Highlander (green card) - 43 (93.48) 1 (2.17) 2 (4.35) Religion of pig growers Buddhism Christianity Islam Spirituality 35 (74.47) 10 (21.28) - 2 (4.25) Formal education of pig growers None 10 14 (28.57) 27 (55.10) - 8 (16.33) Recognition of rat danger No Yes 32 (64) 18 (36) Rat control No Yes 23 (47.92) 25 (52.08) Wild life presence No Yes 39 (79.59) 10 (20.41) 136 Table F-12: Financial status of the respondents Average (SD.) Source of income Primary source of income Labor (5 persons) Merchandise (3 persons) Crop farming (50 persons) Animal farming (2 persons) Secondary source of income Labor (29 persons) Merchandise (1 person) Crop farming (2 persons) Animal farming (21 persons) Unit: Baht 43,776.85 (31,853.63) 35,140 (14,914.69) 81,333.33 (62,010.75) 38,560 (29,089.25) 8,125 (4,419.42) 19,927.50 (34,411.34) 16,317.24 (25,753.87) 65,000 (-) 46,000 (48,083.26) 7,947.62 (7,214.59) Total annual income 58,537.96 (54,758.01) Number (percent) Saving No Yes Unit: Household 14 (25.93) 40 (74.07) What do people do with left over money? Save Buy products Invest in agriculture Provide loan for others Prepare for contingency purposes Prepare for vacation Prepare for kid’s education 40 (74.07) 22 (40.74) 30 (55.56) 3 (5.56) 7 (12.96) 1 (1.85) 12 (22.22) Investment in pig production No Yes Buy pigs Buy deworming drug Buy pig feed Improve pig hygiene 14 (31.11) 31 (68.89) 15 (48.39) 20 (64.52) 5 (16.13) 4 (12.90) Debt/on credit No Yes 12 (22.22) 42 (77.78) Source of fund/loan Neighbor Bank for Agriculture and Agricultural Co- operatives (BAAC) Village fund Government saving bank (GSB) Private leasing Informal leasing 21 (52.50) 11 (27.50) 19 (47.50) 1 (2.50) 4 (10.00) 8 (20.00) 137 Number (percent) Convenient Products Possess TV Radio Stove Fridge Rice cooker VCD/DVD Electric iron Electric fan Satellite dish Water pump Electric generator Bike Motorbike Car/truck Cell phone Cloth washer Laptop Electric pan Water heater Pesticide sprayer Lawn mower Unit: Household 45 (83.33) 21 (38.89) 12 (22.22) 33 (61.11) 27 (50.00) 30 (55.56) 9 (16.67) 34 (62.96) 43 (79.63) 4 (7.41) 3 (5.56) 25 (46.30) 48 (88.89) 5 (9.26) 19 (35.19) 4 (7.41) 2 (3.70) 1 (1.85) 1 (1.85) 1 (1.85) 1 (1.85) 138 Table F-13: Access to medical service of the respondents Number (percent) Right for free medical service None Universal coverage Low income Elderly Social security service Community leader Public health volunteer Private insurance Unit: Person 4 (7.41) 50 (92.59) 1 (1.85) 4 (7.41) 1 (1.85) 1 (1.85) 3 (5.56) 2 (3.70) Place people usually go for medical service Never been treated Buy medicine themselves Nearby clinic Tambon health promoting hospital District hospital Provincial hospital Traditional health care Spiritual treatment Fang hospital (a district hospital in another district) Prasat neurological hospital (provincial hospital) Unit: Household 1 (1.85) 34 (62.96) 23 (42.59) 36 (66.67) 41 (75.93) 7 (12.96) 13 (24.07) 9 (16.67) 5 (9.00) 6 (11.11) Source of information/news Newspaper Radio TV Public announcement Journal/magazine Neighbor Internet 5 (9.26) 18 (33.33) 47 (87.04) 25 (46.30) 1 (1.85) 38 (70.37) 1 (1.85) Services from institutions Frequency Headman Teacher Public health officer Animal health officer Public health volunteer Animal health volunteer Police Heifer officer Tree bank officer Sub district officer District officer Average Days Visited in a Year (SD.) 29.43 (45.44) 13.32 (52.98) 3.17 (7.84) 1.46 (5.32) 27.50 (76.58) 1.62 (4.44) 0.54 (3.32) 0.26 (1.65) 0.02 (0.14) 0.25 (1.64) 0.04 (0.27) 139 Benefit Headman Teacher Public health officer Animal health officer Public health volunteer Animal health volunteer Police Heifer officer Tree bank officer Sub district officer District officer Score rank from 0 to 4 (SD.) 2.09 (1.46) 1.81 (1.61) 1.26 (1.60) 0.87 (1.32) 2.31 (1.41) 0.69 (1.33) 0.09 (0.49) 0.13 (0.67) 0.02 (0.14) 0.22 (0.86) 0.02 (0.14) These numbers are low because we count those who did not receive service in the denominators 140 Appendix G Cost structures and revenue streams of pig production in highlands 141 Table G-1: Cost structures and revenue streams of pig production in highlands Unit: Baht/year Overall Pen Tether Average S.D. n Average S.D. n Average S.D. n Cost structure Fixed cost Pig house cost Pig house cost (depreciation cost/yr) Land rent Variable cost Feed cost Water supply/pig housing maintenance Fine (pigs disturb other properties) Other variable cost Cost per capita Total cost from pig rearing Total cost (with opp.cost) Revenue per capita Total revenue from selling pigs Profit from selling pig Profit per one pig Loss from selling pig Loss per one pig Profit (with opp.cost) Profit per one pig (with opp.cost) Loss per one pig (with opp.cost) 2,158.50 9,219.44 30 274.49 968.88 30 - 4,643.95 5,449.66 42 674.65 810.25 12 300 - 1 130.67 90.72 5 926.41 780.20 43 5,395.49 6,022.53 43 5,521.71 5,880.60 43 2,953.93 1,877.88 35 24,231.00 25,133.81 35 19,484.59 21,229.25 33 2,292.23 1,445.70 7 2,934.78 4,234.02 7 914.70 984.32 44 18,319.96 20,418.08 44 2,208.78 1,379.63 40 796.50 1,100 4 3,981.43 7,948.51 15 391.87 515.80 15 - 5,011.76 6,307.56 17 724.48 983.82 8 - 33.33 28.87 3 1,206.94 965.24 18 5,546.69 6,631.08 18 5,363.26 6,597.85 18 2,477.08 2,113.13 18 21,017.86 27,247.17 18 17,653.09 22,106.48 13 2,308.31 1,197.35 13 2,832.20 4,785.38 5 804.08 1,035.73 5 12,630.45 19,544.03 18 1,963.44 1,075.37 15 917.84 1,314.04 3 1,000 - 1 133.33 - 1 - 1,200 - 1 - - - 808.89 - 1 12,133.33 - 1 12,266.69 - 1 1,500 - 1 22,500 - 1 10,366.67 - 1 691.11 - 1 - - 10,233.31 - 1 682.22 - 1 - 141 142 Fence underneath house Outdoor-located fence Combination of outdoor-located fence and pen Average S.D. n Average S.D. n Average S.D. n Cost structure Fixed cost Pig house cost Pig house cost (depreciation cost/yr) Land rent Variable cost Feed cost Water supply/pig housing maintenance Fine (pigs disturb other properties) Other variable cost Cost per one pig Total cost from pig rearing Total cost (with opp.cost) Revenue per one pig Total revenue from selling pigs Profit from selling pig Profit per one pig Loss from selling pig Loss per one pig Profit (with opp.cost) Profit per one pig (with opp.cost) Loss per one pig (with opp.cost) 135 143.09 5 47.71 25.31 3 - 2,662.50 2,209.68 5 330 381.84 2 300 - 1 - 1,058.88 781.78 5 2,883.13 2,213.53 5 2,974.81 2,213.92 5 1,300 1,204.16 5 9,833.33 7,285.83 5 7,155.62 8,450.41 3 1,196.04 689.60 3 3,191.25 3,972.17 2 1,191.25 1,143.74 2 5,156.49 7,893.04 5 1,145.11 728.01 3 1,216.26 1,108.38 2 - - - 5,375 - 1 - - - 1,343.75 - 1 5,375 - 1 5,391.68 - 1 2,000 - 1 8,000 - 1 2,625 - 1 656.25 - 1 - - 2,608.32 - 1 652.08 - 1 - 1,333.33 763.76 4 262.03 176.69 3 - 3,096.56 644.73 4 - - - 594.36 326.74 4 3,293.09 823.57 4 3,493.08 898.20 4 2,825 1,325.08 4 16,000 2,449.49 4 12,706.91 1,630.49 4 2,230.64 999.27 4 - - 12,506.93 1,555.76 4 2,193.42 975.71 4 - 142 143 Combination of pen and free range Combination of fence underneath house and pen Combination of fence underneath house, pen and free range Average S.D. n Average S.D. n Average S.D. n Cost structure Fixed cost Pig house cost Pig house cost (depreciation cost/yr) Land rent Variable cost Feed cost Water supply/pig housing maintenance Fine (pigs disturb other properties) Other variable cost Cost per one pig Total cost from pig rearing Total cost (with opp.cost) Revenue per one pig Total revenue from selling pigs Profit from selling pig Profit per one pig Loss from selling pig Loss per one pig Profit (with opp.cost) Profit per one pig (with opp.cost) Loss per one pig (with opp.cost) 1,500 - 1 750 - 1 - 21,600 - 1 - - - 1,443.75 - 1 23,100 - 1 23,199.96 - 1 4,000 - 1 64,000 - 1 40,900 - 1 2,556.25 - 1 - - 40,800.04 - 1 2,550 - 1 - 236.67 100.17 4 87.78 25.89 3 - 3,699 2,368.97 4 - 300 - 1 - 501.43 208.18 4 3,839.83 2,417.30 4 4,031.53 2,459.34 4 2,750 2,217.36 4 36,333.33 20,502.03 4 31,513.56 18,773.18 4 3,148.09 1,301.20 4 - - 25,718.47 19,019.90 4 3,468.29 1,263 4 - 50 - 1 4.99 - 1 - 540 - 1 - - - 41.92 - 1 544.99 - 1 644.95 - 1 3,000 - 1 39,000 - 1 38,455.01 - 1 2,958.08 - 1 - - 38,355.05 - 1 2,950.39 - 1 - 143 144 Combination of pen and tether Combination of tether and free range Combination of fence underneath house and tether Average S.D. n Average S.D. n Average S.D. n Cost structure Fixed cost Pig house cost Pig house cost (depreciation cost/yr) Land rent Variable cost Feed cost Water supply/pig housing maintenance Fine (pigs disturb other properties) Other variable cost Cost per one pig Total cost from pig rearing Total cost (with opp.cost) Revenue per one pig Total revenue from selling pigs Profit from selling pig Profit per one pig Loss from selling pig Loss per one pig Profit (with opp.cost) Profit per one pig (with opp.cost) Loss per one pig (with opp.cost) 470 - 3 187.99 - 1 - 5,805 7,445.15 3 840 - 1 300 - 1 120 - 1 792.37 482.48 3 6,287.66 7,297.19 3 6,387.66 7,386.68 3 5,000 3,000 3 38,666.67 43,143.17 3 32,379 35,875.78 3 4,207.63 2,544.10 3 - - 32,279 35,782.44 3 4,193.24 2,534.51 3 - - - - 1,102.50 1,347.04 2 - - - 523.13 713.29 2 1,102.50 1,347.04 2 1,110.84 1,335.24 2 250 353.55 1 4,000 - 1 3,850 - 1 481.25 - 1 - - 4,889.16 1,493.18 2 1,725.83 - 1 - - - - 5,760 - 1 - - - 960 - 1 5,760 - 1 5,960.04 - 1 3,000 - 1 18,000 - 1 12,240 - 1 2,040 - 1 - - 12,039.96 - 1 2,006.66 - 1 - 144 145 Combination of pen, tether and free range Combination of outdoor-located fence, pen and free range Average S.D. n Average S.D. n Cost structure Fixed cost Pig house cost Pig house cost (depreciation cost/yr) Land rent Variable cost Feed cost Water supply/pig housing maintenance Fine (pigs disturb other properties) Other variable cost Cost per one pig Total cost from pig rearing Total cost (with opp.cost) Revenue per one pig Total revenue from selling pigs Profit from selling pig Profit per one pig Loss from selling pig Loss per one pig Profit (with opp.cost) Profit per one pig (with opp.cost) Loss per one pig (with opp.cost) 80 - 1 53.33 - 1 - 6,400 - 1 800 - 1 - 266.67 - 1 470 - 1 7,520 - 1 7,653.36 - 1 4,000 - 1 64,000 - 1 56,480 - 1 3,530 - 1 - - 56,346.64 - 1 3,521.67 - 1 - - 550 - 1 34.37 - 1 - 2,700 - 1 - - - 546.87 - 1 2,734.37 - 1 2,884.36 - 1 1,000 - 1 5,000 - 1 2,265.63 - 1 453.13 - 1 - - 2,115.65 - 1 423.13 - 1 - - 145 146 CURRICULUM VITAE Name Ms. Chalisa Kallayanamitra Date of Birth December 26, 1987 Educational Background M.A., Economics, Chiang Mai University, Chiang Mai Province, Thailand. GPA 4.00 (Expected graduation 2013) B.A., Economics, Chiang Mai University, Chiang Mai Province, Thailand. GPA 4.00 (2010) High School Diploma, Uttaradit Daruni School, Uttaradit Province, Thailand. GPA 3.95 (Study Structure: Arts-Mathematics) (2007) Exchange Student, Ross Sterling High School, Baytown, Texas, USA. (August 2005-June 2006) Junior High School Diploma, Uttaradit Daruni School, Uttaradit Province, Thailand. GPA 3.84 (2002) Academic Honors/Awards/Grants Exceptional Student Oral Presentation Award, 4th Biennial Conference of the International Association for Ecology and Health, Kunming, China. (15-18 October 2012) Participant, 5th International Students’ Forum, Oita University, Oita, Japan. (24- 28 August 2011) 147 Scholarship for Undergraduate Program from Bank of Tokyo-Mitsubishi UFJ. (10 September 2010) Academic Distinction Award (GPA 4.00) Academic Year 2009 Academic Distinction Award (GPA 3.93) Academic Year 2008 Academic Distinction Award (GPA 3.82) Academic Year 2007 Student Body President, Uttaraditdaruni School, Academic Year 2005 First Candidate King’s Bestow Award, Uttaradit Educational Service Area Office 1 (2005) Public Benevolent Student Award, Uttaradit Daruni School, Uttaradit Province, Thailand. (2005) Admired Juvenile Award in National Children’s Day, Ministry of Education, Thailand. (2005) Honorable Mention for the King’s Bestow Award, Ministry of Education, Thailand. (2004) Academic Distinction Award (GPA 3.93), Study Structure: Sciences-Mathematics (2003) Public Benevolent and Courteous Student Award, in National Children’s Day, Ministry of Education, Thailand. (2000) Professional Training/Exchange Programs/Internships Attended Workshop on “GIS and Spatial Analysis in One Health,” Faculty of Veterinary Medicine, Chiang Mai University, Thailand. (22 October-2 November 2012) 148 Attended Workshop on “Integrating EcoHealth in Transboundary Disease Risk Assessment and Management,” EcoHealth-One Health Resource Centre, Chiang Mai University, Chiang Mai Province, Thailand. (25-27 July 2012) Attended EcoHealth-One Health Lecture Series 2011 “Ecological Epidemiology and the Emergence of Zoonotic Diseases: Toward an Integrative Science”, EcoHealth-One Health Resource Centre, Chiang Mai University, Chiang Mai Province, Thailand. (2011) Student Exchange Program at College of Business Administration, Chung Chou Institute of Technology, Chang Hua, Republic of China (Taiwan). (15 March- 14 May 2011) SCG Excellent Internship Program at Concrete Products and Aggregate Co., Ltd. (CPAC), an Affiliate of SCG Cement, Chiang Mai Province, Thailand. (22 March-24 May 2010) Short Term Students Cultural Trip 2009, Kunming, People’s Republic of China. (8-15 October 2009) Work and Travel Program at SeaWorld, Discovery Cove and Aquatica, Florida, United States of America. (10 March-12 June 2008) American Foreign Service (AFS) Intercultural Exchange Program. Attended Ross Sterling High School, Texas, United States of America. (12 August 2005-29 June 2006) Attended Youth Democracy Training Program under the Secretariat of the House of Representatives and King Prajadhipok’s Institute. (20-30 March 2004) Attended the Promotion of Academic Olympiad and Development of Science and Education Foundation (2nd Camp) under the Patronage of Her Royal 149 Highness Princess GalyaniVadhana Krom Luang Naradhiwas Rajanagarindra. (14-29 March 2004) Academic and Community Service Teacher Assistant at Workshop on “Creating a Business: Exploring the Different Steps of How to Set-up and Run a Small Business” in Cooperation with Faculty of Economics, Chiang Mai University and PUM Netherlands Senior Experts. (27 February-9 March 2012) Facilitator at the Global Health Institute Training Course “One Health - Thailand 2012” in Cooperation with the Global Health Institute, the University of Minnesota, Chiang Mai University, USAID, DAI, and Tufts University. (30 January-9 February 2012) Assisted the 5th International Conference of the Thailand Econometric Society, Faculty of Economics, Chiang Mai University, Chiang Mai Province, Thailand. (12-13 January 2012) Assisted the 4th International Conference of the Thailand Econometric Society, Faculty of Economics, Chiang Mai University, Chiang Mai Province, Thailand. (13-15 January 2011)