Militarization, Risk and Community Responses

Health and Human Rights in Post-Conflict Eastern Burma: Militarization, Risk and Community Responses

William W. Davis, MA, MPH

A dissertation submitted to Johns Hopkins University in conformity with the requirements for the degree of Doctor of Public Health

Baltimore, Maryland

January, 2014

Executive Summary Background

Despite recent political reforms in Burma, human rights abuses are ongoing in ethnic regions in the country. Fighting continues in some areas, and ceasefires in others have not brought an end to human rights violations. In response to political reforms, the international community has

decreased political and financial support for ethnic communities in Burma, leaving them more

vulnerable than ever. Continued advocacy for an end to human rights abuses and for continued

humanitarian support for ethnic groups in Burma is greatly needed. The goals of this project

were to gather evidence of human rights abuses in Karen state, eastern Burma, identify health

consequences of these abuses and use the data to advocate for justice for victims.

Methods

We created a framework to describe social determinants of health in Karen state and used it to

generate hypotheses about the effects of militarization and human rights violations on health

and the role of village responses in moderating these effects. A second framework was used to

operationalize international human rights law into a survey instrument applicable in Karen state.

We then conducted a systematic literature review to determine state-of-the-art methods for cluster sampling in conflict areas and to identify cases in the peer-reviewed literature when security precautions may preclude following best practices for sampling.

Using the best practices identified by the literature review, we conducted a two-stage cluster survey of 686 households in eastern Burma in January 2012 that covered health status, access to healthcare, food security, exposure to human rights violations and identification of perpetrators.

Data analysis included descriptive and interpretive components. We used logistic regression to

identify associations between exposure to armed groups, village responses, human rights

abuses and health outcomes.

Results

This project provided timely data that documented human rights abuses in Karen state and the

need for humanitarian assistance. Data was used to advocate for continued humanitarian aid for

ethnic areas, continued pressure on the Burmese government to stop human rights abuses and

for perpetrators of human rights abuses to be held accountable for their crimes.

Household hunger, measured by FANTA-2 scale, was low in 581 (84.7%) households, moderate in 85 (12.3%) households and high in six (0.9%) households. Households reporting food shortages during any month in 2011 ranged from 19.9% to 47.0%, with food insecurity peaking just prior to the harvest. Diarrhea prevalence in children was 14.2% and night blindness in women of child-bearing age was 5.6%.

Forced labor was the most common human rights violation, reported by 185 households

(24.9%); 210 households (30.6%) reported experiencing at least one human rights violation in

2011. Multiple logistic regression analysis identified associations between human rights violations and poor health outcomes.

Novel Findings

This is the first study to assess the latest methods for cluster sampling in conflict areas. We found that researchers doing cluster surveys in conflict areas must sometimes alter methods to ensure safety and security of field surveyors. Common alterations are skipping and replacing

iii

clusters that are insecure, making fewer attempts to revisit empty households, using fewer field

supervisors and making contextual decisions on how best to allocate households to clusters.

Several common security protocols are identified.

This is also the first study to examine militarization and village responses in Burma. Close

proximity to a military base was a predictor of human rights violations, inadequate food

production, inability to access healthcare and diarrhea. Exposure to armed groups predicted these outcomes and also household hunger. In households that reported no human rights violations, risk of household hunger, inadequate food production, diarrhea and child diarrhea increased when neighbors’ households reported human rights violations. Households in villages that reported using any self-protection technique had lower risk of experiencing human rights violations. Households in villages that reported negotiating with the Burmese army had lower risk of human rights violations, household hunger, inadequate food production and diarrhea.

Stratified analysis suggests that self-protection may modify the effect of exposure to armed groups on risk of human rights violations and some health outcomes.

Acknowledgements I have been fortunate in my doctoral experience to have worked with so many intelligent and

inspiring people.

The team at Physicians for Human Rights showed me the value of a strong dedication to the mission and vision of an organization. My supervisor Rick Sollom trusted me to work

independently in the field and encouraged me to develop new projects and examine new angles

of human rights and health in the region. Andrea Gittleman and Hans Hogrefe taught me

everything I know about advocacy. I have never worked with a more open and collaborative

team. Vince Iacopino and Adam Richards gave me new insights into approaching research

design.

I spent a year in Mae Sot, Thailand, planning and implementing this project. There I developed

lasting friendships and working relationships. Collaborating with the partner organizations

reinforced the deep respect that I developed long ago for community-based organizations. I am

grateful to BPHWT, KDHW, CIDKP, KYO and the anonymous partner for collaborating with PHR

despite tremendous workloads. I learned a lot from our partners. Eh Kalu’ friendship, Mahn

Mahn’s candor, Win Kyaw’s dedication and Tah Doh Moo’s experience were all invaluable assets

to me. The surveyor training was successful because of the work of Htu Htike San as training

coordinator. Jen Leigh (of GHAP) and Cate Lee helped me settle into Mae Sot and introduce me

to potential partner organizations. They, along with Charlene, Sam, Tatyana and Myo gave moral

support. Matt Finch and Khu Khu Ju of KHRG gave excellent updated advice on the situation in

Karen state and taught me about village agency. The people at FBR graciously shared their experience inside Karen state with me and helped me to understand what it was like to live

there. Thanks to Matt Smith and the rest of the human rights people on the border for their

open sharing of information. I am glad to have befriended Vit Suwanvanichkij, a master of all

things related to public health and politics in Burma and on the border. Finally, I am grateful for the 22 men and women on the survey team who took two months out of their schedule to train and then implement the survey. Because of them this work was possible.

My colleagues and advisers at JHSPH helped me to grow as a scientist and public health advocate. My advisor Chris Beyrer is responsible for the collaboration between PHR and his

Center for Public Health and Human Rights at Hopkins. Chris taught me that advocacy and science are inseparable and that we always have to push for justice. Luke Mullany mentored me as a scientist and researcher. Bob Lawrence inspired me early on in my academic career to pursue health and human rights. Candid advice about everything Hopkins from Jim Tielsch helped me to navigate through years of grad school. Stef also gave similar advice. Sos K and the

Hopkins biostats clinic helped me through the data analysis. Andrea and Emily guided me through the IRB process. Thanks to Sarah for building the Access database and to Hannah for helping with the literature review. I am especially grateful for the team--my DrPH cohort and our PhD friends--for helping me to maintain the balance between working hard and having fun.

My family’s love and support got me through the difficult stages of the research and of the doctoral program. Long ago I learned from them that anything can be accomplished with hard work and a good sense of humor. Thanks to Beth, Abby, Alma, Dick, Phoebe, Stella and especially Ah Noh.

Table of Contents

Executive Summary ...... ii Acknowledgements...... v Table of Contents ...... vii List of Tables ...... viii List of Figures ...... ix Abbreviations ...... x Project Overview ...... 1 Problem Statement ...... 1 Specific Aims of the Dissertation: ...... 1 Theoretical Perspective ...... 2 Hypotheses and Predictions ...... 3 Experiment ...... 4 Policy Implications of this Research: ...... 4 Role of Bill Davis, DrPH candidate ...... 5 Chapter 1. Historical Context ...... 7 A Short History of Burma ...... 7 History of Karen State ...... 11 Chapter 2. Frameworks ...... 16 Human Rights Frameworks ...... 16 The Human Right to Health ...... 21 A Social-Ecological Model of Health for Karen State ...... 23 Social Ecological Models of Health Identify Distal Determinants of Health ...... 24 Conflict and HRVs are Determinants of Health ...... 26 Self-Protection May Shield Civilians from Effects of Militarization ...... 27 A Social Ecological Model for Post-Conflict Karen State ...... 30 Chapter 3. (Paper1) Methods for Cluster Surveys in Conflict Areas: A Review of the Literature . 39 Chapter 4 Research Methods ...... 63 Chapter 5. (Paper2) Health and Human Rights in Post-Conflict Karen State, Eastern Burma ...... 78 Chapter 6. (Paper 3) Social Ecology of Health in Post-Conflict Karen State, Burma: Militarization, Risk and Community Responses ...... 106

vii

Conclusions and Policy Implications ...... 130 Appendices ...... 131 References ...... 139 Curriculum Vita ...... 153

List of Tables Table 1. Operationalizing Human Rights Instruments for Karen State ...... 19 Table 2. Threats to Health from HRVs ...... 27 Table 3. Health, Militarization and Self-Protection in Karen State: a Social-Ecological Model ..... 34 Table 4. Recommendations for Interventions ...... 38 Table 5. Search Results ...... 53 Table 6. Data From Selected Papers ...... 54 Table 7. Survey Coverage ...... 96 Table 8. Population Structure ...... 96 Table 9. Characteristics of Sampled Population ...... 97 Table 10. Prevalence of Diarrhea ...... 98 Table 11. Prevalence of Night Blindness ...... 98 Table 12. MUAC in Children 6-59 Months ...... 98 Table 13. Responses to FANTA-2 Household Hunger Questions ...... 98 Table 14. Household Hunger ...... 99 Table 15. Months of Adequate Household Food Production ...... 99 Table 16. Access to Healthcare ...... 101 Table 17. Reported Source of Drinking Water ...... 101 Table 18. Human Rights Violations ...... 102 Table 19. Perpetrators ...... 103 Table 20. Associations Between Human Rights Violations and Health Outcomes ...... 104 Table 21. Comparison of Results of Imputed Analysis for Diarrhea ...... 105 Table 22. Survey Coverage, Excluding Tavoy Region ...... 121 Table 23. Health Outcomes ...... 121 Table 24. Human Rights Violations ...... 122 Table 25. Distance to Military Base, in Hours Hiking ...... 122 Table 26. Exposure to Armed Groups ...... 122 Table 27. Self-Protection Activities Reported by Households ...... 122 Table 28. Total HRVs per Village ...... 123 Table 29. Total HHs that Negotiate, per Village ...... 123 Table 30. Total Self-Protection Activities, per Village ...... 123 Table 31. Households with a Neighbor Reporting an HRV ...... 124 Table 32. Proximity to Army Base (in hours hiking) is Associated with HRVs ...... 124

viii

Table 33. Proximity to Army Base (in hours hiking) is Associated with Poor Health ...... 124 Table 34. Exposure to Armed Groups is Associated with HRVs ...... 124 Table 35. Exposure to Armed Groups is Associated with Poor Health ...... 125 Table 36. Health in Households that did not Experience HRVs is Associated with HRVs in Neighbors' Households ...... 125 Table 37. Household Health is Associated with Village HRVs ...... 126 Table 38. Village Self-Protection is Associated with Lower Risk of HRVs in Village ...... 126 Table 39. Village Self-Protection is Associated with Lower Risk of Poor Health...... 126 Table 40. Village Negotiation is Associated with Lower Risk of Poor Health ...... 126 Table 41. Village Negotiation Modifies the Association between Exposure to Armed Groups and Risk of HRVs ...... 127 Table 42. Village Self-Protection Modifies the Association between Exposure to Armed Groups and Risk of HRVs ...... 127 Table 43. Village Negotiation Modifies the Association between Exposure to Armed Groups and Risk of Poor Health ...... 128

List of Figures Figure 1. Household Production of Health ...... 26 Figure 2. Household Production of Health ...... 26 Figure 3. Militarization, Self-Protection and Household Production of Health ...... 27 Figure 4. Village-Level Exposures that may be Distal Determinants of Household Health ...... 32 Figure 5. Village Self-Protection may Modify the Effects of Militarization...... 33 Figure 6. Interactions Examined by this Analysis ...... 33 Figure 7. Population Pyramid ...... 97 Figure 8. Percent of Households Reporting Sufficient Food in Each Month of 2011 ...... 100 Figure 9. Household MAHFP, Stratified by Any HRVs ...... 100

Abbreviations AAAQ Accessibility, Affordability, Availability, Quality BPHWT Backpack Health Worker Team BGF Border Guard Force CIDKP Committee for Internally Displaced Karen People CBO Community-Based Organization COI Commission of Inquiry DKBA Democratic Karen Buddhist Army ESCR Economic, Social, Cultural Rights FANTA-2 Food and Nutrition Technical Assistance HHS Household Hunger Scale HRV Human Rights Violation ILO International Labor Organization KNU Karen National Union KDHW Karen Department of Health and Welfare KHRG Karen Human Rights Group KNLA Karen National Liberation Army KYO Karen Youth Organization MoH Ministry of Health MUAC Middle Upper Arm Circumference NLD National League for Democracy NSAG Non-State Armed Groups PHR Physicians for Human Rights SPDC State Peace and Development Council TBBC Thai-Burma Border Consortium USDP Union Solidarity Development Party

Project Overview Problem Statement Despite recent political reforms in Burma, human rights abuses continue in ethnic regions in the

country. Fighting continues in some areas, and ceasefires in others have not brought an end to

human rights violations. In response to political reforms, the international community has

decreased political and financial support for ethnic minority communities in Burma, leaving

them more vulnerable than ever. Continued advocacy for an end to human rights abuses and for

continued humanitarian support for ethnic groups in Burma is greatly needed.

This project provided timely data that documented human rights abuses in Karen state and the need for humanitarian assistance. Data was used to advocate for continued humanitarian aid for ethnic areas, continued pressure on the Burmese government to stop human rights abuses and for perpetrators of human rights abuses to be held accountable for their crimes.

Specific Aims of the Dissertation: 1) Systematic literature review on methods for cluster sampling conflict settings (Paper 1)

2) Multistage cluster survey in Karen state (Paper 2):

To measure the prevalence of barriers to accessing healthcare, food insecurity and human rights violations in eastern Burma

To identify associations between human rights violations and health outcomes

To identify human rights abuses and health consequences that result from militarization

3) Develop a social-ecologic model to frame militarization and village responses as determinants of health in Karen state (Paper 3)

4) Communicate findings to the general public, academic audiences and governments.

To advocate for victims of abuses and for mechanisms to prevent future abuses

Theoretical Perspective The aims of this project are to measure human rights abuses in Karen state, identify health consequences of militarization, examine the health consequences of abuses and advocate for justice for the victims. The survey tool was designed to measure human rights abuses that are known to occur in Karen state and also health and nutrition status of respondents and their families. The data analysis includes a descriptive component of prevalence of abuses, perpetrators and health status of respondents, and also an interpretive component designed to identify associations between human rights abuses and health outcomes, human rights abuses and areas of administrative control, exposure to armed groups and health outcomes and the role of village self-protection activities in mediating the effects of exposure to armed groups.

The descriptive component includes prevalence of all human rights violations (HRVs), parties identified as responsible for those violations and the health status of respondents, along with

95% confidence intervals. Health status is compared with health indicators from other areas in

Burma.

The first interpretive component explores associations between human rights violations and health outcomes. Previous studies in Karen state have shown that forced labor and other violations are associated with child malnutrition, diarrhea, and night blindness. (BPHWT, 2006)

Studies elsewhere in Burma have found that rights violations related to food security, i.e. theft

or destruction of food, forced to grow crops, and blocked from accessing land were associated with household hunger. (Sollom et al., 2011)

The second interpretive component examines associations between exposures to armed groups, distance to military bases, human rights violations, health outcomes and the role of village self- protection that, when combined, describe a social-ecologic model of health in Karen state.

Peace agreements between the Burmese army and the Karen National Union were recently signed, but a heavy military presence remains in eastern Burma. Qualitative reports suggest that this militarization can result in human rights abuses in the absence of conflict, but that Karen civilians try to reduce the impact of human rights violations by using self-protection techniques.

Militarization and self-protection in addition to human rights violations may be determinants of health for people living in Karen state.

Because the ultimate goal of this research is advocacy, the statistical measures and tests should be easy to explain and understand. Risk ratios and 95% confidence intervals are used to examine the associations between HRVs and health outcomes and also HRVs and administrative area.

Risk ratios have been used in previous studies to show associations between poor health and human rights abuses. (Mullany et al., 2007; Sollom et al., 2011)

Hypotheses and Predictions • Forced labor is still present in Karen state, and based on previous quantitative and

qualitative reports, we predict its prevalence at 12%.

• Households with HRVs will have higher measures of food insecurity, household hunger,

diarrhea, night blindness, and under-5 malnutrition than households that did not

experience HRVs.

• Households living in areas under complete Burmese Army or BGF control will have more

forced labor and pillaging than violations than households living in contested areas.

• Exposure to armed groups (via proximity to army bases or sightings) will increase risk of

human rights violations and poor health outcomes.

• Human rights violations in one household will have negative effects on health in

neighboring households.

• Village self-protection will modify the effect of exposure to armed groups on human

rights violations and also health outcomes.

Experiment A multi stage cross-sectional cluster survey in Karen state, covering Burmese army-controlled areas and areas of mixed administration was used to gather information on the extent of human rights abuses, access to healthcare, health status, and food security of the surveyed population.

Data analysis yielded the prevalence of HRVs, the association of HRVs with health outcomes, and the association of HRVs with different administrative areas.

Policy Implications of this Research: The data gathered in this research was used to advocate for the immediate and long-term needs of Karen people in Burma. Advocacy reports called for continued humanitarian aid, for the

Burmese Army to stop attacks on civilians and for language to stop human rights abuses to be included in ceasefire documents. In the long- term, the data gathered will inform reconciliation

processes and help to push for continued international aid and for the inclusion of community- based organizations and local input in development programs.

Role of Bill Davis, DrPH candidate The dissertation proposal presented here project is part of a collaboration between Physicians for Human Rights, The Center for Public Health and Human Rights at Johns Hopkins Bloomberg

School of Public Health, and five community-based organization working in Karen state:

Backpack Health Worker Team (BPHWT), Karen Department of Health and Welfare (KDHW),

Karen Youth Organization (KYO), the Committee for Internally Displaced Karen People (CIDKP) and one group that wishes to remain anonymous. BPHWT and KDHW operate health clinics in

Karen state, CIDKP gives aid for displaced persons in Karen state and KYO works in community development and civil society activities.

Bill Davis worked for one year in Thailand as the Burma Project Director for PHR and its sole representative in the region. His main objective was designing and implementing the survey described in this proposal, analyzing data and presenting results to the general public, news media and governments.

As Project Director, he was tasked with identifying organizations with which to partner, convincing them to sign onto the project and negotiating memoranda of understanding. Other tasks included assessing the logistical limitations of the implementing partners, determining the sample frame and sample size, piloting the questionnaire, creating training curriculum and implementing the surveyor training, arranging for translations of the questionnaire, all aspects of budget, ensuring surveyor security, surveyor debriefing, data entry, data analysis, report

writing, overall problem solving and advocacy. Advocacy efforts included leading press conferences, writing op-eds and blogs, giving interviews to the press, holding informational meetings with representatives from the UN, INGOs, foreign embassies in the region, US State department, US Congressmen and giving testimony in the Canadian House of Commons

Subcommittee on Human Rights.

Parts of this dissertation have been previously published as an advocacy report, which Bill Davis wrote and is first author, here: https://s3.amazonaws.com/PHR_Reports/burma-karen-rpt-ltr-

2012.pdf

Chapter 1. Historical Context Burma/ Myanmar Nomenclature

In the aftermath of the 1988 pro-democracy uprising, the military Junta that ruled Burma

officially changed the country’s name to Myanmar. Several groups refused to recognize the

name change in protest of the Junta. These groups included the National League for Democracy,

exiled ethnic governments, political prisoner organizations, the international press and most

western democratic countries. The Junta initiated some democratic reforms in 2010 and in

support of the changes foreign governments and the press began calling the country

“Myanmar.” Exiled ethnic nationalities, including all of the partner organizations for this project,

continue to call the country “Burma.” Out of respect for them, “Burma” is used throughout this

dissertation.

A Short History of Burma Burma is the most populous country in Southeast Asia, comprising about 55 million people. The

majority Burman people make up about 70% of the population; they live mostly in the central

plains of the country, often called “Lower Burma,” which includes the major cities of Rangoon

and Mandalay. The other 30% of the population is made of ethnic minority people, who mostly

live in the mountainous areas along the borders with Bangladesh, India, China, Lao, and

Thailand. Most of the ethnic minorities have different religion, language, and identity from the

majority Burmans.

Burma was ruled by a military dictatorship from 1962 until 2011. The Junta regularly repressed

opposition political parties, censored the press and imprisoned thousands of dissidents, including Nobel Peace Prize winner Aung San Suu Kyi. Protests against the government were met with violent repression: over 3000 civilians were killed in student protests in 1988, and a

protest in 2007 led by Buddhist monks resulted in hundreds killed and many more imprisoned

and tortured. The Junta adopted a new constitution in 2008 that set the stage for a new government and held elections in 2010, although many articles in the new constitution ensured

that the military would retain ultimate control. The constitution guaranteed that about 25% of

seats in parliament went to active members of the military, and in an election in 2010 that the

international community decried as unfair, most of the civilian seats went to retired military

officers loyal to the old regime. The constitution did not guarantee ethnic minorities rights and

several of parties representing ethnic minorities were banned from contesting the 2010

election.

In 2011 the new government in Burma enacted several reforms to promote democracy. It released dissident Aung San Suu Kyi from house arrest, released hundreds of political prisoners, eased censorship of local media, lifted the ban on international media, allowed Suu Kyi’s photo to be displayed in public and started ceasefire negotiations with armed ethnic groups.

Though these reforms were met with skepticism from human rights activists, the international community, especially European Union (EU) member nations and United Nations (UN) agencies, were quick to embrace them. Many countries sent high-level diplomats to visit Burma, lifted economic sanctions and increased development aid. Several US politicians who were once outspoken critics of the Junta also joined this group.

The sudden shift in international policy is meant to reward reformists in the Burmese government --there are also hard liners pushing to go back to the old style of rule-- and to encourage continued reform, but an indirect effect is to marginalize ethnic groups. Ethnic groups were once seen as the forces of change and victims of the regime in Burma, and the

international community funded health, education, and development programs, although it refrained from giving military aid. Atrocities committed by the Burma army in ethnic areas were documented and used as advocacy to push for sanctions and to bring attention to the problems in Burma.

Now that international perception towards Burma has changed, donor money is being shifted from programs in ethnic areas to programs in central Burma—in part to reward reforms and in part because they believed aid funneled through central Burma would reach ethnic minority areas. Reports of human rights violations are receiving less media coverage and are met with more skepticism than in the past, possibly because more attention is on reforms in the government. (UNOHCHR, 2011) People in ethnic areas, however, continued to be persecuted by the military and ignored by development programs planned by the central government.

The Burmese government has a long history of committing human rights abuses. Successive military dictatorships under Generals Ne Win and Than Shwe sought to subjugate ethnic people in attempts to create a culturally homogeneous, “Burmanized” country. Ethnic peoples formed their own armies to resist Burmanization and to maintain autonomy, and since 1948 the

Burmese army has been engaged in combat with at least one opposition group. Burmanization is no longer a goal of the central government, but conflict continues to be driven by attempts to control natural resources in ethnic states—many ethnic areas in Burma are rich in timber, jade and gold mines, oil and gas fields, and opium and methamphetamine production. The central government is pursuing hydroelectric and oil pipeline projects in ethnic areas, and this is also contributing to conflict.

The Burmese army allegedly commits human rights violations around extractive industries and economic development projects. (ERI, 2012) In response to reports of widespread human rights violations, in 1996 the International Labor Organization (ILO) launched a Commission of Inquiry

(COI) into forced labor in Burma. (ILO, 1998) It estimated that the Burmese government and especially the military forced 800,000 Burmese citizens, including prisoners, to labor for government projects, including portering, minesweeping, and providing sexual services. (ERI,

2012; ILO, 1998; VOA, 2009)

The ILO commission also found that the Burmese government used forced labor for private enterprises, including to “promote joint venture developments, including the country's oil and natural gas reserves; encourage private investment in infrastructure development, public works, and tourism projects; and benefit the private commercial interests of members of the Myanmar military.” (ILO, 1998)

One such economic development project is the Tavoy/Dawei deep sea port and economic development zone, which was included in this study’s sampling frame. The development projects around Tavoy are currently underway, and the Burmese government has proposed industrial development projects in other parts of Karen state. Some of these projects have been

proposed to promote ceasefire deals because they could provide jobs for displaced people and also enrich local leaders. Local groups, however, have criticized them as they are associated with human rights abuses and local people rarely receive benefit.

The government of Burma is under pressure to negotiate ceasefires with ethnic groups, and this is being done in a hasty manner that bodes ill for success. Ceasefires between ethnic groups and

the Burmese government have a history of failure. A ceasefire signed between the Shan State

Army South (SSA-S) and the Burma Army has been broken twice since January 2012, and a 17-

year ceasefire in Kachin State failed in 2011. Past ceasefires have allowed the Burma Army to

resupply its troops and fortify its bases to prepare for future assaults. Rights of civilians are

frequently overlooked in negotiating agreements, and the Burma Army regularly pillages civilian

property to feed its troops and use civilians for forced laborers, porters and minesweepers.

In the long term, the ceasefires may lead to truces, and the democratic reforms occurring in central Burma may eventually reach ethnic people living in the periphery. But the immediate effects of the reforms—shifts in funding, continued attacks from the Burma Army or bad deals from ceasefires--could have negative impact on civilians living in ethnic areas.

History of Karen State Karen state is in eastern Burma, bordering Thailand. Geography in the area ranges from jungle- covered mountainous regions to coastal plains with paddy fields. The mountainous areas, especially in former conflict areas, have very few roads or other infrastructure. Mobile phone networks are only beginning to be put in place in the main towns, and coverage is extremely low. According to Burmese government maps, the area of Karen state is about 30,000 km2

(Amoz et al., 1998). The Karen National Union (KNU), a Karen nationalist opposition group, calls the areas where Karen people live “Kawthoolei” and defines this area as much larger than the

Burmese maps do, extending into Bago and Tanintharyi divisions to the south and north and

Mon State to the west (see maps in appendix). Karen people also live in the Irrawaddy delta, south of Rangoon. The Karen use different names for villages and administrative areas than the

Burmese use, and although efforts have begun to create a master list of villages with all their names, there is currently much confusion around village names and locations on maps.

Complete census data for Karen state is not available. The last reliable census in this area was completed in 1931; the British estimated that there were about 1,300,000 Karen people (BEWG,

2011). Current estimates suggest about 2 to 3 million people in total live in eastern Burma

(BEWG, 2009), and 5 to 7 million living in Karen worldwide, including in the Irrawaddy delta and

Insein township north of Rangoon. (KBDDF, 2010) Though the exiled KNU makes the generalization “Karen People,” the denizens of Kawthoolei actually comprise several different ethnicities and have different religions and even language from each other.

Multiple opposition groups have operated in Karen state since World War II. The KNU, formed in

1949, is currently one of the more powerful groups. It was founded to promote the establishment of a separate country, but since the 1970s has advocated for a federal system in

Burma. In the early 1990s a group of Buddhists broke off from the predominately Christian leadership of the KNU and established the Democratic Karen Buddhist Army (DKBA).

In 2010 the Burmese government asked all ethnic armies to become part of the Burmese Army in a newly-established Border Guard Force (BGF) that would operate as separate units but under

Burmese Army command. The armed wing of the KNU, the Karen National Liberation Army

(KNLA) refused, but most DKBA units joined. During the time of this study, DKBA units were deserting from BGF and operating on their own.

Fighting between ethnic Karen groups and the Burmese army has taken its toll on the civilian population. Ethnic armies use landmines, recruit child soldiers and levy taxes on civilians. The

Burmese Army regularly takes food and money from civilians, attacks civilian villages, and uses

civilians as porters, guides and minesweepers. Attacks on civilians and pillaging of civilian property have been part of the military’s “four cuts” policy that began in the 1970s. (HRDU,

1998; KDHW, 2012) This policy was aimed at cutting food, funds, information and recruits from insurgent groups with the goal of weakening them. The military allegedly continued the policy in

Karen state and other areas into 2011. (Wai Moe, 2011)

The Burmese Army is responsible for the majority of human rights abuses in Karen state, although armed insurgent groups have used child soldiers and landmines and have been accused of unfair taxation and displacement of civilians. (HRW, 2007) Most abuses by the

Burmese Army tend to occur during troop movements and periods of fighting, and the number of abuses varies seasonally and also from year to year. (KHRG, 2010) During the May to

September rainy season, roads and trails become impassable, restricting movements and making fighting difficult. During this time Burmese Army troops tend to stay near their bases and only go out on short patrols. When the rainy season ends, troops and supplies are moved to forward bases in anticipation of fighting. The Burmese Army seems not to want to risk encountering armed insurgents when it moves supplies, so during these times it will mortar villages to clear them before passing through, and also mortar indiscriminately along roads and around bases.

The Karen Human Rights (KHRG) group published several reports that examined the types of human rights abuses in Karen state and how they vary in different administrative areas. They concluded that the Burmese Army tends to use just enough force to bully a population into compliance. In areas where the Burmese Army has complete control of the population, that is, where there is very little resistance, abuses tend to be more in the form of extorting food and

labor. In areas where the Burmese Army has a weaker presence, such as in areas where resistance movements are strong or remote areas far from roads, the human rights abuses tend

to be violent assaults. (KHRG, 2010; South, Perhult Malin, & Carstensen, 2010) KHRG suggests

that in these areas, more bullying in necessary to control the population. (KHRG, 2010)

Abuses also vary by year, depending on Burmese Army offensives and the extent of open fighting. For example, in 2006 multiple abuses occurred in Dooplaya district after the Burmese

Army broke a ceasefire with the KNU and attacked and burned several towns in KNU territory.

The Burmese Army tactics in this offensive included forced displacement of civilians. (AI, 2008;

KHRG, 2009)

In 2011, during the time of this survey, the Burmese Army was not as active in Karen state as in previous years, possibly because peace negotiations were underway. (KHRG, 2011b) At this time heavy fighting was ongoing in Kachin and Shan States, and the Burmese military was perhaps concentrating its logistics efforts and troop strength in these areas.

Historically, human rights abuses in Karen state were associated with fighting and conflict, such as violent attacks on civilians and villages. Recent reports, however, have suggested that militarization-- the presence of large numbers of troops-- may still result in violations of civilians’ human rights, even if the troops are not engaged in combat operations. (KHRG, 2012) In addition to forced labor (portering, cooking, cleaning and building military bases, etc), and pillaging food, soldiers assigned to protect extractive industries or economic development projects may forcibly relocate civilians or engage in intimidation to remove civilians from land where mining, logging or pipeline construction is ongoing. The ways in which Burmese army

troops are garrisoned and supplied, and rules by which they must abide when interacting with civilians could be negotiated into ceasefire treaties.

Decades of war and human rights violations have taken a toll on health of civilians in Karen

state, and political changes in the central part of Burma have yet to reach this area. Ceasefire

agreements do not necessarily mean an end to human rights abuses or improved healthcare in

Karen state. The government of Burma has not acknowledged its role in past abuses, and this

impunity increases the risk of abuses continuing. The data collected in this survey is used to

advance transitional justice by documenting abuses and perpetrators of the abuses; it links human rights violations to poor health outcomes, to enable advocacy to include health care as part of reparations for abuses; and finally it explores links between economic development

projects and government administration to human rights abuses. The findings have been used

to advocate for protections for civilians around development projects and in ceasefire areas.

In late 2011 the Burmese government engaged in several rounds of ceasefire talks with the KNU.

The KNU submitted an eleven-point proposal of their goals for the talks (KNU, 2012), which included guaranteeing safety and human rights of all civilians specifically mentioning forced labor and extortion. A preliminary ceasefire between the Burmese army and the KNU has been signed, but the Burmese army maintains a heavy troop presence in the region. Data collected by this survey will help to show the severity of human rights violations in Karen state and will be used to convince the Burmese government to include specific language in the ceasefire agreement that would halt human rights violations and hold perpetrators accountable.

Chapter 2. Frameworks Human Rights Frameworks Two of the aims of this project are to measure human rights abuses that have happened in

Karen state and to examine the association between human rights abuses and health outcomes.

The human rights component of the survey was based on qualitative and quantitative data collected previously in Karen state and also PHR surveys that were done in Chin and Shan states.

(KHRG, 2010; KHRG, 2011a; KHRG, 2011b; Mullany et al., 2007; Sollom et al., 2011) The human rights violations assessed by the research instrument include war crimes and crimes against humanity.

Crimes against humanity were defined in the Rome Statute of the International Criminal Court in

2002 as a “Widespread or systematic attack against a civilian population, tolerated or condoned by a state or authority that also has knowledge of the attack.” (International Criminal Court,

2002) The Rome Statute also calls them “a serious attack on human dignity and grave humiliation of a group of humans.” According to the Statute, any of the following eleven human rights violations are serious enough to meet this definition if they are also systematic and condoned by a state:

Murder

Extermination

Enslavement

Deportation or forcible transfer of population

Imprisonment in violation with fundamental rules of international law

Torture

Sexual violence

Disappearance

Apartheid

Other acts causing great suffering

A key component of the Rome Statue defines the jurisdiction of the International Criminal Court

(ICC); it states that ICC can prosecute crimes against humanity when a state is not able or not willing to do so. This is the case in Burma. Because Burma is not a state party to the ICC, only the

UN Security Council could initiate an investigation. One way justice could be upheld is if the ICC decides to prosecute for crimes against humanity, which it can only do if there is data available on these crimes.

A quantitative population-based survey can provide evidence that the attacks were directed against a population and not just a few individuals, which is a key component of the definition of crimes against humanity. This survey asked questions about the perpetrators, so the human rights violations can be linked to the Burmese government or other authorities. This survey cannot show directly that the state condoned attacks on civilians, but statements from Burmese generals and patterns of attacks may show that they intended to target civilians. Furthermore, the state has known about the decades of evidence of attacks on civilians, and at the very least, it has tolerated these acts.

In addition to crimes against humanity, war crimes and other human rights violations have occurred in Karen state and could be documented by the survey. The Fourth Geneva

Conventions and optional protocols I and II to the Geneva Conventions describe protections of civilians in armed conflicts, and they are relevant in Karen state.

Burma is a party to very few international human rights treaties. Of treaties that most apply to the situation in Karen state, Burma has signed, acceded or ratified: Convention on the

Elimination of all forms of Discrimination against Women (CEDAW); Convention on the Rights of the Child (CRC); Protocol to Prevent, Suppress and Punish Trafficking in Persons, Especially

Women and Children; Protocol against the Smuggling of Migrants by Land, Sea and Air;

Convention concerning Forced or Compulsory Labour; and Convention for the Suppression of the Traffic in Persons and of the Exploitation of the Prostitution of Others. (University of

Minnesota Human Rights Library, )

Other treaties to which Burma is a party include: Convention and International Convention for the Suppression of Terrorist Bombing; United Nations Convention against Transnational

Organized Crime; Freedom of Association and Protection of the Right to Organise Convention;

Convention on the Prevention and Punishment of the Crime of Genocide; International

Convention for the Suppression of Unlawful Seizure of Aircraft; and International Convention for the Suppression of the Financing of Terrorism. (University of Minnesota Human Rights Library, )

Documentation of violations of human rights treaties that Burma has signed can provide powerful evidence for advocacy efforts, as it shows Burma is violating international law that it has agreed to follow. That Burma has not signed some treaties, however, does not necessarily mean that they do not apply to Burma. International humanitarian law (IHL) regarding protection of civilians is widely regarded as common law, and does not require a signed treaty to be relevant. (Haenckerts, 2006) The principle of Jus cogens can also apply to human rights violations; it is “a peremptory norm of general international law [which] is . . . accepted and

recognized by the international community of States as a whole as a norm from which no derogation is permitted.” (International Law Commission, 1969)

The table below shows the relationships between the survey questions on human rights, the related human rights and the relevant international human rights instruments. The human rights in the table are defined by the Universal Declaration of Human Rights (UDHR). (UNGA, 1948)

Though this is not a binding treaty, it is the foundation of modern thinking in human rights and is the basis of most international instruments. According to the Proclamation of Teheran from the United Nations International Conference on Human Rights (1968), the UDHR defines fundamental freedoms, customary international law, and “constitutes an obligation for members of the international community.” (International Conference on Human Rights, 1968)

Table 1. Operationalizing Human Rights Instruments for Karen State

*possible crime against humanity **International instrument to which Burma is a party Survey Question Human Right (defined by the Relevant Instruments Universal Declaration of Human Rights) Section 5- forced displacement: Article 13* “Everyone has the right to ICCPR, Article 12 -How many times has your household freedom of movement and residence Vienna Declaration 1993 been forced to move in the last year? within the borders of each State” UN guiding principles on internal -What was the main reason you moved? displacement -How many places have you lived in the Convention on the Status of Refugees last ten years? (non refoulement) Fourth Geneva Convention Protocols I and II of the Geneva Conventions Section 6- forced labor: Article 4* “No one shall be held in slavery “ ‘Forced or compulsory labour’ shall -Has your household been forced to do or servitude” mean all work or service which is exacted labor in the last 12 months and you did from any person under the menace of any because you were afraid? Article 23 “Everyone who works has the penalty and for which the said person has right to just and favourable remuneration not offered himself voluntarily. ” --Forced ensuring for himself and his family an Labor Convention, 1930 existence worthy of human dignity, and Article 7, ICESCR supplemented, if necessary, by other Article 8, ICCPR means of social protection.” Have you been forced to… Article 3 “”Everyone has the right to life, Forced Labor Convention, 1930 -porter liberty and security of person. Article 7, ICESCR Article 8, ICCPR Article 4* “No one shall be held in slavery Customary International Humanitarian or servitude” Law

Fourth Geneva Convention Protocols I and II of the Geneva Conventions -sweep land mines Article 3* “”Everyone has the right to life, Forced Labor Convention, 1930 liberty and security of person. Article 7, ICESCR Article 8, ICCPR Article 4* “No one shall be held in slavery Customary International Humanitarian or servitude” Law Fourth Geneva Convention Protocols I and II of the Geneva Conventions -grow crops Article 4* “No one shall be held in slavery Forced Labor Convention, 1930- or servitude” “Governments may only force citizens to grow crops when it is a “precaution against famine or a deficiency of food supplies” Article 7, ICESCR Article 8, ICCPR -do other labor for the military Article 3 “”Everyone has the right to life, Forced Labor Convention, 1930 liberty and security of person. Article 8, ICESCR Article 8, ICCPR Article 4* “No one shall be held in slavery or servitude” -do any other kind of labor Article 3 “”Everyone has the right to life, Forced Labor Convention, 1930 liberty and security of person. Article 8, ICESCR Article 8, ICCPR Article 4* “No one shall be held in slavery or servitude” -did you also have to pay money for Article 4 “No one shall be held in slavery Forced Labor Convention, 1930 these things? or servitude” Article 8, ICESCR Section 7- Other HRVS: Article 13*“Everyone has the right to ICCPR, Article 12 -Have you been blocked from accessing freedom of movement and residence your land? within the borders of each State” -Has your food or crops been stolen? Article 17 “No one shall be arbitrarily CRC**“No child shall be subjected to deprived of his property.” arbitrary or unlawful interference with his or her privacy, family, [or] home….”

United Nations Convention Against Transnational Organized Crime** Customary International Humanitarian Law** -Has anyone tried to restrict your Article 13*“Everyone has the right to ICCPR, Article 12 movement? freedom of movement and residence within the borders of each State” -Have you been religiously or ethnically Articles 1*, 2, 7 “All human beings are UN Charter** obliges “respect for persecuted? born free and equal in dignity and rights. human rights and for fundamental They are endowed with reason and freedoms for all without distinction as to conscience and should act towards one race, sex, language, or religion” another in a spirit of brotherhood.” Section 8- Violent HRVs Article 3* “Everyone has the right to life, CRC**- “child shall not be forcibly -Has anyone been kidnapped or liberty and security of person.” separated from parents” disappeared? Article 9* “No one shall be subjected to arbitrary arrest, detention or exile.” -Has anyone been injured or killed by Article 3* “Everyone has the right to life, UDHR Article 3 gunshot, explosion, or deadly weapon? liberty and security of person.” Customary International Humanitarian Law Fourth Geneva Convention Protocols I and II of the Geneva Conventions -Has anyone been beaten for longer than Article 9* “No one shall be subjected to U.N. Convention against Torture and ten minutes at a time or tortured? arbitrary arrest, detention or exile.” Other Cruel, Inhuman or Degrading Treatment or Punishment CRC** Protocol Against the Smuggling of

Migrants by Land, Sea and Air**“protect the rights of persons . . . in particular the right to life and the right not to be subjected to torture or other cruel, inhuman or degrading treatment or punishment.”

ICCPR Article 7 -Has anyone been sexually assaulted? Article 3* “Everyone has the right to life, CEDAW**“The definition of liberty and security of person.” discrimination includes gender-based violence, that is, violence that is directed against a woman because she is a woman or that affects women disproportionately. It includes acts that inflict physical, mental or sexual harm or suffering”

Section 3 Health Article 3* “Everyone has the right to life, ICESCR Article 12 and General Comment liberty and security of person.” 14 International Convention on the Article 21 “Everyone has the right of Elimination of All Forms of Racial equal access to public service in his Discrimination Article 5 country.” CEDAW Articles 11,12,14 CRC Article 24 Article 22 “Everyone, as a member of International Convention on the society, has the right to social security Protection of the Rights of All Migrant and is entitled to realization, through Workers and Members of Their Families national effort and international co- Articles 28, 43 operation and in accordance with the Convention on the Rights of Persons with organization and resources of each State, Disabilities Article 25 of the economic, social and cultural rights indispensable for his dignity and the free development of his personality.”

Article 25 “Everyone has the right to a standard of living adequate for the health and well-being of himself and of his family, including food, clothing, housing and medical care and necessary social services, and the right to security in the event of unemployment, sickness, disability, widowhood, old age or other lack of livelihood in circumstances beyond his control.”

The Human Right to Health

The human right to health is defined in the International Covenant on Economic, Social, and

Cultural Rights (ICESCR) as "right of everyone to the enjoyment of the highest attainable standard of health". General Comment 14 of the ICESCR expands this definition to include other rights that address the underlying determinants to health: “rights to food, housing, work, education, human dignity, life, non-discrimination, equality, the prohibition against torture,

privacy, access to information, and the freedoms of association, assembly and movement.”

(CESCR, 2000) General Comment 14 further states that healthcare must be affordable, accessible, available and of high quality (AAAQ). It highlights that states cannot discriminate in healthcare delivery, that all groups including vulnerable population and minorities must have equal access to healthcare. General Comment 14 defines states’ responsibilities to respect, protect and fulfill their citizens’ rights to health. When states do not have sufficient resources to fulfill the right to health, they are required to implement policies that would continuously strive to fulfill the right to health for their citizens.

The AAAQ criteria can be used to identify barriers to healthcare and thus assess the right to health. (Hunt, 2006; H. Potts, 2009) The respect-protect-fulfill framework can help to identify the relationship between the state and the barrier to health in order to overcome the barrier or to hold the state accountable. These frameworks can also be used assess health systems’ abilities to fulfill the right to health. The health system in Karen state is at a much more basic level than most, and the immediate problems that must be addressed are the state’s failure to respect health.

In many cases, the state is failing to respect the health of its people by directly attacking determinants of health—restricting people’s ability to move freely and to grow crops and also through forced labor, which consumes time that otherwise would be spent providing for the family and producing health. If a ceasefire happens and the government stops these activities, then ensuring that the health status of Karen people is on par with the rest of Burma, and ensuring that the government promotes progressive realization of health in Karen state will be priorities.

Single questions can be used to assess the right to health as it pertains to access to healthcare; for example, “have you even been sick and not able to get healthcare.” Stronger documentation, however, comes from analyses and comparison of datasets. These analyses can reveal uneven distributions of health services and health outcomes across geographical, temporal, or ethnic lines.

The analysis for this study will include identifying state-imposed barriers to healthcare, examining the health status of respondents to the survey and comparing them to the health of other groups of people in Burma. Health data is available for Burma from WHO, the Myanmar

Ministry of Health, UNDP and CBOs that are working inside Karen state.

A Social-Ecological Model of Health for Karen State In 2011 the World Development Report estimated that 1.5 billion people lived in conflict areas, fragile states or areas affected by large-scale criminal violence. The report further stated that no country in which this kind of violence was widespread had achieved any Millennium

Development Goals. (World Bank, 2011) Human rights violations against populations are frequently associated with poor health outcomes, and civilian self-protection may play a role in preventing HRVs or mitigating the effects of them. We created a social-ecological model that incorporates self-protection, human rights violations and militarization as factors that affect production of household health in post-conflict Karen state. A social-ecological model illuminates some of the key determinants of health specific to Karen state; this model could inform health interventions and social reconciliation.

Social Ecological Models of Health Identify Distal Determinants of Health Social-ecological models of health are grounded in tenets that the environment and society in which a person lives plays a crucial role in determining that person’s health. (Grzywacz & Fuqua,

2000; Krieger, 2001; Marmot, 2005; McLeroy, Bibeau, Steckler, & Glanz, 1988) These models expand beyond biological paradigms of causality and enable an examination of factors such as economic status, stigma and other characteristics of groups of people as distal determinants of health that can affect multiple disease outcomes. (Mosley & Chen, 2003; Phelan, Link, &

Tehranifar, 2010) This type of analysis allows researchers to identify key environmental factors related to health inequity and to identify social constructs that preclude groups from reaching their highest attainable level of health, which, in turn, may prescribe interventions at a social- environmental level in addition to the level of the individual. (Farmer, Nizeye, Stulac, &

Keshavjee, 2006; Trickett & Beehler, 2013) Research using social-ecological models of health in conflict and post-conflict populations has largely focused on mental health. It has shown that mental and some physical health outcomes are related to daily stressors such as socio-economic status (SES), gender and food security in addition to experiencing traumatic events, as was originally thought. (Araya, Chotai, Komproe, & de Jong, 2007; Assefa, Jabarkhil, Salama, &

Spiegel, 2001; Betancourt, McBain, Newnham, & Brennan, 2013; K. de Jong et al., 2008; K. de

Jong, Kam et al., 2008; Eljedi, Mikolajczyk, Kraemer, & Laaser, 2006; B. Roberts, Felix Ocaka,

Browne, Oyok, & Sondorp, 2009; B. Roberts, Damundu, Lomoro, & Sondorp, 2010; Seino,

Takano, Mashal, Hemat, & Nakamura, 2008; Thapa & Hauff, 2012)

Distal determinants of health are captured in the model of household production of health.

Household production of health is “A dynamic behavioral process through which households combine their (internal) knowledge, resources, and behavioral norms and patterns with

available (external) technologies, services, information, and skills to restore, maintain, and

promote the health of their members.” (Berman, Kendall, & Bhattacharyya, 1994) The household model of health is intended to shift focus from disease-specific models to models that describe practices and environmental influences that result in improved health. (Berman et al., 1994) The model assumes that multiple determinants can affect singular outcomes and that interventions addressing any number of determinants could be successful. It stipulates that household health can be affected on multiple levels: individual-biological, health-producing behaviors of household members, spending household assets on health, local health providers, regional health policy and systems and the macro-economy and society. Conflict and human rights violations have effects on each of these levels.

By identifying distal determinants of health, social-ecologic models can expand the targets for health interventions to include social and environmental factors. (Betancourt et al., 2013; Haar

& Rubenstein, 2012; Thapa & Hauff, 2012; Trickett & Beehler, 2013) Researchers suggest factors that can contribute to success of health interventions in post-conflict settings are interventions that are comprehensive (Gaber & Patel, 2013); (Vervisch, Vlassenroot, &

Braeckman, 2013), equitable (Haar & Rubenstein, 2012) and mitigate factors such as displacement, gender and poor household finances that cause inequality(Bornemisza, Ranson,

Poletti, & Sondorp, 2010) and encourage governments to respect human rights. (Blas et al.,

2008; Kruk, Freedman, Anglin, & Waldman, 2010)

Figure 1. Household Production of Health

Behavioral and biomedical Health outcomes proximate determinants Macro socioeconomic system: Health care Household Exposure system: health susceptibility, Morbidity

Income Accessibility producing Resistance to Mortality Work Availability behaviors Infection Wealth Acceptability Education Quality

(adapted from Berman et al 1994)

Conflict and HRVs are Determinants of Health Conflict is a well-documented determinant of health. It is associated with increased mortality from direct violence, but it also affects distal determinants of morbidity and mortality by limiting access to healthcare, disrupting health services, depressing the economy and creating food insecurity. (Bornemisza et al., 2010; Coghlan et al., 2006; J. T. De Jong, 2010; Haar & Rubenstein,

2012; McDonnell et al., 2004; Tapp et al., 2008; Zwi & Ugalde, 1989)

Human rights violations are also determinants of health. (Mann, 2006) The International

Covenant on Economic Social and Cultural Rights defines states’ obligations for ensuring standards of physical and mental health, specifically mentioning maternal and child health and food security. (ICESCR, 1976) Failures to fulfill these obligations clearly can result in poor health.

During conflict, failures to follow Geneva Conventions regarding treatment of civilians including direct attacks, use of forced labor, looting and forced displacement have been associated with poor health (Amowitz, Reis, & Iacopino, 2002; Assefa et al., 2001; Assefa et al., 2001; Brentlinger

& Hernan, 2007; Johnson et al., 2010; Mullany et al., 2007; Orach et al., 2009; A. Potts, Myer, &

Roberts, 2011; Sollom et al., 2011; Tarantola & Mann, 1995)

Table 2. Threats to Health from HRVs Type Examples Primary Injury from assault; exposure to malaria & diarrhea pathogens from forced portering & forced labor Secondary Food insecurity due to: theft & destruction of food, inability to reach fields, loss of productivity from performing forced labor, limited intra- and inter-village trade due to restrictions on movement—economic depression Limited access to preventive and curative services due to targeting of health workers by armed groups & limits on civilian movement & government obstruction of international aid

Figure 3. Militarization, Self-Protection and Household Production of Health

Behavioral and biomedical Health outcomes proximate determinants Macro socioeconomic system: Health care Household Exposure system: health susceptibility, Morbidity

Income Accessibility producing Resistance to Mortality Work Availability behaviors Infection Wealth Acceptability Education Quality

(adapted from Berman et al 1994)

Policy level State level Household/village level

SelfWhere-Protection HRVs and May self Shield-protection Civilians affect from production Effects of of household Militarization health

Civilians are vulnerable in times of conflict, and protection of civilians has been a priority of the international community since the Geneva Conventions were written. More recently the Sphere project, the International Committee of the Red Cross (ICRC), and United Nations High

Commissioner for Refugees (UNHCR) have highlighted the need for protection of civilians in conflict. (Baines & Paddon, 2012) Despite strong international interest in civilian protection, interventions have had mixed results. (Baines & Paddon, 2012; Gorur, 2013) The UN Secretary

General’s report on protection of civilians in armed conflict called the global state of protection of civilians “abysmal.” (UN Security Council, 2012) The World Development Report cited fragmented aid mechanisms and slow security responses as reasons for limited successes in conflict prevention and recovery, and several reports have criticized the international community’s ability to protect civilians. (Baines & Paddon, 2012; Gorur, 2013; South & Harragin,

2012) Some of these criticisms point out that civilians manage risks of threat in many ways, but interventions from international agencies rarely account for this. Top-down protection activities rarely incorporate civilians’ management of their own risks or function to inform civilians how best to manage risk. (Baines & Paddon, 2012; Gorur, 2013; South & Harragin, 2012) Local strategies to manage risks associated with conflict, also called village agency, resilience or self- protection, are often overlooked in planning protection interventions, but they may play a role in improving safety.

Self-protection is any action that civilians in conflict areas do to reduce risk of attack or exploitation by armed groups. (Baines & Paddon, 2012; Gorur, 2013) It encompasses a wide range of activities generally falling into these categories: avoidance, compliance with demands and confronting the armed group. (Baines & Paddon, 2012; Gorur, 2013; South & Harragin,

2012) One classic example of avoidance is the “night commuters” in northern Uganda—children

who walked from their villages to urban centers every night to sleep where they would have protection from kidnapping by the Lord’s Resistance Army. (Baines & Paddon, 2012) Other examples from Sudan, Zimbabwe, Colombia, Sri Lanka, Uganda and Burma include creating armed home defense groups, using early warning systems for approaching troops, submitting to armed groups demands, hiding food and valuables, establishing local systems justice and rule of law, using traditional medicines when health systems collapsed and negotiating reduction in demands from armed groups. (Baines & Paddon, 2012; Gorur, 2013; South & Harragin, 2012)

There are limits to self-protection and in research there is a risk of romanticizing the capacity of communities to defend themselves. In Burma, recent reports suggest that the demands placed on villages by the military are beginning to exceed villages’ abilities to cope. (KHRG, 2010)

Recent widespread atrocities in Syria, Libya, South Sudan and elsewhere also suggest that there are limits on what communities can do in the face of overwhelming violence. These examples highlight the need for collaboration between international actors whose goal is to protect civilians and the civilians who are already engaged in self-protection.

Self-protection activities, whether they are preventing injuries from violence or contributing in other ways to reducing the consequences of human rights violations, are means of producing household health. (Berman et al., 1994; Christensen, 2004; Harkness & Super, 1994; Schumann

& Mosley, 1994) In addition to prevention, self-protection may have other positive effects such as reducing perceived stigma of persecuted populations and increasing self-efficacy—both of which contribute to the overall dignity and mental health of the persecuted population and indirectly influence physical health. In this sense, self-protection activities in themselves,

regardless of actions of armed groups, may have benefit to villages that use them. (HRW, 2012;

Sapolsky, 2004; Stringhini et al., 2012; Tol, Song, & Jordans, 2013)

A Social Ecological Model for Post-Conflict Karen State In order to gain a more complete picture of factors that influence health, we applied a social- ecological model to examine the relationship between health and human rights in post-conflict

Karen state. Previous studies in this region have shown that households that experience human rights violations have an increased risk of poor health outcomes. (Lee, Mullany, Richards, Kuiper,

Maung, & Beyrer, 2006a; Mullany et al., 2007; Teela et al., 2009) Because the household in any village is not an isolated unit--burdens are shared through economic and social relationships with other households in the village--in this paper we expand the household health and human rights paradigm to the village level and include additional risk and risk-mitigation factors for health outcomes. Our model examines determinants of health at several levels: household and village activities and state, regional and global policy. We highlight what produces health, threats to health from militarization, and protections from these threats, incorporating the

AAAQ framework to assess the rights to health. Six decades of conflict recently ended in Karen state and the international community has begun making development plans for the region, and a model of how health is produced is helpful for planning interventions.

Over the course of the Karen conflict, human rights violations had effects on health: direct injury, loss of food supplies, inability to produce food, loss of capital used to produce health.

(Berman et al., 1994; Mullany et al., 2007; Sollom et al., 2011) Recent studies have suggested that militarization, or the presence of the military even in the absence of conflict, results in human rights violations. (KHRG, 2008; KHRG, 2010; South et al., 2010) Burmese army policies in

which it supplies itself with food and labor from the local population, in addition to counterinsurgency tactics could have health impacts on the population.

In six decades of conflict, and in the absence of any major international intervention, villages evolved mechanisms to protect themselves against human rights violations. This self-protection, applied at household and village levels, includes using early-warning systems to identify approaching hostile troops and fleeing the village temporarily or permanently, managing threats by negotiating a reduction in the amount of forced labor or food demanded, lying to officials about village assets or delaying compliance of demands, refusing to comply with demands or using homemade landmines or Gher der -- “home guard” militia --to protect the village. (KHRG,

2008; KHRG, 2010; South et al., 2010) A qualitative study concluded that communities in Karen state are able to “reduce the economic, social, and humanitarian costs of military rule” with village self-protection. (KHRG, 2008)

The power relationships in Karen state are conducive to self-protection activities. (Finch, 2013)

Corruption and class differences within armed groups, variations among individual officers’ personalities, soldiers’ empathy for villagers and the presence of other armed groups have created a space for community self-protection responses to develop. (Callahan, 2007; Jordt,

2007; KHRG, 2008; KHRG, 2010; Selth, 2002) For example, military commanders strike a balance between demanding enough food and labor to support their commands but not demanding so much as to have villagers refuse or to invite a reprisal attack from an opposition group.

Our social-ecological model of health in Karen state suggests that the presence of armed groups could be a determinant of health, that human rights violations could affect health on both

household and village levels, and that self-protection also could work at both levels. A social- ecological model is useful for understanding the determinants of health in Karen state so that proper interventions can be designed and implemented. We used data from a cross-sectional survey to examine the relationships between these factors, as depicted in figures 4, 5 and 6.

Figure 4. Village-Level Exposures that may be Distal Determinants of Household Health

Figure 5. Village Self-Protection may Modify the Effects of Militarization

Figure 6. Interactions Examined by this Analysis

In this model, we examine production of health at household and community levels, as well as factors that promote health at state, regional and global levels. At the same levels, we also examine how militarization and human rights violations threaten health and how self-protection can reduce health risks associated with conflict and HRVs.

Table 3. Health, Militarization and Self-Protection in Karen State: a Social-Ecological Model Levels Indicators measured by What produces health Threats to health specific Protection from threats us or known to us to Karen state (in context to health of HRVs/ militarization) Household HHH, MAHFP, diarrhea, Health-seeking behaviors HRVs: limits on movement, HH agency self-protection, HRVs, HH economy theft of food, -caching food, flight, distance to clinic, IDP Food security/ production displacement, forced negotiation, refusal to numbers self-protection See models labor, loss of land, comply with demands, of HH health: (Berman et extortion, mental health knowledge of locations al., 1994; Christensen, 2004; outcomes and movements of Mosley & Chen, 2003) armed groups and locations of land mines Community HRVs, self-protection, Water supply and HRVs in neighbors: Self-protection-groups of /village distance to clinic, sanitation, local pharmacy, depress village economy, HHs or village leader, distance to army base, agriculture, economy/trade, reduce resources (labor, caching food, flight, exposure to armed extended family and food) that can be shared negotiation, refusal to groups informal village support or borrowed, mental comply with demands, networks health outcomes village militia and village- deployed land mines Policy- Health and poverty CBOs- preventive and Burmese army sets limits Exiled health Karen state indicators-Karen curative interventions; on movement for CBOs, departments and health compared with Burma MoH- preventive and MoH and civilians; CBOs CBOs; armed groups (as curative interventions; policy/ ability to operate; protection) regional economy; regional lack of rule of law, lack of infrastructure land laws; armed groups; military-run industries MoH has difficulty reaching all areas of Karen- insecurity and lack of resources Policy-Thai Refugee numbers, policy Cross-border relief and Refoulement Thai-Burma-Karen border regarding return and development aid (Partners, government negotiations region resettlement TBBC); refugee camps & politics, refugee camps Policy- int’l Flow of aid money International aid (for MTC, Loss of funding for CBOs, International pressure on BPHWT, CIDKP, MoH-Karen) MoH armed groups, human rights documentation and advocacy, *In addition to threats to health common for households in non-conflict areas

Production of health in Karen state at the household and community levels and through policy at regional and global levels has been discussed elsewhere. ((Berman et al., 1994; Mullany et al.,

2010; TBC, ; USAID, )

Direct threats to health and threats to mechanisms that produce health from militarization occur at all levels of the model. At the household level there are several mechanisms through which human rights violations affect production of health: injury and death, consuming time that would otherwise be spent working for the benefit of the household, and limiting movement and theft of food and possessions (including bribes and extortion). These HRVs impact the household’s ability to produce of buy food; this may result in malnutrition, which increases risk of morbidity and mortality. (Caulfield, de Onis, Blossner, & Black, 2004; SCRIMSHAW, TAYLOR, &

GORDON, 1959) By impeding a household’s ability to gain income through work and by limiting access to trade with other villages, HRVs also stifle village economy. This limits a household’s ability to tolerate catastrophic health events and also to maintain health on a regular basis.

Limits on movement also impede a household’s ability to travel to seek clinical care.

At the village level, threats to household health come from HRVs experienced by other households in the village. HRVs in any household in the village serve to depress the village economy and affect health through the mechanisms discussed above. Households in a village that directly experience HRVs are less likely to have surplus food or wealth (through mechanisms mentioned above) to lend or sell to other households that need it. Moreover, they are more likely to borrow or beg for assets from households that have not experienced HRVs.

The overall village surplus of food and wealth is diminished, and this weakens village welfare and safety nets. Malnutrition and diarrhea in households that have experienced HRVs increases susceptibility of disease for members of that household; this may create reservoirs for infectious disease in the village-- once sick they may infect members of neighbors households.

Policy in Karen state can be a threat to household health. Armed groups limit movement of community-based organizations that deliver healthcare, they limit the ability of the Ministry of

Health to operate, and they limit civilians’ movements. This severely restricts accessibility and affordability of healthcare.

Policy at the regional level affects the refugee situation and the ability of Karen people to have a place to go if the situation inside Burma becomes intolerable. Since 2010, there has been pressure on the UNHCR from the Thai and Burmese governments, as well as several other governments to close refugee camps and repatriate refugees. These actions limit options for

Karen people who want to flee, and it may force them to stay in Karen state where they may have greater risk of poor health outcomes.

At the international level, funding for Karen CBOs and the Burmese Ministry of Health affects household health in Karen state. Since 2010 bilateral donors reduced funding to CBOs working across the Thai-Burma border and began sending it to groups based in Rangoon. Groups in

Rangoon have not been able to deliver healthcare in Karen state and CBOs have cut health programs in response to the loss of funding. Households in Karen state have been left with a weakened health system.

Interventions at all levels can prevent or counter the negative health consequences of militarization. Households may hide food in the jungle so it cannot be stolen, they may flee armed groups, or pay bribes to reduce labor demanded, negotiate reductions in labor or food demanded. Knowledge of which armed groups are operating locally, how best to interact with their leaders and where and when they tend to patrol and place landmines can all help to

prevent or reduce impacts of HRVs. Mental health resilience mechanisms such as family and community structures and religion also play a role here.

At the village level, actions by the village leader may negotiate to reduce forced labor or goods demanded by armed groups. This may adversely affect stronger households, though; when armed groups demand food or labor through a village leader, the leader sometimes asks wealthy households to households with more members to donate more than vulnerable households. (KHRG, 2010)

At the Karen state level, self-protection activities include the formation of community-based groups that deliver healthcare, such as BPHWT, KHRG, CIDKP and others. (BPHWT, ; Karen News,

; KDHW, 2012) Regional policy that can improve health includes regional aid groups working cross-border to help civilians and regional human rights organizations. (FBR, ; KHRG, ; KWO, ;

Partners, ; TBC, )

At the international level, human rights organizations, international pressure and international aid can help to affect Burmese government policy that improves health of Karen civilians.

Hypotheses of relationships between militarization, human rights violations and health are tested in Chapter 6/ Paper 3 in this dissertation.

Table 4. Recommendations for Interventions Levels Possible interventions Household Food security, preventive and curative services Community/village Trainings for self-protection Policy-Karen state Increased MoH presence and cooperation with CBOs; rule of law and land rights; military policies must change; human rights monitoring; infrastructure development Policy-Thai border region Respect int’l humanitarian law for refugees Policy- international Support for CBOs, MoH, ensure that funding mechanisms—via Rangoon or border, get assistance to where it is needed

Chapter 3. (Paper1) Methods for Cluster Surveys in Conflict Areas: A Review of the Literature

Abstract:

Background: Cluster surveys are used to collect population data in humanitarian emergencies. In

the last decade many improvements have been suggested for methods for these surveys, but

implementers of surveys in conflict areas sometimes make trade-offs between best practices

and minimizing security risks for field researchers. This review highlights the evolution of

methods of cluster sampling and summarizes cases in the peer-reviewed literature when

security precautions may preclude following best practices for sampling.

Methods: We identified 34 per-reviewed papers published between 2002 and 2012 that used cluster sampling for surveys in conflict areas. We abstracted data and commentaries on sources of population data, cluster design, design effect, household sampling security protocols, quality control, training, and assessments of confounding.

Results: Researchers doing cluster surveys in conflict areas must sometimes alter methods to ensure safety and security of field surveyors. Common alterations are skipping and replacing clusters that are insecure, making fewer attempts to revisit empty households, using fewer field supervisors and making contextual decisions on how best to allocate households to clusters.

Several common security protocols are identified.

Background:

Armed conflict has severe consequences for public health: direct casualties from violence, disruption of clinical and preventive health services, blocked access to clean water, disruption of

the economy, food supply and government services. (Coghlan et al., 2006; Coupland, 2007;

Degomme & Guha-Sapir, 2010; Haar & Rubenstein, 2012; Murray, King, Lopez, Tomijima, &

Krug, 2002; Pedersen, 2002) NGOs in conflict areas are regularly use survey data to inform programmatic decisions and to measure the success of their interventions.

In addition to planning and monitoring program activities, public health data from conflict areas is also used to protect civilians and to promote justice. (F. Checchi & Roberts, 2008) In 2011

ICRC said that two major challenges the organization faced were protecting internally displaced persons and adherence to international humanitarian law. (ICRC, 2011) Population data from conflicts has been used to monitor violations of IHL, prosecute perpetrators and inform transitional justice efforts. (ICTJ, ; Pham, Vinck, & Weinstein, 2010) Epidemiological data on health and IHL violations was used to inform peace accords in Central America in the 1980s

(Brentlinger, 1996; Brentlinger, Hernan, Hernandez-Diaz, Azaroff, & McCall, 1999; Brentlinger &

Hernan, 2007; Garfield, 1989; Pedersen, 2002; Sabin, Lopes Cardozo, Nackerud, Kaiser, &

Varese, 2003; Siegel, Baron, & Epstein, 1985; Zwi & Ugalde, 1989), for advocacy for protection of civilians in Iraq and DRC (Burnham, Lafta, Doocy, & Roberts, 2006; L. Roberts, 2000), as evidence in ICC proceedings on rape in Sierra Leone (Amowitz et al., 2002) and war crimes in Kosovo

(Iacopino et al., 2001; Spiegel & Salama, 2000), to advocate for women’s rights in conflict

(Amowitz et al., 2002; Amowitz, Reis, & Iacopino, 2002; Johnson et al., 2010; A. Potts et al.,

2011), and to call for a Commission of Inquiry into Crimes Against Humanity in Burma (Sollom et al., 2011).

Cluster surveys are the preferred method for gathering population data in humanitarian emergencies, including in conflict situations, because they can be done under logistical and time

constraints with limited population data. (F. Checchi & Roberts, 2008) Cluster sampling involves selecting a cluster, usually a village, from a list of clusters with known population size, and then sampling units, usually households, within that cluster. Cluster sampling has been common practice for assessments in humanitarian emergencies since the 1970s, but only recently have significant changes been made to improve the sampling design.

Much research has been conducted on the design of cluster surveys and how best to balance logistical costs while sampling enough clusters to keep standard errors low. In the late 1970s the

World Health Program’s (WHO) Expanded Program on Immunization (EPI) assessment protocol recommended sampling seven children in each of 30 clusters to assess immunization coverage.

(Henderson & Sundaresan, 1982; Lemeshow et al., 1985) Researchers found this sampling scheme was not able to measure other outcomes with sufficient precision, and WHO later recommended a 30 x 30 design, specifically aimed at measuring acute malnutrition in children.

(de Ville de Goyet, C, Seaman, & Geijer, 1978; United Nations Administrative Committee on

Coordination – Sub-Committee on Nutrition, 1995) The 30 x 30 sampling design was then applied to a broad range of health outcomes, but this design was criticized because different situations may have different prevalence and distribution of outcomes and may require different sampling schemes. (Salama, Spiegel, Talley, & Waldman, 2004) Methods evolved and researchers have since recommended modifications to the 30 x 30 design that would maintain statistical precision in some circumstances while keeping logistical costs low. (Bilukha, 2008;

Coghlan et al., 2006; Lee, Mullany, Richards, Kuiper, Maung, & Beyrer, 2006b; Mullany et al.,

2007; Spiegel & Robinson, 2010)

Because units within clusters tend to be similar, the grouping of sampling units caused by cluster design, as compared to simple random sampling, can result in higher standard errors and a smaller effective sample size; cluster sampling requires greater sample sizes than simple random sampling to attain similar standard errors. The effect on the standard error due to the clustered design is called design effect; it is the ratio of the variance using cluster sampling to the variance using simple random sampling. Outcomes such as human rights violations or deaths due to violence are not evenly distributed through a population, and when households sampled are clustered, standard errors for these outcomes can increase drastically. Kaiser et al suggest that for child malnutrition, design effects are usually around 1.5, for crude mortality due to violence they are often greater than 4. (Kaiser, Woodruff, Bilukha, Spiegel, & Salama, 2006) They and other groups recommend literature reviews for design effects before performing sample size calculations for specific outcomes in a given setting. (Kaiser et al., 2006; Working Group for

Mortality Estimation in Emergencies, 2007)

The way in which the total sample size is broken into clusters and units per cluster is another key consideration in cluster surveys: the number of households per cluster is proportional to standard error but inversely proportional to survey costs. (Bilukha, 2008; Spagat, 2010; Spiegel

& Robinson, 2010) That is, using fewer clusters with more households per cluster reduces travel time and costs, which reduces the surveyors’ exposure to the conflict and thus risk. But using larger numbers of households per cluster and fewer clusters results in increased intra-cluster correlation, resulting in under-estimates of the true variation in the population. In order to compensate for this under-estimate, an increase in sample size is required. (Kaiser et al., 2006)

Selection of households within clusters is another area in which protocols have evolved over time. The original WHO EPI guidelines recommended starting at the center of the cluster, selecting a random compass direction and selecting houses at random along that direction. This method may result in households in the center of the village having increased probability of being selected, and subjective interviewer judgment of “proximity” in selecting households may introduce bias. (Grais, Rose, & Guthmann, 2007) Though the WHO EPI guidelines were later updated, this method is still used because it avoids the burden of enumerating every household in the sample. (Turner, Magnani, & Shuaib, 1996; WFP, 2005) Using GPS, satellite maps and sampling grids to enumerate and then randomly select houses have also been suggested as improvements on the EPI method. (Grais et al., 2007; Kaiser, Spiegel, Henderson, & Gerber,

2003)

Clusters are sometimes inaccessible due to conflict or destroyed or blocked roads. In anticipation of this, some researchers inflate the sample size and others instruct surveyors to choose the next closest cluster to replace skipped clusters. Both methods have drawbacks: replacement may result in a non-random selection of clusters, inflation may result in an inadequate number of units sampled and data collected may require weighting if many areas were skipped. (Grais et al., 2007; Spagat, 2010)

In reviews of cluster sampling to estimate mortality, malnutrition and HIV prevalence, researchers identified several common errors in methods: sample size calculations were not done, the number of clusters was insufficient, sampling was not done with probability proportional to the size of the clusters, post sampling weighting was not done, and design effect was ignored in sample size calculation, sampling was sometimes not random, response rates

were not reported, strategies to ensure data accuracy not mentioned, interpretation of data was limited, and sites were not revisited to confirm that they were indeed sampled. (Cairns et al., 2009; F. Checchi & Roberts, 2008; Grais et al., 2007; Mills et al., 2008; Spagat, 2010; Spiegel,

2007) In conflict areas, compromises with scientific rigor may be inevitable (Armenian, 1989)

Several best practices papers and manuals on cluster surveys have been published. (Asher,

2013; Cairns et al., 2009; F. Checchi & Roberts, 2005; F. Checchi & Roberts, 2008; Grais et al.,

2006; MSF, 2006; National Research Council, 2007; Purdin, Spiegel, Mack, & Millen, 2009; Rose,

Grais, Coulombier, & Ritter, 2006; Salama et al., 2004; Sanderson, Tatt, & Higgins, 2007; SMART,

; WFP, 2005; WHO, 2000) Challenges remain, however, in implementing surveys, especially in conflict areas.

The chaos that is a result of conflict—manifested as insecurity and violence, migration of people and a lack of physical and governmental infrastructure--presents unique challenges to scientists conducting cluster surveys in conflict areas. Surveys teams working in active conflict areas encounter problems in addition to balancing logistical costs and precision: population denominators are difficult or impossible to determine because of displacements of people of lack of survey data, random selection of households in the field is difficult, security protocols may interrupt or change the sampling scheme, quality control is difficult if movements and communication are restricted, armed groups may not want information on sensitive topics or war crimes exposed, and uneven distributions of violence can affect standard errors. (Armenian,

1989; Bostoen et al., 2007; F. Checchi & Roberts, 2008; Degomme & Guha-Sapir, 2007; Ford,

Mills, Zachariah, & Upshur, 2009; Morris & Nguyen, 2008; Ratnayake, Degomme, & Guha-Sapir,

2009; B. Roberts et al., 2010; Salama & Roberts, 2005; Spiegel & Robinson, 2010) Researchers

who measure and report associations between human rights abuses and other outcomes must address confounding variables of these associations.

Recommendations for best practices for cluster surveys in conflict areas are frequently tempered with caveats regarding logistic feasibility. For example, village mapping is a better way to randomly select households than EPI, but surveyors often only have time for using the EPI method. (Turner et al., 1996) A similar balance must be found between logistic costs and determining how to break down the sample size into clusters and households without significantly increasing variance.

Although much work has been done on optimization of methods for cluster sampling in humanitarian emergencies, reviews specifically on cluster sampling in conflict areas are scant.

The purpose of this literature review is to identify modifications made to cluster surveys in conflict areas and to assess how researchers strike balances between best practices and logistic feasibility.

Methods:

Search strategy and selection criteria

We searched Pub Med, SCOPUS, Global Health and EMBASE databases for peer-reviewed

English-language studies published between October 2003 and October 2013. Prior research suggested that peer-reviewed papers would have the best quality reporting of methods.

(Degomme & Guha-Sapir, 2007; Spiegel, 2007) We limited the search to the last ten years only to reflect state-of-the-art research on methods for cluster sampling.

We applied three search domains; the first domain covered permutations of cluster sampling:

“neighborhood method” [all fields] OR “neighbourhood method” [all fields] OR “sisterhood method” [all fields] OR “cluster.”

The second domain included all countries in which conflict occurred since 2003, derived from

the Peace Research Institute Oslo (PRIO) database. (PRIO, ) The final domain included search

terms to identify conflict and insecurity:

( “war” [mesh] OR “war crime” OR “war affected” OR "complex emergency" [all fields] OR

“humanitarian emergency” [all fields] OR “human rights” [mesh] OR "Human Rights

Abuses/statistics and numerical data"[Mesh] OR “Human Rights” [all fields] OR “altruism”

[mesh] OR "Warfare"[all fields] OR "wars"[all fields] OR "fighting"[all fields] OR "fight"[all fields]

OR "rebellion"[all fields] OR "rebel"[all fields] OR "insurgency"[all fields] OR "insurgent"[all

fields] OR "armed forces"[all fields] OR "armed force"[all fields] OR "armed group"[all fields] OR

"armed groups"[all fields] OR "militia"[all fields] OR "militias"[all fields] OR "revolution"[all

fields] OR "revolutions"[all fields] OR "uprising"[all fields] OR "uprisings"[all fields] OR "riot"[all

fields] OR "riots"[all fields] OR “insecure” [all fields] OR “insecurity” [all fields] or OR “civilians”

[all fields] OR “civilian” [all fields] OR “conflict” [all fields] OR “humanitarian” [all fields] OR

“relief work” [all fields] OR “disaster medicine” [all fields] OR “conflict situation” [all fields] OR

“armed conflict” [all fields] OR “conflict affected” [all fields])

Inclusion criteria were: peer-reviewed original or review papers, with any outcome, published in

English that used cluster sampling design in conflict or post conflict areas. Subjects were human

and any health outcome was included.

Papers were excluded if the topics were foreign soldiers on conflict areas (US soldiers in

Afghanistan, for example), cross sectional studies that did not use cluster sampling, commentaries and opinion papers, grey literature, papers that did not discuss insecurity or violence, papers that were researched during times of peace, secondary analyses of data collected during conflict, and qualitative surveys.

Papers were screened using a three-stage process: first, titles were screened and any paper that would potentially fit the inclusion criteria was included. Duplicates were eliminated at this step.

Next, abstracts were screened independently by two people, and discrepancies were included for the third stage of screening. Finally remaining articles were read completely and data were abstracted from articles that fit the inclusion criteria. We abstracted data and rationale or commentary on sources of population data, cluster design, design effect, household sampling security protocols, quality control, training, and assessments of confounding.

Results

The selected papers included surveys from 11 different countries. The most common outcomes measured were mortality, PTSD, violence, mental health, access to healthcare and health systems/ humanitarian aid performance and infectious disease prevalence and control efforts.

Five papers directly measured human rights violations and presented the results as such; several others measured human rights violations implicitly, such as excess mortality, violations of

Geneva Conventions, or measuring quality of humanitarian aid. Some of these presented

findings in the context of violations of international law (genocide or Geneva Conventions).

Researchers obtained population data from a range of sources: governments, UN agencies, non- governmental humanitarian organizations and community-based service delivery organizations.

When data was not available, researchers consulted with local political and religious leaders, did area mapping and shelter or women counts, used satellite imagery to estimate populations, or used a combination of these techniques with triangulation.

Cluster designs varied; only 3 out of 34 surveys used the EPI 30 x 30 design. Reasons given for using alternate designs were to accommodate increased sample size, to reduce variance, and to limit risk of lost data. Design effects used for sample size calculations ranged from 1.25 to 5.3; however, in 13 surveys design effects were 2. Design effects greater than 2 were used for surveys for trauma, mortality, trachoma, and human rights violations. Design effects of 2 were used for surveys that measured violence, mental health, rape, human rights violations and access to health services.

Twelve surveys used EPI or modified EPI methods to select households. Five selected households randomly from a table (usually generated by local religious leaders), nine surveys selected a random starting point (using GPS, city households numbering systems or mapping), then did proximity or systematic sampling, seven surveys used systematic sampling. In some cases surveys used multiple methods, depending on the layout of households in a specific cluster.

Twenty-one papers made some mention of security protocols in the sampling scheme. Survey designs protected surveyor safety by excluding unsafe areas from the sample frame or sampling everywhere and allowing surveyors to determine if clusters were safe to approach or not. One

study paired clusters to minimize travel time and reduce risk to surveyors. Two studies did not use GPS because possessing GPS units increased risk for surveyors. Two survey protocols specifically stated that the survey design minimized surveyors’ time spent in the field; their surveyors did not revisit empty households, used shorter questionnaires, and used few or no field supervisors.

Protocols in six surveys called for surveyors to skip clusters that they determined were insecure and replace them with the nearest safe cluster, three studies excluded unsafe areas from the sampling frame, eight studies skipped unsafe clusters and did not resample, and five studies oversampled in anticipation of inaccessible areas and weighted data post-sampling to account for the skipped clusters.

Training times for surveyors ranged from two days to three weeks, and surveyor qualifications varied from community health workers to medical doctors and professional surveyors. In nearly all cases surveyors were from local populations.

Fifteen surveys reported statistical associations in their results. Using multiple regression or stratification, analyses were adjusted for gender, financial status, education, age of head of household, displacement, HH size, education of HH, income, damage to home (from conflict), region, demographics, forms of violence experienced, pattern of exposure to traumatic events,

PTSD, depression, area of residence, forced movement, forced labor, age and factors influencing food security.

Discussion

This review identified 34 peer-reviewed papers in which cluster sampling was done in conflict areas. Surveys tended to follow standard guidelines for cluster sampling in how to deal with skipped clusters and in determining households per cluster. Surveys deviated from guidelines in determining design effect and in protocols for household sampling. It is possible that survey protocols deviated from guidelines to limit risk for surveyors in conflict areas; risk reduction protocols due to insecurity also may have changed quality control protocols.

We identified several common themes in security protocols among the papers identified by our search. Individual surveyors tended to be from the areas they surveyed and they had decision- making authority to skip areas they felt were insecure. Several studies mentioned alterations to standard methods that would result in reduced time in the field, including modifications of cluster design, household sampling protocols, and greater acceptability of skipping households.

These alterations could affect variance of the sample and quality control.

Guidelines for cluster surveys indicate that there are no standards for cluster design and design effect; that each scenario warrants an analysis and customized survey design. This was evident in the diversity of cluster designs identified by this search: only three followed the WHO 30 x 30 design.

Design effects used for sample size calculation were not as diverse, and this may or may not represent deviation from standards. Guidelines suggest performing literature searches to help determine design effect, and they also suggest higher design effects are needed if outcomes are not evenly distributed in the population. Few papers justified the choice of design effect, but

one paper (L. Roberts, Lafta, Garfield, Khudhairi, & Burnham, 2004) stated that slightly increased variance was acceptable because wider confidence intervals would not mask the magnitude of the public health problem documented. Use of smaller design effects, and therefore lower sample sizes, may represent a tradeoff between limiting time in the field (by keeping sample size small and more household per cluster) and keeping variance at an acceptable level.

Literature on cluster sampling highlights advantages between replacing clusters that were skipped and oversampling in anticipation of reduced return rates. Most surveys identified by this search replaced skipped clusters with nearby clusters and also oversampled in anticipation of low return rate. Attempting to sample everywhere and then replacing insecure or inaccessible clusters gives a more accurate picture of the amount of inaccessible area in the sample frame. (L. Roberts et al., 2004) No papers generalized results to areas that were not sampled.

Guidelines recommend against using EPI household sampling protocols, however, we identified several papers that used the EPI method. The caveat in the guidelines is that EPI household sampling is the fastest way to do this, and it is possible that researchers decided that this trade off was acceptable in order to minimize time researchers spent in the fiend. No paper justified this explicitly.

Limitations

This literature search included ten years of surveys; however, during this time new recommendations and modifications on how to design and implement surveys were published.

Some surveys included in this literature review were done before recommendations were

published, and could not incorporate these recommendations. There is no clear trend, though, on use of EPI, justification of design effect, what best to do about skipping clusters over time.

We did not include NGO reports in the literature search. Including grey literature would have greatly increased the number of surveys in this review but other literature suggested that these surveys have a lower quality of methodology and likely uninformative.

Chapter 3 Tables and Figures

Table 5. Search Results 495 Papers identified in searches 232 Duplicated excluded 263 Abstracts screened 5 Non-English papers removed 113 Abstracts included (including 17 discrepancies) 34 Papers included

Table 6. Data From Selected Papers Source of Quality Design Survey Survey Population Cluster HH Sampling Security Skipped Control Effect and Outcome(s) Reference Location Date Data Design Methods Protocols Clusters Measures Training Rationale measured (Ahmed, Edward, & Rural Herat provincial Burnham, Province, government used random 4 days, all demographics, 2004) Afghanistan and UNICEF 30 x 10 number table nr nr nr medics Nr MCH

local highly CHW census, literate, Apr- shelter "protocol fit daily checks (Depoortere West Jun counts, area local security for data exposure to et al., 2004) Darfur 2004 mapping 30 x 30 EPI constraints" entry 2 violence, MH

54 minimized

travel by clumping clusters, assigned random clusters w no selection of regard for map security, coordinates sampled 3 then extra in case (L. Roberts Sep MoH Jan 2003 proximity skipped mortality, et al., 2004) Iraq 2004 estimates 33x30 sampling clusters, medics, Nr causes

random selection of a list from village leader; random selection of mosque, then HH from belonging to each mosque; random selection of a UNICEF, point on a Mangarhar Jan- village map and skipped (Scholte et Province, Mar leaders, proximity skipped and not PTSD, al., 2004) Afghanistan 2003 imams 40 x 10 sampling clusters replaced nr nr Nr predictors

UNICEF randomly demographics,

55 UNHCR, data selected from trauma,

(Cardozo et from list created by skipped substituted coping al., 2005) Afghanistan 2-Aug Mosques 50 x 15 local mullah insecure areas nearest nr nr 2.5 mechanisms random selection of main street, balance need then for data with UNDP/ Iraqi residential risk acceptable ministry of street, then to teams, skip pick 2 days, (Burnham Jan-Jul planning numbering clusters if nearest medics w et al., 2006) Iraq 2006 estimates 50 x 40 households insecure cluster nr survey exp Nr mortality systematic random sampling or EPI, excluded prior depending on to cluster experienced how selection, did replace w local nursing Apr- 2004 households not revisit nearest staff , some (Coghlan et Jul government 60 x 20 or were skipped cluster/ w prior al., 2006) DRC 2004 census 30 arranged households zone survey exp 4 mortality (Lee, Mullany, workshops, Richards, Eastern systematic skipped unsafe stratify field training of demographics, Kuiper, Burma 3-Mar CBOs 100x20 sampling clusters and weight supervisors trainers 2 mortality

Maung, & Beyrer, 2006a)

local sampled all authorities, women in food village during distribution concurrent (Doocy, lists, physical nutrition 2.6 Burnham, & Sudan- counts of assessment; substituted calculated- Robinson, Eritrea women or HH selection skipped due to closest trained -3- mortality 2007) border 2004 structures 30 x 30 not done security village nr interviewers effective demographics

one survey page, no weighing systematic equipment, no

56 (Mullany et Eastern see lee interval individual IMR (2) HRVs and

al., 2007) Burma 2004 CBOs 2006 sampling identifiers nr nr 4 day HRVs (5) health

demographics, (Vinck, WFP PTSD, Pham, registration multistage exposure to Stover, & Apr- data 2005 and strategy, 1 week, U violence, Weinstein, Northern May gov't census clusters by skipped cluster replaced if students or definitions of 2007) Uganda 2004 2002 PPS EPI in insecure no security nr professionals 2 peace

field supervisers, Jun- teams stopped interviewers violence and (K. de Jong Kashmir, Aug activities and from mental health et al., 2008) India 2005 nr 30 x 17 modified EPI evacuated nr university nr 2 status

field supervisors, (K. de Jong, Jun- teams stopped interviewers Kam et al., Kashmir, Aug activities and from exposure to 2008) India 2005 nr 30 x 17 modified EPI evacuated nr university nr 2 violence, MH

starting house: random selection (Hassan, from a list of Malik, Kassala households, Okoued, & District bed net then Eltayeb, Eastern distribution proximity trained 2008) Sudan 2006 list 30 x 7 sampling na nr nr personnel Na bed net use (Kim, Griffin, systematic Nadem, random Aria, & Kabul Jan- sampling from demographics, Lawry, Province, Jun 98 clusters list of field heath 2008) Afghanistan 2005 NGO data x about 30 households nr nr supervisors one week 2 knowledge

Southern Sudan guinea excluded

57 Ayod worm modified EPI villages-

(King et al., County, eradication (Turner's disarmament team-prior 2008) Sudan 6-Nov program 20x20 modifications) activities nr experience 7 days 5 trachoma

Sep (Mullany, 2006- proximity or malaria, Lee, Paw et Eastern Jan interval MUAC, al., 2008) Burma 2007 CBOs 40 x 10 sampling nr nr nr 21 days 2 anemia, HRVs (B. Roberts, Ocaka, Browne, 32 cluster traumatic Oyok, & (to reduce events, phys Sondorp, Northern design and MH 2008) Uganda 6-Nov WFP effect) x EPI nr nr nr 6 days 2 outcomes

Oromia and central entire SNNPR statistical section on (Shargie et regions agency of random walk this--field al., 2010) Ethiopia 7-Jan Ethiopia 64 x 25 method nr nr sups, nr 1.2 malaria, nets

systematic random sampling or EPI, 2006 MoH depending on experienced data, or how skipped nursing staff 5.3 and estimates households excluded and as surveyors, 3.4 (based (Coghlan et from local were insecure areas replaced w field on prev. al., 2009) DRC leaders arranged from selection nearest supervisors surveys) mortality

community leaders, UNHCR data, May- shelter malnutrition, (Guerrier et Eastern Oct counts, local mortality and al., 2009) Chad 2007 authorities 30 x 30 EPI nr nr nr 3 days 2 cause

58 systematic exposure to

(Pham, sampling in traumatic Vinck, & randomly U graduates "adjusted events and Stover, Northern chosen with survey included for design war crimes, 2009) Uganda 2005 WFP UNHCR nr direction nr nr experience pilot survey effect" ptsd

random skipped selection prespecified (Donaldson, using insecure Hung, established locations, Shanovich, administrative substituted Hasoon, & numbering using same MoH staff w Evans, Baghdad, 2007 Iraqi HH system for sampling prev survey traumatic 2010) Iraq 2009 SES survey 60 x 20 Baghdad methods experience 3 d 3 injury

mapped, HHs field Oct- census selected with supervisor (Jima et al., Dec enumeration PDAs with visits; see malaria and 2010) Ethiopia 2007 areas 319 x 25 GPS nr nr section nr 1.25 preventions

experienced interviewers, replaced w field previously supervisors, (Johnson et Eastern 10- skipped selected field HRVs, health al., 2010) DRC Mar OCHA 90 x 10 modified EPI villages ones observations 3 days 2 outcomes

(see text) extensive training, short time between training and survey

59 proven

(Mullany et Eastern skipped survey al., 2010) Burma 2008 CBOs 200 x 10 nr insecure areas nr instruments Nr MCH

8 days, tests and mock interviews ongoing 2009 census assessments Ministry of of inter-rater economic experienced reliability exp to (Vinck & Oct- planning and data until full violence, phys Pham, Dec international collectors/ agreement and mental 2010) CAR 2009 2009 cooperation 4 stage EPI nr nr academics on all coding 2 health

random number generation for compass direction, alternative streets to clusters pre- cross, then selected in household to case of pass before skipped demographics, (Burnham national list of starting clusters due to access to et al., 2011) Iraq nr PHCs 17 x 10 household insecurity nr nr 1 week 2 services 60

included no-travel areas in sample frame, then added these clusters to clusters not able to be reached, team could best determine fractions of population

61 missed

because of logistics or security, if skipped for logistics then substituted skipped closest dangerous village; systematic areas , weighted sampling oversampled clusters based on in anticipation with 2003 missed Central Jun - 2003 national household of more census homes 5 days with (A. Potts et African Jul census --data coutns or security data for revisited if experienced al., 2011) Republic 2009 from OCHA 30 x 10 estimation concerns clusters possible interviewers 2 rape, HRVs

skipped insecure neighborhoods and selected replacements using same sampling administrative random methodology, (Shanovich, numbering selection interviewers Donaldson, system- using had authority KAP of Hung, Baghdad Baghdad city to skip emergency Hasoon, & Oct- governorate administrative neighborhoods medical Evans, Baghdad, Nov profile inter- numbering due to MoH w prev systems in 2011) Iraq 2009 agency 60x20 system insecurity nr experience 3 d Na Baghdad

62 MoH website, Landscan Feb- software, UN- random start (Sollom et Chin state, Mar sponsored proximity al., 2011) Burma 2010 MIMU 90 x 8 sampling nr nr nr nr Nr health and HR GPS starting (Kirsch, point and Wadhwani, 80 x 20 + proximity Sauer, 10 clusters sampling or Doocy, & Jan- for GPS and Catlett, Feb comparison systematic substituted aid received, 2012) Pakistan 2011 nr analysis sampling skipped cluster later nr nr 1.5 satisfaction

randomly selected (Kirsch, intersection Leidman, and compass Weiss, & direction then Doocy, government every third previously aid received, 2012) Haiti 11-Jan and UN 60x20 shelter nr nr nr trained 1.5 satisfaction

(Chan & demographics, Kim, 2010) Pakistan 6-Feb NGOs nr nr nr nr nr nr Nr health status

Chapter 4 Research Methods

This study employed a two-stage cluster survey, implemented in January 2012, to measure the prevalence of human rights violations, barriers to health care and food insecurity among civilians in Karen state, eastern Burma.

Sampling Universe The sampling universe included adults and children living in clinical catchment areas served by BPHWT and KDHW in Karen state, eastern Burma, and also adults and children living in three townships around Tavoy town in Mon State, eastern Burma in 2010.

(See maps in appendix for the areas in Karen state that were sampled) This gave a sampling frame of about 80,000 people in 250 villages across the region. Although the Tavoy area is in

Mon state on Burmese government maps, the majority of the population is Karen, and thus it was included in the survey. The areas around Tavoy are administered by the Burmese government and have not had conflict in several years. All of the other areas sampled are administered by the KNU or are contested areas or mixed administration. Although relatively little fighting occurred in Karen state in 2010 (the year covered by the survey questionnaire), conflict was ongoing in the clinical catchment areas over the last decade. The geography in the areas sampled ranges from jungle-covered mountainous regions to coastal plains with paddy rice. The mountainous areas, especially in former conflict areas have very few roads or other infrastructure. Mobile phone networks did not exist, and telephone land lines served only the major population areas.

Sample size and cluster design We calculated the required sample size to be able to detect a prevalence of any human rights violation of 15% and a survey return rate of 82% (both estimated from previous surveys in Karen state), with Type I error of 5%, Type 2 error of 80%

and a design effect of 3.0. (Prudhon & Spiegel, 2007; Rose et al., 2006; Working Group for

Mortality Estimation in Emergencies, 2007) By this calculation, surveyors needed to approach

717 households to ensure that at least 588 households, the required sample size, responded to

the survey.

Although WHO EPI recommends a 30 x 30 design, we determined that a 90 x 8 design was most

appropriate for this survey in this setting. (de Ville de Goyet, C et al., 1978) The primary

exposure measured--human rights violations-- is the result of actions of armed groups, and

exposure to these groups and thus to human rights violations is not evenly distributed through

the population. Like mortality due to violence, human rights violations tend to be clustered. (F.

Checchi, 2010; Depoortere & Checchi, 2006; Purdin et al., 2009; Spiegel & Robinson, 2010;

Working Group for Mortality Estimation in Emergencies, 2007) This uneven distribution of

exposures results in increased variance between clusters, i.e. it inflates the intracluster

correlation coefficient and therefore design effect, and necessitates a greater sample size. (L.

Roberts et al., 2003) By increasing the number of clusters and decreasing the households per cluster we are able to decrease the sample size requirement to one that enabled a balance of logistic constraints of visiting many villages with a total sample size that was feasible to collect.

The 90 x 8 design also minimized the impact of losing entire cluster of data if a surveyor lost data forms if he or she had to flee suddenly due to insecurity.

In the first stage of sampling we selected villages. The partner organizations provided lists of villages and populations in their clinical catchment areas. If village populations were not available, the partner organizations estimated population size based on the number of houses in the village. Using these lists, we randomly selected villages by assigning probabilities of selection

proportional to village population sizes. In the second stage of sampling, surveyors selected eight houses in each village using a modified spin-the-pen technique. (Bennett, Woods,

Liyanage, & Smith, 1991; Grais et al., 2007; Sollom et al., 2011)

Questionnaire

The survey questionnaire was based on questionnaires previously used in Burma and modified for cultural appropriateness for Karen state. (Sollom et al., 2011) The questionnaire was designed to assess common human rights violations, access to healthcare and food security. It included reported exposure to perpetrators and the location of alleged abuses. If a respondent said that they had experienced a human rights violation they were asked several follow-up questions: to confirm that the event happened in the preceding year, to identify the perpetrator, to identify the family member that experienced the violation, and to affirm whether the respondent was an eyewitness or had heard about the violation from a victim or another source. Only human rights violations for which a respondent could answer these questions were included in the analysis.

The questionnaire incorporated the six-question USAID Food and Nutrition Technical Assistance

Project (FANTA) household hunger survey (HHS) and the months of adequate household food production (MAHFP) survey. (Bilinsky & Swindale, 2007a; Deitcher, Ballard, Swindale, & Coates,

2011; Sollom et al., 2011) The FANTA household hunger scale (HHS) was developed and validated to assess household hunger in different cultures so that results from the HHS could be compared across different settings. (Deitchler, Ballard, Swindale, & Coates, 2011) The HHS comprises three questions, how many times in the last four weeks have you: had no food to eat of any kind in your household, gone to sleep at night hungry, and gone a whole day and night

without eating, and four different response codes for each question—never, rarely, sometimes,

and often. The recall period is four weeks. To assess responses, the frequency categories

“never” and “rarely” are combined and assigned a value of 0. “Sometimes” responses are

assigned 1, and “often” assigned 2. The sum of the values for the three questions is calculated

for each household (it is between 0 and 6), and cut-offs are applied to determine the severity of the household hunger. Total scores of 1 and 2 signify little to no household hunger; 2 and 3 moderate household hunger; and 4, 5 and 6 severe household hunger. The percentages of households in each category are reported.

Months of Adequate Household Food Production (MAHFP) was included in the survey because it has been validated to cover a one-year time frame, which, unlike the HHS, would extend to before the most recent harvest in Karen state, and also because it is a measure of a household’s access to food, which could be affected by human rights violations. (Bilinsky & Swindale, 2007b)

The USAID definition of food security has three components: availability, access and utilization.

MAHFP measures access to food. A household’s access to food depends on its ability to grow, find or purchase it. Several different human rights violations could affect a household’s ability to meet its food needs, including theft and destruction of food, forced labor, limited access to land or travel, attacks in civilians and forced displacement.

MAHFP is measured by asking the head of household to consider each month in the preceding year and to recall if there was a time during each month in which the household did not have enough food to meet its needs. From these data we can determine the proportion of households not able to meet their food needs in any one month, or count the number of months over the last year in which each household was not able to meet it food requirements.

Surveyors measured middle-upper arm circumference (MUAC) in all children between 6 and 50

months and asked about diarrhea and night blindness in all household members. The questionnaire also asked about accessibility, affordability, availability, and quality (AAAQ) of health care in Karen state. The AAAQ framework to the right to health is described in General

Comment 14 of the Economic and Social Council, the review committee for the International

Covenant on Economic, Social, and Cultural Rights (ESCR). (CESCR, 2000; Hunt, 2006; H. Potts,

2009) AAAQ is used to assess health services and also underlying determinants of health.

We consulted community-based human rights groups about the content and the wording of the questions to ensure that we were capturing important data and that the survey participants would understand the meanings of the questions. We further refined the questionnaire with the surveyors themselves during the two-week training and field tested it in Mae La refugee camp. It was translated into Sgaw Karen and Burmese and then back-translated to English with a different translator to ensure accuracy of the translation.

If households reported a member who had night blindness or another severe health condition, surveyors referred that person to the nearest clinic.

Surveyors also asked village leaders a short questionnaire about exposure to armed groups, terrain and distance to military bases.

Surveyors and training

Security concerns and military restrictions on movement create difficulties for doing research in

Karen state, and minimizing risk to surveyors necessitated that the surveyors possess in-depth knowledge of local terrain, politics and troop movements of the areas assigned to them. To maximize the safety of the surveyors, the five community-based organizations that were a part of this study identified personnel who were willing to work as surveyors for a three-month period. These staff were already living and working inside Karen state, and the survey clusters to which they would be assigned were within their work catchment areas.

The community-based organizations committed 22 surveyors who worked in 14 different clinic catchment areas. Seven of the surveyors worked with youth groups and the remainder were community health workers. The surveyors lived and worked inside Karen state, were fluent in either Burmese or Sgaw Karen, had knowledge of the terrain, political climate, and local leaders in the area where they surveyed. They were able to do mathematical calculations and were able to travel by local means or by foot through remote areas of the State. The survey team comprised sixteen men and six women aged 20 to 38 who lived and worked in the clinical areas in the sampling frame.

The training team designed and facilitated a two-week course that was translated into Sgaw

Karen and Burmese. The training included lectures and practical sessions on all topics crucial to implementation of the survey. It began with an overview of international human rights, and a discussion on health and human rights in Karen state. Surveyors were trained to further explore answers in the quantitative questionnaire with open-ended follow-up questions. Mathematics practice, MUAC training, the importance of obtaining informed consent and a technique for selecting households to survey were also taught.

A substantial portion of the time covered the content of the survey and the intent of the

questions. We designed these sessions to ensure that the surveyors understood the questions

and that the translations were accurate. Questions were modified and re-translated during these sessions to ensure that they conveyed the research team’s meaning as clearly as possible.

Each surveyor practiced the entire survey protocol several times during a one-day practicum in

Mae La refugee camp and several more times during training sessions and for homework.

Surveyors were required to pass a final check-out test before they went to the field. One surveyor failed the test; this person was partnered with a more experienced surveyor and they implemented the survey as a team.

Security Considerations

Surveyors were responsible for assessing the security situation in a village before approaching it, and then consulting with the village leader on the safety of conducting the survey. If the surveyor determined that the village was not safe to enter, he or she would proceed to the next closest village and implement the survey there. If there was no one available to interview in a household, the surveyor returned only twice over two days before selecting another house. This minimized time spent in the village and thus minimized risk of meeting hostile armed groups.

Surveyors obtained oral consent first from village leaders and then from heads of households.

Survey participants were informed of the purpose of the survey, that they would receive no financial benefit from participating, that there could be risks associated with consenting to the interview, that their responses would remain anonymous and that the surveyor would keep their identities confidential. Participants were informed that they could refuse to answer any question and could stop the interview at any time for any reason. Interviews were conducted in

private, inside respondents’ houses and at least 25 feet away from any high traffic areas from

where passersby might eavesdrop.

Inclusion criteria

Inclusion criteria for the adult surveys required that participants be adult (15 years of age or

older if married, 18 years or older if unmarried) heads of household living in Karen state, who

spoke Burmese or Sgaw Karen, displayed sound psychological state to answer sensitive

questions, and provided informed consent to participate. For mid-upper arm circumference

(MUAC) measurements, inclusion criteria included children 6 to 59 months old who were residents of the enrolled household whose parent had provided informed consent.

Data Collection

Surveyors conducted the study during January 2012. The time period covered by the questionnaire was one year prior to the interview, with the exception of the household hunger section and health status questions. The households hunger section covered only the month prior to the interview, and the health status questions covered two weeks prior to the interview.

Before approaching a village, surveyors would assess security situation by looking for the presence of armed groups. Surveyors would then approach the village leader, introduce themselves, explain the purpose of the survey, and obtain informed consent from the village leader. If the village leader consented to the interview, the surveyor then conducted a short interview with the leader about exposure to the Burma army and access to health case. The surveyor next approached the households, identified the head of household or another adult

(over age 18, over age 15 if married), obtained the same consent and began the interview.

After the informed consent and interview with the village leader, the surveyor and the village leader made a map of the village and determined the center. The surveyor then selected a random direction by spinning a pen (Grais et al., 2007) on the ground and marking the direction of the point. He or she then walked in that direction in a straight line from the center to the end of the village and numbered each house that passed within ten feet of either side of the path.

The surveyor then selected a house a random by writing all of the house numbers on pieces of paper and pulling one out of a hat. The surveyor then consulted the map and selected the seven houses closest to that house. He or she approached initial house and the seven closest houses and asked to speak with the head of household. If the head was not at home, then the surveyor asked to speak with any adult over the age of 18 if they were unmarried or over the age of 15 if they were married. If none of these people were home, the surveyor proceeded to the other houses and returned twice to the empty house. If still no-one was home, the surveyor selected the closest house that had not been interviewed and continued with the protocol at that house.

Surveyors first asked about the composition of the household; anyone who ate meals at the house for the last two months was considered a household member. MUAC of children six months to five years old was measured. If children were not at home, the surveyor asked someone else to find the children and bring them home for the measurement. Surveyors were given one month of survey time and one month of travel time to complete the surveys. Each surveyor had an average of four clusters to visit. Neither the village leaders nor heads of household received compensation for participating in the survey.

Surveyor Debriefing

Surveyors returned to Thailand after collecting the data and met with the project director to check survey forms and debrief. No security incidents occurred, but surveyors skipped ten villages out of 90 because of the presence of Burmese army or BGF troops. Surveyors reported that respondents had no problems understanding the questions.

Quality Assurance

Security concerns precluded several quality assurance steps that are normally taken in the field.

Surveyors were not able to visit villages a second time to repeat the survey, there were no field supervisors to check data as it was being collected and to oversee the sampling process, and surveyors had no communications devices to call with questions or problems with the questionnaire or protocol. We addressed some of these potential problems by holding a two-

week long training that included extensive practical experience under close supervision of

instructors. We also set high standards for the final check-out and did not pass trainees who were not able to select households or conduct the interview properly. Surveyors reported no technical problems or confusion about the survey questions at the debriefings.

Data Entry

Two people entered the survey data separately into two identical Microsoft Access databases.

The databases were designed to minimize errors: they only accepted answers to each question that were in the numerical range expected for that question. We compared the databases with

Dataweighter® software and resolved discrepancies by referring the original survey forms.

Statistical Analysis

Before beginning the analysis we tested for clustering at the household level. STATA 13 svy suite

of commands applies Taylor linearization and accounts for clustering of the survey data at the

cluster level (clustering of households); however, it does not account for clustering at the

household level (clustering of outcomes between individuals in the same household). The outcomes measured on the individual level—MUAC, diarrhea and night blindness—could very likely be clustered. A comparison of Akaike information criteria (AIC) and variance between two multilevel models that accounted for clustering at household or village levels and a model that did not account for any clustering indicated that clustering was not present at the household level, but that it was present at the village level. STATA 13 svy commands are able to account for village level clustering, and mutlilevel models were not used in the final analysis.

The goal of the data analysis was to identify associations between human rights violations and health outcomes. Because villages were skipped for security reasons, and others were substituted, data was weighted at either the clinical area level or the village level by population before analyses were performed.

Survey coverage, participation rates and demographics were estimated. Next, prevalence of human rights violations and alleged perpetrators were estimated, and proportions and 95% confidence intervals were calculated. The prevalence of health outcomes, including diarrhea, under 5 malnutrition, night blindness, household hunger and months of adequate household food production were estimated with percentages and 95% CIs.

Multiple Logistic Regression

We used generalized linear models with log link functions to estimate prevalence rate ratios for

human rights violations and health outcomes, using Poisson regression when models failed to converge. Household hunger was coded into moderate/severe and none/mild categories; MUAC scores for children under 5 were separated into moderate/severe (<12.5mm) and none/mild

(>12.4mm), (based on WHO criteria); and diarrhea and night blindness were recoded as binary variables (present/ not present). We coded MAHFP into a binary variable using 9 months as a cutoff. The distribution of MAHFP peaked at 8 months in households that had experienced

HRVs, and at 10 months for households that had not, and this cutoff best captured the differences between the two populations.

If necessary, adjustments were made for variables known to be confounders or effect modifiers in other surveys in the area: household size, type of water supply, clinical catchment area, religion, topography, female-headed households and exposure to any other human rights violations. (Lee, Mullany, Richards, Kuiper, Maung, & Beyrer, 2006a; Mullany et al., 2007; Sollom et al., 2011) Only adjusted relative risks are reported.

Examination of variance inflation factors (VIF) between variables was used to check for multicollinearity; using a cut-off VIF of 2.5, no multicollinearity was detected. (Allison, ) Only adjusted risk ratios are reported.

Multiple Imputation

About 7% of data for diarrhea and night blindness were missing. We assumed that the data was missing at random (MAR) because logistic regressions suggested that missing data was dependent on observed data and used multiple imputation to estimate values for missing

variables (diarrhea, night blindness, human rights violations). (Allison, 2009) We used multiple imputation to estimate the diarrhea, MAHFP night blindness, household hunger, ‘household member sick not able to access treatment’, and ‘left Karen state for medical treatment’ from household size, human rights violations, village and surveyor. Twenty imputations of data were performed and the mi estimate set of commands, along with svy commands, in STATA 13 was used to analyze the imputed data. We ran multiple logistic regressions first using original, cleaned data and second with imputed variables for all missing data. A comparison of frequency tables between imputed and original values indicated that the proportion imputed values for binary outcomes were consistent with those from the original data set. Results from the two analyses were similar, and only results from the original data set (not imputed) are reported here.

About 15% of data was missing for each of the self-protection activity questions and for distance to base. This proportion was much higher than proportions of missing data for the other survey questions. Analysis of the missing data indicated that the missingness was intermittent, (Yang &

Shoptaw, 2005) and that most of it was due to respondents’ refusals to respond, likely because of the sensitive nature of the questions. We decided that listwise deletion would be the most conservative way to handle the missing data.

Analysis—Paper2

Survey coverage, participation rates, demographics were estimated, prevalence of health outcomes, human rights violations and alleged perpetrators were estimated, and proportions and 95% confidence intervals were calculated.

We used generalized linear models with log link functions (described above) to estimate risk ratios for human rights violations and health outcomes, using Poisson regression when models failed to converge.

Analysis—Paper 3

We first investigated determinants household health. We estimated the risk of adult and child diarrhea and risk of high household hunger or few MAHFP given proximity to the nearest military base in hours hiking. We next estimated risks of the same health outcomes given exposure to types of armed groups: any armed group, the total number of armed groups to which a household was exposed, the Burmese army (BA), or the Burmese army and allied Border

Guard Forces (BABGF).

Next we investigated the relationship between HRVs in the village and poor health outcomes in the household. We estimated the risk of poor health given that at least one household in the village experienced a HRV. Next, in a subpopulation of households that did not report HRVs, we estimated the risk of poor health if other households in the village reported HRVs. We performed these analyses twice, treating village HRVs as a binary outcome (none or any) and also as a continuous one (zero to eight—eight households were interviewed in each village).

Finally we investigated the effects of self-protection on HRVs and health. For each village sampled we combined the reported household self-protection and generated binary self- protection variables (any or none for any self-protection, any or none for negotiation) and continuous self-protection variables (0 to 32 for any self-protection and 0 to 8 for negotiation).

We first examined associations between village self-protection (as a continuous variable) and

household exposure to armed groups, household HRVs and household health outcomes. Next

we stratified self-protection by any or none and examined associations between exposure to armed groups and HRVs and exposure to armed groups and health.

Ethical Approval

The PHR Ethical Review Board, the Institutional Review Board at the Johns Hopkins Bloomberg

School of Public Health and a Karen community advisory team reviewed and approved the research plan.

Chapter 5. (Paper2) Health and Human Rights in Post-Conflict Karen State, Eastern Burma Abstract

Background

Eastern Burma has experienced six decades of conflict, resulting in high prevalence of human rights violations and worse health outcomes than in other areas of Burma. Recent ceasefire agreements resulted in a suspension of conflict in this area, but the threat of human rights violations remains. This study examines health and human rights in eastern Burma during a time when conflict was nearly absent.

Methods and Findings

We conducted a two-stage cluster survey of 686 households in eastern Burma in January 2012

that covered health status, access to healthcare, food security, exposure to human rights

violations and identification of perpetrators. Household hunger, measured by FANTA-2 scale, was low in 581 (84.7%) households, moderate in 85 (12.3%) households and high in six (0.9%) households. Households reporting food shortages during any month in 2011 ranged from 19.9% to 47.0%, with food insecurity peaking just prior to the harvest. Diarrhea prevalence in children was 14.2% and night blindness in women of child-bearing age was 5.6%.

Forced labor was the most common human rights violation, reported by 185 households

(24.9%); 210 households (30.6%) reported experiencing at least one human rights violation in

2011. Multiple logistic regression analysis identified associations between human rights violations and poor health outcomes.

Conclusion

Human rights violations and their health consequences persist despite reduced intensity of

conflict in eastern Burma. Ceasefire agreements should include language that protects human

rights, and reconciliation efforts should address the health consequences of decades of human rights violations.

Background

Karen state, in eastern Burma, has experienced six decades of low-intensity conflict that has had severe impact on the civilian population. The consequences of fighting in eastern Burma have direct effects on the population such as forced displacement, pillaged food stores, injury from violence and forced labor. (Beyrer et al., 2007; Beyrer, 2007; HRW, 2011; KWO, 2007; Mullany,

Lee, Yone et al., 2008; Teela et al., 2009) Indirect effects of war include poor transportation infrastructure, poor supply chains for clinics, and increased risk for healthcare providers. (Beyrer et al., 2007; Beyrer, 2007; Mahn et al., 2008; Mullany, Lee, Paw et al., 2008; Mullany et al., 2010;

Suwanvanichkij, 2008; Teela et al., 2009)

Human rights violations are a determinant of health in eastern Burma. Conflict areas in eastern

Burma have substantially higher infant and under-5 mortality rates compared with the national estimates for the whole country. (Lee, Mullany, Richards, Kuiper, Maung, & Beyrer, 2006a)

Surveys in this region found that child malnutrition and child mortality are associated with forced displacement, and that malaria parasitemia and child malnutrition were associated with theft and destruction of food supply. (Mullany et al., 2007)

Although these studies were done in conflict areas, reports from Burma suggest that human

rights violations can and do occur in areas of low or no conflict. A 2011 survey from Chin State in

western Burma found a 91% prevalence of forced labor with the Burma army responsible for a

majority of the violations, even in the absence of fighting. (Sollom et al., 2011) The same survey

identified associations between household hunger and human rights violations. (Sollom et al.,

2011) Reports from community-based human rights groups suggest that militarization, or the presence of armed groups, regardless of combat status, results in human rights violations in

Karen state. (KHRG, 2008; KHRG, 2010; KHRG, 2012)

Official peace negotiations began in Karen state 2011, a ceasefire was signed in 2012, and open

fighting between the Burma army and the main Karen opposition groups had declined over the

year leading up the ceasefire. The ceasefire talks were accompanied by discussion of return and

resettlement of refugees and IDPs, demining, and construction of “economic development

zones”—factories, extractive industries, hydroelectric and transportation projects-- in the state.

Rights groups have warned that many of these activities are contradictory to rights-based

development, and they have criticized the increased militarization frequently associated with

them. (BEWG, 2011; ERI, 2012; KHRG, 2012) Although the fighting in Karen state may cease,

many challenges remain. Data on health and human rights is needed to mitigate the risk of

future human rights violations, to inform policy that could improve health of Karen people, to

inform reconciliation and transitional justice efforts, and to ensure that Karen people also

benefit from the democratic reforms and opening economy in Burma.

Methods

The project is a collaboration between Physicians for Human Rights, The Center for Public Health

and Human Rights at Johns Hopkins Bloomberg School of Public Health, and five community-

based organization working in Karen state: Backpack Health Worker Team (BPHWT), Karen

Department of Health and Welfare (KDHW), Karen Youth Organization (KYO), the Committee for

Internally Displaced Karen People (CIDKP) and one group that wishes to remain anonymous.

The sampling universe included adults and children living in clinical catchment areas served by

BPHWT and KDHW in Karen state, eastern Burma, and also adults and children living in three townships around Tavoy town in Mon State, eastern Burma in 2010. The sampling frame was about 80,000 people in 250 villages. Geography in the areas sampled ranged from jungle- covered mountainous regions to coastal plains with paddy fields. The mountainous areas, especially in former conflict areas have very few roads or other infrastructure.

We calculated the required sample size to be able to detect a prevalence of any human rights violation of 15% and assumed survey return rate of 82% (both estimated from previous surveys in Karen state), with a design effect of 3.0. (Prudhon & Spiegel, 2007; Rose et al., 2006; Working

Group for Mortality Estimation in Emergencies, 2007)We determined that a 90 x 8 design was most appropriate for this survey in this setting in order to best account for the uneven distribution of outcomes measured and minimized the impact of losing entire cluster of data if a surveyor lost data forms if he or she had to flee suddenly due to insecurity. {(F. Checchi, 2010; de Ville de Goyet, C et al., 1978; Depoortere & Checchi, 2006; Lee, Mullany, Richards, Kuiper,

Maung, & Beyrer, 2006a; Purdin et al., 2009; L. Roberts et al., 2003; Spiegel & Robinson, 2010;

Working Group for Mortality Estimation in Emergencies, 2007) Partner organizations provided population data for their clinical catchment areas or we estimated population size based on the number of houses in the village. We randomly selected villages by assigning probabilities of selection proportional to size, and surveyors selected houses in each village using a modified spin-the-pen technique. (Bennett et al., 1991; Grais et al., 2007; Sollom et al., 2011)

The survey questionnaire was previously used in Burma and modified for Karen state. (Sollom et

al., 2011) It was designed to assess common human rights violations, access to healthcare and food security. If a respondent said that they had experienced a human rights violation they were asked several follow-up questions: to confirm when the event happened, to identify the perpetrator, to identify the family member who experienced the violation, and to affirm whether the respondent was an eyewitness or not. Only human rights violations for which a respondent could answer these questions were included in the analysis. In some cases respondents said civilians were responsible for the human rights violation; these violations were recoded as “no violation” for the analysis because international law dictates that only governments can perpetrate human rights violations.

The questionnaire incorporated the six-question USAID Food and Nutrition Technical Assistance

Project (FANTA) household hunger survey (HHS) and the months of adequate household food production (MAHFP) survey. (Bilinsky & Swindale, 2007a; Deitcher et al., 2011; Sollom et al.,

2011)

We included MAHFP because it has been validated to cover a one-year time frame, which, unlike the HHS, would extend to before the most recent harvest in Karen state. (Bilinsky &

Swindale, 2007b) MAHFP is measured by asking the head of household to consider each month in the preceding year and to recall if there was any time during that month in which the household did not have enough food to meet its needs. The number of months in which any single household was able to meet it food requirements can be calculated (this count is MAHFP).

Surveyors measured middle-upper arm circumference (MUAC) in all children between 6 and 60

months and asked if any household member had diarrhea or night blindness in the two weeks prior to the survey. If households reported a member who had night blindness or another severe health condition, surveyors referred that person to the nearest clinic.

Surveyors and training

Partner CBOs committed 22 surveyors (16 male, 6 female; age range 20-38 years) who were fluent in Burmese or Sgaw Karen, had knowledge of the terrain, political climate and local leaders in the area where they surveyed. They were able to do mathematical calculations and were able to travel by local means or by foot through remote areas of the State. Seven of the surveyors worked with youth groups and the remainder were community health workers.

Surveyors were trained in lectures and practical sessions over two weeks, and they were required to pass a final check-out test before they went to the field.

Before approaching a village, surveyors assessed the security situation , and they sampled the

next closest village if they determined there was a risk. Surveyors first obtained informed

consent from the village leader and interviewed the leader about exposure to armed groups

and access to health care. The surveyor next approached the households, identified the head of

household or another adult, obtained the same consent and began the interview. Neither the

village leaders nor heads of household received compensation for participating in the survey.

Survey participants were 15 years of age or older if married, 18 years or older if unmarried, living in Karen state, who spoke Burmese or Sgaw Karen, displayed sound psychological state to answer sensitive questions and provided informed consent to participate. Children 6 to 59 months old who were residents of the enrolled household whose parent had provided informed consent were included for MUAC measurements. Anyone who ate meals at the house for the two months preceding the survey was considered a household member.

The goal of the data analysis was to identify associations between human rights violations and health outcomes. All analyses were performed using STATA 13 and svy commands to apply

Taylor linearization to the data to adjust for cluster sampling. Data was weighted at either the clinical area level or the village level by population before analyses were performed.

Before beginning the interpretive analysis we tested for clustering at the household level and

found that it was not present. STATA 13 svy commands account for village level clustering, and

mutlilevel models were not used in the final analysis.

We used generalized linear models with log link functions to estimate risk ratios for human

rights violations and health outcomes, using Poisson regression when models failed to converge.

Household hunger was coded into moderate/severe and none/mild categories; MUAC scores for

children under 5 were separated into moderate/severe (<12.5mm) and none/mild (>12.4mm),

(based on WHO criteria); and diarrhea and night blindness were recoded as binary variables

(present/ not present). For analysis we coded MAHFP into a binary variable using 9 months as a

cutoff. The distribution of MAHFP peaked at 8 months in households that had experienced

HRVs, and at 10 months for households that had not, and this best captured the differences

between the two populations.

If necessary, adjustments were made for variables known to be confounders or effect modifiers

in other surveys in the area: household size, type of water supply, clinical catchment area,

religion, topography, female-headed households and exposure to any other human rights

violations. (Lee, Mullany, Richards, Kuiper, Maung, & Beyrer, 2006a; Mullany et al., 2007; Sollom

et al., 2011) Only adjusted risk ratios are reported.

About 7% of data for diarrhea and night blindness were missing. We determined that the data was “missing at random” and used multiple imputation with chained equations to estimate values for missing variables (diarrhea, night blindness, human rights violations). (Allison, 2009)

Results from the original analysis and analysis with imputed data were similar, and only results

from the original data set are reported here.

The PHR Ethical Review Board, the Institutional Review Board at the Johns Hopkins Bloomberg

School of Public Health, and a Karen community advisory team reviewed and approved the

research plan.

Results

Surveyors approached 90 villages; they encountered security risks in 10 villages and substituted eight for a total of 88 village leaders approached. One village leader refused consent, so 87 total villages were sampled. Of the 696 households approached by the surveyors, 686 consented to participate in the survey.

Surveyors assigned to the Tavoy region photocopied their survey forms inside Tavoy town to reduce security risks while crossing into Burma after the training. Two pages of the original survey were not photocopied, and thus some demographic information and data on access to healthcare were not collected in this area. The data was missing systematically, and we excluded variables for which these data were missing in the interpretive analysis.

Demographics

The 686 households sampled represented a total of 3657 people. Household size ranged from one to 16 members, and the mean household size was 5.33 people. Of the households surveyed,

42 (6.16%) fit our criteria for female-headed households. The male-female ratio for 15-25 year olds was 0.88 and that for 15-45 year olds was 0.93.

The population was 13.4% Christian and 59.5% Buddhist, with Atheist, Animist, “other” and “no

response” making up the remaining 7.1%. Seventy-five percent of respondents said they were from the Sgaw Karen ethnic group and 4.4% reported they were Po Karen. Over two thirds of people interviewed were married and most said they were farmers.

Health

Diarrhea prevalence in children was 14.2% and night blindness in women of child-bearing age was 5.6%. Of 423 children aged 6 to 59 months, three (1.0%) had MUAC less than 11.5mm, 10

(3.2%) had MUAC between 11.4 and 12.5, and 24 (8.6%) had MUAC between 12.4 and 13.5.

Surveyors reported difficulty in locating and measuring children, and 108 (24.8%) of eligible children were not measured for MUAC.

Analysis of household hunger questions indicated that 581 (84.7%) of households had low household hunger, the lowest possible rank on this scale, 85 (12.3%) had moderate hunger and six households (0.9%) had severe hunger. Female-headed households experienced household hunger in similar proportions: 36 (87.8%) had low hunger, 4 (9.8%) had moderate hunger and 1

(2.4%) had severe hunger.

For each month in 2011 the proportion of households reporting not having enough food to meet their needs ranged from 19.9% to 47.0%, with the most households reporting having insufficient food in the months just before the harvest. Two hundred eighty eight (43.3%) households reported having adequate food for every month in 2011. The remainder reported at least one

month of inadequate food production, with 99 (14.9%) households reporting that they were not

able to meet their food needs for any month in 2011.

Most households (49.4%) reported that the primary source of their medicine was from a community-based organization. Over a third (37.0%) of households reported sending a household member to any clinic sometime during the last year, 11.8% reported that a member left Karen state to get healthcare and 13.6% reported that a household member was sick but unable to access healthcare. The most commonly reported barriers to healthcare were long distance to travel and high cost of travel.

Human Rights Violations

In 2011, 210 households (30.6%) surveyed reported experiencing at least one human rights violation. Forced labor was the most common violation, 185 households (24.9%) reported any type of forced labor violation (including forced to carry supplied for an armed group, to sweep for mines, to grow crops, to work for the military or other forced labor), while 11 households

(1.4%) reported any kind of assault (including kidnapping, rape, torture and beating). Eighty six households (14.9%) experienced more than one human rights violation, and the most human rights violations reported by a household was six.

Table 19 lists perpetrators by violation. In most cases, the Burma army or other government officials were responsible for the violation.

Table 20 lists associations between health outcomes and human rights violations, adjusted for confounding. There was substantially higher risk for a household member to have diarrhea in

household that had experienced any forced labor (PRR 2.63, 95% CI 1.94 to 3.55), any human

rights violation (PRR 2.73, 95% CI 1.96 to 3.80), forced portering (PRR 2.32, 95% CI 1.50 to 3.58),

theft or destruction of food (PRR 2.10, 95% CI 1.11 to 3.98), restricted movement (PRR 2.61,

95% CI 1.19 to 5.71), or multiple human rights violations (PRR 2.67, 95% CI 1.67 to 4.28).

Similarly, there was higher risk for children under 5 to have diarrhea if their household

experienced theft or destruction of food (PRR 2.91, 95%CI 1.1-3.09), restrictions on movement

(PRR 2.61, 95%CI 1.19-5.71) or multiple human rights violations (PRR 2.67. 95% CI 1.67-4.28).

Households had higher risk of household hunger if they experienced theft or destruction of food

(PRR 2.59, 95% CI 1.38 to 4.88) or were blocked from accessing their land (PRR 2.18, 95% CI

1.03-4.61). Households had an increased risk of inadequate food production if they had

experienced any forced labor (PRR 1.81, 95% CI 1.43 to 2.31), any human rights violation (PRR

1.93, 95% CI 1.50 to 2.46), forced portering (PRR 2.03, 95% CI 1.58 to 2.62), had restricted movement (PRR 1.72, 95%CI 1.14-2.59),or had multiple human rights violations (PRR 1.96, 95%

CI 1.50 to 2.56).

Discussion

This survey identified a 30% prevalence of human rights violations in post-conflict areas Karen state, Burma in 2011. Human rights violations were statistically associated with higher risk of diarrhea, household hunger and household food production.

Prevalence of child diarrhea and night blindness reported by our survey were higher than the average for southeast Asia. (Fischer Walker, Perin, Aryee, Boschi-Pinto, & Black, 2012) (WHO,

2011) Although Karen state has a history of food insecurity that could account for high

prevalence of night blindness, it is possible that heads of household misdiagnosed night

blindness among other household members.

Child malnutrition was lower in this survey than what has been reported previously in eastern

Burma. A 2011 survey reported moderate to severe child malnutrition at 4.7% and mild

malnutrition at 15.6%, and BPHWT reported moderate and severe child malnutrition in Karen

state at 12.5% and mild malnutrition at 28.6%. (TBBC, 2011) It is likely that prevalence of child

malnutrition in our survey was lower because the data was collected immediately after the

harvest when most households had sufficient food supplies. Logistical limitations precluded an

assessment of stunting, which may have been able to capture effects of malnourishment during

development.

Our data indicate that household food security is common. Data for HHH and MAHFP have not

previously been collected in Karen state, but a June 2011 survey found that about half of

households surveyed had enough food stored to last until the harvest in November. (TBBC,

2011)

The prevalence of human rights violations in this survey is similar to what has been previously reported. A 2004 survey found that 25.2% of respondents reported that the military stole or destroyed their food, 8.9% reported forced displacement, 2.1 % were physically attacked and

32.6% of respondents reported forced labor; in total 52.1% of respondents reported having experienced at least one human rights violation. (International Criminal Court, 2002; Mullany et al., 2007) A 2007 survey found that 1.2% of respondents reported that their fields were attacked, 3.1% reported having livestock stolen, 1.9% reported that their food was taken by the

army, 1.5% reported forced labor, and 10.5% reported forced displacement. (Mullany, Lee, Yone et al., 2008)

This survey identified statistically significant associations between human rights violations and several health outcomes: household hunger, months of adequate household food production, diarrhea and night blindness. A survey in Karen state in 2004 also identified associations between food destruction and mortality, food destruction and child malnutrition, and forced labor and mortality. (Mullany et al., 2007) A 2006 study in Karen state found associations between anemia and food security violations. (Mullany, Lee, Yone et al., 2008)

Associations between household hunger and human rights violations have not previously been assessed in Karen state; however, a 2011 report from Chin state, western Burma identified associations between household hunger and forced labor, assault, and human rights violations related to food security. (Sollom et al., 2011)

The results of our survey indicated that human rights abuses continued to be systematic and widespread in Karen state in 2011. We measured a lower prevalence of assaults compared with previous years, but a similar prevalence of forced labor. These findings might be explained because there was less conflict in Karen state in 2011 than in 2004 or 2007. Although there was limited fighting in 2011, the Burma army maintained over 250 outposts or bases in Karen state.

(TBBC, 2011) Army policies dictate that battalions supply themselves from the area they are assigned to patrol, and this often results in forced labor and theft of food from civilians. Our data indicate that militarization in Karen state, even in the absence of fighting, can result in human rights violations and health impacts on the civilian population.

Limitations

Limitations inherent to cluster sampling all apply to this survey and have been discussed in detail elsewhere. (F. Checchi, 2010; Depoortere & Checchi, 2006; Rose et al., 2006; Working

Group for Mortality Estimation in Emergencies, 2007) Cross sectional surveys cannot easily demonstrate causality. In this survey, all health outcomes except MAHFP were recorded from the month prior to the survey, and dates of human rights violations were recorded by surveyors and checked in the data analysis that the violation preceded the health outcome. In this way are able to demonstrate temporality of exposure and outcome, but this remains insufficient to fulfill all criteria for causality. (HILL, 1965)

This survey was done in areas in Karen state where community-based organizations are operating and it did not cover the entire state. , Results cannot be extrapolated to other parts of

Karen state. Families not registered by the CBO census, recent arrivals or displaced villages may not have been in the sample frame. The exclusion of these people from the survey could have biased results either toward or away from the null if there were systematic differences between them and the people registered by the CBO census.

In the past these organizations tended to focus on areas of conflict, and during the time of the survey the sampled population was under mixed administration. The age structure of the population was indicative of a population living in an area of conflict; age structure and proportions of religious and ethnic groups were similar to those reported elsewhere in Karen state. (Mullany et al., 2007; South, 2011). Because of the history of conflict and displacement,

the sampled population maybe more vulnerable in terms of food security and access to

healthcare than people living elsewhere in the state.

Surveyors could not approach ten of the clusters because of the presence of Burmese Army or

other security reasons and we cannot determine if there were systematic differences in human

rights violations or health outcomes between the villages that were skipped and those that were

sampled. If such differences exist, results could be biased.

Security concerns also limited the amount of quality control and follow up that was possible.

Surveyors did not revisit villages and field supervisors were not used during the survey. If a

household was empty, the surveyor returned later in the day, but if it was empty the surveyor

substituted another household. Because of the low level of conflict in the year preceding the

survey, it is unlikely that entire families were killed by armed groups, but it is possible that entire

families fled the area in search of better living conditions or to escape threat of violence.

Surveyors’ inability to follow-up on empty houses may introduce survivor bias into the results.

Additionally, we cannot estimate when household members entered or left the region, and thus we assumed the person-time at risk was the entire year for everyone in the survey.

Social desirability bias may be present if surveyors or participants felt they would benefit from

exaggerating results that would make stronger advocacy. This issue was covered extensively during training and was also written into the informed consent in order to minimize this bias.

Although surveyors lived and worked in the areas they were assigned to cover, it is possible that

interviewees were not comfortable discussing sensitive issues .such as health or human rights

violations. Due to logistical constraints, we did not match surveyors and respondents by sex.

During the informed consent process, surveyors assured respondents of anonymity and

confidentiality, and, it is possible that sensitive information was underreported.

MUAC and household hunger results represent a snapshot of the yearly cycle of malnutrition in

Karen state. Rice is harvested from September to November in Karen state, and at the time of

the survey,--January--it is likely that families were at one of their most food-secure times during the year. A month-by-month analysis of MAHFP confirms this. The nutrition data likely represent a best-case scenario for families over the course of the year.

In this survey, most questions only covered the month prior to the survey, but MAHFP and human rights questions covered an entire year. Herlihy et al suggest recall bias is minimal for traumatic events, but it is possible that recall bias is present in determination of MAHFP.

(Herlihy, Scragg, & Turner, 2002)

Comment

Decades of war and human rights violations have taken a toll on health of civilians in Karen

state, and political changes that began in the central part of Burma in 2010 are slow to reach

this area. Although a ceasefire has been reached in Karen state, this has not meant an end to

human rights abuses or to improved healthcare in the area. As the peace process moves forward in Karen state and violations of civil and political rights cease, more focus should be placed on fulfillment economic social and cultural rights. In Karen state human rights violations are a determinant of health. Poor community health in Karen state has its roots in decades of conflict and abuse that has plagued the region. Peace talks and reconciliation efforts should

acknowledge this and place priority on strengthening health systems in this area. Increased cooperation between the government, international donors and the Karen community is needed to improve health for Karen civilians.

Chapter 5 Figures and Tables

Table 7. Survey Coverage Clusters Households Clinical HHs HHs intended skipped substituted reached % % Area approached consented Doo Tha Tu 8 0 0 8 100 64 64 100 Dooplaya 12 3 2 11 91.7 88 88 100 Pa An 25 4 3 24 96 192 175 91.1 Papun 20 3 3 20 100 160 160 100 Tavoy 17 0 0 17 100 136 135 99.3 Win Yee 8 0 0 8 100 64 64 100 totals 90 10 8 88 97.8 704 686 97.4

Table 8. Population Structure population <5 y 423 population <15 y 1313 population > 65 y 132 males 15-25 y 337 females 15-25 y 379 M/F ratio 15-25 y 0.88 males 15-45 799 females 15-45 857 M/F ratio 15-45 y 0.93 sex, male 1802 sex, female 1802 sex, missing 53 total population 3657

Figure 7. Population Pyramid

Table 9. Characteristics of Sampled Population Religion** n responding percent* Ethnicity** n responding percent* Christian 92 13.4 Sgaw 511 75 Buddhist 408 59.5 Po 30 4.4 Other 49 7.1 other 9 1.4 total 686 total 686

Marital Status*** n responding percent* Work** n responding percent* not married 17 2.5 nothing 14 2 married 488 71.1 farmer 497 72.5 widow/widower 43 6.3 wage labor 17 2.5 separated/divorced 1 0.2 trader 6 0.9 other 10 1.5 total 686 total 686

* percent is adjusted for cluster sampling design

** no response 20% ***no response 12%

Table 10. Prevalence of Diarrhea Everyone 0-59 months yes % yes %

None 3164 86.5 345 81.6

Yes 212 5.8 60 14.2

Missing 281 7.7 18 4.3 Total 3657 423

Table 11. Prevalence of Night Blindness Everyone 0-59 months Women 15-45 years yes % yes % yes % None 3238 88.5 402 95 764 85.4 Yes 155 4.2 3 0.7 50 5.6 Missing 264 7.2 18 4.3 81 9.1 Total 3657 423 895

Table 12. MUAC in Children 6-59 Months

n % WHO classification

Total under 5s 423 - - Total MUACs reported 315 74.5 - Missing data 81 19.1 - Child not at home; refuse to measure 27 6.4 - MUAC 0 to 11.4 3 1 severe

MUAC 11.5 to 12.4 10 3.2 moderate MUAC 12.5to13.4 24 8.6 at-risk MUAC>13.4 278 91.4 none

Table 13. Responses to FANTA-2 Household Hunger Questions % In the last month, how many days did one of these events 1-2 3-10 10+ missing never total missing happen: times times times data data No food in the household because no money 466 66 90 55 9 686 1.3

Household member went to sleep hungry 590 20 47 19 10 686 1.5

Household member went a whole day w/o eating b/c no food 647 11 13 3 12 686 1.8

Table 14. Household Hunger All Households Female-Headed Households

n % n %

Severe HHH 6 0.9 1 2.4

Moderate HHH 85 12.3 4 9.8

Low HHH 581 84.7 36 87.8

Moderate or severe HHH 91 15.8 5 14.3

Table 15. Months of Adequate Household Food Production

Month in 2011 HH had enough food HH had not enough food Total responding** % reporting not enough food

January 531 138 669 20.6

February 531 138 669 20.6 March 525 144 669 21.5

April 522 147 669 22

May 517 153 670 22.8

June 499 168 667 25.2 July 459 210 669 31.4

August 415 254 669 38

September 355 315 670 47

October 366 302 668 45.2 November* 477 191 668 28.6

December* 534 133 667 19.9 >9 months 378 287 665 43.2 *harvest months

** >3% missing data

Figure 8. Percent of Households Reporting Sufficient Food in Each Month of 2011

Figure 9. Household MAHFP, Stratified by Any HRVs

250

200 noHRV 150 anyHRV 100

Number of of Households Number 50

0 0 1 2 3 4 5 6 7 8 9 10 11 12 Months of Adequate HH Food Production for 2011

100

Table 16. Access to Healthcare no yes % yes* HH member went to a clinic in the last year 430 252 37 HH member left Karen state to get healthcare 600 80 11.8 HH member was sick and could not get treatment 500 79 13.6

HH gets medicine from: Government clinic 486 39 7.4 CBO 264 258 49.4 Local pharmacy 83 456 84.6 Traditional healer 378 146 27.9 Other 458 60 11.6

HH faces this barrier to healthcare: Too far 217 330 60.3 High cost (of travel) 210 338 61.7 Not allowed to travel/checkpoints 540 6 1.1 Fighting/ insecurity 521 25 4.6 Other 378 156 29.2 *20% missing due to photocopying errors inside

Tavoy region.

Table 17. Reported Source of Drinking Water

n %*

Unprotected source or untreated 355 20.3%

Protected source or treated 192 28.0%

*20% missing due to photocopying errors inside Tavoy region.

101

Table 18. Human Rights Violations % Households Cases 95% CI 95% CI Type of Violation %* missing responding in 2011 lower upper data** Forced to be porters 672 90 14.4 9.9 20.5 2 Forced to sweep for mines 672 5 0.5 0 1.5 2.8 Forced to grow crops 670 25 2.8 1.4 5.6 3.1 Forced to work for military 556 50 9.5 6.1 14.5 31 Other forced labor*** 671 91 14.1 9.7 20.1 3.1 Blocked from accessing land 663 20 4.3 2.2 8.1 3.9 Food stolen or destroyed 681 24 4.1 3.4 9.8 0.8 Restricted movements 670 28 6 3 11.3 0.7 Religious discrimination 662 6 0.9 0.4 2.3 2.8 Kidnapped 685 1 0.2 0 1.1 0.2 Wounded 675 1 0.2 0 1.7 2.6 Tortured 673 9 1.3 0.1 2.8 2.5 Sexually Assaulted 671 5 0.1 0 2.8 3.3 Any forced labor 684 185 24.9 19.1 31.6 0.3 Any assault 685 11 1.4 0.6 3.1 1 Any HRV 686 210 30.6 22.7 35.6 0 No HRVs 686 476 71.3 64.4 77.3 0 Only one HRV 686 127 14.4 11.5 17.9 0 Two HRVs 686 42 6.9 4.3 10.9 0 Three HRVs 686 29 5.2 3.1 8.6 0 Four HRVs 686 6 1.1 0.4 2.8 0 Five HRVs 686 3 0.6 0.2 2 0 Six HRVs 686 3 0.5 0.1 2.1 0 *calculated using Taylor linearization, so percents may not match a direct calculation **includes refused to answer, not recorded

*** includes cutting wood or bamboo, cleaning compounds, roads, building bridges or buildings

102

Table 19. Perpetrators Households Local NSAG responding Cases in Burma NSAG non- gov’t -- No Human Rights Violation Police cease- BGF*** Other Civilians to HRV 2011 army ceasefire* USDP Response fire** question (VPDC) Forced to be porters 652 90 80 - - - 6 2 2 - - Sweeping for mines 652 5 4 - - - - 1 - - - Forced to grow crops 652 25 25 ------Forced to work for 552 50 33 - 4 - 10 - 1 - 2 military Other forced labor 651 91 38 - 5 - 10 4 30 - 4 Blocked from accessing 644 20 14 - - - - 2 - - 4 land 103 Food stolen or destroyed 660 33 12 ------9 12

Restricted movements 650 28 23 1 1 - - - - - 3 Kidnapped 664 1 1 ------Wounded 656 1 ------1 Tortured 656 9 7 ------2 Sexually Assaulted 651 5 5 ------* non-state armed groups that did not have a ceasefire agreement with the Burmese government in 2011, includes KNLA and some breakaway factions of DKBA ** non-state armed groups that had a ceasefire agreement with the Burmese government in 2011 ** Border guard forces are ethnic armies that signed allegiance to the Burmese army and operated under Burmese army command. Includes most of DKBA

Table 20. Associations Between Human Rights Violations and Health Outcomes

Human Rights Violation Diarrhea Diarrhea in children under 5 Household hunger MAHFP

PRR 95 low 95 high PRR 95 low 95 high PRR 95 low 95 high PRR 95 low 95 high

Any forced labor 2.63 1.94 3.55 1.44 0.90 2.31 1.25 0.76 2.08 1.81 1.43 2.31 Any HRV 2.73 1.96 3.80 1.51 0.93 2.47 1.39 0.85 2.29 1.93 1.50 2.46 Forced to porter 2.32 1.50 3.58 0.90 0.42 1.92 1.24 0.65 2.36 2.03 1.58 2.61 Forced to grow crops 0.52 0.21 1.33 0.42 0.04 4.19 - - - 1.02 0.56 1.84 104 Other forms of forced labor 1.04 0.59 1.84 0.56 0.16 1.93 0.72 0.36 1.46 2.25 1.77 2.86

Blocked from accessing land 1.36 0.61 3.00 1.24 0.24 6.38 2.18 1.03 4.61 1.47 0.86 2.50 Food stolen or destroyed 2.10 1.11 3.98 2.91 1.58 5.34 2.59 1.38 4.88 0.71 0.40 1.28 Restricted movement 2.61 1.19 5.71 3.20 1.74 5.88 1.98 0.73 5.35 1.72 1.14 2.59 Two or more HRVs 2.67 1.67 4.28 1.53 0.64 3.63 1.69 0.93 3.10 1.96 1.50 2.56 Total number of HRVs (as continuous variable) 1.44 1.32 1.58 1.23 1.06 1.44 1.24 1.06 1.45 1.14 1.09 1.19 Adjusted for household size, type of water supply, clinical catchment area, religion, topography, female-headed households and exposure to any other human rights violations.

Table 21. Comparison of Results of Imputed Analysis for Diarrhea

Violation Diarrhea Imputed Diarrhea

PRR 95 low 95 high PRR 95 low 95 high

Any forced labor 2.63 1.94 3.55 2.58 1.98 3.37 Any HRV 2.73 1.96 3.80 2.77 2.05 3.75 Forced to porter 2.32 1.50 3.58 2.49 1.68 3.68 Forced to grow crops 0.52 0.21 1.33 0.56 0.23 1.35 Other forms of forced labor 1.04 0.59 1.84 1.40 0.90 2.17 Blocked from accessing land 1.36 0.61 3.00 1.41 0.65 3.06 Food stolen or destroyed 2.10 1.11 3.98 2.30 1.22 4.32 Restricted movement 2.61 1.19 5.71 3.71 2.08 6.59 Two or more HRVs 2.67 1.67 4.28 3.14 2.16 4.57

Total number of HRVs (continuous variable 1.44 1.32 1.58 1.46 1.36 1.56

105

Chapter 6. (Paper 3) Social Ecology of Health in Post-Conflict Karen State, Burma: Militarization, Risk and Community Responses

Abstract:

Background

Preliminary peace agreements between the Burmese army and the Karen National Union were

recently signed, but a heavy military presence remains in eastern Burma. Qualitative reports

suggest that this militarization can result in human rights abuses in the absence of conflict, and

that Karen civilians try to reduce the impact of human rights violations by using self-protection

techniques.

Methods and Findings

This is a secondary analysis of data from 463 households collected with a two-stage cluster survey of post-conflict areas in eastern Burma in January 2012. We used logistic regression to identify associations between exposure to armed groups, village self-protection, human rights abuses and health outcomes.

Close proximity to a military base was a predictor of human rights violations, inadequate food production, inability to access healthcare and diarrhea. Exposure to armed groups predicted these outcomes and also household hunger. In households that reported no human rights violations, risk of household hunger, inadequate food production, diarrhea and child diarrhea increased when neighbors’ households reported human rights violations. Households in villages that reported using any self-protection technique had lower risk of experiencing HRVs.

Households in villages that reported negotiating with the Burmese army had lower risk of

106

human rights violations, household hunger, inadequate food production and diarrhea. Stratified

analysis suggests that self-protection may modify the effect of exposure to armed groups on risk

of human rights violations and some health outcomes.

Conclusion

Militarization is a determinant of health in eastern Burma. It predicts human rights violations

and also poor health. In some cases, village self-protection activities reduce the risk of human

rights violations and poor health outcomes given exposure to armed groups.

Background:

Social-ecological models of health are grounded in tenets that the environment and society in

which a person lives plays a crucial role in determining that person’s health. (Grzywacz & Fuqua,

2000; Krieger, 2001; McLeroy et al., 1988) These models expand beyond biological paradigms of

causality and enable an examination of factors such as Socio-economic status (SES), stigma and other characteristics of groups of people as distal determinants of health that can affect multiple disease outcomes. (Mosley & Chen, 2003; Phelan et al., 2010) This type of analysis allows researchers to identify key environmental factors related to health inequity and to identify social constructs that preclude groups from reaching their highest attainable level of health, which, in turn, may prescribe interventions at a social-environmental level in addition to the level of the individual. (Farmer et al., 2006; Trickett & Beehler, 2013)

Social-ecological models have shown that mental and physical health outcomes in populations that have experienced conflict are related to daily stressors such as SES, gender and food security in addition to, and in many cases even more so, than experiencing traumatic events.

107

(Araya et al., 2007; Assefa et al., 2001; Betancourt et al., 2013; K. de Jong et al., 2008; K. de Jong,

Kam et al., 2008; Eljedi et al., 2006; B. Roberts et al., 2009; B. Roberts et al., 2010; Seino et al.,

2008; Thapa & Hauff, 2012) Researchers using these models recommend that successful health interventions in these situations should address multiple causes simultaneously, including social and environmental factors. (Betancourt et al., 2013; Thapa & Hauff, 2012; Trickett & Beehler,

2013)

We applied a social-ecological model to examine the relationship between health and human rights in post-conflict Karen state, in eastern Burma. Previous studies in this region have shown that households that experience human rights violations have an increased risk of poor health outcomes. (Vervisch et al., 2013) Because the household in any village is not an isolated unit-- burdens are shared through economic and social relationships--in this paper we expand the household health and human rights paradigm to the village level and include additional risk and risk-mitigation factors for health outcomes. Six decades of conflict recently ended in Karen state and the international community has begun making development plans for the region. We hope that a social-ecological model will help to highlight some of the key determinants of health in the area and that these determinants will inform health interventions and social reconciliation.

Over the course of the Karen conflict, human rights violations were widespread, from extrajudicial killing to forced relocations to arbitrary taxation and forced labor. All of these have effects on health: direct injury, loss of food supplies, loss of capital used to produce health. (Lee,

Mullany, Richards, Kuiper, Maung, & Beyrer, 2006a; Mullany et al., 2007; Teela et al., 2009)

Recent studies have suggested that militarization, or the presence of the military even in the absence of conflict, results in human rights violations. (Berman et al., 1994; Christensen, 2004;

108

Mullany et al., 2007) Burmese army policies in which it supplies itself with food and labor from

the local population, in addition to counterinsurgency tactics have health impacts on the

population.

In six decades of conflict, and in the absence of any major international intervention, villages

evolved strategies to protect themselves against human rights violations. This “village agency,”

or self-protection, applied at household and village levels, includes using early-warning systems

to identify approaching hostile troops and fleeing the village temporarily or permanently,

managing threats by negotiating a reduction in the amount of forced labor or food demanded,

lying to officials about village assets or delaying compliance of demands, refusing to comply with

demands or using homemade landmines or Gher der -- “home guard” militia --to protect the village. (KHRG, 2008; KHRG, 2010; South et al., 2010) The Karen Human Rights Group (KHRG) has conducted village agency trainings since 2005. (KHRG, 2008; KHRG, 2010; South et al., 2010)A

qualitative study concluded that communities in Karen state are able to “reduce the economic,

social, and humanitarian costs of military rule” with self-protection. (KHRG, 2008)

The power relationships in Karen state are conducive to self-protection activities. (KHRG, 2008)

Corruption and class differences in armed groups, variations among individual officers’ personalities, soldiers’ empathy for villagers and the presence of other armed groups have created a space for self-protection to develop. (Finch, 2013) Military commanders may strike a balance between demanding enough food and labor to support their commands but not demanding so much as to have villagers refuse or to invite a reprisal attack from an opposition group.

109

A social-ecological model of health in Karen state suggests that the presence of armed groups

could be a determinant of health, that human rights violations could affect health on both

household and village levels, and that self-protection also could work at both levels. We used

data from a cross-sectional survey to examine the relationships between these factors.

Methods

This study was approved by the Institutional Review Board at Johns Hopkins Bloomberg School

of Public Health, the Ethics Review Board at Physicians for Human Rights, and an ad hoc Karen

community advisory team.

Survey

Data for this analysis were collected in eastern Burma in January 2012. Details of the sampling

scheme, survey tool and survey implementation are described elsewhere. Briefly, we selected

90 villages using probability proportional to size, and selected 8 households in each village using

a modified spin-the-pen technique. The sampling universe included adults and children living in

clinical catchment areas served by community-based health organizations in Karen state,

eastern Burma. We identified 686 households using a two-stage sampling scheme and asked

adult heads-of-household about human rights violations, nutrition, access to healthcare and health status of family members. Of the 686 households interviewed in the original survey, 551 lived in post-conflict areas and were included in the analysis for this paper. Leaders of villages selected for sampling were also interviewed. The time period covered by the survey was Jan

2011 to Jan 2012.

110

The survey questionnaire covered exposure to armed groups, human rights violations, distance

to the nearest Burma army base, and self-protection activities. Six responses were possible for

the village self-protection question: negotiate reduction in labor demanded, pay to reduce the

labor demanded, flee to avoid forced labor, refuse to do the labor demanded, no response and

don’t know.

Health outcomes were measured over different amounts of time preceding the survey: diarrhea

(in last two weeks), household hunger (HHH-over last month) and months of adequate

household food production (MAHFP-over last year). HHH was measured with the six-question

USAID Food and Nutrition Technical Assistance Project (FANTA) household hunger survey

(Callahan, 2007; Jordt, 2007; KHRG, 2008; KHRG, 2010; Selth, 2002) and MAHFP is a count of months in which a household was able to meet its food needs. (Deitcher et al., 2011) For analysis we coded MAHFP into a binary variable using 9 months as a cutoff. The distribution of

MAHFP peaked at 8 months in households that had experienced HRVs, and at 10 months for households that had not, and this cutoff best captured the differences between the two populations.

Statistical analysis

All analyses were performed with STATA 13 using svy commands to apply Taylor linearization to adjust for cluster sampling. Data was weighted at either the clinical area level or the village level by population. First, survey coverage and participation rates were estimated. Next, prevalence of human rights violations, health and nutrition outcomes, self-protection activities were estimated, and 95 % confidence intervals were calculated.

111

We used multilevel models to test for clustering of outcomes at the household level; no

clustering was found and multilevel models were not used in the final analysis. About 15% of the

questions on self-protection and distance to base were missing, likely because of the sensitive

nature of the questions. As these variables were primary exposures or outcomes, we used

listwise deletion instead of imputation to resolve the missing data issue; 463 households were

included in the final analyses.

All interpretive analyses were adjusted for household size, area, type of water supply, religion,

ethnicity, female-headed household and terrain (mountain or plains) if appropriate. Risk ratios

between exposure to armed groups, human rights violations and health outcomes and 95%

confidence intervals were estimated using generalized linear models with log link functions or

Poisson regression when models failed to converge.

Guard Forces (BABGF).

112

performed these analyses twice, treating village HRVs as a binary outcome (none or any) and also as a continuous one (zero to eight—eight households were interviewed in each village).

Finally we investigated the effects of self-protection on HRVs and health. For each village sampled we combined the responses for household self-protection and generated binary self- protection variables (any or none for all self-protection, any or none for negotiation) and continuous village self-protection (0 to 32 for any self-protection and 0 to 8 for negotiation).

We first examined associations between village self-protection (as a continuous variable) and household exposure to armed groups, household HRVs and household health outcomes. Next we stratified village self-protection by any or none and examined associations between exposure to armed groups and HRVs and exposure to armed groups and health. Finally, we examined the relationships between exposure to armed groups and outcomes of HRVs or health, stratifying by any village negotiation.

Results

This is an analysis of data collected in a larger study done in Karen state in January 2012. We excluded some clusters from the original study (the Tavoy region) because they were located in administrative areas that were completely under the control of the Burma army and we felt it was not appropriate to compare these villages with ones in areas of conflict or mixed administration.

In the survey areas included in this analysis, surveyors approached 73 villages, skipped 10 because of security reasons, and substituted 8 for the skipped villages. One village leader refused to consent to the survey, so 70 villages were included in this analysis. Of the households

113

approached, 17 did not consent to the survey (including eight from the village whose leader

refused consent). We dropped 88 households from the analysis because they refused to answer

the self-protection questions, so 463 households were included in the final analysis. This sample

included 2471 people.

Of 463 households surveyed, 57 (12%) reported moderate or severe household hunger, 209

(45%) reported one to eleven months of inadequate household food production, and 45 (10%) reported twelve months of inadequate household food production. Prevalence of adult diarrhea was 7% (167/2471) and diarrhea in children under 5 was 17% (45/269). Out of 463 households reporting, 45 (10%) had a member leave Karen state for healthcare, and 58 (13%) had a household member who was sick but not able to get treatment.

The survey asked about 11 different specific human rights violations (HRVs); HRVs were grouped for this analysis. Of 463 households surveyed, 129 (28%) reported having been subjected to forced labor in the year prior to the survey, 148 (32%) reported experiencing any HRV, and 49

(11%) reported being forced to porter or carry supplies.

Distance from each village to the nearest Burmese army outpost ranged from one to 30 hours hiking. Missing data was high for this question (60 households, 13%). Households reported exposure (seeing) from zero (109 or 24%) to four (12 or 3%) different armed groups in the year prior to the survey; 179 (39%) of households reported seeing the Burma army or allied Border

Guard Forces, and 354 (76%) reported seeing at least one armed group.

114

About half (244 or 53%) of households reported engaging in some kind of self-protection activity: 138 (30%) reported negotiating reductions in forced labor, 141 (30%) reported paying to reduce labor, 18 (4%) reported fleeing to avoid forced labor, and 68 (15%) reported refusing to do forced labor.

On the village level, 279 (54.9%) households were in villages in which one to eight households reported HRVs, 259 (46.6%) of households were from villages in which at least one household reported negotiating with armed groups, and 291 (62.3%) were from villages in which at least one household reported any kind of self-protection. Of 315 (73.2%) households reporting no exposure to HRVs, 131 (38.3%) were from villages in which at least one other household reported HRVs.

Analysis of the data with logistic regression identified several statistically significant associations between exposure to armed groups, HRVs and poor health outcomes. A household’s proximity to an army base was also associated with higher risk of HRVs and poor health outcomes. For each additional hour of hiking from a Burmese army base that a household was located, that household’s risk of experiencing any HRV decreased by 23% (0.77, 95% CI: 0.68-0.87). Similarly, risk of poor MAHFP decreased by 5% (0.95, 95%CI: 0.91-0.99), risk of a household member being sick and not able to receive treatment decreased by 23% (0.77, 95%CI: 0.62-0.96), and risk of a household member reporting diarrhea decreased by 13% (0.87, 95%CI: 0.79-0.95) for each one- hour increase in hiking distance from an army base that the household was located.

Exposure to armed groups followed a similar pattern: households that reported seeing Burmese army or BGF were 5.7 times more likely to report moderate or severe HHH (95% CI: 2.5-13.0)

115

and 12.5 times more likely to report that a household member was sick and not able to get

treatment (95% CI: 5.06-30.69).

In households that did not report HRVs, having neighbors that reported HRVs increased the risk

of poor health outcomes. In households reporting no HRVs, risk of HHH was 5.6 times higher

(95% CI: 1.88-16.91) and risk of diarrhea was 2.53 times higher (95% CI: 1.45-4.42) for

households that had neighbors report HRVs compared with households whose neighbors did

not report HRVs.

Lower risk of HRVs experienced by any household in a village was associated with village self- protection (PRR 0.95, 95% CI: 0.92-0.99) and village negotiation (PRR 0.91, 95% CI: 0.85-0.98).

Village self-protection was not significantly associated with exposure to armed groups (data not shown). Village negotiation was associated with decreased risk of HHH (PRR 0.85, 95% CI: 0.74-

0.96), MAHFP (PRR 0.95, 95% CI: 0.91-1.00) and diarrhea (RR 0.89, 95% CI: 0.82-0.97).

Households that were in villages that practiced any form of self-protection had lower risk of diarrhea (PRR 0.95, 95% CI: 0.91-0.99).

Associations between exposure to armed groups and HRVs were different when stratified by village negotiation or any form of village self-protection. Any form of village self-protection had a modifying effect on associations between exposure to armed groups and experiencing any

HRV (no self-protection PRR 2.96, self-protection PRR 1.63). The effect was opposite, however, when BABGF was the exposure (no self-protection, PRR 1.09, self-protection PRR 1.88). Village- level negotiation had a slight modifying effect on associations between exposure to armed groups and experiencing any HRV (no negotiation PRR 1.74, negotiation PRR 1.66). Again, the

116

opposite effect was present when BABFG was the exposure (no negotiation PRR 1.09

negotiation PRR 1.88).

Sample sizes were small for the analysis of health and exposure to armed groups stratified by

self-protection, and some PRRs were unstable. Village negotiation appeared to have a strong modifying effect on the relationships between exposure to armed groups and health; the greatest differences were in exposure to the number of armed groups and HHH (no negotiation

PRR 2.53, negotiation PRR 1.25) and child diarrhea (no negotiation PRR 2.35, negotiation 0.64) and exposure to any armed group and diarrhea (no negotiation PRR 5.49, negotiation PRR 0.70).

Village negotiation also modified the effect between exposure to Burma Army or Border Guard

Forces and health, although for this exposure the effect was opposite as before: for MAHFP (no negotiation PRR 0.56, negotiation PRR 1.26) and diarrhea (no negotiation PRR 0.92, negotiation

PRR 1.47). Small sample sizes precluded an analysis of interaction terms in multiple logistic regression.

Discussion

This research identified several determinants of household health in Karen state, eastern

Burma: proximity to Burmese army bases, exposure to armed groups and HRVs in neighboring households. Self-protection was associated with decreased risk of HRVs, HHH, MAHFP and diarrhea.

Discussion of the prevalence of HRVs and health outcomes estimated by this survey is found elsewhere. Several prior studies in Burma identified associations between HRVs experienced by a household and health of that household. Studies in Karen state in 2004 and 2006 found

117

associations between food destruction and mortality, food destruction and child malnutrition, forced labor and mortality, anemia and food security violations and forced displacement and access to antenatal care. (Bilinsky & Swindale, 2007b) A 2006 study in Karen state found associations between anemia and food security violations, and also between forced displacement and receiving no antenatal care services. A 2011 survey in western Burma identified associations between household hunger and forced labor, assault, and human rights violations related to food security. (Mullany et al., 2007; Mullany, Lee, Yone et al., 2008)

This study showed that HRVs can affect the health of households that have not experienced them. It also identified associations between exposure to armed groups and poor health. These associations have not been previously documented in Burma, but associations between political violence and population health have been reported in Central America, Africa and the Middle

East. (Sollom et al., 2011)

The effects of self-protection on preventing HRVs or modifying the effect of militarization on health have not previously been measured. We identified several cases in which negotiation weakened the effect of exposure of any armed group on poor health outcomes, and two cases village negotiation strengthened the association between exposure to BA or BGF and poor health. We do not have an explanation for this discrepancy, but unmeasured variables, such as access to transportation routes (for trading), delivery of food or cash aid through CBOs, other unmeasured form of self-protection and dose of exposure to armed group may also affect these associations. Furthermore, the effects of self-protection on health may involve pathways other than the actions of armed groups. Villages that engage in self-protection may have higher self-

118

efficacy or a greater sense of dignity. These villages may have stronger health-seeking behaviors or improved mental health, both of which could affect physical health.

The limitations of the survey and sampling protocols are discussed elsewhere. Specific to this analysis, the number of missing data was high for questions relating to self-protection activities and distance to the nearest military base. It is likely that respondents were reluctant to answer these because of the sensitive nature of the questions. If there was a systematic reason why these questions were unanswered, such as a stronger perceived threat due to high exposure to armed groups, human rights violations or close proximity to military bases, our results could be biased towards the null.

We did not measure mental health or self-reported health. It is likely that these are related to self-protection through self-efficacy. We did not measure proximity to major transportation routes, such as roads or rivers. Although these enhance trade and village economy, they also increase exposure to armed groups. It is possible that proximity to a road could confound relationships between health and exposure to armed groups. We measured household exposure to the number of different armed groups over the course of the year, but we did not measure multiple exposures to the same group. This measurement would have enabled a better dose- response analysis that may have shifted results away from the null hypothesis. Finally, reporting bias between self-protection and HRVs is possible if households who reported HRVs were more likely to remember engaging in self-protection. This bias would strengthen the association between HRVs and self-protection.

119

Comment: Eastern Burma is experiencing peace for the first time in six decades, and this is

accompanied by an influx of development and health interventions from international agencies.

Despite the new ceasefires, militarization remains a determinant of health in post-conflict

eastern Burma. The effects of militarization on health cannot be overlooked when planning or

evaluating public health interventions in this area. Governments that want to improve health in

eastern Burma must address the problem of militarization; they could do this through negotiations with the Burmese army or through behavior change interventions with other armed groups. Evaluations of public health interventions in Karen state should include assessments of human rights violations. Karen villages have learned to cope with militarization, and international agencies should respect and support these coping mechanisms when planning new programs.

120

Chapter 6 Figures and Tables

Table 22. Survey Coverage, Excluding Tavoy Region Clusters Households

Clinical Area intended skipped substituted reached % HHs approached HHs consented %

Doo Tha Tu 8 0 0 8 100 64 64 100% Dooplaya 12 3 2 11 91.7 88 88 100% Pa An 25 4 3 24 96 192 175 91% Papun 20 3 3 20 100 160 160 100% Win Yee 8 0 0 8 100 64 64 100% 121 totals 73 10 8 71 568 551 97%

Table 23. Health Outcomes

HH Hunger HH hunger MAHFP Diarrhea Diarrhea U5 Left Karen for healthcare Sick and cannot access treatment

HHs % HHs % People % People % HHs % HHs %

Low 402 87% none 206 44% no 2257 91% 222 83% 417 90% 400 86%

Moderate/severe 57 12% some 209 45% yes 167 7% 45 17% 45 10% 58 13%

all 45 10%

Missing 4 1% missing 3 1% missing 47 2% 2 1% 1 0% 5 1%

Total HHs 463 463 2471 269 463 463

Table 24. Human Rights Violations

Forced labor Any HRV Portering

HHs % HHs % HHs %

334 72% 315 68% 409 88%

129 28% 148 32% 49 11%

0 0% 0 0% 5 1%

463 463 463

Table 25. Distance to Military Base, in Table 26. Exposure to Armed Groups Hours Hiking Number of groups seen in last HHs % year Hours Hiking HHs % 0 109 24% 1 120 26% 1 194 42% 2 60 13% 2 91 20% 3 98 21% 3 51 11% 4 39 8% 5 32 7% 4 12 3% 6 8 2% Missing 6 1% 10 14 3% Total 463 100%

12 16 3% None 109 24%

15 9 2% Any 354 76%

30 7 2% BA (Burmese army) 148 32% Missing 60 13% BABGF (Burmese army or border 179 39% guard force) Total 463

Table 27. Self-Protection Activities Reported by Households

Negotiate Pay Leave Refuse Any

Response HHs % HHs % HHs % HHs % HHs % 325 70% 334 72% 451 97% 395 85% 237 51% No Yes 138 30% 141 30% 18 4% 68 15% 244 53%

Total 463 463 463 463 463

122

Table 28. Total HRVs per Table 29. Total HHs that

Village Negotiate, per Village

Number Total HHs Total HHs HHs per Number of in villages in villages Village that % HHs that reporting reporting % Reported negotiate, (0-8) HHs (0-8) HRVs HRVs per village that negotiate 0 184 40.5 0 259 60.8

1 55 12.9 1 14 2.9 2 53 12.0 2 20 4.4 3 18 4.4 3 21 4.4 4 30 7.1 4 16 2.9 5 27 5.8 5 30 5.8 6 40 7.3 6 39 7.3 7 8 1.5 7 8 1.5 8 48 8.7 8 56 10.2

Table 30. Total Self-Protection Activities, per Village Number of self-protection HHs in villages reporting (0- activities reported, per % 20) self-protection activities village* 0 172 37.6

1 23 5.8

2 16 2.9

3 11 4.4

4 16 2.9

5 26 5.8

6 13 2.9

7 21 5.8

8 24 5.8

10 6 1.5

11 24 4.4 12 24 4.4 13 16 2.9 14 16 2.9 15 8 1.5 16 39 7.3 20 8 1.5 * each HH could report up to 4 self-protection activities, and there were 8 HHs surveyed in each village.

123

Table 31. Households with a Neighbor Reporting an HRV yes no total % yes

All HHs in survey 404 282 686 58.9

HHs reporting no HRVs 205 271 476 43.1

Table 32. Proximity to Army Base (in hours hiking) is Associated with HRVs 95% CI 95% CI HRV PRR lower upper Any forced labor 0.78 0.69 0.88 Any HRV 0.77 0.68 0.87 No adjustment needed

Table 33. Proximity to Army Base (in hours hiking) is Associated with Poor Health

Health Outcome PRR 95% CI lower 95% CI upper

HHH 0.82 0.67 1.01 MAHFP 0.93 0.88 0.97 Sick and cannot access treatment 0.77 0.62 0.96 Diarrhea 0.87 0.79 0.95 U5 diarrhea 0.93 0.82 1.06 adjusted for mountainous terrain, surveyor

Table 34. Exposure to Armed Groups is Associated with HRVs

HRV PRR 95% CI lower 95% CI upper

Number of armed groups (0-4) Forced labor 1.61 1.25 2.08 Number of armed groups (0-4) Any HRV 1.64 1.31 2.04 Any armed group Forced labor 3.19 1.3 7.83 Any armed group Any HRV 3.08 1.38 6.87 BA Forced labor 1.53 0.86 2.72 BA Any HRV 1.59 0.92 2.75 BAGF Forced labor 1.57 0.88 2.79 BABGF Any HRV 1.6 0.93 2.77 adjusted for mountainous terrain, surveyor

124

Table 35. Exposure to Armed Groups is Associated with Poor Health

outcome PRR 95% CI lower 95% CI upper

Any armed group HHH 2.06 0.88 4.82 MAHFP 1.32 0.72 2.41 Sick and no tx 3.57 1.27 10.02 Diarrhea 1.51 0.8 2.84 U5 diarrhea 1.18 0.62 2.23 Burma Army HHH 6.01 2.73 13.25 MAHFP 0.72 0.48 1.07 Sick and no tx 11.35 5.18 24.88 Diarrhea 1.13 0.58 2.19 U5 diarrhea 1.51 0.79 2.88 Burma Army or BGF HHH 5.69 2.48 13.02 MAHFP 0.78 0.52 1.18 Sick and no tx 12.46 5.06 30.69 Diarrhea 1.13 0.58 2.19 U5 diarrhea 1.48 0.77 2.86

Number of armed groups (0-4) HHH 1.71 1.30 2.23 MAHFP 0.95 0.75 1.20

Sick and no tx 2.40 1.70 3.39 Diarrhea 1.04 0.84 1.27 U5 diarrhea 1.05 0.79 1.38 adjusted for mountainous terrain, surveyor, type of drinking water

Table 36. Health in Households that did not Experience HRVs is Associated with HRVs in Neighbors' Households

Predictor-Village HRVs Outcome PRR 95% CI lower 95% CI upper

Any HHH 5.64 1.88 16.91 Any MAHFP 1.95 1.11 3.41 Any Diarrhea 2.53 1.45 4.42 Any U5 diarrhea 1.92 0.96 3.86 0-8 HHH 1.3 1.13 1.49 0-8 MAHFP 1.25 1.13 1.38 0-8 Diarrhea 1.38 1.22 1.57 0-8 U5 diarrhea 1.2 0.98 1.48 adjusted for area

125

Table 37. Household Health is Associated with Village HRVs

Predictor-Village HRVs Outcome PRR 95% CI lower 95% CI upper

Any HHH 5.2 1.76 15.34 Any MAHFP 1.52 0.95 2.42 Any Diarrhea 3.65 2.17 6.16 Any U5 diarrhea 1.98 1.02 1.23 0-8 HHH 1.12 1.02 1.23 0-8 MAHFP 1.1 1.05 1.16 0-8 Diarrhea 1.27 1.2 1.35 0-8 U5 diarrhea 1.12 1.02 1.23 adjusted for area

Table 38. Village Self-Protection is Associated with Lower Risk of HRVs in Village

PRR 95% CI lower 95% CI upper

Negotiate* 0.89 0.82 0.96 Any self-protection * 0.95 0.92 0.98 * continuous variables

Table 39. Village Self-Protection is Associated with Lower Risk of Poor Health 95% CI PRR 95% CI lower upper HHH 0.95 0.9 1.01 MAHFP 0.96 0.94 0.99 Diarrhea 0.95 0.91 0.99

U5 diarrhea ------

Table 40. Village Negotiation is Associated with Lower Risk of Poor

Health 95% CI PRR 95% CI lower upper HHH 0.85 0.74 0.96

MAHFP 0.93 0.89 0.98 Diarrhea 0.89 0.82 0.97 U5 diarrhea 0.94 0.85 1.04

126

Table 41. Village Negotiation Modifies the Association between Exposure to Armed Groups and Risk of HRVs 95% CI 95% CI Exposure Negotiation PRR lower upper Number of armed groups No 1.74 1.21 2.51 (0-4) Number of armed groups Yes 1.66 1.25 2.23 (0-4) Any armed group No 4.12* 1.49 11.37

Any armed group Yes 1.7* 0.33 8.7 BA No 1.59 0.79 3.2 BA Yes 1.45 0.65 3.26 BAGF No 1.53 0.76 3.1 BABGF Yes 1.57 0.71 3.48 adjusted for surveyor, mountainous terrain *Unstable PRRs

Table 42. Village Self-Protection Modifies the Association between Exposure to Armed Groups and Risk of HRVs Self- 95% CI 95% CI Exposure Protection PRR lower upper Number of armed groups No 1.5 5.82 (0-4) 2.96 Number of armed groups Yes 1.32 2.02 (0-4) 1.63

Any armed group No 19.65* 2.12 182.40

Any armed group Yes 1.97* 0.81 4.80 BA No 1.39 0.63 3.05 BA Yes 1.64 0.81 3.31 BAGF No 1.09 0.49 2.44 BABGF Yes 1.88 0.96 3.66 adjusted for surveyor, mountainous terrain *Unstable PRRs

127

Table 43. Village Negotiation Modifies the Association between Exposure to Armed Groups and Risk of Poor Health Exposure Health outcome Negotiation PRR 95% CI lower 95% CI upper

Any armed group HHH No 16.99* 2.03 141.96 Any armed group HHH Yes 0.69* 0.18 2.65 Number of armed groups HHH No 2.53 1.59 4.04 Number of armed groups HHH Yes 1.25 0.36 4.38 BA HHH No 12.36* 3.65 41.81 BA HHH Yes 2.72 0.87 8.52 BAGF HHH No 11.77* 3.07 45.04 BABGF HHH Yes 2.74 0.87 8.56 Distance to base HHH No 0.57 0.35 0.92 Distance to base HHH Yes 0.9 0.77 1.05 Any armed group MAHFP No 0.9 0.77 1.05 Any armed group MAHFP Yes 0.91 0.53 1.58 Number of armed groups MAHFP No 1.29 0.96 1.74 Number of armed groups MAHFP Yes 0.92 0.69 1.22 BA MAHFP No 0.65 0.39 1.08 BA MAHFP Yes 0.79 0.44 1.42 BAGF MAHFP No 0.56 0.35 0.9 BABGF MAHFP Yes 1.26 0.73 2.18 Distance to base MAHFP No 0.94 0.88 1 Distance to base MAHFP Yes 0.91 0.84 0.99 Any armed group diarrhea No 5.49 1.94 15.54 Any armed group diarrhea Yes 0.7 0.3 1.66 Number of armed groups diarrhea No 1.49 0.99 2.25 Number of armed groups diarrhea Yes 0.89 0.44 1.83 BA diarrhea No 1.04 0.47 2.31 BA diarrhea Yes 1.35 0.72 2.53 BAGF diarrhea No 0.92 0.41 2.08 BABGF diarrhea Yes 1.47 0.78 2.75 Distance to base diarrhea No 0.69 0.55 0.86 Distance to base diarrhea Yes 0.95 0.87 1.04 Any armed group U5 diarrhea No 4.54* 1.04 19.8 Any armed group U5 diarrhea Yes 0.57* 0.15 2.21 Number of armed groups U5 diarrhea No 2.35 0.79 6.92 Number of armed groups U5 diarrhea Yes 0.64 0.26 1.56 BA U5 diarrhea No 2.16 0.77 6.06 BA U5 diarrhea Yes 0.95* 0.39 2.34 BAGF U5 diarrhea No 1.93 0.66 5.67 BABGF U5 diarrhea Yes 0.93* 0.38 2.29 128

Distance to base U5 diarrhea No 0.68 0.5 0.93 Distance to base U5 diarrhea Yes 1 0.89 1.13 adjusted for surveyor, mountainous terrain, area * Unstable PRRs

129

Conclusions and Policy Implications

This research identified associations between militarization, human rights violations and poor

health outcomes. It confirms that human rights violations continue to be widespread in Karen

state, even in the absence of conflict.

An advocacy report based on these was released at an important time for human rights

advocacy in Burma. Changes implemented by the nominally civilian government in Burma had

resulted in reduced international scrutiny on human rights abuses in the country. The Burmese

government had also received much positive press for engaging in ceasefire talks with ethnic

governments; however, at that time far the talks were only focused on cessations of hostilities

and placement of liaison offices and not on protecting civilians from further abuses. The

advocacy report brought international attention back to the situation in Karen state in the

context of ceasefires and reconciliation. It called for the government to acknowledge past

human rights abuses, ensure justice is upheld, give reparations to those who have suffered

abuses, and establish credible and independent institutions that will deter future crimes.

Resolving these issues is key for national reconciliation and to permanently end the mass atrocities that have been ongoing for 60 years in that country.

Advocacy efforts by PHR and Karen partner organizations that used this research helped to maintain US financial support for cross-border aid groups, to reverse other countries’ foreign policy decision to halt cross-border aid, and to continue to argue against bilateral funding decisions that decreased support for cross-border aid groups. The research on economic development and human rights abuses in Karen state continues to be cited by journalists as a major threat facing Karen civilians.

130

Appendices Appendix 1. Map of Burma

131

Appendix 2. Sampling Frame, Disaggregated by District in Karen State

132

Appendix 3. Survey Questionnaire

133

134

135

136

137

138

References

Ahmed, A., Edward, A., & Burnham, G. (2004). Health indicators for mothers and children in rural herat province, afghanistan. Prehospital and Disaster Medicine, 19(3), 221-225.

AI. (2008). Crimes against humanity in eastern myanmar. New York: Amnesty Interntational.

Allison, P. D. When can you safely ignore multicollinearity? Retrieved 10/31, 2013, from http://www.statisticalhorizons.com/multicollinearity

Allison, P. D. (2009). Chapter 4: Missing data. In R. E. Millsap, & A. Maydeu-Olivares (Eds.), The SAGE handbook of quantitative methods in psychology () SAGE Publications Ltd.

Amowitz, L. L., Reis, C., & Iacopino, V. (2002). Maternal mortality in herat province, afghanistan, in 2002: An indicator of women's human rights. JAMA : The Journal of the American Medical Association, 288(10), 1284-1291.

Amowitz, L. L., Reis, C., Lyons, K. H., Vann, B., Mansaray, B., Akinsulure-Smith, A. M., et al. (2002). Prevalence of war-related sexual violence and other human rights abuses among internally displaced persons in sierra leone. JAMA : The Journal of the American Medical Association, 287(4), 513-521.

Amoz, B., Lanjouw, S., Pay Leek, S., Matra, E., Mortimer, G., Smith, A., et al. (1998). Forgotten victims of a hidden war: Internally displaced karen in burma. Chiang Mai: Burma Ethnic Research Group.

Araya, M., Chotai, J., Komproe, I. H., & de Jong, J. T. (2007). Effect of trauma on quality of life as mediated by mental distress and moderated by coping and social support among postconflict displaced ethiopians. Quality of Life Research : An International Journal of Quality of Life Aspects of Treatment, Care and Rehabilitation, 16(6), 915-927.

Armenian, H. K. (1989). Perceptions from epidemiologic research in an endemic war. Social Science & Medicine (1982), 28(7), 643-647.

Asher, J. (2013). Using surveys to estimate casualties post-conflict. In T. Seybolt, J. Aronson & B. Fischoff (Eds.), Counting civilian casualties: An introduction to recording and estimating nonmilitary deaths in conflict (pp. CH 6) Oxford University Press.

Assefa, F., Jabarkhil, M. Z., Salama, P., & Spiegel, P. (2001). Malnutrition and mortality in kohistan district, afghanistan, april 2001. JAMA : The Journal of the American Medical Association, 286(21), 2723- 2728.

Baines, E., & Paddon, E. (2012). 'This is how we survived': Civilian agency and humanitarian protection. Security Dialogue, 43, 231.

Bennett, S., Woods, T., Liyanage, W. M., & Smith, D. L. (1991). A simplified general method for cluster- sample surveys of health in developing countries. World Health Statistics Quarterly.Rapport Trimestriel De Statistiques Sanitaires Mondiales, 44(3), 98-106.

139

Berman, P., Kendall, C., & Bhattacharyya, K. (1994). The household production of health: Integrating social science perspectives on micro-level health determinants. Social Science & Medicine (1982), 38(2), 205-215.

Betancourt, T. S., McBain, R., Newnham, E. A., & Brennan, R. T. (2013). Context matters: Community characteristics and mental health among war-affected youth in sierra leone. Journal of Child Psychology and Psychiatry, and Allied Disciplines,

BEWG. (2009). Environmental protection, indigenous knowledge and livelihood in karen state: A focus on community conserved areas. Chiang Mai: Burma Environmental Working Group.

BEWG. (2011). Burma's environment: People, problems, policies. Chiang Mai: Burma Environmental Working Group.

Beyrer, C. (2007). Burma and the challenge of humanitarian assistance. Lancet, 370(9597), 1465-1467.

Beyrer, C., Villar, J. C., Suwanvanichkij, V., Singh, S., Baral, S. D., & Mills, E. J. (2007). Neglected diseases, civil conflicts, and the right to health. Lancet, 370(9587), 619-627.

Bilinsky, P., & Swindale, A. (2007a). Months of adequate househol food provisioning (MAHFP) for measurement of household food access: Indicator guide. Washington DC: USAID.

Bilinsky, P., & Swindale, A. (2007b). Months of adequate household food provisioning (MAHFP) for measurement of household food access: Indicator guide. Washington, DC: USAID-FANTA.

Bilukha, O. O. (2008). Old and new cluster designs in emergency field surveys: In search of a one-fits-all solution. Emerging Themes in Epidemiology, 5, 7.

Blas, E., Gilson, L., Kelly, M. P., Labonte, R., Lapitan, J., Muntaner, C., et al. (2008). Addressing social determinants of health inequities: What can the state and civil society do? Lancet, 372(9650), 1684- 1689.

Bornemisza, O., Ranson, M. K., Poletti, T. M., & Sondorp, E. (2010). Promoting health equity in conflict- affected fragile states. Social Science & Medicine (1982), 70(1), 80-88.

Bostoen, K., Bilukha, O., Fenn, B., Morgan, O. W., Tam, C. C., ter Veen, A., et al. (2007). Methods for health surveys in difficult settings: Charting progress, moving foward. Emerging Themes in Epidemiology, 4(13)

BPHWT. Home page. Retrieved Dec 23, 2013, from http://www.backpackteam.org/

BPHWT. (2006). Chronic emergency. Mae Sot: Back Pack Health Worker Team.

Brentlinger, P. E. (1996). Health sector response to security threats during the civil war in el salvador. BMJ (Clinical Research Ed.), 313(7070), 1470-1474.

Brentlinger, P. E., & Hernan, M. A. (2007). Armed conflict and poverty in central america: The convergence of epidemiology and human rights advocacy. Epidemiology (Cambridge, Mass.), 18(6), 673-677.

140

Brentlinger, P. E., Hernan, M. A., Hernandez-Diaz, S., Azaroff, L. S., & McCall, M. (1999). Childhood malnutrition and postwar reconstruction in rural el salvador: A community-based survey. JAMA : The Journal of the American Medical Association, 281(2), 184-190.

Burnham, G., Hoe, C., Hung, Y. W., Ferati, A., Dyer, A., Hifi, T. A., et al. (2011). Perceptions and utilization of primary health care services in iraq: Findings from a national household survey. BMC International Health and Human Rights, 11, 15-698X-11-15.

Burnham, G., Lafta, R., Doocy, S., & Roberts, L. (2006). Mortality after the 2003 invasion of iraq: A cross- sectional cluster sample survey. Lancet, 368(9545), 1421-1428.

Cairns, K. L., Woodruff, B. A., Myatt, M., Bartlett, L., Goldberg, H., & Roberts, L. (2009). Cross-sectional survey methods to assess retrospectively mortality in humanitarian emergencies. Disasters, 33(4), 503-521.

Callahan, M. (2007). Of kyay-zu and kyet-zu: The military in 2006. In M. Skidmore, & T. Wilson (Eds.), Myanmar: The state, community and the environment (pp. 36). Canberra: Asia Pacific Press.

Cardozo, B. L., Bilukha, O. O., Gotway, C. A., Wolfe, M. I., Gerber, M. L., & Anderson, M. (2005). Report from the CDC: Mental health of women in postwar afghanistan. Journal of Women's Health (2002), 14(4), 285-293.

Caulfield, L. E., de Onis, M., Blossner, M., & Black, R. E. (2004). Undernutrition as an underlying cause of child deaths associated with diarrhea, pneumonia, malaria, and measles. The American Journal of Clinical Nutrition, 80(1), 193-198.

CESCR. (2000). General comment 14: The right to the highest attainable standard of healthUnited Nations Office of the High Commissioner for Human Rights.

Chan, E. Y., & Kim, J. J. (2010). Characteristics and health outcomes of internally displaced population in unofficial rural self-settled camps after the 2005 kashmir, pakistan earthquake. European Journal of Emergency Medicine : Official Journal of the European Society for Emergency Medicine, 17(3), 136- 141.

Checchi, F., & Roberts, L. (2005). Interpreting and using mortality data in humanitarian emergencies A primer for non-epidemiologists. London: Humanitarian Practice Network/ ODI.

Checchi, F. (2010). Estimating the number of civilian deaths from armed conflicts. Lancet, 375(9711), 255- 257.

Checchi, F., & Roberts, L. (2008). Documenting mortality in crises: What keeps us from doing better. PLoS Medicine, 5(7), e146.

Christensen, P. (2004). The health-promoting family: A conceptual framework for future research. Social Science & Medicine (1982), 59(2), 377-387.

Coghlan, B., Brennan, R. J., Ngoy, P., Dofara, D., Otto, B., Clements, M., et al. (2006). Mortality in the democratic republic of congo: A nationwide survey. Lancet, 367(9504), 44-51.

141

Coghlan, B., Ngoy, P., Mulumba, F., Hardy, C., Bemo, V. N., Stewart, T., et al. (2009). Update on mortality in the democratic republic of congo: Results from a third nationwide survey. Disaster Medicine and Public Health Preparedness, 3(2), 88-96.

Coupland, R. (2007). Security, insecurity and health. Bulletin of the World Health Organization, 85(3), 181- 184.

De Jong, J. T. (2010). A public health framework to translate risk factors related to political violence and war into multi-level preventive interventions. Social Science & Medicine (1982), 70(1), 71-79. de Jong, K., Ford, N., Kam, S., Lokuge, K., Fromm, S., van Galen, R., et al. (2008). Conflict in the indian kashmir valley I: Exposure to violence. Conflict and Health, 2, 10-1505-2-10. de Jong, K., Kam, S., Ford, N., Lokuge, K., Fromm, S., van Galen, R., et al. (2008). Conflict in the indian kashmir valley II: Psychosocial impact. Conflict and Health, 2, 11-1505-2-11. de Ville de Goyet, C, Seaman, J., & Geijer, U. (1978). The management of nutritional emergencies in large populations. Geneva: WHO.

Degomme, O., & Guha-Sapir, D. (2007). Mortality and nutrition surveys by non-governmental organisations. perspectives from the CE-DAT database. Emerging Themes in Epidemiology, 4, 11.

Degomme, O., & Guha-Sapir, D. (2010). Patterns of mortality rates in darfur conflict. Lancet, 375(9711), 294-300.

Deitcher, M., Ballard, T., Swindale, A., & Coates, J. (2011). Introducing a simple method of household hunger for cross-cultural use. Washington DC: USAID.

Deitchler, M., Ballard, T., Swindale, A., & Coates, J. (2011). Introducing a simple measure of household hunger for cross-cultural use. Washington DC: USAID-FANTA.

Depoortere, E., & Checchi, F. (2006). Pre-emptive war epidemiology: Lessons from the democratic republic of congo. Lancet, 367(9504), 7-9.

Depoortere, E., Checchi, F., Broillet, F., Gerstl, S., Minetti, A., Gayraud, O., et al. (2004). Violence and mortality in west darfur, sudan (2003-04): Epidemiological evidence from four surveys. Lancet, 364(9442), 1315-1320.

Donaldson, R. I., Hung, Y. W., Shanovich, P., Hasoon, T., & Evans, G. (2010). Injury burden during an insurgency: The untold trauma of infrastructure breakdown in baghdad, iraq. The Journal of Trauma, 69(6), 1379-1385.

Doocy, S., Burnham, G., & Robinson, C. (2007). Estimating demographic indicators in a conflict-affected population in eastern sudan. Prehospital and Disaster Medicine : The Official Journal of the National Association of EMS Physicians and the World Association for Emergency and Disaster Medicine in Association with the Acute Care Foundation, 22(2), 112-119.

Eljedi, A., Mikolajczyk, R. T., Kraemer, A., & Laaser, U. (2006). Health-related quality of life in diabetic patients and controls without diabetes in refugee camps in the gaza strip: A cross-sectional study. BMC Public Health, 6, 268.

142

ERI. (2012). Where the change has yet to reach: Exposing ongoing earth rights abuses in burma . Chiang Mai: Earth Rights International.

Farmer, P. E., Nizeye, B., Stulac, S., & Keshavjee, S. (2006). Structural violence and clinical medicine. PLoS Medicine, 3(10), e449.

FBR. Home page. Retrieved Dec 23, 2013, from http://www.freeburmarangers.org/

Finch, M. (2013). International ‘humanitarian protection’ and local responses to abuse in eastern burma: Tensions and opportunities. Panel: Local Capacity for Self-Preservation Amid Violence: Recognising and Supporting it, World Conference on Humanitarian Studies, June 2-5 2011, Tufts University.

Fischer Walker, C. L., Perin, J., Aryee, M. J., Boschi-Pinto, C., & Black, R. E. (2012). Diarrhea incidence in low- and middle-income countries in 1990 and 2010: A systematic review. BMC Public Health, 12, 220-2458-12-220.

Ford, N., Mills, E. J., Zachariah, R., & Upshur, R. (2009). Ethics of conducting research in conflict settings. Conflict and Health, 3, 7-1505-3-7.

Gaber, S., & Patel, P. (2013). Tracing health system challenges in post-conflict cote d'ivoire from 1893 to 2013. Global Public Health, 8(6), 698-712.

Garfield, R. M. (1989). War-related changes in health and health services in nicaragua. Social Science & Medicine (1982), 28(7), 669-676.

Gorur, A. (2013). Community self protection strategies: How peacekeepers can help or harm No. Civilians in Conflict Issue Brief #1)Stimson Center.

Grais, R. F., Coulombier, D., Ampuero, J., Lucas, M. E., Barretto, A. T., Jacquier, G., et al. (2006). Are rapid population estimates accurate? A field trial of two different assessment methods. Disasters, 30(3), 364-376.

Grais, R. F., Rose, A. M., & Guthmann, J. P. (2007). Don't spin the pen: Two alternative methods for second-stage sampling in urban cluster surveys. Emerging Themes in Epidemiology, 4, 8.

Grzywacz, J. G., & Fuqua, J. (2000). The social ecology of health: Leverage points and linkages. Behavioral Medicine (Washington, D.C.), 26(3), 101-115.

Guerrier, G., Zounoun, M., Delarosa, O., Defourny, I., Lacharite, M., Brown, V., et al. (2009). Malnutrition and mortality patterns among internally displaced and non-displaced population living in a camp, a village or a town in eastern chad. PloS One, 4(11), e8077.

Haar, R. J., & Rubenstein, L. S. (2012). Health in fragile and post-conflict states: A review of current understanding and challenges ahead. Medicine, Conflict, and Survival, 28(4), 289-316.

Haenckerts, J. (2006). Study on customary international humanitarian law: A contribution to the understanding and respect for the rule of law in armed conflict No. 0860). Geneva: ICRC.

143

Harkness, S., & Super, C. M. (1994). The developmental niche: A theoretical framework for analyzing the household production of health. Social Science & Medicine (1982), 38(2), 217-226.

Hassan, S., Malik, E. M., Okoued, S. I., & Eltayeb, E. M. (2008). Retention and efficacy of long-lasting insecticide-treated nets distributed in eastern sudan: A two-step community-based study. Malaria Journal, 7, 85-2875-7-85.

Henderson, R. H., & Sundaresan, T. (1982). Cluster sampling to assess immunization coverage: A review of experience with a simplified sampling method. Bulletin of the World Health Organization, 60(2), 253- 260.

Herlihy, J., Scragg, P., & Turner, S. (2002). Discrepancies in autobiographical memories--implications for the assessment of asylum seekers: Repeated interviews study. BMJ (Clinical Research Ed.), 324(7333), 324-327.

HILL, A. B. (1965). The environment and disease: Association or causation? Proceedings of the Royal Society of Medicine, 58, 295-300.

HRDU. (1998). Forced relocation and internally displaced persons. Mae Sot: Human Rights Documentation Unit.

HRW. (2007). Sold to be soldiers: The recruitment and use of child soldiers in burma. New York: Human Rights Watch.

HRW. (2011). Dead men walking. New York: Human Rights Watch.

HRW. (2012). Untold miseries: Wartime abuses and forced displacement in kachin state. Washington DC: Human Rights Watch.

Hunt, P. (2006). Report of the special rapporteur on the right of everyone to the enjoyment of the highest attainable standard of physical and mental health, paul hunt . Geneva: UN COmmission on ESCR.

Iacopino, V., Frank, M. W., Bauer, H. M., Keller, A. S., Fink, S. L., Ford, D., et al. (2001). A population-based assessment of human rights abuses committed against ethnic albanian refugees from kosovo. American Journal of Public Health, 91(12), 2013-2018.

ICESCR. (1976). International covenant on economic, social and cultural rights. Geneva: UN.

ICRC. (2011). International humanitarian law and the challenges of contemporary armed conflicts . Geneva: ICRC.

ICTJ. What is transitional justice? Retrieved Dec 23, 2013, from http://ictj.org/about/transitional-justice

ILO. (1998). Report of the commission of inquiry to examine the observance by myanmar of the forced labour convention, 1930 (no. 29), 2 july 1998. Geneva: International Labour Organisation.

International Conference on Human Rights. (1968). Proclaimation of teheran No. A/CONF. 32/41 at 3 (1968)). Teheran: UN.

144

Rome Statute, (2002).

Vienna Convention on the Law of Treaties, (1969).

Jima, D., Getachew, A., Bilak, H., Steketee, R. W., Emerson, P. M., Graves, P. M., et al. (2010). Malaria indicator survey 2007, ethiopia: Coverage and use of major malaria prevention and control interventions. Malaria Journal, 9, 58-2875-9-58.

Johnson, K., Scott, J., Rughita, B., Kisielewski, M., Asher, J., Ong, R., et al. (2010). Association of sexual violence and human rights violations with physical and mental health in territories of the eastern democratic republic of the congo. JAMA : The Journal of the American Medical Association, 304(5), 553-562.

Jordt, I. (2007). Burma's mass lay meditation movement: Buddhism and the cultural construction of power. University of Ohio Press, Research in International Studies Series.

Kaiser, R., Spiegel, P. B., Henderson, A. K., & Gerber, M. L. (2003). The application of geographic information systems and global positioning systems in humanitarian emergencies: Lessons learned, programme implications and future research. Disasters, 27(2), 127-140.

Kaiser, R., Woodruff, B. A., Bilukha, O., Spiegel, P. B., & Salama, P. (2006). Using design effects from previous cluster surveys to guide sample size calculation in emergency settings. Disasters, 30(2), 199-211.

Karen News. About CIDKP. Retrieved Dec23, 2013, from http://karennews.org/tag/cidkp/

KBDDF. (2010). The karen people: Culture, faith and history. Australia: Karen Buddhist Dhamma Dhutta Foundation.

KDHW. (2012). About KDHW. Retrieved May 11, 2012, from http://kdhw.org/department/

KHRG. Home page. Retrieved Dec 23, 2013, from http://www.khrg.org/

KHRG. (2008). Growing up under militarisation: Abuse and agency of children in karen state. Mae Sot, Thailand: Karen Human Rights Group.

KHRG. (2009). Insecurity amidst the DKBA-KNLA conflict in dooplaya and pa'an districts. Mae Sot: Karen Human Rights Group.

KHRG. (2010). Self protection under the strain: Targeting of civilians and local responses in northern karen state. Mae Sot: Karen Human Rights Group.

KHRG. (2011a). Attacks on health and education: Trends and incidents in burma, 2010-2011 No. KHRG #2011-05). Mae Sot, Thailand: KHRG.

KHRG. (2011b). In William Davis (Ed.), Discussion on fighting in karen state Mae Sot, Thailand.

KHRG. (2012). Safeguarding human rights in a post-ceasefire eastern burma. Mae Sot: Karen Human Rights Group.

145

Kim, G., Griffin, S., Nadem, H., Aria, J., & Lawry, L. (2008). Evaluation of an interactive electronic health education tool in rural afghanistan. Prehospital and Disaster Medicine, 23(3), 218-226.

King, J. D., Ngondi, J., Gatpan, G., Lopidia, B., Becknell, S., & Emerson, P. M. (2008). The burden of trachoma in ayod county of southern sudan. PLoS Neglected Tropical Diseases, 2(9), e299.

Kirsch, T. D., Leidman, E., Weiss, W., & Doocy, S. (2012). The impact of the earthquake and humanitarian assistance on household economies and livelihoods of earthquake-affected populations in haiti. American Journal of Disaster Medicine, 7(2), 85-94.

Kirsch, T. D., Wadhwani, C., Sauer, L., Doocy, S., & Catlett, C. (2012). Impact of the 2010 pakistan floods on rural and urban populations at six months. PLoS Currents,

KNU. (2012). Statement on initial agreement between KNU and burmese government. Mae Sot: Karen National Union.

Krieger, N. (2001). The ostrich, the albatross, and public health: An ecosocial perspective--or why an explicit focus on health consequences of discrimination and deprivation is vital for good science and public health practice. Public Health Reports (Washington, D.C.: 1974), 116(5), 419-423.

Kruk, M. E., Freedman, L. P., Anglin, G. A., & Waldman, R. J. (2010). Rebuilding health systems to improve health and promote statebuilding in post-conflict countries: A theoretical framework and research agenda. Social Science & Medicine (1982), 70(1), 89-97.

KWO. Home page. Retrieved Dec 23, 2013, from http://karenwomen.org/

KWO. (2007). State of terror. Mae Sariang: Karen Womens Association.

Lee, T. J., Mullany, L. C., Richards, A. K., Kuiper, H. K., Maung, C., & Beyrer, C. (2006a). Mortality rates in conflict zones in karen, karenni, and mon states in eastern burma. Tropical Medicine & International Health : TM & IH, 11(7), 1119-1127.

Lee, T. J., Mullany, L. C., Richards, A. K., Kuiper, H. K., Maung, C., & Beyrer, C. (2006b). Mortality rates in conflict zones in karen, karenni, and mon states in eastern burma. Tropical Medicine & International Health : TM & IH, 11(7), 1119-1127.

Lemeshow, S., Tserkovnyi, A. G., Tulloch, J. L., Dowd, J. E., Lwanga, S. K., & Keja, J. (1985). A computer simulation of the EPI survey strategy. International Journal of Epidemiology, 14(3), 473-481.

Mahn, M., Maung, C., Oo, E. K., Smith, L., Lee, C. I., Whichard, E., et al. (2008). Multi-level partnerships to promote health services among internally displaced in eastern burma. Global Public Health, 3(2), 165-186.

Mann, J. (2006). Health and human rights: If not now, when? 1997. American Journal of Public Health, 96(11), 1940-1943.

Marmot, M. (2005). Social determinants of health inequalities. Lancet, 365(9464), 1099-1104.

McDonnell, S. M., Bolton, P., Sunderland, N., Bellows, B., White, M., & Noji, E. (2004). The role of the applied epidemiologist in armed conflict. Emerging Themes in Epidemiology, 1(1), 4.

146

McLeroy, K. R., Bibeau, D., Steckler, A., & Glanz, K. (1988). An ecological perspective on health promotion programs. Health Education Quarterly, 15(4), 351-377.

Mills, E. J., Checchi, F., Orbinski, J. J., Schull, M. J., Burkle, F. M.,Jr, Beyrer, C., et al. (2008). Users' guides to the medical literature: How to use an article about mortality in a humanitarian emergency. Conflict and Health, 2, 9.

Morris, S. K., & Nguyen, C. K. (2008). A review of the cluster survey sampling method in humanitarian emergencies. Public Health Nursing (Boston, Mass.), 25(4), 370-374.

Mosley, W. H., & Chen, L. C. (2003). An analytical framework for the study of child survival in developing countries. 1984. Bulletin of the World Health Organization, 81(2), 140-145.

MSF. (2006). Rapid health assessment of refugee or displaced populationsMSF.

Mullany, L. C., Lee, C. I., Paw, P., Shwe Oo, E. K., Maung, C., Kuiper, H., et al. (2008). The MOM project: Delivering maternal health services among internally displaced populations in eastern burma. Reproductive Health Matters, 16(31), 44-56.

Mullany, L. C., Lee, C. I., Yone, L., Paw, P., Oo, E. K., Maung, C., et al. (2008). Access to essential maternal health interventions and human rights violations among vulnerable communities in eastern burma. PLoS Medicine, 5(12), 1689-1698.

Mullany, L. C., Lee, T. J., Yone, L., Lee, C. I., Teela, K. C., Paw, P., et al. (2010). Impact of community-based maternal health workers on coverage of essential maternal health interventions among internally displaced communities in eastern burma: The MOM project. PLoS Medicine, 7(8), e1000317.

Mullany, L. C., Richards, A. K., Lee, C. I., Suwanvanichkij, V., Maung, C., Mahn, M., et al. (2007). Population-based survey methods to quantify associations between human rights violations and health outcomes among internally displaced persons in eastern burma. Journal of Epidemiology and Community Health, 61(10), 908-914.

Murray, C. J., King, G., Lopez, A. D., Tomijima, N., & Krug, E. G. (2002). Armed conflict as a public health problem. BMJ (Clinical Research Ed.), 324(7333), 346-349.

National Research Council. (2007). Tools and methods for estimating Populations at risk from natural disasters and complex humanitarian crises. Washington, DC: National Academy of the Sciences.

Orach, C. G., Musoba, N., Byamukama, N., Mutambi, R., Aporomon, J. F., Luyombo, A., et al. (2009). Perceptions about human rights, sexual and reproductive health services by internally displaced persons in northern uganda. African Health Sciences, 9 Suppl 2, S72-80.

Partners. Home page. Retrieved Dec 23, 2013, from http://www.partnersworld.org/

Pedersen, D. (2002). Political violence, ethnic conflict, and contemporary wars: Broad implications for health and social well-being. Social Science & Medicine (1982), 55(2), 175-190.

Pham, P. N., Vinck, P., & Stover, E. (2009). Returning home: Forced conscription, reintegration, and mental health status of former abductees of the lord's resistance army in northern uganda. BMC Psychiatry, 9, 23.

147

Pham, P. N., Vinck, P., & Weinstein, H. M. (2010). Human rights, transitional justice, public health and social reconstruction. Social Science & Medicine (1982), 70(1), 98-105.

Phelan, J. C., Link, B. G., & Tehranifar, P. (2010). Social conditions as fundamental causes of health inequalities: Theory, evidence, and policy implications. Journal of Health and Social Behavior, 51 Suppl, S28-40.

Potts, H. (2009). Participation and the right to the highest attainable standard of health. UK: Essex Human Rights Centre.

Potts, A., Myer, K., & Roberts, L. (2011). Measuring human rights violations in a conflict-affected country: Results from a nationwide cluster survey in central african republic. Conflict and Health, 5(1), 4.

PRIO. Armed conflict database. Retrieved Dec 31, 2013, from http://www.prio.no/Data/Armed-Conflict/

Prudhon, C., & Spiegel, P. B. (2007). A review of methodology and analysis of nutrition and mortality surveys conducted in humanitarian emergencies from october 1993 to april 2004. Emerging Themes in Epidemiology, 4, 10.

Purdin, S., Spiegel, P., Mack, K. P., & Millen, J. (2009). Surveillance beyond camp settings in humanitarian emergencies: Findings from the humanitarian health information management working group. Prehospital and Disaster Medicine : The Official Journal of the National Association of EMS Physicians and the World Association for Emergency and Disaster Medicine in Association with the Acute Care Foundation, 24 Suppl 2, s202-5.

Ratnayake, R., Degomme, O., & Guha-Sapir, D. (2009). Coming together to document mortality in conflict situations: Proceedings of a symposium. Conflict and Health, 3, 2.

Roberts, L. (2000). Mortality in eastern DRC results from five mortality surveys. New York: IRC.

Roberts, L., Ngoy, P., Mone, C., Lubula, C., Mwezse, L., Zantop, M., et al. (2003). Mortality in the democratic republic of congo: Results from a nationwide survey. New York: IRC.

Roberts, B., Damundu, E. Y., Lomoro, O., & Sondorp, E. (2010). The influence of demographic characteristics, living conditions, and trauma exposure on the overall health of a conflict-affected population in southern sudan. BMC Public Health, 10, 518-2458-10-518.

Roberts, B., Felix Ocaka, K., Browne, J., Oyok, T., & Sondorp, E. (2009). Factors associated with the health status of internally displaced persons in northern uganda. Journal of Epidemiology and Community Health, 63(3), 227-232.

Roberts, B., Ocaka, K. F., Browne, J., Oyok, T., & Sondorp, E. (2008). Factors associated with posttraumatic stress disorder and depression amongst internally displaced persons in northern uganda. BMC Psychiatry, 8, 38-244X-8-38.

Roberts, L., Lafta, R., Garfield, R., Khudhairi, J., & Burnham, G. (2004). Mortality before and after the 2003 invasion of iraq: Cluster sample survey. Lancet, 364(9448), 1857-1864.

Rose, A. M., Grais, R. F., Coulombier, D., & Ritter, H. (2006). A comparison of cluster and systematic sampling methods for measuring crude mortality. Bulletin of the World Health Organization, 84(4), 290-296.

148

Sabin, M., Lopes Cardozo, B., Nackerud, L., Kaiser, R., & Varese, L. (2003). Factors associated with poor mental health among guatemalan refugees living in mexico 20 years after civil conflict. JAMA : The Journal of the American Medical Association, 290(5), 635-642.

Salama, P., & Roberts, L. (2005). Evidence-based interventions in complex emergencies. Lancet, 365(9474), 1848.

Salama, P., Spiegel, P., Talley, L., & Waldman, R. (2004). Lessons learned from complex emergencies over past decade. Lancet, 364(9447), 1801-1813.

Sanderson, S., Tatt, I. D., & Higgins, J. P. (2007). Tools for assessing quality and susceptibility to bias in observational studies in epidemiology: A systematic review and annotated bibliography. International Journal of Epidemiology, 36(3), 666-676.

Sapolsky, R. M. (2004). Social status and health in humans and other animals. Annual Review of Anthropology, 33, 393–418.

Scholte, W. F., Olff, M., Ventevogel, P., de Vries, G. J., Jansveld, E., Cardozo, B. L., et al. (2004). Mental health symptoms following war and repression in eastern afghanistan. JAMA : The Journal of the American Medical Association, 292(5), 585-593.

Schumann, D. A., & Mosley, W. H. (1994). The household production of health. introduction. Social Science & Medicine (1982), 38(2), 201-204.

SCRIMSHAW, N. S., TAYLOR, C. E., & GORDON, J. E. (1959). Interactions of nutrition and infection. The American Journal of the Medical Sciences, 237(3), 367-403.

Seino, K., Takano, T., Mashal, T., Hemat, S., & Nakamura, K. (2008). Prevalence of and factors influencing posttraumatic stress disorder among mothers of children under five in kabul, afghanistan, after decades of armed conflicts. Health and Quality of Life Outcomes, 6, 29-7525-6-29.

Selth, A. (2002). Burma's armed forces: Power without glory. Eastbridge: Norwalk:.

Shanovich, P. K., Donaldson, R. I., Hung, Y. W., Hasoon, T., & Evans, G. E. (2011). Iraqi community members' knowledge, attitude and practice of emergency medical care: Assessing civilian emergency medicine in an area of conflict. Medicine, Conflict, and Survival, 27(3), 151-164.

Shargie, E. B., Ngondi, J., Graves, P. M., Getachew, A., Hwang, J., Gebre, T., et al. (2010). Rapid increase in ownership and use of long-lasting insecticidal nets and decrease in prevalence of malaria in three regional states of ethiopia (2006-2007). Journal of Tropical Medicine, 2010, 10.1155/2010/750978. Epub 2010 Sep 23.

Siegel, D., Baron, R., & Epstein, P. (1985). The epidemiology of aggression. health consequences of war in nicaragua. Lancet, 1(8444), 1492-1493.

SMART. Standardized monitoring and assessment of relief and transitions. Retrieved Dec 31, 2013, from http://www.smartindicators.org/

Sollom, R., Richards, A. K., Parmar, P., Mullany, L. C., Lian, S. B., Iacopino, V., et al. (2011). Health and human rights in chin state, western burma: A population-based assessment using multistaged household cluster sampling. PLoS Medicine, 8(2), e1001007.

149

South, A. (2011). Burma’s longest running war: Anatomy of the karen conflict (2011),Transnational Institute (TNI) and Burma Center Netherlands (BCN).

South, A., & Harragin, S. (2012). Local to global protection in myanmar (burma), sudan, south sudan and zimbabwe No. 72). London: Overseas Development Institute.

South, A., Perhult Malin, & Carstensen, N. (2010). Conflict and survival: Self-protection in south-east burma. London: Chatham House.

Spagat, M. (2010). Estimating the human costs of war: The sample survey approach. In M. Garfinkel, & S. Skaperdas (Eds.), Oxford handbook of the economics of peace and conflict (pp. 318) Oxford University Press.

Spiegel, P. B. (2007). Who should be undertaking population-based surveys in humanitarian emergencies? Emerging Themes in Epidemiology, 4, 12.

Spiegel, P. B., & Robinson, C. (2010). Large-scale "expert" mortality surveys in conflicts--concerns and recommendations. JAMA : The Journal of the American Medical Association, 304(5), 567-568.

Spiegel, P. B., & Salama, P. (2000). War and mortality in kosovo, 1998-99: An epidemiological testimony. Lancet, 355(9222), 2204-2209.

Stringhini, S., Berkman, L., Dugravot, A., Ferrie, J. E., Marmot, M., Kivimaki, M., et al. (2012). Socioeconomic status, structural and functional measures of social support, and mortality: The british whitehall II cohort study, 1985-2009. American Journal of Epidemiology, 175(12), 1275-1283.

Suwanvanichkij, V. (2008). Displacement and disease: The shan exodus and infectious disease implications for thailand. Conflict and Health, 2, 4.

Tapp, C., Burkle, F. M.,Jr, Wilson, K., Takaro, T., Guyatt, G. H., Amad, H., et al. (2008). Iraq war mortality estimates: A systematic review. Conflict and Health, 2, 1.

Tarantola, D., & Mann, J. (1995). AIDS and human rights. AIDS & Society, 6(4), 1, 5.

TBBC. (2011). Displacement and poverty in southeast burma/ myanmar. Bangkok: Thai Burma Border Consortium.

TBC. What we do. Retrieved Dec 23, 2013, from http://theborderconsortium.org/whatwedo/whatwedo.htm

Teela, K. C., Mullany, L. C., Lee, C. I., Poh, E., Paw, P., Masenior, N., et al. (2009). Community-based delivery of maternal care in conflict-affected areas of eastern burma: Perspectives from lay maternal health workers. Social Science & Medicine (1982), 68(7), 1332-1340.

Thapa, S. B., & Hauff, E. (2012). Perceived needs, self-reported health and disability among displaced persons during an armed conflict in nepal. Social Psychiatry and Psychiatric Epidemiology, 47(4), 589-595.

Tol, W. A., Song, S., & Jordans, M. J. (2013). Annual research review: Resilience and mental health in children and adolescents living in areas of armed conflict--a systematic review of findings in low- and

150

middle-income countries. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 54(4), 445-460.

Trickett, E., & Beehler, S. (2013). The ecology of multilevel interventions to reduce social inequalities in health. American Behavioral Scientist, 57(8), 1227.

Turner, A. G., Magnani, R. J., & Shuaib, M. (1996). A not quite as quick but much cleaner alternative to the expanded programme on immunization (EPI) cluster survey design. International Journal of Epidemiology, 25(1), 198-203.

UN Security Council. (2012). Report of the secretary-general on the protection of civilians in armed conflict No. S/2012/376). Geneva: UN Security Council.

UNGA. (1948). Universal declaration of human rights. New York:

United Nations Administrative Committee on Coordination – Sub-Committee on Nutrition. (1995). Report of a workshop on the improvement of the nutrition of refugees and displaced people in Africa . Machakos, Kenya: UN.

University of Minnesota Human Rights Library. Ratification of international human rights treaties - myanmar. Retrieved July/9, 2012, from http://www1.umn.edu/humanrts/research/ratification- myanmar.html

UNOHCHR. (2011, ). Human rights council estalishes new mandates on promoting an equitable international order and on truth, justice and reparation. Message posted to http://www.ohchr.org/EN/NewsEvents/Pages/DisplayNews.aspx?NewsID=11449&LangID=E

USAID. A medical lifeline on the border. Retrieved Dec 23, 2013, from http://www.usaid.gov/results- data/success-stories/medical-lifeline-border

Vervisch, T. G., Vlassenroot, K., & Braeckman, J. (2013). Livelihoods, power, and food insecurity: Adaptation of social capital portfolios in protracted crises--case study burundi. Disasters, 37(2), 267- 292.

Vinck, P., & Pham, P. N. (2010). Association of exposure to violence and potential traumatic events with self-reported physical and mental health status in the central african republic. JAMA : The Journal of the American Medical Association, 304(5), 544-552.

Vinck, P., Pham, P. N., Stover, E., & Weinstein, H. M. (2007). Exposure to war crimes and implications for peace building in northern uganda. JAMA : The Journal of the American Medical Association, 298(5), 543-554.

VOA. (2009, International centre for trade and sustainable development, labour update: ILO, burma to meet on forced labour. Voice of America News,

Wai Moe. (2011, Naypidaw orders new "four cuts" campaign. Irrawaddy,

WFP. (2005). A manual: Measuring and interpreting malnutrition and mortality. Geveva: WFP.

WHO. (2000). The mamagement of nutrition in major emergencies. Geneva: WHO.

151

WHO. (2011). Global prevalence of vitamin A deficiency in populations at risk 1995–2005WHO Global Database on Vitamin A Deficiency. Geneva, World Health Organization.

Working Group for Mortality Estimation in Emergencies. (2007). Wanted: Studies on mortality estimation methods for humanitarian emergencies, suggestions for future research. Emerging Themes in Epidemiology, 4, 9.

World Bank. (2011). World development report 2011: Conflict security and development. Washington, DC: World Bank.

Yang, X., & Shoptaw, S. (2005). Assessing missing data assumptions in longitudinal studies: An example using a smoking cessation trial. Drug and Alcohol Dependence, 77(3), 213-225.

Zwi, A., & Ugalde, A. (1989). Towards an epidemiology of political violence in the third world. Social Science & Medicine (1982), 28(7), 633-642.

152

Curriculum Vita Bill Davis was born in Lancaster and grew up in Parkesburg, Pennsylvania. He attended London Grove Friends Kindergarten in Kennett Square, PA and graduated from Octorara Area High School in Atglen, PA, with one intervening year spent at Peace River High School in Peace River, Alberta, Canada.

He holds a B.A .in Chemistry from Franklin and Marshall College (he attended one semester at Macquarie University, NSW, Australia), an M.A. in Molecular, Cellular and Developmental Biology from the University of Colorado, Boulder and an MPH from Johns Hopkins University Bloomberg School of Public Health.

Bill’s informal training includes five years’ field experience-- in Tanzania and Uganda with the U.S. Peace Corps, in Burma with Medecins Sans Frontieres (MSF) and in Thailand/Burma with Physicians for Human Rights (PHR).

153