USF Libraries
USF Digital Collections

Measuring culture change as an evaluation indicator

MISSING IMAGE

Material Information

Title:
Measuring culture change as an evaluation indicator applying cultural consensus analysis to cultural models of lymphatic filariasis in Haiti
Physical Description:
Book
Language:
English
Creator:
Simpson, Kelly M
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Cultural consonance
Medical anthropology
Methodology
Public health
Explanatory model
Dissertations, Academic -- Community & Family Health -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Introduction: This project explores the links between shared cultural beliefs in the illness domain, specific to lymphatic filariasis, and a support group program implemented in three Haitian towns. The purpose is to introduce an innovative approach to evaluation, the cultural model evaluation technique (CM Evaluation), as well as gain an understanding of the shifting cognitive belief structure around the cultural domain of lymphatic filariasis in the Haitian setting as associated with a support group intervention. Method: The sample population was comprised of 241 women across three sites in Haiti: Archaie, Cabaret, and La Plaine. Data were collected from longitudinal surveys in 2003, baseline, and 2005, outcome. Descriptive statistics and CM Evaluation were utilized to assess the success of the support group program.CM evaluation is a two-pronged approach, comprised of cultural consensus analysis (CCA) and cultural consonance analysis (CC), that differs from standard evaluation tools in that it measures beliefs and behaviors at the shared community level and is culturally contextualized. Results: At baseline, most participants were not single (59%), Catholic (49%), literate (57%), relatively poor (71%), and engaged in selling at home or the market (46%). In the reduced model longitudinal CM comparisons, intervention and control groups, the intervention group had the highest rate of consensus (ER=4.71), significant changes in the culturally correct answer key (chi-sq=5.1, df=1, p<.02) and cultural competence (t=3.63, p<.0006). Alternately, controls exhibited no significant differences in the culturally correct answer key (Fisher's Exact two-tailed P<1.00) or cultural competence (t=.62, p<.5407) from baseline to outcome.Conclusion: Evidence suggests that support group participation does significantly impact the shared illness beliefs surrounding lymphatic filariasis, and that this format is appropriate for resource poor settings lacking clinical support. Also, this study suggests that the CM evaluation approach is an appropriate and effective evaluation indicator for assessing changes in shared belief, cultural consensus analysis, resulting from public health interventions while the behavioral piece, cultural consonance, requires further refinement.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Kelly M. Simpson.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 303 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002047060
oclc - 497135107
usfldc doi - E14-SFE0002744
usfldc handle - e14.2744
System ID:
SFS0027061:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Measuring Culture Change as an Evaluation Indicator: Applying Cultural Consensus Analysis to Cultural Models of Lymphatic Filariasis in Haiti by Kelly M. Simpson A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Community and Family Health College of Public Health University of South Florida Major Professor: Jeannine Coreil, Ph.D. Julie Baldwin, Ph.D. Carol Bryant, Ph.D. Yiliang Zhu, Ph.D. Sara Green, Ph.D. Date of Approval: November 17, 2008 Key Words: Cultural Consonance, Medical Anthropology, Methodology, Public Health, Explanatory Model Copyright 2008, Kelly Simpson

PAGE 2

To my parents: They have always encouraged me to seek out my passion and, upon finding it, stop at nothing to fulfill it.

PAGE 3

Acknowledgements I thank my parents for all of their support as they have lived and learned the process of earning a PhD with me. Dad, I thank you for working so hard to learn about what I do and support me every step of the way. Mom, thanks for keeping me balanced and reminding me how much a woman can accomplish when she approaches her passion with integrity, intellect, and determination. I appreciate the guidance, insight, support and friendship of my many mentors: Judith Farquhar, you helped me find my way to graduate education, Holly Mathews, you saw me through my MA and suggested I work with my current adviser, Carol Bryant, thank you for sharing your experiences in the fiel d and believing in my abilities to work successfully within the comm unity setting, Yiliang Zhu, th anks for sharing your own PhD story with me and for taking advantage of teachable moments in our interactions, Julie Baldwin, thank you for sharing your know ledge of working with cross-cultural communities and cheering me on, Sara Gree n, you gave me a new perspective on disabilities and always move me forwar d in my development with your thoughtprovoking questions. Josephine Beoku-Betts, thank you for your informal support as I attained my degree your friendship, sugges tions, and perspective have great value to me. Donna Evon, thank you for supporting my effo rt to gain my PhD, taking time out of

PAGE 4

your professional duties to help me grow pr ofessionally, and engaging me fully with the UNC team and our project. Jeannine Coreil, your participation in my Ph D endeavor has been invaluable. Through the long journey, I can honestly say I could not have done it without you. You have acted as my department chair, professor, employer, adviser, and, ultimately, friend. I do not think I could have found a more compatible ment or had I tried. Over time, you have both encouraged and challenged me intellectuall y, professionally, and pe rsonally. I attribute my achievement to both my own work and your investment in me as your student. I also thank my fellow doctoral students for the collaborations, happy hours, and informal support. I also thank my rugby fam ily for supporting my PhD efforts. To my coworkers, thank you for listening to me discu ss my work and being excited with me as I achieved each step along the way. To the U SF COPH staff, thanks for both helping me administratively and developing supportive frie ndships with me. Finally, to my friends and family who have tolerated my now I am here now I am not existence for these past several years thanks for loving me a nd sticking with me through it all.

PAGE 5

i Table of Contents List of Tables iv Abstract xiii Chapter One: Statement of the Problem 1 Introduction 2 Global Burden of Disability 2 Global Burden of Lymphatic Filariasis 3 Lymphatic Filariasis 4 Response to the Lymphatic Filariasis Problem 6 Addressing GPELF Aims through Research 8 Evaluating the Impact of Public Health Programs Targeting Lymphatic Filariasis 9 The Argument for Cultural Models 11 Assessment of Cultural Models 12 The Data for the Current Project 13 Preliminary Analyses 14 Building on the Preliminary Analyses 15 Research Questions and Hypotheses 17 The Relevance of the Data to the Research Questions 19 Significance and Innovation 21 Chapter Two: Review of the Literature 24 Rationale 25 Global Disability 25 Models of Disability 26 Defining Disability 28 Cross-cultural Disability 29 Symbolic Interactionism 34 Support Groups 43 Explanatory and Cultural Models 48 Cultural Consensus and Cultural Consonance 58 Chapter Three: Methods 64 Study Design 65 Description of the Data 68 Context of the Data 71 Variables of Interest 76 Data Quality 82

PAGE 6

ii Analyses 84 Demographics 84 Cultural Models 85 Cultural Consonance 94 Sample Size 96 Chapter Four: Results 99 Integrity of the Data 102 Demographic Analyses 103 Cultural Models at Baseline 117 Cultural Consensus Analysis by Baseline Locale 117 Baseline Cultural Consensus Analysis by Matched Group: Intervention and Control 120 Baseline Cultural Consensus Anal ysis: Intervention and Control Groups by Demographic Characteristic 124 Cultural Models at Outcome 128 Outcome Cultural Consensus Analysis Comparison by Matched Group: Intervention and Control 129 Outcome Cultural Consensus Anal ysis: Intervention and Control Groups by Demographic Characteristics 137 Marital Status 141 Religion 141 Literacy 142 Wealth 142 Age 143 Stage of Disease 143 Number of Acute Attacks 144 Changes in Cultural Competency: Baseline to Outcome 144 Changes in Cultural Models: Baseline to Outcome 146 Strength of Cultural Consen sus: Baseline to Outcome 146 Link between Cultural Consensus and Cultural Consonance: Baseline to Outcome 148 Cultural Consonance Analyses 148 Cultural Consonance Analyses by Demographics 150 Marital Status 151 Religion 151 Literacy 151 Wealth 152 Age 152 Stage of Disease 153 Number of Acute Attacks 153

PAGE 7

iii Chapter Five: Discussion 154 Cultural Models at Baseline 155 Baseline Comparisons 155 Baseline Comparisons by Demographic Characteristics 158 Cultural Models at Outcome 160 Outcome Comparisons: Reduced Models (OMC-RM & OMI-RM) 160 Outcome Comparisons: Full Models (OMC-FM & OMI-FM) 165 Outcome Comparisons by Demographic Dimensions 167 Changes in Cultural Competency: Baseline to Outcome 171 Changes in Cultural Models: Baseline to Outcome 174 Strength of Cultural Consen sus: Baseline to Outcome 177 Link between Cultural Consen sus and Cultural Consonance 178 Cultural Model Evaluation 181 Cultural Consensus Analysis Component 181 Cultural Consonance Component 182 Sample Size and CM Evaluation 183 Implications 186 Implications of the Support Group Program 186 Implications of the CM Evaluation Methodology 188 Strengths and Limitations of the Study 191 Future Research 193 Conclusions 196 References 198 Appendices 213 Appendix A Appendix B Appendix C Appendix D About the Author End Page

PAGE 8

iv List of Tables Table 1 Research Ques tions and Hypotheses 18 Table 2 Examples of Propositions in Symbolic Interactionism 39 Table 3 Support Group Meeting Agenda 46 Table 4 Causes of Lymphatic Filariasis: Traditional Beliefs 55 Table 5 Treatments for Lymphatic Filariasis: Traditional Beliefs 56 Table 6 Linking Research Questi ons to Method of Analysis 66 Table 7 Sample Sizes for Baseline Data 70 Table 8 Sample Sizes for Outcome Data 70 Table 9 Sample Sizes for Matched Data 70 Table 10 Requirements for Potential Facilitators 73 Table 11 Restatement of Support Group Meeting Agenda 75 Table 12 Objectives of the Support Group 76 Table 13 Variables of Interest 78 Table 14 Variables used in Cultural Consensus Analysis 92 Table 15 Sample Size Table for Cultural Consensus Analysis 97 Table 16 Review of Research Questions and Hypotheses 101 Table 17 Demographics: Baseline Data 104 Table 18 Demographics: Outcome Data 106 Table 19 Illness Prof ile: Baseline Data 108 Table 20 Illness Profile: Outcome Data 110

PAGE 9

v Table 21 Illness Beliefs: Baseline Data 112 Table 22 Illness Beliefs: Outcome Data 114 Table 23 Culturally Correct Answer Keys by Town 118 Table 24 Comparisons of Culturall y Correct Answer Keys by Town 119 Table 25 Baseline Sample: Consen sus Analysis Results by Group 120 Table 26 Baseline Sample: Culturally Correct Answer Keys by Intervention and Control Groups 121 Table 27 Baseline Sample: Comparisons of Culturally Correct Answer Keys by Intervention and Control Groups 122 Table 28 Baseline Sample: Cultural Comp etency Analysis Results by Group 123 Table 29 Baseline Sample: Comparisons of Cultural Competency by Intervention and Control Groups 123 Table 30 Key Findings in Baseline Data for Demographic Characteristics 126 Table 31 Significant Differences in Culturally Correct Answer Keys in the Baseline Data for Demographic Characteristics 127 Table 32 Significant Differences in Cultu ral Competency in the Baseline Data for Demographic Characteristics 127 Table 33 Outcome Sample: Consensus Analysis Results by Group 129 Table 34 Outcome Sample: Culturally Correct Answer Keys by In tervention (OMI-RM) and C ontrol (OMC-RM) Group 131 Table 35 Outcome Sample: Culturally Correct Answer Keys by Inte rvention (OMI-FM) and Cont rol (OMC-FM) Groups that Include Additional Cultural Model Items only Present in the Outcome Sample 133 Table 36 Outcome Sample: Comparisons of Culturally Correct Answer Keys by Intervention and Control Groups 135 Table 37 Outcome Sample: Cultural Co mpetency Analysis Results by Group 136

PAGE 10

vi Table 38 Outcome Sample: Comparisons of Cultural Competency by Intervention and Control Groups 136 Table 39 Key Findings in Outcome Data for Demographic Characteristics 138 Table 40 Significant Differences in Cultu rally Correct Answer Keys in the Outcome Data for Demographic Characteristics 140 Table 41 Significant Differences in Cu ltural Competency in the Outcome Data for Demographic Characteristics 140 Table 42 Baseline and Outcome: Cultural Competency Analysis Results by Group 145 Table 43 Baseline to Outcome: Compar isons of Cultural Competency by Intervention and Control Groups 145 Table 44 Baseline to Outcome: Comparis ons of Culturally Correct Answer Keys by Intervention and Control Groups 146 Table 45 Baseline and Outcome: Cons ensus Analysis Results by Group 147 Table 46 Baseline and Outcome: Cultu ral Consonance Analysis Results by Group 149 Table 47 Comparisons of Cultural Consonance by Intervention and Control Groups 149 Table 48 Significant Differences in Cultural Consonance for Demographic Characteristics 150 Table A1 Baseline Sample: Consensus An alysis Results by Group and Marital Status 214 Table A2 Baseline Sample: Culturally Correct Answer Keys by Intervention and Control Groups by Marital Status 215 Table A3 Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Contro l Groups by Marital Status 216 Table A4 Baseline Sample: Cultural Competency Analysis Results by Group and Marital Status 217

PAGE 11

vii Table A5 Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Marital Status 217 Table A6 Baseline Sample: Consensus Analysis Results by Group and Religion 218 Table A7 Baseline Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Religion 220 Table A8 Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Religion 221 Table A9 Baseline Sample: Cultural Comp etency Analysis Results by Group and Religion 221 Table A10 Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Religion 222 Table A11 Baseline Sample: Consensus Anal ysis Results by Group and Literacy 222 Table A12 Baseline Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Literacy 223 Table A13 Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Literacy 224 Table A14 Baseline Sample: Cultural Competency Analysis Results by Group and Literacy 224 Table A15 Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Literacy 225 Table A16 Baseline Sample: Consensus An alysis Results by Group and Wealth 226 Table A17 Baseline Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Wealth 227 Table A18 Baseline Sample: Comparison of Culturally Correct Answer Keys for Intervention and Control Groups by Wealth 228 Table A19 Baseline Sample: Cultural Comp etency Analysis Results by Group and Wealth 229 Table A20 Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Wealth 229

PAGE 12

viii Table A21 Baseline Sample: Consensus Analysis Results by Group and Age 230 Table A22 Baseline Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Age 231 Table A23 Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Age 232 Table A24 Baseline Sample: Cultural Co mpetency Analysis Results by Group and Age 232 Table A25 Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Age 232 Table A26 Baseline Sample: Consensus Analysis Results by Group and Stage of Disease 233 Table A27 Baseline Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Stage of Disease 234 Table A28 Baseline Sample: Comparisons of Culturally Correct Answer Keys for In tervention and Control Groups by Stage of Disease 235 Table A29 Baseline Sample: Cultural Co mpetency Analysis Results by Group and Stage of Disease 235 Table A30 Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Stage of Disease 236 Table A31 Baseline Sample: Consensus Analysis Results by Group and Number of Attacks 236 Table A32 Baseline Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Number of Attacks 238 Table A33 Baseline Sample: Comparison of Culturally Correct Answer Keys for Intervention and Control Groups by Number of Attacks 239 Table A34 Baseline Sample: Cultural Co mpetency Analysis Results by Group and Number of Attacks 239 Table A35 Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Number of Attacks 240

PAGE 13

ix Table B1 Outcome Sample: Consensus Analysis Results by Group and Marital Status 242 Table B2 Outcome Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Marital Status 244 Table B3 Outcome Sample: Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Marital Status 245 Table B4 Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Gr oups by Marital Status 247 Table B5 Outcome Sample: Cultural Co mpetency Analysis Results by Group and Marital Status 248 Table B6 Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Marital Status 248 Table B7 Outcome Sample: Consensus Analysis Results by Group and Religion 250 Table B8 Outcome Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Religion 251 Table B9 Outcome Sample: Culturally Corr ect Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Religion 252 Table B10 Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Religion 254 Table B11 Outcome Sample: Cultural Comp etency Analysis Results by Group and Religion 254 Table B12 Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Religion 255 Table B13 Outcome Sample: Consensus Analysis Results by Group and Literacy 256 Table B14 Outcome Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Literacy 257 Table B15 Outcome Sample: Culturally Corr ect Answer Keys for Intervention

PAGE 14

x and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Literacy 258 Table B16 Outcome Sample: Comparisons of Culturally Correct Answer Keys fo r Intervention and Contro l Groups by Literacy 260 Table B17 Outcome Sample: Cultural Co mpetency Analysis Results by Group and Literacy 261 Table B18 Outcome Sample: Comparisons of Cultural Competency for Intervention and Cont rol Groups by Literacy 261 Table B19 Outcome Sample: Consensus An alysis Results by Group and Wealth 262 Table B20 Outcome Sample: Culturally Corr ect Answer Keys for Intervention and Control Groups by Wealth 263 Table B21 Outcome Sample: Culturally Corr ect Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Wealth 264 Table B22 Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Wealth 266 Table B23 Outcome Sample: Cultural Comp etency Analysis Results by Group and Wealth 267 Table B24 Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Wealth 268 Table B25 Outcome Sample: Consensus Analysis Results by Group and Age 269 Table B26 Outcome Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Age 270 Table B27 Outcome Sample: Culturally Corr ect Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items Present only in the Outcome Sample by Age 271 Table B28 Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Age 273 Table B29 Outcome Sample: Cultural Co mpetency Analysis Results by Group and Age 273

PAGE 15

xi Table B30 Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Age 274 Table B31 Outcome Sample: Consensus Analysis Results by Group and Stage of Disease 275 Table B32 Outcome Sample: Culturally Correct Answer Keys for Intervention and Control Groups by Stage of Disease 276 Table B33 Outcome Sample: Culturally Corr ect Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Stage of Disease 277 Table B34 Outcome Sample: Comparisons of Culturally Correct Answer Keys for In tervention and Control Groups by Stage of Disease 279 Table B35 Outcome Sample: Cultural Co mpetency Analysis Results by Group and Stage of Disease 280 Table B36 Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Stage of Disease 280 Table B37 Outcome Sample: Consensus Analysis Results by Group and Number of Attacks 281 Table B38 Outcome Sample: Culturally Corr ect Answer Keys for Intervention and Control Groups by Number of Attacks 282 Table B39 Outcome Sample: Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Number of Attacks 284 Table B40 Outcome Sample: Comparisons of Culturally Correct Answer Keys for In tervention and Control Groups by Number of Attacks 286 Table B41 Outcome Sample: Cultural Co mpetency Analysis Results by Group and Number of Attacks 286 Table B42 Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Number of Attacks 287 Table C1 Cultural Consonance Results by Group and Marital Status 289

PAGE 16

xii Table C2 Comparisons of Cultural Comp etency for Intervention and Control Groups by Marital Status 290 Table C3 Cultural Consonance Analysis Results by Group and Religion 290 Table C4 Comparisons of Cultural Consonance for Intervention and Control Groups by Religion 291 Table C5 Cultural Consonance Analysis Results by Group and Literacy 292 Table C6 Comparisons of Cultural Consonance for Intervention and Control Groups by Literacy 293 Table C7 Cultural Consonance Analys is Results by Group and Wealth 293 Table C8 Comparisons of Cultural Consonance for Intervention and Control Groups by Wealth 294 Table C9 Cultural Competency Analysis Results by Group and Age 294 Table C10 Comparisons of Cultural Cons onance for Intervention and Control Groups by Age 295 Table C11 Cultural Competency Analys is Results by Groups and Stage of Disease 296 Table C12 Comparisons of Cultural Co mpetency for Intervention and Control Groups by Stage of Disease 296 Table C13 Cultural Consonance Analysis Results by Group and Number of Attacks 297 Table C14 Comparisons of Cultural Cons onance for Intervention and Control Groups by Number of Attacks 298

PAGE 17

xiii Measuring Culture Change as an Evalua tion Indicator: Applying Cultural Consensus Analysis to Cultural Models of Lymphatic Filariasis in Haiti Kelly M. Simpson ABSTRACT Introduction: This project explores the links be tween shared cultural be liefs in the illness domain, specific to lymphatic filariasis, and a support group program implemented in three Haitian towns. The purpose is to intr oduce an innovative approach to evaluation, the cultural model evaluation technique (CM Evaluation), as well as gain an understanding of the shifting cognitive belief structure around the cultural domain of lymphatic filariasis in the Haitian setting as associated with a suppor t group intervention. Method: The sample population was comprised of 241 women across th ree sites in Haiti: Archaie, Cabaret, and La Plaine. Data were collected from longitudinal surveys in 2003, baseline, and 2005, outcome. Descriptive statis tics and CM Evaluation were utilized to assess the success of the s upport group program. CM evaluation is a two-pronged approach, comprised of cultural consensu s analysis (CCA) and cultural consonance analysis (CC), that differs from standard ev aluation tools in that it measures beliefs and behaviors at the shared community leve l and is culturally contextualized.

PAGE 18

xiv Results: At baseline, most participants were not single (59%), Catho lic (49%), literate (57%), relatively poor (71%), and engaged in selling at home or the market (46%). In the reduced model longitudinal CM comparis ons, intervention and control groups, the intervention group had the highest rate of consensus (ER=4.71), significant changes in the culturally correct answer key (chi-s q=5.1, df=1, p<.02) and cultural competence (t=3.63, p<.0006). Alternately, controls exhi bited no significant differences in the culturally correct answer key (Fishers Exact two-tailed p<1.00) or cultural competence (t=.62, p<.5407) from baseline to outcome. Conclusion: Evidence suggests that support group participati on does significantly impact the shared illness beliefs surrounding lymphatic filariasis, and that this format is appropriate for resource poor settings lacki ng clinical support. Al so, this study suggests that the CM evaluation approach is an appr opriate and effective evaluation indicator for assessing changes in shared belief, cultura l consensus analysis, resulting from public health interventions while the behavioral piece, cultural consonance, requires further refinement.

PAGE 19

1 Chapter 1: Statement of the Problem

PAGE 20

2 Introduction This project explores the links between shared cultural beliefs in the illness domain, specific to lymphatic filariasis, and a support group program implemented in three Haitian towns: Archaie, Cabaret, and La Plaine. The purpose is to introduce an innovative approach to evaluation, the cultural model evaluation technique (CM Evaluation), as well as to gain an understand ing of the shifting cogni tive belief structure around lymphatic filariasis in the Haitian se tting as associated with a support group intervention. CM Evaluation is a two-pronged approach, comprised of cultural consensus analysis (CCA) and cultural consonance anal ysis (CC), that differs from standard evaluation tools in that it measures beliefs and behaviors at the sh ared community level and is culturally contextualized. The Haitian setting and lymphatic filariasis were identified as appropriate areas of investigation as multiple illness belief syst ems exist in tandem in Haiti. Also, lymphatic filariasis, a parasitic condition, is an infectious disease that causes long-term, irreversible disability. Finally, this illness is preval ent on both the global and local stage. Global Burden of Disability As of 2002, it is estimated that 600 mill ion people have a disability globally (WHO, 2005). Eighty percent of these individuals are livi ng in low income countries (WHO, 2006a), and this number, 600 milli on, is probably underreported as many

PAGE 21

3 disabled individuals do not ev er intersect with the formal health care system. The numbers of people experiencing disabilities worldwide is staggering, and much work must be done to improve the quality of life, standard of care, and human rights afforded this population. Global Burden of Lymphatic Filariasis In this project, Lymphatic Filariasis (LF) is the disabling condition examined. This disease is of great con cern globally as the World Heal th Organization reports that more than a billion people are at risk for the condition worldwide (WHO, 2000). Of these, 120 million people are impacted by LF (WHO, 2000). This number is comprised of individuals who are positive for microfilaria only, positive for adult worms and no microfilaria, and positive for both adult wo rms and microfilaria (P. Lammie, personal communication, June 19, 2007). Approximately one third, or 40 million, of the people impacted by this disease have clinical manifestations of LF disease in the form of hydrocele (25 million people) or lymphoe dema (15 million people) (WHO, 2007b; WHO, 2000). This disease is currently clas sified as the seco nd leading cause of permanent and long-term disability in the world (Ahorlu et al., 1999, 252).

PAGE 22

4 Lymphatic Filariasis Lymphatic filariasis is a parasitic condition. Specifically, it is caused by nematodes, or roundworms, that live in the lymph system and subcutaneous tissues of afflicted individuals (Parker & Parker, 2002). People contract this condition through mosquito bites (Parker & Parker, 2002; CDC DP D, 2004); an infected individual is bitten by a mosquito, and the larvae from this pers on are ingested by the mosquito (Parker & Parker, 2002). At this point, the mosquito is now able to spread LF to others who are uninfected. Once a person has contracted LF, the e ffects can be devastating. Symptoms include lymphedoema, or fluid retention and sw elling, in the arms, breasts, legs, or male genital region (Parker & Pa rker, 2002; CDC DPD, 2004). Also infected individuals may experience difficulty in warding off infecti ons (Parker & Parker, 2002; CDC DPD, 2004). Multiple infections in the skin and lymph syst em can lead to elephantiasis, or hardening of the skin (Parker & Parker, 2002; CDC DP D, 2004). Other possible symptoms include lymphangitis, lymphadenitis, pulmonary tropical eosinophilia syndrome, pruritis, dermatitis, onchocercomata, lymphadenopathies, and ocular lesions (Parker & Parker, 2002). Finally, the filariae can cause internal damage to the kidneys and the lymphatic system (WHO, 2000). An acute attack of adenolymphangitis is one secondary condition that often results in LF patients. These attacks are id entified with acute onset of fever and with localized pain and warmth, with or without sw elling or redness, in th e limb and/or genital

PAGE 23

5 region (WHO, 2006c, 377). Mu ltiple acute attacks c ontribute to worsening lymphedoema symptoms over time. As LF progresses, the disabling effect on afflicted individu als is profound. In biomedical terms, these effects can take th e form of pain, disfigurement, and sexual disability (Parker & Parker, 2002, 12; CDC DP D, 2004; Coreil, Mayard, Louis-Charles, & Addiss, 1998; Person et al., 2007 ). Social consequences of th is diagnosis exist as well. These problems include stigma, social isolati on, difficulty finding a marriage partner, and difficulty working (Parker & Parker, 2002; CDC DPD, 2004; WHO, 2000 LF Fact Sheet; Coreil et al., 1998; Person et al., 2007). Indi viduals infected with LF who are unable to find work, or are limited in their capacity to work, can also be impacted economically as a result. Treatment for people with LF is multi-faceted. First, one should take a yearly dose of medicine that kills the microscopic worms circulating in your blood (Parker & Parker, 2002, 13). These medicines are diethy lcarbamazine and ivermectin (Parker & Parker, 2002), and are % effective in re moving microfilariae from the blood for a full year after treatment (WHO, 2000 LF Fact Sh eet). Drug treatment does not eradicate all the adult worms, but the medicine does prev ent transmission of the disease from an infected individual to others (Par ker & Parker, 2002; CDC DPD, 2004). In addition to medication, other treatments that help mitigate symptoms of LF exist. These treatments include: washing swollen areas daily with soap and water, using anti-bacterial cream on existing wounds, elev ating swollen areas, and exercising swollen arms or legs (Parker & Parker, 2002; CDC DPD, 2004). These efforts, in addition to

PAGE 24

6 using protective footwear, can reduce the number of bacterial infections, prevent additional swelling, and improve lymph flow (Parker & Parker, 2002). As infections are minimized through these self-care behavi ors, the risk of acute attacks of adenolymphangitis, and, ultimately, wors ening lymphedoema, is also reduced. Response to the Lymphatic Filariasis Problem As lymphatic filariasis is a serious co ndition and also a prob lem that is both preventable and eradicable, prevention is necessary on multiple levels. In 1997, the World Health Organization passed re solution WHA50.29 which called for the elimination of lymphatic filari asis, one of only six potenti ally eradicable infectious diseases (WHO, 1997). Out of this resolution, the global progr am for the elimination of lymphatic filariasis was formed (GPELF). GPELF calls for the eradication of lym phatic filariasis by the year 2020 (WHO, 2004), and the program aims to achieve this goal in two ways. These two approaches involve interrupting the transmission of LF and preventing disability (WHO, 2004). In order to achieve the first aim, mass drug administration (MDA) is endorsed (WHO, 2004). This MDA utilizes the same approach as drug treatment for individuals who already have LF. By administ ering a yearly dose of the appropriate medicines, any microscopic worms present in an individual ar e eradicated before the onset of LF. When this process is completed several years in a row, the transmission of MDA is effectively curtailed. Annually eliminati ng the microfilaraie prevents the development of adult

PAGE 25

7 worms within an individual who does not curr ently have LF. In an individual with LF, this treatment aids in preven ting transmission to others. To date, GPELF has successfully implemented MDA programs in 42 countries (WHO, 2007). In 2005 alone, 146 million peopl e worldwide received drug treatment (WHO, 2006b). Of these treated indivi duals, 1,255,476 were Haitians (WHO, 2006b). Though MDA is much needed, it is not comprehe nsive enough in and of itself to address all the concerns related to LF. MDA has so me impact on the quali ty of life (QOL) for people who are already living with lymphatic filariasis, but th ere are other measures that focus more exclusively on QOL dime nsions of the LF experience. With regard to the second goal, prev enting increased disability and improving QOL for people already afflicted with lympha tic filariasis, GPELF focuses on community home-based self care and access to surgery for individuals with hydrocele (WHO, 2004). Hydrocele is one presentation th at male LF patients can experience. This problem occurs when fluid accumulates in the scrotum of the affected individual (Ahorlu, Dunyo, Asamoah, & Simonsen, 2001). In addition, support groups and education are tertiary prevention measures that are aimed at improvi ng morbidity control associated with an LF diagnosis. The self-care treatments outlined above are also included at this level of prevention.

PAGE 26

8 Addressing GPELF Aims Through Research In response to the aims of GPELF, inte rrupting transmission a nd the prevention of disability (WHO, 2004), recent research ha s focused on the dimension of support groups, and the impact these groups have on the pr evention of disability in the LF-infected population. Particularly relevant to this project is the Leogane project, identified as such because it was conducted in Leogane, Haiti (C oreil, Mayard, & Addiss, 2003). This project worked to address the aims set out by GPELF and provides a solid foundation for the current project to build on. Coreil et al. (2003) tested a tertiary preventive approach of a support group intervention in the community of Leogane between 1998 and 2002. Tertiary preventive efforts involve those efforts to help su stain maximal functional and psychological capacity despite the presence of both the dis ease, such as hypertension, and its outcomes, heart disease, stroke, or kidney failure (Mann, 1997, 7). In the case of LF and support groups, the focus is on the prevention of sec ondary conditions resulting from the primary diagnosis and quality of life concerns. Coreil et al.s study was conducted in part nership with Hpital Ste. Croix LF program. This program commenced in the earl y 1990s and was a pioneer in the western hemisphere, after Brazil, in the development of physical therapy demonstration projects for LF. Currently, Hpital Ste. Croix has beco me the Haitis referral site for LF. The collaborative work aimed to assess the appl icability of the chronic disease support group model within a developing country setti ng (Coreil, Mayard, & Addiss, 2002).

PAGE 27

9 Specifically, the project was designed to discern influences on support group participation and to identify the impact of support group membership on individual management of LF (Coreil et al., 2003). In Coreil et al.s work (2003), a longitudinal, matched control group design was implemented, and data were analyzed both qua litatively and quantitat ively. Bivariate and logistic regression analyses indicated that support group members experienced benefits from membership including increased and more accurate knowledge about LF, more regular involvement with home care practices for affected legs, a nd improved quality of life. Other principle findings include that the intervention was cost-effective, inspired enthusiasm, and had high rates of participa tion from individuals invo lved in the support group (Coreil et al., 2002). Evaluating the Impact of the Public Health Programs Targeting Lymphatic Filriasis As research endeavors work to combat the global LF problem, it is important to evaluate the impact of these efforts. Tradit ional evaluation measures target the following dimensions: knowledge, attitudes, be haviors and physiological functioning (McDermott & Sarvela, 1999). MDA programs and their outcomes can be measured directly, physiologically, by a ssessing whether or not micr ofilaria rates below 1% are achieved. This level of success is considered th e threshold at which transmission of LF is effectively interrupted in a community (WHO, 2006d).

PAGE 28

10 It is a more challenging task to achie ve accurate measurements of knowledge, attitudes, and behaviors. In th is project, changes in knowledge are of particular interest. Typically, knowledge is framed in terms of knowledge gained as a result of participating in a health education program (McDermott & Sarvela, 1999). This content knowledge is regularly assessed with selected-response and constructed-respons e items. McDermott and Sarvela (1999) explain that constructed-response items requi re test takers to develop their own answers to questions and selected-response items ask test takers to choose from among an array of possi ble answers to questions. These standard approaches were utilized in Coreil et al.s (2003) study as changes in knowledge, attitudes, and behaviors were targeted. The current pr oject takes evaluation a step further; it aims to broaden the ev aluation toolkit by looki ng at changes in the cultural knowledge of a community, utiliz ing a grounded approach, in addition to changes in biomedical knowledge, identified a priori, that result from a public health intervention. In order to achieve this goal, a new health indicator is introduced to the evaluation regimen, the cultural model (CM). Cultural models of illness are collective understandings of an illness in a community shared within a social group. More formally, cultural models are schemas about a domain that are shared by members of a group having shared problems, shared task solutions and similar life expe riences (Bradway & Barg, 2006). Simply, the aim is to understand the extent to which a support group model intervention can influence shifts in the cu ltural models, and associated behaviors, embraced by a population around the il lness lymphatic filariasis.

PAGE 29

11 The Argument for Cultural Models Cultural models are schemas about a dom ain that are shared by members of a group having shared problems, shared task solutions, and simila r life experiences (Bradway & Barg, 2006); symbolic interactioni sm, as a theoretical framework, aids in understanding how cultural mode ls are identified. This theory focuses on the acquisition and generation of meaning, and it puts forth that meanings are embedded and constructed in our interactions with ot her people and institutions (White & Klein, 2002). These meanings are often taken for-granted, and in terpretation is important (White & Klein, 2002). A clear way of thinking about this theory is that what humans define as real has real consequences (White & Klein, 2002, 60). Thus, if a sample population believes that deviant social behavior causes LF then that belief yields real consequences in the form of understanding the condition, how it is contracted, what the symptoms are, how it is treated, and how it is transmitted. For example, individuals with lymphatic filariasis may receive social sanctions as a result of these understandings in the form of stigma or discrimination. Symbolic interactionism informs an unde rstanding of the im portance of cultural models in assessing public he alth interventions. Cultura l models are a product of ecological processes, interactions on multiple levels. These levels of interaction include intrapersonal, interpersonal, social groups, a nd societal institutional experiences. In each instance, there are judgments made regarding wh at is important to discuss about a topic,

PAGE 30

12 what is accurate information, and what are appropriate cour ses of action in a specific situation. The support group setting examined here provides a microcosm of these multilevel interactions as participants interact w ith other participants, facilitators, animators, the program director, the support group as a whole entity, the local community, and other support groups from surrounding areas. These processes are dynamic in nature, and, as a result, so are the cultural models embraced by a community. As new information or ways of thinking are introduced into a community, or support group, ideas about what is real, important, or accurate information may shift. Norms adjust to incorporate the new information, and the overall cultural model may shift. What the community perceives as real has a real c onsequence; thus, it is argued that the community will adapt self-car e behaviors and attitudes to accommodate this new way of thinking about a condition, in this case LF. Assessment of Cultural Models This research aims to understand how cultural models shif t within a study population from the baseline to outcome poi nts in a support group intervention; a prepost design is implemented. A specialized statis tical analysis was utilized in conjunction with commonly used statistical approaches to achieve this goal. Specifically, cultural consensus analysis was utilized. Romney ( 1999, S103) explains that cultural consensus theory helps describe and measure the extent to which cultural beliefs are shared, and the central idea is to look at the use of the pattern of agreement or consensus among

PAGE 31

13 informants to make inferences in their differential competence in knowledge of the shared information pool constituting cu lture (Romney, Weller, & Batchelder, 1986, 316). Thus, examining cultural models is in line with cultural consensus theory. This type of quantitative an alysis, cultural consensus analysis, aims to deal with two problems: first how can the cultura l knowledge of different informants be estimated, and, second, how can th e correct answers to specif ic questions be inferred and with what degree of conf idence (Romney et al, 1986, 88). In this case, the first aim is of interest as the goal is to better u nderstand the cultural k nowledge in the target population. The Data for the Current Project After Coreil et al. (2003) completed the initial support group study, Gladys Mayard, project director, replicated the pr oject in another region of Haiti that had no exposure to LF educational or clinical programs. This appl ication of the model, grantfunded by Presbyterian Church USA through the Womens Birthday Fund, was designed to evaluate the effectiveness of a community -based educational in tervention both in an abbreviated format and in the absence of a clinical treatment program. A successful outcome would increase the feasibility of th is support group model as a low cost and effective option for rural townships lacking a clinic affiliation. In practice, the support groups more closely replicated the Leogane study conducted by Coreil et al. (2003) as it was not implemented in an abbreviated format.

PAGE 32

14 However, there were four key differences. Th e project was conducted in three locales, La Plaine, Archaie, and Cabaret, and came to be known as the Archaie Project. Also, it was implemented without the suppor t of a local medical clinic In addition, vocational or skills training was not include d in the support group format. Finally, the Archaie Project included both control and intervention gr oups that completed both a baseline and outcome survey. The three sites identified in the Archaie project were chosen for several reasons. First, a national filariasis survey indicat ed each of these towns experience a high prevalence of LF disease (Beau de Rochars et al., 2004). In addition, each of these sites had no previous exposure to clinical LF treatment programs (J. Coreil, personal communication, April 10, 2008). Lastly, the national program was interested in expanding a clinical treatment program into Archaie and was interested in the baseline data produced as part of the Archaie projec t (J. Coreil, personal communication, April 10, 2008). Preliminary Analyses Kanda (2004, viii) conducted a preliminary an alysis on the baseline data from the Archaie project. In these analyses, Kanda re ported on the issues of morbidity control and QOL [quality of life] among lymphedoema patients due to lymphatic filariasis in three rural Haiti towns. In his work, The Quality of Life among Lymphedema Patients Due to Lymphatic Filariasis in Three Rura l Towns in Haiti, Kanda (2004) found that

PAGE 33

15 regional differences in the three towns sampled were significant. Particularly, the way in which the illness was understood varied among those respondents who regularly interacted with western ideas of medicine (more urban popul ation) versus those who did not (more rural population) (Kanda, 2004). Areas that were more rural in nature tended towards understanding LF more along traditio nal cultural and spiritual dimensions and less along the lines of western medica l thought (Kanda, 2004). These varying perceptions of LF influenced utilization of medi cal services, types of services sought after, the likelihood to engage in self-care practices, and leve ls of self-efficacy related to self-care practices (Kanda, 2004). Though this research yielded multiple findings, the outcome most relevant to the current projec t includes the regional differences in beliefs surrounding lymphatic filariasis. Building on the Preliminary Analyses Coreil et al. (2003) and Ka nda (2004) both make important contributions to the study of lymphatic filariasis among Haitian women. This research builds on their work as both the questions and research design vary in important ways. The Leogane Project (Coreil et al., 2003) examines shifts in biomedical knowledge th at result from support groups, is linked with hospital clinic serv ices, was conducted in one geographic region (Leogane), and involved longitudinal mon itoring and a case control study (matched controls). The Archaie Project employed a quasi-experimental, non-equivalent control group design, and these data differ in several wa ys from those data collected in Coreils

PAGE 34

16 work. First, the data were collected from multip le locations (Archaie, Cabaret, La Plaine). Also, these support groups were implemented in areas that were not associated with local, specialized LF hospital clinic services. This wholly community-based approach utilizes lay health advisors, and, in educating and empowering community members to be leaders, builds community capacity. Finally, the research questions of interest in this secondary analysis focus on changes in cultur al knowledge in cont rast to biomedical knowledge. This work also varies from Kandas ( 2004) preliminary analyses of the Archaie Project baseline data. Kanda (2004) focuses on the intersection of quality of life and morbidity control issues in the LF population. He utili zed a cross-sectional design, produced static findings, primarily focused on behavioral, physiological, and mental health measures (EuroQol and CES-D). The current project employs a pre-post design, identifies changes over time, and the primary research questions are distinctly different with an emphasis on cultural models, assesse d both cognitively and behaviorally through cultural consensus and cultural consonance analyses, respectively. These research and design differences o ffer some clear advantages. Firstly, the data from the Archaie Project research a llows for increased ge neralizability of the findings to support groups in Haiti. As it invo lves multiple sites, it is less likely that overall findings are specific to the circumstances present in one locale and more likely to be representative of the Haitian population as a whole. Also, these findings have broader implications for the feasibility of implem enting support groups in a wider range of geographic areas. As these support groups ar e wholly community-b ased, the number of

PAGE 35

17 sites that can be effectively impacted by the support group model increases as clinic affiliation is not an essential element for su ccess. Finally, a focus on changes in cultural knowledge offers a new way to evalua te the success of a program. Research Questions and Hypotheses This research project is expl oratory in nature as the us e of cultural models as an evaluation measure is a new one. Thus, there ar e several key research questions that arise with regards to the data set (Table 1). They are as follows:

PAGE 36

18 Table 1. Research Questions and Hypotheses Research Question Hypothesis Does a cultural model exist in the sample population at baseline? If so, does the CM meet the threshold for cultural consensus at baseline? If so, what is the strength of the consensus? It is expected that a shared cultural model of LF will exist at baseline. Does a cultural model exist in the sample population at outcome? If so, does the model meet the threshold for cultural consensus at outcome? If so, what is the strength of the consensus? It is expected that a shared cultural model of LF will exist at outcome. Are there significant changes in levels of cultural competency existing at baseline and outcome? It is expected that levels of cultural competency will be greater at outcome than baseline. Are there significant differences in the elements included in the CMs from baseline and outcome points? Is the cultural model present at baseline different from the cultural model present at outcome? It is expected that the CM at outcome will be significantly different than the CM present at baseline. The outcome CM is expected to include more elements of western biomedical beliefs than the CM at baseline. It is expected that a bicultural model will exist at outcome incorporating elements of traditional and western biomedical ideas about LF. Is the strength of cultural consensus for the CM greater at outcome than baseline? It is expected that the strength of consensus for the CM at outcome will be greater than the strength of consensus for the CM at baseline. Does a significant link between belief (cultural model) and behavior (cultural consonance) exist in the sample population? It is expected that greater consensus regarding cultural models will be linked to higher rates of self-care behaviors identified in the cultural model.

PAGE 37

19 These research questions and hypothese s provide a framework for understanding whether or not cultural models related to lymphatic filariasis exist within the Haitian population, the strength of these shared models if they exist, and the degree to which these cognitive frameworks can be impacted by a public health intervention. Beyond the findings in aggregate, the results can also be examined along demographic measures. In this way, it is possible to determine di fferential changes among subgroups in the population. As this work is completed, several specific aims will be accomplished. This project allows for an innovative demonstrat ion of a new indicator CM evaluation, for evaluating health outcomes. Also, levels of cultural consensus for cultural models and cultural consonance related to lymphatic filari asis will be examined. Finally, regardless of the findings in this project, this study in troduces methodological a dvances in the areas of cultural consensus analys is, cultural consonance, and public health program evaluation. The Relevance of the Data to the Research Questions In order to address the re search questions, hypotheses, and aims, it is important that the data are appropriate. These data are diachronic in nature and are able to capture changes over time. Utilizing both the baseline a nd outcome data, it is possible to examine the shifts in cultural knowledge central to the project. Also, the data include questions that capture the necessary content informa tion required to examine cultural models. In

PAGE 38

20 addition, there is both an intervention a nd control group sample. Comparisons between the two groups over time are possible. This desi gn also controls for historical biases or a natural evolution of CMs that could be pres ent. Finally, these type s of data are difficult to obtain in a resource poor set ting. Thus, these data are appropriate for the current work. The data were collected in three rural towns in Haiti: Archaie, Cabaret, and La Plaine. Data were gathered at baseline a nd post-intervention point s to better understand illness knowledge, self-care pract ices, symptomology, and acute attacks. The baseline data were collected in September 2003, and the outcome data were obtained between May 2004 and May 2005. A total sample size of 241 participants was attained at baseline, after the exclusion of men and individuals indi cating they were less than 18 years of age at baseline, and the sample sizes were 88, 50, and 103, for Archaie, Cabaret, and La Plaine, respectively. In the outcome data, the design involved an n=100 for both the intervention and control groups, and 87 of these subjects were positively matched with their baseline counterparts: Archaie, 45, La Plaine, 22, and Cabaret, 20. The matching process was necessitated in order to appropr iately examine changes in cu ltural models over time. Of the 100 participants originally in the inte rvention and control gr oups, 60 and 27 people, respectively, were positively matched for this portion of the analys is. In the matched intervention group, Archaie, Cabaret, and La Plaine samples were comprised of 39, 20, and 1 participants, respectively. Likewise, the matched control group was distributed across all three sites as Arch aie included 6, Cabaret 2, a nd La Plaine 19 individuals.

PAGE 39

21 In comparison to standard evaluation tool s, the CM evaluation tool has important new information to convey. First, the curre ntly accepted evaluation methods typically rely on individual outcome measures, not group culture change. This research addresses this gap and proposes CM evaluation as a m easurement technique to assess for this group culture change. The CM evaluation examines sh ifts in ways of understanding an illness. Cultural models, examined through cultural c onsensus analysis and cultural consonance have yet to be used in an evaluation contex t and could provide usef ul insights as to how different cultural communities respond to publ ic health interventi ons. Essentially, the CM evaluation allows for a quantitative examin ation of cultural change within a patient community. Significance and Innovation The findings of this project provide some insight as to the impact a public health program can have on the health-related cu ltural models embraced by a community. If significant changes in cultural models and c onsonance have taken place, the kinds of shifts that occur are of great import to understanding the link between cognitive models and behavioral outcomes. For instance, are cu ltural models resulting after the community support group intervention more in line with biomedical beliefs? Cultural models are produced through social interaction. These ways of understanding illness have very real implica tions in how individuals perceive their condition and seek treatment for it. Insights related to the shift in understanding that

PAGE 40

22 result from an intervention yield multiple im plications. For instance, if future research shows a link between cultural models that have shifted, as a re sult of public health programs, and outcomes such as more help -seeking behavior, more acceptance by the community, more adherence to treatment, fewer secondary conditions, and a better quality of life, it is possible that the CM evaluation could cont ribute in a highly meaningful way. This type of information co uld also aid in culturally tailoring future interventions for a specific population. The proposed research offers an innovative approach to evaluating the impact of public health interventions. Traditionally, know ledge and behavior have been assessed to determine the success or failure of a public health program. This project works to introduce a new health indicator, the CM evaluation, to the battery of traditional evaluation measures. The cultural model approach to evaluation allows for three ke y differences from traditional evaluation measures. The first of these differences is that the CM evaluation approach is sensitive to cross-cultural diffe rences and can assess the impact of a public health approach on cultural beliefs. Stated differently, the cultural model evaluation can measure the cultural impact of a program. Secondly, the cultural model evaluation allows for a more nuanced understanding of the success or failure of an intervention. For instance, significant changes may take place within the cultural consensus around LF models. However, if these changes do not immediately manifest into behavioral change, it is possible traditional impact evaluation instruments would miss these changes in its examination of biomedical knowledge and

PAGE 41

23 behavior. The project could be deemed a failu re, when, in fact, important outcomes have been achieved in the community of interest. Also, although it is necessary to explore further in future research, it is expected that when a program exhibits strong shifts in behavior, biomedical knowledge, and shared cu ltural models there will also be a higher instance of sustainability for the program goals, or a link to positive outcome evaluation results. Thirdly, this approach allows for a m easure of collective knowledge within a group and makes a distinction be tween this type of knowledge and that present at the individual level. In tradit ional public health work, populat ion change is indirectly measured through aggregate individual level da ta in a sample. This approach involves a group-level indicator and direc tly measures culture change that occurs in response to public health efforts. Finally, in addition to providing an added dimension to evaluation, this project is also methodologically important. Examining cultu ral models as an evaluation tool is a new application for cultural consensus and cu ltural consonance modeling; this project expands cultural consensus and cultural c onsonance analyses beyond a purely descriptive method. Also, preand postdata have never been examined through the cultural consensus approach.

PAGE 42

24 Chapter 2: Review of the Literature

PAGE 43

25 Rationale In order to inform this research project appropriately, it is important to review four key theoretical frames: models of disability, symbolic interactionism, support groups, cultural consensus and consonance anal ysis. The two major perspectives in disability studies, biomedical and social, and literature addr essing cross-cultural disability specifically will be reviewed. Next, the symbo lic interactionism framework is discussed as it is the primary theoretical perspective utilized to understand cross-cultural disability. Support groups provide a partic ular context for examining the constructed reality of cross-cultural disability; t hus, additional information on support groups is warranted. Finally, cultural consensus and consonance a pproaches provide an avenue for examining constructed ideas about LF, a disabil ity, within the support group format, and, accordingly, will be described as part of th e theoretical framing of this project. Global Disability Disability is clearly a widespread probl em as it is estimated that, globally, 600 million people have a disability (National Council on Disability, 2002). In 1988, 9.5% of children between 2 and 9 years of age in Ja maica experienced a disability (Thorburn, 1999). Also, in Haiti, 800,000 of the total population experience a disabling condition; this rate indicates that approxi mately 1 out of every 10 people in Haiti claims a disability (Bigelow et al., 2004).

PAGE 44

26 Models of Disability In this research, the socially constructed cultural model of lymphatic filariasis is investigated. The way that disability in ge neral is framed in a society impacts the negotiated social meanings of specific diseas es. As cultural definitions of disability intersect with the construction of cultural models, it is important to review the literature on models of disability and the constr uction of definitions of disability. Historically, two major models of disabil ity have existed: the medical model and the social model. The United States and th e United Kingdom have the richest history of disability research globally, and these countries initially embraced the medical model and social model, respectively. The United States ut ilized the medical model of disability to frame disability for over a hundred years (Scotc h, 2000). In this model, disability is considered an individual problem as a person is labeled with a partic ular disability label (Jeon, 2001). The etiology of a disability is considered bi ochemical or biological in nature, and this model draws upon Cartesian dua lity (Marks, 1999). Thus, disability is considered a chronic functional incap acity whose consequence was functional limitations assumed to result from physical or mental impairment (Scotch, 2000, 214). Superstitious beliefs and holistic approaches to health are disc ounted in the medical model of disability (Marks, 1999). Treatme nt and action focus on terminating the biological basis or concern cau sing the disability, and, if th ese measures are successful, disabling conditions are eradicated (Marks, 1999). For example, in Zieglers work, epilepsy is considered a medical condition (Ziegler, 1982). In this case, if the

PAGE 45

27 biochemical reactions that cause epileptic seizures are stymied with medication, the disability effectively ceases to exist. In contrast, the focus of the social model of disability shifted from the individual to oppressive environments (Jeon, 2001). Included under this umbrella are social structures, attitudes, and other influences th at create disabilities (Marks, 1999). Some other social influences on di sability include class, racism, age, gender, and sexuality (Marks, 1999). Additionally, an emphasis exists on minority status and discrimination issues (Scotch, 2000). Within this framew ork, an impairment only becomes disabling because of social structures and organizati on (Marks, 1999, 77). A disability lies not within an individual but within society a nd is defined as the loss or limitation of opportunities to take part in the normal life of the community on an equal level with others due to physical and so cial barriers (Barnes, 1994, Cited in Marks, 1999, 80). Social norms influence how people cope with impairments and effect when and where these impairments become disabling. An example of this concept in action includes the disabling environment created when there is a lack of ramps for wheelchair users (Marks, 1999). In this instance, wh eelchair users were able to navigate independently as long as ramps existed; withou t the ramps the ability to fully function in society without aid was minimized. The cultural model concept fits nicely w ithin the social model of disability. Negotiated meanings around lymphatic filariasis the disabling condition investigated in this work, are a reflection of the interplay between multiple levels of influence. These shared cultural ideas about lymphatic filariasis may impact the extent to which people are

PAGE 46

28 included or excluded from social participation, affect the de velopment of legislation on disability issues, and ultimately have some bearing on the built environment as well. The degree to which manifestations of these cultur al beliefs occur in one direction or other shapes the degree to which a lymphatic fila riasis diagnosis is a disabling condition. Defining Disability The broad context of how a disability is conceptualized, either the medical or social model, exerts a force on the creation of disability definitions. For instance, in the US, the definition of a disability is as follows : (A) A physical or mental impairment that substantially limits one or more of the major life activities of such individual; (B) A record of such an impairment; or (C) Bei ng regarded as having such an impairment (United States Department of Justice, 1992). Though this governmental definition of disability is clear-cut, the concept is a dynamic one. Since the 1970s, many changes have ta ken place in this concept; it has moved from a biomedical arena to a sociopolitical one. The first major shift occurred with The Rehabilitation Act of 1973. In this act, cong ress recognized that societys accumulated myths and fears about disabil ity and disease are as handicap ping as are the physical or mental limitations that flow from an actual impairment (United States Department of Justice: Disability discrimination law applicab le to federal employment), and ultimately provided more protection for disabled Amer icans against discri mination (Jeon, 2001). These social and political influences im pact the scope of disability definitions; new conditions are added as disability over time. For example, HIV/AIDS became

PAGE 47

29 protected under the Americans with Disabilities Act (ADA) as a physical impairment in the US Supreme Court Case Bragdon vs. Abbott in 1997 (Supreme Court of the United States, 1997). As such, categories of disabi lity may vary over time and across cultural groups. These socially created definitions of disability influence how people cope with disabilities, when and where they access the fo rmal healthcare system, and their status as a compliant or non-compliant patient. In the present research, the aim is to understand lymphatic filariasis not in American terms, as outlined above, but, rath er, in terms that are emic to the Haitian people. Cultural models offer an opportunity to understand how the Haitian people create meaning around lymphatic filariasis. Also, it is possible to examine how these meanings may be influenced by exposure to American ideas of disability via the social support group intervention. Crosscultural Disability As the concept of disability undergoes continuing debate and change in the United States, indigenous populations al so develop their own ideas of disability. These emic definitions do not always adhere to the same cognitive concepts employed in the United States. The following excerpt exemplif ies varied ideas about disability: In some cultures, it was acknowledged that disability represents such a shame that they have to hide the person, to such an extent that it appears that there are no people with disabilities in that communit y. In other cultures, everyone in the community has a soft spot for persons w ith disabilities, and the person with a disability is an active participan t in family and community life. (McCallion, Janicki, & Grant-Griffin, 1997, 350)

PAGE 48

30 Greeson et al. (2001) exemplify this idea through their work with Somali immigrant concepts of disability related to genetic c ounseling. Through structur ed interviews, these authors found that Somali concepts of disability followed six themes: (1) disability refers to both physical and mental conditions, with mental disability generally thought of first and as more seve re; (2) in Somalia, the family cares for disabled family members, treating them as if they were normal; (3) there are major cultural differences between Somalia and the United States in how persons with disabilities are treated; (4) caring for a person with a disability is stressful for the family; (5) Allah determines whether or not a child will be disabled, and this cannot be predicted or altered; and (6) family is the primary life focus, and therefore, risk of disability doe s not affect reproductive decisions (Greeson, Veach, & LeRoy, 2001, 359). In this study there are clear disconnects between the Ameri can and Somali concepts of disability that carry im plications for genetic c ounseling approaches. How a cultural group perceives and acts on a condition impacts the stigma associated with the diagnosis, social support, and family roles and responsibilities. With regards to public health, cross-cultural studie s on disability provide crucial information towards understanding barriers to accessing we stern-style healthcare and noncompliance with biomedical treatment regimens. In addition to the Somali study, Anne Fadimans (1997) work with the Hmong, and Paul Farmers (1990, 1992) work with Haitians are two of the most well known pieces of cross-cultural work in the area of disability. The following examples provide evidence for the intersection between cross-cu ltural concepts of di sability and public health. Epilepsy is one example of a disability that can be socially constructed in different ways. From a biomed ical viewpoint, epilepsy is actually a group of disorders

PAGE 49

31 the common feature is that there is a sudde n paroxysmal or episodic discharges by some neurons within a particular section of th e brain (Ziegler, 1982, 435). Epilepsy is constructed as a strictly soma tic disability, and this conditi on should be treated with drugs that will eliminate the attacks (Ziegler, 1982). Additionally, epilepsy is considered both an acute and chronic illness as the condition is not eradicated but only erupts in acute episodes occasionally (Ziegler, 1982). The previous description is a mainstream understanding from a biomedical perspective. Cross-culturall y, however, different constructions exist. Anne Fadiman (1997) worked among the Hmong population in California, and found concepts of disability to be quite diffe rent that those of western medicine. She followed the experience of a Hmong family who had an epileptic child. Among the Hmong, epilepsy is considered an illness, but the cause of the illness and ramifications of the condition are much di fferent than the Amer ican understanding. The Hmong believe that epilepsy is a condition contracted when a sp irit steals your soul (Fadiman, 1997). Instead of thinking that epilepsy, or the spirit catches you and you fall down, is a detriment to a person, the Hmong fram e the experience in alternate terms. Though this population does acknowledge that epilepsy is a serious condition, the Hmong consider qaug dab peg to be an illne ss of some distinction Hmong epileptics often become shamans. Their seizures are thought to be evidence that they have the power to perceive things other pe ople cannot see (Fadiman, 1997, 21). Public health concerns such as treatment adherence or self-care are affected by these socially constructed meanings of di sability. The way in which a condition is

PAGE 50

32 constructed culturally influences the treatm ents considered appropriate for curing or addressing the health concern. The Lees, Lias family, did not strictly follow the suggested treatment regimen of anti-convulsa nts (Fadiman, 1997). They administered partial amounts of the medications, and quit making Lia take them when she was doing well (Fadiman, 1997). Thus, they were nonc ompliant in western terms. Fadiman suggests that, for the Lees, t he crisis was the treatment, not the epilepsy (Fadiman, 1997, 53). Though the treatment suggested by the biomed ical system was problematic for the Lees, they augmented the anti-convulsant trea tment with traditional medicine. The Lees believed that epilepsy was caused by soul loss, and they proceeded to utilize traditional healing methods to find Lias soul. Th e Lees performed sacrifices, bought healing amulets, used massage, and tried changing Lia s name in order to heal their child of epilepsy (Fadiman, 1997). Thus, both choice of treatment and trea tment adherence are greatly influenced by the cultural cons truction of disability and illness. Another example of cross-cultural work on disability focuses on HIV/AIDS. Paul Farmers work among Haitians provides anot her example of how c oncepts of illness and disability are framed. Haitians suggested tw o ways in which the disease spreads: by the mode of infectious disease and by the pro cess of sorcery (Farmer, 1990, 21). When HIV/AIDS spread by infectious disease, it wa s considered universally fatal (Farmer, 1990). However, if the disease was sent through sorcery, magical intervention is possible (Farmer, 1990). Ultimately, the models proposed in Haiti link "sickness to moral concerns and social relations" (Far mer, 1990, 23). Each case of HIV/AIDS was

PAGE 51

33 determined to be a product of sorcery or infectious disease by examining the social relations of the seropositive individual. The cause of each case of HIV impacted the way treatment was received and utilized. A case of an HIV/AIDS di agnosis as culturally cons tructed exists with Manno Surpris, a local teacher who first contract ed AIDS in Do Kay (Farmer, 1990, 14). Manno was not a native of Do Kay, and he came to the village in the capacity of a teacher (Farmer, 1992, 61-62). Additionally, Manno he ld several paid positions within the community, and this situation caused resent ment among some of the villagers (Farmer, 1992, 62). When Manno contracted AIDS, the community determined the cause of his problem was rooted in sorcery (Farmer, 1992, 67). If a man had enemies, then, at times, individuals would send a sickness to them (Farmer, 1992); this situation presumably occurred to Manno. Manno did adhere to th e treatment regimens he was given by the doctor, but because his condition was constructe d as an act of sorcery, he also sought the help of a traditional healer, or houngan (Farmer, 1990). Currently, there are large bodies of research in the field of disability studies, but cross-cultural and transdisciplinary research on disability is still sparse in the literature. The proposed research suggests that disconnects in cognitive cat egories of disability exist between the Haitian population and mainstream biomedical categories, and that these disconnects have implications for help-seeki ng, treatment adherence, and myriad health care decision-making behaviors.

PAGE 52

34 Symbolic Interactionism Symbolic interactionism remains the primary theoretical approach used towards constructing and understanding cr oss-cultural beliefs about disability. Thus, this theoretical framework will guide the rese arch design. This theory focuses on the acquisition and generatio n of meaning, and it puts forth that meanings are embedded and constructed in our interactions with other people and inst itutions (White & Klein, 2002; Ritzer, 2000; Charon, 2007, Forte, 2001). Thes e meanings are often taken for-granted, and interpretation is important (White & Kl ein, 2002; Ritzer, 2000; Charon, 2007; Forte, 2001). A clear way of thinking about this theory is that what humans define as real has real consequences (White & Klein, 2002, 60). Thus, if a sample population believes that deviant social behavior causes LF then that belief yields real consequences in the form of understanding the condition, how it is contracted, what the symptoms are, how it is treated, and how it is transmitted. For example, individuals with lymphatic filariasis may receive social sanctions as a result of these understandings in the form of stigma or discrimination. Symbolic interactionism is comprised of four assumptions, four concepts, and multiple propositions. In this context, an assumption is defined as a statement about the phenomenon central to the discipline that repres ents the beliefs the theorist holds true (J. Coreil, personal communication, class notes, 20 03). Concepts are not things, but, rather, stand for the abstract cla ss of things, ideas, or entitie s (White & Klein, 2002, 10).

PAGE 53

35 Concepts help to organize experience, are generally stable, and are often measurable (White & Klein, 2002). Finally, propositions exist when a concept is linked in a meaningful way by a relation to another c oncept (White & Klein, 2002, 11). A theory must include a minimum of two propositions, and they often include more (White & Klein, 2002). The assumptions included within the sym bolic interactionism framework are as follows: human behavior must be understood by the meanings of the actor, actors define the meaning of context and situation, indi viduals have minds, and society precedes the individual (White & Klein, 2002; Forte, 2001 ; Ritzer, 2000). The fi rst assumption, human behavior must be understood by the meanings of the actor, addresses the fact that it is important to understand what meaning an actor attaches to each behavior they undertake. If this assumption is not held, then each action taken by an actor can be interpreted myriad ways. Thus, it becomes impossible to in terpret the meaning of an actors behavior without this assumption. The second assumption present within the sy mbolic interactionism perspective is actors define the meaning of context and situ ation. This assumption explains that how we define the situation in which we find ourse lves explains what pr oblems we define and what actions and solutions we should unde rtake (White & Klein, 2002, 64). White & Klein (2002) use the example that a drunken person may believe that a herd of pink elephants is coming towards him/her, and th is person reacts as though elephants exist around him/her even in the absence of their true existence. The emphasis is on how an

PAGE 54

36 individual perceives a situation, ascribes mean ing to it, and undertak es action in response to stimuli. The third of the four assumptions associated with this theory is that individuals have minds (White & Klein, 2002; Ritzer, 2000; Charon, 2007). Symbolic interactionism assumes that the human mind acquires, integr ates, and processes information (White & Klein, 2002, 64; Ritzer, 2000). As the mind pro cesses information, the mind incorporates it into the identity role of both an actor and an object (White & Klein, 2002; Ritzer, 2000; Forte, 2001). Finally, the last assumption included w ithin the symbolic interactionism perspective is that society precedes the i ndividual (White & Klein, 2002; Ritzer, 2000). The idea is that since we cannot conceptuali ze without symbols, and symbols are shared, society in rudimentary form must precede th e individual mind and self (White & Klein, 2002, 64). In order to make sense of the worl d and share a view be tween individuals, there must be a basic level of common understand ings as to what things mean. This last assumption has lead to the focus on socializa tion within symbolic interactionism (White & Klein, 2002). How people are socialized im pacts how they acquire, integrate, and process information; in turn, this proce ss directly influences interpretation. In addition to the assumptions within the symbolic interactionism framework, four core concepts exist: self and mind, socia lization, role, and definition of the situation. White and Klein (2002, 65) describe the self as a symbolic re presentation of that which did an act (I) and that which was acted on (me). In this context, the self can be viewed in two ways; in the first I status, the concep t of self is related to how an individual

PAGE 55

37 perceives themselves as a subject. The sec ond perspective focuses more on an objectified perception of self; how does the other perceive of the self, or me. Thus, the self is constructed by our consciousness from the tw o perspectives of I and me (White & Klein, 2002, 65). The second concept within this framework is socialization. Socialization is defined as the process by which we acquire the symbols, beliefs, and attitudes of our culture (White & Klein, 2002, 66). This con cept includes two stages for acquisition of meanings: the play stage and the game stag e. White and Klein (2002, 66) describe the play stage as learning to take the role of other, and the game stage as being able to incorporate his or her self in an organized activity th rough the generalized other. The third concept identified is role. Role is defined as the given individuals ability to take the roles of, or put himself in the place of, other individuals implicated with him in given social situations (Mea d, as quoted in White & Klein, 2002, 66). An individual takes on roles, and, in doing so, put[s] oneself in the place of the actor, and it includes the rules that the actor is expected to follow. This concept of the role is further explained along thr ee dimensions: expectations, clarity, and role strain (Wh ite & Klein, 2002). Expectations refer to the things an individual is supposed to do when they are taking on a particul ar role, or the rules of the role (White & Klein, 2002). Clarity refers to th e extent to which the e xpectations of a role are clear (White & Klein, 2002). Finally, role strain addresses situations where the actor does not have sufficient resour ces to enact a role or role s (White & Klein, 2002, 67).

PAGE 56

38 The final concept in symbolic interactionism is defin ition of the situation. This concept embraces the idea that what we defi ne as real will have real consequences (White & Klein, 2002, 67). This focus draws attention to the thought that perception influences behavior (White & Klein, 2002). Finally, these pe rceptions are created through interaction between an individual and the envi ronment, and are not solely internal and individual (White & Klein, 2002). In addition to the assumptions and concepts critical to this theoretical approach, symbolic interactionism encompasses multip le propositions as well. Propositions are often used interchangeably with hypothesis to mean any idea or hunch that is presented in the form of a scientific statement (Marriner-Tomey, 1998). Several examples of propositions that can come out of this theory are outlined below in table form (Table 2):

PAGE 57

39 Table 2. Examples of Propositions in Symbolic Interactionism Proposition Explanation The quality of egos role enactment in a relationship positively affects egos satisfaction with the relationship. A person feels better about a relationship when they think they are doing a good job of enacting their role in a relationship. The greater the perceived clarity of role expectations, the higher the quality of role enactment. The clearer the expectations of the role, the easier it is to meet those expectations. The more individuals perceive consensus in the expectation about a role they occupy, the less their role strain. The more consensus there is on role expectation, the more a person knows the rules of their role, the less role strain a person experiences. The greater the diversification of a persons roles, the less consensus the person will perceive in the expectations about those roles. The more roles a person has, the more expectations they must meet, and the more likely that multiple expectations may become ambiguous or contradictory. The greater the perceived role strain that results from performing a role, the less the ease in making a transition into the role and the greater the ease in making a transition out of the role. The more stress a role produces, the more difficult it is to adjust to that role. (White and Klein, 2002, 68-70) In addition to the nuts and bo lts of symbolic interactionism, this theory includes four major variations: struct ural approach, interactional approach, microinteractional

PAGE 58

40 approach, and phenomenology of the family a pproach (White & Klein, 2002). The first variation, the structural approach, is compri sed of three basic notions: position, norm, and role (White & Klein, 2002). In this context, an individua l can have more than one position, and these positions are embedded in a social network of interrelated positions (White & Klein, 2002, 72). Each one has associated norms or expectations (White & Klein, 2002, 72). Essentially, roles are a key focus in this variation, and the central metaphor is the idea of actors on a stage (Wh ite & Klein, 2002). The idea is that these actors are poured into pre-exis ting, inflexible molds, and the only variation lies in how well the actor performs his/her role. In this variation, the Iowa School of thought, oversocialization takes away freedom to adap t to or change the environment (White & Klein, 2002; Ritzer, 2000; Forte, 2001). Thus, the major criticism of this approach is that it is deterministic and views social cha nge as very slow (White & Klein, 2002). An alternate variation of symbolic interactionism is the interactional approach. This version of symbolic interactionism fo cuses on patterns that are developed through interaction (White & Klein, 2002; Ritzer, 2000) Culture and society are created through the interaction of individuals and there is a clear focus on creativity and problem-solving (White & Klein, 2002) As such, the interactio nal approach is less deterministic as it recognizes that social structure and culture provide a broad outline for behavior; there are no specific rules (White & Klein, 2002; Ritzer 2000). Due to its fl uid and unstructured nature, however, there is one major criticism of this variation. This criticism is that interactionists minimize the ro le of structure too much and assume too much power for individuals to create their ow n roles (White & Klein, 2002, pg).

PAGE 59

41 An extension of the interactional appr oach, and the third variation of this theoretical perspective, is the microinteractional approac h. This variant, the Chicago School of thought, focuses on the individual a nd the self, and the central metaphor is a dramaturgical approach (White & Klein, 2002; Forte, 2001). There is an emphasis on complete fluidity in roles; the self shifts within interactional context and conceptualization of the cont ext and its associated rule s (White & Klein, 2002). This approach, though, does not make any attempt to extend their explanations to the macroscopic level; this problem is the major criticism of this variation (White & Klein, 2002). Finally, phenomenology of the family is the final variation. This perspective focuses on the taken-for-granted everyday life world and the ways in which these takenfor-granted meanings are created and maintained (White & Klein, 2002, 76). Phenomenology of the family aims to identify the assumptions and typifications that enter into the construction of the everyday li fe. Each individual s report of subjective experience is influenced by intersubjective components; these co mponents are shared with and communicable to others (White & Klein, 2002). These intersubjective meanings are the foundation of the social world, shared in a set of actors, compose the commonly held and understood meanings of our everyday life, and represent frames of reference (White & Klein, 2002, 76). These four variations describe differe nt ways of thinking about symbolic interactionism, and all intersect with a majo r debate within this framework. The debate discusses two viewpoints regarding interactions between people: (1) interactions are a

PAGE 60

42 product of the expectations resi ding in the social structure, (2) interactions are created and negotiated by the actors in each interaction (White & Klein, 2002, 72). Each variation falls somewhere in the continuum of these two extreme approaches. However, in the current research, neither extreme is appropriate. The two positions, as expressed through the variants of symbo lic interactionism, are both re levant on some level to the current research effort. Symbolic interactionism, and specifically the interactional approach, is relevant to the current endeavor as shared cultural mode ls developed in a support group setting are investigated. Kaye (1997, 19) proposes that t he support group may be conceptualized as a social microcosm in that it is a small, comp lete social world. Within this microcosm, illness meanings in support groups, in this case beliefs about LF, are negotiated on multiple ecological levels including intrapersonal and interpersonal (Kramer & Nash, 1995; Kurtz, 1997). Forte (2001, 31) explains, the pe rson is both a producer of society, a free agent who influences social processe s, and a product of society, a member caught in the necessities of the so cial-historical drama. Support group participants both shape the group dynamic and are shaped by it. As re ality only exists thr ough the lense of human interpretation (Forte, 2001), beliefs about causes of and treatment for LF do not exist absent from human interaction.

PAGE 61

43 Support Groups Support groups are defined as a group of people, sometimes led by a therapist, who provide each other moral support, info rmation, and advice on problems relating to some shared characteristic or experien ce (Support Group, 2007). Schopler & Galinsky (1995, 4) suggest that support gr oups exist in the middle of the continuum between selfhelp groups and treatment groups and are m ember-centered; leader ship is provided by professionals, volunteers, or, at times, by me mbers. They are generally small groups that, as described by Kurtz (1997 4), meet for the purpose of giving emotional support and information to persons with a comm on problem. These groups often involve professional facilitators and affiliation with broader issueoriented groups (Kurtz, 1997; Schopler & Galinsky, 1995). Groups of this nature do not generally charge fees and are not highly structured (Kurtz, 1997). Several key elements are present in s upport groups. Kaye (1997, 21; Schopler & Galinsky, 1995) suggests there are five pro cesses present: giving support, imparting information, conveying a sense of belongi ng, communicating experiential knowledge, and teaching coping methods. These processes aid in attaining goals that are typically found in support groups such as emotional rele ase, validation of concerns, reduction of social isolation, information, improved copi ng, decreased stress, problem-solving, and, at times, advocacy (Schopler and Galinsky, 1995, 6; Kaye, 1997). Participants in support groups are often de viant in some particular way; this deviance may involve a problem central to daily living or with a set of life

PAGE 62

44 circumstances that represents a departure from the norm (Kaye, 1997). Kaye (1997,18) suggests that the individual ty pically views the problem or life circumstance as one that will endure over time, believes it to be socia lly or psychologically isolating, and has had little or no preparation or forewarning in adva nce of its onset. In the case of lymphatic filariasis, the illness is the deviant condition that draw s individuals to take on support group membership. Support groups are generally formed through three stages of development (Hermann, 2005). Coreil & Ma yard (2006, 129) nicely summar ize the three stages of support group development as defined by the American Cancer Society (Hermann, 2005): phase one, where members find commonalitie s, seek out information, and explore alternative ways of allevi ating problems; phase two, where members provide mutual support, share experiences, offer help, rede fine the illness experience, and uncover new ways of coping; and phase three, in which the group may be brought to a close (if goals have been met) or become a long-term project (whether goals have been met or not). As a support group moves through these de velopmental stages, the group provides a microcosm where the tenets of symbolic in teractionism play out. Within these groups, participants are interacting at interpersonal, community, and societal levels. At each level of interaction, meanings are shaped and defi ned. Coreil & Mayards (2006) work in Haiti, as well as the Archaie project, provides insigh ts on the indigenization of illness support groups as they move through these three phases. In this work, indigenization is defined as the process of transformation that often occurs when social institut ions developed in one social context are transplanted into a totally different social cont ext (Coreil & Mayard,

PAGE 63

45 2006, 129). The transformation described exem plifies the symbolic interactionism perspective in action. Interpersonal interactions within the support group resu lted in adaptation of the general support group model. The format and content of each meeting was adjusted to increase effectiveness in the local context; meetings were infused with religious overtones including prayers and hymns (Cor eil & Mayard, 2006). Also, the primary content of interest did not include discussi on of personal illness experience (Coreil & Mayard, 2006). Rather, the women were eager to gain health information and acquire new vocational skills (Coreil & Mayard, 2006). In the Archaie project, the support groups were tailored to the following agenda (Table 3):

PAGE 64

46 Table 3. Support Group Meeting Agenda Support Group Meeting Agenda Pray Sing News about health for each member of the group Teach the assigned subject in the book (educational manual related to lymphatic filariasis and self-care practices) Discussion about the topic for that meeting Exercise and washing the leg Share experiences about the disease and social life Planning for the next meeting Refreshments Amusement gossip, talking Singing Pray Close and Farewell (G. Mayard, personal communication, June, 2007) Again elements of religion and educationa l information were of primary importance. These adaptations closely relate to hallmarks of phase one development. The way that the group tailored the support group on the interpersonal level ties in nicely with the goals of phase one, finding commonalities (religious belief) and seekin g out information (health information and skills).

PAGE 65

47 At the community level, phase two de velopment, redefining the illness experience, is evidenced. First, support gr oups are impacted by the group interaction; in both the Archaie project and the Leogane project ideas about what lymphatic filariasis is, how it is caused, how to treat LF, and ot her illness beliefs are influenced through discussion and the educational modules di scussed. In addition, both support group initiatives interacted with the community at large through an end of year party and increased awareness of lymphatic filarias is (Coreil & Mayard, 2006). The support groups achieved these goals at the organizational level and general local community level, respectively. The last phase of support group developmen t parallels with interactions at the societal level. In the Leogane support group, the original program period was intended to last two years. At the end of this program, participants were enthus iastic about continuing the support group as a long term project. Towa rds this end, group members broadened the scope of the program to focus on general h ealth and development initiatives in Haiti beyond lymphatic filariasis exclusively (Core il & Mayard, 2006). Also, all women in the Leogane community were invited to pa rticipate (Coreil & Mayard, 2006). Through increased membership and partnerships with hospital administrators, the original support groups became subsumed under FADES (Femmes en Action Pour le Developpment et la Sante/Women in Action for Developmen t and Health) (Coreil & Mayard, 2006). Support groups provide an appropriate conduit for the examination of the symbolic interactionism processes. Coreil & Mayards (2006) work as well as the current Archaie project provide an example of how in teractions at multiple levels can shape the

PAGE 66

48 meaning of what a support group is in Hait i as well as what goals can be accomplished through this format. The current project aims to use this perspective to examine phase two development in support groups particular ly. The primary focus will be on how the illness experience of lymphatic filariasis is redefined as a result of support group participation. As the illness experience is redefined, explanatory a nd cultural models of illness shift. Explanatory and Cultural Models One way of understanding illness beliefs is to look for the meaning of a disease experience. Symbolic interactionism allows that these illness beliefs may be created and ascribed meaning through stru ctural norms, individual intera ctions, and the assumptions of everyday life. As meaning organizes so cial behavior (Fulto n, Madden, & Minichello, 1996, 1355), the proposal to find meaning in the lymphatic filariasis (LF) experience is appropriate. There are several key contri butors to the study of mean ing and social behavior related to illness, and Kleinman is among th ese individuals. Kleinm an works to achieve an understanding of illness experiences and social meanings related to those experiences. In his work, Kleinman (1988) distingu ishes between illness and disease. Illness and disease are two terms that play a role in explanatory models, or an individuals understanding of a condition. Di sease involves the biomedical understanding of the condition; personal, soci al and cultural concep ts related to a condition are excluded

PAGE 67

49 (Kleinman, 1988, 5). In Kleinmans work, a dis ease, in biomedicine, is defined in the narrow biological terms of the biomedical model, this means that disease is reconfigured only as an alteration in biol ogical structure or functioni ng (Kleinman, 1988, 5). The pathophysiology of the body and the concept of biological reductionism are key in the western biomedical healthcare delivery syst em (Kleinman, 1988, 6). The body is treated as a machine; if there is something wrong w ith the machine, it is pinpointed, diagnosed, and then fixed, cure. The biomedical construc t of disease is objective in nature, and only captures a portion of the entire experience a patient has when they have a medical condition. The concept of illness is different from disease, and it complements the gaps present in the biomedical definition of diseas e. Illness, according to Kleinman, involves the symptoms, suffering, and response of th e sick individual to a diagnosis (1988, 3). Illness focuses on how the sick person and th e members of the family or wider social network perceive, live with, and respond to symptoms and disability (Kleinman 1988, 3). Sociocultural and psychological f actors are key in understanding illness (Good, 1994). In contrast to the disease conditi on, the illness condition is subjective. Kleinman suggests one way to understand illn ess and disease in a community is to extract meaning from a social occurrence via the explanatory models used by a society. An explanatory model includes the "notions that patients, families, and practitioners have about a specific illness episode" (1988, 121) A more technical definition follows: A set of beliefs about the etiology of an illness, onset of symptoms, the pathophysiology, the course of the sickness, and the appropriate treatment that is used by the individual to interpret sy mptoms and make sense of the illness

PAGE 68

50 experience. (H. Mathews, personal comm unication, class notes, 2000; Kleinman, 1988, 121) Influences including cultural beliefs, popular culture, family and community traditions, biomedicine, and idiosyncratic experiences shape the development of these explanatory models (Kleinman, 1978, 253). Kleinman uses the concept of a headache as an example of an explanatory model (Kleinman, 1988). The symptoms that de fine a headache are culturally defined (Kleinman, 1988). For instance, when throbbing in the temples occurs, is that a headache or a migraine? Understandings about the difference in the symptoms and experience indicate which definition is more appropria te. Once the condition is diagnosed, then culturally defined treatment options are ava ilable. The traditions beliefs, and prior experience associated with the condition shape which treatment option is deemed appropriate. The social categorizations of symptoms, causes, and treatment options for the condition are part of the explanatory model. In order to extract an individual's explanat ory model of an illness, it is necessary to obtain several pieces of information. Ge nerally, the following types of questions are raised: 1. What do you think is wrong with you -what is the problem (sickness) that you are experiencing? 2. What do you think has cause d this problem (sickness)? 3. Why do you think it started when it did? 4. What do you think your sickne ss does to you? How does it work? 5. How severe is your sickness? Will it have a short or long course and

PAGE 69

51 why? 6. What kind of treatment do you think you should receive? 7. What are the most important re sults you hope to receive from this treatment? Can your sickness be cured? If so, how will you know when the cure has taken place? 8. Have you done anything else about your sickness-have you taken any home remedies, seen any other ki nds of healers, etc.? If yes, how are these things working and are you satisfied with them/why or why not? 9. What are the chief problems your sickness has caused for you? 10. What do you fear most about your sickness? 11. What resources do you have availabl e (questions about social support, income, access to community programs) to help you cope with this illness? Will/do you use them -why or why not? (Kleinman, 1988, 121; H. Mathews, pe rsonal communicatio n, class notes 2000) Once the required data have been collected, the information can be synthesized into an individuals explan atory model. Cultural models are a natural extensi on of the explanatory model concept. Cultural models are schemas about a domai n that are shared by members of a group having shared problems, shared task solutions and similar life expe riences (Bradway & Barg, 2006). Where explanatory models are epis odic and rooted in the individuals illness

PAGE 70

52 experience and help-seeking behavior, cultural models are broader in context. In the cultural model, the same dimensions of an illness are of interest, but they are expressed, for example, as more general beliefs about wh at types of things cau se an illness and what types of treatment are appr opriate. In the cas e of the explanatory model personal experience is explored, and, in the case of the cultural model, group level beliefs are investigated. Though there is a distinc tion between explanatory and cultural models, the literature, to date, has ofte n used these terms interchangeably. The following discussion regarding explanatory models and cultural models reflects this circumstance. Currently, no literature exists that summa rizes general Haitian illness and disease concepts. However, disease specific resear ch has been undertaken and several themes surrounding illness consistently emerge. Tw o major categories exist: natural and supernatural (Miller, 2000). Within the natural category, beliefs surrounding disease causation include the blood paradigm, TB pa radigm, microbe paradigm, and humoral (hot/cold) imbalances. Supernat ural causes of illness are r ooted in problematic social relationships between the afflicted and either another person or a displeased spirit/ancestor (Miller, 2000) Miller (2000, 207) e xplains that, in Haiti, the body is considered permeable to the actions, for good or for ill, of outside forces. Farmers (1990) work with cultural mode ls of HIV/AIDS in Haiti exemplifies these themes. In his work, Farmer (1990) looked at how cultural models emerge and change. Farmer suggests that as HIV/AIDS was introduced into Haitian culture, the society at large produced cultural explanations for how the disease arrived and spread.

PAGE 71

53 The Haitians proposed three paradigms to explain the presence of HIV: the blood paradigm, the TB paradigm, and the microbe paradigm (1990, 23). The blood paradigm suggests that there is a link be tween social atmosphere and th e descriptive characteristics of an individuals blood (Farmer, 1990). It was not considered extr aordinary if a person surrounded by social controversy became ill due to bad blood. The tuberculosis paradigm presented as the symptoms of HIV/AIDS were similar to those observed in tuberculosis cases. Th e condition is not only disfiguring but also chronic, sapping the bodys strength over months or years (Farmer, 1990). In addition, a number of individuals diagnosed with HIV/AIDS in Haiti also present with a tuberculosis diagnosis (Farmer, 1990). The microbe paradigm falls in line with bi omedicine; in this paradigm, HIV/AIDS is caused by a microbial source. This beli ef has the official blessing of the local representatives of cosmopolit an medicine (Farmer, 1990). These paradigms suggest ways in which HIV was caused. Alternatively, the same community sugge sted two ways in which the disease spread: by the mode of infec tious disease and by the proce ss of sorcery (Farmer, 1990, 21). When HIV/AIDS spread by infectious dise ase, it was considered universally fatal (Farmer, 1990). However, if the disease wa s sent through sorcery, magical intervention is possible (Farmer, 1990). Ultimately, the models proposed in Haiti link "sickness to moral concerns and social relations" (Far mer, 1990, 23). Each case of HIV/AIDS was determined to be a product of sorcery or infectious disease by examining the social relations of the seropositive individual.

PAGE 72

54 In the case of lymphatic filariasis, belie fs about causation in Haiti draw on these general ideas of illness and disease. Add itionally, multiple studies have explored explanatory models of lymphatic filariasis worldwide and similar disease concepts arise. Illness beliefs regarding cause and treatment of lymphatic filariasis in Haiti, India, Dominican Republic, Ghana, and Thailand ar e reviewed (Coreil et al., 1998, Ramaiah, Vijay Kumar, & Ramu, 1996; Person et al., 2007; Ahorlu et al., 1999; Rauyajin, Kamthornwachara, & Yablo, 1995; Bandyopadhya y, 1996; Eberhard, Walker, Addiss, & Lammie, 1996; Morfit, 1998; Babu, Hazra, C hhotray, & Satyanarayana, 2004; Coreil et al., 2003; Gyapong, Gyapong, Adjei, Vlassoff, & Weiss, 1996). Across all of these studies, only a sma ll percentage of th e sample population identified mosquitoes as the vector for LF (microbe or biomedical paradigm). The popular causes mentioned were rooted in trad itional belief systems, and the treatment options provided leaned prim arily towards traditional me dicine. Trained health care professionals and drug treatment regimens (MDA specifically) retain a minor presence in the treatment spectrum. Traditional beliefs high lighted in several studies are presented in the following tables (Table 4, 5):

PAGE 73

55 Table 4. Causes of Lymphatic F ilariasis: Traditional Beliefs Cause Country Haiti India Dominican Republic Ghana Thailand Humoral Heat X X Excess Coldness in the Body X Blood Accumulation of Bad Fluid X Circulation Problems X X Bad Blood X X Environmental Walking Barefoot in Dirt X Walking Barefoot in Water X Drinking Water from local ponds/Contaminated Water X X X Insect/Spider Bites X X Air Circulation X Contact with fevered individual X Worms X Climate X Supernatural Magic Powder X X X Witchcraft X X Curses X X Bad Spirits X X X Magic Charms X Supernatural Forces X Magic X Fate X Dwarves X Intense Moral Conflict X Occupational Weaving Occupation X Excessive Physical Labor X X X Standing too long at a disagreeable job X Heredity Heredity X X X X X Health, Illness, Injury Injury to the Foot X X X X Poor Nutrition or Dietary Habits X X X Pregnancy X X Fever X Other Cycling too much X Sleeping in the same house with an infected man, and the man gets up and leaves while his partners still sleeps X (Coreil et al., 1998; Ramaiah et al., 1996; Person et al., 2007; A horlu et al., 1999; Rauyajin et al., 1995; Bandyopadhyay, 1996; Eber hard et al., 1996; Kanda, 2004; Morfit, 1998; Babu et al., 2004; Coreil et al., 2003; Gyapong et al., 1996)

PAGE 74

56 Table 5. Treatments for Lymphatic Filariasis: Traditional Beliefs Treatment Country HaitiIndia Dominican Republic Ghana Thailand Herbal Herbal Remedies: Potions (Ex: manioc leaf, custard apple, oak) X X X X Herbal Remedies: Wraps/Poultices (Ex: ashes, animal feces, castor oil, cornmeal) X X X X X Soaking in Salt Water X Powders (Ex: fonsa) X Supernatural/Religious Prayers X X Spiritual Consultation X Magical Ritual X Incantation X Traditional Leeches/Bloodletting/Scarification X X Scrubbing the Surface of Leg with Red Ants X Jolting or Hitting the Affected Area Massage X X Self-Medication Pharmaceuticals (Ex: flanax, tetrazan, penicillin, oral painkillers, chloroqu ine, paracetamol) X X X X Dietary Control X X Injections X X Creams X X X Enema X Surgery Amputation/Surgery X X X (Coreil et al., 1998; Ramaiah et al., 1996; Person et al., 2007; A horlu et al., 1999; Rauyajin et al., 1995; Bandyopadhyay, 1996; Eber hard et al., 1996; Kanda, 2004; Morfit, 1998; Babu et al., 2004; Coreil et al., 2003; Gyapong et al, 1996)

PAGE 75

57 Explanatory and cultural models provide us eful ideas about the social and cultural meaning of a condition. For inst ance, individuals may only cont ract an illness as a result of some socially deviant acti on. Alternatively, the cause for the onset of an illness may be understood in strictly biomedical terms: th e illness was caused by bacteria and was not brought on by deviant behavior. Of the multip le dimensions present in the explanatory model framework, it is arguable that the two most important dimensions are cause and treatment. Varied understandings of causati on are important to the development of the patient's help-seeking and compliance patterns Several studies have made this link between perceived cause of a condition a nd subsequent help-seeking behavior and treatment choices (Coreil, 1983; Rauyajin et al., 1995; Ahorlu et al., 1999; Ramaiah et al., 1996). The strength of cultural models is that th ey allow for the influence of social and cultural variables on an illness to be understood and ha ndled. The social understandings that are incorporated into cultural models, respectively, are drawn from different sources of knowledge. These sources are rooted in shared cultural beliefs and include shared popular knowledge from the media, ethnomed ical or alternative healing traditions, biomedical understandings, and personal or fa mily experiences. Cultu ral models explored here provide insight into the cu lturally shared beliefs about a particular disease and into the processes by which the part icipants incorporate these understandings into their own illness experience. With the insights of the symbolic interac tionism perspective, it is possible to see how cultural models could become a useful tool in evaluating the success of public health

PAGE 76

58 interventions. Cultural models are one manifest ation of social interaction. These ways of understanding illness have very real implica tions in how individuals perceive their condition and seek treatment for it. Insights related to the shift in understanding that resulted from an intervention yield multiple implications. For instance, an intervention shown to influence changes in thinking about illness that lead to more help-seeking behavior, more acceptance by the communit y, more adherence to treatment, fewer secondary conditions, and a better quality of life could be a very powerful indicator indeed. This information could also provide ideas towards culturally tailoring future interventions for a specific population. Cultural Consensus and Cultural Consonance In order to evaluate how explanatory and cultural models of LF have shifted from baseline to outcome points, a specialized statistical analysis will be utilized in conjunction with commonly used statistical approaches. Specifically, cultural consensus analysis will be utilized. Ro mney (1999, S103) explains that cultural consensus theory helps describe and measure the extent to which cultural beliefs are shared, and the central idea is to look at t he use of the pattern of agreement or consensus among informants to make inferences in their differential competence in knowledge of the shared information pool constituting culture (Romney et al., 1986, 316). Thus, examining cultural models is in line with cultural consensus theory.

PAGE 77

59 This type of quantitative an alysis, cultural consensus analysis, aims to deal with two problems: first how can the cultura l knowledge of different informants be estimated, and, second, how can th e correct answers to specif ic questions be inferred and with what degree of conf idence (Romney et al, 1986, 88). In this case, the first aim is of interest as the goal is to better u nderstand the cultural k nowledge in the target community. One goal of this research is to unde rstand if there are shif ts in the consensus of the sample population as to what cultural models are primary. In order to think about cultural consensus with in this project, it is likened to factor analysis. Factor analys is takes sets of survey items and looks for those items that hang together on certain factors. Th ese factors represent more ma cro level latent constructs, and the items that hang on a factor provide an indirect measure of the underlying latent construct. Similarly, cultural consensus analysis examines items related to cultural knowledge and identifies the informants that ha ng together in their responses to the items of interest. These informants who respond to the items similarly are assumed to draw upon the same underlying cultural domain. The key difference in these approaches is that factor analysis identifies survey items that are related, and cultural consensus analysis identifies informants whose response patte rns are related. The items that will be examined through cultural consensus analysis in this research are drawn from the baseline and outcome questionn aires administered in the Evaluation of Support Groups in the Management of Lymphedema Caused by Lymphatic Filari asis project. Initially, cultural consensus analysis wa s only used with dichotomous variables, but Romney, Batchelder, & Weller (1987) have sh own that it is also a robust statistical

PAGE 78

60 approach with true/false, multiple choice, and fill-in-the-blank data. Also, the use of rank order data and interval data in cultural cons ensus analysis has been introduced (Romney et al., 1987; Weller, 1987). The data utilized in this proj ect are appropriate for the application of cultural c onsensus analysis as de fined by these criteria. The data in this project will be anal yzed along the three major components in consensus analysis. These components include the overall shared cons ensus in a sample population, individual levels of cultural competence, and the answer key developed from the data itself (Romney et al., 1986). In the first component, the levels of agreement at baseline, outcome, and between the two points in time will be examined. If the eigenvalue ratio, between the first and second eigenvalues, attains a threshold of 3.0, consensus is assumed (Dressler, 2001). When co nsensus is met, it means that the sample is drawing on the same cultural domain. Dre ssler (2001, 3) defines a cultural domain as some topic or subject found in ever yday life that people talk about. The second component that must be a ddressed is the level of individual knowledge, or cultural competen ce. In this circumstance, co mpetence is a measure of the individuals knowledge as compared to the shared cultural knowle dge base (Weller, 1987). In specific terms, Weller (1987, 181) describes competence as the probability that an individual knows an answer is equi valent to the proportion of shared knowledge that individual has with the unobservable culturally corr ect answers. When high competence and low standard deviation ex ist within a population, high levels of consensus result (Romney et al., 1987).

PAGE 79

61 Finally, the culturally correct answer key must be developed from the data. In order to develop this key, each inform ants competence is assessed. The answers provided by informants who have higher levels of competence are weighted more heavily than individuals who are less competent (Rom ney et al., 1986). Thr ough this process, the informants who best know the culture that is being measured have more influence in identifying the best cultural estimate for each item (Romney et al., 1986). When the most competent informants find high levels of concordance on items, then these items are flagged as key pieces of the cultural doma in being accessed. These key pieces form the cultural model shared by the population. The strengths of this approach are multiple, and this analysis is appropriate to the current endeavor. Per Romney et al. (1987), cultural consensus analysis is appropriate for research circumstances where the cultural competence of each informant, as well as the correct answers to questions posed, are unknown. In addition, Romney at al. (1987, 164) explain that one of the main purposes of cultural consensus theory is to use response data to provide estimates of the corr ect answers as well as tell us how confident we can be of those answers. As the research is undertaken without prior knowledge as to what cultural domains are being assesse d, this approach is appropriate. Results obtained by applying this analysis should also contribute to the solution of some questions about what the cultural beliefs actually are in some cases (Romney et al., 1986, 333). This final application is relevant to the problems at hand. It is important to understand what the beliefs are and how much consensus each belief commands in a population in order to inform culturally tailored public health efforts.

PAGE 80

62 If significant changes in cultural models ha ve taken place, it is of interest to know what kind of changes these include. For instan ce, are cultural models resulting after the community support group intervention more in line with biomedical beliefs or did they become more rooted in traditional medical beliefs associated w ith the diagnosis? If significant changes are not present, is that considered a problem from the evaluation standpoint? An extension of the cultural consensus analytic approach is cultural consonance which examines behaviors. As previously stated, beliefs about causation of an illness have been linked to help-seeking behaviors, and these findings are in line with an anthropological view that culture serves a directive function with respect to behavior (Dressler, Dos Santos, & Balieiro, 1996). Cultural consonance aims to quantify divergence or convergence of i ndividual level behaviors as th ey relate to the cultural model identified in the cultural consensus analys is (Dressler et al., 1996 ). Dressler et al. (1998) and Chavez, McMullin, Shiraz, Hubbell (2001) have shown that the degree of cultural consonance in a community has been linked to health outcomes such as coronary heart disease and cervical cancer screening. In this project, self-care behaviors pr esent in the cultural model of lymphatic filariasis are the elements examined. The degree to which an individuals behaviors correspond with the treatment op tions present in the cultura l model is quantified as a percent coefficient, or percentage of behavior s an individual engages in out of all possible behaviors. Values can range from 0 100%, where a score of 100% indicates an individual is behaving in complete alignm ent with the self-care treatment elements

PAGE 81

63 identified in the communitys cultural model of LF. Also, if the sample is examined as a whole, the average rate of consonance can be calculated, indicating the percentage of ideal behaviors an average person in the st udy population enacts rela ted to self-care and lymphatic filariasis. Cultural consensus and cultural consonance work together as analytic approaches in an evaluation context. Cultural consen sus highlights elements included in local community belief systems, and cultural consonance measures the degree to which community members implement these illn ess beliefs through behavior. Regarding evaluation, both knowledge and be havior resulting from these analyses are grounded in the local community itself and, as such, are cu lturally appropriate. Also, the link between belief and behavior can be examined in an illness and context speci fic way. Finally, if a link between local cultura l models and behaviors is strong, there is evidence that changes in cultural model resulting from the support group intervention do indeed impact health care behaviors. Should the cultural model sh ifts towards a biomedical framework, one could extrapolate that suppor t group participants are mo re likely to engage in biomedically grounded se lf-care behaviors.

PAGE 82

64 Chapter 3: Methods

PAGE 83

65 Study Design The primary focus of this work was to c onduct a test regarding the utility of the cultural model as an evaluation indicator. In order to attain this goal, changes in cultural models before and after part icipation in womens support groups were explored. The central research questions investigated to be tter understand changes in cultural models over time and the utility of CMs for evalua tion purposes are reviewed below (Table 6):

PAGE 84

66 Table 6. Linking Research Ques tions to Method of Analysis Research Question Hypothesis Method of Analysis Does a cultural model exist in the sample population at baseline? If so, does the CM meet the threshold for cultural consensus at baseline? If so, what is the strength of the consensus? It is expected that a shared cultural model of LF will exist at baseline. Cultural Consensus Analysis Does a cultural model exist in the sample population at outcome? If so, does the model meet the threshold for cultural consensus at outcome? If so, what is the strength of the consensus? It is expected that a shared cultural model of LF will exist at outcome. Cultural Consensus Analysis Are there significant changes in levels of cultural competency existing at baseline and outcome? It is expected that levels of cultural competency will be greater at outcome than baseline. Cultural Consensus Analysis Are there significant differences in the elements included in the CMs from baseline and outcome points? Is the cultural model present at baseline different from the cultural model present at outcome? It is expected that the CM at outcome will be significantly different than the CM present at baseline. The outcome CM is expected to include more elements of western biomedical beliefs than the CM at baseline. It is expected that a bicultural model will exist at outcome incor porating elements of traditional and western biomedical ideas about LF. Cultural Consensus Analysis Is the strength of cultural consensus for the CM greater at outcome than baseline? It is expected that the strength of consensus for the CM at outcome will be greater than the strength of consensus for the CM at baseline. Cultural Consensus Analysis Does a significant link between belief (cultural model) and behavior (cultural consonance) exist in the sample population? It is expected that greater consensus regarding cultural models will be linked to higher rates of self-care behaviors identified in the cultural model. Cultural Consensus Analysis Cultural Consonance Analysis

PAGE 85

67 This project involves testi ng for significant differences, changes, from baseline to outcome points, and it is a ppropriate to employ a quasi-exp erimental design (McDermott & Sarvela, 1999, 217). Randomization was not po ssible as the study sp ecifically focused on the LF-infected population; participant availability was a concern (Kanda, 2004). Also, no list existed of eligib le participants (Kanda, 2004). When it is not possible to randomize group assignment to test group diffe rences, the quasi-experimental approach can be used with equal success as long as the groups do not display significant differences prior to the intervention (M cDermott & Sarvela, 1999). Each group is assessed for the presence of cultural models at baseline and outcome as well as any changes in model through cultura l consensus analysis in order to determine the success of the support group program. Additionally, cultura l consonance analysis is employed to examine the degree to which participants behaviors reflect the existing cultural models in the sample population at each point, baseline and outcome. In addressing these research questions it is necessary to run analyses on the reliability and validity of the data as well as demographics, cultural consensus analysis, and cultural consonance analysis. Reliability and validity analyses are necessary to establish the accuracy and integrity of the dataset. Univariate statistics examining demographics are important as they allow fo r a description of the sample population, both in size and distribution. Once these preliminary analyses were conducted, cultural cons ensus and cultural consonance analyses were requ ired. Cultural consensus analys es yielded results regarding group level illness beliefs around the cultural do main investigated, lymphatic filariasis.

PAGE 86

68 The consonance portion complemented the cultu ral consensus work as it explored the link between cultural mode ls and the enactment of self-care behaviors. Description of the Data The data utilized in this work were draw n from support groups in 3 Haitian towns, were gathered in an evaluation context, and are diachronic in nature (data were collected at baseline and post-inte rvention points). Though the nature of the data set requires both a pre and post data point measure, some particip ants were lost to follow-up for the second point of data collection. Partic ipants invoked different aliase s at the baseline and outcome points; however, the original research team successfully matched 87 individuals as they personally knew the aliases they utilized. N on-probability sampling was utilized as the study was focused on a particular condition of a single disease and there was no list or information of the study population (Kanda 2004, 39). Kanda (2004) further explains that this approach to data collection was appropriate as the p opulation available for participation was limited and that potential participants were not easily identified. A quasi-experimental quantitative design was utilized and, as such, involved both control and intervention groups. Also, this descri ption is appropriate as participants were not randomly assigned to their respective groups. The original research team identified all known cases of LF in women across the three si tes; they expected to find approximately 200 cases and intended to enroll 100 peopl e in control and intervention groups, respectively. As cases were identified, each woman was approached and asked to

PAGE 87

69 complete a baseline survey. After completion of the questionnaire, they were invited to participate in the support group; thus, participants self-sel ected themselves into the intervention group. Remaining participants became control memb ers of the population. Members of the control group completed ba seline and outcome surveys, and these individuals were not involved in the support group program. Alternat ely, participants in the intervention portion of th e project were exposed to the support group program and also completed surveys at both baseline a nd outcome points. La Plaine contributed control group participants onl y; Cabaret and Archaie incl uded individuals involved in both the intervention and control groups. The baseline sample was gathered in September 2003, and the outcome sample was obtained in April 2005. These evaluation su rveys, conducted as part of a program evaluation, were gathered over the course of the second year commencing after the first year of participation. A total sample size of 241 participants was attained at baseline, after the exclusion of men and individuals indicati ng they were less than 18 year s of age at baseline, and the sample sizes were 80, 50, and 102, for Archaie, Cabaret, and La Plaine, respectively (Table 7). In the outcome data, there were 200 subjects (Table 8). For the analyses involving change in cultural models over time participants baseline and outcome data were matched. This portion of the analysis includes 60 people in the intervention group and 27 in the control group (Table 9).

PAGE 88

70 Table 7. Sample Sizes for Baseline Data Baseline Data Archaie La Plaine Cabaret 89 102 50 Baseline Data N = 241 Table 8. Sample Sizes for Outcome Data Outcome Data Archaie La Plaine Cabaret 108 65 27 Outcome Data N = 200 Table 9. Sample Sizes for Matched Data Matched Data Archaie (n=46) La Plaine (n=19) Cabaret (n=22) Intervention Control Intervention Control Intervention Control 40 6 0 19 20 2 Total N = 87 Intervention N = 60 Control N = 27 The sample sizes for the matched inte rvention and control groups do meet the power threshold required by cultural consensu s analysis. In a resource poor setting, diachronic data are difficult to gather; thus, this data set is preferable to cross-sectional data sets that may otherwise be available.

PAGE 89

71 Context of the Data The primary study that produced the data examined support groups for women with lymphatic filariasis (LF) in Haiti. Funded by the Presbyterian Womens Birthday Fund, the primary research was an extens ion of a support group program conducted by Coreil et al. (2003) in Leogane. Partners from Hpital Ste. Croix, CDC, the University of South Florida, and the Presbyterian Church USA worked with the project director. The Archaie project involved sample groups in Archaie, Cabaret, and La Plaine. Archaie is the most rural of the three to wns, but it also has the largest estimated population at 100,000 (Kanda, 2004). Cabaret is sm aller than Archaie with approximately 60,000 residents (Kanda, 2004). Also, though Cabaret is more metropolitan in nature than Archaie, participants in the study lived in the more rural ar eas of this town. Lastly, La Plaine is both the smallest and most urba n of the three towns, with an estimated population of 10,000 (Kanda, 2004) The three sites identified in the Archaie project were chosen for several reasons. First, a national filariasis survey indicat ed each of these towns experience a high prevalence of LF disease (Beau de Rochar s et al., 2004). Kanda (2004, 38) further explains, The towns are also located in one of two regions which have the highest prevalence of microfilaremia. Therefore, anal ysis of the data collected in these towns would be one of the most representative info rmation about LF in Haiti. In addition to the rates of lymphatic filarias is, each of these sites had no previous exposure to clinical LF treatment programs (J. Coreil, persona l communication, April 10, 2008). Lastly, the

PAGE 90

72 national program was interested in expanding a clinical trea tment program into Archaie and was interested in the baseline data produced as part of the Archaie project (J. Coreil, personal communication, April 10, 2008). Within these sites, the Cabaret samp le had one support group while Archaie included five support groups (G. Mayard, pe rsonal communication, June 2007). Gladys Mayard provided information regarding the co ntext of the data; a description of this support group program is below. Each support group was conducted by a p eer facilitator. In order to be a facilitator, an individual mu st meet several crite ria. These requirements follow (Table 10):

PAGE 91

73 Table 10. Requirements for potential facilitators Requirements 1. The individual must be a patient herself 2. The individual must particip ate in the baseline survey 3. The individual must have time to invite other members to attend meetings 4. The individual must have time to vis it with sick patients at their homes 5. The individual must be able to produce a report, written in Kreyol, after each support group meeting. 6. The individual must be able to keep the group materials in her house. These materials included items for hygiene (soap, towels, medicines) and it ems for exercise. 7. The individual must be able to respect the rules of the groups. 8. In the event that several people were in terested and met the necessary criteria, facilitator positions were given to poor er patients participating in the group. (G. Mayard, personal communication, June, 2007) Once facilitators were chosen for each of the groups, they completed a training program. The project direct or conducted a week-long training for support group facilitators. Two animators assi sted with this process. Elem ents included in facilitator training included review of th e objectives of support groups the rules for support group meetings, discussion of the duties and respons ibilities of the faci litator position, and a review of all educational activ ities to be covered over the course of the next year. After completion of the first year of th e program, the project director initiated a second training session. Facilita tors, individual support group leaders, were evaluated for their year one performance and reviewed the ke y topics in the initia l training program. If

PAGE 92

74 facilitators received a poor evaluation, they were replaced. Two support groups received new facilitators for year two. Two animators, or assistant directors, were employed to manage multiple support groups; the animators were not diagnosed with LF. The project director trained these two individuals on how to manage a group and how to teach the subject matter in the support group materials. Each animator had a mont hly debriefing meeting with the project director. Finally, the project director also visited each group once a month. Once facilitators and animators were se lected and trained, a location for each support group meeting was identified. The pref erred location for meetings was a patients house. Facilitators conducted a site visit to evaluate th e appropriateness of the home volunteered. If the location was not acceptable, facilitators made arrangements to utilize a local school. Another important quality each meeting location must possess was a position of centrality to the participants. Because some patients had more advanced disease, a walk of no more than 15 minutes to arrive at the meeti ng was desired. Chosen locales included: patient homes, kindergarte n school, national school, and a college (G. Mayard, personal communication, June, 2007). After animators, facilitators, and mee ting locations were selected, the groups began to meet. Facilitators organized each meeting and all communication towards this end was conducted face to face. Then, the animators attended each meeting to teach the subject matter included in the curriculum for each session. Each support group had an assigned meeting day and each meeting was a minimum of two hours long. One hour was allotted for LF curriculu m instruction, and the other hour in volved patients sharing their

PAGE 93

75 experiences, refreshments, and amusement (Table 11). Topics for the next session were agreed upon by support group part icipants at the end of each reunion. These meetings took place two times a month for the first year and weekly during year two of the program. Table 11. Restatement of Support Group Meeting Agenda Support Group Meeting Agenda Prayer Singing News about health for each member of the group Teach the assigned subject in the book Discussion about the topic for that meeting Exercise and washing the leg Share experiences about the disease and social life Planning for the next meeting Refreshments Amusement gossip, talking Singing Prayer Closing and Farewell The objectives of these support grou ps were multiple (Table 12):

PAGE 94

76 Table 12. Objectives of the Support Group Objectives of the Support Group 1) To understand LF etiology and transmission 2) To learn and practice r ecommended self-care regimen 3) To recognize signs and symptoms of complaints needing medical attention 4) Reduce stigma and negative psychosocial imp act of disability (increased self-esteem, etc.) Variables of Interest Several different categories of variables are included in the analyses of the support group program described above. These categories include vari ables relevant to understanding demographic info rmation, cultural models, and cultural consonance within the sample population. The demographic variab les included in this project serve two purposes. The first is to help describe the sample population in this work. The second is to provide a means for understanding differences in cultural models or other evaluation measures along demographic dimensions. Th e leg measurement and illness history questions also provide data that can be used towards the latter purpose. The knowledge about the illness measures provide information that parallel the key questions in Kleinmans (1988) work on explanatory models. The variables that address these topics are utili zed in the cultural co nsensus analysis aimed at understanding cultural models of lymphatic filariasis. Complementing the cultural consensus analysis,

PAGE 95

77 the self-care practices variable allows for examination of the degree of cultural consonance present in the sample. The questionnaires that ar e being analyzed involve several major sections: demographics, illness history, foot exam, knowle dge about the illness, self-care practices, self-efficacy, SF-36 Scale, EuroQol, CES-D, a nd CDC Healthy Days. Only the quality of life and depression segments of the survey wi ll not be utilized. Below is a list of the variables of interest (Table 13):

PAGE 96

78 Table 13. Variables of Interest Variable Construct Scale Response Options Demographics Sex Sex Nominal Female Male Age Age Ratio 18 95 Marital Status Marital Status Nominal Married Engaged Common Law Single Live Together Other Separated/Divorced In Relationship Religion Religion Nominal Catholic No Religion Other Protestant Voudouiste Occupation Occupation Nominal Farmer Other Work Seller at home Unemployed Seller at market Tailor/ seamstress Income Income Nominal Radio Bicycle/Motorbike Living Room Storage Chest Literacy Literacy Nominal Yes No Illness History What did you think you had? Diagnosis Nominal Chill Eczema Gland Pregnancy Magical powder Insect bite Sprain An illness Big foot Filariasis Dont know Other What was the first symptom you noticed? Symptom Nominal Foot swollen Pain Swollen gland Fever Headache Foot hot Other

PAGE 97

79 Table 13. (continued) Variable Construct Scale Response Options Illness History (continued) What did you do to treat the illness? Treatment Nominal Traditional healer Herbal remedy Pommade Herbal leaves on leg Cupping/leeches Put ice on leg Pharmaceutical medicine Other What precautions do you take with your foot? Tell me everything you do for it. Treatment Nominal Apply Dolex or Pomade to the Leg Do Nothing Take Medicines: Pills, Injections Avoid Water: Cold, Rain, Dirty Avoid the Cold Wash the leg: leaves Avoid Walking on the Ground Leg Elevation Bandage Herbal Remedies Wear Clothing: Socks, Pants Massage See a Doctor Do not play football Do not hit the leg Exercises Epson Salts Do not use Leeches Do not use hot remedies Pray Sing when there is rain Wear flat shoes Keep leg clean Keep leg warm Less work in garden During the past year, how many acute attacks did you have? Severity Ratio 0-3

PAGE 98

80 Table 13. (continued) Variables of Interest Variable Construct Scale Response Options Illness History (continued) Did you buy materials for swollen leg in the past year per attack? Treatment Nominal Yes No What did you buy? Treatment Nominal Shoe/Sandal Stool Pomade Basin Bandage Other Leg Measurements Right Leg Measurements: 10 cm from toe Severity Ratio 19-42 Right Leg Measurements: 10 cm from floor Severity Ratio 18-49 Right Leg Measurements: 20 cm from floor Severity Ratio 23-54 Left Leg Measurements: 10 cm from toe Severity Ratio 18.5-38 Left Leg Measurements: 10 cm from floor Severity Ratio 17-57 Left Leg Measurements: 20 cm from floor Severity Ratio 22.5-80 Stage of Illness Right Severity Nominal 0-6 Stage of Illness Left Severity Nominal 0-6 Lesions present Right Severity Nominal Yes No Lesions present Left Severity Nominal Yes No

PAGE 99

81 Table 13. (continued) Variable Construct Scale Response Options Knowledge about the Illness Can you tell me what causes this illness? Illness Cause Nominal Insect bite Chill Magic Worms Vitamin Deficiency Other Sprain/injure foot What kinds of care can help your gwopye? Treatment Nominal Hygiene/washing Exercise Wear shoes Bandage Permanganate Medicine Crme (Salve) Nothing Elevation Other Massage Are there things you can do to prevent acute attacks? Treatment Nominal Yes No What can you do? Treatment Nominal Hygiene/washing Exercise Wear shoes Bandage Permanganate Medicine Crme Nothing Elevation Other Massage What can be done to provide relief during an acute attack? Treatment Nominal Hygiene/washing Exercise Wear shoes Bandage Permanganate Medicine Crme Nothing Elevation Other Massage

PAGE 100

82 Table 13. (continued) 09 Variable Construct Scale Response Options Self-Care Practices Tell me everything you do for your leg and how often. Treatment Nominal Hygiene Massage Wear Sandals Elevation Crme (Salve) Exercise Permanganate Bandage Herbal Remedy Medicine Pommade Other What other things can you do to help your leg that you do not currently do? Treatment Nominal Exercise Hygiene Wear Sandals Bandage Permanganate Medicine Crme Herbal Remedy Elevation Pomade Massage Other Data Quality The data resulting from these questionnair es were entered into Microsoft Excel and they were analyzed with SAS and UNIC ET 6, a program with the consensus analysis capabilities. However, the data were cl eaned before any of the information was examined. This process involved carefully reviewing the data to ensure that they were keyed correctly and are being read correctly by the computer (Hatcher & Stepanski, 1994, 98). A student at the Centers for Disease C ontrol (CDC) input the raw baseline data into a Microsoft Excel database, and the completed database was evaluated for reliability

PAGE 101

83 and validity prior to conducting analyses on the data. First, th e reliability of the measures in the dataset was assessed. In order to evaluate the accuracy of the data entry process, calculations measuring the reliability of the double entry process were conducted. This type of reliability involves capturing the degree to which two or more ratersa re consistent with each other (Tashakkori, 1998, 85). The raw data fr om 10% of the baseli ne respondents (n=25) were re-entered by the principal investigator in this project. Then, a Kappa coefficient, a method for assessing the amount of agreem ent between raters, was calculated to determine if reasonable reliability for the double entry process existed ( 90%, p<.05)(Gwet, 2002). Specifically, a simple ka ppa coefficient was em ployed as only two raters are involved (Gwet, 2002). Validity of the measures included in this study was also reviewed. Content validity was reasonable within the illness hist ory, leg measurement, knowledge of illness, and self-efficacy measures. Content validity i nvolves having experts evaluate whether or not an item measures what it is suppos ed to be measuring (Tashakorri, 1998). The research instrument was team designed by individuals possessing expertise on Haitian culture and lymphatic filariasis, respectively. The team was comprised of a Haitian culture expert, lymphatic filariasis experts, a nd local social science researchers of Haitian descent (former support group staff). Thei r previous work on support groups and lymphatic filariasis reveal local categorie s present regarding causes and treatment of lymphatic filariasis (Coreil et al., 2003; Coreil et al., 1998). In addition, the project director conducted a pilot development proc ess, to aid in instrument development,

PAGE 102

84 through open-ended interviews with provi ders and patients (J. Coreil, personal communication, Dec 3, 2007). Finally, the inst rument was pre-tested and revised accordingly before it was implemented in the current study (J. Coreil, personal communication, Dec 3, 2007). Analyses Once the quality of the data entry has b een established, data analyses aimed at addressing the research questions were c onducted. Demographic analyses are presented to provide a better understanding of the samp le population. Cultural consensus analysis explored the existence of and elements reta ined in the populations cultural model. Cultural consonance analysis examined the degree to which study participants enact behaviors incorporated in the re presentative cultural model. Demographics The demographic data allows for a description of the sample population. Univariate statistics (frequencies, means, stan dard deviations) were utilized to inform the shape of the distribution for select measur es and describe the sample population. Also, bivariate analyses were used to assess for the presence of significant demographic differences between the intervention and control groups at baseline.

PAGE 103

85 Cultural Models In order to understand the cu ltural models present in a community, it is important to understand the domains of inquiry include d in this approach. Kleinman (1998) works to achieve an understanding of illness e xperiences and social meanings through exploration of explanatory models. Explanatory models are defined as: A set of beliefs about the etiology of an illness, onset of symptoms, the pathophysiology, the course of the sickness, and the appropriate treatment that is used by the individual to interpret sy mptoms and make sense of the illness experience. (Kleinman, 1988, 121) Influences including cultural beliefs, popular culture, family and community traditions, biomedicine, and idiosyncratic experiences shape the development of these explanatory models (Kleinman, 1978, 253). The major doma ins addressed in Kleinmans concept of explanatory models included diagnosis, illness cause, reason for the onset of illness, illness manifestation, severity of the condi tion, treatment, possible treatment outcomes, help-seeking (actions taken to address illness), primary problems associated with the illness, fears about the illness, and resources available to address the illness (Kleinman, 1988, 121). The domains present in ones explanator y model were explored through several questions on the survey utilized to provide the current data. However, these questions were asked about general beliefs surroundi ng lymphatic filarias is, not about an individuals specific subjec tive lived experien ce with the same condition. As such, elements of the communitys cultural model are explored. Cultural models are schemas

PAGE 104

86 about a domain that are shared by members of a group having shared problems, shared task solutions, and similar life expe riences (Bradway & Barg, 2006). Specifically, questions regarding cause and treatment were included in this analysis of community cultural models. The in clusion of only two domains is defensible as they are arguably the two most important dimensions. Foster (1976, 775) explains the central role of illness causality: we find that the kinds of curers, the mode of diagnosis, curing techniques, preventive acts, and the relationship of a ll these variables to the wider society of which they are a part, derive from beliefs about illness causality. It is not going to far to say that, if we are given a clear de scription of what people believe to be the cause of illness, we can in broad outline fill in the other elements in that medical system. Foster (1976, 778) further indicat es that the kind of curers found in a particular society, and the curing acts in which th ey engage, stem logically from the etiologies that are recognized. Studies examining a variety of diseas e conditions exemplify the relationship between perceived cause of illness and patterns of help-seeking behavior. HeurtinRoberts & Reisin (1992) found two lay models regarding hypertens ion in an African American population, and these varying health beliefs were significan tly related to rates of treatment compliance. In the case of dia rrheal disease in Haiti, the model of oral rehydration therapy (ORT) an individual iden tified correlated significantly with time delayed before utilizing ORT therapies (Cor eil & Genece, 1988). Additionally, Mathews, Lannin, & Mitchell (1994) linked cultural models of breast cancer to delays in seeking cancer treatment among African American women. Finally, as di scussed previously, several studies on lymphatic filariasis suppor t the relationship between etiology beliefs

PAGE 105

87 and home care practices (Coreil, 1983; Ra uyajin et al., 1995; Ahorlu et al., 1999; Ramaiah et al., 1996). These cultural models of lymphatic filari asis along the dimensions of cause and treatment were investigated in this study through cultural consensus analysis. Initially, cultural consensus analysis was only used with dichotomous variables, but Romney et al. (1987) have shown that it is also a robust st atistical approach with true/false, multiple choice, and fill-in-the-blank data. Also, the us e of rank order data and interval data in cultural consensus analysis has been intr oduced (Romney et al., 1987; Weller, 1987). The data utilized in this project are appropri ate for the application of cultural consensus analysis as defined by these criteria. In this work, the approach to consensus analysis for dichotomous variables was adopted. Romney et al. (1986) provided the table below to explain the matrix of information utilized in cultural c onsensus analysis. In this table: Xik is the ith informants response to the kth question. There are N informants and M questions. The model assumes a que stionnaire where each question has L possible response alternatives with only one correct answer (Romney et al., 1986, 316). Informant Question 1 2 . K . M 1 X11 X12 . X1 k . X1M 2 X21 X22 . X2 k . X2M . . I Xi1 Xi2 . Xi k . XiM . . N XN1 XN2 . XN k . XNM (Romney et al., 1986, 316)

PAGE 106

88 This matrix provides an example of how the raw data in this study will look when it is prepared for analysis. Once the data are pr epared and cleaned, the process of cultural consensus analysis will commence. When conducting cultural cons ensus analysis, there are three key components in the analysis: the overall shared consensus in a sample population, individual levels of cultural competence, and the answer key deve loped from the data itself (Romney et al., 1986). In order to assess for consensus, Romn ey et al. (1986) apply the minimal residual method of factor analysis Factor analysis approaches are utilized as a va riable reduction procedure (Hatcher and Stepanski, 1994). This factor analysis pro cess yields estimates of each informants competence on this set of questions (Romney et al., 1986, 322). As each informants competence is plotted, an underlying factor structure emerges. In cultural consensus analysis, a single factor st ructure is desired. Romn ey et al. (1986, 323) explain that a single factor stru cture in cultural consensus anal ysis indicates that a single underlying all-positive factor, in our case competence, accounts for several times as much variance as the next factor. When a single fa ctor structure is present, it implies that the informants are referring to a shared cultura l domain when they produce responses to the items (Romney et al, 1986.). The second component, cultural competence, is a measure of the individuals knowledge as compared to the shared cultu ral knowledge base (Weller, 1987). To estimate competence, or a persons cultural ex pertise, the process is slightly different from the one used to assess the portion of correct answers for a single individual when the correct answers are know n. When the correct answers are known a priori, it is

PAGE 107

89 possible to assess each indivi duals competence without linki ng it to another informants answers (Romney et al., 1986). However, when there is not an a priori answer key, the only way to estimate the proporti on of correct answers for an informant is to examine one individuals responses in comparison to all other informants responses (Romney et al., 1986). Regarding the answer key, Romney et al. (1986) outline a Bayes probability framework and illustrate the process with an example with only two informants and one true-false question. The authors explain th at, for example, if you already know the individual competencies of each informant ( .00-1.0) and the a priori probability that a question is answered correctly (.5 for each possi ble response, true [1] or false [0]) then it is possible to solve for the probability of a response pattern where the correct answer is true (1, 1 or where both informants pr oduced the correct answer). Once these probabilities are arrived at, then it is possible to assess the ov erall likelihood of a true or false correct answer for all informants in the dataset, not just the two informants used to arrive at the information above. When the posteriori probabilities are calculated, it is possible to discern the culturally correct an swer to each item for the overall informant pool. In this case, the response with the highest posteriori probability is the culturally correct response (Romney et al., 1986). In order to execute the analysis to pr oduce these three components, three major assumptions exist. These assumptions ar e common truth, local independence, and homogeneity of items (Romney et al, 1986). Comm on truth indicates that there is a fixed answer key applicable to all informan ts (Romney et al, 1986, 317), and local

PAGE 108

90 independence assumes that each informants answers are given independently of each other informant (Romney et al., 1986, 317). Finally, homogeneity of items assumes that questions are all of the same diffic ulty level (Romney et al., 1986, 318). In the formal model, items relative to a domain of interest are identified by the population of interest. Romney (1999) explains this proc ess by describing Wellers (1984) work with Guatemalan women in brief. The first step in preparing items that can be analyzed using consensus analysis involved identifying which items provide appropriate measures. In order to achie ve this goal, Weller asked a group of 20 Guatemalan women to free list all known contagious diseases (Romney, 1999). These women identified a number of diseases, and, ulti mately 27 diseases were identified by at least 15% of the population (Romney, 1999). These 27 remaining items from the free listing exercise were incorporated into th e instrument for Wellers work on cultural consensus. The goal of the free-listing component is to produce an exhaus tive list of possible responses in a particular domain. Then res ponses identified by a moderate proportion of community members are retained for the anal ysis. The present data are secondary in nature, and this free listing ex ercise was not possible. Howeve r, the responses included in the instrument were generated in an ethnogr aphic survey conducte d in 1996 (J.Coreil, personal communication, 2008). Thus, previous work has approximated the free-listing process. The formal cultural consensus model is appr opriate in this study as the data were measured at the nominal level and the data can be transformed to a dichotomous response

PAGE 109

91 format (Weller, 1987). The data related to cu ltural models were collected in a multiple choice format, where more than one response selection was allowed for each question; these questions, each of which has L possibl e responses, were first converted to a true/false format. For example, if item 1 has 6 possible response categories, each informant was described as to whether or not they indicated each possible response as an answer (1) or not (0) in the response matri x. Each possible response was converted into a true/false format where only two values (1,0) were possible. In this case, item 1 was converted into six variables, 1a-1f. Once the data were transformed into i ndividual, dichotomous variables, the UNICET 6 statistical software program was used to c onduct the cultural consensus analysis. Now that the process of cultural c onsensus analysis for th is particular problem has been described, it is important to review the variables included in the cultural model portion of the analysis (Table 14):

PAGE 110

92 Table 14. Variables used in Cultural Consensus Analysis Variable Construct Data Available Number of variables after transformation Pre-Post Analyses 1) Can you tell me what causes this illness? Cause Baseline and Outcome 6 variables 2) What treatment options can be used for your foot? Treatment Baseline and Outcome 11 variables Post Analyses Only 3) What kinds of care can help your gwopye? Treatment Outcome Only 10 variables 4) What can you do to prevent acute attacks? Treatment Outcome Only 10 variables 5) What can be done to provide relief during an acute attack? Treatment Outcome Only 10 variables Weller (2007) suggests analyses that involve 20 variables or more provide reasonable estimates in cultural consensus analysis. Seventeen and forty-seven variables were present for analysis in the baseline and outcome data, respectively. Variable two in the table above was devel oped by the combination of the two following variables: tell me everything you do for your leg and how often, what other things can you do to help your leg that you do not curren tly do? Though less data are available in the baseline survey, the analyses were sti ll valid. Weller (persona l communication, May 15, 2008), co-founder of cultural consensus anal ysis, indicated that 17 variables was close enough to 20 to provide reasonable out comes. Pre-post comparative analyses were conducted on the first two items identified in the table; outcome analyses were only run

PAGE 111

93 on all the items in the table. The elements retained in the full and reduced outcome analyses were also compared. Cultural consensus analysis was run on the baseline and outcome data, respectively, by town, Archaie, Cabaret, La Pl aine. Within the baseline data, the results were compared, with independent samples t-te sts, for significant differences in cultural competence. Then, chi-square analyses tested the culturally correct answer keys for significant differences. At base line the participants in the matched intervention and control groups across the three towns were assessed for these differences. If the null hypotheses were supported, the baseline data from all three towns could be combined in additional analyses. In the next step, data were assessed with the baseline data only for all baseline informants, matched intervention only, a nd matched controls only. Similarly, the outcome data were examined for cultural models for matched controls only and matched intervention informants only. Cultural consen sus analysis was run on both the reduced, items available at both times, and full, in cluding additional CM items present in the outcome data only, data sets; the analysis on the reduced data allowed for comparison between the baseline and outcome cultural m odel results. Once these analyses were completed, these same samples were assessed along demographics and other dimensions. These dimensions included: age, marital status religion, wealth, liter acy, stage of disease, and number of acute attacks. The presence of a shared cultural m odel was evaluated in each of these dimensions. In addition, independent samples ttest comparisons were utilized to compare

PAGE 112

94 baseline and outcome data for significant differences in mean cultural competency scores. Paired sample t-tests were used to expl ore significant differences in the matched intervention and control samp les from baseline to outcome. A chi-square comparison assessed for significant differences in the cu lturally correct answer key for matched intervention and control samples from baseline to outcome as well. Cultural Consonance After the cultural consensus analyses we re conducted, it wa s possible to next examine the levels of cultural consonance in the various samples. The degree to which an individuals behaviors corre sponds with elements in the cultural model, cultural consonance, is quantified as a percent coe fficient, or percentage of behaviors an individual engages in out of all possible behaviors. The percentage is an approximation of the degree to which individuals behaviors ar e in line with the cultural ideal, or the elements included in the cultural model iden tified through cultural consensus analysis (Dressler, 1996; Dressler, Bindon, & Neggers, 1998). Values can range from 0 100%, where a score of 100% indicates an individual is behaving in complete alignment with the elements identified in the communitys cultur al model of LF (Dressl er et al., 1996). Also, if the sample is examined as a whole, the average rate of consonance can be calculated, indicating the percentage of ideal behaviors an average person in the community enacts. The cultural consonance methodology has pr eviously been ap plied in several different domains. Dressler utilized this measure to examine the relationship between

PAGE 113

95 cultural consonance, as relate d to cultural model of lifesty les (1996), access to social support (Dressler, Balieiro, & Dos Santos, 1997), and blood pressure. He has also explored the relationship betw een cultural consonance of lif estyles and coronary heart disease (Dressler et al., 1998). In addition, Chavez et al. (2001) investigated the relationship between cultural consonance and cervical cancer-s creening. Though these and several other studies have employed the cultural consonance approach, this methodology has not yet been implemented in either the Haitian population or in a community of individuals w ith lymphatic filariasis. In this project, self-care behaviors pr esent in the cultural model of lymphatic filariasis were examined. The degree to wh ich an individuals behaviors correspond with the treatment options present in the cultu ral model was calculated. Additionally, the samples average rate of consonance was cal culated, indicating the percentage of ideal behaviors an average person in the community enacted relate d to self-care and lymphatic filariasis. Cultural consonance analyses were run on the same samples and sub-samples described in the cultural model segment. In this way, the extent of cultural consonance was established in tandem with the cultura l model findings. For the comparisons between samples and sub-samples, independent and paired samples t-tests were used, as appropriate, to test for signi ficant differences in mean percentage coefficients.

PAGE 114

96 Sample Size In order to determine the sample size needed to appropriate ly conduct cultural consensus analysis, several factors were cons idered: cultural competence of the sample participants, confidence level, and the proportion of questi ons that must be clearly classified (Romney et al., 1986). With a higher level of cultural competence, fewer individuals are needed (Romney et al., 1986). Alternately, the higher the desired confidence level, the larger the sample size must be (Romney et al., 1986). Finally, the more questions that must be correctly classi fied yields a larger sample size (Romney et al., 1986). Romney et al. (1986) explain how the appr opriate sample size was derived. First, the a posteriori probability for a true answer equals the confidence level that a question is decisively classified in the culturally correct answer key. For instance, if the a posteriori probability of a cultura lly correct answer being given is .95, then there is only a 5% chance that a respondent will an swer that question incorrect ly. In order for a question to be considered decisively classified, where one of the two possible responses is considered clearly correct, a minimu m confidence level accepted is .80. Once the culturally correct answers to the instrument are determined at .80 threshold, then it is important to examine th e proportion of questions overall that meet this criteria. If 95% of the questions in th e instrument meet these parameters, then the proportion of questions decisi vely classified is .95.

PAGE 115

97 After the proportion of questions decisi vely classified is determined, then Romney et al.s (1986) table indicating sample size requirements is utilized. This table indicates the sample size required to attain the desired confidence level in cultural consensus data (Table 15). Table 15. Sample Size Table for Cultural Consensus Analysis Proportion of Questions Average level of cultural competence .5 .6 .7 .8 .9 .90 Confidence Level .80 9 4 4 4 4 .85 11 6 4 4 4 .90 13 6 6 4 4 .95 17 10 6 6 4 .99 25 16 10 8 4 .95 Confidence Level .80 9 7 4 4 4 .85 11 7 4 4 4 .90 13 9 6 4 4 .95 17 11 6 6 4 .99 29 19 10 8 4 .99 Confidence Level .80 15 10 5 4 4 .85 15 10 7 5 4 .90 21 12 7 5 4 .95 23 14 9 7 4 .99 20 13 8 6 .999 Confidence Level .80 19 11 7 6 4 .85 21 13 8 6 4 .90 23 13 10 8 5 .95 29 17 10 8 5 .99 23 16 12 7 Note: Well over 30 informants needed. (Romney et al., 1986, 326) Using this table, a sample size of onl y 23 was required when the following conditions were met: .5 level of cultural competence, .999 confidence level, and .90

PAGE 116

98 proportion of questions correctly classified. In this case, each informant was assumed to have a confidence only slightly better than guessing, a higher threshold for questions decisively classified was set, and the highest confidence le vel possible was indicated. If concordance rates are high within the sample and less stringent parameters are set on determining sample size, it is possible a sample size as small as 4 informants could yield the same results. Supplementing the standard cultural consensus sample size parameters, Sue Weller (personal communication, May 4, 2007), co-founder of the cultural consensus approach, recommended a sample size of 60 in this project because change over time was being investigated; the interv ention sample, with N=60, met th is criteria. Thus, for all research components, the sample size was adequate.

PAGE 117

99 Chapter 4: Results

PAGE 118

100 Univariate, bivariate, cultural consensus, and cultural consonance analyses were conducted in order to describe the sample population and to a ddress the research questions of the study. The results will be discus sed in relation to the specific study aims. The research questions are presente d here for reference (Table 16):

PAGE 119

101 Table 16. Review of Research Questions and Hypotheses Research Question Hypothesis Does a cultural model exist in the sample population at baseline? If so, does the CM meet the threshold for cultural consensus at baseline? If so, what is the strength of the consensus? It is expected that a shared cultural model of LF will exist at baseline. Does a cultural model exist in the sample population at outcome? If so, does the model meet the threshold for cultural consensus at outcome? If so, what is the strength of the consensus? It is expected that a shared cultural model of LF will exist at outcome. Are there significant changes in levels of cultural competency existing at baseline and outcome? It is expected that levels of cultural competency will be greater at outcome than baseline. Are there significant differences in the elements included in the CMs from baseline and outcome points? Is the cultural model present at baseline different from the cultural model present at outcome? It is expected that the CM at outcome will be significantly different than the CM present at baseline. The outcome CM is expected to include more elements of western biomedical beliefs than the CM at baseline. It is expected that a bicultural model will exist at outcome incorporating elements of traditional and western biomedical ideas about LF. Is the strength of cultural consensus for the CM greater at outcome than baseline? It is expected that the strength of consensus for the CM at outcome will be greater than the strength of consensus for the CM at baseline. Does a significant link between belief (cultural model) and behavior (cultural consonance) exist in the sample population? It is expected that greater consensus regarding cultural models will be linked to higher rates of self-care behaviors identified in the cultural model.

PAGE 120

102 Integrity of the Data The accuracy of double entry in the baseli ne data was tested for reliability to ensure the secondary data provided at the baseline point was accurate. The overall sample at baseline was comprised of 241 individuals; thus, 10% of th e baseline surveys were reentered from the raw data into an excel da tabase. A simple kappa co-efficient was run on 2 raters, 25 surveys, 45 variables, and high levels of reliability for the double entry process resulted ( =.99, p<.0001). Additionally, the outcome data were re-entered in its entirety into the excel databa se ensuring accurate and consiste nt recording of the values from the raw data. Variables included in the analyses e xhibited good content validity, assessed by asking experts to evaluate whether or not an item measures what it is supposed to be measuring (Tashakorri, 1998), as the cause a nd treatment variables on the survey were developed by a team of experts including a Ha itian culture expert, lymphatic filariasis experts, and local social science research ers of Haitian descent (former support group staff). Additionally, the teams expertise was further informed by previous research on local categories surrounding lymphatic filariasis (Coreil et al., 2003; Coreil et al, 1998), and both a pilot development and pre-test proces s were executed in the current project (J. Coreil, personal communication, Dec 3, 2007).

PAGE 121

103 Demographic Analyses Descriptive statistics provide general in formation on the sample and sub-samples investigated in this project. Groups of interest include all individuals in the baseline data (B-All), baseline matched controls only (B MC), baseline matched intervention only (BMI), outcome matched controls only (OMC ), and outcome matched intervention only (OMI). These five sample groups are desc ribed in the tables below (Table 17):

PAGE 122

104 Table 17. Demographics: Baseline Data Demographics B-All (N=241) BMC (N=27) BMI (N=60) N (%) N (%) N (%) Town Archaie 89(37%) 6(22%) 40(67%) Cabaret 50(21%) 2(7%) 20(33%) La Plaine 102(42%) 19(70%) 0(0%) Marital Status Married, Partnered, Live Together 142(59%) 21(78%) 32(53%) In Relationship, Engaged 17(7%) 2(7%) 1(2%) Single, Separated, Divorced, Widowed 80(33%) 4(15%) 27(45%) Religion Catholic 118(49%) 15(56%) 27(45%) Protestant 96(40%) 6(22%) 28(47%) Voudouiste 10(4%) 2(7%) 3(5%) None, Other 17(7%) 4(15%) 2(3%) Literacy Literate 137(57%) 15(56%) 38(63%) Wealth > 2 Wealth Items 69(29%) 3(11%) 25(42%) (radio, storage set, living room, bicycle/motorcycle) < 2 Wealth Items 172(71%) 24(89%) 35(58%) Occupation Farmer 14(6%) 1(4%) 1(13%) Seller at Home or Market 111(46%) 12(44%) 29(48%) Tailor/Seamstress 14(6%) 2(7%) 2(3%) Other 33(14%) 4(15%) 6(10%) Unemployed 69(29%) 8(30%) 15(25%) Mean(SD) Mean(SD) Mean(SD) Age 47(16.6) 46(14.9) 47(15)

PAGE 123

105 At baseline (B-All), the majority of part icipants were married, Catholic, literate, less wealthy, and engaged in selling at home or the market (Table 17). When baseline intervention (BMI) and cont rol (BMC) groups were compared, the control group was more likely to be married while members of the intervention group tended to be single, separated, divorced, or widowed. The inte rvention and control groups were not significantly different from one another on the dimensions of religion (chi-sq=3.07,df=1, p<.08), where vodouism, no religion, and other were removed from the comparison due to small cell sizes, income(chi-sq=3.72,df= 4,p<.45), literacy(chi-sq=.47,df=1,p<.49), and age(t=-.29,df=85,p<.77).

PAGE 124

106 Table 18. Demographics: Outcome Data Between baseline and outcome, most of th e trends noted in the baseline data remained the same (Table 18). At outcome, intervention (OMI) and control groups (OMC) at outcome were not significantly di fferent on the following dimensions: marital status(chi-sq=.96,df=1,p<.33) where in a rela tionship and engaged were collapsed with the married/partnered/living together category due to small cell sizes, religion(chisq=.66,df=1,p<.42), where voudism, no religion, and other were removed from the OMC (N=27) OMI (N=60) Demographics N (%) Town Archaie 6(22%) 44(73%) Cabaret 3(11%) 15(25%) La Plaine 18(67%) 1(2%) Marital Status Married, Partnered, Live Together 17(63%) 31(52%) In Relationship, Engaged 0(0%) 0(0%) Single, Separated, Divorced, Widowed 10(37%) 29(48%) Religion Catholic 14(52%) 29(48%) Protestant 9(33%) 28(47%) Voudouiste 0(0%) 0(0%) None, Other 4(15%) 3(5%) Literacy Literate 11(42%) 37(62%) Wealth > 2 Wealth Items 3(11%) 25(42%) (radio, storage set, living room, bicycle/motorcycle) < 2 Wealth Items 24(89%) 35(58%) Occupation Farmer 1(4%) 5(8%) Seller at Home or Market 12(44%) 24(40%) Tailor/Seamstress2(7%) 4(7%) Other 12(44%) 27(45%) Unemployed 0(0%) 0(0%) Mean(SD) Age 46 (14.9) 47(14.5)

PAGE 125

107 comparison due to small cell sizes, in come(chi-sq=3.72,df=4,p<.45), literacy(chisq=2.76,df=1,p<.10), and age(t=.37,df=85,p<.71). In addition to the basic demographics of the sample populations, the baseline data included information on each participants person al experience with lymphatic filariasis. The table below provides information regarding the illness profile of the baseline sample population along multiple dimensions: age the in dividual first knew of their illness, their first impression of the illness, first symptom noticed, treatment choices, stage of disease, presence of lesions, number of acute attacks in the past year, and foot, ankle, and leg sizes for each leg, respectively. The outcome sa mple is also profiled in the table below on a more limited basis (Table 19).

PAGE 126

108 Table 19. Illness Beliefs: Baseline Data BL All (N=241) BL Controls (N=27) BL Intervention (N=60) N (%) Cause Insect Bite 7(3%) 0(0%) 3(5%) Magic 24(10%) 4(15%) 7(12%) Sprain 23(10%) 1(4%) 4(7%) Worms 16(7%) 4(15%) 1(2%) Chill 29(12%) 1(4%) 6(10%) Vitamin Deficiency 3(1%) 1(4%) 0(0%) Other 32(13%) 0(0%) 13(22%) Dont Know 139(58%) 17(63%) 37(62%) Treatment Hygiene 189(78%) 22(81%) 45(75%) Sandal 206(85%) 24(89%) 50(83%) Permanganate 36(15%) 5(19%) 6(10%) Crme 44(18%) 5(19%) 12(20%) Elevation 67(28%) 7(26%) 19(32%) Massage 23(10%) 4(15%) 3(5%) Exercise 16(7%) 1(4%) 7(12%) Bandage 49(20%) 6(22%) 9(15%) Medicine 58(24%) 3(11%) 12(20%) Herbal Remedy 132(55%) 14(52%) 31(52%) Pomade 87(36%) 8(30%) 22(37%) Other 102(42%) 9(33%) 34(57%) Overall, the baseline sample (B-All) was, on average, in their late twenties when they first knew of their illness and there were not significant di fferences between the control (BMC) and interv ention (BMI) groups on this measure (t=-.07,df=70,p<.94). These participants had about one acute attack in the past year and exhibited stage one and stage two disease in the right and left legs respectively. Most indi viduals did not have

PAGE 127

109 lesions on their affected legs. Control a nd intervention groups were not significantly different for the number of acute attacks they experienced(t=-1.30,df=84,p<.20), size of the ankles(Left: t=1.67,df=32.2,p<.11; Right: t= 1.77,df=84,p<.08), and size of the right leg (t=.99,df=84,p<.32). When individuals realized they were sic k, most self-reported that their condition was due to a chill or gland issues. The first symptoms most noticed in association with the illness included a swollen foot, pain, a nd swollen glands. Treatment options used most frequently to address this health concer n were herbal remedies, seeking out a health professional, and use of pharmaceutical medicines. With regards to age of onset, number of acute attacks, presence of lesions, and disease stage of the right and left legs, th e intervention (BMI) and control (BMC) groups at baseline followed the same trends as the overall sample (B-All). However, there were some differences between the control and in tervention groups, respec tively, regarding the first impression of the illness and treatment choices. The control group tended to identify their condition as rooted in bad blood or magical powder, while the intervention group attributed their problems to bad blood or a chill. Herbal remedies and cupping/leeching were most used to treat the control group s problems at baseline, and the intervention group identified the same treatment choices as the overall baseline sample: herbal remedies, aid of a health professional, pharmaceutical medicines.

PAGE 128

110 Table 20. Illness Profile: Outcome Data Illness Profile: Outcome Data OMC (N=27) OMI (N=60) Mean(SD) Number of Acute Attacks in the Past Year 1.33(.68) 1.58(.86) Foot Size: Right 25.85(2.26) 23.61(2.27) Foot Size: Left 26.42(3.87) 23.96(2.27) Ankle Size: Right 27.17(4.35) 25.52(3.83) Ankle Size: Left 28.94(7.25) 26.41(4.13) Leg Size: Right 37.19(5.23) 35.9(5.66) Leg Size: Left 39.27(7.54) 35.61(5.44) N(%) Number of Acute Attacks in the Past Year 0 12(44%) 22(37%) 1 11(41%) 29(48%) 2 3(11%) 5(8%) 3 1(4%) 4(7%) Stage of Disease: Right 0 5(19%) 7(12%) 1 3(12%) 18(31%) 2 9(35%) 19(32%) 3 8(31%) 13(22%) 4 0(0%) 2(3%) 5 1(4%) 0(0%) 6 0(0%) 0(0%) Stage of Disease: Left 0 3(12%) 3(5%) 1 8(31%) 20(34%) 2 4(15%) 21(36%) 3 5(19%) 12(20%) 4 2(8%) 3(5%) 5 4(15%) 0(0%) 6 0(0%) 0(0%) Lesions Yes 10(43%) 21(40%) No 13(57%) 31(60%)

PAGE 129

111 In the outcome samples, the control group (OMC) did not generally have an acute attack within the past year while the intervention group (OMI ) had about one acute attack in the same time period (Table 20). Both gr oups showed stage two disease in the right leg, and in the left leg controls (OMC) a nd intervention (OMI) pa rticipants tended to have stage one and two disease, respectiv ely. In the outcome intervention (OMI) and control group (OMC) samples, differences we re assessed; no differences between these two groups for the number of acute att acks experienced in the past year(t=.57,df=85,p<.57), for the sizes of the foot(Right: t=1.92,df=85,p<.06; Left: t=1.75,df=37.9,p<.09), the size of the le gs(Right: t=1.49,df=34.2,p<.14; Left: t=1.82,df=30.8,p<.08), stage of illness(Ri ght: chi-sq=.75,df=1,p<.39; Left: chisq=2.42,df=1,p<.12), or lesi ons(chi-sq=.06,df=1,p<.80). Study participants were asked about bot h their personal il lness experience, baseline samples only, and their general be liefs about what could cause lymphatic filariasis and how one might treat it. The di stribution of these general belief categories are outlined below for both the baseline and outcome samples.

PAGE 130

112 Table 21. Illness Beliefs: Baseline Data Illness Beliefs B-All (N=241) BMC (N=27) BMI (N=60) N (%) Cause Insect Bite 7(3%) 0(0%) 3(5%) Magic 24(10%) 4(15%) 7(12%) Sprain 23(10%) 1(4%) 4(7%) Worms 16(7%) 4(15%) 1(2%) Chill 29(12%) 1(4%) 6(10%) Vitamin Deficiency 3(1%) 1(4%) 0(0%) Other 32(13%) 0(0%) 13(22%) Dont Know 139(58%) 17(63%) 37(62%) Treatment Hygiene 189(78%) 22(81%) 45(75%) Sandal 206(85%) 24(89%) 50(83%) Permanganate 36(15%) 5(19%) 6(10%) Crme 44(18%) 5(19%) 12(20%) Elevation 67(28%) 7(26%) 19(32%) Massage 23(10%) 4(15%) 3(5%) Exercise 16(7%) 1(4%) 7(12%) Bandage 49(20%) 6(22%) 9(15%) Medicine 58(24%) 3(11%) 12(20%) Herbal Remedy 132(55%) 14(52%) 31(52%) Pomade 87(36%) 8(30%) 22(37%) Other 102(42%) 9(33%) 34(57%) In the baseline data, the majority of re spondents indicated that they did not know what caused lymphatic filariasis (Table 21) However, there were some clear thoughts regarding actions people can take to treat this disease including the use of sandals, hygiene, and herbal remedies, respectivel y. Members of the control and intervention groups exhibited the same beliefs about illne ss cause as the entire baseline sample (BAll), and the control group (BMC ) also identified the same three treatment options as the

PAGE 131

113 entire baseline sample. The intervention group (BMI) varied slightly as other was chosen as a key option in addition to the use of sandals and hygiene.

PAGE 132

114 Table 22. Illness Beliefs: Outcome Data OMC (N=27) OMI (N=60) N (%) Cause Insect Bite 1(4%) 37(62%) Magic 1(4%) 4(7%) Sprain 0(0%) 4(7%) Worms 4(15%) 7(12%) Chill 0(0%) 0(0%) Vitamin Deficiency 0(0%) 0(0%) Other 21(78%) 19(32%) Treatment Hygiene 23(88%) 53(88%) Sandal 23(88%) 54(90%) Permanganate 2(8%) 15(25%) Crme 1(4%) 27(45%) Elevation 5(19%) 49(82%) Massage 2(8%) 34(57%) Exercise 1(4%) 46(77%) Bandage 4(15%) 13(22%) Medicine 11(42%) 25(42%) Herbal Remedy 10(38%) 36(60%) Pomade 9(35%) 25(42%) Other 5(19%) 24(40%) Treatment (OC Only) Hygiene 18(67%) 51(85%) What kinds of care can help your gwopye? Sandal 15(56%) 36(60%) Permanganate 1(4%) 17(28%) Crme 3(11%) 24(40%) Elevation 6(22%) 46(77%) Massage 2(7%) 34(57%) Exercise 2(7%) 44(73%) Bandage 3(11%) 12(20%) Medicine 9(33%) 24(40%) Nothing 2(7%) 0(0%) Other 8(30%) 15(25%)

PAGE 133

115 Table 22. Illness Beliefs: Outcome Data (continued) OMC (N=27) OMI (N=60) N (%) Treatment (OC Only) Hygiene 12(44%) 42(70%) What can you do to prevent acute attacks? Sandal 12(44%) 34(57%) Permanganate 1(4%) 7(12%) Crme 1(4%) 11(18%) Elevation 3(11%) 33(55%) Massage 0(0%) 27(45%) Exercise 0(0%) 34(57%) Bandage 0(0%) 6(10%) Medicine 3(11%) 20(33%) Nothing 2(7%) 0(0%) Other 4(15%) 17(28%) Treatment (OC Only) Hygiene 14(52%) 38(63%) What can be done to provide relief during an acute attack? Sandal 12(44%) 25(42%) Permanganate 0(0%) 13(22%) Crme 1(4%) 13(22%) Elevation 4(15%) 31(52%) Massage 1(4%) 27(45%) Exercise 1(4%) 29(48%) Bandage 3(11%) 9(15%) Medicine 13(48%) 26(43%) Nothing 1(4%) 1(2%) Other 7(26%) 18(30%) In the outcome data, the population suggest ed two primary causes of lymphatic filariasis: other causes and insect bites (Table 22). The control group (OMC) chose other as the prominent cause of lymphatic fi lariasis. In contrast, the intervention group (OMI) highlighted insect bites a nd other causes, respectively.

PAGE 134

116 When asked about general treatment, the outcome control group (OMC) selected sandals, hygiene, and pharmaceutical medicine s as important options. The intervention group (OMI) responded differently with these treatment possibilities: sandals, hygiene, elevation, and exercise. The outcome sample was asked three a dditional questions on general beliefs about treating lymphatic filari asis. The first additional ques tion was what kinds of care can help your gwopye (LF). The control group (OMC) chose hygiene, sandal, and the use of pharmaceutical medicines. In contrast, part icipants in the intervention (OMI) sample pointed to hygiene, sandals, elevation, and exercise were selected most often. The next additional treatment question used in the outcome sample was what can you do to help prevent acute attacks? Res ponding to this question, the control group (OMC) suggested hygiene and sandals, a nd the intervention (OMI) group indicated hygiene, sandals, exercise, elevation, and massage as viable alternatives. The final question asked regarding cultural models of LF in the outcome sample was what can be done to provide relief during an acute attack? Spec ific to the control group (OMC) hygiene, medicine, sandal were iden tified as treatment possibilities. Lastly, the intervention group (OMI) s uggested hygiene, elevation, exercise, and massage as ways to provide relief for an acute attack.

PAGE 135

117 Cultural Models at Baseline Cultural Consensus Analysis by Baseline Locale After the demographic analyses were completed, cultural consensus and consonance analyses were conducted. Before grouping the data into intervention and control groups for examination, the cultural mode ls present in each town at baseline were reviewed. This step was necessary in orde r to ensure homogeneity of cultural models existing across the three sites, Archaie, Cabaret, and La Plaine, at baseline. First, a consensus analysis was run on each of the sites; Archaie (n=89) failed to meet consensus as the eigenvalue ratio (ER) between first and second eigenvalues was 2.52, suggesting the lack of a sh ared cultural model in this location. Cabaret (n=50) and La Plaine (n=102) exhibited consensus w ith ER of 4.0 and 3.2, respectively. Though all three towns did not exhibit consensus for a si ngle cultural model, the criterion required to combine the towns for further analyses invol ved a review of the culturally correct answer keys (AK). If the elements of the AK, or elements identified as part of the local CM, were not significantly different from one another by location, regional variation could be disregarded. Chi-square analyses resulted in the interpretation of the Fishers exact statistic as the data involved small ce ll sizes (Hatcher & St epanski, 1994), and the findings suggest no significant differences exist between the AK in the three locales (Table 23).

PAGE 136

118 Table 23. Culturally Correct Answer Keys by Town Archaie (N=89) Cabaret (N=50) La Plaine (N=102) Answer (Weighted %) Answer (Weighted %) Answer (Weighted %) Cause Insect Bite 0(95%) 0(98%) 0(98%) Magic 0(87%) 0(96%) 0(95%) Sprain 0(91%) 0(86%) 0(96%) Worms 0(92%) 0(100%) 0(92%) Chill 0(80%) 0(90%) 0(93%) Vitamin Deficiency 0(96%) 0(100%) 0(100%) Treatment Hygiene 1(87%) 1(91%) 1(96%) Sandal 1(93%) 1(97%) 1(100%) Permanganate 0(92%) 0(90%) 0(73%) Crme 0(83%) 0(93%) 0(72%) Elevation 0(71%) 0(82%) 0(62%) Massage 0(97%) 0(99%) 0(80%) Exercise 0(94%) 0(95%) 0(94%) Bandage 0(86%) 0(86%) 0(70%) Medicine 0(78%) 0(74%) 0(76%) Herbal Remedy 1(66%) 1(63%) 0(52%) Pomade 0(57%) 1(60%) 0(76%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** Weighted % values are derived from the frequency with which a response was given in the sample population to a given item ad justed for the varying degrees of cultural competency of sample particip ants. Increased weight was al located for the responses of sample participants who possessed higher cultural competency.

PAGE 137

119 Table 24. Comparisons of Culturall y Correct Answer Keys by Town Chi-Square df p< Fishers Exact Two-Tailed p< Archaie vs. Cabaret .1799 1 .67 1.00* Archaie vs. La Plaine .2345 1 .63 1.00* Cabaret vs. La Plaine .8095 1 .37 .66* In addition to the evidence provided by th e comparison of the answer keys, each town was also assessed for significant differe nces in mean cultural competency scores (Table 24), where cultural competency desc ribes the degree of cultural expertise any given sample participant has regarding the cu ltural domain in question (e.g.: lymphatic filariasis). Archaie and Caba ret show borderline significant differences in levels of competency ( t=-1.94, df=137, p<.054) while Archaie and La Plaine clearly exhibit a lack of significant differences in individual cultur al competency levels (t=-1.54, df=189, p<.1249). Likewise, Cabaret a nd La Plaine are not significantly different in the average levels of cultural competency present in each town (t=.82, df=150, p<.4119). Taken as a whole, the lack of clearly significant differences between these three towns further suppor ts the grouping of the sample populations in all additional analyses.

PAGE 138

120 Baseline Cultural Consensus Analysis Com parison by Matched Group: Intervention and Control After examining the baseline sample by town, analyses were conducted by intervention and control group designation. C onsensus analysis was run on the following samples: baseline matched controls (BMC), baseline matched inte rvention (BMI), entire baseline sample (B-All). Additionally, groups were compared for significant differences between levels of cultural competency as well as between elements of the culturally correct answer key. Table 25. Baseline: Consensus Analysis Results by Group N First Eigenvalue Second Eigenvalue Eigenvalue Ratio BMC 27 7.829 3.248 2.410 BMI 60 17.038 7.015 2.429 B-All 24171.840 24.892 2.886 *One individual removed from analyses in the outcome control group due to missing data As the eigenvalue ratio between the firs t and second eigenvalues was not equal to or greater than three, the threshold for c onsensus was not met for all three baseline samples (Table 25). Once the degree of consensus in a sample was established, the answer keys were compared for significant di fferences. The answer keys and comparative analyses are below:

PAGE 139

121 Table 26. Culturally Correct Answer Ke ys by Intervention and Control Group Baseline Sample BMC (N=27) BMI (N=60) B-All (N=241) Answer (Weighted %) Answer (Weighted %) Answer (Weighted %) Cause Insect Bite 0(100%) 0(95%) 0(97%) Magic 0(82%) 0(93%) 0(93%) Sprain 0(98%) 0(96%) 0(92%) Worms 0(86%) 0(99%) 0(94%) Chill 0(96%) 0(90%) 0(88%) Vitamin Deficiency 0(96%) 0(100%) 0(99%) Treatment Hygiene 1(93%) 1(91%) 1(92%) Sandal 1(98%) 1(97%) 1(97%) Permanganate 0(81%) 0(88%) 0(84%) Crme 0(83%) 0(76%) 0(82%) Elevation 0(75%) 0(63%) 0(70%) Massage 0(84%) 0(95%) 0(91%) Exercise 0(97%) 0(86%) 0(94%) Bandage 0(77%) 0(86%) 0(80%) Medicine 0(89%) 0(82%) 0(77%) Herbal Remedy 0(51%) 0(53%) 1(58%) Pomade 0(69%) 0(66%) 0(61%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met In the baseline sample, all groups iden tify hygiene and sandals as a treatment choice (Table 26). When the full baseline samp le (B-All) is reviewed, herbal remedies are also indicated as a viable treatment option.

PAGE 140

122 Table 27. Baseline Sample: Comparisons of Culturally Correct Answer Keys by Intervention and Control Groups ChiSquare df p< Fishers Exact Two-Tailed p< BMC vs. B-All .2345 1 .6282 1.00* BMI vs. B-All .2345 1 .6282 1.00* BMC vs. BMI .0000 1 1.0000 1.00* indicates which statistic was interpreted for significant differences between the answer keys **In the BMC vs. B-All comparisons the B-All sample was reduced to an n=214 (removed the control individuals) so answer keys could be examined absent any overlapping individuals *** In the BMI vs. B-All comparisons the B-All sample was reduced to an n=181 (removed the intervention individuals) so an swer keys could be examined absent any overlapping individuals Chi-square analyses were conducted to examine the differences between answer keys (Table 27). In some cases, this an alysis was appropriate; however, in other comparisons, small cell sizes resulted in the interpretation of the Fishers exact statistic instead. As expected, the baseli ne comparisons did not indicate significant differences in the culturally correct answer keys. In addition to analyzing the answer keys for differences, average levels of cultural competency were also assessed (Table 28).

PAGE 141

123 Table 28. Baseline Sample: Cultural Competency Analysis Results by Group N Average Competency BMC27 .51 BMI 60 .49 B-All241.50 *One individual removed from analyses in the outcome control group due to missing data Table 29. Baseline Sample: Comparisons of Cultural Competency by Intervention and Control Groups df t p< BMC vs. B-All239 -.13 .8994 BMI vs. B-All 239 .64 .5200 BMC vs. BMI 85 .44 .6591 *In the BMC vs. B-All comparisons the B-A ll sample was reduced to an n=214 (removed the control individuals) so an indepe ndent samples t-test could be run ** In the BMI vs. B-All comparisons the B-All sample was reduced to an n=181 (removed the intervention individuals) so an independent samples t-test could be run Independent samples t-tests indicate no significant differences in competency between the baseline samples (Table 29).

PAGE 142

124 Baseline Cultural Consensus Analysis : Intervention and Control Groups by Demographic Characteristics Once the baseline intervention and control groups were reviewed for the presence of cultural consensus and si gnificant group differences were explored, the intersection between demographic dimensions and cultu ral models were examined. Consensus analysis was run on the same samples as the comparisons above: baseline matched controls (BMC), baseline matched intervention (BMI), and the entire baseline sample (BAll). In addition to the consensus analyses groups were compared for significant differences between levels of cultural compet ency as well as between elements of the culturally correct answer key on the follo wing demographics: mar ital status, religion, literacy, wealth, age, stage of diseas e, and number of acute attacks. With regard to the demographic analyses, Protestants and Catholics were compared as there were too few individuals who selected vodou, othe r, or no religion to include in the analysis. Along the demographi c dimension of wealth, two categories were identified for comparison. Each individual was asked if they possessed any of the following four items: radio, storage chest, bicycle, living room. These indicators are common markers of material style of life in th e Haitian context, a construct that has been used to indirectly measure relative wealth. If an individual owned two of these items or less, they were classified as less wealthy; in contrast, people owning either three or four of the items were identified as more wealt hy. For age, the median age, 46 years old, was used to divide the sample into younger a nd older groupings. Stage of disease was also

PAGE 143

125 assessed for differences in cultural model. Individuals were groupe d into two categories for comparison: less severe and more se vere. Those people experiencing stage 0-2 disease were grouped with less severe as th ese stages are characterized by either no symptoms or swelling only. In contrast, indi viduals experiencing st age 3-7 disease were assigned a value of more severe as stages 3-7 are associated with skin folds, knobs, lesions, and the inability to care for oneself. In the data, each person had each leg staged for disease severity; the most severe of these values was utilized in this analysis. Finally, the number of acute attacks experienced by each individual in the past year was also explored. People were grouped by fewer and more attacks in the past year due to the small cell sizes present in the alternate comp arison of no acute attacks versus any acute attacks. Thus, individuals having 0-1 acute a ttacks in the past year were identified as having fewer attacks than their counterparts w ith more attacks, 2-3 acute attacks within the past year. The tables below outline the general demographic findings; more detailed demographic data from the baseline cultu ral consensus analyses are available in Appendix A.

PAGE 144

126 Table 30. Key Findings in Baseline Data for Dem ographic Characteristics Group Demographic Characteristic Demographic Category N Presence of Consensus BMC Marital Status Single 4 No Not Single 23 No BMI Single 27 No Not Single 33 No B-All Not Single 159 Yes Single 82 No BMC Religion Protestant 6 No Catholic 15 Yes BMI Protestant 28 No Catholic 27 No B-All Protestant 96 No Catholic 118 Yes BMC Literacy Literate 15 No Not Literate 12 Yes BMI Literate 38 No Not Literate 22 Yes B-All Literate 137 No Not Literate 104 Yes BMC Wealth Less Wealth 24 No More Wealth 3 Yes BMI Less Wealth 35 Yes More Wealth 25 No B-All Less Wealth 172 Yes More Wealth 69 No BMC Age Younger 14 No Older 13 No BMI Younger 28 No Older 32 No B-All Younger 121 No Older 120 No BMC Stage of Disease Less Severe 9 No More Severe 18 Yes BMI Less Severe 36 No More Severe 24 No B-All Less Severe 125 No More Severe 116 No BMC Number of Attacks Fewer Attacks 19 Yes More Attacks 8 No BMI Fewer Attacks 31 No More Attacks 28 No B-All Fewer Attacks 149 Yes More Attacks 89 No

PAGE 145

127 Table 31. Significant Differences in Cultura lly Correct Answer Keys in the Baseline Data for Demographic Characteristics Demographic Characteristic Demographic Comparison BMC BMI B-All Marital Status Single vs. Not Single No No No Religion Protestant vs. Catholic No No No Literacy Literate vs. Not Literate No No No Wealth Less Wealth vs. More Wealth No No No Age Younger vs. Older No No No Stage of Disease Less Severe vs. More Severe No No No Number of Attacks Fewer Attacks vs. More Attacks No No No Table 32. Significant Differences in Cultural Competency in the Baseline Data for Demographic Characteristics Demographic Characteristic Demographic Comparisons BMC BMI B-All Marital Status Single vs. Not Single No No No Religion Protestant vs. Catholic No No No Literacy Literate vs. Not Literate No No No Wealth Less Wealth vs. More Wealth No No No Age Younger vs. Older No No No Stage of Disease Less Severe vs. More Severe No No No Number of Attacks Fewer Attacks vs. More Attacks No Yes Yes

PAGE 146

128 Though consensus was not present in the complete baseline samples (BMC, BMI, B-All), consensus was reached in some s ub-samples when demographic designations were considered (Table 30). Re garding marital status, the fu ll baseline sample (B-All) reached consensus for the non-single portion of the sample. The Catholic segment of the baseline controls (BMC) popul ation showed strong consensu s, and this consensus was also reflected in the full baseline sample. Non-literate individuals exhibited consensus across all three baseline samples (BMC, BMI, B-All). Participants who are wealthier displayed consensus in the baseline control group (BMC), while less wealthy individuals showed consensus in the intervention group (B MI) and the full baseline sample (B-All). With regards to severity, the baseline contro l group (BMC) presented with consensus for people experiencing higher stage disease. Also, the baseline control (BMC) and full baseline sample (B-All) produced cultural c onsensus in the sub-sample reporting fewer acute attacks in the past year. Though some sub-samples presented w ith shared consensus, none of the culturally correct answer keys were significan tly different from one another within each demographic characteristic (Table 31). Ho wever, both the baseline intervention (BMI) and full baseline samples (B-All) for number of attacks did show significant differences in competency along the number of attack s demographic characteristic (Table 32).

PAGE 147

129 Cultural Models at Outcome Outcome Cultural Consensus Analysis Com parison by Matched Group: Intervention and Control Once cultural consensus analyses were conducted on the baseline samples, these analyses were repeated by intervention and control group for the following samples: outcome matched controls (OMC-Reduced Model [RM]) and the outcome matched intervention (OMI-RM) sample. Also, outco me matched controls (OMC-Full Model [FM]) and intervention (OMI-FM) participants were assessed for cultural models with the inclusion of the additional CM questions only present in the outcome survey. Groups were also compared for significant differences between levels of cu ltural competency as well as between elements of the cu lturally correct answer key. Table 33. Outcome Sample: Consensus Analysis Results by Group N First EigenvalueSecond Ei genvalueEigenvalue Ratio OMC-RM* 26 9.135 2.347 3.892 OMI-RM 60 26.897 5.714 4.71 OMC-FM* 26 6.959 2.646 2.630 OMI-FM 60 20.469 4.253 4.812 *One individual removed from analyses in the outcome control group due to missing data

PAGE 148

130 As the eigenvalue ratio between the firs t and second eigenvalues was equal to or greater than three, the thre shold for consensus was met for the following three groups: outcome controls (OMC-RM), outcome inte rvention (OMI-RM), outcome intervention with additional items (OMI-FM) (Table 33). Of the three groups that met consensus, it was greatest in the interv ention groups at outcome. Once the degree of consensus in a sample was established, the answer keys were compared for significant differences (Table 34, 35). The outcome samples suggest that hygiene and sandals remain elements of th e cultural model acro ss both control (OMCRM) and intervention (OMI-RM) groups. However, the cultural model for the intervention (OMI-RM) group is also comprised of the following key treatment elements: crme, elevation, massage, exercise, and herb al remedies. This group also believes that insect bites cause lymphatic filariasis.

PAGE 149

131 Table 34. Culturally Correct Answer Keys by Intervention (OM I-RM) and Control (OMC-RM) Group Outcome Sample OMC-RM (N=26) OMI-RM (N=60) Answer (Weighted %) Cause Insect Bite 0(96%) 1(72%) Magic 0(97%) 0(96%) Sprain 0(100%) 0(96%) Worms 0(85%) 0(89%) Chill 0(100%) 0(100%) Vitamin Deficiency0(100%) 0(100%) Treatment Hygiene 1(98%) 1(97%) Sandal 1(97%) 1(98%) Permanganate 0(94%) 0(77%) Crme 0(97%) 1(50%) Elevation 0(83%) 1(94%) Massage 0(97%) 1(66%) Exercise 0(97%) 1(89%) Bandage 0(93%) 0(81%) Medicine 0(56%) 0(61%) Herbal Remedy 0(58%) 1(42%) Pomade 0(69%) 0(59%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 150

132 In the full outcome model, including items present only in the outcome data, the control (OMC-FM) group sugge sts hygiene and sandals are the only agreed upon courses of treatment. Alternately, th e intervention (OMI-FM) group endorsed the idea of insect bites as the cause of LF and the use of hygiene, sandals, el evation, massage, exercise, and herbal remedies.

PAGE 151

133 Table 35. Culturally Correct Answer Keys by Intervention (OM I-FM) and Control (OMC-FM) Group that Include Additional Cu ltural Model Items only Present in the Outcome Sample Outcome Sample OMC-FM (N=26) OMI-FM (N=60) Answer (Weighted %) Cause Insect Bite 0(94%) 1(74%) Magic 0(94%) 0(98%) Sprain 0(100%) 0(95%) Worms 0(95%) 0(86%) Chill 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) Treatment Hygiene 1(100%) 1(99%) Sandal 1(99%) 1(99%) Permanganate 0(96%) 0(79%) Crme 0(99%) 0(53%) Elevation 0(93%) 1(95%) Massage 0(99%) 1(66%) Exercise 0(99%) 1(93%) Bandage 0(94%) 0(82%) Medicine 0(68%) 0(64%) Herbal Remedy 0(82%) 1(57%) Pomade 0(85%) 0(59%) Treatment (OC Only) Hygiene 1(92%) 1(98%) What kinds of care can help your gwopye? Sandal 1(86%) 1(80%) Permanganate 0(97%) 0(67%) Crme 0(94%) 0(52%) Elevation 0(87%) 1(93%) Massage 0(97%) 1(70%) Exercise 0(97%) 1(92%) Bandage 0(97%) 0(87%) Medicine 0(75%) 0(67%) Nothing 0(98%) 0(100%) Treatment (OC Only) Hygiene 1(73%) 1(89%)

PAGE 152

134 Table 35. (continued) Outcome Sample OMC-FM (N=26) OMI-FM (N=60) Answer (Weighted %) What can you do to prevent acute attacks? Sandal 1(73%) 1(78%) Permanganate 0(97%) 0(88%) Crme 0(97%) 0(79%) Elevation 0(90%) 1(72%) Massage 0(100%) 1(56%) Exercise 0(100%) 1(74%) Bandage 0(100%) 0(94%) Medicine 0(87%) 0(69%) Nothing 0(95%) 0(100%) Treatment (OC Only) Hygiene 1(78%) 1(79%) What can be done to provide relief during an acute attack? Sandal 1(75%) 1(60%) Permanganate 0(100%) 0(78%) Crme 0(99%) 0(76%) Elevation 0(93%) 1(67%) Massage 0(99%) 1(57%) Exercise 0(99%) 1(65%) Bandage 0(94%) 0(89%) Medicine 0(68%) 0(63%) Nothing 0(97%) 0(99%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 153

135 Table 36. Outcome Sample: Comparisons of Culturally Correct Answer Keys by Intervention and Control Groups Chi-SquaredfP< Fishers Exact Two-Tailed p< OMC-RM vs. OMI-RM 5.1000* 1 .0239.0570 OMC-FM vs. OMI-FM 9.5958* 1 .0020.0036 OMI-RM vs. OMI-FM .0003* 1 .98591.00 OMC-RM vs. OMC-FM .2617 1 .60901.00* indicates which statistic was interpreted for significant differences between the answer keys Chi-square and Fishers exact analyses we re conducted to examine the differences between answer keys (Table 36). As expecte d, significant differences existed between the elements of the answer keys in these co mparisons: outcome controls (OMC-RM) vs. outcome intervention (OMI-RM), outcome controls with additional items (OMC-FM) and outcome intervention with additional items (OMI-FM). Also, as hypothesized, the remaining comparisons did not in dicate significa nt differences. In addition to analyzing the answer keys for differences, average levels of cultural competency were also assessed (Table 37).

PAGE 154

136 Table 37. Outcome Sample: Cultural Co mpetency Analysis Results by Group N Average Competency OMC-RM26.56 OMI-RM 60.62 OMC-FM 26.44 OMI-FM 60.51 *One individual removed from analyses in the outcome control group due to missing data Table 38. Outcome Sample: Comparisons of Cultural Competency by Intervention and Control Groups df t P< OMC-RM vs. OMI-RM 84-1.08.2833 OMC-FM vs. OMI-FM 84-1.17.2458 OMI-RM vs. OMI-FM 2.06 .0439 OMC-RM vs. OMC-FM 2.41 .0237 Independent samples t-tests point to signi ficant differences in competency when the outcome intervention a nd control (OMI-RM & OMC-RM) groups were compared to the outcome samples including additional CM items (OMI-FM & OMC-FM); this occurrence suggests that there is a lesser degree of cultura l competency when patients are asked to identify treatment options for specific scenarios, such as how to treat an acute attack, rather than general treatme nt choices for LF (Table 38).

PAGE 155

137 Outcome Cultural Consensus Analysis : Intervention and Control Groups by Demographic Characteristics Once the outcome intervention and control groups were reviewed for the presence of cultural consensus and si gnificant group differences were explored, the intersection between demographic dimensions and cultu ral models were examined. Consensus analysis was run on the same samples as the comparisons above: outcome matched controls (OMC-RM), outcome matched inte rvention sample (OMI-RM), as well as outcome matched controls (OMC -FM) and intervention (OMI-FM) participants with the inclusion of the additional CM questions only present in the outcome survey (Table 39). In addition to the consensus analyses, groups were compared for significant differences between levels of cultural competency as well as between elements of the culturally correct answer key on the following demogr aphics: marital status religion, literacy, wealth, age, stage of disease, and number of acute attacks (Table 40, 41). The tables below outline the general demographic findings ; more detailed dem ographic data from the outcome cultural consensus analyses are available in Appendix B.

PAGE 156

138 Table 39. Key Findings in Outcome Data for Demographic Characteristics Group Demographic Characteristic Demographic Category N Consensus OMC-RM* Marital Status Single 4 Yes Not Single 22 Yes OMI-RM Single 27 Yes Not Single 33 Yes OMC-FM* Single 4 Yes Not Single 22 No OMI-FM Single 27 Yes Not Single 33 Yes OMC-RM* Religion Protestant 6 Yes Catholic 15 Yes OMI-RM Protestant 28 Yes Catholic 27 Yes OMC-FM* Protestant 6 Yes Catholic 15 No OMI-FM Protestant 28 Yes Catholic 27 Yes OMC-RM* Literacy Literate 15 Yes Not Literate 11 Yes OMI-RM Literate 38 Yes Not Literate 22 Yes OMC-FM* Literate 15 No Not Literate 11 Yes OMI-FM Literate 38 Yes Not Literate 22 Yes OMC-RM* Wealth Less Wealth 23 Yes More Wealth 3 No OMI-RM Less Wealth 35 Yes More Wealth 25 Yes OMC-FM* Less Wealth 23 No More Wealth 3 No OMI-FM Less Wealth 35 Yes More Wealth 25 Yes

PAGE 157

139 Table 39. (continued) Group Demographic Characteristic Demographic Category N Consensus OMC-RM* Age Younger 13 No Older 13 Yes OMI-RM Younger 28 Yes Older 32 Yes OMC-FM* Younger 13 No Older 13 Yes OMI-FM Younger 28 Yes Older 32 Yes OMC-RM* Stage of Disease Less Severe 9 Yes More Severe 17 Yes OMI-RM Less Severe 36 Yes More Severe 24 Yes OMC-FM* Less Severe 9 Yes More Severe 17 Yes OMI-FM Less Severe 36 Yes More Severe 24 Yes OMC-RM* Number of Attacks Fewer Attacks 18 Yes More Attacks 8 Yes OMI-RM Fewer Attacks 31 Yes More Attacks 28 Yes OMC-FM* Fewer Attacks 18 Yes More Attacks 8 Yes OMI-FM Fewer Attacks 31 Yes More Attacks 28 Yes *One individual removed from analyses in the outcome control group due to missing data

PAGE 158

140 Table 40. Significant Differences in Culturally Correct Answer Keys in the Outcome Data for Demographic Characteristics Demographic Characteristic Demographic Comparison OMC RM OMI RM OMC FM OMI FM Marital Status Single vs. Not Single No No No No Religion Protestant vs. Ca tholic No No No No Literacy Literate vs. Not Literate No No No No Wealth Less Wealth vs. More Wealth No No No No Age Younger vs. Older No No No No Stage of Disease Less Severe vs. More Severe No No No No Number of Attacks Fewer Attacks vs. More Attacks No No No No Table 41. Significant Differences in Cultura l Competency in the Outcome Data for Demographic Characteristics Demographic Characteristic Demographic Comparison OMC RM OMI RM OMC FM OMI FM Marital Status Single vs. Not Single No No No No Religion Protestant vs. Ca tholic No No No No Literacy Literate vs. Not Literate No No No No Wealth Less Wealth vs. More Wealth Yes No Yes No Age Younger vs. Older No No No No Stage of Disease Less Severe vs. More Severe No No No No Number of Attacks Fewer Attacks vs. More Attacks No No No No

PAGE 159

141 Marital Status An eigenvalue ratio greater than three was present for the outcome control group with additional items (OMC-FM) for the si ngle sample, as well as the control group (OMC-RM), intervention group (OMI-RM) a nd intervention group with additional items (OMI-FM) at outcome regardless of marital status. When a group exhibited consensus in either or both conditions, it was strong er for the single portion of the sample. The highest cultural competency presente d in the intervention group at outcome (OMI-RM) across both marital conditions, and this finding is in line with overall study hypotheses. Additionally, no signifi cant differences in levels of competency exist within each group by marital status. This information, considered with the levels of cultural consensus and lack of significant differen ces in answer keys suggests that single individuals experience more cohesiveness in their cultural model of LF, but that this difference is not great enough to impact th e overall shared cultural model when both single and not single particip ants are grouped together. Religion In the case of religion, Protestants and Catholics were compared; there were too few individuals who selected vodou, other, or no religion to include in the analysis. Consensus was met for several groups: Protesta nts and Catholics in the outcome controls sample (OMC-RM), Protestants and Catholics in the outcome intervention sample (OMI-

PAGE 160

142 RM), Protestants in the outcome controls with additional items group (OMC-FM), and both Protestants and Catholics in the in out come intervention group with additional items (OMI-FM). Consensus was highest in the tw o intervention groups for the Protestant sample. No significant differences between answer keys by religion existed within each of the groups. Regarding cultural competency, no significant differences existed between conditions, Protestant and Ca tholic, within each of the groups analyzed. Literacy In the case of literacy, strongest consensus is present among the outcome controls both reduced and full samples (OMC-RM & OM C-FM). Consensus if also found in all members of the interventi on group in both full and reduced models (OMI-RM & OMIFM); however, consensus is greater among indi viduals who are not l iterate. The answer keys showed no significant differences within each group. The same trend held true when differences in cultural competency were examined within each group. Wealth Within the wealth comparisons, outc ome intervention groups, both full and reduced (OMI-RM & OMI-FM), exhibited the highest level of consensus across both wealth categories. Specifically, individuals with more wealth in the intervention group

PAGE 161

143 showed stronger consensus than their less wealthy counterpart s. For the outcome control groups, only the less wealthy individuals fo r the reduced model (OMC-RM) showed consensus. However, in the outcome in tervention group both stronger consensus and competence was present for wealthier individuals. Age When looking at cultural models presen t in the sample by age, younger outcome intervention members exhibit consensus where their outcome control counterparts do not. Older members show consensus in all f our outcome models, but the younger group indicates the strongest consensus. The younge r portion either had no consensus or the strongest consensus of all outcome groups (OMI-RM). Stage of Disease All groups showed consensus at outcome al beit lesser in the cont rol groups. In the reduced outcome model, consensus is highest in the portion of the sample experiencing more severe disease. In the outcome inte rvention model assessing general treatment options, consensus existed for individuals more severely impacted by LF. However, when specific treatment scenarios were introduce d, the intervention model with additional items present in the outcome survey only (OMI-FM), those with less severe disease indicated greater consensus.

PAGE 162

144 Number of Acute Attacks Consensus is also present in all outcome groups for num ber of acute attacks, and it is greater in the outcome interventi on groups (OMI-RM & OMI-FM). Within the intervention groups, people with fewer attacks exhibited slightly more consensus than the portion experiencing more acute attacks. The highest cultural consensus for individuals having fewer attacks was found in the outcome intervention sample (OMI-RM), and the highest overall consensus presented in the outcome control sample (OMC-RM) for those with more attacks. Changes in Cultural Competency: Baseline to Outcome In addition to assessing th e cultural models through cons ensus analysis at each data point, baseline and outco me, longitudinal changes were also examined. One change investigated included the degree of change in cultural competency le vels between groups before and after support group participation (Table 42).

PAGE 163

145 Table 42. Baseline and Outcome: Cultura l Competency Analysis Results by Group N Average Competency BMC 27.51 BMI 60.49 OMC-RM26.56 OMI-RM 60.62 *One individual removed from analyses in the outcome control group due to missing data Table 43. Baseline to Outcome: Comparisons of Cultural Competency by Intervention and Control Groups Dft p< BMC vs. OMC-RM .62 .5407 BMI vs. OMI-RM 3.63.0006 Paired samples t-tests measured the de gree of change over time in cultural competency for both the control and inte rvention samples (Table 43). The findings support the key hypotheses that no significant change in cultural competency would occur from baseline to outcome in the contro l group and, in contrast that the intervention group would show significant improvement of cultural competency at outcome.

PAGE 164

146 Changes in Cultural Models: Baseline to Outcome The second component of the cultura l consensus analyses examined longitudinally was a comparison of culturally correct answer ke ys between the two baseline and outcome samples. Table 44. Baseline to Outcome: Comparisons of Culturally Correct Answer Keys by Intervention and Control Groups Chi-Squaredfp< Fishers Exact Two-Tailed p< BMC vs. OMC-RM .0000 1 1.00001.00* BMI vs. OMI-RM 5.1000* 1 .0239 .0570 indicates which statistic was interpreted for significant differences between the answer keys Chi-square and Fishers exact analyses were conducted, as appropriate, to examine the differences between answer ke ys (Table 44). As expected, significant differences existed between the baseline and outcome intervention samples (BMI & OMI-RM). Also, as hypothesized, the rema ining comparison, baseline and outcome control samples (BMC & OMC-RM) did not indicate signif icant differences. Strength of Cultural Consen sus: Baseline to Outcome The final component of the cultural consensus analyses involved a review of the degree of shared consensus w ithin samples at baseline and outcome. These comparisons

PAGE 165

147 are descriptive only as no test of statisti cal significance currently exists to assess for significant differences in the eigenvalue ra tio from baseline to outcome points. Table 45. Baseline and Outcome: C onsensus Analysis Results by Group N First EigenvalueSecond Ei genvalueEigenvalue Ratio BMC 27 7.829 3.248 2.410 BMI 60 17.038 7.015 2.429 OMC-RM* 26 9.135 2.347 3.892 OMI-RM 60 26.897 5.714 4.71 *One individual removed from analyses in the outcome control group due to missing data As the eigenvalue ratio between the firs t and second eigenvalues was equal to or greater than three, the thre shold for consensus was met for the following three groups: outcome controls (OMC-RM) and the outcome intervention sample (OMI-RM) (Table 45). Overall, consensus increased from ba seline to outcome for both the control and intervention groups. However, the degree of shared consensus remained greatest for the intervention group at outcome (OMI-RM).

PAGE 166

148 Link Between Cultural Consensus and Cultu ral Consonance: Baseline to Outcome Cultural Consonance Analyses After the cultural consensus analyses were completed, cultural consonance (CC) was then assessed. For each of the comparis on groups, baseline control (BMC), baseline intervention (BMI), baseline all (B-All), outcome controls (O MC-RM), and outcome intervention (OMI-RM), tested in the first portion of the CM evaluation, a CC analysis was also conducted. The outcome control and intervention models including the additional items (OMC-FM & OMI-FM) cannot be assessed for cultural consonance as the questions elicited information regardi ng what a person could do in each treatment scenario and did not inquire as to which tr eatment behaviors were actually enacted in these more specific treatment circumstances In each model tested, treatment items identified in each culturally correct answer keys are the items tested for consonance. Thus, if four treatment options are identified as part of the cultural model in a sample, each respondent could exhibit a consonance of 0, 25%, 50%, 75%, or 100%. The average consonance of each sample is reported in this section where average consonance indicated average degree of beha vioral enactment of elements incorporated in the cultural model within the sample population (Table 46).

PAGE 167

149 Table 46. Baseline and Outcome: Cultura l Consonance Analysis Results by Group N Average Consonance BMC 27 .83 BMI 60 .70 B-All 241.63 OMC-RM*26 .88 OMI-RM 60 .56 One individual removed from analyses in the outcome control group due to missing data In these comparisons, outcome controls (OMC-RM) showed the highest rates of consonance followed by the baseline control gr oup (BMC). This finding is alternate to the expected relationship between consonan ce and consensus. Independent samples ttests were also executed in order to assess for significant differences in levels of cultural consonance, and the results are presented in the table below: Table 47. Comparisons of Cultural Consona nce by Intervention and Control Groups df t p< BMI vs. BMC 85 1.39 .1687 B-All vs. BMC 266 -3.35<.0009 B-All vs. BMI 73.3-1.24.2184 OMC-RM vs. OMI-RM35.55.07 <.0001 BMC vs. OMC-RM 51 -.56 .5776 BMI vs. OMI-RM 83 2.20 .0309

PAGE 168

150 Though the findings related to average ra te of consonance were unexpected, at outcome the consonance rates between the control (OMC-RM) and intervention (OMIRM) groups were significantly di fferent as were the rates between the baseline (BMI) and outcome (OMI-RM) intervention samples (Table 47). Cultural Consonance Analyses by Demographics Once cultural consonance analyses were reviewed for the key comparison samples, consonance was then tested al ong demographic dimensions for significant differences (Table 48). More detailed resu lts from these analyses are available in Appendix C. Table 48. Significant Differences in Cultural Consonance for Demographic Characteristics Demographic Characteristic Demographic Comparisons BMC BMIB-All OMCRM OMIRM Marital Status Single vs. Not Single No No Yes No Yes Religion Protestant vs. Cat holic No Yes No No No Literacy Literate vs. Not Literate No No No Yes Yes Wealth Less Wealth vs. More Wealth No Yes No No Yes Age Younger vs. Older No No No No No Stage of Disease Less Severe vs. More Severe No No No No No Number of Attacks Fewer Attacks vs. More Attacks Yes Yes Yes No Yes

PAGE 169

151 Marital Status Independent samples t-tests suggest that significant differences exist between individuals by marital status in the full baseline sample (B-All) as well as the outcome intervention (OMI-RM) sample. Taken as a w hole, it seems that people who participate in the support group program and are partnere d are more likely to engage in treatment behaviors than their si ngle counterparts. Religion The only group to display significant diffe rences in consonance is the baseline intervention (BMI) sample. At baseline, Protes tants exhibited signifi cantly higher rates of consonance, but, after participation in the support group, Catholics increased in consonance greatly; this shift eliminated any significant diffe rences in the two groups at outcome. The data allow that Catholics effec tively incorporated self -care practices into their treatment regimen at a rate much im proved than was evidenced at baseline. Literacy When independent samples t-tests were run, the only signi ficant difference highlighted was found in the outcome interven tion group. In this case, it looks as though, at outcome, non-literate indivi duals were not as greatly impacted by the support group program as literate members of the sample population.

PAGE 170

152 Wealth Through t-test comparisons, significant di fferences in consonance were revealed in both baseline and outcome intervention groups. This observation suggests that less wealthy people were less amenable to enacti ng the self-care practices, other than the ones present at baseline, they endor sed in the cultural model. In contrast, wealthier people were able to incorporate newl y introduced self-care practices into both their belief and behavioral systems. Age After looking at the consonance rates by age, t-tests comparisons support the finding that there are no significant differences between age categories within any of the samples tested. The data indicate that olde r individuals in the intervention sample increased in their willingness to enact trea tment practices indicated in the CM after support group participation. Younger individual s in this sample, however, endorsed the beliefs present in the CM, but were not as quick to add additional behaviors to their treatment regimen as evidenced by their reduced rate of consonance at outcome.

PAGE 171

153 Stage of Disease Independent samples t-tests show no significant differences between the consonance rates of people with more or less severe disease within each of the samples. After a review of these data, it appears th at people with less severe disease were less likely to include the additi onal self-care practi ces they endorsed, indicated in the culturally correct answer ke ys, after support group partic ipation than support group participants with more severe disease. Number of Acute Attacks When examined for significant differen ces in consonance by number of acute attacks within the past year baseline controls (BMC), ba seline intervention (BMI), and outcome intervention (OMI-RM) groups met the threshold for significance. Individuals with less acute attacks were more likely to enact self-care behaviors in the treatment regimen after support group participation.

PAGE 172

154 Chapter 5: Discussion

PAGE 173

155 This chapter aims to discuss the study results within the parameters of the key research questions outlined in earlier chap ters. Additionally, a brief assessment of the cultural evaluation methodology and the s upport group program is also presented. Finally, implications of the research findings and future directions in research are presented. Cultural Models at Baseline Baseline Comparisons The first research question posed in th is study aimed at understanding key beliefs about lymphatic filariasis and the degree to wh ich these ideas were sh ared in the baseline sample populations. It was postulated that a sh ared cultural model, meeting the threshold for consensus, would exist in these samp les at baseline. This hypothesis was not supported. At baseline, all three samples, contro ls, intervention, and full baseline sample, failed to meet the threshold for consensus, an eigenvalue ratio between the first and second factors greater than three. However, each of these sample s (BMC, BMI, B-All) did identify the same key elements in their culturally correct an swer keys: hygiene and wearing sandals. The full sample population (B -All) also selected herbal remedies, but this choice was not decisively classified. Competency leve ls for all three groups were also present at a level barely above the th reshold of guessing. Thus, regarding the first

PAGE 174

156 research question, it seems that there wa s no clear cohesiveness in beliefs about lymphatic filariasis. First, it is of interest to note that no causal categories were clearly endorsed as part of the culturally correct answer key at ba seline. This finding c ould be rooted in the broader Haitian cultural contex t. As Farmer (1990) noted in his work with HIV/AIDS in Haiti, the cause of an illness carries great social meaning. Attribution of a lymphatic filariasis diagnosis to biomedical causes connot es a treatable illness that doesnt carry a negative moral implication; how ever, the disease is not considered curable within this paradigm. Alternately, traditional causal ca tegories such as sorcery, if endorsed, may reflect poorly on the individual socially and morally; this type of illness, though, can be cured (Brodwin, 1996; Farmer, 1990). The lack of clearly identified cau se for LF within the baseline sample answer ke ys coupled with the high numbe r of participants identifying other or dont know as the cause of their illn ess at baseline (Kanda 2004) could reflect a resistance to categorize ones own illness as either incurable (biomedical) or morally compromising (traditional). An alternate possibility regarding the causal patterns found in the culturally correct answer keys may be explained by multiple studies on lymphatic filariasis globally. In several studies, biomedical causes of LF (e.g.: insect bi tes) presented in the data at very minimal levels (Coreil et al ., 1998; Ramaiah et al., 1996; Person et al., 2007; Ahorlu et al., 1999; Rauyajin et al., 1995; Bandyopadhyay, 1996; Eberhard et al., 1996; Morfit, 1998; Babu et al., 2004; Coreil et al., 2003; Gyapong et al., 1996). These findings suggest it is plausible that study participants did not make a conscious choice between the

PAGE 175

157 morally impacted biomedical or traditional causal categories. In this scenario, it is possible that participants endorse traditiona l causes of LF, due to lack of awareness regarding biomedical alternatives, but sti ll fail to share consensus regarding which traditional causes most likely result in an LF diagnosis. Regarding the overall model identified in the answer keys, three possible explanations are presented. First, these resu lts may indicate tensi on between traditional and biomedical concepts of lymphatic filarias is. However, this expl anation is not likely. It is more likely that there were no shared traditional beliefs about LF in the sample populations. The rationale for this assertion lies in the elements identified in the culturally correct answer key. One would expect both tr aditional and biomedical elements to be highlighted if the two models were in clea r opposition or tension, but this finding was not born out in the data. Instead, two key biomedi cal ideas about lymphatic filariasis were selected. It seems that the groups had some exposure to the biomedical perspective prior to the intervention. This tende ncy towards biomedical treatme nts at baseline may indicate that participants in the support groups were primed to receive additional biomedical information about lymphatic filariasis. Also, since there are no heavily competing shared traditional ideas about lymphatic filariasis, resistance to support group content may have been minimized. Alternately, the results may represent a weak bicultural model comprised of both traditional and biomedical concepts of ly mphatic filariasis. The support for herbal remedies, a traditional treatment option, is weaker than that of the biomedical concepts in the baseline samples, but it remains present. Th is integrated approach would allow for the

PAGE 176

158 easy integration of biomedical concepts into the overall cult ural model of LF while not excluding the possibility of herb al remedies to exist within the model. This explanation result in the same environment of minimi zed resistance to the support group content. A third possibility addresses the strong pr esence of hygiene and sandals within the baseline cultural models. It is reasonable that study participants had no clear thoughts regarding cause and treatment of LF, as supported by the lack of overall consensus. In lieu of clear beliefs about LF, the genera l biomedical ideas of hygiene and wearing sandals may have been chosen as they are biomedical treatment options for a number of diseases. Lastly, these results could reflect the stig matized nature of LF in Haiti. Within resource poor settings, disabi lities are often hidden and not discussed openly (McCallion et al., 1997; Halcon, Blum, Beuhring, Pate Campbell-Forrester, & Venema, 2003; Mayhew, 2003; Jacobson, 2003). As a result, little to no dialogue between social entities, as described within the social interactioni sm framework, may occur. This possibility allows for the presence of varied, non-cohesive or transitional ideas regarding LF; such a landscape would certainly yiel d a lack of consensus rega rding cause and treatment categories of lymphatic filariasis. Baseline Comparisons by Demographic Dimensions When the groups were examined by demographic dimensions, several subsamples showed consensus. These sub-samples include: partnered, Catholic, and non-

PAGE 177

159 literate individuals. People experiencing more severe disease and fewer acute attacks in the past year also showed consensus. C onsensus was also present along the wealth dimension, but the results varied within the baseline samples. Finally, these sub-samples generally endorsed the use of hygiene and w earing sandals as treatment options while herbal remedies receive d moderate support. At baseline, it is possible that partnered individuals were more likely to show consensus as, even in an environment where disability is not openl y discussed, partners can negotiate beliefs about lym phatic filariasis interpersonall y. Also, LF is often found among more economically disadvantaged segments of the population; it is reasonable that poorer, non-literate people have more exposure to shared ideas of LF leading to higher rates of consensus at baseline. For those that experience more severe disease, they may be more likely to believe in appl ying the same general treatment options to LF as they do to multiple other illnesses; the need to alleviat e symptoms in a landscape of severe disease may encourage shared belief in thes e general treatment categories. The presence of consensus in these dem ographically separated groups at baseline suggests that participants experiencing thes e designations may experience more impact from the biomedical content of the support group program. This suggestion is supported as these individuals endorsed both hygiene and sandals with a high level of shared belief at baseline. However, the grea ter impact these participants may experience probably does not significantly affect the overall findings. This assertion is supported as each sample was tested for significant differences betw een the two demographic conditions (e.g.: younger, older) for each characteristic assessed : marital status, religion, literacy, wealth,

PAGE 178

160 age, stage of disease, number of attacks. These comparisons resulted in no significant differences between the two conditions for a ll the demographic characteristics. As no significant differences were evident, it is probable that no one demographic segment of the population over-influenced the content of the culturally correct answer key or the degree of consensus in the baseline populations. Cultural Models at Outcome Outcome Comparisons: Reduced Models (OMC-RM & OMI-RM) Following the research question assessi ng the presence of cultural models at baseline, the same question was posed rega rding the outcome data. Again, a shared cultural model was expected in both th e intervention and control groups. It was hypothesized that the control sample would s how consensus, but the consensus was not expected to increase from baseline. Also, the it ems in the answer key were expected to be more traditional in nature. In contrast, it was hypothesized that the intervention group would show an increased consensus from base line, that additional biomedical cause and treatment ideas would be included in the CM, and that some traditional elements may also be present in the CM. Supporting the alternative hypothesis, bot h the outcome controls and outcome intervention samples displayed consensus at ou tcome. Also, as expected, the intervention group shared a higher degree of consensus th an the control group. Hygiene and sandals

PAGE 179

161 were indicated in both groups as key elements of the CM, but the intervention group also incorporated insect bite as a cause of LF and the following as treatment alternatives: crme, elevation, massage, exercise, herbal remedies. Of these treatment options included, insect bite, crme, massage, and herbal remedies were not decisively classified in the model. The items resulting in the intervention CM provide strong support for the hypothesis that biomedical categories would be strongly represente d in the culturally correct answer key and some traditional item s may retain their presence in the model (i.e.: herbal remedies). The presence of higher shared levels of consensus in the intervention group was both expected and reasonable when consider ed within an interactionist framework. Members of the support group were both exposed to the same information about LF and had an opportunity to openly discuss the content with one an other. Through the intrapersonal, interpersonal, and inter-group exchanges, th e support group format allowed for ideas to be negotiated between members on multiple ecological levels. Ultimately, the cultural model embraced by support group participants reflects the program content to varying degrees. The variation with which several of the items in the outcome intervention cultural model presented (by way of a posteriori proba bilities) can be discussed in a broader context. First, it is not surprising that insect bite as cause of disease did not reach the .80 threshold for definitive classi fication. At baseline, there wa s as strong shared belief, .95 a posteriori probability, that ins ect bites do not cause lymphatic filariasis. However, after support group participation, this cause of LF presented with a probability of .72. Though

PAGE 180

162 this item did not reach the threshold, it was both indicated as part of the cultural model and increased greatly as a causal belief fr om baseline to outcome. This shift could indicate that shared belief in insect bites as the cause of LF are tr ansitional or still in development after participati on in the support group. As insect bite globally has not been frequently cited as a cause of LF in previous studies (Coreil et al., 1998; Ramaiah et al., 1996; Person et al., 2007; Ahorlu et al., 1999; Rauyajin et al., 1995; Bandyopadhyay, 1996; Eberhard et al., 1996; Morfit, 1998; Babu et al., 2004; Coreil et al., 2003; Gyapong et al., 1996), it is possible th at support group members, while integrating the idea that insect bites cause the conditi on, are still negotiating what ro le insect bites play when simultaneously considered with ot her traditional cause categories. Another possibility relate s to potential confusion between the cause categories worm and insect bite. It is possible that the worms category should be grouped with the insect bite category as the mosquitoes inject a tiny worm through their bites; some people may identify this cause as worms and others insect bite. If these two categories were combined, the cause category insect bite/wor m may have met the .80 threshold indicating decisive classification. An alternate explanation for the insect bi te trend in the outcome data refers back to infectious disease and traditional/supe rnatural causal frameworks. The increased reception of insect bite as a cause of LF c ould reflect an increased awareness that, while not curable, LF is very treatable and that a ny future disease progress can be arrested by implementing self-care practices. As suc h, endorsing an infectious disease causal

PAGE 181

163 category removes negative moral implications of the diagnosis and still provides that treatment steps can be taken to improve ones condition. As insect bites are identified as a cau se of LF in the intervention group CM, treatment alternatives endorsed in the sample are expected to be li nked to that causal belief. Foster (1976) argued that illness cau sality provides the base for medical decisionmaking regarding treatment options, and multiple studies have born out this link between illness cause and treatment (Heurtin-Robe rts & Reisin, 1992; Coreil & Genece, 1988; Mathews et al., 1994; Coreil, 1983; Rauyajin et al., 1995; Ahor lu et al., 1999; Ramaiah et al., 1996). Since insect bites are a biomedic al cause category, it is expected that biomedical treatment alternatives will have a strong presence in the CM. This trend is present in the intervention sample as biom edical elements of the support group program were clearly embraced: the use of hygiene, wear ing sandals to prevent injury to the foot, elevating the leg, and exercising the leg to improve circulation. Additionally, crme and massage are also part of the biomedical regi men and were part of the culturally correct answer key though to a lesser degree. Herbal remedies were also indicated in the intervention culturally correct answer key. This finding is contrary to the expected outcome. Herbal remedies were not present in the baseline intervention sample, so it is not likely that the degree of belief in herbal remedies remained static over time in the in tervention group. However, it is possible that participating in the support group program led to an overall increase in awareness of local treatment possibilities for LF Within the support group, info rmal discussions between members could have led to members gaining beli ef in herbal remedies as an option. Also,

PAGE 182

164 while herbal remedies were not introdu ced in the support group program, if these remedies do not contraindicate with the trea tment regimen proposed in the support group, herbal remedies may not have been heavily discouraged either. Since herbal remedies only presented with a .42 a posteriori probab ility in the outcome intervention group, it is clear that this idea about treatment, while present, is not pervasive. Where the outcome intervention group s howed consensus, so did the outcome control group. This result supports the hypothesis that both outcome control and intervention group would yield shared consensus; however, this hypothesis, for the control group, was predicated on the expectation of consensu s at baseline for the control group as well. An increase in consensus was not expected from baseline to outcome in the control group. Thus, this outcome was unexp ected, and it is uncertain why this shift emerged. Multiple plausible explanations for this fi nding exist. First, it is possible that some diffusion of information from the support groups occurred within the control sites in Archaie and Cabaret as these communities ha d both an interventi on and control group. If this contamination did occur, it was not to a great degree. If a large amount of contamination was present, one would expe ct items other than hygiene and sandal to receive shared support in the control samp le. Instead, the two ite ms present in the baseline sample (hygiene, sandal) strengthen ed in consensus without the introduction of new treatment concepts. An alternate possibility is related to th e fact that the majority of the control sample came from La Plaine, a control site only. It is not unreasona ble to think that the

PAGE 183

165 two treatment options present in the baseline CM would receive more support at outcome in a locale that is the most urban of the th ree towns assessed in this research. As it is a more urban location, it is possible that contro l members in this town may have had more exposure to western medical id eas during the course of th e support group program than their more rural counterparts. Also, though the di fferences at baseline were not significant between locales for cultural model, Kanda (2004) did find some regional differences in his previous examination of the three towns. This examination suggests that people living in La Plaine, at baseline, were more likely to utilize health servi ces and routine health care practices. This is likely explained by the greater accessibility to the capital (Kanda, 2004, 118). Outcome Comparisons: Full M odels (OMC-FM & OMI-FM) In the outcome survey, several additiona l cultural model questions were included that were not present in the baseline surv ey (OMI-FM & OMC-FM). As such, these items cannot be compared over time to the baseline data. However, these data still provide interesting information. The reduced outco me intervention model (OMI-RM) showed consensus as predicted, and the outcome in tervention model including the additional items (OMI-FM) also yielded cultural consensus. This finding is in line with the original hypothesis; the outcome intervention sample is expected to show shared cultural consensus.

PAGE 184

166 The larger model (FM) examined some of the same treatment alternatives in the reduced outcome model (RM), but different tr eatment scenarios were presented. First, a general question assessing what kinds of car e can help with lymphatic filariasis was posed, and this question is similar to the info rmation gathered about general treatment in the reduced model. Secondly, pa rticipants were asked what can be done to prevent acute attacks. Finally, the survey included the following question for support group members: what can be done to provide relief during an acute attack. After running cultural consensus analysis similar results were found in the full model as those in the reduced model for all questions. The only differences present involved the inclusion of massage and excl usion of herbal reme dies. Participants exhibited higher rates of shared consensus for hygiene, sandal, elevation, massage, and exercise for the general treatment questions th an for the questions addressing treatment of acute attacks. These findings suggest that support group participants have strong agreement on treatment alternatives when asked generall y how one should treat lymphatic filariasis. However, when acute attacks are addresse d specifically, study members transfer the general treatment options to the prevention and treatment of acut e attacks with less certainty. It is possible that the higher consensus revealed the degree to which specific content was discussed in the program. T hus, the support group program could aim to spend more time on acute attacks specifically ; additional time spent on discussing these more specific scenarios allow for more interaction, negotiation, and development of shared agreement on treatment choices. Alternat ely, this trend may reflect the quality of

PAGE 185

167 the program content such that the findings may po int to improving support group program content relevant to acute attacks, he lping participants to better understand how general treatment options can be best app lied towards preventing or treating an acute attack. Thirdly, the lack of high a posteriori probabilities present in the acute attack specific data may reflect the difficulty explai ning the link between the self-care practices and prevention of acute attacks. Outcome Comparisons by Demographic Dimensions Demographic dimensions were assessed in the outcome samples in an effort to control for these variables. Though tradition regr ession analyses are not applicable in this evaluation methodology, it is possibl e to examine the differences in cultural models for each demographic condition (e.g.: single vs. pa rtnered sub-samples). These analyses assess for significant differences within dem ographic characteristics (e.g.: marital status) along both the culturally correct answer keys and cultural compet ency. If significant differences are found, it may point to a greate r than random influen ce of a particular demographic variable and require additional consideration. In the outcome samples for the reduc ed model (OMC-RM & OMI-RM), multiple sub-samples showed consensus when tested al ong demographic charac teristics. All four sub-samples showed consensus for the follo wing demographic characteristics: marital status, religion, literacy, stage of disease, and number of attacks. Additionally, along the wealth dimension, all groups, except the wealthier outcome controls (OMC-RM),

PAGE 186

168 demonstrated consensus. For age, all s ub-samples except younger outcome controls (OMC-RM) also met the cr iteria for consensus. Within the control sub-samples, hygiene and sandal were universally endorsed for groups who had shared consensus. Participants who were literate and experienced more acute attacks in the past year also endorse d pharmaceutical medicines. Literate controls also selected herbal remedies. The str ong support for the use of hygiene and wearing sandals are in line with the results of th e reduced and full model control samples at outcome. One possible explanation for the a ddition of pharmaceutical medicines to the CM for literate individuals may be that thes e participants had more opportunity to read western medical materials. Through the introduction of additional information, literate members would then have to in corporate this new information into their beliefs about LF. Also, as most of the controls resided in La Plaine, closer to an urban center, people experiencing more attacks may have had more exposure and access to pharmaceutical medicines. Regarding intervention participants, all sub-samples showed consensus at outcome and all of these samples supported th e following as part of the cultural model: insect bite, sandal, elevation, massage, and exercise. All samples except wealth also selected hygiene as part of the CM. Single intervention participants also chose crme, herbal remedies, and pomade, while Protestant s and Catholics indica ted herbal remedies and crme, respectively. Literate individuals endorsed crme as a treatment alternative while non-literate individuals believed in herbal remedies and pomade. Poorer intervention members believed in herbal remedies and crme while younger people

PAGE 187

169 supported crme only. People experiencing more severe disease included crme as a part of their CM and their counterparts indi cated pharmaceutical medicines and pomade. Lastly, individuals experienci ng fewer attacks incorporated crme in their CM and those with more acute attacks chose herbal remedi es. Overall, these findings are similar to those of the full intervention sample CM. The only exception to this trend is the appearance of both pomade and pharmaceutical medicines into the CM for some subsamples. Further discussion of these trends by demographic dimension follows below. Considering the CM elements included for marital status, single people seem more likely to take an integrated appro ach to treating their lymphatic filariasis, incorporating primarily biomedical treatme nt choices followed by some use of more traditional approaches such as herbal remedies and pomade. Within the Haitian context, single individuals may have le ss economic support than their non-single counterparts; as such, they may experience more urgency to reso lve symptoms in order to retain or restore personal economic viability. With regards to religion, these elements included in the culturally correct answer key suggest that both Cathol ics and Protestants benefited from the content of the support group program while Protestants are slightly more likely to integrate some traditional approaches, herbal remedies, in their belie fs about treatment. Generally, though, religion does not seem to heavily influence the type of elements, biomedical vs. traditional, involved in the CM. Taken as a whole, non-literate members of the support group were more likely to identify traditional healing options than th eir literate counterpa rts after support group

PAGE 188

170 participation. The exposure of this segment of the sample to less education, and thus less western ideas, may indicate increased openness to traditional treatment options. While non-literate participants may be more open to traditional concepts, it is important to note that they also endorsed similar biomedical c oncepts to their litera te counterparts. Large differences between literate and non-literate study members may have been minimized by the use of educational materials that include d descriptive pictures in addition to text. In addition to the explanati on above, it is interesting to note that both crme and pomade are endorsed by literate and non-literate individuals, respectively. The presence of pomade in the CM at outcome may be a result of conflated ideas between crme, biomedical, and pomade, traditional. These conc epts are similar, and it is possible that traditional terminology was broadened in the intervention sample community to include crme under the umbrella of the pomade concept. Along the wealth dimension, less wealthy intervention participants also endorsed crme and herbal remedies as treatment altern atives; this finding is similar to the nonliterate outcome intervention sample. It is reasonable to suggest that less wealthy members of the community may also be the se gment of the populati on that has the least access to educational opportunities; thus, it is not surprising that wealth and literacy demographic dimensions produced linked outcomes. For stage of disease, if pomade and cr me were conflated, and both terms are used at outcome to indicate the crme concept, th en people who experience less severe disease may be slightly more open to an integrated a pproach to treating LF. If these two concepts are not conflated, stage of disease does not seem to exert seri ous pressure towards

PAGE 189

171 biomedicine or traditional treatments as both groups, more and less severe, in the intervention sample incorporated at leas t one traditional treatment alternative. In this case, it seems that people experiencing more acute attacks are open to integrated approaches to treating LF as they also belie ve herbal remedies may be helpful. This willingness to endorse both biomedical and more traditional options may be a result of the degree of symptomology they experience; in this situation, people may be more willing to try any treatment they if they beli eve there is a possibility of alleviating the symptoms. When the sample populations were examined by demographics for the model including additional outcome survey items, the trends remained essentially the same. Not all of the same groups met consensus in the full model, but those that met consensus in both the full and reduced model presented w ith the same types of items in the CM. Changes in Cultural Competency: Baseline to Outcome After the samples were examined for the presence of consensus and the items that make up the cultural model, the next rese arch question asked are there significant changes in levels of cultural competency existing at baseline and outcome? It was expected that members of the intervention group would display signi ficant increases in cultural competency as a result of support gr oup participation. The key comparisons were between the baseline and outcome control a nd intervention samples, respectively, and the

PAGE 190

172 alternative hypothesis, that signifi cant changes in competency would occur in the intervention group, was supported. The control group did show some increas e in cultural competency, but there was no statistically significant change over the course of the study. In contrast, the intervention sample displayed increases in cultural competency from baseline to outcome, and these shifts were clearly statis tically significant. However, there were not significant differences in the level of cultural competency between controls and intervention at outcome. These findings furthe r support the idea that controls either received some information through diffusion or proximity to an urban center as suggested earlier. Though there was an increase in competency across both comparison groups, participants in the support group clearly showed larger gains. This increase in competency within the context of more complex cultural models (more items included in the CM) indicates that the support group program was implemented successfully when considering its impact on levels of cultura l expertise and shared belief structures. In addition to being a stand alone succe ssful finding, the significant increase in cultural competency in the intervention gr oup contributes to the body of literature on support groups and changes in knowledge. Firs t, the findings that there are significant changes in knowledge as a result of support group participation support the trends present for other disease categories such as di abetes and cancer (Clark, 2008;Gottlieb & Wachala, 2007; Ferlic, Goldman, & Kenne dy, 1979; Heinrich & Schag, 1985; Cain, Kohorn, Quinlan, Latimer, & Schwartz, 1986; Grahn, & Danielson, 1996; Carlsson & Strang, 1998; Lepore, Helgeson, Eton, & Schul z, 2003; Taylor et al., 2003; Norris,

PAGE 191

173 Engelgau, & Narayan, 2001; Deakin, McShane, Cade, & Williams, 2005). Specifically, this LF support group program lasted longer th an a year, and the significant change in cultural knowledge of LF supports previous findings that sup port groups of longer duration showed more improved knowledge (Clark, 2008; Gottlieb & Wachala, 2007). Also, the presence of significant changes in knowledge resulting from this support group suggests that the CM evaluation tool, al ong the dimension of cultural competency, has something important to offer. Standard evaluation approaches assess the pre-post changes in knowledge in refe rence to pre-determined knowle dge measures. While this is useful to address certain que stions, it is not culturally contextualized. The CM tool, regarding cultural competency, allows for a di rect assessment of i ndividual and aggregate levels of expertise in reference to locally shared cultural beliefs. Another interesting findi ng regarding cultural competency was the significant difference in cultural competency between the outcome intervention sample and the intervention sample including additional items. It was expected that no significant difference would be present between these samp les. As the degree of cultural competency in the full model was much lower than the reduced model and the a posteriori probabilities for items in the CM were also lower when acute attack scenarios were introduced, it is likely that the support group program was mo re effective in addressing general treatment categories for LF than those related to acute attacks. Finally, when demographics were revi ewed, it became apparent that baseline intervention members experiencing fewer attack s displayed higher levels of competency. Since this same trend was not significantly pr esent at outcome, it is likely that people

PAGE 192

174 with more attacks gained more LF specific knowledge from the support group program than those with fewer attacks. People with more attacks may face more challenges with daily living or be less able to participate fu lly in their local communities; as such, they may have had less exposure to local ideas rega rding how to treat LF before participating in the support group. Alternat ely, participants experienci ng more severe disease may have been more highly motivated to adopt health behaviors (Cameron, Leventhal, & Leventhal, 1995); as such, individuals experiencing more attacks may have attended more support group meetings or paid higher atte ntion to the educatio nal content due to a higher perceived need. Changes in Cultural Models: Baseline to Outcome The fourth key research question addre ssed whether or not th e culturally correct answer keys were significantly different fr om baseline to outcome for the comparison groups: baseline controls, baseline interventi on, outcome control, outcome intervention. It was postulated that the in tervention group would show a significant difference in the culturally correct answer keys between the ba seline and outcome poi nts. Additionally, the CM at outcome was expected to reflect both more elements of western biomedical approaches and still retain some traditional elements as well. As expected, the control group, while incr easing in degree of cultural consensus, did not display a significant difference in th e elements included in the CM at outcome. This finding bolsters the id ea that if diffusion of information from the support group

PAGE 193

175 counterparts did occur, this diffusion was not highly si gnificant. If it was highly significant, one would expect ad ditional items to become part of the CM; this situation did not occur. The items included in the CM remained the same. An alternate explanation for the newly developed consensus in the c ontrol group is that diffusion of information from support group counterparts in Archaie and Cabaret occurred or, for La Plaine controls, proximity to an urban center result ed in additional expos ure to western ideas around treatment. A final explanation for the unexpected c onsensus in the control group at outcome includes an element of hist oricity. All members of all groups may have had some increased support for hygiene and sandals as treatment options as a result of a natural evolution in the belief system around LF separate from the support group. Though historicity could have accounted for some of the strong supp ort in all groups for hygiene and sandals, the intervention group also selected additional items as culturally correct at outcome. As such, historicity is not likely to have accounted for all the changes in the intervention group. When the answer keys were compar ed over time, the intervention group did produce significantly different results. Inse ct bites were thought to cause lymphatic filariasis. Additionally, hygiene and sandals continued to receive strong support as treatment possibilities. New items in the model include clear belief in elevation and exercise as well as moderate belief in the use of crme, massage, and herbal remedies. These findings support the hypothesis that west ern ideas would present in the outcome CM. Also, herbal remedies, a traditional treat ment for LF, became a part of the CM at

PAGE 194

176 outcome, and this result supported the sec ond relevant hypothesis, that traditional elements would also present in the CM. Thought s as to why herbal re medies, a traditional treatment, would present at outcome and not at baseline in the intervention sample were previously discussed. As touched upon above, the elements presen t in the culturally correct answer key contribute to the literature on symbolic interactionism. Th is theoretical framework has been the primary approach used to examine cr oss-cultural disability, and it continues to appear applicable when the findings of this study are considered. The answer key reflects participants newly developed belief in severa l of the components in cluded in the support group material that, in comparison to the control group, seem to result from the interactions between study me mbers and the material (intra personal), each other, the facilitators, and others, resp ectively. Also, the introduction of herbal remedies into the answer key specifically speaks to the tenets of the SI framework. This item was not a component of the support group educational conten t. As such, it could only gain in value and consensus within the intervention gr oup through informal interactions between members; one component of each support group meeting, the time for members to share experiences about their disease, lends its elf well to these negotiations of meaning. Additionally, it is possi ble that the components of the e ducation program that were not included in the final answer key were not able to be integrated with the LF paradigm present at baseline as easily as other components included in the culturally correct answer key at outcome.

PAGE 195

177 The final comparison revealed that, at outcome, the control and intervention group displayed significant differences in thei r respective answer keys. This finding both minimizes the likelihood that historicity played a significant role in any of the results and supports the idea that key cha nges are directly related to participation in the support group intervention. Strength of Cultural Consen sus: Baseline to Outcome The fifth question of interest in this study inquires as to whether or not the strength of the cultura l consensus, or the ER, changed fr om baseline to outcome. It was hypothesized that members of the interven tion group would experience changes in the degree of cultural consensus observed between baseline and outcome measurements. This hypothesis was fully supported by the data. Changes in cultural consensus took pl ace in both the control and intervention group over time. These changes could in part be due to reasons st ated above including historicity, diffusion of information (controls only), or proximity to urban centers with more exposure to western medical ideas (c ontrols only). However, even if these influences played a role in increasing the degr ee of consensus for a ll study participants, it is not likely that the change in consensus can be fully attributed to these possibilities. Taking into account the changes in the cultura lly correct answer keys, level of cultural competence, and change in consensus as a whole, it becomes clear that much of the change is due to the support group program. Additionally, the change in the degree of

PAGE 196

178 consensus for the intervention sample was gr eater than that present for the control sample. Lastly, the intervention sample at ou tcome displayed the highest rate of cultural consensus for the four samples (BMC, BMI, OMI-RM, OMC-RM). Link Between Cultural Consen sus and Cultural Consonance The final study question examines th e link between cultu ral consensus and cultural consonance, or the link between belief and behavior. It was expected that greater consensus regarding cultural models woul d be linked to higher rates of self-care behaviors identified in the cultural model. This hypothesis was not supported in the data. In the control sample, consensus increased from baseline to outcome, but consonance decreased at a significant leve l in this same time period. Significant differences in consonance were also present in the intervention samples over time, but the direction of consonance presented in revers e of expectations; c onsonance lessened over time. There are at least two possible explanations for this trend. First, the baseline control sample (BMC) retained two items in the cult urally correct answer key at outcome (OMCRM), but consonance still decreased over ti me. Regarding the control sample, it is possible that both the diffusion and proximity to urban center issues, speculated on previously, led to reinforcement of the two items in the existing CM. This explanation would explain the increased consensus found at outcome. Also, if new treatment ideas were introduced into the cont rol group through these two fo rces, diffusion and proximity to an urban center, it is possi ble that some control members moved away from consistent

PAGE 197

179 behavioral enactment of hygiene and wearing sa ndals in favor of experimenting with new treatment possibilities. If this occurred, thes e new ideas would not be expected to appear as part of the CM because they were not introduced in a formal way and, thus, probably not pervasive enough to be shared on the group level in the control group. A second alternative relevant to the interv ention model is relate d to the number of items present in the model. All groups e xhibited a higher consonance rate than the outcome intervention sample; these three gr oups also had fewer items comprising the cultural model. Only two items were involve d in the CM for both baseline samples (BMI & BMC) as well as the outcome contro l (OMC-RM) population. In contrast, the intervention sample (OMI-RM) included se ven elements. With a higher denominator determining the average rate of consonance, an individual would have to engage in a larger proportion of the behavi ors in the outcome intervention sample to attain the same consonance level that was present at baseline. For instance, at baseline, a member of the intervention group would only have to engage in one behavior to di splay a consonance of .50. At outcome, this same individual would ha ve to engage in three to four of the behaviors to attain the same level of consonance. It is possible that members in the intervention group execute more treatment beha viors than their cont rol counterparts and still display a lower rate of consonance. Another explanation for the consonance findings in the inte rvention group is linked to the absence of clini cal resources in addition to th e support group program. It is possible that members of the support group engaged in less overall health-promoting behavior when health care prof essionals were not present to reinforce the components of

PAGE 198

180 the support group content. This alternative is reasonable as Haitians hold physicians in high esteem (Voyer, Rail, Laberge, & Purne ll, 2005); generally, physicians advise is accepted as truth. If the patient-doctor dynamic was present in this intervention, as opposed to a patient-lay health adviser inte raction, the behavioral outcomes may have manifested differently. Finally, a review of the demographic char acteristics revealed that members of the outcome intervention group (OMI-RM) who were partnered were more likely to have a higher average rate of consonance. This tre nd was also present for Protestants in the baseline intervention (BMI) sample. Literate people in the outcome intervention (OMIRM) showed more consonance than their non-li terate counterparts in the outcome control group (OMC-RM), and, at baseline, both cont rol (BMC) and interv ention (BMI) group members experiencing fewer acute attacks s howed higher consonance. In contrast, at outcome, intervention members (OMI-RM) havi ng more attacks displayed the greater rates of consonance. Lastly, intervention members who were less wealthy at baseline (BMI) and wealthier at outcome (O MI-RM) had higher consonance. These demographics results suggest th at the support group program was more effective for people who were partnered when behavior enactment is considered. This finding is in line with other st udies that have shown that an individuals engagement in health-promoting behaviors can be positivel y impacted by marital status (Falba & Sindelar, 2008; Parruti et al., 2006; Doherty, Schrott, Metcal f, Iasiello-Vailas, 1983). Also, this trend may result as people who are partnered may be more likely to have two household incomes, and additional monies are av ailable to purchase supplies to enact the

PAGE 199

181 treatment regimen. This interpretation complements the finding that wealthier intervention members had higher consonance levels at outcome. Additionally, it seems that when provided with LF specific self-c are practice information, people experiencing more severe symptoms utilized more trea tment options. This finding is reasonable as increased perceived severity of a disease is often linked to greater amounts of healthpromoting intention and behavior (Sherm an, Pennington, Simonton, Latif, Arent, & Farley, 2008; Iriyama, Nakahara, Jimba, Ichikawa, & Wakai, 2007; de Wit, Vet, Schutten, & van Steenbergen, 2005). Prior to expo sure to the support group content, it is possible that people experiencing more seve re symptoms were simply unaware of additional self-care practices that could improve their condition. Cultural Model Evaluation Cultural Consensus Analysis Component Overall, the cultural consensus analysis component performed well within the evaluation context. Changes in consensus were observed as a result of the support group program. Also, this analytic tool was able to identify the shared elements of a cultural model and assess levels of cu ltural competency; it was also able to detect significant changes over time in the latter two measures. When culture is defined as patterned ways of thought and behavior that characterize a so cial group, which are learned through socialization processes and persist through time (Coreil et al., 2001, 29), the cultural

PAGE 200

182 consensus analysis was able to quantitativel y measure shifts in the degree of shared patterned ways of thought about lymphatic fi lariasis in the sample population. Thus, it seems the cultural consensus analysis com ponent of the cultural model evaluation is appropriate and useful when examining longitudi nal changes in shared illness beliefs in a cross-cultural context. Cultural Consonance Component The cultural consonance component provided more mixed results. While the percent coefficients were assessed for degree of behavioral enactment of the corresponding treatment options present in the CM for each sample, the results presented in a trend opposite from that hypothesized. It is possible that th e cultural consonance analysis as it is currently implemented accura tely reflects the behavi oral patterns in the sample data. However, alternate e xplanations may also be relevant. Though the results born out in the consona nce analysis could be accurate, it is likely that this component was implemented in a less sensitive manner than is needed to appropriately assess consonance. For instan ce, if a control sa mple indicated two treatment options in the CM, then that sample was assessed for its rate of behavioral enactment of those two items. However, if the corresponding inte rvention group selected seven items, this group would calculate the degree of consonance based on seven treatment behaviors instead of two. Thus, in this evaluation methodology, it may be more accurate to assess members of both the intervention and control samples for their degree

PAGE 201

183 of behavioral enactment of the items retained in the CM of the intervention sample. In this way, a more direct comparison between the two groups is possible as both groups now share a common denominator. Also, this approach highlights the overall differences in the amount of health-promoting behavior produced in both groups as a result of exposure or non-exposure to the supp ort group intervention, respectively. Sample Size and CM Evaluation Sample Size in the CM evaluation met hod is also reviewe d. After meeting the sample sizes indicated in the power analysis table put forth by Romn ey et al. (1986) in order to conduct the cultural consensus analysis included in this project, a post-hoc review of the key groups was executed. The four comparison groups, baseline controls (BMC), baseline intervention (BMI), outcome controls (OMC-RM), outcome intervention (OMI-RM), were examined for the number of questions d ecisively classified and cultural competency levels. The following proportion of questions wa s decisively classified for baseline controls, baseline intervention, outcome controls, and outcome intervention, respectively: .76, .76, .82, .59.These proportions are less than th e .80 proportion of que stions that need to be decisively classified in order to appl y the power analysis table in Romney et al. (1986). Average cultural competencies are pr esented in the same order: .51, .49, .56, and .62. It is expected that answer s may be given at a rate of guessing among the three groups that present with cultural competencies at or around the .5 level; all three of these groups

PAGE 202

184 were not exposed to the support group intervention. The last group, the outcome intervention group does show an increase in cultural competency that lends itself towards purposeful responses to the questions. When the information in the data is examined and the power analysis table is utilized, it is necessary to use the lowest number of propor tion of questions decisively classified that is available, .80. One of the four groups met this threshold and the other three were below it. With .80 proportion of quest ions correctly answere d, the next step is to examine cultural competency levels in the ta ble. As three of the four groups exhibited at or about .5 competency levels, this value wi ll be designated. In th is scenario, even at a .99 confidence level, a sample size of only 15 is required to produce appropriate results. The control samples were almost twice this size, and the intervention samples were quadruple this size. Thus, it is expected that reaching the .80 threshold would have been easily attained. Multiple explanations for these results exist. The first is that a larger number of items may have resulted in support for this sample size designation with a minimum of .80 questions decisively classi fied. While seventeen items is considered a reasonable number to obtain estimates, a larger number of items allow for a larger proportion to not exhibit a posteriori probabilitie s of .80 or greater and sti ll meet the criterion of .80 proportion of questions clearly classified requ ired to apply the power analysis table. The second possibility is that this po wer analysis table is appropriate for examining a community that has clear ideas about the cultural do main in question, and that this sample size table does not acc ount for communities that are in transition

PAGE 203

185 regarding a cultural domain. Both the outcome control (OMC-RM) and outcome intervention (OMI-RM) groups e xhibited cultural cons ensus (ER >3) while their baseline counterparts (BMC & BMI) did not. Howeve r, several items did not meet the .80 a posteriori probability threshold and result ed in less than .80 questions decisively classified in the outcome intervention samp le. It is possible that beliefs about the treatment items with lower a posteriori pr obabilities, pomade, herbal remedies, and pharmaceutical medicines, are in transition within the community and the conflicted beliefs are reflected in the lower probabiliti es observed. Thus, there is agreement that these items are part of the cu ltural model for which there is consensus, but the importance of their presence in the cultural model is uncertain or in tran sition. It is possible that the power analysis tables presently used are appr opriate to cross-secti onal data only; it may be necessary to further examine power issues when using this analysis to examine longitudinal data. A third possibility is that four of the items, crme/pomade and worms/insect bite, were conflated in the results. In the case of the crme/pomade confusion, each concept, one biomedical (crme) and one traditional (pomade), refers to an ointment like treatment. Therefore, it would not be surpri sing that study participan ts would confuse the two ideas. As neither of these items, crme or pomade, was present in the baseline data, this explanation gains credibility. Additionally, for worms/insect bite, insect bite reflects a biomedical cause of LF and worms are c onsidered a traditional cause category, albeit not one generally associated with LF. Howeve r, it is possible that, by mentioning the tiny worms transmitted to the host through the insect bites, that patients invoked traditional

PAGE 204

186 ideas about worms causing illness, leading to su bsequent confusion of the two terms. If the definitively classified proportion of questi ons was adjusted to reflect these changes, the outcome intervention (OMI-RM) sample w ould meet the .80 threshold required in the sample size table, .82. Only one of the outcome samples met the .80 threshold for proportion of questions definitively classified to obtain accurate results, making it difficult to directly apply the sample size table parameters. Howe ver, the sample sizes required by utilizing Romney et al.s (1986) power table at the lowe st confidence interval are greatly exceeded in the study samples. Additionally, the items reta ined in the cultural models in the control group did not change significantly during the course of the study. This finding further supports that it is possible to be confident that the overall impact of the support group program reported here is not threatened or in accurate do to sample size concerns. Finally, this position is strengthene d as the elements retained in the CM, when run by demographic dimensions and using smaller sa mple sizes, yielded similar results to the overall main comparisons. Implications Implications of the Support Group Program The findings from this support group intervention suggest that, regarding knowledge around LF, this program is a succes s. Changes in the amount of shared

PAGE 205

187 consensus, degree of cultural expertise, and elements included in the CM were observed. These significant results yield multiple implications. The shift in cultural knowledge in the community may act as an indicator of the long-term impact of the support group program, and the resulting behavioral trends displayed by the consonance analysis suggest two possibilities. The first implication is relevant to the community readiness for ch ange. The link between behavior and belief may be weak in this study because the s upport group program was able to influence the community only in its shared thought processe s. In this instance, it seems an additional segment may be required to supplement the support group program content in order to translate the shift in shared cultural knowledge into enacted behavior. An alternative possibility addresses the potential flaw in the consonance analysis. If the methodology requires modification, it is possible that th e support group program did affect both shared cultur al knowledge and behavior; in this case, the consonance analysis, as conducted, may not have been sensitive enough to capture the changes effectively. This scenario suggests that th e support group intervention may have been even more effective than that suggested in the first possible explanation. Thus, in either scenario, the support group program proved its utility. Also important to note is one implica tion related to the support group format in general. As support groups are their own microcosm a nd opportunities for sharing information and negotiation of illness meanings ex ist within this setti ng, it is possible that ideas not directly reflective of the educational content may pr oliferate in this setting. For example, the presence of herbal remedies in the culturally correct answer key at outcome

PAGE 206

188 is a product of these soci al interactions. In th is case, a shared belief in herbal remedies does not contraindicate with the self-car e practices emphasized in the support group. However, in other situations, the dissemi nation of non-biomedical beliefs might be counterproductive. This introduc tion of this kind of misinfor mation may lead some health professionals to be reluctant in endorsing the support group format (Gray, Carroll, Fitch, Greenberg, Chart, & Orr, 2001). In contrast to this potential drawba ck, an additional benefit of support group participation may be increased empowerment on the part of the patient. As lay health advisers presented the information in a mo re egalitarian setting, support group members may take a more active role in negotiati ng meaning around the LF diagnosis. After participation, patients may take a more active role in their own healthcare decisionmaking process. Implications of the CM Evaluation Methodology The CCA portion of the CM evaluati on methodology appears useful in this research endeavor. Changes in cultural knowle dge and behaviors associated with that knowledge were investigated. The knowledge portion of the analysis yields stronger results regarding its utility as an evaluati on tool. However, with some modification, the consonance piece may prove more appropriate than it did in this part icular application. Overall, the test of this methodology s uggests that the CM evaluation can allow for more appropriate tailoring of a public health program or inte rvention. For instance,

PAGE 207

189 herbal remedies became a part of the CM at outcome for the intervention group even though it was not present at baseline. This fi nding suggests that the support group content in this context may need to address the role of herbal remedies in treating LF. Also, the data point to possible confus ion between the terms crme/pomade and insect bite/worms; the program content may need to adjust to better explain what thes e two categories are, how they differ, and which alternatives are appropriate self-care practices for LF. This type of data is valuable as the support group represents a microcosm (Kaye, 1997), and, in this case, the support group microcosm is comprised of persons sharing many characteristics with the larger community. As such, the interactions or confusion between new terms and previously existing categories, described above, may provide insight on how the educational content presented will be integrated into the larger community. Additionally, the implication of this CM ev aluation demonstration is that this tool offers different strengths than those present in traditional evaluation approaches. Traditional evaluation measures target the following dimensions: knowledge, attitudes, behaviors and physiological functioning (McDermott & Sarvela, 1999). These evaluative approaches gather data at the indi vidual level and indir ectly assess community level change by examining the data in aggreg ate. Also, outcome variables are determined prior to program implementation. For example, a traditional evaluation would not have assessed beliefs in herbal remedies at outco me as this content area was not included in the educational materials u tilized in the support group. The CM evaluation methodology, in contra st, was able to assess changes in cultural knowledge with cultura lly meaningful categories. Th e CM approach allowed for

PAGE 208

190 the revelation of local cultural categories of cause and treatment to be included in the evaluation instrument; some of these categor ies, herbal remedies and pomade, were highlighted in the findings. A traditional tool, which typically identifies outcome variables a priori, may not be as sensitive as th e CM approach and it is apparent that local ideas are important as born out in this study. Also, the cultural impact of the program can be assessed in this approach because the cultural knowledge of LF was measur ed in the study popula tion over time. Thus, unlike traditional evaluation tools that meas ure individual knowledge and analyze it in aggregate, thereby making an indirect in ference regarding comm unity level culture change, the CM approach directly measures shared cultural beliefs. These changes, as evidenced in the culturally correct answer keys, proved significant, and, because they are measured at the population level, this appro ach is directly releva nt for assessing public health programs. The final implication of testing this CM evaluation approach is that a new way to evaluate programs has been introduced into the public health disciplin e. This approach to program evaluation is adaptable for interv entions with varied goals; in different scenarios, specific components of the methodol ogy may retain more va lue. For instance, in this study, both the answer key and cultural consonance re sults were important as the goal was to introduce new ideas about cause and treatment of LF through the support group format that ultimately resulted in behaviors reflecting the beliefs introduced through the program. The degree of consensu s and competency were of secondary importance in that they served to strengthe n confidence in the changes within the CCA

PAGE 209

191 component. Another program may intend to increase expertise around existing health care beliefs; in this scenario, the items incl uded in the culturally correct answer key become secondary in importance to the degree of change in cultural competency within the study sample. Consonance may also provide secondary information if they program evaluated is focused on change in knowledge without a behavioral program goal. Per these examples, both consensus analysis and c onsonance analysis have been broadened in their scope methodologically and can be applied with utilit y in a number of different ways. Strength and Limitations of the Study Multiple strengths and limitations are pres ent in the study. First, these data are valuable as they were collected in the setting of a developing country. In these environments, it is challenging to rigorously collect large amounts of interview data. In addition, these data are diachroni c in nature and, as such, in clude a baseline and outcome sample. As a result, it is possible to examine changes over time in this sample as well as assess the impact of the support group interventi on itself. Finally, the data include both a control group and intervention group. With the involvement of a control group it is easier to understand what changes over time are due to the intervention its elf and what trends may be influenced by historical events. Though the dataset itself includes multiples strengths, there are also limitations. First, as the data are secondary in nature, th e limitations related to this kind of research

PAGE 210

192 are inherent. These limitations include that the data were not collected for one of the primary research questions in this work and th at elements of releva nt information may be missing; the original study was designed as a traditional evaluation intervention. Other challenges often posed by secondary data an alysis have been mitigated by working closely with the original rese arch team to understand the context of the data and research process employed to gather the information in this dataset. Another key limitation related to the cultural consensus analysis involves the development of items on the questionnaire th at are used to assess the cultural model. These items were not created through the fr ee listing exercise. Thus, some items also present in the cultural model may not be present in the current data. However, this concern is minimized as the variables incl uded in the cultural model questions were developed through previous ethnographic research in Haiti (J. Coreil, personal communication, 2008). Also, although two components relevant to Kleinmans explanatory models are included in these data, other dimensions are not addressed at all. In order to better assess cultural models in a sample, items relevant to all segments of CMs should be measured. That said, arguably the two most important dimensions of cu ltural models are cause and treatment; both of these concepts are captured in the current data. Related to the inclusion of two dimens ions of the cultural model, the limited dimensions examined resulted in fewer items assessed in the cultural consensus analysis. The inclusion of more items related to the underlying constructs of treatment and cause may have been helpful. That said, as treatm ent and cause constructs related to lymphatic

PAGE 211

193 filariasis are relatively concrete concepts, fewer items are necessary to appropriately assess these ideas than more abstract ideas. In addition, though the sample size is adequate for the analyses, a larger sample of matched participants would be preferable. In itially there were 100 pa rticipants involved in control group and 100 particip ants in the interv ention group. With the involvement of the original research team, however, onl y 87 individuals total (60 intervention, 27 control) could be matched between their baseline and outcome data. Selection bias is also an issue worth no ting. Since participants were not randomly assigned to the intervention group, it is possible that the people who chose to participate in the support group were somehow distinct ively different than their control group counterparts. It is unknown if members of the control group would have responded differently to the support group program conten t than the intervention sample discussed. Future Research This research project introduced a new evaluation indicator. As such, much additional research is needed to grow the methodology and provide additional support that it is indeed a useful tool. This additio nal research is also required to expand the application for this method. Severa l key areas immediately arise. First, the CM evaluation t ool needs to be applied to data that was collected primarily to measure cultural models. Al so, dimensions beyond that of cause and treatment need to be included in the analys is. To address the sample size discussion

PAGE 212

194 above, the instruments used in the CM analys es should include more items for analysis than those present in this st udy. Larger sample sizes are desi rable so that varying sample sizes can be used to test the accuracy of th e CM tools in an environment of fewer and larger samples. Additionally, the CM approach should be applied to cultural domains that are both fully developed and those that are in transition to assess the degree to which the CM approach can accurately assess cultural dom ains that are in transition over time. The consonance portion of this evaluation method also needs to be investigated further. Modifications to the c onsonance tool need to be test ed for appropriateness as part of the CM toolkit. Also, specific to this cultural domain and c ontext, CM evaluation should be applied to sample populations expe riencing the support gr oup with the support of a local clinic and samples without that exposure. A comparative analysis between these two groups may provide additional info rmation regarding the surprising consonance findings resulting in this study do people with more exposure to an LF clinic display alternative degrees of consonance from their co unterparts in rural towns with no clinic available? Another area for future investigation rela tes to outcome evaluation. It is suggested that changes in cultural knowledge reveal ed in the CM evaluation may act as a proxy indicator of long-term sustainable change in resource poor settings when outcome evaluation is not financially supported or easily conducted. Thus, some study needs to explore this idea further and examine the su stainability of the cultural changes in knowledge present at outcome in this study; do these beliefs remain strong in the

PAGE 213

195 community once the program has ended? Do these beliefs in time translate into behavioral change if behavi oral change was not observed in the impact evaluation? Studies utilizing both traditional and cu ltural model evaluation methodologies are also suggested. This undertaking would al low for triangulation of overall outcomes between the two methodologies. Though traditional and cultural model evaluation methods have different strengths and lim itations, confidence in the CM evaluation method could be strengthened if both appro aches yield the same general conclusions. Additionally, specific to the support group itself, the program was implemented over a period of two years. Though originally developed for implementation over a shorter course of time, the support group progr am clearly did not test this idea. Thus, further inquiry on the duration of the support group program required to produce significant changes in belief a nd behavior is warranted. Als o, additional examination of the potential confusion of crme/pomade and in sect bite/worms categories may be useful; the data can be reviewed to assess whethe r or not respondents rare ly selected both of these items as cause categories. If there is little overlap, th is finding would lend itself to the assertion that the terms were conflated in the sample population. Exploration of potential conf ounders to the findings may also be of interest. Ways to examine these confounders may involve the in clusion of additional questions related to issues like diffusion of information and the imp act of more urban locales. Refinement of the methodological design to minimize the li kelihood of confounders like diffusion of information (e.g.: one town is a control gr oup and one is an intervention group), may aid in this process.

PAGE 214

196 Also, inquiry into the types of interv entions that may be appropriate for the application of cultural mode l evaluation is of interest Support group programs are appropriate for this evaluation approach for several reasons; one of these reasons includes the presence of interactio n on several ecological leve ls within the support group. However, other public health interventions reta in different key qualitie s and, as yet, it is unknown the degree to which the cultural model ev aluation is a compatible tool for these alternate contexts. Finally, development of the statistical anal yses to further the sophistication of the cultural model approach is necessary. As it cu rrently stands, the shifts in the degree of cultural consensus over time provide descri ptive information onl y; development of statistical methods that can assess the degree of difference in consensus over time would greatly enhance this methodology. Increased sensitivity in the chi-square analysis implemented in the answer key comparisons may broaden the types of programs that can be assessed with cultural model evaluation and improve its current application. Lastly, the introduction of techniques to assess the im pact of various demographic dimensions on a more nuanced level would also be a usef ul addition to the cultural model evaluation approach. Conclusions Overall, the CM evaluation is an inn ovative evaluation method for public health professionals. The cultural consensus analysis portion exhibits strong evidence for its

PAGE 215

197 utility, while the consonance piece warrant s further investigation. Additionally, the results of this study suggest that support gr oup programs such as the one implemented in this study can be successful in a resource poor setting lacking the support of a lymphatic filariasis hospital clinic.

PAGE 216

198 References Ahorlu, C.K., Dunyo, S.K., Asamoah, G., & Si monsen, P.E. (2001). Consequences of hydrocele and the benefits of hydrocelec tomy: a qualitative study in lymphatic filariasis endemic communities on the coast of Ghana. Acta Tropica 80, 215-221. Ahorlu, C.K., Dunyo, S.K., Kwadwo, A.K., Nkrumah, F.K, Aagaard-Hansen, J., & Simonsen, P.E. (1999). Lymphatic filarias is related perceptions and practices on the coast of Ghana: Implicati ons for prevention and control. Acta Tropica 73, 251-264. Babu, B.V., Hazra, R.K., Chhotray, G.P., & Satyanarayana, K. (2004). Knowledge and beliefs about elephantiasis and hydrocele of lymphatic filariasis and some sociodemographic determinants in an en demic community of Eastern India. Public Health 118, 121-127. Bandyopadhyay, L. (1996). Lymphatic filariasis and the women of India. Social Science and Medicine 42(10), 1401-1410. Barnes, C. (1994). Disabled people in Br itain and discriminati on: A case for anti discrimination legislation London: Hurst and Company in Association with BCODP. Cited in Marks, D. (1999). Disability: Controve rsial debates and psychosocial perspectives London: Routledge.

PAGE 217

199 Beau de Rochars, M.V.E, Milord, M.D., St Jean, Y., Desormeaux, A.M, Dorvil, J.J., Lafontant, J.G., Addiss, D.G., & Streit, T.G. (2004). Geograp hic distribution of lymphatic filariasis in Haiti. American Journal of Tropi cal Medicine and Hygiene 71(5), 598-601. Bigelow, J., Korth, M., Jacobs, J., Anger, N ., Riddle, M., & Giffor d, J. (2004). A picture of amputees and the prosthet ic situation in Haiti. Disability and Rehabilitation 26(4), 246-252. Bradway, C., & Barg, F. (2006). Developing a cultural model for long-term urinary incontinence. Social Science and Medicine 63, 3150-3161. Brodwin, P. (1996). Medicine and morality in Haiti: The contest for healing power Cambridge, MA: Cambridge University Press. Cain, E.N., Kohorn, E.I., Quinlan, D., Latimer, K., & Schwartz, P.E. (1986). Psychosocial benefits of a cancer support group. Cancer 57, 183-189. Cameron, L., Leventhal, E.A., & Leventhal, H. (1995). Seeking medi cal care in response to symptoms and life stress. Psychosomatic Medicine 57, 37-47. Carlsson, M.E., & Strang, P.M. (1998) Educational suppor t programme for gynaecological cancer patients and their families. Acta Oncologica 37(3), 269275. Centers for Disease Control, Department of Parasitic Diseases (CDC DPD) (2004). Lymphatic filariasis fact sheet Retrieved February 25, 2007 from http://www.cdc.gov/ncidod/dpd/parasites/ly mphaticfilariasis/f actsht_lymphatic_fi lar.htm#treatment

PAGE 218

200 Charon, J.M. (2007). Symbolic interactionism: An introduction, an interpretation, an integration, 9th Edition Upper Saddle River, New Je rsey: Pearson, Prentice Hall. Chavez, L.R., McMullin, J.M., Shiraz, I.M, H ubbell, F.A. (2001). Beliefs matter: Cultural beliefs and the use of cervi cal cancer-screening tests. American Anthropologist 103(4), 1114-1129. Clark, M. (2008). Diabetes self-management education: A review of published studies. Primary Care Diabetes 2, 113-120. Coreil, J. & Mayard, G. (2006). Indigeniza tion of illness support groups in Haiti. Human Organization 65(2), 128-139. Coreil, J., Mayard, G., & Addiss, D. (2003) Support groups for women with lymphatic filariasis in Haiti. World Health Orga nization, Social, Economic and Behavioral Research, Report Series, No. 2, [http://www.who.int/tdr/research/finalreps/no56.htm]. Coreil, J., Mayard, G., & Addiss, D. (2002). Support groups for women with lymphatic filariasis in Haiti Final report to World Health Organization, Special Programme for Research and Training in Tropical Diseases. Brief Repor t on WHO website: http://www.who.int/tdr/re search/finalreps/no56.htm Coreil, J., Bryant, C., & Henderson, J.N. (2001). Social and behavioral foundations of public health Thousand Oaks, CA: Sage Publications Inc. Coreil, J., Mayard, G., Louis-Charles, J., & Addiss, D. (1998). F ilarial elephantiasis among Haitian women: Social context a nd behavioural factors in treatment. Tropical Medicine and International Health 3(6), 467-475.

PAGE 219

201 Coreil, J., & Genece, E. (1988). Adoption of oral rehydration therapy among Haitian mothers. Social Science and Medicine 27(1), 87-96. Coreil, J. (1983). Allocation of family re sources for health care in rural Haiti. Social Science and Medicine 17(11), 709-719. Deakin, T., McShane, C.E., Cade, J.E ., & Williams, R.D.R.R. (2008). Group based training for self-management strategies in people with type 2 diabetes mellitus (Review). The Cochrane Library Issue 4. De Wit, J.B., Vet, R., Schutten, M., & va n Steenbergen, J. (2005) Social cognitive determinants of vaccination behavior against hepatitis B: An assessment among men who have sex with men. Preventive Medicine 40(6), 795-802. Doherty, W.J., Schrott, H.G., Metcalf, L., & Ia siello-Vailas, L. (1983) Effect of spouse support and health beliefs on medication adherence. Journal of Family Practice 17(5), 837-841. Dressler, W.W. (2001). Culture and individual adapta tion: Research report 1 University of Alabama. Dressler, W.W. (1996). Culture and blood pressu re: Using consensus analysis to create a measurement. Field Methods 8, 6. Dressler, W.W., Bindon, J.R., & Neggers, Y. H. (1998). Culture, socioeconomic status, and coronary heart disease risk factor s in an African American community. Journal of Behavioral Medicine, 21(6), 527-544.

PAGE 220

202 Dressler, W.W., Balieiro, M.C., & Dos Santos J.E. (1997). The cultu ral construction of social support in Brazil: Associ ations with health outcomes. Culture, Medicine, & Psychiatry 21, 303-335. Dressler, W.W., Dos Santos, J.E., & Ba lieiro, M.C. (1996). St udying diversity and sharing in culture: An example of lifestyle in Brazil. Journal of Anthropological Research 52(3), 331-353. Eberhard, M.L., Walker, E.M., Addiss, D. G., & Lammie, P.J. (1996). A survey of knowledge, attitudes, and perceptions (KAPs) of lymphatic filariasis, elephantiasis, and hydrocele among reside nts in an endemic area in Haiti. American Journal of Tropic al Medicine and Hygiene 54(3), 299-303. Fadiman, A. (1997). The spirit catches you and you fall down: A Hmong child, her American doctors, and the collision of two cultures New York: Farrar, Straus, and Giroux. Falba, T.A., & Sindelar, J.L. (2008). Spousal concordance in hea lth behavior change. Health Services Research 43, 96-116. Farmer, P. (1992). AIDS and accusation: Haiti and the geography of blame Berkeley and Los Angeles: University of California Press. Farmer, P. (1990). Sending sickness: sorcery, po litics, and changing concepts of AIDS in rural Haiti. Medical Anthropology Quarterly 4(1): 6-27. Ferlic, M., Goldman, A., & Kennedy, B.J. (1979). Group counseling in adult patients with advanced cancer. Cancer 43, 760-766.

PAGE 221

203 Forte, J.A. (2001). Theories for practice: Symbol ic interactionist translations Lanham, Maryland: University Press of America, Inc. Foster, G.M. (1976). Disease etiologi es in non-western medical systems. American Anthropologist 78(4), 773-782. Fulton, G., Madden, C., & Minichello, V. (1996). The social construction of anticipatory grief. Social Science and Medicine 43(9), 1349-1358. Good, B. J. (1994). Medicine, rationality, experience: an anthr opological perspective New York: Cambridge University Press. Gottlieb, B.H., & Wachala, E.D. (2007). Can cer support groups: A critical review of empirical studies. Psycho-oncology 16, 379-400. Grahn, G., & Danielson, M. (1996). Coping with the cancer experience. II. Evaluating an education and support program me for cancer patients and their significant others. European Journal of Cancer Care 5, 182-187. Gray, R.E., Carroll, J.C., Fitch, M., Greenber g, M., Chart, P., & Orr, V. (2001). Cancer self-help groups and family physicians. Cancer Practice 7(1), 10-15. Greeson, C.J., Veach, P.M., & LeRoy, B.S. ( 2001). A qualitative investigation of Somali immigrant perceptions of disability: Implications for genetic counseling. Journal of genetic counseling 10(5), 359-378. Gwet, K. (2002). Computing inter-rate r reliability with the SAS system. Series: Statistical Methods for InterRater Reliability Assessment No. 3.

PAGE 222

204 Gyapong, M., Gyapong, J.O., Adjei, S., Vlassoff, C., & Weiss, M. (1996). Filariasis in northern Ghana: Some cultural beliefs and practices and their implications for disease control. Social Science and Medicine 43(2), 235-242. Halcon, L., Blum, R.W., Beuhring, T., Pate, E., Campbell-Forrester, S., & Venema, A. (2003). Adolescent health in the Caribbean: A regional portrait. American Journal of Public Health 93(11), 1851-1857. Hatcher, L., & Stepanski, E.J. (1994). A step-by-step approach to using the SAS system for univariate and multivariate statistics Cary, NC: SAS Institute Inc. Heinrich, R.L., & Schag, C.C. (1985). Stre ss and activity management: Group treatment for cancer patients and spouses. Journal of Consulting and Clinical Psychology 53(4), 439-446. Herman, J. (2005). Cancer support groups: a guide for facilitators American Cancer Society. Heurtin-Roberts, S. (1992). The relation of culturally influenced lay models of hypertension to compliance with treatment. American Journal of Hypertension 5(11), 787-792. Iriyama, S., Nakahara, S., Jimba, M., Ichika wa, M., & Wakai, S. (2007). AIDS health beliefs and intention for sexual abstin ence among male adolescent students in Kathmandu, Nepal: A test of perceive d severity and susceptibility. Public Health 121(1), 64-72.

PAGE 223

205 Jacobson, E. (2003). An introduction to Ha itian culture for re habilitation service providers. Center for International Re habilitation Research Information and Exchange. Jean, Y., & Haider-Markel, D.P. (2001). Trac ing issue definition and policy change: An analysis of disability issu e images and policy response. Policy Studies Journal 29(2), 215. Kanda, K. (2004). The quality of life among lymphede ma patients due to lymphatic filariasis in three rural towns in Haiti Unpublished masters thesis, University of South Florida, Tampa, Florida. Kaye, L.W. (1997). Self-help support groups for ol der women: Rebuilding elder networks through personal empowerment Taylor & Francis. Kleinman, A. (1988). The illness narratives: Suffering, healing, and the human condition New York: Basic Books, Inc. Kleinman, A. (1978). Culture, illness, and car e: clinical lessons from anthropologic and crosscultural research. Annals of Internal Medicine 88, 251-258. Kramer, K.D., & Nash, K.B. (1995). The Un ique Social Ecology of Groups: Findings from Group for African Americans Af fected by Sickle Cell Disease. In Support groups: Current perspectives on theory and practice New York: The Haworth Press, Inc. Kurtz, L.F. (1997). Self-help and support groups: A handbook for practitioners Thousand Oaks: SAGE Publications.

PAGE 224

206 Lepore, S.J., Helgeson, V., Eton, D., & Schulz, R. (2003). Improving quality of life in men with prostate cancer: A randomized controlled trial of group education interventions. Health Psychology 22(5), 443-452. Mann, J. (1997). Medicine and public health, ethics and human rights. The Hastings Center Report 27(3), 6-13. Marks, D. (1999). Disability: Controversial deba tes and psychosocial perspectives London: Routledge. Marriner-Tomey, A. (1998). Introduction to an alysis of nursing theories. In Marriner Tomey, A., Allgood, A., & Raile, M., Nursing theorists and their work (4thed). St. Louis: Mosby. Mathews, H.F., Lannin, D.R., & Mitchell, J. P. (1994). Coming to te rms with advanced breast cancer: Black womens narrati ves from eastern North Carolina. Social Science and Medicine 38(6), 789-800. Mayhew, L. (2003). Disability: Global tr ends and international perspectives. The European Journal of Social Research 16(1), 3-28. McCallion, P., Janicki, M., & Grant-Griffin, L. (1997). Explor ing the impact of culture and acculturation on older families caregiving for persons with developmental disabilities. Family Relations 46(4), 347-357. McDermott, R.J., & Sarvela, P.D. (1999). Health education evaluat ion and measurement: A practitioners perspective WCB/McGraw-Hill.

PAGE 225

207 Mead, G.H. (1934). Mind, self, and society (C. Morris Introduction and Ed.). Chicago: University of Chicago. Cited in White, J.M., & Klein, D.M. (2002). Family theories (2nd ed.). Thousand Oaks, CA: Sage Publications, Inc. Miller, N.L. (2000). Haitian ethnomedical systems and biomedical practitioners: Directions for clinicians. Journal of Transcultural Nursing 11(3), 204-211. Morfit, V. (1998). Lymphatic filariasis infecti on among men in a Haitian town: explanatory models Unpublished masters thesis, University of South Florida, Tampa, Florida, United States. National Council on Disability (2002). Understanding the role of an international convention on the human rights of people w ith disabilities: an analysis of the legal social, and practical implications for policy makers and disability and human rights advocates in the United States: a white paper Retrieved November 25, 2004, from NCD website via GPO Access: http://www.ncd.gov/newsroom/ publications/unwhitepaper%5F05-2302.html Norris, S.L., Engelgau, M.M., & Venkat Nara yan, K.M. (2001). Effectiveness of self management training in Type 2 diabetes : A systematic review of randomized trials. Diabetes Care 24(3), 561-587. Parker, J.N., & Parker, P.M. (2002). The official patients sourcebook on lymphatic filariasis: A revised and updated directory for the internet age San Diego, CA: ICON Health Publications.

PAGE 226

208 Parruti, G., Manzoli, L., Toro, P.M., DAmico, G., Rotolo, S., Graziani, V. et al. (2006). Long-term adherence to first line highly active antiretroviral th erapy in a hospitalbased cohort: Predictors and impact on virologic response and relapse. AIDS Patient Care STDS 20(1), 48-56. Person, B., Addiss, D., Bartholomew, L.K., Me ijer, C., Pou, V., Gonzalvez, G. et al. (2007). A qualitative study of the psyc hosocial and health consequences associated with lymphedema among women in the Dominican Republic. Acta Tropica 103, 90-97. Ramaiah, K.D., Vijay Kumar, K.N., & Ram u, K. (1996). Knowledge and beliefs about transmission, prevention, a nd control of lymphatic filariasis in rural areas of South India. Tropical Medicine and International Health 1(4), 433-438. Rauyajin, O., Kamthornwachara, B., & Yablo, P. (1995). Socio-cultural and behavioral aspects of mosquito-borne lymphatic filari asis in Thailand: A qualitative analysis. Social Science and Medicine 41(12), 1705-1713. Ritzer, G. (2000). Modern sociological theory, 5th edition Boston: McGraw-Hill. Romney, A.K. (1999). Culture cons ensus as a statistical model. Current Anthropology 40(Supplement), S103-S115. Romney, A.K., Batchelder, W.H., & Weller, S. (1987). Recent applications of cultural consensus theory. The American Behavioral Scientist 31, 2, 163. Romney, A.K., Weller, S.C., & Batchelder, W. H. (1986). Culture as consensus: A theory of culture and informant accuracy. American Anthropologist 88(2), 313-338.

PAGE 227

209 Scotch, R. (2000). Models of disability a nd the American with Disabilities Act. Berkeley Journal of Employment and Labor Law 21, 213-222. Schopler, J.H., & Galinksy, M.J. (1995). E xpanding Our View of Support Groups as Open Systems. In Support groups: Current perspectives on theory and practice New York: The Haworth Press, Inc. Sherman, A.C., Pennington, J., Simonton, S., Lati f, U., Arent, L., & Farley, H. (2008). Determinants of participation in cancer support groups: The role of health beliefs. International Journal of Behavioral Medicine 15(2), 92-100. Support Group. (2007). The American Heritage Dictionary of the English Language, Fourth Edition Houghton Mifflin Company. Supreme Court of the United States (1997). Sy llabus: Bragdon vs. Abbot et al. Retrieved May 24, 2004, from http://supct.law.co rnell.edu/supct/html/97-156.ZS.html Tashakkori. T. (1998). Mixed methodology Thousand Oaks, CA: Sage Publications. Taylor, K.L., Lamdan, R.M., Siegel, J.E., Shelby, R., Moran-Klimi, K., & Hrywna, M. (2003). Psychological adjustment among African American breast cancer patients: One-year follow-up results of a randomized psychoeducational group intervention. Health Psychology 22(3), 316-323. Thorburn, M.J. (1999). The role of family: disa bility and rehabilitati on in rural Jamaica. The Lancet 354(9180), 762-763. United States Department of Justice. (1992). Office of the attorney general: Nondiscrimination on the basis of disabi lity in state and local government services. Retrieved October 18, 2003, from www.usdoj.gov/cr t/ada/reg2.html.

PAGE 228

210 United States Department of Ju stice. (n.d.). Disability disc rimination law applicable to federal employment. Retrieved October 18, 2003, from www.bop.gov/hrmpg/lmr/hrmlmrxs.pdf Voyer, P., Rail, G., Laberge, S., & Purnel l, L. (2005). Cultural minority older womens attitudes towards medication and implic ations for adherence to a medicine regimen. Diversity in Health and Social Care 2(1), 47-62. Weller, S. (2007). Cultural consensus theo ry: Applications and frequently asked questions. Field Methods 19, 339. Weller, S. (1987). Shared knowledge, intracult ural variation, and knowledge aggregation. American Behavioral Scientist 31, 2, 178-193. Weller, S. (1984). Consistency and consensus among informants: Disease concepts in a rural Mexican village. American Anthropologist 86(4), 966-975. White, J.M., & Klein, D.M. (2002) Family Theories (2nd ed.) Thousand Oaks, CA: Sage Publications, Inc. World Health Organization (WHO) (2007). Th e global program to eradicate lymphatic filariasis (GPELF). Re trieved January 10, 2007 from http://www.who.int/lymphatic_filariasis/disease/en/ World Health Organization (WHO) (2007b). Ly mphatic filariasis: The disease and its epidemiology. Retrieved January 14, 2007 from http://www.who.int/lymphatic_f ilariasis/epidemiology/en/ World Health Organization (WHO) (2006a). Disa bility and rehabilita tion team. Retrieved July 13th, 2006 from http://www.who.int/dis abilities/in troduction/en/index.html

PAGE 229

211 World Health Organization (WHO) (2006b). Global program to eliminate lymphatic filariasis. Weekly epidemiological record 81 (22), 221-232. World Health Organization (WHO) (2006c). Informal consultation on preventing disability from lymphatic fila riasis, WHO, Geneva, August 2006. Weekly epidemiological record 81(40), 373-384. World Health Organization (W HO). (2006d). A Workshop to revi ew the impact of mass drug administration (MDA) on lymphatic fila riasis (LF) disease, infection and transmission: 11-13 December 2006, Sa lle B, WHO/HQ Geneva, Switzerland. Retrieved January 15th, 2008 from http://www.who.int/neglected_diseases/massdrug/en/ World Health Organization (WHO) (2005) Disability, including prevention, management, and rehabilitation. Proceedings of the 58th World Health Assembly, Agenda Item 13.13. World Health Organization (WHO) (2004). Global program for the elimination of lymphatic filariasis: strategic pl an 2003-2005, challenges of scaling up Geneva: World Health Organization. World Health Organization (WHO) (2000). Lym phatic filariasis: Fact sheet. Retrieved July 13th, 2006 from http://www.who.int/med iacentre/factsh eets/fs102/en/ World Health Organization (WHO) (1997) Disability, including prevention, management, and rehabilitation. Proceedings of the 50th World Health Assembly, Agenda Item 20.

PAGE 230

212 Ziegler, R.G. (1982). Epilepsy: Individual illness, human predicament, and family dilemma. Family Relations 31(3), 435-444.

PAGE 231

213 Appendix A

PAGE 232

214 Appendix A General demographic results from the cultu ral consensus analyses are presented in the body of Chapter 4. Specific results of th e statistical analyses and comparisons summarized previously are captured in the tables below. Demographic dimensions reviewed here include: marital status, religion, literacy, wealth, age, stage of disease, and number of acute attacks. Marital Status In the baseline analyses of marital status an eigenvalue ratio greater than three was present for the full baseline sample (B -All) when participants were not single. Table A1. Baseline Sample: Consensus Anal ysis Results by Group and Marital Status N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Not Single Single Not Single Single Not Single Single Not Single Single BMC 23 4 6.811 .596 2.587 .238 2.633 2.504 BMI 33 27 9.444 7.815 3.810 2.987 2.479 2.616 B-All 159 82 47.97324.45615.7109.146 3.054 2.674 The culturally correct answer keys were comprised of different items by marital status and group. Both the baseline contro l groups highlighted hygiene and sandal as treatment options, but only the single group al so indicated herbal remedies and pomade

PAGE 233

215 Appendix A as well. The intervention group showed a si milar trend, but the si ngle group suggested elevation for treatment instead of herbal reme dies or pomade. Finall y, the entire baseline sample endorsed hygiene, sanda ls, and herbal remedies. Table A2. Baseline Sample: Culturally Corre ct Answer Keys by Intervention and Control Group by Marital Status BMC BMI B-All N=23 N=4 N=33 N=27 N=159 N=82 Answer(Weighted %) Not Single Single Not Single Single Not Single Single Cause Insect Bite 0(100%) 0(100%) 0(93%) 0(96%) 0(97%) 0(97%) Magic 0(85%) 0(77%) 0(91%) 0(96%) 0(91%) 0(96%) Sprain 0(100%) 0(78%) 0(92%) 0(100%) 0(90%) 0(96%) Worms 0(83%) 0(100%) 0(100%) 0(98%) 0(93%) 0(98%) Chill 0(100%) 0(63%) 0(84%) 0(97%) 0(86%) 0(91%) Vitamin Deficiency 0(100%) 0(63%) 0(100%) 0(100%) 0(99%) 0(97%) Treatment Hygiene 1(94%) 1(78%) 1(88%) 1(95%) 1(93%) 1(91%) Sandal 1(97%) 1(100%) 1(94%) 1(99%) 1(97%) 1(98%) Permanganate 0(82%) 0(83%) 0(100%) 0(68%) 0(87%) 0(78%) Crme 0(80%) 0(100%) 0(80%) 0(72%) 0(80%) 0(84%) Elevation 0(75%) 0(83%) 0(73%) 1(50%) 0(74%) 0(61%) Massage 0(81%) 0(100%) 0(96%) 0(92%) 0(88%) 0(95%) Exercise 0(96%) 0(100%) 0(94%) 0(76%) 0(95%) 0(92%) Bandage 0(72%) 0(100%) 0(80%) 0(92%) 0(73%) 0(93%) Medicine 0(87%) 0(100%) 0(80%) 0(86%) 0(80%) 0(69%) Herbal Remedy 0(51%) 1(59%) 0(51%) 0(52%) 1(58%) 1(57%) Pomade 0(72%) 1(59%) 0(61%) 0(73%) 0(65%) 0(55%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met

PAGE 234

216 Appendix A Comparisons of the answer keys belo w also indicate a lack of significant differences within each group wh en compared by marital status. Table A3. Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Marital Status Chi-Square df P< Fishers Exact Two-Tailed p< BMC: Not Single vs Single .8095 1 .3683 .6562* BMI: Not Single vs Single .2345 1 .6282 1.00* B-All: Not Single vs Single .0000 1 1.00 1.00* indicates which statistic was interpreted for significant differences between the answer keys In addition to examining levels of c onsensus and analyzing answer keys for differences, average levels of cultura l competency were also assessed.

PAGE 235

217 Appendix A Table A4. Baseline Sample: Cultural Compet ency Analysis Results by Group and Marital Status N Average Competency Not SingleSingleNot Single Single BMC 23 4 .51 .37 BMI 33 27 .51 .47 B-All 159 82 .52 .49 *One individual removed from analyses in the outcome control group due to missing data Table A5. Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Marital Status df t p< BMC: Not Single vs Single 25 1.47 .1531 BMI: Not Single vs Single 42.2 .66 .5140 B-All: Not Single vs Single 133 .87 .3886

PAGE 236

218 Appendix A Religion In the case of religion, Protestants and Catholics were compared; there were too few individuals who selected vodou, other, or no religion to include in the analysis. Consensus was met for Catholics in the fu ll baseline sample (B-All) as well at the baseline controls (BMC). Interestingly, Cat holics exhibited a higher level of consensus across all baseline samples, and this trend is reversed in all bu t one of the outcome samples. Table A6. Baseline Sample: Consensus Analysis Results by Group and Religion N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Protestant Catholic Protestant Catholic Protestant Catholic Proteestant Catholic BMC 6 15 1.658 4.694 1.637 .918 1.013 5.114 BMI 28 27 8.073 7.915 3.414 2.996 2.365 2.642 B-All 96 118 26.81237.32410.60811.672 2.527 3.198 Once the degree of consensus in a sample was established, the answer keys were examined. The baseline control samples sele cted hygiene and sandal as salient features, and the Protestant control sample also incl uded herbal remedies. Intervention groups at

PAGE 237

219 baseline also believe hygiene and sandals are appropriate treatment choices for LF; additionally, the Catholic portion of the in tervention sample endorses herbal remedies and pomade. Overall, both Protestants and Cat holics in the full baseline sample identified hygiene, sandals, and herbal reme dies as ways to treat LF.

PAGE 238

220 Appendix A Table A7. Culturally Correct Answer Ke ys for Intervention and Control Group by Religion Baseline Sample BMC BMI B-All N=6 N=15 N=28 N=27 N=96 N=118 Answer(Weighted %) Protestant Catholic Protestant Catholic Protestant Catholic Cause Insect Bite 0(100%) 0(100%) 0(97%) 0(93%) 0(96%) 0(98%) Magic 0(100%) 0(93%) 0(93% ) 0(93%) 0(96%) 0(97%) Sprain 0(100%) 0(100%) 0( 95%) 0(96%) 0(92%) 0(92%) Worms 0(65%) 0(93%) 0(100% ) 0(97%) 0(93%) 0(95%) Chill 0(83%) 0(100%) 0(84% ) 0(94%) 0(86%) 0(90%) Vitamin Deficiency 0(83%) 0(100%) 0(100%) 0(100% ) 0(98%) 0(99%) Treatment Hygiene 1(100%) 1(99%) 1( 96%) 1(87%) 1(92%) 1(96%) Sandal 1(100%) 1(99%) 1(100% ) 1(93%) 1(98%) 1(97%) Permanganate 0(91%) 0(71%) 0(87%) 0(85%) 0(83%) 0(80%) Crme 0(78%) 0(86%) 0(72% ) 0(80%) 0(79%) 0(82%) Elevation 0(78%) 0(71%) 0( 58%) 0(72%) 0(70%) 0(65%) Massage 0(85%) 0(78%) 0(96%) 0(100% ) 0(85%) 0(93%) Exercise 0(88%) 0(100%) 0( 86%) 0(86%) 0(91%) 0(96%) Bandage 0(91%) 0(63%) 0( 78%) 0(93%) 0(75%) 0(80%) Medicine 0(100%) 0(78%) 0( 89%) 0(75%) 0(84%) 0(68%) Herbal Remedy 1(63%) 0(61%) 0(65%) 1(64%) 1(51%) 1(58%) Pomade 0(61%) 0(79%) 0( 83%) 1(52%) 0(65%) 0(60%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met

PAGE 239

221 Appendix A After review of the elements in each of the answer keys, these answer keys and levels of cultural competency were compar ed for significant differences. These analyses are below: Table A8. Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Religion Chi-Square df p< Fishers Exact Two-Tailed p< BMC: Protestant vs Catholic .2345 1 .6282 1.00* BMI: Protestant vs Catholic .8095 1 .3683 .6562* B-All: Protestant vs Catholic .0000 1 1.00 1.00* indicates which statistic was interpreted for significant differences between the answer keys Table A9. Baseline Sample: Cultural Co mpetency Analysis Results by Group and Religion N Average Competency ProtestantCatholicProtestant Catholic BMC 6 15 .50 .53 BMI 28 27 .49 .49 B-All 96 118 .49 .52

PAGE 240

222 Appendix A Table A10. Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Religion Df t p< BMC: Protestant vs Catholic 19 .26 .7965 BMI: Protestant vs Catholic 53 .06 .9520 B-All: Protestant vs Catholic 212 1.15 .2511 No significant differences between answer keys by religion existed within each of the groups. Regarding cultural competency, no significant differences existed between conditions, Protestant and Catholic, within each of the groups analyzed. Literacy In the case of literacy, the threshold for consensus is met in all baseline samples among those who are not literate. Table A11. Baseline Sample: Consensus An alysis Results by Group and Literacy N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Literate Not Literate Literate Not Literate Literate Not Literate Literate Not Literate BMC 15 12 3.967 3.875 2.165 .989 1.832 3.917 BMI 38 22 10.9356.235 5.110 1.907 2.140 3.270 B-All 137 104 40.47732.20414.54710.161 2.782 3.169

PAGE 241

223 Appendix A In all the baseline groups, elements of the cultural model include hygiene and sandal. Treatment with herbal remedies was al so identified as a key element in several of the baseline samples. Table A12. Culturally Correct Answer Keys for Intervention and Control Groups by Literacy Baseline Sample BMC BMI B-All N=15 N=12 N=38 N=22 N=137N=104 Answer(Weighted %) Literate Not Literate Literate Not Literate Literate Not Literate Cause Insect Bite 0(100%) 0(100%) 0(92%) 0(100%) 0(96%)0(99%) Magic 0(86%) 0(77%) 0(95%) 0(89%) 0(94%)0(90%) Sprain 0(97%) 0(100%) 0(94%) 0(100%) 0(91%)0(92%) Worms 0(82%) 0(95%) 0(98%) 0(100%) 0(91%)0(99%) Chill 0(95%) 0(100%) 0(84%) 0(100%) 0(90%)0(86%) Vitamin Deficiency 0(95%) 0(100%) 0(100%) 0(100%) 0(99%)0(99%) Treatment Hygiene 1(93%) 1(95%) 1(90%) 1(94%) 1(93%)1(92%) Sandal 1(100%) 1(95%) 1(95%) 1(100%) 1(98%)1(96%) Permanganate 0(77%) 0(83%) 0(91%) 0(83%) 0(82%)0(85%) Crme 0(71%) 0(95%) 0(76%) 0(76%) 0(74%)0(91%) Elevation 0(71%) 0(78%) 0(57%) 0(77%) 0(64%)0(76%) Massage 0(66%) 0(100%) 0(93%) 0(100%) 0(85%)0(96%) Exercise 0(100%) 0(95%) 0(89%) 0(83%) 0(92%)0(96%) Bandage 0(59%) 0(91%) 0(76%) 0(100%) 0(69%)0(93%) Medicine 0(86%) 0(91%) 0(80%) 0(85%) 0(74%)0(81%) Herbal Remedy 0(65%) 1(60%) 1(58%) 0(71%) 1(55%)1(61%) Pomade 0(74%) 0(68%) 0(64%) 0(73%) 0(68%)0(55%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met

PAGE 242

224 Appendix A When the answer keys were compared for differences by literacy, no significant differences were found. Table A13. Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Cont rol Groups by Literacy ChiSquare df p< Fishers Exact Two-Tailed p< BMC: Literate vs Not Literate .2345 1 .6282 1.00* BMI: Literate vs Not Literate .2345 1 .6282 1.00* B-All: Literate vs Not Literate .0000 1 1.00 1.00* In addition to the lack of significant difference between answer keys within each group, no significant differences in levels of cultural competency within each group by literacy were present either. Table A14. Baseline Sample: Cultural Comp etency Analysis Results by Group and Literacy N Average Competency Literate Not Literate Literate Not Literate BMC 15 12 .48 .53 BMI 38 22 .50 .47 B-All 137 104 .51 .50

PAGE 243

225 Appendix A Table A15. Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Literacy df t p< BMC: Literate vs Not Literate 25 -.60 .5534 BMI: Literate vs Not Literate 58 .54 .5927 B-All: Literate vs Not Literate 189 .28 .7815 In the case of literacy, the answer keys showed no significant differences within each group. The same trend held true when di fferences in cultural competency were examined within each group. Wealth Within the wealth comparisons, less wealthy members of the baseline intervention (BMI) and full (B-All) samples showed consen sus. Also, wealthier baseline controls (BMC) also met the criteria for consensus.

PAGE 244

226 Appendix A Table A16. Baseline Sample: Consensus An alysis Results by Group and Wealth N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Less Wealth More Wealth Less Wealth More Wealth Less Wealth More Wealth Less Wealth More Wealth BMC 24 3 6.863 .668 2.672 .166 2.569 4.034 BMI 35 25 9.661 7.860 3.182 3.565 3.036 2.205 B-All 172 69 51.027 21.021 17.001 7.999 3.001 2.628 Answer keys were also compared and examined for significant differences within each group by wealth. All baselin e samples identified hygiene and sandal as appropriate treatment options. Additionall y, all of the groups with a more wealth designation highlighted herbal remedies and poma de as viable treatment choices.

PAGE 245

227 Appendix A Table A17. Culturally Correct Answer Keys for Intervention and Control Groups by Wealth Baseline Sample BMC BMI B-All N=24 N=3 N=35 N=25 N=172 N=69 Answer (Weighted %) Answer (Weighted %) Answer (Weighted %) Less Wealth More Wealth Less Wealth More Wealth Less Wealth More Wealth Cause Insect Bite 0(100%) 0(100%) 0(97%) 0(93%) 0(99%) 0(94%) Magic 0(86%) 0(59%) 0(92%) 0(95%) 0(91%) 0(96%) Sprain 0(98%) 0(100%) 0(92%) 0(100%) 0(94%) 0(87%) Worms 0(87%) 0(71%) 0(100%) 0(97%) 0(95%) 0(93%) Chill 0(96%) 0(100%) 0(100%) 0(73%) 0(91%) 0(79%) Vitamin Deficiency 0(96%) 0(100%) 0(100%) 0(100%) 0(99%) 0(97%) Treatment Hygiene 1(97%) 1(71%) 1(96%) 1(86%) 1(94%) 1(89%) Sandal 1(99%) 1(100%) 1(97%) 1(98%) 1(97%) 1(99%) Permanganate 0(76%) 0(100%) 0(80%) 0(95%) 0(82%) 0(90%) Crme 0(78%) 0(100%) 0(69%) 0(88%) 0(82%) 0(83%) Elevation 0(67%) 0(100%) 0(58%) 0(76%) 0(71%) 0(66%) Massage 0(87%) 0(71%) 0(93%) 0(99%) 0(89%) 0(94%) Exercise 0(96%) 0(100%) 0(80%) 0(96%) 0(94%) 0(95%) Bandage 0(72%) 0(100%) 0(93%) 0(71%) 0(82%) 0(77%) Medicine 0(87%) 0(100%) 0(88%) 0(72%) 0(80%) 0(69%) Herbal Remedy 0(63%) 1(100%) 0(59%) 1(60%) 1(57%) 1(58%) Pomade 0(76%) 1(71%) 0(78%) 1(51%) 0(66%) 1(50%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met

PAGE 246

228 Appendix A After the answer keys were reviewed for the elements included in each cultural model, the answer keys were compared be tween more and less wealthy individuals within each group. These anal yses indicate that there ar e no significant differences between answer keys within each group. Table A18. Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Cont rol Groups by Wealth ChiSquare df p< Fishers Exact Two-Tailed p< BMC: Less Wealth vs More Wealth .8095 1 .3683 .6562* BMI: Less Wealth vs More Wealth .8095 1 .3683 .6562* B-All: Less Wealth vs More Wealth .1799 1 .6715 1.00* Further analyses were conducted to asse ss rates of cultural competency in each group and whether or not signi ficant differences in competency exist between less and more wealthy individuals in each sample. No significant differences were present in the baseline sample.

PAGE 247

229 Appendix A Table A19. Baseline Sample: Cultural Comp etency Analysis Results by Group and Wealth N Average Competency Less Wealth More Wealth Less Wealth More Wealth BMC 24 3 .50 .47 BMI 35 25 .46 .53 B-All 172 69 .50 .52 Table A20. Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Wealth df t p< BMC: Less Wealth vs More Wealth 25 .29 .77 BMI: Less Wealth vs More Wealth 58 -1.04 .3025 B-All: Less Wealth vs More Wealth 239 -.58 .5655

PAGE 248

230 Appendix A Age When looking at cultural models present in the sample by age, the median age, 46 years old, was used to divide the sample into younger and older groupings. With this grouping, there is no consensus present in any of the baseline samples. Table A21. Baseline Sample: Consensus Analysis Results by Group and Age N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Younger Older Younger Older Younger Older Younger Older BMC 14 13 3.451 4.384 1.477 2.057 2.337 2.131 BMI 28 32 7.354 9.481 3.411 3.737 2.156 2.537 B-All 121 120 34.81737.68311.62413.119 2.995 2.872 A review of the answer keys indicates which elements were salient across different groups. All baseline groups indicate hygiene and sandals as viable treatment options. Additionally some of the groups also identify herbal remedies as a treatment option.

PAGE 249

231 Appendix A Table A22. Culturally Correct Answer Keys for Intervention and Control Groups by Age Baseline Sample BMC BMI B-All N=14 N=13 N=28 N=32 N=121 N=120 Answer (Weighted %) Answer (Weighted %) Answer (Weighted %) Younger Older Younger Older Younger Older Cause Insect Bite 0(100%) 0(100%) 0(100%) 0(91%) 0(99%) 0(96%) Magic 0(94%) 0(73%) 0(94%) 0(94%) 0(93%) 0(92%) Sprain 0(100%) 0(96%) 0(95%) 0(96%) 0(91%) 0(92%) Worms 0(96%) 0(79%) 0(98%) 0(100%) 0(92%) 0(97%) Chill 0(93%) 0(100%) 0(87%) 0(92%) 0(89%) 0(87%) Vitamin Deficiency 0(93%) 0(100%) 0(100%) 0(100%) 0(99%) 0(98%) Treatment Hygiene 1(94%) 1(96%) 1(89%) 1(94%) 1(93%) 1(91%) Sandal 1(98%) 1(100%) 1(97%) 1(97%) 1(98%) 1(97%) Permanganate 0(82%) 0(80%) 0(96%) 0(80%) 0(86%) 0(83%) Crme 0(86%) 0(82%) 0(77%) 0(74%) 0(74%) 0(89%) Elevation 0(86%) 0(67%) 0(67%) 0(59%) 0(64%) 0(76%) Massage 0(71%) 0(92%) 0(96%) 0(94%) 0(86%) 0(95%) Exercise 0(100%) 0(94%) 0(82%) 0(90%) 0(91%) 0(97%) Bandage 0(61%) 0(88%) 0(85%) 0(86%) 0(72%) 0(87%) Medicine 0(82%) 0(94%) 0(82%) 0(83%) 0(75%) 0(78%) Herbal Remedy 0(67%) 1(57%) 0(57%) 1(51%) 1(53%) 1(62%) Pomade 0(71%) 0(67%) 0(77%) 0(58%) 0(71%) 0(52%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met In an examination of culturally correct answer keys and cultural competency, each group exhibited no significant differences by age.

PAGE 250

232 Appendix A Table A23. Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Age ChiSquare df p< Fishers Exact Two-Tailed p< BMC: Younger vs Older .2345 1 .6282 1.00* BMI: Younger vs Older .2345 1 .6282 1.00* B-All: Younger vs Older .0000 1 1.00 1.00* Table A24. Baseline Sample: Cultural Compet ency Analysis Results by Group and Age N Average Competency YoungerOlder Younger Older BMC 14 13 .44 .57 BMI 28 32 .47 .50 B-All 121 120 .50 .52 Table A25. Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Age df T p< BMC: Younger vs Older 19.5 1.78 .0906 BMI: Younger vs Older 58 .57 .5715 B-All: Younger vs Older 239 .65 .5139

PAGE 251

233 Appendix A Stage of Disease At baseline, only the control group (BMC) afflicted with more severe symptoms showed consensus. Table A26. Baseline Sample: Consensus Analys is Results by Group and Stage of Disease N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Less Severe More Severe Less Severe More Severe Less Severe More Severe Less Severe More Severe BMC 9 18 2.482 5.426 1.694 1.395 1.465 3.890 BMI 36 24 10.4246.392 3.606 3.383 2.891 1.890 B-All 125 116 37.23134.65013.11411.861 2.839 2.921 Answer key comparisons reveal that the all members of baseline control groups identify hygiene and sandal as key treatment opt ions, but people with less severe disease also believe that herbal re medies and pomade are appropr iate. Members of the baseline intervention groups also identified hygiene and sandal as treatment choices; less severely afflicted members of the intervention group at baseline also believe herbal remedies are relevant for treatment. In the overall base line sample, hygiene, sandals, and herbal remedies are key treatment choices.

PAGE 252

234 Appendix A Table A27. Culturally Correct Answer Keys for Intervention and Control Groups by Stage of Disease Baseline Sample BMC BMI B-All N=9 N=18 N=36 N=24 N=125 N=116 Answer (Weighted %) Answer (Weighted %) Answer (Weighted %) Less Severe More Severe Less Severe More Severe Less Severe More Severe Cause Insect Bite 0(100%) 0(100%) 0(95%) 0(96%) 0(97%) 0(97%) Magic 0(75%) 0(87%) 0(95%) 0(90%) 0(92%) 0(94%) Sprain 0(100%) 0(97%) 0(93%) 0(100%) 0(91%) 0(92%) Worms 0(90%) 0(84%) 0(98%) 0(100%) 0(95%) 0(94%) Chill 0(87%) 0(100%) 0(91%) 0(88%) 0(85%) 0(91%) Vitamin Deficiency 0(87%) 0(100%) 0(100%) 0(100%) 0(98%) 0(99%) Treatment Hygiene 1(86%) 1(96%) 1(91%) 1(94%) 1(92%) 1(93%) Sandal 1(96%) 1(99%) 1(96%) 1(99%) 1(97%) 1(98%) Permanganate 0(68%) 0(89%) 0(86%) 0(90%) 0(85%) 0(83%) Crme 0(76%) 0(91%) 0(82%) 0(64%) 0(79%) 0(85%) Elevation 0(76%) 0(78%) 0(64%) 0(59%) 0(73%) 0(65%) Massage 0(92%) 0(80%) 0(95%) 0(96%) 0(92%) 0(88%) Exercise 0(100%) 0(95%) 0(88%) 0(83%) 0(93%) 0(95%) Bandage 0(68%) 0(83%) 0(85%) 0(88%) 0(81%) 0(79%) Medicine 0(76%) 0(95%) 0(87%) 0(75%) 0(75%) 0(79%) Herbal Remedy 1(76%) 0(64%) 1(50%) 0(58%) 1(58%) 1(57%) Pomade 1(64%) 0(84%) 0(72%) 0(59%) 0(54%) 0(70%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met Answer key comparisons within each group by stage revealed no significant differences in the elements comprising the cultural model by disease stage.

PAGE 253

235 Appendix A Table A28. Baseline Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Stage of Disease ChiSquare df p< Fishers Exact Two-Tailed p< BMC: Less Severe vs More Severe .8095 1 .3683 .6562* BMI: Less Severe vs More Severe .2345 1 .6282 1.00* B-All: Less Severe vs More Severe .0000 1 1.00 1.00* The level of cultural competency present in each sample by disease stage was also assessed, and no significant differences were found within each sample. Table A29. Baseline Sample: Cultural Compet ency Analysis Results by Group and Stage of Disease N Average Competency Less Severe More Severe Less Severe More Severe BMC 9 18 .49 .52 BMI 36 24 .51 .44 B-All 125 116 .51 .50

PAGE 254

236 Appendix A Table A30. Baseline Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Stage of Disease df t p< BMC: Less Severe vs More Severe 25 -.31 .7590 BMI: Less Severe vs More Severe 36.9 1.02 .3153 B-All: Less Severe vs More Severe 239 .17 .8652 Number of Attacks Consensus at baseline was indicated in the baseline control (BMC) and full baseline (B-All) samples for people with fewer attacks. Table A31. Baseline Sample: Consensus Anal ysis Results by Group and Number of Attacks N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Fewer Attacks More Attacks Fewer Attacks More Attacks Fewer Attacks More Attacks Fewer Attacks More Attacks BMC 19 8 6.002 1.989 1.003 1.740 5.986 1.143 BMI* 31 28 10.1806.471 3.722 3.189 2.735 2.029 B-All 149 89 47.17423.40214.27910.039 3.304 2.331

PAGE 255

237 Appendix A Culturally correct answer keys were produced for each sample along the dimension number of acute attacks within th e past year. Individuals in the control group experiencing more attacks suggest that worm s are the cause of lym phatic filariasis. All groups indicate that hygiene and sandals ar e appropriate treatment choices. Herbal remedies are also highlighted in some of the baseline samples.

PAGE 256

238 Appendix A Table A32. Culturally Correct Answer Keys for Intervention and Control Groups and Number by Attacks Baseline Sample Control Intervention Baseline All N=19 N=8 N=31 N=28 N=149 N=89 Answer(Weighted %) Attacks Fewer More Fewer More Fewer More Cause Insect Bite 0(100%) 0(100%) 0(95%) 0(95%) 0(97%) 0(97%) Magic 0(82%) 0(91%) 0(88%) 0(100%) 0(91%) 0(96%) Sprain 0(97%) 0(100%) 0(93%) 0(100%) 0(92%) 0(93%) Worms 0(95%) 1(54%) 0(100%) 0(97%) 0(96%) 0(92%) Chill 0(100%) 0(85%) 0(97%) 0(78%) 0(89%) 0(83%) Vitamin Deficiency 0(100%) 0(85%) 0(100%) 0(100%) 0(100%) 0(96%) Treatment Hygiene 1(95%) 1(80%) 1(97%) 1(81%) 1(94%) 1(88%) Sandal 1(98%) 1(96%) 1(95%) 1(98%) 1(98%) 1(96%) Permanganate 0(79%) 0(95%) 0(86%) 0(90%) 0(83%) 0(86%) Crme 0(91%) 0(75%) 0(76%) 0(76%) 0(82%) 0(82%) Elevation 0(79%) 0(75%) 0(58%) 0(76%) 0(73%) 0(67%) Massage 0(82%) 0(95%) 0(91%) 0(100%) 0(91%) 0(89%) Exercise 0(100%) 0(89%) 0(91%) 0(80%) 0(95%) 0(92%) Bandage 0(73%) 0(95%) 0(87%) 0(82%) 0(81%) 0(78%) Medicine 0(89%) 0(95%) 0(82%) 0(81%) 0(77%) 0(76%) Herbal Remedy 0(51%) 1(61%) 0(51%) 0(51%) 1(56%) 1(62%) Pomade 0(78%) 1(63%) 0(57%) 0(76%) 0(60%) 0(61%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met Within the number of attacks demogra phic characteristic, culturally correct answer keys were tested for significant differences. No si gnificant differences existed.

PAGE 257

239 Appendix A Table A33. Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Number of Attacks ChiSquare df p< Fishers Exact TwoTailed p< BMC: Fewer Attacks vs More Attacks 1.6190 1 .2032 .3983* BMI: Fewer Attacks vs More Attacks .0000 1 1.00 1.00* B-All: Fewer Attacks vs More Attacks .0000 1 1.00 1.00* indicates which statistic was interpreted for significant differences between the answer keys In addition to analyzing the answer keys for differences, average levels of cultural competency were also assessed. The baseline intervention (BMI) and baseline all (B-All) groups exhibited significant differences in cultural competency at baseline. Table A34. Baseline Sample: Cultural Comp etency Analysis Results by Group and Number of Attacks Fewer Attacks More Attacks Fewer Attacks More Attacks BMC 19 8 .53 .44 BMI 31 28 .55 .42 B-All 149 89 .53 .46 One individual removed from analyses in the intervention group due to missing data

PAGE 258

240 Appendix A Table A35. Comparisons of Cultural Competen cy for Intervention and Control Groups by Number of Attacks df t p< BMC: Fewer Attacks vs More Attacks 25 1.12 .2728 BMI: Fewer Attacks vs More Attacks 44.9 2.51 .0156 B-All: Fewer Attacks vs More Attacks 157 2.49 .0139

PAGE 259

241 Appendix B

PAGE 260

242 Appendix B General demographic results from the cultu ral consensus analyses are presented in the body of Chapter 4. Specific results of th e statistical analyses and comparisons summarized previously are captured in the tables below. Demographic dimensions reviewed here include: marital status, religion, literacy, wealth, age, stage of disease, and number of acute attacks. Marital Status An eigenvalue ratio greater than three was present for the outcome control group with additional items (OMC-FM) for the si ngle sample, as well as the control group (OMC-RM), intervention group (OMI-RM) a nd intervention group with additional items at outcome (OMI-FM) regardless of marital status. When a group exhibited consensus in either or both conditions, it was stronger for the single portion of the sample. Table B1. Outcome Sample: Consensus Anal ysis Results by Group and Marital Status N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Not Single Single Not Single Single Not Single Single Not Single Single OMC-RM* 22 4 7.471 .867 1.998 .201 3.739 4.309 OMI-RM 33 27 14.62412.1303.228 1.802 4.530 6.732 OMC-FM* 22 4 5.696 .645 2.426 .149 2.348 4.329 OMI-FM 33 27 10.7159.709 2.184 1.963 4.907 4.947 *One individual removed from analyses in the outcome control group due to missing data

PAGE 261

243 Appendix B The culturally correct answer keys were comprised of different items by marital status and group. Regarding the outcome sample s, the control groups produced the same items in the cultural model as were present at baseline. In contrast the intervention group cultural model included additional elements. Both married and single individuals selected insect bite as the cause of LF and indi cated hygiene, sandals, elevation, massage, and exercise for treatment. However, single people in the intervention group also chose crme, herbal remedies, and pomade.

PAGE 262

244 Appendix B Table B2. Culturally Correct Answer Keys for Intervention and Control Groups by Marital Status Outcome Sample OMC-RM OMI-RM N=22 N=4 N=33 N=27 Answer (Weighted %) Answer (Weighted %) Not Single Single Not Single Single Cause Insect Bite 0(100%) 0(75%) 1(75%) 1(68%) Magic 0(96%) 0(100%) 0(96%) 0(97%) Sprain 0(100%) 0(100%) 0(96%) 0(96%) Worms 0(82%) 0(100%) 0(81%) 0(96%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(98%) 1(100%) 1(97%) 1(97%) Sandal 1(96%) 1(100%) 1(98%) 1(98%) Permanganate 0(98%) 0(82%) 0(83%) 0(69%) Crme 0(96%) 0(100%) 0(56%) 1(56%) Elevation 0(85%) 0(82%) 1(93%) 1(94%) Massage 0(96%) 0(100%) 1(66%) 1(65%) Exercise 0(96%) 0(100%) 1(86%) 1(92%) Bandage 0(91%) 0(100%) 0(83%) 0(78%) Medicine 0(58%) 1(58%) 0(61%) 0(63%) Herbal Remedy 0(50%) 1(58%) 0(50%) 1(69%) Pomade 0(67%) 0(82%) 0(67%) 1(49%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data In the case of the full models, including the items present in the outcome sample only, both single and not single people identified the same key elements with the exception of pharmaceutical medicines; this item was only present in the control group

PAGE 263

245 comprised of single participants. For the di fferent treatment questions presented, the members of the intervention group, both singl e and not single, agreed on the following elements: insect bite, hygiene, sandal, elevation, massage, ex ercise. Single individuals in the intervention condition al so included herbal reme dies, pomade, and crme. Table B3. Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Marital Status Outcome Sample OMC-FM OMI-FM N=22 N=4 N=33 N=27 Answer(Weighted %) Not Single Single Not Single Single Cause Insect Bite 0(100%)0(64%) 1(75%) 1(73%) Magic 0(92%) 0(100%) 0(98%) 0(100%) Sprain 0(100%)0(100%) 0(96%) 0(94%) Worms 0(94%) 0(100%) 0(78%) 0(95%) Chill 0(100%)0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%)0(100%) 0(100%) 0(100%) Treatment Hygiene 1(100%)1(100%) 1(98%) 1(99%) Sandal 1(100%)1(100%) 1(98%) 1(99%) Permanganate 0(100%)0(83%) 0(85%) 0(73%) Crme 0(99%) 0(100%) 0(58%) 0(52%) Elevation 0(96%) 0(83%) 1(92%) 1(98%) Massage 0(99%) 0(100%) 1(64%) 1(67%) Exercise 0(99%) 0(100%) 1(89%) 1(98%) Bandage 0(92%) 0(100%) 0(87%) 0(77%) Medicine 0(72%) 0(53%) 0(61%) 0(68%) Herbal Remedy 0(85%) 0(73%) 0(52%) 1(69%) Pomade 0(87%) 0(83%) 0(69%) 1(53%) Treatment (OC Only) Hygiene 1(93%) 1(90%) 1(98%) 1(98%)

PAGE 264

246 Table B3. (continued) OMC-FM OMI-FM N=22 N=4 N=33 N=27 Answer(Weighted %) Not Single Single Not Single Single What kinds of care can help your gwopye? Sandal 1(85%) 1(90%) 1(83%) 1(76%) Permanganate 0(100%) 0(83%) 0(64%) 0(72%) Crme 0(96%) 0(83%) 0(57%) 1(53%) Elevation 0(89%) 0(83%) 1(90%) 1(96%) Massage 0(96%) 0(100%) 1(70%) 1(58%) Exercise 0(96%) 0(100%) 1(87%) 1(80%) Bandage 0(96%) 0(100%) 0(82%) 0(77%) Medicine 0(82%) 1(54%) 0(62%) 0(62%) Nothing 0(98%) 0(100%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(70%) 1(90%) 1(89%) 1(90%) What can you do to prevent acute attacks? Sandal 1(70%) 1(90%) 1(75%) 1(81%) Permanganate 0(100%) 0(83%) 0(87%) 0(91%) Crme 0(100%) 0(83%) 0(79%) 0(79%) Elevation 0(93%) 0(83%) 1(65%) 1(80%) Massage 0(100%) 0(100%) 1(55%) 1(57%) Exercise 0(100%) 0(100%) 1(65%) 1(85%) Bandage 0(100%) 0(100%) 0(95%) 0(94%) Medicine 0(93%) 0(64%) 0(65%) 0(62%) Nothing 0(96%) 0(90%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(81%) 1(73%) 1(73%) 1(87%) What can be done to provide relief during an acute attack? Sandal 1(77%) 1(73%) 1(53%) 1(69%) Permanganate 0(100%) 0(100%) 0(72%) 0(86%) Crme 0(99%) 0(100%) 0(75%) 0(79%) Elevation 0(92%) 0(100%) 1(60%) 1(76%) Massage 0(99%) 0(100%) 1(54%) 1(62%) Exercise 0(99%) 0(100%) 1(57%) 1(75%) Bandage 0(92%) 0(100%) 0(86%) 0(94%) Medicine 0(72%) 0(53%) 0(56%) 0(74%) Nothing 0(96%) 0(100%) 0(100%) 0(98%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 265

247 Appendix B Comparisons of the answer keys belo w also indicate a lack of significant difference within each group when compared by marital status. Table B4. Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Marital Status Chi-Square df p< Fishers Exact Two-Tailed p< OMC-RM: Not Single vs Single .8095 1 .3683 .6562* OMC-FM: Not Single vs Single .0718* 1 .7887 1.00 OMI-RM: Not Single vs Single 1.074* 1 .3001 .4905 OMI-FM: Not Single vs Single .3837* 1 .5356 .6799 indicates which statistic was interpreted for significant differences between the answer keys In addition to examining levels of c onsensus and analyzing answer keys for differences, average levels of cultura l competency were also assessed.

PAGE 266

248 Appendix B Table B5. Outcome Sample: Cultural Co mpetency Analysis Results by Group and Marital Status Outcome Sample N Average Competency Not Single Single Not Single Single OMC-RM 22 4 .54 .46 OMI-RM 33 27 .63 .61 OMC-FM 22 4 .42 .37 OMI-FM 33 27 .52 .50 *One individual removed from analyses in the outcome control group due to missing data Table B6. Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Marital Status Df t p< Outcome Controls: Not Single vs Single 24 .79 .4396 Outcome Control with Additional Items: Not Single vs Single 24 .33 .7452 Outcome Intervention: Not Single vs Single 58 .24 .8078 Outcome Intervention with Additional Items: Not Single vs Single 45 .29 .7742 The highest cultural competency presente d in the intervention group at outcome across both marital conditions, and this findi ng is in line with overall study hypotheses. Additionally, no significant differences in leve ls of competency exist within each group

PAGE 267

249 Appendix B by marital status. This informa tion, considered with the leve ls of cultural consensus and lack of significant differences in answer keys suggests that single i ndividuals experience more cohesiveness in their cu ltural model of LF, but that this difference is not great enough to impact the overall shared cultura l model when both single and not single participants are grouped together. Religion In the case of religion, Protestants and Catholics were compared; there were too few individuals who selected vodou, other, or no religion to include in the analysis. Consensus was met for several groups: Protesta nts and Catholics in the outcome controls sample (OMC-RM), Protestants and Catholics in the outcome intervention sample (OMIRM), Protestants in the outcome controls with additional items group (OMC-FM), and both Protestants and Catholics in the in out come intervention group with additional items (OMI-FM). Consensus was highest in the tw o intervention groups for the Protestant sample. Interestingly, Catholics exhibited a hi gher level of consensu s across all baseline samples, and this trend is reversed in all but one of the outcome samples.

PAGE 268

250 Appendix B Table B7. Outcome Sample: Consensus Analysis Results by Group and Religion N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Protestant Catholic Protestant Catholic Protestant Catholic Proteestant Catholic OMC-RM 6 15 1.259 5.135 .393 1.4999 3.201 3.425 OMI-RM 28 27 13.19812.0271.866 3.105 7.071 3.874 OMC-FM 6 15 1.188 3.969 .288 1.644 4.120 2.415 OMI-FM 28 27 9.377 10.0771.441 2.552 6.508 3.949 Once the degree of consensus in a sample was established, the answer keys were examined. In the outcome sample, elements for cause and treatment of lymphatic filariasis were high lighted. Within the c ontrol sample, hygiene and sandals are the key treatment pieces and are supported in both the Protestant and Catholic portions of the population. Regarding the intervention samples, in sect bites are seen as the cause of LF. Protestants and Catholics in the intervention gr oup think that hygiene, sandal, elevation, massage, and exercise are ways to treat their disease. Additionally, Protestants and Catholics endorse herbal remedi es and crme, respectively.

PAGE 269

251 Appendix B Table B8. Culturally Correct Answer Keys for Intervention and Control Groups by Religion Outcome Sample OMC-RM OMI-RM N=6 N=15 N=28 N=27 Answer (Weighted %) Answer (Weighted %) Protestant Catholic Protestant Catholic Cause Insect Bite 0(84%) 0(100%) 1(75%) 1(65%) Magic 0(100%) 0(94%) 0(98%) 0(94%) Sprain 0(100%) 0(100%) 0(97%) 0(94%) Worms 0(71%) 0(86%) 0(89%) 0(90%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(92%) 1(100%) 1(95%) 1(100%) Sandal 1(100%) 1(94%) 1(98%) 1(97%) Permanganate 0(92%) 0(93%) 0(85%) 0(70%) Crme 0(100%) 0(95%) 0(57%) 1(62%) Elevation 0(92%) 0(73%) 1(94%) 1(93%) Massage 0(100%) 0(95%) 1(69%) 1(63%) Exercise 0(100%) 0(95%) 1(89%) 1(92%) Bandage 0(100%) 0(87%) 0(83%) 0(79%) Medicine 0(64%) 0(56%) 0(66%) 0(62%) Herbal Remedy 0(52%) 0(65%) 1(66%) 0(53%) Pomade 0(72%) 0(56%) 0(58%) 0(57%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met

PAGE 270

252 Appendix B Within the full models, including additional items in the outcome data only, controls, both Protestant and Catholics thought that hygiene and sandals are appropriate ways to treat lymphatic filariasis. For th e intervention group, the following treatment options were indicated in one or both re ligious groups: hygiene, sandal, elevation, massage, exercise, crme, herbal remedy. Both intervention groups agree that insect bites cause LF. Table B9. Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Religion Outcome Sample OMC-FM OMI-FM N=6 N=15 N=28 N=27 Answer (Weighted %) Answer (Weighted %) ProtestantCatholic Protestant Catholic Cause Insect Bite 0(73%) 0(100%)1(76%) 1(70%) Magic 0(100%) 0(90%) 0(99%) 0(98%) Sprain 0(100%) 0(100%)0(96%) 0(94%) Worms 0(93%) 0(94%) 0(87%) 0(86%) Chill 0(100%) 0(100%)0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%)0(100%) 0(100%) Treatment Hygiene 1(100%) 1(100%)1(97%) 1(100%) Sandal 1(100%) 1(98%) 1(99%) 1(98%) Permanganate 0(100%) 0(94%) 0(87%) 0(72%) Crme 0(100%) 0(98%) 0(59%) 1(57%) Elevation 0(100%) 0(86%) 1(95%) 1(95%) Massage 0(100%) 0(98%) 1(70%) 1(62%) Exercise 0(100%) 0(98%) 1(91%) 1(98%) Bandage 0(100%) 0(89%) 0(87%) 0(79%) Medicine 0(66%) 0(67%) 0(65%) 0(68%) Herbal Remedy 0(86%) 0(83%) 1(66%) 0(56%) Pomade 0(93%) 0(76%) 0(55%) 0(59%)

PAGE 271

253 Appendix B Table B9. (continued) OMC-FM OMI-FM N=6 N=15 N=28 N=27 Answer (Weighted %) Protestant Catholic Protestant Catholic Treatment (OC Only) Hygiene 1(86%) 1(95%) 1(98%) 1(99%) What kinds of care can help your gwopye? Sandal 1(86%) 1(84%) 1(75%) 1(84%) Permanganate 0(100%) 0(94%) 0(54%) 0(85%) Crme 0(100%) 0(88%) 0(51%) 1(51%) Elevation 0(100%) 0(77%) 1(91%) 1(95%) Massage 0(100%) 0(94%) 1(74%) 1(67%) Exercise 0(100%) 0(94%) 1(93%) 1(94%) Bandage 0(100%) 0(94%) 0(86%) 0(91%) Medicine 0(72%) 0(68%) 0(55%) 0(84%) Nothing 0(93%) 0(100%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(86%) 1(68%) 1(83%) 1(94%) What can you do to prevent acute attacks? Sandal 1(86%) 1(68%) 1(73%) 1(84%) Permanganate 0(100%) 0(94%) 0(83%) 0(95%) Crme 0(100%) 0(94%) 0(74%) 0(84%) Elevation 0(100%) 0(83%) 1(72%) 1(69%) Massage 0(100%) 0(100%) 1(65%) 0(55%) Exercise 0(100%) 0(100%) 1(78%) 1(67%) Bandage 0(100%) 0(100%) 0(93%) 0(98%) Medicine 0(73%) 0(89%) 0(64%) 0(79%) Nothing 0(100%) 0(100%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(86%) 1(78%) 1(81%) 1(78%) What can be done to provide relief during an acute attack? Sandal 1(86%) 1(72%) 1(61%) 1(61%) Permanganate 0(100%) 0(100%) 0(65%) 0(90%) Crme 0(100%) 0(98%) 0(70%) 0(83%) Elevation 0(100%) 0(87%) 1(75%) 1(59%) Massage 0(100%) 0(98%) 1(72%) 0(58%) Exercise 0(100%) 0(98%) 1(67%) 1(63%) Bandage 0(100%) 0(89%) 0(82%) 0(98%) Medicine 0(65%) 0(67%) 0(61%) 0(71%) Nothing 0(100%) 0(100%) 0(98%) 0(100%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met

PAGE 272

254 Appendix B Table B10. Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Religion Chi-Square df p< Fishers Exact Two-Tailed p< OMC-RM: Protestant vs Catholic .0000 1 1.00 .3988* OMC-FM: Protestant vs Catholic .0000* 1 1.00 1.00 OMI-RM: Protestant vs Catholic .0000* 1 1.00 1.00 OMI-FM: Protestant vs Catholic .0429* 1 .8360 1.00 indicates which statistic was interpreted for significant differences between the answer keys Table B11. Outcome Sample: Cultural Comp etency Analysis Results by Group and Religion N Average Competency ProtestantCatholicProtestant Catholic OMC-RM 6 15 .44 .56 OMI-RM 28 27 .64 .62 OMC-FM 6 15 .36 .45 OMI-FM 28 27 .52 .53

PAGE 273

255 Appendix B Table B12. Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Religion Df t p< OMC-RM: Protestant vs Catholic 19 1.57 .1340 OMC-FM: Protestant vs Catholic 19 .70 .4906 OMI-RM: Protestant vs Catholic 53 -.36 .7227 OMI-FM: Protestant vs Catholic 53 .17 .8631 No significant differences between answer keys by religion existed within each of the groups. Regarding cultural competency, bo th religious groups experienced increases in average cultural competency between baseline and outcome in the intervention samples. However, no significant differences existed between conditi ons, Protestant and Catholic, within each of the groups analy zed. These findings in general suggest that individuals in the intervention groups gained more cultural co mpetency than their control counterparts as a result of the support group, a nd that Protestants may have been more receptive to the content of th e support group module as rates of consensus were greatest among these participants at outcome.

PAGE 274

256 Appendix B Literacy In the case of literacy, strongest consensus is present among the outcome controls both reduced and full samples (OMC-RM & OMCFM). Consensus if also found in all members of the interventi on group in both full and reduced models (OMI-RM & OMIFM); however, consensus is greater am ong individuals who are not literate. Table B13. Outcome Sample: Consensus An alysis Results by Group and Literacy N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Literate Not Literate Literate Not Literate Literate Not Literate Literate Not Literate OMC-RM* 15 11 5.603 3.978 1.226 .231 4.571 17.206 OMI-RM 38 22 18.8698.831 4.008 1.563 4.708 5.651 OMC-FM* 15 11 3.039 3.856 2.163 .199 1.405 19.391 OMI-FM 38 22 13.3387.695 2.816 1.498 4.737 5.136 *One individual removed from analyses in the outcome control group due to missing data With regards to the reduced model out come samples, hygiene and sandals are present in the cultural model. Additional tr eatment items highlighted in the outcome intervention samples include the use of crme, elevation, massage, exercise, herbal remedies, and pomade. One cause, insect bite, is included in these groups.

PAGE 275

257 Appendix B Pharmaceutical medicine and herbal remedies were identified in the outcome control sample comprised of l iterate individuals. Table B14. Culturally Correct Answer Keys for Intervention and Control Groups by Literacy Outcome Sample OMC-RM OMI-RM N=15 N=11 N=38 N=22 Answer (Weighted %) Literate Not Literate Literate Not Literate Cause Insect Bite 0(94%) 0(100%) 1(75%) 1(67%) Magic 0(100%) 0(91%) 0(97%) 0(95%) Sprain 0(100%) 0(100%) 0(99%) 0(91%) Worms 0(71%) 0(100%) 0(91%) 0(83%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(95%) 1(100%) 1(100%) 1(90%) Sandal 1(93%) 1(100%) 1(100%) 1(93%) Permanganate 0(96%) 0(93%) 0(73%) 0(83%) Crme 0(94%) 0(100%) 1(64%) 0(78%) Elevation 0(74%) 0(93%) 1(94%) 1(93%) Massage 0(94%) 0(100%) 1(67%) 1(66%) Exercise 0(94%) 0(100%) 1(91%) 1(87%) Bandage 0(94%) 0(90%) 0(75%) 0(90%) Medicine 1(55%) 0(69%) 0(84%) 0(60%) Herbal Remedy 1(70%) 0(93%) 0(58%) 1(88%) Pomade 0(57%) 0(82%) 0(64%) 1(51%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 276

258 Appendix B For the outcome samples including additional items, insect bite is an element of the cultural model among both interventi on groups. Across the various treatment scenarios presented, the intervention samples selected hygiene, sandals, crme, elevation, massage, exercise, herbal remedy, pomade, and medicine as treatment options. Within the control groups, hygiene and sandals were consis tently chosen in re sponse to the various treatment situations. Table B15. Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Literacy Outcome Sample OMC-FM OMI-FM N=15 N=11 N=38 N=22 Answer (Weighted %) Literate Not Literate Literate Not Literate Cause Insect Bite 0(88%) 0(100%) 1(76%) 1(72%) Magic 0(100%)0(88%) 0(98%) 0(98%) Sprain 0(100%)0(100%) 0(100%) 0(88%) Worms 0(83%) 0(100%) 0(88%) 0(83%) Chill 0(100%)0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%)0(100%) 0(100%) 0(100%) Treatment Hygiene 1(97%) 1(100%) 1(100%) 1(95%) Sandal 1(97%) 1(100%) 1(100%) 1(96%) Permanganate 0(98%) 0(94%) 0(76%) 0(86%) Crme 0(97%) 0(100%) 1(61%) 0(79%) Elevation 0(85%) 0(94%) 1(94%) 1(97%) Massage 0(97%) 0(100%) 1(66%) 1(69%) Exercise 0(97%) 0(100%) 1(94%) 1(92%) Bandage 0(97%) 0(90%) 0(75%) 0(96%) Medicine 0(57%) 0(73%) 0(66%) 0(57%) Herbal Remedy 0(57%) 0(94%) 0(59%) 1(85%) Pomade 0(76%) 0(87%) 0(64%) 1(52%) Treatment (OC Only) Hygiene 1(74%) 1(100%) 1(99%) 1(97%)

PAGE 277

259 Appendix B Table B15. (continued)\ OMC-FM OMI-FM N=15 N=11 N=38 N=22 Answer(Weighted %) Literate Not Literate Literate Not Literate What kinds of care can help your gwopye? Sandal 1(64%) 1(95%) 1(76%) 1(87%) Permanganate 0(100%) 0(94%) 0(70%) 0(61%) Crme 0(90%) 0(94%) 1(51%) 0(55%) Elevation 0(85%) 0(83%) 1(92%) 1(94%) Massage 0(90%) 0(100%) 1(71%) 1(72%) Exercise 0(90%) 0(100%) 1(93%) 1(89%) Bandage 0(90%) 0(100%) 0(87%) 0(85%) Medicine 0(73%) 0(74%) 0(79%) 1(59%) Nothing 0(95%) 0(100%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(57%) 1(79%) 1(88%) 1(92%) What can you do to prevent acute attacks? Sandal 1(57%) 1(79%) 1(76%) 1(83%) Permanganate 0(100%) 0(94%) 0(85%) 0(94%) Crme 0(100%) 0(94%) 0(72%) 0(94%) Elevation 0(97%) 0(83%) 1(66%) 1(82%) Massage 0(100%) 0(100%) 0(51%) 1(72%) Exercise 0(100%) 0(100%) 1(67%) 1(87%) Bandage 0(100%) 0(100%) 0(94%) 0(93%) Medicine 0(86%) 0(90%) 0(75%) 0(53%) Nothing 0(87%) 0(100%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(66%) 1(82%) 1(73%) 1(90%) What can be done to provide relief during an acute attack? Sandal 1(57%) 1(82%) 1(51%) 1(77%) Permanganate 0(100%) 0(100%) 0(70%) 0(94%) Crme 0(97%) 0(100%) 0(69%) 0(92%) Elevation 0(92%) 0(90%) 1(59%) 1(83%) Massage 0(97%) 0(100%) 0(52%) 1(77%) Exercise 0(97%) 0(100%) 1(56%) 1(83%) Bandage 0(97%) 0(90%) 0(86%) 0(94%) Medicine 0(55%) 0(73%) 0(66%) 0(57%) Nothing 0(92%) 0(100%) 0(98%) 0(100%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 278

260 When the answer keys were compared for differences by literacy, no significant differences were found. Table B16. Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Cont rol Groups by Literacy ChiSquare df P< Fishers Exact Two-Tailed p< OMC-RM: Literate vs Not Literate .8095 1 .3683 .6562* OMC-FM: Literate vs Not Literate .0000* 1 1.00 1.00 OMI-RM: Literate vs Not Literate .1193* 1 .7298 1.00 OMI-FM: Literate vs Not Literate .3837* 1 .5356 .6799 indicates which statistic was interpreted for significant differences between the answer keys In addition to the lack of significant difference between answer keys within each group, no significant differences in levels of cultural competency within each group by literacy were present either.

PAGE 279

261 Appendix B Table B17. Outcome Sample: Cultural Comp etency Analysis Results by Group and Literacy N Average Competency LiterateNot Literate Literate Not Literate OMC-RM 15 11 .58 .56 OMI-RM 38 22 .68 .55 OMC-FM 15 11 .39 .54 OMI-FM 38 22 .55 .47 One individual removed from analyses in the outcome control group due to missing data Table B18. Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Literacy df t p< OMC-RM: Literate vs Not Literate 24 .18 .8588 OMC-FM: Literate vs Not Literate 24 -1.53 .1383 OMI-RM: Literate vs Not Literate 30.7 1.66 .1072 OMI-FM: Literate vs Not Literate 30.9 .87 .3919 In the case of literacy, the answer keys showed no significant differences within each group. The same trend held true when di fferences in cultural competency were examined within each group. Though there were no significant differences within each group, each of the two linked samples, baseli ne controls and baseline intervention, showed increases in average cultural co mpetency at outcome. These changes were greatest in the literate portion of the samp le. However, the strongest consensus was

PAGE 280

262 Appendix B present in the not lite rate samples for both control and intervention groups. Thus, literacy does not seem to significantly impact the elem ents highlighted in the cultural models, but individuals who are not literate seem to share a stronger buy-in to the shared model than their literate counterparts. That said, literate participants in the intervention group showed the greatest overall increase in comp etency from baseline to outcome. Wealth Within the wealth comparisons, outc ome intervention groups, both full and reduced, exhibited the highest level of c onsensus across both wealth categories. Specifically, individuals with more wealth in the intervention group showed stronger consensus than their less wealthy counterpar ts. For the outcome c ontrol groups, only the less wealthy individuals for the re duced model showed consensus. Table B19. Outcome Sample: Consensus An alysis Results by Group and Wealth N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Less Wealth More Wealth Less Wealth More Wealth Less Wealth More Wealth Less Wealth More Wealth OMC-RM* 23 3 8.550 .027 1.991 .021 4.295 1.268 OMI-RM 35 25 14.648 12.163 3.608 1.872 4.060 6.499 OMC-FM* 23 3 6.408 .004 2.278 .003 2.813 1.044 OMI-FM 35 25 11.461 8.871 2.748 1.693 4.170 5.241 *One individual removed from analyses in the outcome control group due to missing data

PAGE 281

263 Appendix B The outcome samples all highlighted hygiene and sandal as key pieces of the cultural model. Both interven tion groups identified eleva tion, massage and exercise as well. Specific to the less wealthy intervention individuals, crme and herbal remedies are also selected. Intervention members chose inse ct bite as a cause of lymphatic filariasis. Table B20. Culturally Correct Answer Keys for Intervention and Control Groups by Wealth Outcome Sample OMC-RM OMI-RM N=23 N=3 N=35 N=25 Answer(Weighted %) Less Wealth More Wealth Less Wealth More Wealth Cause Insect Bite 0(96%) 0(100%) 1(66%) 1(79%) Magic 0(96%) 0(100%) 0(94%) 0(99%) Sprain 0(100%) 0(100%) 0(97%) 0(94%) Worms 0(84%) 0(100%) 0(95%) 0(80%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(98%) 1(100%) 1(95%) 1(99%) Sandal 1(100%) 1(52%) 1(97%) 1(99%) Permanganate 0(94%) 0(100%) 0(70%) 0(84%) Crme 0(97%) 0(100%) 1(53%) 0(54%) Elevation 0(85%) 0(52%) 1(92%) 1(96%) Massage 0(97%) 0(100%) 1(68%) 1(63%) Exercise 0(97%) 0(100%) 1(87%) 1(91%) Bandage 0(92%) 0(100%) 0(76%) 0(86%) Medicine 0(56%) 0(52%) 0(64%) 0(58%) Herbal Remedy 0(59%) 0(52%) 1(67%) 0(54%) Pomade 0(70%) 0(52%) 0(60%) 0(59%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 282

264 Appendix B Less wealthy members of th e control group suggest that hygiene and sandals are the appropriate treatment choices for lymphatic filariasis in a variety of scenarios. More wealthy controls selected hygiene, elevat ion, medicine, herbal remedy, and pomade. Intervention members who were less wealthy believe that lymphatic filariasis is caused by insect bites and can be treated with hygien e, sandal, elevation, massage, exercise, and herbal remedies. More wealthy intervention me mbers also believe that insect bites cause lymphatic filariasis. They believe that tr eatment options include: hygiene, sandal, elevation, massage, exercise, crme. Table B21. Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Wealth Outcome Sample OMC-FM OMI-FM N=23 N=3 N=35 N=25 Answer(Weighted %) Less Wealth More Wealth Less Wealth More Wealth Cause Insect Bite 0(93%) 0(100%) 1(68%) 1(81%) Magic 0(93%) 0(100%) 0(97%) 0(100%) Sprain 0(100%) 0(100%) 0(96%) 0(95%) Worms 0(94%) 0(100%) 0(93%) 0(76%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(100%) 1(63%) 1(97%) 1(100%) Sandal 1(100%) 0(100%) 1(98%) 1(100%) Permanganate 0(96%) 0(100%) 0(75%) 0(85%) Crme 0(99%) 0(100%) 0(49%) 0(58%) Elevation 0(92%) 1(63%) 1(94%) 1(96%) Massage 0(99%) 0(63%) 1(71%) 1(59%) Exercise 0(99%) 0(100%) 1(93%) 1(93%) Bandage 0(93%) 0(63%) 0(77%) 0(89%) Medicine 0(66%) 1(63%) 0(66%) 0(61%) Herbal Remedy 0(80%) 1(63%) 1(65%) 0(53%) Pomade 0(84%) 1(100%) 0(59%) 0(60%)

PAGE 283

265 Appendix B Table B21. (continued) OMC-FM OMI-FM N=23 N=3 N=35 N=25 Answer (Weighted %) Less Wealth More Wealth Less Wealth More Wealth Treatment (OC Only) Hygiene 1(91%) 0(100%) 1(98%) 1(99%) What kinds of care can help your gwopye? Sandal 1(85%) 0(100%) 1(76%) 1(83%) Permanganate 0(96%) 0(100%) 0(71%) 0(63%) Crme 0(93%) 0(100%) 0(58%) 1(55%) Elevation 0(86%) 1(63%) 1(90%) 1(96%) Massage 0(96%) 0(100%) 1(70%) 1(70%) Exercise 0(96%) 0(100%) 1(93%) 1(91%) Bandage 0(96%) 0(100%) 0(94%) 0(78%) Medicine 0(74%) 1(63%) 0(68%) 0(67%) Nothing 0(98%) 0(100%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(71%) 0(100%) 1(87%) 1(91%) What can you do to prevent acute attacks? Sandal 1(71%) 0(100%) 1(70%) 1(87%) Permanganate 0(96%) 0(100%) 0(93%) 0(82%) Crme 0(96%) 0(100%) 0(78%) 0(79%) Elevation 0(90%) 1(63%) 1(72%) 1(72%) Massage 0(100%) 0(100%) 1(54%) 1(58%) Exercise 0(100%) 0(100%) 1(75%) 1(73%) Bandage 0(100%) 0(100%) 0(98%) 0(89%) Medicine 0(87%) 1(63%) 0(76%) 0(61%) Nothing 0(94%) 0(100%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(77%) 0(100%) 1(78%) 1(80%) What can be done to provide relief during an acute attack? Sandal 1(73%) 0(100%) 1(58%) 1(61%) Permanganate 0(100%) 0(100%) 0(89%) 0(64%) Crme 0(99%) 0(100%) 0(82%) 0(69%) Elevation 0(93%) 1(63%) 1(63%) 1(71%) Massage 0(99%) 0(100%) 1(58%) 1(57%) Exercise 0(99%) 0(100%) 1(67%) 1(62%) Bandage 0(93%) 0(100%) 0(95%) 0(81%) Medicine 0(65%) 1(63%) 0(67%) 0(60%) Nothing 0(96%) 0(100%) 0(98%) 0(100%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 284

266 Appendix B After the answer keys were reviewed for the elements included in each cultural model, the answer keys were compared be tween more and less wealthy individuals within each group. These anal yses indicate that there ar e no significant differences between answer keys within each group. Table B22. Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Wealth ChiSquare df p< Fishers Exact Two-Tailed p< OMC-RM: Less Wealth vs More Wealth .0000 1 1.00 1.00* OMC-FM: Less Wealth vs More Wealth .5937* 1 .4410 .6083 OMI-RM: Less Wealth vs More Wealth .4857* 1 .4858 .7283 OMI-FM: Less Wealth vs More Wealth .0000* 1 1.00 1.00 indicates which statistic was interpreted for significant differences between the answer keys Further analyses were conduc ted to assess rates of cultu ral competency in each group and whether or not signi ficant differences in competency exist between less and more wealthy individuals in each sample. The wealthier portion of the intervention sample showed the highest rates of cultural co mpetency. Significant differences did exist

PAGE 285

267 Appendix B in the cultural competency levels of the le ss wealth vs more wealthy members of both control models. However, the sample of thes e groups is comprised of only three people; cultural consensus analysis can theoretically be conducted with a sample of three but even the lowest recommended sample in the power analysis table previously discussed suggests a minimum sample size of four people. Thus, the significant differences suggested in this an alysis are suspect. Table B23. Outcome Sample: Cultural Comp etency Analysis Results by Group and Wealth N Average Competency Less Wealth More Wealth Less Wealth More Wealth OMC-RM 23 3 .59 .07 OMI-RM 35 25 .59 .75 OMC-FM 23 3 .46 .0007 OMI-FM 35 25 .48 .55 One individual removed from analyses in the outcome control group due to missing data

PAGE 286

268 Appendix B Table B24. Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Wealth Df t p< OMC-RM: Less Wealth vs More Wealth 24 5.41 <.0001 OMC-FM: Less Wealth vs More Wealth 21.9 7.71 <.0001 OMI-RM: Less Wealth vs More Wealth 58 -1.21 .2327 OMI-FM: Less Wealth vs More Wealth 58 -.90 .3710 In the outcome intervention group both stronger consensus and competence was present for wealthier individua ls. Also, the change in competence level from baseline to outcome in the intervention group was greater for this group. In contrast, less wealthy controls indicated greater consensus and gr eater competence in th e control groups. That said, the sample size was only 3 individuals in the wealthier segmen t of the control group sample; thus, trends observed in these samp les may be suspect. Considering the evidence in the intervention samples, then, it looks as though the in tervention had greater impact on wealthier participants. Age When looking at cultural models present in the sample by age, the median age, 46 years old, was used to divide the sample into younger and older groupings. With this grouping, younger outcome intervention members exhibit consensus wh ere their outcome

PAGE 287

269 Appendix B control counterparts do not. Older members s how consensus in all four outcome models, but the younger group indicates the strongest consensus. Table B25. Outcome Sample: Consensus Analysis Results by Group and Age N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Younger Older Younger Older Younger Older Younger Older OMC-RM* 13 13 4.382 4.572 1.887 1.043 2.322 4.385 OMI-RM 28 32 13.67013.2572.637 2.888 5.183 4.591 OMC-FM* 13 13 3.187 3.692 1.430 .775 2.229 4.766 OMI-FM 28 32 10.30810.2081.911 2.527 5.395 4.039 *One individual removed from analyses in the outcome control group due to missing data For the reduced models at outcome, the an swer keys suggest different items are important within the cultural model. Both younger and older control groups selected hygiene and sandals as the only treatment choices. Alternately, intervention groups highlighted insect bites as the cause of lym phatic filariasis and hygi ene, sandal, elevation, massage, exercise, and herbal remedies for treatment. Younger members of the intervention sample also added crme as a treatment.

PAGE 288

270 Appendix B Table B26. Culturally Correct Answer Keys for Intervention and Control Groups by Age Outcome Sample OMC-RM OMI-RM N=13 N=13 N=28 N=32 Answer(Weighted %) Younger Older Younger Older Cause Insect Bite 0(92%) 0(100%) 1(85%) 1(59%) Magic 0(100%) 0(93%) 0(97%) 0(97%) Sprain 0(100%) 0(100%) 0(95%) 0(97%) Worms 0(80%) 0(91%) 0(90%) 0(87%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(96%) 1(100%) 1(96%) 1(98%) Sandal 1(100%) 1(94%) 1(99%) 1(96%) Permanganate 0(96%) 0(93%) 0(86%) 0(66%) Crme 0(94%) 0(100%) 1(51%) 0(50%) Elevation 0(80%) 0(87%) 1(96%) 1(91%) Massage 0(94%) 0(100%) 1(74%) 1(58%) Exercise 0(94%) 0(100%) 1(88%) 1(90%) Bandage 0(94%) 0(92%) 0(79%) 0(83%) Medicine 0(55%) 0(57%) 0(66%) 0(56%) Herbal Remedy 0(60%) 0(58%) 1(56%) 1(61%) Pomade 0(60%) 0(78%) 0(67%) 0(52%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data Models were also run at outcome with additional treatment items included only in the outcome data. In these full models, both younger and older control groups looked to hygiene and sandals to treat lymphatic filarias is across multiple treatment scenarios. In contrast, insect bites were seen as the cause of LF in both younger and older intervention samples. Intervention groups also feel that hygiene, sandal, elevation, massage, exercise,

PAGE 289

271 Appendix B and herbal remedy were appropriate in di fferent circumstances. Younger and older individuals also identified cr me and pomade, respectively. Table B27. Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items onl y Present in the Outcome Sample by Age Outcome Sample OMC-FM OMI-FM N=13 N=13 N=28 N=32 Answer(Weighted %) YoungerOlder Younger Older Cause Insect Bite 0(87%) 0(100%) 1(85%) 1(63%) Magic 0(100%)0(88%) 0(98%) 0(98%) Sprain 0(100%)0(100%) 0(94%) 0(96%) Worms 0(92%) 0(97%) 0(89%) 0(83%) Chill 0(100%)0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%)0(100%) 0(100%) 0(100%) Treatment Hygiene 1(99%) 1(100%) 1(98%) 1(99%) Sandal 1(100%)1(99%) 1(100%) 1(98%) Permanganate 0(99%) 0(94%) 0(86%) 0(72%) Crme 0(98%) 0(100%) 1(53%) 0(57%) Elevation 0(92%) 0(93%) 1(98%) 1(93%) Massage 0(98%) 0(100%) 1(80%) 1(54%) Exercise 0(98%) 0(100%) 1(93%) 1(93%) Bandage 0(98%) 0(89%) 0(81%) 0(84%) Medicine 0(64%) 0(73%) 0(71%) 0(57%) Herbal Remedy 0(80%) 0(84%) 1(54%) 1(60%) Pomade 0(79%) 0(90%) 0(68%) 1(49%) Treatment (OC Only) Hygiene 1(95%) 1(90%) 1(99%) 1(98%)

PAGE 290

272 Appendix B Table B27. (continued) OMC-FM OMI-FM N=13 N=13 N=28 N=32 Answer (Weighted %) Younger Older Younger Older What kinds of care can help your gwopye? Sandal 1(81%) 1(90%) 1(83%) 1(76%) Permanganate 0(100%) 0(94%) 0(61%) 0(73%) Crme 0(92%) 0(94%) 1(52%) 0(55%) Elevation 0(92%) 0(82%) 1(95%) 1(91%) Massage 0(92%) 0(100%) 1(74%) 1(67%) Exercise 0(92%) 0(100%) 1(93%) 1(90%) Bandage 0(92%) 0(100%) 0(91%) 0(83%) Medicine 0(79%) 0(72%) 0(72%) 0(63%) Nothing 0(100%) 0(97%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(66%) 1(80%) 1(97%) 1(82%) What can you do to prevent acute attacks? Sandal 1(66%) 1(80%) 1(86%) 1(70%) Permanganate 0(100%) 0(94%) 0(84%) 0(92%) Crme 0(100%) 0(94%) 0(64%) 0(92%) Elevation 0(100%) 0(82%) 1(81%) 1(63%) Massage 0(100%) 0(100%) 1(62%) 1(51%) Exercise 0(100%) 0(100%) 1(76%) 1(73%) Bandage 0(100%) 0(100%) 0(93%) 0(95%) Medicine 0(87%) 0(88%) 0(69%) 0(69%) Nothing 0(93%) 0(97%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(73%) 1(84%) 1(86%) 1(74%) What can be done to provide relief during an acute attack? Sandal 1(66%) 1(84%) 1(60%) 1(60%) Permanganate 0(100%) 0(100%) 0(67%) 0(88%) Crme 0(98%) 0(100%) 0(61%) 0(90%) Elevation 0(97%) 0(88%) 1(76%) 1(59%) Massage 0(98%) 0(100%) 1(70%) 0(54%) Exercise 0(98%) 0(100%) 1(75%) 1(56%) Bandage 0(98%) 0(89%) 0(87%) 0(91%) Medicine 0(63%) 0(73%) 0(68%) 0(60%) Nothing 0(93%) 0(100%) 0(100%) 0(98%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 291

273 Appendix B Though there are different elements pres ent in the cultural models discussed above, each group exhibited no significant differences by age. Table B28. Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Age ChiSquare df p< Fishers Exact Two-Tailed p< OMC-RM: Younger vs Older .0000 1 1.00 1.00* OMC-FM: Younger vs Older .0000* 1 1.00 1.00 OMI-RM: Younger vs Older .1193* 1 .7298 1.00 OMI-FM: Younger vs Older .1703* 1 .6799 .8367 indicates which statistic was interpreted for significant differences between the answer keys Competency scores were also assessed for significant differences by age within each group. No significant di fferences were produced in these analyses. Table B29. Outcome Sample: Cultural Compet ency Analysis Results by Group and Age N Average Competency YoungerOlder Younger Older OMC-RM* 13 13 .55 .55 OMI-RM 28 32 .65 .59 OMC-FM* 13 13 .42 .45 OMI-FM 28 32 .53 .49 One individual removed from analyses in the outcome control group due to missing data

PAGE 292

274 Appendix B Table B30. Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Age df t p< OMC-RM: Younger vs Older 24 .07 .9443 OMC-FM: Younger vs Older 24 .25 .8052 OMI-RM: Younger vs Older 58 -.91 .3644 OMI-FM: Younger vs Older 58 -.53 .5999 The older segment of the sample showed c onsensus in all four groups at outcome. The younger portion, by comparison, either had no consensus or the strongest consensus of all outcome groups (outcome intervention model). Coupled with the fact that the younger sample in the intervention group s howed a greater change in cultural competency from baseline to outcome than the older people, it seems that, while both groups benefited from the support group pr ogram, younger individuals were both more receptive to the material presented and more lik ely to quickly integrate new material into their cultural model of lymphatic filariasis. Stage of Disease All groups showed consensus at outcome al beit lesser in the cont rol groups. In the reduced outcome model, consensus is highest in the portion of the sample experiencing more severe disease.

PAGE 293

275 Appendix B Table B31. Outcome Sample: Consensus An alysis Results by Group and Stage of Disease N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Less Severe More Severe Less Severe More Severe Less Severe More Severe Less Severe More Severe OMC-RM* 9 17 1.955 6.071 .611 1.789 3.198 3.393 OMI-RM 36 24 15.65711.4383.458 1.831 4.527 6.247 OMC-FM* 9 17 1.558 4.483 .338 1.423 4.615 3.151 OMI-FM 36 24 11.8128.626 2.152 2.141 5.489 4.029 *One individual removed from analyses in the outcome control group due to missing data In the outcome samples, hygiene and sa ndals are consistently identified as important elements of the cultural model ac ross control and intervention groups as well as by age. Individuals in the intervention group believe that insect bites cause lymphatic filariasis and that elevation, massage, exerci se, and herbal remedies aid in treating LF. Also, the younger and older intervention gr oups selected crme and pomade as well, respectively.

PAGE 294

276 Appendix B Table B32. Culturally Correct Answer Keys for Intervention and Control Groups by Stage Outcome Sample OMC-RM OMI-RM N=9 N=17 N=36 N=24 Answer(Weighted %) Less Severe More Severe Less Severe More Severe Cause Insect Bite 0(88%) 0(100%) 1(79%) 1(63%) Magic 0(100%) 0(95%) 0(98%) 0(94%) Sprain 0(100%) 0(100%) 0(93%) 0(100%) Worms 0(93%) 0(81%) 0(92%) 0(84%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(100%) 1(96%) 1(95%) 1(100%) Sandal 1(100%) 1(95%) 1(96%) 1(100%) Permanganate 0(100%) 0(90%) 0(68%) 0(88%) Crme 0(93%) 0(100%) 1(54%) 0(53%) Elevation 0(93%) 0(77%) 1(92%) 1(96%) Massage 0(93%) 0(100%) 1(67%) 1(65%) Exercise 0(93%) 0(100%) 1(92%) 1(85%) Bandage 0(93%) 0(94%) 0(79%) 0(82%) Medicine 0(66%) 0(51%) 0(73%) 1(56%) Herbal Remedy 0(65%) 0(55%) 1(53%) 1(64%) Pomade 0(80%) 0(63%) 0(68%) 1(52%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data In the outcome models including the additi onal outcome only items, hygiene and sandals remain key elements across all samples, a nd the intervention group continues to clearly identify insect bites as a cau se of lymphatic filariasis. In addition to hygiene and sandals, intervention samples also identified the following as possible treatment options

PAGE 295

277 Appendix B in varying circumstances: el evation, massage, exercise, me dicine, herbal remedies, pomade, crme. Table B33. Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Stage of Disease Outcome Sample OMC-FM OMI-FM N=9 N=17 N=36 N=24 Answer(Weighted %) Less Severe Most Severe Less Severe Most Severe Cause Insect Bite 0(82%) 0(100%) 1(79%) 1(67%) Magic 0(100%)0(91%) 0(99%) 0(98%) Sprain 0(100%)0(100%) 0(92%) 0(100%) Worms 0(99%) 0(93%) 0(90%) 0(80%) Chill 0(100%)0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%)0(100%) 0(100%) 0(100%) Treatment Hygiene 1(100%)1(99%) 1(97%) 1(100%) Sandal 1(100%)1(99%) 1(98%) 1(100%) Permanganate 0(100%)0(94%) 0(72%) 0(89%) Crme 0(99%) 0(100%) 0(51%) 0(56%) Elevation 0(99%) 0(89%) 1(92%) 1(99%) Massage 0(99%) 0(100%) 1(69%) 1(62%) Exercise 0(99%) 0(100%) 1(93%) 1(93%) Bandage 0(99%) 0(91%) 0(82%) 0(82%) Medicine 0(72%) 0(67%) 0(74%) 1(51%) Herbal Remedy 0(85%) 0(80%) 1(52%) 1(63%) Pomade 0(94%) 0(80%) 0(68%) 1(54%) Treatment (OC Only) Hygiene 1(95%) 1(89%) 1(99%) 1(97%)

PAGE 296

278 Appendix B Table B33. (continued) OMC-FM OMI-FM N=9 N=17 N=36 N=24 Answer (Weighted %) Less Severe Most Severe Less Severe Most Severe What kinds of care can help your gwopye? Sandal 1(86%) 1(85%) 1(87%) 1(69%) Permanganate 0(100%) 0(95%) 0(61%) 0(77%) Crme 0(99%) 0(91%) 0(55%) 1(53%) Elevation 0(99%) 0(80%) 1(88%) 1(99%) Massage 0(99%) 0(96%) 1(68%) 1(73%) Exercise 0(99%) 0(96%) 1(93%) 1(90%) Bandage 0(99%) 0(96%) 0(84%) 0(90%) Medicine 0(81%) 0(72%) 0(66%) 0(69%) Nothing 0(100%) 0(97%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(77%) 1(72%) 1(92%) 1(84%) What can you do to prevent acute attacks? Sandal 1(77%) 1(72%) 1(80%) 1(75%) Permanganate 0(100%) 0(95%) 0(83%) 0(95%) Crme 0(100%) 0(95%) 0(80%) 0(77%) Elevation 0(100%) 0(85%) 1(65%) 1(82%) Massage 0(100%) 0(100%) 1(57%) 1(55%) Exercise 0(100%) 0(100%) 1(69%) 1(81%) Bandage 0(100%) 0(100%) 0(93%) 0(95%) Medicine 0(82%) 0(90%) 0(73%) 0(63%) Nothing 0(91%) 0(97%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(77%) 1(78%) 1(81%) 1(77%) What can be done to provide relief during an acute attack? Sandal 1(77%) 1(75%) 1(60%) 1(60%) Permanganate 0(100%) 0(100%) 0(74%) 0(82%) Crme 0(99%) 0(100%) 0(81%) 0(70%) Elevation 0(99%) 0(89%) 1(60%) 1(78%) Massage 0(99%) 0(100%) 1(55%) 1(61%) Exercise 0(99%) 0(100%) 1(60%) 1(73%) Bandage 0(99%) 0(91%) 0(91%) 0(85%) Medicine 0(72%) 0(66%) 0(67%) 0(58%) Nothing 0(91%) 0(100%) 0(98%) 0(100%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control group due to missing data

PAGE 297

279 Appendix B Answer key comparisons within each group by stage revealed no significant differences in the elements comprising the cultural model by disease stage. Table B34. Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Groups by Stage of Disease ChiSquare df p< Fishers Exact Two-Tailed p< OMC-RM: Less Severe vs More Severe .0000 1 1.00 1.00* OMC-FM: Less Severe vs More Severe .0000* 1 1.00 1.00 OMI-RM: Less Severe vs More Severe .1176* 1 .7316 1.00 OMI-FM: Less Severe vs More Severe .3830* 1 .5360 .6802 indicates which statistic was interpreted for significant differences between the answer keys The level of cultural competency present in each sample by disease stage was also assessed. In the outcome samples, cultural co mpetency was greater in the portion of the sample experiencing more severe symptoms. The differences in cultural competency within each group by stage were tested and no significant differences were present.

PAGE 298

280 Appendix B Table B35. Outcome Sample: Cultural Compet ency Analysis Results by Group and Stage of Disease N Average Competency Less Severe More Severe Less Severe More Severe OMC-RM* 9 17 .42 .57 OMI-RM 36 24 .60 .64 OMC-FM* 9 17 .33 .44 OMI-FM 36 24 .50 .52 One individual removed from analyses in the outcome control group due to missing data Table B36. Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Stage of Disease df t p< OMC-RM: Less Severe vs More Severe 24 -1.93 .0658 OMC-FM: Less Severe vs More Severe 24 -.95 .3537 OMI-RM: Less Severe vs More Severe 58 -.57 .5709 OMI-FM: Less Severe vs More Severe 58 -.28 .7768 In the outcome intervention model assessi ng general treatment options, consensus existed for individuals more severely impact ed by LF. However, when specific treatment scenarios were introduced, the intervention m odel with additional items present in the outcome survey only, those with less seve re disease indicated greater consensus. Additionally, there were changes in the cultu ral competency levels of both the baseline and outcome intervention groups for both c onditions. Thus, it seems that both groups, severely and less severely afflicted, benef it from the support group program while people

PAGE 299

281 Appendix B with less severe disease are better able to ag ree on what treatment options are appropriate in more specific scenarios. Number of Acute Attacks Consensus is present in all outcome groups, and is greater in the outcome intervention groups. Within the intervention gr oups, people with fewer attacks exhibited slightly more consensus than the po rtion experience more acute attacks. Table B37. Outcome Sample: Consensus Anal ysis Results by Group and Number of Attacks N First Eigenvalue Second Eigenvalue Eigenvalue Ratio Fewer Attacks More Attacks Fewer Attacks More Attacks Fewer Attacks More Attacks Fewer Attacks More Attacks OMC-RM* 18 8 5.951 3.188 1.868 .418 3.187 7.620 OMI-RM 31 28 14.60812.7412.955 2.641 4.944 4.825 OMC-FM* 18 8 4.893 1.482 1.246 .429 3.927 3.452 OMI-FM* 31 28 10.51810.1382.178 2.153 4.828 4.710 *One individual removed from analyses in th e outcome control group due to missing data Answer keys in the outcome sample sugge st that insect bites cause lymphatic filariasis in the intervention groups. Hygiene and sandals conti nue to be salient across all outcome models. Elevation, massage, crme, ex ercise, and herbal remedies are also

PAGE 300

282 Appendix B present in intervention group cultural mode ls. Pharmaceutical medicines are mentioned as well in the outcome control gr oup experiencing more attacks. Table B38. Culturally Correct Answer Keys for Intervention and Control Groups by Number of Attacks Outcome Sample OMC-RM OMI-RM N=18 N=8 N=31 N=28 Answer(Weighted %) Fewer Attacks More Attacks Fewer Attacks More Attacks Cause Insect Bite 0(100%) 0(87%) 1(68%) 1(77%) Magic 0(95%) 0(100%) 0(96%) 0(97%) Sprain 0(100%) 0(100%) 0(97%) 0(94%) Worms 0(89%) 0(77%) 0(88%) 0(90%) Chill 0(100%) 0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%) 0(100%) 0(100%) 0(100%) Treatment Hygiene 1(97%) 1(100%) 1(99%) 1(96%) Sandal 1(95%) 1(100%) 1(99%) 1(97%) Permanganate 0(90%) 0(100%) 0(60%) 0(94%) Crme 0(100%) 0(92%) 1(60%) 0(60%) Elevation 0(77%) 0(92%) 1(94%) 1(95%) Massage 0(100%) 0(92%) 1(79%) 1(52%) Exercise 0(100%) 0(92%) 1(95%) 1(80%) Bandage 0(94%) 0(92%) 0(81%) 0(79%) Medicine 0(68%) 1(65%) 0(56%) 0(67%) Herbal Remedy 0(54%) 0(62%) 0(56%) 1(73%) Pomade 0(62%) 0(78%) 0(55%) 0(64%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control and intervention group due to missing data

PAGE 301

283 Appendix B The answer keys for the outcome samp les with additional items were also reviewed. Both intervention samples identified insect bite as the cause of lymphatic filariasis. Control groups believe that hygiene, sandals, and pharmaceutical medicines are the appropriate treatment choices in a number of different circumstances relevant to LF. Intervention groups, however, identified hygi ene, sandals, elevation, massage, exercise, and herbal remedies in various circumstances.

PAGE 302

284 Appendix B Table B39. Culturally Correct Answer Keys for Intervention and Control Groups that Include Additional Cultural Model Items only Present in the Outcome Sample by Number of Attacks Outcome Sample OMC-FM OMI-FM N=18 N=8 N=31 N=28 Answer(Weighted %) Fewer Attacks More Attacks Fewer Attacks More Attacks Cause Insect Bite 0(100%)0(81%) 1(72%) 1(76%) Magic 0(91%) 0(100%) 0(97%) 0(100%) Sprain 0(100%)0(100%) 0(97%) 0(94%) Worms 0(97%) 0(89%) 0(85%) 0(86%) Chill 0(100%)0(100%) 0(100%) 0(100%) Vitamin Deficiency 0(100%)0(100%) 0(100%) 0(100%) Treatment Hygiene 1(100%)1(100%) 1(99%) 1(98%) Sandal 1(100%)1(100%) 1(100%) 1(99%) Permanganate 0(95%) 0(100%) 0(65%) 0(93%) Crme 0(100%)0(97%) 1(54%) 0(61%) Elevation 0(91%) 0(97%) 1(94%) 1(97%) Massage 0(100%)0(97%) 1(77%) 1(54%) Exercise 0(100%)0(97%) 1(97%) 1(88%) Bandage 0(92%) 0(97%) 0(84%) 0(79%) Medicine 0(79%) 1(57%) 0(56%) 0(71%) Herbal Remedy 0(84%) 0(76%) 0(55%) 1(68%) Pomade 0(86%) 0(83%) 0(54%) 0(65%) Treatment (OC Only) Hygiene 1(92%) 1(92%) 1(99%) 1(98%) What kinds of care can help your gwopye? Sandal 1(89%) 1(80%) 1(73%) 1(87%) Permanganate 0(95%) 0(100%) 0(63%) 0(71%) Crme 0(92%) 0(97%) 0(51%) 0(53%) Elevation 0(83%) 0(97%) 1(92%) 1(94%) Massage 0(97%) 0(97%) 1(84%) 1(54%) Exercise 0(97%) 0(97%) 1(96%) 1(87%) Bandage 0(97%) 0(97%) 0(75%) 0(99%) Medicine 0(76%) 0(78%) 0(58%) 0(77%) Nothing 0(100%)0(92%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(82%) 1(53%) 1(87%) 1(91%)

PAGE 303

285 OMC-FM OMI-FM N=18 N=8 N=31 N=28 Answer (Weighted %) Fewer Attacks More Attacks Fewer Attacks More Attacks What can you do to prevent acute attacks? Sandal 1(82%) 1(53%) 1(72%) 1(84%) Permanganate 0(95%) 0(100%) 0(81%) 0(95%) Crme 0(95%) 0(100%) 0(76%) 0(82%) Elevation 0(86%) 0(100%) 1(66%) 1(80%) Massage 0(100%) 0(100%) 1(69%) 0(60%) Exercise 0(100%) 0(100%) 1(73%) 1(76%) Bandage 0(100%) 0(100%) 0(89%) 0(99%) Medicine 0(91%) 0(81%) 0(66%) 0(71%) Nothing 0(98%) 0(87%) 0(100%) 0(100%) Treatment (OC Only) Hygiene 1(87%) 1(56%) 1(85%) 1(73%) What can be done to provide relief during an acute attack? Sandal 1(84%) 1(53%) 1(53%) 1(66%) Permanganate 0(100%) 0(100%) 0(71%) 0(85%) Crme 0(100%) 0(97%) 0(74%) 0(79%) Elevation 0(91%) 0(97%) 1(63%) 1(73%) Massage 0(100%) 0(97%) 1(68%) 0(55%) Exercise 0(100%) 0(97%) 1(68%) 1(63%) Bandage 0(92%) 0(97%) 0(83%) 0(95%) Medicine 0(79%) 1(57%) 0(51%) 0(77%) Nothing 0(100%) 0(87%) 0(98%) 0(100%) *0=No response and indicates that this variable is not part of the cultural model when the threshold for consensus is met ** 1 = Yes response and indicates that this variable is part of the cultural model when the threshold for consensus is met *** One individual removed from analyses in the outcome control and intervention group due to missing data

PAGE 304

286 Appendix B Table B40. Outcome Sample: Comparisons of Culturally Correct Answer Keys for Intervention and Control Gr oups by Number of Attacks ChiSquare df p< Fishers Exact TwoTailed p< OMC-RM: Fewer Attacks vs More Attacks .2345 1 .6282 1.00* OMC-FM: Fewer Attacks vs More Attacks .2749* 1 .6001 .7939 OMI-RM: Fewer Attacks vs More Attacks .0000* 1 1.00 1.00 OMI-FM: Fewer Attacks vs More Attacks .1722* 1 .6782 .8358 indicates which statistic was interpreted for significant differences between the answer keys In addition to analyzing the answer keys for differences, average levels of cultural competency were also assessed. In all the outcome samples, no significant differences were found between those with fewer and more attacks within each group. Table B41. Outcome Sample: Cultural Comp etency Analysis Results by Group and Number of Attacks N Average Competency Fewer Attacks More Attacks Fewer Attacks More Attacks OMC-RM 18 8 .53 .62 OMI-RM 31 28 .65 .62 OMC-FM 18 8 .42 .40 OMI-FM 31 28 .53 .52 One individual removed from analyses in the outcome control a nd intervention group due to missing data

PAGE 305

287 Appendix B Table B42. Outcome Sample: Comparisons of Cultural Competency for Intervention and Control Groups by Number of Attacks df t p< OMC-RM: Fewer Attacks vs More Attacks 24 -1.06 .2999 OMC-FM: Fewer Attacks vs More Attacks 24 .15 .8810 OMI-RM: Fewer Attacks vs More Attacks 57 .44 .6603 OMI-FM: Fewer Attacks vs More Attacks 57 .22 .8302 For the outcome intervention group, pe ople experiencing fewer acute attacks showed higher eigenvalue ratios than those with more attacks. The highest cultural consensus for individuals having fewer attack s was found in the outcome control sample, and the highest overall consensus was found in the outcome control sample for those with more attacks. Interestingly, while the hi ghest cultural competence was found in the outcome intervention group for those having fewe r acute attacks, the largest changes in cultural competence levels from baseline to outcome presented in the portion of the sample having more acute attacks. It seems that people experiencing more attacks were able to overcome the deficit in competency they expressed at baseline and quickly integrate support group material s into their knowledge base such that competency and consensus levels were approximately the sa me between the two segments in outcome intervention samples.

PAGE 306

288 Appendix C

PAGE 307

289 Appendix C General demographic results from the cu ltural consonance analyses are presented in the body of Chapter 4. Specific results of the statistical analyses and comparisons summarized previously are captured in the tables below. Demographic dimensions reviewed here include: marital status, religion, literacy, wealth, age, stage of disease, and number of acute attacks. Marital Status Table C1. Cultural Consonance Resu lts by Group and Marital Status N Average Consonance Single Not Single Single Not Single BMC 4 23 .56 .83 BMI 27 33 .56 .76 B-All 82 159 .54 .67 OMC-RM* 4 22 .69 .86 OMI-RM* 27 33 .53 .66 One individual removed from analyses in the outcome control group due to missing data Along the demographic dimension marital st atus, rates of consonance were higher for not single individuals than for those who were single. Average rates were the highest in the control groups.

PAGE 308

290 Appendix C Table C2. Comparisons of Cu ltural Competency by Interven tion and Control Groups and Marital Status df t p< BMC: Single vs Not Single 25 1.28 .2116 BMI: Single vs Not Single 58 1.90 .0623 B-All: Single vs Not Single 239 3.30 .0011 OMI-RM: Single vs Not Single 58 2.41 .0193 OMC-RM: Single vs Not Single 24 1.09 .2884 Independent samples t-tests suggest that significant differences exist between individuals by marital status in the full baseline sample as well as the outcome intervention sample. Taken as a whole, it seems that people w ho participate in the support group program and are not single are more likel y to engage in more treatment behaviors than their single counterparts. Religion Table C3. Cultural Consonance Analysis Results by Group and Religion N Average Consonance CatholicProtestantCatholic Protestant BMC 15 6 .87 .83 BMI 27 28 .51 .77 B-All 118 96 .64 .60 OMC-RM* 15 6 .90 .92 OMI-RM* 27 28 .66 .58 One individual removed from analyses in the outcome control group due to missing data

PAGE 309

291 Appendix C For the religious dimension, control populations had the highest rates of consonance. In the intervention sample, Pr otestants showed higher consonance than Catholics at baseline and reve rsed the trend in the outcome sample. The controls samples showed the inverse of the intervention samp le regarding average rates of consonance; Catholics had higher consonance at baseline and Protestants were slightly higher at outcome. Table C4. Comparisons of Cultural Consona nce for Intervention and Control Groups by Religion df t p< BMC: Catholic vs Protestant 19 .22 .8296 BMI: Catholic vs Protestant 46.3 -2.89 .0059 B-All: Catholic vs Protestant 212 .94 .3479 OMI-RM: Catholic vs Protestant 53 1.45 .1533 OMC-RM: Catholic vs Protestant 19 -.13 .8968 The only group to display significant diffe rences in consonance is the baseline intervention sample. At baseline, Protesta nts exhibited significantly higher rates of consonance, but, after participation in the support group, Catholics increased in consonance greatly; this shift eliminated any significant diffe rences in the two groups at outcome. The data allow that Catholics effec tively incorporated self -care practices into their treatment regimen at a rate much im proved than was evidenced at baseline.

PAGE 310

292 Appendix C Literacy Table C5. Cultural Consonance Analys is Results by Group and Literacy N Average Competency LiterateNot Literate Literate Not Literate BMC 15 12 .83 .75 BMI 38 22 .61 .73 B-All 137 104 .65 .60 OMC-RM* 15 11 .65 .91 OMI-RM 38 22 .68 .48 One individual removed from analyses in the outcome control group due to missing data Literate individuals in the intervention sample increased their consonance levels at outcome while the non-literate portion act ually decreased in c onsonance from baseline to outcome. In this sample, literate indivi duals had lower consona nce at baseline than non-literate participants; at outcome this finding reversed itself. The control sample yielded that non-literate indi viduals had lower consonance than their counterparts at baseline and higher at outcome.

PAGE 311

293 Appendix C Table C6. Comparisons of Cultural Consona nce for Intervention and Control Groups by Literacy df t p< BMC: Literate vs Not Literate 25 .68 .5048 BMI: Literate vs Not Literate 30.6 -1.21 .2371 B-All: Literate vs Not Literate 239 1.39 .1655 OMI-RM: Literate vs Not Literate 58 3.52 .0008 OMC-RM: Literate vs Not Literate 24 -2.41 .0239 When independent samples t-tests were run, the only significant difference highlighted was found in the outcome interven tion group. In this case, it looks as though, at outcome, non-literate individuals were not as receptive to adding additional treatment behaviors to their regimen as literate members of the sample population. Wealth Table C7. Cultural Consonance by Group and Wealth N Average Consonance Less Wealth More Wealth Less Wealth More Wealth BMC 24 3 .85 .58 BMI 35 25 .73 .54 B-All 172 69 .62 .55 OMC-RM* 23 3 .93 .5 OMI-RM 35 25 .55 .69 One individual removed from analyses in the outcome control group due to missing data When wealth was examined, consonance was higher for less wealthy individuals across all groups except for th e outcome intervention sample Less wealthy participants in this group decreased in consonance while more wealthy people incr eased after being in the support group program.

PAGE 312

294 Appendix C Table C8. Comparisons of Cultural Consona nce for Intervention and Control Groups by Wealth df t p< BMC: Less Wealth vs More Wealth 25 1.30 .2064 BMI: Less Wealth vs More Wealth 56.6 2.13 .0374 B-All: Less Wealth vs More Wealth 239 1.78 .0767 OMI-RM: Less Wealth vs More Wealth 58 -2.61 .0114 OMC-RM: Less Wealth vs More Wealth 2.11 1.49 .2694 Through t-test comparisons, significant di fferences in consonance were revealed in both baseline and outcome intervention groups. This observation suggests that less wealthy people were less amenable to enacti ng the self-care practices, other than the ones present at baseline, they endor sed in the cultural model. In contrast, wealthier people were able to incorporate newl y introduced self-care practices into both their belief and behavioral systems. Age Table C9. Cultural Consonance Analysis Results by Group and Age N Average Consonance YoungerOlder Younger Older BMC 14 13 .71 .79 BMI 28 32 .70 .59 B-All 121 120 .64 .61 OMC-RM* 13 13 .88 .88 OMI-RM 28 32 .58 .57 One individual removed from analyses in the outcome control group due to missing data

PAGE 313

295 A review of the age dimension in this sample population revealed that younger individuals at a higher and lower consonance rate at baseline for the intervention and control samples, respectively. However, at outcome consonance rates were virtually the same regardless of designation, younger or older. Table C10. Comparisons of Cultural Consonance for Intervention and Control Groups by Age df t p< BMC: Younger vs Older 18.6 .58 .5690 BMI: Younger vs Older 43.1 -1.08 .2856 B-All: Younger vs Older 239 -.66 .5097 OMI-RM: Younger vs Older 58 -.18 .8601 OMC-RM: Younger vs Older 24 .00 1.00 After looking at the consonance rates by age, t-tests comparisons support the finding that there are no significant differences between age categories within any of the samples tested. The data indicate that olde r individuals in the intervention sample increased in their willingness to enact trea tment practices indicated in the CM after support group participation. Younger individual s in this sample, however, endorsed the beliefs present in the CM, but were not as quick to add additional behaviors to their treatment regimen as evidenced by their reduced rate of consonance at outcome.

PAGE 314

296 Appendix C Stage of Disease Table C11. Cultural Consonance Analysis Results by Group and Stage of Disease N Average Consonance Less Severe More Severe Less Severe More Severe BMC 9 18 .64 .86 BMI 36 24 .65 .58 B-All 125 116 .65 .60 OMC-RM* 9 17 .89 .88 OMI-RM 36 24 .55 .56 One individual removed from analyses in the outcome control group due to missing data Along the dimension stage of disease, people with less severe disease enacted more behaviors in the CM on average than thos e with more severe disease in the baseline intervention sample; at outcome, both designa tions had approximately the same cultural consonance. At baseline, people with hi gher stage disease had a higher rate of consonance; however, at outcome, both portions of the control sample exhibited similar levels of consonance. Table C12. Comparisons of Cultural Consonance for Intervention and Control Groups by Stage of Disease df t p< BMC: Less Severe vs More Severe 25 -1.71 .1002 BMI: Less Severe vs More Severe 31.4 .61 .5485 B-All: Less Severe vs More Severe 239 1.31 .1902 OMI-RM: Less Severe vs More Severe 57.6 -.23 .8222 OMC-RM: Less Severe vs More Severe 24 .05 .9582

PAGE 315

297 Appendix C Independent samples t-tests show no significant differences between the consonance rates of people with more or less severe disease within each of the samples. After a review of these data, it appears th at people with less severe disease were less likely to include the additi onal self-care practi ces they endorsed, indicated in the culturally correct answer ke ys, after support group partic ipation than support group participants with more severe disease. Number of Attacks Table C13. Cultural Consonance Analysis Re sults by Group and Number of Attacks N Average Consonance Fewer Attacks More Attacks Fewer Attacks More Attacks BMC 19 8 .87 .55 BMI 31 28 .84 .54 B-All 149 89 .65 .58 OMC-RM* 18 8 .83 .88 OMI-RM 31 28 .67 .56 One individual removed from analyses in the outcome control group due to missing data Across all samples, consonance rates were higher for individuals with fewer acute attacks in the past year. Within the contro l groups, consonance rates remained relatively static from baseline to outcome for people with fewer attacks while it increased for people with more acute attacks. Interv ention group members saw a decrease in consonance over time for participants experi encing fewer attacks, and those with more attacks displayed the same le vel of consonance over time.

PAGE 316

298 Appendix C Table C14. Comparisons of Cultural Consonance for Intervention and Control Groups by Number of Attacks df t p< BMC: Fewer Attacks vs More Attacks 25 2.45 .0217 BMI: Fewer Attacks vs More Attacks 57 2.82 .0065 B-All: Fewer Attacks vs More Attacks 236 1.93 .0545 OMI-RM: Fewer Attacks vs More Attacks 57 1.98 .0524 OMC-RM: Fewer Attacks vs More Attacks 24 -.32 .7493 When examined for significant differenc es in consonance by number of acute attacks within the pas year, baseline cont rols, baseline intervention, and outcome intervention groups met the threshold for signi ficance. Individuals with less acute attacks were more likely to enact self-care behavior s in the treatment regi men after support group participation.

PAGE 317

299 Appendix D

PAGE 318

300

PAGE 319

301

PAGE 320

302

PAGE 321

303

PAGE 322

About the Author Kelly Simpson is a Research Associate I at Fa mily Health International in Durham, North Carolina. She has completed a BA in psyc hology & anthropology and an MA in medical anthropology in addition to the PhD in Pub lic Health. Kelly is a member of Phi Kappa Phi Honor Society, and she has won several competitive awards: student honorary award for research and practice ( 2008), twice won a maternal a nd child health traineeship (2004-2005, 2005-2006), and the USF university graduate fellowship (2003-2004). Noncompetitive awards received include a stipe nd from the conference presentation and grant program (2008) and an AmeriCorps*NCCC edu cation award, after performing a year of community service within the AmeriCorps *NCCC program. Kelly also completed a study abroad program in Northern Irela nd during her undergraduate tenure (1998). Finally, Kellys key focus areas are cross-cu ltural health, cultural models, infectious disease (HIV, Hepatitis C, TB, Lymphatic Filari asis), disability, international health, and Haitian communities.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200397Ka 4500
controlfield tag 001 002047060
005 20100112133527.0
007 cr mnu|||uuuuu
008 100112s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002744
035
(OCoLC)497135107
040
FHM
c FHM
049
FHMM
090
RA425 (Online)
1 100
Simpson, Kelly M.
0 245
Measuring culture change as an evaluation indicator :
b applying cultural consensus analysis to cultural models of lymphatic filariasis in Haiti
h [electronic resource] /
by Kelly M. Simpson.
260
[Tampa, Fla] :
University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 303 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
520
ABSTRACT: Introduction: This project explores the links between shared cultural beliefs in the illness domain, specific to lymphatic filariasis, and a support group program implemented in three Haitian towns. The purpose is to introduce an innovative approach to evaluation, the cultural model evaluation technique (CM Evaluation), as well as gain an understanding of the shifting cognitive belief structure around the cultural domain of lymphatic filariasis in the Haitian setting as associated with a support group intervention. Method: The sample population was comprised of 241 women across three sites in Haiti: Archaie, Cabaret, and La Plaine. Data were collected from longitudinal surveys in 2003, baseline, and 2005, outcome. Descriptive statistics and CM Evaluation were utilized to assess the success of the support group program.CM evaluation is a two-pronged approach, comprised of cultural consensus analysis (CCA) and cultural consonance analysis (CC), that differs from standard evaluation tools in that it measures beliefs and behaviors at the shared community level and is culturally contextualized. Results: At baseline, most participants were not single (59%), Catholic (49%), literate (57%), relatively poor (71%), and engaged in selling at home or the market (46%). In the reduced model longitudinal CM comparisons, intervention and control groups, the intervention group had the highest rate of consensus (ER=4.71), significant changes in the culturally correct answer key (chi-sq=5.1, df=1, p<.02) and cultural competence (t=3.63, p<.0006). Alternately, controls exhibited no significant differences in the culturally correct answer key (Fisher's Exact two-tailed P<1.00) or cultural competence (t=.62, p<.5407) from baseline to outcome.Conclusion: Evidence suggests that support group participation does significantly impact the shared illness beliefs surrounding lymphatic filariasis, and that this format is appropriate for resource poor settings lacking clinical support. Also, this study suggests that the CM evaluation approach is an appropriate and effective evaluation indicator for assessing changes in shared belief, cultural consensus analysis, resulting from public health interventions while the behavioral piece, cultural consonance, requires further refinement.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Jeannine Coreil, Ph.D.
653
Cultural consonance
Medical anthropology
Methodology
Public health
Explanatory model
690
Dissertations, Academic
z USF
x Community & Family Health
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2744