UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Alternative and complementary therapy use by women living with breast cancer : a test of three models Balneaves, Linda Georgie 2002

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
ubc_2002-749894.pdf [ 18.05MB ]
Metadata
JSON: 1.0090653.json
JSON-LD: 1.0090653+ld.json
RDF/XML (Pretty): 1.0090653.xml
RDF/JSON: 1.0090653+rdf.json
Turtle: 1.0090653+rdf-turtle.txt
N-Triples: 1.0090653+rdf-ntriples.txt
Original Record: 1.0090653 +original-record.json
Full Text
1.0090653.txt
Citation
1.0090653.ris

Full Text

ALTERNATIVE AND COMPLEMENTARY THERAPY USE BY WOMEN LIVING WITH BREAST CANCER: A TEST OF THREE MODELS by LYNDA GEORGIE BALNEAVES BSc, The University of Manitoba, 1990 BN, The University of Manitoba, 1994 MN, The University of Manitoba, 1996 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (School of Nursing, Faculty of Applied Science) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 2002 © Lynda Georgie Balneaves, 2002 In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Nursing The University of British Columbia Vancouver, Canada Date GtJvIo^ 10 , 7QQ2 II Abstract The overall aim of this study was to enhance knowledge of the experience of alternative and complementary therapy (ACT) use in women living with breast cancer. This was achieved by developing and testing three cognitive models of the causal relationships between selected health beliefs, sociobehavioural factors, demographic characteristics, and ACT utilization among women with breast cancer. The prevalence and patterns of ACT use were also examined. A retrospective, correlational survey design was used in this study. A random sample of 650 women with stage I or II breast cancer was selected from the British Columbia Cancer Registry, of which 577 women were eligible for study participation. Completed self-report questionnaires were received from 334 women. The survey included questions to assess perceived risk of breast cancer recurrence, symptom distress, perceived efficacy of ACTs, barriers to ACT use, perceived control, and ACT utilization. Descriptive statistics were used to describe ACT utilization. Structural equation modelling was used to test the three models of ACT use across the contexts of preventive, ameliorative, and restorative health behaviour. A substantial proportion of women with breast cancer was found to be using a variety of ACTs. Vitamin/mineral supplements, herbal remedies, and spiritual therapies were the most commonly reported ACTs, with the majority of women using fewer than five therapies following their breast cancer diagnosis and spending under $50.00 a month on ACTs. Women most often sought information about ACTs from lay sources, including family and friends and print media. The majority of women had disclosed their use of ACTs to at least one of their conventional health care provider(s). Women who had used ACTs prior to their breast cancer diagnosis and had received encouragement from significant others to use ACTs were found to have greater commitment to ACTs. Health beliefs were found to explain a minimal amount of variances in women's commitment to ACTs. Exceptions included perceived efficacy of ACTs with regards to restoring well being and perceived control over well being. Women who believed ACTs to be efficacious in improving physical and mental well being and perceived themselves to be responsible for their well being were more likely to be committed to ACTs. The study findings suggest that ACT use is a widespread phenomenon in breast cancer populations that is influenced most strongly by past health behaviour and the norms and preferences that exist within women's social groups. These findings also contribute to our understanding of ACT use by women with breast cancer as being a reflection of their commitment to self-care and wellness. The need for further research that explores the roles of family members and health care providers in treatment decisions related to ACTs and educational and counseling strategies that support informed treatment decision making are implications of this study. iii Table of Contents Abstract iTable of Contents iii List of Tables x List of Figures xiAcknowledgements, xiiCHAPTER ONE-INTRODUCTION 1 Background to the StudyConceptual Issues 3 Defining Alternative/Complementary Therapies 3 Classifying Alternative/Complementary Therapies 6 Why Study Alternative/Complementary Therapy Use in Women with Breast Cancer? 6 Research Purpose 9 SummaryCHAPTER TWO - LITERATURE REVIEW 11 Alternative and Complementary Therapy Use in General and Cancer Populations 11 Prevalence of Alternative/Complementary Therapy Use in the General Population 12 Assessment of Alternative/Complementary Therapy Use 12 Sampling Issues 1Retrospective Self-Report Measures 13 Alternative/Complementary Therapy Utilization Patterns 13 Prevalence of Alternative/Complementary Therapy Use in Cancer Populations 16 Alternative/Complementary Therapy Use by Women Living with Breast Cancer 22 Characteristics of Consumers of Alternative/Complementary Therapies 24 Demographic Profile within General Populations 25 Demographic Profile within Cancer Populations 6 Previous Health Experiences 27 Health Beliefs and Alternative/Complementary Therapy Use 2Health Beliefs within General Population 28 Health Beliefs within Cancer Populations 33 Health Beliefs within the Breast Cancer Population 35 Perceptions of Need for Care 36 Summary of the Role of Health Beliefs 37 iv Other Factors Associated with Alternative/Complementary Therapy Use 38 Social Support 3Control 40 Treatment Decision Making by Women Living with Breast Cancer 42 Conventional Treatment Decision Making 4Treatment Decision Making and Alternative/Complementary Therapies 45 Chapter Summary 47 CHAPTER THREE - THEORETICAL MODELS 50 Models of Health Behaviour 5Behavioural Model of Health Care Utilization 51 Health Locus of ControlTheory of Reasoned Action 52 Summary 5The Health Belief ModelTheoretical Models of Commitment to Alternative/Complementary Therapies 55 The Preventive Model of Alternative/Complementary Therapy Use 55 The Ameliorative Model of Alternative/Complementary Therapy Use 61 The Restorative Model of Alternative/Complementary Therapy Use 65 Chapter Summary 7CHAPTER FOUR - METHODS AND PROCEDURES.. 72 Research DesignSample 7Setting and Participants 72 Sample Size , 73 Recruitment of Participants 7 Survey Completion Rates 78 Data Collection 81 Modified Total Design Method 8Pilot Testing 2 Operationalization of Study Constructs 83 Demographics 84 Encouragement to Use Alternative/Complementary Therapies 84 Perceived Likelihood of Recurrence 85 Perceived Symptom Distress 6 Perceived Risk of Harm 87 Perceived Severity of HarmPerceived Efficacy of Alternative/Complementary Therapies 88 Perceived Barriers to Alternative/Complementary Therapy Use 89 Perceived Control 90 Commitment to Alternative/Complementary Therapies 92 Data Analysis 101 Missing DataStructural Equation Modelling 10Ethical Considerations 102 CHAPTER FIVE - DESCRIPTIVE FINDINGS 105 The Sample 10Demographic, Disease, and Treatment Characteristics of Non-Respondents 105 Demographic Characteristics of the Respondents 10Disease Characteristics of the Respondents 106 Conventional Cancer Treatment History of the Respondents 10Differences between the Respondents and Non-Respondents 109 Use of Alternative/Complementary Therapies by Women with Breast Cancer 110 Prevalence of Alternative/Complementary Therapy Use 11Prevalence of Alternative/Complementary Therapy Use by Therapy Type 112 Effect of Demographic Characteristics on Alternative/Complementary Therapy Use .113 Patterns of Alternative/Complementary Therapy Use 114 Number of Alternative/Complementary Therapies Used 114 Prior Experience with Alternative/Complementary Therapies 115 Initiation of Alternative/Complementary Therapy Use 11Current Usevof Alternative/Complementary Therapies 117 Frequency of Alternative/Complementary Therapy Use 118 Effort Involved in Using Alternative/Complementary Therapies 11Cost of Alternative/Complementary Therapy Use 119 Sources of Information about Alternative/Complementary Therapies 120 Disclosure of Alternative/Complementary Therapy Use 121 CHAPTER SIX - MODEL PREPARATION 122 Preventive Model of Alternative/Complementary Therapy Use 12Examination of Missing Data 12vi Tests of Normality 126 Univariate Normality 12Bivariate Normality 8 Confirmatory Factor Analysis of the Measurement Model 131 Inclusion of Women with Breast Cancer Recurrence and Other Cancer Diagnosis.... 132 Measurement Scaling and Reliabilities 134 Summary of the Preparation of the Preventive Model of Alternative/Complementary Therapy Use 137 Ameliorative Model of Alternative/Complementary Therapy Use 13Examination of Missing Data 139 Tests of Normality 140 Univariate Normality 14Bivariate Normality 2 Confirmatory Factor Analysis of the Measurement Model 143 Inclusion of Women with Breast Cancer Recurrence and Other Cancer Diagnosis.... 146 Measurement Scaling and Reliabilities 148 Summary of the Preparation of the Ameliorative Model of Alternative/Complementary Therapy Use 14Restorative Model of Alternative/Complementary Therapy Use 151 Examination of Missing Data 15Tests of Normality 153 Univariate Normality 15Bivariate Normality 5 Confirmatory Factor Analysis of the Measurement Model 156 Inclusion of Women with Breast Cancer Recurrence and Other Cancer Diagnosis.... 157 Measurement Scaling and Reliabilities 159 Summary of the Preparation of the Restorative Model of Alternative/Complementary Therapy Use 160 CHAPTER SEVEN - STRUCTURAL EQUATION MODELLING RESULTS 162 Preventive Model of Alternative/Complementary Therapy Use 16Structural Model 1a 16Model Modifications to Model 1a 164 Structural Model 1b 165 Model Modifications to Model 1b 16Structural Model 1c 166 Model Modifications to Model 1c 16vii Structural Model 1d 166 Model Modifications to Model 1d 16Structural Model 1e 167 Model Modifications to Model 1e 16Structural Model 1f 16Model Modifications to Model 1f 168 Bootstrapping of Model 1fDirect and Indirect Effects in Structural Model 1f 170 Summary of Model Testing of the Preventive Model of Alternative/Complementary Therapy Use 173 Ameliorative Model of Alternative/Complementary Therapy Use 174 Structural Model 2b 17Model Modifications to Model 2b 176 Structural Model 2c 17Model Modifications to Model 2c 177 Structural Model 2d 17Model Modifications to Model 2d 177 Structural Model 2e 178 Model Modifications to Model 2e 17Bootstrapping of Model 2e 17Direct and Indirect Effects in Structural Model 2e 181 Summary of Model Testing of the Ameliorative Model of Alternative/Complementary Therapy Use 183 Restorative Model of Alternative/Complementary Therapy Use 18Structural Model 3a 18Model Modifications to Model 3a 185 Structural Model 3b 18Model Modifications to Model 3b 186 Structural Model 3c 18Model Modifications to Model 3c 186 Structural Model 3d 187 Model Modifications to Model 3d 18Bootstrapping of Model 3d 188 Direct and Indirect Effects in Structural Model 3d 189 Summary of Model Testing of the Restorative Model of Alternative/Complementary Therapy Use 192 viii CHAPTER EIGHT-DISCUSSION AND IMPLICATIONS 193 Discussion of the Findings 19Alternative/Complementary Therapy Use by Women Living with Breast Cancer 193 Prevalence of Alternative/Complementary Therapy Use 194 Commonly Used Therapies 195 Patterns of Use 197 Cost of Alternative/Complementary Therapy Use 199 Sources of Information about ACTs and Disclosure of Use 200 The Role of Health Beliefs in Alternative and Complementary Therapy Use 202 Preventive Model of Alternative and Complementary Therapy Use 203 Ameliorative Model of Alternative and Complementary Therapy Use 207 Restorative Model of Alternative and Complementary Therapy Use 209 Comparison of the Three Models of Alternative/Complementary Therapy Use 212 Contributions and Strengths of the Study 215 Theoretical ConsiderationsMethodological Considerations 216 Limitations of the Study 217 Study DesignRetrospective Data 21Measurement Issues 8 Generalizability 221 Implications for Practice 2 Recommendations for Future Research 224 Conclusion 229 REFERENCES 230 APPENDICES 248 Appendix 1 - Family Physician Letter 24Appendix 2 - Letter of Invitation 9 Appendix 3 - Consent Form 250 Appendix 4 - Questionnaire Cover Letter 252 Appendix 5 - Study Questionnaire 3 Appendix 6 - Follow-up Letter 279 Appendix 7 - Survey Assessment Form 280 Appendix 8 - Complete Alternative/Complementary Therapy List (Liberal Definition) 281 Appendix 9 - Covariance Matrices for Models of Alternative/Complementary Therapy Use.. 283 ix List of Tables 1. Classification of Alternative/Complementary Therapies 7 2. Summary Data from Prevalence Studies of ACT Use in Cancer Populations 18 3. Survey Completion Rates 74. Reasons for Ineligibility5. Inclusion and Exclusion Selection Criteria 79 6. Reasons for Non-Participation of Eligible Women 77. List of Scales/Items in Final Study Questionnaire 95 8. Study Constructs and Associated Instruments in the Preventive, Ameliorative, and Restorative Models 100 9. Demographic Characteristics of the Respondents.... 107 10. Conventional Cancer Treatment History of the Respondents 108 11. Discrepancies between Respondents' Self-Report and BC Cancer Registry Data on Conventional Treatment History 109 12. Demographic and Disease Characteristics of Respondents and Non-Respondents ..110 13. Prevalence of Alternative/Complementary Therapy Use 111 14. Most Frequently Reported Alternative/Complementary Therapies 112 15. Prevalence of Alternative/Complementary Therapy Use by Therapy Type 113 16. Association of Demographic Characteristics with Alternative/Complementary Therapy Use 114 17. Number of Alternative/Complementary Therapies Used 115 18. Association between Prior Alternative/Complementary Therapy Use and the Use of Alternative/Complementary Therapies Following Breast Cancer Diagnosis.... 116 19. Alternative/Complementary Therapy Initiation by Therapy Type 116 20. Comparison of Therapy Initiation by Therapy Type 117 21. Association between Therapy Initiation and Current Use of Alternative/ Complementary Therapies at Time of Survey Completion 118 22. Frequency of Alternative/Complementary Therapy Use 1123. Effort Involved in Using Alternative/Complementary Therapies by Therapy Type 119 24. Reported Sources of Information about Alternative/Complementary Therapies 121 25. Summary of Missing Data in the Preventive Model of Alternative/Complementary Therapy Use 124 X 26. Frequency of Imputed Missing Data for Alternative/Complementary Therapy Use Variables by Therapy Type 125 27. Univariate Statistics for Exogenous Variables (Preventive Model - Model 1a) 126 28. Univariate Statistics for Endogenous Variables (Preventive Model - Model 1a) 127 29. Bivariate Normality Violations for Exogenous and Endogenous Variables in the Preventive Model (Model 1a) Following Transformations 129 30. Univariate Statistics for Revised and New Variables (Preventive Model - Model 1a). 130 31. Bivariate Normality Violations for Exogenous and Endogenous Variables in the Preventive Model (Model 1a) Following Transformations 131 32. Goodness-of-Fit Indices for the Measurement Model of the Preventive Model (Model 1a 132 33. Factor Loadings and t-Values for the Measurement Model of the Preventive Model.. 133 34. Mean Differences between Women with and without Breast Cancer Recurrence on Preventive Model (Model 1a) Variables 134 35. Mean Differences between Women with and without Other Cancer Diagnosis on Preventive Model (Model 1a) Variables 135 36. Summary of Missing Data in the Ameliorative Model of Alternative/Complementary Therapy Use (Model 2a) 140 37. Univariate Statistics for Exogenous Variables (Ameliorative Model - Model 2a)... 141 38. Univariate Statistics for Endogenous Variables (Ameliorative Model - Model 2a) 142 39. Bivariate Normality Violations for Exogenous and Endogenous Variables in the Ameliorative Model (Model 2a) Following Transformations 143 40. Goodness-of-Fit Indices for the Measurement Model of the Ameliorative Model (Model 2a and Model 2b) 144 41. Factor Loadings and f-values for the Measurement Model of the Ameliorative Model (Model 2a and Model 2b) 145 42. Mean Differences between Women with and without breast cancer recurrence on the Ameliorative Model (Model 2b) Variables 146 43. Mean Differences between Women with and without other cancer diagnosis on the Ameliorative Model (Model 2b) Variables 147 44. Error Variances in the Ameliorative Model of Alternative/Complementary Therapy Use.(Model 2b) 149 45. Summary of Missing Data in the Restorative Model of Alternative/Complementary Therapy Use (Model 3a) 152 46. Univariate Statistics for Exogenous Variables (Restorative Model - Model 3a) 153 47. Univariate Statistics for Endogenous Variables (Restorative Model - Model 3a) 154 XI 48. Bivariate Normality Violations for Exogenous and Endogenous Variables in the Restorative Model Following Transformations 156 49. Goodness-of-Fit Indices for the Measurement Model of the Restorative Model (Model 3a) 1550. Factor Loadings and f-values for the Measurement Model of the Restorative Model (Model 3a) 7 51. Mean Difference between Women with and without Breast Cancer Recurrence on the Restorative Model (Model 3a) Variables 158 52. Mean Difference between Women with and without Other Cancer Diagnosis on the Restorative Model (Model 3a) Variables 159 53. Error Variance in the Restorative Model of Alternative/Complementary Therapy Use (Model 3a) 160 54. Goodness-of-Fit Indices for Sequential Modifications of the Preventive Model of Alternative/Complementary Therapy Use 165 55. Parameter Estimates for Model 1f (Preventive Model) 171 56. Direct and Indirect Effects from Structural Model 1f (Preventive Model) 171 57. Goodness-of-Fit Indices for Sequential Modifications of the Ameliorative Model of Alternative/Complementary Therapy Use 174 58. Parameter Estimates for Model 2d (Ameliorative Model) 179 59. Direct and Indirect Effects from Structural Model 2d (Ameliorative Model) 182 60. Goodness-of-Fit Indices for Sequential Modifications of the Restorative Model of Alternative/Complementary Therapy Use 186 61. Parameter Estimates for Model 3d (Restorative Model) 188 62. Direct and Indirect Effects from Structural Model 3d (Restorative Model) 190 63. Major Study Findings and Implications 225 xii List of Figures 1. Preventive Model of Alternative/Complementary Therapy Use by Women Living with Breast Cancer 56 2. Ameliorative Model of Alternative/Complementary Therapy Use by Women Living with Breast Cancer 63 3. Restorative Model of Alternative/Complementary Therapy Use by Women Living with Breast Cancer 7 4. Flow Diagram of the Development of the Final Dataset prior to Randomization 74 5. Flow Diagram of the Recruitment Process 80 6. Transformation of Perceived Control Items 91 7. Total Monthly Cost of Current Alternative/Complementary Therapy Use (Liberal Definitions 128. Model 1a - Preventive Use of Alternative/Complementary Therapies 138 9. Model 2b - Ameliorative Use of Alternative/Complementary Therapies 150 10. Model 3a - Restorative Use of Alternative/Complementary Therapies 161 11. Structural Model 1 a - Preventive Use of Alternative/Complementary Therapies 163 12. Structural Model 1f- Preventive Use of Alternative/Complementary Therapies 172 13. Structural Model 2b - Ameliorative Use of Alternative/Complementary Therapies 175 14. Structural Model 2d - Ameliorative Use of Alternative/Complementary Therapies 180 15. Structural Model 3a - Restorative Use of Alternative/Complementary Therapies 184 16. Structural Model 3c - Restorative Use of Alternative/Complementary Therapies 191 xiii Acknowledgements Although there is just one person listed at the front of this dissertation, there are many others that have contributed to what you are about to read. Foremost, I would like to express my sincere gratitude to my doctoral dissertation committee, Drs. Joan Bottorff, Pamela Ratner, Greg Hislop, and Carol Herbert for their kind guidance and exceptional mentoring during the past six years. It was a long road, but your encouraging words and thoughtful comments pushed me through the tough times and challenged me to keep open to different perspectives. To my partner Allan Brett, you were there from the beginning of this journey and have cheered me on through the late nights, the never-ending deadlines, and the times when putting a sentence together was a feat unto itself. Thank you for your love and patience - I promise no more degrees! To my parents, Georgie and Les Balneaves, this dissertation reflects the determination and confidence you instilled in me to pursue my dreams, even the ones that scare me. Thank you for always being there and cheering me on. To my "BC" parents, Bill and Dianna Brett, I came to BC to get my PhD and never thought that I would end up with an extended family. Thank you for opening your hearts to me and sharing your home. And for my friends, Brenda, Carol, Ruth, Caroline, and Anne Marie, my sincere gratitude for the wine, whine, and love that you were always ready to share. I would also like to acknowledge the generous support I received throughout my studies from a variety of organizations, including the National Health Research Development Program, the British Columbia Health Research Foundation, the Canadian Red Cross, the Canadian Nurses Foundation, the Canadian Association of Nurses in Oncology, the National Cancer Institute of Canada, the University of British Columbia, and UBC School of Nursing. And to the women who took the time to participate in this research, my deepest thanks for your willingness to share your experiences and for helping us better understand the treatment choices that you are faced with. Your stories continue to inspire me to work towards finding ways to better care for you and your family members. 1 Chapter 1 Introduction In recent years, there has been a remarkable increase in interest and use of therapies that are considered to be beyond the realm of conventional medical care. In a recent survey conducted by the Fraser Institute (Ramsay, Walker, & Alexander, 1999), nearly three quarters of Canadians surveyed had used treatment that was alternative or complementary to their allopathic care at some point in their lives. One group that has been vocal in its support and advocacy of alternative and complementary therapies (ACTs) has been women with breast cancer. In a document recently published by the BC/Yukon Chapter of the Canadian Breast Cancer Foundation (Trussler, 2001), consumers identified ACTs as playing a central role in the supportive breast care strategy. With women with breast cancer comprising one of the largest categories of cancer survivors in Canada (National Cancer Institute of Canada [NCIC], 2001), the impact of ACTs on provincial health care systems may be significant. As the movement towards integration of ACTs into conventional cancer care has gained momentum, researchers have focussed on determining what social and behavioural factors influence the treatment decisions of individuals living with cancer. While much has been uncovered by this research, what remains to be developed is a comprehensive, theoretical model of ACT utilization. Background to the Study Interviews with women living with breast cancer have revealed ACTs to be integral to their experiences with this life-threatening disease (Boon et al., 1999; Gray et al., 1997; Truant, 1998). For some women, ACTs are a means of preserving hope when faced with the uncertainty of their prognosis (Truant & Bottorff, 1999). For others, the decision to use ACTs reflects a desire to regain control over their health and to assert their independence within the biomedical health-care system (Montbriand, 1995a). Women with breast cancer who choose ACTs have also been motivated by specific goals related to their disease and recovery, 2 including the improvement of their immune system, the management of adverse effects of conventional cancer treatments, and the restoration of their physical and emotional well being (Crocetti et al., 1998). Despite the limited number of women who have reported using ACTs in the hope of curing their breast cancer (Morris, Johnson, Homer, & Walts, 2000) and as a replacement for conventional cancer care (Burstein, Gelber, Guadagnoli, & Weeks, 1999), conventional health professionals and researchers have expressed concern regarding the use of these therapies (Beyerstein, 1997; Damkier, Elverdam, Glasdam, Jensen, & Rose, 1998; Davidoff, 1998; Durant, 1998). Their concern is not without justification because there has been limited study of the efficacy and safety of ACTs in relation to cancer care (Tagliaferri, Cohen, & Tripathy, 2001) and some suggestion that potentially harmful interactions with conventional treatments may exist (Decker, 2000; Jacobson & Verret, 2001). A recent study of women with early-stage breast cancer (Burstein et al., 1999) raised additional concerns when women using ACTs were found to report higher levels of psychological distress and anxiety than non-users. Researchers have been cautious in relating ACT use to increased psychological morbidity (Holland, 1999), instead suggesting that cancer patients may turn to ACTs to treat the distress that has been not addressed by the conventional health-care system. With nearly one half of women with breast cancer choosing not to discuss their treatment decisions about ACTs with their conventional health-care providers (Adler & Fosket, 1999; Balneaves, Kristjanson, & Tataryn, 1999), opportunities to discuss concerns about ACTs and satisfaction with conventional care are often not realized in clinical settings. Concerns about the safety of ACTs and the possibility that ACT use may function as a marker for distress and dissatisfaction with care have resulted in a large body of research focussed on determining which individuals are most likely to utilize ACTs. In both the general population (Eisenberg, 1997; Kelner & Wellman, 1997a; Millar, 1997) and across cancer diagnoses (Crocetti et al., 1998; Downer et al., 1994; Ernst & Cassileth, 1998), a distinct profile of ACT consumers has been revealed. Sociodemographic factors, such as age, gender, education, income, and a history of chronic illness, have been associated with ACT utilization (Blais, Maiga, & Aboubacar, 1997; Eisenberg et al., 1998; Lerner& Kennedy, 1992). Research has also examined the role of cognitive factors and belief systems in the decision to use ACTs. Most striking has been the association between beliefs about health and illness and treatment decisions related to ACTs (Furnham & Bond, 2000; Furnham & Kirkcaldy, 1996; Kelner & Wellman, 1997a; Risberg, Wist, & Bremnes, 1998; Vincent & Furnham, 1996). Recent work with women with breast cancer has revealed the importance of these beliefs, in addition to such sociobehavioural factors as perceived control, social support, previous use of ACTs, and perceptions of need (Balneaves et al., 1999; Boon et al., 2000; Crocetti et al., 1998; Truant & Bottorff, 1999). While this research has provided a preliminary and descriptive understanding of 3 ACT use in the context of breast cancer, what remains unanswered is how beliefs and sociobehavioural variables interrelate and motivate individuals' treatment decisions related to ACTs. This gap in knowledge underscores the need for innovative research that moves beyond the descriptive level and results in the development and testing of comprehensive theories of ACT use. The recent emphasis on the association between health beliefs and ACT use (Blais et al., 1997; Boon, Brown, Gavin, Kennard, & Stewart, 1999; Furnham & Kirkcaldy, 1996; McGregor & Peay, 1996; Millar, 1997; Vincent & Furnham, 1996; Yates et al., 1993) points towards the possible relevance of social cognitive theories in explaining ACT utilization. Conceptual Issues Defining Alternative/Complementary Therapies Studies exploring the prevalence and motivations of ACT use have been marked by a lack of consensus with regards to the manner in which therapies are described, defined, and categorized. A variety of labels has been applied to ACTs, including alternative, complementary, adjunctive, unproven, unconventional, unorthodox, questionable, and quackery. The language used to describe ACTs has been described as reflective of the cultural or political position of these therapies within a society's health care system rather than a description of the care provided (Turner, 1998). Lerner (1994) further suggested that the terminology used relates primarily to the degree to which a therapy has achieved "acceptability" within the medical establishment. As such, the language used to describe ACTs is by no means benign in intent or meaning. Care is needed in selecting nomenclature that accurately and appropriately describes ACTs and the positions these therapies hold within the cultural, economic, medical, and educational domains of a society (Eskinazi, 1998). At the beginning stages of ACT research, the term "alternative medicine" gained prominence within the biomedical community, most notably because of its use in two national surveys on ACT use conducted in the United States (Eisenberg et al., 1998; Eisenberg et al., 1993). This term was used to describe the following: • Interventions neither taught widely in medical schools, nor generally available in US hospitals (Eisenberg et al., 1998; Eisenberg et al., 1993), • Therapies used instead of Western medicine (Spiegel, Stroud, & Fyfe, 1998), • Therapies characterized by their sociopolitical marginality (i.e., lack of support from medicine and/or government) (Kelner & Wellman, 1997b), and • Therapies that are physiologically active, potentially harmful, and in conflict with mainstream care (Cassileth, 1998). 4 These descriptions of "alternative medicine" were problematic in several ways. Foremost, in placing "alternative medicine" at the margins of conventional medicine, there existed an underlying assumption that the therapies were used solely within a culture in which the biomedical paradigm was dominant. This assumption failed to consider health care practices on a worldwide basis where other distinct belief and practice systems exist. Secondly, the way in which "alternative medicine" was defined suggested that any therapy or intervention not claimed by conventional Western medicine would be classified as alternative (Wardell, 1994). With physicians and other health care professionals beginning to incorporate ACT content within their curricula, which therapies should be included under the rubric "alternative medicine" becomes increasingly difficult to determine. The use of the term "medicine" within this description is also of concern because it lacks neutrality and juxtaposes these therapies with a paradigm focussed on disease rather than well being. There also has been the suggestion that "alternative medicine" is representative of the co-optation of selected therapies by the conventional medical community. Another popular label used to define ACTs has been "complementary" therapies or medicine (Ernst, 1995; Fairfield, Eisenberg, Davis, Libman, & Phillips, 1998; Furnham, Vincent, & Wood, 1995; Pietroni, 1994; Truant & Bottorff, 1999). Considered to be more respectful than "alternative medicine" (Turner, 1998), "complementary" therapies have been described in numerous ways: • Therapies that complement the intelligent use of conventional approaches deemed scientifically to be efficacious (Lerner, 1994), • Therapies that involve cooperative effort between biomedicine and alternative practitioners (Nienstedt, 1998), • Therapies used in addition to and to balance the shortcomings of mainstream medicine (Spiegel et al., 1998), • Therapies that are adjunctive to medical care (Kelner & Wellman, 1997b), and • Therapies that are pleasant, non-toxic, and beneficial to quality of life (Cassileth, 1998). Special emphasis has been placed on the distinction in meaning between "alternative" and "complementary" therapies and the coexistence of both kinds of therapies (Cassileth, 1998; Spiegel et al., 1998). Complementary therapies, for the most part, are considered to be more adjunctive and supportive than alternative therapies. In particular, Milton (1998) suggested that within cancer care, complementary therapies are used to manage cancer symptoms and the side effects of conventional treatment. Accordingly, when complementary therapies are used to promote healing and quality of life, they are believed to work in tandem with conventional medicine. The use of the complementary therapy term as a descriptor of ACTs, however, does not encompass those therapies that are used in place of conventional medicine. Most recently, a movement towards terminology that encompasses both alternative and complementary approaches to health care has occurred. In 1998, the US National Institutes of Health (NIH) Office of Alternative Medicine was re-established as the National Center for Complementary and Alternative Medicine (NCCAM). The following definition and description of complementary and alternative medicine (CAM) is used by the NCCAM: A broad domain of healing resources that encompasses all health systems, modalities, and practices and their accompanying theories and beliefs, other than those intrinsic to the politically dominant health system of a particular society or culture in a given historical period. CAM includes all such practices and ideas self-defined by their users as preventing or treating illness or promoting health and well being. Boundaries within CAM and between the CAM domain and the domain of the dominant system are not always sharp and defined. (Anonymous, 1997, p. 50) CAM has also been defined by Ernst and Cassileth (1998) as the: Diagnosis, treatment and/or prevention which complements mainstream medicine by contributing to a common whole, by satisfying a demand not met by orthodoxy, or by diversifying the conceptual frameworks of medicine, (p. 777) Together, these definitions position CAM within a social, political, and cultural context and acknowledge the holistic nature putatively embodied by many traditional healing systems. However, concerns have been raised regarding the broadness of the term CAM and its application to therapies that have been accepted as adjunctive interventions within conventional medicine (e.g., group therapy, relaxation techniques) (Cassileth, 1998). Determining what therapies are accepted and by whom continues to be an elusive task because beliefs, attitudes, and knowledge surrounding ACTs continue to unfold within medical and lay communities. Although the use of the term CAM represents a progressive step towards a comprehensive and contextual definition of therapies not traditionally included within a society's dominant health care system, a medical bias in language is still evident. Balneaves (1996) attempted to avoid this bias by developing the acronym ACTs (alternative and complementary therapies). This terminology not only considers therapies that are used to supplement or replace treatments offered by the dominant health care system, but also symbolizes choice and action on the part of health care consumers. In addition, the term ACTs allows both alternative 6 systems of healing (e.g., Traditional Chinese Medicine) and individual therapies (e.g., acupuncture) to be included under one appellation and acknowledges the intervention of practitioners other than physicians. Which therapies are considered to be alternative, complementary, or conventional, however, is dependent upon the relative hegemony of the biomedical paradigm within the society under study and the personal experiences of health care consumers. Notwithstanding this ambiguity, the ACT terminology is used in this study. Classifying Alternative/Complementary Therapies A variety of strategies have been proposed to organize and classify the hundreds of existing ACTs. The most often quoted method is the seven category approach recommended by the NCCAM (Nienstedt, 1998), which includes the following: diet, nutrition, and lifestyle changes; mind/body interventions; alternative systems; bioelectromagnetic interventions; manual healing; pharmacological and biological therapies; and herbal medicine. These categories, however, have been criticized as being too broad and ambiguous to permit definitive classification (Nienstedt, 1998). For example, although relaxation and meditation techniques are traditionally classified under the rubric of mind/body interventions, they may also play an integral role within lifestyle changes, such as stress reduction. A more specific categorization system has been proposed by the Ontario Breast Cancer Information Exchange Project (1994) and consists of 15 distinct classes of ACTs (see Table 1). This classification is particularly useful within the context of this dissertation research, because it was developed specifically to inform Canadian women living with breast cancer1. Being able to classify therapies into their respective traditions allows ACT use to be explored in a more refined manner and acknowledges the range of therapies that exists. Why Study Alternative/Complementary Therapy Use in Women with Breast Cancer? The utilization of ACTs has been examined across general and disease-specific populations, with cancer (Boon et al., 2000; Downer et al., 1994; Montbriand, 1995a; Yates et al., 1993), multiple sclerosis (Fawcett, Sidney, Hanson, & Riley-Lawless, 1994), and HIV/AIDS (Pawluch, Cain, & Gillett, 1994; Singh et al., 1996) being a few examples. That ACTs are being used to address a variety of life-threatening and chronic conditions, in addition to general well being, is apparent. The development and testing of cognitive models of ACT use could occur in populations other than women living with breast cancer and provide important insights into the underlying health beliefs and sociobehavioural factors that influence treatment choice. Testing the application of the developed models within these populations is an important future research endeavour, however, for the purposes of this study, women with breast cancer are the focus. 1 This classification system includes therapies that are of particular interest to women living with breast cancer and, as such, is not a comprehensive list of ACT therapies. For example, spiritual therapies, such as prayer and laying on of hands, were not included. f 7 Table 1. Classification of Alternative/Complementary Therapies Category Therapies Natural Health Practices Aromatherapy Ayurvedic medicine Herbalism Naturopathy Homeopathy Native North American healing Traditional Chinese medicine Herbal Therapies Aloe Carnivora Coffee enema Essiac Hoxsey method Pau d'Arco Iscadora Ginseng Dietary Therapies Alkaline/acid cleansing diet Gerson therapy Grape cure Metabolic therapy Macrobiotic diet Vitamins Vitamin A B vitamins Beta-carotene Vitamin C Vitamin D Vitamin E Megavitamin therapy Minerals Calcium Selenium Zinc From Earth and Sea Alkylglycerols Canthaxanthin Evening primrose Mushrooms Seaweed Shark Cartilage Energy Life Force Therapies Acupuncture/acupressure Reiki Polarity therapy Therapeutic touch Movement Therapies T'ai Chi Yoga Physical Therapies Chiropractic Massage Reflexology Psychological Therapies Hypnosis Imagery A/isualization Meditation Psychotherapy and counselling Relaxation Expressive Arts Therapies Art therapy Music therapy To and From the Body Cell extraction therapy Colonic irrigation Urea Oxygen Therapies Ozone Hydrogen Peroxide Drugs 714-X Aspirin DMSO Enzyme therapy Bezaldehyde Chelation therapy Immune Boosters Bestatin Chondriana Coley's toxins Immuno-augmentative therapy Isoprinosine Note. Adapted from Ontario Breast Cancer Information Exchange Project (1994). A guide to unconventional cancer therapies. Toronto, ON: Author. This population has been chosen as a starting point to test models of ACT utilization for several reasons. Foremost, breast cancer is a pervasive disease in Canadian society. In the past decade, the incidence of breast cancer has risen steadily. One in nine women is now expected to develop this disease at some point in her lifetime, with approximately 19,500 new cases of 8 breast cancer being diagnosed in 2001 (NCIC, 2001). While breast cancer mortality has declined slightly in recent years, breast cancer remains the second leading cause of cancer death for Canadian women. In British Columbia, approximately 2,500 women were diagnosed in 2001 (NCIC, 2001). These statistics suggest that a significant number of women in Canada are living with, and dying from, this disease. Secondly, the effects of breast cancer on women and their families are extensive and profound. For many women, living with breast cancer is a traumatic, life-altering experience. The time from diagnosis to survival is fraught with uncertainty, dramatic physical changes, and emotional upheaval (Bleiker, Pouwer, van der Ploeg, Leer, & Ader, 2000; Pelusi, 1997; Wainstock, 1991). The side effects of conventional cancer treatments experienced by some women contribute to these feelings of distress. Breast cancer patients are also faced with existential issues, such as ascribing meaning to their illness (Luker, Beaver, Leinster, & Owens, 1996) and confronting a terminal prognosis (Wainstock, 1991). Given the psychosocial distress and potential mortality associated with breast cancer, women may seek treatment options other than those offered by conventional medicine. Recent prevalence studies have confirmed that ACTs are used by women living with breast cancer, with reported prevalence rates between 17% to 67% (Balneaves et al., 1999; Boon et al., 2000; Burstein et al., 1999; Crocetti et al., 1998; Morris et al., 2000; Moschen et al., 2001; Rees et al., 2000; Salmenpera, 2002; VandeCreek, Rogers, & Lester, 1999). Compared to patients with other forms of cancer, women with breast cancer are significantly more likely to use some form of alternative or complementary treatment (Morris et al., 2000). With such a substantial number of women using ACTs, the breast cancer population offers an interested, willing, and relevant milieu in which to conduct research examining the role of health beliefs and sociobehavioural factors associated with ACT use. Another reason for researching ACT use in women with breast cancer is because they have demanded it. At the National Breast Cancer Forum in 1993, breast cancer survivors and advocates rated ACTs as one of the top priorities in terms of both efficacy studies and behavioural research (National Forum on Breast Cancer, 1994). A recent inquiry into the state of breast cancer care in British Columbia has given further support to the importance of ACTs to women living with this disease (Trussler, 2001). Recent initiatives through the National Cancer Institute of Canada's Canadian Breast Cancer Research Initiative (CBCRI) have also emphasized the relevance and importance of research exploring ACT use within the Canadian breast cancer community (Jacobson, 1996). Research Purpose The main purpose of this research study was to develop and test three cognitive models of ACT utilization in women living with breast cancer. In testing these models, the influence of selected health beliefs (including perceived risk, perceived efficacy of ACTs, perceived barriers to ACT use, and perceived control) and sociobehavioural factors (including previous ACT use, encouragement to use ACTs, sociodemographic characteristics) on women's treatment decisions related to ACTs was explored. Guided by the theoretical underpinnings of the Health Belief Model (Janz & Becker, 1984; Rosenstock, 1974b), the three models tested in this research allowed women's use of ACTs to be examined across three different health contexts: prevention, amelioration, and restoration. In developing three unique models, the relevance of health beliefs and sociobehavioural factors within different contexts of ACT use was explored and the intent of therapy use was investigated. A secondary aim of the study was to determine the types of therapies used by women with breast cancer and the degree of commitment extended towards ACTs. The following research questions were addressed in the study: • How are health beliefs and selected sociobehavioural factors associated with the use of ACTs by women living with breast cancer within preventive, ameliorative, and restorative health contexts? • To what extent do the preventive, ameliorative, and restorative models explain ACT use in a breast cancer population? • What are the prevalence, pattern, and financial cost of ACT use by women living with breast cancer in British Columbia? Summary As the prevalence of and interest in alternative and complementary therapies (ACTs) have grown within general and cancer populations, it has become increasingly important to understand why individuals make such choices and what underlying beliefs influence their treatment decision making. Such knowledge will inform future research priorities in ACTs, assist in the development of appropriate counselling and educational strategies for patients and conventional health-care providers, and improve the clinical care of people using ACTs. Focussing on the experience of women living with breast cancer, this research study was undertaken to develop and test three cognitive models that examined the effects of selected health beliefs and sociobehavioural factors on women's decisions to use ACTs. In the following chapter, an overview of the literature pertaining to ACT use in general and cancer populations, including the characteristics of ACT users and motivating factors of ACT use, is provided. Later chapters review the theoretical framework guiding the development of the three cognitive models, the design and procedures of the study, and the descriptive and structural modelling research findings. The final chapter provides a concise discussion of the results and implications of the study. 11 Chapter 2 Literature Review The first section of this review provides a summary of the diverse literature on alternative and complementary therapy (ACT) use. This section focusses on recent sociobehavioural research that has examined ACT use within the general, breast cancer and other cancer populations. Research findings related to the prevalence of ACT use within general and cancer populations, the demographic and other predisposing characteristics of ACT consumers within general and cancer populations, and the motivations underlying the decision to use ACTs by individuals with cancer and within the general population are discussed. In the second section, a concise review of research examining treatment decision making in women with breast cancer is presented. Together, these two sections provide the substantive foundation for the development and testing of three cognitive models of ACT use within a sample of women living with breast cancer. The majority of literature reviewed was drawn from a search of the research literature published in the past two decades, encompassing the fields of nursing, medicine, psychology, and sociology, (i.e., CINAHL, Medline, PsychLit, and Sociofile). Excluded from this review, for the most part, are non-English and unpublished literature. Non-research articles (e.g., editorials, letters to the editors, opinion articles) included in the review consist primarily of discussion pieces that provide a theoretical or social context to the research exploring the use of ACTs. Alternative and Complementary Therapy Use in General and Cancer Populations The following section provides a detailed review and critique of sociobehavioural ACT research. The main areas of discussion include the prevalence of ACT use across general and cancer populations, the personal and social characteristics of ACT consumers, the influencing role of health beliefs in determining treatment choice, and other motivators of ACT use. 12 Prevalence of Alternative/Complementary Therapy Use in the General Population Previous epidemiological research has revealed that between 15% to 68% of the general population in North America, Europe, and the Middle East have used at least one type of ACT in their lifetime (Angus Reid Group, 1997; Bernstein & Shuval, 1997; Eisenberg et al., 1998; Eisenberg et al., 1993; Grenfell, Patel, & Robinson, 1998; MacLennan, Wilson, & Taylor, 1996; Millar, 1997; Murray & Shepherd, 1993; Ramsay et al., 1999). Despite the wide variation in utilization estimates, this research suggests that the use of ACTs is a substantial health care phenomenon within industrialized countries. The disparity among utilization estimates can be attributed, in part, to several methodological issues, including the methods of assessment of ACT use (Harris & Rees, 2000), sampling issues, and the retrospective nature of the studies. Each of these issues is discussed, followed by a comparison of ACT utilization patterns in the general population. Assessment of Alternative/Complementary Therapy Use There has been a lack of consensus among researchers about how ACT use should be assessed. For some researchers, the use of therapies that are marginal to conventional medical treatment and that are generally unavailable within the majority of mainstream health-care institutions has been considered to be indicative of ACT use (Eisenberg, 1997; Eisenberg et al., 1993; Murray & Shepherd, 1993; Ramsay et al., 1999). In contrast, others researchers have used only consultations with alternative/complementary practitioners as being representative of ACT use (Millar, 1997). With over 50% of ACT consumers using treatments without consulting an alternative/complementary practitioner (Eisenberg et al., 1998), examining only the use of alternative/complementary practitioners may significantly underestimate the actual use of ACTs. Further, there has been variation in the way in which individuals are asked about ACT use. Although the majority of researchers have used pre-determined lists of ACTs and alternative practitioners to uncover ACT use (Eisenberg et al., 1998; Eisenberg et al., 1993; Millar, 1997; Murray & Shepherd, 1993; Ramsay et al., 1999), researchers have also used open-ended questions about treatment choices (Grenfell et al., 1998). This diversity in data collection methods limits comparisons across studies and biases participants' responses. Sampling Issues Several issues arise related to the sampling methods used in previous surveys of ACT use. While random sampling was employed in the majority of prevalence studies examining ACT use in general populations (Angus Reid Group, 1997; Bernstein & Shuval, 1997; Eisenberg et al., 1998; Eisenberg et al., 1993; Ramsay et al., 1999; Thomas, Carr, Westlake, & Williams, 1991), some researchers (Grenfell et al., 1998; Kristof, Schlumpf, Wyss, & Sailer, 1998; Murray & Shepherd, 1993) used convenience samples to determine the pervasiveness of ACT use in the general public. The latter sampling strategy raises concerns regarding the generalizability of 13 the findings. Further, the reliance on convenience samples to determine prevalence rates of ACT use, coupled with the identification of ACT use as being the subject of interest of the research, may have resulted in an over-representation of individuals who had used ACTs. Eisenberg et al. (1998, 1993) attempted to avoid this bias by making no mention of ACTs in their initial recruitment of participants. Another methodological issue is the recruitment methods used in some of the prevalence studies. The use of telephone and/or household interviews that were limited to those participants able to speak English may have restricted the participants recruited in terms of socioeconomic status, ethnicity, and residency. As a consequence, the study findings may have limited generalizability to disadvantaged and ethnic populations (Eisenberg et al., 1998; Eisenberg et al., 1993; Millar, 1997; Ramsay et al., 1999). Further, the recruitment of participants through conventional or alternative health-care systems (Grenfell et al., 1998; Kristof et al., 1998; Murray and Shepherd, 1993; Thomas, et al., 1991) may have potentially under- or over-represented the prevalence of ACT use within general populations. For example, it is not surprising that Kristof et al. (1998) found over 96% of attendees at an alternative health fair had used ACTs at some time in their lives. A final sampling issue that has not been addressed well in past prevalence studies has been whether the sample size was adequate to produce prevalence rates of ACT use that could be applied with confidence to the general population. Sample size calculations or confidence intervals are provided in only a few of the research articles reviewed (Eisenberg et al., 1998; Eisenberg et al., 1993; MacLennan et al., 1996) Retrospective Self-Report Measures The use of retrospective self-report measures in the majority of ACT survey research may have limited the validity of the findings. For example, Murray and Shepherd (1993) asked participants recruited from a general medical practice about the ACTs they had tried in the previous 10 years. Such a lengthy time span increases the possibility of memory distortion (i.e., recall bias) and inaccurate reporting of ACT use. A further limitation of retrospective studies on ACT use has been the cross-sectional approach of the research, which limits causal interpretations of the data and prevents cohort effects from being clearly identified. Prospective research on ACT use is needed to capture the dynamic nature of ACT use and to provide more precise measures of prevalence and incidence rates. Alternative/Complementary Therapy Utilization Patterns Given the methodological limitations of past prevalence research, Eisenberg et al.' s (1998, 1993) nationally-representative household telephone surveys in the United States have been the most scientifically rigorous investigations to date. Using random-digit sampling and the random selection of one household member over the age of 18 years, Eisenberg et al. (1993) 14 conducted telephone interviews with 1,539 participants about their use of ACTs. One in three participants was found to have used at least one type of ACT (selected from a pre-determined list of 16 ACTs) in the past 12 months. The most commonly used ACTs included relaxation techniques (13%), chiropractics (10%), and massage (7%). Of those using ACTs, 64% did so without consulting an alternative/complementary practitioner (e.g., over-the-counter herbal products) and 72% chose not to inform their primary care physician about their use of ACTs. Overall, Eisenberg et al. (1993) estimated out-of-pocket expenditures for therapies and practitioner visits in the United States to total $13.7 billion (US funds). In a similarly designed survey conducted in 1997, Eisenberg et al. (1998) found that ACT use increased by 25% since 1990, with 42% of participants over the age of 18 years (N = 2,055) reporting the use of ACTs in the previous 12 months. The most frequently reported ACTs included relaxation techniques (16%), herbal medicine (12%), massage (11%), and chiropractics (11%). The total number of visits to an alternative/complementary practitioner increased by 47%, with 46% of ACT consumers reporting consultations with an alternative/ complementary practitioner. No difference was noted in the follow-up study with regards to disclosure of ACT use to conventional health-care providers. Expenditures for ACTs were found to have increased by 45% since 1990, with an estimated $21.2 billion (US funds) being spent on therapies and practitioners. Research conducted in Canada has found prevalence rates similar to Eisenberg et al. (1998). In 1997, the Angus Reid Group released to the public the findings of a nation-wide telephone survey of 1,200 Canadian adults 18 years and older (Angus Reid Group, 1997).2 Over four in ten (42%) Canadians reported using ACTs, with British Columbia residents being most likely to report ACT use (56%). Chiropractics (59%), herbology (23%), acupuncture (22%), and homeopathy (18%) were the most commonly reported ACTs. Further information on the design of the study, particularly related to sampling procedures, would have been helpful to further evaluate the scientific merit of this survey. In a similar study, the Fraser Institute conducted 1,500 telephone interviews with a randomly selected sample of Canadian adults in 1997 (Ramsay et al., 1999). Households were selected using random-digit dialing, with respondents being randomly selected based on the "birthday method" in which the person with the most recent birthday was interviewed. The interview was structured similar to Eisenberg's (1998, 1993), surveys with respondents being asked to choose the therapies they had used in the past 12 months and in their lifetime from a randomized list of 22 ACTs. Nearly three quarters of the respondents indicated that they had 2 While specifics were not provided in the press release related to the sampling procedure, the data were weighted statistically to ensure the sample's age/sex composition reflected that of the actual Canadian population based on the 1996 Census. The researchers reported that with the Canadian-wide sample of 1,200, surveys such as this were accurate within +2.8 percentage points, 19 times out of 20. 15 used at least one ACT in their lifetime, with British Columbians (85%) being the most likely to report the use of ACTs. Chiropractic care was the most frequently reported therapy (36%), followed by relaxation and massage (23%), prayer (21%), and herbal remedies (17%). The findings of this survey were compromised by a low response rate (25.7%). A comparison between respondents and non-respondents on selected demographic, behavioural, and attitudinal characteristics was conducted and no selection bias was observed, however, this comparison must be considered with caution as less than 4% of non-respondents took part in the follow-up interview. Highlighting the discrepancies in prevalence rates as a result of study design issues, Millar (1997) found that only 15% of Canadians 15 years and older reported consulting an alternative/complementary practitioner in the previous year. Using data from the National Population Health Survey (1994-1995) (Catlin & Will, 1992; Tambay & Catlin, 1995), Millar (1997) assessed the use of practitioners such as massage therapists, homeopaths, naturopaths, acupuncturists, and chiropractors. Regional differences in the use of alternative/complementary practitioners were apparent, with only 5% of respondents living in the Atlantic provinces reporting alternative practitioner consultations compared to 12% of respondents in Ontario, and 21% of respondents in both the prairie provinces (Manitoba, Saskatchewan and Alberta) and British Columbia. The discrepancy in prevalence rates found in this study compared to Eisenberg et al.'s studies (1998, 1993), the Angus Reid Group poll (1997) , and the Fraser Institute survey (Ramsay et al., 1999) can be attributed to differences in how ACTs were assessed (i.e., practitioner vs. therapies). In assessing only the use of alternative/complementary practitioners, Millar (1997) may have substantially underestimated the use of ACTs in Canada. Although the ACT prevalence rates obtained by Murray and Shepherd (1993) from a nonrandom convenience sample (N = 233) in the United Kingdom must be viewed with caution, the percentage of individuals who had used ACTs was similar to rates reported in the United States (Eisenberg et al., 1998; Eisenberg et al., 1993) and Canada (Angus Reid Group, 1997; Ramsay et al., 1999). Just over one third of men (34%) and 46% of women reported trying at least one form of ACT in the past 10 years, with manipulative treatments (i.e., massage), homeopathy, and acupuncture being the most commonly reported therapies. Despite being limited by a nonrandom, convenience sample of patients (N = 300) attending three outpatient medical clinics (diabetes, rheumatology, and chest), Grenfell et al.'s (1998) study of ACT use in the United Kingdom provides further insight into the use of ACTs within the general population. Approximately 68% of respondents were found to have used at least one type of ACT in the past 12 months, with acupuncture, homeopathy, herbal therapies, osteopathy, and Ayurvedic remedies being the most popular therapies. Prevalence of ACT use 16 varied by ethnicity, with Black (sic) (78%) and Asian (77%) patients reporting higher use of ACTs than White (sic) patients (53%). Caucasian patients were found to prefer acupuncture and homeopathy, while Blacks and Asians used more herbal therapies. The high prevalence of ACT use reported by participants was attributed to the population studied, which consisted of the chronically ill and had a high proportion of ethnic minorities. Further information about the number of recent immigrants within the sample would have been helpful in evaluating the high prevalence of ACT use. Despite numerous methodological limitations, past prevalence research has revealed ACT use to be a clinically significant occurrence within the general population. Use of therapies that are situated outside of the conventional medical system appears to be increasing, with physical therapies being most prominent within North American populations. Prevalence of Alternative/Complementary Therapy Use in Cancer Populations Compared to general populations, greater divergence in the prevalence of ACT use has been found in people living with cancer, with estimates ranging between 7% to 81% (Boon et al., 2000; Coss, McGrath, & Caggiano, 1998; Crocetti et al., 1998; Downer et al., 1994; Eidenger & Schapira, 1984; Ernst & Cassileth, 1998; Feigen & Tiver, 1986; Goldstein, Chao, Valentine, Chabon, & Davis, 1991; Lerner& Kennedy, 1992;M. Miller et al., 1998; Montbriand, 1995a; Oneschuk, Fennell, Hanson, & Bruera, 1998; Rees et al., 2000; Risberg, Kaasa, Wist, & Melsom, 1997; Risberg, Lund, Wistetal., 1995; Salmenpera, 2002; Swisher et al., 2002; Yates et al., 1993). This variability can be attributed, in part, to the methodological limitations previously outlined, including measurement and sampling issues. More specifically, Ernst and Cassileth (1998) cited the lack of specificity and inconsistent definitions of ACTs used by cancer patients and researchers as "contributing significantly to this variability" (p. 780). For example, by labelling all therapies that are received outside of conventional health-care settings as ACTs (e.g., self-help groups, counselling, home remedies, folk practices, and wellness regimens), prevalence rates of ACT use within cancer populations may have been artificially inflated (e.g., Montbriand, 1995a; Risberg, Lund, Wist et al., 1995; Risberg, Lund, Wist, Kaasa, & Wilsgaard, 1998). Conversely, defining ACTs as only those treatments used specifically to cure cancer (e.g., Cassileth, Lusk, Strouse, & Bodenheimer, 1984) may have significantly underestimated the prevalence rate. A further criticism of ACT use research within cancer populations has been the lack of differentiation between ACTs that are used as adjuncts to conventional cancer care and those that are used for curative purposes and in the place of conventional cancer treatment (Ernst & Cassileth, 1998). The latter type of ACTs are most often used by individuals who have removed themselves from the conventional health-care system, and as such, are rarely represented within prevalence statistics developed from clinically-situated samples. 17 Concern has been raised about the choice of study samples and the potential for selection bias within research exploring ACT use by individuals with cancer. The inclusion and exclusion of participants based on specific cancer diagnoses, stage of disease (i.e., palliative), and treatment experience (i.e., conventional vs. alternative/complementary) provides only a limited view of ACT use by individuals with cancer and restricts the generalizability of the findings. Further, the reliance on convenience samples within this area of research has made it difficult to reach conclusions about the prevalence of ACT use in cancer populations. Recently, Ernst and Cassileth (1998) attempted to provide a summary of existing prevalence data on ACT use within cancer populations. Table 2 represents a modified version of the published table that excludes non-English articles and pediatric research. Several recent studies not included in Ernst and Cassileth's (1998) review have been added. Very few of the prevalence studies conducted in the 1990s utilized random sampling techniques, the exceptions3 being Coss et al.'s (1998) survey of cancer patients in California and Lerner and Kennedy's (1992) national telephone survey of cancer patients in the United States. Similar prevalence rates were reported in both studies (6% and 9%, respectively), however, closer examination reveals several methodological limitations. Foremost, Coss et al. (1998) examined only the use of alternative/complementary practitioners, and as such, may have grossly under-estimated the use of ACTs implemented independently by individuals living with cancer. Secondly, the inclusion of proxy respondents, both friends and family members, in Lerner and Kennedy's (1992) survey raises questions regarding the accuracy of the findings, more so considering the potentially covert nature of ACT use (Balneaves et al., 1999; Eisenberg et al., 1998). Further, the terminology used by Lerner and Kennedy (1992) to describe ACTs (i.e., "questionable treatment methods") may have introduced a pejorative bias to the survey, resulting in the under-reporting of ACT use. More recent prevalence studies on ACT use in cancer populations have addressed some of the sampling concerns that have plagued studies in this area by employing random sampling techniques (Boon et al., 2000; Rees et al., 2000). Within Canada, the prevalence rate of ACT use within cancer populations has been reported to range from 7% to 81% (Boon et al., 2000; Eidenger & Schapira, 1984; Montbriand, 1995a; Oneschuk et al., 1998). Eidenger and Schapira (1984) conducted interviews with a volunteer sample of 315 cancer patients about whether unconventional treatments such as Laetrile, vitamins, or special diets could cure cancer. Only 7% of participants reported using "medications to treat their cancer other than those prescribed by their physicians" (p. 2739); however, given the question wording (i.e., "cure") used to assess ACT use, the estimated 3 Crocetti et al. (Crocetti et al., 1998) recruited women with breast cancer through the archives of the Tuscany Cancer Registry. Details regarding the recruitment and sampling procedures, however, are not sufficient to determine if random sampling was utilized. Table 2. Summary Data from Prevalence Studies on ACT Use in Cancer Populations Authors (Year) Sample (Country) Method Most Commonly Used ACTs Prevalence Arkko, Arkko, Kari-Koskinen, & Taskinen (1980) 151 consecutive outpatients with cancer (Finland) Personal interviews Herbs, vitamins, diets 56% of female and 30% of male patients had used ACTs Balneaves et al. (1999) Convenience sample of 54 women with breast cancer (Canada) Personal interviews with questionnaire Relaxation, meditation, vitamins/tonics, spiritual therapies 67% had used at least one ACT since diagnosis Begbie, Kerestes, & Bell (1996) 507 outpatients with cancer (Australia) Questionnaire Diets, psychological methods 22% had used some form of ACT Boon et al. (2000) Random sample of 422 women with breast cancer (Canada) Mailed questionnaire Vitamins/mineral, herbal products, green tea, special diet or foods 66.7% had used at least one ACT once in their life. Burke & Sikora (1993) 100 new consecutive patients with various cancers (UK) Personal interviews Counseling, meditation, relaxation, visualization 32% had used some form of ACT Cassileth et al. (1984) 304 inpatients with a variety of cancer diagnoses and 356 cancer patients of ACT practitioners (US) Personal interviews Metabolic treatments, diets, megavitamins, imagery, spiritual healing, immune stimulants 54% of patients receiving conventional medical treatment used ACTs Clinical Oncology Group (Anonymous, 1987) 463 outpatients with cancer (New Zealand) Personal Interviews Diets, vitamins, herbal products, laetrile 37% had sought advice about ACTs Coss etal. (1998) 503 randomly selected patients with a variety of cancer diagnoses (US) Telephone interviews Nutritionists, counselors, herbalists, massage therapists 6% had seen an alternative/ complementary practitioner Crocetti et al. (1998) 473 women with breast cancer identified through a cancer registry (Italy) Mailed questionnaire Homeopathy, manual healing, herbalism, and acupuncture 16.5% began using ACTs after diagnosis. 8.7% had used ACTs previously Downer et al. (1994) 600 patients with a variety of cancer diagnoses (UK) Mailed questionnaire and face-to-face interviews Relaxation, visualization, diets, homeopathy, vitamins, herbalism 16% had used or were using ACTs 19 TABLE 2. Summary Data from Prevalence Studies on ACT Use in Cancer Populations (cont.) Authors (Year) Sample (Country) Method Most Commonly Used ACTs Prevalence Eidenger & Schapira (1984) 315 consecutive patients with a variety of cancer diagnoses (Canada) Personal interviews Interview questions • focussed on Laetrile, vitamins, diets 7% had tried some type of ACT Feigen & Tiver (1986) 202 consecutive patients with a variety of cancer diagnoses (Australia) Questionnaire Vitamins, herbalism, protein supplements, naturopathy, faith healing, meditation 13% had used diet supplements; 13% had consulted ACT practitioners Goldstein et al. (1991) Convenience sample of 40 radiation cancer patients with a variety of diagnoses (US) Personal interviews Diets, metabolic therapy, mental imagery 12% were using ACTs while receiving treatment Lerner & Kennedy(1992) 5047 cancer patients (2855 of which were proxy interviews) (US) Telephone interviews Imagery, hypnosis, psychic therapy, diets, alternative drugs 9% had used at least one ACT in their lifetime Liu etal. (1997) 100 consecutive Chinese patients with advanced cancer (Taiwan) Questionnaire Chinese medicine 81% were using ACTs along with conventional medical treatment Montbriand (1995b) Convenience sample of 48 patients and 252 randomly selected patients with respiratory/digestive system cancers (Canada) Personal interviews Vitamins, minerals, herbal products 64% used ACTs Oneschuk et al. (1998) 143 consecutive patients with a variety of cancer diagnoses (Canada) Personal interviews Herbalism, vitamins, minerals 37% using ACTs Rees et al. (2000) Random sample of 714 women with breast cancer (UK) Mailed questionnaire Massage, aroma-therapy.chiroprac-tics, osteopathy, relaxation/medita tion, spiritual healing 31.5% had used at least one ACT since diagnosis 20 TABLE 2. Summary Data from Prevalence Studies on ACT Use in Cancer Populations (cont.) Authors Sample (Country) Method Most Commonly Prevalence (Year) Used ACTs Risberg, Lund, 252 patients with a Questionnaire Laying on of hands, 19.6% had used & Wist (1995) variety of cancer homeopathy, ACTs for their diagnoses herbalism, vitamins, cancer (Norway) diets Risberg, Lund, 642 patients with a Questionnaire Laying on of hands, 20% had used one Wist, et al. variety of cancer homeopathy, zone or more ACTs (1995) diagnoses therapy, herbalism, (Norway) diets, Nitter therapy, Iscador Risberg et al. 252 patients with a Questionnaire Faith healing, laying Prevalence (1998) variety of cancer with follow-ups at on of hands, increased over diagnoses. 60 4, 12, 24, 60 homeopathy, zone follow-up: 26% at month follow-up months and therapy, bergs, 4 months, 24% at with 110 patients. telephone vitamins, diets, 12 months, 18% at (Norway) interviews Iscadore, Nitter 24 months, and therapy 27% at 60 months. Salmenpera 229 women with Mailed Vitamins/minerals, 30% and 28% of (2002) breast cancer and questionnaire diets, natural health women and men, 209 men with products, spiritual respectively, had prostate cancer healing used ACTs (Finland) following diagnosis. Swisher et al. 113 women with Questionnaire Faith healing, 49.6% reported (2002) gynecologic therapeutic touch, using ACTS since cancers seen in imagery, meditation/ diagnosis; 46% of an outpatient clinic relaxation, herbal users ingested (US) products, vitamins/ some type of ACT; minerals 79% of users used a psychological or spiritual therapy Yates et al. Convenience Questionnaire Vitamins/tonics, 40% had used at (1993) sample of 152 meditation, relax least one ACT or patients with a ation, special foods, consulted at least variety of meta faith healing one ACT practi static cancer tioner for their diagnoses cancer (Australia) Note. Modified from Ernst, E. & Cassiletth, B.R. (1998). The prevalence of complementary/alternative medicine in cancer: A systematic review. Cancer, 83:777-782. prevalence rate may have been biased downward. Sampling has also been problematic in Canadian prevalence studies, with convenience sampling (Oneschuk et al., 1998) or mixed convenience and random sampling approaches (Montbriand, 1995a) being used to recruit 21 participants. An exception has been Boon et al.'s (2000) survey of breast cancer survivors in Ontario, which is described in greater detail in the following section on prevalence of ACT use in breast cancer populations. Risberg et al.'s (1998) study of ACT use in Norway is worthy of mention because of the longitudinal nature of the research. Two hundred and fifty-two individuals, with a variety of cancer diagnoses, were followed over a 5-year period to assess how ACT use changes over the cancer trajectory. Reported prevalence at each data collection time varied between 17.4% and 27.3%. The estimated cumulative risk of being a consumer of ACTs over a 5-year period was 45%. This research illustrates the potential limitations of cross-sectional studies of ACT use and suggests that prevalence rates of ACT use may fluctuate depending on the disease status of the population and the stage of the cancer trajectory. Further, Risberg et al. (1998) demonstrated that the majority of users (61%) started using ACTs three to four months following their cancer diagnosis. This finding emphasizes the importance of allowing several months to elapse after diagnosis prior to estimating prevalence rates within cancer populations. Longitudinal research is of particular interest in relation to ACT use because it provides information about the long-term implications of cancer patients' alternative or complementary treatment decisions on their health and the conventional and alternative health-care systems. Earlier work by Cassileth et al. (1984) has provided additional evidence of the impact of the cancer trajectory on ACT use. Of the 660 cancer patients interviewed from across the US, 64% sought conventional medical treatment first, 18% received conventional treatment and ACTs simultaneously, and 18% began using ACTs before accessing conventional medical care. Moreover, 77% reported using ACTs prior to beginning chemotherapy or radiotherapy. With 24% of the sample using metabolic therapies and 20% undertaking major dietary changes, the potential for interactions between conventional treatment and ACTs is of significant concern. Also of interest in this study was the fact that 43% of participants did not begin using ACTs until distant spread of their disease occurred, suggesting that individuals with advanced cancer may be a unique population with regard to attitudes and treatment behaviour. These findings, however, must be considered with caution because of the use of grouped data, in which patients from both conventional (N = 304) and alternative/complementary health-care systems (N = 356) were included in the study. Slight geographical differences in ACT use among cancer patients are apparent from previous prevalence research. In North America, metabolic therapies, special diets, and vitamins are the most commonly reported ACTs (Boon et al., 2000; Cassileth et al., 1984; Coss et al., 1998; Goldstein et al., 1991; Lerner& Kennedy, 1992). One exception is Montbriand's (1995a) study, in which the majority of cancer patients (71%) reported using physical alternative practices. A preference for physical ACTs was also reported in two Australian studies (Feigen & 22 Tiver, 1986; Yates et al., 1993). In contrast, research from Norway (Risberg, Lund, & Wist, 1995; Risberg, Lund, Wist et al., 1995; Risberg, Lund et al., 1998), the United Kingdom (Downer et al., 1994; Rees et al., 2000), and Italy (Crocetti et al., 1998) revealed a preference for spiritual healing, mind-body therapies (i.e., visualization), and traditional systems of healing (i.e., homeopathy, herbalism). For example, in Downer et al.'s (1994) exploration of ACT use by cancer patients receiving conventional treatment (N = 415), 65% of those individuals using ACTs practiced healing therapies and 25% used homeopathy remedies. Geographical differences in ACT use may reflect philosophical and health-care funding disparities between North American and European health-care systems. While the allopathic tradition has gained dominance within North America, many European countries have retained a pluralistic view of medicine (Lerner, 1994). Individuals from these countries may have better access to therapies from many traditions and may experience greater acceptance of their use of therapies that address more than physical needs. Such differences, however, may also reflect the lack of an internationally agreed upon definition of ACTs and should be interpreted with caution. Alternative/Complementary Therapy Use by Women Living with Breast Cancer While there is some evidence that the type of cancer has limited influence on the prevalence of ACT use (Lerner & Kennedy, 1992; Oneschuk et al., 1998; Risberg, Lund et al., 1998), the use of ACTs by women with breast cancer has been the subject of six recent studies (Balneaves et al., 1999; Boon et al., 2000; Crocetti et al., 1998; Gray et al., 1997; Rees et al., 2000; VandeCreek et al., 1999). In Crocetti et al.'s (1998) study of 473 women with breast cancer selected from a population-based Italian cancer registry, just over 16% of the participants reported using ACTs following diagnosis. The ACTs most commonly used were homeopathy (24%), manipulative therapies (i.e., massage) (16%), herbalism (14%), and acupuncture (7%). The majority of women reported using ACTs either independently of conventional treatment (30%) or following conventional treatment (30%), with 17% reporting simultaneous use of ACTs with conventional treatment. The women who used ACTs were found to be significantly younger, better educated, and had a previous history of ACT use, with the latter characteristic being the only independent significant predictor of ACT use. These findings must be interpreted with caution, however, because the respondents were significantly younger than non-respondents. Turning to the United Kingdom, Rees et al. (2000) undertook a population-based survey of 714 women who had been diagnosed with breast cancer between 1990 and 1996. Using a mailed questionnaire that had been pre-tested with outpatients at an alternative cancer therapy clinic, respondents were asked if they had ever visited or received therapy from one or more of a list of alternative practitioners. The women were also asked about their use of ACTs in the past 12 months and since their breast cancer diagnosis. Nearly 32% of the respondents had 23 consulted an alternative practitioner since diagnosis, with just over one half of these women reporting using more than one type of therapy. Over-the-counter ACTs (e.g., vitamins/minerals, herbal products) had been used by 33.2% of the women, with 14.8% using these ACTs since their breast cancer diagnosis. The most commonly used therapies in the previous year were massage/aromatherapy, chiropractics/osteopathy, relaxation/yoga/meditation, and spiritual healing. Turning to North America, Balneaves et al. (1999) explored the health beliefs and treatment practices of a convenience sample of 54 women living with breast cancer in Manitoba, Canada. In marked contrast to the findings of Crocetti et al. (1998) and Rees et al. (2000), over 67% of the participants reported using at least one ACT since being diagnosed with breast cancer. This discrepancy was attributed to several factors, including differences in ACT definitions, instrument design, and sample selection methods. The use of face-to-face interviews may have also facilitated rapport between the participants and the researcher and, as a consequence, increased women's willingness to report ACT use. The most frequently reported ACTs were meditation/relaxation therapies (60%), vitamins/tonics (57%), and spiritual/faith healing (53%). This research provides preliminary insight into the use of ACTs by Canadian women with breast cancer. In another Canadian study, Boon et al. (2000) used the Ontario Cancer Registry to obtain a random sample of 422 women who were diagnosed with breast cancer in either 1994 or 1995. Nearly 40% of the respondents reported visiting an ACT practitioner and an additional 62% reported using at least one ACT at sometime. Overall, 66.7% of women with breast cancer reported using some type of ACT at least once in their life.4 Interestingly, statistics specific to the use of ACTs following breast cancer diagnosis were not provided. Only 16.4% of ACT users in this sample reported that they were currently adhering "completely" to an ACT treatment regimen. Vitamins and minerals were the most commonly used therapy, with 13.4% of respondents reporting these supplements as being the only ACT they had ever used. Other common therapies included herbal products (including green tea, Essiac), special diets or foods, bodywork, and meditation. Additional support for the high prevalence rates reported by Balneaves et al. (1999) and Boon et al. (2000) was provided by VandeCreek, Rogers, and Lester (1999), who examined ACT use within 112 breast cancer outpatients randomly selected from a breast cancer clinic in the Midwestern United States. Using a pre-determined list of ACTs developed by Eisenberg et al. (1993), 91% of the sample reported using at least one of 18 therapies.5 The most frequently 4 Boon et al. (Boon et al., 2000) noted that the therapies reported by the women were "not necessarily in an attempt to treat their cancer" (p. 2517). 5 The researchers noted that in contrast to previous prevalence studies (e.g., Eisenberg et al., 1993), prayer and exercise were included as ACTs in the analysis. 24 reported therapies included prayer (84.5%), exercise (75.8%), and other spiritual therapies (48.3%). While researchers have questioned the inclusion of prayer within the alternative health care paradigm (Eisenberg et al., 1998; Eisenberg et al., 1993), the authors of this study suggested that the morbidity and mortality associated with breast cancer may increase the importance of spirituality in women's response to their disease. Similar findings related to the use of spiritual therapies have been reported in other cancer populations (Risberg, Lund et al., 1998). Overall, women spent approximately $42 per visit, with the mean number of visits reported to be 5.3.6 Similar to the research conducted in general populations, exploration of ACT use in cancer populations has been hindered by inconsistent conceptualization of ACT use, convenience samples, and selection biases. Accordingly, wide ranges of prevalence rates and therapies have been reported. Some evidence suggests the ACT use within cancer populations fluctuates across the cancer trajectory and is influenced by geographical location. While prevalence studies of ACT use in breast cancer populations have, in general, been better designed with regards to sampling and measurement issues, a variety of prevalence rates have been reported. In addition, the lack of distinction between women at different stages of the cancer trajectory (i.e., newly diagnosed, undergoing conventional cancer treatment, survivorship, and palliative) may have influenced the estimation of the prevalence of ACT use and the types of therapies reported. Characteristics of Consumers of Alternative/Complementary Therapies As the use of ACTs has become more prevalent within general and cancer populations, researchers have been interested in identifying those individuals most likely to use these types of therapies. In the following sections, the broad range of literature that describes the personal and social characteristics of ACT consumers within general and cancer populations, including the demographic profile of ACT users, and the influence of previous health experiences on ACT use, is discussed. Demographic Profile within General Populations Past empirical research has provided a consistent profile of ACT consumers. Sociodemographic factors found to be associated with complementary therapy use include age, gender, level of education, income, and health status (Eisenberg et al., 1998; Eisenberg et al., 1993; Murray & Shepherd, 1993; Thomas et al., 1991). The typical consumer of ACTs within the general population has been identified as female, under the age of 65 years, and with a high socioeconomic status. A similar profile has been found within the Canadian general population 6 The time period in which ACT use was assessed was not reported. 25 (Angus Reid Group, 1997; Berger, 1993; Blais et al., 1997; Kelner & Wellman, 1997a; Millar, 1997; Ramsay et al., 1999). There has been evidence supporting an association between ethnicity and the prevalence of ACT use (Eisenberg et al., 1998; Eisenberg et al., 1993; Grenfell et al., 1998). Eisenberg et al. (1993, 1998) reported that ACT use in the United States was least prevalent within the African-American population. In contrast, Grenfell et al. (1998) found ACT use in the United Kingdom to be more prevalent within Black [sic] (78%) and Asian (77%) populations than in Caucasian populations (53%). Limited investigation into the impact of immigration history on ACT use has been conducted (Hilton et al., 2001; Ma, 1999). Further epidemiological research is needed to clarify the nature of the relationship between ethnicity, immigration, acculturation, and ACT use, particularly within multicultural societies where popular, folk, and professional sectors of health care may overlap (Kleinman, 1988). While a demographic profile of ACT consumers has been revealed, recent evidence shows that the relationships between demographic characteristics and ACT use may be dissipating. In a secondary data analysis of Canada's National Health Population Survey 1994-1995 (Statistics Canada, 1995), Balneaves and Ratner (under review) found no significant relationships between gender, education, income, and ACT use. Both men and women from a variety of educational and socioeconomic backgrounds reported consulting alternative/ complementary practitioners. The lack of a distinct demographic profile was attributed to the growing salience, availability, and acceptance of ACTs within the Canadian health-care system (Balneaves & Ratner, under review). The recent Fraser Institute survey of ACT use (Ramsay et al., 1999) also found no significant relationship between reported annual income and ACT use in Canada. This research suggests that as selected ACTs gain legitimacy through scientific research, education, or widespread use and if they become accessible through insurance funding, ACT use may become prevalent across all segments of society. Demographic Profile within Cancer Populations Contradictory evidence of a distinct profile of ACT users also exists within cancer populations. In the majority of research studies, the typical consumer of ACTs is female, under the age of 65 years, and from a high socioeconomic class (Cassileth et al., 1984; Coss et al., 1998; Downer et al., 1994; Lerner & Kennedy, 1992; Oneschuk et al., 1998; Risberg, Lund et al., 1998; Swisher et al., 2002; Yates et al., 1993). As well, there has been some evidence to suggest that individuals with advanced cancer are more likely to use ACTs than patients with less advanced cancer (Lerner & Kennedy, 1992; Risberg, Lund, Wist et al., 1995). Preliminary research has also revealed significant associations among ethnicity and the prevalence of ACTs (Maskarinec, Shumay, Kakai, & Gotay, 2000) and the types of therapies used (Alferi, Antoni, Ironson, Kilbourn, & Carver, 2001; Lee, Lin, Wrensch, Adler, & Eisenberg, 2000). Contradictory findings exist, however, that challenge the influence of demographic characteristics on cancer 26 patients' use of ACTs. In a multi-centre study of Norwegian cancer outpatients, Risberg, Lund, et al. (Risberg, Lund, Wist et al., 1995) found no significant differences between users and non-users of ACTs with regards to gender or education. Cassileth et al. (1984) also reported no significant associations between gender, education, marital status, ethnicity, or stage of disease at diagnosis and ACT use by patients attending a conventional cancer facility. These discrepancies underscore the need for further study of the relationships between ACT use and demographic characteristics within cancer populations. Opposing views also exist regarding the underlying processes that explain the relationships between demographic characteristics and the use of ACTs by individuals living with cancer. For example, Lerner and Kennedy (1992) suggested that ACT use was more prevalent within higher income groups because of the out-of-pocket costs associated with ACT use in the United States. In contrast, Yates et al. (1993) attributed the effect of socioeconomic status on ACT use by Australian cancer patients to differences in health beliefs. Individuals situated in higher socioeconomic groups were found to hold more positive beliefs about alternative causes of cancer and the efficacy of ACTs than individuals reporting lower education and income levels. In addition, Yates et al. (1993) found younger cancer patients, who used ACTs more frequently than older individuals, to be more skeptical about conventional medical care. Thus, while the development of a profile of ACT consumers has been helpful in identifying potential users of ACTs in cancer populations, the demographic characteristics of ACT users may be more indicative of variations in health beliefs rather than treatment preferences. Further study is needed to clarify the associations between demographic characteristics, ACT use, and health beliefs. Within the breast cancer population, preliminary work has revealed associations between ACT use and selected demographic characteristics. In their study of 242 Italian women with breast cancer, Crocetti et al. (1998) found that women using ACTs were significantly younger, better educated, and had used ACTs prior to their breast cancer diagnosis. This study is unique compared to other epidemiological research on ACT use because it recognizes the inter-relationship among demographic characteristics; the proportion of highly educated women was reported to be significantly greater among younger women, along with previous use of ACTs (Crocetti et al., 1998). Balneaves et al. (1999) also reported a significant relationship between education level and the use of ACTs by Canadian women with breast cancer. The lack of additional significant associations between demographic factors and ACT use in this study reflects the limited power of this study resulting from the restricted sample size (N = 54). More recent studies of ACT use in breast cancer survivors have further supported the role of age, education, income, and cancer treatment history (Boon et al., 2000; Rees et al., 2000; Salmenpera, 2002). No theories regarding the cognitive or social processes through which 27 demographic characteristics influence treatment decisions specific to ACT use, however, have been put forth specific to women with breast cancer. In summary, past epidemiological research within general and cancer populations has provided contradictory evidence regarding the demographic profile of ACT consumers. Additional research with larger populations is needed to clarify the relationship between gender, age, socioeconomic status, chronic illness and the decision to use ACTs, with a special emphasis on the interrelationships among demographic characteristics and health beliefs in relation to the use of ACTs by women living with breast cancer. Previous Health Experiences Beyond demographic characteristics, evidence exists of a relationship between chronic disease occurrence and the use of ACTs (Balneaves & Ratner, under review; Kelner & Wellman, 1997a; Millar, 1997; Murray & Shepherd, 1993; Ramsay et al., 1999). In a recent Canadian survey of ACT use (Ramsay et al., 1999), between 60% and 71% of individuals reporting a variety of chronic conditions (e.g., back problems, arthritis, lung problems) had used at least one ACT in the past year. Millar (1997) also observed a progressive trend in ACT use in relation to chronic illnesses, with 26% of individuals having three or more chronic conditions reporting ACT use in the previous 12 months. This trend is of significance given the increasing prevalence of chronic conditions in an aging Canadian population. In addition to chronic illness being a potential predictor of ACT use, some support is found for an association between previous ACT use and the decision by individuals faced with cancer to use ACTs. For example, in Risberg, Lund, et al.'s (1995) exploration of ACT use among Norwegian cancer patients, individuals who had used ACTs prior to diagnosis were significantly more likely to use ACTs as part of their cancer care. Montbriand (1995b) also found previous use of ACTs to be predictive of ACT use by individuals with cancer, with 87% of patients with a high interest in ACTs reporting using ACTs prior to diagnosis. An association between ACT use and previous ACT use in women with breast cancer has been reported by Crocetti et al. (1998) and Rees et al. (2000). Additional research, however, is needed to understand the predictive value of illness history and previous ACT use with regard to the use of ACTs by cancer patients and the influence of health experiences on the treatment decision making process. Health Beliefs and Alternative/Complementary Therapy Use One area of ACT research that has received much attention in recent years has been the role of cognition in the decision to use ACTs. Of particular interest has been the relationship between beliefs about health and illness and the use of therapies beyond the scope of conventional medicine. In the following section, the potential associations between health 28 beliefs and ACT use within both general and cancer populations is discussed, with special emphasis being placed on beliefs reflecting perceived need for care (i.e., perceived risk). Health Beliefs within General Population Researchers have expressed interest in uncovering not only the demographic characteristics of ACT consumers, but also the attitudes and beliefs of individuals within the general population who use ACTs. In particular, extensive research has been conducted on the effect of lay perceptions of health and illness on the decision to use ACTs. The most prominent program of research has been led by Furnham et al. (Furnham, 1994; Furnham & Beard, 1995; Furnham & Bhagrath, 1993; Furnham & Forey, 1994; Furnham & Kirkcaldy, 1996; Furnham & Smith, 1988; Furnham, Vincent, & Wood, 1995; Vincent & Furnham, 1996, 1997), who have explored the relationships among lay health beliefs and the use of ACTs in the United Kingdom. Beginning in 1988, Furnham and Smith (1988) examined the health beliefs of a convenience sample of 87 individuals who were attending either a general practitioner or a homeopathic provider. Homeopathic patients were found to hold more negative beliefs about the efficacy of conventional medicine and more positive beliefs in the body's ability to heal. In contrast, patients receiving conventional medicine held more positive beliefs about the effects of conventional medicine therapies and, correspondingly, had more confidence in the care provided by general practitioners. No differences were found between the two patient groups with regard to perceived illness susceptibility. Although this research was limited in scope and could not be generalized to the larger population of ACT users, it provided preliminary support for health beliefs being antecedent variables associated with the decision to use ACTs. A follow-up study of homeopathic and conventional medical patients (N = 160) by Furnham and Bhagrath (1993) examined a broader range of health beliefs and behaviour. Controlling for the effects of age and income, homeopathic patients were found to be highly skeptical of conventional medical care and to hold strong beliefs about the tenets of homeopathy. These individuals were also found to be more aware of health issues than were conventional medical patients and more likely to believe in the role of lifestyle in preventing illness (i.e., stress reduction, relaxation and meditation techniques). These findings led to the hypothesis that individuals who use homeopathy have an intrinsically greater interest in health care than individuals attending general practitioners. This interpretation, however, must be considered with caution because participants' past medical history and current health status and the extent to which participants had used practitioners from different health paradigms were not taken into account. Furnham and Forey's (1994) research attempted to address, in part, these limitations by comparing the health beliefs of patients attending general practitioners and patients drawn from a variety of ACT providers (N = 160). ACT consumers were again found to be more cynical 29 about conventional medicine, perceived greater efficacy of ACTs, and had a higher health consciousness than individuals sampled from the offices of general practitioners. ACT and general practitioner patients, however, reported using treatments from both the conventional and alternative health-care systems. In both study groups, serious illnesses were brought first to the attention of conventional medical providers. This research is significant because it suggests that ACT users select therapies based on the nature of their illness and view ACTs as being on a continuum of care, which includes conventional medical care. Extending previous work, Furnham, Vincent, and Wood (1995) examined the unique beliefs held by conventional medicine patients (N = 58) and three groups of ACT consumers, including osteopathic (N = 65), homeopathic (N = 76), and acupuncture (N = 57) patients. This research provided further support for the presence of a continuum in treatment choice and, correspondingly, in health beliefs. ACT consumers and conventional medicine patients were found to vary in their beliefs about the quality of the doctor-patient relationship, the efficacy of conventional medicine and ACTs, and the importance of a healthful lifestyle. Those patients using acupuncture were found to be the least satisfied with their relationships with conventional care providers, to be the most skeptical of the efficacy of conventional medicine, and to place the greatest emphasis on lifestyle as a change agent in health than other ACT users. Also of interest was the association between chronic illness and ACT use, with acupuncture patients being most likely to have a chronic illness in comparison to other patients. Unfortunately, this research was limited by significant crossover between study groups in terms of treatment history. In a follow-up study, Furnham and Beard (1995) grouped study participants (N = 187) by therapy use rather than clinic attendance. Not only did this study explore the health beliefs of conventional, complementary, and alternative therapy patients, but also their general beliefs, such as coping styles7 and "Just World" beliefs".8 Although the sampling design (both convenience and random) and the restricted range of ACTs explored (acupuncture and Shiatsu) limited the generalizability of the findings, this research is important because it revealed that health-specific beliefs discriminated more clearly between treatment choice than general beliefs. Furnham and Beard (1995) suggested that the lack of significant associations between coping styles, "Just World" beliefs and ACT use illustrated the irrelevance of personality factors in the 7 Coping styles were measured using Miller's (1987) Behavioral Style Scale in which individuals are classified as being either monitors (actively seek information about a potential threat) or blunters (avoid relevant threat information). 8 "Just World" Beliefs were first introduced by Lerner (1965) and are based on the tendency of people to blame others for their own misfortunes. A 6-item scale, the "Just World" Beliefs measure assesses the extent to which individuals perceive the world to be orderly, stable, and just. In the context of Furnham and Beard's (1995) research, individuals who use ACTs are hypothesized to perceived the world as being less stable and just. 30 decision to use ACTs. However, they acknowledged that further research was needed to explore the relationship between trait characteristics and ACT use as well as the association between health beliefs and ACT use. It remains to be demonstrated if health beliefs not only lead to ACT use but are also a consequence of attending an ACT provider or independently using ACTs. Vincent and Furnham (1996) examined the motivation of ACT consumers to use therapies that were not offered by conventional medicine. Two hundred and sixty-eight patients from three ACT practices (acupuncture, osteopathy, and homeopathy) were asked to rank 20 potential reasons for seeking ACTs. Following factor analysis, five factors were identified in order of importance: a) a positive valuation of complementary treatment, b) the ineffectiveness of orthodox treatment for their chronic illness, c) concern about the adverse effects of orthodox medicine, d) concern about communication with physicians, and e) the availability of complementary medicine. Vincent and Furnham (1996) concluded that ACT consumers were being both "pulled" towards ACTs through positive beliefs about the efficacy and nature of ACTs and "pushed" by the failure of conventional medicine to address their chronic illness and by inadequate doctor-patient relationships. With 82% of the sample initially consulting a physician about their complaint, these researchers concluded that a "wholesale disillusionment" about conventional medicine was not apparent on the part of ACT consumers; rather, the decision to use ACTs was a reflection of the inadequacy of conventional medicine in treating chronic conditions. Vincent and Furnham (1997) have further explored beliefs about the efficacy of conventional medicine held by ACT consumers. Acupuncture patients (N = 82) were asked to complete a questionnaire about the perceived efficacy of selected ACTs for four categories of illness (major, minor, chronic, and psychological) and attitudes towards conventional medicine and health. Conventional medicine was seen by all participants as being more effective in treating major, life-threatening illness, such as heart disease and cancer. In contrast, ACTs were seen as being more effective in the treatment of chronic and minor illness. Not surprisingly, positive attitudes towards science were associated with a stronger belief in the efficacy of conventional medicine, while beliefs in the importance of psychological factors in health were associated with positive attitudes towards ACTs. Despite this research being limited to individuals undergoing acupuncture treatment, Vincent and Furnham (1997) suggested that health beliefs and their associated health behaviours were linked to wider beliefs about science and illness causation. More sophisticated research is needed, however, to examine not only the predictive power of health beliefs in relation to ACT use, but also the complex interactions among health beliefs, previous health experiences, demographic characteristics, and ACT use across a range of treatment options and within both general and disease-specific populations. 31 Influenced by the movement towards ACT effectiveness research that has pitted alternative paradigms against the biomedical hegemony present in Western health-care systems, Furnham has focussed his research program in recent years on the disparity in health beliefs between the two health-care systems. In a vignette-based study of the perceived efficacy of homeopathy and conventional medicine (Furnham & Bond, 2000), personal treatment history was found to play a significant role in how participants perceived the effectiveness of the different treatments. Those individuals with experience with ACTs perceived homeopathy to be more effective than conventional medicine, with the opposing trend found for individuals who attended a general practitioner. In a more recent study, Furnham (2002) examined the assertion that ACT use is representative, or an outcome, of an individual's rejection of empiricism and positivism (e.g., Beyerstein, 1997). Using beliefs about predicting the future9 as a proxy for "non proven" or postmodern belief systems, Furnham (2002) hypothesized that individuals interested and confident in the efficacy of ACTs would be more likely to believe in what he termed "future-ologies". Regression analysis revealed that attitudes towards and knowledge of ACTs were related to attitudes and knowledge about future-ologies, more so than attitudes towards science and conventional medicine. In other words, the higher the efficacy of ACTs was rated, the higher the efficacy of future-ologies was also rated. However, a factor analysis revealed that participants made clear distinctions between ACTs and future-ologies and were highly skeptical of the latter. Furnham (2002) concluded that individuals interested in ACTs might be more open and sympathetic to alternative belief systems than would be conventional medicine consumers. It could be argued, however, that the distinction between alternative and conventional health care systems is becoming more blurred with the increasing use of ACTs. For example, in the Fraser Institute survey (Ramsay et al., 1999), the majority of individuals who used ACTs (72%) believed that alternative and conventional healthcare should be used concurrently rather than individually. Further research is thus needed to examine the predictive capacity of health and general beliefs in determining treatment choices. The program of research developed by Furnham and colleagues has provided substantial evidence of the relationship between health beliefs and health behaviour within the general population. Individuals who used ACTs were found to be both pulled and pushed towards these therapies as a consequence of their positive valuation of ACTs and their dissatisfaction with conventional medicine. Being health conscious and placing greater emphasis on lifestyle and its influence on health was also found to be associated with the use of ACTs. ACT consumers, however, were not in a "flight from science," instead, ACTs were being used as part of a continuum of care that included both conventional and alternative health care. 9 Methods of predicting the future presented to participants included astrology, palmistry, tarot, clairvoyance, graphology, oriental astrology, dowsing, and necromancy. 32 In Canada, researchers also have examined the effect of health beliefs on ACT use within the general population. Using a comprehensive health care utilization model (Andersen, 1968; Andersen, 1995), Kelner and Wellman (1997a) interviewed 300 individuals attending five different practitioners: family physicians, chiropractors, acupuncturists, naturopaths, and Reiki practitioners. ACT users were found to not only share a unique demographic profile (female, younger, highly educated) and to have experienced more chronic illness, but they also had a greater sense of personal responsibility for their health. The association between health beliefs and ACT use was less clear. While some individuals using ACTs expressed confidence in the principles of ACTs, other individuals were motivated by more pragmatic reasons, including desperation. ACT consumers were found to use a mixture of treatments, both alternative and conventional. Kelner and Wellman (1997a) concluded that while some ACT consumers may ascribe to an "alternative ideology," others consider conventional medicine and ACTs to be part of a multidimensional health-care system. The existence of a continuum of care, ranging from conventional to alternative, was further supported by Kelner and Wellman (1997b), who compared the characteristics of patients across the five modes of treatment (family physician, chiropractic, acupuncture, naturopathy, and Reiki). The heterogeneity of ACT consumers was revealed in this research, with the differences between patients who used conventional treatment and those who used ACTs becoming more pronounced as therapies became more alternative. For example, individuals who used Reiki were found to be more educated and had higher incomes in comparison to other conventional and ACT consumers. Conventional and ACT consumers were also found to differ in terms of health profile, with individuals further along the continuum of care using ACTs for a broader range of health issues (i.e., chronic pain, emotional health, health promotion). Kelner and Wellman (1997b) further suggested that the continuum not only existed in terms of the profile of consumers, but also with regard to public recognition and institutional legitimacy of treatments. Researchers from the United Kingdom also have explored the association between health beliefs and ACT use within the general population (Murray & Shepherd, 1993). Despite its numerous design limitations, this research revealed the range in beliefs, health profiles and ACT use that exists in individuals attending ACT providers. Individuals who used ACTs were found to use a variety of ACTs for both minor ailments and for recurrent health problems, such as colds, headaches, and gastrointestinal disorders. Many ACT consumers expressed distrust of conventional medicine with regard to unknown long-term dangers and regarded ACTs as being "safe" and "non-invasive." However, it was interesting to note that non-users of ACTs reported lower consultation rates with general practitioners than ACT users. This finding provides further support for the proposition that the use of ACTs does not preclude conventional 33 medicine utilization but instead represents some of the many treatment options deemed appropriate and acceptable in today's health-care system. Health Beliefs within Cancer Populations Turning towards ACT use within cancer populations, researchers have explored the motivations of ACT use, including the association between health beliefs and health behaviour. In a study of 660 cancer patients in the United States who were receiving conventional medical therapy and ACTs, Cassileth et al. (1984) found significant differences between study groups with regard to their beliefs about illness and treatment. The majority of participants receiving ACTs, with or without conventional treatment, believed that their cancer could have been prevented, primarily through diet, stress management, and environmental changes. These individuals also perceived conventional cancer treatment, including chemotherapy and radiotherapy, to be more harmful than helpful and ACTs to be beneficial. Conventional treatment patients held opposing beliefs. Due to the retrospective nature of the study, it is difficult to determine whether the beliefs of ACT consumers were established prior to their illness or treatment, were a consequence of their exposure to alternative and complementary practitioners, or were an attempt to reduce cognitive dissonance. However, this research provided preliminary evidence of the unique belief systems of ACT consumers within cancer populations. Downer et al. (1994) interviewed 48 cancer patients in the United Kingdom who were using ACTs. When asked to explain their attraction to ACTs, over one half of the sample reported feeling more hopeful when using ACTs than when using conventional medicine alone. Forty-two percent were attracted to ACTs because of their belief in the non-toxic and "holistic" nature of the therapies. Twelve patients also reported that conventional medicine had been unable to offer any further treatment for their disease. This research points to the potential role of both positive beliefs in the efficacy of ACTs and hope in the decision to use ACTs. For some individuals living with cancer, ACTs may not only correspond to their health beliefs about treatment, but also may address a psychological need for optimism within their cancer care. Drawing from the health behaviour theories of Fabrega (1974) and Becker (1974), Yates et al. (1993) explored the effect of selected health beliefs on cancer patients' treatment decisions related to ACTs. In a study of 152 Australians living with advanced cancer, significant predictors of ACT use included a belief in "alternative" causes10 of cancer and being optimistic (operationalized as "will to live"). This research builds upon Cassileth et al.'s (1984) work and provides further empirical evidence of a relationship between beliefs about cancer etiology and treatment choice. The importance of maintaining hope in the face of a potentially life-threatening 10 Alternative causes of cancer included such statements as "My cancer was caused by pollution" and "My cancer was caused by stress". 34 illness was also emphasized as being a key motivator in the decision to use ACTs and points to a potential gap in the care provided to cancer patients through conventional medicine. Although this research was limited to individuals with advanced cancer, it provides preliminary evidence that ACT use within cancer populations may be unique with regard to the underlying motivations and health beliefs held by ACT consumers. Research by Risberg and colleagues (Risberg, Lund, & Wist, 1995; Risberg, Wist et al., 1998) also highlights the unique experience of cancer patients relative to their use of ACTs. In a comparative study of Norwegian patients with non-malignant (N = 305) and malignant disease (N = 252), patients without cancer expressed more positive beliefs about the potential benefits of ACTs in cancer care than individuals living with cancer (Risberg, Lund, & Wist, 1995). It was postulated that because of the personal implication of such beliefs to patients with malignant disease, cancer patients might be more circumspect in their attitudes towards ACTs. Compared to the previous work of Yates et al. (1993), this finding creates an interesting paradox because it suggests that while cancer patients may use ACTs to foster hope, they may be unwilling to place too much confidence in the potential benefits of ACTs. This difference, however, may reflect the greater knowledge held by cancer patients with regard to treatment of their disease and the potential benefits of ACTs. The discrepancy between Yates et al.'s (1993) and Risberg, Lund, and Wist's (1995) findings may also highlight the differences in beliefs among patients with early-stage versus late-stage cancer. Risberg, Wist et al. (1998) found additional evidence of a difference between cancer and non-cancer patients with respect to beliefs about the cause of cancer. Non-cancer patients were found to be more dogmatic in their beliefs about cancer etiology, expressing positive beliefs in the role of the environment and lifestyle in causing cancer. In contrast, cancer patients were more ambiguous in their beliefs about cancer etiology. This lack of certainty mirrors past research with cancer patients that has revealed similar ambiguity in ascribing meaning to and causation of cancer (Blaxter, 1983; Linn, Linn, & Stein, 1982; Yates et al., 1993). The uncertainty held by cancer patients about the cause of their disease may limit the explanatory power of beliefs about causation in predicting ACT use. Further research is needed to clarify the nature of the relationship between the use of ACTs and etiological beliefs. With regards to Canadian research on the health beliefs of cancer patients using ACTs, few studies have focussed on a broad range of cancer diagnoses. Gray et al.'s (1997) study, however, did examine the beliefs and attitudes of 32 cancer survivors, of whom 35% had been diagnosed with a malignancy other than breast cancer. For many of the respondents, their interest in ACTs was motivated by their desire to make certain that they were not missing any treatments that could be important in their recovery. Some respondents reported using ACTs to prevent a recurrence and to increase the likelihood of a healthy future. For many respondents, 35 however, their use of ACTs following diagnosis was simply an extension of their existing health beliefs and practices. Health Beliefs within the Breast Cancer Population Limited research has been conducted on the health beliefs of ACT consumers within the breast cancer population. Boon et al.'s (1999) qualitative study of the health beliefs of breast cancer survivors using ACTs (N = 36) revealed four main themes: survival, reacting to a bad experience with conventional medicine, prevention of further illness, and a belief that there was "nothing to lose." Many of the women believed that ACTs could enhance their chance of survival by "boosting the immune system," stabilizing their current disease, or preventing a possible recurrence. Further, the women cited negative experiences with conventional medicine, including side effects, as motivating them to seek ACTs. It was concluded that rather than being pulled or pushed towards ACTs, women with breast cancer were simply "hedging their bets" in an attempt to facilitate their future survival. A follow-up study by Boon et al. (2000) with 422 women diagnosed with breast cancer further revealed that ACT consumers were less likely to believe that conventional cancer treatments would cure their cancer, prevent a spread of disease, assist other treatments to work, boost their immune system, or be perfectly safe. Instead, women using ACTs expressed concern that conventional therapy had side effects and would weaken their bodies' natural reserves. In contrast, ACTs were perceived to be safer than conventional treatments and more likely to assist the body's natural forces to heal. In contrast to Boon et al.'s (1999; 2000) findings, Balneaves et al. (1999) failed to find significant associations between ACT use and health beliefs in 54 women living with breast cancer. Irrespective of treatment choice, the majority of women held positive beliefs about the outcomes of conventional care and the supportive nature of ACTs. Although these beliefs did not appear to influence treatment decisions, their co-existence highlights the open-mindedness and flexibility with which women living with breast cancer consider multiple treatment options and choose those most relevant to their health care and support needs. However, the study by Balneaves et al. (1999) was limited by its sample size, making it difficult to generalize to the larger breast cancer population and to rule out a Type II error11. Qualitative researchers have provided additional evidence of the consistency of health beliefs across treatment choices within the breast cancer population (Brown & Carney, 1996). In semi-structured interviews with 20 women with breast cancer who had used either conventional therapies alone or in conjunction with ACTs, women in both study groups 11 Type II error is defined as the failure to find a statistically significant association when one exists. 36 attributed their illness to environmental factors and believed recovery to be possible through healthful diets and lifestyle changes. Women using ACTs differed only in the emphasis they placed on stress as a causative factor in illness. Also of interest were the women's perceptions about the development of their health beliefs. Almost all of the women using ACTs described their health beliefs to be recently formed and perceived these beliefs to have been influenced by their experiences of having cancer. In contrast, women using only conventional medicine reported their beliefs to be "lifelong" and influenced primarily by family or religious experiences. This difference may be understood within the context that the women using ACTs had been living with cancer twice as long as women in the conventional medicine group, suggesting that beliefs about health and illness may change overtime. Brown and Carney (1996) postulated that as cancer becomes a more permanent part of a person's life, beliefs may shift to assist in the development of meaning within the illness experience. Thus, in order to explicate the relationship between health and illness beliefs and treatment choice, research is needed that encompasses the breast cancer trajectory, from early diagnosis to survival (5 year or 10 year). Perceptions of Need for Care A special kind of health belief that has received minimal attention within general and cancer populations has been the impact of individuals' perceptions of need for care on the decision to use ACTs. Need for care has been defined as both a subjective perception of risk (e.g., risk of cancer recurrence, quality of life) and an objective perception of need (e.g., cancer staging information) (Andersen, 1968; Andersen, 1995). Only one study has examined the specific effect of need for care factors on ACT use. In their research on the motivations of Canadian patients using ACTs, Kelner and Wellman (1997a) found that the majority of ACT consumers perceived their health problems to be serious (83%) and disruptive to their daily functioning (89%). In comparison, 67% of conventional medicine patients believed their condition was negatively affecting their daily lives. These findings suggest that as individuals' perceived need for care increases, they may be more willing to use therapies that are outside the conventional medical system. How need for care factors interrelate with other elements of the treatment decision-making process, such as health beliefs, remains to be determined. Further support for a potential link between ACT use and need for care has been provided by a number of studies that have reported a link between anxiety and ACT use. In particular, ACT consumers have been found to report higher levels of psychological distress than non-users (Burstein et al., 1999; Furnham & Bhagrath, 1993; Furnham & Smith, 1988). This difference has been attributed to ACT consumers' longer illness careers and greater physical dysfunction, resulting in more pronounced distress (Furnham & Smith, 1988). Montbriand (1995b) also reported an association between ACT use and anxiety, with those individuals expressing a high interest in ACTs being more likely to describe stress as a 37 response to their cancer diagnosis and treatment. A recent study of the psychological characteristics of 117 Austrian breast cancer outpatients using ACTs provided additional evidence of an association between distress and ACT use, with women using more than three ACTs tending to adopt a more depressive coping style than women who used fewer ACTs (Moschen et al., 2001). High users of ACTs were also found to report poorer emotional functioning than low users. Researchers have been cautious, however, in suggesting that the use of ACTs be used as a screening test for clinically significant distress and have encouraged further prospective, longitudinal research to examine the relationship between treatment choice and psychological distress (Burstein & Weeks, 1999). Contradictory evidence of the association between clinical evaluations and ACT use does exist, however, in the literature. For example, in Risberg et al.'s (1995) comparison of patients with malignant and non-malignant disease, cancer patients who used ACTs were more likely to be receiving palliative treatment than non-users, but their performance status12 was not significantly associated with ACT use. In conclusion, some empirical support exists for an association between perceived need for care and ACT use. In particular, perceived seriousness of a disease, chronicity of the illness, and anxiety have been implicated as motivating factors of ACT use within cancer and general populations. The lack of prospective research, however, has precluded researchers from conclusively determining the directionality of the relationship between psychological distress and ACT use. Possibly, commitment to a restrictive, detailed regimen of ACTs may result in excessive attention to illness and its outcomes, resulting in higher anxiety in high users of ACTs (Moschen et al., 2001). Further research is warranted to examine the potential psychological side effects of ACT use. Summary of the Role of Health Beliefs There is preliminary evidence to suggest that individuals who use ACTs as part of their cancer care may hold specific beliefs about their disease and the benefits of ACTs. ACT consumers within cancer populations may use ACTs as a means of maintaining hope and optimism, particularly when faced with advanced or terminal cancer. ACT use by individuals living with cancer may also reflect their dissatisfaction with or concerns about conventional cancer treatments. What is striking is that, with the exception of studies by Kelner and Wellman (1997a) and Yates et al. (1993), the research conducted to explore the role of health beliefs in predicting ACT use has occurred without direction from an explicit theoretical framework. Given the many well-established theories of health behaviour, this gap is surprising and may account 12 Assessed by the ECOG (European Cooperative Oncology Group) Performance Scale, which is used to assess how a patient's disease is progressing and how the disease affects the daily living abilities of the patient, as well as to determine appropriate treatment and prognosis. 38 for the broad array of health beliefs that have been tested. Research informed by theory would allow treatment decisions related to ACTs to be examined within the general rubric of health behaviour and allow comparisons with conventional health behaviour. While there is a danger of oversimplifying reality in the development and testing of theoretical models (Weiss, 1995), the application of health behaviour theory in ACT research would provide a tentative foundation for the prediction and explanation of this complex phenomenon. Other Factors Associated with Alternative/Complementary Therapy Use Beyond health beliefs, the role of other factors in the decision to use ACTs has been examined within both general and cancer populations. These factors include quality of life (Jordan & Delunas, 2001; Moschen et al., 2001; Paltiel et al., 2001), personality characteristics (Owens, Taylor, & Degood, 1999; Sturm, 2000; Sugimoto & Furnham, 1999), coping styles (Moschen et al., 2001), and conventional medicine utilization (Kaboli, Doebbeling, Saag, & Rosenthal, 2001). Two additional concepts that have received much attention within ACT research in cancer populations, and are the focus of discussion in this section, are social support and control. Social Support Within general populations, limited empirical work has explored the association between social support and ACT use. One exception has been Kelner and Wellman's (1997a) examination of the process through which ACT consumers gain information about therapies and practitioners. The majority of treatment and practice referrals were reported to be from family members, acquaintances, co-workers, and other ACT providers. Very few ACT consumers (3%) reported receiving a recommendation from their physician regarding ACTs. Overall, one half of all patients using ACTs cited referrals by others as the primary motivation underlying their choice in therapy. While treatment and practice referrals may be a questionable proxy measure of social support, this research points to the social context in which individuals make treatment decisions specific to ACTs. Additional evidence of the social context of treatment decision-making in cancer populations is apparent in the literature. Qualitative work by Truant (1998), Gray et al. (1997), and Montbriand (1995a) has highlighted the influence of family members, friends, cancer survivors, co-workers, and health food store workers on treatment decisions made by women with breast cancer who use ACTs. In fact, many women reported feeling overwhelmed with the amount of information provided by well-meaning individuals about ACTs (Truant, 1998). Particularly persuasive in encouraging ACT use have been members of cancer support groups, with several studies reporting support group participation as being an important determinant of ACT use (Boon et al., 2000; Gray et al., 1997; Lee et al., 2000). 39 The possibility of cancer patients being burdened by an abundance of information on ACTs led Risberg, Kaasa, Wist, and Melsom (1997) to inquire about the referral patterns of 126 patients in Norway. The majority of ACT consumers received their primary information about therapies from relatives and friend (64%), with 5% stating that the media was their main referral system. Only 14% reported feeling some pressure from significant others to use ACTs, with those aged 30-45 years reporting the greatest sense of obligation. Risberg et al. (1997) suggested that for cancer patients and their family members, ACT use might be one way of coping with the uncertainty surrounding illnesses such as cancer. Other researchers focussing on women with breast cancer have also found relatives and friends to be primary sources of information about ACTs (Crocetti et al., 1998), although how influential family and friends are on treatment decisions has been questioned (Salmenpera, Suominen, Lauri, & Puukka, 2001). Surprisingly, general practitioners were also identified as being an important source of knowledge regarding therapies not offered by conventional medicine (Crocetti et al., 1998). Earlier work by Lerner and Kennedy (1992) revealed the role of physicians in introducing the idea of ACTs to non-cancer and cancer patients, with 31% of participants identifying physicians as being their primary source of information about ACTs. This finding is surprising given the number of participants (61.5%) in Eisenberg et al.'s (1998) study who did not disclose their use of ACTs to their physician. These discrepancies point to the need for further research to make clear the distinction between social support and referral patterns. Based on available knowledge, Yates et al. (1993) conducted one of the few studies that has attempted to directly study the effect of social support on ACT utilization rates within cancer populations. Describing social support as "encouragement to use," a significant association between encouragement and the use of ACTs was demonstrated. Those individuals who had received support in using ACTs were four times more likely to use ACTs than those who had received no encouragement. Further analysis, however, failed to confirm the independent effect of encouragement to use ACTs on utilization. The limited research on the influence of social support on the decision to use ACTs by individuals with cancer provides contradictory evidence of an association between these two concepts. However, rich qualitative data have highlighted the involvement of significant others, especially family members, friends, and health professionals in treatment decisions, including the use of ACTs. Further study is required to elucidate the nature of the relationship between ACT use and social support and to determine the mechanism through which cancer patients learn about ACTs and make the decision to use these therapies. Exploration of the social and psychological consequence of giving and receiving encouragement to use ACTs is also warranted. 40 Control One variable that has attracted considerable interest as a potential motivator of ACT use within general and cancer populations has been the concept of control. Despite wide variation in conceptualization and instrumentation, the majority of research has supported a positive relationship between ACT use and the desire for control. How control is conceived or manifested by patients who use ACTs, however, has yet to be conclusively determined. The most focussed program of research on the association between ACT use and control has been developed by Montbriand (1995a; 1995b; Montbriand & Laing, 1991). In an ethnographic study of ACT use by 75 acute care patients in Canada, Montbriand and Laing (1991) uncovered three themes of control: 1) perceived control; 2) internal-external control; and 3) illusion of control. The first theme encompassed patients who perceived themselves as having the ability to escape or regain freedom from the control of biomedicine. For these individuals, ACT use represented a strategy through which they recovered control of their health and covertly separated themselves from the conventional health-care system. ACT use was also examined through the lens of Rotter's (1966) internal-external locus of control theory. Individuals who used therapies that were more psychological in nature appeared to keep control to themselves, illustrating an internal orientation. In contrast, individuals who used physical and/or spiritual therapies gave away their control to practitioners or to a "Higher Being", representing an external locus of control. Control was also found to fluctuate as patients progressed through the disease trajectory, suggesting control to be a flexible, rather than static, entity. Lastly, Montbriand and Laing (1991) drew inspiration from gambling studies to present ACT use as being an illusion of control, in which patients discount the influence of chance on their health outcomes. Such conceptualization lends support to an additional theme, desire for control, which has been the subject of much research within cancer populations (Degner & Russel, 1988; Degner & Sloan, 1992; Degner, Sloan, &Venkatesh, 1997; Hack, Degner, & Dyck, 1994). In the face of an uncontrollable disease, such as cancer, individuals may seek to increase their chance of survival and control of their illness through the use of ACTs. In considering the larger body of literature on ACT use within general and cancer populations, control has been linked consistently with the decision to use ACTs. In one of the first examinations of cancer patients using conventional medicine and ACTs, ACT consumers were found to be more likely to accept an active role in their health care than patients using only conventional medicine (Cassileth et al., 1984). The personal responsibility held by ACT consumers was reflected in their choice of therapies that were life-style oriented rather than practitioner-controlled. While this research did not specifically consider control as an intervening variable in the. decision to use ACTs, it provided preliminary evidence of an association between desire for control and health behaviour. 41 In a survey of 125 outpatient oncology patients, Hiratzka (1985) found that individuals who held positive attitudes towards ACTs and had greater knowledge of ACTs scored higher on the internal locus of control subscale of the Multi-Dimensional Health Locus-of-Control Scale (Wallston, Wallston, & DeVellis, 1978). Despite the criticisms of this scale for it being too static and limiting (Furnham & Beard, 1995), Hiratzka (1985) concluded that individuals with an internal orientation were more likely to seek out and to use information about ACTs than individuals holding a "powerful others" or "chance" orientation. The implications of this research, however, are limited because only attitudes towards and knowledge about ACTs were examined rather than actual treatment utilization. Furnham (Furnham & Bhagrath, 1993; Furnham & Forey, 1994; Furnham & Smith, 1988) extended Hiratzka's (1985) research by exploring the association between health locus of control and ACT use within the general population. Using Lau and Ware's (1981) health-specific locus of control scale, Furnham and Smith (1988) found homeopathic patients to have significantly lower scores on the provider control subscale than patients attending a general practitioner. This difference was attributed to homeopathic patients' general dissatisfaction with, and distrust of, conventional medicine. In a follow-up study of a larger sample of homeopathic and conventional medical patients, a significant difference between the two study groups on the health-specific locus of control scale was reported (Furnham & Bhagrath, 1993). This difference, however, was found on the internal locus of control subscale rather than the provider control scale. Homeopathic patients were found to believe more strongly in their ability to control their health than conventional medicine patients. This belief was attributed, in part, to homeopathic patients' disappointment with conventional medicine, which pushed them towards greater responsibility for their own health. Furnham and Forey (1994) provided further support for the association between ACT use and the desire for personal control over health decisions and less reliance on conventional medicine practitioners. More recent work, however, has challenged the hypothetical relationship between locus of control and ACT use, with significant differences among ACT and conventional medicine consumers being attributed to demographic differences (i.e., age and illness history) rather than control orientation (Furnham et al., 1995). This discrepancy emphasizes the need to consider the interactions among personal, social, and cognitive factors in conducting ACT utilization research. Yates et al. (1993) improved on previous research by exploring the predictive value of the concept "need for control" with regards to ACT use by cancer patients. Need for control was conceptualized as being one's attitude towards having control over decisions about one's cancer and its treatment. ACT users were found to express a stronger desire for control over health-care decisions in comparison with individuals using only conventional treatments. In contrast to Furnham et al.'s (1995) findings, need for control was found to be an independent 42 predictor of ACT use within this population. The validity of Yates et al.'s (1993) findings, however, was tempered by the use of an investigator-developed measure of control.13 Balneaves et al. (1999) addressed this limitation by using a validated measure of desired control (i.e., Control Preference Scale, Degner & Sloan, 1992; Degner et al., 1997)14 in their study of ACT use in women with breast cancer. The findings of this research, along with Truant's (1997) ethnographic study of the decision-making process of women with breast cancer who use ACTs, are discussed in greater detail in the following section on treatment decision making. In summary, research within general and cancer populations has provided preliminary evidence of an association between control and the use of ACTs. The lack of consistency in the definition and measurement of control has made comparisons across studies difficult and has highlighted the need for more definitive theory development. Determining whether control is best operationalized as a desire for control, an internal/external orientation, or an illusion of control will contribute significantly to research exploring the relationship between ACT use and control. Examination of the role of control in relation to other health beliefs would further current understanding of the complexities associated with treatment decisions specific to ACTs Treatment Decision Making by Women Living with Breast Cancer This section provides a concise summary of research examining treatment decision making by women living with breast cancer. Women's experiences and preferences related to conventional treatment decision making are examined, focussing on women's preferred role in decision making, demographic and cognitive factors associated with treatment decision making, and the ways in which treatment decisions are made. The limited literature on treatment decision making within the context of ACT use also is examined. Conventional Treatment Decision Making The majority of studies examining the conventional treatment decision-making process of women with breast cancer has focussed on what role women want to play in their treatment decisions (Beaver et al., 1996; Bilodeau & Degner, 1996; Degner, Kristjanson et al., 1997; Degner & Sloan, 1992; Hack et al., 1994; Kenny, Quine, Shiell, & Cameron, 1999; Street & Voigt, 1997). The findings have been contradictory, with some researchers suggesting that women with breast cancer prefer to take on an active or collaborative role in their treatment decisions (Degner, Kristjanson et al., 1997; Keating, Guadagnoli, Landrum, Borbas, & Weeks, 13 Yates expressed concern regarding the reliability of the 2-item Encouragement to Use ACT measure and suggested that a more reliable measure of control should be used in future ACT research. (P. Yates, personal communication,. October 31, 1994) Degner & Sloan's (1992) Control Preference Scale has been well validated through a series of studies and across a range of populations (Beaver et al., 1996; Bilodeau & Degner, 1996; Degner, Kristjanson et al., 1997; Degner & Sloan, 1992; Degner, Sloan et al., 1997). 43 2002; Mastaglia & Kristjanson, 2001) and other researchers suggesting that when faced with a life-threatening diagnosis such as breast cancer, many women prefer to defer treatment decisions to their physicians (Beaver et al., 1996; Bilodeau & Degner, 1996; Degner & Sloan, 1992). This discrepancy, along with data that suggests that being offered an active role in treatment decision may either positively (Keating et al., 2002; Street & Voigt, 1997) or negatively (Fallowfield, Hall, Maguire, Baum, & A'Hern, 1994; Reaby, 1998) influence women's physical and psychological well being, has emphasized the importance of individualizing decision-making strategies. The need for improved communication between health-care providers and women with breast cancer regarding expectations of the decision-making process has been further supported by research identifying discrepancies in women's preferred and actual role in treatment decisions (Bilodeau & Degner, 1996; Degner, Kristjanson et al., 1997; Keating et al., 2002). For example, in Bilodeau and Degner's (1996) survey of 74 women newly diagnosed with breast cancer, only 19% of women preferring a collaborative role were able to assume such a role in the clinical setting. While barriers to achieving a more active role in decision making have been identified (e.g., lack of information, lack of time, physical and mental distress, poor patient-provider communication skills (Reaby, 1998; Sainio, Eriksson, & Lauri, 2001), no known studies have examined the difficulties faced by women preferring a more passive role in treatment choice in achieving their desired level of involvement in treatment decisions. Factors that have been significantly associated with the preferred decisional role of women with breast cancer include age, education, and income level (Beaver et al., 1996; Bilodeau & Degner, 1996; Degner & Sloan, 1992; Hack et al., 1994; Pierce, 1993). In general, young women with a high socioeconomic status have been found to prefer active roles in treatment decisions. In contrast older women and those from lower socioeconomic groups have preferred less control in their treatment choices. The amount of variance in preferred decisional role explained by demographic factors, however, has been of limited clinical significance (Beaver et al., 1996). Despite the controversy over the role of women with breast cancer in treatment decisions, researchers have developed some insights into the types of information women would like during the decision-making process. Hack et al. (1994) examined the information needs of 17 women diagnosed with stage I or II breast cancer who were two to six months post-diagnosis. Women who preferred an active role in choosing their breast cancer treatment were found to desire detailed information regarding their diagnosis, treatment alternatives, and treatment procedures. All women, notwithstanding their preferred decisional role, were found to want information about the side effects of treatment and their prognosis. The findings of this study, however, are limited by the small sample size. In a larger study of 74 women diagnosed 44 with breast cancer (Bilodeau & Degner, 1996), information needs included (listed in order of perceived relevance): stage of disease, likelihood of cure, treatment options, physical and emotional side effects, risk to relatives, impact on social activities, self-care issues, and sexuality. Women's ability to assimilate technical information about their diagnosis, treatment, and prognosis and physicians' ability to communicate such information in an understandable manner, however, has been questioned (Kenny et al., 1999). In terms of the actual decision made with regards to breast cancer treatment, the majority of studies has focussed on women's decision to have either mastectomy or breast-conserving surgery (i.e., lumpectomy) (Graling & Grant, 1995; Mastaglia & Kristjanson, 2001; Stafford, Szczys, Becker, Anderson, & Bushfield, 1998; Street & Voigt, 1997; Ward, Heidrich, & Wolberg, 1989). The proportion of women choosing breast-conserving surgery over mastectomy has varied considerably in the literature. Factors found to discriminate between surgical choice include age, education, income level, geographic residency, and physician preference (Graling & Grant, 1995; Hughes, 1993; Mastaglia & Kristjanson, 2001; Stafford et al., 1998). Young urban women with high socioeconomic status were found to prefer lumpectomy to mastectomy. These differences raise interesting questions regarding women's perceptions of risk and the perceived availability of breast cancer treatment based on financial and travel considerations. Conflicting results have been found regarding the role of information and physician communication in women's surgery decisions (Hughes, 1993; Mastaglia & Kristjanson, 2001). Few studies have examined the actual decision-making process of women with breast cancer. One exception is Pierce (1993), who used grounded theory to explore the decision making experiences of 48 women newly diagnosed with early-stage breast cancer. Three patterns of decision making were uncovered: a) deferrer; b) delayer; and c) deliberator. Just over 40% of the women were classified as being "deferrers", which was defined by the women's lack of conflict over treatment options, their preference to defer to their physician, and limited deliberation. For many of these women, they did not perceive themselves as having a treatment choice. In contrast, 44% of the women were classified as "delayers", who vacillated between treatment options and had difficult distinguishing between the benefits and costs of the different choices. These women often made a decision using the "first difference rule", in which the treatment option that had the first detectable benefit over the other treatments was chosen. For only 15% of the women, a "deliberator" role in decision making was identified. These women accepted a personal responsibility for their decision and developed a structured deliberation plan to sort through the available treatment options and to seek information. Although this latter pattern of decision making was closest to the normative models presented in the decision making literature (Janis & Mann, 1977; Slovic, Fischhoff, & Lichtenstein, 1977; Tversky & Kahneman, 1981; Tversky & Shafir, 1992), women who were deliberators experienced the 45 greatest psychological distress in making their treatment decisions compared to women who used other decisional styles. Pierce (1993) was cautious in judging one decision-making style as being better and encouraged long-term follow-up to determine women's satisfaction or regret with their treatment decisions. In summary, research on treatment decisions by women with breast cancer has revealed that while some women may prefer an active role in the decision-making process, other women may be more comfortable in being less involved in treatment choices. That the decision-making experience is stressful for all women, notwithstanding the preferred decisional role, is apparent and emphasizes the need for clear communication regarding treatment options and the decision-making process. Health-care providers must also provide sufficient information to allow those women interested in having more control over their treatment decisions to make informed choices. Although the role of control in treatment decision making has been well studied, further research is warranted on how women with breast cancer make decisions and what are other influencing factors. Treatment Decision Making and Alternative/Complementary Therapies To date, few studies have been conducted on the decision-making process of consumers of ACTs, particularly individuals diagnosed with cancer. Exceptions include Montbriand (1995b), Truant (Truant, 1997; Truant & Bottorff, 1999), and Balneaves et al. (1999). Although much of this work appears to overlap the research previously discussed on the association between control and ACT use, these studies are discussed here in the context of treatment decision making. Given the body of research that has focussed on the role of control in the treatment decisions of women with breast cancer (Bilodeau & Degner, 1996; Degner, Kristjanson et al., 1997; Degner & Sloan, 1992; Hack et al., 1994; Mastaglia & Kristjanson, 2001), it is not surprising to find such commonality. In the first study mentioned, Montbriand (1995a) found the theme of control interwoven throughout her examination of cancer patients' decisional strategies specific to ACTs. The first phase of her study involved the development of a decision tree based on interviews with 48 individuals diagnosed with respiratory or digestive system cancers. The decisional tree consisted of 21 questions that differentiated individuals who expressed high interest in ACTs, regular interest in ACTs,15 or were interested in biomedical treatments only. The main themes of the decision tree included the following: (1) preferred treatment methodology, (2) secrecy about alternatives, (3) social group influence, (4) considerations of cost, (5) perceived stress, (6) desired decisional control, (7) judgements about cure, (8) change and searching, and (9) faith in 15 High and regular interest groups were differentiated by the high interest respondents intending on using ACTs as a consequence of their cancer diagnosis and regular interest respondents continuing to use those ACTs they had been using prior to diagnosis. 46 the practice. The decision tree's predictive value was tested in the second phase of the study, in which 252 cancer patients were interviewed. Overall, 90.4% of respondents were correctly classified using the decision tree, with 22% being high interest group members, 48% classified as regular interest group members, and 29.8% responding in a manner consistent with the biomedical only group. Those individuals included in the high interest group were more likely to be young and female, have a more severe cancer diagnosis, and report a higher level of education and income than those in the biomedical only group. High interest members were also more secretive about their use of ACTs, received support from their social group to use ACTs, wanted a high level of decisional control, and held health beliefs supportive of alternative treatment paradigms. Interestingly, high interest members perceived stress as being a motivating factor in their decision to use ACTs. With regards to control, Montbriand (1995a) found high interest members' desire for control over health care was associated with several factors, including social, cultural, and economical influences. She suggested that making the decision to use ACTs allowed some cancer patients to express their independence from the conventional health-care system. The amount of variance in ACT use explained by preferred role in treatment decisions, however, was not clear from the model developed by Montbriand (1995a). Further predictive work is needed to model the concept of control along with other health beliefs and sociobehaivoural variables in relation to the decision to use ACTs. Qualitative work by Truant (Truant, 1998; Truant & Bottorff, 1999) has provided further rich description of the dynamic nature of treatment decision making in women with breast cancer who use ACTs. Central to the decision-making process was the issue of control. The women (N = 16) who participated in this grounded theory study described their use of ACTs as being a means though which control could be regained. The motivations underlying the need to regain control, however, varied as women moved through the cancer trajectory. Newly diagnosed women expressed a desire to reduce their feelings of loss and to take action. These women reported making quick, and often uninformed, decisions about ACTs before beginning conventional treatment. For many women at this point in their disease, decisions related to conventional cancer treatments took precedence over decisions related to ACTs. Following surgery, a desire to take back control from the conventional health-care system and to increase control over the cancer and one's health became paramount. Women at this point in the trajectory experienced less anxiety and were able to take more time to research treatment options and to chose therapies that fit best with their lifestyle and belief systems. Beyond increasing a sense of control, ACTs were also used to manage the side effects of conventional cancer therapies, boost the immune system, promote well being, and prevent a recurrence of breast cancer. Once conventional treatment had ended and the women began to regain a sense of "normality" in their lives, they continued to use ACTs in an attempt to enhance their illusion or 47 perception of control. In using ACTs, women felt they were continuing to exert some measure of control over their cancer and general health. Although all women using ACTs reported taking an active role in their decisions related to ACTs, they described more passive roles in the treatment decision-making process within the conventional health-care system. This finding, in part, contradicts past research on women's preferred role in treatment decisions (Bilodeau & Degner, 1996; Degner & Sloan, 1992) and suggests that desire for control may fluctuate not only across the cancer trajectory but also across treatment decisions. The decision to use ACTs may be one way in which women with breast cancer are able to play a more active role in treatment decisions while maintaining their relationships with conventional health-care providers. Balneaves et al. (1999) also examined ACT decision making from the perspective of control by using Degner and Sloan's (1992; 1997) Control Preference Scale to determine the preferred decisional role of women with breast cancer using ACTs. Of the 52 women interviewed, the majority of ACT users (94%) were found to prefer an active or collaborative role in treatment decision making, whereas only 56% of women using only conventional medicine were found to prefer similar roles. Similar to Truant and Bottorff (1999), Balneaves et al. (1999) concluded that the decision to use ACTs may be an important mechanism through which women with breast cancer gain control, knowledge, and responsibility over an illness that appears, at times, uncontrollable. The research on treatment decisions related to ACT use in cancer populations has provided insight into the role of control as both a motivator and an outcome of the decision to use ACTs. The influence of other factors, including stage of disease and conventional treatment, psychological distress, and social support on ACT decisions has been also suggested. Additional decision-making research is needed, however, that examines the variability of ACT use beyond a "yes-no" dichotomy and the role of both social and cognitive factors in influencing the manner in which ACTs are used. Chapter Summary This chapter has reviewed and critiqued the two main areas of literature: alternative and complementary therapy use in general and cancer populations and treatment decision making in women living with breast cancer. With regards to the ACT literature, three major areas of ACT research within general and cancer populations were reviewed: the prevalence of ACT use, the characteristics of ACT consumers, and the social and cognitive motivations of ACT use. Although lacking in concise and consistent terminology, this body of literature suggests ACT use to be a growing health care phenomenon within industrialized nations. The increase in the use of ACTs within general populations has been mirrored within cancer populations, with a limited body of literature 48 suggesting that ACTs play a significant and integral role in the experiences of women living with breast cancer. Further research is needed to develop more specific and reliable statistics on the prevalence of ACT use within breast cancer populations. The literature on the characteristics of ACT consumers within general and cancer populations has provided inconclusive evidence of associations among personal and social factors and the use of ACTs. Gender, age, socioeconomic status, health experiences, and health beliefs have all been implicated as predisposing factors in the decision to use ACTs. The literature has also pointed towards potential associations between demographic characteristics, health beliefs, and the use of ACTs by individuals living with cancer. These preliminary findings underscore the importance of moving from descriptive research to inquiry that is grounded in theory to explain and understand the potential interrelationships among sociodemographic characteristics, personal belief systems, and ACT use within cancer populations. In particular, the research on ACT use by women with breast cancer has raised questions regarding the role of health and illness beliefs in the decision to use ACTs and the interactions among beliefs and sociodemographic factors, and ACT use. Research on the utilization of ACTs has also revealed several factors that potentially motivate individuals to search for treatment alternatives outside of the conventional health-care system. Despite the lack of prospective, longitudinal research, strong associations have been identified between social support and control and the use of ACTs. The limited research exploring the motivations of ACT use in women with breast cancer has provided some support for these factors; however, further study is required to fully elucidate the nature and quality of the relationships among these concepts and women's use of ACTs. Greater specificity is also required regarding the context within which ACTs are used in breast cancer populations. While the majority of prevalence research has considered ACT use to be singular in purpose (i.e., curative), qualitative work has suggested that ACTs are used more discretely throughout the breast cancer trajectory (Truant, 1998). Research on treatment decision making has shed further light on the way in which women with breast cancer make decisions and what social and cognitive factors may affect their treatment decisions. The desire for control has been found to be an important variable in women's decision-making process, with treatments such as ACTs representing a means of regaining control from the biomedical establishment and asserting one's independence. Much of the work on treatment decision making specific to ACTs has occurred, however, without a strong theoretical framework. By examining the treatment choices of women with breast cancer within an accepted and recognized cognitive framework, those factors associated with treatment decisions related to ACTs are uncovered. 49 In summarizing the literature, ACT use has been described as a consequence of sociobehavioural (i.e., demographic factors, previous experience, social support) and cognitive factors (beliefs about health, illness, and control). The descriptive nature of past research, however, has hindered attempts to test the interrelationships among these factors and ACT use, and examine ACT use within a recognized theory of health behaviour. The proposed research will address these limitation by developing and testing three causal models of ACT use by women with breast cancer, focussing on the personal, social and cognitive factors that have been implicated in the literature as being associated with ACT use. The following chapter delineates more fully the proposed models of ACT use and the theoretical underpinnings of the hypothesized relationships. 50 Chapter 3 Theoretical Models Despite the existence of a vast body of literature describing the use of alternative/ complementary therapies (ACTs) by general and cancer populations, few researchers have attempted to examine ACT use within the context of a theoretical framework of health behaviour. This has resulted in ACT use being presented in a disjointed and fragmented manner that has prevented the causal relationships between sociodemographic characteristics, attitudes, health beliefs, and the use of ACTs from being fully elucidated within a comprehensive theory. Without this knowledge, health professionals are unable to understand which individuals are most likely to use ACTs, their motivations for such health behaviour, and the context in which ACT use is most prevalent. The purpose of this study is to develop and test three hypothetical models of ACT use by women with breast cancer. Using key constructs derived from the Health Belief Model (Janz & Becker, 1984; Maiman & Becker, 1974; Rosenstock, 1974b), the use of ACTs by women who had been diagnosed and treated for breast cancer was examined within the context of preventive, ameliorative, and restorative health behaviour. In developing and testing three distinct models of ACT use, the purpose underlying women's use of therapies beyond conventional medicine and the role of health beliefs in motivating ACT use across a range of health contexts were examined. The following discussion provides an overview of the most well known models of health behaviour, including the Health Belief Model, as well as a detailed description of the three hypothetical models of ACT use that were tested in this study. Empirical support for the postulated models and structural relationships is also provided. Models of Health Behaviour Since the 1960s, numerous sociobehavioural models have been developed in an attempt to further understanding of the decision-making process underlying the use of selected 51 health services. In this section, a brief outline of three of the most familiar models of health behaviour is provided, describing how each model conceptualizes health action. The advantages and disadvantages of each model also are presented. Behavioural Model of Health Care Utilization One of the first models of health behaviour to be advanced was Andersen's (1968; 1995) behavioural model of health care utilization. Originally developed to explain the use of health services by families, the model has been applied across a variety of behaviours, including utilization of conventional health-care services (Barrilleaux & Miller, 1992; Bazargan, Bazargan, & Baker, 1998; Cohen, 1993; Potvin, Camirand, & Beland, 1995), self-care activities (Fleming, Giachello, Andersen, & Andrade, 1984; Fosu, 1989) and the use of ACTs (Kelner & Wellman, 1997a). The behavioural model conceptualizes health behaviour as being a product of complex interrelationships among predisposing (personal and social structures), enabling (resources and barriers to care), and need-for-care (perceived risk and need) factors (Andersen, 1968; Andersen & Newman, 1973; Andersen, 1995). Of these factors, perceived need-for-care has been found to be the strongest predictor of health behaviour (Andersen, 1995). While ambitious in its attempt to account for total health service utilization, Andersen's model has been described as a theoretical framework of personal and social factors rather than an actual theory of health behaviour (Kirscht, 1974). The model lacks specificity and is, therefore, restricted in its application to exploring the social cognitions (i.e., health beliefs) associated with health-care decisions. Health Locus of Control In an attempt to better explain and predict health behaviour at the social cognitive level, several theories have been developed using the construct of locus of control. Defined as the degree to which individuals perceive events in their lives to result from their own actions and therefore controllable (internal control), or as being unrelated to their own behaviour and beyond their control (external control) (Lefcourt, 1976), locus of control originated within the traditions of social learning theory (Rotter, 1966; Wallston et al., 1978). Social learning theorists conceptualize behaviour, including health action, to be a function of both an individual's expectancies about the consequences of one's actions and ability to perform a behaviour and the value of the expected outcome (Oberle, 1991; Rosenstock, Strecher, & Becker, 1988). Wallston et al. (1978) further refined the locus of control construct by developing the multidimensional health locus of control (MHLC) scale, which measures the degree to which individuals perceive their health to be a consequence of their own action, the action of powerful others, or chance. Despite wide application to a variety of health behaviours (Rosenstock et al., 1988; Zindler-Wernet & Weiss, 1987), health locus of control theories have been able to explain only a minimal amount of variance in health behaviour (Conner & Norman, 1996). The theories 52 have been criticized for their lack of stability across health behaviours and their failure to incorporate perceived value of health. Theory of Reasoned Action The theory of reasoned action, also known as the theory of planned behaviour, has extended the scope of social cognitive theories by examining not only personal beliefs and attitudes towards health, but also the social norms associated with health behaviours (Ajzen & Fishbein, 1980; Lauver, 1992). Within this model, health behaviour is proposed to be a consequence of one's intentions, or health motivations, which are in turn derived from beliefs about the perceived consequences of an action, the normative value of those consequences, and one's capacity to perform the action. Individuals are believed to engage in health behaviour if they perceive the action to result in an outcome that they value, if they believe that significant others desire they should act, and if they believe they have the needed resources to successfully perform the behaviour (Conner & Norman, 1996). The theory of reasoned action has been used in examinations of a variety of health behaviours, including smoking (Marin, Marin, Perez-Stable, Otero-Sabogal, & Sabogal, 1990), sexual behaviour (Chan & Fishbein, 1993), and health screening (Lauver & Angerame, 1993). The model and its associated propositions, however, have not received unequivocal support within the literature. Foremost, health beliefs and attitudes have been found to have direct, rather than indirect, effects on health behaviour (Pender & Pender, 1986). This finding calls into question the importance of intention in predicting health behaviour. The theory of reasoned action has also failed to consider the impact of factors such as previous experience with a behaviour, which has been shown to be predictive of health action (Lauver, 1992). Summary The previous discussion provided a concise overview of selected theories that have been used in past research to examine the motivations and cognitions underlying the decision to engage in health behaviour. A wide variety of factors, including beliefs about expectancy, value, control, and self-efficacy have been implicated as being central to motivating health actions. Within the context of this study, however, the limitations of these models create doubt about the applicability of these theories in research examining the use of ACTs. Given the nascent stage of development of this field of study, a theory is needed that provides an elementary foundation for conceptualizing the social cognitions associated with ACT use. With health beliefs forming the most basic component of the above models, the Health Belief Model and its potential application in the present study is considered in the following section. The Health Belief Model The Health Belief Model (HBM) arose in the 1950s in response to a series of research problems faced by researchers in the United States Public Health Services. Frustrated by their 53 inability to explain the failure of many preventive health services and screening programs, the HBM was developed as a means of predicting and modifying health behaviour. Drawing from Lewin's (1951) theory of valence, which considers actions to be determined by positive, negative, and neutral forces within the environment, the HBM hypothesizes that health behaviour is a consequence of an individual's evaluation of the perceived threat of an illness and the potential benefits and costs of a specific health action (Janz & Becker, 1984; Rosenstock, 1974b). Similar to other expectancy-value models (Slovic et al., 1977), the HBM conceptualizes health behaviour as arising from subjective perceptions of probability and utility and as being the consequence of an evaluative, decision-making process. Within the HBM, the perceived threat of an illness is dependent upon two beliefs, perceived susceptibility to illness and the perceived severity of the consequences of an illness (Sheeran & Abraham, 1995). Perceived susceptibility is defined as "one's subjective perception of risk of contracting a condition" (Janz & Becker, 1984, p. 2) and captures an individual's sense of vulnerability to illness, resusceptibility, and belief in diagnosis. Perceived severity is defined as "feelings concerning the seriousness of contracting an illness" (Janz & Becker, 1984, p.2) and includes not only evaluations of the clinical consequences of illness (i.e., death, disability, and suffering), but also the social consequences of a condition (i.e., work, family life, and relationships). Together, perceived susceptibility and severity provide the force or motivation to act (Rosenstock, 1974b). The course of action that is taken is determined by two sets of beliefs about behaviour, including the perceived benefits or efficacy of a health behaviour and the perceived barriers to enacting a behaviour. Perceived benefits are defined as "beliefs regarding the effectiveness of the various health actions available in reducing the disease threat" (Janz & Becker, 1984, p. 2) and refer to a lessening in either perceived susceptibility to or perceived seriousness of an illness. In contrast, perceived barriers represent the "potential negative aspects of a particular health action" (Janz & Becker, 1984, p. 2) and include such factors as inconvenience, expense, pain, and effort. In considering which health behaviour to undertake, Janz and Becker (1984) suggested that a kind of "cost-benefit" analysis occurs, with individuals weighing the perceived efficacy of health behaviour against its associated costs. While perceptions of susceptibility, severity, benefits, and barriers provide the motivation and direction underlying health behaviour, the HBM also proposes that a cue to action must be present to provoke an individual to act. Despite the limited research examining the role of cues to action in the HBM (Janz & Becker, 1984; Sheeran & Abraham, 1995), it is suggested that cues serve as instigating events that set behaviour in motion (Rosenstock, 1974b). Cues may be both internal (e.g., symptoms, pain) or external (e.g., encouragement from others, media) and vary in the level of intensity based on one's perceived susceptibility and severity. For those individuals with little acceptance of their susceptibility to, or the severity of an illness, an intense 54 cue to action would be needed to provoke behaviour. Individuals who perceive themselves as being highly vulnerable to a serious illness would require only a minor cue for action to engage in health behaviour. The majority of studies utilizing the HBM have focussed on the applicability of the theory within the context of preventive health behaviour, which has been defined as behaviour that is undertaken in an attempt to prevent disease or to detect disease in an asymptomatic stage (Rosenstock, 1974a). Some researchers have confirmed, however, the relevance of the HBM to sick role behaviour and illness behaviour (Becker, 1974; Kirscht, 1974). Sick role behaviour has been described as those actions used by individuals who are socially recognized as being ill to restore health or to prevent the further progression of disease (e.g., compliance with a treatment regimen) (Janz & Becker, 1984). In contrast, illness behaviour refers to those activities undertaken by individuals who feel ill for the purposes of determining their health status and discovering possible remedies for their perceived distress (e.g., clinic utilization) (Kasl & Cobb, 1966). It has been acknowledged by some theorists that the line between sick role and illness behaviour is blurred and, at times, difficult to discern (Kirscht, 1974). The distinction between sick role and illness behaviour as operationalized in this study becomes apparent later in this chapter when the use of ACTs by women with breast cancer is considered from the context of symptom management (i.e., sick role behaviour) versus the response to ambiguous and ill-defined concerns related to general well being (i.e., illness behaviour). Despite criticisms of the HBM as being insufficiently conceptualized and lacking clear and empirically supported relationships among the model constructs (Davidhizar, 1983; Mikhail, 1981; Sheeran & Abraham, 1995), the HBM has been used in a wide range of health research (Janz & Becker, 1984). The HBM has been particularly useful in cancer research, providing a means of exploring a variety of health behaviours, including breast self-examination (Calnan & Rutter, 1986; Champion, 1991; Champion & Miller, 1996; Rutledge, 1987; Stillman, 1977), cervical cancer screening (Burak & Meyer, 1997), and compliance with conventional cancer treatment (Newell, Price, Roberts, & Baumann, 1986). In addition, research comparing the predictive capacity and parsimony of a variety of health behaviour theories, has supported the use of a modified HBM in predicting health actions (Mullen, Hersey, & Iverson, 1987; Murray & McCMillan, 1993). These studies offer support for the potential application of the HBM within research that explores the health behaviour of women living with breast cancer, including their use of ACTs. While the HBM is limited in its accounting for the variance in health behaviours explained by attitudes and health beliefs, the theory could potentially provide valuable insight into the motivations and beliefs that underlie breast cancer patients' commitment to using ACTs. 55 Theoretical Models of Commitment to Alternative/Complementary Therapies Previous research on the use of ACTs by cancer patients, including women living with breast cancer, has revealed numerous motivations underlying the decision to use therapies that are not offered as part of conventional medical care. For some individuals, ACTs are used to increase one's chances of survival and to prevent a recurrence of cancer (Downer et al., 1994; Millar, 1997; Truant, 1997). Other cancer patients find themselves struggling with the adverse effects of conventional cancer treatment and use ACTs as a means of coping with their physical, emotional, and spiritual distress (Crocetti et al., 1998; Furnham & Kirkcaldy, 1996). To capture the diverse health contexts in which individuals with cancer may use ACTs, three theoretical models of ACT use have been developed. These models, derived from the HBM (Janz & Becker, 1984; Rosenstock, 1974b), consider women's use of ACTs from a preventive, sick role, and illness behaviour perspective. While the three models of behaviour are not discontinuous (Rosenstock, 1974a), testing each of the models separately allows the health beliefs and attitudes underlying the decision to use ACTs to be most clearly conceptualized within an appropriate theoretical and contextual framework. The following discussion presents a theoretical description of each of the proposed models of ACT use, with the Preventive Model of ACT Use providing a template for the Ameliorative and Restorative Models of ACT Use. Empirical evidence supporting the postulated relationships among study constructs is also provided. The Preventive Model of Alternative/Complementary Therapy Use Figure 1 illustrates the Preventive Model of ACT Use. Using key constructs from the HBM, as well as sociobehavioural variables identified in past research to be associated with preventive health behaviour, the Preventive Model of ACT Use explains ACT use within the context of preventing a recurrence of disease. Previous research using the HBM has investigated such health behaviour as flu vaccination (Rundall & Wheeler, 1979), preventive dental behaviour (Chen & Land, 2002), breast self-examination (Champion & Miller, 1996; Champion, 1991), mammography screening (Clarke, Lovegrove, Williams, & Macpherson, 2000; Holm, Frank, & Curtin, 1999), genetic screening (Becker, Kaback, Rosenstock, & Ruth, 1975) and smoking cessation (Weinberger, Greene, Mamlin, & al., 1981). To date, this research has focussed on individuals who are not ill but who wish to avoid injury or disease. Kirscht (1983) acknowledged that preventive health behaviour is not limited to individuals who are well, but is also undertaken by those experiencing sickness. In this study, women who have been previously diagnosed and treated for breast cancer are the subjects of interest. Within this context, preventive health behaviour is not directed towards an initial diagnosis of breast cancer but is instead used in the prevention of a 57 recurrence of breast cancer. Fear of recurrence has been reported to play a significant role in the emotional distress and uncertainty experienced by breast cancer survivors (Northouse, 1981; Pelusi, 1997; Wainstock, 1991). Such fear may motivate women to seek treatment alternatives that may increase the success of conventional cancer therapies and prevent a recurrence of breast cancer. In this sense, ACTs are used preventively to avoid disease. The outcome construct, Commitment to ACTs, represents women's preventive use of ACTs within the context of breast cancer recurrence and is defined as the degree of effort and expense one is willing to extend in using ACTs. While the majority of past relevant research has examined behaviour deemed by the biomedical community to be efficacious in reducing or identifying the threat of disease, lay health practices, such as the use of ACTs, are important to study as they provide insight into the popular health beliefs held by the general public (Kirscht, 1983). In the Preventive Model, Commitment to ACTs is hypothesized to be a direct consequence of three key health beliefs, women's perceived control over recurrence, their prior history of ACT use, and the social support they received related to ACTs. The health beliefs include: (1) Perceived Risk of Breast Cancer Recurrence, (2) Perceived Efficacy of ACTs, and (3) Perceived Barriers to ACT Use. These beliefs are analogous to the HBM constructs of perceived susceptibility, perceived benefits, and perceived barriers, respectively. The HBM construct, perceived severity, has been omitted from the model because it is assumed that all women in the study will perceive a recurrence of breast cancer to be a serious health threat. The irrelevance of severity in models of preventive breast health behaviour has also been supported in the theoretical and empirical literature (Clark, Hill, Rassaby, White, & Hirst, 1991; Janz & Becker, 1984; Sheeran & Abraham, 1995; Stillman, 1977). Perceived Risk of Breast Cancer Recurrence is defined as the degree to which one feels susceptible to a recurrence of breast cancer. A variety of beliefs have been used in prevention research to measure perceptions of susceptibility, including perceived likelihood of carrying the Tay-Sachs gene (Becker et al., 1975), perceived vulnerability for recurrent infections secondary to cystic fibrosis (Abbott, Dodd, & Webb, 1996), and perceived susceptibility to the side effects of hypertension (heart disease, stroke, and kidney disease) (Kirscht & Rosenstock, 1977). In research that has examined women's breast health practices (i.e., breast self-examination [BSE] and mammography utilization), perceived susceptibility has been measured by the subjective perception of breast cancer risk (Calnan & Rutter, 1986; Champion & Miller, 1996; Champion, 1999; Hyman, Baker, Ephraim, Moadel, & Philip, 1994; Stillman, 1977). Positive associations between women's perception of susceptibility and both BSE practice and mammography screening have been reported (Champion, 1991; Hallal, 1982; Stein, Fox, Murata, & Morisky, 1992; Stillman, 1977). These associations suggest that women who perceive themselves to be at high risk for breast cancer are more likely to engage in preventive health practices than 58 women who perceive themselves to be at low risk. Extending these findings to the Preventive Model of ACT Use, it is theorized that women who perceive their risk of recurrence to be high are more likely to commit to ACTs in the hope of increasing their chances of a cancer-free survival than women who perceive themselves to be at relatively low risk. Significant associations between perceived benefits of and barriers to preventive health behaviour and cancer screening have also been identified (Burak & Meyer, 1997; Fulton et al., 1991; Hyman et al., 1994; Rutledge, 1987). In three studies of women's BSE practices, beliefs about the perceived benefits of BSE were found to be predictive of preventive health behaviour (Calnan & Rutter, 1986; Hallal, 1982; Rutledge, 1987). Women who considered BSE to be effective in the early detection of breast cancer were found to be more likely to engage in BSE than women who did not believe BSE to be beneficial. Mammography screening studies also have revealed negative associations between beliefs about perceived barriers to mammography and compliance with screening (Hyman et al., 1994; Stein et al., 1992). In a study that examined enrollment in a breast cancer prevention trial, non-participants were found to be more concerned about perceived barriers to participation (e.g., effects of tamoxifen, receiving a placebo) than women enrolled in the trial (Yeomans-Kinney et al., 1995). Similar findings in relation to the influence of perceived benefits and barriers on the enactment of preventive health behaviour have been reported within the ACT cancer literature (Cassileth et al., 1984; M.Miller et al., 1998; Yates, 1991). In one study, nearly 60% of cancer patients using ACTs held the belief that their treatments would effect a remission or prevent metastatic growth (Cassileth et al., 1984). Kelner and Wellman (1997a) reported that healthy users of ACTs perceived few barriers to their use of alternative treatments. These findings lend support to the inclusion of perceived benefits and barriers within the Preventive Model of ACT Use. In this model, Perceived Efficacy of ACTs is defined as the degree of confidence one has in the ability of ACTs to prevent a recurrence of breast cancer. Perceived Barriers to ACT Use are defined as the negative deterrents associated with using ACTs. Together, it is hypothesized that women who hold positive beliefs about the effectiveness of ACTs in preventing a recurrence of breast cancer and perceive few barriers to accessing ACTs will demonstrate greater commitment to ACTs than women to whom these types of therapies are believed to be ineffective or inaccessible. In response to past criticisms of the HBM that have decried the lack of consideration of such constructs as control and self-efficacy (King, 1983; Lauver, 1992; Rosenstock et al., 1988; Sheeran & Abraham, 1995), Perceived Control over Recurrence is included as a construct within the Preventive Model. Defined as the degree of control one has over the prevention of a recurrence of breast cancer, Perceived Control over Recurrence is hypothesized to positively influence Commitment to ACTs. Women who believe that they have some control over a 59 recurrence of breast cancer are more likely to take preventive action and commit to using ACTs than women who believe that they have limited control over their future prognosis. Support for this hypothesis has been provided within both the preventive health literature and previous ACT research (Abbott et al., 1996; Balneaves et al., 1999; Furnham & Bhagrath, 1993; Furnham & Smith, 1988; Hallal, 1982). For example, in Hallal's (1982) study of breast health practices, women who performed BSE were found to be less likely to have a locus of control that was dependent on powerful others than women who failed to practice BSE. Similarly, individuals who used homeopathic remedies reported greater perceived control over their health than individuals receiving only conventional medical care (Furnham & Bhagrath, 1993). Both these studies suggest that individuals who participate in preventive health behaviour accept greater responsibility for their health and perceive themselves as playing an active role in their health care. Despite being equivocally conceptualized within the HBM, researchers have found sociodemographic characteristics to be significantly associated with preventive health behaviour (Kirscht, 1983; Sheeran & Abraham, 1995). In particular, education and income have been positively associated with such preventive health behaviour as immunizations (Cummings, Jette, & Brock, 1979), genetic screening (Becker et al., 1975), prenatal care (Stout, 1997), breast cancer screening (Champion & Miller, 1996; Fink, Shapiro, & Roester, 1972; Thomas, Fox, Leake, & Roetzheim, 1996), and ACT use (Balneaves et al., 1999; Cassileth et al., 1984; Downer et al., 1994; Lerner & Kennedy, 1992; M. Miller et al., 1998; Yates et al., 1993). In addition to socioeconomic factors, age has been negatively correlated with both breast health behaviour (Champion, 1994; Rutledge, 1987; Yeomans-Kinney et al., 1995) and ACT use in cancer populations (Crocetti et al., 1998; Downer et al., 1994; M. Miller et al., 1998; Yates et al., 1993). Although the inclusion of ethnicity and immigration history in the Preventive Model would have been informative, there was a lack of diversity in the study sample to ensure sufficient variability in these constructs. Over 89% of the respondents reported their ethnicity to be Caucasian, Canadian, or "nothing in particular".16 In addition, more sophisticated and specific data (i.e., country of birth, number of years living in Canada, first language spoken) would have been required to determine the effect of acculturation on ACT use. While the majority of past research has directly associated sociodemographic factors to preventive health behaviour, there has been some suggestion that age, education, and income may indirectly influence health behaviour through their effects on health beliefs (Sheeran & Abraham, 1995; Janz & Becker, 1984). For example, in Champion's (1984, 1994) studies on 16 The only feasible statistical comparison of ACT use across ethnic groups would have required ethnicity to be recoded as a binary variable (i.e., Caucasian/Canadian and other ethnic group). Comparing ACT use based on being Caucasian or not, however, does not have any substantive support in the literature. 60 BSE and mammography compliance, age was found to be inversely correlated with women's perceived susceptibility to breast cancer. Age also has been negatively associated with perceived efficacy of preventive practices, including the use of ACTs (Nakazono, Davidson, & Andersen, 1997; Yates et al., 1993). Significant associations have been identified between income and beliefs about perceived barriers and control (Stein et al., 1992; Yates et al., 1993), and between education and beliefs about perceived efficacy of health behaviour (Nakazono et al., 1997). These relationships suggest that the differences found across age cohorts and social classes in relation to preventive health practices may reflect the mediating influences of health beliefs on the performance of health behaviour. Drawing from previous research that highlights the indirect relationships among socoiodemographic variables and preventive health behaviour, the following relationships are proposed within the Preventive Model of ACT Use. Foremost, it is theorized that Age is inversely related to Perceived Risk of Breast Cancer Recurrence, Perceived Efficacy of ACTs, and Perceived Control over Breast Cancer Recurrence. Younger women are hypothesized to perceive themselves as having a higher risk of breast cancer recurrence, to perceive ACTs as more efficacious, and to perceive themselves as having greater control over breast cancer recurrence. In turn, younger women are postulated to be more committed to using ACTs. Secondly, Income is indirectly associated with Commitment to ACTs through its effects on Perceived Barriers to ACT Use and Perceived Control over Breast Cancer Recurrence. Women with higher socioeconomic status are predicted to confront few deterrents to using ACTs and to have access to resources that increase their sense of control over their health. Lastly, Education is proposed to positively influence Perceived Efficacy of ACTs and Perceived Control over Breast Cancer Recurrence. Women with higher educational attainment are hypothesized to have greater knowledge of the potential benefits of ACTs and to have greater access to health and informational resources that increase their sense of control over recurrence of breast cancer. In addition to the relationships among the sociodemographic variables and health beliefs, Age, Income, and Education are allowed to covary within the model. Some theorists have further suggested that preventive health behaviour is influenced by past experiences with prevention activities (Sheeran & Abraham, 1995). While few HBM studies have included previous health behaviour as a study construct (some that have include Cummings et al., 1979; Fajardo, Saint-Germain, Meakem, Rose, & Hillman, 1992; Fulton et al., 1991), an association between past preventive health behaviour and utilization of preventive health services has been noted in the literature. For example, in Fulton et al.'s (1991) study of mammography utilization, women's past history of breast screening behaviour predicted mammography screening. Previous use of ACTs, prior to the diagnosis of cancer, has been positively associated with the use of ACTs following cancer diagnosis (Crocetti et al., 1998; M. 61 Miller et al., 1998). Similar to the associations among sociodemographic variables and preventive health practices, the influence of previous use on behaviour has been suggested to be a consequence of indirect effects on health beliefs, particularly perceived efficacy (Sheeran & Abraham, 1995). In the Preventive Model of ACT Use, Previous Use of ACTs is indirectly related to Commitment to ACTs through a positive effect on Perceived Efficacy of ACTs. Women who have used ACTs prior to their breast cancer diagnosis are hypothesized to be more confident in the outcomes of ACTs, thus increasing their commitment to ACTs. It is further theorized that previous experience with ACTs negatively influences how women perceive barriers to using ACTs. As the effect of previous use of ACTs on an individual's commitment to ACTs cannot be explained entirely through its influence on perceived efficacy and barriers, a direct positive relationship is also proposed. This relationship captures other factors potentially associated with previous use, including cultural traditions and habitual health behaviour. Previous ACT use is also permitted to covary with the sociodemographic variables. Several cues to actions have been identified in previous research as being potential triggers of women's preventive health behaviour, including such factors as social support, physicians' recommendations, and health insurance (the latter being particularly relevant in the United States) (Burak & Meyer, 1997; Champion, 1991; Fulton et al., 1991; Stein et al., 1992). The effect of social influence on health behaviour has been of particular interest within the ACT literature, with numerous studies reporting encouragement from significant others as being an important motivator in the decision to use ACTs (Crocetti et al., 1998; Kelner & Wellman, 1997a; Truant, 1998; Yates et al., 1993). Based on this preliminary evidence, the construct Encouragement to Use ACTs was included in the Preventive Model of ACT Use. Defined as the social support received from significant others regarding the use of ACTs, Encouragement to Use ACTs is theorized to have both direct and indirect effects on Commitment to ACTs. Foremost, receiving support from significant others regarding ACTs is hypothesized to increase the perceived efficacy of ACTs and to reduce perceived barriers to using ACTs, resulting in a heightened commitment to ACTs. Secondly, Encouragement to Use ACTs is postulated to have a direct effect on Commitment to ACTs, reflecting the influence of social desirability on preventive health behaviour. Encouragement to Use ACTs and Previous Use of ACTs were permitted to covary with each other and with the sociodemographic factors in the model. The Ameliorative Model of Alternative/Complementary Therapy Use The second theoretical model to be examined is the Ameliorative Model of ACT Use. This model incorporates key constructs of the HBM and related sociobehavioural variables and focusses on women's use of ACTs within the context of experiencing symptom distress as a consequence of conventional cancer treatments. The Ameliorative Model is most closely aligned to previous HBM research that has examined the impact of health beliefs on sick role 62 behaviour, which has been previously defined as any activity undertaken by an individual who is socially recognized as being ill, for the purposes of becoming well (Kasl & Cobb, 1966). Often, sick role behaviour occurs following the diagnosis of a disease or the prescription of conventional medical treatment. In this context, the Ameliorative Model examines how women with breast cancer respond to medically recognized side effects of conventional cancer therapies (e.g., fatigue, lymphadema) through the use of ACTs. Although the majority of research that has utilized the HBM to examine sick role behaviour has focussed on individuals' compliance with conventional treatment recommendations (Abbott et al., 1996; Becker, 1974; Becker & Maiman, 1975; Kirscht & Rosenstock, 1977; Paskett, Carter, Chu, & White, 1990), the Ameliorative Model of ACT Use takes a less pejorative stance and examines the self-care activities of women who have chosen to use ACTs as a means of managing their symptom distress. Past research suggests that symptom distress is experienced by a considerable number of breast cancer survivors during and following their conventional cancer treatment (Cimprich & Ronis, 2001; Ehlke, 1988; Hoskins, 1997; Knobf, Cimprich, & Ronis, 2002; Love, Leventhal, Easterling, & Nerenz, 1989; Wainstock, 1991). Symptoms most prevalent within the breast cancer population include fatigue, nausea, vomiting, pain, and sleep disturbances (Ehlke, 1988; Hoskins, 1997; Lindley, Vasa, Sawyer, & Winer, 1998; Love et al., 1989, Oberst, 1991). Emotional distress has also been reported as being a serious consequence of the diagnosis and treatment of breast cancer (Mast, Lindley, Vasa, Sawyer, & Winer, 1998; Wainstock, 1991). These symptoms and distress are proposed within the Ameliorative Model to be motivating factors behind women's commitment to ACTs. The Ameliorative Model of ACT Use and its related constructs are illustrated in Figure 2. While similar to the Preventive Model of ACT Use with regards to the inclusion of health beliefs and sociobehavioural variables, several key differences exist between the models. The following discussion focusses on these differences and provides empirical evidence of the specified causal relationships. Foremost, the outcome construct within the Ameliorative Model, Commitment to ACTs, is hypothesized to be representative of women's ameliorative use of ACTs within the context of symptom distress secondary to conventional cancer treatment. Defined as the degree of effort and expense one is willing to extend in using ACTs, this construct reflects women's sick role behaviour following the diagnosis and treatment of breast cancer. To capture the impact of symptom distress on women with breast cancer's commitment to ACTs, the central construct in the Ameliorative Model is the health belief, Perceived Symptom Distress. This construct, defined as the degree of discomfort felt in relation to the symptoms 63 64 being experienced (McCorkle & Young, 1978), is analogous to the HBM concept of Perceived Severity. Previous research has suggested that perceived severity is a key predictor of sick role behaviour, more so than beliefs about perceived susceptibility (Sheeran & Abraham, 1995; Becker & Maiman, 1975; Kirscht, Becker, & Eveland, 1976). The importance of the perceived severity construct within models of sick role behaviour has been partially explained by individuals' ability to conceptualize the consequences of illness when they are experiencing symptoms (Janz & Becker, 1984). Perceived susceptibility may be of lesser importance within models of sick role behaviour because perceptions of risk become superfluous when individuals are confronted with symptoms or a disease. In other words, individuals who consider themselves to be sick and have been diagnosed with an illness may no longer perceive themselves to be at risk. Given these conjectures, the perceived susceptibility construct has been omitted from the Ameliorative Model of ACT Use. Evidence for the potential relationship between Perceived Symptom Distress and Commitment to ACTs has been provided within the ACT cancer literature (Crocetti et al., 1998; M. Miller et al., 1998; Yates, 1991). For example, in one study of ACT use by breast cancer patients, physical and psychological distress was cited as the main reason for using ACTs (Crocetti et al., 1998). Recent intervention research on the effects of guided imagery on women's adjustment during conventional breast cancer treatment is also suggestive of a positive relationship between perceived symptom distress and commitment to ACTs (Kolcaba & Fox, 1999). As such, it is theorized within the Ameliorative Model of ACT Use that women who experience heightened perceptions of symptom distress are more likely to commit to ACTs than women experiencing few or no adverse effects from conventional breast cancer treatment. Similar to the Preventive Model, Commitment to ACTs within the Ameliorative Model is hypothesized to be a consequence of two other health beliefs, Perceived Efficacy of ACTs and Perceived Barriers to Using ACTs. However, in the Ameliorative Model of ACT use, Perceived Efficacy of ACTs is defined as the degree of confidence one has in the ability of ACTs to address the adverse side effects of conventional treatment. The construct, Perceived Barriers to ACT Use, is defined as the deterrents of ACT use. The relevance of these constructs within models of sick role behaviour has been illustrated in a meta-analysis conducted by Janz and Becker (Janz & Becker, 1984), who found that perceived benefits and barriers were significant predictors of sick role behaviour in the majority of past research (Janz & Becker, 1984). Research on the use of ACTs within the general population has also revealed significant associations among the Perceived Efficacy of ACTs and Perceived Barriers constructs and the use of ACTs in the management of such symptoms as anxiety, headaches, and dermatitis (Furnham & Forey, 1994; Furnham & Smith, 1988). Given this evidence, it is theorized that women who consider ACTs to be effective in managing the side effects of conventional cancer 65 treatment and who perceive few barriers to ACT use are more likely to commit to ACTs than women who do not hold these beliefs. Researchers have also implicated perceived control as being influential in decision making related to sick role behaviour, including the use of ACTs (Furnham & Bhagrath, 1993; Furnham & Forey, 1994; Kirscht & Rosenstock, 1977). Within the Ameliorative Model of ACT Use, the control construct, Perceived Control over Adverse Effects is defined as the degree of control one has over the adverse effects of conventional cancer treatment. It is postulated that women who believe in their ability to control the side effects of cancer therapies and their subsequent symptom distress are more committed users of ACTs than women who do not consider symptom management to be within their control. Like the Preventive Model of ACT use, health beliefs are postulated to mediate the effect of the sociobehavioural constructs of Age, Education, and Previous Use of ACTs on women's commitment to ACTs. Based on past breast cancer research, younger women are hypothesized to experience greater symptom distress than older women, hence motivating them to seek out alternative forms of symptom management (Love et al., 1989; Pozo et al., 1992; Wainstock, 1991). Younger women are also hypothesized to perceive ACTs as being more efficacious and to perceive greater control over the adverse effects of conventional cancer treatment. The remaining relationships among Income, Education, Previous Use of ACTs, and the HBM constructs are identical to those proposed within the Preventive Model, with the exception of the effect of Education on Perceived Symptom Distress. Previous research on the effect of education on symptomatology and distress in cancer patients has suggested that individuals who report a high level of education experience significantly greater levels of distress and difficulty as a consequence of their cancer treatment (Love et al., 1989). Given this finding, women with a higher level of education are predicted to experience greater perceived symptom distress than women reporting less education. The latent constructs of Age, Income, Education, and Previous Use of ACTs also are allowed to covary within the Ameliorative Model. The last construct in the Ameliorative Model, Encouragement to Use ACTs, functions in the same manner as in the Preventive Model. The support received from significant others regarding the use of ACTs is hypothesized to affect women's perceptions of the efficacy of ACTs and perceived barriers to ACT use. As well, social influence is suggested to directly influence commitment to ACTs through the desire to meet the expectations of one's social network. The Encouragement to Use ACTs, sociodemographic factors, and the Previous Use of ACTs construct were permitted to covary. The Restorative Model of Alternative/Complementary Therapy Use The final model to be tested in this study is the Restorative Model of ACT Use, which examines women's use of ACTs within the context of addressing non-specific concerns about 66 the state of their general well being following their breast cancer diagnosis. This model is closest in theory to past HBM research that has examined illness behaviour. As previously mentioned, illness behaviour refers to those actions undertaken in response to vague symptoms or health concerns of illness for the purposes of determining one's health status and identifying possible remedies for perceived distress (Kirscht, 1974). While the boundary between illness and sick role behaviour has been acknowledged to be indistinct at times, illness behaviour is specific to those situations in which the nature of the perceived symptom or health threat is ambiguous or the cause is unknown (Kirscht, 1974). The HBM has been used sparingly in exploring illness behaviour, with the exception of research on clinic utilization (Berkanovic, Telesky, & Reeder, 1981; Kirscht et al., 1976; Leavitt, 1979). However, Kirscht (1974) recommended using the HBM to examine additional illness behaviour, including promptness or delay in seeking care and the use of "non-medically approved remedies". It is the latter issue that is of interest within the context of the Restorative Model of ACT Use. In comparison to the Preventive and Ameliorative Models, the Restorative Model of ACT Use is unique in that it considers how women define their general well being following the diagnosis and treatment of breast cancer. Rather than reflect definitive perceptions of risk or symptom distress that are specific to the disease process, the Restorative Model examines the ambiguous threat of the cancer experience on well being. In contrast to previous research that has examined health behaviour from the perspective of illness (Kirscht et al., 1976; Leavitt, 1979), perceived threat to well being may occur in the absence of well-defined symptomatology. Previous research has shown cancer patients to be concerned about the generalized effect of cancer and conventional treatment on physical, emotional, and spiritual well being (Boon et al., 1999; Cassileth et al., 1984; Crocetti et al., 1998; Downer et al., 1994). According to Cassileth et al. (1984), individuals with cancer are particularly fearful of the effect of conventional cancer therapies on the "body's reserve" or immune system. In fact, using ACTs to "boost the immune system" is frequently cited by cancer patients as being one of the main reasons for using these types of therapies (Cassileth etal., 1984; Downer et al., 1994; Risberg et al., 1997; Yates, 1991). Research specific to women with breast cancer has also revealed concerns about the failure of conventional medicine to provide holistic care that considers not only the disease, but also emotional and spiritual well being (Boon et al., 1999; Brown & Carney, 1996). Given this context, the Restorative Model attempts to explain those actions undertaken by women in response to the perceived threat of the breast cancer experience to their general well being. Figure 3 illustrates the Restorative Model of ACT Use. Similar to the Preventive and Ameliorative Models, the Restorative Model considers commitment to ACTs to be a consequence of key constructs from the HBM as well as selected sociobehavioural variables. However, the Restorative Model is unique in several important ways. The following discussion highlights these differences and provides, where possible, empirical evidence of the specified causal relationships. Due to the limited number of studies that have tested the illness behavioural model, evidence is drawn primarily from the care-seeking literature. As in the previous two models of ACT use, the outcome construct Commitment to ACTs is defined as the degree of effort and expense one is willing to extend in using ACTs. This construct, however, is operationalized within the context of women's illness behaviour following the diagnosis and treatment of breast cancer. Specifically, women's commitment to ACTs in the Restorative Model reflects women's attempts to define and restore their health in the face of a cancer experience. Previous research examining women's care-seeking behaviour in response to breast symptoms has suggested that a variety of cognitive factors are associated with the intent to seek care. These factors include fear related to breast symptom discovery, perceived seriousness of symptoms, and knowledge of cancer (Facione, Dodd, Holzemer, & Meleis, 1997; Facione & Giancarlo, 1998; Lauver & Angerame, 1993; Worden & Weisman, 1975). Anxiety has also been positively associated with prompt care-seeking behaviour in women who have detected breast abnormalities, with women reporting a high level of anxiety being more likely to seek care than women with minimal anxiety (Lauver & Angerame, 1993). These findings are suggestive of a positive association between perceptions of threat and care-seeking behaviour. Some support for the role of perceived threat of illness in seeking care has been also provided by the limited research that has used the HBM to explore illness behaviour (Kirscht et al., 1976; Leavitt, 1979). Given this evidence, the constructs Perceived Risk of Harm and Perceived Severity of Harm are included in the Restorative Model of ACT Use. Analogous to the HBM constructs of perceived susceptibility and perceived severity, these constructs reflect women's perceptions of harm resulting from their experiences with breast cancer. Perceived Risk of Harm is defined as the degree to which one believes it is likely that their well being has been, or will be, harmed by their breast cancer experience. The concept of "harm" is derived from stress and coping theory, where harm is conceptualized as being either damage or loss (Lazarus & Folkman, 1984). Not only is physical well being affected by the breast cancer experience, but so too is a woman's psychological, social, and spiritual well being. Implicit within the concept of harm is the notion of threat, reflecting the future implications of the losses in one's life (Lazarus & Folkman, 1984). In the Restorative Model of ACT Use, women who feel it is very likely that their well being has been, or will be, harmed by their cancer experience are theorized to use ACTs as a means of restoring or reducing the risk of harm to their well being. Accordingly, women who perceive 69 themselves to be at high risk of harm will demonstrate greater commitment to ACT than will women with low perceived risk of harm. Perceived Severity of Harm is defined as the perceived degree of personal threat felt in relation to one's well being as a consequence of the breast cancer experience. This construct represents women's evaluations of how seriously their physical, psychological, social, and spiritual well being has been harmed by their diagnosis and conventional cancer treatment. Women who believe that their well being has been, or will be, seriously damaged by their breast cancer experience are postulated to use ACTs as a means of restoring or protecting their well being. A positive, direct relationship between Perceived Severity of Harm and Commitment to ACTs is thus theorized within the Restorative Model of ACT Use. Research exploring illness behaviour also has demonstrated the significant influence of both beliefs about efficacy of treatment and barriers to care on the decision to seek conventional medical care (Kirscht et al., 1976; Leavitt, 1979). For example, in Kirscht et al.'s (1976) examination of the use of medical services by low-income mothers, beliefs about the efficacy of medical care were positively related to clinic utilization. Within the literature examining women's responses to breast symptoms, beliefs about the effectiveness of cancer treatment and barriers to care are also associated with care-seeking behaviour (Facione & Dodd, 1995; Facione et al., 1997; Worden & Weisman, 1975). In light of this research, it is probable that these beliefs may also influence decisions to use ACTs. The constructs Perceived Efficacy of ACTs and Perceived Barriers to ACT Use are therefore included in the Restorative Model of ACT Use. Within this model, Perceived Efficacy of ACTs is defined as the degree of confidence one has in the ability of ACTs to enhance or protect one's physical, psychological, social, and spiritual well being. Perceived Barriers to ACT Use is defined as the negative costs and constraints associated with using ACTs. It is postulated that women who hold positive beliefs in the efficacy of ACTs to address their concerns about well being and who perceive few barriers to accessing these types of therapies are more committed to using ACTs in comparison to women who hold opposing beliefs. Few studies in the care-seeking literature have examined the potential relationship between control and the decision to seek conventional medical attention. One exception has been Timko (1987), who used the theory of reasoned action to examine women's intentions to delay seeking care in response to a hypothetical breast symptom. Women who perceived themselves as having greater control over their own health and lives were found to have greater intention to delay than women who believed that they had less control. While this relationship appears to contradict the positive associations proposed in previous models of ACT use between perceived control and health behaviour, this finding must be considered within the context of the conventional health-care system. Delay to seek care in Timko's (1987) study 70 referred to accessing conventional medical care and did not encompass therapies that might be obtained from other sources or self-care activities. For example, in Facione and Giancarlo's (1998) narrative analysis of women's care-seeking behaviour following detection of a breast symptom, some women used ACTs rather than consult a conventional medical provider. Given that a positive relationship between Perceived Control and Commitment to ACTs may exist within the context of illness behaviour, the construct Perceived Control over Well Being is included in the Restorative Model of ACT Use. Defined as the degree of control one has over the maintenance and improvement of well being, women who believe that they have control over their well being are theorized to have greater commitment to ACTs than women who perceive themselves to have little control. This relationship suggests that individuals who believe that they can influence their well being are more likely to enact health behaviours, such as using ACTs. The indirect relationships among sociobehavioural variables and women's commitment to ACTs are similar to those discussed in the Preventive and Ameliorative models. Due to the paucity of research examining the associations among health beliefs and personal characteristics within the context of illness behaviour, the proposed relationships are primarily drawn from previous preventive and sick role behaviour research (Champion, 1994; Nakazono et al., 1997; Stein et al., 1992; Wainstock, 1991). The assumption is thus made that the effects of Age, Income, and Education on health beliefs are consistent across the Preventive, Ameliorative, and Restorative Models. Younger women are proposed to feel more threatened by the breast cancer experience, to perceive ACTs as being more effective in addressing their concerns about well being, and to consider themselves as having greater control over their well being. Women with a high level of income are also hypothesized to perceive fewer barriers to accessing ACTs and to experience greater perceived control over well being. Educational level is further postulated to be positively related to perceived efficacy of ACTs and perceived control over well being. Overall, women who are young, educated, and affluent are committed to ACTs in an attempt to restore their well being following the diagnosis and treatment of breast cancer. The constructs of Age, Income, and Education are also permitted to covary in the model. Also included in the Restorative Model of ACT Use are the constructs Previous Use of ACTs and Encouragement to Use ACTs. Some evidence exists in the care-seeking literature to support potential associations between prior experiences, social support, and illness behaviour (Bottorff et al., 1998; Facione & Dodd, 1995; Facione et al., 1997; Facione & Giancarlo, 1998; Lauver & Angerame, 1993). For example, Facione et al. (1997) found that health care utilization habits are significantly related to help-seeking intentions related to breast symptoms. The importance of receiving social sanctions from significant others prior to seeking care for breast symptoms has also been highlighted within specific cultural groups (Bottorff et al., 1998). Direct 71 effects are thus proposed within the Restorative Model between Previous Use of ACTs and Encouragement to Use ACTs and Commitment to ACTs. Women who have used ACTs prior to diagnosis and who have received support from family and friends to pursue ACTs are presumed to be more committed to ACTs, more so than women without previous experience or support. The previous use and encouragement constructs are also proposed to have similar indirect effects on selected health beliefs (i.e., perceived benefits and barriers), as proposed in the preventive and ameliorative models. Previous Use of ACTs is allowed to covary with the demographic characteristics, as well as with the Encouragement to Use ACTs construct. Chapter Summary This chapter has described in detail the three proposed models of ACT use by women with breast cancer. Using the HBM as a guiding theoretical framework, these models conceptualize ACT use within the context of preventive, ameliorative, and restorative health behaviour and as a consequence of key health beliefs, sociobehavioural factors, and demographic characteristics. The following chapter provides a discussion of the research methods, the instrumentation of the study constructs, and the data analyses. 72 Chapter 4 Methods and Procedures Research Design This research study utilized a retrospective, correlational survey design to collect data on the use of alternative/complementary therapies (ACTs) by women living with breast cancer. Self-report questionnaires were mailed to a random sample of women with breast cancer living in British Columbia. Structural equation modelling (SEM) was used to test three proposed theoretical models of ACT utilization. In the following sections, a description is provided of the sampling and recruitment strategies, operationalization of the constructs of interest, and data collection and analysis methods. A discussion of the ethical considerations of the study concludes this chapter. Sample Setting and Participants The population of interest in this study was women living in British Columbia that had been recently diagnosed with breast cancer and were 18 years or older. The sample was limited to women who had been diagnosed with stage I or II breast cancer and excluded women who had experienced a recurrence of breast cancer.17 Individuals with advanced cancer were ineligible because of the possible introduction of bias or unmeasured sources of confounding effects in relation to their use of ACTs (Cassileth et al., 1984; Truant & Bottorff, 1999). In addition, only women diagnosed between July 1, 1997 and December 31, 1998 (9 to 29 months 17 Because the BC Cancer Registry does not record all cancer diagnoses (i.e., skin cancer) and is limited to British Columbia residents, women who had been diagnosed with other forms of cancer were not excluded from the initial sampling frame. Instead, women were asked in the questionnaire to indicate if they had ever been diagnosed with another type of cancer other than breast cancer. 73 after diagnosis at time of sampling) were recruited to limit recall bias and to access women when they were most likely to be using ACTs (Truant & Bottorff, 1999). To prevent too much research burden for the potential participants, women who had participated in a national complementary therapy survey of individuals living with cancer (undertaken by the National Cancer Institute of Canada's Sociobehavioural Cancer Research Network [SCRN]) and a genetic testing for breast cancer risk study (Bottorff et al., 2002) were also excluded from the sampling frame.18 Women were recruited through the British Columbia Cancer Registry, a provincial registry that records approximately 95% of all new cancer diagnoses (excluding non-melanoma skin cancer) in British Columbia. Because staging information was complete for only those women referred to a British Columbia Cancer Agency centre or clinic, sample selection was further limited to women who had received conventional cancer treatment and were referred to the British Columbia Cancer Agency.19 A small subset of women were found to fit the eligibility criteria but had been entered twice in the Registry because of having more than one breast cancer diagnoses (i.e., bilateral breast cancer diagnosed on the same and different dates). For women diagnosed with bilateral breast cancer on the same date, the case that was entered first in the registry was used. For women with bilateral breast cancer diagnosed on different dates, the case with the earliest diagnosis date was used in the dataset. A total of 1,997 women met the eligibility criteria and comprised the population from which the random sample was generated (see Figure 4 for further details of the development of this final dataset). Additional inclusion and exclusion criteria that could not be determined prior to the random selection are provided on Table 5. Sample Size Decisions related to sample size are important to researchers who must balance limited resources against findings that accurately represent the population of interest. Too small a sample may result in questionable conclusions while too large a sample causes unnecessary waste and subject burden. Unlike multivariate statistical tests, there are no satisfactory approaches for determining an adequate sample size for research using SEM techniques. The following discussion provides a brief overview of current trends in sample size estimation within 18 1,654 women with breast cancer had been randomly selected from the BC Cancer Registry and recruited for the Sociobehavioural Cancer Research Network (SCRN) and Bottorff et al. (2002) studies and were excluded from this study's sampling frame. It is important to note that the SCRN survey was not limited to breast cancer but included other cancer diagnoses. 19 The BC Cancer Registry estimates that between 70% to 90% (depending on stage of disease) of women diagnosed with breast cancer in British Columbia are referred to the BC Cancer Agency for treatment (W. Robb, personal communication, July 24, 2001; T. G. Hislop, personal communication, July, 26, 2002). 74 Included women who were: 18 years or older Currently alive BC resident Diagnosed between 01/07/1997 and 31/12/1998 No recurrence of breast cancer. Excluded 938 women that had no staging information (i.e., not referred to the BC Cancer Agency for treatment). Excluded 567 women who had higher than stage I or II breast cancer. Excluded 1,654 women who had participated in other research studies to avoid research burden. Excluded 61 women who were entered twice in Registry because of bilateral breast cancer. Figure 4. Flow diagram of the development of the final dataset prior to random selection SEM research. Given the design of this study, sample size calculations for sample survey research provided preliminary estimates of the necessary sample size. In survey research, a sample size is sought that provides an estimate of a population value (e.g., means, proportions) within a certain degree of precision. Typically, the degree of precision expected within survey research is within ±5% or less of population estimate. Despite this margin of error, a sample may be selected that is in error by more than the specified level of precision. Researchers are thus compelled to state how confident they are that a sample 75 estimate falls within the selected margin of error of the true population parameter. In survey research, confidence levels ranging from 95% to 99% have been accepted as illustrating reasonable confidence in the data (Lemeshow, Hosmer, & Lwanga, 1990). More specifically, a 95% confidence interval would suggest that there is only a 5% probability that a sample estimate is beyond +1.96 standard errors of the population parameter, or alternatively, a 95% probability that the sample estimate is within ±1.96 standard errors of the population parameter. For survey research concerned with population proportions, the standard error is approximately equal to the standard deviation of the proportion (Vpq/n, where p is the population parameter, q is 1-p, and n is the sample size). If, for example, we wanted the sample estimate to fall within ± 5% of the population proportion with 95 percent probability, then: [1] 0.05 = 1.96 Vpq/n This leads to the following equation to solve for n: [2] n = 1.962 pq .052 When p is unknown, which frequently occurs in studies exploring innovative areas of research, a population estimate of maximum variability should be used (Kalton, 1983; Lohr, 1999). Because pq reaches its maximal value when p = .50, this value will provide a conservative estimate of sample size. Given this, we can solve equation 2, with n = 384. Equation 2 is considered to be an appropriate estimate of sample size when the population is large. However, when the population size is small, the finite population correction (FPC) term should be considered. [3] FPC = (N-n) (N-1) The FPC adjusts for population size and provides a more precise estimate of sample size that reflects the finite nature of the population (Kalton, 1983). Equation 2 is thus transformed to the following: [4] n= 1.962pg«N .052(N-1) + 1.962pq Given that the estimated population size N for women diagnosed with breast cancer between July 1, 1997 and December 31, 1998 would be approximately 3,900 (NCIC, 1998), then n = 35020. The degree of non-response anticipated in a survey must also be considered when determining the required sample size. For example, if a response rate of 70% were predicted, then a selected sample size needed to generate the achieved sample of 350 would be 500. An 20 Sample size estimates were calculated prior to the commencement of the study and made use of predicted population estimates. 76 additional consideration in the calculation of sample size for survey research is the design of the survey. For simple random sampling designs, such as used in this study, the design effect is equal to 1.0. Turning towards this study, the estimate of primary interest was the proportion of women with breast cancer who were using ACTs. Given a margin of error of ±4.5%, a confidence interval of 95%, and a conservative population estimate of 50%, a sample of 475 women was required. Because the population of women from which the sample was selected was relatively small (i.e., N = 3,900 according to National Cancer Institute of Canada, 1998) the FPC term reduced the sample size to 423 women. With previous survey research in this population resulting in response rates of approximately 65%, it was projected that a total sample of 650 women should be selected from the BC Cancer Registry. Although the previous discussion provides a preliminary suggestion of the sample size required to estimate the proportion of women with breast cancer who used ACTs, it fails to consider the complexity posed by the three structural equation models under question. Not only did these models involve the estimation of numerous parameters, but the goodness of fit of each model was also tested. To date, no statistical technique has been developed that has achieved consensus as being an appropriate estimator of required sample size for SEM research. Intuitively, the larger the sample, the more confident one can be that a model accurately represents the population. As a model increases in complexity with regards to the number of observed variables and covariances among variables, the sample size must also increase for confidence in the data to be maintained. The statistical theory that underlies SEM, however, is asymptotic in nature. That is, the sample from which confident conclusions can be drawn is assumed to approach infinity (Tanaka, 1987). SEM is thus considered to be a large sample technique. Problems arise, however, when one considers the interplay between large samples and statistical power in testing SEM (Hayduk, 1987). Simply put, if a sample is too large, trivial deviations between observed and estimated values will result in the rejection of a model that in fact "fits" the data. Too small a sample will mask errors and lead to the inappropriate acceptance of a model. Simulation research on the effect of sample size on structural equation models has revealed the importance of other model characteristics in determining the required sample size for SEM research. In a Monte Carlo study of maximum likelihood confirmatory factor analysis, Anderson and Gerbing (1984) found significant effects on the variance of goodness-of-fit measures related to the number of indicators per latent construct, the number of factors, and 77 increasing non-normality. Hu and Bentler (1995) also suggested that estimation methods and violations of multivariate normality assumptions might further influence the power of SEM analyses. Tanaka's (1987) Monte Carlo examinations of the effect of sample size on normal and non-normal estimation methods supported the importance of the ratio of subjects to the number of estimated parameters in determining sufficient sample size. Despite the extensive research that has been conducted on sample size determination in structural equation modelling, there exists no definitive answer to the question posed by Tanaka (1987) "How big is big enough?" Rules of thumbs, such as having five subjects per number of free parameters when using normal estimation methods and ten subjects per number of free parameters when using arbitrary estimation methods (Bentler & Chou, 1987) or a minimum sample size of 200 when using maximum likelihood estimation (Boomsma, 1983) provide researchers with approximations of required sample size. Given this uncertainty and taking into consideration the nature of the models to be tested in this study, the decision was made to use the sample size estimate provided by the population survey method. With over 2,500 new cases of breast cancer diagnosed each year in British Columbia, the recruitment of 423 women with breast cancer was feasible and appropriate. Recruitment of Participants Of the total population who met the study criteria (N = 1,997), a simple random sample of women with breast cancer was selected from the British Columbia Cancer Registry on November 17, 1999. Cases were randomly selected from the population dataset using SPSS select case command (random sample). Previous survey research within the breast cancer population in British Columbia has produced response rates between 60% and 70%.21 Given this predicted response rate and the possible loss of participants as a consequence of the mailed survey protocol, language difficulties, and advanced disease or death, 650 women were randomly selected from the provincial cancer registry. Following selection from the cancer registry, the status of the potential respondents and the appropriateness of their inclusion in the study were confirmed through their family physicians (see Appendix 1). Women were deemed ineligible for study participation by family physicians for a variety of reasons, including being unable to participate in survey research because of poor physical or mental health (n = 15) or language ability (n = 1), being uninterested in the study (n = 3), or having moved out of the province (n = 1). Other reasons included the women being involved in other research studies (n = 1) or the physician disapproving of the study recruitment strategy (n = 1). Women who were deemed ineligible by 21 T. Greg Hislop (personal communication, June, 23,1999). 78 their family physician were removed from the study database. A letter of invitation and a consent form was then mailed to all remaining eligible women (see Appendices 2 and 3). The representativeness of the sample with regards to selected demographic characteristics (i.e., age, marital status, education, ethnicity, and income) was to be determined by comparing the study sample with data from the British Columbia Cancer Registry. This comparison was not possible, however, because of the limited demographic information collected from women treated for breast cancer by the British Columbia Cancer Agency. A general comparison of demographic and disease characteristics of non-respondents and respondents was conducted and provided insight into the generalizability of the study findings. Survey Completion Rates Table 3 illustrates the completion rates for the study sample. A total of 334 completed surveys were returned. The gross completion rate, comparing the 334 completed surveys to the 650 women randomly selected from the BC Cancer Registry was 51.4%. However, of the original sample of women, 73 women (9.8%) were ineligible because of refusal by their family physician, being unable to read English, having moved outside of British Columbia, or having died. Further, nine letters of invitation were returned without forwarding addresses. Exact statistics for ineligibility are provided on Table 4. Table 3. Survey Completion Rates Type of Rate Comparison Numerator/ Denominator Rate (%) Gross Completion Rate Most reasonable completion rate (conservative) Most reasonable completion rate (liberal) All completions / All women in the sample All completions/ All possible eligibles3 All completions / All eligiblesb 334/650 334/577 334/464 51.4% 57.9% 72.0% includes 113 no contacts in w hich eligibility was not determined bExcludes 113 no contacts in which eligibility was not determined Table 4. Reasons for Ineligibility Reasons Frequency (%) Physician declined 24 (7.2) Language issues 24 (7.6) Died 9(2.7) Letter of invitation returned 9 (2.7) Moved out of the province 7(2.1) Note. N = 73 79 Accordingly, a more reasonable but conservative estimate of the completion rate would be 57.9% (334 completed surveys / 577 all possible eligibles). Table 5 provides a complete list of all inclusion and exclusion selection criteria for the study. Table 5. Inclusion and Exclusion Selection Criteria • Breast primary site of cancer • No participation in SCRN study and Bottorff et • >17 years of age at diagnosis al. (2002) studies • Women currently alive • No refusal by family physician • BC resident at time of diagnosis and survey • English language ability (reading) completion • Locatable through postal service • Stage l/lI breast cancer • Informed consent provided • No reported breast cancer recurrence • Diagnosed between July 1, 1997 and • Received cancer care through BCCA December 31, 1998 Of the remaining 577 eligible women, 97 women declined to participate in the study for a variety of reasons (see Table 6 for exact statistics), including being not interested, too ill, too busy, and on holidays (n = 4). Seven women believed that their participation was not appropriate because they "no longer had breast cancer" or "had no other treatment" than conventional medical care for their breast cancer. Fifteen women declined to participate without providing specific reasons. Consent to participate was received from 33 women who did not return their completed questionnaire despite multiple follow-up attempts by telephone.22 Despite multiple attempts by telephone23 and one follow-up letter, 113 women provided no response to the letter of invitation to participate in the study. If it was assumed that these women were also ineligible for study participation, a more liberal completion rate of 72.0% is achieved. Thus, the actual completion rate of the questionnaire is between 51.4% and 72.0%, with at least one of every two eligible participants being successfully surveyed. A flow diagram of the recruitment process and response rates is provided in Figure 5. Table 6. Reasons for Non-Participation o F Eligible Women Reasons Frequency (%) Declined consent Not interested 33 (10.4) Too III 24 (7.6) Too busy 14 (4.2) Not appropriate 7(2.1) Out of town 4(1.3) No reason given 15 (4.5) Consent given Survey not returned 33(10.4) No Response 113(33.8) Note. N = 243 22 A minimum of three and a maximum of six attempts were made by telephone at a range of times throughout the day and evening. 23 See above footnote. 80 Included women who were: Stage I or II breast cancer 18 years or older Currently alive BC resident Diagnosed between 01/07/1997 and 31/12/1998 Not involved in selected research studies All Possible Eligible Participants (n = 5771 Excluded 73 ineligible women who: • were declined by their family physician • lacked English language ability • had died • had their letters returned as "undeliverable" • had moved out of the province. 97 women declined to participate. Excluded 113 women who did not return consent form and eligibility was not confirmed. 33 women did not return their questionnaires. Figure 5. Flow diagram of the recruitment process 81 Data Collection The data were collected through mailed self-administered questionnaires. The following section describes the implementation of the mailed survey based on a modified version of Dillman's (1978) Total Design Method. A brief description of the pilot test of the study questionnaire is provided in addition to a full discussion of the operationalization of the study constructs. Modified Total Design Method The Total Design Method (TDM) (Dillman, 1978) was developed to increase response rates to mail surveys and to address the reasons why people do and do not respond to questionnaires. Past survey research using the TDM has reported an average response rate of 74%. Drawing from the theoretical perspective of social exchange theory, the TDM proposes that participants in survey research are motivated by whether they expect the rewards of responding to a survey to exceed the costs of participation (Dillman, 1978). The TDM attempts to reward potential respondents by giving verbal appreciation, showing positive regard, and making the questionnaire interesting. Making the task appear manageable, preventing embarrassment, and eliminating any direct monetary expenses reduce the costs of participating in a survey. The development of trust is also important to the success of the TDM. Trust between the participant and the researcher is achieved partly through the development of legitimacy by identifying with a known organization and by providing a token of appreciation in advance. Addressing these issues, however, is not sufficient to guarantee high response rates. The administration of a survey must be carefully organized and detailed to ensure that questionnaires and follow-ups are received in a timely and accurate manner. Dillman (1978) recommended the following four step-process, which is discussed within the context of the completed study. Following receipt of each woman's signed consent form (described in the above section on patient recruitment), the first mail-out occurred, consisting of a cover letter (see Appendix 4), a study questionnaire (see Appendix 5), a token of appreciation (i.e., non-medicinal herbal tea bag), and a stamped return envelope. In the personalized cover letter, the women were thanked for agreeing to participate and provided another brief explanation of the study. The women were encouraged to complete and return their questionnaires via the self-addressed, stamped envelope within the next few days. The inclusion of an herbal tea bag was used as a token of appreciation and as an incentive for women to take a moment to complete the questionnaire. After the first mailing, Dillman (1978) recommended sending a follow-up letter to remind potential participants about the questionnaire and the importance of their participation in the survey. According to Dillman (1978), follow-up mailings hold the potential to increase the 82 response rate two-fold. To increase the personal nature of the reminder and to answer any questions women had regarding the questionnaire, a maximum of six attempts were made to contact by telephone all women who had not returned their survey within two weeks of the original mailing. Dillman (1978) had noted the effectiveness and appropriateness of telephone contact for follow-up. Women who did not want to participate in the study were thanked for their time and assured of no further contact. Four weeks following the original mailing, all non-respondents (irrespective of whether telephone follow-up contact had been made) were sent a follow-up letter (see Appendix 6) restating the invitation to participate. The cover letter emphasized the importance of each participant to the success of the survey and the value of their response. A contact telephone number was provided for those individuals requiring a replacement questionnaire. While Dillman (1978) suggested including a second survey in the package to replace original questionnaires that had been lost or misplaced, this step was omitted because of financial and environmental concerns. Dillman (1978) recommended that a third follow-up, in the form of a registered letter, be sent to all non-respondents seven weeks following the original mail-out. This final step was not carried out because of concerns by the British Columbia Cancer Agency Clinical Investigation Committee regarding the potential "harassing" nature of this contact. A further modification to the TDM was the addition of follow-up phone calls to clarify missing data following the receipt of the completed questionnaires. The respondents were asked at the end of the questionnaire whether they would agree to being contacted by the principal investigator to clarify missing information on their survey. Five women refused any further contact. Initially, a total of 188 women returned questionnaires with missing data, six of which had such extensive missing data (> 20 items), that the decision was made to not pursue follow-up. Successful follow-up of missing data was completed with 159 women (84.6% of surveys with missing data). Follow-up was unsuccessful with 27 women, of which 19 were unable to be contacted despite repeated attempts (mean number of attempts = 2.7 times), four were too sick or had died, and four had telephone numbers that were no longer in service. Pilot Testing A pilot study was conducted to evaluate the appropriateness, feasibility, and acceptability of the study questionnaire and planned data collection protocols. Quantitative and descriptive data were collected to address the following research questions: (1) Recruitment: Was the modified TDM for mailed surveys appropriate and feasible for this sample of women? What response rate could be expected? (2) Data Collection: Was the questionnaire readable and understandable by the participants? How long did it take the women, on average, to complete the questionnaire? 83 Twenty-five women were selected to take part in the pilot study in numeric order (1 through 25) from the randomly selected sample of the provincial cancer registry. As in the full study, letters of explanation were sent to family physicians to determine the women's eligibility. Consent was also sought from all participants before the distribution of the pilot study questionnaire. Once consent was gained, the modified TDM for mailed surveys was implemented as previously discussed. A one-page survey assessment form (see Appendix 7) was also included in the research package to allow participants to evaluate the questionnaire and study protocols. A total of 13 women participated in the pilot study, resulting in a response rate of 52.0%. Of the 12 women who did not participate, 5 women (20.0%) were too ill or had died, 5 women (20.0%) did not respond following repeated attempts to contact them, one woman had moved out of the province, and one woman reported being "not interested". On average, it took participants 32.3 minutes (SD = 10.1) to complete the questionnaire. Several revisions were made to the questionnaire based on the women's comments and suggestions. These revisions are discussed in greater detail in the following sections: Encouragement to Use ACTs Scale (p. 85); Perceived Risk of Harm Scale (PRHS) (p. 87); and Perceived Efficacy of ACTs Scale (PEACTS) (pp. 89). Operationalization of Study Constructs The study questionnaire was comprised of a number of self-report scales and items that, for the most part, were modified from scales used in previous social science research with cancer populations and that had sound psychometric properties. The majority of the scales used close-ended questions with Likert-type responses. The questionnaire was 25 pages long. A complete list of all scales and items used in the final questionnaire developed following the pilot study is found in Table 7. In developing a study questionnaire that incorporated modified and investigator-developed scales and items, the question of validity and reliability arose. Validity is described as the "fit" between a construct of interest and the true score on a corresponding measure or instrument (Knapp, 1998). In contrast, reliability is the corresponding fit between the true score, which itself is a theoretical entity, and the observed score (Knapp, 1998). Most often, reliability is considered to be an estimate of the consistency or variability of a measurement. Several steps were taken to ensure the reliability of the survey and its corresponding scales. Foremost, wherever possible, multiple rather than single indicators were used to operationalize the study constructs. According to MacCallum (1995), multiple indicators are preferred in structural equation modelling as they better define the constructs and limit the biases in parameter estimates caused by measurement error. To determine the internal consistency of the scales with multiple indicators and the average inter-item correlations, reliability estimates (i.e., 84 Cronbach's alpha) were calculated. Confirmatory factor analyses were also conducted prior to beginning the structural equation modelling to ensure the dimensionality of the study measures and the appropriate loading of the indicators on their corresponding constructs. Demographics The sociodemographic and disease characteristics of the participants were assessed with 12 items, using a variety of open- and closed-ended formats (see Table 7). Demographic characteristics that were measured included age, marital status, ethnic/cultural background (open-ended), level of education, and total household income. The majority of demographic items were based on items from the National Population Health Survey, 1994-995 (Statistics Canada, 1995). Demographic variables used as measures in the three models of ACT use are described in greater detail below. For the study construct Age, respondents were asked to report their date of birth. This value was subsequently subtracted from the date on which the questionnaire was received, providing an approximate measure of age in years at time of data collection. This continuous variable (AGE) was used as an indicator of the construct Age. The latent construct, Income, was assessed by asking respondents, "What is your best estimate of the total income, before taxes and deductions, of your household, from all sources, in the past 12 months?" Responses ranged from 1 = "less than $10,000" to 10 = $100,000 or more" (see Table 7). Responses for "no response" and "refused to answer" were coded as missing values. This categorical variable (TOTINC) was used as an indicator of the construct Income. Education was measured by responses to the question, "What is the highest level of education that you have received?" The responses where coded from 1 = "grade 8 or less" to 12 = "earned doctorate", with no response being coded as a missing value. This variable (EDUCATE) was used as an indicator of the latent construct Education. Disease characteristics measured included date of breast cancer diagnosis, date of recurrence (if applicable), other cancer diagnoses, and conventional breast cancer treatment history, including surgery type, chemotherapy, radiation, and hormone therapy (see Table 7). Women were also asked to report whether they had completed conventional cancer treatment at the time of survey completion. Encouragement to Use Alternative/Complementary Therapies The construct, Encouragement to Use ACTs, was assessed with the modified Encouragement to Use Alternative Cancer Therapies Scale (Yates et al., 1993). This three-item scale was originally developed to assess the degree to which cancer patients were encouraged by others to use alternative cancer therapies. Conceptualized as representing the social pressure associated with using ACTs, respondents were asked whether their family, friends, 85 and people important to them had tried to convince them to use ACTs. Respondents were presented with a four-point Likert-type response format (1 = "Very true"to 4 = "Not true at all'). Inter-item correlations among scale items were high, with Kendall Tau B correlations24 of at least 0.37 (Yates et al., 1993). Revisions to the Encouragement to Use Alternative Cancer Therapies Scale were undertaken by Balneaves et al. (1999) to increase the comprehensiveness of the instrument. As a consequence of the redundancy between items assessing support received from family and friends and that received from "people important to them," the latter item was omitted from the scale. Two items were added to the scale to assess the potential encouragement received from the biomedical community. These items included, "My doctor has encouraged me to use alternative/complementary therapies" and "Other health professionals have encouraged me to use alternative/complementary therapies". The response format was also revised to a five-point scale (1 = "Strongly disagree"to 5 = "Strongly agree'). In this study, additional revisions were made to the Encouragement to Use Alternative Cancer Therapies Scale in response to comments received during the pilot study. Some women commented on the difficulty they experienced with the response scale (strongly disagree to strongly agree). Instead of measuring encouragement in terms of agreement, the women suggested that the response scale be modified so that the amount of encouragement that had been received could be assessed. Accordingly, the response scale was revised to a 4-point Likert-type scale ranging from 1 = "Not at all' to 4 = "Very much" (see Table 7). Items on this scale (ENCOUR1-4) were used as multiple indicators25 of the Encouragement to Use ACTs construct, with higher scores indicating greater encouragement from others to use ACTs. Perceived Likelihood of Recurrence The construct, Perceived Likelihood of Recurrence, was assessed by a six-item modified version of Champion's Health Belief Model - Susceptibility Scale (Champion, 1984, 1999). The original instrument, comprised of five scales measuring perceived susceptibility, seriousness, benefits, barriers, and health motivations, was developed to apply the components of the HBM to breast screening behaviour. Within this framework, perceived susceptibility was defined as "the subjective risks of contracting a specific condition within a specified time period" (Champion, 1984). The original scale included such items as, "My chances of getting breast cancer are great' and "Within the next year I will get breast cancer." All items on the scale were measured on a Likert-type response scale, ranging from "1" (strongly disagree) to "5" (strongly 24 Kendall Tau B is a nonparametric measure of association for ordinal or ranked variables. The sign of the coefficient indicates the direction of the relationship, and its absolute value indicates the strength, with larger absolute values indicating stronger relationships. Possible values range from -1 to 1. 25 As previously mentioned, multiple indicators are preferred in structural equation modelling as they better define the constructs and limit the biases in parameter estimates caused by measurement error. 86 agree). Total scale scores were calculated by summing item responses, with higher scores indicating greater perceived likelihood of breast cancer. Content and construct validity of the original version of the Health Belief Model -Susceptibility Scale has been determined through extensive developmental work and pilot testing of the instrument. In Champion's (1984, 1999) work on the Susceptibility Scale, content validity was determined by an expert panel of judges familiar with the HBM who were asked to rate the relevance of the items to the concept. Factor analysis results reported by Champion (1984, 1999) further validated the Susceptibility Scale as an independent and unidimensional factor. Despite some concerns related to the causal interpretation of correlations between perceived susceptibility and health behaviour (Sheeran & Abraham, 1995), the Susceptibility Scale has been shown to be predictive of breast health behaviour (Champion & Miller, 1996). Reliability coefficients for the scale have shown promise, ranging from .78 to .93 (Champion, 1984; Champion, 1991; Champion, 1999; Champion & Miller, 1996; Sortet & Banks, 1997). For the purposes of the study, slight modifications to the Susceptibility Scale (renamed the Perceived Likelihood of Recurrence Scale or PLRS) were necessary to make the items (PLRS1-6) appropriate to women diagnosed with breast cancer (see Table 7). For example, the original item, "My chances of getting breast cancer are great" was revised to read, "My chances of having breast cancer come back are great. "These changes reflected women's perceived susceptibility to a recurrence of breast cancer. The six items were used as multiple indicators of the latent variable Perceived Likelihood of Recurrence in the Preventive Model. Perceived Symptom Distress In the Ameliorative Model, the study construct Perceived Symptom Distress was assessed with McCorkle and Young's (1978) Symptom Distress Scale (SDS). The SDS is a 13-item scale that measures the subjective distress experienced by cancer patients as a consequence of their disease or conventional cancer treatments. The SDS assesses the level of distress associated with such symptoms as nausea, pain, fatigue, and sleep disturbances. The instrument measures the frequency of symptoms and the intensity of the distress on five-point Likert-type response scales. Response scales range from "1" (least distress) to "5" (most distress). The researchers first assessed symptoms experienced by individuals with cancer through qualitative interviews with 26 patients undergoing chemotherapy and radiation (McCorkle & Young, 1978). Follow-up interviews with 60 men and women undergoing conventional cancer treatment further confirmed the content validity of the scale. Extensive cancer research with the SDS has revealed reliability coefficients ranging from .78 to .89 (Ehlke, 1988; Mast et al., 1998; McCorkle & Young, 1978). While a total score is typically obtained for the SDS by summing item responses, with higher scores indicating greater distress, the 87 decision was made to use the 13 items as multiple indicators of the latent construct Perceived Symptom Distress. Perceived Risk of Harm The Perceived Risk of Harm Scale (PRHS) was developed by the investigator to measure perceptions of susceptibility to harm, a concept specified within the Restorative Model of ACT Use. In this model, perceptions of likelihood (or risk) were specific to the harm caused by the breast cancer experience to a woman's well being. No measures currently exist that operationalize these perceptions within the context of well being. Previous research that has examined the dimensionality of general health (Ratner, Johnson, & Jeffery, 1998) and the HBM construct of susceptibility (Champion, 1984, 1999) was instrumental in the development of this scale. Research by Ratner et al. (1998) illustrated that physical, emotional, social, and spiritual health are unique dimensions of well being that cannot be adequately measured with a single question about perceived health status. To capture the multidimensionality of health, multiple indicators are needed that consider each element of well being. Items were thus developed for the PRHS that measured not only the anticipated impact of cancer and conventional cancer treatment on physical well being, but also on emotional and mental health, social relationships, and spiritual well being. The original PRHS items (PRHS1-8) were developed based on the format of two items from Champion's (1984, 1999) Health Belief Model - Susceptibility Scale (i.e., "There is a good possibility that I will get breast cancer"and "The chances of getting breast cancer are great"). These items were revised to reflect perceptions of potential harm specific to women's physical, mental, social, and spiritual well being (e.g., "There is a good possibility that my emotional well being has been harmed by my breast cancer experience" and "The chances that my emotional well being has been harmed by my breast cancer experience are great"). Participants responded to each item on a five-point Likert-type scale, ranging from 1 (strongly disagree) to 5 (strongly agree). The pilot study participants, however, noted a redundancy between items PRHS1-4 and PRHS5-8, resulting in PRHS5-8 being omitted from the final questionnaire. The four remaining items (PRHS1-4) were modelled as multiple indicators of the latent construct Perceived Risk of Harm in the Restorative Model, with higher scores on each item indicating greater perceived risk of harm (see Table 7). Perceived Severity of Harm Drawing from previous research that examined perceptions of threat (Becker & Maiman, 1975; Folkman, Lazarus, Dunkel-Schetter, DeLongis, & Gruen, 1986), the four-item Perceived Severity of Harm Scale (PSHS) was developed by the investigator to measure the construct Perceived Severity of Harm specified in the Restorative Model of ACT Use. According to Becker 88 and Mainman (1975), beliefs about one's personal susceptibility to harm are not sufficient to motivate health behaviour. An individual must also perceive the threat to be serious and to have significant consequences before action is taken. Using flu vaccination as an example, while individuals may perceive the likelihood of contracting the flu to be great, if they do not perceive the consequences of the illness (e.g., missing work/school, feeling sick) to be great, they are unlikely to make a doctor's appointment to request immunization. Perceptions of severity have been measured in several ways, with the majority being disease-specific (e.g., breast cancer) (Champion, 1984, 1999; Champion & Miller, 1996; Fulton et al., 1991). To measure the more ambiguous concept of harm to well being, generic questions about the perceived seriousness of harm were needed. Women were asked to rate on a four-point Likert-type response scale (1 = "not at air to 4 = " very much") how much their physical, emotional and spiritual well being and their relationships with others were harmed by their breast cancer experience (see Table 7). Items on the PSVHS (PSVHS1-4) were used as multiple indicators of the latent construct Perceived Severity of Harm in the Restorative Model of ACT Use, with higher scores on each item indicating greater perceived severity of harm to well being following breast cancer diagnosis and conventional treatment. Perceived Efficacy of Alternative/Complementary Therapies The seven-item Perceived Efficacy of Alternative/Complementary Therapies Scale (PEACTS) was developed by the investigator to measure the construct Perceived Efficacy of ACTs, which was included in the Preventive, Ameliorative, and Restorative Models of ACT Use. The items were derived from both the Belief in the Efficacy of Alternative Cancer Treatments Scale (Yates, 1991; Yates et al., 1993) and from qualitative research on the treatment practices of women living with breast cancer (Boon et al., 1999; Truant, 1998). The Belief in the Efficacy of Alternative Cancer Treatments Scale (Yates, 1991; Yates et al., 1993) was originally a 12-item instrument that measured cancer patients' beliefs about the efficacy, adverse effects, and relative importance of alternative cancer treatments. Factor analysis resulted in a six-item scale that included such items as "they will cure my cancer" and "they will assist other treatments to work". This scale, however, was developed with an advanced cancer population in mind and focussed on the curative aspects of ACT use. Qualitative studies of women with less advanced breast cancer have revealed a different set of beliefs about the efficacy of ACTs (Boon et al., 1999; Truant, 1998). These beliefs include using ACTs to manage the side effects of conventional cancer treatments, to promote feelings of well being, and to prevent recurrence. Items in the PEACTS attempted to capture these beliefs about the potential benefits of ACTs. The first item (PEACTS1), "Alternative/complementary therapies will prevent a recurrence of my breast cancer", measured women's beliefs in the preventive role of ACTs. This item was used as a single indicator of the Perceived Efficacy of ACTs construct in the 89 Preventive Model. The second and third items (PEACTS2-3), "Alternative/complementary therapies will relieve my symptoms" and ''Alternative/complementary therapies will relieve the side effects of medical treatment", measured perceived efficacy of ACTs within the context of ameliorative behaviour. Slight modifications were made to the wording of PEACT2-3 as a result of pilot testing, with the items altered to read "ACTs will relieve some of my symptoms/the side effects...." . This revision was made in response to one woman's observation that the ACTs she had used had relieved "some symptoms but not all". These two items served as multiple indicators of perceived efficacy of ACTs in the Ameliorative Model of ACT Use. The final four items on the PEACTS (PEACTS4-7) examined women's beliefs about the effectiveness of ACTs in restoring women's physical, mental, social, and spiritual well being following the diagnosis and treatment of breast cancer (e.g., Alternative/complementary therapies will improve my physical well being). These items were used as multiple indicators of the Perceived Efficacy of ACTs construct within the Restorative Model of ACT Use. Response formats for all items on the PEACTS ranged from 1 to 5 (strongly disagree to strongly agree), with a higher value indicating a higher degree of confidence in the efficacy of ACTs (see Table 7). Perceived Barriers to Alternative/Complementary Therapy Use The construct, Perceived Barriers to ACT Use was mea