1 SYMPTOM CLUSTERS IN BREAST CANCER SURVIVORS: PREVALENCE, PREDICTORS, AND CONSEQUENCES By LOIS META RITZ ELLIS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE RE QUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Lois Meta Ritz Ellis
3 To my husband, children, and mentor for their encouragement and endless support
4 ACKNOWLEDGMENTS I would like to acknowledge and extend my deepest gratitude and heartfelt thanks to all those who have made the completi on of my dissertation possible. I am deeply indebted to my committee chair and mentor, Dr. Ann Horgas, whose incredible suggestions, kn owledge, and experience, as well as unfailing support and devotion of her time, were invaluable in helping me achieve my goal. I am also very grateful to my committee members, Dr. Deidre Pe reira, Dr. Beverly Roberts, Dr. Carmen Rodriguez, and Dr. Saunjoo Yoon for their guidance towards the completion of my dissertation. Also, a very special thanks to my family. To my husband, John, for his unshakable belief in me and commitment to keeping me on track. To my son, Sean, for his belief in my success. To my d aughter, Sara, for her phone calls of confidence in completing my d ream. And to my parents, Rudy and Meta, who instilled within me the belief that I could do and be anything I wi shed, and who I know are beaming down upon me with pride.
5 TABLE OF CONTENTS ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 BACKGROUND AND SIGNIFICANCE AND THEORETI C AL FRAMEWORK ........ 13 Introduction ................................ ................................ ................................ ............. 13 Background and Significance ................................ ................................ ................. 13 Conceptual Issues ................................ ................................ ................................ .. 15 Definition of a Survivor ................................ ................................ ..................... 15 Definition of Age: Younger Versus Older for Women Survivors of Breast Cancer ................................ ................................ ................................ .......... 15 Definition of a Symptom Cluster ................................ ................................ ....... 18 Theory ................................ ................................ ................................ ..................... 19 Purpose ................................ ................................ ................................ .................. 20 Research Aims ................................ ................................ ................................ ....... 20 2 REVIEW OF LITERATURE ................................ ................................ .................... 24 Introduction ................................ ................................ ................................ ............. 24 Symptom Research in the Oncology Literature ................................ ...................... 24 Prevalence of Symptoms in Oncology ................................ .............................. 24 Oncology Single, Multiple, and Symptom Cluster Research ............................ 26 Relationship Between Background Characteristics and the Symptom Experience in Oncology ................................ ................................ ................ 29 Relationship Between the Symptom Experience and Outcomes in Oncology .. 31 Summary of Symptom Research in the Oncology Literature ............................ 32 Symptom Research in the Breast Cancer Literature ................................ ............... 33 Prevalence of Symptoms in Breast Cancer ................................ ...................... 33 Breast Cancer Single, Multiple, and Symptom Cluster Research ..................... 36 Relationship Between Background Characteristics and the Symptom Experience in Breast Cancer ................................ ................................ ......... 39 Relationship Between the Symptom Experience and Outcomes in Breast Cancer ................................ ................................ ................................ .......... 41 Summary of Symptom Research in the Breast Cancer Literature .................... 43
6 3 METHODS ................................ ................................ ................................ .............. 46 Introduction ................................ ................................ ................................ ............. 46 Design ................................ ................................ ................................ ..................... 46 NIH PROMIS Dataset ................................ ................................ ............................. 47 Overview ................................ ................................ ................................ .......... 47 PROMIS Sample ................................ ................................ .............................. 48 Study Sample ................................ ................................ ................................ ......... 49 Measures ................................ ................................ ................................ ................ 50 Original PROMIS Measurement Bank ................................ .............................. 50 Measurement of Key Study Variables ................................ .............................. 51 Independent variables ................................ ................................ ................ 51 Dependent variables ................................ ................................ .................. 52 Background characteristics ................................ ................................ ........ 53 Procedures ................................ ................................ ................................ ............. 53 Protection of Human Subjects ................................ ................................ .......... 53 Data Acquisition ................................ ................................ ............................... 54 Statistics ................................ ................................ ................................ ................. 54 Aim 1 ................................ ................................ ................................ ................ 54 Aim 2 ................................ ................................ ................................ ................ 55 Aim 3 ................................ ................................ ................................ ................ 55 Aim 4 ................................ ................................ ................................ ................ 56 4 RESULTS ................................ ................................ ................................ ............... 59 Aim 1: To Describe the Prevalence and Number of Self Reported Physical Health Symptoms (Pain, Fatigue) and Mental Health Symptom (Depression/Anxiety/Anger) Among Women Survivors of Breast Ca ncer. .......... 59 Aim 2: To Determine Whether and How Symptoms Combine to Create Identifiable Clusters In A Sample Of Women Survivors Of Breast Cancer. ......... 59 Bivariate Analysis ................................ ................................ ............................. 59 Multivariate Cluster Analysis ................................ ................................ ............ 60 Aim 3: To Investigate the Relationships Between Backgro und Characteristics (Age, Ethnicity, Education, Relationship Status, Employment Status, Comorbid Conditions) and Symptom Clusters. ................................ .................... 62 Aim 4: To Investigate the Relationships Between Symptom Cl usters and the Functional Outcomes (Physical Function, Social Role Function). ........................ 64 5 DISCUSSION ................................ ................................ ................................ ......... 70 Summary of Results ................................ ................................ ................................ 70 Symptom Experience of Women Survivors of Breast Cancer ................................ 70 Symptom Clusters in Women Survivors of Breast Cancer ................................ ...... 72 Background Characteristics as Predictors of Symptom Clusters in Women Survivors of Breast Cancer ................................ ................................ .................. 76 Outcomes or Consequences of Symptoms Clusters in Wo men Survivors of Breast Cancer ................................ ................................ ................................ ...... 78
7 Limitations ................................ ................................ ................................ ............... 79 Implications for Clinical Practice ................................ ................................ ............. 80 Implications for Future Research ................................ ................................ ............ 81 Conclusion ................................ ................................ ................................ .............. 81 LIST OF REFERENCES ................................ ................................ ............................... 83 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 92
8 LIST OF TABLES Table page 1 1 Theoretical constructs, concepts, and empirical indicators for t he study ............ 23 3 1 Description of sample (N=103) ................................ ................................ ........... 58 4 1 Prevalence and intensity (0 1=none) of symptoms (N=103) ............................... 66 4 2 Frequency distribution of number of symptoms reported (N=103) ...................... 66 4 3 Final cluster analysis z score means on pain, fatigue, and mental h ealth variables ................................ ................................ ................................ ............. 66 4 4 Final cluster analysis relationship to background characteristics ........................ 67 4 5 Physical function responses: a ble to carry out physical activities (N=103) ......... 68 4 6 Social role function responses: satisfaction with activities and relationships (N=103) ................................ ................................ ................................ .............. 68 4 7 Final cluster analysis of physical function ................................ ........................... 68 4 8 Final cluster analysis of social role function ................................ ........................ 69 4 9 M ultiple comparisons of social role function in clusters ................................ ...... 69
9 LIST OF FIGURES Figure page 1 1 Diagram of the New Symptom Management Model (Brant et al., 2010) ............. 21 1 2 Adapted version of the New Symptom Management Model ............................... 22 3 1 PROMIS sample design ................................ ................................ ..................... 57
10 LIST OF ABBREVIATIONS ACS American Cancer Society NIH National Institutes of Health PROMIS Patient Reported Outcomes Measurement Information System
11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SYMPTOM CLUSTERS IN BREAST CANCER SURVIVORS: PREVALENCE, PREDICTORS, AND CONSEQUENCES By Lois Meta Ritz Ellis May 2013 Chair: Ann Horgas Major: Nursing Science s There are currently 2.9 million women survivors of breast cancer in the United States. Many breast cancer survivors report experiencing multiple simultaneous symptoms. The majority of research, however, has focused on single symptoms. Thus, the purpose of this study was to investigate symptom clusters, their demographic predictors and functional consequences, among women survivors of breast cancer. The following aims were addressed: (1) To describe the prevalence and number of self reported physical (pain, fatigue) and mental (depression/anxiety/anger) health symptoms. (2) To determine whether and how symptoms combine to create identifiable clusters. (3) To investigate the relationships between background characteristics and symptom clusters. (4) To investigate the rela tionships between symptom clusters and functional outcomes (physical function, social role function). The research aims were investigated using a descriptive, exploratory, cross sectional, secondary analysis study design. Data from 103 women breast cancer survivors from the Patient Reported Outcomes Measurement Information System (PROMIS) from the N ational Institutes of Health were analyzed. The women in this sample were mainly White, highly educated, partnered, employed, with a mean age of
12 60.4 years (Ran ge = 30 85). Data w ere analyzed with descriptive and non parametric bivariate analyses. Symptom clusters were analyzed with correlation and cluster analysis. The majority of women survivors of breast cancer in this sample reported experiencing symptoms: 67% pain, 62% fatigue, and 63% depression/anxiety/anger. Most women (n=68, 66%) experienced two or three mildly severe symptoms concurrently. Three symptom clusters wer e identified: Cluster 1 (All Minimal Symptoms) (n=53), Cluster 2 (All Mild Symptoms) (n=42 ), and Cluster 3 (All Moderate Symptoms) (n=8). The background characteristic predictors revealed no significant relationships with symptom clusters. There was a trend that women in the more symptomatic cluster s were younger. Symptom clusters were significantly associated with the physical and social role functional outcomes. Women in the more symptomatic clusters had lower physical and social role function. Symptoms do cluster and differ based on symptom intensity. Wo men in clusters with even mil d to moderate intensity symptoms reported worse physical and social role funct ioning. P ractitioners need to recognize that even mild, concurrent symptoms may require treatment for optimal daily functioning of women survivors of breast cancer.
13 CHAPTER 1 BACKGROUND AND SIGNIFICANCE AND THEORETI C AL FRAMEWORK Introduction Cancer survivors are growing in numbers and surviving longer. Cancer occurs predominantly in older women with breast cancer among the most common. One in eight women in the United States develops breast cancer in their lifetime. Most cancer survivors report simultaneous occurrence of multiple symptoms which may negatively impact their health outcomes and daily functioning. The majority of symptom management research, however, has focuse d on single symptoms. Thus, it is relevant that more research be done to understand the prevalence, predictors, and c onsequences of multiple symptoms, or symptom clusters for wome n survivors of breast cancer. Any new knowledge acquired may be used by hea lthcare providers to resolve symptoms and facilitate more positive daily physical and social functioning. The following sections present the background and significance of the problem, the theoretical framework for the study, the purpose stat ement, and th e research aim s. Background and Significance Over 13 million Americans are cancer survivors (ACS, 2012b). Cancer occurs predominantly in the older adult: 75% of people who develop cancer are age 55 or older, and 60% of newly diagnosed cancers occur in a dults over the age of 65 (Bond, 2010a). Breast cancer is the most common cancer diagnosis for women in the United States, excluding skin cancer (ACS, 2012a). Over 2.9 million women in the United States are living with b reast cancer, as they have a 90% fi ve year overall survival rate from diagnosis (ACS, 2012a; ACS, 2012b).
14 In the oncology literature, the emphasis has been on investigating single symptoms. Single symptom research in oncology has focused on pain, fatigue, depression, anxiety, and sleep d isturbance. There is growing recognition that symptoms do not occur in isolation. People diagnosed with cancer present with multiple symptoms and/or develop symptoms while undergoing treatment, which may negatively impact their health outcomes and daily functioning (Barsevick, 2007a; Barsevick, 2007b; Dodd, Cho, Cooper, & Miaskowski, 2010; Miaskowski Aouizerat, Dodd, & Cooper quality of life, adjustment to the cancer diagno sis, functional status, and survival (Brant, Beck, & Miaskowski, 2010). Pervasive in the nursing literature is the need to define and clarify the symptom cluster concept, determine etiologies, determine consequences, and develop appropriate intervent ions to advance the science (Barsevick 2007a; Barsevick, 2007b; Dodd et al., 2010; Dodd, Miaskowski, & Lee, 2004: Fan, Filipczak, & Chow, 2007; Gift, 2007; Kim, Barsevick, Tulman, & McDermott, 2008; Kim McGuire, Tulman, & Barsevick 2005; Miaskowski et al. 2 007; Xiao, 2010). The gap in the literature identified for this study is the need to first examine the symptom experience by identifying if symptoms interact to form symptom clusters in women survivors of breast cancer. Background characteristics primar ily age as a risk factor for women survivors of breast cancer, can be investigated to determine if age impacts the symptom experience. T he symptom experience and/or the symptom clusters identified can be studied to ascertain if relationships exi st between the symptoms and physical or social functioning outcomes. This research is needed to determine if there are discernable
15 symptom clusters, so that the symptom experience and/ or sy mptom clusters can be treated simultaneously with treatment strategies imple mented by healthcare providers to positively impact physical and social functioning as women deal with breast cancer survivorship. Conceptual Issues There are a number of conceptual issues inherent in this work These include the definition of a survivo r, the designation of younger versus older age, and the number of symptoms that designate a symptom cluster. These conceptual issues are discussed in the following section. Definition of a Survivor The definition of a survivor of breast cancer is debatabl e. The National Cancer Institute defines a person as a cancer survivor at the time of diagnosis and discusses survival based on how long people live after diagnosis (Howlader et al., 2012). American Cancer Society (2011) discusses breast cancer survival as relative survival rates from diagnosis. Investigators in this area of inquiry studying women with breast cancer define a survivor as a woman who has been diagnosed with breast cancer (Grunfeld, 2009; Knobf, 2007). For this study women survivors of b reast cancer will be operationally defined as women from the time of diagnosis and continuing throughout their lives, consistent with the current literature review. Definition of Age: Younger Versus Older for Women Survivors of Breast Cancer Another concep tual issue arises with the term age and the definition of older versus younger. Being female is the greatest risk factor for breast cancer, with increasing age being the second greatest risk factor for a diagnosis of breast cancer (ACS, 2012a). The lifeti me risk of a breast cancer diagnosis for a woman living in the
16 United States has increased from 1 in 11 in the 1970s to 1 in 8 currently (ACS, 2011). This is a 12.15% lifetime risk for women to develop breast cancer (ACS, 2011). The American Cancer Societ y (2011) discusses breast cancer incidence as increasing with age, with 95% of new cases occurring in women age 40 or older. The median age for a breast cancer diagnosis is 61 years, with women in the last half of their seventieth decade having the highes t breast cancer incidence (ACS, 2011). Age definitions regarding younger or older women with breast cancer varies between studies. Some researchers use age 50 as the cut off due to this being the average age for menopausal onset (Howard, Anderson, Ganz, Bower, & Stanton, 2012), while others use age 65 based on retirement age set by Social Security (Albert & Freedman, 2010). Bond (2010b) calls for considering physiological age rather than chronological age. The National Cancer Institute (Howlader et al., 2012) relays data for breast cancer for age ranges of under 20, 20 to 34, and then 10 year age ranges of 35 to 44, 45 to 54, 55 64, 65 to 74, 75 to 84, and 85 and above. Data tables by the National Cancer Ins titute also give information comparing under a ge 65 to age 65 and older as well as other tables comparing under age 50 to age 50 and older. No rationale is given for the age differentiations. The American Cancer Society (2011) relays data on women with breast cancer using age 50 as the cut point fo r older versus younger women. In the United States, full Social Security benefits are awarded at age 65 and this chronological age is often used as a marker for old age (Albert & Freedman, 2010; Quadagno, 2008). Social gerontologists frequently split olde r people into categories of young old age 65 to 74, middle old age 75 to 84, and oldest old age 85 and older (Quadagno, 2008). The PROMIS datasets use the age groupings of 18 to 29, 30 to 39,
17 40 to 49, 50 to 59, 60 to 6 4, 65 to 84, and 85 and older. Thus there are inconsistencies in how young versus older age are defined. The literature review on oncology and breast cancer revealed ten articles that specifically addressed younger or older patients with cancer or breast cancer. Six of these ten articl es gave no rationale or used the median age of the sample for defining younger and older cancer patients (Berger, Visovsky, Hertzog, Holtz, & Loberiza, 2012; Fabro et al., 2012; Kenefick, 2006; Khan, Amatya, Pallant, & Rajapaksa, 2012; Miaskowski et al., 2 006; Palesh et al., 2010). Two articles referenced age 65 as older due to the increased occurrence of cancer in this population (Bond, 2010a; Yeom & Heidrich, 2009). One article used age 60 as the cut off for older versus younger, but gave no rationale f or the designation (Cheung, Le, Gagliese, & Zimmermann, 2011). The final article used age 50 as the division between older based on menopausal status (e.g., mean age of menopause is 50) (Howard, Anderson, Ganz, Bower, & Stanton, 2012). Many other researc h articles on women breast cancer survivors include d age as part of the background characteristic s but made no reference to any delineation based on age. For this study, women will be operationally defined as younger if they are below age 50, and as older if age 50 or older. This definition correlates with the American Cancer Society, many of the datasets of the National Institute s of Health, and many breast cancer research article s that relay the age background characteristic related to menopausal onset. The PROMIS dataset fits well with this definition. If the sample has enough older women, then there will also be a comparison based on social gerontology of age 65 to 84 as the young and middle old to those age 85 and older as the oldest old. This compa rison could provide additional clarification regarding older
18 age categories and the symptom experience since age is second only to being a woman as the greatest risk factor for a breast cancer diagnosis. Definition of a Symptom C luster The nursing literat ure began to investigate the association between symptoms, especially in oncology, in the 1990s, but did not label these coexisting symptoms as a et al. 2005; Lenz, Pugh, Milligan, Gift, & Suppe, 1997). Dodd, Miaskowski, and Paul (2001) ar e recognized as the first authors in the nursing literature to identify the concept of a symptom cluster in oncology (Kim et al., 2005; Xiao, 2010). A symptom cluster was defined as having three or more coexisting symptoms with some relationship to each o ther, with or without the same etiology (Dodd, Miaskowski, et al., 2001). The original authors of the symptom cluster concept did not elucidate the nature, intensity, or temporal aspect of the concurrent symptoms. An alternate definition of a symptom clu ster of two or more symptoms that have a relationship to each other, occur simultaneously, remain constant, are somewhat autonomous, and may show precise core characteristics, with or without the same etiology was proposed by others (Kim et al., 2005). B ecause there has only been a decade of research in this area, symptom cluster research is in its infancy. The operational definition of a symptom cluster needs further development and clarification as a means to assist in assessment and management of symp toms among cancer survivors (Dodd et al., 2004; Fan et al., 2007; Xiao, 2010). For this study, the symptom prevalence, the number of symptom s, and the symptom experience were examined to ascertain if symptom interactions occur in any identifiable pattern to form symptom clusters. A cluster was operationally identified as having two or more symptoms based on the most recent definition. The symptom
19 experience was examined to determine any relationships between this experience and background characteristic s, primarily age, as well as physical and social functioning outcomes. These proposed relationships were explored using theory. Theory Theory is a methodical expression of a set of statements articulated to communicate reality with the intent of describin g, explaining, or predicting associations or outcomes (Meleis, 2007). The New Symptom Management Model was used to guide the proposed research (Brant Beck, & Miaskowski 2010). This model incorporates multifaceted characteristics of symptom interactions allowing for symptom cluster analysis. Figure 1 1 is a diagram of the original model (Brant et al., 2010) showing symptom inter actions over time, anteced ents, and consequences of symptom clusters. The New Symptom Management Model was adapted by the inves tigator to explore a subset of constructs. Specifically, an adapted version of this model was used to explore the constructs of the symptom exper ience (symptom clusters) antecedents (background characteristics ), and consequences (function) among women su rvivors of breast cancer (see Figure 1 2 ) The constructs of antecedents, symptom experience, and consequences were theoreticall y conceptualized as background characteristics (age, ethnicity, education, relationship status, employment status comorbid con ditions ), symptom c lusters (physical health symptoms : pain, f atigue and mental health symptom: depression/ anxiety/ anger), and function (physical, social role). The relationships between the constructs, theoretical concepts, and the empirical indicators a re shown in Table 1 1.
20 Purpose The purpose of this study was to investigate the symptom experience demographic predictors, and functional consequences, among women survivors of breast cancer. This study will increase the body of knowledge of the symptom experience in women survivors of breast cancer by identifying if there is clustering of two or more symptoms, by describing relationships of symptoms or clustering of symptoms and function, and by investig ating the impact of background characteristics on the symptom experience T he following aims will be addressed. Research Aims Aim 1 : To describe the prevalence and number of self reported physical health symptoms (pain, fatigue) and mental health symptom (depression/anxiety/anger) among women survivors of breast cancer Aim 2 : To determine whether and how symptoms combine to create identifiable clusters in a sample of women survivors of breast cancer Aim 3: To investigate the relationships between background characteristics (age, ethnicity, education, relationship status, employment status, comorbid conditions) and symptom clusters. Aim 4 : To investigate the relationships between symptom clusters and the functional o utcomes ( physical function social role function )
21 Figure 1 1. Diagram of the New Symptom Management Model (Brant et al., 2010) QOL Survival Function Adjustmen t Symptom slope Symptom intercept Interventions Symptom Experience Timing, Distress, Intensity, Quality Antecedents Demographics Physiological Factors Psychological Factors Situational Factors Consequences S ymptom slope Symptom slope Symptom management strategies: What? When? Where? Why? Interaction Patient Nurse family provider Symptom intercept Symptom intercept Time 1 Time 3 Time 2 Symptom Trajectory Time x
22 Figure 1 2. Adapted v ersion of the New Symptom Management Model Antecedents Symptom Experience Consequen ces Background Characteristics Primary Age Secondary Ethnicity Education Relationship Status Employment Status Comorbid Conditions Symptom Clusters Physical Health Symptoms Pain Fatigue Mental Health Symptoms Depression Anxiety A nger Function Physical Social Role
23 Tab le 1 1. Theoretical c onstructs, c oncepts and empirical i ndicators for the s tudy Constructs Concepts Empirical Indicators Antecedents Background Characteristics Primary Age Secondary Ethnicity Education Relationship Status E mployment Status Comorbid Conditions PROMIS Measures Age in years Age Groups 50 50 White Black Other HS/Associate of Arts College/Advanced Degree Partnered Not Partnered Employed Not Employed Number of Conditions Symptom Experience Symptom Clusters Physical Health Symptoms Pain Fatigue Mental Health Symptom Depression/Anxiety/Anger PROMIS Measures Pain Fatigue Depression/Anxiety/Anger Consequences Function Physical Function Social Role Functi on PROMIS Measures Physical Function Social Role Function
24 CHAPTER 2 REVIEW OF LITERATURE Introduction Cancer occurs predominantly in the older woman, with one in eight women in the United States developing breast cancer in their lifetime (Hor ner et al., 2008; Perry, Kowalski, & Chang, 2007). People diagnosed with cancer present with multiple symptoms and/or develop symptoms while undergoing treatment, which may negatively impact their health outcomes and daily functioning (Barsevick, 2007a; B arsevick, 2007b; Dodd et al., 2010; Xiao, 2010). The majority of symptom management research focuses on a single symptom when most cancer patients report simultaneous occurrence of multiple symptoms, indicating the need for increased symptom cluster resear ch in outcomes (Gift, 2007; Miaskowski et al., 2007). Using an adapted New Symptom Management Model as an organizing framework, the symptom experience of women survivors of bre ast cancer, potential antecedents of background characteristics, and consequences of physical and social role functioning are explored. First, the symptom experience in oncology will be presented, followed by diagnosis specific breast cancer research. Sym ptom Research in the Oncology Literature Oncology symptom prevalence and symptom research is presented. This will be followed by a presentation and synthesis of how background characteristics relate to the symptom experience and how this experience relate s to the outcomes. Prevalence of Symptoms in Oncology Symptoms occurring most frequently in oncology patients are pain, fatigue, sleep
25 disturbance, depression, and anxiety. In a literature review containing 18 research studies pain was reported at a prev alence of 36% in cancer patients (Kim, Dodd, Aouizerat, Jahan, & Miaskowski, 2009). At a pain clinic, neuropathic pain was found to occur in 59% of the 131 patients (Donovan, Taliaferro, Brock, & Bazargan, 2008). Sixty nine percent of 80 lung cancer pati ents complained of pain (Hoffman, Given, von Eye, Gift, & Given, 2007). The most common physical symptom report ed was pain in 46% of 192 patients studied experiencing this symptom (Breen et al., 2009). Fatigue is reported in literature reviews as being the most prevalent symptom experienced by cancer patients (Donovan & Jacobsen, 2007; Kim Dodd, et al. 2009). In a literature review, Donovan and Jacobsen (2007) reported a 70 100% occurrence of fatigue regardless of where a patient was in the treatment trajectory. This literature review also included the finding that fatigue could be managed by decreasing pain (Donovan & Jacobsen, 2007). In 80 patients with lung cancer, 97% reported experiencing fatigue (Hoffman et al., 2007). Sleep disturbance is g enerally reported to be prevalent in 30 50% of cancer patients by several researchers (Donovan & Jacobsen, 2007; Evangelista, & Santos, 2012; Palesh et al., 2010; Savard, Ivers, Villa, Caplett Gingras, & Morin, 2011), with a wider range of 23 61% by others (Bardwell et al., 2008). A lung cancer study reported 51% occurrence of sleep disturbance (Hoffman et al., 2007) with a literature review describing a 41% occurrence (Kim Dodd, et al. 2009). Another study found a 59% incidence of sleep disturbance, al though this result was preoperatively with the prevalence decreasing to 36% at 18 months postoperatively (Savard et al., 2011). In a literature review, a prevalence of 72% for sleep disturbance was reported at a pain clinic
26 (Fiorentino, Rissling, Liu, & A ncoli Israel, 2011). In a study of 823 cancer patients, those with sleep disturbance had an increased incidence of depression and fatigue compared to patients without any sleep disturbance (Palesh, et al., 2010). Depression and anxiety are considered th e most common mood disturbances in cancer patients (Fiorentino et al. 2011). Depression is reported at a 10 25% prevalence rate in cancer patients, but has been reported as high as 59% (Donovan & Jacobsen, 2007). Researchers found that if depression was managed, that fatigue and sleep disturbance symptoms decreased (Donovan & Jacobsen, 2007). In a study of 192 cancer patients, depression was found to occur in 45% of the sample (Breen et al., 2009). In this same study, anxiety was prevalent in 45% of th e cancer patients. Oncology Single, Multiple, and Symptom Cluster Research In the oncology literature, the emphasis has been on investigating isolated symptoms. A literature review examining symptom cluster research in light of symptom management resear ch was conducted with findings confirming that the majority of clinical studies in oncology on pain, fatigue, and depression focused on only one symptom (Miaskowski, Dodd, & Lee, 2004). The authors noted that patients rarely present with just one symptom a nd emphasized the need to assess prevalence rates of clusters of symptoms and their effect on patient outcomes. The following year a literature review indicated the need to identify the characteristics, antecedents, and consequences of symptom clusters (K im et al., 2005). Fan, Filipczak, and Chow (2007) reviewed cancer symptom cluster research between 1997 and 2006 and found 12 studies testing statistically for a symptom cluster. The authors noted that some publications involved symptoms of pain, fatigue, sleep disturbance, and depression or these four symptoms together, but they did not statistically validate the symptom cluster.
27 Pain as a single symptom was compared in male and female patients with unspecified cancer diagnoses (Donovan, Taliaferro, et al. 2008). Fatigue was reviewed in a systematic review of the literature of 55 studies (Donovan, & Jacobsen, 2010). Sleep disturbance was researched in advanced cancer patients receiving palliative care, cancer patients receiving chemotherapy, and cance r patients prior to surgery consecutively (Delgado Guay, Yennurajalingam, Parson, Palmer, & Bruera, 2011; Palesh et al., 2010; Savard et al. 2011). Multiple symptoms were studied in oncology with symptom distress studied in cancer patients, with a contr ol group of non cancer patients, with pain and fatigue identified as most severe (Cleeland et al., 2000). Symptom burden was studied as a predictor of depression and anxiety prior to chemotherapy with percentages of occurrence given for pain, sleep distur bance, depression, and anxiety (Breen et al., 2009). Donovan, Hartenback, and Method (2005) studied women with ovarian cancer specifically and reported 12 concurrent symptoms with fatigue, bowel disturbances, and peripheral neuropathies as the most preval ent. These multiple symptom studies are differentiated from studies that follow in which the authors specifically addressed symptom clusters in oncology. Sy mptom clusters studied in 30 research articles reviewed by Gift (2007) revealed the need to focus o n one type of cancer to delineate specific symptom clusters for a specific cancer. Gift recommended looking at the interactive or additive effects of symptoms, since relieving one symptom could reduce the burden of another co occurring symptom. Conceptua lly examining symptom cluster research over a six year timeframe led to the conclusion that symptom management research was focused on
28 single symptoms rather than on the simultaneous occurrence of symptoms in cancer patients (Miaskowski et al. 2007). The authors proposed that symptom clusters needed to be identified within and across cancer diagnoses, treatments, and stage s of disease. Donovan and Jacobsen (2007) completed a literature review on articles published between 1998 and 2004 to try to documen t evidence for fatigue, sleep disturbance, and depression to be designated as a symptom cluster in cancer with multiple studies found using individual tools to measure these symptoms. Pain and fatigue were most prevalent with sleep disturbance occurring l ess frequently in the symptom cluster identified in patients receiving chemotherapy (Dodd, Miaskowski, & Lee, 2004; Dodd, Miaskowski, & Paul, 2001; Hoffman et al. 2007). Data were relayed in two research articles using the same dataset of oncology patien ts receiving radiation with symptom clusters identified as mood cognitive, sickness behavior, or treatment related (Kim et al., 2009a; Kim et al., 2009b). The three clusters designated included the individual symptoms of pain, fatigue, sleep disturbance, nausea, lack of appetite, dyspnea, sweats, and urinary difficulty. Various combinations of symptom clusters were noted in advanced or metastatic cancer (Stage IV or distant stage) patients, but overall the symptom clusters dealt with pain, sle ep disturban ce, depression, anxiety, gastrointestinal symptoms affecting appetite, and dyspnea with no relationship to outcomes (Cheun, Le, Gagliese, & Zimmermann, 2011; Cheung, Le, & Zimmermann, 2009; Jimenez et al., 2011). Symptom cluster studies dealing with activ ity influenced by pai n and by the support or relationships with others r elated to quality of life as the outcome (Hadi et al., 2008; Hird et al., 2010).
29 Single symptom studies may be helpful if the healthcare provider is able to extrapolate what is needed to treat individual symptoms. Since this would rarely give an integrated view of the oncology patient who is experiencing multiple symptoms, their usefulness is limited. Since cancer patients present most frequently with multiple symptoms, an examinatio n of these symptoms in light of clustering brings a more complete picture of what a cancer patient is experiencing leading to the management of these symptoms concurrently. In addition to looking at oncology from the perspective of single, multiple, and c lustering of symptoms, researchers need to be aware of the effect background characteristic s have on the symptom experience as well as how symptoms relate to outcomes. Relationship Between Background Characteristics and the Symptom Experience in Oncology The background characteristics of age, ethnicity, educ ation, relationship status, employment status, and comorbid conditions have been associated with the symptom experience in oncology. Age was related to pain with younger ( 60) metastatic cancer patient s experiencing worse pain than older ( 60) patients (Cheun, Le, Gagliese, & Zimmermann, 2011). Outpatients undergoing treatment (N=191) were questioned regarding pain, as well as fatigue, sleep disturba nce, and depression (Miaskowski et al., 2006). Subgr oups were designated based on high to low occurrence of the symptoms. Those in the all high group reported a high occurrence of all four symptoms of pain, fatigue, sleep disturbance, and depression. The all low group described low occurrences of all four symptoms of pain, fatigue, sleep disturbance, and depression. The outpatients in the all high group were significantly younger at 54.4 (SD 12.8) than those in the all low group at age 62.4 (SD 12.3). Pain, fatigue, and sleep disturbance
30 were also assess ed in a study of 80 lung cancer patients with age in this study showing no significant effect on the symptom experience (Hoffman et al. 2007). Sleep disturbance was significantly lower in older ( 58) versus younger patients (Palesh et al., 2010). A poten tial rationale for this finding was given that younger patients may experience an increased symptom burden as well as more aggressive tumors. In a literature review of 18 studies only two of the studies looked at the relationship of age and symptoms (Kim, Dodd, et al., 2009). The two studies revealed conflicting results, with one study noting that age was weakly correlated to distress with younger patients having greater symptom distress than older patients, while the second study found patients older tha n age 70 reported higher symptom distress than patients younger than age 40 (Kim, Dodd, et al., 2009). In a study looking at physical symptoms of long term survivors (N=863) who were more than five years post diagnosis, with 73% age 50 or older, it was fo und that 17.6% experienced two or more symptoms with sleep disturbance the most prevalent symptom at 13.1% (Zucca, Boyes, Linden, & Girgis, 2012). In addition to looking at age in relationship to cancer symptoms, ethnicity, education, relatio nship status, and employment status are also presented. Ethnicity was highlighted with a statement that being an ethnic minority was associated with the increased likelihood of having pain inadequately treated (Donovan, Taliaferro, et al. 2008). Education was noted as being associated with inadequate pain treatment (Donovan, Taliaferro, Brock, et al. 2008). Being married or partnered was associated with being in a group reporting low symptoms of pain, fatigue, sleep disturbance, and depression (Miaskowski et al., 2006). In this same study (N=191), those reporting high symptoms of pain, fatigue, sleep disturbance, and depression were
31 less likely to be employed than those reporting low symptoms of pain, fatigue, sleep disturbance, and depression (29% versus 36%, re spectively). Comorbid conditions need to be addressed to note any relationships between these and the symptom experience. Comorbity has been defined as a person having two or more health conditions (Albert & Freedman, 2010). In the United States, over o ne third of adults aged 65 to 79 and over two thirds of adults 80 and older have two or more chronic conditions (Albert & Freedman, 2010). Researchers studying pain, fatigue, and sleep disturbance assessed 80 lung cancer patients and found no difference i n the symptom experience related to comorbidities (Hoffman et al., 2007). Relationship B etween the Symptom Experience and Outcomes in Oncology The most commonly studied symptoms in the oncology literature are pain, fatigue, sleep disturbance, depression, and anxiety. Pain was assessed and treated in 348 patients with metastatic bone pain with subsequent improvement in function and overall quality of life (Hadi et al., 2008). Cancer outpatients who reported low levels of pain, fatigue, sleep disturbance, and depression also reported the highest functional status and quality of life compared to outpatients who had high levels of these four symptoms (Miaskowski et al., 2006). Kim, Dodd, Aouizerat, Jahan, and Miaskowski (2009) systematically reviewed 18 stu dies from 1990 to 2007 that met criteria of oncology patients undergoing active treatment experiencing multiple symptoms with negative effect on patient outcomes with sample sizes ranging from 26 to 727 with results indicating that approximately 40% of the patients experienced more than one symptom. Only five of the 18 studies looked at symptoms in relationship to functional status and quality of life. Two of these five studies looked at number of symptoms related to functional status. Functional status decreased with an increase in number of
32 symptoms or if symptom distress increased. Four of the five studies found that poorer quality of life was associated with reports of more symptoms or symptom distress. Xiao (2010) reviewed oncology symptom clusters in published literature from 1950 through January 2010. The author located 61 articles and noted that the majority of researchers selected specific clusters and then tested for these symptoms, e.g. pain, fatigue, sleep disturbance, and depression. Xiao ( 2010) indicated that most clusters identified had two or three symptoms with functional status or quality of life as the major outcomes researched. Summary of Symptom Research in the Oncology Literature The oncology research literature emphasizes the ne ed for more symptom cluster research since cancer patients present most frequently with multiple symptoms. If the symptoms are treated holistically, then patients may experience greater improvement in their overall functioning. Pain, fatigue, sleep distu rbance, depression, and anxiety are the symptoms reported to be most prevalent in the oncology literature. Across studies, the prevalence of symptoms is as follows: pain, 36 59%; fatigue, 70 100%; sleep disturbance, 30 50% (with reports as high as 72%); depression, 10 25% (with reports as high as 45%); and anxiety 45%. The study of background characteristics in relation to symptoms is limited in the published research. Generally older patients, with older age not consistently defined, experience less p ain, less sleep disturbance, and less distress than younger patients, but data is conflicting. Ethnic minority status and having less education were associated with inadequate pain treatment in one study, with no other available information in other resea rch articles. Only one article linked relationship status to symptoms with being married or partnered associated with fewer symptoms. Employment was only mentioned in the same article referring to relationship status with
33 more women working who had lower levels of symptoms. Comorbidities were mentioned in relation to symptoms in only a few articles, showing no assoc iation to symptoms. Two articles related the need to be aware of the physiological age of older adults when considering symptoms for managem e nt of care This is definitely a gap in the literature that the current study will address. The symptom experience related to outcomes in oncology relates symptoms to outcomes of functional status and/or quality of life and not to social role function s pecifically The oncology literature notes a need for cancer specific research. Using an adapted New Symptom Management Model as an organizing framework, the symptom experience of women survivors of breast cancer, potential antecedents of background char acteristics, and consequences of physical and social role functioning are explored. Symptom Research in the Breast Cancer Literature Breast cancer symptom prevalence and symptom research is presented. This will be followed by a presentation and synthesi s of how background characteristics relate to the symptom experience and how this experience relates to the outcomes. Prevalence of Symptoms in Breast Cancer Pain is reported as a symptom in 40 50% of breast cancer patients. A study (N=32) reported 54% pain prevalence one month following needle biopsy, lumpectomy, or mastectomy for women survivors of breast cancer (Starkweather, Lyon, & Schubert, 2011). In another study with a sample of 127 women with breast cancer who had undergone a mastectomy and che motherapy, 47% reported pain (Gaston Johansson, Fall Dickson, Bakos, & Kennedy, 1999). Six months following mastectomy, with or without adjuvant therapy, 52% of women survivors of breast cancer (N=174) reported pain (Fab ro et al., 2012). On average at 26 months after unilateral lumpectomy or
34 mastectomy, with or without adjuvant therapy, 47% of women survivors of breast cancer (N=3 754) reported having pain (Gartner et al., 2009). Following lumpectomy or mastectomy, with or without adjuvant therapy, 75% o f 85 women who averaged two years since a breast cancer diagnosis stated that they had pain (Khan, Amatya, Pallant, & Rajapaksa, 2012). This higher percentage for this study may possibly be explained by noting that only a visual analog scale was used to a scertain pain, whereas the other four studies used multiple questions (intensity, location, severity). Fatigue is found to occur in 60 80% of breast cancer patients. Fatigue was found in 64% of study participants (N=103) post treatment, which consisted o f surgery, with or without adjuvant therapy (Bower et al., 2011). Kenefick (2006) studied 55 women after surgery at discharge with 70% reporting fatigue. Six months later, reports of fatigue had decreased to 60% for these same women. In another study of 154 women with various stages of breast cancer having undergone surgery, with or without adjuvant therapy, 83% reported fatigue (Bender Ergyn, Rosenzweig, Cohen, & Sereika, 2005). Fatigue was reported at 91% (N=127) by women survivors of breast cancer t hat had undergone surgery, with or without adjuvant therapy (Gaston Johansson et al., 1999). This higher occurrence may be explained by the fact that fatigue was measured with a visual analog scale, with the other three studies using multiple item questio nnaires for assessment of fatigue. One study looking specifically at chronic fatigue six to 42 months off adjuvant treatment (N=304) found a much lower occurrence with 9% of women reporting fatigue six months after treatment and 13% reporting fatigue at 4 2 months post treatment (Andrykowski, Donovan, Laronga, & Jacobsen, 2010).
35 Sleep disturbance is reported to occur in 50 70% of women survivors of breast cancer (Bower et al., 2011; Fiorentino et al., 2011; Van Onselen et al., 2012). In a study of 398 pa tients, sleep disturbance was measured prior to surgery and monthly for six months (Van Onselen et al., 2012). Three groups were distinguishable using growth mixture modeling: high sustaining (high levels of sleep disturbance that continued throughout th e time trajectory of the study); decreasing (high levels that decreased over time); and low sustaining (low levels of sleep disturbance that continued). The high sustaining group reported a 55% prevalence of continued sleep disturbance (Van Onselen et al. 2012). In a study of 103 post treatment (surgery, with or without adjuvant therapy) women breast cancer survivors, 60% reported sleep disturbances (Bower et al., 2011). Mosher and Duhamel (2010) found that 70% (N=173) of metastatic breast cancer patien ts had sleep disturbance In another study of 154 women with various stages of breast cancer having undergone surgery, with or without adjuvant therapy, 89% reported sleep disturbance (Bender et al., 2005). This last study utilized an expert consensus p anel to select items for a secondary analysis, which could explain the high sleep disturbance percentage, while the other three studies utilized standardized sleep disturbance psychometric scales (Pittsburgh Sleep Quality Index or the General Sleep Distur bance Scale). With regard to mental health symptoms, depression occurs in 20 30% of women with breast cancer (Bower et al., 2011; Knobf, 2007). In a study of 103 women who had completed treatment, 25% reported depressive symptoms (Bower et al., 2011). During treatment, 36% of 215 women with breast cancer reported being depressed (So et al., 2009). Knobf (2007) reported the occurrence of anxiety at 20 30% in a literature review
36 on women survivors with breast cancer. Consistent with this range, anxiety was assessed in 21% of 215 women during treatment (So et al., 2009). In 154 women there was a 79% finding of anxiety (Bender et al., 2005). This higher anxiety prevalence in this study might be explained by the use of an expert panel to select items for representation of the symptom, while the other two studies consisted of a literature review and the use of a multiple item fatigue inventory. In women post chemotherapy, anger was found in some women and reported to be due to a preoccupation with death, a concern for recurrence of disease, and an uncertain future, but no prevalence was noted (Evangelista & Santos, 2012). Using principal component factor analysis, anger was associated with depression and fatigue. Breast Cancer Single, Multiple, and Sympto m Cluster Research In the breast cancer literature, the emphasis has been on investigating isolated symptoms. Single symptom research in breast cancer has included research into pain, fatigue, sleep disturbance, depression, and anxiety. Anger has also b een considered as a symptom. Pain was studied in women with early stage breast cancer prior to treatment (Starkweather et al., 2011) and in women post mastectomy (Fabro et al., 2012). Fatigue was studied in women with early stage breast cancer receiving adjuvant therapy (Andrykowski, 2009; Donovan, Jacobsen, Small, Munster, & Andrykowski, 2008). Another study of fatigue in women breast cancer survivors was conducted comparing this cohort to healthy women in a control group (Andrykowski et al., 2010). Fa tigue has also been studied in women breast cancer patients receiving hormonal treatment (Glaus et al., 2006). Pain and fatigue were studied individually in women awaiting surgery to determine predictors of these symptoms postoperatively (Schnur et al., 2 007). Preoperative anxiety was associated with expectations of higher
37 postoperative pain and fatigue (Schnur et al., 2007). Older participants expected less pain, while more educated participants expected more fatigue postoperatively (Schnur et al., 2007 ). Three studies on sleep disturbance in women survivors of breast cancer were conducted: post treatment for Stage I to IIIA (Bardwell et al., 2008); post treatment with local or regional (Taylor et al., 2011), and post surgery (Van Onselen et al., 2012) Depression was assessed in women who were long term survivors of breast cancer (Brunault et al., 2012). Women breast cancer survivors receiving chemotherapy were tested for depression using a questionnaire (Akin Odanye, Chioma, & Abiodun, 2011). A stu dy of women compared at diagnosis and at six months post diagnosis of breast cancer had results indicating a significant improvement over time in anxiety scores for women who perceived that their treatment would be effective and that their lives would retu rn to what they considered normal pre cancer diagnosis (McCorry et al., 2012). In addition to single symptom research, multiple symptom research has been conducted in women survivors of breast cancer. Multiple symptoms consisting of 23 symptoms from the Memorial Symptom Assessment S cale were reported by women breast cancer survivors with Stage I or II post surgery and undergoing chemotherapy with high and low symptom prevalence groups identified, with the high prevalence symptom group reporting significa ntly poorer quality of life (Gwede, Small, Munster, Andrykowski, & Jacobsen, 2008). Women survivors of breast cancer were assessed post treatment with any combination of surgery, chemotherapy, and/or radiation and were found to have symptoms of fatigue, s leep disturbance, and depression (Berger, Visovsky, Hertzog, Holtz, & Loberiza, 2012; Bower et al., 2011). Mosher and DuHamel
38 (201 0 ) studied women with metastatic breast cancer (Stage IV or distant stage) and assessed distress associated with fatigue, sle ep disturbance, depression, and anxiety with functional impairment linked to poorer sleep quality and higher levels of fatigue and depression. Some researchers specifically designated and studied multiple symptoms as symptom clusters in breast cancer. A literature review of 50 articles was conducted on pain, fatigue, sleep disturbance, depression, and anxiety as a symptom cluster prevalent in women survivors of breast cancer (Fiorentino et al. 2011). Percentages of occurrence of these symptoms were giv en for women survivors of breast cancer: pain, 52%; fatigue, 30 60%; sleep disturbance, 20 70%; depression (linked to anxiety with no percentage given); and anxiety, 33%. In the review of articles, pain was associated with sleep disturbance; fatigue was a ssociated with pain, sleep disturbance, depression, and anxiety; sleep disturbance was associated with p ain, fatigue, depression, and anxiety; and depression and anxiety were associated with each other and with sleep disturbance (Fiorentino et al., 2011). These authors concluded that breast cancer patients often have more than one symptom simultaneously or a symptom cluster. This literature review did not actually give a definitive definition of a symptom cluster nor did it confirm a specific symptom clus ter, but indicated the relations between the five symptoms studied without any significance levels discussed Three symptom clusters were identified through confirmatory factor analysis in a study of women (N=93) undergoing radiotherapy for breast cancer (Matthews, Schmiege, Cook, & Sousa, 2012). The three clusters were pain insomnia fatigue, cognitive disturbance outlook, and gastrointestinal. Three stages of breast cancer were explored in patients using secondary analysis:
39 women with early stage; women with Stage I, II, or III, and women with Stage IV metastatic (Bender et al. 2005). The authors indicated the study was unique since symptoms in a cluster were not predetermined. A symptom cluster was defined as having three or more symptoms. Hierarchica l cluster analysis identified three consistent symptoms clustering, although not at a significant level, across all stages of breast cancer: fatigue (lack of energy), cognitive impairment (memory problem or loss of concentration), and depression and anxiet y (Bender et al., 2005). Individual symptoms were prevalent at a significant level, but the clustering was not at a significant level. The authors recommended future studies be completed linking this symptom cluster to functional ability and quality of l ife. Pain, fatigue, and depression in advanced breast cancer patients were assessed as a symptom clu ster with shared variance linking these three symptoms (Thornton, Andersen, & Blakely, 2010). The authors defined a symptom cluster as three or more concur rent indictors of physiological or psychological disturbance that relate to one another. They indicated the need to identify symptom clusters to aid in symptom management. Recognizing that single, multiple, and symptom cluster research gives an overview of the most common symptoms exhibited by women survivors of breast cancer, the question arises of how background characteristics relate to the symptom experience of these women. Relationship Between Background Characteristics and the Symptom Experience i n Breast Cancer The background characteristics of age, ethnicity, education, relationship status, employment status, and comorbid diseases have been associated with the symptom
40 experience in breast cancer. Age was related to symptoms in several research articles. In a study of 418 women awaiting surgery, a significant negative correlation between age and expected pain was found with older women more likely to expect less postoperative pain than younger women (Schnur et al., 2007). Younger women ( 40) in a sample of 174 women post mastectomy were found to have a significantly increased risk of pain lasting beyond the expected three months healing (Fabro et al., 2012). In a study of fatigue (N=373), with 80% of the participants 50 years, menopausal sympt oms related to higher levels of fatigue (Glaus et al., 2006). In a study of sleep disturbance (N=398), women who had high levels of sleep disturbance prior to surgery and for six months after surgery were younger at age 53 (SD 10.9) than the women at age 57.7 (SD 12.1) who had low levels of sleep disturbance throughout this same timeframe (Van Onselen et al., 2012). Three studies reported younger women had greater depression and anxiety than older women with ages noted for older women as over age 50 (Howa rd, Anderson, Ganz, Bower, & Stanton, 2012), over age 57 (Khan et al., 2012), and no age given (Mosher & Danoff Burg, 2005). Age was found to have no association with depression in a study of 120 women after cancer treatment (Brunault et al., 2012). Age was related to distress or general physical and psychological problems not defined in one study (Matthews et al. 2012) and was defined as being age 50 or older in two literature reviews (Knopf, 2007; Knopf, 2011). Mosher and DuHamel (2010) found age to not be significantly associated with distress level. Et hnicity was addressed in a few studies. In a study of 139 women post adjuvant treatment, Hispanic women were mor e likely to report multiple symptoms than Black or
41 White women (Fu et al., 2009). Mosher and Duhamel (2010) found ethnicity was not associated with distress in 173 metastatic breast cancer patients comprised of Hispanic, Black, and White ethnic background s. Education was linked to symptoms in three studies. Lower educational level was associated with increased sleep disturbance (Bardwell et al., 2008), higher levels of depression (Akin Odanye, Chioma, & Abiodun, 2011), and with less distress (Kenefick, 2006). Relationship status was associated with symptoms in two studies. Being married or living with a partner produced greater symptom distress in one study (Kenefick, 2006), while marital status was not associated with depression in two other studies (Akin Odanye et al., 2011; Brunault et al., 2012). Employ ment status was discussed in three studies. Being unemployed was associated with persistent treatment related symptoms in one study (Fu et al., 2009) and was not associated with depression in anot her study (Akin Odanye et al., 2011). Being unable to return to work was independently associated with the sequelae of an advanced stage of breast cancer in a study of 96 women survivors of breast cancer (Villaverde et al., 2008). Comorbid conditions also need to be add ressed to note any relationship between these covariates and the breast cancer symptom experience. No articles were found that expressly considered any relationship. Relationship B etween the Symptom Experience and Outcomes in Breast Cancer Several studies have looked at the relationship of symptoms to outcomes in women survivors of breast cancer. Pain, fatigue, and sleep disturbance symptoms were negatively correlated with optimism, self transcendence and positive mood in 93
42 women receivi ng radiotherapy (Matthews et al. 2012). Women survivors of breast cancer were questioned about the symptoms of pain, fatigue, sleep disturbance, and depression and how these symptoms affected activity and life satisfaction (Berger et al., 2012). Pain wa s consistently associated with severity of the other symptoms and with lower functioning. In another study, women with breast cancer receiving chemotherapy, with or without radiation, were assessed for pain, fatigue, sleep disturbance, and depression rela ted to functional status and quality of life at three time points: the week prior to the second chemotherapy cycle, immediately following the conclusion of chemotherapy, and one year from the initiation of chemotherapy (Dodd, Cho, Cooper, & Miaskowski, 201 0). Subgroups of mild, moderate, and all high symptoms were identified through cluster analysis. At one year after treatment initiation, the women with all high symptoms reported significantly poorer physical functional status and quality of life. At the conclusion of chemotherapy women were questioned regarding pain, fatigue, and depression with moderately, significant correlations found between these three symptoms to each other and with pain and fatigue moderately and significantly correlating with hea lth status (Gaston Johansson et al. 1999). Pain, fatigue, depression, and anxiety were assessed related to quality of life (So et al., 2009). Women (N=215) receiving chemotherapy had higher levels of these four symptoms with poorer quality of life score s than those receiving radiotherapy. The authors found that there were significant, moderate correlations to support a cluster for the four symptoms. Pain, depression, anxiety, and stress symptoms were assessed in a study of 85 women post treatment and were associated with functional independence and well being (Khan et al. 2012). Women in this study reported on average two years post
43 treatment: breast related pain, 75%; depression, 22%; anxiety, 19%; and stress, 19%. The majority of these women repor ted minimal change in their physical functioning with one third noting the highest impact on their psychological well being. Fatigue, sleep disturbance, and depression were studied in 90 metastatic breast cancer patients with higher levels of these symp toms resulting in greater functional impairment (Mosher & DuHamel, 2010). Fatigue, weight gain, and altered sexuality reported following chemotherapy adversely affected quality of life (Wilmoth, Coleman, Smith, & Davis, 2004). Treatment related lymphede ma symptoms in women survivors of breast cancer were assessed with a cluster identified with five symptoms including fatigue, limb sensation, loss of body confidence, decreased physical activity, and psychological distress (Ridner, 2005). Quality of life was the outcome in this study with overall poor quality of life reported. Pain, fatigue, sleep disturbance, depression, and anxiety have been studied in various combinations, as well as with additional symptoms, in women survivors of breast cancer with a nger not investigated in any detail. The outcomes of quality of life and physical functioning have been used most frequently. However, a variety of other outcomes have also been utilized. Summary of Symptom Research in the Breast Cancer Literature The l iterature review of research studies on women survivors of breast cancer illustrates the variety of symptoms that various researchers considered during diverse timeframes, various treatment modalities, and various stages of the breast cancer experience. Th e research to date has included studies of single symptoms and multiple symptoms. Women survivors of breast cancer most often present with multiple symptoms. Several different symptom clusters were identified in various studies. There
44 was no definitive number in a cluster, with cluster size ranging from two to five symptoms within the cluster. Pain, fatigue, sleep disturbance, depression, and anxiety are the symptoms reported to be most prevalent in the breast cancer research. The relationships of back ground characteristics to symptoms are limited in the breast cancer research. Older women, with various ages denoted for older, were found to more likely expect less postoperative pain, have higher levels of fatigue, and less sleep disturbance, depression and anxiety than younger women. Studies conflicted on whether distress was higher or lower in older women. Studies were few and inconsistent rega rding ethnicity, education, relationship status, and employment status. Hispanic women were found to more likely report multiple symptoms than Black or White ethnicities in one study with ethnicity not associated with distress in another study. Lower educational levels were associated with increased sleep disturbance and depression but with lower distress in various studies. Being in a relationship was associated with greater symptom distress in one study, with marital status having no association to depression in two other studies. Being unemployed was linked to persistent symptoms in one study and had no a ssociation to depression in another study. The symptom experience related to outcomes showed symptoms related to a variety of outcomes. Outcomes included optimism, well being, quality of life, activity or functioning, and illness pe rc eption, but not spe cifically social role function. The sample sizes in the research studies reviewed ranged from 23 418, with the majority of the sample sizes below 135. There were two outliers, a study of 2,645 participants who were part of a larger dietary study, and 3, 754 participants who were
45 part of a nationwide study. Hierarchical, factor, regression, and cluster analysis were the most common methodologies used. The significant findings that reveal the gaps in the literature requiring a production of new knowledg e include a more inclusive view of the symptom experience looking at the physical hea lth symptoms of pain, fatigue, sleep disturban ce and the mental health symptoms of depression, anxiety, and anger in relationship to the outcomes of physical and social r ole functioning. There needs to be a consistent designation of symptom clusters for consistent reporting. There is a known higher incidence of older women surviving breast cancer, but little research linking age to the symptom experience. Many questions remain about the symptom experience of women survivors of breast cancer and the relationship of the symptom experience to background characteristics, especially age, and to functional physical and social outcomes. This study begins to address this gap in the empirical literature.
46 CHAPTER 3 METHODS Introduction The research aims ar e investigated using a descriptive, exploratory, cross sectional, secondary analysis study design. This meets the criteria for choosing a study design that is feasible, interesting, novel, ethical, and relevant (Hulley, Cummings, Browner, Grady, & Newman, 2007 ) The study is feasible with a secondary analysis data set utilized. The study content is interesting and novel as indicated by the literature review of limited re search on sympt om clusters in oncology Ethical considerations were met by obtaining University of Florida Health Science Center Institutional Review Board approval and maintaining data security. The study is definitely relevant with the increase of wome n breast cancer survivors and the need to identify and desc ribe the symptom experience. By identifying symptom clusters in women survivors of breast cancer, this knowledge base can be expanded. The symptom clusters can be addressed and treatment strategi es implemented by healthcare providers for the most positive physical and social functioning as women deal with breast cancer survivorship Design Th is was a descriptive, exploratory, cross sectional, secondary analy sis of the data set from the Patient Repo rted Outcomes Measurement Information System (PROMIS) Wave 1 from the National Institutes of Health (NIH) (PROMIS information retrieved from http://www.nihpromis.org/ ). The study describe s and explore s the symptom e xperience of women breast cancer survivors. The relationship between the symptom experience and the outcomes of phys ical and social role function were
47 investigated. Background characteristics w ere identified and associations with the symptom experience w ere examined NIH PROMIS Dataset Overview In 2004 a group of scientists from several United States academic institutions joined with the NIH to develop and evaluate measurements of p atient reported outcomes for a wide variety of chronic diseases to be ma de publicly available to the clinical research community (Cella et al., 2010). Based on the World Health Organization framework domains were developed that resulted in three general components: physical, mental, and social health (Cella et al., 2007). Gl obal items 21,133 participants who self reported on these components within the domains (Cella et al., 2007; Cella et al., 2010). The five subdomains selected for initial item development were pain, fatigue, mental or emotional distress, physical functioning, and social role functioning (Cella et al., 2007). The physical health component included the domains of physical function, fatigue, pain, and sleep disturbance. The mental health component included the emotional distress domain including depression, anxiety, and anger. The social health component included the domains of social role performance and social role satisfaction. In addition to the domains, single item glo bal indicators of pain, fatigue, mental or emotional distress, physical functioning, and s ocial functioning were added. These are referred to in the PROMIS dataset as global health items (Cella et al., 2010).
48 PROMIS Sample The PROMIS Wave 1 total sample dataset included 21,133 participants with data collected from July 2006 to March 2007. These participants were recruited from primary PROMIS research sites (n=1,532) or fro m YouGovPolimetrix (n=19,601). PROMIS network research sites included Duke Univer sity University of Pittsburgh, Stanford University and Un iversity of North Carolina Recruitment from YouGovPolimetrix (Polimetrix) involved selection from a panel of over one million respondents who regularly participated in online surveys and had give n pertinent personal information such as names and addresses. Polimetrix is a polling firm with a web portal allowing participants to express their views regarding public policy and current issues. These panelists were recruited via e random digit dialin g, web newsletter invitation, or through the internet where participants agree to participate in a survey for a compensation of less than $10. Polimetrix uses sample matching to obtain representative samples in a target population. The target selected fo r PROMIS was by gender (50% female), age (20% in e ach of five age groups: 18 29, 3 0 44, 45 59, 60 74, over 75), race/eth nicity (12.3% African American and 12 .5% Latino/Hispanic to match the U. S. census), and education (10% less than high school graduate), with the added target of being representative of the U. S. general population (based on the 2000 U. S. census). The overall sample (N =21,133) consisted of 52% female, median age of 50 years, with 12% age 18 29, 12% age 30 39, 16% age 40 49, 32% age 50 64, and 28% age 65 and older. Ethnicity included 82% White, 9% Black, 8% multi racial, and 1% Asian/Pacific Islanders and Native Americans. Hispanic or Latino comprised 9% of the sample population. Education was 3% less than high school, 16% high school di ploma, 39% some college, 24% college degree, and 19% advanced degree.
49 A non clinical sample was designated as the general population. A clinical sample was designated if the participant reported receiving a specific diagnosis from a physician. The diagn osis or diagnoses related to the presence and degree of limitation of 25 chronic medical conditions such as coronary artery disease, stroke, diabetes, or cancer. The clinical sample (n=7,883) was recruited from three universities who participated as PROMI S network research sites ( University of Pittsburgh, Stanford University and Duke University ) and from Polimetrix. The mean age of this subsample was 57.4, with 4.1% age 18 29, 7.6% age 30 39, 13.9% age 40 49, 27.2% age 50 59, and 15.3% age 60 64, 31.2% 6 5 84, and 0.7% age 85 and older. In this clinical sample, 46.7% were female Ethnicity was reported as 88.6% White, 5% Black, 5.2% multi racial, 0.7% Asian/Pacific Islanders and Native Americans, and 0.5% Asian. In addition, 3.8% of the sample w as Hispa nic or Lati no A subsample from the clinical sample (n=7,883) consisted of participants with breast cancer (n=106). The breast cancer participants consisted of 1.3% of the total clinical sample. Figure 3 1 gives the overview of the total PROMIS sampling design. When the data set was obtained from PROMIS, the breast cancer participants were found to be from the clinical sample from Polimetrix only (n=7,080 ). Three of these participants were male, so the final number of women survivors of breast cancer use d for the study was 103. Study Sample The sample of 103 women survivors of breast cancer ranged in age from 30 to 85, with a mean age of 60 .4 years (SD = 10.1). The vast majority of the sample (n=92, 89%) were between the ages of 50 and 85 The majority we re ethnically W hite (n=98,
50 95%) highly educated with a college or advanced college degree (n=77, 75%) partnered (n=62, 60%) and employed (n=55, 53%) One third of the sample had no comorbid conditions, another third had only one comorbid condition, wit h the final third having two to five comorbid conditions. Table 3 1 provides a detailed description of the sample of 103 women survivors of breast cancer. Measure s Original PROMIS Measurement Bank All PROMIS instrument items were identified from a system atic search of measures with established psychometrics, as well as new item development by experts working in each domain area. The items were selected, classified, reviewed, and revised by domain experts. Focus groups confirmed definitions and assisted in identifying any new areas for item development in the future. Items were then sent for field testing. Final revisions were made after field testing. Each of the domains had an investigative research team of experts in measuring and assessing each sp ecific domain area. Each team completed an extensive literature review. Item response theory was utilized. Item development was accomplished using existing questionnaires as well as new item generation. Psychometric results were analyzed and presented t o each domain working group. Response options use a five option scale: 1=poor or not at all or never or none, 2=fair or a little, or rarely or mild, 3=good or moderate ly or sometimes or moderate, 4= very good or mostly or often or seve re, 5= excellent or com pletely or always or very severe. It was determined that there was good discrimination of item fit with these five options. Respondent burden was considered in keeping responses available consistent. The pain intensity scale had an 11 point intensity sc ale, ranging from 0=no pain to 10=worst imaginable pain. The
51 response categories were pre tested to confirm patient understanding followed by field testing for item calibration (Cella et al., 2010) Measurement of Key Study Variables For the purpose of this s tudy, a subset of variables was selected representing the study concepts. These variables and their measurement are described below. Five global variables from t he PROMIS Wave 1 testing were analyzed. The three independent variables include d the p hysical health symptoms of pain and fatigue and the mental health symptom of emotional distress ( depre ssion/anxiety/ anger ) The two dependent variables were physical functioning and social role functioning The physical health symptom of sleep disturbanc e was not available for analysis since this variable was not included in the Polimetrix clinical sample which included the breast cancer participants Background character istics include d age, ethnicity, education, relationship status, and employm ent statu s. Covariates include d the number of comorbid disease conditions. Independent v ariables Pain : Pain was measured by a single item indicating the self reported average pain intensity rating in the past seven days. Pain intensity was rated on an ordina l 11 point scal e, ranging from 0=no pain to 10= worst imaginable pain. This variable was recoded to evaluate the prevalence of pain (0=no pain, 1 =pain). Fatigue : Fatigue was measured by a single item on which participants rated their average level of fa tigue over the past seven days. The response choices ranged ordinally from one to five: 1=none, 2=mild, 3=moderate, 4=severe, and 5=very severe. This variable was recoded to evaluate the prevalen ce of fatigue (0=no fatigue, 1 = fatigue).
52 Depression/Anxiet y/ Anger : Depression, anxiety, and anger were measured been bothered by emotional problems, such as fee The resp onse choices range d ordinally from one to five: 1=never, 2=rarely, 3=sometimes, 4=often, and 5=always. This variable was recoded to evaluat e the prevalence of depression/anxiety/anger (0=no depression/anxiety/anger, 1=depression/anxiety/ anger) Dependent v ariable s Physica l Functioning : Physical function was measured by self report of the extent that participants were able to carry out everyday physical activities such as walking, climbing stairs, carrying groceries, or moving a chair. Th is was a single item, with possibl e responses ranging ordinally from one to five: 1=not at all, 2=a little, 3=moderately, 4=mostly, and 5=complet ely Since only th r e e participants rated their functioning in the 1 3 categories this variable was recoded. No participant responded were recoded as missing data T was group. This left 101 participants who could either mostly or completely carry out thei r everyday physical activ ities resulting in a binary variable. Social Role Functioning : Social role functioning was measured by a single item ascertaining self report of satisfaction with social activities and relationships. The responses ranged ordin ally from one to five: 1=poor, 2=fair, 3=good, 4=very good, and 5=excellent. The variable was left in its original ordinal form.
53 Background c haracteristics Demographic variables were assessed via a standardized clinical form. age, ethnic ity, level of education, relationship status, a nd employment status were collected This same clinical form asked participants to report a diagnosis confirmed by a physician of any of 25 chronic medical comorbidities or conditions such a s hypertensio n or heart attack Age was coded as a continuous variable by actual chronological age as well as by a binary ordinal variable of 50 or 50 years. Ethnicity was recoded as a categorical, nominal variable of White, Black or other. L evel of education was reco ded into a binary variable of high school graduate/AA/technical degree or college/advanced degree. Relationsh ip status was recoded into a dichotomous variable of partnered (married or living with partner in committed relationship) or not partnered (never married, separated, divorced, or widowed). Employment status was recoded as a dichotomous variable of employed (full time or part time) or not employed (unemployed, homemaker, retired, or on disability). This was the only variable that had missing data, with four participants not having employment status determined. Como rbidy was recoded from zero to five indicating the number of comorbid conditions as a categorical variable. Procedures Protection of Human Subjects Institutional Review Board approval f rom the University of Florida Health Science Center was obtained for this study via electronic submission The protocol number is IRB201200247 and exempt status was received for data collection determined to be non human/exempt, completely anonymous infor mation, with de identified subjects.
54 Florida myIRB. Data Acquisition A research proposal was submitted to the National Institute s of Health (NIH) requesting the PROMIS Wave 1 database This proposal was approved in November 2012. T he NIH PROMIS data were transmitted to the investigator in an encrypted, password protected file The data were de identified by the NIH staff prior to transmission. The data are stored and secure d on a n encrypted share drive in the University of Florida College of Nursing with access only by the researcher and the also encrypted by the Universit y of Florida College of Nursing Information T echnology Department to comply wit h all university and college security policies. The data are accessible only by passcode. Statistics The statistical software, PAWS Statistics version 18 (formerly SPSS) was used for the univariate, bivariate, and multi variate analyses. Data were transmitted in SAS format and imported into the PAWS database. Results were considered statistically significant based on a p value of less than .05. The analysis plan and procedures are presented below. Aim 1 T he prevalenc e and number of self reported physical health symptoms (pain, fatigue) and mental health symptom (depression/anxiety/anger) among women survivors of breast cancer were determined using univariate, descriptive statistics including frequency data of numbers and percentages, means, and standard deviations.
55 The three independent variables were each recoded into dichotomous, categorical variables to determine prevalence. These same variables were used with their ordinal values to determine symptom intensity. Aim 2 Bivariate correlational analysis examined relationships between symptoms. Since the independent variables of physical and mental health symptoms were not norm ally distributed, the non s correlation was used for the analysis. Associations were confirmed and multivariate analysis was conducted to determine how symptoms combined. H ierarchical cluster analysis revealed preliminary information leading to nonhierarchical k means cluster analysis to determine the identifiable sympto m clusters in women survivors of breast cancer. Aim 3 Univariate analysis determined that none of the background characteristics (age, ethnicity, education, relationship status, employment status, comorbid conditions) were normally distributed. The dat a were categorical ordinal or nom inal, with the exception of age, which was a continuous variable. Non parametric bivariate analysi s which is distribution free, indicated for categorical data, with the assumption of independence of data, was therefore employed to investigate the relationships between the background characteris tics and the symptom clusters. The Kruskal Wallis test (non parametric equivalency of the one way independent ANOVA ), contingency tables, chi square, and were the statistics utilized to determine any relationships between the background characteristics and the symptom clusters. Due to being unable to run more than a 2x2 contingency table in PAWS (version 18), SAS ( version 9.3 ) was used
56 for tables over 2x2, so assumption of no more than 10% of the cells having expected frequencies below 5 was violated. Aim 4 Univariate analysis with frequencies and percentages was accomplished for description of the phy sical and social role function dependent outcomes. The physical function variable was not normally distributed and was recoded to a categorical, ordinal variable. Bivariate analys is with a contingency table, chi square was used t o investigate the relationship between symptom clusters and the physical function outcome. The social function variable was also not n ormally distributed. T he categorical, ordinal variable was retained in its original form. Univariate analysis compared mean s and standard deviations of level of social function in the clusters. Bivariate analysis with the Kruskal Wallis test (non parametric equivalency of the one way independent ANOVA), followed by the Wilcoxon rank sum test (non parametric equivalent of the independent t test) was utilized to investigate the relationships between the symptom clusters and the social role function outcome.
57 Clinical N=7,883 Breast Cancer n=106 PROMIS Total Sample N=21,133 Full Bank Testing n=7,005 Network Site n = 329 Polimetrix n = 6,676 Network Site n=1,203 Block Testing n= 14,1 28 Polimetrix n=12 ,925 Clinical n=7,080 General Population n=5,845 Clinical n=803 General Population n=400 General Population n=6,676 General Population n=329 Figure 3 1. PROMIS sample d esign
58 Table 3 1. Description of sample (N=103) Variables N (%) Mea n (SD) Ag e, age ranges: 30 85 Age Group 50 50 103 (100%) 11 (11%) 92 (89%) 60.4 (10.1) Ethnicity White Black Other 98 (95%) 2 ( 2%) 3 ( 3%) Hispanic Yes No 4 ( 4%) 99 (96%) Education High s chool grad/AA or tech nical degree College/ advanced degree 26 (25%) 77 (75%) Relationship status Partnered Not partnered 62 (60%) 41 (40%) Employment status Employed Not employed Not determined 55 (53%) 44 (43%) 4 ( 4%) Number of comorbid condit ions 0 1 2 3 4 5 35 (34%) 33 (32%) 21 (20%) 12 (12%) 1 ( 1%) 1 ( 1%) *Relationship status is coded as partnered or not partnered. Partn ered included being married or living with a partner in a committed relat ionship. Not partne red indicated being never married, separated, divorced, or widowed.
59 CHAPTER 4 RESULTS Th e following sections present the results of the statistical analyses of the symptom experience among women breast cancer survivors. Results are organized according to the study research aims. Aim 1: To Descr ibe the Prevalence and Number of Self R eported Physical Health S ymptoms (Pain, Fatigue) and Mental Health Symptom (Depression/Anxiety/Anger) Among Women Survivors of Breast Cancer S ymptoms for women survivors o f breast cancer included both physical and mental health symptoms. Pain was the most common symptom reported with 67% of the sample experiencing this symptom. Fatigue was reported by 62% of the sample. Similarly, 63% of the women in this sample reported experiencing the mental health symptom (depression/anxiety/ anger ) The mean intensity of these symptoms was 1.6 (SD= 1.8) for pain (0 10 scale ) 1.8 (SD= 0.7) for fatigue (1 5 scale) and 1.9 (SD=0.8) for mental health (1 5 scale) These intensity scores indicate low symptom severity (see Table 4 1). The majority of women survivors of breast cancer in this sample report ed experiencing two or three symptoms (n=68, 66%). Eleven women (11%) reported no symptoms and 24 (23%) reported experiencing only one sym ptom (see Table 4 2). Aim 2: To Determine Whether and How Symptoms Combine t o Create Identifiable Clusters In A Sample Of Women Survivors Of Breast Cancer Bivariate Analysis The results of correlation analysis revealed statistically significant associat ions between symptoms. Since the independent variables were not normally distributed, the non was used to examine relationships C orrelation s w ere as follow s : pain mental health (r s =.26, p .01), pain fati gue (r s =.42,
60 p 001), and fatigue mental health (r s =.34, p .001) indicating small to moderate relationships between these symptom variables, and leant support for the iden tification of symptom clusters Multivariate Cluster Analysis Hierarchical cluster analysis wa s selec ted to analyze the data using the independent symptom variables to reveal any natural groupings or clusters. This type of cluster analysis lends itself to smaller samples of less than a few hundred ( PAWS, version 18 ) The agglomerative method was selecte d with each participant with their specific symptom(s) forming the initial individual clusters, progressing to similar clusters merging to form new clusters based on similarity or proximity based on distance (Hair & Black, 2000; Meyers, Gamst, & Guarino, 2 013). One goal of cluster analysis is to maximize the distance between cluster s. This facilitates detection of true, independent clusters In this study, the squared Euclidean distance which is the distance between any two clusters determined by the sum of the score differences across all participants was used (Hair & Black 2000; Meyers et al. 2013) The clustering algorithm used in the majority of the cluster research reviewed was used for this study: average linkage or between group linkage. Th is average Since the pain measurement (0 10 scale) varied from the fatigue (1 5 scale) and the mental health (1 5 scale) measurement, standardization procedures were used to create z scores. These z scores are standardized to a mean of 0 and a SD of 1. The analysis revealed from two to eight clusters with no convergence into a parsimonious solution. Analysis of the cluster dendrograms revealed that a three
61 cluster solutio n showed the most discrimination and reduced the outliers. Thus, the decision was made to proceed to the nonhierarchical k means cluster analysis which has been noted to com plement hierarchical clustering (Hair & Black 2000). A second type of cluster a nalysis, k means clustering is indicated with a larger number of cases or participants with fewer variables (Meyers et al., 2013 ). Given that only three symptoms were being analyzed, this method was thought to be appropriate. This clustering method uses the agglomerative procedure with clusters joined at each stage. The process is also considered iterative as each stage begins over again based on the results of the prior stag e. The Euclidian distance between the center of each cluster or an average of the clustering variables and the non clustered variables occurs with the cluster with the smallest distance absorbed within the next closest cluster (Meyers et al., 2013) The major difference between hierarchical and k means clustering is that the numb er of clusters is specified a priori Three clusters were stipulated by the researcher based on the preliminary hierarchical clustering results and symptom intensity score s. Two and four clusters were also examined and rejected, with two clusters exhibiting less discrimination and little variability and four c lusters adding a one participant outlier. Using k means cluster analysis, results are presented based on the final cluster centers and cluster membership as the significant compone nts (Meyers et al., 2013). The final cluster centers are the z score means of the cluster variables (Meyers et al., 2013) The symptom variables were standardized and transformed into z scores. Iteration is recommended to be set at 50 to assure a criter ion threshold is reached. Iteration continues until 50 is reached or until the threshold is revealed. In effect, a
62 reassignment of membershi p occurs at each iteration with the cluster centers based on each new cluster reconfiguration until convergence is achieved due to no change or a very small change in the final cluster centers. Based on the k means approach, a 3 cluster solution emerged from the data. Convergence was reached in six iterations. Univariate ANOVAs indicated that the clustered groups differed significantly on all three symptom variables (all p values < .001). The final cluster centers together with the nu mber of cases in each cluster are shown in Table 4 3. The range of n for each cluster was 8 to 53, with two clusters r el atively equivalent in size ( 42 and 53 cases) Cluster 1 (n=53) can be characterized as all minimal symptoms for pain, fatigue and mental health Cluster 2 (n=42) had all mild symptoms. Cluster 3 (n=8) had all moderate symptoms. Aim 3: To Investigate t he R elationships Between Background Characteristics (A ge, Ethnicity, Education, Relationship Status, Employment Status, Comorbid Conditions) and Symptom C lusters The investigation of the relationships between background characteristics (age, ethnicity, e ducation, relationship status, employment status, comorbid conditions) and the three identified clusters was completed using bivariate analysis. Non parametric statistics were used since the variables were not normally distributed, were categorical and m et the assumption of independence of data. These measures included the independent samples Kruskal Wallis test, contingency tables chi exact test value when the frequency assumption was violat ed with sparseness of the cells not meeting the assumption of no more than 10% of the cells having the expected frequency below 5 (Portney & Watkins, 2009). No significant relationships between any of the background characteristics and the three symptom cl usters were found (see Table 4 4)
63 Age was left as a continuous variable when looking at the cluster groupings, as well as compar ing younger ( < 50 years ) versus older ( 50 years ). The independent samples K ruskal Wallis test indicated that age was not sig nificantly different across the three clusters ( 2 =4.7, p=. 0 97). The data show ed ( All Moderate Symptoms ) were 10 years younger than the women in Cluster 1 ( All Minimal Symptoms ) Due to the small sample size there was limi ted powe r to detect an age difference. When comparing younger (< 50 years ) to older ( 50 years ) women survivors of breast cancer via contingency tables, chi trend of women being younger as symptom severity incre ased from Cluster 1 to Cluster 2 to Cluster 3 ( 2 =6.6, p=.0 6 ). Ethnicity and cluster membership was evaluated with and no significant relationship was found ( 2 =1.4 p=.89). Educatio n was examined with contingency tables and ch i squ are and revealed no significance between education and cluster groupings ( 2 =. 03 p = .98 ). The assumption of frequency of no more than 10% cells below 5 was met, so no further testing was required. Relationship s tatus revealed through contingency tables a nd chi square analysis a need to run due to low cell frequencies. N o significant relationship between being partnered or not partnered and the cluster groups was found ( 2 =. 40 p=.86). Contingency tables and chi square analysis ad relationship between employment and cluster groupings ( 2 =.17 p=1.00). Employment status was the only background characteristic for which the entire sample did not give complete information. The st atistics were evaluated based on 99 rather than 103 women survivors o f breast cancer. Contingency tables and chi square analysis
64 between the number of comorbid conditions and the cluster groups ( 2 = 17.36 p=.29). Aim 4: To Investigate the Relationships Between Symptom C lusters and the F unctional O utcomes ( P hysical F unction Social Role F unction ) The relationships between the identified symptom clusters and the functional outcomes of physical function and social role function sh owed significant relationships. Univariate analysis showed the frequencies and percentages of the sample responses for physical function in Table 4 5 for the extent the women were able to carry out physical activities such as walking, climbing stairs, or carrying groceries. The majority carry out physical activities. Social role function of satisfaction with soc ial activities and relationships scoring with frequencies and perce ntages is exhibited in Table 4 6 for this study sample The majority of women (n=92, 89%) were satisfied with their social role activities and relationships by rating social role function Bivariate analysis w as conducted with c ontingency tables and chi square tests and the expected cell count frequencies less than 10% required A significant relationship between the three identif ied cluster groups and the physical function outcome ( 2 =6.9 p=.03) was revealed As the symptom severity incre ased from Cluster 1 ( All Minimal Symptoms ) to Cluster 2 ( All Mild Symptoms ) to Cluster 3 ( All Moderate Symptoms ) the level of physical functio n ing was lower. See the final cluster analysis of physical function in Table 4 7 Social role function was examined with univariate analysis of means and standard deviations The independent samples Kruskal Wallis test revealed a
65 significant difference ( 2 =11.9, p=.00) between the cluster groups. The mean social role function values were lower in clusters with higher symptom severity (see Table 4 8). With this significant result, a multiple comparison of cluster groups was conducted using the bivariat e analysis of Wilcoxon rank sum test (non parametric equivalent of the independent t test) to identify which clusters were different. Clusters 1 and 2 were not significantly different related to social role function ( W s =1783.5, p=.07). Clusters 1 and 3 w ere significantly different related to social role function ( W s =103.5, p=.00). Clusters 2 and 3 were also significantly different related to social role function ( W s =120.0, p=.0 3 ). When comparing the means and standard deviation in the three cluster grou ps regarding social role function the means were significantly lower from Cluster 1 to Cluster 3 and from Cluster 2 t o Cluster 3, indicating that higher symptom severity was associated with lower social role func tioning (see Table 4 9 ).
66 Table 4 1. Prevalence and intensity (0 1=none) of symptoms (N=103) Symptom N (%) Mean (SD) Pain Prevalence Intensity (0 10) 69 (67%) 1.6 (1.8) Fatigue Prevalence Intensity (1 5) 64 (62%) 1.8 (0.7) Mental health symptom (depression/anxiety/ange r) Prevalence Intensity (1 5) 65 (63%) 1.9 (0.8) Table 4 2 Frequency distribution of number of symptoms reported (N=103) Number of symptoms Frequency % 0 11 11 1 24 23 2 30 29 3 38 37 Total 103 100 Table 4 3. Final clust er analysis z score means on pain, fatigue, and mental health variables Variable Cluster 1 All Minimal Symptoms n=53 Cluster 2 All Mild Symptoms n=42 Cluster 3 All Moderate Symptoms n=8 Pain Intensity (0 10 with 0: no pain and 10: worst pain imaginable) .60 .36 2.11 Fatigue Intensity (1 5 with 1: none and 5: very severe) .65 .47 1.86 Mental health (depression/anxiety/ anger) Intensity (1 5 with 1: never and 5: always) .43 .25 1.52
67 Table 4 4. Final cluster a nalysis relationship to background characteristics Background Characteristic Cluster 1 All Minimal Symptoms n=53 Cluster 2 All Mild Symptoms n=42 Cluster 3 All Moderate Symptoms n =8 Chi square/ p value Age ranges: 30 85 M (SD) 62.0 (10.5) M (SD) 60.1(8.5) M (SD) 51.9 (11.5) 2 =4.7 p= .10 Age Group : 50 50 n= 4, 7.5% n=49, 92.5% n= 4, 9.5% n=38, 90.5% n= 3, 37.5% n= 5, 62.5% 2 =6.6 p =.06 Ethnicity : White Black Other n=50, 94 .3 % n= 2 3.7% n= 1, 2.0% n=39, 92.8% n= 1, 2.4% n= 2, 4.8% n= 8, 100% n= 0, 0% n= 0, 0% 2 =1.4 p=.89 Education : High school grad/AA/ technical degree College/advanced degree n=13, 24.5% n=40, 75.5% n=11, 26.2% n=31, 73.8% n= 2, 25% n= 6, 75% 2 =.03 p=.98 Relationship Status : Partnered Not partnered n=32, 60.4% n=21, 39.6% n=26, 61.9% n=16, 38.1% n= 4, 50% n= 4, 50% 2 =.40 p=.86 Employment Status : (N=99) Employed Not emplo yed N=51 n=28, 54.9% n=23, 45.1% N=40 n=22, 55% n=18, 45% N=8 n= 5, 62.5% n= 3, 37.5% 2 =.17 p=1.0 0 Number of comorbid c onditions 0 1 2 3 4 5 n=21, 39.6% n=18, 33.9% n= 10, 18.9 % n= 4, 7.6 % n= 0, 0.0% n= 0, 0.0 % n=12, 28.6 % n=13, 30.9% n=10, 23.8% n= 6, 14.3% n= 0, 0.0% n= 1, 2.4% n= 2, 25.0% n= 2, 25.0% n= 1, 12.5% n= 2, 25.0% n= 1, 12.5% n= 0, 0.0% 2 =17.36 p=.29 Kruskal Wallis test was used for chi square and p value for th e age continuous variable Contingency tables were used with chi square for the age categorical variable, ethnicity, education, relationship status, employment status, and number of comorbid conditions. The p value for education only was associated with the chi square, with the remaining variables using the p (due to cell sparseness).
68 Table 4 5 Physical function responses: able to carry out physical activities (N=103) Response Frequency % 5: completely 81 79 4: mostl y 19 18 3: moderately 1 1 2: a little 2 2 1: not at all 0 0 Total 103 100 Table 4 6 Social role function responses: satisfaction with activities and relationships (N=103) Response Frequency %. 5: excellent 35 34 4 : very good 34 33 3: good 23 22 2: fair 9 9 1: poor 2 2 Total 103 100 Table 4 7. Final cluster analysis of physical function Dependent Variable Cluster 1 All Minimal Symptoms n=53 % (n) Cluster 2 All Mild Symptoms n=42 % (n ) Cluster 3 All Moderate Symptoms n=6 % (n) Chi square/ p value Physical function 4 (Mostly) 5 (Completely) 11 .3 % (6) 88 .7 % (47) 26 .2 % (11) 73.8 % (31) 50% (3) 50% (3) 2 =6.9 p=.03
69 Table 4 8 Final cl uster a nalysis of social ro le function Dependent Variable Cluster 1 All Minimal Symptoms n=53 Mean (SD) Cluster 2 All Mild Symptoms n=42 Mean (SD) Cluster 3 All Moderate Symptoms n =8 Mean (SD) Chi square/ p value Social role function ( 1=poor to 5=excellent) 4.15 (.91) 3.7 6 (1.06) 2.75 (1.04) 2 =11.9 p= .00 Table 4 9. Multiple comparisons of social role function in clusters First Cluster for Comparison Second Cluster for Comparison Wilcoxon (W s) / p value Cluster 1 All Minimal Symptoms n=53 C luster 2 All Mild Symptoms n=42 Mean (SD) 4.15 (.91) 3.76 (1.06) W s =1783.5 p=.07 Cluster 1 All Minimal Symptoms n=53 Cluster 3 All Moderate Symptoms n =8 Mean (SD) 4.15 (.91) 2.75 (1.04) W s =103.5 p=.00 Cluster 2 All Mild Sympt oms n=42 Cluster 3 All Moderate Symptoms n =8 Mean (SD) 3.76 (1.06) 2.75 (1.04) W s =120.0 p=.03
70 CHAPTER 5 DISCUSSION Summary of Results This study explored the prevalence, predictors, and consequences of symptom clusters in a secondary analysis of 103 w omen survivors of breast cancer using NIH PROMIS data. An adapted version of the New Symptom Management Model was used as the theoretical framework and orga nization for this exploration. The results indicated that the majority of women survivors of breast cancer experienced symptoms: 67% reported experiencing pain, 62% reported fatigue, and 63% reported mental health symptoms (depression/anxiety/anger). In addition, t he majority of women (n=68, 66%) in this sample expe rienced two or three sympto ms concurrently. The symptom intensity scores indicat ed overall low symptom severity Three symptom clusters were identified that give insight into the symptom experience of women survivors of breast cancer based on the severity of the symptoms in the cl usters : Cluster 1 ( All Minimal Symptoms ) (n=53), Cluster 2 ( All Mild Symptoms ) (n=42), and Cluster 3 ( All Moderate Symptoms ) (n=8). The background characteristic predictors (age, ethnicity, education, relationshi p status, employment status, como rbid condi tions) revealed no significant relationships with the three identified symptom clusters. There was a trend that women in the more symptomatic cluster were younger. Symptom clusters did significantly relate to t he physical and social role functional conse quences or outcomes Women in the more symptomatic clusters had lower physical and social role function. Symptom Experience of Women Survivors of Breast Cancer The symptom experience of women survivors of breast cancer is multifaceted and includes the c onceptual issue of operationally defining a breast cancer survivor, the
71 prevalence of symptoms, the severity or intensity of the symptoms, and the number of self reported symptoms. For this study, women su rvivors of breast cancer were operationally define d as women from the time of diagnosis and continuing throughout their lives, consistent with the current literature review. The women in this sample were part of a clinical sample recruited for a large NIH study Breast cancer diagnosis was self reported by the sample participants as being confirmed by a physician. Approximately two thirds of the women in this sample reported experiencing the three symptoms that were the focus of this study: pain, fatigue, and mental health symptoms (depression/anxiety/an ger). The results of this study are consistent with the existing literature on fatigue (range in literature= 60 80%) and higher than the published prevalence ranges for pain (40 50%) and depression and anxiety (20 30%). Anger has not been extensively stud ied in relation to bre ast cancer, thus there is no information about the prevalence of anger in this population in the literature. The higher prevalence rate for pain may be indicative of using only one question in this study for that score, whereas othe r studies in the literature used multiple question s (intensity, location, severity). The literature did have one comparable study of 85 women with breast cancer in which pain was scored by one measure using a visual analog scale resulting in a higher pain prevalence of 75% (Khan et al., 2012). The high prevalence for the mental health symptom may reflect that the depression, anxiety, and anger symptoms were combined into one question for the current study, so the percentage may be cumulative based on this combination. This finding, however, is relatively consistent with one study that reported 79% prevalence of anxiety in a sample
72 of 154 women with breast cancer (Bender et al., 2005), with an expert panel selecting the items for representation of the sym ptom. In this study, the range of symptom intensity scores indicated overall low symptom severity. For each symptom investigated, the means were below 2, indicating mild symptoms. As such the pain experienced by most study participants did not meet th e threshold of clinically significant pain (e.g., 4 on a 0 10 scale). These findings indicate that the majority of women breast cancer survivors experience mild, persistent symptoms. Surprisingly, most articles in the empirical literature on breast cance r described the symptom or symptoms experienced without addressing the se verity of the symptoms. Thus, it is difficult to put the findings of this study into the extant literature on this topic. Symptom Clusters in Women Survivors of Breast Cancer Until recently, study of the symptom experience of cancer survivors has focused on single symptoms (e.g., pain or fatigue or depression or sleep disturbance ). Within the past decade, attention has shifted to the multidimensional symptom experience, referred to as symptom clusters. Several studies in this area have confirmed the presence of symptom clusters, although the composition of the clusters varies from study to study. In the present study, three symptom clusters were found that were characterized by th e intensity of pain, fatigue, and mental health symptoms experienced. This finding was comparable with one other research study on breast cancer in the literature. That study was a secondary ana lysis of 112 women with breast cancer receiving chemotherapy with or without radiation, that identified four clusters based on pain, fatigue, depression, and sleep disturbance clustered by all low, mild, moderate, or all high symptom severity (Dodd et al., 2010) Using hierarchical cluster
73 analysis the all high s ymptom group experienc ed poorer physical functional status and quality of life. There are several conceptual and methodological issues in the symptom cluster literature, and each of these was addressed in the current study. One issue is the definition o f a symptom cluster, and how many symptoms must be present to constitute a cluster. A symptom cluster in the nursing oncology literature initially was conceptually defined as having three or more coexisting symptoms (Dod d, Miaskowski, et al., 2001). This initial conceptual definition did not reflect the nature, intensity, or temporal aspect of concurrent symptoms. This original definition was challenged several years later (Kim et al., 2005) with the alternate definition of a symptom cluster having two o r more symptoms related to each other and occurring simultaneously. A symptom cluster was operationally defined as having two or more symptoms for this study. This definition is congruent with the prevailing literature on this topic, but warrants further development and clarification as a means to assist in assessment and management of symptoms among cancer survivors (Dodd et al., 2004; Fan et al., 2007; Xiao, 2010). Symptom cluster research has also focused on the number of symptoms in a cluster and wh at specific symptoms interact to constitute a cluster. The number and type of symptoms has varied across studies, and there is no clear consensus on this question. In addition, the characteri stics of symptom clusters varied across type of cancer. Furthe r, few studies have examined the symptom experience based on the intensity or severity of symptoms in the cluster formation. Despite these conceptual questions, the results of th is study are congruent with one other research s tudy that found four clusters of symptoms based on intensity or severity of a group of sympt oms
74 ( Dodd et al., 2010) This latter view is consistent with the theoretical framework of this study that indicates the symptom experience relates to more than just the presence; it also relat es to the intensity, distress, or quality of the experience. Three symptom clusters were identified in this study using three independent variables of pain, fatigue, and mental health (depression/anxiety/anger) symptoms: Cluster 1 ( All Minimal Symptoms ) ( n=53), Cluster 2 ( All Mild Symptoms ) (n=42), and Cluster 3 ( All Moderate Symptoms ) (n=8). This study result is consistent with a research study that looked at symptom severity in their analysi s (Dodd et al ., 2010). This study revealed that the symptom ex perience varied based on the intensity of the symptoms in the clusters. One d istinct difference between this published st udy and the present study is that the study sample involved women survivors of breast cancer undergoing active treatment. In the curr ent study, it was not possible to ascertain where the women were in their treatment trajectory as that data was not collected in the PROMIS data. Proximity to treatment might impact the number and severity of symptoms experienced by women with breast canc er (Dodd et al., 2010) In the study by Dodd and colleagues, the symptoms were milder at the beginning of treatment trajectory, progressing to moderate to severe symptoms at the conclusion of treatment with a return to milder level of symptoms six months later. This is a consistent finding related to active treatment in an oncolo gy sample of patients, over half of whom had moderate to severe levels of two or more symptoms while undergoing treatment ( Miaskowski et al., 2006). It is possible that the resul ts of the current study might be different if it were known where in the treatment trajectory the women were.
75 A second issue is the method of establishing a cluster. In the empirical literature on this topic, two statistical methods have been used, factor analysis and cluster analysis. Factor analysis and cluster analysis have similarities in that b oth analyses seek to identify groups. Factor analysis is a data reduction statistical method to combine variables to a manageable size ( Field, 2009), and i t h as been used in the oncology literature. This approach has been most commonly used in studies that had a large cadre of symptoms to analyze. For example, two oncology research studies used factor analysis with 22 to 26 symptoms (Breen et al., 2009 ; Cleela n d et al., 2000) One combined sample of breast and prostate cancer patient data used factor analysis with 16 symptoms (K im et al., 2009a), while a breast cancer only sample factor analyzed 16 symptoms (Fu et al., 2009). In comparison, cluster analysis i s used to sort groups or cases based on a small set o f variables (Meyers et al., 2013). In the general oncology literature, hierarchical cluster analysis has been used with four symptoms (Miaskowski et al., 2006). In the breast cancer literature, four sy mptoms were also investigated using hierarchical cluster analysis (Dodd et al., 2010). Based on the two statistical options of factor or cluster analysis, and the oncology and breast cancer research literature cluster analysis was selected for this study Hierarchical followed by k means cluster analysis was selected as the most appropriate way to look at relating symptom variables and combining these symptom variables into realistic, applicable numbers and types of clusters to contribute to the science and to find meaningful answers to help women survivors of breast cancer towards productive functioning.
76 Background Characteristics as Predictors of Symptom Clusters in Women Survivors of Breast Cancer In this study, t he b ackground characteristics (age, e thnicity, education, relationshi p status, employment status, como rbid conditions) were not significantly associated with the symptom clusters. That is, these background variables were not significantly influ encing whether women had all minimal, all mild, or all moderate symptoms. The fact that background characteristics did not influence the identified symptom clusters is consistent with the majority of the breast cancer research literature. Other studies found Hispanic women more likely to report multip le symptoms compared to Black or White women (Fu et al., 2009), while Mosher and Duhamel (2010) found ethnicity not associated with distress. Education level associated with symptoms was equivocal in the literature (Akin Odanye et al., 2011; Bardwell et a l., 2008; Kenefick, 2006), as was relationship status (Akin Odanye et al., 2011; Brunault et al., 2012; Kenefick, 2006). Employment status was not related to symptomotology in some studies and related in others (Akin Odanye et al., 2011; Fu et al., 2009; Villaverde et al., 2008). Thus, the role of these background variables remains equivocal. No resea rch articles were found that examined the relations hip of comorbid conditions to symptoms Descriptive statistics on the number of comorbid conditions was me ntioned but no studies examined this variable in relation to the symptom experience Perhaps the prior research on this topic did not address comorbidity because the focus was primarily on identifying clusters and their outcomes since the sample particip ants were undergoing active cancer treatment ( Dodd et al., 2010; Miaskowski et al., 2006).
77 It would be easy however, to attribute differences in cluster group membership to the health status of the women For instance, one could argue that women with mor e symptoms were simply more ill. The results of this study, however, revealed no significant differences across symptom clusters in the number of diagnosed comorbid conditions experienced by the women in this sample. This finding reflects the fact that o verall, the women in this sample were relatively healthy and high functioning. It may also indicate that the 25 diagnoses examined in this study (e.g. high blood pressure asthma) may have clinical relevance to overall health, but may not be associated wi th physical symptoms of pain or fatigue, or mental health symptoms. The results of this study did not show a significant association between age and symptom clusters. However, t ( All Moderate Symptoms ) we re 10 years younger than the women in Cluster 1 ( All Minimal Symptoms ) Wh en comparing younger (< 50 years ) to older ( 50 years) women there was a trend for younger women to be in more highly symptomatic clusters ( 2 =6.6, p=.0 6 ). Several research studies revealed that younger women tended to have more severe symptoms for pain, depression, anxiety, and distress Younger women ( 40) in a sample of 174 women post mastectomy were found to have a significantly increas ed risk of pain (Fabro et al., 2012). Thre e studies reported younger women had greater depression and anxiety than older women (Howard et al. 2012 ; Khan e t al., 2012; Mosher & Danoff Burg, 2005). Distress was noted to be lower in older women (Knopf, 2007; Knopf, 2011 ; Matthews et al. 2012). Fa tigue was h igher in older women ( 50 years) survivors of breast cancer in 80% of the study participants (N=373) undergoing hormonal treatment (Glaus et al., 2006). The mean age of the women in this current
78 study was 60.4 years (SD 10.1) which approximates the median age for a breast cancer diagnosis which is 61 years (ACS, 2011). This is a somewhat encouraging finding for older women who are breast cancer survivors in that their symptom experience may be reduced relative to younger women. This may also reflect that younger women often have more invasive forms of breast cancer. Outcomes or Consequences of Symptoms Clusters in Women Survivors of Breast Cancer One of the fundamental question s addressed in this study was whether the experience of living wi th multiple symptoms mattered in terms of everyday functioning. The results of this st udy confirm these relationships: symptom clusters were significantly associated with lower physical functioning and lower satisfaction with social role functioning. The majority of women s urvivors of breast cancer in this study (n=100 97%) symptom clusters identified as more intense reported significantly lower physical functioning. This result is consistent with the breast cancer research associating high symptom experiences with poorer physical functional status (Dodd et al., 2010) and greater physical functional impairment (Mosher & Duhamel, 2010). In addition, t he majority of wome n (n=92, 89%) were satisfied with their social role activit ies and relationships and rated their However, there were significant associations between the symptom clusters and satisfaction with t heir social role functioning. Post hoc analysis revea led that Cluster 3 (All Moderate Symptoms) was significantly different from the other two clusters and women in this group ha d lower satisfaction with social role
79 functioning. This is a new finding to add to the science of cluster analysis. Social role function was not examined in any of the empirical breast cancer literature related to symptom clusters and outcomes. Outcomes included in the literature typically included quality of life, well being, o ptimism, act ivity or physical functioning, and illness perception, but no reference specifically to social role function. Not surprisingly, more frequent and intense symptoms are associated with lower quality of life (Dodd et al., 2010) and lower well bei ng (Khan et al., 2012). Thus, these findings on physical functioning and social role functioning are consistent with the broader study of quality of life and well being outcomes, and highlight the fact that these constructs are multidimensional, incorpora ting both physical and social functioning. Taken together, however, the results of this study highlight several important findings. First, symptom clusters are discernible and they differ according to the intensity of the symptom experience. Second, sym ptom clusters have an impact on the daily lives of women breast cancer survivors. Even in this relatively small sample with a restricted range of symptom intensity (mean intensity scores for pain, fatigue, and mental health concerns were all in the mi l d r ange), women in clusters identified as representing elevated symptoms reported worse functioning. Thus, the experience of having three symptoms as opposed to just one, even when they are mild in intensity, has an impact on daily life. This has implicatio ns for s urvivorship that healthcare practitioners need to be aware that even mild, concurrent pain, fatigue, and mental health symptoms may require treatment for optimal functioning. Limitations The re are several limitations of this study that must be ack nowledged. First, the study was a secondary analysis of collected data. Thus, there were limitations on the
80 number and type of data that was available for analysis. For instance, there was no data on sleep disturbances, which is often considered an impo rtant symptom among cancer survivors. In addition, the measurement of the study variables was predetermined and could not be influenced by the author of this study. Further, some variables were not included in the dataset, such as stage of breast cancer or time since diagnosis or treatment. Second, the study design was cross section al Thus, it is not possible to establish causality. Third, the sample size was fairly small and may not be representative of the population of women survivors of breast can cer. Even thoug h the initial data set had 21,133 participants with representation similar to the U. S. census characteristics, the 103 women survivors of breast cancer in thi s study were fairly homogenous in terms of their personal characteristics. They w ere relatively healthy predominantly White, highly educate d partnered and e mployed. A ny generalization to other women survivors of breast cancer must be made cautiously. Finally, the women in this sample had only mild symptoms that did not have much v ariability, which limited cluster discriminations. Thus, conclusions must be tempered and study results should be replicated in a larger sample with more symptom variability. Implications for Clinical Practice The implications for clinical practice of th is study revolve around healthcare practitioners being aware of current research concerning symptom clusters in women survivors of breast cancer. With the knowledg e that the majority of women survivors of breast cancer present with multiple symptoms and n ot single symptoms, practitioners may assess and treat the multiple symptoms concurrently to positively impact physical and social role functioning Even with the low intensity symptoms in this study, physical and social role functioning outcom es were sig nificantly affected Decreasing symptoms
81 in women survivo rs of breast cancer can lead to positively impact ing physical and social function as women deal with breast cancer survivorship. Implications for Future Research Future research could yield pros pective data through the use of the PROMIS surveys available from NIH. A larger sample could be selected to obtain a more diverse sample for comparison of background characteristics and the effect on symptom clusters and the effect of the symptom clusters on physical and social role outcomes Separate questionnaires could be used for each of the variables with the addition of the sleep disturbance variable. Data regarding stage of disease and place in the treatment or post treatment survival trajectory c ould be collected. A long itudinal study examining change over time could be completed. Data could be collected from women survivors of breast cancer following concurrent treatment for their symptoms related to the physical and social role functional outc omes. Conclusion This study gives insight into the symptom experience of women survivors of breast cancer and the functional out comes of the symptom clusters for these women. This study adds to the growing body of literature on the importance of sympto m cluster s by highlighting that even low intensity symptom clusters significantly influence physical and social role functional outcomes. Thus, healthcare practitioners should consistently assess and treat even mild symptoms to facilitate optimal health. Cancer survivors are growing in numbers and surviving longer. One in eight women in the United States develops breast cancer in their lifetime. Symptoms do cluster in these women and affect their physical and social role functional outcomes. It
82 is impe rative that research continue in this area to enhance the survivorship years for women survivors of breast cancer to productively continue their daily lives.
83 LIST OF REFERENCES Akin Odanye, E. O., Chioma, C. A., & Abiodun, O. P. (2011). Measured effect o f some socio demographic factors on depression among breast cancer patients receiving chemotherapy in Lagos State University Teaching Hospital (LASUTH). African Health Sciences 11(3), 341 345. Retrieved from http://www .ncbi.nln. nih .gov.lp.hscl.ufl.edu/pubmed/22275922 Albert, S. M., & Freedman, V. A. (2010). Public health and aging New York, New York: Springer Publishing Company. American Cancer Society. (2011). Breast cancer facts & figures 2011 2012. Atlanta, Georgia: American Cancer Society, Inc. Retrieved from: http://www.cancer.org/Research/CancerFactsFigures/BreastCancerFactsFigures/ breast cancer facts and figures 2011 2012 American Cancer Society. (2012a). Cancer facts & figures 2012. Atlanta, Georgia: American Cancer Soc iety, Inc. Retrieved from http://www.cancer.org/acs/groups/content/ @epidemiologysurveilance/documents /docum ent/acspc031941.pdf American Cancer Society. (2012b). Cancer treatment & survivorship facts & figures 2012 2013. Atlanta, Georgia: American Cancer Society, Inc. Retrieved from: http://www.cancer.org/acs/groups/content/@epidemiologysurveilance/documents /document/acspc 033876.pdf Andrykowski, M. A. (2009). Mag nitude and correlates of response shift in fatigue ratings in women undergoing adjuvant therapy for breast cancer. Journal of Pain Symptom Management, 37, 341 351. doi:10.1016/j.jpainsymman.2008.03.01 Andrykowski, M. A., Donovan, K. A., Laronga, C., & Jaco bsen, P. B. (2010). Prevalence, predictors, and characteristics of off treatment fatigue in breast cancer survivors. Cancer 116, 5740 5748. doi:10.1002/cncr.25294 Bardwell, W. A., Profant, J., Casden, D. R., Dimsdale, J. E., Ancoli Israel, S., Natarajan, relative importance of specific risk factors for insomnia in women treated for early stage breast cancer. Psycho Oncology 17, 9 18. doi:10.1002/pon.1192 Barsevick, A. M. (2007a). The con cept of symptom cluster. Seminars in Oncology Nursing 23, 89 98. doi:10.1016/j.soncn,2007.01.009 Barsevick, A. M. (2007b). The elusive concept of the symptom cluster. Oncology Nursing Forum 34, 971 980. doi:10.1188/07.ONF.971 980 Bender, C. M., Ergyn, F. S., Rosenzweig, M. Q., Cohen, S. M., & Sereika, S. M. (2005). Symptom clusters in breast cancer across 3 phases of the disease. Cancer Nursing 28, 219 225.
84 Berger, A. M., Visovsky, C., Hertzog, M., Holtz, S., & Loberiza, F. R., Jr. (2012). Usual and wors t symptom severity and interference with function in breast cancer survivors. The Journal of Supportive Oncology Advance online publication. doi:10.1016/j.suponc.2011.11.001 Bond, S. M. (2010a). Crossing the threshold of cancer and aging: Integrating ger ontological and oncological nursing knowledge and skills. Journal of Gerontological Nursing 36, 3 4. doi:10:3928/00989134 20100108 04 Bond, S. M. (2010b). Physiological aging in older adults with cancer: Implications for treatment decision making and toxi city management. Journal of Gerontological Nursing 36, 26 37. doi:10.3928/00989134 20091103 98 Bower, J. E., Ganz, P. A., Irwin, M. R., Kwan, L., Breen, E. C., & Cole, S. W. (2011). Inflammation and behavioral symptoms after breast cancer treatment: Do fa tigue, depression, and sleep disturbance share a common underlying mechanism? Journal of Clinical Oncology 29, 3517 3522. doi:10.1200/JCO.2011.36.1154 Brant, J. M., Beck, S., & Miaskowski, C. (2010). Building dynamic models and theories to advance the sci ence of symptom management research. Journal of Advanced Nursing 66, 228 240. doi:10.1111/j.1365 2648.2009.05179.x Breen, S. J., Baravelli, C. M., Schofield, P. E., Jefford, M., Yates, P. M., & Aranda, S. K. (2009). Is symptom burden a predictor of anxiet y and depression in patients with cancer about to commence chemotherapy [Supplemental material]. The Medical Journal of Australia 190, 99 104. Retrieved from http://www.mja.com.au/ Brunault, P., Suzanne, I., Trzepidur Edom, M., Garaud, P., Calais, G., Toledano, A., & Camus, V. (2012). Depression is associated with some patient perceived cosmetic changes, but not with radiotherapy induced late toxicity, in long term breast cancer survivors. Psycho Oncology Advance onlin e publication. doi:10.1002/pon.3038 Initial Adult health item banks and first wave testing of the patient reported outcomes measurement information system (PROMIS) netw ork: 2005 2008. Journal of Clinical Epidemiology 63, 1179 1194. doi:10.1016/ j.jclinepi. 2010.04. 011 (2007). The patient reported outcomes measurement information system (PR OMIS): Progress of an NIH roadmap coopera tive group during its first two years Medical Care 45, S3 S11. doi:10:1097/01.mlr.0000258615.42478.55. Cheung, W. Y., Le, L. W., Gagliese, L., & Zimmermann, C. (2011). Age and gender differences in symptom intensi ty and symptom clusters among patients with metastatic cancer. Supportive Care in Cancer 19, 417 423. doi:10.1007/s00520 010 0865 2
85 Cheung, W.Y., Le, L. W., & Zimmermann, C. (2009). Symptom clusters in patients with advanced cancers. Supportive Care in C ancer 17, 1223 1230. doi:10.1007 /s00520 009 0577 7 Cleeland, C. S., Mendoza, T. R., Wang, X. S., Chou, C., Harle, M. T., Morrissey, M., &Engstrom, M. C. (2000). Assessing symptom dist ress in cancer patients: The M. D. Anderson Symptom Inventory. Cancer 89, 1634 1646. Retrieved from ttp://www.canceronlinejournal.com/view/0/index.html Delgado Guay, M., Yennurajalingam, S., Parson, H., Palmer, J. L., Bruera, E. (2011). Association between self reported sleep disturbance and other symptoms in patients w ith advanced cancer. Journal of Pain and Symptom Management 41, 819 827. doi:10.1016/j.jpainsymman.2010.07.015 Dodd, M. J., Cho, M. H., Cooper, B. A., & Miaskowski, C. (2010). The effect of symptom clusters on functional status and quality of life in wome n with breast cancer. European Journal of Oncology Nursing 14, 101 110. doi:10.1016/ j.ejon.2009.09 .005 Dodd, M. J., Miaskowski, C., & Lee, K. A. (2004). Occurrence of symptom clusters. Journal of the National Cancer Institute Monographs 32, 76 78. doi: 10.1093 /jncimonographs/Igh008 Dodd, M. J., Miaskowski, C., & Paul, S. M. (2001). Symptom clusters and their effect on the functional status of patients with cancer. Oncology Nursing Forum 28, 465 470. doi:10.1016/j.soncn,2007.01.009 Donovan, H. S., Harte nback, E. M., & Method, M. W. (2005). Patient provider communication and perceived control for women experiencing multiple symptoms associated with ovarian cancer. Gynecologic Oncology 99, 404 411. doi:10.1016/j.ygyno.2005.06.062 Donovan, K. A., & Jacobse n, P. B. (2007). Fatigue, depression, and insomnia: Evidence for a symptom cluster in cancer. Seminars in Oncology Nursing 23, 127 135. doi:10.1016/j.soncn.2007.01.004 Donovan, K. A., & Jacobsen, P. B. (2010). The Fatigue Symptom Inventory: A systematic r eview of its psychometric properties. Supportive Care in Cancer Advance online publication. doi:10.1007/s00520 010 0989 4 Donovan, K. A., & Jacobsen, P. B., Small, B. J., Munster, P. N., & Andrykowski, M. A. (2008). Identifying clinical meaningful fatigue with the Fatigue Symptom Inventory. Journal of Pain and Symptom Management 36, 480 487. doi:10.1016 /j.jpainsymman.2007.11.013
86 Donovan, K. A., Taliaferro, L. A., Brock, C. W., & Bazargan, S. (2008). Sex differences in the adequacy of pain management amo ng patients referred to a multidisciplinary cancer pain clinic. Journal of Pain and Symptom Management 36, 167 172. Doi:10.1016/j.jpainsymman.2007.09.013 Evangelista, S. L., & Santos, E. M. (2012). Cluster of symptoms in women with breast cancer treated w ith curative intent. Support Care Cancer Advance online publication. doi:10.1007/s00520 011 1238 1 Fabro, E. A. N., Bergmann, A., e Silva, B. do A., Ribeiro, A. C., P., Abrahao, K. de S., mastectomy pain syndrome: Incidence and risks. The Breast Advance online publication. doi:10.1016/j.breast.2012.01.019 Fan, G., Filipczak, L., & Chow, E. (2007). Symptom clusters in cancer patients: A review of the literature. Current Oncology 14, 173 179. Retriev ed from http://www .current oncology.com/index.php/oncology Field, A. (2009). Discovering statistics using SPSS Thousand Oaks, California: SAGE Publications, Inc. Fiorentino, L., Rissling, M., Liu, L., & Ancoli Israel, S. (2011). The symptom cluster of sleep, fatigue and depressive symptoms in breast cancer patients: Severity of the problem and treatment options. Drug Discovery Today: Disease Models, 8, 167 173. doi:10.1016/j.ddmod.2011.05.001 Gartner, R., Jensen, M. B. Nielsen, J., Ewertz, M., Kroman, N., & Kehlet, H. (2009). Prevalence of and factors associated with persistent pain following breast cancer surgery. The Journal of the American Medical Association 18, 1985 1992. Retrieved from http://jama.jamanetwork.co m/article.aspx?articleid=184861 Gaston Johansson, F., Fall Dickson, J. M., Bakos, A. B., & Kennedy, M. J. (1999). Fatigue, pain, and depression in pre autotransplant breast cancer patients. Cancer Practice 7, 240 247. Retrieved from http:// onlinelibrary.wiley.com /journal/10.1111/(ISSN)1523 5394/issues Gift, A. G. (2007). Symptom clusters related to specific cancers. Seminars in Oncology Nursing 23, 136 141. doi:10.1016/j.soncn. 2007.01.003 Glaus, A., Boehme, C., Thurlimann, B., Ruhstaller, T., Hsu Schmitz, S. F., Morant, breast cancer undergoing hormonal cancer treatment. Annals of Oncology 17, 801 806. doi:1 0.1093/annonc/md1030 Grunfeld, E. (2009). Optimizing follow up after breast cancer treatment. Current Opinion in Obstetrics and Gynecology, 21, 92 06. doi:10.1097/GCO.0b013e328321e437
87 Gwede, C. K., Small, B. J., Munster, P. N., Andrykowski, M. A., & Jacobs en, P. B. (2008). Exploring the differential experience of breast cancer treatment related symptoms: A cluster analytic approach. Supportive Care in Cancer 16, 925 933. doi:10.1007/s00520 007 0364 2 Hadi, S., Fan, G., Hird, A. E., Kirou Mauro, A., Filipcz ak, L. A., & Chow, E. (2008). Symptom clusters in patients with cancer with metastatic bone pain. Journal of Palliative Medicine 11, 591 600. doi:10.1089/jpm.2007.0145 Hair, J. F., Jr., & Black, W. C. (2000). Cluster Analysis. In L. G. Grimm & P. R. Yarol d (Eds.), Reading and understanding more multivariate statistics (pp. 147 205). Washington, DC: American Psychological Association. Hird, A., Wong, J., Zhang, L., Tsao, M., Barnes, E., Danjoux, C., & Chow, E. (2010). Exploration of symptoms clusters within cancer patients with brain metastases using the Spitzer Quality of Life Index. Supportive Care in Cancer 18, 335 342. doi:10.1007/s00520 009 0657 8 Hoffman, A. J., Given, B. A., von Eye, A., Gift, A. G., & Given, C. W. (2007). Relationships among pain, f atigue, insomnia, and gender in persons with lung cancer. Oncology Nursing Forum 34, 785 792. doi:10.1188/07.ONF.785 792 Horner, M.J., Ries, L.A.G., Drapcho, M., Neyman, N., Aminou, R., Howlader, N., Edwards, B.K. (Eds.), SEER Cancer Statistics Review, 1975 2006. National Cancer Institute, 1975 2006 http://seer.cancer.gov/csr/1975_2006/ based on November 2008 SEER data submission, posted to the SEER website, 2009. Howard Anderson, J., Ganz, P. A., Bo wer, J. E., & Stanton, A. L. (2012). Quality of life, fertility concerns, and behavioral health outcomes in younger breast cancer survivors: A systematic review. Journal of the National Cancer Institute Advance online publication. doi:10.1093/jnci/djr541 Cronin KA (eds) (2012). SEER Cancer Statistics Review, 1975 2009 (Vintage 2009 Populations) National Cancer Institute Retrieved from http://seer.cancer .gov /csr/1975_2009_pops09/ Hulley, S. B., Cummings, S. R., Browner, W. S., Grady, D. G., & Newman, T. B. (2007). Designing clinical research. Philadelphia, PA: Lippincott Williams & Wilkins. Jimenez, A., Madero, R., Alonso, A., Mar tinez Feliu, J. (2011). Symptom clusters in advanced cancer. Journal of Pain and Symptom Management 42, 24 31. doi:10.1016/j.jpainsymman.2010.10.266 Kenefick, A. L. (2006). Patterns of symptom distress in older wome n after surgical treatment for breast cancer. Oncology Nursing Forum 33, 327 335. doi:10.1188/06.ONF.327 335
88 Khan, F., Amatya, B., Pallant, J., & Rajapaksa, I. (2012). Factors associated with long term functional outcomes and psychological sequelae in wom en after breast cancer The Breast Advance online publication. doi:10.1016 /j.breast.2012.01.013 Kim, E., Dodd, M. J., Aouizerat, B. E., Jahan, T., & Miaskowski, C. (2009). A review of the prevalence and impact of multiple symptoms in oncology patients. Jo urnal of Pain and Symptom Management 37, 715 736. doi:10.1016/j.jpainsymman.2008 .04.018 Kim, E., Jahan, T., Aouizerat, B. E., Dodd, M. J., Cooper, B. A., Paul, S. patients undergoing radiation the rapy. Supportive Care in Cancer 17, 1383 1391. doi:10.1007/s00520 009 0595 5 Kim, E., Jahan, T., Aouizerat, B. E., Dodd, M. J., Cooper, B. A., Paul, S. occurrence rates versus sy mptom severity ratings in patients at the end of radiation therapy. Cancer Nursing 32, 429 436. doi:10.1097/NCC .0b013e3181b046ad Kim, H. J., Barsevick, A. M., & Tulman, L. (2009). Predictors of the intensity of symptoms in a cluster in patients with brea st cancer. Journal of Nursing Scholarship 41, 158 165. doi:10.1111/j.1547 5069.2009.01267.x Kim, H. J., Barsevick, A. M., Tulman, L., & McDermott, P. A. (2008). Treatment related symptom clusters in breast cancer: A secondary analysis. Journal of Pain and Symptom Management 36, 468 479. doi:10.1016/j.jpainsymman.2007.11.011 Kim, H. J., McGuire, D. B., Tulman, L., & Barsevick, A. M. (2005). Symptom clusters: Concept analysis and clinical implications for cancer nursing. Cancer Nursing 28, 270 282. doi:10. 1016/j.soncn.2007.01.004 Lenz, E. R., Pugh, L. C., Milligan, R. A., Gift, A., & Suppe, F. (1997). The middle range theory of unpleasant symptoms: An update. Advances in Nursing Science 19, 14 27. Retrieved from http://journals.lww.com /advancesinnursingscience/pages /default.aspx Matthews, E. E., Schmiege, S. J., Cook, P. F., & Sousa, K. H. (2012). Breast cancer and symptom clusters during radiotherapy. Cancer Nursing 3 5, E1 E11. doi:10.1097/NCC.0b013e3182277222 (2012). Illness perception clusters at diagnosis predict psychological distress among women with breast cancer at 6 months p ost diagnosis. Psycho Oncology Advance online publication. doi:10.1002/pon.3054
89 Meleis, A. I. (2007). Theoretical Nursing: Development and progress Philadelphia, PA: Lippincott Williams & Wilkins. Meyers, L. S., Gamst, G., & Guarino, A. J. (2013). App lied multivariate research Thousand Oaks, California: SAGE Publications, Inc. Miaskowski, C., Aouizerat, B. E., Dodd, M., & Cooper, B. (2007). Conceptual issues in symptom clusters research and their implications for quality of life assessment in patients with cancer. Journal of the National Cancer Institute Monographs 37, 39 46. doi:10.1093/jncimonographs/lgm003 A. (2006). Subgroups of patients with cancer with differen t symptom experiences and quality of life outcomes: A cluster analysis. Oncology Nursing Forum 33, E79 E89. doi:10.1188/06.ONF.E79 E89 Miaskowski, C., Dodd, M., & Lee, K. (2004). Symptom clusters: The new frontier in symptom management research. Journal o f the National Cancer Institute Monographs 32, 17 21. doi:10.1093/jncimonographs/lgh023 Mosher, C. E., & Danoff Burg, S. (2005). A review of age differences in psychological adjustment to breast cancer. Journal of Psychosocial Oncology 23, 101 114. doi:1 0.1300/J077v23n02_07 Mosher, C. E., & DuHamel, K. N. (2010). An examination of distress, sleep, and fatigue in metastatic breast cancer patients. Psycho Oncology 21, 100 107. doi:10.1002 /pon.1873 Palesh, O. G., Roscoe, J. A., Mustian, K. M., Roth, T., Sa vard, J., Ancoli Morrow, G. R. (2010). Prevalence, demographics, and psychological associations of sleep disruption in patients with cancer: University of Rochester Cancer Center Community Clinical Oncology Program. Journal of Clinical Onco logy 28, 292 298. doi:10.1200/JCO.2009.22.5011 Perry, S., Kowalski, T.L., & Chang, C H. (2007). Quality of life assessment in women with breast cancer: Benefits, acceptability and utilization. Health and Quality of L ife Outcomes, 5, 1 14. doi:10.1186/1477 7525 5 24 Portne y L. G., & Watkins, M. P. (2009). Foundations of clinical research: Applications to practice Upper Saddle River, New Jersey: Pearson Education, Inc. PROMIS information retrieved from http://www.nihp romis.org/ Quadagno, J. (2008). Aging and the life course New York, New York: McGraw Hill. Ridner, S. H. (2005). Quality of life and a symptom cluster associated with breast cancer treatment related lymphedema. Supportive Care in Cancer 13, 904 911. do i:10.1007/s00520 005 0810 y
90 Savard, J., Ivers, H., Villa, J., Caplette Gingras, A., & Morin, C. M. (2011). Natural course of insomnia comorbid with cancer: An 18 month longitudinal study. Journal of Clinical Oncology 29, 3580 3586. doi:10.1200/JCO.2010.33 .2247 Schnur, J. B., Hallquist, M. N., Bovbjerg, D. H., Silverstein, J. H., Stojceska, A., & Montgomery, G. H. (2007). Predictors of expectancies for post surgical pain and fatigue in breast cancer surgical patients. Personality and Individual Differences 42, 419 429. doi:10.1016/j.paid.2006.07.009 So, W. K., Marsh, G., Ling, W. M., Leung, F Y., Lo, J. C., Yeung, M., & Li, G. K. (2009). The symptom cluster of fatigue, pain, anxiety, and depression and the effect on the quality of life of women receiving tr eatment for breast cancer: A multicenter study. Oncology Nursing Forum 36, E205 E214. doi:10.1188/09.ONF.E205 E214 Starkweather, A. R., Lyon, D. E., & Schubert, C. M. (2011). Pain and inflammation in women with early stage breast cancer prior to induction of chemotherapy. Biological Research for Nursing Advance online publication. doi:10.1177 /1099800411425857 Taylor, T. R., Huntley, E. D., Makambi, K., Sween, J., Adams Campbell, L. L., Frederick, W., & Mellman, T. A. (2011). Understanding sleep disturbanc es in African American breast cancer survivors: A pilot study. Psycho Oncology Advance online publication. doi:10.1002/pon.2000 Thornton, L. M., Andersen, B. L., & Blakely, W. P. (2010). The pain, depression, and fatigue symptom cluster in advanced breast cancer: Covariation with the hypothalamic pituitary adrenal axis and the sympathetic nervous system [Brief report]. Health Psychology 29, 333 337. doi:10.1037/a0018836 Miask owski, C. (2012). Identification of distinct subgroups of breast cancer patients based on self reported changes in sleep disturbance. Support Care Cancer Advance online publication. doi:10.1007/s00520 012 1381 3 Villaverde, R. C., Batlle, J. F., Yllan, A. V., Gordo, A. M. J., Sanchez, A. R., Valiente, B. S. J., & Baron, M. G. (2008). Employment in a c ohort of breast cancer patients. Occupational Medicine, 58, 509 511. doi:10.1093/occmed/kqn092 Wilmoth, M. C., Coleman, E. A., Smith, S. C., & Davis, C. (2004 ). Fatigue, weight gain, and altered sexuality in patients with breast cancer: Exploration of a symptom cluster. Oncology Nursing Forum 16, 1069 1075. doi:10.1188/04.ONF.1069 1075 Xiao, C. (2010). The state of science in the study of cancer symptom cluste rs. European Journal of Oncology Nursing 30, 1 18. doi:10.1016/j.ejon.2010.05.011
91 Yeom, H., & Heidrich, S. M. (2009). Effect of perceived barriers to symptom management on quality of life in older breast cancer survivors. Cancer Nursing 32, 309 316. doi:10.1097/NCC.0b013e31819e239e Quality of life and physical symptom clusters in long term cancer survivors across cancer types. Journal of Pain and Symptom Managemen t 43, 720 31. doi:10.1016/j.jpainsymman. 2011.04.023
92 BIOGRAPHICAL SKETCH Lois Meta Ritz Ellis is employed as the director of a nursing program educating nursing assistants, practical nurses, associate nurses and registered nurses completing the ir bachelor in nursing. She graduated from University of Florida with her Bachelor of Science in Nursing in 1972 and with her Master of Science in care prior to medical surgi cal nurs nursing program. In 1996, she became the coordinator of a nursing simulated lab at the same community college and remained in that position until becoming the director of all of the nursing programs in 20 07. While ea rning her Doctorate of Philosophy in Nursing Sciences she focused on researching women survivors of breas t cancer, with a minor in aging.