The Body Image After Breast Cancer Questionnaire The Design and Testing of a Disease Specific Measure

Nancy Baxter, MD

A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy Graduate Department of Medical Sciences University of Toronto

Copyright by Nancy Baxter 1998

IJFl

,cm,

National Library

Bibliothèque nationaIe du Canada

Acquisitions and Bibliographie Services

Acquisitions et seMces bibliographiques

395 Wellington Street Ottawa ON K IA ON4

395, rue Wellington OttawaON K1AON4

Canada

Canada

The author has granted a nonexclusive licence allowing the National Library of Canada to reproduce, loan, distribute or sell copies of this thesis in microform, paper or electronic formats.

L'auteur a accordé une Licence non exclusive permettant à la BibIiothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/fIm, de reproduction sur papier ou sur fomat électronique.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantid extracts fiom it may be printed or othewise reproduced without the author's permission.

L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.

The Body [mage a€ter Breast Cancer Questionuaire. The design and testiog of a disease specific measure. Nancy Noel k t e r , Doctor of Philosophy, Graduate Department of Medical Sciences, University of Toronto, 1998.

Abstract Research has been impeded by the lack of a self-report measure of body image suitable for use in the breast cancer population. The objective of this thesis was to create such a measure, the Body Image after Breast Cancer (BiBC) questionnaire.

575 items were generated fkom a review of preexisting measures, and interviews with 30 experts, 51 wornen wiîh breast cancer and 4 spouses. The number of items was reduced to 216 by a judgemental review for acceptability by 26 women, and elhination of obviously redundant items. The 216 remaining items were formatteci and administered to 360 women with breast cancer. Analysis of the 309 responses enabled M e r elimination of items. A sample of 164 women cornpleted the rernaining 111 items. Further item reduction to 53 items was achieved by factor amlysis and statistical analysis. Six factors were identified (and thus a conceptualization of the aspects of body image salient to women with breast cancer proposed);Vulnerability; Body Stigma, Limitations; Body Concerns; Transparency;and Arm Concerns.

Six scales were formed consisting of the items belonging to the 6 factors, Reliability was tested by administration of the questionnaire twice. Good reliability @oth interna1 consistency and test-retest reliability) was found for the 6 scales (ranging from 0.75 to 0.86). The BIBC was reviewed by 20 women and found to be acceptable in terms of length, comprehensibility and suitability. An examination of the validity of the BIBC was undertaken. Scores on the BXBC were found to correlate with similar measures but not with a masure of social desirability. S a l e s of the BIBC were found to discriminate between women treated for breast cancer with lumpectomy and those treated with mastectomy, and between women with breast cancer and a control group, supporting the validity of the BIBC as a descriptive tool. A brief self-report measure of body image (the BIBC)has been developed, suitable for use in the breast

cancer population. The BIBC has been found to be acceptable, reliable, and valid.

Acknow ledgements The generosity of many individuals was required to complete this thesis. I am grateful to all and would iike to take this oppominity to thank some by name: To Dr Trudeau, Dr Ambus,

Dr Koven, Dr Sidlofsky, Dr Bates, and Dr McGowan, for aUowing me to recruit patients at their busy clinics; To the women who participate. in my study, for their enthusiasm and honesty; To Claire Bombardier, for her time and guidance; To Gerry Devins, for his ability to push me to higher achievement; To Pam Goodwin, for her constant support and fnendship; To the Department of Surgery and Division of General Surgery, for the opportunity to participate in the Surgical Scientist Program and the support 1 received in my excursion down a road

lesser travelled; To Brenda and Paul, for their help in the trenches; To MJ, Pete, and Terry Baxter, for aiways providing encouragement without tw many questions; To Alice, Beverly, Carole, and Sophie, for always being ready with a fnendly ear; To 'The Dude', for Ioving me without conditions or limits; and, to Robin McLeod, for being both my mentor and my number one fan.

This work was supported by p t s from the Cancer Research Society Incorporated and the

Medical Research Council of Canada.

iii

TABLE of CONTENTS

Chapter One Introduction Chapter Two Background Literature Review 2.1 General Introduction 2.2 Body Image Aspects of Salience to the Breast C a n e r Population 2.3 Measurement Methodology 2.4 Measurement of Body image 2.5 Body image Studies in the Breast Cancer Population 2.6 Goals 2.7 Potential ReIevance

Chapter Three Item Generation 3.1 Theory 3 -2 Methodology 3.3 Results 3.4 Summary

Chapter Four Item Reduction 4.1 Introduction 4.2 Item Reduction 1: Theory 4.3 Item Reduction 1: Methodology

4.4 Item Reduction 1: R d t s 4.5 EIimination of Redundant Items 4.6 Item Reduction 2: Theory 4.7 Item Reduction 2: Methodology Fomt Sample and Recniitment StatisticaI Methods 4.8 Item Reduction 2: Results 4.9 Siimmary

Chapter Five Descriptive Statistics for Sample Two 5.1 Introduction 5.2 Methodology Format Sample Recniitment 5.3 Results 5.4 Summary

TABLE of CONTENTS Chapter Six Factor Analysis 6.1 Introduction 6.2 Theory Condensation Rotation 6.3 Methodology Sample Technique 6.4 Results 6.5 Discussion

Chapter Seven Reliability 7.1 Tbeory 7.2 Methodology 7.3 Results 7.4 Summary

Chapter Eight Sensibility 8.1 Theory

8.2 Methodology 8.3 Results Description of the Sampte Purpose Population aad Setting Face Validity Content Validity Feasibility 8.4 Discussion

Chapter Nine Construct Validation 9.1 Introduction 9.2 Methods Selection of Sample of Women with Breast Cancer Selection of Sample of Control Groups Convergent and Divergent Validity Measures Adcninistered Hypotheses Discriminate Validity Cornparison of subjects with 1 breast to subjects with 2 breasts Cornparison of Cancer Group to Control Group

123 124 124

125 130 132 134

TABLE of CONTENTS Chapter N i e Continued 9.3 R d t s

Convergent anci Divergent Validity Dernographics of SarnpIe Hypothesized CorreIations Caiculated Correlations Discriminate VaIidity Demographics and Scores of the 1 Breast and 2 Breast Groups Demographics and Scores of the Contrd Subjects Hypothesized Differences Calculateci Differences Specific Items Power 9.4 Discussion

Chapter Ten

Discussion 10.1 Summary of Measurement Development and Testing 10.2 Development of a Theoretid Framework

10.3 Potential Uses of the Instrument 10-4 Future Directions 10-5 Contribution to the Measurement of Body Image 10.6 Limitations

10.6 ConcIusions

References Appendices

Table 2.1

List of Tables Body Images in Hypothetical Femde Patient with Recently DiagnosePrimary Breast Cancer treated with Mastectomy and Chernotherapy

Table 2.2

Studies of the impact of mastectomy

Table 2.3

Studies cornparhg the impact of lumpectomy to rnastectomy

Table 2.4

Studies of the impact of breast reconstruction after mastectomy

Table 6.1

Eigenvalues of Correlation Matrix for 92 Common Items

Table 6.2

Unrotated Factor Structure for 7 Factor Solution

Table 6.3

Rotated 7 Factor Solution for 92 Common Items

Table 6.4

Comparison of Factor Solutions

Table 6.5

Unrotated Factor Stucture for 6 Factor Solution for 55 Common Items

Table 6.6

Rotated Factor Structure for 6 Factor Solution for 55 Common Items

Table 6.7

Correlation Matrix for 55 Items

Table 7.1

Items with Item-Total Correlation less than 0.40

Table 7.2

Items Remaining for Reliaiblity Assessrnent

Table 7.3

Reliability Results

Table 8.1

Comparison of Domains Sampled by the BlBC to those Recomrnended by Varnos, Hopwood and Harris

Table 9.1

Distribution of Scores for Scales used in the Validity Study

Table 9.2

Hypothesized Correlation Matrk

Table 9.3

Calculated Correlation Coefficients

Table 9.4

Demographics of 1 Breast and 2 Breast Groups

Table 9-5

Distribution of Scores for Scales used in the Validity Study - Cancer O

Table 9.6

Distribution of Scores for Scales used in the Validity Study - c o n t d ~ m u p

Table 9.7

Hypothesized Differences Between Groups

Table 9.8

Calculated Differences Between 1 Breast Group and 2 Breast Group

Table 9.9

Calculated Differences Between Breast Cancer and Control Groups

Table 9.10

Calculated DifTerences Between Breast Cancer Control Groups for 7 - Point Scale

vii

~ O U ~

List of Fiares Figure 1.1

OveMew of Thesis

Figure 3.1

Summary of Item Generation

Figure 4.1

Summary of Item Reduction

Figure 6.1

Scree Plot

L i s t of Appendices Appendix 1 Appendix 2 Appendix 3 Appendix 4 Appendix 5 Appendix 6 Appendix 7 Appendix 8 Appendix 9 Appendix 10 Appendix 11 Appendix 12 Appendix 13 Appendix 14 Appendix 15 Appendix 16 Appendix 17 Appendix 18

Standards for Evaluating Studies of Clinical Course and Prognosis Semi-Stnictured Interview Scales Identified for Item Generation Content Experts h t e ~ e w e d Sample Content Expert InteMew Demographics of Sample of Women with Breast Cancer I n t e ~ e w e dfor Item Generation Items Generated Items Identifmi as Ambiguous, Offensive or Incomprehensible Items Identified, Included Unmodifieci Redundant Items Eliminated Listing of Retained Items by Domain Items Reduced Using Statistical Methods by Domain Sample 2 Results Rotated Factor Solution for Single Breast Subjects Rotated Factor Structure for Two Breast Subjects Comment Form for the Sensibility Patient Sample Items Administered to the Control Group by Scale Distribution of Scores on Validity Questionnaires Body Image after Breast Cancer Questionnaire Roseoberg Self Esteem Inventory impact of Evenî Scale Beck Depression hventory Marlowe-Crowne Social Desirablity Scale Derogatis Intewiew for S e d Functioning EORTC QLQ-30 Multidimensiouai Body Self Rehions Questionnaire

Appendix 20 Final Version of the Body Image after Breast Cancer Questionnaire

CHAPTER ONE

Introduction Traditiondy, the focus of research in onwlogy has been survival and disease free sumival. However, over the past 15 years, the importance of quality of Life has been increasingly realized and the measurement of quality of Life in clinical triais in oncology has become routine (Organization et al., 1995; Moinpour et al., 1990). The most frequenuy utilized quaiity of life

measures in oncology are functiond statu measures (ICamofslty et al., 1949; Oken et al., 1982) and genenc m a u r e s (some cancer specific) (Moiqmur et al., 1989; Aaronson and Bullinger et al., 1988; Ceiia et ai., 1990) which are designed to sarnple the physical, psychological and social

impact of cancer and cancer therapies on daQ living. Although this approach is ofkn appropriate, by emphasizing features of quality of life of importance to the majority of individuals with cancer, feahires salient to more homogenous subgroups of patients may not be

sarnpled. Thus, more tailoresi measures are required for specific research purposes. Such measures are more likely to discriminate between individuals in the population for which they

were designed. These measures are also more likely to be responsive to small, clinically important changes. (Kessler et al., 1995; Guyatt and Valdhuyzen et al., 1989) It is weU recognized that treatment for breast cancer can have an adverse effect on body image (Anderson, 1985; Derogatis, 1980; Hopwood et al., 1988; Polivy, 1977; Schain, 1985; Hopwood, 1993).

It is not surprishg that in Our breastconscious society, the diagnosis and treatment of breast

cancer can have signifiant psychosexual consequences for atfected women. The potential negative impact of breast loss has led to the adoption of breast presewhg approaches in order to maintain body integrity and satisfaction with appearance. However, even in those women treated with breast presenhg techniques, signifcant problems with body image cm occur (Kiebert et al., 1991). Difficulties can be related to breast surgery or to other aspects of therapy,

such as chemotherapy related alopecia (Baxiey et al., 1984). During interviews with women with breast cancer, feelings commonly expressed by the women afkr diagnosis and treatment

included a decreased sense of attractiveness, an increased feeling of vulnerability, a markedly

decreased sense of control over the body, a feeling of being off-balance, a feeling of being

mutilateci, as well as a marked alteration of sexual functioning. Generic q d t y of iife measures for the oncologic population have not been able to discriminate between groups of breast cancer patients in terms of body image and have been insensitive to change in trials wmparing lumpectomy and mastectomy (Kïebert et al., 1991). As body image is an important focus for research in individuals with breast cancer, an instrument designed to measure body image in this population is required. A review of the current literature reveals that there is no validateci bnef self-report measure of body image suitable for use in breast cancer populations (Hopwood, 1993). Some instruments

designed to measure body image in the general population are currently available. Most have

been designed to measure body size or shape distortions and/or body cathexis in the normal population or body image distortion in patients with eating disorders (Hopwood et al., 1988; Hopwood, 1993; Offman et al., 1992 McCrea et al., 1982; Cash, 1994; Mable et al., 1988; Ben-Tovirn et ai., 1991).

As these measures fail to sample many aspects of body image which are salient to

women with breast cancer (eg. stigma related to mastectomy, feelings of a loss of control over the body etc.), they have limited applicability to these women. Some scales of quality of iife and psychosocial reactions in cancer patients include subscales measuring body image (Schain and Weiiisch, 1994, Aaroason, Bartelink et al., 1988). One scale focuses on the physical

appearance of the breast and surgical scar only, f a n g to sarnple other areas of importance (Aaronson, Baneiink et al., 1988). The other instrument is constructed in a fashion which does

not allow summation of items and thus does not lend itself for use in clinical research (Schain and Weilisch, 1994). Neither masure has items generated directiy from patients and there is no

evidence as to the validity of either measure. Thus, a global measure of body image for breast cancer patients is not currently available and is needrd. Specifc Aims 1. To develop a brief self-report measure of body image in breast cancer which has descriptive

properties and is suitable for use in clinical research.

2. To c o n h the conceptual definition of body image used in measurement construction. 3. To establish the reliability of the measure that has been developed. 4. To establish the validity of the measure that has been developed in a preliminary fashion.

Thesis Format

The thesis will be presented in 7 components. The first component (Chapter 2) is a summary of the relevant background as well as a rewiew of the literature. The second component (Chapter 3) describes the thmry, methodology and results of item generation. The third

component (Chapter 4) describes the theory, methodology and results of the 2 methods of item reduction utilized. Chapter 5 describes the sample recruited for the reliability and validity components. The fourth component (Chapter 6) describes the theory, methodology and results of a principal component factor analysis. The fifth component (Chapter 7) describes the

theory, methodology and results of the assessment reliability (internai consistency and testretest reliabiliw). The sixth component (Chapter 8 and 9) describes the theury, methodology and results of the assessment of the validity of the measure. The seventh component (Chapter 10) is a discussion of the findings and implications for future research. For clarity, an overview of the thesis is presented in Figure 1.1 to provide a framework which may be

referred to throughout the thesis.

Figure 1.1 Overview of Thesis Stage

Cbapter

Technique

Sample

# of Items Remaining

Adoption of

3

4 d o m a h constnict adopted Cornpetence, Cornfort, Appeafance and Predictabiiity

NIA

NIA

3

Items generated h m preexisting

30 content experts 51 women with breast cancer 4 spouses

575

MYImage Constnict

Item Generation

measures a d interviews with content experts, patients and spouses Qualitative Item Rduction

4

Items reviewed for acceptability by wonaen with breast cancer Redundat items elmiinnted

26 women with breast cancer

2 16

Statistical Item Reduction 'SampIe 1'

4

Statisticai anaiysis of 3û9 responses (Frequency of Endorsement, Homogeneity of Items, Percentage Nonresponse, Range of Responses,

360 women with breast cancer recruited

111

R Y ) Recntitrnent of 'Sample 2'

5

309 completed questiomaks 'Sample 1' 187 women with breast cancer recruited to complete the BIBC twice at a 2 week interval + 7 other questionmires

111

165 completed questionnaires 'Sample 2' Factor Analysis

6

Principal Component Factor Anaiysis

Sample 1 Sample 2

Statistical Item Reduction

5& 7

Statistical analysis of 165 responses (Homogeneity of Items)

Sample 2

Reiiability

7

Assessrnent of Interna1 Consistency and

Sample 2

+

Test-Retest Reliability

Sensibility

8

Assessrnent of the Quaiitative Validity of

the BIBC

Vaiidity

9

P r e w assessment of the Constmct Validity of the BIBC

20 Women with breast cancer asked to complete questiondm and comment

53

Sample 2

53

CHAPTER 'Iwo

Background Literature Review 2.1 GeneraI Introduction

Although body image is recognized as an important component of selfconcept, consensus on a singular definition of body image is Iacking in the fiterature. To the present day, the specific

meaning of body image remains obscure, and despite wide application of the term, a universally accepted operationahai definition of this concept and its dimensions is Iacking.

Body image was first described by Henry Head, a neurologist, who in 1920 concephialued body image as the neuronal representation determinhg bodily expenence (Head, 1920). Head was particularly interested in the impact of brain damaging lesions and bodily injury on body perception. He proposed that the body image enables the coordination and integration of sensory experiences such as the perception of size, shape, and position in space. Although the

term body image was coined in 1920, the first written account of the impact of bodily injury on the integration of sensory experiences was documented in the seventeenth century (Pare as reported in Arieti, 1959) when the phantom limb phenornenon was first described. Afier Head,

other neurologists found numerous centrai nervous system disorders which manifestai as body

image distortions. Disorders of the parietai lobe of the brain or the terminus of the sensory pathways for the body are associated with disturbances in the perception of the body. Patients with large lesions in the parietal lobe may be unaware of the hemiplegia and hemianesthesia

which often acwmpany such lesions. This neglect of the affecteci side is termed anosognosia. Other related disorders due to neurological disorders include the lack of recognition of a body

part, the inability to distinguish nght from le&and the inability to identiQ one's own fingers (Harrison's Principles of Intemal Medicine, BraunwaId et al., 1987).

Other investigators becarne interested in the impact of psychiatrie disorders on body image. A

range of body image distortions similar to that found with neurologie lesions were found to be present in some schizophrenic patients (Bychowski, 1943). Distortions described by psychotic patients include feelings that the body is disintegrating, feelings that the body is depersonalized

or alien, and a loss of body boundaries or a f d u r e to differentiate one's body fiom other objects or individuals (Fisher and Cleveland, 1968). Other psychiatrie disorders characterized by body image distortions include the eating disorder anorexia nervosa. In anorexia nervosa, emaciated individuals perceive their bodies to be fat. These individuals negatively evaluate their body appearance and tend to overestimate their body size (Piazza et ai., 1983).

W e many investigators focused on body image in individuals with physical or mental illnesses, the modem conceptualization of body image as an essential component of psychological functioning in nonai individuals, is attribut& to Schilder (1935). Schilder conceptualized body image as "our own body which we fonn in our mi& - a conrcious picture comtrrccfed not onlyftom semry impressions but also from unconrciour libidinous elements

andfrom sociallyfonned images of the body." This definition incorporates the psychological and social meaning of the body both for an individual as well as for an individual in a societal

context ( h c e y , 1986). This definition however, lacks the operational detail necessary for systematic research, and although this lack of sufficient detail has been recognized since 1956, the situation largely persists (Silverstein et ai., 1956). As stated by Thompson, 'the phrase body

image has been used as an wilirella fabel with its specifc meaning depenùing on an individual

researcher's definition' ('ï'hompson et ai., 1990) leading to difficulties comparing the results of research . Despite the lack of a detailed accepteci definition of body image, there does appear to be consensus in the fiterature that body image can be separateci into at least 2 main components, a component, Le. the mental representation of bodily features and actions (such as

.

.

size, shape, and position in space) and an amnidinal component which incorporates the psychological meaning of the body (the feelings and thoughts regarding each bodily feature

and action) (Garner and GarfinLel, 198 1; Onman et al., 1992; Thompson et al., 1990). To test this hypothesis, a number of authors have compared results on percephni measures of body size and shape distortions to attitudinal measures of body image in the sarne study subjects. Few signifiant relationships have ken found between the perceptual and aîtitudinal measures of

body image in the studies performed in the normal population (Cash et al., 1986; Keeton et al., 1990; Thompson and Spana, 1988) and in the eating disorder population (Fabian et al., 1989). This

failure to achieve high correlation between the measures of the perceptual component and the attitudinal component of body image provides evidence that the 2 components of body image

are distinct.

The perceptual component of body image, particularly the ability to evaluate accurately the body's size and shape is extremely important to certain populations. In the eating disorder population, an overestimation of body size has found to be related to a number ofindependent measures of psychopathology and such overestimation has also been found to be associated with poor prognosis in anorexia nervosa putton et al., 1976; Slade et al., 1973). However, in the oncologic population, such perceptual distortions of body image do not appear to be a central feature for the majonty of individuals (Hopwood, 1993). In individuals with cancer, thoughts

and feelings about the body appear to be very important and may be influenced by diagnosis

and treatment (Hopwood, 1993; Katz et al., 1995; Mock, 1993). Because of the importance of the attitudinal component of body image to the oncologic populations, measures of this component

WU be the focus of this thesis. The attitudinal component of body image is not static but can change in response to a number

of variables. For example, changes in feelings of attractiveness have been found in subjects before and after exposure to pictures of very attractive women (Cashet al., 1983). In another study, body satisfaction as measured by a body parts satisfaction measure was found to change

in undergraduate female students with various "guided imagery' situations such as trying on

bathing suits and getting dressed for school (Haimovitz et ai., 1993). Also satisfaction with appearance has been shown to fluctuate over the course of the menstmal cycle (Al*

et al.,

1990). Some components of body image may Vary daily "respondingtofluctuatiom in mood,

stress levez,feelings about the self. water refemnnon and current eoring behaviour" (Sankowsky, 1993, pg. 78). Thus, body image can be thought of as a 'loose mental represenruiion of the

body' (Slade, 1994), which may be influenced by a variety of factors in the normal individual.

As descnbed previously, sensory input and changes in sensory input over time may influence

an individual's perception of and attitude towards herlhis body. The treatment of breast cancer

can cause numerous sensory changes including the removal of a breast, surgical scars over the chest wall, breast and or axilla, and arm changes including numbness, s w e b g and paresthesias. These physical changes may have an influence on an individual's perception of and feelings about their body. Cultural and social norms have an impact on body image. Women particularly appear to be strongly influenced by cultural and social factors with respect to ideal body shape and size.

" Women are more likely than men to eqirare self-wonh wirh what they think they look like and whut they believe oîher people think they look likew (Fallon, 1990, pg. 8 1). Ideal body form varies from culture to culture and has varied greatly over the decades and centuries. Women

have attempted to conform to these norms in, at times, a self-destructive manner. The breast

is a central component of the societal definition of feminity. Therefore, the impact of illness, in pariîcular an ibess which threatens the breast, could be predicted to be a signifiant threat to body image. The impact rnay be profound, given the greater salience of appearance to women and women's greater underlying dissatisfaction with the body. Cognitive and affective variables may have an impact on one's perception of the body. Interestingly, individuals may have dissonance between their cognitive response to the body (how they think about their body) and the affective response to the body (how they feel about their body). For example, a number of research studies have demonstrated that individuals tend to feel that their body size is larger than they actually believe their body size to be (Thompson et al., 1987, 1988 and 1989).

Body satisfaction has b e n found to be related to personal happiness (Berscheid et al., 1973). In fact. body image distortion is included as a cognitive symptom of depression in Beck's cognitive theory of depression (Beck, 1973). The relationship between depression and body satisfaction has been found by numerous authors (MarseUa et al., 1981; McCaulay et al., 1988; Mintz et al., 1986; Noles et al.. 1985; OLTet al., 1989; Sweeney et al., 1989). Depressed individuals

tend to view their bodies more negatively than non-depressed individuals (Sweeney, 1989).

Weight, attitudes towards weight and shape, and history of weight change have been found to influence body image. Higher levels of adiposity have been found to be related to higher levels of body-dissatisfaction and a greater desire for weight control (Bmdie et al., 1988). Obese women have been found to see themselves as less fit than women of normal weight (Faubel, 1989). However, women who were overweight were not found to differ in terms of their feelings of attractiveness or healthiness. Determinhg the factors affecting body image through a review of the literature is confounded

by the lack of a single definition of body image. Just as there is no u n i v e r d y accepted operationalized definition of body image, there is no accepted definition of the attitudinal component of body image. The attitudinal component has been theorized to be multidirnensional (Ben-Tovim et ai., 1991; Cash, 1994; Shona, 1974; Vamos, 1993) and this has

been wnfirmed through research using factor analytic techniques (Ben-Tovim et al., 1991; Brown et al., 1990; Champion et al., 1982; Franzoi et al., 1984; Mahoney et al., 1976; Mendelsohn et al., 1994;

Reed et al.. 1991; Rosen et al., 1991; R y c b n et al., 1981; Slade et al., 1990; Theodorakis et al., 1991; Tucker, 1981). Unfortunately, no uniQing theory regarding the dimensionaiity of the attitudinal

component of body image has emerged from these studies. Different researchers have produced difierent dimensional stnictures of body image, reflecting at least in part, underlying differences in the measures used by the researchers and the populations studied. The instruments used in the studies were designed for a variety of populations (eating disorder populations, normal population, adolescent population) and for a variety of purposes (to measure satisfaction, to measure attitudes, to measure appearance related anxiety, and to measure behavioural tendencies accompanying body image disturbance). Thus differences in the dimensions identifmi for various measures may reflect the different aspects of body image sampled by the measures. For illustration, Cash has designed a measure of body image for use in the generai population. He hypothesized various dimensions of body image to be important in this population and

designed a measure to sample these dimensions (Cash, 1994). A factor analysis of a large

number of responses largely confirmed the factor structure that Cash had originally proposecl. Seven factors were identified; the affective response to physical appearance; the cognitive and behavioural response to physical appearance; the affective response to the fiîness; the cognitive and behavioural response to fitness; the affective response to health; the cognitive and

behavioural response to health; and the cognitive and behavioural response to illness (Brownet al., 1990). In contrast, Ben-Tovim and Walker developed a measure of body image primarily

for use in eating disorder research and tested the dimensionality of body image using the measure. Six factors were identified; feeling fat; body disparagement; strength and fitness; salience of weight and shape; attractiveness; and Iower body fatness. Three weight related dimensions were identified in the Ben-Tovim and Walker measure as compared to the Cash general population measure, where no separate dimension for weight was found. DifTerences in the dimensions of body image identified through these studies iikely reflect clifferences in the measures used and populations tested. In this example, a measure designed for and tested

in the eating disorder population placed a greater emphasis on weight as a central area of salience in body image than a measure designed for the general population. There are a large number of body image measures available for use. The large number of measures would indicate that there is not a single correct way of measunng this construct. Given the complexity of body image, a single measure wuld not be capable of m e a s u ~ gall

aspects of the construct. Therefore, depending on the intended purpose and target population of the measure, different aspects of body image may be sampled. For exarnple, a mesure designed for clinical research in eating disorders would be expected to have a greater number of items related to body shape, size, and weight than an instrument designed to measure body image in patients undergoing limb amputation for soft tissue sarcoma. A researcher choosing a measure, or designing a measure for a population must decide which aspects of body image are important to the population king tested and should be measured given the purpose of the

research study.

2.2 Body Image Aspects of Salience to the Breast Cancer Population

In order to evaluate or design a measure of body image for breast cancer patients, it is necessary to decide which aspects of body image are essential to evaluate in this population. The impact of illness on body image is an important but poorly explored area. Body image is

felt to be essential to the maintenance of self-concept and psychological functioning, and thus the threat to the body image pose. by physical illness, by its effect on body appearance and

function, rnay have an impact on a wide range of psychosocial functions. Physical iUness may also cause bodily changes resulting in the need for adaptation of the body image. However, little research has been done in the attempt to determine the aspects of body image most salient to people with medical, surgical or oncologic ilinesses. Several authors have attempteû to develop a theoretical framework for evaluating the impact of illness on body image. Shontz (1974) proposed a cornplex structure of body image comprised

of a number of body functions acting at various levels of bodily expenence. Shontz proposed 7 functions that the body performs (the body functions); The body is a =ory

processor of sensory information; The body is an instrument to

reg&.a

and

out b&aviours from

simple reflexes to abstract thinking; The body is a source of nez& for example, sleep, food etc; The body w

s a pnvate woru and dows for individual experience; The body is a

of sfimuü to the self, for example in proprioception; The body is a

of social

stimuli,for example physical appearance is an important social stimuli; The body is an

.veinstniment . enabling communication though gesture, posture, facial movement etc.

These 7 functions of the body that incorporate body image are not discrete. Each function can

act at 4 levels of body experience in an integrated and hierarchial manner. Four Levels of

me Body Sche-: This level of experience enables the perception of the body as an object in space; me Body Self: This level allows differentiation of expenences

bodily expenence are;

into those which happen to the self and those which do not; n e Body F m : This level of

expenence incorporates the symbolic and metaphoncal images of the body;-hT -:

This level embodies knowledge of the body aquired through education versus experience. The 7 functions of the body may act at any level of body expenence.

Shontz proposes that disability or illness may affect an individual's body image at any function

as well as at any level and outlines 5 main physical disorders which may influence body image; Disorders resulting from damaee to the b h , for example changes in gait due to brain metastases from breast cancer; Disorders due to

to 0 - 3 - o m

without loss of body parts particularly as they affect posture and movement, for example permanent neuropathy with sensory loss caused by chemotherapy neurotoxicity; Disorders due to x u t e ,to

for example the loss of a breast due to breast cancer; Disorders due

. -ormetabo for example problems related to chemotherapeutic toxicity; *

for example chemotherapy e Disorders due primarily to f i m

induced menopause in the treatment of breast cancer. This framework for evaluating the impact of illness on body image is extremely complex, and this complexity makes testing difficult. The use of this framework in the breast cancer

population would require the evaluation of the impact of diagnosis and treatment on the 5 disorders which would then influence body image through 7 functions at any of 4 levels of experience. The wmplexity of this schemata precludes its use in evaluating or developing clinical rneasurements. Aiso, the framework emphasizes a perceptual and cognitive view of body image versus the affective reaction of an individual to the body. The affective reactions

to the body are felt by many authors to be an essential of body image in many populations (Brown et al., 1990; Gallagher, 1986; Garner et al., 198 1; Offman et al., 1992; van der Velde et al., 1985).

Vamos (1993, pg. 171) has attempted to created a simpler framework for evaluating the aspects of body image most salient to the chronically ill. As stated by the author "body image in

chronic physical illness is trandcued into a compler array of infernal represeniaiions or body images. niese will Vary both in t e m of the specifc characteristics of each disorder, and ako in r e m of pananentage, gender, life tqerience and 1(suai coping mechcuiisms." Four aspects or dimensions of body image are proposed, and these four dimensions are theorized to incorporate all the areas of body experience affected by chronic physical illness. The first dimension, Cornfort, is defiiied as the personal experience of the alteration in sensory

experiences because of disease or treatment including such items as pain, tiredness and nausea. is defined as the evaluation of changes in functional ability and includes items such as cognitive ability, mobility and sexual functioning.

is defined as changes

in outward appearance because of disease or treatment and the obviousness of the disorder, and includes self and other evaiuated aspects of appearance.

is the final dimension

and is defined as the stability of ilhess over time, and includes the degree of variability,

suddenness of change, age appropriateness of disability, fear of recurrence and feelings of control over the body functions. This framework for evaluating the impact of chronic illness on body image is relatively simple and is directly applicable to the oncologic population. The fiamework is focused on the illness

experience and was designed for ease of application and maximum relevance to patients and clhicians. However, the application of the framework to the measurement of body image in ill populations has not been detailed in the literature. Despite this, as the framework is simple, sensible and compelling, when attempting to constmct a measure of body image for the breast

cancer population or evaluating such a measure, these aspects should be sampled. Table 2.1 illustrates how such a format might be applied to evaluate the impact of illness on body image, in this case for a woman with recently diagnosed primary breast cancer who has had a mastectomy and is undergoing chemotherapy. Of note is the interaction between the individual

effects of illness and treatrnent (change atîributed to cancer), the severity of the effects (magnitude and direction) and the patient interpretation of the meaning of these effects in terms of the impact on body image.

-

Table 2.1 Body Images in Hypothetical Fernale Patient with Recently Diagnosed Primary Breast Cancer treated with Mastectomy and Chemotherapy

Direction Cornfort

Improving

Constant reminder of cancer A m feels alien

Tiredness

Severe

Feeling old, sick

Nausea

Moderate

Body is being

Surgical Pain

Improving

poisoned Cornpetence

A m Function

Improving

Body is healing

Sexual Function

Very Poor

Normal Duties

Unable to Perform

No longer a "realn woman Feels useless and like an old wornan

Worsening Worsening

Looks Sick

Very Poor

Feels deformed, Mutilated, revolting

Severe

May not live to see children grow up

Feelings of Body being Invaded

Severe

Thoughts of cancer taking over body

Worry about Aches and Pains

Worsening

Hyper vigilance Cannot trust body to tell her when something is wrong

Feelings of Sexual Attractiveness

As body image implies awareness of the self in relation to both the body and the whote sociocultural environment (Wassoer, 1982), it is not surpnsiig that visible malformations and physical changes caused by surgery or other traumas may have a drarnatic impact on body image. Using qualitative methodology, Harris (1982) attempted to conceptualize the impact of abnormal appearance in a study of 54 post-operative patients; 14 subjects had congenital malformations, 11 had abnormalities resulting from disease or injury, 6 had abnormalities

resulting from physiological processes and 23 had abnomalities resulting from developmental disproportion. The subjects were asked to write anecdotal accounts of their expenences with abnormal appearance. The data were reviewed qualitatively (although the qualitative methodology used was not stated) and a common 6 part sequential pattern of coping with abnormal appearance was identifieci: 1. Induction and deveiopment of self-consciousness Subjects described the point at which they became aware of the abnormality, and the point at

which they became setf-conscious of the abnonnality. This self-consciousness was reinforced

both by self-criticism and the criticism of others.

2. Defence mechanisms Subjects described going to great lengths to hide their abnormalities. Three major techniques emerged. a. Camouflage techniques, including changes in posture and dress. b. Restrictions to lifestyle, avoiding any situation which is potentially embarrassing.

c. Amficial behaviour adopted to conceai the degree of self-consciousness and distress experienced because of the abnormality 3. Unavoidable distresshg activities

Subjects described the stress assuciated with activities which exposed their abnormality which could not be avoided. 4. Downgrading of sekoncept This included feelings of inferiority, unattractiveness, insecunty and loss of femininity which

were descnbed by two-thirds of the subjects.

5. Difficulties with interpersonai relationships

Subjects descnbed difficulties in relationships with strangers. Difficulties with sexual and

marital relationships were commoniy mentioned by women with abnormaiities of the breasts. Subjects also d d b e d a feling of isolation and inability to obtalli understanding from others. 6. Rationalization of disabiiity and associated distress

Subjects in the study wmmonly attempted to rationalize their abnormality; however most were unable to do so successfully. This is one of the few studies in the fiterature which attempts to understand the impact of abnormal appearance on self-concept and body image in an experiential, patient focused manner. It provides a cogent and compelhg theoretical framework, and would be applicable to the breast cancer population in t m s of the adjustment to the physical deformity,

particularly mastectomy. Therefore, the six areas covered in this framework should be sampled when evaluating or developing a body image measure for the breast cancer population. H o p w d (1988, 1993) has examined the importance of the effect of cancer on body image. The sequelae of cancer treatment on body image cm be marked especially "when body inîegrity is breached or body fwicîion fis) uolrered as a resulr of medical inletvention."

Hopwood identifies 7 areas essential to explore in the evduation of body image in the oncologic population: 1) Dissatisfaction with appearance (dressed) 2) Loss of femininity/rnasculinity

3) Reluctancelavoidance to look at self naked 4) Feeling less attractivdsexually attractive

5) Adverse effect of treatment/loss of body integrity

6) Self-consciousness about appearance 7) Dissatisfaction with scar/prosthesis It is unclear how Hopwood has identifid these areas of salience. However, they provide the only available fiamework for evaiuating body image specifically in the oncologic population in

the literature. Therefore, these areas must be carefully considered when evaluating or constructhg a suitable measure for women with breast cancer.

Despite the attempts by several authors to develop a theoretical framework for evaluating the impact of illness on body image, research evaluating the validity of the various theones is not available. Studies of body image in the medically ill have been conducted using a variety of measures; some validated in the general population; some created ad hoc. The measures of body image which have been previously use. or are available for use in the breast cancer population will be cntically reviewed. The criticai review WU focus on general measurement principles (which wili be briefly outlined to provide a fmmework for evaluation) and the suitability of content. 2.3 Measurement Methodology

With the increasing focus on quality of life issues in clinical practice and research in medicine, the measurement of 'subjective' concepts such as body image has taken on greater salience. The need for precise methodology in such measurement has thus become essential. As stated by Streiner and Norman "therupeuticefforts in many disciplines of medicine...are direcred

eqirolly ifnot primari& to the improvement of qualiry, not quamiry of life. If the efforts of these disciplines are to be placed on a sound scienn!$c bais, methodr m m be devised to measure whut was previousiy thoughr to be umeasurable, and msess in a reproducibk and

valid f&on those subjective srares which cannot be converted imo the posirion of a needle on a diain (Streiner and Norman, 1989, pg. 1-2). It is possible to measure such subjective States and psychological concepts in a valid and reproducible manner. The prhciples goveming the measurement of body image are the sarne as those goveming any psychological concept. These principles should form the basis foi the evaluation of existing measures and should guide

the creation of any new measure. Cronbach defines a psychological measure as "systematic

procedure for observing behaviour and descn'bing it wïth the aid of numerical scales orfied eutegories" (Cronbach, 1990, pg. 32). This definition incorporates a broad scope of assessments including 'repons on penonulity, procedures for observing social behaviour, apparahû tests rneasuring coordinutïon, and even records of output on a production line." The hindamental features of psychological measures have been described by Anastasi (1988) and provide a framework for the evaluation and development of measures.

1. Psychological measures are a

of b b v i o u r . A rneasure is constmcted such that

responses to a select number of items are extrapolated to make judgements about a broader phenomenon. For example, a bnef rneasure o f body image would include oniy some of the many items which could be used to measure body image. Thus, such a measure would consist of a sarnple of items (or behaviours) which are then extrapolated to make judgements about the concept body image. The items included in a measure should therefore be chosen to be representative. The sample of items rnay be utilized to predict future performance, to diagnose, to detect change, to determine relationships to other phenomenon and to classify (or differentiate) (Anastasi, 1988).

2. Psychologid measures must be .-

It is essential that the results of

psychological measures reflect the attributes of the individuals evaluated as opposed to differences in test circumstances or the biases of observers. Standardization implies that psychological measures must have prescribed conditions for administration. This also implies that an individual taking the measures would obtain the same score if evaluated by different observers. Objectivity can be maxirnized by the use of scoring systems which rninimize hterpretation (such as multiple choice questions). Although perfect standardkation

and objectivity are not attainable in a practical world, all measures must strive to maximize

these characteristics. 3. Psychological measures must be m.Anastasi (1988, pg. 109) defines reliability as "theexteru ro which individual differences in tesr scores are arfribmble îo 'true' differences in

the characce&ics under comi&rmanon Md the w e m ro which they are mburoble to chance" or random error.

To be useful, an instrument must be reliable, providing precise and consistent information. Measures of reliability quanti@ the amount of variance in the observed scores which is due to

true variance versus the amount of variance in the observed scores which is due to random error (Cronbach, 1990). If there is no random error, reliability will be 1.O. The most common rnethods of assessing measurement error, (or reliability) include intemal consistency, test-retest reliability, and inter-rater reliability. The aspects of reliability of importance to this thesis will be discussed in more detaü in chapter 7.

Measures of intecial consistency estimate the amount of error due to the sampling of items,

and hence estimate reliabiüty based on the domain-sampling mode1 of test theory. The most commonly used measure of intemal consistency is coefficient alpha. Coefficient alpha ranges from O to 1, with O indicating no homogeneity and 1.0 indicating perfect homogeneity. Its value depends both on the average correlation of items and the number of items. Reliability of a multidimensional instrument can be estimated by calculating the coefficient alpha for each homogeneous domain. st

Re-

. ..

Test-retest reliabiüty assesses the stability of a measure on two or more separate occasions (Anastasi, 1988; NuoaaiIy et ai., 1994). Differences in scores between the test and retest

correspond to random fluctuations in responses over time and thus are an estimate of the amount of variation in the observed score which is due to random error. Test-retest reliability does not depend on the assumption of homogeneity and thus would be supenor to coefficient alpha for heterogeneous measures. However, in some cases the experience in the first testing may influence the responses to the second, leading to a change in the correlation between test and retest. Also, if real change in a characteristic has occurred between test and retest, this variation will be incorrectly considered error. The intracIass correlation coefficient is the most appropriate statistic to determine the degree of concordance between the test and retest and varies from -1.0 to 1.0, with -1.0 indicating perfect discordance and 1.O indicating perfect concordance (Streiner and Norman, 1989).

The reiiability of measures which require the judgement of raters, suck as interview assessments of quality of life requires special consideration. Variation in the observed score WUbe a combination of true variance and error variance which will be composed of error

variance fiom domain sampling and the error inherent in the subjective assessrnent of individual judges (Anastasi, 1988). Such an instrument is considered to have inter-rater reliability if different judges, using the measure independently, obtain similar scores.

Standards for reliability d a e r accurding to the proposed use for the measure. When test results are to be used to corne to conclusions about groups, the standard of reriability is felt to be lower than when decisions are to be made about an individual. Thus, for research

purposes, a standard of 0.70 is felt to be sufficient for rnaking decisions about groups, while when making decisions about individuah, a reliability level of 0.90 is recommended (Nunnally, l994).

4. FinaUy, a psychological measure must be

a; it must measure what it daims to

measure. Reliability is a minimum criterion for vaiidity. (If the observed variation in test scores is due largely to error variation, the rneasure cannot be vatid.) However, even if a measure has perfect reliability it may not be valid. For example a very reliable rneasure of depression would not be a valid measure of functional limitation in cancer patients. If there is a standard (or accepteci criterion), the cntenon validity of a rneasure may be established by comparing the results of the measure ufider study to this standard. This may be done concurrently, by wmparing results on the study rneasure to a gold standard of measurement. If the rneasure is designed to predict an outcome, then results can be compareci

to the actual outcome. Unfortunately, in the case of most psychologie measures, there is no gold standard and thus other approaches to validation must be undertaken and evidence supporting the validity of a measure gradually accumulated. The most important approaches to establishing validity (face validity, content validity, construct validity and responsiveness) will be discussed (Anastasi, 1988; Cronbach, 1990;Nunnally, 1994; Streiner and Norman, 1989).

..

e V&&y

This refers to the apparent validity of a measure, ie. the degree to which a measure looks like it is assessing what it purports to measure to the clhicians and patients who use it. Although

an instrument which 'looks good' may fail to meet other more fundamental components of validity, face validity is important. If an instrument does not make sense to clinicians and patients, no matter how reliable and valid it is, it is unlikely to be widely used.

As described by Hays et a1 (1993), content validiîy 'is the extent to which a measure samples a

representative range of the behaviour d e r s@.'

A measure should examine a representative

sample of dl areas of importance to the domain to be measured. There should be a sufficient number and range of items in a measure to adequately sample the domain of interest. Conversely, areas which are not salient to the measured domain should be excluded from the measure. The extent to which a measure fülfils these objectives determines its content validity. There is no standard procedure to assess content validity; it is largely done using expert judgement.

Thus, in order to establish content validity, a clear concept of what is

being measured is essential. Without an adequate definition of a construct, it is impossible to determine if the domain is adequately sampled. Face and content validity are discussed in

more detail in chapter 8.

.

Construct v&&y *

If an instrument is measuring what it purports to measure, then one may predict how the instrument should behave under given circumstances. This relationship foms the bais for constnict validity. If predictions about how an instrument should behave under prescnbed conditions, based on the theoretical understanding of both the instrument and the 'COI~SCTUC~' being measured, are proven to be true through research, then evidence of constnict validity of the instrument is provided. In order to establish that an instniment has constnict validity,

hypotheses about how a measure should behave are generated and are then confumed or refuted through testing. Validity is supported if the findings of research support the proposed hypothesis. In constnict validation, both the instrument and the underlying theoretical construct

for the instrument are tested. Construct validation is a gradua1 process and requires the testing of multiple hypotheses. A number of different methods for establishing constnict validity exist. These methods will be described in further detail in chapter 9. Measures of subjective phenornenon may be used to evaluate the effect of treatment. If the measures are to be utilized in this fashion, it is essential to establish the ability of a measure to detect clinicdy important change when it occurs. If this characteristic is not estabfished, a

failure to detect a change in a study may either reflect a lack of real change in the characteristic under study or a failure in the ability of the measure to detect a change when it

was indeed present. Although not accepter by all, responsiveness is considered to be an aspect

of construct validity by some authors (Tulsky, 1990). In summary, the fundamental properties of a subjective measure test have been bnefly

presented. These properties wiil be reviewed in hirther detaii throughout the thesis. These

properties wiU be used in the evaluation of existing body image measures and must be considered during the development of any new measure. 2.4 Measurement of Body Image

The fundamental principles governing the measurement of subjective phenornenon in general

are directly applicable to the measurement of body image. A measure of body image must be constmcted in a manner which ensures it is standardized . The measure must appear valid and must sample appropriate content in a thorough manner. Given the difficulties in definhg body image, a clear statement of the hypothesized constnict of body image utilized in measurement development is required in order to fairly judge the content validity of the measure. The instrument must be show to reliably sample the hypothesized construct. Finally, any instrument must be shown to measure the constnict it purports to measure; that is, the instrument must be proven valid in terms of the measurement of body image. Many authors have attempted to develop measures of body image constmcts. The measures developed can be grossly subdivided into perceptual measures and attitudinal measures of body image, corresponding to the accepteci two main components of body image described in Chapter 2.1; the

. . v; and the attitudinal CO-

. Perceptual body image

measures attempt to determine an individud's ability to perceive body size and shap (Le. bodily dimensions) accurately. Attitudinal measures of body image attempt to evaluate an individual's subjective appraisal of the physical self. Cash descnbes the assessment of self-

attitudinal aspects of the body image constnict. "Here,bodj image is conceived as one's atrrWtudinaldispositions toward the pphycal self: As ~ h ï d e srhese , dispositions imlude

and behavioural components. Moreover, the affective/evaluative, cognitive/1~tentr'oml,

physical self encompasses not on& the aesrhercs of one's physical size/appearance but ais0 its cornpetence or @ness' und its biological i~egriryor 'health/illness'"(Cash,1994, pg. 1). Thompson States that although the majority of measures of the attitudinal dimension of body image focus on satisfaction with the physical self other aspects of this dimension include

"subjectiveconcem, cognitionr, anxieq and unnnncipated avoidance of cerrain sinc~ul~otls" (Thornpson et al., 1990) and behaviours. As descnbed in Chapter 2.1, the main impact of an m e s s such as breast cancer on body

image is felt to be largely in the attitudinai versus the perceptual component of body image. Although some individuals with breast cancer rnay have perceptual distortions of body image, and some effects of treatment for breast cancer rnay lead, in some individuals, to perceptual distortions (Le. treatment related weight gain), such distortions are unlikely to occur as a result of illness diagnosis and treatment in the majority of women with breast cancer. Conversely,

feelings and thoughts about the body rnay be dramatically altered by the diagnosis and treatment of b r m cancer. Thus, the remainder of the discussion will focus on the measurement of the attitudinal component of body image. The attitudinal component of body image is multifaceted (Thompson, 1990) and can be grossly divided into 2 major constituent parts; an affective dimension (how one feels about the body) and a cognitive dimension (how one thinks about the body.) Either or both of these dimensions rnay be sampled by a measure. In turn, a measure may assess satisfaction with the body, concems or anxiety about the body, behaviours based on feelings or thoughts about the body or some combination of these aspects of body image. In addition, a body image instrument rnay concentrate on an assessrnent of individual aspects of the body, such as body

parts (Berscheid et al., 1972; Fraazoi and Shields, 1984; Secord and Jourard, 1953; Slade et al., 1990), it rnay focus on a single dimension of a body image constnict (Derogatis, 1975; Fitts, 1%5), it

rnay be a multidimensional measure based on a global constnict of body image (Cash,1994), or it rnay be developed for a specific population (Ganz et al., 1990;Schag et al., 1990;Schain and Weliisch, 1994). For review, the measures will be divided into 2 broad categories; measures developed for the general population; and measures developed for oncologic populations.

24

Several body image mea~ures,(Berscheid et al., 1972; Franzoi and Shields, 1984; Secord and Jourard, 1953; SIade et al., 1990) including some of the oldest measures, have assessed satisfaction with

individual body parts with the assumption that the mean satisfaction rating with various body parts is a valid rneasure of the o v e d 'body cathexis' or satisfaction with the body. These

instruments consist of a listing of individual body parts which the respondent must consider. The respondents must then denote the degree to which they are satisfied with each body part. The body parts sampled by these measures range from the chin to the ankles. This type of

body image measure has limited utility for the breast cancer population. Many of the body parts listed are unlikely to be of salience to the breast cancer population (for example c h s and

ankles) and many areas of great salience to the breast cancer population are excludecl (for example, breasts). Also, the relationship between body-parts satisfaction and body image as a

theoretical constnict is unclear. The validity of measures of satisfaction with individual body parts as global body image measures has not ken established. As stated by Ben-Tovim and

Waker (1990, pg. 164), 'Zr appears that the emphars on sa&isf~cn'onwirh the body, which is the legacy of Jourard ond Secord's work, may be a comtruct imposed on women by

researchers, rather than a dimension thus women necessariiy hotd ro be importani in their Mews of their bodies.' Also the body-parts satisfaction instruments are unlikely to be valid

measures of broader body image constnicts, such as that proposed by Vamos for the chronically ilI, as these measures do not assess body image in a multidimensional fashion. By focusing only on satisfaction with parts of the body, the measures fail to sarnple essential areas such as feelings of being in control of the body, feelings about symptoms related to cancer or feelings about iunctional limitations attributed to cancer or treatment of cancer. By sampling

only a single aspect of body image, satisfaction with body parts, these measures are of lirnited use in clinical research in the breast cancer population. Despite the limitations of body-parts satisfaction m a u r e s , the measures have been used to some degree in the breast cancer population. A study examining differences between cholecystectorny and mastectomy patients (Ray, 1977) failed to show a difference in body image between these populations utiliEng the Body Cathexis Scale (Jourard and Secord, 1953). Pohy

(1977) modified the Bershied Walster and Bohmstedt Body Image Scale (l972), increasing the

relevance of the scale to the breast cancer population, in a shidy of women undergoing mastectorny. Although significant change was demonstrated in the score for the mastectomy group over time using this measure, no direct comparison was made between the group undergoing mastectomy and the control group. Thus the scores in the mastectomy group on the scde may not have been significantly different fiom the scores of the controls. In a study comparing women who had undergone mastectomy to women with gynaecologicd cancer, a higher score OF.the Bershied s a l e was found in the gynaecological cancer group (Krouse and Krouse, 1981). The authors conclude that the gynaecoiogical cancer group had a significantly

poorer body image than a group of women who had mastectomy for breast cancer. In reaching their conclusions, the authors did not note the inherent limitations of the scale in assessing an oncologic patient 's body image. Polivy's modified Body Image Scale was utilized in several other studies (Penman et al., 1987; Reaby et al., 1994). In one study, no difference between mastectomy and lumpectomy patients

in terms of body image was found using this measure, despite the generai consensus that women who have had a lumpectomy after breast cancer have a more positive body image than women who have had a mastectomy (bine et al., 1991; Kiebert et al., 1991; Shover, 1991). In the second study, a non-cancer control group actually expressed

body image dissatisfaction

than women with breast cancer! These findings indicate that assessing satisfaction with body parts is not an adequate method of assessing a global constnict of body image in women after breast cancer, lacking face validity, content validity and constnict validity for such purposes. Severai measures of body image have been constructeci which measure a single aspect of the body image constmct (Derugatis, 1975;Fim, 1965). The Tennessee Self-Concept Scaie (Fitts, 1965) is a well validated measure of self-concept which contains a physicai self subscale. This

subscale measures the physical self-concept component of body image. This measure was utilized in a study comparing subjects with mastectomy to control subjects and failed to show any difference between these groups in terms of body image (Jenkins, 1980). The s a l e was

also used to compare individuals with lumpectomy, mastectomy and reconstruction and failed to show any differences between these groups (MOCL,1993).

The body image subscale of the Derogatis Sexud Functioning Inventory (Derogatis, 1975) measures the body image component of sexuality and sexual attractiveness. This subscale has been used in 2 studies in the breast cancer population (Kornblih et al., 1993; Wolberg et al., 1989) and has not discriminated between women with mastectomy and women with lumpectomy or

women with metastatic breast cancer at widely varying doses of megestrol acetate. Fisher and Cleveland (1958) have developed a projective measure of the attitudinal component of body image, known as the Banier and Penetration Score. Scores for the measure are based on responses to Roschach inkbots. Fischer (1986) noted that people Vary ' w l h respect to the finnness or definiteness they amibe to their bo@ boundaries. ' The Barrier and Penetration

scores attempt to quant@

this body boundary construct. The scores purport to 'indicate

subjects' (rttl2udes towards th& own bodies as barnoersagainst the outside world' (Ben-Tovim and Walker, 1991). This measure has b e n used in one study in the breast cancer population

comparing 20 wornen who had had lumpectomy to 20 women who had had rnastectomy for the treatment of breast cancer (Sanger, 1981). Women with lumpectomy were found to have significantly higher barrier scores than women with mastectomy. No significant difference was found in penetration scores. The Barrier and Penetration scores have not been widely used and are not suitable for general use in clinical research. Completion of the measure is time consurning, requiring an interview with trained personnel. The measure also relies on the subjective interpretation of patient responses by the researcher, again making widespread use in clinical research impossible. Thus it appears that these generic measures of specific components of body image are of limited utility in the breast cancer population, lacking face validity, content validity and constnict validity. These measures fail to sample a broader constnict of body image as recommended by Vamos, Hopwood, and Harris. As descnbed in chapter 2.1, Cash had developed a multidimensional body image measure for

use in the general population. It is felt to be the most comprehensive and extensively validated body image measure available (Thompson et al., 1990); however the measure has not been

utiiized in the breast cancer population. On inspection, some wmponents of the measure would appear to be of limited salience to the breast cancer population, (for example, questions regarding physical fitness and participation in sports). In fact, of the 7 areas felt to be

essential in the evaluation of body image in the oncologic population by Hopwood, only 3 are addressed in this instrument. Applying the construct of body image in the chronically ill proposed by Vamos, no dornain is found to be adequately sampled by the instrument. The cornpetence domain is sampled oniy by questions relating to physical fitness and exercise. No questions sample the comfort domain. Aspects of the appearance dornain are not sampled (for exarnple, the impact of disease and treatment on appearance as well as the obviousness of the disorder), and the predictability dornain is sampled only through general questions about health. Thus, the Multi-Dimensional Body Self Relations questionnaire lacks content validity for use in the breast cancer population. Four measures relevant to this discussion have been created for use in the oncologic population. One quality of life questionnaire has ben constructed for the oncologic population and ùicludes a subscale measuring body image (Ganzet al., 1990; Schag et al.. 1990), one measure of psychosocial reactions to breast surgery includes a body image component (Schain and Wellisch 1994) and two body image measures have been constructed for use in the breast

cancer population (Feather et ai., 1989; Lasry et al., 1987). The CARES measure (Ganr et al., 1990; Schag et al., 1990) is an instrument designed to assess

the rehabilitation needs of cancer patients and has been proposed to be a measure of quality of life. It has a 3 item body image subscale focusing on embarrassrnent and discornfort with body and b o d y changes. This subscale has not b e n assessed for validity in rneasuring body image; however, it did prove capable of discnminating between individuals who had undergone lumpectomy and mastectomy (Ganz et al., 1990). Although this measure does appear to sarnple a component of body image, the scope of body image which is assessed is very limited, and not included are many aspects of body image considered to be essential in the

breast cancer population. Thus the measure lacks content validity. Also, given its brevity ,

with only three items rated on a 5 point Likert scale, this measure may have lirnited

discriminative ability, Le. a large difference between groups may be detectable by this instrument (such as the difference between lumpectomy and mastectomy) w hereas smaüer differences may not be. Another measure, Psychosocial Reactions to Breast Surgery, has been developed for the breast

cancer population (Schain and Wellisch, 1994) and includes many items relating to body image rated on a 5 point Likert scale from strongly agree to strongly disagree;

I. 1 like my loohjw the way they are 2. 2feel good uboru my M y 3. 1 would l i k to chunge some pans of my body

4. I feel less physicdy attradve now than before my breasf cancer 5. I feel less sexually desirable now than before my b r e m cancer treament 6. My breast c m e r n e m e n t has made me feel ashamed of my body 7. My body looh us good as it did before my breast cancer

The measure also includes items summated on a 4 point Likert scale ranging From occumng

not at ail to a lot; I. Feelings of not being able to corurol evenrs in your life 2. Concem t h a cancer will recur

3. Negan've feelings about your appearance nude 4. Negative feelings abou your appearance dressed 5. Problem carrying oui household chores

6. Problem carrying o u work ncn'viries 7. Problems participuting in sports 8. Problem in your stzrual relmions 9. Problems infeeling self-comcious in groups of women

10. Problems infeeling self-consciour in mixed groups As well there are a number of questions relating to breast appearance which Vary according to

the type of breast surgery an individual has undergone. The items are not summated in any fashion but are analysed on an item by item basis.

The measure appears to sample a broad body image construct, including cornpetence and appearance. However the cornfort and predictability domains of the constnict of body image for the chronically ili are not adequately sarnpled. Also the failure to create some means of summating the items in any fashion iimits the usefulness of the measure in research. There is

a lower reliability inherent in comparing individuals based on items versus scales made up of a number of items. Additionally, when comparing groups, given the fact that a large number of

items must be compared versus the score of a single scde, the chance of a type 1 error (the probability of finding a difference between groups when a difference does not tmly exist) greatly increases. The interpretation of disparate results on items would also be difficult. Thus this maure is Lirnited in its usefulness in chical research in the breast cancer population. A third instrument was developed in the breast cancer population for a study comparing the

body image of individuais who had undergone lumpectomy and mastectomy (Lasry, 1987). The items sample the following areas; 1. Satisfaction with breast appearance 2. Satisfaction with breast texture 3. Satisfaction with body appearance 4. Attractiveness of patient according to others 5. Attractiveness change due to operation

6. Description of scar (revolting - beautifil) 7. Fear of not king sexually attractive

The reliability of the scale has been assesseci and a coefficient alpha of 0.83 was found (Lasry, 1995). The sale has also b e n found to discriminate between women with mastectomy and

women with lumpectomy. However, the scale appears to sample a narrow constnict of body

image, focushg on appearance only. Three of the seven questions relate to the breast or surgical scar, and this limits the applicability of this scale in comparing groups differing in nonsurgical parameters. Also it is unclear how women with a mastectomy should answer questions about breast appearance and texture. Thus, this sale does not appear adequate for measuring a broader body image constmct in clinical research not limite. to the evaluation of surgical outcornes.

The final scale, the Mastectomy Attitude Sale (Feather, 1989) is specific for populations of

women who have undergone mastectomy, and thus is not applicable to individuals who have had other surgical procedures for the treatment of breast cancer. In summary, there does not appear to be an adequate measure of body image available for

clinical research in the breast cancer population. Measures designed for the general population lack content validity. Measures designed for the oncologic population also lack content validity and in some cases are lirnited to specifc populations only. Therefore, the development of such a measure is justified. After a review of what is currently known about the impact of breast cancer on body image, the rernainder of the thesis will describe the

development and testing of a body image measure developed for use in the breast cancer population. 2.5 Body Image Shidies in the Breast Cancer Population

The breast is a symbol of femininity and sexuality in Our society, and thus it would not be

surprishg if breast cancer and treatment for breast cancer had a significant impact on body image. There is a literature evaluating the effects of breast cancer on body image. The majonty of studies attempt to determine if charactenstics of treatment (eg. lumpectomy versus

mastectomy) predict body image status after varying periods of time. The studies Vary greatly in quality. The authors in general do not articulate the conceptualization of body image which

is being studied, the studies utilize a wide variety of measures (frorn percephal evaluations to body esteem measures), and tend to have small sample sizes. Because the majority of studies attempt to determine the clinical course of individuals with breast cancer in terms of body image outcorne, the studies will be summarized and evaluated using methodologic standards outlined by Sackett and colleagues for criticaliy appraising studies of clinical course and progsosis (Department of Clinical Epidemiology and Biostatistics 1981; Laupacis et al., 1994; Sacken et al., 1985). Sackett proposes 6 criteria for evaluaîhg articles to

learn the clinical course and prognosis of disease, in this case, determinhg the clinical course of breast cancer in terms of impact on body image.

For the purposes of this iiterature review, these criteria have been modified for use.(Appendix 1) The following 5 criteria have been used as a framework for critical appraisal. 1.

Assembly of Cohort: Methods for selection of subjects for study should be clearly identified such that the results are interpretable and selection bias is minimized.

2.

Description of Population: The referral pattern and the patient population included

should be throughly described to enable the evaluation of the potential impact of referral bias on study outcome. 3.

Respoose Rate: It is essential that a high rate of response (or foilow-up) is achieved. An 80% rate of response has been suggested as a standard (Irvine, 1991).

4.

Outcome Measure: It is essential that the outcome measures utilized in the studies be

reliable and valid. If this is not the case, the meaning of the study results becorne uninterpretable. Also, if the measure requires the subjective evaluation of an individual other than the subject, (i.e. the measure is not a self-report inventory) the evaluator should be biinded to major prognostic factors.

5.

Adjustment for Prognostic Factors: Adjustment for possible confounding variables,

such as age and time since diagnosis, should be underiaken to minirnize the influence of these factors on results.

For the review, the studies of body image in the breast cancer population will be divided into 4 major categories, reflecting the aspects of treatment which have studied. Most research evaluating the impact of breast cancer on body image has focused on surgical treatment, thus the fist 3 categories evaluate the impact of types of surgical treatment. The surgical studies were in general designed to determine the impact of the various surgical options on body image (mastectomy, lumpectorny, or mastectomy with reconstxuction) or to compare outcome after mastectomy and lumpectorny. The review has been structured to reflect this: 1. Studies of the impact of mastectomy 2. Studies comparing the impact of lumpectorny to mastectomy

3. Studies of the impact of breast reconstruction after mastectomy 4. Other studies

1. Studies of the impact of mastectomy (Table 2.2) Six studies have atternpted to evaluate the impact of mastectomy on body image (Feather et al.,

1988; Jenkins, 1980; Krouse and Krouse, 1981; Penman et al., 1987; Polivy, 1977; Ray, 1977) and these

are presented in Table 2.2. The studies have had conflicting results. Most of the studies had smaU sarnple sizes, used body-parts satisfaction rneasures and found no difference in body image between women who had undergone mastectomy and controls, or did not perfom a

forma1 comparison. The one large study which compared an inception cohort of women who had had a mastectomy to a control group demonstrateci that women who had a mastectomy and adjuvant therapy expressed significantly more body image dissatisfaction than a nonsurgicd group (Penman et al., 1987).

2. Studies cornparhg the impact of lumpectomy to rnastectomy (Table 2.3) Twenty-one studies have compared the psychosocial outwme including an assessrnent of body image in breast cancer patients who have undergone lumpectomy to those who have undergone mastectomy. These are presented in Table 2.3. Three overviews of the study results have also

been pubLished (Irvine et al., 1991; Kiebert et ai., 1991; Schover, 1991). The sample size of the studies ranges from 22 to 257. Response rate when stated ranges from 26 % to 100% . The studies utilized a variety of measures of body image as endpoints. Some studies used measures designed for the generd population (Mock, 1993; Sanger et al., 1981; Wolberg et al., 1989) or generai oncologic population (Ganz et ai., l992), some utilized an unbiinded i n t e ~ e w technique, some utilized a breast cancer specific questionnaire which allowed comparison of individual items but not sale total (Kemeny et al., 1988; Schain et al., 1983; Schain, d'Ange10 et al., 1994; Weilisch et al., 1989; ), and in some studies specific instruments were constnicted for use

in the particular study (brtelink et al., 1985; de Haes et al., 1985; Lasry et al., 1987). In summary, 18 of the 21 studies detected differences in body image between the two groups, and in ail 18 studies the body image of subjects who had had a mastectomy appeared to be more negatively affecteci than the body image of subjects having breast conserving therapy for

breast cancer. Women who had breast wnserving therapy felt more comfortable with their nude appearance, felt more satisfied with theu appearance, felt more attractive and more

sexually desirable than did women who had had a mastectomy. Three studies (Omne-Ponten et al., 1994; Sanger et ai., 198 1; Wolberg et ai., 1989) ilailecl to show any differences in body image

between women who had lumpectomy and those who had mastectomy. Two of these studies utilized body image measures designed for the general population, and the other included ody 19 women who had had lumpectomy.

Most of the 21 studies comparing women who had undergone mastectomy to those undergoing

lumpectorny also included a number of other psychosocial measures. Of note, the only consistent differences found behveen these groups in terms of psychosocial functioning were in the areas of body image and sexual functioning (hine et al., 1991; Kieben et al., 1991; Schover 1991).

1981

Lay.

lOvomc75 % one levei 6 items

---&

I

Homogeneity Item-Domain CaneMion

J.

Range

> 10%

Not fully Endorsed

0.70 reduced

122 Items

11 Items with Item-Item Codafion bit 0.64.7 with judgemental reduction 111 Items 12 Items Specific for Women with 2 Breasts 7 Items Specific for Women with 1 Breast

CHAPTERFIVE Descriptive Statisties for Sample Two 5.1 Introduction

To increase the number of subjects available for factor analysis and to carry out the reliability and validity phase of the study, a second sample of women was required to complete the 111 items remaining after item reduction. Women were asked to complete the revised BIBC questionnaire, shortened to 111 items, twice with a 2 week interval between administrations. As part of the validation study, the women completed seven other related questionnaires dong

with the first administration of the BIBC. These 7 questionnaires will be described further in

Chapter 9. 5.2 Methodology

Format The format of the prototype questionnaire utilized in the Item Reduction phase of the study

appeared to be adequate, well understood and easy to complete. However, comments were received about the 7 point scale. Several individuals felt 7 points to be excessive, thus the response scale was collapsed to a 5 point scale in the following manner; 1. The frequency response scale was coiiapsed from almst never / rarely / infequenfly /

somerimes / O - n/ very oftn / almst always t o almost never / Nlfiequenrly / sometimes / open / almost always. 2. The Likert response sale was coliapsed from very stronglj disogree / srrongly disagree /

disagree / neirher / agree / strongly agree / very sîrungly agree to scrongly disagree / disagree / neither / agree / strongly agree.

Scores ftom the 7 point sale u t i l h i for sample 1 may be collapsed to a 5 point scales by

cornbined the first and second category and similarly, the sixth and seventh category. Sample Sample size for the reliability study can be estimated based on the expected value of the intra-

class correlation coefficient (ICC) and the length of the desired confidence interval around the

parameter. At a value of the ICC of 0.86, to have a 95% confidence interval with a length of 0.1 (ranging from 0.81to 0.9 l), 142 patients are required (Streiner, 1994). Assuming an

80% overall response rate to the tirst administration of the questionnaire and a 95 % response

rate to the second administration , 187 women were recruited to participate in the test-retest reliability assessrnent and validity study using the following cntena; Inclusion Cnteria; Definite histological or cytological diagnosis of breast cancer Subject is aware of diagnosis Understanding of English Over the age of 18 Greater than 3 months since diagnosis No change in medical condition anticipated over a 2 week period Exclusion Criteria; Obvious psychotic or confusional state Mental inability to complete questionnaire Undergoing adjuvant chemotherapy Recruitment The subjects were selected in a consecutive fashion from medical, radiation and surgical oncology clinics al Mount Sinai Hospital, the Toronto Hospital, Women's College Hospital and P ~ c e s Margaret s Hospital. The physician involved in patient w e identified potentidy eligible subjects in clinic. After expressing interest in participating in the study, the subjects were approached by study personnel. The potential subjects underwent an intake interview at which time subject eligibility was assessed in tems of the inclusion and exclusion criteria. The study was explained in detail to eligible subjects and signed consent for participation was obtained. The subjects were asked to wmplete the package of questionnaires with standardized instructions in the home and retum it within one week to the study centre using a pre-addressed stamped envelope. Measures to maximize cooperation rate were instituted. After retum of the package of questionnaires, the retest administration of the BIBC was rnailed

to the subjects. Individuals were asked to note the date of cornpletion of both BIBC questionnaires allowing calculation of the time interval expiring between test and retest. AU identiQing material was destroyed after the questionnaires had b e n retumed.

5.3 RmIts

One hundred and eighty-seven women were recruited, 165 women retumed the fist package of questionnaires and 161 retumed the second administration of the BIBC questionnaire, for a total response rate of 86%. The average age of subjects returning questionnaires was 60, the median age 61 and the range was fiom 3 1 to 81 years of age. Thirty percent of the sarnple had undergone mastectomy and 5%had undergone a reconstructive procedure. Twenty-one percent had been administered chemotherapy in the pst. Twenty-nine percent were currently on tamoxifen. Sixty-eight percent were married and 49 % considered themselves sexually active. Seven percent stated that they had expenence local recurrence of breast cancer and 5% stated they had experienced distant recurrence of breast cancer. Sixty-one percent of the sample was native to Canada. Average time since diagnosis was 5.1 years with a median time of 3.6 years

and a range fiom 2 months to 43 years. The second administration of the questionnaire was completed a median of 15 days after completion of the fist questionnaire, with a mean of 21 days +1- 15 days and a range of 7 to 97 days for the 153 women who gave a date of completion for both administration. One hundred and eleven subjects (73%) completed the second administration between 1 and 3 weeks after the fust questionnaire was completed. Fifteen percent of women stated that they had experienced 'some change' in body image between the administrations. No one stated that they had experienced a 'big change' in body image. The results for the 2 administrations of the questionnaire are presented in Appendix 13. These

results were evaluated using the statistical ikm reduction cnteria descnbed in Chapter 4.

1. Range-:

The full range of responses was endorsed for each item in either the

fust or the second administration of the questionnaire.

: - 2. -F

Six items had one response category endorsed by over 75 % of

the respondents in both administrations of the questionnaire. As the group of subjects recniited

for this phase of the research was more homogeneous than the group of women recniited for the item reduction phase, and as the number of response categones had been reduced to 5 ftom

7, this finding was not unexpected. Thus, despite meeting this critena for item elhination, the six items were retained for further testing.

3. N o m e m:The highest nonresponse rate for an item was 7% and for the majority of the items (84%), the nonresponse rate was l a s than 2%. Thus, no item met the cntena for

item elimination based on nonresponse rate, indicating that the subjects found the items acceptable.

1 Ikm-Domain : correlations are evaluated in Chapter 7. 5.4 Summary

A sample of 187 subjects with breast cancer was recniited to participate in the reliability and

validity phase of the study. Subjects included were anticipated to be stable in terms of body image over a two week period. Eighty-eight percent of women completed the first administration of the questionnaire dong with 7 related questionnaires, and 86% completed the second administration, a mean of 2 1 days d e r the first. The items were acceptable to the

subjects, with low rates of nonresponse to items. This sample of responses will form the basis of further testing of the masure. For the purposes of cl*,

this sarnple wiiI be referd to in

the remainder of this thesis as 'Sample 2'. The 309 women retuming a completed

questionnaire described in chapter 4 will be referred to as 'Sample 1' in the remainder of the thesis.

CHAPTER SIX

Factor AnaIysis 6.1 Introduction

One hundred and eleven items remained after statistical item reduction. A factor analysis was thus undertaken in an attempt explore the underlying factor structure of the data and to possibly eliminate fuxther items. Gorsuch describes the aim of factor analysis as a method 'to s m a r i z e the inrerrelationships

among variables in a concise but accurate manner ar an aid in concepçualization' (Gorsuch, 1974,pg. 2) and the author goes on to describe 3 major uses of factor analysis as a technique;

1.

To reduce the number of variables for future research while maximuing the amount of information or variance of the onginal data set descnbed by this srnaller number of variables

2.

To explore the data for relationships mong variables leading to the development of

new consmicts and theories for testing in further research 3.

To test hypothesized constructs or theones by examining the relationships among variables.

There are two major types of factor analysis; exploratory factor analysis and confirmatory factor anaiysis. As NumalIy states 'an exploratory analysis deflnesfacors in the pure&

mathemm.cal tenns of bestjtt, typically ïnost variance accounted for, "and e v e ~ l l leah y to facors which the investigator then imeprets...In contrasr, focors are defned directly in a conflnnatory anulysis. nie intent is to have thefactors incorporute the properties t h have been hypothesized and then &termine how well thesefit the data' (Nunnally and Bernstein, 1994, pg. 450-451). Thus the first two aims of factor analysis are best attained by exploratory factor analysis while the final aim is attained by c o n h a t o r y factor analysis. Comrey describes an iterative approach to studies using factor analysis (Comrey, 1978). He recommends that such an approach starts with a 'temative conception of whar the ultimate factor stwture is apt to

be.. .The iniiial investigmahon gives an approxiinaîion tu what thefucîor smcture shouId be. On the buis of rhis conception. modifcah'ons in m'sting vakb1e.s are made, new variables are added, unù so on. and predictiom are made about whaf will huppen in the nerf snuiy as a

result of these changes. nie neu sBcdy verifis orfails to venfy these predicn'om' and this cycle continues until the factor structure of the concept is clarified (Comrey, 1978, pg. 649). Having undertaken item generation for the body image measure with a 'tentative conception' of body image (that of a 4 domain mode1 including cornpetence, appearance, comfort and predictabsty), an exploratory factor analysis of the data was undertaken to give an approximation of the underlying factor structure in order to refine the conceptwlization of body image and to guide any further elimination of items (ie variables). 6.2 Factor AnaIysis Theory

Kleinbaum (1992) describes 3 goals of an individual factor analysis: 1.

To achieve parsimony by incorporating the maximum amount of the information contahed in the original data in a minimal number of factors

2.

To obtain factors which are independent of each other

3.

To obtain a factor structure which is conceptually meaningful

Ln keeping with these goals, NunnaUy (1994) describes 2 components of an exploratory factor analysis. In the first component, the shared variance amongst the items is condensed into a smaller number of factors, commonly in a manner which atternpts to achieve parsimony and independence. The second component, the rotation of factors, attempts to facilitate the interpretation of the factors, assistîng in the development of a meaningful factor stnicture.

J. CondmSabl There are three major approaches to the condensation of the variation of the items into factors: centroid analysis; principal components analysis; and maximum likelihood analysis. Two of these approaches, principal component (PCA) and maximum likelihood analysis (MLA), are in

common use. In the principal components rnethod of analysis, factors are extracted in a manner that maximizes the amount of variance explained by the resulting factor structure. In the maximum likelihood method, the goal is to use existing data to predict the factor structure

for the population.

In both PCA and MLA, factors are defined as linear combinations of the variables of the original data ma&.

Factors are derived mathematically in a manner that optimizes some

property in the analysis, generally from a correlational matrix of the data. PCA optimizes the amount of total variance explained by a factor structure. MLA explains as much as possible of the population correlation ma&

as estimated from the sample correlation matrix. While there

are differences in the two techniques, large differences are generaily not found in the resulting factor solutions. PCA is routinely used in exploratory factor analysis as it is rnathematically simpler than MLA and it thus more easily understood and computed. Also the MLA technique does not result in a stable solution in a l l cases. In both methods, after the k s t factor is defined mathematidy, 'partialing' of the data matrix

is performed, whereby that portion of the value of a variable which can be explained by the first factor is 'partialed' or removed fkorn the variable, leaving a residual value for the variable. A correlation ma&

is then created from the residual values. This rnatrix is terrned

a partial correlation maûix. The second factor is then extracteci from the partial correlational matrix using the same mathematical technique which defuied the first factor. This process continues until aU factors have been defmcd. A matrix of factor loadings rnay then be created. Factor loadings represent the correlation

behkreen t!!e score of a given factor with the scores of each variable in the data matrix. The

square of such a factor loading represents the proportion of variance of a given variable which is explained by a given factor. The higher the value of a factor loading, the closer the relationship between the variable and the factor. In the matrix of factor loadings, generally

each of the columns represents a given factor and each of the rows represents a given variable

Factor,

Variables

var3

Factor,

Factor,

...

Factor,

The surn of the ail the squared factor loadings for a given factor (ie the sum of the squared loadings of a colurnn in the above matrix) represents the total amount of variance of the data explained by that factor. This is termed the eigenvalue for the factor. The total amount of variance in the data is equal to the number of variables. The sum of the a l l squared factor loadings for a single variable (ie the sum of the squared loadings of a row in the factor loading

rnatrix) tells the proportion of variance of a given variable explained by the factor structure on which the ma& is based. This is termed the communality for the variable. The sum of all the squared factor loadings for a given matrix gives the amount of variance explained by the

factor structure under examination. To achieve perfect explmation of variance in a data set, one would need to extract as many factors as variables. This is of course, inconsistent with the first goal of factor analysis, that is to achieve parsimony by explaining the maximum amount of the variance of a data set with the minimal number of factors.

Deciding how many factors to retain is difficult. There is no

single method for deciding on the optimal number of factors for a particular analysis. One rule of thumb is to include any factor with an eigenvalue above 1.O (Guttman, 1954). However,

as the number of variables in the original data matrix increases, the l e s va.riance a factor needs to account for to mach the cutoff for inclusion (Nunnally and Bernstein, 1994). In such cases, this method has been found to suggest too many factors and thus is not recommended by sume authors (Gorsuch, 1983; Nunaally and Bernstein, 1994). Another method of selecting the number of factors involves the use of scree plots. Scree is a geological term which refers to the rubble which accumulates at the base of a cliff and is not included in the calculation when the height of the cliff is being measured. The use of a scree plot in factor analysis alIows the differentiation of 'the cliff or the important factors (explaining substantial amounts of variance) from 'the rubble' or the unimportant factors (explaining insignifiant arnounts of variance.) In a scree plot, the eigenvalue for a factor (found on the y axis) is plotted against the ordinal number of the factor (found on the x axis). The scree plot is used to define the point where any further factors would explain markedly

less variance in the data, Nunnally terms this the 'transition point' in the scree plot function

(Nunnally and Bernstein, 1994). Using scree plots typically results in the selection of fewer factors then does the eigenvalue= 1 rule and the exciuded factors tend to be unimportant. However, because the transition point is defined visuaily, the determination of the number of factors using this method is subjective. Nunnaily states however that 'the major critenon (for determinhg the number of factors) is

that the resulting factors should be meaningFl afrer rotmion. Unfortunate&, extrachng t w

many or toofew factors cm cause problem of this type' (Nunnally and Bernstein, 1994). Nunnally recommends that a factor should have at least 4 variables which correlate above 0.5 with the factor. Nunnally also suggests that a factor should have 'some' variables which

correlate almost exclusively with that factor. Nunnally defuies this as a variable which has less than a 0.3 correlation with any other factor. 'Unless ofactor is ïzt l e m t h s~ung,it is

best to ignore it. '

2. Roeation Once the nurnber of factors to be included in the factor structure has been determined commonly the factors are rotated to aid in the interpretation of the factor structure. Rotations are linear combinations of the factors which change the factor loadings of the variable but not the communality (the proportion of variance of a given variable explained by the factor structure) of the variable. Rotation generally accomplishes 3 goals (Nunnaliy and Bernstein, 1994): 1.

Rotation should strengthen the relationship between variable and factor. Afier rotation, the relationship between a factor and the variables which belong to it will be increased while the relationship between a factor and the variables which do not belong to it will be decreased. This may aid greatly in the interpretation of a factor. If a variable loads

very highly on one factor and insignificantly on other factors, insight into the rneaning of the factor may be gained by examining the content of the highly loading variable (Anastasi, 1988).

2.

Rotation should concentrate the variance shared by highly correlated variables on a single factor. After rotation, highly correiated variables will load on a single factor versus several factors. This aiso aids in interpretation, by clustenng related variables

around a single factor. Each factor cm then be describeci in tems of what is common between the variables. 3.

Rotation should make the variances of the factors more equal in magnitude. After condensation, the amount of variation accounted for by the factors identifid tends to Vary widely, with the first factor generally accounting for far more variance than later

factors. Rotation produces factors which are more equal in value. The rotation of factors may be done orthogonally, in which case the factors are rotated to

remain at 90 degrees (or orthogonal) from each other, or obliquely, in which case the rotated factors do not remain at 90 degrees from each other. Numally suggests the use of orthogonal rotations in exploratory analysis for several reasons (Nunnally and Bernstein, 1994); 1.

Orthogonal rotations are mathematically simpler than oblique rotations.

2.

Orthogonal and oblique rotations have commonly been shown to lead to similar conclusions.

3.

The interpretation of an oblique factor structure is more difficult.

In general, orthogonal rotations are performed using a Varimax method, a technique which achieves rotation by maximizing the sum of variances of squared elements of the columns of the factor matrix (Kaiser, 1958). Once condensation and rotation are wmplete, one must decide which variables load on to which factors (or which variables have saiient loadings on a factor) in order to interpret the solution in a meaningful fashion. Gorsuch describes a salient loading as 'one which is s u n c i e l y high to assume t h a relutionrhip exisfs benveen îhe variuble and the facor. In

addition,it usualiy meam thnr the relnrionship is high enough so thar the variable can aid in inrepreting the factor and vice versa" (ûorsuch, 1974, pg. 184). One general rule for

determinhg the minimum value of a salient loading considers the standard error of the correlation coefficient which determines the factor loading. At a given sample s i x , for a given level of significance, one can determine the minimum value of a correlation coefficient that is statistically different fiom zero (Gorsuch, 1974). For example, for a sample of 100

individuals, a correlation coefficient must be at lzast 0.2 to be significantly different than zero at the p < 0.05 level of significance. Accordhg to the rule of thurnb, a factor loading is

considered salient if it is at least twice this value. In the case of the previous example, a factor loading must be at least 0.4 to be considered salient for a sample of 100 individuals. However, for large sample sizes, the standard error of a correlation coefficient decreases, meaning that smaller coefficients will become significantly different from zero. For example, for a sample size of 175, the value of a salient factor loading would be'0.3. As the sample size becomes very large and correlation coefficients become statistically significant at smaiier values, the coffesponding factor loadings deemed salient by this rule may become too smali to be interpretable. Thus, in such cases a lower limit for a salient factor loading of 0.3 has been set (Gorsuch, 1974).

The distribution of factor loadings for a variable is also important in determining the meaning of a particular factor loading. 'Ifthe variable loads on one and only one factor, rhen the inrepretarion is simplifep (Gorsuch, 1974, pg . 186). Variables which have significant factor

loadings on several factors may be uninterpretable. Therefore, a rule of thumb in interpreting factor loadings is to compare the factor loading of a variable to the comrnunality estimate for the variable. If the value of the factor loading squared is not at least half of the of the value of the communality of the variable, it is not considered significant. Thus, the factor loading under

consideration must comprise at least half the variance accounted for by the factor structure for

a given variable. Variables which do not fulN these cnteria are termed complex. The results of a factor andysis may be used to reduce the number of variables in a measure.

For a given factor analysis, one rnay consider variables which do not have significant factor loadings and variables which have complex factor loadings to inadequately fit the factor structure. These variables may thus be considered for elhination, leaving only variables with significant factor loadings on single factors.

6.3 Methodology Sample

The women with breast cancer who completed the body image questionnaire for the item reduction phase as detatied in chapter 4 (Sample 1) and the women who completed body image questionnaire for the reliability/validity phase as detailed in chapter 5 (Sample 2) formed the

sample for factor analysis. There were 111 items (or variables), 92 items common to alI subjects, 7 items specific for wornen with one breast and 12 items specific for women with two breasts. This sample resulted in a subject to variable ratio for common items of 5.1 to 1. Gorsuch suggests the ratio of 5 to 1 as the minimum number of subjects per variable as a ' d e of rhumb' (Gorsuch, 1974) and Comrey suggests that there should be at l e s t 200 subjects, in order to provide a sample large enough to 'give stable correlation coeflcienrs' (Comrey, 1978). Thus the 472 subjects available when the factor analysis was undertaken were

considered to comprise an adequate sample on which tc base an exploratory factor analysis. The subjects in the Sample 1 utilized a 7 point response scale whereas the subjects in Sample 2

utilized a 5 point response scale. The 7 point scale was thus collapsed to a 5 point scale in the fashion described in chapter 5.

Technioue A principal components factor analysis was conducted with a varimax rotation The number of

factors was selected by examination of the eigenvalues, the scree plot and by reviewing the resulting factor structures for solutions with different numbers of factors in keeping with Kleinbaum's third goal, conceptual meaningfulness. Factor loadings greater than 0.30 were considered to be salient when the squared factor loading was greater than 50% of the communality for the item. The variables which were found to have a loading meeting these critena were assigned to the factor on which the loading was salient. Variables which did not meet these critena were considered cornplex. Because there were two versions of the questionnaire (one version for women with a single breast and one version for women with two breasts ) in no case had ail items been administered to a given subject. Thus a factor analysis with a l l subjects (number of subjects = 472) and all common items (number of items =92) was undertaken initially. Two separate subgroup analyses were then undertaken, one for women who had cornpleted the single breast specific questions (nurnber of subjects = 152) and one analysis for women who had cornpleted the two breasts items (number of subjects = 309). As the sample size for the subgroup

analysis decreased substantially, it was decided conclusions about the overall factor structure would be based on results for the common items in the entire group. The subgroup analysis would allow the determination of the factor loadings of the 1 and 2 breast specific items.

6.4 Results

on A correlation matrix was calculated for the 92 common items. The matrix was inspectai to see if factoring would be worthwhile in keeping with Nunnally's recommendations (Nunnally and Bernstein 1994). A substantial number of high correlations were found between items (77

correlations above 0.5) suggesting that there was enough common variance behveen the items to make factoring worthwhile. Some groupings of sirnilar variables have high correlations with each other and low correlations with dissirnilar variables indicating the potential for

extracting factors fiom the data. For example, the items ' Ifeel serunlly atrucnve when Z am nude', ' 2feel sexuaity amactive', 'I would avoid a new sexual relan'onship because of my

body', and '1feel desirable as a woman' are conceptually related and have high item-item correlations with each other, while these items have low item-item correlations with the conceptually unrelated items ' A m pain is aproblem for me', ' n efeeling in my a m is

normal' and 'Swelling of my a m is a problem for me'. Thus there appeared to be suff~cient evidence that the data warranted a factor analysis. Co-ts

a(PCA)

The PCA was undertaken. The resulting eigenvalues are presented in table 6.1 for factors explainhg at least 1% of the variance. Eighteen factors had Eigenvalues above 1 indicating that statistically, a single factor solution would not be adequate. If the Eigenvalue greater than one criterion for selecting the appropriated number of factors were used as a method for sekcting the optimal number of facton, 18 should be chosen. When a factor structure which

included 18 factors was evaluated, only 4 factors were found to each explain more than 5 percent of the variance of the original data. This would indicate that 18 factors would be too great a number to result in a meaningful factor structure, and that the Eigenvalue greater than one rule of thumb for the selection of the number of facton was not appropriate for this analysis.

Table 6.1 Eigenvalues of Correiation MptriK for 92 Cornmon Items Expiaking Over 1%of Variance Factor #

Eigmvaiue

% Variance

1

25.22

27

2

4-81

5.2

3

3.92

4.3

The scree plot method of determining the optimal number of factors was thus evaluated. The

scree plot of the eigenvalues was generated. (Figure 6.1)

Scree Plot

%y examining the scree plot, it appears the dope of the cuve of the scree plot changes

between 1 and 2 factors and between 3 and 11 factors, with a levelling off between 8 to 11 factors. The transition point appeared to be at 6 or 7 factors.

An iterative approach to detennining the optimal factor structure and further item reduction was undertaken. The 3, 4, 5, 6 , 7 and 8 factor solutions were generated using a principal components analysis and varimax rotation and evaluated in order to determine which factor solution provided the most meaningful interpretation of the data. As will be discussed in detail, the 7 factor solution was selected as statistically optiinal for the 92 common items. The unrotated factor structure for the 7 factor solution with common items is presented in Table 6.2. The rotated factor structure for the 7 factor solution organized with variables with salient loadings assigned to the appropriate factor, or classified as complex, is presented in table 6.3.

Tnblc 6.2 Unroticed Factor Stnrturc for 7 fictor miution Continual

Ibn 1 feel cotnfort*blo about h c wny 1 look whcn 1 cxcrck 1 would feel comforbblc changing in a public chango-m 1 worry ihimy wcilht will change 1 un &fMd with (hC h p e of my body Iun~fiadwith(heipparu~xofmyhipii Iwonyabfnamyhir Bcing tIred mîcrfem witb my Lire 1 un brpW wiih my level of tncrgy 1 have problemi 1 dJalr a&ut breuc anccr 1worxyiboriihrrcharodp.Lu 1 f d them m a tima bomb h i d a of me 1 fodmy body ianbbto fghtdUaK 1 fœi h t MmcLhing u over my body 1 wony ihic u i a r ir cprcading Ifeclu8gyatmybody 1 fetl in conml of my budy 1 fœl my body b failing nput Iocod rciuunooaibwtmy bahô IfaelIuaincooimlofancer 1 fa1 pmnc LO cancer 1 f d my body hi bctn invadd 1 fbcl my body hiICI mc domi 1 fcel h&y 1 ftcl lopiidd 1 wony about my body 1 feel ugly 1 fttl dunrgad 1 fccl uhamod of my body 1 f e i disfigurai 1 fecl g d about my body 1 fecl normal 1 fecl 1- funininc rince cariccr 1 ükcmybody 1 feel oki bcforc my lime 1 fccl comforrrble king MUI in my b n 1 a n waik iround my baimom in my undcmcar 1 avoid lodring a! my rcin h m b w î rurgcry 1 M&ficd wiih the nhapc of my buaockn 1 fa1 1 will ba hcahhy in ihe future Swclling of my n m n a pmblcm for me Ann p h m a pmblan for me 1 un f o q d The feeling in my ann M normal I Mllœpy dufins lhc diy

V&e explainad by cach f r t o r % Vlrirnce cxphkied by eich factor

Factor l a d i n g i FACTOR4 FACTORS -0.30125 -0.W092 0.13992 0.13284 0.40722 0.02486 -0.26844 O. 15771 -0.37255 0.02493 0.04650 0.1951 1 -0.03790 0.27459 0.03258 -0.17934 -0,05051 0.31241 0.26844 -0.22388 0.20297 *O. 13036 0.01072 9,272rU) O. 10222 O, 18340 -0.W028 -0.20554 O. 12099 -0.17958 0.09008 9,24879 0.01251 0.13222 0.07973 -0.1 1081 O. 13955 -0.09250 0.04148 0.24671 0,08265 -0,21427 -0,04381 -0.34637 -0.03952 -0.31038 0.14135 -0.12736 O. 10575 0.05888 0.21764 -0.12080 0,10916 0.06041 0.03342 -0,14293 O. 18770 -0,07756 0.05692 -0.11621 -0.06116 0.053 19 O. 13063 -0.09365 -0.07665 -0,13116 -0.20413 0.08146 -0.07917 -0.15036 O. 15045 -0,11836 0.20853 -0.lS38l -0.23039 -0.14310 -0.40879 0.10463 0.13317 0,17479 -0,04209 0,24558 -0,00713 0.21592 -0,03370 0.25852 -0,03117 -0.10822 -0.08330 0.23534

Table 6.3

Rohted 7 Factor Solution for 92 Comma 1Factor iuadiqr

FACTOR1 FACMR2 FACTOR3 FACMR4 FACTORS

h r1 I fctl t h t puc of mc m u t namin hiddai 1 hide my body whcn chrngbg clouicr 1 woulâ avoiâ a ncw uxurl rclrrianrhip b c a w of tny body Itrylohidcmyboây 1 fœi lm feminiae ibanm IwouldLrccpmy~covdduringrcxuilin~ 1 rvoid bokiag at my wan f m b r a i t ~ e r y 1WOUc h phydcd coarici rÿcb u hugging

1avoiâ phpicd iotimr)l 1.m.frridofrouchirythewrnfmmb~~cry Ifctlu~itlrinivcwbciix.mnudc

1woukl f d wmforlablc chmghg In a publio d w g ~ r o o m

h r2 1thinlr.bwtbrcuta~cu

Iwonyoutc*nceririprrding 1fœlthiKimdbbigirrii;ioloverrnybody 1 f a l m y body hu klmtdown Ifalmybodyhubccainvdd 1 feel k m t r lime bomb inride of mc 1 wony iboui my body I necd m u r u n c c about my bcahh 1 foc1 pmnc lo M C C ~ 1 wony about minor rcba uid prinr 1 fœl dunrgcd 1 fecl uigry 8t my body Wbcn 1 much my b m t ( s ) , I foel uuriour Factor 3 1 M niiafd w i l thc ipparuicc of rny h i p 1 am ratiikd with ibc s h p e of my buuocIrr 1 rm uttficd wilb the b p c of my body 1 fccl comfomiblc iibout tbc way 1 look whcn 1 excrcuc 1 lüre my body 1 likc my loob jurt h c way thcy arc Fictor 4 1 fccl thrt pcoplc arc looking nt my chat 1 nccd 14 bc ruuurcd about the rppcamce of my buii 1 feel h t people ue h k i q nt me 1 feel people an icU my brarti ue not n o r d 1 think my brtuii iippar uneva ta oîhcn 1 wony about the wiy I look 1 wony about my b.u

Table 6.3 Continuai

Factor 5 û t b c n have had ta u k o ovcr my dutia B c i drod ~ micrfcm witb my lire My body rcop me €rom doky -8 1 wani io do I am i l q y durhq the diy 1 have pmbkriu concaltl8iry 1 un forgelful

Iunuliifiedwiîhmy.~parurcuq>ccL1ocariom 1 cin puticipiic in nomiil &itia 1 am hppy witû my level of

E.ccor 7 h pùa n a proôlun for me

Swclling of my um t a problan for me 1 am setirfiad with the appcinnce o f my a m i 1 would facl d o r t a b l c m a rlœvclar mp 'Ibe feeling in rny arm L n o r d

1 feel unuilty mariaive 1 fccl daimblc u iwomm 1 worry ibwl my .bility io takc crre of orben I womy ahut my ability io tJ;e Gare of m p l f IhwirtrwryFromodiu,wbcaibcygctcloiciomycba 1 w d feel acrvcnir r e l f - e x d d q my brcut(r) 1feeldolt8blcbddry~mybody

Iocodiobercumuatibwtmyapparuw 1 fec1gOod lboutfny rpptu*nco

-

M Y ~ W = h wry 1 bok witbout my clotha on 1 lika t MY b WY 1 hava fût ro W M m y i i p p a ~ c c1, bave w m k â ia cry 1 im uxnfortablo bciug #ai in id m ruit iainlting ibout my .ppe~ace hi hccrfcred with my coactatrmtim 1 fœl d o r t a b h io dre nudc IwonythrlIrpptuwerlt 1 rm wlf-eoniieioui ibouc my q p a r u i c c 1 WOW my ippcuUrt whai Ultding 8 ~ g C I ' l 1 avoid clothing ifui d a mc iwam of thc ihpc of my b u t 1 rvoid rbopping for clotha 1 wwld fœl comfortlble in a low cut I O ~ I would fecl coairort~bIcin tight nvatcri 1 woy b t my weight wiU change 1 fccl my body b able io f i ~ b dt k 1 fccl ia conml of my body 1 foc1 my body ir fdiimg rpan 1 r u 1 1 un m conml of cancer 1 f d halthy 1 fccl iopridbd 1 f#l ugly 1 f d a d w n a i of my body 1 f a 1 dbfigurcd 1 fetl good about my body 1 f u l old beforc my timc I fecl comfoniiblc bchg roai in my b n 1 a n wilk uound my bcdroom in my undcnvar

90

There are a number of reasons why the 7 factor solution was chosen as optimal and the 3, 4, 5 , 6 and 8 factor solutions were rejected for the 92 varïables.(Table 6.4)

Table 6.4 Cornparison of Rotated Factor Solutions for 92 Variables Number of Factors in Solution 1

7

8

45 56

48 %

49 %

67

64

55

59

25

23

22

15

16

22

6

6

5

3

4

47

42

47

52

45

50

3

4

5

Percentage of Variance accounted for

37%

4096

43 %

Nurnber of Items with Saiient Loadhgs

81

68

Number of Items with Significant Crossloadings

35

Minimum Number of Items per Factor Number of Items with Loadings above 0.50

.6

The 3, 4, 5 and 6 factor solution have a large number of variables with high cross loadings. Cross loading variables meet the criterion for having salient loadings on a single factor but have loadings greater than 0.3 on additional factors. The number of cross loading variables was 35, 25, 23 and 22 variables for the 3, 4, 5 and 6 factor solutions respectively. The high number of cross loading variables indicates that the factors in these solutions do not achieve independence, and that the structure does not best represent the underlying relationships between the variables. In comparison, the 7 factor solution had 15 variables with signifiant cross loadings. In addition, the 3 and 4 factor solutions do not produce conceptually

meaningful results. In the 8 factor solution, 2 factors, the sixth and seventh factor identified have only 4 variables with salient loadings each and in the case of the seventh factor, only one of the variables has a

loading of over 0.5. Thus the 8 factor solution fails to meet Nunnally's criteria for a rneaningful factor that there be at least 4 variables with high loadings per rneaningful factor. Thus, the 8 factor solution was rejected. As solutions with greater than 8 factors would be

expected to have even smaller number of variables per factor, these solutions were not considered further. There were marked similarities behveen the 5, 6, and 7 factor solutions. In all three solutions, variables loaded ont0 5 factors similarly; variables sampling content related to vulnerabiüty of the body to disease or invasion of the body loaded ont0 a factor (termed the vulnerability factor); variables sampling feelings of stigma about the body loaded ont0 a factor (termed the body stigrna factor); variables samphg the assessment of the capacity of the body loaded ont0

a factor (termed the limitations factor); variables sampling feelings about the general appearance of the body loaded ont0 a factor (termed the body concems factor); and variables

sampling feelings of concem about cancer related appeamnce and obviousness of cancer loaded ont0 a factor (termed the transparency factor). In the 6 factor solution, an additional factor with variables sampling arm concerns was produced (temed the arm concerns factor). In the

7 factor solution, an arm concerns factor was generated and in addition, a seventh factor was

formed with the items ' Ifeel Z will be healthy in thefurure', '1feel normal', and '1feel attractive when clothed' (terrned the seventh factor).

The 5 and 6 factor solutions did not achieve statistical independence, as both solutions had high numbers of cross loading items. However, when compared to the 5 factor solution, the 6

factor solution is superior conceptuaily. The emergence of an arm concerns factor, the major difference between the 5 and 6 factor solutions, is consistent with the unique problems and symptoms of women with breast cancer. An axillary lymph node dissection is a standard component of the surgical management of breast cancer patients and is associated with substantial morbidity (ie. a m swelling, pain and paraesthesia). Twenty-one unique items pertaining to arm concerns were generated in the item generation component of this research,

indicating that arm concems are important to women with breast cancer. Conceptually, one would expect arm concems to be more related to each other than to other concems, and thus a separate arm concerns factor is sensible, and hence the 6 factor solution is preferable to the 5 factor solution.

In the 7 factor solution there is a substantial reduction in the number of variables with salient loadings which have high cross loadings onto an additional factor. Thus the 7 factor

solution is statistidy the preferred solution. However, concepW y it is difficult to hypothesize a unifying concept for the seventh factor, which has only 3 items, 'I feel 2 will be

healthy in thefuntre', ' Ifeel wmial',and ' I feel ~ttractivewhen clothed'. The 7 factor solution was therefore not conceptuauy an optimal solution. To deterrnine the optimal factor structure for the data, a second analysis was performed with the aim of achieving a conceptuaUy meaningfûl factor solution in a stepwise fashion. As the

factor analysis is exploratory in nature, this iterative approach is appropnate. Factor analysis

WU form the bais of fûrther item reduction. Items which do not load saliently on the factor solution selected as optimal will be eliminated. In the case of the 7 factor solution, fbrther analyses would be based on the 55 items which loaded saliently in the original solution. Thus, the effect of including only the 55 items loading saliently in the 7 factor solution (versus the

original 92) in an additional factor analysis was investigated. A 7 factor solution for the 55

variables onIy (versus the original 92 variables) was generated using the methodology previously described. In the resulting 7 factor solution for the 55 variables, the seventh factor

now consisted of oniy one variable 'Z avoid Iooking ar my scars J%ITZ brem surgery '. This indicates that for the 55 variables, the 7 factor solution was not an optimal solution statistically or conceptually. The 6 factor solution for the 55 variables was then generated. The unrotated factor structure for the 6 factor solution with common items is presented in table 6.5. The roiated factor structure organized with variables with salient loadings assigned to the appropriate factor, with complex variables presented separately for the 6 factor solution is presented in table 6.6.

Tabk 6.5 Unmtmtd hcror Stmeturc for 6 Fador Solution for 55 Commoa It-

FACMRl FACMRZ F A C M i U FACTOR4 FACTORS FACTOR6

. 3

a9

8~ d

-1

4:s

-4 - a a e

-+j

E

$4

g

8

g., Q s J j # f

ialnllg:

e i i p a a dg&

di;.: PgBPr $ 1i11911 g ~ 3 3

Factor Lvndiqi FACïûRl FACTOR2 FACTOR3 FACTOR4 FACTORS F A O R 6 ha

Facior 6 (Ann C-rnr) A m pain ir r problan for me Swof my nam b r problan for me Iam&fieâ wilthc.ppeu*ncoofmyum ïbfccüry i n m y u m n a o d

Muhihaoriiil Whaa 1 much my brcut(8). 1 fecl inrioui 1 fecl atlnctive wben cknbad 1 would fetl comfodle m 8 i l t v c l a i top I feel d u n q d 1 fœl 1 will be h d h y in the future

In the 6 factor solution for the 55 items the factors correspond to those generated in the original 6 (92 variable) factor solution however the number of cross loadings in the solution has been substantially reduced. Only 4 variables with a salient loading have cross loadings above 0.3, and none have cross loadings above 0.35. This indicates that the factors are statistically independent. AU factors have at least 4 variables with loadings above 0.5 in keeping with Nunnally's criteria for a meaningful factor. Conceptually, with the exclusion of the seventh factor, this solution is more sensible than the original 7 factor solution. Thus, for the 55 items, the 6 factor solution was selected as the optimal factor solution. The correlation

ma& for the 50 common variables having salient loadings in the final 6 factor solution is presented in table 6.7. The efiect of the addition of the one breast specific items and two breast specific items to the 6 factor solution was evaluated by repeating the factor analysis with the specific items. Included in the analysis were the 50 common items which had salient loadings in the previous 6 factor

solution (ie. items which were found to have complex Ioadings were eliminated from this analysis) and specific items in the single breast subgroup (total number of items = 57) as weU in the two breast subgroup (total number of items = 62). These analyses are found in Appendices 14 and 15 respectively. The addition of the specific items resulted in Little change to the overall factor structure. Four of the seven 1-breast specific items loaded ont0 one of the 6 factors in a salient fashion in the mastectomy subgroup analysis. Eleven of the twelve 2breast specific items loaded onto one of the 6 factors in a salient in the lumpectorny subgroup

analysis. Therefore, a total of 65 items were found to have salient loadings in the 6 factor solution, 50 common items, 4 one-breast specific items and 11 two-breast specific items. The remaining 46 items were factorially complex and rnay be eliminated.

6.5 Discussion

An exploratory factor analysis of the data from the Body Image after Breast Cancer

Questionnaire was undertaken to approximate the true factor structure in order to hirther refine the conceptualkation of body image in the breast cancer population. The factor analysis was also undertaken to guide any further elimination of items. -The analysis was performed in

keeping with Kleinbaum's goals of factor analysis; parsimony; independence; and conceptual

meaningfuhess. A 6 factor solution was selected as best achieving these goals. The factors identifiai by the factor analysis were a body stigma factor; a vulnerability factor; a limitations factor; a body concems factor; a transparency factor; and an arm concerns factor. Thirteen items (eleven common items and two 2-breast specific items) load ont0 the vulnerability factor. Items loading on to the vulnerability factor sample feelings of vulnerability of the body to illriess and cancer, as well as feelings of invasion of the body and a loss of trust in the body as a healthy and functioning organism. All items load pusitively on the factor. Endorsement of the seventeen items would indicate more feelings of vulnerability. Twenty-two items (twelve common items, seven 2-breast specific items and three 1-breast

specific items) load onto the body stigma factor. Twelve items load positively onto the factor. These items sample feelings of a need to keep the body hidden and avoiding physical intimacy. The endorsement of these items indicates a higher level of body stigma. Ten items load

negatively onto the factor. These items sarnple feelings of cornfort with nude appearance and sexual attractiveness. Endorsement of these items would indicate lower levels of body stigma.

Ten items load ont0 the Limitations factor, 5 positively and 5 negatively. The positively loading factors indicate a increased sense of limitation ( 0 t h have had to take over my dwies, My

b d y stops mefrorn doing things I wunt to do) while those loading negatively indicate a lesser sense of limitation (1feel nonnul, Z can pamDcipatein nomial acn'vities.) Six items load ont0 the body concems factor, all positively, with endorsement of the items

indicating a more positive view of general body appearance (1 m sans-ed with the shape of MY

body, I like my Iooks just the way they are.)

Ten items load onto the transparency factor, seven common items and one 2-breast and one 1breast specific items, ali loading positively ont0 the factor. Endorsement of the items indicates

a greater sense of concem about the obviousness of cancer related changes to appearance or the ability of others to detect such changes ( for example, 2feel fhar people are lookîng af my

chest, Z wony about my hair.) The final factor, the arm factor consists of five items, four cornmon items, and one 2-breast

specific item. Three items load positively on the factor and two negatively. Endorsement of positively loading items would indicate a greater degree of arm syrnptoms and concerns whereas endorsement of negatively loading items would indicate a lesser degree. These items are unified by their focus on the a m , in terms of concems and symptoms. The 2-breast specific item w hich loads ont0 this factor is 'My breart is painful

IO

touch'. Arm symptoms

tend to occur more frequently soon afkr surgery at which time breast pain is also more likely to occur. This may explain the association of these items. Forty-six items were identified as factorially cornplex. As these items did not seem to fit with the proposed 6 factor conceptuakation of body image in the breast cancer population, they were selected for elimination. On review of the items to be eiirninated in this fashion, none appears to sample a unique concept. Because of this, the effect of the elimination of the 46 items in this manner on the vaüdity of the measure should be minimal. The factor structure identified in this analysis is conceptuaily meaningful. Women interviewed in the item generation phase of this research spoke at length about the factors identified, particularly about the body stigma and vulnerability factors. However, despite the apparent conceptual rneaning of the factor structure identified, the content-related validity of the underlying factors has not been estabfished. In the subsequent validity chapter, an initial investigation of the validity of this measure is presented, with a focus on the validity of each factor as a sale, discussing the content and construct validity of the scales. A number of authors have performed exploratory factor analyses on body image measures (Ben-Tovim et al., 1991; Brown et al., 1990; Champion et al., 1982; Franzoi et ai., 1984; Mahoney et al., 1976; Mendelson et al., 1995; Reed et ai., 1991; Rosen J et al., 1991; Rychlanan et al., 1982; Slade et

al., 1990; Theodorakis et al., 1991; Tucker 1981). AU analyses have found body image to be a

multifactorial constmct, with between 2 and 7 factors. However, the factors identified by these analyses are not consistent and to a large degree reflect the original formulation of questions

for the measures. For example, Brown et al. (1990) perfonned a factor analysis on the MultiDimensional Body-Self Relations Questionnaire (Cash 1994), a widely utilized and validated body image measure developed for the general population. Items for the questionnaire were

generated to fit a constnict of body image which proposai 3 attitudinal dimensions (affect, cognition, and behaviour) related to three domains (appearance, fitness, and health/illness). Thus a 9 factor structure was hypothesized. A principal components factor analysis with varirnax rotation was performed with 2000 subjects cornplethg the questionnaire. In this analysis, a 7 factor solution was found to be optimal, as the cognition and behaviour dimensions were found to merge into a single attitudinal dimension. Thus 7 factors were identified; an appearance evaluation factor (affective response to physical appearance); an appearance orientation factor (cognitive and behavioural response to appearance); a fitness evaluation factor (affkctive response to physical effectiveness), a fitness orientation factor ; a

health evaluation factor (affective response to physical integrity), a health orientation factor; and an unpredicted factor, illness orientation (the cognitive and behavioural response to illness). Thus the factor structure was largely in keeping with the underlying conshuct of body image which guided item generation for the measure.

In cornparison, items for a body image measure created for use in the eating disorder population, the Ben-Tovim Walker Body Attitudes Questionnaire (Ben-Tovim et al., l982), were generated through interviews (number unspecified) with a range of female subjects including women with eating disorders. Two hundred and fifteen potential items were identified in this manner. Three hundred and twenty eight subjects completed the questionnaire and a principal components factor analysis with varimax rotation was undertaken. Five factors were found to be useful in this analysis: feelings of fatness; feelings of disgust with the body; self-assessed

physical strength and fitness; the importance of weight and shape in the person's life; and selfperceived physical attractiveness. The number of itenis in the questionnaire was reduced to 48

based on the results of the factor analysis. A second factor analysis was undertaken on a new sample of subjects. Six interpretable factors emerged, with a new lower-body fat factor idegtified.

Differences in the items included in the various questionnaires, (ie. the differences in the content sampled by the questionnaires) may have resulted in the differences between the factor structures identified by the factor analyses of these measures. Also, differences in the samples

on which factor analysis was bas& may have resulted in differences in factor structure. Anastasi States 'factors are not staric enîities bur are themelves the product of the individual's of erperiences Vary cwnulative erperiential history. Znrofur as the interrelananottships

long

individuak and groups, d i j f e n t factor patterns rnay be expected umong h m ' (Anastasi, 1988, pg. 390). Thus, it is not surprishg that the factor analysis of the Ben-Tovim Walker Body

Attitudes Questionnaire, developed for use in the eating disordered population including many questions about weight, shape and feelings about fatness, derived fiom a sample which included a high proportion of women with eating disorders would propose a different factor structure for body image than a body image measure developed and tested on the general population, or a body image m a u r e developed and tested on the breast cancer population. In keeping with Comrey's recommendation for the use of factor analysis (Comrey 1978)'

the initial factor analytic investigation derives 'an u p p r o x i ~ i o 'nof the true constmct. Modification of the measure must be undertaken in keeping with this approximation and the resulting measure then submitted to further testing. Thus an exploratory factor analysis should not be thought of as conclusive, but rather as hypothesis generating. Factor analyses of the Body Image after Breast Cancer Questionnaire should be repeated independently by other

investigators in different populations of women with breast cancer in order to support or refute the validity of a 6 factor model. To move fiom hypothesis generating to hypothesis testing, a different factor analytic approach, confirmatory factor analysis, should be undertaken. In confumatory factor analysis, one starts with an underlying theory about a constnict such as body image, and hypothesk the underlying factor structure a priori. The confumatory factor

analysis assesses, in an hypothesis testing fashion, if the data sufficiently fits the theory based

on tests of statistical significance. Thus, testing the 6 factor model of body image using confirmatory factor analysis would be an essential future step in the exploration of the underlying constnict of body image.

CHAYrER SEVEN Reiiabiiity The BLBC is a self-adrninistered instrument, without altemate foms. Thus intemal consistency

and test-retest reliability formed the bais of the assessment of reliability for this measure. 7.1 Theory

Co-

This form of reliability is based on the domain-sampling model of test theory (NunnaIiy, 1994). Items included in any measure *ui be considered a randorn sample of alI possible items which

evaluate a particu1a.r attribute. Because the sample of items is Iirnited in any measure, the observed score will always differ from the true score by an amount of error due to item selection. For example, a measure of depression consisting of 20 items would include only a smaU number of aIl the possible items which could comprise such an assessment. If scores on the items included in the measure were generalized as a measure of the depression construct,

variation in the observed score on the depression scale will be due to a combination of true differences in depression as weU as differences due to the limited sarnpling of alI possible items measunng aIl possible aspects of depression. Errors due to the 'sampling' of the variety of situations that accompany test administration (ie. errors due to skipping items, fatigue, misreading of questions, etc.) are also generalized in the domain-sampling model. One may conceptualize measurement error variance as the error inherent in the random samphg of items from any domain as weii as the random sampling of alI possible situational factors. Measures of intemal wnsistency estimate reliabüity based on the average correlation among items within a measure (Anastasi, 1988; Cronbach, 1990; Nunnally, 1994). In a measure of a unidimensional constmct, because aU the items should be measuring the same thing, the average correlation between the items should be high, ie. items in such a rneasure should 'hang together' . If the average correlation between items is not high, some items in the measure must be sampling extranaus phenomena. The most comrnonly used masure of intemal consistency is coefficient alpha. Coefficient alpha ranges from O to 1, with O indicating no homogeneity and 1.0 indicating perfect homogeneity. Its value depends both on the average

correlation of items and the number of items. (Thus a common technique for improving the reliability of a s a l e is to increase the number of items.) Intemal consistency is expected in instruments that masure single domains (ie homogeneous instruments). For heterogeneous instruments, a high interna1 consistency for the entire measure would not be expected. Rather, one would expect the domains sampled in the measure ( Le. subscales which measure single

aspects of a construct) to have high intemal consistency. est Re-

- ..

Test-retest reliability assesses the stability of a measure on N o or more separate occasions. Differences in scores between the test and retest correspond to random fluctuations in

responses over time and thus are an estimate of the arnount of variation in the observed score which is due to random error. Differences in testing conditions may also affect the consistency of measurement and hence the test-retest reliability. Test-retest reliability does not depend on the assumption of homogeneity. However, in some cases the experience in the first testing rnay influence the responses to the second, leading to a change in the correlation between test and retest. Also, if real change in a characteristic has occurred between test and retest, this variation wiU be incorrectly considered error. Therefore, the time interval between test and retest must be selected to minirnize the possibility of real change occumng. The assessrnent of test-retest reliability or stability is important for measures which have been designed for use in clinical trials. The ability of an instrument to measure individuais in a steady state with consistency influences the ability of a measure to detect change- Sensitivity to change is an essential property of measures to be used in clinical trials. Test-retest stability has

an influence on calculated sarnple size for such studies. Thus for measures which may be used in clinical trials, or for another purpose where the detection of change is important, test-retest stabüity must be assessed. The intraclass correlation coefficient is the most appropriate statistic to determine the degree of concordance between the test and retest and varies from -1.0 to 1.O, with -1.O indicating perfect discordance and 1.O indicating perfect concordance.

It is essential that s d e s of the BIBC have minimal measurement error due to item sarnpling. It is also important that in the test has adequate stability in stable populations. Thus, it

is important that both interna1 consistency and test-retest stability be assessed for the BIBC For research purposes, a measure should achieve a reliability of at Ieast 0.70, w hile for use with individuais, a reliability level of 0.90 is recommended in groups which are stable in terms

of the quality king measufed (Nuaaally iuxi Bernstein, 1994). 7.2 Methodoiogy As descnbed previously, Sarnple 2 (wnsisting of 165 women with breast cancer) was recruited to çomplete the BIBC questionnaire twice, with a 2 week interval between administrations.

The interval of 2 weeks for readminisüation was selected as little real change in body image is expected in stable subjects over a two week period. Thus, a change in score over this p e n d of time is unlikely to represent an important clinical change. It is also unWrely after an

interval of two weeks that individuai questionnaire items would be recalled (Streiner and Norman, 1989). Only women who were expected to be stable in tems of body image over a 2

week period were selected. The inclusion of women who had a high possibility of change in body image over a 2 week period (such as women undergoing adjuvant chemotherapy) would have led to an underestirnate of the stability of the measure in the steady state. Women were

asked if they felt they had experienced a change in body image over the 2 week peiiod. Coefficient alpha was calculaîed f a each scale-(forcommon items only) in the entire sample. Coefficient alpha was also calculated for each d e s including 1 breast or 2 breast specific

items in the appropriate subgroup. The following formula was use&

Test-retest stability was estimated for each sale and the entire test for common items in ail

women, and in the subgroups of women with 1 breast and women with 2 breasts for d e s when appropriate. As the study confonned to a one-way random effects model, the ICC was c a l c u k d h m the standard ANOVA table using the following formula (Fleiss, 1986):

BMS - WMS

R = BMS - (16-1)WMS

Where BMS = Mean square between subjects WMS = Mean square within subjects k, = Number of replicate readings Ninety-five percent, one-sided confidence intervals were calculated using the formula described by Fleiss (1986). A secondary analysis was planned to estimate the ICC in the subgroup of women who did not

report a change in body image over the interval between the test and retest. 7.3 Results

Sample 2 formed the buis for reliability testing. As described in Chapter 5, 165 women completed the first administration and 161 women completed both administrations of the BIBC questionnaire. Itemdomain correlations, as described in Chapter 4, were calculated for each item within its scale (the 6 d e s corresponding to the 6 factors derived fkom factor analysis). Using the previous critena for item reduction, items which had less than a 0.40 correlation with the total score for a sale were identified for elimination. Twelve items in total had item-

sale correlations less than 0.40; 5 common items; 5 two breast specific items; and 2 one breast specific items. These items are listed in Table 7.1 and were excluded from further

analysis.

Table 7-1 Items with Item-Total Correlation less than 0.40

-

Fador 1 Vuinerabîlity 1 two breast specific item I am always mare of my bremts (L)

-

Factur 2 Body Stigma 1 Common item I would moid a new scxzurl relan'mh@ becaure of my body

3 two breast specific items I feel cotgtiortable looking at my brurrt (L) I am sarirjUd wirh the appearrmcc of my breast (LJ I wortldfeel comfortcrbh wirh a s d parmer touciiling my breasr (ZI

2 one breast specific items I feel cornfortab& wlien others sec my martec~omy(M" I feel t h my marrecromy is rcgty (M)

-

Factor 3 Limitations 2 common items I am sorisfied w3h my appeamncr at specicll occc~siom

I am forge@

-

Fador 4 Body Concerus No items

-

Fador 5 Transparency 2 common items I worry abw my hair I worry abw the way I look 1 two breast specifk item A di&rencc between rny n i p p k can be sem in clorhing (Z)

Factor 6 - Arm Concerns

No items

The remaining 53 items, assigned to the appropriate scale, are listed in Table 7.2. Coefficient

alpha was calculated for each scale. This analysis was repeated for the subgroup of women

with 1 breast (n =44) and the subgroup with 2 breasts (n = 120). The average scores for each

sale for the fist and second administration of the s a l e dong with the values for coefficient alpha are presented in table 7.3.

Table 7.2 Items Remaining for Rel'mbility Asscssment Vuinerability (Factor 1)

1worry that cancer is sprwding I thidc about bteasr cancer 1 feel rbere is a tirnt bomb h i d e of m 1 feel ihat roaic~hingis taking over my body 1 fee1 my body has becn invaded 1 nced rcassurance about my heaith 1 feel my body has let me down 1 woy about my body I w o y about minor aches and pains 1 ftel prone to cancer I fecl angry at my body I think about my brcast (L) Body Stigma (Factor 2) 1 feel that pari of me m u t cemain hidden 1 bide my body whcn changing clolhes 1 try ta hide rny body 1 would keep my chest covered during semai intimacy 1 avoid physical intirnacy 1 feel less feminine since cancer 1 avoid looking at my scars h m b m sugcry 1 avoid close physical contact such as hugging 1 am afraid of touching the scars fmm brcast surgery 1 feel scxually attractive when 1 am nude 1 would feel cornfortable changing in a public change-roam 1 feel cornforcable Iooking at my miutectomy OuL) 1 feel comfomble when othem iee my brcast (L) 1 am happy with the position of my nipple (L) 1 am natitificd with the sizt of my bteast (L) The appeamnce of my b w s t could disturb orhem (L)

Limitations (Factor 3) k i n g tired intrrferes with my life My body stops me from doing things 1waat [O do Ohers have had to takc ovcr my dutics 1 am sleepy during the &y I have problems conccntniting 1c m participate in nomial activities 1am happy wiih my level of cacrgy 1 am forgetfiil Body Concems (Factor 4)

I am satisfied with rhe appearance of my hips 1am iratisfied with ihe shape of my buütxlcr f am satisfied with Ihe shape of my body 1 feel c o r n f o ~ b l cabout Ihe way 1 Iook whcn 1 exercisc 1 Iikc my body 1 like my Iooks just the way thty arc Tmnsparrncy (Factor 5) 1 frel that people a n looking at my chest 1 fcel that people are looking at me 1 fcel people can tell my bmasts arc Dot nonnal 1 aced to be ~ s s u r e about d the appeuance oFmy bua 1 think my brrarts appear uacven to othcn 1worry about my prosthesis or padding slipping (M)

Arm pain is a ptoblem for me Swcliing of my .rm is a pcoblem for me

1 am sstisfied with ihe appearance of my arm n i e fetihg in my a m is nomial My breast is painfil to couch (L)

Table 7.3 Reliability Results

#of Items

Mean Score

Standard Deviation

Coefficient Alpha

test retest test retest test retest

11

27 26.5 29 29 27 27

8.13 8.84 8.45 9.18 8.70

0.86 0.90 0.87 0.84 0.83

9.34

0,91

test retest test retest test retest

11

26 23

8-47 7.67

0,85

15

30

9.34 8.52

12

30 31 30

test rçtest

8

16

test retest

6

Scale

Mcdian Score

Range

ICC

Vulnerability

Al1 Subjects 2 Breast Group 1 Breast Group

12 II

0.84 0,83 0.87

Body Stigma

Ail Subjects 2 Breast Group 1 Breast Group

0,85

0.83 0.85

0.81

8.90 8.87

0.83 0.91 0.84

0.92

6.02 5.94

0,84 0.84

0.85

16.5

19.5 20

4.59 4.76

0.84

0.77

Limitations

Al1 Subjects

Body Concerns All Subjects

0.86

ICC 95%Iowa Confidence Limit

Table 7.3 Reliability Results Continued

Transparency

AI1 Subjects 2 Breast Group 1 Breast Group

test retcst test retesl test retest

5 5

6

Arm Concerna

ALI Subjects

test retest

4

2 Breast Group

test retest test retest

5

I Breast Group

Total Score

test retest

4

45

NIA

NIA

For common items, the coefficient alpha ranged from 0.78 for the Transparency and the Arm Concerns scaies to a high of 0.86 for the Vulnerability sale in the first administration of the questionnaire. The values for the coefficient alpha were the same or higher for the second adrninistration of the questionnaire in all but one case. In subgroup analysis, the coefficient alphas for common plus specific items ranged from 0.68 for the 2 breast specific version of the Transparency s a l e to 0.93 for the 1 breast version of the same scale. Of note, women in the 2 breast subgroup seemed to respond to the items in the Transparency s a l e in a more uniform fashion than women in the 1 breast subgroup (standard deviation 2.33 versus 4.33, range 13 versus 18 for the fust administration respectively). Greater homogeneity of the 2 breast group

may explain the lower reliability of the s a l e in these women. In the second administration of the questionnaire, the coefficient alpha is higher (O.73), likely resulting from a higher

variability as evidenced by a higher standard deviation in the retest administration (sd =3.13). The ICC for each scale is also presented in Table 7.3. The ICC for common scale items ranges from a low of 0.77 for the Body Concems scde to a high of 0.85 for the Body Stigrna and Limitations scales. For the subgroups of women, including specific items, the ICC ranged from a low of 0.75 for the Transparency sale in the 2 breast subgroup to a high of 0.92 for the Body Stigma s a l e in the 1 breast subgroup. The ICC for all common items was 0.88. For those women who indicated they had experienced no change in body image over the test-retest

period, the ICC was 0.90 while for those who indicated that they had expenenced sorne change, the ICC was 0.75. 7.4 Summary

The reliability of the BIBC questionnaire has been assessed in a sample of 165 women with breast cancer. Good reliability, as measured by intemal consistency and test-retest stability has

been found for the six scales defuied by factor analysis, indicating that the BIBC masure has adequate reliability for use as a research tool. The scales currently do not have adequate reliability for use in decision making for individual patients, although in some cases (the Vulnerability scale and the scale total) reliability approaches these values.

CHAPTER EIGHT

Sensibility Introduction As stated in chapter 2.3, an instrument is valid if it truly measures what it purports to measure. When there is a gold standard of measurement, results from a new instrument may be compareci to the gold standard. As there is no gold standard for the measurement of body

image in the breast cancer population, other means must be used to establish the validity of the

BIBC. Thus the validity of the measure will be evaluated in this thesis by examination of sensibility (including content validity) and constnict validity. 8.1 Theory

Feinstein (1987) divides the assessment of validity into two broad categones; the quantitative assessment of validity, for example construct, cntenon and predictive vaüdity; and the qualitative assessment of validity, for example content and face validity. Feinstein terms the qualitative assessment of validity, sensibility. According to Feinstein , sensibility is compnsed of 5 aspects; purpose and framework; overt format; face validity; content validity; and ease of usage. For purposes of the evaluation of measures, Bombardier (1994) has modified Feinstein's model, and outlines 4 topics which should be examined in detemining a measure's sensibility: purpose, population and setting; face validity; content validity; and feasibility. 1. Eurpase, P o -

S e m

For a measure to be valid, it is essential that it is clearly specified for which populations and settings the measure has b e n designed to be used. The concept or attribute which the measure purports to assess should be well defined. A clear statement of these very basic charactenstics is essential prior to any M e r validation.

..

2. &ce V a i u Bombardier defines 3 essential components of face validity;

a. Items should be suitably phrased and should appear to be appropnate to those complethg the measure. This component attempts to evaluate the widely accepted definition of face validity, descnbed by Numaiiy (1994) as 'the extent to which the test taker or sorneone else

(urualljtsomeone who is not trained ro lookfor fonnal evidence of validiry)feels the imn~nent masures what it is inteptded to m u r e . ' ui addition, items should also be phrased in a

manner which is inoffensive. b. The response categories for the items should be appropriate. In addition to the evaluation of

individual items, Bombardier extends face validity to include the evaluation of the suitability of the response categones used in a measure.

c. The aggregation of global ratings should be appropriate. If an aggregate rating is used, the suitability of the method of aggregation must be evaluated. 3. -nt

..

Valrdity

Bombardier describes 3 essentid components to the evaluation of the content validity of a measure. a. The relevant aomains and exclusions should be appropnate. The measure should examine a representative sample of all areas of importance of the constmct to be measured, and there should be no important omissions. Similarly, there should be no inappropriate inclusions of

areas which are not salient to the measure. constmct. This is in keeping with the generally accepteci definition of content validation (Streiner and Norman, 1995). What women with breast cancer actuaiiy consider to be appropriate for the measurement of body image in terms of included domains and exclusions has not been reported in the literature. However, as detailed in chapter 2.2, three authors have proposai content which should be evaluated in a measure of body image populations such as wornen with breast cancer. The content of the BIBC measure

may thus be compared to the content recommended by these authors in a form of content validation. Hopwood (1993) identifies 7 areas essential to explore in the evaluation of body image in the oncologic population; dissatisfaction with appearance (dressed); loss of femininity/masculinity; reluctance/avoidance to look at self naked; feeling less attractive/sexually attractive; adverse effect of treatrnentlloss of body integrity; self-consciousness about appearance; and dissatisfaction with scar/prosthesis. Thus, a body image m a u r e should sample content in these seven areas. Vamos (1993) conceptualùed body image in the chronidy ill as being composai of a 4-domain substnicture as outlined in chapter 2.2; comfort; cornpetence;

appearance; and predictability, and thus a body image measure should sample these 4 areas. Harris (1982) conceptuallled the impact of abnomai appearance on self-concept. He identified a common 6 sequential part pattem of coping with abnormal appearance; induction and development of selfconsciousness; defence mechanisrns; unavoidable distressing activities; downgrading of selfconcept; difficulties with interpersonal relationships; and r a t i o n h t i o n of

disability and associated distress. Although this conceptualization incorporates the impact of abnormal appearance on self-concept, a broader constmct than body image, some components of the sequential pattem should be sampled in the body image measure. b. The domain structure of the measure should be suitable. The individual domains specified

to comprise a measure should be m u W y exclusive, ie. each domain should sample unique content. This may be assessed both qualitatively, using judgement, as well as quantitatively, using a factor analytic technique.

c. The method of selecting items for inclusion in the measure should be appropriate. Depending on the purpose of the measure, some method of selecting items may be more prone to the exclusion of relevant domains than others. For exarnple, a measure of quality of life to b e used in a specific patient population may fail to sample d l important aspects if the opinions

of affectai patients are not solicited at some point dunng item generation. Thus the marner in which items were generated should be considered when assessing the sensibility of a measure.

4. &a&&,! Five components are descnbed for the evaluation of the feasibility of a measure.

a. The items, scaling and scoring of the measure should be easy to comprehend. If items are not easily understood by subjects, responses may be subject to error based on the misinterpretation of items. The scaling of items should also be easily understood by subjects,

again avoiding error based on interpretation of the outcorne scale. The s c o ~ of g the measure should be easily understood and performed by those administering the measure. b. The measure should be simple to use, with standardized procedures and instructions for use.

The failure to ensure a standardized, simple procedure for the use of a measure and data collection rnay lead to error based on misinterpretation of instructions on the part of subjects completing the measure, or administrators wllecting and analysing the data.

c. Completion of the measure must be acceptable to subjects. If subjects will not complete a measure, it will not be a usefd measure. d. The format of administration should be simple. The need for special tests or skills should be assessed*

e. The administration time should be acceptable. A long administration time may affect the acceptability of a measure to subjects and thus may limit the use of the measure, especially in

instances when several masures are administered together. Feinstein notes that although sensibility is essential, sensibility alone does not determine the validity of a measure. If an instrument is found to be sensible, then quantitative methods of validation are necessary to detennine if the instrument is measuring what it purports to measure. 8.2 Methodology

Feinstein does not describe formal techniques for the appraisal of sensibility. Rather, he recommends the use of the guidelines as a form of 'checklist' for appraising sensibility using a judgemental approach. Two approaches to the assessrnent of sensibility were taken for this thesis. The first approach consisted of a thorough judgemental review of the 53 item body image measure. The judgemental approach also included a review of the negative comments about the questionnaire received from individuals cornplethg the questionnaire in the item reduction and reliability sarnples. In the second approach, 20 women with breast cancer were recruited from breast cancer clinics at the Mount Sinai Hospital to complete the questionnaire and to comment on the questions, instructions and time required to complete the questionnaire (Appendix 16). Only some of the aspects of sensibility could be assessed by the patient sample; face validity; content validity; and feasibility.

8.3 Results n-1.

. .

of h S a m &

Of 23 women approached to complete the questionnaire, 20 agreed to participate in the study. Of those who did not participate, 2 individuals could not read English, and 1 individual did not wish to participate. The 20 women who received the questionnaire a l l completed and

retumed the comment sheet. The average age of the sample was 57 with a median age of 60. Four women who answered the questionnaire spoke English as a second language. Most women (84%)had at least finished high schwl although only 32% had attended university. 2. Eurpose. P o -

Se-

Judgemenrai Review The BIBC has currently been developed as a descriptive measure of the concept 'body image'

and will be evaluated for its descriptive properties as part of this thesis. The BIBC rnay in the future serve other purposes; for example, the measure may be used in an evaluative role for c h i c a l trials (ie. the ability of the BIBC to detect change when change has occurred); however the validity of using the BIBC for such purposes must be evaluated pnor to such use. The population to which the BIBC may be applied is well described. The masure has been

developed for use in women with breast cancer who are able to read English. The instrument is to be used at any time after diagnosis and surgical treatment (the setting).

..

3. Eace Validity

-

a. Appropriateness of items Judgementul Review

Techniques of item elimination utilized in this thesis were applied in an attempt to maximize the acceptabiiity and face validity of the resulting instrument. In the initial component of item reduction (chapter 4)' items were eliminated or modified based on review by women with breast cancer. Unsuitable items were identified, and modified or eliminated (number of items eliminated = 158). In the statistical cornponent of item reduction (chapter 4), poorly phrasai items (ambiguous, offensive or not felt to be applicable to those completing the questionnaire) were identified as those having a nonresponse rate of greater than 10%. Nineteen items were identified in this fashion and eliminated. In the sample of 360 wornen recruited for the item reduction study, 19% commenteci negatively on the questions or topics included in the questionnaire, for example one individual stated ' I am retuming your questionnaire

u m e r e d because it is inapossibly inapplicabie' and another stated 'some of the questions

were a w e r e d n/a since they were not relative to rny situation'. In the sarnple of 187 women recniited in the reliability sample who completed approximately half the number of items as

the previous sarnple, the number of negative comments on the questions or topics included in the questionnaire decreased in number to 8%, indicating that the shortened questionnaire was

found to be more acceptable to the women completing it.

Patient Sample Women recruited to comment on the questionnaire were asked if they felt the questionnaire really measured body image. Nineteen of the 20 women responded affînnatively (ie. that they felt the questionnaire had face validity). The individual who responded negatively was undergoing chernotherapy. This individual commentai on the lack of chemotherapy specific items in the measure 'ï7t.e complemenring questionnoire of effects of rudiaion / chemofherapy etc. is very relevant and shuuId be adminisrered together, pam*cular&in the case of Zmpectomy'. The exclusion of chemotherapy-specific items may explain the lack of face validity of the BIBC to this woman. No respondent found an item offensive. Two respondents

found some items unclear; in one case the item '1 avoid physical Nlnmacy ' was found to be unclear by a 70 year old woman. Another women found 2 items unclear. Unfortunately she did not indicate which items. In summary, items appear to be appropnate, having face validity for the majority of women.

-

b. Appropriateness of response categories Judgemental Review

In the item reduction sarnple, severai women commented about having difficulties with a 7-

point scale, and recommended fewer alternatives. For example, one individual stated 'a sevenpoinr scale seems unnecessary. Ifind rhat peoplefind iî easier to respond fo ajive-poinî scale' and another individual felt that 'very strongly agree or disagree seems too intense. I wou[d

drop the MY'. Thus, for the reliability sarnple, the number of response categones was reduced to 5. Only one individual commented on the questionnaire format in this sample, 'The questionnuireformat is not my favourire - if is limiting in the response. I ofren don 't like any of the caregories. I have a m e r e d to the best of my &il@ - given the format'. This seems to be a

general comment about questio~airesversus a specific comment about the number of responses in the questionnaire.

Patient Sample

AU the respondents found the response scales easy to use. However, despite finding the d e s easy to use, one individual noted that at times she would have preferred a not appropriate / not

applicable response - '1 um mt NI a sexual reZmionrhip, so some of these questions weren 'c

relevant' .

-

c. Appropriateness of global rating Judgemental Review

N o global rating score is proposed for the measure. Rather, each scale of the measure will have a score based on the sum of the scores on the individual items, with positive items reversai scaled. Based on the results of the factor analysis (chapter 6) this method of aggregating the scores of the items would seem sensible. The validity of the scdes using such

a scoring technique will be evaluated in chapter 9, constmct validation.

..

4. C o r n V&&y

-

a) Appropriate inclusions and exclusions Judgementd Review

The domain structure of the body image questionnaire was compared to the content

recomrnended by three authors Vamos, Hopwood and Harris for important omissions or inappropnate inclusions (see table 8.1). Al1 content considered to be important by Varnos, Hopwood or Harris was represented by at least one scale of items from the body image questionnaire, with the exception of Harris' 'Rationalization' domain. Of note, Harris States that unlike the other 5 aspects of his construct, not al1 individuals report rationalizing abnormal appearance. Because the rationalization domain has limited applicability, it is appropriate that this content is not sampled in the BIBC.

The limitations factor sampled in the body image questionnaire was identified as important content only in the Varnos' constnict of body image in the chronically iii. As items for the

BIBC were generated from a sernistnictured interview based in part on Vamos' conrtruct, the presence of the limitations factor is not surprising. The validity of this scale will be examined in chapter 9.

119 Table 8.1 Cornparison of Domninc Sampled by the B

1

E to those Recommeded by Vamos, Hopwooci and Harris

Body Image in Breast Cancer Questionnaire Scale VuinerabiIity

Body Stigma

Limitations

Body

Transparency

Concerns

A m Concems

X

Cornpetence

X

X

X

X

X X

X

X

X

X

X X Naked Self

X

X X

integrity

Self-

X

X

X

X

X

X

Consciousness Scar 1

Prosthesis SelfConsciousness

Defense Mechanisms

X

Unavoidable Activities

X

X

X Interpersonai Relations

X

X

X

X

l

X

Paîient Sample The 20 respondents were asked if they felt that 'here were topics not covered in the questionnaire which should have been covered. Six respondents felt this was the case; 2 respondents felt that chemotherapy specific topics should be inciuded; 1 respondent felt a bathing suit question should be included; 1 respoadent felt that age should be an included topic; 1 individual felt 'whether the cancer chunged body Nnage' should be an included topic;

and 1 individual felt that the reaction of the spouse should be an included topic. Of note, items incorporating bathing suits, age, spousal reaction and chemotherapy had been generated but were eliminated in the item reduction phase for a number of reasons. Most importantly, many of the items were applicable to subsets of the target population (for example, women in marriage-like relationships) and had lirnited applicability to others.

Only 1 individual felt that there were topics covered in the questionnaire which should not have been covered. The two items which the respondent indicated were inappropriate @ut not offensive) were ' I avoid physical imtLtlmacy' and ' I would keep my chest covered during sexual

inhacy'. This respondent was 71 years of age.

-

b. Suitabüity of domain struchire Judgementaf Review

The factor structure was determineci by a Principal Components Analytic technique with orthogonal rotation and thus statistically the factors are independent.

-

c. Appropriateness of method of item selection Judgemental Review The main method of item generation for the masure was patient focused, through interviews with women with breast cancer, spouses of women with breast cancer and individuals with expertise in caring for women with breast cancer (chapter 3). The involvement of women with

the disease during item generation reduced the potential for missing important domains of

content. Subjects interviewed for item generation were not limited to the 4 domain construct of

body image proposed by Varnos which was utilized in item generation. Rather, any items that the individual felt was important in terms of body image after breast cancer was included for

further testing, again lessening the potential for missing important concepts.

5-.

. ..

-

a. Comprehension of items, scaling and scoring Judgemental Review An analysis of the readability of the 53 remaining items was performed. There are an average

of 7.5 words per item. Four different methods of estimating the level of reading difficulty, the

Flesch Reading Ease, The Flesch-Kincade Grade Level, The Coleman-Liau Grade Level and the Bormuth Grade Level (Microsot? 1993), placed the reading difficulty of the 53 items

between a grade 4 and a grade 7 levei, indicating that the items could be understood by most women. The scaling of the measure is simple. Six scores are formed by the summation of the

items belonging to the 6 d e s . The average item score per s a l e is calculateci by dividing the s a l e score by the number of items in the scale. This ailows comparisons across scales to be made. Reverse scoring is required for 17 items. Patient S m p l e

AU respondents found the instructions clear, and nineteen of twenty respondents found the instructions helpful.

-

b. Simplicity of use Judgemenrol Review

The instructions for the questionnaire are standardized and the form that is used by the patient serves as the data collection sheet further easing usage.

c. Acceptability of completion - Jdgemental Review The measure is acceptable to subjects. The item reduction study had an 85% response rate and the reliability study had an 89% response rate, indicating that the questionnaire was acceptable

to women and that they were willing to omplete the BIBC in high numbers. Patient S m p l e Completion of the questio~aireappared to be acceptable to the women; al1 respondents stated that they would complete the BIBC again if asked. d. Simplicity of Administration - Judgemental Review

The measure requires no speciai tests or skiils to adrninister or complete.

-

e. Administration Time P&*ent Sample Time for cornpletion of the BIBC ranged h m 4 to 25 minutes, with a median of 7.5 minutes. Eighty-five percent of the sample completed the questionnaire within 10 minutes. Only 1

respondent felt that the questionnaire was too long. She had taken 10 minutes for completion. The remaining 19 respondents felt that the length of the BIBC was 'about righr'. 8.4 Discussion

After a preliminary evduation of the BIBC by patients and a judgemental review, the rneasure

appears to have adquate 'sensibility' and appears to be acceptable to women with breast

cancer. Thus proceeding with quantitative methods for the validation of the BIBC i s warranted. The construct validation of the masure will be assessed in Chapter 9. Several issues were identified by the patient sample. Women undergoing active therapy do not feel that all aspects of body image are measured in the current form of the BIBC. This is not surprising, as rnost items specific for women undergoing active therapy were eliminated in the item reduction phase. These items were eliminated as they were not relevant to the majority of women with breast cancer. Thus, the creation of an active therapy form of the BIBC specific for wornen undergoing active therapy must be considered. The items in the questionnaire which sample the sexual cornponent of body image appear to pose difficulties for some women, especially women who are older and women who are not sexuaIly active. Although these items are phrased in such a fashion that all women, sexually active or not, would be able

to give a response, the items appear to be problematic to some individuals. The inclusion of these items, which assess sexual stigma after breast cancer diagnosis and treatment, is however, an extremely important component of body image to a large number of women and thus the items cannot be eliminated. The feasibility of including a not applicable response for these items could be considered in future research, especiaily in older samples of women.

CHAPTERNINE Coustruct Validation 9.1 Introduction

To demonstrate constnict validity, hypotheses regarding how a m a u r e shodd behave are generated and are then supported or refiited through research. Validity is supported if the fmdings of research support the proposed hypothesis. Constmct validation is a gradual process and requires the testing of multiple hypotheses by numerous independent researchen. Thus, only preliminary validation of the BIBC may be undertaken as part of this thesis. The preliminary validation will examine the convergent I divergent ( N u n ~ i i yand Bernstein, 1994)

..

and discriminant validity of the BIBC. m v e r m t v a w is demonstrated when results of the

measure under investigation correlate with measures of related concepts. D i v e m t va-

..

is

demonstrated when the results of the measure under investigation correlate poorly with

. . .

..

measures of unrelated concepts. Discriminant- (as defined for this thesis) is demonstrated when groups expected to differ in terms of the construct being measured differ in score on the measure being tested. This is often investigated by the use of extreme groups. To demonstrate this type of constnict validity one must hypothesize groups which should differ

in terms of the characteristic under study (for example, the body image of women after lumpectomy as compared to those after mastectomy). The demonstration of the hypothesizeû difference between the groups using the measure under study supports the construct validity of the measure. Nunnaliy describes both divergent and discriminant validity as dif5erent aspects

of divergent validity. For clarity, in this thesis, these 2 aspects of 'divergent' validity as described by Nunnaily wiii be investigated separately using the terms divergent and discriminant as described. The validation of the BIBC described in this thesis rnust be thought of as preliminarily only.

Many further studies, in different populations of women with breast cancer will be required to establish the validity of the BIBC. 9.2 Methods

To assess the convergent validity of the BIBC, correlations in scores of the BIBC and measures of related constructs were investigated. To assess the divergent validity of the BIBC, 1

L

correlations in the score of the BIBC and a measure of social desirability were investigated. To assess the discriminant validity of the BIBC, the ability of the BIBC to distinguish between

women with cancer and women without cancer, as well as women who had had mastectomy

and those who had had lumpectomy or mastectomy with reconstruction was investigated. 9.2.1 Selection of Samples for Validity Study For the validity study, 2 samples of women were used; a sample of women with breast cancer

(Sample 2); and a sample of women without breast cancer recruited for the study. Selection of the Sample of Women with Breast Cancer As descnbed in chapter 5, Sample 2 was recruited fiom breast cancer clinics for the reliability and validity study. For the validity aspect, the subjects completed the BIBC questionnaire plus

7 other questionnaires, the Multi-Dimensional Body Self Relations Questionnaire, the Beck Depression Inventory, the Rosenberg Self Esteem Inventory, the Marlowe-Crown Social Desirability Questionnaire, the Derogatis InteMew for Sexual Functioning, the Impact of Event Scale, and the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire. The format and psychornetric properties of each questionnaire will be briefly described. The questionnaires were randornly ordered. In addition, women answered demographic information including age, sexual activity and date of diagnosis.

Selection of the Sample of Control Groups A control group was recruited including women attending farnily practice clinics and women

without breast cancer who had had a benign breast biopsy in the past.

Control Group One hundred and sixteen patients over the age 30 were recruited fiom the clinics of 2 breast surgeons and family practitioners. One halfof the sarnpie had no history of breast disease and one

half had a history of a benign biopsy for breast disease. The women were selected using the following entry criteria;

Inclusion Criteria: No previous diagnosis of breast cancer

No obvious psychotic or confusional state

Understanding of English

Mental abiiity to complete questionnaire

For the sample, women were selected in a consecutive fahion after signing informeci consent. Because not all items were applicable, women in the control groups did not complete all53 items of the BIBC.The items and Scala completed by the contml groups are listed in Appendix 18. The control group was recmited prior to the decision to reduce the response

s a l e from a 7 point scale to a 5 point scale. Thus women.in the control group completed a version of the BIBC with a 7 point response scale. As previously descnbed in chapter 5, the 7 point s a l e was collapsed to a 5 point sale to allow comparisons to be made to the group of

women with breast cancer recruited for the reliability / vaLidity study. 9.2.2 Convergent and Divergent Validity

The seven questionnaires which were administered to the wornen in Sample 2 in addition to the

BIBC are briefiy described; Measures Admhktered 1. m e MuIfi-DUnensiod Body SeCf Relations Questionnoire (MBSRQ):

The MBSRQ is a measure of the self-attitudinal aspects of the body-image construct developed for the general population (Cash,1994). The measure consists of 69 items producing 7 factors (Appearance Evaluation, Appearance Orientation, Fitness Evaluation, Fitness Orientation,

Health Evaluation, Health Orientation and Illness Orientation) as well as 3 scales (Body-Areas Satisfaction, Overweight Preoccupation and Self-Classified Weight.) The MBSRQ has been found to be a reliable instrument with intemal consistency ranging from 0.73 for the BodyAreas Satisfaction Scale to 0.9 for the Fitness Orientation Scale and 1 month test-retpct

stability ranging fkom 0.74 for the Body Areas Satisfaction Scale to 0.94 for the Fitness Orientation Scale. Constnict validation of the measure has included the evaluation of convergent and discriminant validity as well as sensitivity to change ( A b b e and Thompson, 1990; Butters and Cash, 1987; Cash and Green, 1986; Noles et al., 1985). The MBSRQ has not been used in

the cancer population. Of the muItidimensional body image measures available, the MBSRQ is regarded as the ' m s t comprehensive and widely validared...Therefore. ir wouid appear

reusonable to Niclude it in any comprehenrive body-image ussessrnent barrery' (Thompson et al. 1990, pg. 43).

Although the BIBC and the MBSRQ have been created to rneasure body image in different populations ushg different constnicts of body image, both measures have been created to measure the attitudinal component of a multidimensional constmct of body image. Thus, clinically and statistically signifiant correlations would be expected between the two

measures. Some scales of the 2 measures assess very similar components of body image, for exarnple the Body Areas Satisfaction s a l e of the MBSRQ and the Body Concems scale of the

BIBC,both of which sample satisfaction with body parts. Responses to these scales would thus be expected to be highly related. Some of the scales of the 2 measures however, sample less

clearly related components of body image, for example, the Fitness Orientation scak of the

MBSRQ and the Transparency scale of the BIBC. Thus although clinically and statistically significant correlations could be predicted between the BLBC and the MBSRQ, such correlations would not be expected to occur between al1 scales of the measures.

2. The Beck Depression Iaventoty 0: The BDI is a measure of the severity of depression in adolescents and adults (Beck and Steer, 1993). It consists of 21 symptoms and attitudes assessing the various aspects of depression.

First developed in 1961, the BDI has been widely utilized and has well-established reliability and validity. The BDI has been found to be reliable with an internal consistency of 0.81 for nonpsychiatric samples (Beck et al., 1988). Test-retest stability in nonpsychiatnc patients has been found to range h-om 0.6 to 0.9 (Beck et al., 1988). There is a large body of evidence supporting the validity of the measure, including its use in oncology patients (Ekck et al., 1988; Cavanaugh et al., 1983; Heim and Oei, 1993; Elsen et al., 1995; Plumb and Holland, 1977; Steer et al.,

1986; Steer et al., 1987). Available research suggests that depressed individuals have a more

negative body image than nondepressed individuals (Pruzinsky, 1990). These findings appear to be consistent in studies with a variety of measures of depression and body image. Thus

statistically and clinically significant correlations would be expected between the BIBC and the BDI, although this may be tme for only some of the scales of the B E C

3. The Rosenberg Sew-Esteem Inventov (RSE) This is a self-report measure of self-esteem consisting of 10 items. The measure has been in use for over 20 years (Crandail, 1973; Rosenberg, 1979). Good reliability has been found in oncologic populations for this measure with an interna1 consistency mging from 0.76 to 0.87

and a test-retest reliability of 0.74 (CU~OW and Somerfield, 1991). The measure has been found to have convergent and discriminant validity and has been widely used as an outcome measure in research in the medical literature for a broad range of medical conditions. Kigh correlations

have been found between body image and self-image, between body cathexis (i.e. the degree of feeling of satisfaction or dissatisfaction with the various parts or processes of the body) and

self cathexis, and body satisfaction and self-esteem by multiple authors using a variety of measures including the Rosenberg Self-Esteem scale (Freedman, 1990; Lemer et al., 1973; Orr et al., 1989; Rosen and Ross, 1968; Secord and lourard, 1953). This is consistent with the view of

body image as an integral component of self-concept (van der Velde, 1985). Thus statisticaily and c l i n i d y signifiant correlations should be found between the BIBC and the RSE, although these may not be found for ali scales of the BIBC. 4. The Marlo we-Crown Socid DesimbiIity Seule (MCSD)

This scale was developed to determine the degree to which subjects need to respond to questions in culturally sanctioned or socially desirable ways (Crowne and Marlow , L96O). An intemal consistency of 0.88 and a one month test-retest correlation of 0.89 has b e n found for

the s a l e (Crowne and Marlow, 1960). Consûuct validation has included the evaluation of the convergent, divergent and discriminant properties of the measure (Crowne and MarIow, 1960; Duquette et al., 1994; Lane et al., 1990; Toner et al., 1992). Streiner and Norman (1989, pg.56)

recommend that social desirability 'should be rninimized whenever possible. and the person 's

propensity to respond in this mariner should be arsessed whenever if may affecî how he or she

amers.' The authors also recommend the use of the MCSD in the assessrnent of social desirability, 'unless there are compelling reasonr fo use anorher inder' as the MCSD is the most widely used measure for this purpose. A minimal effect of social desirability on the responses to the BIBC is important to the validity of the results of the BIBC. Thus, although

the relationship between social desirability and body image has not been elucidated in the literature, no statistically or clinically significant correlations should be found between scales of the BIBC and the MCSD. 5. The Demgatis Intetview for Sexual Functioning @ISI;1

The DISF was designed to provide an estimate of the quality of sexual functioning (Derogatis, 1995). It consists of 25 items which form 5 subscales (Sexual CognitionlFantasy, Sexual

Arousal, Sexual Behaviour and Experience, Orgasm, and Sexual Drive and Relationship). The

internal consistency of the subscales range from 0.7 to 0.8 and the test-retest reliability ranges

from 0.8 to 0.9 (Derogatis, 1995). The DISF has been found to have discriminant validity in one group of patients ( Z m i c h et al., 1990a; Zinreich et al., 1990b). Further evidence of the validity of this measure is limited because of the recent development of the measure. However, the DISF was developed as a short form of a widely used and validated measure of sexual functioning, the Derogatis Sexual Functioning Inventory (Derogads and Melisarantos, 1979). The longer questionnaire consists of 256 items and has a long history of use with established reliability and validity. However, the length of the 256 item questionnaire prohibited the use of the questionnaire in this study; the more established validity of the longer version was not felt to compensate adequately for the added respondent burden of the longer questionnaire. A relationship exists between body experience, body image and sexual functioning. Among

physically traumatized persons, body-image variables influence sexual experience. Positive relationships have been found between the quality and quantity of sexual experience and degree of body satisfaction. It is hypothesized by a number of authors that if individuals are uncomfortable with their bodily appearance then it would be difficult to enjoy sexual contact or

be comfortable with sexual expression (Hagen and Cash, 1990; MacCorquodale and DeLamater, 1979; Pnuinsky, 1990; W i u t h , 1987). Thus clinically and statistically significant correlations

would be expected between the DISF and the BIBC,although this may be true for only some subscales.

129

6. TIte Impact of Event S d e (TES): This s a l e was developed to m a u r e the subjective distress for a life event. It consists of 15 items which form 2 s u b d e s ; Intrusion; and Avoidance (Horowitz and al., 1979; ZiIberg et al., 1982). Internal consistency of the scales range from 0.78 to 0.92 and test-retest reliabîiity

ranges from 0.79 to 0.87 (Horowitz and al., 1979; Zilberg et al., 1982). A number of studies have demonstrated the convergent and discriminant validity of the measure (Baider et ai., 1992; Deahl et ai., 1994; Horowitz and ai., 1979; Kent and KWet, 1992; Zilberg et al., 1982). The measure has

also been found to be sensitive to change (Baider et al. 1994; Kosten et al. 1991). Although the

relationship between intrusion and avoidance and body image has not been evaluated in the literature, there are significant similarities in the content sampled by the IES and the BIBC, specifically, between the Vuherability sale of the BIBC and the Intrusion scde of the ES. Thus a statistidy and clinically significant relationship would be anticipated for some scales

of the BIBC and the scores on the scales of the IES based on an examination of item content alone. 7. The Eumpean OrganuaaOn for Researeh and Treatment of Càncer Cure Qua& of Lÿe

Questionnaire (EORTC QL@30) This sale was developed as a cancer specific quality of life questionnaire (Aaronson et ai., 1991). It has 30 items which form 5 fùnctional scales (Physical, Role, Cognitive, Emotional,

and Social), a global quality of life sale, 3 symptom scaies (Fatigue, Pain, and Nausea and Vorniting), 5 single-item symptom measures and one financial impact question (EORTC,1995). Two factors have been identified for the psychosocial variables of the BIBC;the Emotional Distress factor (consisting of the Emotional and Cognitive functioning scales) and the

Functional Ability factor (consisting of the Role, Social and Global functioning scales) (personal communicatinn,SA Mchchlan). Internal consistency for the functioning scales has been

found to range from 0.58 to 0.94 (Aaroason et al., 1991; Osoba et al., 1994). Construct validation of the scales has consisted of the evaluation of the discriminant validity as weli as the convergent validity of the measure (Aaronson et al., 1991; Osoba et al., 1994). The measure has

also been found to be sensitive to change (Osoba et ai., 1994). The EORTC QLQ-30has gained wide acceptance as a measure of quaüty of life in oncologic populations. It has been accepted

as the standard quality of iife measure by the National Cancer Institute of Canada.

The relationship between body image and quality of life is unclear. Of note, although the majority of investigations comparing women who have had lumpectomy to women who have had mastectomy for breast cancer have demonstrated a difference in body image between the two groups, the two groups have not been found to differ in tems of quality of life or other

aspects of psychological functioning ( h i n e et al., 1991; Kieben et al., 1991). Although this may provide some evidence that body image and 'quality of life' as measured in these studies are not related, the actual relationship between body image and quality of life has not been studied

in detail. Thus it was not possible to base a relationship between the scales of the BIBC to scales of the EORTC QLQ-30on a review of the literature.

Generation of Hypotheses for the Convergent and Divergent Validity Study

For depression, self-esteem, social desirability, body image and sexuality, adequate evidence was available in the literature to predict statistically and clinically signifiant relationships with body image. These relationships, however, were predicted on the bais of prior research conducted with preexisting measures of body image. Preexisting measures did not utilize the 6 domain construct of body image which forms the bais for the BIBC scales. Thus although significant correlations between results of these measures with the BIBC could be predicted based on information available in the literature, the pattern of correlations between the scales of the BIBC and depression, self-esteem, body image (as measured by the MBSRQ), and sexuality could not be elucidated directly. Expert opinion was thus solicited to interpret the literature in terms of predicting the patterns of correlations between the BDI, the RSE, and the MBSRQ and the scales of the BIBC. The relationship between subjective distress, and quality of life and body image could not be determined by the literature. Thus expert opinion was solicited to predict the relationships between the BIBC and quality of life as measured by the EORTC QLQ-30, and subjective distress as measured by the Intrusion and Avoidance scdes of the ES. Correlations between the scales of the BBC and the BDI, RSE, MBSRQ, EORTC

QLQ-30,DISF,IES were predicted a priori by 3 individuals with expertise in quality of life measurement and experience with the breast cancer populations VJG, SAM, NNB). The scales of the BIBC were predicted to have no relationship to the Marlowe-Crown Social Desirability

sale as a precondition of the validity of the BIBC scales.

The 3 expert judges reviewed the available literature descnbing the relationship of body image to self-esteem, quality of iife, depression, social desirability, subjective distress and sexuality. The judges, based on their interpretation of the literature, predicted the pattern of correlations

between the scaies of the BIBC and the various measures. An ordinal sale was used by the j udges to quantify the dinical significance of the hypothesized correlations; weakhegligible;

moderate; and strong. The ordinal sale was based on recommendations by Bumand et al (1990) after modifications. Burnand recommends that correlation coefficients under 0.32 be

considered weak or negligible correlations and correlation coefficients above 0.60 be considered strong correlations. Burnard suggests an additionai cut point at 0.45 which demarcates 'subsramiaIly signifcanr ' correlations, however, this cut point is suggested

'tenrorivety' only. Therefore, for hypothesized correlations in this thesis, the judges defined a

weak or negligible correlation as under 0.32, a moderate correlation as between 0.32 and 0.6, and a strong correlation as greater than 0.60. The judges recorded their hypothesized correlations between the BIBC scales and the measures

independently using structured data collection forms. The independently generated hypotheses were reviewed, and discrepancies were resolved through a consensus process. A hypothesized correlational matrix was constnicted between the 6 scales of the BIBC and the 7 other measures based on the opinion of the experts. In general, correlations between similar constnicts were expected to be strong (for example, the Body Areas Satisfaction s a l e of the MBSRQ and the Body Concems scde of the BIBC),correlations between related yet distinct

constmcts were expected to be moderate (the Physical Functioning s d e of the EORTC QLQ-

30 and the Limitations scale of the BIBC), and correlations between unrelated constmcts were expected to be weakhegligible (the MCSD and all scales of the BIBC).

The actuai correlations between measures were evaluated using the Spearman's rank correlation coefficient. The statistical significance of a correlation coefficient is detennined by the magnitude of the coefficient and the sample size. For correlations involving the full sample available in the validity study, (n= 165), a correlation coefficient above 0.154 would be statisticdy significant at the 0.05 level, a correlation coefficient above 0.202 would be significant at the 0.01 level, and a coefficient of 0.256 would be significant at the 0.001 level

of statistical significance (Sachs ,1982). Because of the relatively large sample size, very modest correlations would be highly statistically signifiant. Thus the use of an index of clinical significance for the correlation coefficient is supported. The analysis was first conducted with common items (items completed by both women with 1 breast and women with 2 breasts). The analysis was then repeated including the items specific for women in these groups. 9.2.3 Diseriminant Vaiidity

Cornparison of subjects with 1 breast to subjects with 2 breasts

Hypotheses: As described in chapter 2.5, there is general consensus in the literature that signifiant differences in body image exist between groups of women who have a lumpectomy and women who have a mastectomy after treatment for breast cancer. Thus, a minimum

criteria for the constmct validity of the BIBC is that the masure is capable of discriminating between these two groups. However, as described in chapter 2.5, previous researchers have utilized ad hoc measures of body image, or measures sampling specific components of body image only. Thus, although it is clear that scores on some scales of the BIBC should differ significantly between women with 1 and 2 breasts, it is unclear if all scales should differ. For example, it is not clear that women with different forms of surgical treatment for breast cancer should differ in terms of functional ability. Thus a difference in score on the Limitations s a l e may not be expected. Therefore, to test the ability of the BIBC to discnminate between women

with lumpectomy and women with mastectomy, 3 independent judges hypothesized a priori which scales should differ signifimtly in score between the hvo groups, after reviewing the available literature.

Comprison Groups: The subjects who completed the BIBC in the validity sample were separated into those with 1breast (individuah who had had mastectomy without breast reconstruction) and those with 2 breasts (individuals who had had lumpectomy or mastectomy with breast reconstruction). Demographic information was calculated for the 2 groups and

cumpared using the student's t test for continuous variables and Chi square test for categorical variables. The scores on the scales of the BIBC were calculated and the statistical significance of the difference in score between the groups was determined using the Wilcoxon rank sum test (a nonparametric method of comparing the means of 2 groups) and the student's t test (a

133

parametric method of comparing the means of 2 groups) for each scale. These findings were compared to the hypothesized differences. Because of the potential confounding by demographic differences between the two groups if present, when demographic differences were found, a second analysis was undertaken to control for the influence of the potential confounder. An analysis of variance was conducted to d e t e d e the impact of the confounders on the relationship between the factor scores and group membership. This was performed in a parametric and a non-parametric (Kruskal-Wallis k-sample test) fashion. There are advantages and disadvantages to the use of parametric or non-parametric methods for the analysis of the validity study data. Pararnetric methods are powerful, widely used and

understood methods of analysing data. However, the use of parametric rnethods requires the assumption that the data conforms to an underlying distribution. For example, the use of the student's t test for the cornparison of the means of two groups assumes that the underlying sampling distribution of the means for both groups conforms to the student's t distribution. The use of parametnc methods for ordinal data assumes the existence of a uniform relationship between categones, Le. the difference between 1 and 2 in a five-point Likert scale is assumed to be the same as the difference between 4 and 5. Conversely, non-parametric methods do not require underlying assumptions about the distribution of the data or the uniformity of the relationship between categories. However, as non-parametric methods are generally based on

ranks (versus values), information is lost and these methods are therefore less powefil than parametric methods. When applicable, both parametric and non-parametric statistics will be used in the validity study. Conctusions should be supportai by both methods. There are a

number of reasons why both methods are presented; 1.

The underlying distributions of scores for the scales of the BIBC are unknown.

2.

While the use of an ordinal scaie would support the application of non-parametrïc statistics, pararnetnc statistics have b e n widely applied and considered acceptable for such data.

3.

The limited number of subjects included in the validity study may limit the usefulness

of non-parametric statistics.

4.

The vaiidity study is preliminary in nature. While convergence of the results of pararnerric and non-parametric methods would support hypotheses, substantial differences between the conclusions of the methods would require further research and might lead to recommendations regarding the choice of statistics for analyses of the

BIBC in the future. Cornparison of Cancer Group to Control Groups

The 3 independent judges generated a priori hypotheses about the ability of the BIBC scales to differentiate between respondents with breast cancer and respondents frorn the control group (women who had had a benign breast biopsy and wornen fiam family practice chic). Demographic information was calculated for the groups and compared as descnbed previously. The mean scores on the BIBC scales as administered to the control and cancer groups were

then calculated. The statistical difference between the scores for each s a l e was determined using the Wilcoxon rank sum test and the students t test. A second analysis controlling for the

potential confounding by demographic variables which differed behveen the two groups was performed in a parametric and non-parametric fahion. The analyses were first conducted with items completed by all women. The analyses were then repeated including items specific to women with 2 breasts. As this validation study is preliminary only, no adjustment for multiple cornparisons was made. Thus a p value of l e s than 0.05 is supportive of the discriminant validity of the BIBC. Defuiitive proof of the discriminant validity of the BLBC wiil require further research studies producing consistent results. 9.3 Results 9.3.1 Convergent and Divergent Validity

Demographics of Sample

The dernographics of the sarnple of women with breast cancer recruited for the reliability I validity study has been described in chapter 5. The distributions of scale scores for the scales

used in the validity study are descnbed in Table 9.1 and fquency histograms for the measures are presented in Appendix 19. Most scores have skewed distribution with signifiant deviation for the normal distribution with the exception of the MCSD. In the majority of cases

the distributions are skewed, with a higher concentration of responses in the range of scores

indicating better function or adjustment. A positivel;/ skewed distribution is found for the d e s of the BIBC (with the exception of the Body Concms scale), the RSE, the BDI and the

IES Avoidance and Intrusion d e s . For these scales, a higher score indicates worse function or adjustment. A negatively skewed distribution was found for the scales of the MBSRQ (with the exception of the Appearance Orientation and Body Satisfaction s d e s ) and the EORTC QLQ-30 scaies. For these scales, a lower score indicated worse bnction or adjustment. The scales of the DISF

are positively skewed; however in this case, the concentration of scores in the lower range of the scales corresponds to a lower quality of sexual functioning.

The distribution for the Transparency scale of the BIBC demonstrates a ceiling e f b t in this

sample, with the majonty of responses concentrated at the most positive extreme of the possible range of scores. The distribution of 4 of the 5 subscales of the DISF demonstrate a flwr e k t in this sample with the majority of responses being concentrated at the most negative extreme of the possible range of scores. Because of the deviations from normality found in the distribution of scores, it is important

that nonparametric statistid methods are used in addition to parametric methods in the statistical analyses.

Table 9.1 Distribution of Scores for Scaies used in the VaIidity Stridy Max

I I Higher

%O=

Skcwncss

I

Kumsis

Dev. h m

Normal

Yes

Yes

Yes Yes Dcrogatis

Interview for

Yes

Scxual

Yes

Functioning Drive

1 124 1 9.8 1 10 1 5.6 Yes

~ppcarancc

162

3.3

3.3

0.7

163

3.8

3.8

0.6

Yes

Evaluation Appcarance

Orientation

Yes MU&

Yes

Dimensional MY

Self

Yes

Relations Questionnaire

Yes

Yes

Table 9.1 (continuedl Distribution of Scores for Scales used in the Valiciity ShrrJy Min

Max

Skewness

Kunosis

Role Eutopean

Orgabtion

For Rescarch

Emofiod COgNtivc

a d Trcatment of Caocer

Social

Quality of Lifc

~ l ~ b ~ l

Questionnaire worse

Disvtss Factor Ability Factor

I Body Stiglry worse Body Concerna

Questionnaire 1

worse -

-

9.3.1 Hypothesized Correlations The hypothesized correlation r n a h produced by the 3 judges is presented in Table 9.2. Consensus amongst the judges was easily achieved when discrepancies had occurred. In

summary, consistent with the available literature, moderate correlations were expected between scales of the BIBC (Vulnerability, Body Stigma, Body Concems and Transparency) and the RSE, scales of the BIBC (Vulnerability, Body Stigma and Limitations) and the BDI, and the Vulnerability s a l e of the BIBC and the Dnve/Relationship scale of the DISF. Strong and moderate correlations were predicted between the scales of the BIBC (with the exception of Arm Concems) and the d e s of the MBSRQ (with the exception of the Fitness Evaluation and

Fitness Orientation scaies). No signifiant correlations were predicted between any scales of the BIBC and the MCSD scale. No predictions could be made by the judges about correlations

between the Overweight Preoccupation and Self-Classified Weight scales of the MBSRQ and

the scales of the BIBC (with the exception of the Body Concerns scale) and the scales of the

DISF (with the exception of the Sexuai Drive / Relationship scale) and the scales of the BIBC. For the IES and EORTC QLQ-30,with limited evidence available in the literature, the judges predicted that strong and moderate correlations would be found between the BIBC s d e s

(Vulnerability , Body Stigma, and Transparency) and the Intrusion scale of the ES, and moderate correlations wouid be found for the BIBC sale (Body Stigma) and the Avoidance scale of the IES.For the EORTC QLQ-30questionnaire, moderate correlations were predicted between scales of the BIBC (Vulnerability and Limitations) and scaies of the EORTC QLQ-30 (Physical Function, Emotional Function and Global Heaith Status) and the Emotional Distress

and Functional Ability factors of the questionnaire.

Table 9.2 Hypothesized Correlation Matrix

weak weak

weak weak

weak

Rosenberg Self Eatcem Sale

Apeea-e

dente

moderate

moderate

weak

wcak

weak

weak

weak

high

wak

Evaluation weak

Orientation weak

Evaluation

Fitness

weak

weak

Orientation Heallh

weak

weak

weak

weak

weak

wcak

wak

high

wcak

weak

weak

Evaluation HcaIih

weak

Orientation üiness

weak

Orientation Body-A~s

Satisfaction

Table 9.2 Hypothesized Correlation Matrix Continued Body

Vulaerability

Transparcncy

M y

Limitations

Stigma

Concerna

Phytical Fn

weak

weak

moderate

wcak

weak

weak

Rote Fn

weak

weak

wcak

weak

weak

weak

Emotionat

moderate

WC&

wtak

weak

wesk

weak

Function

1

1

1

1 weak

Social Function

1 Weak

1 w4c

1

lVak

1 weak

1 wu*

weak weak

Emotional

moderate

wcak

weak

wtak

weak

weak

weak

GS

ibIe only

Calculated Correlations The calculated Spearman's rank correlation coefficients are presented in Table 9.3. As predicted fiom the Literature and the opinion of the expert judges experts, moderate correlations were found between the RSE and scales of the BIBC. Also as predicted, moderate correlations were found between the BDI and the sales of the BIBC. For one scale of the

BIBC (Limitations) a strong (0.63) correlation was found with the BDI. Moderate and strong correlations were found between the scales of the MBSRQ and the scales of the BLBC, particularly between the Appearance Evaluation and the Body Areas Satisfaction scales of the

MBSRQ and the Body Concems sale of the BIBC. The Tramparency scde of the BIBC did not correlate with the MBSRQ as predicted.

Table 9.3 Calculated Correlation Coefficients Limitations

Beck Depression hvtntory

Appearancc

Orientation Fitneos

Evalwtion

Fitnesa Orientation Heallh Evalwtion

üinesa Orientation Body- Arcao

Satisfaction

Table 9 3 Calculritcd Comlation CoefficientsContinucd

I

Limitations

1

Body

Couuni

Physical Function

Ernotionai Function Cognitive Function

Global Heailh Status

Cognition

11 0.04

1 -0.12

1 4.01

1 4.05

0.04

1 0.30*

1 4.17

1 4.16

Toul Score

Kev

0 0.05

IX

0.0001

>p

2 p

> 0.01

Negtigible / W d Correlation Predicted

1 No Prediction Made

As predicted, no clinically or statistically signifiant correlations were found between the

MCSD scaie and the d e s of the BIBC. Contrary to the a priori hypothesis, only weak I negligible correlations were found between the Dnve / Relationship scale of the DISF and the

scales of the BIBC. The only clinically and statistically signifiant correlation between the

BIBC and the DISF was found for the Drive I Relationship sale of the DISF and the Body Stigma scale of the BIBC,with significant correlations for the Arousal and Behaviour scales of the DISF and the Body Stigrna s a l e of the BIBC. As predicted fiom the judgment of the 3 content experts, moderate correlations were found

between the IES subscaies and the d e s of the BIBC. Also as predicted, the correlations

tended to be higher for the Intrusion sale, however, the correlation between the Intrusion

sale of the IES and the Vuinerability of the BIBC (r=0.55) did not achieve the predicted strong correlation. Correlations for the Avoidance scale tended to be higher than expected. No signifiant correlations was found for Body Stigma and Subjective Distress and this negative

füiding was unexpected. Correlations between the EORTC QLQ-30and the BIBC tended to be

higher than predicted, however as predicted, few moderate correlations were found between the Body Concerns, Transparency, and Arm Concems scales of the BIBC and the scales of the

EORTC QLQ-30. Moderate wrrelations behveen the Vulnerability scale of the BIBC and the subscaies of the EORTC QLQ-30 were found as expected, however a sirnilar pattern was found between the Body Stigma scale and the EORTC scales. This had not been predicted. Moderate or strong correlations were found between the Limitations scale of the BIBC and ail scales of the EORTC QLQ-30.These correlations were higher than predicted but generally foliowed the pattern of correlations hypothesized a priori. Because of the ceiling effects found for the Transparency scale of the BIBC and the floor effect found for the scales of the DISF, the Kendall's tau correlation coefficient was calculated for these scales. Kendall's tau correlation coefficient is denved non-pararnetrically based on ranks. This sîatistic is recommended when many tied ranks are present. The Kendall's tau correlation coefficient was compared to the Spearrnan's correlation coefficient. Correlations between the scales tended to be lower using this statistic, however a i l correlations which had been statistically signifiant remained so and no additional correlations were found to be statistically or clinicaLly significant for the Transparency scale of the BIBC or the scales of the DISF.

9.3.2 Discriminant Validity

Demographics and Scores of the 1 Breast and 2 Breast Groups

Forty-two women fiom Sample 2 had had mastectomy with no reconstruction. These women formed the 1 breast group. One hundred and twenty two women had had lumpectomy or

mastectomy with breast reconstruction and formed the 2 breast group. Demographic information for groups is presented in Table 9.4. Table 9.4

Demographics of 1 Breast and 2 Breast Groups I Breast Group

2 Breast Group

Significance of

Difference I

Average Age

62. I

59.7

NS

% Married

72 %

68%

NS

% Sexuaily Active

40%

52%

NS 1

Time Since

83 months

53 months

p=0.02

Diagnosis

The 1 breast group was found to have a significantly longer average time since diagnosis than the 2 breast group (83 versus 53 months respectively). No other demographic variable

significantly differed between the two groups. The distribution of the scores on the scales of the BIBC for the 2 groups are presented in Table 9.5.

Table 9.5 Distribution of Swres for Scales used in the Vdidity Study Scom for the B K items given to Cancer Group (Comrnon Items Only)

Median

l

i

1 Std

1 Min

Ï Max

Skewness

Kurtosis

2 Breast Group

1

Body Stigma

1 Breast Group

2 Breast G T O U ~

2 Breast Group 1 Breast Group

2 Breast Grwp 1 Breast Group

2 Breast Group 1 Breast Gtoup

2 Breast Group

Demographics and Scores of the Control Subjects

Fifty two o f 60 w o m e n with breast biopsies and 52 of 56 wornen from family practice clinics

who were recruited retunied the questionnaire, for a response rate of 90 % . Wornen in the control group were significantly younger than the women in the cancer group, and statisticaily more likely to be sexually active. The scale scores for items administered to subjects in the

cancer and control groups are presented in Table 9.6.

Table 9.6 Distribution of Scores for Scales used in the Validity Study Score for the BIBC items given to Cancer and Conîrol Groups (Common Items Only)

SU Dev VuhmbW

1 1 1 in

~ a xSicemers

1 Cancer GIwp 1 154 1 24.6 1 24 1 Control Gmup 1 94 1 22.3 1 22

Limitations

Cancer Group

Body Coafems

Control Group

Hypothesized differences Between Groups

The hypothesized group differences are summarized in Table 9.7. Significant differences were predicted between the scores of the 1 breast and 2 breast groups in the Body Stigma sale and the Transparency scale. Significant differences were predicted between the scores of the cancer

and conuol groups in the Vulnerability, Body Stigma, Transparency scales and possibly the

Body Concems scale. Table 9.7 Hypothesized Dfierences Betweeo Groups

BIBC Scale

1 Breast vs 2 Breast Group

1 I

Cancer vs Control Group

1 Signiticant, Cancer Wone M y Stigma

1 Limitations Body Concerns

1 Traosparency Arm Concerns

Signifïcant, Mastectomy Worse

1 No Diffsrence

Significant,Cancer Worse

1 No Différence

No Difference

Significant, Cancer Better

1 Significatzt, Mastectomy Worse

1 Significant, Cancer Worse

No DNerence +

kdiction Quafificd as PossibIe

1 *

1

147

Calculated Differences Between Groups The calculateci c o m p ~ s o n of s BIBC scores behveen the groups are presented in Table 9.8. As predicted, statistically significant differences were found between the 1 breast and 2 breast groups for the Body Stigma and Transparency scales. A statistically significant difference was also found between these 2 groups for the AMI Concems Scale. This was not hypothesized a

priori. As there was a signifiant difference in the time since diagnosis between the two groups, a second analysis was undertaken controhg for this variable. Controlling for tirne

since diagnosis did not dramatically alter the differences in scores between the groups for any scale. The statistical sigriificance of the difference between the groups found for the Arm Concerns scale decreased slightly d e r controlling for time since diagnosis (from p=0.02 to p =O.O6 using non-parametric statistics, and fiom 0.05 to p =O.O7 using parametnc statistics). Table 9.8 Calculated Differences Between 1 Breast Group and 2 Breast Group

BXBC Scale

Before Adjustrnent for Time Since

With Adjustment for Time Since Diagnosis - -

NonParametric

Parametric

p a ) sa F w o n in R r w . Master of Science in Clinical Epidemiology Dissertation. University of Toronto 1996. Margolis G, Goodman RL, Rubin A. Psychological effects of breast-conserving cancer treatment and rnastectomy. psychosomatics 1990; 31:33-39. Marsella A, Shizuru L, Brennan J, et al. Depression and body image satisfaction. J Crossw s y c h o @ 1981; 12:36û-371. McCaulay M, Mintz L, Glenn A. Body image, self-esteem, and depression-proneness: closing the gender gap. Sex 1988; I8:38l-W. McCrea CW, Summerfield AB, Rosen B. Body image: a selective review of existing measurement techniques. Pr J Med Ps 1982; 55:225-233.

Mendelson , White D. Relation between body-esteem and self-esteem of obese and normal children. S m . 1982; 54:899-905. Mendelson B, White D, Mendelson M. Selfesteem body est=: E e t s of -es -. Research Bulletin, Center for Research in Human Development Concordia University 1994-1995; vol 13. Meyer L, Aspegren K. Long-term psychological sequelae of mastectomy and breast conserving treatment for breast cancer. Acta 1989; 28: 13-18. Microsoft Corporation. m o s o f t Wwd 1Jw's Gu&

1993- 1994

Mintz L, Betz N. Sex daerences in the nature, realism and correlates of body image. 1986; 15~185-195.

Mittal B, Balasubramanian S. Teshg the dimensionality of the self-consciousness scales. 1 1987; 51:53-68.

Mock V. Body image in women treated for breast cancer. Nurs Res-

1993; 42: 153.

Moinpour C, Feigl P, Metch B, Hayden K, Meyskens F, Crowley J. Quality of life endpoints 1989; 81:485-495. in cancer clinical trials: reviews and recommendations. J Nat1 Cancer Moinpour C, Hayden K, Thompson 1, Feigel P, Metch B. Quality of life assessrnent in Southwest Oncology Group trials. Qnc&gy 1990; 4:79-85. Nishisato N, Torii Y. Effects of categorizing continuous normal distributions on the productmoment correlation. 1970; 13:45-49. Noguchi M, Kitagawa H, Kinoshita K, Earashi M, Miyazaki 1, Tasukuchi S, Saito Y, et al. Psychologic and cosmetic self-assessments of breast conserving therapy compareci with mastectomy and immediate breast reconstniction. J S U ~ Ont; P 1993; 54:260-266. Noles S, Cash T, Winstead B. Body image, physical attractiveness and depression. J C Ps-

1985; 53:88-94.

Nunndly J. ESycho&c Numally J, Bernstein 1.

t

u

.

McGraw Hill Inc. New York. 1978.

W McGraw-Hill . Inc. New York.

1994

Offman HJ, Bradley SI. Body image of children and adolescents and its measurement: an overview. Çan J Psychiatq! 1992; 37:417-422.

m

Oken M, Creech R, Tormey D et al. Toxicity and response critena of the Eastern Cooperative Oncology Group. Am J C h O& 1982; 5:649-655. Ornne-Ponten M yHolmberg L, Sjoden P. Psychosocial adjustment among women with breast cancer stages 1 and II: Six-year foilow-up of consecuive patients. J J 1994; 12: 17781782. Organization D, Till JE, Pater JL, Young JR. Health-related quality of life: measurement and clinicai application. A workshop report. Çan J O= 1995; 5:338-343. Orlofsky J. Relationship between sex role attitudes and personality traits and the sex role behavior scale-1: A new masure of masculine and feminine role behaviors and interests. 1 Soc1981; 40:927940. Orr D, Reznikoff M, Smith G. Body image, self-esteem and depression in bum-injured 1989; 10:454-461. adolescents and young adults. J- B Osoba D, Zee B, Pater J, Warr D, Kaizer L, Lareille J. Psychomeûic properties and responsiveness of the EORTC Quality of Life Questionnaire (QLQ-C30)in patients with breast, ovarian and lung cancer. 1994; 3:353-364. Pare, A. In Arieti S. (ed)

of Psyahiatqr, vol 1. New York: Basic Books.

Penman D, Bloom J, Fotopoulos S, Cook M, Holland J, Gates C, Ramer D, et al. The impact of mastectomy on self-concept and social function: A combined cross-sectional and 1987; 101-130. longitudinal study with cornparison groups.

Piazza E, Rollins N, Lewis F. Measunng severïty and change in anorexia nervosa. A d o w 1983; 8:293. Phelan P. Cognitive cone1ates of bulimia: The bulimic thoughts questionnaire. D i s o r d a 1987; 6:593-607. Plumb M, Holland J. Comparative studies of psychological function in patients with advanced . . 1977; 39:264-279. cancer, 1: Self-reported depressive symptoms. PsychosoPolivy J. Psychological effects of mastectomy on a women's feminine self-concept. Ment 1977; 16477-87.

Jm

Posavac E, Waker R, Foley J, Sannito T. Further factor-analytic investigation of the Thorne femininity study. J 1977; 33:25-3 1.

Pruzinsky T. Psychopathology of body experience: expanded perspectives. In body^ . -e Cash TF, Pruzinsky T. Guilford Press, New York. 1990, pp 170-189.

Ray C. Psychological implications of mastectomy. Br J Soc Clh Ps16:373-377.

1977;

Reaby L, Hort L, Vandervord J. Body image, self-concept, and self-esteem in women who had a mastectomy and either wore an extemal breast prosthesis or had breast reconstruction 1994; 15:361-375. and women who had not experienced mastectomy. Reed D, Thompson J, Brannick M, Sacco W. Development and validation of the physical m 1991; 5:323-332. appearance state and trait anxiety s a l e (PASTAS). u e t y D Reed D, Thompson J, Brannick M, Sacco W. Bpdy ytThompson J, 1994.

scale. Obtained from

Reitman E, Cleveland S. Changes in body image following sensory deprivation in S o W s y M . 1964; 68: 168-176. schizophrenia and control groups. J AbRobinson J, Shaver P. of s o Institute for Social Research, University of Michigan, 1973. Roid G, Fitts W. Tennessee Self-C-t SeMces, Los Angeles 1991.

Sc&

. Ann Arbor, Michigan:

flSCS) M a d . Western Psychological

Rosen G, Ross A. Relationship of body image to self-concept. J Comdt Clin -P 1968; 32: 100. Rosen J, Vara L, Wendt S, Leitenberg H. Validity studies of the eating disorder examination. 1990; 9:s19-528. Rosen J, Srebnik D, Saltzberg E, Wendt S. Development of a body image avoidance questionnaire. J Comlt C i l i n 1991; 3:32-37. Rosen J, Reiter J. M y Dyfrom the authors.

D-ion

.

.

User's

u, 1992, obtained

Middletown. , Connecticut: Wesleyan Rosenberg ,M. 0 University Press, 1989. Rosenberg M. ÇPaCei-. 1979.

..

New York, New York. Basic Books, Inc., Publishers,

RosenRoth D, Armstrong J. Feelings of fahiess questionnaire: A measure of the crosssituational variability of body expenence. IaJat Disorders 1993; 14:349-358. Roth D, Armstrong J. Feelings of F

. Obtained fiom the authors, 1994

v

Ryckman R, Robbins M, Thomton B, CantreU P. Development and validation of a physical selfefficacy scale. J 1982;42:891-900.

. - . Springer-Verlag New York. 1982. .. Sackett D, Haynes R, Guyatt G, Tugweii P. W c a l F.oidemiologv, a Basic S .. cal M W . Little, Brown and Company, Boston, 1985. pp 173-185.

Sachs L.

k e f~

Sanger CK, Reznikoff M. A cornparison of the psychological effects of breast-saving procedures with the modified radical mastectomy. 1981 ; 48:2341-2346. Sankowsky M. The effect of a treatment based on the use of guided visuo-kinestheic imagery on the alteration of negative body-cathexis in women. Dissertation Abstracts International 1981;42, (University Microfilms International No. 8203904) as found in Haimovitz D, Lansky L, O'Reilly P. Fluctuations in body satisfaction across situations. Int J Fat Disordm

1993; 13:77-84. Sannito T, Walker R, Foley J, Posavac E. A test of female sex identification: The Thome 1972;2853 1-539. femininity scale. J Clin PsScaramucci M, Laylon J. Androgyny. A function of degree? J De-

1990;25-29.

Schag C, Heinrich R, Aadland R, Ganz P. Assessing the problems of cancer patients: Psychometric properties of the Cancer Inventory of Problem Situations. &&h P s 1990; ~

9:83-102. Schain W,Edwards B, Gorreil C, et al. Psychosocial and physical outcornes of primary breast cancer therapy: Mastectomy versus excisional biopsy and irradiation. m a t -

1983;3:377-382. Schain W. Breast cancer surgenes and psychosexual sequelae: implications for remediation. s Onc N u r a 1985; 1:200-205. Schain W, d'Ange10 T, Dunn M et al. Mastectomy versus conservative surgery and radiation therapy , psychological consequences. 1994;73:1221-8. Schain WS, Wellisch DK. Unpublished. Obtained fiom the authors, 1994.

tto

Brmt S

m Ou-.

Schilder P. Imageof o f Body. Kegan Paul, Trench, Trubner and Company, London, 1935. Schontz F. Body image and its disorders.

Ia J P s y c.u in. M d

1974; 5: 150-161.

Schover L. The Impact of breast cancer on sexuality, body image, and intimate relationships. ÇB 1991; 41~112-121.

Schover L, Yetman R, Tuason L, Meisler E, Esselstyn C, Hermann R, Broniatowski S, et al. 1995; 75:54-64. Partial mastectomy and breast reconstruction. Secord P. Objectification of word association procedures by the use of homonyms: A measure of body cathexis. J Pers 1953; 2 l:479-495. Secord P, Jourard S. The appraisal of body-cathexis and the self. J C

o 1953;

17:343-347.

Sheatsley P. E[andbook.Academic Press 1983, pp 225-227. Shontz F. Body image and its disorders. m'l J Ps-

.

*

Me&cine 1974; 5:46 1-472.

Silverstein A, Robinson H. The representation of orthopedic disability in children's figure drawing. $ C o a t P m 1956; 20:333-341. Slade P. What is body image. &ha-

Th= 1994; 32:497-502.

Slade P, Russell G. Experimental investigations of bodily perceptions in anorexia nervosa and obesity. W o t h e r Psvchosam. 1973; 22:359-363. Slade P, Dewey M, Newton T, Brodie D, Kiemle G. Development and preliminary validation of the body satisfaction s a l e (BSS). 1990; 4:213-220. Slade P. The Rodv Satisfaction W. Obtained from the author, 1994. Sneeuw K, Aaronson N, Yarnold J, Broderick M, Regan J, Ross G,Goddard A. Cosmetic and functional outcornes of breast conserving treatment for early stage breast cancer. 2. Relationship with psychological functioning. -y 0 1992; 25: 160-166. Solnit A, Pnel B. Psychological reactions to facial and hand burns in young men: cm 1 see myself through your eyes -dy ChiU 1975; 30549466. Steer R, Beck A, Riskind J, Brown G. Differentiation of depressive disorders from generalized anxiety by the Beck Depression Inventory. J Clin P s y c u 1986; 40:475-478.

Steer R, Beck A, Brown G, Berchick R. SeIf-reported depressive symptoms differentiating Clin 987;43:246-250. major depression frorn dysthymic disorders.

Steinberg MD, Juliano MA, Wise L. Psychological outcome of Iumpectomy versus 1985; 142:34-39. mastectomy in the treatment of breast cancer. & J Ps Streiner D, Norman G . H e a l t h - ~ c a l e s . A d Oxford University Press Oxford. 1989. pp 39-42.

Chide kUh& D e v e w

a

Streiner DL. Sarnple-size fomulae for parameter estimation. 78:275-284. Streiner D,Norman G. HealthScales. A P UE. Oxford University Press Oxford. 1995.

1994;

m Guide to th& D e v e i o m

Sudman S, Bradbum N. Askine Questions. A P d c d ciilide to Questicrnnaire Desigxl. Jossey-Bass Inc. San Francisco. 1982. Sweeney M, Zionts P. The "second s b " : Perceptions of disturbed and nondisturbed early 1989; XXW:411-420. adolescents on clothing, self-concept and body image. -e Taylor S, Lichtman R, Wood J, Blurning A, Dosik G, Leibowitz R. Illness-related and treatment-related factors in psychosocial adjustment to breast cancer. Caux 1985; 55:25062513.

Theodorakis Y,Doganis G,Bagiatis K. Age differences and structural validity for the Greek 1991; 68:43-49. version of the Body Cathexis Scale. Psyc-hplaeical Thomas C, Freeman R. The body esteem scale: Constnict validity of the female subsdes. 1 1990; 54:204-212. Thompson J, Fabian L, Moultone D, Dunn M, Altabe M. Development and validation of the physical appearance related teasing scale. J P e r m - A m 1991; 56:s 13-521. Thornpson J, Dolce J, Spana R, et al., Ernotionally versus intellectually based estimates of body size. I aJ F w D m 1981; 6:507-5 14.

Thompson J, Psaltis K. Multiple aspects and correlates of body figure rathgs: a replication J F w Disorders 1988; 7:813-817. and extension of FaUon and Rozin (1985). Thompson J, Spana R. The adjustable light beam method for the assessment of size estimation accuracy: description, psychometrics, and normative data. W'I J v. 1988; 7521-526.

Thompson I, Dolce J. The discrepancy between emotional versus rational estimates of body size, actual size, and ideal body ratings: theoretical and clinical implications. J clin 1989; 45:473-478. Thompson J, Penner L, Altabe M. Procedures, problems, and progress in the assc~mentof dc m . body images. In: Cash T, Pnuinsky T eds. Guilford Press, New York 1990. pp 80-109.

Thome F. The measurement of femininity. J Clin Psychology 1977; 3 3 5 1 0 . Toner B, Koyama E, Garfinkel P,Jeejeebhoy K, Gasbarro 1. Social desirability and irritable .. bowel syndrome. Ia J -P in Medicine 1992; 22:99-103. Tucker L. Intemal structure, factor satisfaction, and reliability of the Body Cathexis Scale. ercephlitl Motor S m 1981; 53:891-896. Tulsky D. An introduction to test theory.

1990; 443-48.

Vamos M. Body image in chronic iUness - A reconceptualization. 23: 163-178.

J Ps-

M d 1993;

van der Velde C. Body images of one's seif and of others: development and dinical significance. JP 1985; 142527436. van Strien T, Fnjters J, Bergers G,Defares P. The Dutch eating behavior questionnaire (DEBQ)for assessrnent of restrained, emotional, and extemal eating behavior. Disorders 1986; 5:295-3 15.

--

Wassner A. The impact of rnutilating surgery or trauma on body-image. 29: 86-90.

XOLNLW&1982;

Wellisch D, Schain W, Noone B, Little J. Psychosocial correlates of immediate versus delayed reconstniction of the breast. 1985; 76:713-718. Wellisch D,DiMatteo R, Silverstein M,et al. Psychosocial outcornes of breast cancer therapies: lumpectorny versus mastectomy. Psvchoromatics 1989; 30:365-373. Wilmuth M. Sexuality after spinal cord injury: A criticai review. çlio Ps7:389-412.

Rev 1987;

Wolberg W, Romsaas E, Tanner M, Malec J. Psychosexual adaptation to breast cancer surgery. Q m x 1989; 63: 164-1655.

Yilmazer N, Aydiner A, Ozkan S, Aslay 1, Bilge N. A cornparison o f body image, selfesteem and social support in total mastectomy and breast-conserving therapy in Turkish women. -e C a 1994; 2:238-241. Zilberg N, Weiss D, Horowitz M. Impact of event scale: a cross-validation study and some empuical evidence supporthg a conceptual mode1 of stress response syndromes. J CoPsycM 1982;50:407-414.

. .

Zinreich E, Derogatis L, Herpst J, Auvil G, Piantadosi S, Order S. b t l J R&aQon 0nwhg.y Bi01 P h u 1990; 19:729-732. Zinreich E, Derogatis L, Herpst J, Auvil G, Piantadosi S, Order S. Pretreatment evaluation . * of sexual functioning patients with adenocarcinorna of the prostate. h t J &&&lm O n a b g y 1990; 19:1001-1004.

APPENDICES

Appendix I Standards for Evaluating Studies of Clinical Course and Prognosis Wàs an 'inceptioncohon' assembled? 1. When evaluating a study of prognosis, it is essential that patients be identified at a very early course in their treatment, such that d l individuais, those who are cured, those who have died of their disease, as well as those whose disease has persisted, are included. Clearly a study of body image beginning with an inception cohort would provide excellent information about the effect of illness on all patients, however this critena may not be relevant in al1 populations. For instance, we may be interested in the body image of long tenn survivors, versus a l l individuals affected (some of whom will die of disease). An inception cohort would not be necessary in such a case. Thus for this critical review of the Literature, the critena has been altered from 'patients should be i d e n ~ f eor d the tinte of their diagnosis so that those who moy mi adjust are Nicluded with those who do adjut', (Irvine, 1991) to methods for selection of subjects for study should be clearly identified such that the results are interpretable and selection bias is minimized. This cnteria will be termed assembly of cohort.

WCLF rhe referral puttem described ? Subjects for any study must be selected from a pool of potential patients. Characteristics of this population vary from site to site, for exarnple patients with a particular disorder referred to a major academic centre have been shown to differ in terms of clinical course and prognosis from ail patients with that disorder.(Elienberg, 1980) Patients at different centres may differ in terms of socioeconomic status, ethnicity and treatment and this could influence the results of research exarnining the impact of breast cancer on body image in these populations. Therefore the referral pattern should be described in the study and the patient population included should be throughly delineated in order to evaluate the impact referral bias may have had on study outcome. This cnteria will be termed description of population. 2.

Was complete followup achieved ? 3. If follow up is not complete, it is possible that study participants differed significantly from those who did not participate. In questionnaire research, responders are known to differ from non-responders in many respects which affect outcorne (Kelsey, 1986). Thus, it is essentid that a high rate of response (or follow-up) is achieved. An 80% rate of response has been suggested as a standard.(Irvine, 1991) This critena will be temed response rate. Were objeciive outcorne criteria developed and wed and was the outcorne assessrnenf blind a) Were objective outcorne criteria developed and ured? It is essentiai that the outcome measures utilized in the studies be reliable and vaiid. If this is not the case, the meaning of the study results become uninterpretable. 4.

Appendix 1 Continued b) War the outcome assessment 'blind'? If researchers measuring the outcome of disease are not blinded to major prognostic factors, bias (diagnostic-suspicion bias and expectation bias) is easily introduced. Thus, blind evaluation of outcorne, or the utilization of self-report measures, which do not necessitate the subjective assessment of an extemal examiner, is essential to the validity of study results. Many of the commonly used measures of body image used are self-report inventories completed by the subjects. In such cases bhding of the investigaton is not necessarily essential to the vaüdity of the studies and blinding of patients to type of treatment received would be impossible. However if the measure requires the subjective evaluation of an individual other than the subject, blinding is essential. The 2 Sackett Cntena were objective oucome crireria developed and used and was the outcorne ussessrnerit blind will be combined and termed outcome measure. If the outcorne measure requires an investigator's evaluation, then blinding will be noted. In this review, the outcome of interest to be measured is body image.

5. W u &jusment for exîraneous progmstic facrors carried out? Extraneous factors, such as concurrent iliness, age, and time since diagnosis rnay impact on the results of studies of body image. Groups of women who have had different forms of treatment for breast cancer (eg. lumpectomy or mastectomy) may also differ in terms of other factors such as age or time since diagnosis. Such differences may influence prognosis in terms of body image. Thus adjustrnent for possible confounding variables should be undertaken to minirnize the influence of these factors on results. This cnteria wiU be termed adjustment for prognostic factors.

Appedk 2 SEMSTRUCTURED INTERVIEW

AGE, MARITAL STATUS, SURGERY,ADJUVANT THERAPY, METASTATIC DISEASE, WEIGHT CATEGORY 1-Have you experienced m y probiems with body image since your breast cancer diagnosis? If so wbat bave the problems beea? 2.Ha.s the way you view your body cbanged since your breast cancer diagnosis? If so, how has your view changeci? el*Mer developing breast cancer some women become uncornfortable with the appearance of their breasts or scars. Have you experienced this? Couid you e b r a t e ? 2.After developing breast cancer, some women notice other changes in their appearance, such as hair o r skin changes. Have you experienced this? Could you eiaborate? 3 .Som0 women are l e s satisfied or are more concerned with their appearance d e r developing breast cancer. Have you noticed this? Could you elaborate? 4.Some women feel more selfconscious after developing breast cancer* Have you? Could you eiaborate? 5.Some women develop weight problems, weight loss or gain after developing breast cancer. Have you? Could you elaborate? 6.After breast cancer diagnosis a d treafment, some women become more concemd about how others will react to this? Could you elaborate? their appearance than before treaûnent. Have you ndindid 7.Some women feel l e s attractive after breast cancer t-ent. Have you noticd this? Codd you elaborate? 8.Some women notices changes in the way their spouses react to them &er breast cancer treatment. Have you noticed tbis? CouIci you elaborate? 9.Some women change th8 way they dress d e r developing breast cancer. Have you? CouId you elaborate? 1.Some women change their activities, such es exercising, afler developing breast cancer. Have you? Couid you elaborate? 2 .Some women notice changes in their sexual relations a d sexual drive. Have you noticed any changes? Coulci you elaborate? 3.Some women notice changes in their abiiities to work at home o r outside the home. Have you noticeù aay changes? Coulci you elaborate? 4.Have you encountered any dificuities doing normal activities since developing cancer? Could you eIaborate? 1.Do you have any problems with pain at the surgical site or elsewhere? Coulci you elaborate? 2.Have you experienced any arm problems since your breast cancer diagnosis? Can you elaborate? 3.Have you been more fatigued since your breast cancer diagnosis? How has this effected your Me? 4.Have you noticed menopausa1 symptoms since your breast cancer diagnosis? Do you feel these are related to you treatment? How do you feel about this? 5.Did you have other side effects of chemotherapy (eg nausea)? What were they? How did this effect you? 6.Have you noticed any other symptoms since your breast cancer diagnosis? Do you feel these are relaieci to you treatment? How do you feet aboui t h ? 1.Are you more aware of yout body than befofe your diagnosis (for example aches and pains)? Do you worry about cancer recurrence/spread? Couki you elaborate? 2.Do you feel as capable as before your surgery? Do you feel more vulnerable than before your surgery? Could you elaborate? 3.Have you had to have other procedm doue which relate to the breast cancer (eg hic& catheter insertion)? How has this affected you? 4.Some women notice changes in their feelings of femininimininity after breast cancer treatmeat. Has this affected you? Could you elaborate? S.Do you feel in control of your body? Do you feel in control of cancer? Can you elaborate?

-

Appedix 3 Scales Identifid for Item Generations

The Muftidimensional Body Seif-Relations QuestioMaire (Cash 1994;Cash et ai., 1986; Brown et ai,, 1990) The Body Cathexis &aie (Secord and J o d 1953)) The Body Satisfaction Scale (Slade e& al., 1990;SIade 1994) The Body Image Anxiety Scale ( R e d et al., 1991, 1994) The Body Esteern Scale (Frauzoi anci Shields 1984;Thomas et ai., 1990) The BoJy Image Mex (Lasry et ai., 1987) The M y image Scale (Berscheiû, Walster ancl Bohnistedt 1972) The Body Focus Questionnaire (Fischer 1970) The Semrintic Differential M m r e of Body image (Champion et al., 1982) The Ben-Tovim W a k Body Attitudes Questionnaire (Ben-Tovim and Walker, 1991) The Body Image Avoiciance Questionnaire (Rosen J et al., 1991) The Body Dysmorphic Disorder Examioation Self-Report a d interview (Rosen I et al., 1992)

1. 2. 3. 4. 5.

6. 7. 8. 9.

10. 11. 12.

-eR The Body Shape Questionrviire (Cooper et al., 1987;Evans et al., 1993) The Dutch Eating Behaviour QuestionriRite (Strien et al., 1986) The Eating Disorciers C d o n (Cooper et al,, 1987; Rosen J 1990) The Golclfarb Fear of Fat S d e (Goldfh et al., 1985) The Binge Eating ScaIe (Gormally et ai., 1982) The Hennan and Polivy Restraint Scaie (Herman, Poiivy et al-, 1978) The Bulimic Thoughts Questionmire (Phelan 1987) The Eating Disorcier hventory (Garner 1990) The Feelings of Fatness Q u e s t i o ( ~ Rd et al., 1993; Roth et al., 1994) The F d , Fitness and h k s Questio~aire(Hall et al., 1983)

1. 2. 3. 4. 5.

6. 7. 8.

9. 10. t

l. 2.

3. -1

1. 2.

1. 2.

3.

Cancer Smcific M Psychosociai Reactions to Bteast Surgeries (Schain et ai., 1994; WeUisch et ai., 1985; Kemeny et ai., 1988) The Mastectomy Attitude Scale (Feather 1989) The EORTC Rating of The Cosmetic ResuIts of Breast-Conserving Treatment (Aaronson et al., 1988) . . FitncThe Derogatis S e 4 Fuactioniag inventoxy (Derogatis 1975) The Sexual Desife Conflict Scaie (Kaplan 1991)

The Thome Femininity Scale (Sannito et ai., 1972;Thorne 1977;Prosavac et al., 1977) The Sex RoIe Behaviwr Scale (Orlofsky 1981) The BEM Sex Role hventary (Scaramucci et al., 1990;Giigen et al., 1992)

el f-Ca 1. 2. 3. 4.

5. 6.

The Physicai Self-Efficacy Scde ( R y c b et ai., 1982) The Janis Field Self Esteem Scale (Robinsonet ai., 1973) The Texas Social Behavior Inventory (Heimreich et al., 1974) The Tennessee Self-Concept Scaie (Foid aad Fitts 1991) The Rosenberg Self-Esteem Scale (Rosenberg, 1979, 1989) The Coopersmith Seif-Esteem Inventory (Robinson et ai., 1973)

Self-Con%iQuxws The Self Consciousness Scaie (Mittal et al., 1987) 1. 2. The Physical Appearance Related Teasing Scale (Thompson J et ai., 1991)

Appedix 4 Content Experts Intervieweci

RADIATION ONCOLOGY D r Pam Caüan, Prulcess Margaret Hospital Dr G d Rawlings, Princess Margaret Hospital MEDICAL ONCOLOGY D r Carol Sawka, Bayview Regional Cancer Centre Dr E h Warner, Bayview Regionai Cancer Centre Dr Maureen Trudeau, Henrietta Banting Breast Centre GENERAL SURGERY Dr U Ambus, Toronto Hospital Dr E Fiih, Heurietta Banting Breast Centre Dr L Lickley, H e ~ e t t Banting a Breast Centre Dr D M c C d y , Henrietta Banting Breast Centre

PLASTIC SURGERY Dr I Semple, H e ~ e t t aBanting Breast Centre Dr J Mahoney, St Michad's Hospital

PSYCHIATRY Dr B Dorian, Mount Sinai Hospital Dr Moscareiio, Women's CoUege Hospihi

PSYCHOLOGY

MJ Esplen, Group Psychotherapist, Toronto Hospitai, Mount Sinai Hospital Dr J Polivy, Psychologist, York University Dr B Dom, Clinical Psychologist, Sunnybrook Medical Centre Dr K Jaspir, Eathg Disorclen Unit, Toronto Hospital NURSING AM Blair, Nurse Oncologist, Bayview Regional Cancer Centre Dr Marg Fitch, Oncology Research, Bayview Regional Cancer Centre Chris Barret, Nurse Oncologist, Womn's College Cancer Centre Leslie Vincent, Head of Nurse Oacology, Mount Sinai Hospital Mount Sinai Nurse Oncology Group

SOC= WORK Mount Sinai Social Work Group Jan h g s t r u m , Eating Disorclers Unit, Toronto Hospital Joanne Avery, Toronto Hospital, Oocology REHABILITATION MEDICINE Mount Sinai Rehabilitation Medicine Group OTHERS Prosthesis Expert: Pameia Stein Reach for Recovery Volunteer: D r Francis Doane Weiispring Counseiior. S h e h O'Leary Look Goal Feel Better Coordinator: Patrick Blakiey

Appedix 5 S q l e Content Expert Interview

This is a cihicai psychoIogist who has a broad experience counseliing women d e r the diagnosis a d treatment of breast cancer. He states that the response of women to breast cancer in tenns of M y image is quite variabte. Women who have a greater investmenî in their looks pnor to diagnosis tend to have a more difficult time in djusting to the effects of diagnosis and treatment. Single younger women and women whose partner has a big investment in theü looks aIso seem to have more difficulty in acijusting, Although age is a factor, with older women tending to adjust to the effects of surgery beüer than younger women, this is by no means entirely consistent, with some older women having very adverse reactions to treatment. In general, women who have had lumpectomy, issues amund body image tend to resolve over t b . For women who have had mastectomy, this is less consistent; some resolve their issues by having breast reconstruction. Arm swelling arwl niimbness rnay trigger fears of recurcence in some women a d the sweUing serves as a constant reminder of the cancer diagnosis. The tictuai swelling of the arm may cause women to feel negatively about their M i e s . A decrease in the mobiiity of the shoulder may be very difficult for active women who use theu arms. The issue of s e d fùnctioning &et breast cancer diagnosis and treatment is very important. It is important to detennine not just how the women look ancl feel but aiso how they fiction. Changes in the breast may cause s e d dificulties. Some women experieace niunhness or tingling in the breast wbich may decrease sexual pleasure. Also partners may feel unsure of how to incorporate the a l t e d breast into sexual relations. This problem may be compouadd when there is a hck of c o d ç a t i o n between the women with breast cancer and her partner. The women may feel rejected by her partner because of this. Treatment imluceci menopause rnay cause a change in sexual desire especiaily in the k t year afbr treatment. Mechanical difficulties with vaginal lubrication as weU as atrophy of the vaginal epithetium may cause discornfort during intercourse. This in tum may affect the relationslip of the women with breast cancer to her partner. When asked, many women wiU comment that sex is not as important to them any longer. For many this reflects a reluctance to confront the practical issues of sexual functioning. Women experience a lack of energy which can be due to chemotherapy as weli as sleep disturbance because of hot flashes from chemotherapy induced menopause or h m tamoxifen treatment. The alopecia which women experience during chemotherapy can be devastating for women, even women who are weil adjusteci to changes in the breast. The loss of pubic hair may &O be disconcerting as they are in general unprepared for this. Cognitive dyshction is a common effect of chemotherapy and c m last for 8 to 16 months d e r therapy. Women cornplain that they can't do caiculations, can't balance a checkbook, can't make decisions etc. This can be devastating, especidy for career women. Because of the cognitive impairment, women can lose confidence in themeIves a d may hold back at work. Weight gain is another major issue for women It becomes quite traumatiung for some women to see themselves, having gainecl weight, having lost their hair a d generally feeling a&I. Weight gain can become a vicious circle for some women, the women gain weight, they become depressed, they are put on anti4epressants, they gain weight because of the medication and h e w they feel even wcrse about themselves.

Many women feel out of control because of the cancer. This affects their image of themselves. They feel out of control emotiondy and over what is happenhg to their bodies. The women generally become unsure of themselves amuncI health issues. For example, women who previously felt self-confident about breast self-examination, may Iose self-efficacy, they are n d sure what they are feeling in their breasts d e r treatment, anci not sure what to do about lumps. Beccurse of the cancer, a portion of women feel their mortality in ways they harln't examined before. Some women feel diminished because of this. Taking drugs at aU becomes a major issue for some heaIth conscious women. They can feel that they am taking thbgs that are foreign, a d because of their ambivalence, these women may feel Iess diminished or less of hmselves as people.

In women with metastatic disease issues change as survival takes on a much more immediate and promulent d e . The issues become more "basic". Minor aches and pains in the whole breast cancer population become remiders to the patient of the fact that they had cancer, and may trigger a series of thoughts including the fear of rsurrence. Survivors seem to resolve most of these issues over tirne.

1am happy with my hair My a m look normal My arm feek nonnal 1 worxy about my ann 1 worry about cancer recurrence My body reminds me chat I have harl cancer 1 feeI badly about my body 1 enjoy breast touching during sexual intimacy A man woukd feel cornfortable touching my breast(s) 1 worxy about minor aches and pains 1 am less of a person since cancer 1 am less of a woman since cancer 1 feel confident whea 1examine my breast(s) 1am in control of my body 1 am happy with my weight 1 feel good about myself I can think nonnaiiy I can concentrate I can remember chings n o d y 1 have energy Vaginal dryness bothem me 1 enjoy sexual intercourse 1 have discornfort during sexual intercourse My mind is hctioning aormaiiy

Appedix 6 Demographics of Sample of Women with Breast Cancer Intervieweci for Item Generation

70=8

>

Single = 8 Marcied = 30 DivorcdSepatatd = 7 Widowed = 6

Lumpectomy = 23 Mastectomy = 23 Reconstruction = 4 No Surgery = 1

Weight (as Radiation = 27 Chemotherapy = 22 Tamoxifen = 22 Other = 6

ffiown = 7 None = 42 Possible = 2

bv the interviewer)

Overweight = 22 Normal Weight = 24 U d e r Weight = 5

Appeadk 7 Ikms Genemkd A differcnce between my bmsu can be seea in

clothing A diffcrcnce between my nipplcr can bc scen in clothing A man would feel cornfortable iauching my breast(s) A man would fmd m attractive Aftcr dressing for the &y, 1 f-1 sccurr about my aPPe"-e Afùr 1 am drcased for the &y, 1 am pleased with my appearancc Arm pain is a problem for me Beforc a sexwl eqcrience with a paancr, 1bccomt anxious k i n g tired affects my daily living king tircd holds me back king tired interferes with my daidy routine Being rùcd intetfcrri with my iife Being tired timits my activiiies Bccast pauis i r e a problem for me Changes in my ricin colour have woy me Commemi about my weight upret me Differrnccs bctween my brcasra bothcr me Differenccs bctween my uipplts botbcr me During close contact with others, 1 try to avoid contact with my chca During the day 1 think about my mastcctomy Facial hair is a problcm for me Having o d y one nipple bothem me 1 am a prisoncr to my appearance 1 am able to allow &ers to look my scars 1 am able to care for myaclf 1 am able to carc for &ers 1 am able to conccnlratc m d y 1 am able to do my work wrmaUy 1 am able to do normal thingr 1 am able to fiinction normally at home 1 am able to ttnction w r d l y wben working 1 am able to panicipacc in normal physical activities 1 am able io read normally 1 am afraid o f becoming fat 1 am afmid o f my bread(s) 1 am afmid o f touching the scar 1 am always nwarc of my brcasu 1 am anxious about my body 1 am ashamed of how 1look 1 am attractive 1 am a w a n o f other people's bmstrr 1 am awarc o f the position of my brrarru 1 am cornforcable being huggcd 1 am comfortabIe being ieen in the nude 1 am cornforcable chaaging clother in fiont of othen 1 am cornfortable looking at my scan 1 am comfoctable looking at mywlf 1 am desirable 1 am easily distractcd 1 am eficient 1 am embanasscd whcn d e r people sec me exercising 1 am forgeth1 1 am fnistnrted with my body 1 am good to my body 1 am happy wiih my appcarancc 1 am happy with my appetitt 1 am happy with my hair

107.

108. 109. 1IO. Ill. 112. 113.

1 am happy with my tevel of energy 1 am happy with my pmsihesis 1 am happy with my skin 1 am happy with my waist line 1 am happy with my weight 1 am happy wiih the appearance of my belly 1 am happy with the appeamnce of rny hair 1 am happy with the appeatançe of my skin 1 am happy with the position of rny nipple 1 am happy with the shape of my belIy 1 am happy with the shape of my hips 1 am happy with the shape of my thighs 1 am happy with the teldure of my skin 1 am happy with the way 1 look 1 am intimidated by the thought of a s e m l encounkr 1 am limited by my body 1 am more c o n c e a d with my appcarance than other womcn 1 am prroccupied with my hair 1 am proud of my body 1 am pmud o f m y bust 1 am satisfied with my appearance 1 am satisfied with rny appearance at special occasions 1 am satisfied with my appearance in clothcs 1 am salisfied with my appearance naked 1 am satisfied with my body 1 am satisfied with my hait 1 am satisfied with my level of semial desire 1 am satisfied with my sex drive 1 am saiisfied with my rcx Iife t am satisfied with my sexual relations 1 am satisficd with my shape 1 am satisfied with my skin 1 am satisfied with my waist Iine I am satisfied with the appearancc of my a m I am ratisfied with the appearance of my belly 1 am satisficd with the appcarance of my b m s t 1 am satisficd with the appearance of my bust when cfolhcd 1 nm saiisfied with the appcarance of my buttoclcs I am satisfied with the appearance of my fiagernaüs 1 am satisfied with the appearancc of my hips 1 am satisfied with the appearance of my maatcctomy 1 am satisfird with the appearance of my nipple 1 am satisfied with the appearance of my radiation lattoos 1 am satisfied with the appearance of my scar 1 am satisficd with the appearance of my thighs 1 am satisfied with the appeamnce of my waist 1 am satisfied with the feeling in my brcast 1 am satisficd with the miction of &en, CO my brcast 1 am satisfied with the reactions of othcrs to my mastectomy 1 am satisfird with the shape of my belly 1 am uitisfied with the shape of rny b m t 1 am satisficd with the shapc of my buuocb 1 am satisfied with the shape of my hips 1 am satisficd with the ohape of my thighs 1 am satisficd wilh the sizc of my brcast 1 am satisficd with the slün in the area of my mastectomy 1 am satisfitd wiih the texture of my breast 1 am satisfird with the way my nipple(s) look in clothing

1am sclfconscious about my appcarnacc I am ~elfcoruciwsabout my appeararice outside the home 1am self-conscious of my appeataace at home I am scxually amclive 1 am scxually desirnble 1am sexy 1 am slecpy during the &y 1 am tircd 1 am iruitcd likt a p e m n 1 am usually awarc of my appeatance 1 appuir compeicnt 1 appnxiatt my body 1 avoid close physical contact such as hugging 1 avoid clothes shopping 1 avoid cloihing which maker me awarc of the shape of my bust 1 avoid going out because of my appeanincc I avoid letting othem set tny scar for fcar of frighttning thcm 1 avoid looking at my scan h m breast surgery 1 avoid meeting people b e c a m of my appeamncc I avoid people because of my appearance 1 avoid physical intimacy 1 bccome anxious a h r scx 1 bccomc anxious during wx 1 can change clothes in a change m m with othcrs 1 can change clothes in a private change m m 1 can concentrate I can [iinction at home 1 can hnction at work 1 can move my a m normaUy 1 can participatt in normal activitits f can perfonn my duties 1 can mmember ihings normally 1 can sleep through the night 1 can think normally I c m use my a m normaily 1 can walk amund the house in my bm 1 can walk amund the house in my night clouies 1 can walk amund the houe in rhe nude 1 can Wear clothes which show my tigure 1 catch discases easily 1 check my brcast 1 check my mastectorny scar 1 couid allow somuioe to touch my brcast 1 couid cnjoy sex 1 cover my chest during sexual iatimacy 1 dmss to hide my brcasts 1 enjoy being sccn in a w i m suit 1 cnjoy brcam touching during iicxual -Uitimacy i enjoy semai inkrcounc 1enjoy sexual intimacy 1 feel adequate 1 feel angry at my body 1 feel anxious touching my maaectomy scar 1 fetl as cornfortable swimming m w as beforc cancer 1 feel as if my brcast ia aili thcre 1 fecl ashamed of my body 1 feel attractive 1 feel attractive when clothcd 1 feel attractive when naked 1 fer1 bad when 1 look at my hair 1 fiel badly about being tattood 1 fetl badly about my body

1 feel badly about my mastectorny 1 feel badly about my radiation tattoos 1fcel cancer is in my body 1 fccl comfortablc about my body wnilc cxercWmg 1 fcel comfonablc being sccn in my b n 1 fecl comfomble examining my bteast(s) 1 fttl comfonable having medical professiods sce my chcst 1 feel comfortable in the nude 1 fecl comfortable in the nude in front of &ers 1 feel comfoxtable looking at rny body 1 feel comforiable looking at my breast 1 feel comfortablc looking at my chest 1 fer1 comfombk looking at my masrectomy 1 feel comfomble looking at mysclf in the nudc 1 &el comfortable meeting strangeni 1 kel comfortable the way 1 look when 1am uakcd 1 feel cornfoctable undressing in fmnt of others 1 feel comfomble when others sec my bccast 1 feel comfortable when othcnr see my mastcctomy 1 feel comfortable when others set my scars 1 feel comfortrtble with close physical c o m c t such as hugging 1 fecl comforiable with my body I feel competcnt 1 feel comptece I feel confident about my hair 1 fetl confident examining my breast(s) 1 feei confident in my appearancc 1 fcel confident in my body 1 feel c o ~ e c t e dto my body 1 fetl covcring my chcst during sex makcs me more desirable 1 feel cmoked 1 feel cursed 1 feel damaged 1 feel deformcd 1 fccl desirable 1 feel desirable as a woman 1 fecl different 1 fer1 differrnt from oihers 1 feel dirty I feel disfigured 1 feel disfigured by my mdiation tattoos 1 feel distorted 1 fer1 embarrascd about my body 1 feci fat 1 kt1 fcminine 1 leel good about my appearance I feel good about my body 1 feel good about my hair 1 feel good about my weight 1 feel htallhy 1 feel hclplcss 1 feel 1 am going to die soon I feel 1 am in control of cancer 1 feel 1 can fight cancer 1 fecI I need to pmtcct my chcst 1 feel Iwill be healthy in the friture 1 feel il1 1 feel in control of my body 1 fect injured 1 feel invaded 1 feel legs of a person since cancer 1 feel Iess of a woman since cancer

240. 241. 242, 243. 244. 245. 246. 247. 248. 249. 250. 25 1. 252. 253. 254. 255. 256. 257. 258. 259. 260. 26 1. 262. 263. 264. 265. 266. 267. 268. 269. 270. 271. 272. 273. 274. 275. 276. 277. 278. 279. 280. 28 1. 282. 283. 284. 285. 286. 287. 288. 289. 290. 29 1. 292. 293. 294. 295. 296. 297. 298. 299. 300. 301. 302. 303. 304.

1 feel like a pan of me ir missikg 1 feel like a woman 1 feel like 1am only half here 1 feel iike my body has becn p o b n e d 1 feel Like my bmast is r n part ~ ~of me 1 feel like somcihing is cating my body up 1 feel Likt soorthing u p w i n g iaside my body 1 feel lopsidcd 1 feel mutilatcd 1 feel my body has bccn chan@ by carrer 1 feel my body has becn invadcd 1 feel my body h s betrayed mt 1 fecl my body har dcceived me 1 feel my body hm disappointcd me 1 feel my body has failed me 1 feel my body has let me down 1 fecl my body has iurncd againnt me 1 feel my body is able to fight cancer 1 fcel my body W able to fight d i i 1 feel my body is auractive 1 feel my body is bad 1 feel my body W differcnt 1 feel my body is falliag apcur 1 feel my body is pione to d'irrcaile 1 fecl my body ia mn d o m 1 feel my body is shu#ing down 1 feel my brcast is deformcd 1 feel my breasts Iook the mme 1 feel my brrasts look the mme in cloching 1 fccl my hips are too large 1 feel my mind ia du11 1 feel my thighs arc too large 1 fecl My body U curscd 1feelmrmaI 1 f e l off balance 1 feel old 1 feel old beforc my time 1 f e l pains wherc my breaa should be 1 feel part of' my body is misaing 1 fecl people can tell 1 Wear a prosthesis or p a d d i i 1 fcel people can tell my brraru are not normal 1 feel powerless 1 feel prone COcancer 1 feel pmne to discase 1 feel satisfied with my weigtit 1 feel secure about my body 1 feel self-conscious about my body 1 feeI self-conscious about my hair 1 feeI self-conscious about my naked b d y 1 feel self+onscious about my weight 1 feel sensual 1 feel ~cxuallyattractive 1 feel semtaIIy attractive when 1 am nude 1 feel sick I feel sick to my stomach I fccl suange 1 fecl ttnsc 1 feel lhat cancer ia growing in my body I feel lhat cancer is hiding in my body 1 fecl that cancer is sprcadii in my body 1 feel that cancer will always be in my systcm 1 feel ihat everyone is Iooking at my bust 1 feel that 1 belong in my body Ifeelrhatmybrtastisugly I feel Lhat my bust appears normal to othen

1 fer1 that my masttctomy is ugly 1 fcel that pan of me mua remain hiddcn 1 fwl that people are always looking at my brcasts 1 fecl that people are loolcing at me 1 feel th41 people are looking at my chest 1 feel that people c m tell 1 have o d y one aipple 1 feel that somelhing alien W in my body l feel that somtthing is taking over my body I fecl that sonuthihg W wrong with my body 1 fctl the appearance of my body might diirhirb &ers i feel the applrarance of my body migbt frightcn othcrs 1 fecl the rcan frorn breast surgery arc ugly 1 fcei thcre is a timc bomô inside of me 1 feel tired 1 fed tircd al1 ihe tirne 1 feel trwred like 1 have a disease others can catch 1 fed ugly 1 feei wcary 1 feel wholc 1 feel womady 1 fiddle wirh the position of my bra 1 fi& routine menut1 tssks difficult 1 get upset when 1 Iook at rny hair 1 have a good encrgy levrl 1 have changed the way 1 d e s s because of my radiation tauoos 1 have dilticulty achicving orgasrn 1 have dificulty rcading 1have discornfort during scxual intercourse 1 have energy 1 have felt so badly about my appearance, I have cned 1 have good posture 1 have let my body down 1 have lost interest id sex 1 have pmblems concentrathg 1 have problems with my attention span 1 have problcms with my memory 1 have restfùl sleep 1 have sex a p p d 1 have thought about hanning my body 1 hide my body when changing clothes 1 hidc my tattoo 1 kctp a part of my body hidden fmm othcnr 1 keep my chest covcred during sex 1 keep my htad covetcd at home 1 keep my head covered in bed 1 h o w what is happening in my body 1 like my body 1 likc my look just Che way lhey arc 1 like lhe way 1 look 1 like the way 1 look wiihout my clothes on 1 like what 1 sec whcn 1 look in the mimr 1 look awhl 1 look good 1 look htelthy 1 look il1 1 look like 1 have cancer 1 look likt mysclf 1 look normal 1 look normal in a swim suit 1 look okay 1 look oId 1 look oldçr than 1 sbould 1 look pale

1 look sick

395.

396. 397. 398.

399. 400. 401.

402403. 404. 405-

406. 407. site

408. 409. 410. 411. 412. 413. 414. 415. 416. 417. 418. 419. 420. 421.

422. 423.

1 look tired 1 look well 1 make adjustments to h e position of my bra 1 nted massurancc about my heallb 1 need to bç reassund about my appearance 1 nced to be ceasouttd about the appearance of my b-(s) 1 prokct my chest when king huggtd 1 protcct the sidc that 1had b w r t cancer on 1 atay in monz because of my appearance 1take cace of my body 1think about bnas! cancer 1t h i d about my brca.st 1th- abwt my b m rurgery 1 think about my hair 1 think about my martectomy 1 lhink about my radiation tactoos 1 ihink about the fact ihat 1am missing a bmmt 1think 1 deserve ihe attention of the opposit~MX 1 trcasurc my body 1tricat my body beuer than 1 w c d to 1trust my body 1try to hide my body 1try ta hide my bust 1try to hidç my bust with my arms in public 1try to hide my scars 1twist away f i m &ers when ihey get close to my chast 1 walk wiui my a m covering my chca 1 Wear loose titting clolbiag 1 wtar low cut tops 1 Wear rrvealing cloihing 1 Wear sleeveless ciothes 1 Wear swim suiu 1 Wear tight sweaters 1 Wear topa with V arck lines 1 wish 1 looked b e u r 1 worry about cancer coming back 1 worry about cancer cccurrcnce 1 worry about cancer mumence in my brrast 1worry about cancer rrcurrence in my mastectomy 1 worry about looking il1 I worry about losing ïntenrt in rex 1worry about minor aches d p a h 1 worry about my ability to care for mysdf 1 worry about my ability to care for orhem 1worry about my appearance 1worry about my appetitc 1 worry about my a m 1wony about my body 1worry about my brcast 1worry about my cating habits 1worry about my hair 1worry about my proethesis o r padding dipping 1wony about my weight 1 worry about & e n rejecting me because of my body 1womy about the appeatance o f my bust wben clothtd 1 worry about the appcarance o f my aipple(s) 1 worry about the food 1 crt 1 worry about the position of my brcaas 1 worry about the position of my pmsthesis o r padding

I worry about the way 1 look 1 worry that I appear weak 1worry lhat 1 look sick 1 worry Lhat my breasts don't look the samc in clothes 1 worry that my breasts look the same 1 worry Lhat my pmsthesis o r padding shows 1 worry ihat my prosthesis o r padding will move 1 worry that my weight will change 1 worry ihat ouiers can tell 1 have a mawectomy 1 worry b a t oihcrs can tell 1 have cancer 1 worry rhat ohers will touch my breast 1 worry lhat people can tell 1 am wcaring a prosthesis o r padding 1 worry that people can tell 1 have a rrastectomy 1 worry that peuple wiü touch my prosthesis or padding 1 worry h t the cancer is sprcading 1 worry when oîhets touch my brwst a c c i d e d y 1would avoid a ncw sexual rclationship because of my b m s f cancer 1would avoid an intima& relationship because of my cancer 1 would avoid an intimate rehtionship because of rny body 1 would be acceptable a s a sexual partner 1 would be comfortablc ~ k e in d front of othen 1 would be embarrassed to be secn nude by a lover 1 would be sexually attractive to a man 1 would enjoy brrast touching during scxual intimacy 1 would feel anxious examining my brrast(s) 1 would fecl comfoctabtc changing clorhes in fmnt of ohers 1 would feel comfortable changing in a public changem m

1 would fcel comfortablt changing in fmnt of other womm 1 would fer1 comfortable in a low cut top 1 would fed comfortable in a sle~velesstop 1 would feel comforrable in a swim mit 1 would feel comfoctable in loosc fitting c l o h h g 1 would feel comfortable in revealing clothing 1 wouId feel comfortabte id the nude with a lover 1 would feel comfortablc in tight sweaters 1 would feel comforrable naked during sex 1 would fceI comfortable wearing a top with V nrc)tli 1 would fccl comforiable wirh othcrs touching my breast 1 would feel comfohable with others touching rny mastectomy scar 1 would feel self-conscious if 1 were naked in front of a lover 1 would keep my chest lovered during sex 1 would turn my back towards othcrs whila undcessing It is dificult to care for my appcarance It upszts me to have othen ser my breast 1; upsets n u to have othets set my mastectomy Joint stiffncss borhets m e tooking at my head makes me feel bad Looking at my hcad upseu me Lumps and bumps on my body look ugly My appearance could bother oihtr people My a m bothers me My a m feels diff'rcnt

My a m feeh normal My a m feels smnge to me My a m is fiabby My a m is normal My a m is ugly My a m is weak My a m looks nomial My a m moves n a d y My body disgusts mt My body feels balançed My body feels bloated My body feels durable My body feels heaIthy My body feels sick My body feels m n g e to me My body feels swollea My body feels iough My body fceb weak My body feels whole My body fmstrates m My body haa bcen changcd by cancer My body has been damaged My body hoMo me back My body is diffenat My body is ugly My body l o o k good My body l o o k normal My body reminds me of my bmast cancer My body stops me €rom doing rhings 1 want to do My body will k U mc whea mnicihing is wmng My body would be amctive to a mm My breast cancer treatmeat has made me feel ashamed of my body My breast feels differcat to me My brrast feels heavy My brrast feels like it belongs to sameone tlsc My breast fcels ~ r m a t My bmsf is painhl to touch My breast l o o b normal My bnasts appear lopsided to othcn My bnasts appear uneven to oihecs My brcasts look off balance My enjoymtnt of sex has dccrcased My face has definition My hair adds to my appearance My level of sexual dtsire womcs me My l o o k upsct mc My mastectomy is in the back of my mind My memory is normal My mind has bctn affkctcd by cancer trcatrnents My mind is working normaily My pmsthesis ia comforuble My pmsthesis is ugly My scnse of smell is normal My s e n s of taste is normal My skin is irritated My lattoo bothem me My underann feel normal My undcrann lwks n o d Nausca inkrfemri with my life Othen CM tell 1 am misshg a bOthcrs can tell 1 have cancer Others have had to take ovcr my dutiu Others people are comfortable looking at rny body O~herswould bc disturûed by sechg my scar

Pain affccta rny ability to d o my normal activities Pain in my body affécu my activities Pain in my b r w t bolhers me Pain in rhe a m of my rnasrccmmy botbers mc Pain in the a r a of my mastcctomy is a pmbIem for mi:

Parts of my body feel foreign Parts of my body feel like thcy bclong to someooc tlse

Parts of my body ftcl numb People can teil 1 am iil People can trU 1 have cancer People feel s o q for me because of my apptarance People ihink of cancer when rbey x e me Peopk m a t me d i f f e r r d y because of the way I look Scting my ccflcction in a mirrot has made me feel bad Sex is paintiil for me Skin dryness is a problem for me Strangem know îherc is somdhing wmng witb me Swelling of my a m is a problem for me The appamnce of my body could disturb othen The appeacanct of my body couId frighkn othcrs The appearance of my breaa could disairb orhcrs n i e appamnce of my brwst disturbs me The feeling in my a m is normal The look of my malrtectomy distuhs me Tho way my maslcctorny loob could d'urtuh othen n i e wtight of my prosthesis boihers me There is extra tissue under rny arm Thinking about my appearance has interfercd with my concentration Vaginal discharge is a problem for me Vaginal dryness boihers me Vaginal drynasa is a problem for me When I touch my b ~ a s t ( s ) , feei anxious

Appendix 8 Items Idcntified as Ambiguous, Offensive o r Incomprehensible Item A man would feel comfortable touchiig my brtast(s)

Aftcr drtssing for the day, I fecl secure about my apptarancc Changes in my skin colour have worry me Differences bttween my breasts bother me Having only one nipplc bothers me 1 am a prisoner to my appearance

I am able to care for mysclf 1 am able to care for others 1 am able to do normal things 1 am a f i d of my brtast(s) 1 am anxious about my body I am aware of the position of my breasts I am comfortable being hugged

I am corn fortable changing clothes in fmnt of othcrcr I am w i I y distractcd 1 am efficient 1 am frustrateci with my body I am happy with my appttitc I am happy with my skin 1 am happy with rny waist iine 1 am happy with the shape of my beUy

1 am limiteci by my body 1 am more concernai with my appeamct than other women

I am proud of my bust 1 am satisfied with my appearance 1 am satisfied with my body 1 am satisfied with my hair 1 am satisfied with my ievel of sexual dcsire 1 am satisficd with my shape I am satisficd with the appearance of my brcast i am satisfied with the appearance of my nipple

I am satisfied with the appcarancc of my bclly 1 am satisfied with the feeling in my brcast I am satisfied with the reactions of othen to rny mastcctomy 1 am self-conscious about my appearanct outside the home 1 am sexy

Reason/Comments Offensive - offensive to Icsbians, should be changcd to a sexual partner Poor Wording would be bctîer worded I am pleascd versus 1 feel stcurt Unclcas Unclear Unclear UncIcar Offensive term is 'melodramatic' changing prisoncr to limiteci would improvc item Unclear Ambiguous tmotionally o r physically carc Ambiguous - emotionally care for o r physicdy care for Unclear what are normal things Unclear - afraid of actual breasts or what breast and breast cancer is d o h g to the body or afraid of rccumnce Unclau - anxious about attractiveness o r recurrence Unclcar Ambiguous - hugged by who? This would make a big diffcrence as to how this was answered UncIear - how undrrssed naked, panialiy UncIear - distracted from what Multiple Meanings efficient at what? lob, houst, organizing t h e Multiple Meanuigs - the look o r hnction Poor Wording happy did not rnake sense in this context - satisficd would be beuer Poor Wording - if happy was changed to satisfied, item would b t casier to answcr Poor Wording satisfied is more appropriate wording than happy for this item Poor Wording - felt that appearance of my beUy would bbe cluirer than shape Ambiguous - is this due to limitations resulting from treatment or baseline limitations of body Poor Wording - would be beüer worded 1 think 1 am Unclear Unclea r Poor Wording - subject felt wording would be improved if satisfied was changed to accept my appeamce Arnbiguous too open ended Poor Wording - should be hair o r scalp Unclear Multiple Meanhgs - what shape, of body, of breast etc Unclear Unclear - wnat does this refer to during sex, when out with people etc Unclcar UncIear sensation or comfort UncIear who are others Unclcar Ambiguous is this passive o r active, do you feel sexy o r do you s 10% of responses missing 19.1 wony aboui iosing ùtterest in sex Rcason: > 10% of riisponsei missing 20. My enjoyment of sex hm decreased Reason: > 10% of responses missing

21.1 am capabk of erïjoyùag sex Reason: > 10% of rcsponses missing 22.1 worry when orhers louch my prosîhesis or padding occidenMy Reason: 1-D comlation undcr 0.4

23. I would feel conCforîable with a sexdpartner touching my mostectomy scar R a s o n : > 10% of responscs missing

Appcndix 12 Continued Domab: Appesirance 1. I wony about my weight Reason: > 0.7 Item-Item corrclation with Item I wowy ihm my weiglu will change Lower 1-D correlations 2.1 keep my heaà covered at b m c Reason: Frequency of Endorsement over 75% for one IeveI Fuii range of responscs not endorscd > 1096 of responses misshg 1-D correlation undtr 0.4 3.1 worry about hair bss Reason: 1-D correhtion under 0.4 4.1 wony thai oîhers con tellI have cancer Reason: Frequency of Endorsement over 75% for one level 5. I ge! upset when I look ut my haÙLrc& Reason: 1-D comiation undcr 0.4 6.1 con walk around my bedroom in tite nude Reason: 1-D comIation under 0.4 7 . 1 worry îhaî I look sick Reason: 1-D corretation under 0.4 8.1 wony abou my ealùtg Miils Reason: 1-D comzlation under 0.4 9.1 am awwe of other people's bre& Reason: 1-D corretation under 0.4 10. I feel fd Reason: 1-D correlation under 0.4 I l . I feel secure about my appeanurce when I ana dressed Reason: 1-D comiation under 0.4 12.1 keep my head CO vered in bed Reason: F q u e n c y of Endorsement over 75%for one level > 10% of mponses misshg 1-D comlrition under 0.4 13.1 ûy tb hide my b u t witir my anns ùr public Reason: Frequency of Endorsement over 75% for one level 14. I feel bad whe~zI look ot my habbdp Reason: > 10% of responses rnissing 1-D comlation undcr 0.4 15. I fddk wirh the position of my bru Reason: 1-D correlation under 0.4 16. I feel cojtfrdnî irr my apperucurce Reason: Item-Item correlation with Item I feel good about my appearance blt 0.M.7 17.1 wony îhaî a dfirence beîween my breasfs cm be seen in cloîhing (L) Reason: > 0.7 Item-1tem comlation with item I think my breasts appear uneven ro ofhers Lower 1-D correlation, Wome distniution, L oniy 18. I feel my breasf is deformed (L) Reason: Item-Item correlation with Item Ifecl disfzgured blt 0.6-0.7 19.1 wony lhar people can tell I have had a matectorny ( M ) Reason: Fuii range of responses not endormi 20.1 worry about rhe posr'lion of my prosthesis or pa&iUtg ( M ) Rcason: > 0.7 Item-Item correlation with iiem I worry abou my prosthesis orpadding siipping Lower 1-D correlations 21. 1 womy lirat people con &CI I am wecuing o prosthesk o r @hg ( M ) Reason: Item-Item correlation with Item lfeel people can tell my bremu are nor normal blt 0.6-0.7 22.1 fcel selfconscious about my hair Reason: 1-D correlation under 0.4 23.1 am sCZILFfud wwiLh ihe way my bel& boks Rcason: 1-D correlation under 0.4 24. I feel treated as though I have a diseuse I can give û~others Reason: 1-D correhtion under 0.4 25.1 bokpa& Reason: 1-D correlation under 0.4

Appcndix 12 Continued Appearance Continued 26.1 a m scrtisfid wwirh the shape of my thïghs Rason: 1-D correlation undcr 0.4 27.1 am preoccupied wifh my haù Rauran: 1-D correlation undcr 0.4 28. I feel salisfïed wwilh my wekht Reason: 1-D correlation undcr 0.4 29. Skin dryness ir a probkm for me Rtason: 1-D comlation undcr 0.4 30. Facial kir k a problem for me Raison: 1-D cordation undcr 0.4 31.1 woucdfeel c o m f o ~ l Ur e a s w h suil Reason: > 0.7 Item-Item Comiation with Itcm I am cornfortable being seen in a swh su& Lower 1-D Comiation, No differcnct bctween L and M 32. Seeing my reffeclion in a mirror has mude me feel bad Reason: Item-Item comiation with Item I feel rhatparr of me must remah hidden bit 0-6-0.7 33.1 prefer to tunt rny bock tuwrvds oliiers whüè wtdrossbag Rcason: Item-Item comlation with Itmi I feel rharpart of me m u t remin hÿiden blt 0A-û-7 34.1 avoid ktîbrg others see my swvsfor fear of disturbing thent Reason: Item-item correlation with Itmi I would feel cornforfable changing in a public change-room blt 0.6-0.7 35.1 would feel comforhzble weruing a top wiLh a V neckltre Reason: Item-Item Correlation with Itcm (I wouldfeel cornfortable in a low CU top b/t 0 . 0 . 7 36. I feel rhat my bretrst ir ug& (L) Ratson: Item-Item correlation with Item I am s w e d with the appearance of my b r e m blt 0.6-0.7 37.1 am s&fid wüh the sttcrpe of my breast Reason: > 0.7 Item-Item Comlation with Item I am surirfied with the appearance of my b r e m Stightly lower 1-D comiation Wording less satisfactory 38. The way my masîecîomy Cook could disrurb orhers R a o n : 1-D comiation under 0.4 39.1 am sotisfred wwirh the skin bt the area of my marteclomy Reason: 1-D comiation undcr 0.4 40.1 am saîisfid wiih the appearance of my mustectomy Reason: 1-D comlation under 0.4

Domain: Predictability 1, I worry abouî my ann Reason: 1-D corrclation undcr 0.4

2. 1feel femirrùie Reason: 1-D correlation under 0.4

3 . 1 &ke care of my body Reason: 1-D comlation undcr 0.4

4 . 1 worry about ccwcer recunence LI my breast (L) Reason: > 0.7 Item-Item Correlation with Item Z thurk about breast cancer Lower 1-D comlation, Oniy Lumpectomy scalc 5. My mastecfomy Ls Ut the bock of my mind Reason: 1-D comlation under 0.4 6 . 1 wony about caracer recurrence ot my martectomy site ( M ) Rcason: > 0.7 Item-Item Correlation with Itcm I worry rhar the cancer is spreading Lower 1-D comlation, OnIy Mastcctomy Scale 7. I feel my body is abiè io &ht cancer Raison: 1-D correlation under 0.4 8. I feel my body has deceived nre Rcason: > 0.7 Item-Item corrclation witfi Item Zfeel my body has tumed againsr me Lower 1-D comiation, Grcakr Gcqucncy missing 9. I feel my body has tumed agczinsf me Raison: > 0.7 Item-Item correlation with Item I feel my body hm let me down Slightly lower 1-D comiation Wording las satisfactory

Appcndix 12 Continued

Domain: Predictability Coatinueci 10, I feel as llrough somethhg is gtawirig h i d e my body Reason: > 0-7 Item-Item corrtlation with Item Ifeel rhar somerhhg K raking over my 60& Lower I-D correlation

11. My body wïü tell me when somerhutg ir wrong Rcason: I-D corrclation under 0.4 12. l feet Ih4t cancer wül aLways be in rny sy-m Rcason: > 0.7 Item-Item correlation with Item Ifeel th& cancer ïs h d h g ih my b e Lowtr I-D corrclation

13.1 catch disease easüy Rcason: I-D correlation under 0.4 14, I om q f d of beconring fcrt Rcason: I-D corrclation under 0.4 15. I feetpart of my body is &sikg Rcason: > 0.7 Item-Item correlation with Item Ifeel ihuf pan of me mut remain hidden Lower 1-0 Comlation 16. I feel that cancer is hidihg in my body Reason: > 0-7 Item-Item corrclation with Item Ifeel fiasomerhhg is faking over my body Lower I-D correlation 17, I feel my body has beirayed me Raison: > 0.7 Item-Item corrclation with Item I feel rny body has Ief me down Lowcr I-D correlation 1 8 , l feel bad abou being ldiooedfor raduirion Reason: > 10% of responses missing I-D comlation undcr 0.4

19. I feel morked by my rudialion tu!îoos Rcason: > 10% of responsw missing I-D corrthtion under 0.4

Domain: Cornfort 1. I am able îo read n o m i & Reason: I-D correlation undcr 0.4 2. My a m bolhem me Reason: > 0.7 Item-Item comlation with Item A m pain ir a problem for me Lower I-D comhtion 3. 1 c m thiizk rzomally Reason: FuU range of responses not cndoned I-D correhtion under 0.4 4.1 have probLems wirh my atte~îionspon Raison: > 0-7 Item-Item correlation with Itcm I have problems concenfrafing Lower 1-0 correhtion 5. Pain d e c i s my &l'lin( & do my n o d acliv&s Rcason: Item-Item corrclation with Itcm My body srops mefiom doing fiings I wanf ro do blt 0.M.7 6. Nausea uzrerferes wirh my v e Reason: I-D comlation under 0.4 7 . 1 have problems with my memory Rcason: > 0-7 Item-Item correlation with Itcm I mnforgefi1 Lowcr I-Dcorrelation, Worst Distniution 8. 1 have physical discomJori dwulg s e x d uttercourse Reason: > 10% of rtsponscs missing 9. Being tired lUnils my activities Remon: > 0.7 Item-Item comlrrrion with Item Being tùed interferes with my Life Siightly less satisfactory wording 10,1 wony ubouî my appetiîe Reason: I-D comlation under 0.4 11. I feel sick to my stomach Rcason: I-D correlation under 0.4; Round 2

Apjxmdix 12 Continued

Domain: Cornfort Continued 12. My brearifeek heavp Rcason: 1-D correlation undcr 0.4 13. The weight of my prosîhesis bothers me Raison: > 10% of rtsponsts misshg 1-D correlation undcr 0.4 14. I fee L as g m y bre& ir SMthere Raison: 1-D correlation undcr 0.4 15. I feelpah where my bread should be Raison: Full range of nsponscs not endorscd 1-D correlation under 0.4 16. My underarnt feels normal Rcason: 1-D correlation undcr 0.4 17. Vaginal discharge ir a probkm for me Reason: 1-D comlation undtr 0.4 1%.My sense of smell is normal Reason: 1-D correlation undcr 0.4

19. I con larfefood nonna& Reason: 1-D correlation under 0.4 20. Vaginal drytress i s a problem for me Reason: 1-D correlation under 0.4 21. I am sat&fred wijh my qpe& Reason: 1-D correlation under 0.4

22. My prosthesk is cornforiable (M) Reason: > 10%of responses misshg 23. Breast pairrs are a p r o b k for me &) Reason: Item-Itemcorrelation with Item My breasf is painficl

to rouch bit 0.6 to 0.7

Mmin 1 Mmin 2

the wav thcu.rre

Man 4,23 4.15 1,38 1.37

2.17 2.23 1 A2

1.46 2.16

2.00

2.04 2.06 2,16 2,17

2,24 2.25 2.57

2.47 1.63 1.53 1.42 1.45

2.29 2.25 1.31 1.35 1-44 1.50 1.72 1 .7l 2.90 2.91

3.64 3.73

2.62 2.71 3,91 4.06 1,63 1,62

Appadix 14 Roud Factor Solution for Single Broail Subjccts (50 Cornmon Ilaai uid 7 S p i T i I

w

Ficior Poacrn FACTOR1 FACTOR2 FACTOR3 FACTOR4 FACTORS F A O R 6 ha

Factor 1 1 hide my body whm chuying clolha 1 foc1 ttut prct of me murt rcmnin h i d m 1 feel Ihilmy mu(eciomy n ugiy (M) 1 ûy to hide my body I fecl leu femiaiac mince canar 1 avoid lookLy mt my ran from bricut murgery 1 would avoid a aew rexuai W o a i b i p b u r e of my body 1 w d kotp my cbcd c o v d duMg uxud inI avoid phpial mtimrcy 1mdd oftouchiry tbamhrmi bnutiurgcry 1 feel nümdve wben I un nide 1 f d canforhbk wbea oihen rw my mriilcccomy (hi) 1 feel comfort.ble lookiq aî rny mm(cc(omy (M) 1 would f d d o r c . b l a chinging in a public cbuigeroom Fmor 2 wony rhri an= L rprcdtry f#l~iratimcbombinridoofme Waùwbrtulanccr feel my body hnr btai invdcd fcel my body bu kt ma d o m f d thai mm&6 L taking ovcr my body worry about my body

wony about minor actm ud pina ntcd reuruiance about my batth ftel proae & cancer Famr 3 1 ftcl hlpcoplc rn lookity r i me 1 fcel hipeople arc looking at my chai 1 chink my bttuti a p p a r m c v a to othen 1 wony n b u t tny prorihau or pdding mlipping (M) I ncad to bc r t u i u d about tbe appcomcc of my buni 1 foc1 poapb can teII my b m t r are not n o m 1 1 wony aboui îhc way 1 look 1 worry about my hair

for Two Brtut Subjccca (50 Commoo litmi d 12 Spwifw Itani

Fwr1 1 fœl comforiible whui othcn rte my brcuil (L) 1 would feel comfomble changing in a public c h i u i ~ w w m 1 un happy with the pooiîion of my nipplc (L) 1 fccl icxudy irinctivc whcn 1 un audc 1 am utirfied with I c a p p m c e of my b i r u i (L) 1 would f 4 aunfortable witb a u x u d m e r roucbky my b r a i t (L) 1 am d f i o d witb tbc a h of my brrrr( (LJ 1 focl d o r t a b l e lookin8 i( my brcvt (I$ 1 w w l d k c q my ch- w v c d dur@ acnuril kirimircy 1 ivoid clore phyaksi cootnct iuch uhuggiq 1 would tvoid r ncw drtlriiomhip h i i r a o f my body 1 mvoid phyrial intiaucy Tbe i p p a ~ c of e my brcam couid dbturb oihcn (iJ 1 fœl hot put of ma mwt runoin hiddcn 1 hido my body whai chnging ciotha

Factor 2 1 ihinL &out b n u t anccr 1 a d remsaurana about my btrhh 1 wony ch.r cracCr m r p d i n g 1 fœl th.1 m e t h i n 8 h taking ovcr my bod 1 fcel my body hu bwr m v d d 1 fœl iherc L time bomb hide of me 1 ftcl my body h u la me dawn 1 wony about m h r rchca and paina 1 wony mbui my body 1 fecl mgry i t my body 1 feel pmno to cancer 1 think about my b m t (L) 1 un ù w i y r awirc of my brasri (L) Facior 3 &hg t i d inlcrfcru wiih my lifc My body stop me frorn doing t h i g a 1 w ~to tdo Olhcn &ve hnd ta rikc ovcr my dutia 1 Mrlttpy during the day 1hnvc problcrm conmtmling 1 om forgelhl 1am anhiid wilh my ippearance ni rpccial ocaiona 1 can paniciptc in n o m 1 activitia 1 mm happy wiih my Icvd of cncrgy

ï l a n k you for completing the Body h g e in Brraot Cancer Q u u i i o ~ a i r c(cbe 'BIBC'), Your comments wil1 be emrrmcly ureftl in cnnrrbg Chat the questionnaire is relevant and acceptable to womcn wicb brcast cancer. Ptecornpeîe the following comment sheets and rcturn thcm

using the erivclope providcd. 1. How long did the questionnaire take to comptete? 2. How did you find the Iengih of the questionnaire?

-

3.

4. 5.

Werc the instructions clcar? Werr the instructions helpfiil? Was it easy to UK

minutes

-

Toohng Too Short About Right

-

No Yer

-

the acalu for your rcqoarer? -

No Ym

No Yes The questionnaire is supposcd to meaaire body image. Do you feel that the questionnaire reaUy did masure body image? 6. No ya Did you feci that there werc topics not c o v e d in the questionnaire which should have b a n covcred? 7. No Yes If yea, what were the topic rbat should have k n c o v e d ?

-

-

Did you feel that there were topics covercd in the quedoruraire which ahould mot have been coveccd? No Yes If yes, what werr the topic rbat shouId not have b e n c o v e d ? 8.

-

-

Werc any of the ~triremeatriin the questionnaire nrrlPPR No yKf yes, which Jtatcments were unclcar?

9.

-

Were any of the statemnts in the q u d o n n a i r c offensive? No Yer If yes, which statementa werc offensive? 10.

11.

-

Would you comptete the questionnak again if asked?

-

No y1would l i e to know a tiüle bit about you. This information is confideaial. 12, What is your age? Yearr Is English your native tongue? 13. No Yes What is your level of formal education? 14. Grade 8 or Less Som High School High SchooI Complettd Some Collcge o r Trade School Trade Cercificate or Coiiege Diploma S o m University Education, No Degrce University Degrec Professional or Post-ûraduate Degrte Thank you vcry much for your participation. Your f d b a c k WUbe e m m c l y usetU1. Pleasc rtnirn the comment sheet in the cnvctope providcd. You may also ceturn your completcd q u e d o n n a k if you wish. If you have any fiirther commcrirs, please use the following page to include them. Additional commcnts:

-

-

-

-

Appendix 17 Items Adrninistered to the Control Groups Listed by Subscale

Yen

Ya Ya No Ya

Yes Ya Yu

Appendix 18 Distribution of Scores on Q u ~ s ~ ~Used o M in Validity ~ Study Distribution of Scores for the Body Image after Breast Cancer Questionnaire

Appendk 18 Continueû Distribution of Scores for the RSE, the IES, the BDI and the MCSD Questionnaires 2s

,

-

IES l n ~ - o Sii)iurrk n

1

Appendix 18 Continued Distributions for the Derogatis Interview for Sexual Functioning Questionnaire

O

S 10 15 Sonwl ~ o

20 2s 30 35 uEtlicpaiaia l Score

Appendix 18 Continuai Distribution for the EORTC QLQ30

-

EORTC -al 140

FuncCidng -Ql)reale I

Appendix 18 Continued Distribution for the Multidimensional Body Self Relations Questionnaire

Appendix 18 Continued Distribution for the Multidimensiond Body Self Relations Questionnaire Continued

Appendix 19 Final Version of the Body Image afkr Breast Cancer Questionnaire INSTRUCTIONS - PLEASE READ CAREFULLY The following 4 pages contain statements about how people might think, feel, o r behave after to you ~~EQMUYover developing breast cancer. You are asked to indicate &e way the past month. Please read each statement carefdiy and decide how it applies to you. When answering, consider how you have felt Qver Your answers are confidentid so please do not &te your name on any of the pages. Using the scales listed below, indicate your answers by writing them to the lefi of the statements. There are two types of statements. For the f h t type of statement the following scale is used;

m.

1 = Strongly Disagree 2 = Disagree 3 = Neither Agree nor Disagree 4 = Agree 5 = Strongly Agree

Example

2

1.

Skin Dryness is a problem for me,

In the blank space enter 1if you strongly disagree with the statement, 2 if you disng~eewiîh the staternent, 3 if you neither agree nor disrigree with the statement, 4 if you agree with the statement and 5 if you strongly agree with the statement. In this case the answer is 2, the person disagrees with the statement.

In the second type of statement the following scale is used; 1 = Never/Almost Never 2 = Infiequentiy 3 = Sometimes 4=ûften 5 = AlwaysIAlmost Always

Example 1-

I can use my arm normally.

In the b1ank space enter 1if the statement is never or almost never true, 2 if the statement is infrequently true, 3 if the statement is sometimes m e , 4 if the statement is often tnie and 5 if the statement is aiways or iilmûst always true. In this case the answer is 4, the person can often use their arm normal1y. Remember that there are no right or wrong answers, just give the answer that is true for you over the past month. Some questions may seem to be more important to you than others. Try to answer all questions to the best of your abiity. There should be an answer that is m e for you. It is important that you answer every item. Please be completely honest. Your responses are confidential. Your name will never appear on this survey and once y o w survey is retumed, anything that could identiS, you will be destroyed.

TYPE ONE STATEMENTS RESPONSES 1 = StrongIy Disagree 2 = Disagree 3 = Neither Agree nor Disagree 4 = Agree 5 = Strongly Agree 1 try to hide my body.

The feeling in my arm is normal. 1 avoid looking at my scars from breast surgery. 1 feel there is a time bornb inside of me.

1 am sleepy d u ~ the g day. 1 am happy with my Level of energy. 1 feel prone to cancer. 1 am satisfied with the shape of my body.

1 feel less ferninine since cancer. 1 iike my body.

1 feel comfortable about the way 1 look when 1 exercise. 1 would feel comfortable changing in a public change-room. 1 feel my body has been invaded.

I am satisfied with the appearance of my m. 1 feel my body has let me down. 1 like my looks just the way they are.

Others have had to take over my duties. 18.

1 feel that part of me must remain hidden.

TYPE ONE STATEMENTS

RESPONSES 1 = Strongly Disagree 2 = Disagree 3 = Neither Agree nor Disagree 4 = Agree 5 = Strongly Agree 19.

1 am afraid of touching the scars from breast surgery.

20.

I am satisfied with the appearance of my hips.

2 1.

I avoid close physical contact such as hugging.

22.

I feei that something is taking over my body.

23.

I am satisfied with the shape of my buttocks.

The following questions pertain to your feelings about your breast or mastectomy site. If you are missing a breast(s) (if you have had a mastectomy without breast reconstruction) please answer question 24. If you are not missing a breast (if you have had a lumpectomy, a mastectomy with breast reconstruction, or no surgical treatment to your breasts) please skip questions 24 and answer questions 25 to 27.

The foliowing item should be answered by women who are misshg one o r both breasts. 24.

1feel cornfortable looking at my mastectomy.

The following items should be answered by women who are not missing a breast. 25.

I am happy with the position of my nipple.

26.

1 am satisfied with the size of my breast.

27.

I feel cornfortable when others see my breasts.

28.

The appearance of my breast could disturb others. & mt section contairu TYPE TW0 statements

TYPE W O STATEMENTS 1 2 3 4

5

RESPONSES = NevedAlmost Never = Infrequently = Sometirnes = Often = AlwayslAlmost Always

I feel that people are looking at my chest. 1avoid physical intimacy.

1 feel that people are looking at me. 1hide my body when changing clothes.

1 worry that the cancer is spreading. 1need to be reassured about the appearance of my bust.

1 think about breast cancer.

Being tired interferes with my life. 1feel sexually attractive when I am nude.

S w e b g of my arm is a problem for me. 1worry about my body. 1would keep my chest covered during sexual intimacy.

1 feel angry at my body.

1 ne& reassurance about my health. 1 on participate in normal activities. 1have problems concentrating.

My body stops me from doing things 1want to do.

TYPE TWO STATEMENTS RESPONSES 1 = Never/Almost Never 2 = Infrequently 3 = Sometimes 4 = Often 5 = Always/Almost Always

I think my breasts appear uneven to others. A m pain is a problem for me.

1worry about minor aches and pains.

I feel normal.

I feel people can tell my breasts are not normal. The foiiowing questions pertain to your feelings about your breast or mastectomy site. If you are rnissing a breast(s) (if you have had a mastectomy without breast reconstruction) please answer question 51. If you are not missing a breast (if you have had a lumpectomy, a mastectomy with breast reconstruction, or no surgical treatment to your breasts) please skip questions 51 and answer questions 52 and 53.

The foliowing item should be answered by women who are missing one or both breasts. 5 1.

1 worry about my prosthesis or padding slipping.

The following items should be answered by women who are not missing a breast. 52.

1 think about rny breast.

53.

My breast is pauiful to touch.

Thank you for completing the questionnaire. Your time and effort are greatly appreciated Your thoughts about the survey are important. If you have any geneml comments or comments on specific statements, we would appreciate you taking the time to write them down in the space provided (or continue on the back if necessary). Thanks again!

IMAGE WALUATION TEST TARGET (QA-3)

APPLIED

IMAGE.lnc

-.-----,

1653 East Main Street Rochester. NY 14609 USA Phone: 716/4821)30 Fax: 716/288-5989

Q 1993. Applied Image. lm.Ail Rights Resenred