Although several cancer related body image scales have been developed,1-4 most have not been validated in long-term breast cancer survivors or in women who underwent both mastectomy/lumpectomy and chemotherapy. This new body image scale builds on previous work by incorporating body image into both a quality of life framework with three domains (physical, psychological, and social) and a global measurement of well-being based on a breast reconstruction framework.5
Objective
The purpose of this study was to develop and psychometrically test a body image scale for breast cancer survivors. Psychometric analyses were guided by a theoretical framework that defines associations between body image and proximal (physical, psychological, and social) and distal (Index of Well-Being) quality of life outcomes. Research hypotheses guiding analyses included the following:
- Internal consistency reliability will be measured in the following ways:
- The Cronbach alpha for body image will be 0.8 or above,
- inter-item correlations will be less than 0.7,
- and the corrected item-total correlations will exceed 0.4 for all items.
- Construct validity will be demonstrated by the following:
- Exploratory factor analysis will identify a unidimensional latent variable,
- Body image will be significantly associated with constructs in the physical, psychological, and social quality of life domains,
- Proximal variables and body image will predict overall quality of life.
Methods
Data for this study were obtained via mailed questionnaires and telephone as part of a larger quality of life study that compared breast cancer survivors (n=1127) with acquaintance controls (women who were not diagnosed with breast cancer).6 After items were initially developed by modifying existing body image instruments, they were reviewed by two focus groups (composed of 6-8 breast cancer survivors at least 3 years post-treatment) for relevance, clarity, and inclusiveness of issues BCS face during survivorship. To establish content validity, content experts provided judgment of the items confirmed by the focus groups. Based on comments from content experts, three items were deleted and item index of content validity (item-CVI) was assessed and found to be 1.00 (an excellent modified kappa statistic).
Measures
Physical Quality of Life Domains
The 10-item Physical Functioning Scale (PF-10) measures physical activity limitations,7 and the 13-item Functional Assessment of Cancer Therapy–Fatigue (FACT-F) measured fatigue.8
Psychological Quality of Life Domains
The 20-item Center for Epidemiologic Studies–Depression (CES-D) Scale measured depression.9 The State-Trait Anxiety Inventory (STAI) measured anxiety with two self-report scales measuring state and trait anxiety.10 The Concerns About Recurrence Scale (CARS) was developed specifically for breast cancer survivors.11 The CARS includes four items assessing overall fear of recurrence, and twenty-nine items grouped into four subscales that assess the reasons for anxiety regarding recurrence, including worries about death, health, role, and womanhood.
Social Quality of Life Domains
The 15-item ENRICH Marital Satisfaction Scale (ENRICH-MSS),12 measured marital satisfaction including intrapersonal issues such as communication, conflict resolution, sexual satisfaction, commitment, and roles. The 7-item Sexual Index Scale measured sexual functioning. The Sexual Index Scale was modified from existing instruments and was composed of subscales for sexual difficulty and sexual enjoyment. The Northouse Spouse/Support measures the relationship between a spouse and breast cancer survivor regarding the breast cancer experience.13
Distal Quality of Life Domain
The Index of Well-Being (IWB) is a 9-item scale often used to measure global well-being and been used as a measure of quality of life in breast cancer survivors.14
Data Analyses
Reliability. Internal consistency reliability was measured with Cronbach alpha coefficients, inter-item correlations, and corrected item-total correlations. Cronbach alpha coefficients above 0.8, inter-item correlations below 0.7, and item-total correlations above 0.3 and below 0.7 were considered adequate.
Validity. Construct validity was measured through exploratory factor analysis and the correlation of body image with quality of life domains. A common-factor model was estimated using squared multiple correlations as the initial communalities and the principal component method of estimation. Rotation was not used because only one component was extracted. Factor loadings above 0.4 were considered adequate.
Pearson correlation coefficients of body image with selected measures (PF-10, FACT-F, CES-D, STAI, CARS, ENRICH-MSS, Sexual Index Scale, and Northouse Spouse/Support) of each quality of life domain (physical, psychological, and social) were first estimated to determine the relationships between body image and quality of life variables. A linear regression measured the impact of covariates, body image, and proximal quality of life variables on overall quality of life (as measured by the Index of Well-Being).
Results
Reliability. Individual items of the body image scale were assessed for high inter-item correlations (above 0.7) that would indicate redundancy of items. One item had an inter-item correlation of 0.7 or greater with other items of similar content. The redundant item was, “I am satisfied with the appearance of my body.” Internal consistency reliability analysis was conducted with the redundant item deleted and a Cronbach alpha of 0.88 was obtained. Item-total correlations ranged from 0.367-0.829.
Validity. Exploratory factor analysis revealed Eigenvalues for the first four factors as 4.602, 0.942, 0.675, and 0.608, indicating a marked leveling off of Eigenvalues after the first factor in the scree plot. The first factor explained 93% of the variance shared between the seven items and 57.53% of the total observed variance. The factor loadings for the one-factor solution ranged from 0.446–0.889, well above the 0.4 threshold, indicating a one-factor measurement model for the body image latent dimension.
Construct validity was tested with Pearson’s correlations by correlating each of the dimensional variables with body image unadjusted and then adjusted for covariates. Body image was a significant predictor of well-being even after adjusting for demographic covariates and other domain-specific quality of life variables. Judging from the t and p values, the variables that were most strongly associated with well-being were physical functioning (t=1.998, p=.046), body image (t=2.775, p=.006), worry about recurrence (t=2.776, p=.006), trait (t=-6.424, p=.000) and state anxiety (t=-2.447, p=.015), depression (t=-1.931,p=.054) and spousal support (t=5.902, p=.000). The combination of all variables in the model explained a substantial amount of the variance [F(14,784) = 48.35, p<0.000] in well-being.
Conclusions:
The body image scale demonstrated high internal consistency reliability and construct validity as demonstrated through factor analyses and theoretical prediction. We found that body image is related to physical functioning, fatigue, depression, fear of recurrence, anxiety, sexuality, marital satisfaction, and well-being; thus this concise and easily administered scale could be used clinically to identify survivors who have symptoms related to these concepts. The scale could also be used to determine effectiveness of interventions targeting survivor’s perception of their body following treatment. This new body image scale could be used as a screening tool during a primary care visit. If a survivor indicated that she had problems related to body image (e.g. feeling bad about her body), interventions on this construct could improve other symptoms, such as anxiety, depression, fatigue, or sexual/relationship dysfunctions.
See more of: Invited Posters