Validation of the instrument “ Body image after breast cancer ” in Brazil

The body image is an important aspect to be considered during the treatment of women with breast cancer. Therefore, we understand the importance of the quality of an instrument that evaluates this condition. The instrument validated in this study can bring new possibilities of studying the effects of cancer treatments on body image. Validation of the questionnaire, translation, cultural adaptation and pre-tests were done with women with breast cancer residents of the States of Alagoas, Pernambuco and São Paulo. A Confirmatory Factor Analysis (CFA) demonstrated that the original model of the instrument is valid, reliable and can be an important tool for identifying the impact of breast cancer on body image of women with this health condition, allowing adjustments in clinical treatment, psychological support and actions of other professionals involved in the treatment.


Introduction
Body image can be deeply impacted by psychological and physical factors which can cause damage to the self-identity of a person.Especially for oncological patients, changes caused by both, the diagnosis of cancer and its treatment can affect these women functionally and emotionally-most of the time in an irreversible manner (Annunziata, Giovannini, & Muzzatti, 2012;White & Hood, 2011).Changes in appearance and body functions, such as scars, hair loss, rashes, amputation of the breast, fatigue, and impact on sexuality are some of the elements that affect the per-ception individuals have of themselves.Secondary dysfunction in a woman's sexual life can be caused specifically by three factors: the decrease of self-esteem, impaired reproductive ability, and the side effects caused by the treatment and /or surgery (APA, 1994;Baron et al., 2000;Luskin, Warner, & Knapik, 2006;Vamos, 1993).
In oncology, body image must be studied in a multi-dimensional way.Specifically for women with breast cancer, doctors and researchers have to understand the meaning of the breast in a patient's life, their concerns regarding appearance and body functions, their beliefs, feelings, behaviors and life experiences, rather than just a diagnosis at the end of a treatment (White & Hood, 2011).
Body image is considered an important component in the quality of life in patients with breast cancer (Hopwood, Fletcher, Lee, & Al Ghazal, 2001).Cash and Pruzinsky (2002) consider that the investigation of body image is associated with quality of life, and researchers must take into account particular effects of a disease or specific treatments on body image.Another issue to consider is that studies on quality of life often associate the assessment of body image to sexuality.Although sexuality is related to body identity, it cannot be forgotten as each type of cancer is associated with specific body image concerns, and sexuality will not always be the principal focus.This situation does not allow a researcher to say that for each patient body image representation is preserved as before the diagnosis (Cash & Pruzinsky, 2002;Hopwood et al., 2001;White, 2002).
Psycho-oncology research works on these issues in order to develop specific instruments for assessing body image in cancer patients.For women with breast cancer, the following questionnaires are available: (a) The Body Image and Relationship Scale, developed by Hormes et al. (2008); (b) The Sexual Adjustment Scale developed by Dalton et al. (2009); (c) The Body Image Scale, created by Hopwood et al. (2001); and (d) The Body Image after Breast Cancer Questionnaire, created by Baxter et al. (2006).The last one, the "Body Image after Breast Cancer Questionnaire" (BIBCQ; Baxter et al., 2006) is the focus of our research, because it meets the multidimensional perspective required in the research of oncology psychosocial features.Being multidimensional, the BIBCQ has six different factors (or dimensions): vulnerability (V), transparency (T), body stigma (BS), arm concerns (AC), body concerns (BC) and limitations (L).Therefore, the scale was designed to cover different aspects of the body image in an oncologic perspective, and allows us to monitor the long-term impact of breast cancer in women.The instrument is founded in theoretical basis with direct information about the target population, and also is grounded on earlier information in literature about body image in people with chronic diseases (Baxter et al., 2006).
In Brazil, breast cancer accounts for 22% of new cancer cases each year, with a very high mortality rate, due to its diagnosis in advanced stages.Women who survive undergo aggressive treatment and suffer with procedures that changes appearance and body functions (INCA, 2011).The literature on breast cancer in Brazil indicates that the diagnosis of breast cancer is followed by anguish, suffering, and anxiety throughout the treatment, financial insecurity caused by the cost of drugs, surgery, hospitalization, depression and a decrease in self-esteem (Venancio, 2004).The high incidence of the illness and the disruption diagnosis and treatment of breast cancer (during and at the end of the protocols) cause enormous impact in women's lives.Emphasis on understanding these changes caused by treatments would help planning interventions that may improve survival rate after cancer remission.Therefore, a specific instrument to evaluate body image after breast cancer would give researchers and doctors some relevant information about their patients' psychological health and wellbeing, and the impact of any intervention.
Base on these arguments, the purpose of this study is to investigate and cross-culturally adapt the psychometric properties of the Body Image after Breast Cancer Questionnaire (Baxter et al., 2006) to Brazil's Portuguese language.This tool will safely assess the impact of treatment and intervention programs to support women with breast cancer.Around in the world, research on breast cancer provides its contribution to possibilities of cross-cultural influences.Recently, a study from the Western University of England has investigated the psychosocial impact of breast cancer in South Asian and Black women (Patel, 2013).The study assessed the body image through the "Body Image after Breast Cancer Questionnaire" and the results have shown that the experiences due to breast cancer diagnosis and treatment can be influenced by different ethnicities/cultures.The author also has investigated women with no proficiency in English and in need of assistance of translators due to the absence of a validated questionnaire in the language of the target group.This study reinforces the importance of cross-cultural studies, and specially the validation of the questionnaires in different languages.

Participants
The sample is comprised of a non-probabilistic selection of 393 women, who attended treatment centers in several cities in the Southeastern and Northeastern region of Brazil.Patients' average age was 54.6 years (SD = 12.01), and they were diagnosed at 65.20 months (SD = 62.62).Most of the women reported being sexually active (59%), 59% were married, 15.4% single, 13.3% widowed, and 12% divorced.Most women (93.2%) have only one breast affected by cancer.A mastectomy with reconstruction surgery was the most frequent procedure (55%), followed by lumpectomy.In order to avoid research bias, the study did not include illiterate women or women who were undergoing chemotherapy during the data collection period.We believe that women undergoing chemotherapy could experience mental confusion, and illiterate women could misinterpret the questions if somebody read the questionnaire for them.

Instruments
The instrument Body Image after Breast Cancer Questionnaire -BIBCQ (Baxter et al., 2006), is a multidimensional scale designed to assess the impact of breast cancer on body image.In the original study, the questionnaire was answered by women who were treated at the University of Toronto Hospital, Ontario, Canada.The questionnaire has 45 general items, two optional items for women who have removed one or both breasts, and six items for women who did not remove either breast.
The BIBCQ questionnaire consists of six scales: (1) one scale of vulnerability (V), which assesses the sense of vulnerability of the body to other diseases and cancers; (2) a scale for body stigma (BS), which assesses the need to hide the body; (3) a scale of limitations (L), which assesses feelings of competence and skills; (4) a scale of concerns with the body (BC), which assesses satisfaction with appearance and shape of the body; (5) a scale of transparency (T), which assesses concerns with the changes (ob-viously cancer-related) in appearance, and (6), a scale of concerns with the arm (AC), which assesses the worries with symptoms related to the arm and its appearance (Baxter et al., 2006, p.222).
Questions 1 to 28 include an agreement scale of 5 points (1 = strongly disagree and 5 = strongly agree).Questions 29 to 53 use a frequency scale of 5 points (1= never to 5= always).In the demographic questionnaire, participants report their age, marital status, sexual activity, and their perceptions about quality of sexual life, as well as their perceptions about financial security.These two variables include the following crude measures: (1) "Evaluate your sexual activity in the last 30 days from 0 to10" (0 = extremely poor to 10 = very satisfactory), and (2) "How financially secure have you felt for the last 30 days"? (1= unsure to 4 = safe).

BIBCQ cross-cultural adaptation
The Portuguese version of the BIBCQ questionnaire was based on the studies of Beaton, Bombardier, Guillemin and Ferraz (2002) and Ferreira, Campana, Campana and Tavares (2011).First, the scale was translated into Portuguese by two native Brazilian individuals who were fluent in English.The translators worked independently of each other, producing two different translations (T1 and T2).A simplified version (T12) was produced by the agreement of the two translators coordinated by a neutral judge.Later, two independent translators (BT1 and BT2) translated the questionnaire back into English, resulting in a simplified version (T12).Both translators were native English speakers, but fluent in Portuguese.The complete material (T1, T2, T12, BT1 and BT2) was submitted to a committee of experts that included an oncologist, a physical therapist who was trained in psychoanalysis, a methodologist, a linguist and a member of the target population of this study.The committee prepared the pre-test version with a scale that observed semantic, idiomatic, conceptual and cultural aspects.This version was assessed by six participants for clarity of the items and their instructions, the alternatives for responses, and the layout.Situations for misunderstandings were identified in the instructions, and items 7, 11 and 32 (semantic aspects), and also in the layout.Revision of statements was discussed with the members of the committee of experts, and then, nine participants were interviewed.All critical issues were resolved, except item 7.This item was revised by members of the committee of experts and a new simplified version was proposed.In the third pre-test, all of the items were understood by the participants regarding instructions and response options.The layout was approved as well.After this process, the Brazilian version of BIBCQ was ready for data collection.

Procedures
Prior data collection, this study was approved by the Ethics Committee of the University of Campinas (date of approval: June 22, 2010; approval number: 503/2010), as well as by each ethics committee of each participating hospital/clinic.Participants were invited from cancer treatment centers of several cities of Northeastern and Southeastern States in Brazil.The same researcher collected all the data for this study.The researcher personally invited patients to participate in the research, and explained its purpose.
Both the Consent and Awareness form and the BIBCQ questionnaire had to be read and answered by the patient, without the help of others.The time to complete the questionnaire varied greatly among participants, according to their educational level.The women who attended elementary school took an average of 30 minutes to respond the questionnaire and form.Those with high school or college degree took in average 10 minutes to fill out the questionnaire.Participation in the research was non-paid.
Model adjustment was assessed by multiple indicators: Goodness-of-Fit Index (GFI), Adjusted Goodness-of-Fit Index (AGFI), Normed Fit Index (NFI), Non-normed Fit Index (NNFI), and Comparative Fit Index (CFI).According to the literature (Hu & Bentler, 1999), these indexes are expected to be equal or above 0.95.The Root Mean Square Error of Approximation (RMSEA) was also considered with acceptance values below 0.08.In the present study, ideal values for standardized Chi-square should be smaller than 5 (Maruyama, 1998).Initial models with poor adjustment, factorial loadings, residual items, and LISREL modification rates were submitted to further adjustments.Based on Dr. Baxter recommendations, 45 common items were submitted to analysis.Items 24 and 51 were specific to women with mastectomy, without reconstruction; and items 25, 26, 27, 28, 52 and 53 were specific to women who did not undergo surgery, or had a lumpectomy, or a mastectomy performed with reconstruction.Due to specificity of these items, their inclusion on the CFA would generate an awkward matrix.After adjusting the general model, we used Spearman correlation to evaluate those items that could be in the scale that uses the total-item score correlation, and that was considered statistically relevant for correlations above 0.30 (Nunally, 1978).
To establish the internal consistency, Cronbach's alpha and construction reliability (Hair Jr. et al. 2005) were calculated.Reliability of the construction (RC) was calculated as follow: Where λ j is the standardized factor load (assertion) of the j indicator, and E j is the measurement error of the j indicator, calculated as 1 -reliability of the indicator.Values greater than or equal to 0.70 were accepted.
To establish the convergent validity, t-values and factorial loads of the observable variables were analyzed.Item factorial loads greater than 0.50 were desired, however, due to our sample size, values of 0.30 were accepted (Hair et al., 2005).T-values higher than or equal to 1.96 were considered acceptable (Garver & Mentzer, 1999).

Hypotheses
As part of the construction validation using a nomological network approach, we hypothesized that: 1) Considering the demographics of the sample, the BIBCQ factor related to body concerns would be associated with sexual life (Preston, 2010;White, 2000;2002).
2) Compared to the rest of the sample with lower levels of sexual activity, the BIBCQ would differentiate groups according to their sexual activity (APA, 1994).
Our analytical decision for accepting the covariance errors between items was based on LISREL's modification indices.Common reasons for this decision include item redundancy (caused by similar content or social desirability) and/or omission of an exogenous factor (Schumacker & Lomax, 2004).Acceptance of a covariance error should be justified on theoretical grounds, rather than sole on statistical criterion (e.g., enhancement of a model adjustment; Silvia & Maccallum, 1988).It is unlikely that social desirability is a factor associated to error covariance, since we ensured conditions for reducing bias (voluntary and anonymous participation) during data collection.It is also unlikely that a latent variable (factor) was ignored, since similar models were already proposed and investigated in previous studies.It is, therefore, possible that the existence of an error covariance was due to similar content between items, meaning that the participants perceived the content of the items to be similar.

Internal reliability
Cronbach's alpha test and construction reliability values are presented in Table 2.
The bivariate correlation Spearman test was used to assess relationship between sexual life satisfaction and the scores related to body stigma and the scores related to body concern factors.A negative correlation was found between sexual life satisfaction and both scores.Women who underwent a mastectomy without reconstruction showed significant correlation between satisfaction with sexual life and the scores of body stigma factor (r= -.44, p <.0001), and women who underwent a mastectomy with reconstruction, or lumpectomy, or had no surgery exhibited significant correlation between satisfaction with sexual life and the scores of body stigma factor (r= -.21, p <.0001).However, a negative correlation was found between satisfaction with sexual life and the scores of the body concern factor (r= -.21, p <.0001).
With regard to discriminant validity, evidences suggest that the studied scale successfully discriminates the sampled women for changes in sexual life, and for different perceptions of financial security.There were statistical differences between scores of the body concern factor (U= 15176, p= .016,r= -.12), limitation (U= 14535.5, p= .003,r= -.15) and body stigma, both for women who had mastectomy with reconstruction, lumpectomy or had no surgery, and for those who underwent mastectomy without reconstruction (U= 14535.5, p= .003, r= -.18).
Considering the four levels of financial security perception (insecure, a little insecure, a little safe, and safe) for both women who had mastectomy without reconstruction, and for women who had mastectomy with reconstruction /lumpectomy /no surgery, financial security was statistically different for scores of all factors (Table 3)., 3, 9, 12, 18, 19, 25, 26, 27, 28, 30, 32 and 40.For the V factor, the statements are: 4, 13, 15, 22, 39, 41, 42, 48 and 52.For the T factor: the statements are: 29, 31, 34, 46 and 50.Values placed on the right refer to the internal reliability factor calculated with the statements that comprise the factor for the M group.For the factor BS, the statements are: 1,3,9,12,18,19,24,30,32 and 40. For the V factor,the statements are: 4,13,15,22,35,39,41,42 and  This questionnaire could also provide information about the differences between women who underwent breast reconstruction of women who did not lose either breast, and women who did not undergo reconstruction surgery.The questionnaire scales (vulnerability, body concerns, limitations, transparency, arm concerns and body stigma) can be used separately to evaluate a specific factor.
The BIBCQ questionnaire is an important tool for research in the health area and can be used by doctors, nurses, physiotherapists, physical educators and other professionals.Also, the BIBCQ questionnaire is an important tool to evaluate attitudes and perceptions towards the disease or its consequences (surgeries, collateral effects), to measure and follow the rehabilitation results related to body image or compare the effects of different kinds of exercise programs focused on body image (on a short or long term basis).
Nevertheless, a number of limitations must also be highlighted.This is the first psychometric study of BIBCQ in Brazilians, and it is limited to a non-probabilistic sample, so it does not represent the whole population.Future work in Brazilians and people from other cultures could offer a better understanding of body image in breast cancer survivors.Other psychometric properties such as, stability, time, predictive validity, and other evidence of construction validity should be investigated in future research.For developing countries, creating a reduced version of the scale would be interesting to ensure interviews with illiterate and operationally literate women.These women usually have limited access to health care, which make researchers' efforts to better understand them a priority.As a group of additional risk, unprivileged women who are survivors of breast cancer need specific actions to facilitate their process of identity reconstruction.
Despite these limitations, the study is a reliable scale and confirms its validity for assessing the impact of breast cancer on body image of Brazilian women.

Discussion
The results of this study suggest that the Brazilian version of BIBCQ is a valid and reliable instrument to assess body image in women with breast cancer.The results confirmed the six factor model proposed by Baxter et al. (2006), including specific issues exhibited by women who underwent mastectomy without reconstruction, or lumpectomy, mastectomy with reconstruction or for those who did not perform any surgery.
The BIBCQ have adequate levels of internal consistency and construction validity.Our results suggest that the quality of sexual life is associated with some aspects of body image.Our results support the notion that active sexual life is a reflection of personal action and resources (such as resilience) which help to maintain a cohesive body identity (APA, 1994).As suggested by Hilton (1988), Stanton and Snider (1993), Lam and Fielding (2003), the financial security of women who survived breast cancer can be an important variable.The differences found between scores of groups with different perceptions of financial security also suggest that financial resources to purchase medication and hire professional assistance may have helped women during an early or a late recovery, making the structuring of identity after breast cancer an easier process.
The elimination of nine items of the original questionnaire could be due to translation problems.However, it seems unlikely, given the fact that an extremely careful process was performed to make sure the Brazilian BIBCQ version was equivalent to the original version in its idiomatic, semantic, cultural and conceptual terms, and to ensure that the version would be understood by the target population.It is more likely that cultural differences may have caused this situation.Some experiences with breast cancer are specific to Brazilian women.In this perspective, the removed items can be regarded as less relevant for analyzing the impact of breast cancer on body image of Brazilian women.In fact, other studies found differences in scales produced in other cultures when validated for Brazilians (Campana et al., 2009;Gouveia, Singelis, & Coelho, 2002;Swami et al., 2011).
In general, the BIBCQ questionnaire could represent the insertion of findings for the Brazilian population in the international scenario of body image studies.Some studies have investigated the behavior towards cancer in people from

Table 1 .
Adjustment values in the BIBCQ questionnaire scale.

Table 2 .
(Patel, 2013)factor, the statements are:29, 31, 34, 46, 50 and51.**Calculatedonlywith statements common to all respondents.Source: LISREL ® 8.51 e SPSS 15 Values of the reliability tests.differentculturessuchas, North Americans, Europeans, Australians, Japanese, Chinese, Koreans and Africans(Patel, 2013), but seldom with Brazilians.A questionnaire validation for the Brazilian people could enable cross-cultural research.It also could expand cross-cultural research on women's health and, more specifically, on breast cancer.

Table 3 .
Comparison of factor scores in relation to financial security.