Study setting and design
A descriptive cross-sectional study using quantitative method was conducted. Data were collected in the three Swedish cities Falun, Gävle and Uppsala. They are the capitals of Dalarna, Gävleborg and Uppsala counties with populations of 280 000, 76 000 and 376 000, respectively, and with high mortality of women with breast cancer .
Women with breast cancer were selected by use of the registrations of the Regional Cancer Centres (RCCs) in Uppsala and Örebro. The RCCs were established to build up national cancer registers for notification, planned therapy and follow-up . The criteria for selection were: women (1) with breast cancer diagnosis since at least one year, (2) having undergone mastectomy, (3) of age at least 18 years, (4) living in Falun, Gävle or Uppsala, and (5) willing to participate in the study. In total 481 out of 975 eligible women participated in the study and returned their questionnaires.
The questionnaire comprised four parts devoted to (1) socio-demographic characteristics, (2) sources of information, (3) body image, and (4) life satisfaction. The first two parts, developed by the authors for this study, are provided as Supplementary File 1.
The socio-demographic characteristics concerned age, civil status, educational level, religion, cultural/ethnic minority, underlying disease (chronic), duration of diagnosed breast cancer, types of treatment (mastectomy, chemotherapy, radiation therapy, Herceptin (HER-2), and hormone therapy), and breast reconstruction.
The sources of information concerned information support from physicians, nurses, internet, partner, family and friends, and other sources. It had 54 questions, nine for each source. This part was developed for this study by the investigators. Each question provided a score of zero or one. Therefore, the score ranged from zero to nine for each source of information, and the total score for sources of information ranged from zero to 54, a higher score indicating a larger number of sources of information. This part had Cronbach’s alpha = 0.89.
Body image was evaluated by using the Body Image Scale (BIS) , a ten-item scale with four possible responses: 0 (not at all), 1 (a little), 2 (quite a bit), and 3 (very much). The range of possible scores was from zero to thirty, a higher score indicating more dissatisfaction with body image. This scale was translated forward-backward (i.e. from English to Swedish and from Swedish to English) and tested for validity by the investigators and rechecked by the two research experts (one was an oncology nurse and another was a psychologist). It had been tested for reliability by the authors on breast cancer patients not participating in this study with Cronbach’s alpha = 0.93.
Life satisfaction was measured by using a life satisfaction questionnaire (LSQ) . It was constructed to measure life satisfaction/quality of life in women with breast cancer. It had 34 items with six dimensions: physical symptoms, sickness impact, quality of everyday activities, socio-economic situation, quality of family relation, and quality of close- friend relationship. Each item had a 7-point scale, ranging from 1 to 7. An example of an item is “How much have you been troubled by tiredness during the last week?” Its scale is: 1 to a very high degree, 2 to a high degree, 3 to a fairly high degree, 4 to some degree, 5 to a low degree, 6 almost not at all, and 7 not at all, the last alternative representing the highest satisfaction. The raw scores of the items were added, divided by the highest point in that scale and multiplied by 100. This normalization makes it possible to compare factors with different numbers of items in their dimensions, and 100 represents the maximum quality of life in each dimension. A higher score indicated a better life satisfaction. The scale had been tested by the authors for reliability on breast cancer patients not participating in this study with a Cronbach’s alpha ≥ 0.70 for each dimension.
We created hypothetical Directed Acyclic Graphs (DAGs) based on reviewed literature [21, 25, 26] with the aim to demonstrate what possible factors are associated with body image (Fig. 1a) and with life satisfaction (Fig. 1b). A maximum of 30 participants per independent variable was considered suitable for test parameters with a power of 80% . As there were 12 possible factors from the DAGS, the minimum number of participants was 360. To prevent missing data, data had been analysed from the women participating in the study.
After the heads of the clinics of surgery/oncology and plastic surgery in Falun, Gävle and Uppsala had been informed about the research project, they gave permission to conduct it. Also the nurses of the clinics were informed about the study in order to be able to answer questions from the participants. Written information about the study and its purpose, a consent letter and a questionnaire were sent to the selected women by post. They were assured anonymity and confidentiality and were told that they could drop out at any time. The ethical requirements of the Declaration of Helsinki-Ethical Principles for Medical Research Involving Human Subjects were fulfilled. Each questionnaire had a code number to facilitate reminders. The women interested to participate signed the consent letter, answered the questionnaire and returned these documents in a pre-stamped envelope, while those who did not want to participate returned the consent letter and the questionnaire without filling them out. A maximum of two postal reminders were sent after two weeks and one month if the women had not returned the envelopes.
We analyzed data using descriptive and inferential statistics. Descriptive statistics were used to summarize socio-demographic characteristics of the participants by, e.g., frequency and mean. Inferential statistics applied correlation and linear regression analyses. The level of statistical significance for all analyses was set at p = 0.05. Pearson’s correlation was performed to determine the differences between BIS scores and LSQ scores from six dimensions.
Multiple linear regression analyses were used to estimate relationships between socio-demographic characteristics, each kind of information support, total sources of information, treatment variables and outcome variables (i.e. body image and life satisfaction). The outcome variables were continuous variables, where the BIS score represented the body image and the LSQ score represented the life satisfaction. The LSQ score had seven dimensions: physical symptoms, sickness impact, quality of everyday activities, socio-economic situation, family relation, close-friend relationship, and overall life satisfaction. Therefore, there were eight outcome variables. Assumptions were satisfied before the analyses (e.g., auto-correlation, multi-collinearity, homoscedasticity, linearity, and multivariate normality). Socio-demographic variables included age, time since diagnosis, underlying disease, culture/ethnic minority, civil status, education level, and treatment variables including chemotherapy, radiation therapy, hormone therapy, Herceptin, and breast reconstruction. Age, time since diagnosis, each information support and total sources of information were continuous variables. Dummy variables (categorization to zero and one) were underlying disease, culture/ethnic minority (no = 0, yes = 1), civil status (married/lived together = 0, the others = 1), education level (high school or above = 0, secondary school/others = 1), and treatment variables (no = 0, yes = 1). First, we inserted each socio-demographic characteristic, each information support, total sources of information and treatment variable into simple linear regression for each outcome variable. Significant socio-demographic characteristic, information support and total sources of information and treatment variables from the simple regression retained in the multiple stepwise linear regression analyses. We adjusted the civil status and education level as confounders in the multiple regression for all outcome variables. We provided adjusted R2 and a standardized partial regression coefficient (β) and 95% confidence interval (CI) to demonstrate the fitness and strength of association of each outcome variable.