Study setting and design
A descriptive cross-sectional study was carried out in three cities in Sweden: Uppsala, Gävle and Falun.
Participants
Based on the Regional Cancer Centre (RCC) in Uppsala and Örebro, registered women were invited to participate in the study. The inclusion criteria were: women who (1) had been diagnosed with breast cancer at least one year before data collection, (2) were at least 18 years old, (3) lived in Uppsala, Gävle or Falun, and (4) were willing to participate. Women who reported a history of mental disorder or dementia were excluded. In total, 481 out of 975 eligible women with breast cancer agreed to participate in the study.
Instruments
Use was made of a questionnaire containing four parts: (1) sociodemographic characteristics, (2) social support, (3) psychological distress, and (4) HRQoL. Sociodemographic characteristics concerned age, marital status, education, religion, belonging to a cultural/ethnic minority, having an underlying disease, duration of diagnosed breast cancer, methods of treatment (e.g., chemotherapy, radiotherapy, Herceptin, and hormone therapy).
The part social support concerned six sources of information support, viz. physicians, nurses, the Internet, partner, family members and friends, and the patient’s institution. It was created by PCL. Each source comprised nine questions, each of which gave a score of zero if the answer was “No” and one if the answer was “Yes”. Therefore, each source could give a total score ranging from zero to nine, and a higher total score indicated more information support. This part had a Cronbach’s alpha coefficient of 0.89.
The part psychological distress comprised of anxiety and depressive symptoms. Anxiety and depressive symptoms were measured by use of the Hospital Anxiety and Depression Scale [22]. The scale had 14 items divided into 2 subscales; one measured anxiety (HADS-A) and the other measured depressive symptoms (HADS-D). Each subscale had seven items with a four-Likert scale. The total possible score for each subscale ranged from zero to 21, and a higher score indicated more symptoms. HADS-A and HADS-D had Cronbach’s alpha coefficients of 0.89 and 0.84, respectively, for Swedish women with breast cancer [13].
HRQoL was measured using the European Organization for Research and Treatment of Cancer Breast Cancer-Specific Quality of Life Questionnaire (QLQ-BR23) [23]. It is a disease-specific questionnaire with 23 questions, each of which had four options assigned by a number (not at all=1, a little=2, quite a bit=3, and very much=4). It assessed eight dimensions: body image (BRBI), sexual functioning (BRSEF), sexual enjoyment (BRSEE), future perspective (BRFU), side effects of systemic therapy (BRST), breast symptoms (BRBS), arm symptoms (BRAS), and indignation by hair loss (BRHL) [24]. All dimensions were transformed to 100-percent scores, and higher scores indicated lower quality of life. This questionnaire was translated to Swedish and tested before data collection among other breast cancer patients with acceptable Cronbach’s alpha score in each sub-scale [13].
Procedure
The heads and nurses of clinics of surgery/oncology and plastic surgery in Uppsala, Gävle and Falun were informed about the study. The heads of the clinics gave permission to conduct the study, and the nurses understood the study and were able to answer questions if the participants asked. Thereafter, written information about the study and its purpose together with a consent letter and a questionnaire was sent by ordinary mail to the eligible women. They were assured of their anonymity and of confidentiality, and they were told that they could drop out at any time. The Declaration of Helsinki for medical research was fulfilled. The women who agreed to participate in the study signed a consent letter, responded to the questionnaire, and returned these documents in a stamped envelope. Women who did not wish to participate in the study returned the documents without filling in any information. A reminder was sent twice by post (after two weeks and one month) to women who had not returned the envelope in due time.
Directed Acyclic Graphs (DAGs) [25] were constructed based on previous studies in order to demonstrate what factors were associated with psychological distress [26 – 28], and with HRQoL [13]. See Figures 1a, and b.
Analyses
We analyzed data using descriptive and inferential statistics. Descriptive statistics summarized sociodemographic characteristics, social support, HADS-A scores, HADS-D scores, and QLQ-BR23 scores for participants in terms such as frequency, mean, and standard deviation (SD). Inferential statistics applied Pearson’s correlation and linear regression analyzes.
Pearson’s correlation was used to determine the correlation between the scores of the HADS-A, HADS-D, and each dimension of QLQ-BR23. Whilst linear regression analyses were performed to determine relationships between sociodemographic factors, treatments, and social support factors, and outcome variables.
In multiple linear regression analyses, outcome variables were the scores of psychological distress, and QLQ-BR23, all of which were continuous variables. Sociodemographic and social support were considered independent variables. Age, duration of diagnosed breast cancer, and each information support were continuous variables. Dummy variables (categorization to zero and one) were marital status (married/lived together=0, the others=1), education (high school or above=0, secondary school/others=1), belonging to a cultural/ethnic minority (no=0, yes=1), having an underlying disease (no=0, yes=1), and methods of treatments (no=0, yes=1). Religion was excluded because of a low number in its sub-group. Assumptions were satisfied before the analyses (i.e. auto-correlation, multi-collinearity, homoscedasticity, linearity, and multivariate normality). First, we inserted each independent variable into a simple linear regression analysis for each outcome variable. Significant independent variables from the simple analyses remained in multiple linear regression analyses using stepwise selection method (alpha-to-enter of 0.05, alpha-to-remove of 0.10). 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. The level of statistical significance for all analyses was set at p<0.05.