Data source
We recruited both breast cancer outpatients and inpatients in Sichuan Oncology Hospital from November 2017 to May 2018. Ethical permission was granted by the Ethics Committee, West China School of Medicine/West China Hospital, Sichuan University(approval number 2017 − 255). We were authorized the right to use FACT-B ( Simplified Chinese version) and EQ-5D-5L (Simplified Chinese version).
Study Participants
Inclusion criteria were as follows. First, participants were clinically and/or pathologically diagnosed with breast cancer. Second, patients were aged 18 and above. Third, patients should not have any mental problems and have the ability to express. In addition, patients should agree to participate this study. Informal consent were obtained from all participants. We excluded patients that had comorbidities such as cardiovascular disease and mental health problems. We interviewed 451 breast cancer patients with 5 respondents who did not complete the interview. Overall, 446 participants were included in data analysis.
Measures
We measured participants’ quality of life by FACT-B, which measures quality of life from five dimensions including physiological conditions(PWB), social and family support (SWB), emotional conditions (EWB), functional status(FWB), and additional symptoms with breast cancer (BCS). The FACT-B scale consists of 37 questions. Since the scale of scores of these five dimensions differed, we standardized them into a scale of 100. The validity and reliability of FACT-B at Chinese version was examined by prior investigators [21].
In parallel, patients’ health utility was measured by EQ-5D-5L (Simplified Chinese version), which measures five health dimensions including mobility, self-care, usual activities, pain/discomfort, and anxiety/depression with five-level severity from no problems, slight problems, moderate problems, severe problems to extreme problems/unable. The severity of each dimension was coded from 0 to 4 with as the reference group. For example, 0 in mobility represents that individuals have no problems with walking, and 4 represents that they could not walk. In addition, participants were required to report self-rated health status ranging from 0 to 100 with 100 represents the best health status one can imagine(EQ-VAS). In contrast, 0 represents the worst health status. The validity and reliability of 5Q-5D-5L (Simplified Chinese version) were examined by prior Chinese investigators as well [22]. We calculate health utility by employing the value set based on Chinese data [23].
The main independent variable of interest isdisease states, i.e.,P, R, S, and M. In addition, we introduced covariates including TNM stage (0, I, II, III, and IV), surgical approaches (breast conserving surgery, modified radical surgery vs. no surgery), menopaus state(yes vs.no), radiotherapy (yes vs. no), chemotherapy(yes vs. no), targeted therapy(yes vs. no), endocrine therapy(yes vs. no), and inpatients(vs. outpatients) to control for clinical confounders which may affect patients’ health via adverse effects, which are not mediated by disease states. [12, 19].
Furthermore, we included patients’ demographic attributes (age and marital status)and socioeconomic characteristics including education attainment, household income, location of Hukou (urban vs rural), occupation, and medical insurance type as covariates to controlling for the effect of social deprivation on health[15, 17].
Data analysis
To gauge the difference of variables in four kinds of disease states, descriptive analyses including Chi-square test, Fisher exact probability test, and ANOVA test were performed according to variables’ characteristics. We calculated health utilities by a value set developed based on China’s population in prior investigation[23]. To clarify the degree of overlap between instruments, Spearman's rank correlation coefficient is calculated not only between each instrument, but also between the FACT-B domains.
Univariate regression analysis was conducted to figure out potential predictors of participants’ health, which was reflected as overall scores of FACT-B, scores of each dimensions in FACT-B, self-rated health, health utility from EQ-5D-5L, and the degree of five dimensions in EQ-5D-5L, respectively (results available from authors). In this regard, we performed multiple regression models according to dependents’ characteristics. Linear regression model was performed for overall scores of FACT-B since it was normally distributed. Ordinal logistic regression models were respectively performed for BCS, FWB, EWB, SWB, and PWB, self-rated health, and the degree of mobility, self- care, usual activities, pain, and depression, as the distribution of BCS, FWB, EWB, SWB, PWB, and self-rated health were highly skewed, and the degree of mobility, self-care, usual activities, pain, and depression are ordinal data. For BCS, FWB, EWB, SWB, and PWB, we divided each variable into four balanced groups coded as 0 to 3 with 0 representing the lowest group and the reference group in the model, each group consisting of similar number of participants. Similar treatment was used for self-rated health status with five balanced groups coded as 0 to 4 with 0 representing the group of the worst health. Tobit model was performed for health utilities, as they were right-censored.
Independent variables of statistical significance (p < 0.05) in univariate analysis were then introduced to multivariate analysis. Variance inflation factors were performed to examine multicollinearity among independent variables in multivariate models.
Furthermore, we analyzed correlation between quality of life and health utilities from EQ-5D-5L by employing rank correlation test. Lastly, we estimated health utilities from quality of life from FACT-B by employing a mapping function derived from Singaporean population[20], and conducted rank correlation test between estimated health utilities and those directly measured from participants with EQ-5D-5L.
The mapping functionbased on Singaporean patients was below[20]:
Estimated health utility = 0.2846 + 0.0121 × PWB + 0.0044 × FWB + 0.0034 × BCS
Data analyses were performed with SPSS 23.0. and SAS University Edition. p value less than 0.05 was considered statistically significant.