Article Text
PDF
Abstract
Objectives The purpose of this study was to investigate the quality of life (QoL) of older Chinese patients with breast cancer and to explore further the associations of functions, symptoms, financial burdens and comorbidities with global health/quality of life (gQoL).
Design This was a cross-sectional study carried out following the Strengthening the Reporting of Observational Studies in Epidemiology checklist.
Setting This study was conducted in two hospitals in Beijing from October 2021 to November 2022.
Participants Patients with breast cancer aged over 65 years were included in the final analysis, which comprised a total of 481 patients.
Primary and secondary outcome measures The European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 and the Charlson Comorbidity Index (CCI) were used to assess patients’ QoL and comorbidities. The data were analysed using network analysis and path analysis.
Results Out of the 136 possible edges in the final networks, 84 (61.8%) were non-zero. ‘Fatigue’ was the central symptom and indirectly decreased the gQoL, which was mediated by increasing ‘financial difficulties’, ‘CCI’ and ‘role function’ (β = −0.35, p<0.001). ‘Physical function’ was also an important and direct intervention node that was indirectly related to gQoL, and this was mediated by ‘role function’ (β = −0.15, p=0.006). Path analysis accounted for 32.0% of the total effect.
Conclusions The various dimensions of QoL are highly interrelated and mutually reinforcing. These results highlight the importance of improving the fatigue and physical function of older patients with breast cancer. Interventions targeting these symptoms may lead to an overall improvement in gQoL.
Trial registration number ChiCTR2200056070; Public title: Frailty and Comorbidity in the Elderly Study (FACE Study).
- Aged
- Quality of Life
- Breast tumours
- Nursing Care
Data availability statement
Data are available upon reasonable request. In order to protect patient data privacy, we do not share data unless the request is reasonable and with the consent of the corresponding author.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
The study was supported by a national project with a good preliminary design.
Data were collected using standardised and validated procedures and instruments, increasing its credibility.
Both network and path analysis methods were used to leverage their respective strengths and address their limitations.
One limitation of our study is that it employed a cross-sectional study design with heterogeneous subjects.
The combined analysis of metastatic and nonmetastatic patients underestimated these differences.
Introduction
With increased awareness of the importance of individualised, patient-centred care, as well as the increased rates and duration of breast cancer survival, quality of life (QoL) is becoming the central parameter of breast cancer survivorship.1 2 QoL is affected by a variety of factors in older patients with breast cancer, as they face tremendous declines in both physical and psychological functions, with issues ranging from organ failure and neuropathy to depression.3 4 Moreover, with the rapid ageing of the population, 41.4% of patients with breast cancer in China are estimated to be aged 60 years or older by 2030.5 There is a need to focus on this group of patients in China.
The European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30) is a widely validated scale for assessing QoL and includes multidimensional concepts such as symptoms, functions and financial burden.6 7 Most previous studies have focused on how specific factors affect QoL, such as the impact of cognitive function and physical function on QoL, the impact of comorbidities on QoL and the impact of treatment-related symptoms on QoL.8–10 However, the different dimensions of QoL tend to influence each other and are highly correlated, but there have been no studies that fully illustrate this relationship. Network analysis is often used in the study of multiple items to understand the interrelationships among them and to identify important targets, called ‘central symptoms’ for clinical intervention.11 However, network analysis cannot establish causality or directional relationships between nodes; this limitation can be overcome by path analysis. Path analysis provides insight into the pathways between nodes (predictors and mediators) that lead to the resulting variables. By exploring the interaction of different dimensions of QoL, we can identify the specific processes that affect QoL.
Thus, the purpose of this study was to understand the relationships among multiple dimensions of QoL in older Chinese patients with breast cancer using network and path analysis. Specifically, we used network analysis to examine the relationships among the QoL subscales and to identify the core symptoms. Next, we used path analysis to investigate the relationships between core symptoms and global health/quality of life (gQoL). By elucidating the symptom-function-financial burden-comorbidity-gQoL relationship, we hope to provide insights for developing effective interventions and rehabilitation programmes for older patients with breast cancer to ensure their QoL.
Patients and methods
Study design and setting
This study was part of a multicentre, prospective, cross-sectional registry study designed to establish a clinical database for older patients. This breast cancer cohort included 510 participants from two hospitals in Beijing from October 2021 to November 2022. The reporting in this study follows the Strengthening the Reporting of Observational Studies in Epidemiology checklist for cross-sectional studies (online supplemental table S1).12 All the participants were informed of the aim of the study and signed an informed consent form.
Supplemental material
Participants
The inclusion criteria for the participants were as follows: (1) had pathologically confirmed breast cancer; (2) were over 65 years of age; (3) were able to understand the purpose and content of the survey and cooperate with the survey; and (4) had complete medical records.
The exclusion criteria were as follows: (1) refused to complete the scale; (2) incomplete questionnaires; and (3) medical records that were incomplete and could not be analysed statistically.
Measures
QoL was assessed using the EORTC QLQ-C30.6 It consists of several programmes: five functional domains (physical, role, cognitive, emotional and social); three domains of physical symptoms (fatigue, nausea/vomiting and pain); several individual symptoms (dyspnoea, insomnia, appetite, constipation, diarrhoea and financial difficulties); and gQoL. The gQoL includes two questions, namely, (1) the subjective assessment of the patient’s physical health status and (2) the overall QoL. Each item scale is then converted into a standard score for analysis. Higher scores for functional domains and general health indicate better functional status and QoL. Higher scores for symptom domains indicate more symptoms or problems (and poorer QoL).
The Charlson Comorbidity Index (CCI) is currently the most widely used comorbidity index, with a total of 14 items (online supplemental table S2).The total score is calculated by adding the weights. To avoid duplicate assessments, breast cancer was included in the clinical data for statistical analysis but not in the CCI. The above sinicisation scales have been validated in the Chinese population.13 14
Clinicopathological data, including age at the time of the survey, education level, stage, molecular subtype and treatment, were collected from the included patients after the questionnaire survey. The treatment was identified as the treatment the patients had received at the time of the survey.
Data analysis
Clinical data Statistics
SPSS (V.26.0) software was used for descriptive statistical analyses, with means and SD for continuous variables and frequencies and percentages for categorical variables. Additionally, the median was provided for the QoL subscales to facilitate comparison with other published data.15
Network estimation
Network analysis was performed using the qgraph, networktools, ggplot, mgm and bootnet packages of R software (V.4.1.2; using the R Foundation for Statistical Computing). The least absolute shrinkage and selection operator and extended Bayesian information criterion were used to regularise the correlation matrix to reduce the margin of possible spurious correlations. The blue connections between nodes represent positive connections and the red connections represent negative connections. The network was visualised using the Fruchterman-Reingold algorithm and strongly connected nodes were usually close to each other. The importance of each node was quantified by calculating the expected influence.16 The greater the expected influence was, the more important the node was in the network model. To more intuitively identify specific symptoms that are directly related to gQoL, we used the ‘flow’ function in the R package qgraph for plotting.17 Predictability refers to the extent to which the variance of a node can be explained by all of its neighbouring nodes.17 The average predictability of all nodes in the network reflects the extent to which the network was affected by external factors.
To estimate the stability of the centrality measure, the case‒drop subset bootstrap method was used, in which an increasing proportion of subjects were randomly removed from the dataset and the centrality index was recalculated. To quantify the stability of the centrality index, the correlation stability (CS) coefficient was calculated. Ideally, centrality estimates should be greater than 0.5. In addition, to measure the accuracy of the edge, estimated 95% CIs for the region containing the true regularised partial correlation (edge) were calculated by ‘nonparametric’ bootstrapping (n bootstrap=1000).18
Path analysis
Based on the results of the network analysis, multiple linear analyses were performed on factors directly related to gQoL to identify independently related factors. Path analysis was used to identify direct and indirect effects between variables and to determine the overall fit of the model. A well-fitting model is needed to meet the following conditions: test χ2, χ2/df ratio <3, goodness-of-fit index (GFI)>0.90, comparative fit index (CFI)>0.90, adjustment goodness-of-fit index (AGFI)>0.95 and root mean square error of approximation (RMSEA)<0.10.19 Based on the multivariate Lagrangian multiplier test, the model was modified twice to add new paths where necessary. The significance of all direct and indirect effects was assessed to determine which variables had direct or indirect effects on gQoL. The significance level was set to 0.05. Standardised beta coefficients (β) were derived for each explanatory variable to allow comparison and estimation of the relative importance of each measure. The R-squared value was calculated to determine the proportion of variance that the model was able to explain.20 Path analysis was performed using IBM SPSS Amos software (V.23).
Patient and public involvement
None.
Results
Clinicopathological characteristics
A total of 510 questionnaires were sent out, and 491 were returned, of which 10 were incomplete. Therefore, the examination results of 481 patients were included in the analysis. The median age at enrolment was 69 (range 65–91) years. There was a greater percentage (33.7%) of patients with 10–12 years of education and fewer patients (9.4%) with only 6 years of education. Most participants (76.7%) had early-stage BC, and HR+/HER2− BC accounted for the majority of cases (68.6%). Most patients had undergone surgery (91.1%). More than half of the patients had received chemotherapy or endocrine therapy (56.5% and 65.3%, respectively). A total of 30.1% had received radiotherapy. Among the 369 patients with early-stage breast cancer, 86 (23.3%) were in the pre-chemotherapy phase of treatment, and only 40 (10.8%) were in the chemotherapy phase. A total of 243 (65.9%) patients were in the post-chemotherapy phase or had already received endocrine therapy at the time of enrolment. The median time from active treatment was 44 (1–336) months (table 1).
Clinicopathological data
Distribution of function and symptom scores
The distributions of the function and symptom scores are shown in table 2. The highest score was found for social function(86.9±22.7), whereas cognitive function(80.2±21.3) was rated much lower. Most of the patients’ functional scores were greater than 50, and only a small number of patients had scores less than 50, among whom the role function ratio was the highest (9.6%). Insomnia and fatigue were the most frequently reported symptoms, with the highest scores (32.3±33.6 and 20.1±22.7, respectively).
Scores of EORTC QLQ-C30 questionnaires
Network analysis results
As shown in figure 1A, 84 of the 136 possible edges (61.8%) were non-zero, indicating significant interconnectedness between symptoms. The strongest correlation between the variables in the network analysis was between ‘fatigue’ and ‘social function’ (−0.722), and the strongest relationship between the variables in the network analysis and gQoL was ‘fatigue’ and ‘gQoL’ (−0.513) (online supplemental table S3). The symptom with the greatest expected influence was ‘fatigue’(1.617), followed by ‘role function’(1.052) and ‘social function’(0.873), indicating that these symptoms and functions had the greatest impact on the overall network (figure 1B, online supplemental table S4). In contrast, the variables with the least expected effects were ‘age’ (0.048), ‘CCI’ (0.382) and ‘diarrhoea’ (0.492). Moreover, the node predictability values ranged from 45.0% to 96.5%, with an average of 77.2%, indicating that, on average, 77.2% of the variance in nodes from the network could be explained by their neighbouring nodes (online supplemental table S4). ‘Age’ (0.965), ‘CCI’ (0.901) and ‘diarrhoea’ (0.893) had the highest predictability in the model, whereas ‘fatigue’ (0.450), ‘role function’ (0.625) and ‘physical function’ (0.665) had the lowest predictability.
(A) Network structure of quality of life and characteristics. (B) Expected influence of each node on the quality of life and characteristics network.
To better illustrate the relationships between the items and gQoL scores, we plotted the QoL flow network (figure 2). Among the total symptoms, 11 were directly related to gQoL and the remaining symptoms, including age, were indirectly related to gQoL. Multiple linear regression analysis revealed that ‘CCI’, ‘financial difficulties’ and ‘role function’ were independently correlated with ‘gQoL’ (online supplemental table S5).
Flow network of quality of life and characteristics.
The expected influence of the bridge and the strength of the bridge also had high stability, and the coefficient CS values were both 0.672 (online supplemental figure S1). The accuracy of edge weight estimation between nodes is shown in online supplemental figure S2.
Supplemental material
Path analysis model
Based on network analysis and multiple linear regression analysis results, we further examined the relationships between ‘fatigue’, ‘physical function’, ‘CCI’, ‘financial difficulties’, ‘role function’ and ‘gQoL’. The final model revealed significant correlations among the variables (figure 3). ‘Financial difficulties’ (β = −0.11, p=0.009) and ‘role function’ (β = 0.26; p<0.001) had direct effects on ‘gQoL’. ‘CCI’ had an indirect effect on ‘gQoL’ and was mediated by ‘financial difficulties’(β = −0.12; p<0.001). ‘Fatigue’(β = −0.35, p < 0.001) had an indirect effect on ‘gQoL’ and was mediated by ‘financial difficulties’, ‘role function’and ‘CCI’. ‘Physical function’ had an indirect effect on ‘gQoL’, which was mediated by ‘role function’ (β = −0.15; p=0.006). ‘Physical function’ and ‘fatigue’ were also significantly correlated (β = 0.69; p<0.001). The multivariate linear regression final model for the mediation showed good global adjustment: fit indices: χ2 (9) = 3.870 (p>0.424); χ2/df = 0.968; GFI = 0.997; AGFI = 0.986; CFI = 1.000; RMSEA = 0.000 (95% CI = (0.000; 0.068)). The R-square value indicated that this model explained 32.0% of the variance in the general health score.
Path analysis with standardised direct effects. CCI, Charlson Comorbidity Index; Fnd, financial difficulties; Ftg, fatigue; gQoL, global health/quality of life; PhF, physical function; RlF, role function.
Discussion
The balance between treatment effectiveness and QoL in older patients with breast cancer is an important issue for clinicians. To our knowledge, this is the first cross-sectional study to investigate QoL in older Chinese patients with breast cancer in the real world, using network and path analyses to help understand the factors influencing QoL to guide and inform social, health and education policies. As expected, functions, symptoms and comorbidities had a significant impact on gQoL through different paths.
The results of the network analysis visualised the complex network of variables and provided clues to further understand the relationships among functions, symptoms, comorbidities and gQoL. ‘Fatigue’ as the central symptom, was closely related to patient function, including ‘role function’,‘physical function’ and ‘social function’. A recent meta-analysis revealed that the incidence of fatigue after chemotherapy in patients with cancer was approximately 49%.21 Studies have shown that older subgroups experience increasingly severe levels of fatigue.22 The high centrality of fatigue in this study indicates that it is highly associated with other symptoms, which has very important clinical intervention value. Compared with changing sociodemographic and clinical factors, which can be challenging, it seems more effective to address fatigue and systemic therapy side effects to improve long-term survivors' QoL. Therefore, the assessment and management of fatigue should be considered in daily nursing practice.
In addition to clinical drug interventions, the results of the path analysis also provide valuable insights into ways to improve fatigue. Physical function and fatigue were significantly correlated. The low predictability of physical function indicates that it is also a good direct intervention node and is independently related to ‘role function’. Previous studies have also shown that physical activity can have a positive effect on patients' quality of life by preventing a decline in physical function.23 In addition, chronic disease management has also been shown to be important. In this pathway analysis, fatigue was an independent variable and comorbidities were the regulators. However, in cross-sectional studies, this is not absolute. These results suggest that fatigue in older patients can be improved by addressing comorbidities.
Women with breast cancer have a similar risk of developing chronic diseases or comorbidities as women without cancer because of the natural effects of ageing. However, comorbidities have a significant effect on cancer treatment decisions and healthcare costs. These patients are likely to experience inadequate treatment or perioperative complications.24 Our data also revealed that the presence of chronic disease had a negative effect on gQoL, that is mediated by ‘financial difficulties’. We included patients over 65 years of age, almost all of whom were retired, did not have high sources of income and were likely to experience greater financial burden.21 This not only causes patients to delay seeking care and forgo necessary treatments but also affects treatment decisions and patient compliance.25 This requires the participation of the whole society’s medical insurance to reduce the financial burden on patients with cancer. This is not easy, but it may partly explain why the QoL of patients with breast cancer in China is lower than that of patients in Western countries.23 24
All identified core symptoms had a high predictive value, with an average predictive value of 74.8% for all nodes in the network. In other words, most of the nodes in the network could be interpreted and controlled by their neighbours. When the CS was greater than 0.5, the network had good repeatability. The model evaluation indices of the path analysis also met the expected requirements and showed that the overall fit of the model was good.
The limitation of this study is that it was a cross-sectional study, meaning that the relationships observed are not directional. They cannot confirm the causal relationship between symptoms. Path analysis compensates for this to some extent, but time series data should still be collected in the future. Second, the drivers of quality of life differ according to metastatic status. The combined analysis of metastatic and non-metastatic patients underestimated these differences. Finally, the participants in this study were recruited from two hospitals, and this group may not be representative of the older breast cancer population across different regions of China.
Conclusion
The various dimensions of QoL are highly interrelated and mutually reinforcing. These results highlight the importance of improving the fatigue and physical function in older patients with breast cancer. Interventions targeting these symptoms may lead to overall improvement in gQoL.
Data availability statement
Data are available upon reasonable request. In order to protect patient data privacy, we do not share data unless the request is reasonable and with the consent of the corresponding author.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved as part of a multicentre, cross-sectional registry study (registration number: ChiCTR2200056070). The study was carried out according to a named standard approved by the Ethics Committee of the National Cancer Center/Cancer Hospital,Chinese Academy of Medical Sciences and Peking Union Medical College (approval number: 22/216-3418). Participants gave informed consent to participate in the study before taking part.
References
Footnotes
MiX and LJ contributed equally.
Contributors PZ: Conceptualisation, Methodology, Supervision, Writing—Review and Editing. MiX and LJ: Software, Formal analysis, Investigation, Data Curation, Writing—Original Draft. XQ, MeX, ZL, JZ and HC: Data Curation, Investigation. ZL, QingL, QiaoL, XW, JW, YL, SC, XZ, JZ: Resources. The guarantor of the study is PZ, who takes full responsibility for the final work and the conduct of the study, had access to the data and controlled the decision to publish.
Funding This study was funded by National Key Research and Development Program of China (grant number 2020YFC2004803) and Medical Oncology Key Fundation of Cancer Hospital Chinese Academy of Medical Sciences (grant number CICAMS-MOMP2022007).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.