Next Article in Journal
Very Delayed Acute Hepatitis after Pembrolizumab Therapy for Advanced Malignancy: How Long Should We Watch?
Previous Article in Journal
Managing Opioids and Mitigating Risk: A Survey of Attitudes, Confidence and Practices of Oncology Health Care Professionals
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Current Oncology
Volume 28
Issue 1
10.3390/curroncol28010087
curroncol-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Longitudinal Symptom Burden Trajectories in a Population-Based Cohort of Women with Metastatic Breast Cancer: A Group-Based Trajectory Modeling Analysis

by
Suman Budhwani
1,*,
Rahim Moineddin
2,3,
Walter P. Wodchis
3,4,5,6,
Camilla Zimmermann
7,8 and
Doris Howell
7,9
1
Institute for Health System Solutions and Virtual Care, Women’s College Hospital, Toronto, ON M5S 1B2, Canada
2
Department of Family and Community Medicine, University of Toronto, Toronto, ON M5G 1V7, Canada
3
Institute of Clinical Evaluative Sciences, Toronto, ON M4N 3M5, Canada
4
Institute of Health Policy, Management & Evaluation, University of Toronto, Toronto, ON M5T 3M6, Canada
5
Health System Performance Network, University of Toronto, Toronto, ON M5T 3M6, Canada
6
Institute for Better Health, Trillium Health Partners, Mississauga, ON L5M 2N1, Canada
7
Princess Margaret Cancer Centre, University Health Network, Toronto, ON M5G 2C1, Canada
8
Department of Medicine, University of Toronto, Toronto, ON M5S 1A8, Canada
9
Lawrence S. Bloomberg Faculty of Nursing, University of Toronto, Toronto, ON M5T 1P8, Canada
*
Author to whom correspondence should be addressed.
Curr. Oncol. 2021, 28(1), 879-897; https://doi.org/10.3390/curroncol28010087
Submission received: 27 November 2020 / Revised: 29 December 2020 / Accepted: 31 January 2021 / Published: 14 February 2021
Browse Figure
Versions Notes

Abstract

Understanding the symptom burden trajectory for metastatic breast cancer patients can enable the provision of appropriate supportive care for symptom management. The aim of this study was to describe the longitudinal trajectories of symptom burden for metastatic breast cancer patients at the population-level. A cohort of 995 metastatic breast cancer patients with 16,146 Edmonton Symptom Assessment System (ESAS) assessments was constructed using linked population-level health administrative databases. The patient-reported ESAS total symptom distress score (TSDS) was studied over time using group-based trajectory modeling, and covariate influences on trajectory patterns were examined. Cohort patients experienced symptom burden that could be divided into six distinct trajectories. Patients experiencing a higher baseline TSDS were likely to be classified into trajectory groups with high, uncontrolled TSDS within the study follow-up period (χ2 (1, N = 995) = 136.25, p < 0.001). Compared to patients classified in the group trajectory with the highest relative TSDS (Group 6), patients classified in the lowest relative TSDS trajectory group (Group 1) were more likely to not have comorbidities (97.34% (for Groups 1–3) vs. 91.82% (for Group 6); p < 0.05), more likely to receive chemotherapy (86.52% vs. 80.50%; p < 0.05), and less likely to receive palliative care (52.81% vs. 79.25%; p < 0.0001). Receiving radiotherapy was a significant predictor of how symptom burden was experienced in all identified groups. Overall, metastatic breast cancer patients follow heterogeneous symptom burden trajectories over time, with some experiencing a higher, uncontrolled symptom burden. Understanding trajectories can assist in establishing risk-stratified care pathways for patients.

1. Introduction

In Canada, approximately 5% of all breast cancers in women were diagnosed as Stage IV between 2011 and 2015 [1]. Although constituting a smaller percentage of overall breast cancer diagnoses, Stage IV or metastatic breast cancer can pose a significant burden on both patients and the healthcare system [2,3,4,5,6]. Advances in treatment have resulted in a growing number of advanced cancer patients living longer [7,8,9,10]. Consequently, management of ongoing and long-term symptom burden from disease progression and treatment (chemotherapy, radiotherapy) by the patient, their families, healthcare providers and the healthcare system is essential [2,5,11,12]. While healthcare services such as home and palliative care can provide some support based on available capacity [13,14,15], advanced metastatic breast cancer can create significant challenges in the management of the disease and symptoms both for patients and providers, and result in unmet patient care needs over time [2,5,16,17,18,19,20]. As such, there is a need to better understand chronic, long-term patient symptom burden for metastatic breast cancer patients longitudinally. With this understanding, healthcare providers can better advise patients on appropriate treatment, provide supportive care such as self-management support and palliative care for patients earlier in the course of the disease trajectory, as well as create personalized and risk-stratified care pathways, thereby meeting patient and health system needs [21].
However, symptom burden in the literature has typically been examined longitudinally for cancer patients early in the course of the disease [22,23,24,25] or at the end of life [26,27,28,29,30]. Some studies have incorporated more extensive longitudinal measurement of symptom burden in patients at all stages of breast cancer, rather than focusing specifically on Stage IV metastatic breast cancer [6,31,32,33,34,35,36,37,38]. While some research has examined symptom burden longitudinally for patients with metastatic breast cancer, this research has been limited to a few assessments or time points of measurement of symptom burden [18,39,40], or has depicted average symptom burden across a largely heterogenous patient group [41]. Hence, the aim of this study was to identify and describe the longitudinal, heterogeneous trajectories of symptom burden in a population-based sample of metastatic breast cancer patients using data from multiple symptom burden assessments over time. A secondary aim was to examine the influence of covariates on symptom burden trajectory patterns.

2. Methods

2.1. Participants and Procedure

A longitudinal, observational approach was utilized, using retrospectively collected population-level administrative data from linked health administrative databases housed at Cancer Care Ontario in the province of Ontario, Canada. This study was approved by the University of Toronto Health Sciences Research Ethics Board, with patient consent considered implied due to the use of health administrative data.
From health administrative databases, a cohort of all adult (18 years of age or older at diagnosis), female patients diagnosed de novo with primary breast cancer and distant metastases (Stage IV) between 1 January 2010 and 31 December 2014 were identified using the Ontario Cancer Registry (OCR), a provincial registry of all residents with new diagnoses or deaths from cancer [42]. Stage IV metastatic breast cancer patients were identified using the tumour-node-metastasis (TNM) staging classification system and International Classification of Disease (ICD) codes utilized in the OCR. Individuals with any indication of multiple primary cancer diagnoses were excluded.
The constructed cohort of patients with Stage IV metastatic breast cancer was linked to the Interactive Symptom Assessment and Collection (ISAAC) database to obtain Edmonton Symptom Assessment System (ESAS) assessments for each patient from date of diagnosis until death or 31 December 2015, the end of the study follow-up period. The ESAS is a validated and reliable symptom screening tool for cancer patients [43,44] that assesses the severity of nine commonly experienced symptoms, namely pain, tiredness, drowsiness, nausea, lack of appetite, shortness of breath (dyspnea), depression, anxiety, and wellbeing on a scale of 0–10, with 0 indicating no symptoms (or best score for the symptom scale of wellbeing), and 10 serving as an indication of the most severe symptom experienced at a given point in time [44]. Patients complete the ESAS at each visit to an Ontario Cancer Centre and thus, data is collected mostly from ambulatory patients, with each patient potentially having multiple ESAS assessments longitudinally during the course of their care [45,46].
Inclusion and exclusion criteria for the study cohort were applied at the patient and ESAS assessment level. At the patient level, the following exclusions were made: (a) since longitudinal assessment of symptoms was required, 256 patients with fewer than three ESAS assessments were excluded from the study cohort; (b) six patients were found to have symptom scores of zero on all individual symptom scores for all collected ESAS assessments that after further examination of the dataset were assumed to be data error; these six patients were excluded from the study cohort, (c) three patients with missing values on the explanatory variable of neighbourhood income quintile were also excluded from the study cohort.
At the ESAS assessment level, a total of 58 ESAS assessments were excluded if more than 50% of the symptom scores on the assessment were missing. ESAS assessments with less than or equal to 50% of symptom scores missing in the assessment were used to impute total symptom distress score based on the mean of the non-missing scores available. ESAS assessments were also excluded from the cohort if patients had two ESAS assessments on the same day. In these cases, the ESAS assessment with the higher summative total symptom distress score (a sum of the nine individual scales comprising the ESAS) was included (n = 603 duplicate assessments excluded).

2.2. Measures

Total symptom distress score (TSDS) was the main outcome variable for this study, representing the symptom burden experienced by patients at a single time point of measurement [44]. It has been used previously in clinical trials to measure patient symptom severity [47,48]. TSDS scores were calculated by summing the nine ESAS physical and psychological symptom assessment scores from each unique ESAS assessment (pain, tiredness, drowsiness, nausea, appetite, dyspnea, anxiety, depression, and wellbeing) for each patient, providing a potential score of 0–90 for each assessment [44]. The TSDS did not follow a normal distribution; as such, a λ = 0.25 transformation was performed to normalize the TSDS using the Box–Cox transformation methodology to control for the skewness of data without losing the data’s heterogeneity [49,50,51], inclusive of a shift of +1 to ensure positive values. Transformed TSDS sores were analyzed and are reported.
Date of diagnosis until date of death or study endpoint for each patient were used to calculate months since diagnosis as the main time variable within the analysis. Time-fixed (baseline) covariates included age at diagnosis and year of diagnosis derived from the OCR; variables of neighbourhood rurality and neighbourhood income quintile were calculated using postal codes derived from linkage to the Registered Persons Database, which captures all registered Ontarians for publicly funded healthcare. Postal codes were linked to Statistics Canada census data to determine neighbourhood rurality and income quintile (a quintile of 1 having the lowest income, and of 5 having the highest income) [52,53,54]. Lastly, the Charlson Comorbidity Index, a measure of comorbidities, was derived from methodology based on patient hospitalization records [55,56,57] and excluding scoring for cancer and metastases [56,58].
To address temporality, we used time-varying covariates, which consisted of receipt of any treatment or supportive care 0–7 days prior to each ESAS assessment date. Treatment and supportive-care variables were selected based on their noted impact on symptoms for cancer patients in the literature [3,13,59,60,61]. Treatment variables included chemotherapy derived from the Activity Level Reporting System (ALR), and radiotherapy derived from the National Ambulatory Care Reporting System (NACRS) database, capturing all ambulatory and emergency department visits to hospitals by patients, and the ALR database. Supportive care included receipt of in-home, non-end-of-life home care services excluding case management, nurse practitioner palliative care, placement, and uncategorized ‘other’ services, derived from the Home Care Database (HCD). Supportive care also included receipt of palliative care services, which were identified based on a previously published methodology [15], including only services provided by physician providers using the Ontario Health Insurance Plan (OHIP) database and end-of-life home-care services from HCD excluding case management, placement, and uncategorized ‘other’ services. Finally, all emergency department visits for each patient were identified using the NACRS database. A rate of emergency department visits per person month was calculated.

2.3. Statistical Analysis

The statistical analysis focused on identifying the patterns and distinct number of longitudinal symptom trajectories of patients over time, using data collected through repeated measurement of patient symptom burden. For all individuals in the cohort, the first 30 ESAS assessments were selected for analysis due to the wide range of total number of ESAS assessments for each patient in the cohort. Standard growth models are useful to study variability over time through calculation of an average developmental trajectory, where all individuals in the study population are expected to change in the same direction and by the same strength over time. However, in certain variables such as symptom scores, patients may experience symptoms in different patterns over time, and capturing this variability becomes important. As such, the semi-parametric statistical technique of group-based trajectory modelling (GBTM) was selected and utilized for the purpose of identifying distinct subgroups of patients in the cohort following similar patterns of change over time on the outcome variable of TSDS [62,63,64,65,66]. The use of GBTM enables identification of distinct subgroups of patients following similar patterns of change over time, thereby improving estimates of experienced symptom burden (TSDS) by patients [64,66,67,68,69]. Trajectories are fitted using maximum-likelihood methods [64,69].
All analyses were performed using SAS v.9.4 software [70]. A SAS procedure to run the GBTM analysis was downloaded [71,72]. p-values less than 0.05 for two-sided tests were considered significant, and confidence intervals were calculated. An uncensored normal (CNORM) model was selected in the GBTM due to the continuous nature of the outcome variable of the TSDS, with minimum and maximum values specified outside the range of observed values due to no expected or demonstrated clustering [63,66,69]. Next, the number of groups and the order of polynomials defining the shape of the trajectory for each group as flat (0), linear (1), quadratic (2), cubic (3), or quartic (4) were iteratively fitted and specified. Due to limited a priori knowledge of the potential number of trajectory subgroups expected, the number of groups and each group’s polynomial order was decided by starting with a group number of one and highest polynomial order of quartic (4), and iterations were continued to find the optimal number of groups and polynomial order [64,66,73]. The optimal model was determined based on changes in Bayesian information criteria (BIC) values between models [69]. The BIC is a statistical measure of model fit based on assessment of increased model parameters balanced by increasing model complexity, with higher numbers indicating an improved model fit [64]. The BIC is used to approximate a second measure of model parsimony, the log of the Bayes factor, with >10 being considered strong evidence in favour of model parsimony [66,69,73]. However, in certain situations the BIC may not always clearly indicate the optimal number of groups and order of polynomials for each group, and in such cases the objective of seeking the greatest model parsimony should be balanced with the objective of reporting a representative number of distinctive trajectories from the data [64]. Other criteria considered in selecting the optimal model included the fit of the 95% confidence intervals to estimated trajectories and points of overlap [63], the statistical significance at a p-value of 0.05 for the highest-order standard coefficient in the results of the model, and the percentage of cohort in each group [66,71,74]. Following the selection of the optimal number of groups and polynomial order for each group, the final model fit was assessed using (a) average posterior probability (AvePP) (group assignment probability) >0.7, (b) odds of correct classification (OCC) >5 and (c) |π−Ρ| (estimated group probabilities versus the proportion of sample assigned to the group) ~ 0 [64,66]. A profile plot for each trajectory group in relation to covariates was created, and chi-square statistics or logistic regressions were conducted to explore whether patient demographic and clinical characteristics were associated with trajectory groups identified [64,68,75].
Finally, a three-stage process was followed to select baseline covariates to include in the above selected model as hypothesized predictors of group membership [64,68]. The first stage was the process by which the above unadjusted model was selected. In the second stage, a multivariable multinomial logistic model was run to examine the association between trajectory group assignment and time-fixed baseline covariates to understand baseline predictors of group membership. No statistically significant covariates at the p = 0.05 level were found. As such, no time-fixed baseline covariates were included in the re-estimation of trajectories in the third and final stage, to jointly estimate trajectories and associations between trajectories and covariates. Hence, only time-varying covariates (chemotherapy, radiotherapy, home care, and palliative care) were included in the estimation at this stage as predictors of the patterns of symptom burden for each group’s trajectory [64].
Two types of sensitivity analyses were performed. The first sensitivity analysis explored the relationship between each patient’s baseline median TSDS score and subsequent probability of being assigned to a higher symptom burden trajectory (Groups 4–6) or a changing, lower symptom burden trajectory group (Groups 1–3) over the disease trajectory, through calculation of chi-square statistics. The second sensitivity analysis assigned patients censored due to death in the follow-up time period with a maximum transformed TSDS to better understand the influence of death (or subsequent missing values of TSDS) on observed trajectories.

3. Results

A total of 995 patients with 16,146 unique ESAS assessments were included in the study cohort (Table 1). Through iterative GBTM analysis, a six-group model with a polynomial order of 444000 was selected, demonstrating the lowest BIC value among all iterations and a log Bayes factor >100, indicating very strong evidence of model parsimony (Table 2). Figure 1 depicts the six trajectories for TSDS over time with percent group memberships for each group before adjustment with covariates. The 95% confidence intervals modelled for this trajectory demonstrated a close fit to the depicted trajectories and limited points of overlap, suggesting distinct trajectories (Supplementary Figure S1). Model-fit statistics demonstrated good model fit for the model selected: group membership was >5% for each group, AvePP for each group was above the recommended criteria of 0.7, the OCC was >5 for each group, and |π−Ρ| was also close to 0 for each group. Overall, trajectory Groups 4–6 demonstrated relatively higher, uncontrolled TSDS scores over time, while trajectory Groups 1–3 demonstrated changing symptom burden patterns over time and had relatively lower TSDS scores. The first sensitivity analysis indicated that for patients starting with a higher than median TSDS baseline score, there was an increased probability of being assigned to a high, uncontrolled TSDS group trajectory (Groups 4–6) in the follow-up time period (χ2 (1, N = 995) = 136.25, p < 0.001). The second sensitivity analysis demonstrated similar patterns as observed in Figure 1, other than for Groups 2 and 3.
Table 3 shows patient baseline characteristics by each of the six trajectory groups. Compared to individuals in Group 6 (the highest relative TSDS trajectory overall), individuals in Group 1 (the lowest relative TSDS trajectory overall) were more likely to not have any comorbidities (97.34% (for Groups 1–3) vs. 91.82% (for Group 6); p < 0.05), more likely to receive chemotherapy (86.52% vs. 80.50%; p < 0.05), and less likely to receive palliative care (52.81% vs. 79.25%; p < 0.0001). However, none of these baseline characteristics were significant in predicting a patient’s higher or lower symptom burden group membership.
Table 4 describes parameter estimates for the original trajectory group model adjusted for time-varying covariates of group trajectory. This allowed us to determine how time-varying covariates of radiotherapy, chemotherapy, home care, and palliative care influenced the actual pattern or trajectory of symptom burden experienced by patients within each group. Receipt of radiotherapy was a significant predictor of TSDS trajectory patterns over time for all groups, while receipt of home care and palliative care each predicted TSDS patterns in four of the six groups. All covariates were significant predictors for the trajectory observed for Group 5. For Group 3, which showed a trajectory of most relative improvement in TSDS over time, radiotherapy and palliative care were both significant predictors, while for Group 2, which showed a trajectory of relative increase in TSDS over time, chemotherapy and radiotherapy were significant predictors.
Table 5 shows the clinical parameters for the adjusted group membership. Lower percentages of patients in Groups 1 to 3 (lower TSDS) received chemotherapy and palliative care in comparison to Groups 4 to 6 (higher TSDS). Groups 2 and 3 also generally had lower percentages of patients with death within three years of diagnosis in comparison to Groups 4 to 6.

4. Discussion

This study characterized the progression of symptom burden in a population-based cohort of women with advanced breast cancer based on the TSDS from ESAS assessments conducted over the disease trajectory. Six distinct symptom burden trajectories were identified for this population group that can be further grouped into a low symptom burden and changing trajectory over time (Groups 1 to 3) and a high, uncontrolled symptom burden trajectory over time (Groups 4 to 6). These results are similar to those observed in other studies that demonstrated differences in how patients with breast cancer experience symptom burden [31,33]. The identification of two subgroups of cancer patients, one experiencing low symptom burden and the other experiencing relatively higher symptom burden, has also been found in other studies of different cancer populations [25,26]. Findings from this study reiterate these findings specifically for the patient population of advanced metastatic breast cancer, suggesting similarities across diverse cancer population groups.
Our analysis also showed that there were no baseline demographic and clinical characteristics that were associated with a patient’s membership in a particular trajectory group; however, the symptom burden trajectory for patients in each group was influenced by covariates of chemotherapy, radiotherapy, home care, and palliative care in different ways. Our analysis demonstrates that demographic and clinical characteristics, such as comorbidities, do not have an influence on how patients experience symptom burden, potentially suggesting the independent effect of treatment on symptom burden experienced. However, this might also be because demographic and clinical characteristics, such as functional status and disease severity, were not included in the analysis as they could not be derived from our utilized data sources. Moreover, while home care and palliative care were modelled in this analysis as predictors of symptom burden, they are likely proxies for symptom burden, that is the reason why patients may be eligible for home care and palliative care due to high symptom burden.
Our study also found that patients in Groups 4 to 6 had relatively higher symptom scores and an uncontrolled symptom burden trajectory over time. This might be because these patients may be on a closer trajectory towards death, as was also demonstrated by a larger percentage of patients dying within the first three years of diagnosis in Groups 4 to 6 (all high, uncontrolled symptom trajectory groups). Additionally, a greater proportion of patients in these groups also resided in lower income and urban neighbourhoods, suggesting possible challenges to accessing breast cancer screening and a cancer diagnosis [76,77]. Findings from the sensitivity analysis demonstrated that patients with a TSDS higher than or equivalent to the median at diagnosis, were more likely to follow a high, uncontrolled symptom burden trajectory over time, a finding also demonstrated in other studies [29,31]. This demonstrates the importance of the assessment of symptom burden from point of diagnosis and the need for appropriate symptom management and supportive care, such as palliative care and home care from disease onset, particularly for individuals with a high, uncontrolled symptom burden. Moreover, this study found that some patients have changing (increasing or decreasing) symptom burden over time (Groups 1 to 3), a finding also demonstrated by Kang and colleagues for the general cancer population [29]. Groups 1 to 3 had a larger percentage of patients who received chemotherapy in comparison to other groups, suggesting receipt of active treatment may be the reason for changing symptom trajectories over time [78]. This subset of patients would benefit from appropriate supportive care, such as symptom self-management strategies to control symptom burden associated with active treatment over time. Overall, these results suggest that there are sub-sets of patients that may benefit from proactive symptom-support intensification and help set the foundational groundwork for the creation of personalized risk-stratified cancer care pathways [21].
While this study has shown distinct trajectories of symptom burden in women with metastatic breast cancer, there are some limitations that must be considered. First, while distinct trajectories were estimated for this cohort, they should not be considered reified groups or absolute certainties [64,73]. The use of administrative data precluded examination of demographic and clinical characteristics such as functional status, specific sites of metastasis, biological subtype of breast cancer, and disease severity, which may influence symptom trajectory patterns [36,41]. Moreover, certain symptom scores on the ESAS may not be as sensitive to changes in symptom burden over time (e.g., appetite) [79]. In addition, model selection, although based on model fit statistics indicating parsimony, was also based on other recommended criteria requiring judgment. This is inherently a part of GBTM [64], thus presenting a need for more specific delineation of the model selection process in the literature [80]. Lastly, this study examined TSDS, which may mask a symptom score change on single symptoms over time [81]. Future research examining individual symptom score trajectory patterns over time, particularly differences in physical and psychosocial symptoms, will be important for identifying populations at risk who may require intensification of symptom management for specific symptoms.

5. Conclusions

Overall, this study provides important findings by identifying the nuances of symptom burden trajectories experienced by metastatic breast cancer patients; the relative contribution of chemotherapy, radiotherapy, home care, or supportive care on experienced symptom burden; and the importance of focusing attention on patients who are diagnosed with high symptom burden. These patients may potentially live with high symptom burden for the duration of the disease with the assistance of targeted palliative care and self-management support strategies and education via a personalized, risk-stratified care delivery approach. Further assessment by healthcare providers of patient symptom burden (high/uncontrolled or low/changing over time) would facilitate symptom control in this population of metastatic breast cancer patients with advanced disease who are in a dual state of survivorship and limited mortality [16,82,83]. This study helps set the foundational groundwork for the creation of personalized risk-stratified cancer-care pathways considered critical for cancer care in the future [21]. Future research can build on this research to examine other factors that were not able to be examined in this study, such as disease severity, functional status, and caregiver burden, among others [21], and their role in and contribution to patient symptom burden trajectories over time. Future research can also examine potential variations in trajectories of individual symptom scores for patients [81]. This will enhance the understanding of symptom burden over time in the metastatic breast cancer patient population leading to better symptom burden control, symptom management, and improved quality of life for patients.

Supplementary Materials

The following are available online at https://www.mdpi.com/1718-7729/28/1/87/s1: Supplementary Figure S1—Total Symptom Distress Score Trajectories over Time (95% Confidence Intervals)

Author Contributions

Conceptualization, S.B., R.M., W.P.W., C.Z. and D.H.; methodology, S.B., R.M., W.P.W., C.Z. and D.H.; software, S.B. and R.M.; validation, S.B., R.M., W.P.W., C.Z. and D.H.; formal analysis, S.B., R.M. and D.H.; investigation, S.B., R.M., W.P.W., C.Z. and D.H.; resources, S.B., W.P.W. and D.H.; data curation, S.B., W.P.W. and D.H.; writing-original draft preparation, S.B.; writing-review and editing, S.B., R.M., W.P.W., C.Z. and D.H.; visualization, S.B. and R.M.; supervision, W.P.W. and D.H.; project administration, S.B. and D.H.; funding acquisition, S.B. and W.P.W. All authors have read and agreed to the published version of the manuscript.

Funding

The primary author was supported by trainee grants from the Ontario SPOR Support Unit and the Ontario Ministry of Health and Long-term Care to the Health System Performance Network, as well as funding from the School of Graduate Studies, University of Toronto. The views expressed in this paper are the views of the authors and do not necessarily reflect those of the funders. The funders had no influence on the design and conduct of the study; the collection, management, analysis and interpretation of the data; the preparation, writing, review or approval of the manuscript; or in the decision to submit the article for publication.

Acknowledgments

We would like to thank Phongsack Manivong, Brooke Filsinger, Nancy Shukla, Entela Zaffino, and Lisa Ellison from Cancer Care Ontario for facilitating the acquisition of data for this study. Parts of this material are based on data and information compiled and provided by the Canadian Institute of Health Information, Health Shared Services Ontario, and the Institute of Clinical Evaluative Sciences. However, the analysis, conclusions, opinions, and statements expressed herein are those of the authors, and not necessarily those of the Canadian Institute of Health Information, Health Shared Services Ontario, or the Institute of Clinical Evaluative Sciences.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Canadian Cancer Statistics Advisory Committee. Canadian Cancer Statistics 2018; Canadian Cancer Society: Toronto, ON, Canada, 2018; Available online: http://www.cancer.ca/Canadian-Cancer-Statistics-2018-EN (accessed on 26 August 2018).
  2. Reed, E.; Simmonds, P.; Haviland, J.; Corner, J. Quality of life and experience of care in women with metastatic breast cancer: A cross-sectional survey. J. Pain Symptom Manag. 2012, 43, 747–758. [Google Scholar] [CrossRef]
  3. Mittmann, N.; Isogai, P.K.; Saskin, R.; Liu, N.; Porter, J.M.; Cheung, M.C.; Leighl, N.B.; Hoch, J.S.; Trudeau, M.E.; Evans, W.K.; et al. Population-based home care services in breast cancer: Utilization and costs. Curr. Oncol. 2012, 19, 383–391. [Google Scholar] [CrossRef] [PubMed]
  4. Mittmann, N.; Porter, J.M.; Rangrej, J.; Seung, S.J.; Liu, N.; Saskin, R.; Cheung, M.C.; Leighl, N.B.; Hoch, J.S.; Trudeau, M.; et al. Health system costs for stage-specific breast cancer: A population-based approach. Curr. Oncol. 2014, 21, 281. [Google Scholar] [CrossRef] [PubMed]
  5. Reed, E.; Corner, J. Defining the illness trajectory of metastatic breast cancer. BMJ Support 2015, 5, 358–365. [Google Scholar] [CrossRef] [PubMed]
  6. Hamer, J.; McDonald, R.; Zhang, L.; Verma, S.; Leahey, A.; Ecclestone, C.; Bedard, G.; Pulenzas, N.; Bhatia, A.; Chow, R.; et al. Quality of life (QOL) and symptom burden (SB) in patients with breast cancer. Support Care Cancer 2017, 25, 409–419. [Google Scholar] [CrossRef] [PubMed]
  7. Andre, F.; Slimane, K.; Bachelot, T.; Dunant, A.; Namer, M.; Barrelier, A.; Kabbaj, O.; Spano, J.P.; Marsiglia, H.; Rouzier, R.; et al. Breast cancer with synchronous metastases: Trends in survival during a 14-year period. J. Clin. Oncol. 2004, 22, 3302–3308. [Google Scholar] [CrossRef]
  8. Thomas, A.; Khan, S.A.; Chrischilles, E.A.; Schroeder, M.C. Initial Surgery and Survival in Stage IV Breast Cancer in the United States, 1988–2011. JAMA Surg. 2016, 151, 424–431. [Google Scholar] [CrossRef]
  9. Cancer Care Ontario. Survival Improving for the Four Most Common Cancers in Ontario; Cancer Care Ontario: Toronto, ON, Canada, 2016; Available online: https://archive.cancercare.on.ca/common/pages/UserFile.aspx?fileId=361287 (accessed on 14 January 2019).
  10. Cancer Care Ontario. Ontario Cancer Statistics 2018; Cancer Care Ontario: Toronto, ON, Canada, 2018; Available online: https://www.cancercareontario.ca/en/statistical-reports/ontario-cancer-statistics-2018-report (accessed on 26 August 2018).
  11. Mosher, C.E.; Johnson, C.; Dickler, M.; Norton, L.; Massie, M.J.; DuHamel, K. Living with metastatic breast cancer: A qualitative analysis of physical, psychological, and social sequelae. Breast J. 2013, 19, 285–292. [Google Scholar] [CrossRef]
  12. Park, E.M.; Rosenstein, D.L. Living with Advanced Cancer Unmet Survivorship Needs. North Carol. Med. J. 2014, 75, 279–282. [Google Scholar] [CrossRef][Green Version]
  13. Gomes, B.; Higginson, I.J.; McCrone, P. Effectiveness and cost-effectiveness of home palliative care services for adults with advanced illness and their caregivers. Cochrane Database Syst. Rev. 2013. [Google Scholar] [CrossRef]
  14. Brumley, R.; Enguidanos, S.; Jamison, P.; Seitz, R.; Morgenstern, N.; Saito, S.; McIlwane, J.; Hillary, K.; Gonzalez, J. Increased satisfaction with care and lower costs: Results of a randomized trial of in-home palliative care. J. Am. Geriatr. Soc. 2007, 55, 993–1000. [Google Scholar] [CrossRef]
  15. Tanuseputro, P.; Budhwani, S.; Bai, Y.Q.; Wodchis, W.P. Palliative care delivery across health sectors: A population-level observational study. Palliat Med. 2017, 31, 247–257. [Google Scholar] [CrossRef]
  16. Budhwani, S.; Wodchis, W.P.; Zimmermann, C.; Moineddin, R.; Howell, D. Self-management, self-management support needs and interventions in advanced cancer: A scoping review. BMJ Support 2018. [Google Scholar] [CrossRef]
  17. Harley, C.; Pini, S.; Bartlett, Y.K.; Velikova, G. Defining chronic cancer: Patient experiences and self-management needs. BMJ Support 2015, 5, 343–350. [Google Scholar] [CrossRef] [PubMed]
  18. Lam, W.W.T.; Tsang, J.; Yeo, W.; Suen, J.; Ho, W.M.; Yau, T.K.; Soong, I.; Wong, K.Y.; Kwong, A.; Suen, D.; et al. The evolution of supportive care needs trajectories in women with advanced breast cancer during the 12 months following diagnosis. Support Care Cancer 2014, 22, 635–644. [Google Scholar] [CrossRef]
  19. Schulman-Green, D.; Bradley, E.H.; Knobf, M.; Prigerson, H.; DiGiovanna, M.P.; McCorkle, R. Self-management and transitions in women with advanced breast cancer. J. Pain Symptom Manag. 2011, 42, 517–525. [Google Scholar] [CrossRef] [PubMed]
  20. Aranda, S.; Schofield, P.; Weih, L.; Yates, P.; Milne, D.; Faulkner, R.; Voudouris, N. Mapping the quality of life and unmet needs of urban women with metastatic breast cancer. Eur. J. Cancer Care (Engl.) 2005, 14, 211–222. [Google Scholar] [CrossRef]
  21. Mayer, D.K.; Alfano, C.M. Personalized Risk-Stratified Cancer Follow-Up Care: Its Potential for Healthier Survivors, Happier Clinicians, and Lower Costs. JNCI J. Natl. Cancer Inst. 2019, 111, 442–448. [Google Scholar] [CrossRef]
  22. Barbera, L.; Seow, H.; Howell, D.; Sutradhar, R.; Earle, C.; Liu, Y.; Stitt, A.; Husain, A.; Sussman, J.; Dudgeon, D. Symptom burden and performance status in a population-based cohort of ambulatory cancer patients. Cancer 2010, 116, 5767–5776. [Google Scholar] [CrossRef] [PubMed]
  23. Bubis, L.D.; Davis, L.; Mahar, A.; Barbera, L.; Li, Q.; Moody, L.; Karanicolas, P.; Sutradhar, R.; Coburn, N.G. Symptom burden in the first year after cancer diagnosis: An analysis of patient-reported outcomes. J. Clin. Oncol. 2018, 36, 1103–1111. [Google Scholar] [CrossRef]
  24. Deshields, T.L.; Potter, P.; Olsen, S.; Liu, J. The persistence of symptom burden: Symptom experience and quality of life of cancer patients across one year. Support Care Cancer 2014, 22, 1089–1096. [Google Scholar] [CrossRef] [PubMed]
  25. Shi, Q.; Smith, T.G.; Michonski, J.D.; Stein, K.D.; Kaw, C.; Cleeland, C.S. Symptom burden in cancer survivors 1 year after diagnosis: A report from the American Cancer Society’s Studies of Cancer Survivors. Cancer 2011, 117, 2779–2790. [Google Scholar] [CrossRef] [PubMed]
  26. Seow, H.; Barbera, L.; Sutradhar, R.; Howell, D.; Dudgeon, D.; Atzema, C.; Liu, Y.; Husain, A.; Sussman, J.; Earle, C. Trajectory of performance status and symptom scores for patients with cancer during the last six months of life. J. Clin. Oncol. 2011, 29, 1151–1158. [Google Scholar] [CrossRef] [PubMed]
  27. Tai, S.-Y.; Lee, C.-Y.; Wu, C.-Y.; Hsieh, H.-Y.; Huang, J.-J.; Huang, C.-T.; Chien, C.-Y. Symptom severity of patients with advanced cancer in palliative care unit: Longitudinal assessments of symptoms improvement. BMC Palliat. Care 2016, 15, 32. [Google Scholar] [CrossRef]
  28. Hui, D.; dos Santos, R.; Chisholm, G.B.; Bruera, E. Symptom expression in the last seven days of life among cancer patients admitted to acute palliative care units. J. Pain Symptom Manag. 2015, 50, 488–494. [Google Scholar] [CrossRef] [PubMed]
  29. Kang, J.H.; Kwon, J.H.; Hui, D.; Yennurajalingam, S.; Bruera, E. Changes in symptom intensity among cancer patients receiving outpatient palliative care. J. Pain Symptom Manag. 2013, 46, 652–660. [Google Scholar] [CrossRef] [PubMed]
  30. Ekström, M.; Johnson, M.; Schiöler, L.; Kaasa, S.; Hjermstad, M.; Currow, D. Who experiences higher and increasing breathlessness in advanced cancer? The longitudinal EPCCS Study. Support Care Cancer 2016, 24, 3803–3811. [Google Scholar] [CrossRef] [PubMed]
  31. Bidstrup, P.E.; Christensen, J.; Mertz, B.G.; Rottmann, N.; Dalton, S.O.; Johansen, C. Trajectories of distress, anxiety, and depression among women with breast cancer: Looking beyond the mean. Acta Oncol. 2015, 54, 789–796. [Google Scholar] [CrossRef]
  32. Davis, L.E.; Bubis, L.D.; Mahar, A.L.; Li, Q.; Sussman, J.; Moody, L.; Barbera, L.; Holloway, C.M.; Coburn, N.G. Patient-reported symptoms after breast cancer diagnosis and treatment: A retrospective cohort study. Eur. J. Cancer 2018, 101, 1–11. [Google Scholar] [CrossRef]
  33. Dunn, L.B.; Cooper, B.A.; Neuhaus, J.; West, C.; Paul, S.; Aouizerat, B.; Abrams, G.; Edrington, J.; Hamolsky, D.; Miaskowski, C. Identification of distinct depressive symptom trajectories in women following surgery for breast cancer. Health Psychol. 2011, 30, 683. [Google Scholar] [CrossRef]
  34. Avis, N.E.; Levine, B.J.; Case, L.D.; Naftalis, E.Z.; Van Zee, K.J. Trajectories of depressive symptoms following breast cancer diagnosis. Cancer Epidemiol. Prev. Biomark. 2015, 24, 1789–1795. [Google Scholar] [CrossRef]
  35. Donovan, K.A.; Gonzalez, B.D.; Small, B.J.; Andrykowski, M.A.; Jacobsen, P.B. Depressive symptom trajectories during and after adjuvant treatment for breast cancer. Ann. Behav. Med. 2013, 47, 292–302. [Google Scholar] [CrossRef] [PubMed]
  36. Ecclestone, C.; Chow, R.; Pulenzas, N.; Zhang, L.; Leahey, A.; Hamer, J.; DeAngelis, C.; Bedard, G.; McDonald, R.; Bhatia, A.; et al. Quality of life and symptom burden in patients with metastatic breast cancer. Support Care Cancer 2016, 24, 4035–4043. [Google Scholar] [CrossRef] [PubMed]
  37. Kim, Y.; Yen, I.H.; Rabow, M.W. Comparing symptom burden in patients with metastatic and nonmetastatic cancer. J. Palliat Med. 2016, 19, 64–68. [Google Scholar] [CrossRef] [PubMed]
  38. Bower, J.E.; Wiley, J.; Petersen, L.; Irwin, M.R.; Cole, S.W.; Ganz, P.A. Fatigue after breast cancer treatment: Biobehavioral predictors of fatigue trajectories. Health Psychol. 2018, 37, 1025. [Google Scholar] [CrossRef]
  39. Stanton, A.L.; Wiley, J.F.; Krull, J.L.; Crespi, C.M.; Hammen, C.; Allen, J.J.B.; Barrón, M.L.; Jorge, A.; Weihs, K.L. Depressive episodes, symptoms, and trajectories in women recently diagnosed with breast cancer. Breast Cancer Res. Treat. 2015, 154, 105–115. [Google Scholar] [CrossRef]
  40. Cleeland, C.; von Moos, R.; Walker, M.S.; Wang, Y.; Gao, J.; Chavez-MacGregor, M.; Liede, A.; Arellano, J.; Balakumaran, A.; Qian, Y. Burden of symptoms associated with development of metastatic bone disease in patients with breast cancer. Support Care Cancer 2016, 24, 3557–3565. [Google Scholar] [CrossRef]
  41. Walker, M.S.; Hasan, M.; Yim, Y.M.; Yu, E.; Stepanski, E.J.; Schwartzberg, L.S. Retrospective study of the effect of disease progression on patient reported outcomes in HER-2 negative metastatic breast cancer patients. Health Qual. Life Outcomes 2011, 9, 46. [Google Scholar] [CrossRef] [PubMed]
  42. Cancer Care Ontario. Ontario Cancer Registry; Cancer Care Ontario: Toronto, ON, Canada; Available online: https://www.cancercareontario.ca/en/cancer-care-ontario/programs/data-research/ontario-cancer-registry (accessed on 24 August 2018).
  43. Bruera, E.; Kuehn, N.; Miller, M.J.; Selmser, P.; Macmillan, K. The Edmonton Symptom Assessment System (ESAS): A simple method for the assessment of palliative care patients. J. Palliat Care 1991, 7, 6–9. [Google Scholar] [CrossRef] [PubMed]
  44. Chang, V.T.; Hwang, S.S.; Feuerman, M. Validation of the Edmonton symptom assessment scale. Cancer 2000, 88, 2164–2171. [Google Scholar] [CrossRef]
  45. Ontario Ministry of Health and Long-Term Care. The Ontario Cancer Symptom Management Collaborative (OCSMC); Ontario Ministry of Health and Long-Term Care: Toronto, ON, Canada, 2014.
  46. Cancer Care Ontario. Symptom Assessment Tools: Your Symptoms Matter; Cancer Care Ontario: Toronto, ON, Canada; Available online: https://www.cancercareontario.ca/en/guidelines-advice/symptom-side-effect-management/symptom-assessment-tool (accessed on 20 March 2019).
  47. Zimmermann, C.; Swami, N.; Krzyzanowska, M.; Hannon, B.; Leighl, N.; Oza, A.; Moore, M.; Rydall, A.; Rodin, G.; Tannock, I.; et al. Early palliative care for patients with advanced cancer: A cluster-randomised controlled trial. Lancet 2014, 383, 1721–1730. [Google Scholar] [CrossRef]
  48. Bakitas, M.; Lyons, K.D.; Hegel, M.T.; Balan, S.; Barnett, K.N.; Brokaw, F.C.; Byock, I.R.; Hull, J.G.; Li, Z.; McKinstry, E.; et al. The project ENABLE II randomized controlled trial to improve palliative care for rural patients with advanced cancer: Baseline findings, methodological challenges, and solutions. Palliat Support Care 2009, 7, 75–86. [Google Scholar] [CrossRef]
  49. Box, G.E.; Cox, D.R. An analysis of transformations. J. R. Stat. Soc. Ser. B (Methodol.) 1964, 26, 211–252. [Google Scholar] [CrossRef]
  50. SAS Institute Inc. Box Cox Transformations; SAS Institute Inc.: Cary, NC, USA, 2015. [Google Scholar]
  51. Osborne, J.W. Improving your data transformations: Applying the Box-Cox transformation. Pract. Assess. Res. Eval. 2010, 15, 2. [Google Scholar]
  52. Statistics Canada. Postal Code Conversion File; Statistics Canada: Ottawa, ON, Canada, 2018; Available online: http://www5.statcan.gc.ca/olc-cel/olc.action?objId=92-154-X&objType=2&lang=en&limit=0 (accessed on 31 August 2018).
  53. Statistics Canada. Census Metropolitan Influenced Zone (MIZ); Statistics Canada: Ottawa, ON, Canada, 2015; Available online: https://www12.statcan.gc.ca/census-recensement/2011/ref/dict/geo010-eng.cfm (accessed on 31 August 2018).
  54. Statistics Canada. Postal CodeOM Conversion File Plus (PCCF+) Version 6B, Reference Guide; Statistics Canada: Ottawa, ON, Canada, 2014; Available online: https://mdl.library.utoronto.ca/sites/default/files/mdldata/open/canada/national/statcan/postalcodes/pccfplus/2011/2014nov/82-F0086-XDB-2014v6b-eng.pdf (accessed on 31 August 2018).
  55. Charlson, M.E.; Pompei, P.; Ales, K.L.; MacKenzie, C.R. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J. Chronic Dis. 1987, 40, 373–383. [Google Scholar] [CrossRef]
  56. Lix, L.; Smith, M.; Pitz, M.; Ahmed, R.; Quon, H.; Griffith, J.; Turner, D.; Hong, S.; Prior, H.; Banerjee, A.; et al. Cancer Data Linkage in Manitoba: Expanding the Infrastructure for Research; Manitoba Centre for Health Policy: Winnipeg, MB, Canada, 2016; Available online: http://mchp-appserv.cpe.umanitoba.ca/reference//Candata_web_final.pdf (accessed on 27 August 2018).
  57. Manitoba Centre for Health Policy. Concept: Charlson Comorbidity Index; Manitoba Centre for Health Policy: Winnipeg, MB, Canada, 2010; Available online: http://mchp-appserv.cpe.umanitoba.ca/viewConcept.php?conceptID=1098 (accessed on 22 March 2019).
  58. Barbera, L.; Atzema, C.; Sutradhar, R.; Seow, H.; Howell, D.; Husain, A.; Sussman, J.; Earle, C.; Liu, Y.; Dudgeon, D. Do patient-reported symptoms predict emergency department visits in cancer patients? A population-based analysis. Ann. Emerg. Med. 2013, 61, 427–437.e5. [Google Scholar] [CrossRef] [PubMed]
  59. Azim, H.A.; De Azambuja, E.; Colozza, M.; Bines, J.; Piccart, M. Long-term toxic effects of adjuvant chemotherapy in breast cancer. Ann. Oncol. 2011, 22, 1939–1947. [Google Scholar] [CrossRef] [PubMed]
  60. Canadian Cancer Society. Side Effects of Chemotherapy; Canadian Cancer Society: Toronto, ON, Canada, 2019; Available online: http://www.cancer.ca/en/cancer-information/diagnosis-and-treatment/chemotherapy-and-other-drug-therapies/chemotherapy/side-effects-of-chemotherapy/?region=on (accessed on 21 March 2019).
  61. Canadian Cancer Society. Side Effects of Radiation Therapy; Canadian Cancer Society: Toronto, ON, Canada, 2019; Available online: http://www.cancer.ca/en/cancer-information/diagnosis-and-treatment/radiation-therapy/side-effects-of-radiation-therapy/?region=on (accessed on 21 March 2019).
  62. Jones, B.L.; Nagin, D.S. Proc TRAJ: A SAS Procedure for Group-Based Modeling of Longitudinal Data. In Proceedings of the 135st APHA Annual Meeting and Exposition 2007, Washington, DC, USA, 3–7 November 2007. [Google Scholar]
  63. Jones, B.L.; Nagin, D.S. Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociol. Methods Res. 2007, 35, 542–571. [Google Scholar] [CrossRef]
  64. Nagin, D.S. Group-Based Modeling of Development; Harvard University Press: Cambridge, MA, USA, 2005. [Google Scholar]
  65. Nagin, D.S. Analyzing developmental trajectories: A semiparametric, group-based approach. Psychol. Methods 1999, 4, 139. [Google Scholar] [CrossRef]
  66. Andruff, H.; Carraro, N.; Thompson, A.; Gaudreau, P.; Louvet, B. Latent class growth modelling: A tutorial. Tutor. Quant. Methods Psychol. 2009, 5, 11–24. [Google Scholar] [CrossRef]
  67. Raudenbush, S.W. Comparing personal trajectories and drawing causal inferences from longitudinal data. Annu. Rev. Psychol. 2001, 52, 501–525. [Google Scholar] [CrossRef] [PubMed]
  68. Collins, J.E.; Katz, J.N.; Dervan, E.E.; Losina, E. Trajectories and risk profiles of pain in persons with radiographic, symptomatic knee osteoarthritis: Data from the osteoarthritis initiative. Osteoarthr. Cartil. 2014, 22, 622–630. [Google Scholar] [CrossRef]
  69. Jones, B.L.; Nagin, D.S.; Roeder, K. A SAS procedure based on mixture models for estimating developmental trajectories. Sociol. Methods Res. 2001, 29, 374–393. [Google Scholar] [CrossRef]
  70. SAS Institute Inc. SAS 9.4; SAS Institute Inc.: Cary, NC, USA, 2018. [Google Scholar]
  71. Jones, B. Traj Group-Based Modelling of Longitudinal Data—SAS Examples; Carnegie Mellon University: Pittsburgh, PA, USA, 2017; Available online: https://www.andrew.cmu.edu/user/bjones/example.htm (accessed on 6 June 2018).
  72. Jones, B. Traj Group-Based Modelling of Longitudinal Data—SAS Download; Carnegie Mellon University: Pittsburgh, PA, USA, 2017; Available online: https://www.andrew.cmu.edu/user/bjones/download.htm (accessed on 6 June 2018).
  73. Arrandale, V.; Koehoorn, M.; MacNab, Y.; Kennedy, S.M. How to Use SAS® Proc Traj and SAS® Proc Glimmix in Respiratory Epidemiology; University of British Columbia: Vancouver, BC, Canada, 2006. [Google Scholar]
  74. Gilmour, H.; Stranges, S.; Kaplan, M.; Feeny, D.; McFarland, B.; Huguet, N.; Bernier, J. Longitudinal trajectories of sleep duration in the general population. Health Rep. 2013, 24, 14–20. [Google Scholar] [PubMed]
  75. Kaup, A.R.; Byers, A.L.; Falvey, C.; Simonsick, E.M.; Satterfield, S.; Ayonayon, H.N.; Smagula, S.F.; Rubin, S.M.; Yaffe, K. Trajectories of depressive symptoms in older adults and risk of dementia. JAMA Psychiatry 2016, 73, 525–531. [Google Scholar] [CrossRef]
  76. Vahabi, M.; Lofters, A.; Kumar, M.; Glazier, R.H. Breast cancer screening disparities among urban immigrants: A population-based study in Ontario, Canada. BMC Public Health 2015, 15, 679. [Google Scholar] [CrossRef]
  77. Borugian, M.J.; Spinelli, J.J.; Abanto, Z.; Xu, C.L.; Wilkins, R. Breast cancer incidence and neighbourhood income. Health Rep. 2011, 22, 7. [Google Scholar] [PubMed]
  78. Hurvitz, S.; Guerin, A.; Brammer, M.; Guardino, E.; Zhou, Z.Y.; Viau, D.L.; Wu, E.Q.; Lalla, D. Investigation of adverse-event-related costs for patients with metastatic breast cancer in a real-world setting. Oncologist 2014, 19, 901–908. [Google Scholar] [CrossRef] [PubMed]
  79. Zimmermann, C.; Cheung, W.Y.; Lo, C.; Rodin, G. Edmonton Symptom Assessment System Screening and Depression at the End of Life. J. Clin. Oncol. 2011, 29, 3107–3108. [Google Scholar] [CrossRef]
  80. Baumgartner, S.E.; Leydesdorff, L. Group-based trajectory modeling (GBTM) of citations in scholarly literature: Dynamic qualities of “transient” and “sticky knowledge claims”. J. Assoc. Inf. Sci. Technol. 2014, 65, 797–811. [Google Scholar] [CrossRef]
  81. Howell, D. It is unclear whether specialist palliative care teleconsultation leads to an improvement in patient symptom scores. Evid. Based Nurs. 2018, 21, 1. [Google Scholar] [CrossRef] [PubMed]
  82. Bailey, D.E., Jr.; Steinhauser, K.; Hendrix, C.; Tulsky, J.A. Editorial: Pairing self-management with Palliative Care: Intervening in life-limiting illness. J. Nurs. Healthc. Chronic Illn. 2011, 3, 1–3. [Google Scholar] [CrossRef] [PubMed]
  83. Kirsten, L.; Hobbs, K. Supportive care in advanced breast cancer. In Cancer Forum; The Cancer Council Australia: Sydney, Australia, 2017; p. 20. [Google Scholar]
Curroncol 28 00087 g001
Figure 1. Total Symptom Distress Score (TSDS)* group trajectories over time. *Transformed values are reported. Magnitude of TSDS differences within group trajectories are relative, and not absolute values.
Figure 1. Total Symptom Distress Score (TSDS)* group trajectories over time. *Transformed values are reported. Magnitude of TSDS differences within group trajectories are relative, and not absolute values.
Curroncol 28 00087 g001
Table 1. ESAS assessments per patient.
Table 1. ESAS assessments per patient.
All Patients in Cohort (n = 995)
n%
Age Group a
18–39717.14%
40–4917217.29%
50–5925425.53%
60–6924124.22%
70–7917017.09%
80–99878.74%
Neighborhood Income Quintile a
Quintile 1 (Low Income)19519.60%
Quintile 219619.70%
Quintile 319319.40%
Quintile 421221.31%
Quintile 5 (High Income)19920.00%
Rurality a
Urban88088.44%
Rural11511.56%
Diagnosis Year
201019619.70%
201119419.50%
201220420.50%
201321321.41%
201418818.89%
Charlson Comorbidity Index a
094995.38%
1–3464.62%
Receiving Treatment—Chemotherapy b
No14914.97%
Yes84685.03%
Receiving Treatment—Radiotherapy b
No54454.67%
Yes45145.33%
Receiving Care—Home Care b
No48548.74%
Yes51051.26%
Receiving Care—Palliative Care b
No30730.85%
Yes68869.15%
Survival Time
Between 0 to 1 Year13513.57%
Between 1 to 3 Years27527.64%
Between 3 to 5 Years777.74%
Greater than 5 Years90.90%
Censored c 49950.15%
ESAS Scores Baseline d,e
Mean Total Symptom Distress Score (SD, Range, N)20.33 (15.73, 0.00–79.00, 995)
Mean Pain Score (SD, Range, N)2.33 (2.63, 0.00–10.00, 992)
Mean Tiredness Score (SD, Range, N)0.86 (1.84, 0.00–10.00, 995)
Mean Lack of Appetite Score (SD, Range, N)1.79 (2.44, 0.00–10.00, 994)
Mean Shortness of Breath Score (SD, Range, N)1.47 (2.37, 0.00–10.00, 993)
Mean Nausea Score (SD, Range, N)3.45 (2.81, 0.00–10.00, 995)
Mean Drowsiness Score (SD, Range, N)2.44 (2.85, 0.00–10.00, 991)
Mean Depression Score (SD, Range, N)1.75 (2.51, 0.00–10.00, 995)
Mean Anxiety Score (SD, Range, N)3.04 (2.90, 0.00–10.00, 993)
Mean Wellbeing Score (SD, Range, N)3.21 (2.69, 0.00–10.00, 990)
ED Visits f
Rate of ED Visits per Patient Month (Standard Error, Confidence Interval)0.12 (0.02, 0.12–0.12)
a. Baseline. b. 0–7 days prior on any of 30 ESAS assessments. c. Censored as at end of follow-up time period of December 31, 2015. d. Untransformed values. e. N for symptom scores may differ from sample cohort due to missing symptom scores on some assessments. f. Patient month is calculated as total number of months patient was observed in study period (not limited to period of first 30 ESAS assessments). SD = Standard Deviation, ED = Emergency Department.
Table 2. Summary of model selection results.
Table 2. Summary of model selection results.
Number of GroupsPolynomial OrderBIC alog Bayes Factor b
14−33,389.92-
244−29,166.78>1000
3444−27,620.32>1000
44440−27,075.20>1000
544401−26,746.86>100
6444000−26,486.85>100
a. BIC=Bayesian information criteria. Model with lowest BIC value is preferred. b. A log Bayes factor of >10 indicates very strong evidence of model parsimony.
Table 3. Patient characteristics by trajectory group (unadjusted).
Table 3. Patient characteristics by trajectory group (unadjusted).
All Patients in Cohort
Group 1Group 2Group 3Group 4Group 5Group 6
Total N89138111233265159
n%n%n%n%n%n%
Age Group a,b
18–3977.87%139.42%109.01%198.15%145.28%85.03%
40–491415.73%2014.49%2320.72%4318.45%4015.09%3220.13%
50–592528.09%2820.29%3027.03%6126.18%7528.30%3522.01%
60–691921.35%3928.26%2623.42%6126.18%6424.15%3220.13%
70–791820.22%2820.29%1614.41%3414.59%4115.47%3320.75%
80–9966.74%107.25%65.41%156.44%3111.70%1911.95%
Neighbourhood Income a
Quintile 1 (Low Income)1921.35%2719.57%1210.81%4519.31%5320.00%3924.53%
Quintile 21112.36%2316.67%2522.52%5121.89%5520.75%3119.50%
Quintile 32022.47%2820.29%1917.12%4820.60%5219.62%2616.35%
Quintile 42325.84%3021.74%2825.23%4720.17%4717.74%3723.27%
Quintile 5 (High Income)1617.98%3021.74%2724.32%4218.03%5821.89%2616.35%
Rurality a
Urban7483.15%12389.13%9989.19%20788.84%23086.79%14792.45%
Rural1516.85%1510.87%1210.81%2611.16%3513.21%127.55%
Diagnosis Year a
20101516.85%2518.12%2118.92%4619.74%5119.25%3823.90%
20111921.35%2014.49%2018.02%5021.46%5320.00%3220.13%
20121617.98%2316.67%2623.42%5121.89%5520.75%3320.75%
20132123.60%3021.74%2320.72%4921.03%5922.26%3119.50%
20141820.22%4028.99%2118.92%3715.88%4717.74%2515.72%
Charlson Comorbidity Index c,d
032997.34%22797.42%24793.21%14691.82%
1–392.66%62.58%186.79%138.18%
Receiving Treatment—Chemotherapy b,d
No1213.48%1712.32%98.11%2812.02%5219.62%3119.50%
Yes7786.52%12187.68%10291.89%20587.98%21380.38%12880.50%
Receiving Treatment—Radiotherapy b
No5359.55%8561.59%6154.95%12051.50%14153.21%8452.83%
Yes3640.45%5338.41%5045.05%11348.50%12446.79%7547.17%
Receiving Care—Home Care b
No4752.81%7151.45%5347.75%10545.06%13350.19%7647.80%
Yes4247.19%6748.55%5852.25%12854.94%13249.81%8352.20%
Receiving Care—Palliative Care b,d
No4247.19%5136.96%4136.94%7331.33%6725.28%3320.75%
Yes4752.81%8763.04%7063.06%16068.67%19874.72%12679.25%
ER Visits/Person Month e
Mean ER Visits/Person Month0.100.100.080.120.140.17
a. Baseline b. On first 30 ESAS assessments following diagnosis c. Data for Groups 1–3 have been merged due to small cell numbers d. Bivariate relationship between group and covariate significant at p = 0.05 level e. Bivariate relationship not examined.
Table 4. Parameter estimates, errors, tests, and p-values for adjusted model.
Table 4. Parameter estimates, errors, tests, and p-values for adjusted model.
Group Number (Membership %)ParameterEstimateStandard ErrorTestp-Value
1 (11.49%)Chemotherapy−0.100.06−1.670.0944
Radiotherapy *0.600.134.71<0.0001
Home Care *0.350.093.820.0001
Palliative Care *0.380.075.35<0.0001
2 (9.86%)Chemotherapy *−0.150.06−2.660.0079
Radiotherapy *0.430.113.760.0002
Home Care0.110.081.470.1412
Palliative Care0.120.061.890.0592
3 (12.10%)Chemotherapy−0.060.06−1.100.2734
Radiotherapy *0.430.114.010.0001
Home Care0.020.100.190.8466
Palliative Care *0.370.065.68<0.0001
4 (24.63%)Chemotherapy−0.070.04−1.740.0822
Radiotherapy *0.250.083.360.0008
Home Care *0.170.062.810.0050
Palliative Care *0.340.057.47<0.0001
5 (26.48%)Chemotherapy *−0.160.04−3.800.0001
Radiotherapy *0.240.083.020.0025
Home Care *0.220.063.920.0001
Palliative Care *0.150.043.430.0006
6 (15.44%)Chemotherapy0.010.050.180.8568
Radiotherapy *0.300.112.830.0047
Home Care *−0.140.07−2.020.0430
Palliative Care−0.0040.06−0.070.9469
* Significant at p = 0.05.
Table 5. Clinical parameters by trajectory group for adjusted model.
Table 5. Clinical parameters by trajectory group for adjusted model.
Group 1 (n = 121)Group 2 (n = 93)Group 3 (n = 114)Group 4 (n = 245)Group 5 (n = 271)Group 6 (n = 151)
Chemotherapy (Y)88.43%86.02%87.72%88.98%81.18%80.13%
Radiotherapy (Y)36.36%43.01%47.37%48.16%46.86%45.03%
Home Care (Y)52.89%49.46%50.88%53.06%49.45%51.66%
Palliative Care (Y)66.12%58.06%62.28%66.53%74.17%78.81%
Radiotherapy + Chemotherapy (Y)33.88%36.56%40.35%42.45%37.64%38.41%
Radiotherapy + Home Care (Y)23.14%20.43%25.44%26.94%25.09%25.17%
Radiotherapy + Palliative Care (Y)25.62%24.73%32.46%35.1%35.06%37.09%
Home Care + Chemotherapy (Y)50.41%45.16%46.49%48.16%43.17%42.38%
Home Care + Palliative Care (Y)38.84%32.26%33.33%37.55%39.85%41.06%
Palliative Care + Chemotherapy (Y)60.33%52.69%54.39%61.22%62.73%62.91%
Chemotherapy + Radiotherapy + Home Care + Palliative Care (Y)17.36%12.90%15.79%17.96%16.97%18.54%
Time to Death <= 3 Years44.63%19.35%20.18%41.22%47.23%56.95%
Time to Death >3 Years5.79%15.05%10.53%8.98%6.64%8.61%
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Budhwani, S.; Moineddin, R.; Wodchis, W.P.; Zimmermann, C.; Howell, D. Longitudinal Symptom Burden Trajectories in a Population-Based Cohort of Women with Metastatic Breast Cancer: A Group-Based Trajectory Modeling Analysis. Curr. Oncol. 2021, 28, 879-897. https://doi.org/10.3390/curroncol28010087

AMA Style

Budhwani S, Moineddin R, Wodchis WP, Zimmermann C, Howell D. Longitudinal Symptom Burden Trajectories in a Population-Based Cohort of Women with Metastatic Breast Cancer: A Group-Based Trajectory Modeling Analysis. Current Oncology. 2021; 28(1):879-897. https://doi.org/10.3390/curroncol28010087

Chicago/Turabian Style

Budhwani, Suman, Rahim Moineddin, Walter P. Wodchis, Camilla Zimmermann, and Doris Howell. 2021. "Longitudinal Symptom Burden Trajectories in a Population-Based Cohort of Women with Metastatic Breast Cancer: A Group-Based Trajectory Modeling Analysis" Current Oncology 28, no. 1: 879-897. https://doi.org/10.3390/curroncol28010087

APA Style

Budhwani, S., Moineddin, R., Wodchis, W. P., Zimmermann, C., & Howell, D. (2021). Longitudinal Symptom Burden Trajectories in a Population-Based Cohort of Women with Metastatic Breast Cancer: A Group-Based Trajectory Modeling Analysis. Current Oncology, 28(1), 879-897. https://doi.org/10.3390/curroncol28010087

Article Metrics

Back to TopTop