|Home | About | Journals | Submit | Contact Us | Français|
Economic burden is emerging as a crucial dimension in our understanding of adjustment to cancer during treatment. Yet, economic burden is rarely examined in cancer survivorship. The goal of this paper is to describe the effect of economic hardship and burden among women with breast cancer.
We examined baseline and follow-up (3 and 6 month) data reported by 132 stage I and II breast cancer survivors assigned to the Wait Control arm of the Breast Cancer Education Intervention (BCEI), a clinical trial of education and support interventions. Repeated measures models fitted with linear mixed models were used to examine relationships between aspects of economic burden and overall quality of life (QOL) scores. Structural equation models (SEM) were used to examine the relationship between overall economic burden and QOL.
Nineteen economic events were reported. The proportion of survivors who reported increase in insurance premiums increased in the 6-month study period (p=.022). The proportion of survivors reporting change in motivation (p=.016), productivity (p=.002), quality of work (p=.01), days missed from work (p<.001) and sacrificing other things (p=.001) declined. An increase in economic events was significantly associated with poorer quality of life at each of the study time points.
Economic burden of breast cancer extends into post-treatment survivorship. Better understanding of economic impact and managing economic burden may help maintain QOL.
About 2.5 million women in the United States are living with a history of breast cancer 1. Survival from breast cancer is high with 89% survival at five years 2. While cancer survivorship care plans often include information about cancer surveillance, health maintenance, and psychosocial care, 3,4,5 economic burden and financial distress are not well described. Yet, economic burden is emerging as a crucial dimension in our understanding of cancer and cancer survivorship. 6–9
Economic burden is defined as “the loss of economic resources and opportunities associated with the occurrence of cancer.” 6–10, Loss of economic resources may lead to reduction in income or change in economic lifestyle, need to borrow money, depletion of savings, declaring bankruptcy, sacrificing family plans, and other troublesome events. Loss of opportunities may lead to difficulty in keeping or finding a job, and decreased productivity and quality of work 11. Since economic burden may be associated with changes in quality of life (QOL), we collected economic data within the Breast Cancer Education Intervention (BCEI), a randomized trial of quality of life psychoeducational support interventions for breast cancer survivors. In this paper, we conducted a secondary analysis of economic burden and its relationship to QOL among early stage breast cancer survivors in the Wait Control arm of the BCEI.
Data from 132 early-stage breast cancer survivors, enrolled in the Wait Control arm of the BCEI, a randomized trial of quality of life psychoeducational support interventions for breast cancer survivors in the first year after primary treatment, were used in this study. The overall design and goals of the BCEI were previously published 12. Briefly, breast cancer survivors were recruited from a regional cancer center and private oncology offices in the Southeastern United States. Eligible participants were: at least 21 years of age, had histologically confirmed Stage 0-II breast cancer with no evidence of recurrent or metastatic disease, within two years of diagnosis, and a minimum of one month since completion of surgery, radiation therapy and/or chemotherapy. Participants were also eligible if they were receiving hormonal therapy (i.e., aromatase inhibitor or tamoxifen) at study entry. Institutional review boards at all participating institutions which complied with Health Insurance Portability and Accountability Act Guidelines, approved the study.
After completing baseline measures, 261 BCEI participants were randomly assigned to either the Experimental or Wait Control arm. The Experimental intervention arm (n=129) received three weekly education and support sessions delivered by an Intervention Nurse. The sessions focused on differential aspects of QOL including: (1) physical (fatigue, pain, menopausal symptoms, and change in body image); (2) psychological adjustment, social and family relationships, work and financial concerns; and (3) spirituality and meaning in illness. They were also given a 168-page binder of written information, and three audiotapes of instruction to supplement their learning. They were contacted for monthly follow up telephone calls or in person visits for the next five months. Total participation time was six months.
Participants assigned to the Wait Control arm (n=132) received baseline and monthly telephone calls and visits. At six months of study participation, they received the education and support intervention as described above. The analyses for this paper included only participants in the Wait Control arm, in order to avoid confounding effects on the relationship between QOL and economic burden resulting from the experimental intervention.
Baseline data included breast cancer treatment including stage of disease, type of surgery, radiation therapy, chemotherapy, hormonal, and anti-HER2 therapy were collected by survey. Sociodemographic variables included age, race, ethnicity, education, marital status, work status, family income, and type of insurance.
Breast cancer survivors’ quality of life was assessed using the QOL-BCS, which is a 50 item, 10 point ordinal scale that evaluates overall QOL and four individual QOL domains of physical, psychological, social, and spiritual well being. The QOL-BCS was adapted from the QOL-Cancer Survivors Scale (QOL-CS) 13. Test-retest reliability was 0.89 and Cronbach’s alpha was 0.93; both scores were established using the original QOL-CS. In the present study, baseline alpha coefficients for overall QOL and domain scores were as follows: Overall QOL = 0.92; Physical well-being = 0.74; Psychological well-being = 0.91; Social well-being = 0.73; and Spiritual wellbeing = 0.67. Each QOL item is rated on a 10 point scale asking participants to rate their QOL from 0 (= no problem) to 10 ( = worst problem). Thus, a lower score closer to 0 reflected fewer problems and higher QOL; while a higher score reflected more problems and lower QOL. For the present study, the overall QOL score was calculated as the average of the item scores in the physical, psychological, and social domains. For each of these domains, subscale scores were calculated as well, as averages of the item scores within each domain. The spiritual domain was not included in the computation of the overall QOL score or included in the analyses. The alpha coefficient for the 43 items used to compute the overall QOL score remained at 0.92 after exclusion of the seven spiritual domain items.
Breast cancer finances were assessed using the BCFS Inventory, a 42 item questionnaire containing questions about crucial aspects of cancer-related economic burden including work (i.e., changes in motivation, productivity, quality and quantity of work), financial hardship (i.e., changes in self and spouse/partner’s income, applying for unemployment benefits, finding second jobs, selling property, borrowing money, using savings, declaring bankruptcy, changing economic lifestyle, or missing bill payments, and increases in insurance premiums or reaching health coverage limits), and out of pocket expenses 14. Of particular importance to this analysis were 19 BCFS economic burden items that were specific to changes in work and financial hardship. The BCFS inventory was adapted from Given and colleagues 14, one of the first economic inventories specific for cancer patients. We used the BCFS in a prior analysis of the BCEI that examined out of pocket costs between minority and Caucasian breast cancer survivors 15.
Baseline socio-demographic and treatment characteristics were tabulated. Patterns of missing data were explored. Descriptive statistics for variables of interest were computed at each of the three study time points.
To examine changes in economic burden over the study period, the responses to each of the 19 BCFS economic burden items were tabulated at each of the three study time points. Generalized linear mixed models fitted with binomial distributions and logit links and variance/covariance structures that accounted for lack of independence among repeated measures on the same participants were used to test for differences in reporting of events between baseline and the two follow-up time points. All generalized linear mixed models were fitted using the GLIMMIX procedure in SAS v9.2 software 16.
Structural equation models (SEM) were used to explore the relationship between economic burden items and QOL. An SEM model allows testing of a conceptual model on collected data by using the hypothesized conceptual model to try and reproduce the sample variance/covariance or correlation matrix of the observed variables. Because it is the whole variance/covariance (or correlation) matrix that is modeled (as opposed to simpler linear models in which a single outcome is modeled as a function of some predictors), SEM models allow simultaneous testing of multiple hypothesized relationships among the observed variables, including intermediate relationships. Two types of relationships can be modeled between two variables: hypothetical causal or co-variation (non-causal) relationships. The models are presented graphically, with one-way arrows indicating causal relationships, and two-way arrows indicating co-variation relationships. The algorithms for estimating and testing the model parameters are dependent on the underlying distributions of the observed variables, with multivariate normality commonly assumed. The main objective of the modeling process is to attain a parsimonious model that accurately reproduces the sample variance/covariance (or correlation) matrix of the data, and, at the same time, makes sense from a conceptual standpoint.
Goodness of fit is examined using a χ2 test, with the null hypothesis being that the model provides good fit to the data; consequently, a significant result, i.e. a p-value smaller than 0.05, indicates inadequate fit. Because the results from a χ2 goodness-of-fit test may be influenced by the sample size, other indices commonly used to assess model fit include CFI (comparative fit index), GFI (goodness-of-fit index), and RMSEA (root mean square error of approximation). Typically, CFI and GFI values greater than 0.95, and RMSEA values smaller than 0.06, are all indicative of good fit 17.
The sum of affirmative responses to the 19 BCFS economic burden items was computed to serve as a proxy for overall economic burden. In the fitted SEM models, the physical, psychological, and social domain scores were used as indicators of a latent QOL variable. Multivariate normality could be safely assumed for the QOL physical, psychological, and social domain scores, since these are computed as averages. However, the baseline characteristics were either binary indicators or ordinal variables, and the number of reported economic burden items was a right-skewed discrete count variable. Thus, SEM model parameters were estimated using Browne's asymptotically distribution-free method implemented in the SAS procedure TCALIS, instead of the maximum likelihood fitting algorithms which assume multivariate normality for all variables in the model.
The initial SEM model included socio-demographic and treatment characteristics and the number of BCFS economic burden items at baseline as hypothetical causes for baseline QOL. Baseline QOL and number of reported BCFS economic burden items at Month 3 were hypothesized as causative for Month 3 QOL, while Month 3 QOL and economic burden items at Month 6 were hypothesized as causative for Month 6 QOL. Initial fit of the model was improved by an iterative process that consisted of trimming and adding reasonable relationships until an acceptable fit was achieved.
We examined the time-averaged association between overall QOL scores and the following baseline socio-demographic and treatment characteristics: race, education, age, rural residence, marital status, employment status, income, type of surgery received, radiation therapy, and chemotherapy. Next, repeated measures models fitted with linear mixed models were used to examine time-averaged relationships between each of the 19 BCFS economic burden items and overall QOL scores, controlling for the baseline socio-demographic and treatment characteristics that were significantly associated to QOL scores. Then, an initial model for QOL scores was fitted using as predictors the baseline socio-demographic and treatment characteristic and the BCFS economic burden items that were individually significantly associated to the QOL scores. A time effect was included to adjust for differences in QOL scores among the three time-points. This model was trimmed to include only significant predictors. The linear mixed models were fitted with variance/covariance structures that accounted for lack of independence among repeated observations on the same participants using the MIXED procedure in SAS.
A total of 132 breast cancer survivors were included in the present analyses. Data from only one participant were missing at Month 6. Table 1 shows participants' baseline socio-demographic and treatment characteristics. The modal participant was Caucasian, between 46–65 years of age, married, having some college education, working full time, with a family income of more than $50,000. The majority of survivors were within the first year of diagnosis, received combined lumpectomy and radiation therapy as primary treatment, and adjuvant chemotherapy. Over 78% of survivors were on hormonal therapy.
Descriptive statistics for QOL overall, Physical, Psychological, and Social QOL domain scores at baseline, Month-3, and Month-6 are listed in Table 2. Mean scores for the QOL overall and domain mean scores ranged from 1.91 to 3.90. The possible range for the scores is 0–10. Lower scores indicate fewer problems and higher perceived QOL.
Survivors reported a mean of 2.94 economic burden items at baseline (range 0–11), 2.45 at month 3 (range 0–13), and 2.25 at month 6 (range 0–14). Overall, the median was 2 at baseline and 1 at Months 3 and 6. Table 3 shows the proportion of survivors reporting each of the 19 BCFS economic burden items at the three time points. Participants' responses differed significantly from baseline in six of the 19 items. The proportion of survivors reporting changes in motivation to work, productivity and quality of work, and missing days of work consistently declined from baseline to Month 6. More than 40% of women at baseline reported sacrificing things like vacations, while fewer reported this item at Months 3 and 6. On the contrary, a significantly higher percentage of participants reported an increase in insurance premiums at Months 3 and 6 compared to baseline.
Figure 1 illustrates the standardized path coefficients of the final SEM model for QOL and economic burden. The final model provides good fit to the data, as indicated by the non-significant χ2 goodness-of-fit test and the values of the three additional fit indices. From the hypothesized variables affecting QOL at baseline (i.e, number of economic burden events, education, age, marital status, income, and chemotherapy) only the paths from number of economic burden events and chemotherapy were significant. The effect of the number of events (a proxy for overall economic burden) on QOL appeared to be stronger at baseline and affecting overall quality of life. At Months 3 and 6, economic burden affected mostly the social dimension of QOL.
For the baseline socio-demographic and treatment characteristics, the effects on QOL were as follows: higher education, never married status, and receipt of chemotherapy were associated with lower perceived QOL (higher QOL scores). Conversely, age older than 66 and higher family income (i.e., higher than $50,000) were associated with higher perceived QOL (lower QOL scores). Six BCFS economic burden items (motivation to work, productivity of work, quality of work, days missed from work, change in spouse/partner's income, and sacrificing things like vacations) were each significantly associated with QOL after controlling for baseline socio-demographic and treatment characteristics. Results of the initial and final linear mixed models are shown in Table 4. The final model for QOL scores included three BCFS economic burden items and two socio-demographic and treatment characteristics: reports of adverse changes in motivation to work, missed days from work, and sacrificing things like vacations were significantly associated to higher QOL scores (lower perceived QOL), controlling for marital status and chemotherapy.
In this group of mainly insured breast cancer survivors, we found that in the six month period following completion of breast cancer treatment, more than 50% reported at least one economic burden event related to either work or financial hardship. More than a quarter reported changes in income or sacrificing things like family plans over a 6 month period, and among those who worked, more than 15% reported changes in motivation, productivity or quantity (missed days) of work. These events, in turn, were negatively associated with QOL.
Even though the majority of our participants had health insurance, they were not spared financial hardship associated with the disease. In fact, over time more women reported increases in insurance premiums. Our findings are consistent with Arozullah and colleagues who noted that survivors with comprehensive health insurance continued to report having economic burden 18. Yet, cancer survivorship plans rarely include a discussion of economic burden related to insurance premiums and how to manage out of pocket costs for exceeded insurance benefits.
In general, data indicate that a cancer diagnosis leads to higher economic burden 19–22. Cancer survivors, for example, have higher out-of-pocket costs for medical care than persons who do not experience cancer 20,21,23. Cancer survivors also have a higher risk for high economic burden compared with patients with other chronic illnesses 22. A diagnosis of cancer is one of the major reasons for declaring bankruptcy due to medical events which was one of the economic burden events that we also measured 24. Therefore, it is reasonable to assume that the economic burden events reported by our BCEI participants were related to the cancer diagnosis. Moreover, breast cancer can be particularly burdensome due to its sequelae. Management of breast cancer complications such as lymphedema 25 and recurrence 26 have been associated with an increase in economic costs and burden. Similarly, chemotherapy treatment is a predictor of greater financial problems after treatment has ended 27,28. Our study findings also supported that chemotherapy was associated with more economic burden events.
Our results indicated that survivors reported persistent economic events in the months after initial active treatment. These findings are consistent with Gordon and colleagues who found that increased health service expenditures and lost income were the most common sources of economic burden identified by breast cancer survivors up to 18 months after diagnosis. 29
Most BCEI participants worked either full or part time during treatment, and continued to work after treatment ended. Changes in motivation, productivity, and quality of work declined significantly over time. Six months after study entry, they regained work productivity. These findings are similar to Rasmussen et al 11 who conducted qualitative interviews with cancer survivors to examine diverse patterns of work, and found that survivors described work as important in establishing their identity, and creating social relationships. Survivors who maintained regular work schedules during treatment were able to retain productivity 11. On the other hand, Maunsell found that breast cancer survivors who interrupted work schedules during treatment and later returned to work after cancer treatment ended, reported negative outcomes including job loss, demotion, unwanted changes in work, and problems with their employer. These survivors also reported a personal change in attitudes and diminished physical capacity 30. These findings suggest the need for future studies to help survivors continue to work or modify work while on treatment, rather than to take a leave of absence and reenter the workplace after treatment ends.
While BCEI participants reported economic burden, other investigators found extreme economic hardship reported by minority and underserved women with cancer. Darby and colleagues noted that underserved African American breast cancer survivors reported economic hardship because of lack or inadequate health insurance coverage 31. Ell and colleagues found medical cost, income, and financial stress were high in a population of low income Hispanic women receiving cancer treatment 32. Gray reported financial burden among rural Canadian breast cancer survivors 33.
The impact of increased economic costs can result in a “trade-off” between paying for breast cancer or paying for regular family expenses 34. Sherwood interviewed 22 breast and ovarian cancer survivors who reported having to access retirement or savings to pay for breast cancer expenses. Survivors worried about their future financial expenses and the impact on their work income 34. Our study showed similar findings in that more than a quarter of survivors reported “sacrificing other things” like vacations, although this proportion declined significantly over the six month period from 40% to 30%. Financial difficulties often affect the cancer survivorship experience and can lead to psychological distress 32,35–37. Gupta et al. also found an association between perceptions of financial difficulty and lower satisfaction with QOL 37. Kobayashi and colleagues in Japan reported the negative impact of lower family income and loss of employment on QOL among Japanese cancer survivors 38. Yet, Miller and colleagues found that when economic concerns were addressed, QOL among advanced cancer survivors improved 39.
While our study population focused on breast cancer survivors, the findings have relevance to the practice and research in gynecologic oncology. First, the Society of Gynecologic Oncology (SG0) Breast Cancer Task Force Mission Statement published in 2008 is “To promote the provision of comprehensive breast health and cancer care of women, including education, research, screening, prevention and treatment (p7).” 40 Authors of the SGO Statement further argue for a comprehensive model of care among physicians who can encompass all physical and psychological aspects of diagnosis, initial and adjuvant treatment, and surveillance. This statement resonates with breast cancer survivors who neither routinely nor exclusively resume cancer surveillance and follow-up with their medical oncologists. Survivors may choose to return for surveillance to see their regular primary care provider or gynecologist. In support of this evidence, a 2007 SGO Strategic Planning Survey found that 67% of gynecologic oncology respondents reported seeing five patients having a personal history of breast cancer per month, with 26% reported seeing up to eleven breast cancer survivors per month40.
BCEI economic burden findings may also have relevance for the growing numbers of gynecologic oncology survivors who are likewise younger, employed, and have a full plate of family and social responsibilities. Data, however, are sparse and yet are vitally needed to determine whether gynecologic oncology survivors face similar or other concerns in economic burden and hardship that have not yet been identified.
Several strengths are identified in our study. First, this is one of the first studies to examine changes in economic burden among women in survivorship after treatment has ended. Thus, the findings add to our understanding of economic burden occurring in survivorship. Second, survivors are living longer and may face financial hardship and economic burden that are neither planned nor expected. Thus, QOL interventions that specifically include managing the potential for hidden or unknown expenses or economic downfalls are warranted.
Researchers interested in evaluating financial concerns and economic burden can begin to test interventions that can satisfactorily address financial concerns during survivorship. With a longer expectation of survival and differential changes over time and their influence on quality of life is a rich arena for future prospective, longitudinal evaluations of the effects on quality of life.
Several limitations are also acknowledged. First, the homogenous study participants (e.g., Caucasian, working, and income) as noted earlier, do not reflect the economic burden experienced by women of color, rural, and/or poor women. While the BCFS added rich descriptive data, we were not able to address changes over time because of the binary yes/no nature of the questions. Moreover, the surveys had inherent recall difficulties, but these may apply more specifically to the out of pocket costs. With worse toxicity from chemotherapy, survivors may also report worse recollection of financial impact. If there were recall difficulties, we may have underestimated economic burden. Thus, future investigations may warrant additional cancer survivor-specific survey development. We also acknowledge Gupta and colleagues’ recommendations that there is not yet a single assessment survey that explores the many facets of economic burden and financial difficulty 37.
Economic burden is an essential component that contributes to our understanding of cancer survivorship, and provides a glimpse into the everyday life and financial pressure facing breast cancer survivors. This analysis provides additional evidence that economic burden associated with breast cancer treatment continues into post-treatment survivorship. Future interventions for other cancer survivors that include a discussion about economic impact of cancer, and ways to manage economic burden may help maintain or improve QOL outcomes over time.
Economic burden is emerging as a crucial dimension in cancer survivorship. We explored economic burden facing 132 breast cancer survivors over a six month period. An increase in economic events was associated with poorer quality of life.
Grant support from the National Institute of Nursing Research (NINR) and the Office of Cancer Survivorship at the National Cancer Institute (#5R01NR5332).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest Statement
The authors declare that there are no conflicts of interest.
Financial Disclosure: None