Abstract

Background

Wage losses after breast cancer may result in considerable financial burden. Their assessment is made more urgent because more women now participate in the workforce and because breast cancer is managed using multiple treatment modalities that could lead to long work absences. We evaluated wage losses, their determinants, and the associations between wage losses and changes for the worse in the family's financial situation among Canadian women over the first 12 months after diagnosis of early breast cancer.

Methods

We conducted a prospective cohort study among women with breast cancer from eight hospitals throughout the province of Quebec. Information that permitted the calculation of wage losses and information on potential determinants of wage losses were collected by three pretested telephone interviews conducted over the year following the start of treatment. Information on medical characteristics was obtained from medical records. The main outcome was the proportion of annual wages lost because of breast cancer. Multivariable analysis of variance using the general linear model was used to identify personal, medical, and employment characteristics associated with the proportion of wages lost. All statistical tests were two-sided.

Results

Among 962 eligible breast cancer patients, 800 completed all three interviews. Of these, 459 had a paying job during the month before diagnosis. On average, these working women lost 27% of their projected usual annual wages (median = 19%) after compensation received had been taken into account. Multivariable analysis showed that a higher percentage of lost wages was statistically significantly associated with a lower level of education ( Ptrend = .0018), living 50 km or more from the hospital where surgery was performed ( P = .070), lower social support ( P = .012), having invasive disease ( P = .086), receipt of chemotherapy ( P < .001), self-employment ( P < .001), shorter tenure in the job ( Ptrend < .001), and part-time work ( P < .001).

Conclusion

Wage losses and their effects on financial situation constitute an important adverse consequence of breast cancer in Canada.

CONTEXT AND CAVEATS
Prior knowledge

Wage losses for breast cancer patients and the factors associated with the proportion of wages lost because of breast cancer had not been assessed adequately.

Study design

In this prospective cohort study, breast cancer patients were interviewed via telephone about their cancer treatment, work absence, and compensation received. Analysis of variance was used to identify factors associated with the proportion of annual wages lost in the year after diagnosis.

Contribution

This study provides an estimate of the proportion of wages lost for a large number of women that takes compensation into account and provides needed information on some potential determinants of the proportion of wages lost.

Implications

Lost wages are an important issue for women with breast cancer, one that patients should be made aware of. The results should inform reflection in Canada and elsewhere on changes to public policies concerning insurance that would help workers facing severe illness.

Limitations

Potential wage losses due to failure to obtain job promotions or to receive cost-of-inflation pay increases were beyond the scope of this study.

Assessment of the economic burden of breast cancer has mainly concentrated on costs to the health-care system ( 1 ), and costs from the patient's perspective have generally been overlooked. However, economic issues for cancer patients are now correctly recognized as important (2,3), and more attempts have been made in recent years to evaluate the economic impact of cancer.

Several observations suggest that salary losses are an important component of the economic burden of breast cancer for patients. First, in previous studies, wage losses represented a substantial proportion of the cancer patient's total costs ( 4–9 ). These results were supported by findings from a survey conducted by a Canadian breast cancer advocacy group, in which wage losses were also identified as a source of financial strain among women with breast cancer ( 10 ). In a qualitative study, some women judged them to be the most important financial cost of the disease ( 11 ). Second, wage losses may now affect a larger proportion of women than previously because more women work than in earlier cohorts ( 12 ) and more are diagnosed with breast cancer before age 65 ( 13 , 14 ). Third, breast cancer is now managed with multiple treatment modalities, and, as a result, a substantial proportion of women receive two or more kinds of adjuvant treatment. These changes entail a more protracted treatment course and may increase the time women need to take off work. Indeed, in a recent population-based study Canadian women reported a mean work absence of 6.7 months in the first year following diagnosis ( 15 ).

Although the effects of breast cancer on working conditions have been studied ( 14 , 16–18 ), wage losses to patients with early-stage breast cancer have rarely been quantified. Some of the previous assessments of wage losses for breast cancer patients also included patients with other types of cancer (for whom costs may differ considerably) without presenting specific estimates for breast cancer ( 8 , 9 , 19–22 ). In recent Australian ( 5 ) and Swedish ( 4 ) studies, wage losses were estimated for an entire year, but most other studies among breast cancer patients have assessed wage losses for limited periods of time ( 6 , 24 , 25 ) or only for a specific event in the treatment trajectory ( 23 , 26 , 27 ).

Our aim was to estimate the burden from wage losses for Canadian patients by 1) determining the proportion of annual wages lost over the first 12 months after a diagnosis of early-stage breast cancer; 2) identifying personal, disease, treatment, and employment characteristics that were associated with wage losses; and 3) describing effects of wage losses on changes for the worse in the family's financial situation during this period.

Subjects and Methods

Study Design

We conducted a prospective cohort study assessing financial costs for breast cancer patients and their caregivers in the 12 months after diagnosis using cost identification, an analytic approach that attempts to account for all costs incurred by a condition ( 28 ). Although we measured both out-of-pocket costs and wage losses, we focus here only on wage losses experienced by patients.

Subjects

Consecutive series of women with a histologically confirmed new diagnosis of nonmetastatic breast cancer first treated between January 1, 2003, and December 23, 2003, inclusively, in one of eight hospitals in several cities in the province of Quebec who met study eligibility criteria were invited to participate. The eight hospitals were selected because they serve women from both urban and rural areas and travel from home to treatment center to obtain treatments may affect costs. Potentially eligible participants were identified at each hospital through examination of operating lists for breast surgery and pathology reports. A specially trained nurse provided a detailed explanation of the study to potential participants. Each participating hospital's Ethics Review Committee approved the study.

Women with a previous history of breast or other cancer or distant metastasis at diagnosis or those for whom a telephone interview was impossible (this was due to lack of a telephone, insufficient fluency in French, hearing problems, or other physical or psychological problems) were ineligible. Nuns were considered to be ineligible because they receive no wages as individuals. Women enrolled in the study who died or experienced recurrence or a new primary cancer during the study period were not considered in this analysis. All participants provided signed informed consent.

Cost Interview Development and Data Collection

When we considered undertaking this study, no interview developed from a cost identification perspective was available to measure costs of cancer resulting from wage losses or out-of-pocket expenses. Thus, over a 2-year period, we developed and tested new interviews to be administered by telephone to measure both out-of-pocket costs and wage losses as comprehensively and accurately as possible. We referred to a number of sources to develop interview prototypes, including published literature ( 8 , 9 , 24 , 29–32 ), our previous work, and cost measurement questionnaires used by other investigators ( 9 , 24 , 31 , 33 ). We pretested interview prototypes individually and then used focus groups to test them further to ensure that costs that were frequent, important, and burdensome were comprehensively covered and that questions used to assess them were easily understandable and worded in language appropriate for a wide range of potential respondents ( 11 , 34 ). We also conducted a small prospective feasibility study with 6 months of follow-up to develop and test all computer programming required to aggregate information from specific questions collected during the interviews into specific cost estimates, thus verifying that we had all elements essential for cost estimation ( 7 ).

The interview questions that resulted from these different steps were structured around key events in the cancer trajectory––notably diagnosis, breast and axillary surgery, radiotherapy and/or brachytherapy, chemotherapy, and hormone therapy––to tie questions about components of costs to the events in the patient's own diagnosis and treatment path and thus facilitate recall. (Additional information about the questionnaires and programming required to calculate costs from questionnnaire responses can be obtained by contacting the corresponding author.) We showed previously that breast cancer survivors who were 3 years past diagnosis accurately recall summary treatment and prognostic information ( 35 ). An aide-memoire we developed was also mailed to participants before interviews to refresh their memory about the main cost-generating events and situations. To avoid, as much as possible, reliance on mental calculation and to increase the validity of information collected, women were only rarely asked to directly report dollar amounts spent. Instead, they were questioned about facts and events. Thus, questions with respect to wage losses concentrated on dates of work absence, type of compensation received (this was elicited using specific prompts), and percentage of the usual wage covered by each compensation type. The combination of this information to estimate wage losses was performed by the research team, not the respondent ( 36 ).

The telephone interviews were conducted centrally by specially trained interviewers 1, 6, and 12 months after the start of definitive treatment. The 1-month interview collected information on personal characteristics that might influence wage losses including age, education, whether the patient was living with a spouse or partner, any health conditions other than breast cancer that limited daily activities, personal and family income, and social support. Social support was measured using questions (on the numbers of individuals available as confidants, who could provide practical assistance and who were close to the respondent and expressed affection for her) from the index of social support developed by Santé Québec for their 1992–1993 population-based survey of Quebecers ( 37 , 38 ). Using the predefined response categories, scores were calculated as the weighted sum of subjects’ answers, expressed as a percentage of the total possible score with higher scores indicating higher support. For women working 1 month before diagnosis, work conditions at that time were assessed, including whether the patient had been self-employed, job title, type of industry, length of tenure in the job held at diagnosis (measured by month and year of start of employment for that job), average number of paid hours per week , days worked per week, and, finally, daily, weekly, monthly, or annual before-tax wages (depending on the metric used by the women to report them).

In interviews conducted 6 and 12 months after the start of definitive treatment, we collected information on work absence(s) since diagnosis. For each full-time absence of 1 week or more, information was collected on reason for absence, start date, and, if the patient was back at work when interviewed, end date.

For women who did not experience full-time absence lasting 1 week or more, similar information was collected for periods of reduced working hours if the period lasted at least 4 weeks. Reduced hours of work were estimated only for women who reported no absence lasting 1 week or more. Self-employed women were also questioned about the need to hire and pay for replacement labor.

Women were also questioned about types of compensation received. In Canada, workers may have access to different types of compensation in case of illness. In 2003, nearly 90% of Canadian workers were eligible in principle for compensation through the federal government salary insurance program, which can cover up to 55% of usual salary or equal a maximum payment (if salary is higher than the maximum eligible salary established by government regulations) for a total of 15 weeks ( 39 ). Only workers who are employees have access to this type of insurance, and eligibility is based mainly on the previous number of hours worked. Unionized workers [about 32% of workers in Canada and 40% in Quebec ( 40 )] generally have access to salary insurance through their employer. This form of insurance usually covers longer periods of absence at more generous rates of compensation than that provided by the government salary insurance. When available, this type of insurance takes precedence over that available through the government program. Some employees also have annual paid sick leave and/or annual paid holidays as part of their fringe benefits that can be used as salary compensation during absence, as can any overtime that they may have accumulated. However, self-employed workers do not have access to these types of compensation, and some take out personal private salary insurance. To calculate wage losses, we collected information on the types of compensation received, duration, and amount received, expressed as either a proportion of the wages covered or the dollar amount received (according to how the patient recalled it).

Information about medical characteristics that might influence wage losses was collected from women's medical files. This information included prognostic factors (size, histological type, grade, and hormone receptor status of the tumor and lymph node status), surgeries performed (mastectomy, lumpectomy, sentinel node biopsy, axillary dissection, breast reconstruction), and adjuvant treatments received (radio-, chemo-, and hormone therapy).

Information on the women's perceptions of their financial situation in the past 12 months was collected with eight questions administered at both the 1- and 12-month interviews. These questions were adapted from published studies and questionnaires ( 41–45 ) and pertained specifically to the patient's satisfaction with her financial situation and worry about, and perception of, her own financial situation compared with that of others of the same age. Adequacy of family income to meet regular expenses (for food, housing, paying bills, and purchasing drugs) and to deal with unforeseen expenses was also assessed. We also asked patients the following question: “How costly has breast cancer been for you and your family?”—with four response options: not at all, a bit, quite, or very costly.

Statistical Analyses

Wage losses were estimated for every woman who was working during the month before diagnosis and who had either a full-time absence or a period of reduced hours of work because of breast cancer. Wage losses were calculated in several steps that combined information on weekly wages before diagnosis, duration of absence, and compensation received. Specifically, we first calculated the duration of each full-time absence lasting 1 week or more using absence start (or the date of treatment initiation if the absence started before treatment initiation) and end dates (the date of the 12-month interview was used as the end date if the woman had not yet returned to work). Second, we estimated the wages the woman would have earned for this period had she not been absent by multiplying the duration of absence by the average weekly before-tax wage the woman reported earning in the month before diagnosis. Third, we estimated wages actually received during the woman's absence by summing compensation received from all sources. Finally, to obtain the wage loss for the absence, we subtracted the actual total dollar amount received in compensation from the amount that would have been earned had the woman not been absent. A similar approach was used to estimate wage losses because of reduced hours of work. Wage losses were then used to calculate the main outcome for this analysis, namely, the proportion of usual annual wages lost. This proportion was calculated by dividing the wage loss during the 12 months following diagnosis by the projected annual wages the woman would have earned had she not been absent from work. This latter was calculated as the average before-tax weekly wage earned in the month before diagnosis, multiplied by 52.

We chose to estimate wage losses only for the portion of absences occurring after the date of treatment initiation to ensure that wage losses were not related to the period preceding diagnosis. When a woman reported retirement or definitively stopping work for some reason, the calculation of her wage losses started at treatment initiation but ceased at retirement or definitively stopping work. Wage losses were not estimated when any element required for their calculation (ie, weekly wages before diagnosis, duration of absence, compensation received) was missing.

We used descriptive statistics to describe the proportion of annual wages lost (mean, standard deviation, median, and interquartile range). Analysis of variance using the general linear model procedure was used to identify personal, medical, and employment characteristics associated with the proportion of annual wages lost from either a full-time absence or a period of reduced hours of work ( 46 ). A number of sociodemographic, disease and treatment, and employment characteristics were chosen a priori as potential determinants, including age, level of education, whether the patient lived with a partner, distance from the treatment center, level of social support, type of mastectomy (total, partial), type of breast cancer (invasive disease or ductal carcinoma in situ), chemotherapy (yes, no), radiotherapy and/or brachytherapy (yes, no), hormone therapy (yes, no), presence of health conditions other than breast cancer that limited daily activities, being self-employed (yes, no), working full time (yes, no, using 30 hours or more per week as the definition of full-time work) ( 47 ), and the number of years of experience in the job held when diagnosed. Age was categorized as less than 50 and 50 years or more to generally represent pre- and postmenopausal women and thus somewhat different life stages. Educational level was classified according to established levels in the Canadian education system because the diplomas obtained after completing these different levels may determine access to some jobs and associated employment conditions and fringe benefits. In Quebec, education up to the completion of high school represents 11–12 years of schooling; collegial level represents the 2–3 years after high school but not university-level education; those in the third group had university training beyond collegial education. Women who reported having one or more medical problems that limited daily activities were compared with those having none. To have categories with meaningful differences in the level of support and still have adequate numbers of subjects in the lowest support group, low social support was defined by scores in the lowest quintile of scores for the entire cohort of 800 women studied. Distance from the hospital where first treatment was received was dichotomized as less than 50 km and 50 km or more from the treatment center. This classification was used to capture the small but conceptually important group of women who had to travel substantially greater distance to obtain treatment. For years of experience in the job held at diagnosis, the categories compared were 0–4, 5–9, 10–19, and 20–40 years in that job; these categories were chosen to ensure sufficient numbers at each level.

We first assessed mean differences in the proportion of annual wages lost according to levels of each of these characteristics individually. Then, all characteristics were entered simultaneously into a single multivariable model and removed one by one, starting with the one with the highest P value from the two-sided F test, until only variables with a P value less than .10 remained. This exploratory approach and this level of statistical significance were chosen because the identification of determinants of cancer wage losses is still a new area of inquiry. Mean differences and P values did not differ substantially when the effect of each potential determinant was assessed while adjusting simultaneously for the effects of all other potential determinants. Thus, we present only the multivariable model that included the variables with P values less than .10.

To better understand the pathways that underlie the association between each determinant and proportion of annual wages lost, the relationships between these variables and each parameter used to produce the estimates of wage losses (weekly wages before diagnosis, total absence duration, and total average percentage compensated) were also assessed in multivariable models. Because weekly wages before diagnosis is a component of the total wage loss that would be difficult to modify in the short term, we present analyses only for absence duration and total percentage compensated. Linear trend across levels of characteristics having three or more levels was tested using a linear contrast for equally spaced categories ( 46 ).

Effects of annual wage losses on participants’ financial situations were assessed by calculating the percentages of women who experienced a change for the worse in financial situation during the year since diagnosis in terms of satisfaction with the financial situation, worry about and perception of her own financial situation compared with that of others of the same age, and adequacy of family income to meet regular expenses and to deal with unforeseen expenses. Percentages of patients who had experienced a change for the worse were compared according to the proportion of projected annual wages lost dichotomized in two categories, 33% or more (which corresponded roughly to the highest quartile) and less than 33% (corresponding to the other three quartiles). Relative risks (RRs) were computed using log-binomial regression models (the GENMOD procedure). All analyses were performed using SAS software (SAS Institute, Cary, NC).

All statistical tests were two-sided. Power was examined for the comparison of the mean proportion of annual wages lost in a group of 50 patients to that in a group of 353 patients (50 patients was among the smallest groups compared) using a two-sided t test at an alpha level of 0.05 and an SD of ±26%. Using these parameters, there was 80% power to detect a difference of 11 percentage points in the mean of the proportion of annual wages lost.

Results

Participants

During the study period, 1397 women with breast cancer were identified. Of the 962 patients who met eligibility criteria, 829 (86.2%) consented to participate and completed the 1-month interview and 800 of the 962 initially eligible women (83.2%) completed all three interviews. Reasons for not completing interviews were lack of time for or interest in the study (20 women), loss to follow-up ( 2 ), move from Canada ( 1 ), health problems ( 4 ), and death ( 2 ). Interviews planned for 1, 6, and 12 months after start of treatment were conducted on average 36 (SD = 17 days), 184 (SD = 15 days), and 365 (SD = 12 days) days after the start of treatment, respectively.

Among the 800 women who completed all three interviews, 459 (57.4%) had a paying job during the month before diagnosis, and the following analyses are restricted to the latter. Mean age at diagnosis was 50 years, and most (71.7%) were living with a partner ( Table 1 ). Virtually all were first treated with breast surgery (97.8%) that for most of those having surgery consisted of partial mastectomy (83.5%). Sixty-two percent of the 390 women who had a sentinel node biopsy or axillary dissection had disease without nodal involvement. A high proportion (85.4%) received two or more types of adjuvant treatment. At diagnosis, 15.9% were self-employed and most women (73.9%) were working full time. On average, women had been in the job held at diagnosis for 13 years and earned Can$703 weekly before taxes (SD = Can$538).

Table 1

Sociodemographic, medical, and employment characteristics of 459 women with newly diagnosed nonmetastatic breast cancer, who were working 1 month before diagnosis *

Characteristics No. (%) or value 
Sociodemographic characteristics  
    Age at start of definitive treatment, y  
        23–49 201 (43.8) 
        50–71 258 (56.2) 
        Mean ± SD 50.3 ± 7.2 
    Highest level of completed education †  
        Primary to high school 181 (39.4) 
        Collegial level 135 (29.4) 
        University 143 (31.2) 
    Lives with a partner 329 (71.7) 
    Family income (Canadian dollars) ‡  
        <$30000 70 (15.3) 
        $30000–$49999 124 (27.0) 
        $50000–$79999 116 (25.3) 
        ≥$80000 137 (29.8) 
        Unknown 12 (2.6) 
    Distance from hospital where first treatment was received, km  
        0.30–49.9 397 (86.5) 
        50.0–1045 km, No. (%) 62 (13.5) 
        Median 13.3 
    Social support score §  
        Low (26.7–66.6) 66 (14.4) 
        High (66.7–100.0) 393 (85.6) 
Medical characteristics  
    First treatment received  
        Breast surgery alone or with axillary dissection 449 (97.8) 
        Neoadjuvant chemotherapy 10 (2.2) 
    Type of mastectomy  
        Partial 375 (81.7) 
        Total 84 (18.3) 
    Had axillary sentinel node biopsy or axillary dissection 390 (85.0) 
    Presence of axillary invaded nodes  
        No 240 (52.3) 
        Yes 150 (32.7) 
        Unknown 69 (15.0) 
    Type of breast cancer  
        Ductal carcinoma in situ 57 (12.4) 
        Invasive disease 400 (87.1) 
        Phyllodes 2 (0.0) 
    Tumor size, cm  
        ≤2.0 295 (64.3) 
        >2.0 140 (30.5) 
        Unknown 24 (5.2) 
    Hormone receptors (estrogen or progesterone)  
        Positive 370 (80.6) 
        Negative 70 (15.3) 
        Unknown 19 (4.1) 
    Had chemotherapy 262 (57.1) 
    Had radiotherapy and/or brachytherapy 408 (88.9) 
    Received hormone therapy 348 (75.8) 
    No. of types of adjuvant treatments  
        0 8 (1.7) 
        1 59 (12.9) 
        2 217 (47.3) 
        3 175 (38.1) 
    No. of medical problems that limited daily activities  
        0 422 (91.9) 
        ≥1 37 (8.1) 
Employment characteristics  
    Self-employed ‖ 73 (15.9) 
    Skill level usually required for the job held at diagnosis ‖¶  
        Primary to high school 149 (32.5) 
        Collegial level 125 (27.2) 
        University or management occupations 184 (40.1) 
        Unknown 1 (0.0) 
    Experience in the job held in the month before diagnosis, y ‖  
        0–4 147 (32.0) 
        5–9 66 (14.4) 
        10–19 117 (25.5) 
        20–40 127 (27.7) 
        Unknown 2 (0.0) 
        Mean ± SD 13 ± 11 
    Worked full time (≥30 h/wk) ‖  
        Yes 339 (73.9) 
        No 115 (25.1) 
        Unknown 5 (1.1) 
    Hours worked per week ‖# , mean ± SD (range)  34 ± 14 (1–168) 
        Days worked per week ‖# , mean ± SD (range)  5 ± 1 (1–7) 
    Before-tax weekly wage (Canadian dollars) ‖**  
        Mean ± SD 702.63 ± 537.88 
        Median (range) 625.45 (20.00–6003.00) 
    Held a second job in the month before diagnosis 24 (5.2) 
Characteristics No. (%) or value 
Sociodemographic characteristics  
    Age at start of definitive treatment, y  
        23–49 201 (43.8) 
        50–71 258 (56.2) 
        Mean ± SD 50.3 ± 7.2 
    Highest level of completed education †  
        Primary to high school 181 (39.4) 
        Collegial level 135 (29.4) 
        University 143 (31.2) 
    Lives with a partner 329 (71.7) 
    Family income (Canadian dollars) ‡  
        <$30000 70 (15.3) 
        $30000–$49999 124 (27.0) 
        $50000–$79999 116 (25.3) 
        ≥$80000 137 (29.8) 
        Unknown 12 (2.6) 
    Distance from hospital where first treatment was received, km  
        0.30–49.9 397 (86.5) 
        50.0–1045 km, No. (%) 62 (13.5) 
        Median 13.3 
    Social support score §  
        Low (26.7–66.6) 66 (14.4) 
        High (66.7–100.0) 393 (85.6) 
Medical characteristics  
    First treatment received  
        Breast surgery alone or with axillary dissection 449 (97.8) 
        Neoadjuvant chemotherapy 10 (2.2) 
    Type of mastectomy  
        Partial 375 (81.7) 
        Total 84 (18.3) 
    Had axillary sentinel node biopsy or axillary dissection 390 (85.0) 
    Presence of axillary invaded nodes  
        No 240 (52.3) 
        Yes 150 (32.7) 
        Unknown 69 (15.0) 
    Type of breast cancer  
        Ductal carcinoma in situ 57 (12.4) 
        Invasive disease 400 (87.1) 
        Phyllodes 2 (0.0) 
    Tumor size, cm  
        ≤2.0 295 (64.3) 
        >2.0 140 (30.5) 
        Unknown 24 (5.2) 
    Hormone receptors (estrogen or progesterone)  
        Positive 370 (80.6) 
        Negative 70 (15.3) 
        Unknown 19 (4.1) 
    Had chemotherapy 262 (57.1) 
    Had radiotherapy and/or brachytherapy 408 (88.9) 
    Received hormone therapy 348 (75.8) 
    No. of types of adjuvant treatments  
        0 8 (1.7) 
        1 59 (12.9) 
        2 217 (47.3) 
        3 175 (38.1) 
    No. of medical problems that limited daily activities  
        0 422 (91.9) 
        ≥1 37 (8.1) 
Employment characteristics  
    Self-employed ‖ 73 (15.9) 
    Skill level usually required for the job held at diagnosis ‖¶  
        Primary to high school 149 (32.5) 
        Collegial level 125 (27.2) 
        University or management occupations 184 (40.1) 
        Unknown 1 (0.0) 
    Experience in the job held in the month before diagnosis, y ‖  
        0–4 147 (32.0) 
        5–9 66 (14.4) 
        10–19 117 (25.5) 
        20–40 127 (27.7) 
        Unknown 2 (0.0) 
        Mean ± SD 13 ± 11 
    Worked full time (≥30 h/wk) ‖  
        Yes 339 (73.9) 
        No 115 (25.1) 
        Unknown 5 (1.1) 
    Hours worked per week ‖# , mean ± SD (range)  34 ± 14 (1–168) 
        Days worked per week ‖# , mean ± SD (range)  5 ± 1 (1–7) 
    Before-tax weekly wage (Canadian dollars) ‖**  
        Mean ± SD 702.63 ± 537.88 
        Median (range) 625.45 (20.00–6003.00) 
    Held a second job in the month before diagnosis 24 (5.2) 
*

SD = standard deviation.

Completed high school represents 11–12 years of schooling and collegial level the 2–3 years after high school but before university-level education.

For women not living with a partner, personal income was used as family income.

§

Based on six items from the index of social support developed by Santé Québec for their 1992–1993 population-based survey. Low support indicates that the patient's score was in the lowest quintile of scores for the entire cohort of 800 women studied.

For the job at which the woman worked the most hours if more than one job.

Based on the National Occupational Classification used by Human Resources and Skills Development Canada, Government of Canada (2006).

#

Information not available for five women.

**

Information not available for 10 women.

Wage Losses

Six of the 459 women working at diagnosis stopped working definitively, either to retire (n = 2) or for other reasons (n = 4), without experiencing any absence because of breast cancer. Among the 453 remaining women, 34 (7.5%) were never absent from work for 1 week or more because of breast cancer, eight (1.8%) reported only a period of reduced hours of work but no absence of 1 week or more because of breast cancer, while 411 women (90.7%) were absent from work for 1 week or more because of breast cancer. Most women had a single full-time absence in the first year after diagnosis that lasted an average of 32.3 weeks (7.5 months) after treatment initiation ( Table 2 ). At the 12-month interview, 21.6% of women who had had a full-time absence were still not back at work (data not shown). Among the 316 women whose initial absence from work was over, just over half (167) reported having made a progressive return to work.

Table 2

Characteristics of absences of 1 week or more and reduction of work hours of 4 weeks or more and associated wage losses during the 12 months after first treatment for 457 women working at diagnosis *†

Characteristic No. Mean (SD) Median (IQR) 
Work absence of 1 week or more ‡     
    No. 411 1.2 (0.7) 1.0 (1.0–1.0) 
    Duration, wk 411 32.3 (16.3) 33.0 (18.6–48.9) 
Reduction of work hours of 4 wk or more §     
    Duration, wk 17.8 (17.1) 9.4 (5.2–31.9) 
    No. of hours reduced per week 32.6 (53.6) 11.5 (7.5–27.5) 
Total percentage of wages compensated considering only full-time absence 396 58.0 (32.3) 68.5 (35.2–79.8) 
Estimated wage losses for both absences and reduced hours of work, Canadian dollars  403 ‖ 9311 (19056) 5502 (2338–9905) 
Percentage of annual wages lost for absences and reduced hours of work  403 ‖ 27 (26) 19 (9–34) 
Characteristic No. Mean (SD) Median (IQR) 
Work absence of 1 week or more ‡     
    No. 411 1.2 (0.7) 1.0 (1.0–1.0) 
    Duration, wk 411 32.3 (16.3) 33.0 (18.6–48.9) 
Reduction of work hours of 4 wk or more §     
    Duration, wk 17.8 (17.1) 9.4 (5.2–31.9) 
    No. of hours reduced per week 32.6 (53.6) 11.5 (7.5–27.5) 
Total percentage of wages compensated considering only full-time absence 396 58.0 (32.3) 68.5 (35.2–79.8) 
Estimated wage losses for both absences and reduced hours of work, Canadian dollars  403 ‖ 9311 (19056) 5502 (2338–9905) 
Percentage of annual wages lost for absences and reduced hours of work  403 ‖ 27 (26) 19 (9–34) 
*

SD = standard deviation; IQR = interquartile range.

Information on work absences missing for two women who were absent from work 1 month before diagnosis and never returned to work during the study period.

If a woman held more than one job, work absences and wage losses were described for the main job, which is the job where the woman worked the most hours.

§

Reduced hours of work were estimated only for women having no work absence of 1 week or more.

Missing values for 16 women because of missing information either on before-tax weekly wage usually earned in the month before diagnosis or type or amount of compensation received.

Women reported using various types of compensation, sometimes in combination, including employer salary insurance (n = 238 women), annual paid sick leave (n = 153), government employment insurance (n = 94), annual paid holidays (n = 49), and, for the self-employed, private salary insurance (n = 12). Considering all these types of compensation, women on average were compensated for 58.0% of their projected usual wages during absence (calculated on the basis of their wages in the month before diagnosis). Women who reported use of the government salary insurance were compensated for 53% of their usual wages for 14.6 weeks. A small proportion, 14.8%, reported no compensation of any kind.

We calculated that the 403 women who had an absence or reduced hours of work had a mean loss of 27% of the annual before-tax wages that they would have earned had they not been absent from work (median = 19%) ( Table 2 ). We also calculated that 10.0%, 16.1%, and 74.0% of women lost 66% or more, between 33% and 65%, and less than 33%, respectively, of their normal annual earnings. Wage losses were highly variable and the median wage loss was Can$5502 (mean = Can$9311) in the first year after diagnosis ( Table 2 ). In addition to their own wage losses, we estimated that 18 self-employed women also paid an average of Can$5294 (SD = Can$5309) to a substitute worker during their own absence. This latter expense, representing 16% of their usual annual salary, was a cost for which the women received no direct compensation.

Characteristics Influencing Wage Losses

A higher proportion of annual wages lost was statistically significantly (at a threshold of P = .10) associated with a lower level of completed education ( P = .0018), living less than 50 km from the hospital where surgery was performed ( P = .070), lower level of social support ( P = .012), self-employment ( P < .001), part-time work ( P < .001), having shorter tenure in the job held at diagnosis ( P < .001), having invasive disease ( P = .086), and chemotherapy ( P < .001) ( Table 3 ). Together, these eight variables explained 34% of the variance in the proportion of annual wages lost.

Table 3

Proportion of annual wages lost for 403 women working at diagnosis and having a breast cancer–related absence lasting 1 week or more, or reduced hours of work for 4 weeks or more, during the first 12 months after first treatment, according to sociodemographic, medical, and employment characteristics from univariate and multivariable adjusted analyses *†

  Univariate Multivariable 
Characteristics No. Mean % of annual wages lost (95% CI) P‡ Mean % of annual wages lost (95% CI) P‡ 
Age at start of definitive treatment, y      
    23–49 181 29 (26 to 33) .055 – – 
    50–71 222 24 (21 to 28)  – – 
Highest level of completed education      
    Primary to high school 153 31 (27 to 35) <.001 30 (27 to 34) .0018 
    Collegial level 121 26 (22 to 31)  26 (22 to 30)  
    University 129 21 (17 to 25)  22 (19 to 26)  
Lives with a partner      
    Yes 290 27 (24 to 30) .61 – – 
    No 113 26 (21 to 30)  –  
Distance from hospital where first treatment was received, km      
    0.30–49.9 351 25 (23 to 28) .024 26 (24 to 28) .070 
    50.0–1045 52 34 (27 to 41)  32 (26 to 37)  
Social support score      
    Low (26.7–66.6) 62 33 (27 to 40) .022 33 (28 to 38) .012 
    High (66.7–100.0) 341 25 (23 to 28)  25 (23 to 28)  
Type of mastectomy      
    Partial 326 25 (22 to 28) .011 – – 
    Total 77 33 (28 to 39)  –  
Type of breast cancer      
    DCIS 46 10 (3 to 18) <.001 21 (14 to 28) .086 
    Invasive disease 357 29 (26 to 31)  27 (25 to 30)  
Chemotherapy      
    Yes 237 35 (32 to 38) <.001 34 (31 to 36) <.001 
    No 166 15 (11 to 18)  17 (13 to 20)  
Radiotherapy and/or brachytherapy      
    Yes 360 26 (24 to 29) .83 – – 
    No 43 27 (20 to 35)  –  
Hormone therapy      
    Yes 310 27 (24 to 29) 1.00 – – 
    No 93 27 (21 to 32)  –  
Medical problems that limit daily activities, No.      
    0 372 26 (24 to 29) .37 – – 
    ≥1 31 31 (22 to 40)  –  
Self-employed      
    Yes 61 43 (37 to 49) <.001 41 (36 to 47) <.001 
    No 342 24 (21 to 26)  24 (22 to 26)  
Experience in the job held in the month before diagnosis, y      
    0–4 128 35 (31 to 40) <.001 33 (30 to 37) <.001 
    5–9 60 26 (20 to 32)  27 (21 to 32)  
    10–19 98 25 (20 to 30)  24 (20 to 28)  
    20–40 116 18 (14 to 23)  21 (17 to 25)  
Worked full time (≥30 h/wk)      
    Yes 307 23 (21 to 26) <.001 24 (22 to 27) <.001 
    No 96 37 (32 to 42)  34 (29 to 38)  
  Univariate Multivariable 
Characteristics No. Mean % of annual wages lost (95% CI) P‡ Mean % of annual wages lost (95% CI) P‡ 
Age at start of definitive treatment, y      
    23–49 181 29 (26 to 33) .055 – – 
    50–71 222 24 (21 to 28)  – – 
Highest level of completed education      
    Primary to high school 153 31 (27 to 35) <.001 30 (27 to 34) .0018 
    Collegial level 121 26 (22 to 31)  26 (22 to 30)  
    University 129 21 (17 to 25)  22 (19 to 26)  
Lives with a partner      
    Yes 290 27 (24 to 30) .61 – – 
    No 113 26 (21 to 30)  –  
Distance from hospital where first treatment was received, km      
    0.30–49.9 351 25 (23 to 28) .024 26 (24 to 28) .070 
    50.0–1045 52 34 (27 to 41)  32 (26 to 37)  
Social support score      
    Low (26.7–66.6) 62 33 (27 to 40) .022 33 (28 to 38) .012 
    High (66.7–100.0) 341 25 (23 to 28)  25 (23 to 28)  
Type of mastectomy      
    Partial 326 25 (22 to 28) .011 – – 
    Total 77 33 (28 to 39)  –  
Type of breast cancer      
    DCIS 46 10 (3 to 18) <.001 21 (14 to 28) .086 
    Invasive disease 357 29 (26 to 31)  27 (25 to 30)  
Chemotherapy      
    Yes 237 35 (32 to 38) <.001 34 (31 to 36) <.001 
    No 166 15 (11 to 18)  17 (13 to 20)  
Radiotherapy and/or brachytherapy      
    Yes 360 26 (24 to 29) .83 – – 
    No 43 27 (20 to 35)  –  
Hormone therapy      
    Yes 310 27 (24 to 29) 1.00 – – 
    No 93 27 (21 to 32)  –  
Medical problems that limit daily activities, No.      
    0 372 26 (24 to 29) .37 – – 
    ≥1 31 31 (22 to 40)  –  
Self-employed      
    Yes 61 43 (37 to 49) <.001 41 (36 to 47) <.001 
    No 342 24 (21 to 26)  24 (22 to 26)  
Experience in the job held in the month before diagnosis, y      
    0–4 128 35 (31 to 40) <.001 33 (30 to 37) <.001 
    5–9 60 26 (20 to 32)  27 (21 to 32)  
    10–19 98 25 (20 to 30)  24 (20 to 28)  
    20–40 116 18 (14 to 23)  21 (17 to 25)  
Worked full time (≥30 h/wk)      
    Yes 307 23 (21 to 26) <.001 24 (22 to 27) <.001 
    No 96 37 (32 to 42)  34 (29 to 38)  
*

CI = confidence interval; DCIS = ductal carcinoma in situ.

Information required to calculate the proportion of annual salary lost was available for 403 of 418 women.

F tests or, for ordinal variables, test for linear trend.

These same eight characteristics were associated with one or both of the two individual parameters contributing to wage losses presented in Table 2 , namely, absence duration and compensation received (Table 4). A longer work absence was associated with living further away from the hospital ( P = .0081), having invasive disease ( P = .091), receiving chemotherapy ( P < .001) and not being self-employed ( P = .0045). Getting a lower percentage of compensation for lost wages was statistically significantly associated with six of the eight determinants of the proportion of annual wages lost: lower educational level ( P < .001), lower social support ( P = .011), chemotherapy ( P = .0085), self-employment ( P < .001), shorter tenure of the job held at diagnosis ( P < .001), and working part-time ( P < .001).

Effects of Wage Losses on the Family's Financial Situation

Similar proportions of women said that breast cancer had been very costly or not at all costly (14.7% and 13.4%, respectively). Thus, the great majority of women found it costly to some degree. Women who reported that breast cancer was not at all costly lost 13.4% of their usual annual wages compared with 22.3%, 32.5%, and 38.4% among women who said that breast cancer was a bit, quite, or very costly, respectively (analyses adjusted for education, self-employment, job tenure, and full-time work; data not shown). Compared with women who had lost less than 33% of their projected annual wages, those who had lost 33% or more were statistically significantly more likely to report a change for the worse since diagnosis in each of eight indicators of their financial situation (adjusted RRs ranged from 1.9 to 5.9 for these indicators) ( Table 5 ). Although relatively small numbers of women reported a change for the worse in their financial ability to meet basic needs, those who had lost 33% or more of their usual income were at much higher risk of reporting a lack of funds for necessities such as food (RR = 3.8, 95% confidence interval [CI] = 1.9 to 7.5), lodging (RR = 5.9, 95% CI = 2.8 to 12.6), and paying bills (RR = 3.5, 95% CI = 2.1 to 5.8).

Table 4

Relationships between determinants of wage losses and total absence duration and average percentage compensated during absence for 396 women working at diagnosis and who were absent from work for 1 week or more because of breast cancer *

  Total absence duration, wk Percentage compensated during absence, % 
Characteristics No. Mean (95% CI) P† Mean (95% CI) P† 
Highest level of completed education      
    Primary to high school 150 32 (30 to 34) .85 53 (49 to 57) <.001 
    Collegial level 121 32 (30 to 34)  59 (55 to 63)  
    University 125 32 (30 to 35)  63 (59 to 68)  
Distance from hospital where first treatment was received, km      
    0.30–49.9 344 31 (30 to 33) .0081 59 (56 to 61) .25 
    50.0–1045.0 52 37 (33 to 40)  55 (48 to 61)  
Social support score      
    Low (26.7–66.6) 60 34 (31 to 38) .14 51 (45 to 57) .011 
    High (66.7–100.0) 336 32 (30 to 33)  59 (57 to 62)  
Type of breast cancer      
    DCIS 45 29 (25 to 33) .091 62 (54 to 70) .29 
    Invasive disease 351 33 (31 to 34)  58 (55 to 60)  
Chemotherapy      
    Yes 237 40 (38 to 41) <.001 55 (52 to 58) .0085 
    No 159 21 (19 to 23)  63 (58 to 67)  
Self-employed      
    Yes 58 28 (24 to 31) .0045 20 (14 to 27) <.001 
    No 338 33 (32 to 34)  64 (62 to 67)  
Experience in the job held in the month before diagnosis, y      
    0–4 126 31 (29 to 34) .19 48 (44 to 53) <.001 
    5–9 58 31 (28 to 34)  55 (48 to 61)  
    10–19 97 33 (30 to 35)  61 (56 to 65)  
    20–40 114 33 (31 to 35)  68 (64 to 73)  
Worked full time (≥30 h/wk)      
    Yes 301 32 (31 to 33) .76 62 (59 to 65) <.001 
    No 95 33 (30 to 35)  46 (41 to 51)  
  Total absence duration, wk Percentage compensated during absence, % 
Characteristics No. Mean (95% CI) P† Mean (95% CI) P† 
Highest level of completed education      
    Primary to high school 150 32 (30 to 34) .85 53 (49 to 57) <.001 
    Collegial level 121 32 (30 to 34)  59 (55 to 63)  
    University 125 32 (30 to 35)  63 (59 to 68)  
Distance from hospital where first treatment was received, km      
    0.30–49.9 344 31 (30 to 33) .0081 59 (56 to 61) .25 
    50.0–1045.0 52 37 (33 to 40)  55 (48 to 61)  
Social support score      
    Low (26.7–66.6) 60 34 (31 to 38) .14 51 (45 to 57) .011 
    High (66.7–100.0) 336 32 (30 to 33)  59 (57 to 62)  
Type of breast cancer      
    DCIS 45 29 (25 to 33) .091 62 (54 to 70) .29 
    Invasive disease 351 33 (31 to 34)  58 (55 to 60)  
Chemotherapy      
    Yes 237 40 (38 to 41) <.001 55 (52 to 58) .0085 
    No 159 21 (19 to 23)  63 (58 to 67)  
Self-employed      
    Yes 58 28 (24 to 31) .0045 20 (14 to 27) <.001 
    No 338 33 (32 to 34)  64 (62 to 67)  
Experience in the job held in the month before diagnosis, y      
    0–4 126 31 (29 to 34) .19 48 (44 to 53) <.001 
    5–9 58 31 (28 to 34)  55 (48 to 61)  
    10–19 97 33 (30 to 35)  61 (56 to 65)  
    20–40 114 33 (31 to 35)  68 (64 to 73)  
Worked full time (≥30 h/wk)      
    Yes 301 32 (31 to 33) .76 62 (59 to 65) <.001 
    No 95 33 (30 to 35)  46 (41 to 51)  
*

CI = confidence interval; DCIS = ductal carcinoma in situ.

F test and test for linear trend used for ordinal variables.

Table 5

RRs of reporting a change for the worse in financial situation over the year after diagnosis comparing women with a high proportion of annual wages lost to those with low proportion *

 No. (%) reporting change for the worse in financial indicator during the year after diagnosis   
Indicator of financial situation  ≥33% annual wages lost (N = 105) †  <33% annual wages lost (N = 297) †  RR ‡ (95% CI)   RR § (95% CI)  
Satisfaction with financial situation 44 (41.9) 53 (17.9) 2.3 (1.7 to 3.3) 2.3 (1.6 to 3.4) 
Worries about financial situation 44 (41.9) 70 (23.6) 1.8 (1.3 to 2.4) 2.0 (1.4 to 2.7) 
Perception of financial situation compared   with that of others of the same age 37 (35.2) 56 (19.0) 1.9 (1.3 to 2.6) 1.9 (1.3 to 2.8) 
Ability to meet daily needs for:     
    Food 20 (19.1) 14 (4.7) 4.0 (2.1 to 7.7) 3.8 (1.9 to 7.5) 
    Lodging 18 (17.7) 11 (3.8) 4.6 (2.3 to 9.4)  5.9 (2.8 to 12.6)  
    Paying bills 28 (26.7) 30 (10.1) 2.6 (1.7 to 4.2) 3.5 (2.1 to 5.8) 
    Purchasing drugs 12 (12.2) 16 (5.7) 2.2 (1.1 to 4.4) 2.5 (1.1 to 5.8) 
    Dealing with unforeseen expenses 24 (22.9) 40 (13.5) 1.7 (1.1 to 2.7) 2.0 (1.2 to 3.4) 
 No. (%) reporting change for the worse in financial indicator during the year after diagnosis   
Indicator of financial situation  ≥33% annual wages lost (N = 105) †  <33% annual wages lost (N = 297) †  RR ‡ (95% CI)   RR § (95% CI)  
Satisfaction with financial situation 44 (41.9) 53 (17.9) 2.3 (1.7 to 3.3) 2.3 (1.6 to 3.4) 
Worries about financial situation 44 (41.9) 70 (23.6) 1.8 (1.3 to 2.4) 2.0 (1.4 to 2.7) 
Perception of financial situation compared   with that of others of the same age 37 (35.2) 56 (19.0) 1.9 (1.3 to 2.6) 1.9 (1.3 to 2.8) 
Ability to meet daily needs for:     
    Food 20 (19.1) 14 (4.7) 4.0 (2.1 to 7.7) 3.8 (1.9 to 7.5) 
    Lodging 18 (17.7) 11 (3.8) 4.6 (2.3 to 9.4)  5.9 (2.8 to 12.6)  
    Paying bills 28 (26.7) 30 (10.1) 2.6 (1.7 to 4.2) 3.5 (2.1 to 5.8) 
    Purchasing drugs 12 (12.2) 16 (5.7) 2.2 (1.1 to 4.4) 2.5 (1.1 to 5.8) 
    Dealing with unforeseen expenses 24 (22.9) 40 (13.5) 1.7 (1.1 to 2.7) 2.0 (1.2 to 3.4) 
*

RR = relative risk; CI = confidence interval.

All information on financial situation was missing for one woman among those who lost <33% of annual wages. Information on financial indicators was missing or not applicable for 0–7 women and 0–15 women among those who lost 33% or more and <33% of annual wages, respectively. Thus, percentages were not always calculated using 105 or 297, respectively.

Crude estimates.

§

Adjusted for prediagnosis variables associated with proportion of annual salary lost: education (≤high school, >high school), self-employment (yes, no), tenure in the job held at diagnosis (<10 years, ≥10 years), full-time work (yes, no).

Adjusted only for self-employment (yes, no), tenure in the job held at diagnosis (<10 years, ≥10 years), full-time work (yes, no).

Discussion

In this study of women who had been diagnosed with nonmetastatic breast cancer, the vast majority of those who were working at the time of diagnosis had a single full-time absence from work that was due to breast cancer that lasted on average 7 months and that resulted in a mean loss of 27% of their projected usual annual salary (median = 19%) after taking into account financial compensation received. As the proportion of annual wages lost increased, so did the proportion of women who experienced a worsening of their satisfaction with, perception of, and worries about their financial situation as well as of the adequacy of family income to meet regular expenses. We identified subgroups of women who were more likely to lose higher proportions of their annual wage; these included women with lower levels of education, further distance to travel to the hospital, a lower levels of social support, and who were self-employed, working part-time, or had had shorter tenure (before the time of diagnosis) in the job they held when breast cancer was diagnosed. Those who had invasive disease or received chemotherapy also lost a higher proportion of their annual wage. Overall, these findings point to wage losses from breast cancer in Canada as an important adverse consequence of this disease.

There are few other studies with which to compare ours. Among the few previous studies of wage losses because of breast cancer ( 4–6 , 24–27 ), only one was prospective and measured wage losses for the entire treatment and recovery period (losses were assessed among a cohort of Australian women during the first 18 months after diagnosis) ( 5 ). However, only absolute wage losses were presented, and the lack of information on absence duration and losses related to usual annual income makes these data difficult to interpret in terms of burden experienced by the woman. A US study that included different groups of women also indicated that salary losses could be considerable as they represented 19% of the monthly household income for the group of women diagnosed less than 6 months earlier and 28% for another group diagnosed 6–12 months earlier ( 6 ).

Our study has several strengths. First, its findings are based on women from consecutive series of patients diagnosed and treated in different hospitals and geographical regions. Participation among eligible women and retention in the study over the year of follow-up were high, 86% and 97% respectively. Second, data were collected with telephone interviews specifically designed to assess costs to breast cancer patients, interviews that we extensively pretested and refined using a variety of methods ( 7 , 11 , 34 ). Interview structure and questions were designed to facilitate recall and increase validity, as was the use of an aide-memoire. Data that we obtained on key parameters of wage losses correspond very closely to what we expected on the basis of prior knowledge. For example, women were absent for 7.5 months in the first year after diagnosis, similar to the 6.7 months reported by women in our population-based study of work experience after breast cancer ( 15 ). Consistent with our observations from this study, clinicians and patients in Quebec report that women are generally absent for a single period that lasts from treatment initiation to completion ( 11 ). Also, Canadian government salary insurance usually covers up to 55% of usual wages for women who qualify for a period of 15 weeks. Our corresponding estimates for the same insurance using the information provided by women were very similar—53% of usual wages compensated for 14.6 weeks.

This study also had some potential limitations. First, our assumption when estimating wage losses––that wages in the month before diagnosis were typical and would have been stable during the year of follow-up had the woman not been off work––may not always have been adequate for certain categories of workers, such as the self-employed or seasonal workers. That this assumption applies to the great majority of salaried women is supported by findings from our previous population-based study of work experience after breast cancer, in which work conditions were found to be stable over a 3-year period ( 16 ). However, this study also suggests that we would have needed different and longer interviews to adequately assess the effects of breast cancer on the income of self-employed or seasonal workers. Second, wage losses may have been underestimated for women who had a full-time absence and reduced their work schedule before or after this absence. This was because losses resulting from a temporarily reduced work schedule were assessed only among women who did not have a full-time absence of 1 week or more, a decision necessitated by the need to reduce interview length. However, 90% of women had a relatively long full-time absence from work, and this absence probably represents the major source of their wage losses. Third, we did not consider potential wage losses due to failure to obtain job promotion or to receive cost-of-inflation pay increases. These sources of lost income were beyond the scope of this study, and to have evaluated them adequately would have required the design of additional sequences of questions and thus longer interviews.

Our study makes several original contributions to research on breast cancer costs. First, this is the only study to date to assess the proportion of annual salary lost for a large number of working women from consecutive series of patients studied right from start of treatment and over a full year. This proportion may be a better indicator of financial burden than absolute wage losses, for two reasons. First, it relates losses to the level of financial resources available to the woman, namely, her annual wages, and it relates them to a standard period of time, independent of the length of work absence or reduction. Had we focussed only on absolute wage losses, we might have underestimated the economic burden borne by some women ( 48 ). For example, although women with a higher level of education lost more income, the proportion of their annual salary lost was lower because their absolute salary level was much higher and they were better compensated. Second, our study provides information on types of compensation received, and we have factored compensation into the evaluation of the real extent of wage losses. Third, by examining determinants of wage losses and the two potentially modifiable parameters that contribute to them––absence duration and compensation––we obtained information on possible pathways leading to important wage losses. This contribution to understanding of pathways leading to wage losses could guide the identification of targets for interventions aimed at reducing the economic consequences of breast cancer ( 49 ).

Indeed, the pathways to financial hardship after breast cancer are multifactorial and the factors involved may interrelate in complex ways to produce wage losses. Two examples from our study illustrate this complexity. First, receipt of chemotherapy was strongly associated with a higher proportion of annual salary lost because of its associations with both longer absence from work and a reduced level of compensation. Chemotherapy is an aggressive treatment that extends over several months with a number of side effects that may contribute to longer absence from work. It is often followed by radiotherapy. If the resulting absence is prolonged, compensation may not be adequate, resulting in a lower percentage compensated on average for these women. As a second example, lower levels of compensation were associated with several characteristics of the job at diagnosis, namely, self-employment, lower educational level, shorter job tenure, and part-time status. Women in these categories may have more limited access to employer or federal government illness insurance coverage. For example, self-employed women took less time off work than other employed women but reported lower compensation levels resulting in much higher losses. Currently, the self-employed cannot buy into Canada's government illness insurance program, and private insurance is very expensive. Part-time workers and women with shorter job tenure may also fail to qualify for government or employer illness insurance because they may not meet certain requirements (eg, the number of hours worked).

These results concerning financial burden after breast cancer among Quebec women should be useful for increasing our understanding of the level of wage losses of other Canadians diagnosed with breast cancer or other serious illness that results in a relatively long but temporary absence from work. The information provided about the possible pathways that act to increase wage losses through their associations with absence duration and level of compensation may apply to patients treated both in and outside Canada. Nonetheless, even in contexts where the approach to breast cancer management is similar to that in Canada, national norms and practices related to absence from work while ill and availability of governmental and industrial wage compensation could differ considerably from country to country. For example, American women may have different levels of wage losses than those observed in this study because of generally shorter absence from work after breast cancer diagnosis ( 50 ) and lack of a national program for cash sickness benefits ( 51 ).

Our results have implications for health-care professionals and women facing this disease. First, these findings should sensitize clinicians to the real extent to which wage losses resulting from breast cancer can substantially and negatively affect the financial situation of working women and their families. Second, they suggest that the subject of work absence and its possible financial consequences should be explicitly discussed by the clinician with the patient early on in the treatment trajectory. Our results should reassure those groups of women identified as less likely to suffer from substantial salary losses, namely, those who get less burdensome treatments and/or who have more favorable working conditions. But perhaps more important, particularly for the groups of women that we have identified as being at higher risk of losing a substantial proportion of their wages, information about wage losses might help to inform women in their thinking and planning about how to handle work and work absence to lessen potential losses ( 9 , 52 , 53 ). Depending on a woman's work and financial situation and her own preferences, being more informed may prompt her to investigate all sources of compensation available to her, contribute to her making efficient use of her current financial and social resources, and help her negotiate work absence and/or work reorganization during treatments with her employer.

On a policy level, our results should add to the current reflection in Canada and elsewhere on reasonable changes to public policies concerning illness insurance that would help workers facing severe illness ( 10 , 54 ). Special attention should be given to self-employed and part-time workers because they have the lowest levels of financial compensation to offset wage losses. Attention should also be given to the duration of coverage. The combination of evidence-based treatments that offers the best chance for cure for many women with breast cancer ( 55 ) can result in absence from work that can considerably exceed the 15-week maximum currently allowed in Canada. Adaptation of public policies concerning illness insurance programs may help prevent high financial strain from becoming an additional consequence of severe illness for workers and their families.

Funding

Canadian Breast Cancer Research Alliance (#010318, #013324, #017317); Canadian Institutes of Health Research (Investigator award to E.M., PhD Fellowship Award to S.L.); Fondation de l’Université Laval (PhD Fellowship Award to S.L.).

References

1.
Will
B
Berthelot
J-M
Le Petit
C
Tomiak
E
Verma
S
Evans
W
Estimates of the lifetime costs of breast cancer treatment in Canada
Eur J Cancer
 , 
2000
, vol. 
36
 
6
(pg. 
724
-
735
)
2.
Aziz
NM
Cancer survivorship research: challenge and opportunity
J Nutr
 , 
2002
, vol. 
132
 
11 suppl
(pg. 
3494S
-
3503S
)
3.
Wingo
PA
Howe
HL
Thun
MJ
, et al.  . 
A national framework for cancer surveillance in the United States
Cancer Causes Control
 , 
2005
, vol. 
16
 
2
(pg. 
151
-
170
)
4.
Lidgren
M
Wilking
N
Jonsson
B
Rehnberg
C
Resource use and costs associated with different states of breast cancer
Int J Technol Assess Health Care
 , 
2007
, vol. 
23
 
2
(pg. 
223
-
231
)
5.
Gordon
L
Scuffham
P
Hayes
S
Newman
B
Exploring the economic impact of breast cancers during the 18 months following diagnosis
Psychooncology
 , 
2007
, vol. 
16
 
12
(pg. 
1130
-
1139
)
6.
Arozullah
AM
Calhoun
EA
Wolf
M
, et al.  . 
The financial burden of cancer: estimates from a study of insured women with breast cancer
J Support Oncol
 , 
2004
, vol. 
2
 
3
(pg. 
271
-
278
)
7.
Lauzier
S
Maunsell
E
Drolet
M
Coyle
D
Brisson
J
Hébert-Croteau
N
Costs to women during the first 6 months after a diagnosis of breast cancer
Psychooncology
 , 
2003
, vol. 
12
 
4 suppl
(pg. 
S161
-
S162
)
8.
Houts
PS
Harvey
HA
Simmonds
MA
, et al.  . 
Characteristics of patients at risk for financial burden because of cancer and its treatment
J Psychosoc Oncol
 , 
1985
, vol. 
3
 
2
(pg. 
15
-
22
)
9.
Houts
PS
Lipton
A
Harvey
HA
, et al.  . 
Nonmedical costs to patients and their families associated with outpatient chemotherapy
Cancer
 , 
1984
, vol. 
53
 
11
(pg. 
2388
-
2392
)
10.
Canadian Breast Cancer Network
Results of the National Survey on the Financial Cost of Having Breast Cancer
2004
Ottawa: ON
Canadian Breast Cancer Network
 
11.
Lauzier
S
Maunsell
E
De Koninck
M
Drolet
M
Hébert-Croteau
N
Robert
J
Conceptualization and sources of costs from breast cancer: findings from patient and caregiver focus groups
Psychooncology
 , 
2005
, vol. 
14
 
5
(pg. 
351
-
360
)
12.
Langlois
S
Baillargeon
J-P
Caldwell
G
Fréchet
G
Gauthier
M
Simard
J-P
La société québécoise en tendances 1960–1990
 , 
1991
Québec, Canada
Diffusion Prologue Inc, Institut québécois de la recherche sur la culture
13.
Chirikos
TN
Economic impact of the growing population of breast cancer survivors
Cancer Control
 , 
2001
, vol. 
8
 
2
(pg. 
177
-
183
)
14.
Chirikos
TN
Russell-Jacobs
A
Cantor
AB
Indirect economic effects of long-term breast cancer survival
Cancer Pract
 , 
2002
, vol. 
10
 
5
(pg. 
248
-
255
)
15.
Drolet
M
Maunsell
E
Mondor
M
, et al.  . 
Work absence after breast cancer diagnosis: a population-based study
CMAJ
 , 
2005
, vol. 
173
 
7
(pg. 
765
-
769
)
16.
Maunsell
E
Drolet
M
Brisson
J
Brisson
C
Masse
B
Deschenes
L
Work situation after breast cancer: results from a population-based study
J Natl Cancer Inst
 , 
2004
, vol. 
96
 
24
(pg. 
1813
-
1822
)
17.
Chirikos
TN
Russel-Jacobs
A
Jacobsen
PB
Functional impairment and the economic consequences of female breast cancer
Women Health
 , 
2002
, vol. 
36
 
1
(pg. 
1
-
20
)
18.
Bradley
CJ
Bednarek
HL
Neumark
D
Breast cancer survival, work, and earnings
J Health Econ
 , 
2002
, vol. 
21
 
5
(pg. 
757
-
779
)
19.
O'Brien
BJ
Rusthoven
J
Rocchi
A
, et al.  . 
Impact of chemotherapy-associated nausea and vomiting on patients’ functional status and on costs: survey of five Canadian centres
CMAJ
 , 
1993
, vol. 
149
 
3
(pg. 
296
-
302
)
20.
Moore
K
Out-of-pocket expenditures of outpatients receiving chemotherapy
Oncol Nurs Forum
 , 
1998
, vol. 
25
 
9
(pg. 
1615
-
1622
)
21.
Longo
CJ
Fitch
M
Deber
RB
Williams
AP
Financial and family burden associated with cancer treatment in Ontario, Canada
Support Care Cancer
 , 
2006
, vol. 
14
 
11
(pg. 
1077
-
1085
)
22.
Lloyd
K
Engebretson
A
Adamczyk
P
Tomori
C
Lyons
T
Bennett
C
Out-of-pocket costs and lost wages associated with colorectal, breast, and prostate cancer care
Proc Am Soc Clin Oncol
 , 
2001
, vol. 
20
  
Abstract 961
23.
Yabroff
KR
Davis
WW
Lamont
EB
, et al.  . 
Patient time costs associated with cancer care
J Natl Cancer Inst
 , 
2007
, vol. 
99
 
1
(pg. 
14
-
23
)
24.
Given
BA
Given
CW
Stommel
M
Family and out-of-pocket costs for women with breast cancer
Cancer Pract
 , 
1994
, vol. 
2
 
3
(pg. 
187
-
193
)
25.
Butler
L
Downe-Wamboldt
B
Melanson
P
, et al.  . 
Prevalence, correlates, and costs of patients with poor adjustment to mixed cancers
Cancer Nurs
 , 
2006
, vol. 
29
 
1
(pg. 
9
-
16
)
26.
Grunfeld
E
Gray
A
Mant
D
, et al.  . 
Follow-up of breast cancer in primary care vs specialist care: results of an economic evaluation
Br J Cancer
 , 
1999
, vol. 
79
 
7–8
(pg. 
1227
-
1233
)
27.
Maguire
P
Pentol
A
Allen
D
Tait
A
Brooke
M
Sellwood
R
Cost of counselling women who undergo mastectomy
BMJ
 , 
1982
, vol. 
284
 
6333
(pg. 
1933
-
1935
)
28.
Hillner
B
Economic and cost-effectiveness issues in breast cancer treatment
Semin Oncol
 , 
1996
, vol. 
23
 
1 suppl 2
(pg. 
98
-
104
)
29.
Lansky
SB
Cairns
NU
Clark
GM
Lowman
J
Miller
L
Trueworthy
R
Childhood cancer: nonmedical costs of the illness
Cancer
 , 
1979
, vol. 
43
 
1
(pg. 
403
-
408
)
30.
Birenbaum
LK
Clarke-Steffen
L
Terminal care costs in childhood cancer
Pediatr Nurs
 , 
1992
, vol. 
18
 
3
(pg. 
285
-
288
)
31.
Houts
PS
Lipton
A
Harvey
HA
Martin
BA
A method for collecting detailed data on direct and indirect medical expenses of cancer patients
Prog Clin Biol Res.
 , 
1983
, vol. 
120
 (pg. 
193
-
199
)
32.
Moore
KA
Breast cancer patients’ out-of-pocket expenses
Cancer Nurs
 , 
1999
, vol. 
22
 
5
(pg. 
389
-
396
)
33.
Agency for Healthcare Research and Quality (AHRQ)
Medical expenditure panel survey [questionnaire]
2001
Rockville, MD
Agency for Healthcare Research and Quality
 
http://www.meps.ahrq.gov/whatis.htm . Accessed January 17, 2008
34.
Lauzier
S
Maunsell
E
De Koninck
M
Drolet
M
Hébert-Croteau
N
Women's and caregivers’ costs from breast cancer: findings from a focus group study
Psychooncology
 , 
2003
, vol. 
12
 
4 suppl
pg. 
S163
 
35.
Maunsell
E
Drolet
M
Ouhoummane
N
Robert
J
Breast cancer survivors accurately reported key treatment and prognostic characteristics
J Clin Epidemiol
 , 
2005
, vol. 
58
 
4
(pg. 
364
-
369
)
36.
Drummond
M
Sculpher
M
Torrance
G
O'Brien
B
Stoddart
G
Drummond
M
Sculpher
M
Torrance
G
O'Brien
B
Stoddart
G
Economic evaluation using patient-level data
Methods for the Economic Evaluation of Health Care Programmes
 , 
2005
3rd ed
Oxford, UK
Oxford University Press
(pg. 
247
-
275
)
37.
Audet
N
Lemieux
M
Cardin
J-F
Enquête sociale et de santé 1998—Cahier technique et méthodologique: Définition et composition des indices
 , 
2001
, vol. 
Vol. 2
 
Montréal, QC
Institut de la Statistique du Québec
(pg. 
191
-
193
)
38.
Institut de la statistique du Québec
Enquête sociale et de santé 1998
 , 
2001
Québec, Canada
Les publications du Québec
39.
The Canada Employment Insurance Commission
Employment insurance 2005 monitoring and assessment report. Report No.: SP-102-04-06E
2006
Ottawa, ON
Human Resources and Development Canada
40.
Statistics Canada
Summary tables: Average hourly wages of employees by selected characteristics and profession, unadjusted data, by province (monthly) - data for September 2006
2006
Ottawa, ON
Statistics Canada
 
41.
Société Canadienne du Cancer
Rapport final sur les besoins des personnes atteintes de cancer au Canada
1992
Toronto, ON
42.
Keith
VM
Gender, financial strain, and psychological distress among older adults
Research on Aging
 , 
1993
, vol. 
15
 
2
(pg. 
123
-
147
)
43.
Krause
N
Jay
G
Liang
J
Financial strain and psychological well-being among the American and Japanese elderly
Psychol Aging
 , 
1991
, vol. 
6
 
2
(pg. 
170
-
181
)
44.
Schulz
R
Williamson
GM
Knapp
JE
Bookwala
J
Lave
J
Fello
M
The psychological, social, and economic impact of illness among patients with recurrent cancer
J Psychosoc Oncol
 , 
1995
, vol. 
13
 
3
(pg. 
21
-
45
)
45.
Bellerose
C
Lavallée
C
Chénard
L
Levasseur
M
Santé Québec: Et la santé, ça va en 1992–1993? Rapport de l’Enquête sociale et de santé 1992–1993
 , 
1995
, vol. 
Vol. 1.
 
Montréal, QC
Ministère de la santé et des services sociaux
46.
Kleinbaum
D
Kupper
L
Muller
K
Nizam
A
Applied regression analysis and other multivariable methods
1998
3rd ed
Pacific Grove, CA
Duxbury Press
47.
Statistics Canada
Guide to the Labour force Survey
 , 
2007
 
48.
Hornbrook
MC
Lipscomb
J
Gotay
CC
Snyder
C
On the definition and measurement of the economic burden of cancer
Outcomes Assessment in Cancer: Measures, Methods, and Applications
 , 
2005
Cambridge, UK
Cambridge University Press
(pg. 
480
-
502
)
49.
Emanuel
EJ
Fairclough
DL
Slutsman
J
Emanuel
LL
Understanding economic and other burdens of terminal illness: the experience of patients and their caregivers
Ann Intern Med
 , 
2000
, vol. 
132
 
6
(pg. 
451
-
459
)
50.
Bradley
C
Oberst
K
Schenk
M
Absenteeism from work: the experience of employed breast and prostate cancer patients in the months following diagnosis
Psychooncology
 , 
2006
, vol. 
15
 
8
(pg. 
739
-
747
)
51.
United States Social Security Administration
Social Security Programs Throughout the World: the Americas, 2005
 , 
2006
Geneva, Switzerland
International Social Security Association
 
52.
Maunsell
E
Coyle
D
Brisson
J
, et al.  . 
Family costs of breast cancer: extent, determinants, and relation with quality of life
 
Canadian Breast Cancer Research Alliance (grants 010318, 013324, 017317) and Quebec Ministry of Health; $991,114; 1999–2008
53.
Bradley
CJ
The need for online information on the economic consequences of cancer diagnosis, treatment, and survivorship
J Med Internet Res.
 , 
2005
, vol. 
7
 
3
pg. 
e29
 
54.
Prince
MJ
Filling a gap in Canadians’ economic and social security: the case for a medium-term sickness/disability income benefit. Paper prepared for Human Resources and Social Development Canada
2006
55.
Steering Committee on Clinical Practice Guidelines for the Care and Treatment of Breast Cancer
Clinical practice guidelines for the care and treatment of breast cancer
CMAJ
 , 
1998
, vol. 
158
 
3 suppl
(pg. 
S1
-
S82
)
E. Maunsell, as the principal investigator, had overall responsibility for all aspects of the study, including its conception and design, the acquisition, analysis and interpretation of data, and drafting of the manuscript. S. Lauzier also contributed to the design, acquisition, analysis, and interpretation of data and drafted the manuscript. M. Drolet contributed to the design of questionnaire and to the acquisition of data. B. Abdous contributed to the statistical analysis. D. Coyle, N. Hébert-Croteau, J. Brisson, B. Mâsse, A. Robidoux, and J. Robert contributed to the conception and design of the study and to the interpretation of data. All of the authors revised the manuscript critically for important intellectual content. We thank the women and caregivers who participated in the different phases of the Family Costs of Breast Cancer Study and the study coordinator, interviewers, data manager, statisticians, and the clinician collaborators responsible for recruitment: Michèle Brie (Centre hospitalier régional de Baie-Comeau, Baie-Comeau), Judith Gaudreault (Centre hospitalier régional du Grand-Portage, Rivière-du-Loup), Marie-Hélène Girouard (Centre hospitalier régional de Trois-Rivières, Trois-Rivières), Jean Robert (Hôpital du Saint-Sacrement, Québec), André Robidoux (Centre hospitalier de l’Université de Montréal, Montréal), Mathieu Roy (Centre hospitalier Pierre-Boucher, Longueuil), Renée Simon (Hôpital Charles-Lemoyne, Greenfield Park), and Éric Imbeau (Hôtel-Dieu de Lévis, Lévis).