Health system costs for stage-specific breast cancer: a population-based approach

Original Article

Medical Economics

Health system costs for stage-specific breast cancer: a population-based approach

N. Mittmann , PhD , * ,, § , J.M. Porter , MSc || , J. Rangrej , MSc || , S.J. Seung , BSc * , N. Liu , MSc || , R. Saskin , MSc || , M.C. Cheung , MD SM # , N.B. Leighl , MD MSc ** , J.S. Hoch , PhD § †† , M. Trudeau , MD MA # , W.K. Evans , MD ‡‡ , K.N. Dainty , PhD †† , C. DeAngelis , BScPhm PharmD # , C.C. Earle , MD MSc || #

*Health Outcomes and PharmacoEconomics ( hope ) Research Centre, Sunnybrook Health Sciences Centre, Toronto, ON.
Department of Pharmacology, University of Toronto, Toronto, ON.
International Centre for Health Innovation, Richard Ivey School of Business, Western University, London, ON.
§ Applied Research in Cancer Control, Cancer Care Ontario, Toronto, ON.
|| Institute for Clinical Evaluative Sciences, Toronto, ON.
# Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON.
**Princess Margaret Cancer Centre, University Health Network, Toronto, ON.
†† Keenan Research Centre of the Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Toronto, ON.
‡‡ McMaster University, Hamilton, ON.




The objective of the present analysis was to determine the publicly funded health care costs associated with the care of breast cancer ( bc a) patients by disease stage.


Incident cases of female invasive bc a (2005–2009) were extracted from the Ontario Cancer Registry and linked to administrative datasets from the publicly funded system. The type and use of health care services were stratified by disease stage over the first 2 years after diagnosis. Mean costs and costs by type of clinical resource used in the care of bc a patients were compared with costs for a matched control group. The attributable cost for the 2-year time horizon was determined in 2008 Canadian dollars.


This cohort study involved 39,655 patients with bc a and 190,520 control subjects. The average age in those groups was 61.1 and 60.9 years respectively. Most bc a patients were classified as either stage i (34.4%) or stage ii (31.8%). Of the bc a cohort, 8% died within the first 2 years after diagnosis. The overall mean cost per bc a case from a public payer perspective in the first 2 years after diagnosis was $41,686. Over the 2-year time horizon, the mean cost increased by stage: i , $29,938; ii , $46,893; iii , $65,369; and iv , $66,627. The attributable cost of bc a was $31,732. Cost drivers were cancer clinic visits, physician billings, and hospitalizations.


Costs of care increased by stage of bc a. Cost drivers were cancer clinic visits, physician billings, and hospitalizations. These data will assist planning and decision-making for the use of limited health care resources.

KEYWORDS: Breast cancer , costs , population-based analysis , disease stage


All permanent residents of the province of Ontario (a population of 13.2 million) are covered by a publicly funded health care system. The system pays for hospitalizations, most physician services, and emergency department ( ed ) services, and for selected prescription medications for the subset of the population more than 65 years of age or receiving social assistance. The provincial government authority collects data about those services and the service providers. These population-level data provide researchers with a unique opportunity to analyze the types of health services delivered within the system.

Breast cancer ( bc a) is a leading cause of morbidity and mortality in Canadian women1 and has a significant financial impact. In 2014, approximately 24,400 women will have been diagnosed with breast cancer, representing 26% of all new cancer cases in women2. Because health care management for bc a occurs across acute care, institutional care, and community settings, significant care and cost is assumed by the public health care system. Identification of the costs and the key resource utilization drivers will assist health system administrators in making informed policy decisions. Unfortunately, very few publications have determined bc a lifetime costs in Canada; the reported range is $309–$454 million3,4.

Several recent studies have determined overall costs for several cancers5,6 and have examined utilization and costs of specific modalities of health care, such as home care in colorectal cancer7,8 and home care costs in bc a, which were determined and stratified by disease stage9.

The objective of the present analysis was to determine the costs incurred in a publicly funded health care system for the management of bc a, by disease stage, in the first 2 years after diagnosis.


Incident cases of female invasive bc a (ICD-9 174.x) diagnosed between January 1, 2005, and December 31, 2009, were extracted from the Ontario Cancer Registry. The bc a cases in the registry were linked by their encrypted health card number to a spectrum of administrative datasets held at the Institute for Clinical Evaluative Sciences, an independent notfor-profit organization whose core business is to conduct research that contributes to the effectiveness, quality, equity, and efficiency of health care and health services in Ontario. The Institute for Clinical Evaluative Sciences Registered Persons Database includes information on patient characteristics (age, sex, etc.). The Ontario Ministry of Health and Long-Term Care holds data on reimbursements for hospitalizations (inpatient, day surgery), ed visits, physician visits, home care services, long-term care services, and prescription drug claims. Cancer Care Ontario holds data in its activity level reporting ( alr ) system on cancer services provided in the province (chemotherapy, radiation) through regional cancer centres and most, but not all, of the facilities that administer chemotherapy to patients.

In the present analysis, all health system services that were provided to individuals who met the eligibility criteria and that were reimbursed by the health system were included. All patients were followed from index date to death or to March 31, 2010, whichever came first. A control group selected from a population of women never diagnosed with cancer—that is, women without a record in the provincial cancer registry—were matched by age, income, prior health system use, and region to the women diagnosed with bc a. Cases and potential controls had to match exactly on birth year, health region of residence, modified income variable, and resource utilization banda. Income quintile assignment was based on Statistics Canada’s Postal Code Conversion File, pccf + (version 5E). The income variable was modified to account for potential misclassifications of neighbourhood income quintile derived from postal codes in rural areas. In addition, the Adjusted Clinical Group softwareb was used to assign a resource utilization band score to patients and control subjects alike. Control subjects who had an invalid health card number or who died before the patient’s breast cancer diagnosis date were excluded. The ratio of control subjects to bc a patients was up to 5:1.

For patients, bc a stage was based on a central staging algorithm that incorporates both pathologic and clinical staging information10. Women in the case group who had an invalid health card number were excluded.

Follow-up periods in the study population were variable because of the varying index dates (2005– 2009). The analysis considered the period of the first 2 years after diagnosis because women newly diagnosed with bc a would be likely to have experienced sequential treatment with some combination of surgery, radiation, and chemotherapy during that time. Table i describes the public health system services evaluated in the analysis.

TABLE I   Source and definition of cost components


Demographic characteristics for the bc a and control cohorts were summarized. The overall cost of care for the entire bc a population and the cost of care for the matched cohort, the cost per bc a patient (by stage) and per control subject, and the cost differences between the groups were calculated. The cost for the bc a cohort alive at the end of 2 years was also determined. The cost of each health care resource by each bc a patient who used a provincially funded health care resource was calculated, as was the percentage of the health care resource used by disease stage. Finally, the attributable cost for bc a patients (after comparison with control subjects) was determined. All cost data are presented in descriptive form (means, medians, standard deviations, and quartiles 1 and 3) over the time horizon of 2 years post-diagnosis using 2008 Canadian dollars. All analyses were performed using SAS 9.2 (SAS Institute, Cary, NC, U.S.A.).


The study included 39,655 bc a patients and 190,520 control subjects. Table ii shows that the average age of bc a patients was 61.1 years, with most of the cohort being 65 years of age or younger. The bc a patients resided predominantly in urban settings. Among the bc a patients for whom staging information was available, most were diagnosed with stage i (34.4%) or ii (31.8%) disease. Although not shown in the table, 8% of the bc a group ( n = 3253) died within 2 years of diagnosis; 2% of the control subjects died during the same period. By stage, the proportion of bc a patients who died was 2% (stage i ), 5% (stage ii ), 8% (stage iii ), and 49% (stage iv ).

TABLE II   Demographic information for the study group


Table iii shows that, from a public payer perspective, the overall mean cost per bc a case in the first 2 years after diagnosis was $41,686 (based on 39,655 bc a cases). Mean cost of care for stage iii and iv patients was at least twice that for stage i patients. The overall mean cost declined slightly to $40,426 for women who remained living ( n = 36,402) during the entire 2-year time horizon.

TABLE III   Costs for breast cancer cohort and living breast cancer cohort


Table iv presents the mean and median costs for all bc a patients and for those who used a given health care resource. Some notable cost trends included an increase in mean cost with advancing disease stage for resources such as hospitalization, ed visits, medications, homecare, and Ontario Health Insurance Plan ( ohip ) physician billings. The mean cost of same-day surgery declined with advancing stage. The mean costs of cancer clinic visits and radiation therapist time increased from stage i to stage iii , but decreased for stage iv disease.

TABLE IV   Health care resource-specific costs (total cohort and those who used the resource)


In terms of resource utilization by bc a patients who used health care resources, 54 women had no physician visitsc. In terms of resource utilization, other results showed that 85.3% of patients had at least 1 cancer clinic visit, 74.5% received at least 1 publicly funded homecare service, 72.7% underwent same-day surgery, 64.0% had at least 1 visit with a radiation therapist, 60.2% had at least 1 hospitalization, 58.2% made at least 1 ed visit, and 42.8% received at least 1 chemotherapy treatment.

Figure 1 illustrates the dollar amounts of the individual health care resources used by the bc a cancer cohort, overall and at each disease stage. Although the greatest number of patients were stage i at diagnosis, stage ii incurred the largest overall cost ($590,996,657) of all disease stages, chiefly as a result of cancer clinic visits (25.3%), followed by ohip physician billings (17.4%) and hospitalizations (15.9%).



FIGURE 1 Cost of health care resources, total and by stage, for the breast cancer cohort ( n = 39,655). a Includes physician billing, family health network or family health organization capitation, nonphysician, and diagnostic and laboratory (physician component) costs. b Includes dialysis, rehabilitation, mental health hospitalization, and the Assistive Devices Program. ohip = Ontario Health Insurance Plan; odb = Ontario Drug Benefit.

Figure 2 shows the differences in mean costs between the bc a patient cohort and the control cohort, disaggregated by health care resource. The largest cost difference between patients and control subjects was that for cancer centre visits (+$10,510 for bc a cases); chemotherapies (+$6,563) and physician billings (+$5,013) were second- and third-most costly. Concomitant drug costs (Ontario Drug Benefit Formulary) were $1,257 higher in the bc a patients than in the control group. Long-term care was the sole health care resource whose costs were higher in the control group (−$584).



FIGURE 2 Net mean cost of health care resources (breast cancer cases – controls). a Includes physician billing, family health network or family health organization capitation, nonphysician, and diagnostic and laboratory (physician component) costs. b Includes dialysis, rehabilitation, mental health hospitalization, and the Assistive Devices Program. odb = Ontario Drug Benefit; ohip = Ontario Health Insurance Plan.


This Canadian analysis is the first to examine stage-based costs for a population-based cohort of women with a diagnosis of bc a in a publicly funded system. The results presented here represent one of the largest Canadian bc a cohorts with disease staging, and almost half the women in our cohort were less than 65 years of age. Using a conservative but comprehensive costing approach, the overall mean cost of managing women for 2 years after a bc a diagnosis was found to be $41,686. That cost translates into $1.7 billion for the first 2 years of care after diagnosis for the 39,655 bc a patients in our study cohort. In terms of attributable costs, the bc a patients used $31,732 more in public health system resources than did matched control subjects without any cancer.

The overall mean cost increased by disease stage because of higher resource utilization. Compared with women having stage i or ii bc a, those with advanced bc a had higher proportions of hospitalizations, cancer clinic visits, ed visits, and homecare. For example, the proportion of patients with at least 1 ed visit during our 2-year timeframe increased from 50% in stage i to almost 80% in stage iv . In contrast, almost 84% of stage i bc a patients underwent same-day surgery, a proportion that declined to 36% for stage iv bc a patients (likely because of the limited procedures available to patients with advanced disease). Different trends were observed for chemotherapies and radiation. In patients receiving chemotherapy, utilization increased with disease stage: It was highest for those diagnosed at stage iii (75%), declining to 48% at stage iv (again because of limited options for treating advanced disease). Because radiation therapist time was used as a surrogate for radiation therapy, 72%, 68%, and 85% utilization was found for stages i , ii , and iii respectively; utilization then dropped to 53% for stage iv , indicating that radiation is a key component of the treatment armamentarium for our bc a patient cohort.

Recent Canadian work5,6 using population-based cohorts has provided overall costs for a number of cancers, including bc a, but those analyses did not evaluate the costs of all health care resources by stage of disease or determine the attributable bc a cost compared with matched control subjects.

Previous publications of bc a costs3,4 used different methodologies for determining lifetime bc a costs. Our work led to a substantially higher cost, based on fewer women, and representing only the first 2 years after diagnosis. We hypothesize that the discrepancies are a result of different data sources, inclusion or exclusion of certain health care resources, inflation, and the availability of more (and more expensive) medications to manage bc a.

Lifetime costs for bc a have previously been modelled in a Canadian setting4. Using the Statistics Canada Population Health Model (a microsimulation model), Will and colleagues4 estimated the average undiscounted lifetime cost per women by stage (1995 Canadian dollars), finding an estimated lifetime medical cost per woman that was substantially lower than our stage-based 2-year cost. The main differences are the result of approach (treatment utilization algorithms being modelled rather than patient-level data being obtained from the health care system), of information on resources and costs becoming outdated, and again, of more (and more expensive) medications becoming available to manage bc a.

A review11 of bc a treatment costs from other countries also reported lifetime costs that were lower than our 2-year costs, generally because only specific types of resources (treatment12,13, treatment-related adverse effects14, surgery15) were used. In one study from the United Kingdom, Remák and Brazil16 used regional administrative databases and physician questionnaires to reported a lifetime cost of £12,500 (in 2000 currency) for the management of stage iv breast cancer.

Our work is subject to a number of limitations. The data sources were Ontario databases collected for administrative purposes; they might therefore not contain all variables of interest with respect to the medical management of bc a. Screening (for example, mammography, ultrasonography) was not considered in our analysis because only costs after diagnosis were examined. Stage information was missing or unknown for 17.9% of the bc a cohort, and we therefore did not consider those patients in our stage-based costing. The collection of stage data is still improving at the provincial level. Information on hormone receptor and her2 (human epidermal growth factor receptor 2) status was not available at a population level, and those data can influence the type of treatment offered and, consequently, the cost implications.

Costing using the alr database required several assumptions and probably resulted in an underestimation of the total systemic therapy costs ($260 million). For example, alr data largely represent the cancer centres in the province, and doses outside the 5%–95% range were excluded. Drugs in the alr that were administered non-intravenously were excluded from the analysis because of inconsistencies in how they were entered into the system (that is, some sites entered oral therapies into their computerized order entry systems; other sites did not). Drug wastage was not considered in our analysis because we applied costs only to the dose administered and not to the vials that would actually be used. Also, because not all facilities that administer systemic chemotherapy report through the alr database, we anticipate that the systemic chemotherapies administered in the province are underestimated.

We cannot estimate the use and cost of all oral medications (women under 65, living in community, not on social assistance), because private and out-of-pocket payments are not covered in the public insurer database. However, we did include all systemic medications and expensive medications for all women in the population and oral medications for women meeting provincial eligibility requirements.

The ohip physician billing category included physician billing, family health network or family health organization capitation services, nonphysician billing, and the physician component for diagnostic and laboratory tests. The cost of the technical component for diagnostic tests (for example, technician time) was not included in our analysis because hospitals are responsible for that component as part of their global budget; the professional component of diagnostic tests (for example, physician) was captured in the total ohip costs. The foregoing exclusion also applies to laboratory tests conducted at hospital institutions.

Another limitation is that radiation costing consisted only of radiation therapist time; it did not consider equipment, physicist time, and administrative costs. Future studies will consider other radiation-related resources for costing. Our study evaluated only direct medical costs and not indirect, lost productivity, or out-of-pocket costs that are not available in the administrative data. Other work has shown that indirect costs are substantial, accounting for well over 50% of the total cost of cancer1719. Lastly, it is evident that analyzing data within 2 years after the initial diagnosis might not accurately identify all costs and utilization of breast cancer management, because, for many patients, treatment and survival can extend beyond those 2 years.

Despite the described limitations, we have provided a comprehensive cost study, by stage of disease, based on administrative data for an entire population of women with a diagnosis of bc a in the first 2 years after a diagnosis. Our work provides critical utilization and cost data to governments, industry, private payers, and academia. In particular, our data will be useful for decision-makers in the health care system examining the burden of illness at different stages; health economists generating health technology assessments for first-, second-, and third-line interventions; modellers building decision analytic models and microsimulations for health economics; and policymakers investing in publicly funded resources across various disease severities.


Our study is the first to examine the impact of disease stage on the initial publically funded provincial health system resources and costs for a bc a cohort in Canada. We found that, in the 2 years after a bc a diagnosis, significant direct health care costs ($41,686 per patient) were spent by the publicly funded health system, of which $31,732 per patient are attributable to bc a-related care (cost differential compared with matched controls). The attributable cost is based on significant resource utilization associated with cancer clinics, hospitalizations, same-day surgery, and ed visits for all women with a bc a diagnosis. In our analysis, women with stage ii bc a account for one third of the overall bc a cost. Future analyses will examine various timeframes or phases throughout bc a management and disaggregate the stage-based resources even further.

The methods used here will help with further costing work for bc a and other disease sites. These data are critical to understanding stage-based resources and funding in bc a. When designing public policies for the treatment of bc a, it is important to consider the type and extent of publicly funded health care services utilized. Such data will inform the future planning of health care for women with bc a.


The investigators thank Nelson Chong (programmer at the Institute for Clinical Evaluative Sciences) and Rene Robitaille (programmer at Cancer Care Ontario) for their help in untangling and linking datasets. Thanks also go to Angie Giotis for her assistance in costing medication regimens. We are grateful to Thi Ho and Katrina Chan (grant coordinators) for their administrative insight. We also acknowledge Shazia Hassan, Peggy Kee, and Grace Bannon for costing and administrative assistance.


All authors declared no perceived conflict of interest regarding the development of the present document. In the interest of being completely transparent: NM declared educational programs, unrestricted funding, and consultancies through the Health Outcomes and PharmacoEconomics ( hope ) Research Centre, a group that consults to the pharmaceutical industry. SJS declared consultancies through the hope Research Centre, a group that consults to the pharmaceutical industry. WKE declared consultancy contracts with Roche and Boehringer–Ingelheim and work on advisory boards for Bristol–Myers Squibb and Astellas.


1. The Conference Board of Canada (cboc). Home > How Canada Performs > International Rankings > Health > Mortality Due to Cancer [Web page]. Ottawa, ON: cboc; 2012. [Available at:; cited February 26, 2012]

2. Canadian Cancer Society. Home > Cancer information > Cancer type > Breast > Statistics. Breast Cancer Statistics [Web page]. Toronto, ON: Canadian Cancer Society; 2014. [Available at:; cited June 18, 2014]

3. Patra J, Popova S, Rehm J, Bondy S, Flint R, Giesbrecht N. Economic Cost of Chronic Disease in Canada 1995–2003. Toronto, ON: Ontario Chronic Disease Prevention Alliance; 2007. [Available online at:; cited January 15, 2014]

4. Will BP, Bertheolot JM, Le Petit C, Tomiak EM, Verma S, Evans WK. Estimates of the lifetime costs of breast cancer treatment in Canada. Eur J Cancer 2000;36:724–35.
cross-ref  pubmed  

5. de Oliveria C, Bremner KE, Pataky R, et al. Understanding the costs of cancer before and after diagnosis for the 21 most common cancers in Ontario: a population-based descriptive study. CMAJ Open 2013;1:E1–8.

6. de Oliveria C, Bremner KE, Pataky R, et al. Trends in use and cost of initial cancer treatment in Ontario: a population-based descriptive study. CMAJ Open 2013;1:E151–8.

7. Mittmann N, Liu N, Porter JM, et al. End-of-life home care utilization and costs in patients with advanced colorectal cancer. J Community Support Oncol 2014;12:92–8.

8. Mittmann N, Liu N, Porter J, et al. Utilization and costs of home care for patients with colorectal cancer: a population-based study. CMAJ Open 2014;2:E11–17.
cross-ref  pubmed  pmc  

9. Mittmann N, Isogai PK, Saskin R, et al. Population-based home care services in breast cancer: utilization and costs. Curr Oncol 2012;19:e383–91.

10. Cancer Care Ontario (cco). Stage Data Resolution Interim Solution for User Access. Version 1.0. Toronto, ON: cco; 2011.

11. Campbell JD, Ramsey SD. The costs of treating breast cancer in the US—a synthesis of published evidence. Pharmacoeconomics 2009;27:199–209.

12. Breast cancer stage cost analysis in a managed care population. Based on a presentation by Kenneth L. McDonough, md, ms. Am J Manag Care 1999;5(suppl):S377–82.

13. Legorreta AP, Brooks RG, Leibowitz AN, Solin LJ. Cost of breast cancer treatment: a 4-year longitudinal study. Arch Intern Med 1996;156:2197–201.
cross-ref  pubmed  

14. Hassett MJ, O’Malley AJ, Pakes JR, Newhouse JP, Earle CC. Frequency and cost of chemotherapy-related serious adverse effects in a population sample of women with breast cancer. J Natl Cancer Inst 2006;98:1108–17.
cross-ref  pubmed  

15. Lamerato L, Havstad S, Gandhi S, Jones D, Nathanson D. Economic burden associated with breast cancer recurrence: findings from a retrospective analysis of health system data. Cancer 2006;106:1875–82.
cross-ref  pubmed  

16. Remák E, Brazil L. Cost of managing women presenting with stage iv breast cancer in United Kingdom. Br J Cancer 2004;91:77–83.

17. Lidgren M, Wilking N, Jõnsson B, Rehnberg C. Resource use and costs associated with different states of breast cancer. Int J Technol Assess Health Care 2007;23:223–31.
cross-ref  pubmed  

18. Broekx S, Den Hond E, Torfs R, et al. The costs of breast cancer prior to and following diagnosis. Eur J Health Econ 2011;12:311–17.

19. Gordon L, Scuffham P, Hayes S, Newman B. Exploring the economic impact of breast cancers during the 18 months following diagnosis. Pyschooncology 2007;16:1130–9.

Correspondence to: Nicole Mittmann, 2075 Bayview Avenue, Room E240, Toronto, Ontario M4N 3M5. E-mail:

(Return to Top)

aWe used the Johns Hopkins Adjusted Clinical Groups System ( to classify patients into health resource utilization bands. The system uses a multistep algorithm to assign International Classification of Diseases codes to 32 aggregated diagnosis groups, which are then combined with age, sex, duration and severity of disease, and number of diseases to categorize patients into 1 of 102 clinically similar disease groups (“adjusted clinical groups”) that describe patients in terms of the totality of their previous disease history. The system then groups patients who might not be clinically similar, but who are expected to place a similar burden on the health care system, into quintiles of predicted health resource utilization. The resource utilization bands are 0 (none), 1 (healthy users), 2 (low), 3 (moderate), 4 (high), and 5 (highest)8. ( Return to Text )

bThe Adjusted Clinical Groups software uses a methodology designed to measure the intensity of resource use over a defined period of time. For resource utilization band scoring, the service utilization look-back period was 2 years for patients and control subjects alike. ( Return to Text )

cWe suspect that this observation reflects a miscoding issue, because to reach a diagnosis of bc a, a physician should have been involved, and at least 1 physician visit should therefore have been found. ( Return to Text )

Current Oncology , VOLUME 21 , NUMBER 6 , December 2014

Copyright © 2019 Multimed Inc.
ISSN: 1198-0052 (Print) ISSN: 1718-7729 (Online)