1 University Hospital, Ghent; 2 HEDM, Health Economics and Disease Management, Meise; 3 Free University Brussels, Brussels, Belgium
Received 16 October 2002; revised 20 January 2003; accepted 11 March 2003
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Since the incidence of breast cancer is growing, prevention programs can be expected to have a large economic impact on the health care system. From a health economic point of view, one is interested in the costs saved by disease prevention.
Patients and methods:
To predict 10-year cumulative incidence-based costs of postmenopausal breast cancer, a state transitional model was developed based on published clinical data. The model simulates disease progression and includes nine health states of 1 year: node-negative and node-positive early cancer; local relapse; metastasis, each with its follow-up states; and death. The cost per state was obtained from a chart review in 118 patients with different disease states. Costs were calculated from the health insurance perspective and discounted at 3%.
Results:
The cumulative 10 year cost per patient was equal to €31 774 [95% confidence interval (CI) €30 53633 012] of which 30% was hospital costs, 28% systemic treatment, surgery and radiotherapy and 14% testing. Costs were at their highest following diagnosis and before death.
Conclusions:
This incidence-based approach identified the cost of postmenopausal breast cancer over time and may serve as a valid baseline for assessment of new interventions in prevention or early treatment.
Key words: breast cancer, cost of illness, postmenopausal, retrospective chart review
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Prevention programs can be expected to have a large economic impact on the health care system given the size of the population addressed. To optimize their effect, they can be applied in high-risk populations; for example, in women aged 5070 years.
From a health economic point of view, in addition to the efficacy in terms of the number of cases prevented, one is also interested in the costs saved by preventing the disease from developing. In this context, the objective of the current study is to calculate the long-term direct medical costs of postmenopausal breast cancer.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The model starts at the time of breast cancer diagnosis. Based on epidemiological data in the literature, 13% of breast cancer patients enter the model in the state treatment of metastatic breast cancer [5]. Of the remaining patients, 40% start in treatment of node-positive early breast cancer and 60% in treatment of node-negative early breast cancer [6]. After treatment, all patients with early breast cancer can move to follow-up, have locoregional relapse, develop metastatic disease or die.
If a patient experiences locoregional relapse, they move to treatment of locoregional relapse, and subsequently either to follow-up after locoregional relapse, metastatic disease or death. Once patients have developed metastatic disease, they enter the state treatment of metastatic disease. Thereafter, patients can move to follow-up after metastatic breast cancer or death. At any time during the 10 year period, each patient starting in the model is allocated to one of the health states. Probabilities of disease progression in the different cancer stages were based on literature data with currently used treatment patterns.
To each health state a cost was then attributed. These cost data were based on a multicenter retrospective chart review performed in 119 women with different disease stages at breast cancer diagnosis and at different points of disease progression. By attributing to each state the probability of a patient being in that state at each annual cycle and the average annual cost of being in that state, the average accumulated 10 year costs of postmenopausal breast cancer was calculated.
Clinical data input
Calculation method. Annual rates of disease progression (transition probabilities) were calculated from published risks [4]. All figures are reported in Table 1 and explained hereafter.
|
Since an additional reduction in recurrence rates from combination therapy as opposed to chemotherapy or endocrine therapy alone has not been quantified, combined treatment was attributed the lowest of both risks reported by the EBCTCG. The EBCTCG total recurrence rates include contralateral breast cancer and locoregional breast cancer, as well as metastatic relapse. Therefore, the local and contralateral relapse rates were subtracted in order to obtain specific rates of metastatic relapse (Table 1).
Probability of contralateral breast cancer. The EBCTCG reported an annual risk of 0.30% in women receiving endocrine treatment and estimated an annual risk of 0.43% with adjuvant chemotherapy [7, 8].
Probability of locoregional recurrence. The 10-year cumulative incidence of ipsilateral locoregional recurrence after endocrine treatment in patients with node negative early breast cancer is 4% [9]. This figure is based on a relatively small study (150 patients) but the risk in patients not receiving systemic therapy was comparable with the control risk in the very large study by Fisher et al. [10]. Although the risk of locoregional relapse after chemotherapy in node negative patients has been shown to depend on the type of chemotherapy [cyclophosphamide, methotrexate and fluorouracil (CMF) versus methotrexate and fluorouracil (MF)], this difference was less apparent in the age group >50 years. Therefore we used the average reported risk of 0.2% [11]. The annual risk for locoregional recurrence in node positive postmenopausal patients was equal for endocrine therapy and for chemotherapy (0.4%) [10, 12].
Probability of disease progression after locoregional relapse. A retrospective analysis of 756 patients with low risk (node negative) breast cancer observed a relationship between local and distant recurrence rates [13]. It was concluded that the incidence of distant recurrence was increased 4.4-fold in patients having shown a locoregional relapse. Other studies in broader patient groups confirm this relationship by a significantly shorter median time to metastasis after local relapse [13, 14].
Disease progression after relapse was also shown to depend on initial lymph node status, the risk being significantly higher in node-positive patients compared with node-negative patients [15]. The calculations of transition probabilities show an annual risk of progression to metastatic disease of 13.6% for initially node-negative patients and 27.6% for initially node-positive patients.
Mortality in primary early breast cancer and locoregional relapse. Patients who die from breast cancer die from metastatic disease and the few women recorded in clinical trials as having died from breast cancer or from an unknown cause without any record of recurrence (2% of breast cancer-related deaths) are assumed to have had recurrence just before they died [8]. Therefore, the overall probability of death within health states related to early breast cancer is considered equal to the probability of non-breast cancer-related death. Based on the rates of death as first event (prior to progression) in clinical trial populations with early breast cancer (0.6% per year according to EBCTCG 1998 based on data from 44 300 women-years), this mortality rate can be assumed to be equal to the natural mortality for the age and sex matched general population (average age in model = 64 years). This is 0.92% per year at the start of the model and 1.48% per year as from year 6 [16].
Remission, reactivation and mortality in metastatic breast cancer. In a randomized multinational study comparing cyclophosphamide, epirubicin and 5-fluorouracil (CEF) with CMF in 460 metastatic breast cancer patients, median overall survival was comparable in both groups: 20.1 and 18.2 months, respectively. However, CEF showed significantly better response rates with a median time to disease progression of 8.9 versus 6.3 months [17]. Comparable survival and recurrence rates have been reported in other metastatic breast cancer trials [1820].
The cost of supplementary treatment episodes in case of metastatic recurrence has been included in the follow-up state of metastatic breast cancer.
Cost data input
Retrospective chart review. A total of 118 patient charts from seven regionally spread Belgian centers were reviewed. In each center a random selection of patient files was made based on listings of ICD9 codes for breast cancer and metastatic breast cancer (174 or 198.9, respectively) in 1997. Patients were included based on the following criteria:
Charts were reviewed by two independent physicians for a duration of 1 year, or until the occurrence of relapse or further progression, because after progression the data were no longer representative of the health state in which the patient entered the study. The cost of treating disease progression was obtained from a separate patient sample included at the time of progression. All resources directly or indirectly related to breast cancer diagnosis were recorded during the study period. Allocation of resources to breast cancer was based on the physicians notes and other documentation in the patient charts. In case of doubt with regard to allocation the treating physician was consulted.
At the end of the 1-year study period, the planned follow-up schedule or, if available, the actual follow-up during the subsequent year was recorded. Thus, we measured the costs during the actual treatment period and the costs of follow-up. Ambulatory costs were measured only if mentioned in the hospital files. Therefore, it is likely that some costs at the general practitioners level went undetected. No attempt was made to estimate an average cost of breast cancer-related resource use at the general practitioner level.
One hundred and eighteen patients with an average age of 64 years were included. Nineteen per cent were aged <55 years, 39% between 55 and 65 years and 42% >65 years. Disease stages were distributed as follows: 25 patients with node-negative disease; 28 with node-positive disease; 12 patients with locoregional relapse; and 43 patients were included with metastatic breast cancer. Ten additional patients had end-stage metastatic disease.
Of 43 patients with metastatic disease, 16 patients had their first diagnosis of metastatic progression, 16 had a metastatic relapse after previous remission and 11 patients had metastatic lesions at the moment of breast cancer diagnosis.
Unit costs. Costs were calculated by multiplying for each procedure the number of procedures with the respective unit cost for 1998. The unit cost for hospital or day clinic was the average cost per day recorded in the patient invoices. The cost per procedure for interventions, diagnostic investigations and medical visits were derived from the official listings of the Belgian Health Insurance INAMI/RIZIV.
Medication costs were based either on patient invoices or on unit costs derived from the official listings by the Belgian Center for Pharmacotherapeutic Information.
Sensitivity analysis. Sensitivity analyses were conducted whereby input variables in the model were varied over a range of possible values in order to assess their impact on the final outcomes. All transition variables were relatively increased or decreased by 20%; all costs related to locoregional relapse and advanced disease were increased or decreased by 20% as well; the probability of having metastasis at diagnosis was varied from 0% to double that of the baseline value. This methodology of simultaneous variation of model variables by 20% to test robustness of outcomes was applied previously [21]. The extent of variation chosen (20%) is very likely to cover the confidence intervals surrounding transition probabilities which include variations around 10% [8, 17]. For example, Ackland et al. [17] reported a median survival rate in metastatic cancer patients of 20.1 months [95% confidence interval (CI) 1823 months], from which an annual mortality rate of 33.9% was calculated with a 95% CI from 30.4% to 37.0%, a variation of up to 10.5%.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
In the subgroup of patients with locoregional relapse the sample size for measuring resource use during follow-up was very small. Therefore, the costs were assumed to be equal to those incurred by follow-up of node-positive breast cancer.
In the chart review there was no significant difference in total follow-up costs between the second and first treatment episode of metastatic breast cancer; therefore, all metastatic patients were pooled. The results of calculating follow-up costs are shown in Tables 3 and 4. Table 5 shows the cost calculations for hospitalization, day clinic and physician visits as an illustration of methods. Details on cost calculations for other items of resource use can be provided upon request.
|
|
|
Ten year cost prediction as from diagnosis
The Markov model was run to estimate the costs of breast cancer from diagnosis over a period of 10 years, thereby discounting at an annual rate of 3%.
This resulted in a total cumulative cost estimate at 10 years of €31 774. Via Monte Carlo simulation for 1000 patients diagnosed with breast cancer we calculated a 95% CI of €30 536€33 012 [4].
In order to simplify the reporting of results, a regrouping of costs was performed, resulting in seven main cost categories: tests; radio/chemotherapy and surgery; endocrine treatment; hospital stay; visits; and day clinic and terminal care. The overall cost consists mainly of treatment costs (radio/chemotherapy and surgery) and costs associated with hospital stay (Figure 1). The evolution of these costs over time for an average patient shows a decreasing slope that can be explained by the increasing number of patients dying, and by discounting future costs (Figure 2).
|
|
The model predicts a 10-year overall mortality of 42.4%. The absolute divergence from the observed overall 10-year mortality from breast cancer in Europe (45% [22]) is only 2.6%, suggesting good model validity.
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We neither collected data at the general practitioner level nor looked at the patients own contributions. Therefore, the current analysis probably represents the lower limit of the actual average medical cost of managing postmenopausal breast cancer.
To our knowledge, very few cost-of-illness studies related to female breast cancer have been conducted in Europe. Most reports are related to the costs of acute treatment of specific cancer stages. For example, Bercez et al. [25] published data with regard to the treatment costs for metastatic recurrence and local breast cancer recurrence in France. The average cost of treatment, based on retrospective data analysis in 146 patients, was FF 175 168 (€26 704) for metastatic recurrence and FF 115 705 (€17 639) for local recurrence. In our sample, based on a 1-year observation, the respective costs were 40% and 30% lower. However, Bercez et al. [25] also included non-medical costs, such as patient transportation.
Wolstenholme et al. [26] used a modeling approach to assess the implications of treatment costs for different stages at diagnosis on breast cancer screening cost-effectiveness. Screening programs reduce the stage at diagnosis, and improve breast cancer survival by 24% by shifting to earlier stages at diagnosis. Similar to previous reports, the study illustrates that late stage cancer is more expensive than early breast cancer, improving the expected cost-effectiveness of screening [27]. However, the cost difference between cancer stages loses significance when lifetime costs are considered. This was also the case in our analysis, in which changing the probability of metastatic disease at the time of diagnosis had only a modest influence on the overall total 10-year costs.
Thus, the results are in accordance with those reported by Wolstenholme et al. [26]; namely, that reducing the stage at diagnosis by breast cancer screening is unlikely to result in large treatment cost economies. The high costs of palliative care for late stage breast cancer management could be counterbalanced by high costs for aggressive initial cancer treatment and for long-term follow-up.
The medical profession shows great interest in any treatment that may reduce the incidence of breast cancer, such as screening or preventive medication. In this respect, selective estrogen receptor modulators show some promise in prevention, but their cost-effectiveness still needs to be analyzed further [28].
The results from our analysis may be of interest for the estimation of the possible savings incurred by breast cancer prevention and early treatment, and to identify populations, for example based on their risk profile, in which a specific intervention is both effective and cost-effective.
![]() |
Acknowledgements |
---|
![]() |
Footnotes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Nationaal Kankerregister. Borstkanker 1995; 14.
3. Sonnenberg FA, Beck JR. Markov models in medical decision making: a practical guide. Med Decis Making 1993; 13: 322338.[ISI][Medline]
4. Briggs A, Sculpher M. An introduction to Markov modelling for economic evaluation. Pharmacoeconomics 1998; 13: 397409.[ISI][Medline]
5. Launois R, Reboul-Marty J, Henry B, Bonneterre J. A cost-utility analysis of second-line chemotherapy in metastatic breast cancer. Docetaxel versus paclitaxel versus vinorelbine. Pharmacoeconomics 1996; 10: 504521.[ISI][Medline]
6. Hansen N, Morrow M. Breast disease. Med Clin North Am 1998; 82: 203222.[ISI][Medline]
7. Early Breast Cancer Trialists Collaborative Group. Tamoxifen for early breast cancer: an overview of the randomised trials. Lancet 1998; 351: 14511467.[CrossRef][ISI][Medline]
8. Early Breast Cancer Trialists Collaborative Group. Polychemotherapy for early breast cancer: an overview of the randomised trials. Lancet 1998; 352: 930942.[CrossRef][ISI][Medline]
9. Dalberg K, Johansson H, Johansson U, Rutqvist LE. A randomized trial of long term adjuvant tamoxifen plus postoperative radiation therapy versus radiation therapy alone for patients with early stage breast carcinoma treated with breast-conserving surgery. Cancer 1998; 82: 22042211.[CrossRef][ISI][Medline]
10. Fisher B, Anderson S, Redmond CK et al. Re-analysis and results after 12 years of follow-up in a randomised clinical trial comparing total mastectomy with lumpectomy with or without irradiation in the treatment of breast cancer. N Engl J Med 1995; 333: 14561461.
11. Fisher B, Digman J, Mamounas EP et al. Sequential methotrexate and fluorouracil for the treatment of node-negative breast cancer patients with estrogen receptor-negative tumors: eight-year results from national sur-gical adjuvant breast and bowel project (NSABP) B-13 and first report of findings from NSABP B-19 comparing methotrexate and fluorouracil with conventional cyclophosphamide, methotrexate, and fluorouracil. J Clin Oncol 1996; 14: 19821992.[Abstract]
12. Wils JA, Bliss JM, Marty M et al. Epirubicin plus tamoxifen versus tamoxifen alone in node positive postmenopausal patients with breast cancer: a randomised trial of the international collaborative cancer group. J Clin Oncol 1999; 17: 19881998.
13. Cowen D, Jacquemier J, Houvenaeghel G et al. Local and distant recurrence after conservative management of very low risk breast cancer are dependent events: a 10-year follow-up. Int J Radiat Oncol Biol Phys 1998; 41: 801807.[CrossRef][ISI][Medline]
14. Veronesi U, Marubini E, Del Vecchio M et al. Local recurrences and distant metastases after conservative breast cancer treatments: partly independent events. J Natl Cancer Inst 1995; 87: 1927.[Abstract]
15. Jatoi I, Hilsenbeck SG, Clark GM, Osborne CK. Significance of axillary lymph node metastasis in primary breast cancer. J Clin Oncol 1999; 17: 23342340.
16. Tafforeau J, Van Oyen H. Indicatozen van mortaliteit in de Franse en Vlaamse Gemeenschap. Brussels, Belgium: Centrum vor Operationeel Onderzoek in de Volksgezondheid 2002.
17. Ackland SP, Anton A, Breitbach GP et al. Dose intensive epirubicin-based chemotherapy is superior to an intensive cyclophosphamide, methotrexate and fluouracil regimen in metastatic breast cancer: a randomised multinational study. J Clin Oncol 2001; 19: 943953.
18. Pacini P, Rinaldini M, Algeri R et al. FEC (5-fluouracil, epidoxorubicin and cyclophosphamide) versus EM (epidoxorubicin and mitomycin-C) with or without lonidamine as first line treatment for advanced breast cancer. A multicenter randomised study. Final results. Eur J Cancer 2000; 36: 966975.[CrossRef][ISI][Medline]
19. Rose C, Kamby C, Mouidsen HT et al. Combined endocrine treatment of elderly postmenopausal patients with metastatic breast cancer. A randomised trial of tamoxifen vs tamoxifen + aminoglutethimide and hydrocortisone and tamoxifen fluoxymesterone in women above 65 years of age. Breast Cancer Res Treat 2000; 61: 103110.[CrossRef][ISI][Medline]
20. Jassem J, Pienkowski T, Pluzanska A et al. Doxorubicin and paclitaxel versus fluorouracil, doxorubicin, and cyclophosphamide as first-line therapy for women with metastatic breast cancer: final results of a randomized phase III multicenter trial. Central and Eastern Europe and Israel Paclitaxel Breast Cancer Study Group. J Clin Oncol 2001; 19: 17071715.
21. Annemans L, Moeremans K, Mutimer D et al. Modeling costs and cost-effectiveness of different CMV management strategies in liver transplant recipients as a support for current and future decision-making. Value Health 2002; 5: 347358.[CrossRef][ISI][Medline]
22. Berrino F, Sant M, Verdecchia A et al. Survival of cancer patients in breast cancer: The Eurocare study 1995. IARC Scientific Publication No. 132. Lyon, France: International Agency for Research on Cancer 1995.
23. Forbes JF. The incidence of breast cancer: the global burden, public health considerations. Semin Oncol 1997; 24: S1-20S1-35.
24. Koopmanschap MA, Rutten FFH. A practical guide for calculating indirect costs of disease. Pharmacoeconomics 1996; 10: 460466.[ISI][Medline]
25. Bercez C, Lebrun T, Bonneterre ME et al. Advanced breast cancer: an evaluation of the cost of recurrence. Cancer 1999; 86: 585590.
26. Wolstenholme JL, Smith SJ, Whynes DK. The costs of treating breast cancer in the United Kingdom: implications for screening. Int J Tech Assess Health Care 1998; 14: 277289.[ISI]
27. Legoretta AP, Brooks RJ, Leibowitz AN, Solin LJ. Cost of breast cancer treatment. A 4-year longitudinal study. Arch Intern Med 1996; 156: 21972201.[Abstract]
28. Mitlak BH, Cohen FJ. Selective estrogen receptor modulators: a look ahead. Drugs 1999; 57: 653663.[ISI][Medline]