Annals of Oncology Advance Access originally published online on November 9, 2006
Annals of Oncology 2007 18(2):293-298; doi:10.1093/annonc/mdl410
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2006 European Society for Medical Oncology
breast cancer |
Modeling for cost-effective-adjuvant aromatase inhibitor strategies for postmenopausal women with breast cancer
1 Department of Medicine, Dalhousie University at Queen Elizabeth II Health Sciences Centre
2 Cancer Care Nova Scotia, Halifax, Nova Scotia, Canada
* Correspondence to: Dr T. Younis, Department of Medicine, Dalhousie University at Queen Elizabeth II Health Sciences Centre, 454 Bethune Building, 1278 Tower Road, Halifax, NS B3H 2Y9 Canada. Tel: +1 902-473-6054; Fax: +1 902-473-6186; E-mail: tallal.younis{at}cdha.nshealth.ca
 |
Abstract |
|---|
 Top Abstract introductionmaterials and methodsresultsdiscussionAcknowledgementsReferences |
|---|
Background: To determine cost-effective (CE) strategies comparing adjuvant upfront aromatase inhibitor (AI) with sequential tamoxifen (TAM) AI in postmenopausal (PM) women with breast cancer (BC).
Design: A Markov model was constructed to calculate cumulative costs and quality-adjusted life year (QALY) gains for upfront AI and TAM-AI in a hypothetical cohort of 60-year-old PM women with BC. Costs, utilities and probabilities were derived from the literature. The hazard ratios (HRs) of AI strategies were applied to a baseline cancer recurrence risk (RR) to determine CE strategies at the $50,000/QALY gain threshold. A direct payer perspective is utilized, and costs and benefits were discounted at 3%.
Results: Two-way sensitivity analyses are presented to determine CE strategies across a wide range of HRs and in different clinical scenarios including varying RRs (low, average, high and very high). TAM-AI is the preferred CE strategy at low and average RR, while upfront AI is CE at very high RR. The CE strategy in patients with high RR was dependent on the scenario examined.
Conclusions: This model may help health care providers select CE-adjuvant AI strategies in PM women with BC, until further direct evidence is available from randomized clinical trials.
Key words: adjuvant therapy, aromatase inhibition, breast cancer, costs, utility
|
introduction |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionAcknowledgementsReferences |
|---|
Aromatase inhibitors (AIs) are currently included in most treatment guidelines for the adjuvant therapy of hormone receptor-positive breast cancer (BC) in postmenopausal (PM) women [1, 2]. Strategies incorporating AIs are associated with small but statistically significant improved efficacy outcomes, different toxicity profiles and higher upfront costs compared with a standard course of tamoxifen (TAM) [3–14]. Cost–utility (CU) analyses, however, have reported favorable costs per quality-adjusted life year (QALY) gains [i.e. cost-effective (CE) ratio] for AI strategies compared with 5 years of TAM [15, 16].
A number of studies have attempted to identify the preferred AI treatment strategy. Two models have examined efficacy outcomes, such as time to recurrence (TTR) and disease-free survival (DFS), but did not include a cost analysis [17, 18]. A third included costs but assumed similar improvements in recurrence risk (RR) across all AI strategies [19]. We have previously examined the costs per QALY gained for upfront anastrazole and sequential TAM–exemestane based on the outcomes of the Arimidex, Tamoxifen, Alone or in Combination (ATAC) and Intergroup Exemestane Study (IES) studies, respectively [20]. The currently reported AI trials, however, present variable estimates of relative efficacies for the upfront AI and sequential TAM-AI strategies [3–12, 21]. As well, AI strategies are often recommended to a heterogeneous group of patients with different characteristics, including variable baseline risks for disease recurrence. As such, a model examining the CU of upfront AI and sequential TAM-AI over a range of efficacy hazards, including those reported in all the relevant AI clinical trials, and in patients with variable baseline risks of disease recurrence may provide additional information in regard to CE AI strategies for PM women with hormone receptor-positive BC.
|
materials and methods |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionAcknowledgementsReferences |
|---|
A Markov model was constructed to calculate cumulative costs and QALY gains for upfront AI and sequential TAM-AI for 10- and 20-year horizons. Subsequently, two-way sensitivity analyses were generated to predict CE strategies across a wide range of efficacy hazard ratios (HRs).
the model
A Markov model [22] was constructed in Microsoft Excel® (Microsoft Corporation, Redmond, WA, USA) to calculate cumulative costs and QALY gains for a hypothetical cohort of 60-year-old PM women with hormone receptor-positive BC undergoing adjuvant hormonal therapy. Five health states were included in the baseline model (Figure 1): well on therapy, well off therapy, local relapse, distant relapse and dead (with or without relapse). The adverse events of hormonal therapies were not accounted for in the baseline analysis, as they were not reported consistently in the relevant AI studies, but were examined in a sensitivity analysis. All patients entered the model in perfect health (well on therapy) and could move to other health states on the basis of transition probabilities and assumptions outlined in Table 1 for the baseline model and the analysis incorporating the adverse events. Costs and utilities were assigned according to the time spent in each health state (Table 2). The generated cumulative costs and QALY gains were calculated in monthly cycles, for 10- and 20-year horizons. The model examined outcomes for upfront AI and sequential TAM-AI by applying the HR associated with the respective AI strategy to baseline event rates for TAM alone starting at the relevant time point (Figure 2).
|
|
|
|
A meta-analysis reported HR of recurrence = 0.55 [probabilities, costs and utilities
Event rates for cancer recurrence, death with or without recurrence and adverse events were derived from the literature and the relevant AI trials (Table 1). Event rates for cancer recurrence were derived to reflect estimated 10-year RR without adjuvant systemic therapy assuming a one-third relative risk reduction in cancer recurrence with TAM alone. Four hypothetical cohorts were examined with variable 10-year risks of cancer recurrence in the absence of adjuvant systemic therapy—(i) low risk: node-negative disease with an
25% projected disease RR at 10 years; (ii) average risk: reflecting a mixed cohort of node-negative (60%) and node-positive (40%) disease with an aggregate estimated risk of disease recurrence of 35%; (iii) high risk: reflecting an estimated 50% risk of recurrence for node-positive disease and (iv) very high risk: representing node-positive patients with a high nodal burden and an 75% risk of disease recurrence. Event rates for adverse events in the sensitivity analysis were derived from the ATAC and IES studies (see Appendix; online only) [20]. These included a higher risk of venous thromboembolism, vaginal bleeding and endometrial cancer, but lower risk of bone fractures for TAM-AI relative to upfront AI [20]. Utilities and costs were derived from the literature (Table 2). Quality weights (utilities) associated with each health state ranged from 1.0 (perfect health) to 0.0 (death). Costs of adjuvant hormonal therapy, BC follow-up and cancer recurrence management were considered [26, 27] and those due to adverse events were accounted for in the sensitivity analysis [28–32]. This analysis took a direct payer perspective, with drug acquisition costs based on average wholesale prices in Nova Scotia, Canada, in 2004. Costs were adjusted for inflation [33], and are reported in 2005 Canadian dollars. Costs and benefits were discounted at 3%.
determination of CE strategies
Outcomes in the model were primarily influenced by the HRs of cancer recurrence for AI strategies, relative to TAM alone, which determine cumulative benefits achieved. We calculated the costs and QALY gains for a wide range of HRs (0–1) for both upfront AI and sequential TAM-AI. Two-way sensitivity analyses graphs were generated to illustrate the CE strategy for every combination of HRs on the basis of a $50,000/QALY gained threshold. Subsequently, the actual HRs reported in the AI trials were plotted in the graphs to determine the CE strategies. An estimated HR of 0.73 for the upfront AI strategy compared with TAM alone (average 0.74 in ATAC and 0.72 in Breast International Group (BIG) 1-98) was utilized in the analysis. For the sequential TAM-AI approach, variable HRs have been reported in different AI trials and the HR for DFS, but not TTR, was reported in one study [7–9]. In this analysis, the HR of 0.60 for recurrence in the Arimidex-Nolvade (ARNO)/Austrian Breast Cancer Study Group (ABCSG) study was utilized (Figure 2). For Estrogen Receptor (ER)+/Progesterone Receptor (PR)+ versus ER+/PR– subsets, HRs of 0.84 and 0.65 versus 0.43 and 0.42 for upfront AI and TAM-AI were utilized, respectively [3–5, 10].
Other analyses conducted include the following: (i) hormone receptor expression profile (ER+/PR+ versus ER+/PR–), (ii) adverse events perspective, (iii) variable age (±10 years), (iv) longer horizon (20 versus 10 years), (v) higher CE threshold ($75 000 versus 50 000/QALY gains), (vi) shorter duration of carryover benefit for AI (3 versus 5 years), (vii) variable utility estimates (±10%) and (viii) variable downstream costs (±25%).
|
results |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionAcknowledgementsReferences |
|---|
Results from the two-way sensitivity analyses are presented in four graphs representing the different cohorts examined (Figure 3). In the baseline analysis, upfront AI is a CE option in the very high risk cohort, while sequential TAM-AI appears to be the CE strategy in the low-, average- and high-risk cohorts. Upfront AI is associated with more favorable costs per QALY gained in patients with higher compared with lower risks of cancer recurrence.
|
Results for other scenarios are presented in Table 3 compared with the baseline model. Sequential TAM-AI appears to be the CE strategy in most of the scenarios examined, especially for the low- and average-risk cohorts.
|
From the hormone receptor expression perspective, sequential TAM-AI was the preferred CE strategy in the ER+/PR+ subgroup across all baseline RRs while upfront AI appears to be a CE strategy in the ER+/PR– subgroup regardless of the baseline risk of cancer recurrence. For the latter subgroup, the higher relative efficacy of upfront AI in the ER+/PR– subgroup resulted in improved cumulative downstream benefits compensating for the higher upfront drug costs.
Incorporating adverse events result in the upfront AI strategy becoming a CE option in the high-risk cohort. Older age (70 years) favors sequential TAM-AI and younger age (50 years) favors upfront AI, but neither changed the preferred CE strategies predicted at the reported HRs. A shorter life expectancy may result in less cumulative benefit, thereby making the higher drug costs of the upfront AI strategy less CE. A longer horizon perspective (20 versus 10 years) also favors sequential TAM-AI, which becomes the CE strategy regardless of baseline risk of cancer recurrence. Similar to previously reported efficacy models [17, 18], this CE model projects a slightly lower 10-year absolute recurrence rate for TAM-AI compared with upfront AI at the HRs examined, translating into further incremental benefits at the 20-year horizon.
As expected, accepting a higher CE threshold ($75 000 versus 50 000/QALY gained) favors the upfront AI strategy. Assuming a shorter carryover benefit for AI strategies (3 versus 5 years) also favors the upfront AI strategy. In these two scenarios, upfront AI is a CE option in the very high and high-risk cohorts, while TAM-AI remains the CE strategy in the low- and average-risk cohorts (Table 3).
The model was robust to reasonable changes in key assumptions and input data including the range of utilities and costs examined. A later switch from TAM to AI in the sequential TAM-AI strategy (3 versus 2.5 years), however, favors upfront AI, which becomes the CE strategy in the high-risk cohort. An earlier switch (2 versus 2.5 years) does not influence the conclusions.
|
discussion |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionAcknowledgementsReferences |
|---|
Adjuvant AI therapies, such as upfront AI and sequential TAM-AI, are often recommended for PM women with BC [1, 2] but the optimal strategy remains undefined as more data are awaited from direct comparisons in randomized clinical trials. This model provides two-way sensitivity analyses to help predict CE strategies in relevant clinical scenarios including variable baseline risks of cancer recurrence. Sequential TAM-AI appears to be the CE strategy in low- and average-risk patients, while upfront AI appears to be a CE option in very high risk patients. The CE strategy in high-risk patients was dependent on the scenario examined.
The magnitude of benefit, in absolute terms, derived from adjuvant therapy depends on the predicted baseline risk of disease recurrence, which commonly drives decision making regarding choice of therapeutic modalities [34]. Given the constant relative benefits of a therapeutic strategy, the estimated absolute benefits derived would be greater for disease with the highest baseline risk of recurrence. In this analysis, an upfront AI strategy was associated with improved incremental benefits, and a favorable CU ratio, for patients with the highest estimated baseline risk of cancer recurrence. For the entire patient population examined, upfront AI was CE in patients with the highest risk of cancer recurrence, while sequential TAM-AI was the CE strategy in patients with lower RRs.
In most of the scenarios examined, sequential TAM-AI was the preferred CE strategy. Upfront AI, however, was a CE strategy in a number of scenarios. For patients with ER+/PR– BC, upfront AI appears to be a CE strategy relative to TAM-AI. This conclusion, however, should be interpreted with caution as the higher relative efficacy of AI treatment in the ER+/PR– subgroup was observed inconsistently in the AI clinical trials [3–6, 10]. For the high-risk cohort (estimated 50% risk of disease recurrence in 10 years), upfront AI was a CE option when incorporating adverse effects, assuming a shorter carryover benefit, or accepting a higher CE threshold. Adverse events are important considerations, but they were not reported consistently in all AI clinical trials [3–14]. Compared with TAM [23], the duration of carryover benefit for AI strategies beyond the hormonal treatment period remains uncertain due to the relatively short follow-up in AI clinical trials [3–14]. A higher CE threshold may be more acceptable if payers are willing to accept higher costs for QALY gains. As such, the CE strategy for patients with high-risk disease should be considered in the context of all these variables.
The upfront AI strategy was not CE when a 20-year horizon perspective was examined. A 20-year horizon allows greater opportunities to accrue cumulative survival benefits for the strategy with the most favorable absolute recurrence estimates at the 10-year horizon. The 10-year horizon was utilized in our baseline analysis to conform to the commonly employed 10-year estimates of RR when discussing adjuvant therapy for BC [34] as well as to allow for comparisons with other published models [17, 18].
Two previously published models [17, 18] examined the 10-year DFS or TTR outcomes for AI strategies on the basis of relative benefits reported in the AI trials but did not account for costs or adverse effects. Similar to our model, the cumulative benefits, but not absolute recurrence rates, were slightly better for upfront AI relative to TAM-AI at the 10-year horizon [17, 18]. The analysis we present here examines costs per cumulative benefit as opposed to cumulative or absolute benefit alone. Costs are not uncommonly discussed when considering a more expensive therapeutic strategy [35]. In the CE model by Lønning [19], a lower recurrence rate with upfront AI compared with sequential TAM-AI was required for upfront AI to be a CE strategy. Our analysis, along with other models [17, 18], does not predict a lower recurrence rate for upfront AI compared with TAM-AI at the 10-year horizon.
This study has limitations. As in all prediction models, this cost–benefit analysis relies on key assumptions to reduce the infinite number of possible clinical outcomes to allow a feasible analysis incorporating indirect comparisons across different clinical trials. The driving factors in this model were the relative efficacies of the AI strategies examined and the estimated baseline risks of cancer recurrence. The relative efficacy of each of the AI strategies, as reported in the AI clinical trials, was assumed to be constant across cohorts with variable risks of cancer recurrence. A large meta-analysis for AI therapy, similar to that conducted for adjuvant TAM [23], would, however, be required to confirm this assumption. As well, the estimates of relative efficacy for AI strategies continue to evolve and are often updated [3–14]. We, therefore, provide a two-way sensitivity analyses across a wide range of HRs for possible future analyses. Other assumptions, and point estimates for costs and utilities, were tested in the sensitivity analyses and the conclusions derived were robust across a reasonable range of uncertainties. Lastly, this analysis reflects a Canadian health care cost perspective, which may not necessarily apply to all other health care systems.
In summary, this model provides two-way sensitivity analyses to predict CE AI strategies across a wide range of efficacy HRs and variable baseline cancer RRs. Upfront AI appears to be a CE option in very high risk patients, while sequential TAM-AI is the CE strategy in low- and average-risk patients. This model may help health care providers select the CE-adjuvant AI strategies until further direct evidence, and longer follow-up, is available from randomized clinical trials.
|
Acknowledgements |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionAcknowledgementsReferences |
|---|
This study has been supported in part by a grant from the Capital Health Research Fund. This study was presented as a poster presentation at the San Antonio Breast Cancer Symposium (SABCS) 2005 annual meeting.
Received for publication July 30, 2006. Revision received September 26, 2006. Accepted for publication October 5, 2006.
|
References |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionAcknowledgementsReferences |
|---|
1. Carlson RW, Brown E, Burstein HJ, et al. (2006) National Comprehensive Cancer Network (NCCN) Task Force Report: adjuvant therapy for breast cancer. J Natl Compr Canc Netw 4:Suppl 1, S1–S26.
2. Winer EP, Hudis C, Burstein HJ, et al. (2005) American Society of Clinical Oncology technology assessment on the use of aromatase inhibitors as adjuvant therapy for postmenopausal women with hormone receptor-positive breast cancer: status report 2004. J Clin Oncol 23:3619–629.
3. Arimidex Tamoxifen Alone or in Combination Trialists' Group. (2002) Anastrozole alone or in combination with tamoxifen versus tamoxifen alone for adjuvant treatment of postmenopausal women with early breast cancer: first results of the ATAC randomized trial. Lancet 359: pp. 2131–2139.[CrossRef][Web of Science][Medline]
4. Trialists' ATAC Group. (2003) Anastrozole alone or in combination with tamoxifen versus tamoxifen alone for adjuvant treatment of postmenopausal women with early-stage breast cancer: results of the ATAC trial efficacy and safety update analyses. Cancer 98:1802–1810.[CrossRef][Web of Science][Medline]
5. Trialists' ATAC Group. (2005) Results of the ATAC trial after completion of 5 years' adjuvant treatment for breast cancer. Lancet 365:60–62.[CrossRef][Web of Science][Medline]
6. Thurlimann B, Keshaviah A, Coates AS, et al. for the BIG 1-98 Collaborative Group. (2005) A comparison of letrozole and tamoxifen in postmenopausal women with early breast cancer. N Engl J Med 353:262747–2757.
7. Coombes RC, Hall E, Gibson LJ, et al. (2004) A randomized trial of exemestane after two to three years of tamoxifen therapy in postmenopausal women with primary breast cancer. N Engl J Med 350:1081–1091.
8. Coombes RC, Hall E, Snowdon CF, et al. (2004) The Intergroup Exemestane Study: a randomized trial in postmenopausal patients with early breast cancer who remain disease-free after two to three years of tamoxifen—updated survival analysis. Breast Cancer Res Treat 88:57 (Abstr 3).[CrossRef]
9. Coombes RC, Paridaens R, Jassem J, et al. (2006) First mature analysis of the Intergroup Exemestane Study. Proc Am Soc Clin Oncol 24:18 Suppl, 9S (Abstr 527).
10. Jakesz R, Jonat W, Gnant M, et al. (2005) Switching of postmenopausal women with endocrine responsive early breast cancer to anastrozole after 2 years' adjuvant tamoxifen: combined results of ABCSG trial 8 and ARNO 95 trial. Lancet 366:455–462.[CrossRef][Web of Science][Medline]
11. Boccardo F, Rubagotti A, Puntoni M, et al. (2005) Switching to anastrozole versus continued tamoxifen treatment of early breast cancer: preliminary results of the Italian Tamoxifen Anastrozole trial. J Clin Oncol 23:5138–5147.
12. Jonat W, Gnant M, Boccardo F, et al. (2005) Switching from adjuvant tamoxifen to anastrozole in postmenopausal women with hormone-responsive early breast cancer: a meta-analysis of the ARNO 95 trial, ABCSG trial 8, and the ITA trial. Breast Cancer Res Treat 94:S11 (Abstr 18).
13. Goss PE, Ingle JN, Martino S, et al. (2005) Randomized trial of letrozole following tamoxifen as extended adjuvant therapy in receptor-positive breast cancer: updated findings from NCIC CTG MA.17. J Natl Cancer Inst 97:171262–1271.
14. Goss PE, Ingle JN, Martino S, et al. (2003) A randomized trial of letrozole in postmenopausal women after five years of tamoxifen therapy for early-stage breast cancer. N Engl J Med 349:1793–1802.
15. Hillner BE. (2004) Benefit and projected cost-effectiveness of anastrozole versus tamoxifen as initial adjuvant therapy for patients with early-stage estrogen receptor-positive breast cancer. Cancer 101:61311–1322.[CrossRef][Web of Science][Medline]
16. Karnon J. (2006) Aromatase inhibitors in breast cancer: a review of cost considerations and cost effectiveness. Pharmacoeconomics 24:3215–232.[CrossRef][Web of Science][Medline]
17. Punglia RS, Kuntz KM, Winer EP, et al. (2005) Optimizing adjuvant endocrine therapy in postmenopausal women with early stage breast cancer: a decision analysis. J Clin Oncol 23:5178–5187.
18. Cuzick J, Sasieni P, Howell A. (2006) Should aromatase inhibitors be used as initial adjuvant treatment or sequenced after tamoxifen? Br J Cancer 94:4460–464.[CrossRef][Web of Science][Medline]
19. Lønning PE. (2006) Comparing cost/utility of giving an aromatase inhibitor as monotherapy for 5 years versus sequential administration following 2–3 or 5 years of tamoxifen as adjuvant treatment for postmenopausal breast cancer. Ann Oncol 17:2217–225.
20. Skedgel C, Rayson D, Dewar R, et al. (2006) Cost-utility of adjuvant hormone therapies for breast cancer in post-menopausal women: sequential tamoxifen-exemestane and upfront anastrozole. Breast Cancer Res Treat August 2; [Epub ahead of print].
21. Electronic Medicines Compendium. (2005) http://emc.medicines.org.uk/emc/assets/c/html/displayDocPrinterFriendly.asp?documentid=2484 (23 February 2006, date last accessed). Pharmacia, Kent, UK.
22. Sonnenberg F and Beck R. (1993) Markov models in medical decision making: a practical guide. Med Decis Making 13:322–338.
23. Early Breast Cancer Trialists' Collaborative Group. (1998) Tamoxifen for early breast cancer: an overview of the randomised trials. Lancet 351:1451–1467.[CrossRef][Web of Science][Medline]
24. Statistics Canada. Life Tables—Canada, provinces and territories, 1995 –1997. Cat. No. 84–537. Government of Canada. 2003 Statistics, Canada.
25. Tufts-New England Medical Center, Institute for Clinical Research and Health Policy Studies. The CEA Registry. http://www.hsph.harvard.edu/cearegistry/data/phaseIIpreferenceweights.pdf (1 June 2005, date last accessed).
26. Will BP, Berthelot JM, Le Petit C, et al. (2000) Estimates of the lifetime costs of breast cancer treatment in Canada. Eur J Cancer 36:6724–735.[CrossRef][Web of Science][Medline]
27. Potvin K, Younis T, Sellon M, et al. (2005) Patterns of trastuzumab use and cost in a single Canadian cancer institute. Proc Am Soc Clinic Onc 23:16 Suppl, 545s (Abstr 6070).
28. Al-Fozan H, Dufort J, Kaplow M, et al. (2002) Cost analysis of myomectomy, hysterectomy, and uterine artery embolization. Am J Obstet Gynecol 187:51401–1404.[CrossRef][Web of Science][Medline]
29. Lessard C and Framarin A. (2003) Endometrial Ablation Techniques in the Treatment of Dysfunctional Bleeding. (Agence d'evaluation technologies et des modes d'intervention en sante, Quebec).
30. Skedgel C, Goeree R, Pleasance S, et al. (2004) The cost-effectiveness of extended antithrombotic prophylaxis following total hip arthroplasty. Value Health 7:6693 (Abstr PCV26).
31. Coyle D, Cranney A, Lee K, et al. (2001) Cost effectiveness of nasal calcitonin in postmenopausal women. Pharmacoeconomics 19:5565–575.[CrossRef][Web of Science][Medline]
32. Grima D, Burge T, Becker D, et al. (2002) Short-term cost-effectiveness of bisphosphonate therapies for postmenopausal osteoporotic women at high risk of fracture. P & T 27:9448–455.
33. Statistics Canada. Consumer Price Index (Health Care). http://cansim2.statcan.ca/cgi-win/cnsmcgi.exe?Lang=E&RootDir=CII/&ResultTemplate=CII/CII_pick&Array_Pick=1&ArrayId=326-0002 (4 June 2005, date last accessed).
34. Olivotto IA, Bajdik CD, Ravdin PM, et al. (2005) Population-based validation of the prognostic model ADJUVANT!for early breast cancer. J Clin Oncol 23:122716–2725.
35. Reeder CA and Gordon D. (2006) Managing oncology costs. Am J Manag Care 12:1 Suppl, S3–S16.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||