© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Clin Transplant 2014: 28: 1084–1091 DOI: 10.1111/ctr.12421
Clinical Transplantation
Cost-effectiveness analysis of early vs. late autologous stem cell transplantation in multiple myeloma Pandya C, Hashmi S, Khera N, Gertz MA, Dispenzieri A, Hogan W, Siddiqui M, Noyes K, Kumar SK. Cost-effectiveness analysis of early vs. late autologous stem cell transplantation in multiple myeloma. Abstract: Background: Autologous stem cell transplant (ASCT) is the current standard of care for most patients with multiple myeloma (MM) who are transplant eligible, yet the timing of ASCT is disputed due to a similar overall (OS) and progression-free survival with an early ASCT (eASCT) or a delayed ASCT (dASCT) approach. Objective: We developed a decision analytic model to perform costeffectiveness analysis of the two commonly used treatment strategies for MM. Methods: Data on disease progression and treatment effectiveness came from 2001 to 2008 cohort treated at the Mayo Clinic and from published studies. Cost analysis was performed from a third-party payer perspective. Results: The Consumer Price Index adjusted 2012 costs of eASCT and dASCT were $249 236 and $262 610, respectively. eASCT cohort had a benefit of 1.96 quality-adjusted life years (QALYs), 0.23 QALYs more than dASCT, implying that eASCT is preferred (dominant) over dASCT. The most critical variables in one-way sensitivity analysis were treatmentrelated mortality and OS associated with eASCT strategy. Conclusions: We conclude that eASCT could potentially be a relatively cost-effective treatment option for appropriate patients with MM, and these results would help patients, providers, and payers in decision making for timing of ASCT.
Chintan Pandyaa, Shahrukh Hashmib,c, Nandita Kherad, Morie A. Gertzb,c, Angela Dispenzierib,c, William Hoganb,c, Mustaqeem Siddiquib,c, Katia Noyese and Shaji K. Kumarb,c a
Department of Public Health Sciences, University of Rochester, Rochester, NY, b Transplant Center, Mayo Clinic, Rochester, MN, cDivision of Hematology, Department of Internal Medicine, Mayo Clinic, Rochester, MN, dHematology Oncology/Blood and Marrow Transplant, Mayo Clinic Arizona, Phoenix, AZ and eSurgical Health Outcomes and Research Enterprise (SHORE), University of Rochester, Rochester, NY USA Key words: autologous stem cell transplant – cost-effective analysis – decision tree – early vs. delayed stem cell transplant – multiple myeloma Corresponding author: Chintan Pandya, MD, MPH, Division of Health Policy and Outcomes Research, Department of Public Health Sciences, University of Rochester Medical Center, 265 Crittenden Blvd, PO Box 420644, Rochester, NY 14642-0644, USA. Tel.: 631-413-6172; fax: 585-461-4532; e-mail: [email protected] Conflict of Interest: No relevant conflict of interest to declare. Accepted for publication 13 July 2014
Multiple myeloma (MM) remains the most common indication for hematopoietic stem cell transplants (HSCT) in the United States (US) (1), yet a controversy exists about the timing of HSCT, because early compared to delayed autologous HSCT (ASCT) results in similar progression-free survival (PFS) and overall survival (2, 3). The current expert guidelines for MM management, that is, Mayo Stratification for Myeloma And Riskadapted Therapy (mSMART) (4), UK Myeloma Forum (UKMF) (5), National Comprehensive
1084
Cancer Network (NCCN) (6), and the European Society for Medical Oncology (ESMO) Guidelines Working Group (7) do not clearly specify a preference of early or delayed ASCT in newly diagnosed patients with MM. With approximately 5000 HSCT for MM being currently performed in the United States annually, the estimated costs of autologous HSCT in the United States (adjusted to 2012) (8, 9) can be up to millions of US dollars annually. This costly procedure may add to the growing economic constraints
Cost analysis of marrow transplant in myeloma to the public health systems in many countries including the United States. Whether early or a delayed transplant strategy can result in substantial cost benefits is unknown. The two most recent retrospective studies (2, 10) have compared only the clinical benefits of early vs. delayed ASCT in patients with MM. Both the studies reported the absence of statistically significant difference in the clinical outcomes such as overall survival and progression-free survival between the two strategies. There is no published randomized trial to date addressing the issue of timing of ASCT, which is an important question in the era of novel agents (thalidomide, lenalidomide, and bortezomib). To our knowledge, this is the first economic analysis of timing of ASCT for patients with MM. The American College of Physicians has advocated for cost-effectiveness analysis to be on the agenda for comparative effectiveness research (11), which itself is a new arena of research funded by the 2010 US Health Care Reform. Thus, a health economic analysis of autologous HSCT in MM is warranted and may help the clinicians and policy makers in establishing guidelines in the MM treatment paradigm. In medical decision making, there are many situations where decisions must be made effectively and reliably when adequate evidence-based data are unavailable. Decision trees are a reliable and effective decision making technique that provide high classification accuracy with a simple representation of currently available research (12). Here we designed a study, based on the best current data, to model a decision tree of early ASCT (eASCT) vs. delayed ASCT (dASCT) strategies in MM to estimate their cost-effectiveness.
Methods and model inputs
Chemotherapy for 6 months
Transplant
Time to Progression: 20 months
Therapy for 22 months
Fig. 1. Reference case for early transplant: probabilities of four yr events in a newly diagnosed multiple myeloma patient.
Decision tree model
Fig. 2. Reference case for delayed transplant: probabilities of four yr events in a newly diagnosed multiple myeloma patient.
We constructed a decision analytic model to compare cost-effectiveness of early vs. delayed transplant from published data on MM transplant studies. The reference case (which forms the base case for derivation of the cost-effectiveness ratio [13]) for the decision tree was based on assumptions from published data of a retrospective study undertaken by Mayo Clinic on early vs. delayed transplants in newly diagnosed patients with MM who were treated with immunomodulatory drugs (IMiDs) from 2001 to 2008 at the Mayo Clinic (2), as depicted in Figs. 1 and 2. In this study, patients who underwent a stem cell harvest attempt were considered transplantation eligible, and if the ASCT was performed within 12 months of diagno-
sis and within two months of harvest, it was considered to be eASCT group; the delayed group consisted of those patients who had ASCT performed any time after 12 months of diagnosis. The overall survival rate and the time to progression after ASCT were found to be comparable in both eASCT and dASCT cohorts. Patients in this study received lenalidomide or thalidomide along with dexamethasone as the chemotherapy regimen. Melphalan was used as a preparative regimen in most and no patients received any post-ASCT maintenance therapy in this study. Almost all transplants at Mayo Clinic during this time period were performed as an outpatient.
1085
Pandya et al. Cost data
The cost analysis was performed from a US thirdparty payer perspective. Although the Harmonized Index of Consumer Prices (HICP) is a well-known indicator of inflation in many countries, the Consumer Price Index (CPI) is still the most commonly utilized inflation adjustment tool in the US healthcare studies; thus, costs of ASCT and chemotherapy were converted into 2012 US$ for this analysis using the standard CPI (14). Cost of transplantation was obtained from study performed by Khera et al. (9), in which the estimated cost of autologous SCTs ranged from $ 36 000 to 88 000. We used the average cost of $62 000 (2012 CPI) for the transplantation, which is similar to the currently reported cost (CPI unadjusted) of autologous transplants in the United States (15, 16). Estimates of treatment costs in myeloma have been published by many groups (17–19). For our model, the cost of chemotherapy was determined from the study performed by Fullerton et al. (19) in which average wholesale price (AWP) was utilized for costs. The drug and medical cost for one cycle of chemotherapy was estimated to be $6687 (adjusted for 2012 CPI). Using the one cycle cost, total cost of chemotherapy was calculated for early and delayed ASCT based on the total number of cycles administered to each group. Early SCT totaled 28 cycles, delayed SCT totaled 30 cycles for a four-yr model as per the original publication (2). As lenalidomide is much more commonly utilized for initial therapy compared with thalidomide in the Western Hemisphere, only lenalidomide costs were considered for the model input. Quality of life data
The outcomes were expressed in the units of health utility values obtained from phase III HOVON trial, in which the efficacy of intensified chemotherapy followed by ASCT was compared with intensified chemotherapy alone in patients newly diagnosed with multiple myeloma (20). The quality of life (QoL) in this study was assessed using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire C30 (EORTC QLQ-C30) and the EuroQol-5D (21). All scales and items were linearly transformed to range from 0 to 100. The utilities measured by the EuroQol-5D ranged from 0.40 before induction therapy to 0.70 after ASCT and translated to QALYs using area under the curve method. QALY for both the early and delayed ASCT cohorts was derived by multiplying the QoL with survival. Incremental cost-effectiveness ratio (ICER) was derived as a
1086
Table 1. Base case inputs (2, 3, 9, 19, 20) Input variable Delayed SCT variables Treatment-related mortality Response rate Overall survival Complete remission (CR) Relapse rate in CR Partial remission (PR) Relapse rate in PR Early SCT variables Treatment-related mortality Response rate Overall survival Complete remission (CR) Relapse rate in CR Partial remission (PR) Relapse rate in PR Quality-adjusted life years Transplantation related death Overall mortality after response No relapse after CR Relapse after CR No relapse after PR Relapse after PR Overall mortality after no response Costs Chemotherapy Transplant cost
0.08 0.87 0.56 0.37 0.14 0.63 0.12 0.02 0.85 0.65 0.35 0.14 0.65 0.12 0 1 2.8 2 2.6 2 0 $6687 $62 000
ratio of the change in costs to incremental benefits of eASCT vs. dASCT as described by original methods (22). Tree Age Pro 2012 software was used to perform the decision analysis (Table 1). Cost-effectiveness analysis
For cost-effective analysis interpretations, we used the generally accepted cost-effectiveness range of $50 000–$100 000 per QALY in the United States, which is considered as societally acceptable. In congruence with majority of economic evaluations, we discounted future costs and outcomes equally at the rate of 0–7% annually. For the treatment strategies, we calculated the incremental cost-effectiveness ratios (ICERs) expressed as cost per quality of life year with each treatment option. Sensitivity analysis
Tornado analysis (Figure not shown) was performed to determine model stability to parameter estimation and to elicit which of the variables in the decision analysis had significant impact on results. One-way and two-way sensitivity analysis of the variables found influential in tornado analysis was performed to assess the influence of variability in these parameters on the model’s effect.
Cost analysis of marrow transplant in myeloma Results
As the decision tree model was based on our published data on IMiD patients undergoing ASCT, the patient characteristics are given in the original publication (2). The cohort of patients in this study received lenalidomide at standard dose of 25 mg daily, days 1–21 along with dexamethasone in a 28-d cycle. Due to the generic availability of dexamethasone, its cost was considered negligible and not included in the model input. The CPI adjusted cost of eASCT (28 cycles of chemotherapy + ASCT) was $249 236, whereas the cost of dASCT (30 cycles of chemotherapy + ASCT) was $262 610. Patients who underwent eASCT had an expected benefit of 1.96 QALYs following treatment, which was 0.23 QALYs more than the patients who received dASCT implying that eASCT is preferred over dASCT in regard to cost-effectiveness (Fig. 3). The ICER was calculated for the early vs. delayed approach; however, its utility becomes secondary because eASCT dominated over dASCT in the decision tree. Sensitivity analysis
Fig. 4 indicates that for all the probabilities of transplant-related mortality (TRM) of eASCT below 0.135, eASCT is the preferred option.
Similarly as per Fig. 5, we found that eASCT is the preferred option for all the OS rates of early SCT above 0.538. That is, eASCT showed dominance even if the probability of OS was lowered to 53.8% from the OS of reference case of 65%. Variability in the overall response rate (ORR) of early SCT did not have significant impact on the preferred outcome of the decision tree analysis as indicated in Fig. 6. Two-way sensitivity analysis of OS and ORR shows that eASCT is the preferred treatment option for all probabilities except probabilities of OS 0.7 (Fig. 7). Thus, eASCT remained the preferred treatment option across all the plausible ranges of TRM, OS, and ORR in one-way and two-way sensitivity analyses. The baseline results are shown for undiscounted (0%) outcomes and costs, while sensitivity analysis was performed for the discount range of 3–7%. Discounted rates did not have an effect on the results of preferred intervention. Discussion
Significant advances in MM treatment paradigm have increased the initial drug therapy options in newly diagnosed patients with MM, which consequently may have an impact on the definitive role of ASCT in these patients. However, the timing of ASCT has become even more controversial in the
Fig. 3. Decision tree analysis of early vs. delayed transplant in newly diagnosed multiple myeloma patients.
1087
Pandya et al.
Fig. 4. Sensitivity analysis of transplant related mortality.
Fig. 5. Sensitivity analysis of overall survival.
Fig. 6. Sensitivity analysis of overall response rate.
novel era agent after the publication of Mayo Clinic (2) results on OS and PFS with respect to early and delayed ASCT. Given nearly similar mortality and toxicity with either approach, our
1088
analysis supports the decision of early transplant approach due to substantial cost benefits. Besides being the first cost-effectiveness analysis on the timing of ASCT in MM, our analysis has
Cost analysis of marrow transplant in myeloma
Fig. 7. Two-way sensitivity analysis.
many strengths—the sensitivity analysis captured most of clinically relevant variables, that is, procedure mortality, survival, and the response rates; the study was not supported or solicited by any pharmaceutical industry; data from patients treated with novel agent were utilized. However, a few limitations remain. Some practitioners are currently using post-ASCT maintenance therapy based on the results from IFM (23) and the CALGB (24) studies. We did not incorporate maintenance therapy in either of the early or delayed group. Some myeloma experts caution against the routine use of post-ASCT maintenance therapies based on multiple factors including cost (25, 26). Badros et al. (25) indicated that the yearly costs of lenalidomide maintenance therapy would approach US$ 163 281, which would translate into an annual societal cost of 816 million US$ (given approximately 5000 annual transplants for MM in the United States)—for an incurable treatment. Our analysis was limited to dexamethasone and lenalidomide regimen only. We did not account for other regimens (particularly bortezomib based regimens) in this analysis. If we were to incorporate the data from studies incorporating other novel agents, it may result in different costs, but would be unlikely to result in significant changes in costeffectiveness of early vs. late strategy of ASCT because the denominator for the ICERs would largely be unchanged given excellent safety profiles of novel agents and a nearly similar OS. This analysis focused on the timing of ASCT but not on the choice of pre-ASCT induction therapy in newly diagnosed patients with MM, whose therapy we believe should be stratified according to the cytogenetic risk groups (4, 27). Our analysis was limited to direct costs of drug and the inpatient hospitalization costs. Unlike the cost-effectiveness models in cancer screening stud-
ies (28, 29), an accurate modeling of indirect costs is limited by the absence of data on measurement of costs in clinical trials in hematologic malignancies. Recently, Garrison et al. (30) compared the cost-effectiveness of novel agent based regimens in newly diagnosed patients with MM including indirect costs. Although we could not capture the data for monitoring of blood counts or the data for the requirement of anticoagulation in small number of patients who developed thrombosis during drug therapy, it is likely that the potential cost burden of the dASCT cohort will be much more compared to eASCT, as the total drug therapy duration is more in the former. We used AWP in the final model. While the best surrogate for the actual cost of the price of a drug in United States is controversial (31, 32), currently the average sales price (ASP) is considered to be an appropriate benchmark for the monetary drug cost by many health economists due to the current Centers for Medicare & Medicaid Services (CMS) reimbursement policy (33). As lenalidomide was the only non-generic drug used for cost analysis in this study, a small difference in the cost of lenalidomide between ASP and AWP is unlikely to skew the results in either direction with respect to ICERs for the timing of transplants. A significant source of uncertainty in the study stems from lack of direct RCT evidence on the intervention. Until such data are available to conduct robust costeffective analyses, indirect comparison to guide future studies could be conducted using registry data such as that obtained from Tufts Medical Center Cost-Effectiveness Analysis (CEA) Registry (34). The results of cost-effectiveness and decision tree modeling in this study strengthen the case of utilization of ASCT early in a newly diagnosed (transplant eligible) appropriate patient with MM, and
1089
Pandya et al.
this information would help in decision making for the patients, providers, and the third-party payers. The cost-effectiveness of early ASCT in this study is contingent on the critical interpretation of published studies used to populate model and assumptions concerning model parameters. However, implementing this approach, in wake of the absence of direct RCT evidence, we can parse out the uncertainties in the evidence base that could influence the cost-effectiveness of the intervention. This is highlighted in the one-way and two-way sensitivity analysis indicating the importance of transplant-related mortality, overall survival, and response rate in eASCT, which may influence the cost-effectiveness of the intervention. Our study is the first CEA to our knowledge evaluating early vs. delayed ASCT and serves as a valuable approach to structuring the decision problem as well as bringing the current available evidence to bear on the problem for future observational or economic studies for ASCT in MM. Future studies in the MM treatment paradigm must focus on the cost benefits of these drugs with relationship to transplant and survival, and the cost-effectiveness ratios need to be continually reevaluated in light of these advancements. Authors’ contributions
Chintan Pandya and Shahrukh Hashmi: Contributed equally to concept/design of the study, data analysis and interpretation, drafting the article, critical revision of the article, and approval of the article; Morie A. Gertz, Nandita Khera, Angela Dispenzieri, Mustaqeem Siddiqui, William Hogan, Katia Noyes and Shaji K. Kumar: Contributed to the editing of the paper and approval of final version of the article. We would also like to thank the editor and anonymous reviewers of Clinical Transplantation for their insightful comments and suggestions to improve the manuscript. References 1. PASQUINI MC, WANG Z. Current use and outcome of hematopoietic stem cell transplantation: CIBMTR Summary Slides, 2013. Available at: http://www.cibmtr.org. 2. KUMAR SK, LACY MQ, DISPENZIERI A et al. Early versus delayed autologous transplantation after immunomodulatory agents-based induction therapy in patients with newly diagnosed multiple myeloma. Cancer 2012: 118: 1585. 3. FERMAND JP, RAVAUD P, CHEVRET S et al. High-dose therapy and autologous peripheral blood stem cell transplantation in multiple myeloma: up-front or rescue treatment? Results of a multicenter sequential randomized clinical trial. Blood 1998: 92: 3131.
1090
4. CONSENSUS GUIDELINES TO MANAGEMENT OF PLASMA CELL DISORDERS. Stratification for Myeloma And Risk-adapted Therapy (mSMART). http://www.msmart.org/msmart_ma r09_002.htm (accessed 03/2013). 5. British Committee for Standards in Haematology in conjunction with the UK Myeloma Forum (UKMF) Guidelines on the Management and Diagnosis of Multiple Myeloma September 2010. http://www.bcshguidelines. com/documents/MYELOMA_Mngmt_GUIDELINE_RE VISION_Sept_2010.pdf (accessed 03/2013). 6. NATIONAL COMPREHENSIVE CANCER NETWORK. Myeloma: http://www.nccn.org/professionals/physician_gls/f_guide lines.asp#myeloma (accessed 03/2013). 7. HAROUSSEAU JL, DREYLING M. The ESMO Guidelines Working Group Multiple myeloma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2010: 21: v155. 8. FREY P, STINSON T, SISTON A et al. Lack of caregivers limits use of outpatient hematopoietic stem cell transplant program. Bone Marrow Transplant 2002: 30: 741. 9. KHERA N, ZELIADT SB, LEE SJ. “Economics of hematopoietic cell transplantation.” Blood 2012: 120: 1545. 10. CORSO A, MANGIACAVALLI S, COCITO F et al. Long term evaluation of the impact of autologous peripheral blood stem cell transplantation in multiple myeloma: a cost-effectiveness analysis. PLoS ONE 2013: 8: e75047. 11. AMERICAN COLLEGE OF PHYSICIANS. Information on costeffectiveness: an essential product of a national comparative effectiveness program. Ann Intern Med 2008: 148: 956. 12. BRIGGS A, CLAXTON K, SCULPHER M. Decision Modelling for Health Economic Evaluation. Oxford: Oxford University Press, 2006. 13. RUSSELL LB, GOLD MR, SIEGEL JE et al. The role of costeffectiveness analysis in health and medicine. Panel on Cost-Effectiveness in Health and Medicine. JAMA 1996: 276: 1172. 14. U.S. BUREAU OF LABOR STATISTICS. http://www.bls.gov/bls/ inflation.htm (accessed 04/2013). 15. SMITH TJ, HILLNER BE, SCHMITZ N et al. Economic analysis of a randomized clinical trial to compare filgrastimmobilized peripheral-blood progenitor-cell transplantation and autologous bone marrow transplantation in patients with Hodgkin’s and non-Hodgkin’s lymphoma. J Clin Oncol 1997: 15: 5. 16. LEE SJ, KLAR N, WEEKS JC, ANTIN JH. Predicting costs of stem-cell transplantation. J Clin Oncol 2000: 18: 64. 17. ARMOIRY X, FAGNANI F. Management of relapsed or refractory multiple myeloma in French hospitals and estimation of associated direct costs: a multi-centre retrospective cohort study. J Clin Pharm Ther 2011: 36: 19. 18. ISHAK J, RODRIGUES F. Cost effectiveness of treatments for relapsed/refractory multiple myeloma: response to a methodology. Eur J Haematol 2011: 87: 95. 19. FULLERTON DP, TRAUTMAN H, HUANG H. A budget impact model comparing resource utilization of four approved therapies for multiple myeloma (MM) in the US. Blood 2007: 110: 3324. 20. SEGEREN CM, SONNEVELD P, VAN DER HOLT B et al. Overall and event-free survival are not improved by the use of myeloablative therapy following intensified chemotherapy in previously untreated patients with multiple myeloma: a prospective randomized phase 3 study. Blood 2003: 101: 2144. 21. VAN AGTHOVEN M, SEGEREN CM, BUIJT I. A cost-utility analysis comparing intensive chemotherapy alone to inten-
Cost analysis of marrow transplant in myeloma
22.
23.
24.
25. 26.
27.
28.
sive chemotherapy followed by myeloablative chemotherapy with autologous stem-cell rescue. Eur J Cancer 2004: 40: 1159. FOLLAND S, GOODMAN AC, STANO M. Chapter 4: Economic efficiency and cost-benefit analysis. In: The Economics of Health and Health Care, 6th edn. Boston, MA: Prentice Hall, 2010: 68. ATTAL M, CANCES L, MARIT G. Lenalidomide maintenance after stem-cell transplantation for multiple myeloma. N Engl J Med 2012: 366: 1782. MCCARTHY PL, OWZAR K, HOFMEISTER CC. Lenalidomide after stem-cell transplantation for multiple myeloma. N Engl J Med 2012: 366: 1770. BADROS AZ. Lenalidomide in myeloma — a high-maintenance friend. N Engl J Med 2012: 366: 1836. KYLE ROBERT A. Role of maintenance therapy after autologous stem cell transplant for multiple myeloma: lessons for cancer therapy. Mayo Clin Proc 2011: 86: 419. CAVO M, RAJKUMAR SV, PALUMBO A et al. International Myeloma Working Group consensus approach to the treatment of multiple myeloma patients who are candidates for autologous stem cell transplantation. Blood 2011: 117: 6063. GOLDIE SJ, GAFFIKIN L, GOLDHABER-FIEBERT JD et al. For the alliance for cervical cancer prevention cost working group cost effectiveness of cervical cancer screening in five developing countries. N Engl J Med 2005: 353: 2158.
29. KHANDKER RK, DULSKI A, KILPATRICK JB. A decision model and cost-effectiveness analysis of colorectal cancer screening and surveillance guidelines for averagerisk adults. Int J Technol Assess Health Care 2000: 16: 799. 30. GARRISON LP, WANG ST, HUANG H et al. The cost-effectiveness of initial treatment of multiple myeloma in the U.S. with bortezomib plus melphalan and prednisone versus thalidomide plus melphalan and prednisone or lenalidomide plus melphalan and prednisone with continuous lenalidomide maintenance treatment. Oncologist 2013: 18: 27. 31. CURTISS FR, LETTRICH P, FAIRMAN KA. What is the price benchmark to replace average wholesale price (AWP)? J Manag Care Pharm 2010: 16: 492. 32. DEPARTMENT OF HEALTH AND HUMAN SERVICES. Office of Inspector General (OIG) Report ‘Replacing Average Wholesale Price: Medicaid Drug Payment Policy’. https:// oig.hhs.gov/oei/reports/oei-03-11-00060.pdf (accessed 03/ 2013). 33. MEDICARE PART B DRUG AVERAGE SALES PRICE: http:// www.cms.gov/Medicare/Medicare-Fee-for-Service-Part-BDrugs/McrPartBDrugAvgSalesPrice/index.html?redirect= /mcrpartbdrugavgsalesprice/ (accessed 03/2013). 34. THORAT T, CANGELOSI M, NEUMANN PJ. Skills of the trade: the Tufts cost-effectiveness analysis registry. J Bene-Cost Anal 2012: 3: 1.
1091
Clin Transplant 2014: 28: 1084–1091 DOI: 10.1111/ctr.12421
Clinical Transplantation
Cost-effectiveness analysis of early vs. late autologous stem cell transplantation in multiple myeloma Pandya C, Hashmi S, Khera N, Gertz MA, Dispenzieri A, Hogan W, Siddiqui M, Noyes K, Kumar SK. Cost-effectiveness analysis of early vs. late autologous stem cell transplantation in multiple myeloma. Abstract: Background: Autologous stem cell transplant (ASCT) is the current standard of care for most patients with multiple myeloma (MM) who are transplant eligible, yet the timing of ASCT is disputed due to a similar overall (OS) and progression-free survival with an early ASCT (eASCT) or a delayed ASCT (dASCT) approach. Objective: We developed a decision analytic model to perform costeffectiveness analysis of the two commonly used treatment strategies for MM. Methods: Data on disease progression and treatment effectiveness came from 2001 to 2008 cohort treated at the Mayo Clinic and from published studies. Cost analysis was performed from a third-party payer perspective. Results: The Consumer Price Index adjusted 2012 costs of eASCT and dASCT were $249 236 and $262 610, respectively. eASCT cohort had a benefit of 1.96 quality-adjusted life years (QALYs), 0.23 QALYs more than dASCT, implying that eASCT is preferred (dominant) over dASCT. The most critical variables in one-way sensitivity analysis were treatmentrelated mortality and OS associated with eASCT strategy. Conclusions: We conclude that eASCT could potentially be a relatively cost-effective treatment option for appropriate patients with MM, and these results would help patients, providers, and payers in decision making for timing of ASCT.
Chintan Pandyaa, Shahrukh Hashmib,c, Nandita Kherad, Morie A. Gertzb,c, Angela Dispenzierib,c, William Hoganb,c, Mustaqeem Siddiquib,c, Katia Noyese and Shaji K. Kumarb,c a
Department of Public Health Sciences, University of Rochester, Rochester, NY, b Transplant Center, Mayo Clinic, Rochester, MN, cDivision of Hematology, Department of Internal Medicine, Mayo Clinic, Rochester, MN, dHematology Oncology/Blood and Marrow Transplant, Mayo Clinic Arizona, Phoenix, AZ and eSurgical Health Outcomes and Research Enterprise (SHORE), University of Rochester, Rochester, NY USA Key words: autologous stem cell transplant – cost-effective analysis – decision tree – early vs. delayed stem cell transplant – multiple myeloma Corresponding author: Chintan Pandya, MD, MPH, Division of Health Policy and Outcomes Research, Department of Public Health Sciences, University of Rochester Medical Center, 265 Crittenden Blvd, PO Box 420644, Rochester, NY 14642-0644, USA. Tel.: 631-413-6172; fax: 585-461-4532; e-mail: [email protected] Conflict of Interest: No relevant conflict of interest to declare. Accepted for publication 13 July 2014
Multiple myeloma (MM) remains the most common indication for hematopoietic stem cell transplants (HSCT) in the United States (US) (1), yet a controversy exists about the timing of HSCT, because early compared to delayed autologous HSCT (ASCT) results in similar progression-free survival (PFS) and overall survival (2, 3). The current expert guidelines for MM management, that is, Mayo Stratification for Myeloma And Riskadapted Therapy (mSMART) (4), UK Myeloma Forum (UKMF) (5), National Comprehensive
1084
Cancer Network (NCCN) (6), and the European Society for Medical Oncology (ESMO) Guidelines Working Group (7) do not clearly specify a preference of early or delayed ASCT in newly diagnosed patients with MM. With approximately 5000 HSCT for MM being currently performed in the United States annually, the estimated costs of autologous HSCT in the United States (adjusted to 2012) (8, 9) can be up to millions of US dollars annually. This costly procedure may add to the growing economic constraints
Cost analysis of marrow transplant in myeloma to the public health systems in many countries including the United States. Whether early or a delayed transplant strategy can result in substantial cost benefits is unknown. The two most recent retrospective studies (2, 10) have compared only the clinical benefits of early vs. delayed ASCT in patients with MM. Both the studies reported the absence of statistically significant difference in the clinical outcomes such as overall survival and progression-free survival between the two strategies. There is no published randomized trial to date addressing the issue of timing of ASCT, which is an important question in the era of novel agents (thalidomide, lenalidomide, and bortezomib). To our knowledge, this is the first economic analysis of timing of ASCT for patients with MM. The American College of Physicians has advocated for cost-effectiveness analysis to be on the agenda for comparative effectiveness research (11), which itself is a new arena of research funded by the 2010 US Health Care Reform. Thus, a health economic analysis of autologous HSCT in MM is warranted and may help the clinicians and policy makers in establishing guidelines in the MM treatment paradigm. In medical decision making, there are many situations where decisions must be made effectively and reliably when adequate evidence-based data are unavailable. Decision trees are a reliable and effective decision making technique that provide high classification accuracy with a simple representation of currently available research (12). Here we designed a study, based on the best current data, to model a decision tree of early ASCT (eASCT) vs. delayed ASCT (dASCT) strategies in MM to estimate their cost-effectiveness.
Methods and model inputs
Chemotherapy for 6 months
Transplant
Time to Progression: 20 months
Therapy for 22 months
Fig. 1. Reference case for early transplant: probabilities of four yr events in a newly diagnosed multiple myeloma patient.
Decision tree model
Fig. 2. Reference case for delayed transplant: probabilities of four yr events in a newly diagnosed multiple myeloma patient.
We constructed a decision analytic model to compare cost-effectiveness of early vs. delayed transplant from published data on MM transplant studies. The reference case (which forms the base case for derivation of the cost-effectiveness ratio [13]) for the decision tree was based on assumptions from published data of a retrospective study undertaken by Mayo Clinic on early vs. delayed transplants in newly diagnosed patients with MM who were treated with immunomodulatory drugs (IMiDs) from 2001 to 2008 at the Mayo Clinic (2), as depicted in Figs. 1 and 2. In this study, patients who underwent a stem cell harvest attempt were considered transplantation eligible, and if the ASCT was performed within 12 months of diagno-
sis and within two months of harvest, it was considered to be eASCT group; the delayed group consisted of those patients who had ASCT performed any time after 12 months of diagnosis. The overall survival rate and the time to progression after ASCT were found to be comparable in both eASCT and dASCT cohorts. Patients in this study received lenalidomide or thalidomide along with dexamethasone as the chemotherapy regimen. Melphalan was used as a preparative regimen in most and no patients received any post-ASCT maintenance therapy in this study. Almost all transplants at Mayo Clinic during this time period were performed as an outpatient.
1085
Pandya et al. Cost data
The cost analysis was performed from a US thirdparty payer perspective. Although the Harmonized Index of Consumer Prices (HICP) is a well-known indicator of inflation in many countries, the Consumer Price Index (CPI) is still the most commonly utilized inflation adjustment tool in the US healthcare studies; thus, costs of ASCT and chemotherapy were converted into 2012 US$ for this analysis using the standard CPI (14). Cost of transplantation was obtained from study performed by Khera et al. (9), in which the estimated cost of autologous SCTs ranged from $ 36 000 to 88 000. We used the average cost of $62 000 (2012 CPI) for the transplantation, which is similar to the currently reported cost (CPI unadjusted) of autologous transplants in the United States (15, 16). Estimates of treatment costs in myeloma have been published by many groups (17–19). For our model, the cost of chemotherapy was determined from the study performed by Fullerton et al. (19) in which average wholesale price (AWP) was utilized for costs. The drug and medical cost for one cycle of chemotherapy was estimated to be $6687 (adjusted for 2012 CPI). Using the one cycle cost, total cost of chemotherapy was calculated for early and delayed ASCT based on the total number of cycles administered to each group. Early SCT totaled 28 cycles, delayed SCT totaled 30 cycles for a four-yr model as per the original publication (2). As lenalidomide is much more commonly utilized for initial therapy compared with thalidomide in the Western Hemisphere, only lenalidomide costs were considered for the model input. Quality of life data
The outcomes were expressed in the units of health utility values obtained from phase III HOVON trial, in which the efficacy of intensified chemotherapy followed by ASCT was compared with intensified chemotherapy alone in patients newly diagnosed with multiple myeloma (20). The quality of life (QoL) in this study was assessed using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire C30 (EORTC QLQ-C30) and the EuroQol-5D (21). All scales and items were linearly transformed to range from 0 to 100. The utilities measured by the EuroQol-5D ranged from 0.40 before induction therapy to 0.70 after ASCT and translated to QALYs using area under the curve method. QALY for both the early and delayed ASCT cohorts was derived by multiplying the QoL with survival. Incremental cost-effectiveness ratio (ICER) was derived as a
1086
Table 1. Base case inputs (2, 3, 9, 19, 20) Input variable Delayed SCT variables Treatment-related mortality Response rate Overall survival Complete remission (CR) Relapse rate in CR Partial remission (PR) Relapse rate in PR Early SCT variables Treatment-related mortality Response rate Overall survival Complete remission (CR) Relapse rate in CR Partial remission (PR) Relapse rate in PR Quality-adjusted life years Transplantation related death Overall mortality after response No relapse after CR Relapse after CR No relapse after PR Relapse after PR Overall mortality after no response Costs Chemotherapy Transplant cost
0.08 0.87 0.56 0.37 0.14 0.63 0.12 0.02 0.85 0.65 0.35 0.14 0.65 0.12 0 1 2.8 2 2.6 2 0 $6687 $62 000
ratio of the change in costs to incremental benefits of eASCT vs. dASCT as described by original methods (22). Tree Age Pro 2012 software was used to perform the decision analysis (Table 1). Cost-effectiveness analysis
For cost-effective analysis interpretations, we used the generally accepted cost-effectiveness range of $50 000–$100 000 per QALY in the United States, which is considered as societally acceptable. In congruence with majority of economic evaluations, we discounted future costs and outcomes equally at the rate of 0–7% annually. For the treatment strategies, we calculated the incremental cost-effectiveness ratios (ICERs) expressed as cost per quality of life year with each treatment option. Sensitivity analysis
Tornado analysis (Figure not shown) was performed to determine model stability to parameter estimation and to elicit which of the variables in the decision analysis had significant impact on results. One-way and two-way sensitivity analysis of the variables found influential in tornado analysis was performed to assess the influence of variability in these parameters on the model’s effect.
Cost analysis of marrow transplant in myeloma Results
As the decision tree model was based on our published data on IMiD patients undergoing ASCT, the patient characteristics are given in the original publication (2). The cohort of patients in this study received lenalidomide at standard dose of 25 mg daily, days 1–21 along with dexamethasone in a 28-d cycle. Due to the generic availability of dexamethasone, its cost was considered negligible and not included in the model input. The CPI adjusted cost of eASCT (28 cycles of chemotherapy + ASCT) was $249 236, whereas the cost of dASCT (30 cycles of chemotherapy + ASCT) was $262 610. Patients who underwent eASCT had an expected benefit of 1.96 QALYs following treatment, which was 0.23 QALYs more than the patients who received dASCT implying that eASCT is preferred over dASCT in regard to cost-effectiveness (Fig. 3). The ICER was calculated for the early vs. delayed approach; however, its utility becomes secondary because eASCT dominated over dASCT in the decision tree. Sensitivity analysis
Fig. 4 indicates that for all the probabilities of transplant-related mortality (TRM) of eASCT below 0.135, eASCT is the preferred option.
Similarly as per Fig. 5, we found that eASCT is the preferred option for all the OS rates of early SCT above 0.538. That is, eASCT showed dominance even if the probability of OS was lowered to 53.8% from the OS of reference case of 65%. Variability in the overall response rate (ORR) of early SCT did not have significant impact on the preferred outcome of the decision tree analysis as indicated in Fig. 6. Two-way sensitivity analysis of OS and ORR shows that eASCT is the preferred treatment option for all probabilities except probabilities of OS 0.7 (Fig. 7). Thus, eASCT remained the preferred treatment option across all the plausible ranges of TRM, OS, and ORR in one-way and two-way sensitivity analyses. The baseline results are shown for undiscounted (0%) outcomes and costs, while sensitivity analysis was performed for the discount range of 3–7%. Discounted rates did not have an effect on the results of preferred intervention. Discussion
Significant advances in MM treatment paradigm have increased the initial drug therapy options in newly diagnosed patients with MM, which consequently may have an impact on the definitive role of ASCT in these patients. However, the timing of ASCT has become even more controversial in the
Fig. 3. Decision tree analysis of early vs. delayed transplant in newly diagnosed multiple myeloma patients.
1087
Pandya et al.
Fig. 4. Sensitivity analysis of transplant related mortality.
Fig. 5. Sensitivity analysis of overall survival.
Fig. 6. Sensitivity analysis of overall response rate.
novel era agent after the publication of Mayo Clinic (2) results on OS and PFS with respect to early and delayed ASCT. Given nearly similar mortality and toxicity with either approach, our
1088
analysis supports the decision of early transplant approach due to substantial cost benefits. Besides being the first cost-effectiveness analysis on the timing of ASCT in MM, our analysis has
Cost analysis of marrow transplant in myeloma
Fig. 7. Two-way sensitivity analysis.
many strengths—the sensitivity analysis captured most of clinically relevant variables, that is, procedure mortality, survival, and the response rates; the study was not supported or solicited by any pharmaceutical industry; data from patients treated with novel agent were utilized. However, a few limitations remain. Some practitioners are currently using post-ASCT maintenance therapy based on the results from IFM (23) and the CALGB (24) studies. We did not incorporate maintenance therapy in either of the early or delayed group. Some myeloma experts caution against the routine use of post-ASCT maintenance therapies based on multiple factors including cost (25, 26). Badros et al. (25) indicated that the yearly costs of lenalidomide maintenance therapy would approach US$ 163 281, which would translate into an annual societal cost of 816 million US$ (given approximately 5000 annual transplants for MM in the United States)—for an incurable treatment. Our analysis was limited to dexamethasone and lenalidomide regimen only. We did not account for other regimens (particularly bortezomib based regimens) in this analysis. If we were to incorporate the data from studies incorporating other novel agents, it may result in different costs, but would be unlikely to result in significant changes in costeffectiveness of early vs. late strategy of ASCT because the denominator for the ICERs would largely be unchanged given excellent safety profiles of novel agents and a nearly similar OS. This analysis focused on the timing of ASCT but not on the choice of pre-ASCT induction therapy in newly diagnosed patients with MM, whose therapy we believe should be stratified according to the cytogenetic risk groups (4, 27). Our analysis was limited to direct costs of drug and the inpatient hospitalization costs. Unlike the cost-effectiveness models in cancer screening stud-
ies (28, 29), an accurate modeling of indirect costs is limited by the absence of data on measurement of costs in clinical trials in hematologic malignancies. Recently, Garrison et al. (30) compared the cost-effectiveness of novel agent based regimens in newly diagnosed patients with MM including indirect costs. Although we could not capture the data for monitoring of blood counts or the data for the requirement of anticoagulation in small number of patients who developed thrombosis during drug therapy, it is likely that the potential cost burden of the dASCT cohort will be much more compared to eASCT, as the total drug therapy duration is more in the former. We used AWP in the final model. While the best surrogate for the actual cost of the price of a drug in United States is controversial (31, 32), currently the average sales price (ASP) is considered to be an appropriate benchmark for the monetary drug cost by many health economists due to the current Centers for Medicare & Medicaid Services (CMS) reimbursement policy (33). As lenalidomide was the only non-generic drug used for cost analysis in this study, a small difference in the cost of lenalidomide between ASP and AWP is unlikely to skew the results in either direction with respect to ICERs for the timing of transplants. A significant source of uncertainty in the study stems from lack of direct RCT evidence on the intervention. Until such data are available to conduct robust costeffective analyses, indirect comparison to guide future studies could be conducted using registry data such as that obtained from Tufts Medical Center Cost-Effectiveness Analysis (CEA) Registry (34). The results of cost-effectiveness and decision tree modeling in this study strengthen the case of utilization of ASCT early in a newly diagnosed (transplant eligible) appropriate patient with MM, and
1089
Pandya et al.
this information would help in decision making for the patients, providers, and the third-party payers. The cost-effectiveness of early ASCT in this study is contingent on the critical interpretation of published studies used to populate model and assumptions concerning model parameters. However, implementing this approach, in wake of the absence of direct RCT evidence, we can parse out the uncertainties in the evidence base that could influence the cost-effectiveness of the intervention. This is highlighted in the one-way and two-way sensitivity analysis indicating the importance of transplant-related mortality, overall survival, and response rate in eASCT, which may influence the cost-effectiveness of the intervention. Our study is the first CEA to our knowledge evaluating early vs. delayed ASCT and serves as a valuable approach to structuring the decision problem as well as bringing the current available evidence to bear on the problem for future observational or economic studies for ASCT in MM. Future studies in the MM treatment paradigm must focus on the cost benefits of these drugs with relationship to transplant and survival, and the cost-effectiveness ratios need to be continually reevaluated in light of these advancements. Authors’ contributions
Chintan Pandya and Shahrukh Hashmi: Contributed equally to concept/design of the study, data analysis and interpretation, drafting the article, critical revision of the article, and approval of the article; Morie A. Gertz, Nandita Khera, Angela Dispenzieri, Mustaqeem Siddiqui, William Hogan, Katia Noyes and Shaji K. Kumar: Contributed to the editing of the paper and approval of final version of the article. We would also like to thank the editor and anonymous reviewers of Clinical Transplantation for their insightful comments and suggestions to improve the manuscript. References 1. PASQUINI MC, WANG Z. Current use and outcome of hematopoietic stem cell transplantation: CIBMTR Summary Slides, 2013. Available at: http://www.cibmtr.org. 2. KUMAR SK, LACY MQ, DISPENZIERI A et al. Early versus delayed autologous transplantation after immunomodulatory agents-based induction therapy in patients with newly diagnosed multiple myeloma. Cancer 2012: 118: 1585. 3. FERMAND JP, RAVAUD P, CHEVRET S et al. High-dose therapy and autologous peripheral blood stem cell transplantation in multiple myeloma: up-front or rescue treatment? Results of a multicenter sequential randomized clinical trial. Blood 1998: 92: 3131.
1090
4. CONSENSUS GUIDELINES TO MANAGEMENT OF PLASMA CELL DISORDERS. Stratification for Myeloma And Risk-adapted Therapy (mSMART). http://www.msmart.org/msmart_ma r09_002.htm (accessed 03/2013). 5. British Committee for Standards in Haematology in conjunction with the UK Myeloma Forum (UKMF) Guidelines on the Management and Diagnosis of Multiple Myeloma September 2010. http://www.bcshguidelines. com/documents/MYELOMA_Mngmt_GUIDELINE_RE VISION_Sept_2010.pdf (accessed 03/2013). 6. NATIONAL COMPREHENSIVE CANCER NETWORK. Myeloma: http://www.nccn.org/professionals/physician_gls/f_guide lines.asp#myeloma (accessed 03/2013). 7. HAROUSSEAU JL, DREYLING M. The ESMO Guidelines Working Group Multiple myeloma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2010: 21: v155. 8. FREY P, STINSON T, SISTON A et al. Lack of caregivers limits use of outpatient hematopoietic stem cell transplant program. Bone Marrow Transplant 2002: 30: 741. 9. KHERA N, ZELIADT SB, LEE SJ. “Economics of hematopoietic cell transplantation.” Blood 2012: 120: 1545. 10. CORSO A, MANGIACAVALLI S, COCITO F et al. Long term evaluation of the impact of autologous peripheral blood stem cell transplantation in multiple myeloma: a cost-effectiveness analysis. PLoS ONE 2013: 8: e75047. 11. AMERICAN COLLEGE OF PHYSICIANS. Information on costeffectiveness: an essential product of a national comparative effectiveness program. Ann Intern Med 2008: 148: 956. 12. BRIGGS A, CLAXTON K, SCULPHER M. Decision Modelling for Health Economic Evaluation. Oxford: Oxford University Press, 2006. 13. RUSSELL LB, GOLD MR, SIEGEL JE et al. The role of costeffectiveness analysis in health and medicine. Panel on Cost-Effectiveness in Health and Medicine. JAMA 1996: 276: 1172. 14. U.S. BUREAU OF LABOR STATISTICS. http://www.bls.gov/bls/ inflation.htm (accessed 04/2013). 15. SMITH TJ, HILLNER BE, SCHMITZ N et al. Economic analysis of a randomized clinical trial to compare filgrastimmobilized peripheral-blood progenitor-cell transplantation and autologous bone marrow transplantation in patients with Hodgkin’s and non-Hodgkin’s lymphoma. J Clin Oncol 1997: 15: 5. 16. LEE SJ, KLAR N, WEEKS JC, ANTIN JH. Predicting costs of stem-cell transplantation. J Clin Oncol 2000: 18: 64. 17. ARMOIRY X, FAGNANI F. Management of relapsed or refractory multiple myeloma in French hospitals and estimation of associated direct costs: a multi-centre retrospective cohort study. J Clin Pharm Ther 2011: 36: 19. 18. ISHAK J, RODRIGUES F. Cost effectiveness of treatments for relapsed/refractory multiple myeloma: response to a methodology. Eur J Haematol 2011: 87: 95. 19. FULLERTON DP, TRAUTMAN H, HUANG H. A budget impact model comparing resource utilization of four approved therapies for multiple myeloma (MM) in the US. Blood 2007: 110: 3324. 20. SEGEREN CM, SONNEVELD P, VAN DER HOLT B et al. Overall and event-free survival are not improved by the use of myeloablative therapy following intensified chemotherapy in previously untreated patients with multiple myeloma: a prospective randomized phase 3 study. Blood 2003: 101: 2144. 21. VAN AGTHOVEN M, SEGEREN CM, BUIJT I. A cost-utility analysis comparing intensive chemotherapy alone to inten-
Cost analysis of marrow transplant in myeloma
22.
23.
24.
25. 26.
27.
28.
sive chemotherapy followed by myeloablative chemotherapy with autologous stem-cell rescue. Eur J Cancer 2004: 40: 1159. FOLLAND S, GOODMAN AC, STANO M. Chapter 4: Economic efficiency and cost-benefit analysis. In: The Economics of Health and Health Care, 6th edn. Boston, MA: Prentice Hall, 2010: 68. ATTAL M, CANCES L, MARIT G. Lenalidomide maintenance after stem-cell transplantation for multiple myeloma. N Engl J Med 2012: 366: 1782. MCCARTHY PL, OWZAR K, HOFMEISTER CC. Lenalidomide after stem-cell transplantation for multiple myeloma. N Engl J Med 2012: 366: 1770. BADROS AZ. Lenalidomide in myeloma — a high-maintenance friend. N Engl J Med 2012: 366: 1836. KYLE ROBERT A. Role of maintenance therapy after autologous stem cell transplant for multiple myeloma: lessons for cancer therapy. Mayo Clin Proc 2011: 86: 419. CAVO M, RAJKUMAR SV, PALUMBO A et al. International Myeloma Working Group consensus approach to the treatment of multiple myeloma patients who are candidates for autologous stem cell transplantation. Blood 2011: 117: 6063. GOLDIE SJ, GAFFIKIN L, GOLDHABER-FIEBERT JD et al. For the alliance for cervical cancer prevention cost working group cost effectiveness of cervical cancer screening in five developing countries. N Engl J Med 2005: 353: 2158.
29. KHANDKER RK, DULSKI A, KILPATRICK JB. A decision model and cost-effectiveness analysis of colorectal cancer screening and surveillance guidelines for averagerisk adults. Int J Technol Assess Health Care 2000: 16: 799. 30. GARRISON LP, WANG ST, HUANG H et al. The cost-effectiveness of initial treatment of multiple myeloma in the U.S. with bortezomib plus melphalan and prednisone versus thalidomide plus melphalan and prednisone or lenalidomide plus melphalan and prednisone with continuous lenalidomide maintenance treatment. Oncologist 2013: 18: 27. 31. CURTISS FR, LETTRICH P, FAIRMAN KA. What is the price benchmark to replace average wholesale price (AWP)? J Manag Care Pharm 2010: 16: 492. 32. DEPARTMENT OF HEALTH AND HUMAN SERVICES. Office of Inspector General (OIG) Report ‘Replacing Average Wholesale Price: Medicaid Drug Payment Policy’. https:// oig.hhs.gov/oei/reports/oei-03-11-00060.pdf (accessed 03/ 2013). 33. MEDICARE PART B DRUG AVERAGE SALES PRICE: http:// www.cms.gov/Medicare/Medicare-Fee-for-Service-Part-BDrugs/McrPartBDrugAvgSalesPrice/index.html?redirect= /mcrpartbdrugavgsalesprice/ (accessed 03/2013). 34. THORAT T, CANGELOSI M, NEUMANN PJ. Skills of the trade: the Tufts cost-effectiveness analysis registry. J Bene-Cost Anal 2012: 3: 1.
1091
