This study is a component of a larger project investigating the use of evidence to support resource allocation decisions in cancer care at the BC Cancer Agency (BCCA). The overall goal of this project is to pilot evidence-based methods for priority setting and resource allocation with a Steering Committee composed of senior BC Cancer executives.
EBMA project’s steering committee, consisting of senior executives of the BCCA and other key decision-makers, identified five different program areas for cost-effectiveness analysis, including the focus of this study, the use of bevacizumab as a first- and second-line systemic therapy treatment for advanced colorectal cancer.
Bevacizumab (Avastin; Genetntech/Roche; San Franscisco, CA, USA), given as a first- or second-line systemic therapy in combination with otherhas been found to confer longer overall survival among patients with metastatic disease when compared to 5-FOLFOX or FOLFIRI, CAPOX or CAPIRI alone. The cost-effectiveness of bevacizumab was assessed in a technology appraisal by the National Institute for Health and Clinical Excellence (NICE) in 2007 (National Institute for Health and Clinical Excellence, 2007). The appraisal used a simple-state transition model, populated data from published survival curves, cost, and utility estimates from Genentec. No discounting was used. The base-case cost per quality adjusted life year (QALY) gained was £62,857 for bevacizumab combined with FOLFIRI and £88,436 for bevacizumab combined with 5-FU/LV (National Institute for Health and Clinical Excellence, 2007). The authors conclude that the probability that bevacizumab is cost-effective given a £30,000 willingness-to-pay threshold is effectively zero, and thus the use of the drug as a first line therapy for metastatic CRC is not recommended (National Institute for Health and Clinical Excellence, 2007). A recent Cochrane review found that among the five randomized trials, which included a total of 3101 patients, the addition of bevacizumab to routine systemic therapy prolongs overall survival from 17.7 to 20.5 months, and progression-free survival from 7.1 to 9.7 months when used as a primary treatment (Wagner, et al. 2009). The authors note that the magnitude of the observed survival benefit varied depending on the type of chemotherapy to which it was associated (Wagner, et al. 2009).
BCCA has included bev as systemic therapy protocol for mCRC in combination with:Irinotecan and capecitabine (UGICIRB);Oxaliplain and capecitabine (UGICOXB);Irinotecan, Fluorouracil and Folinic Acid (UGIFFIRB); andOxaliplatin, 5-Fluorouracil, and Folinic Acid (UGIFFOXB)A true post-market cost-effectiveness analysis of this systemic therapy is not possible, since only the healthier patients, free from complications, and diagnosed under age 70 are eligible to receive it. Doing a simple comparison between patients who were treated with bevacizumab versus those who were not would therefore be inherently biased in favour of the bevacizumab, which would nullify the value of the calculated ICER.
This incremental cost-effectiveness assessment used a payer (health system) perspective to compare the era of first- and second-line bevacizumab with the pre-bevacizumab era. Using R-Gui, a Markov transition model (presented in Figure 1) with five health states was created with a one-month cycle length in order to evaluate long-term health outcomes, quality of life effects, and costs associates with the two eras of treatment for metastatic CRC. The model was run on a twelve year time horizon. Results are presented as incremental cost-effectiveness ratios (ICERs) in cost per quality adjusted life year (QALY) gained, as well in cost per life year gained (LYG). Patients begin with a diagnosis of Stage IV (metastatic) CRC. The majority of patients go on to receive some form or forms of systemic therapy (classified as first line, second line, or third line) before eventually transitioning to death; however, a small percentage of patients are not treated with chemotherapy. These patients transition to death at a faster rate than those who receive systemic therapy. Patients in any of the systemic therapy or no chemotherapy states may be receiving other forms of treatments simultaneously, including radiotherapy or surgery. The costs and benefits of these additional treatments have been modelled within the context of the systemic therapy states.
Data for the era-based comparison were generated using a complete cohort of patients diagnosed with stage IV CRC in British Columbia, identified using the BCCA’s Cancer Agency Information Service (CAIS). Included in the pre-era are those patients who received their initial diagnosis between January 1st, 2003 and December 31st, 2004. Follow-up continued to death, or censoring on October 31st, 2005. Included in the bev-era are all patients diagnosed between January 1st, 2006 and December 31st, 2006, and followed until October 31st, 2008 or until death, whichever occurs first. These inclusion and cut-off dates were purposively selected to ensure equal follow-up between the two cohorts, and based on the most recent mortality data available from BC Vital Statistics. Patients who were diagnosed in 2005 were intentionally excluded from either cohort to reduce contamination of the pre-era with patients receiving bevacizumab as part of a clinical trial. Patients who were participants in other clinical trials were also excluded, since treatment for these individuals does not represent a cost to the Agency. Additionally, because of our inability to accurately discern if and when patients diagnosed with early stage disease suffer relapse or progress to metastatic disease using the CAIS data system, only those patients who suffered Stage IV disease on presentation were included. The resulting cohorts consisted of 611 patients diagnosed during the pre-era and 332 diagnosed in the post-era. Probabilities for the era-based analysis are included in Table 1 below.
Probabilities for the era-based analysis are included in Table.Used the sample listed in previous slide to track individuals’ transitions through the study states, and to death, creating transition probabilities according to statistical models.Time-dependent survival probabilities were derived based on Weibull models. Time-varying chemotherapy probabilities were derived based on Exponential models.
Diagnosis and staging: Includes lab tests*, radiology and other imaging**, and staffing costs.Day Surgery: Includes direct costs (nursing, operating room/ICU time, diagnostic imaging, pharmacy, labs) and indirect costs (administration, finance, HR, plant operations).Inpatient costs: Includes direct costs (nursing, operating room/ICU time, diagnostic imaging, pharmacy, labs) and indirect costs (administration, finance, HR, plant operations).Systemic therapy: Includes administration costs (nursing and other staff time, materials etc.)Radiation therapy:Includes direct costs (nursing, operating room/ICU time, diagnostic imaging, pharmacy, labs) and indirect costs (administration, finance, HR, plant operations).
This finding is supported by the survival curves in Figures 3a and 3b. Overall, median survival improved from 15.56 months in the pre-era to 19.45 months in the bev-era. Figure 3a shows survival curves for individuals in each era who were treated with some form of systemic therapy, excluding those individuals who were not treated with this modality. This curve demonstrates a statistically significant improvement in median survival associated with being treated in the bev-era, from 17.2 months to 20.1 months (Chi-square: 4.8, p-value: 0.0285). FOR NO CHEMO:Survival curves corresponding to those patients in each era who were not treated with systemic therapy were also created (Figure 3b). Although there a small increase in median survival associated with being treated in the bev era, this increase was not statistically significant (p=0.198).
Survival improvement more than in clinical trials, but not out of wack
To address the second objective of this analysis, a selected subset of the cohorts identified for the era-based analyses were used in a nested case-control comparison. Included in this comparison were any patients from either era who were diagnosed before age 70, and who received some form of doublet chemotherapy as a first-line treatment. Doublet chemotherapy protocols include FOLFOX, FOLFIRI, CAPOX, and CAPIRI in the pre-era and all those plus UGICIRB, UGICOXB, UGIFFIRB and UGIFFOXB in the bev-era. These selection criteria were applied with the intention of restricting the pre-era and bev-era samples to include only those patients who were or would have been eligible to receive a bevacizumab-based systemic therapy protocol. The limitations of this approach are presented in the discussion section below. As with the era-based analysis, time-dependent survival probabilities were derived based on Weibull models and chemotherapy probabilities were derived based on Exponential models.
Base-case for the nested case-control comparison are presented in Table 8. Treatment during the bev-era resulted in a gain of an average of 0.04 QALYs at a cost increase of $3,791 per patient. The resulting cost-effectiveness ratio is $43,058/QALY or $10,764/LYG, which is substantially more cost-effective than the simple era-based comparison. This is primary due to a reduction in incremental cost of treatment. The average per-patient treatment cost in the pre-bev era is dramatically increased (by almost $10,000) when the sample is restricted. Although this is also true of the average bev-era treatment cost, the increase in cost is not as dramatic (cost increases by about $7,000).
The BC Cancer Agency has elected to include bevacizumab on its drug treatment formulary for advanced CRC. This report has attempted to address, from the perspective of the Agency and greater health care system, the cost-effectiveness of the eras of treatment for CRC, comparing the era before bevacizumab to the era where bevacizumab is now used. Thus, this report directly evaluates the Agency’s decisions to fund bevacizumab. The cost-effectiveness of the bev-era compared to the pre-bev era is $62,468.68/QALY or $15,617/LYG, based on a 3.9 month per-patient improvement in survival, and a $3,791 per-patient increase in cost (when full cohorts of individuals being treated for advanced CRC are assessed). This cost-effectiveness ratio cannot be directly inferred as the cost-effectiveness of bevacizumab because it is not a simple comparison of costs and benefits for individuals who received bevacizumab versus those who did not. A true post-market cost-effectiveness analysis of this systemic therapy is not possible, since only the healthier patients, free from complications, and diagnosed under age 70 are eligible to receive it. Doing a simple comparison between patients who were treated with bevacizumab versus those who were not would therefore be inherently biased in favour of the bevacizumab, which would nullify the value of the calculated ICER.
To get as close as possible to that elusive case-control comparison, the cohorts were restricted to include only those patients who either were or would have been eligible to receive the drug, and all the analyses were reproduced. The cost-effectiveness improves to $43,058/QALY or $10,764/LYG when the cohort is restricted to include only those individuals diagnosed before age 70 who received doublet chemotherapy. The ICERs for the pseudo case-control analysis represent an improvement in cost-effectiveness over the full era-based analysis. The reason for this improvement is in the cost of treatment. Although restricting the sample increased the cost of treatment in both the pre-bev and bev era cohorts, the increase was more dramatic in the pre-era. This is because a higher proportion of patients in the pre-era than the bev era who were not treated with any systemic therapy. When the values for these individuals were removed, the average treatment cost in the pre-era increased dramatically. Both the era-based and case-control analyses produced ICERs lower than those previously published by NICE. There are four main reasons why this finding is not unexpected. First, use of bevacizumab is expanded compared to what was reported in the trial. Several patients over age 65 in the cohort were treated with bevacizumab, and list of precluding factors is less strict during care delivery than it is in a clinical trial. Second, the first cohort of patients to receive bevacizumab at the BCCA were restricted to a maximum of six months of treatment <I need to check with Hagen to find out how long this restriction was in place.> before transitioning onto a non-bevacizumab-based chemotherapy regimen. Truncating the length of time patients received this drug also obviously truncates the allowable cost. Third, not all patients in the bevacizumab era were treated with the drug (in fact, only about 35% of patients received it) and the extra cost is washed out across the remaining cohort. Fourth, and finally, there was in improvement in median survival from the pre-era to the bev-era that was unrelated to the introduction of bevacizumab. This improvement is visible in the survival curve produced for patients who were not treated with systemic therapy (Figure 3b). It is likely that this small, but notable, improvement in survival resulting from an increase in curative surgical treatments and an increase in the proportion of patients receiving systemic therapy, made the bev-era look more effective.
An Application of Evidence-Based Marginal Analysis: Assessing the ...
An Application of Evidence-Based Marginal Analysis: Assessing the Incremental Cost Effectiveness of Eras of Metastatic Colorectal Cancer Therapy in British Columbia, Canada: Pre- and Post-Bevacizumab Introduction <br />Lindsay Hedden<br />Priorities 2010, Boston, MA<br />
Priority Setting and Resource Allocation at BCCA<br />This bevacizumab study is part of a larger program of research into Evidence Based Marginal Analysis<br />Goal: to develop and pilot novel evidence-based methods for priority setting and resource allocation within the context of cancer control and care in British Columbia<br />A key objective: evaluate the effectiveness of priority setting decisions using utilization, mortality, and quality of life data<br />
STEERING COMMITTEE<br />Established and refined decision criteria<br />Identified three areas for potential resource reallocation<br />Reviewed results of cost-effectiveness analyses<br />Made recommendations for resource reallocation <br />PROGRAM PANELS<br />Provide clinical and data expertise on model building<br />Validate results<br />EMBA Study Structure<br />
Bevacizumab (bev): given as a first- or second-line systemic therapy in combination with other regimens to treat metastatic colorectal cancer (mCRC)<br />2.8 month average improvement in overall survival <br />2.6 months average improvement in progression-free survival<br />National Institute for Health and Clinical Excellence (UK) <br />£62,857-£88,436 per QALY gained <br />Use of bev as first-line therapy is NOT recommended<br />Bevacizumab (Avastin): Background<br />
To estimate the incremental cost-effectiveness of bevas a systemic therapy treatment for mCRC, accounting for the differences in costs and health outcomes associated with bevand standard of care treatments<br />BUT: Cannot directly compare costs and outcomes for patients treated vs. not treated with bevacizumab because of selection bias<br />Goal<br />
Compare eras of treatment for mCRC: <br />pre-bevacizumab introduction and post-bevacizumab introduction<br />secondary pseudo case-control comparison<br />Objectives<br />1) To assess the cost-effectiveness of the era of bev protocols in the treatment of mCRC compared with the pre-bev era<br />2)to evaluate the incremental cost-effectiveness of a first- and second-line bevamong the subset of patients receiving “doublet” chemotherapy (5-FU plus irinotecan or oxaliplatin) <br />Approach and Objectives<br />
Complete cohort of patients presenting with mCRC at diagnosis, identified using BCCA’s Information Service (CAIS)<br />Pre-era: Diagnosed Jan 1, 2003-Dec 31, 2004; followed to death, censoring, or Oct 31, 2005<br />Bev-era: Diagnosed Jan 1, 2006-Dec 31, 2006; followed to death, censoring, or Oct 31, 2008<br />611 cases in pre-era & 332 in the post-era<br />Sample<br />
Survival: derived based on Weibull models<br />Chemotherapy: derived based on Exponential models<br />Transition Probabilities<br />
Utility Values<br />*Source: Ness, R.M., et al., Outcome states of colorectal cancer: identification and description using patient focus groups. The American Journal of Gastroenterology, 1998. 93(9): p. 1491-1497<br />
Survival for individuals who initiated chemotherapy<br />
Subset of era-based analysis:<br />1) Diagnosed before age 70<br />2) Treated with first-line doublet chemotherapy<br />Intent: include only patients who wereorwould have been eligible for a bev-based protocol<br />Restricted Analysis<br />
Era-based: $62,468.68/QALY or $15,617/LYG 3.9 month/patient improvement in survival & $3,791/patient increase in cost <br />Not directly inferred as cost-effectiveness of bev<br />Other factors my have led to improvements in survival, increases in cost<br />Interpretation<br />
Restricted Analysis: $43,058/QALY or $10,764/LYG 4.4 month/patient improvement in survival & $1,894/patient increase in cost <br />Closer to a true incremental cost-effectiveness comparing bev with standard of care, but not perfect<br />Both methods produced ICERs demonstrating better cost-effectiveness than estimated by NICE<br />Interpretation (2)<br />
As a 1st or 2nd line treatment for mCRC, bev may be relatively cost-effective, considered as part of a suite of available treatments<br />the era-based ICER of $62,468 is well in-line with cost-effectiveness ratios reported for other therapies for metastatic cancer therapies<br />Implications<br />
Acknowledgements<br />Project team:<br />Dr. Stuart Peacock<br />Dr. Diego Villa<br />Dr. Hagen Kennecke<br />Funding sources:<br />CIHR Partnerships in Health Systems Improvement<br />