Anemia is common in patients with cancer and is associated with negative clinical outcomes. Treatment with erythropoiesis-stimulating agents (ESA) effectively increases hemoglobin and reduces the need for blood transfusions, improves short-term disease-specific quality of life (QOL),1 and is preferred over transfusions by patients,2, 3 but is costly. In addition, recent evidence indicates that traditional administration of ESA in patients with anemia related to cancer (ARC) is associated with increased risk of mortality and adverse events.4-6
The conclusions of previous economic evaluations of ESA for the treatment of ARC have varied, although many have concluded that it is generally not cost-effective.1, 7, 8 However, all previous analyses have evaluated use of this agent as commonly used before recent guidelines from the American Society of Clinical Oncology/American Society of Hematology (ASCO/ASH), which recommend initiation of ESA when hemoglobin is <10 g/dL, a target hemoglobin of ≤12 g/dL, and use in chemotherapy-induced anemia only.9 Guideline adherence is likely to result in lower cumulative ESA dose, the major driver of costs. Furthermore, although increased risk of thromboembolic events and death have been reported with traditional ESA use in this population,4-6 the risk of these events is less clear with more conservative ESA use. As such uncertainty exists concerning the balance of costs, adverse events, and clinical benefit of ESA in this patient population. Many patients and clinicians seek to use ESA because of its convenience, effectiveness in reducing transfusions, and improvements in QOL. We sought to determine the cost-effectiveness of ESA when used in a conventional fashion and in accordance with ASCO/ASH guidelines.
- Top of page
- MATERIALS AND METHODS
- CONFLICT OF INTEREST DISCLOSURES
In the base case, we modeled a cohort of patients similar to those identified in a recent systematic review.4 The average age of the cohort was 62 years, their mean initial hemoglobin was 10.4 g/dL, and 80% were treated with chemotherapy. The type of cancer among patients in this cohort was 21% hematological, 63% solid, and 17% mixed malignancy. Compared with no ESA and supportive transfusion, treatment with epoetin resulted in incremental costs of $8643 and incremental benefits of 0.03 QALYs over 15 weeks, resulting in an incremental cost per QALY gained of $267,000. During a 17-month time frame, the ESA strategy was dominated with incremental costs of $8643 and lower QALYs (−0.086) compared with the no ESA strategy.
The base-case results were robust in sensitivity analysis (Tables 4 and 5). By using the extremes of 95% confidence intervals for short-term and long-term mortality, both favoring ESA, resulted in an incremental cost per QALY gained of $101,000. A best-case quality-of-life scenario in which all assumptions favored the ESA arm over the short-term led to an incremental cost effectiveness of $125,000/QALY (Table 6). Making the assumption that there is no increased risk of mortality or adverse events with ESA did not significantly alter the results of the QOL scenarios. Incremental cost per QALY gained remained >$180,000 under the assumption that transfusion required 1 day of hospitalization, disutility for transfusion reaction, weekly dose and dose duration of epoetin, and cost per unit of ESA (Table 5). Varying the cumulative dose and cost per unit of ESA had a larger quantitative impact on results. Use of the most up-to-date Canadian estimates of the cost of transfusing 1 unit of erythrocytes (including collection, production, distribution, administration, and transfusion reactions)37 resulted in an incremental cost per QALY gained of $271,000/QALY.
Table 4. Sensitivity Analyses Parameters and Results for Relative Risk of Mortality at 15 Weeks and at 1 Year (Base-Case ESA Compared With No ESA)
|Model Time Horizon||RR Short-Term Mortality, 15 Wk||RR Long-Term Mortality, Annual||Incremental $/QALY Gained|
|Short-Term, 15 wk||1.15 (Point estimate)||na||$267,346 (Base Case)|
| Sensitivity analysis||1.29 (Upper 95% CI)||na||$315,336|
| Sensitivity analysis||1.03 (Lower 95% CI)||na||$236,497|
|Long-Term, 1.3 y||1.15 (Point estimate)||1.33 (Point estimate)||Dominated|
| Sensitivity analysis||1.29 (Upper 95% CI)||1.88 (Upper 95% CI)||Dominated|
| Sensitivity analysis||1.03 (Lower 95% CI)||0.79 (Lower 95% CI)||$100,984|
Table 5. Base-Case Model of Additional 1-Way Sensitivity Analyses
|Variable||Point Estimate||Range||Incremental Cost||Incremental Effectiveness QALY||Incremental $/QALY Gained|
|Base case|| || ||$8634||0.03||$267,000|
|Cost per 1000 U epoetin||$14.40||−25% to +25%||$6367-$10,919||0.0323-0.0323||$196,946-$337,746|
|Unit of epoetin per wk||42,148||30,000||$6019||0.0323||$186,183|
|Baseline risk of mortality||0.131||0.087 to 0.175||$8643-$8643||0.0360-0.0287||$240,062-$301,628|
|Incremental units of blood transfused||−0.80||−0.99 to −0.61||$8534-$8753||0.0323-0.0323||$263,961-$270,732|
|Cost of transfusion||$576||−25% to +25%||$8758-$8528||0.0323-0.0323||$270,910-$263,783|
|Inclusion of thrombotic event costs||—||—||$8898||0.0323||$275,230|
|Admission to hospital for transfusion, cost per day||$934-$1330|| ||$7905-$8466||0.0323-0.0323||$244,076-$256,255|
Table 6. Sensitivity Analyses and the Impact of Alternate Methods of Estimating QOL and Other Assumptions on Incremental Costs and QALYs of ESA Compared With Standard Treatment
|Source of QOL Estimate||Duration of Benefit||Incremental Hemoglobin ESA vs Standard, g/dL||QOL Standard Treatment||QOL ESA||Incremental Cost||Incremental Effectiveness QALY||Incremental $/QALY Gained|
|Ossa prediction||15 wk||1.50||0.78||0.93||$8643.2||0.0323||$267,346|
|Ossa originala||15 wk||1.50||0.78||0.86||$8643.2||0.0152||$569,411|
|Ossa prediction||15 wk||1.72a||0.78||0.96||$8643.2||0.0397||$217,824|
|Ossa prediction||26 wka||1.72a||0.78||0.96||$8643.2||0.0688||$125,668|
To estimate the cost effectiveness of ESA use concordant with ASH/ASCO guidelines,9 model inputs from studies that met components of these guidelines were summarized where available (Table 3). The initial and actual dose of ESA was substantially lower than the base-case analysis, although infrequently reported. The results were quantitatively lower than the base case, but all exceeded $70,000/QALY, and ESA was dominated (more costly with less benefit) in 3 of 10 models. It should be noted that because of the poor quality of studies, inconsistent reporting, and wide confidence intervals of the model inputs, there is considerable uncertainty in these estimates.
- Top of page
- MATERIALS AND METHODS
- CONFLICT OF INTEREST DISCLOSURES
Recent evidence clearly demonstrates that increased risk of mortality and adverse events are associated with conventional ESA use in patients with anemia related to cancer.4-6 A currently unanswered question facing clinicians and policy makers is whether the benefits in quality of life, on balance, outweigh risks and costs where the risk of adverse consequences of ESA are attenuated or eliminated, which is possible under more conservative ESA administration practices. Several Canadian provinces currently fund ESA for the treatment of ARC when used according to the ASCO/ASH guidelines, despite the lack of evidence to support this practice.4 Our results suggest that treatment with ESA remains economically unattractive even when used according to the ASCO/ASH guidelines41, 42 —similar to the conclusions reached by previous economic evaluations evaluating more liberal use of ESA.1, 7, 8
To our knowledge, no RCTs used a preference-based measure of overall quality of life, which is required to perform cost-utility analyses. This is especially important when considering strategies to treat anemia, as the major benefit is in QOL. Disease-specific measures, which are commonly used in clinical studies of cancer patients with anemia, may have an increased sensitivity to detect small differences in certain aspects of quality of life. Although changes in disease-specific measures do not necessarily lead to clinically important changes in utility-based measures, the magnitude of the changes found4, 40 meet thresholds for minimal clinically important differences (MCID).43, 44 In addition, the sole study that reported a generic (not disease-specific) measure of overall quality of life28 found greater improvements among ESA recipients in the physical components summary of the Short Form 36 (SF-36) (3.4 vs −0.7 for ESA and no ESA, respectively) and in a global measure of QOL that was assessed by using a visual analog scale (11.0 vs −0.4 for ESA and no ESA, respectively). Although this trial was not eligible for inclusion in the systematic review, the magnitude of this improvement also met the MCID threshold.45 Therefore, it appears reasonable to assume that ESA may improve quality of life among cancer patients, at least in the short term.
We used assumptions that may exaggerate the QOL gains associated with differences in hemoglobin, favoring the ESA strategy. In the base-case method of estimating QOL, the difference in the utility score between the ESA and the no ESA arm in surviving patients was 0.15. This large difference in utility is far greater than the minimal clinically important difference of 0.03 for the EQ-5D46 and is similar to the change in QOL that is associated with a patient with end-stage kidney disease on dialysis who receives a kidney transplant.47 Despite this possible overestimation of QOL in the ESA arm, the incremental cost-utility ratio (ICUR) remained >$100,000/QALY, even in multiway sensitivity analysis.
We did not incorporate patient preferences with respect to avoiding blood transfusion (travel, time requirement, intravenous administration, aversion to receipt of blood products) and avoiding the potential cyclic nature of this treatment (gradually dropping hemoglobin followed by a rapid rise with blood transfusion). However, 2 published studies used willingness-to-pay (WTP) and discrete-choice experiment (DCE) methods to estimate the value that participants place on these considerations. Ortega3 reported the willingness-to-pay for ESA therapy (compared with no ESA therapy) of $587 and $613 (1997 US dollars), respectively, from patients who were receiving cisplatin or noncisplatin therapy. By using DCE methods, Ossa2 elicited WTP from laypersons in the United Kingdom of £368 (2004 British pounds). These results indicated the preference of patients and laypersons for ESA over transfusion only, although these studies did not inform respondents of the increased risk of adverse outcomes that have recently come to light. Regardless, incorporation of the monetary value of these preferences would not result in meaningful changes in the conclusions, suggesting that in private-payor systems, this treatment strategy would be unattractive.
We performed additional scenario analyses to simulate the use of ESA in accordance with the ASCO/ASH criteria.9 However, very few studies met even individual components of these guidelines, and the selective reporting of outcomes might have introduced bias. In light of these inherent uncertainties, we performed 10 analyses simulating the use of ESA in accordance with components of ASCO/ASH criteria. In 3 of these analyses, ESA therapy was dominated (more costly with less benefit than the no ESA strategy); in 4 the incremental cost per QALY gained was >$100,000/QALY; and none were <$70,000/QALY. These analyses included scenarios where the mortality risk was similar or lower than the no ESA strategy (and the cost of thrombotic events were not considered). Overall, these findings suggest that even with more conservative administration and dosing, ESA therapy for patients with cancer is unlikely to be economically attractive when using a commonly accepted threshold of economic attractiveness.41, 42
As in most economic evaluations, our models and results are limited by the availability and quality of existing evidence. Our evaluation has been strengthened by its rigorous methods and our state-of-the-art systematic review. The lack of strong evidence on the incremental changes in utility-based QOL was addressed by modeling approaches and sensitivity analyses as described above. The model did not incorporate other considerations that may affect QOL, including cyclical hemoglobin values, aversion to intravenous administration, and receipt of human-blood products, but this was informally assessed by using willingness-to-pay estimates from other sources.
This is the first economic evaluation to determine the ICUR of ESA use in accordance with recently published ASCO/ASH guidelines. Although considerable uncertainty concerning the precise incremental costs and benefits of ESA therapy in this subgroup remains, there is no evidence to suggest that ESA is more economically attractive in this subgroup than in any other. Finally, this economic evaluation considered subjects that were candidates for blood transfusion. Specific subpopulations, such as those in whom blood transfusions are difficult or unacceptable (ie, iron overload, Jehovah's Witnesses) and patients with poor access to transfusion support, were not examined.
In summary, ESA use in patients with ARC does not appear to be economically attractive, even when used in the more conservative fashion recommended by current guidelines. Available evidence suggests that using ESA to treat ARC does not represent a good value for the money.