By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Dr. Hornberger has a consulting arrangement with federal agencies (Veterans Administration, U.S. Health Resources and Services Administration, National Institutes of Health) and with private companies (Roche, Allergan, Boston Scientific, Amgen, and Genomic Health).
F. Hoffmann-La Roche Ltd. supported the development of the model and the subsequent analyses. The sponsor had no role in the model design, cost-effectiveness analysis, or interpretation of the results. All investigators had unlimited access to the model and data. No limitations on publication were imposed. The authors made the final decisions on all aspects of the current article.
For patients with anthracycline-pretreated metastatic breast carcinoma, capecitabine plus docetaxel significantly increased overall survival compared with docetaxel alone. The current study evaluated the cost-effectiveness of the capecitabine/docetaxel combination versus docetaxel monotherapy, comparing the gain in quality-adjusted survival with associated health care costs.
Patients were randomized to receive 21-day cycles of oral capecitabine 1250 mg/m2 twice daily, on Days 1–14, plus docetaxel 75 mg/m2 Day 1 (n = 255), or docetaxel 100 mg/m2 on Day 1 (n = 256). Health and cost outcomes in the two arms were compared, and cost-effectiveness was estimated. Data on survival time and medical care resource use were prospectively collected in the trial. Costs associated with medical care resource use and quality-of-life adjustments were obtained from the published literature. The incremental cost-effectiveness ratio was calculated as the cost per quality-adjusted life year (QALY) gained.
Capecitabine/docetaxel increased the median overall survival by 3 months compared with docetaxel alone (14.5 vs. 11.5 months). The mean quality-adjusted survival was increased by 1.8 months in the capecitabine/docetaxel group. The total medical-resource utilization cost per patient was 8.9% higher with the combination: $24,475 for combination therapy versus $22,477 for single-agent docetaxel. The mean cost per QALY gained with combination therapy was $13,558 (standard deviation, $6742). Cost savings due to reduced docetaxel dose and hospital use were the major cost offsets with the combination. Sensitivity analyses showed that varying the mean hospital cost per day from the 5th to the 95th percentile resulted in cost-utility ratios ranging from $20,326 to as low as $6360.
A recent randomized, Phase III trial demonstrated that a regimen combining the oral fluoropyrimidine, capecitabine (Xeloda; F. Hoffmann-La Roche, Basel, Switzerland), with docetaxel resulted in significantly improved survival, compared with single-agent docetaxel in patients with anthracycline-pretreated metastatic breast carcinoma (MBC).1
In the current study, capecitabine/docetaxel resulted in significantly improved time to disease progression (TTP), the primary end point, compared with single-agent docetaxel (log-rank test, P = 0.0001; hazard ratio = 0.652; 95% confidence interval [CI], 0.545–0.780). The median TTP was 6.1 months (95% CI, 5.4–6.4; 93% of events) in the combination arm and 4.2 months (95% CI, 3.4–4.5; 98% of events) with docetaxel alone. Furthermore, the combination demonstrated a manageable safety profile, with health-related quality of life (HrQOL) maintained in the patients receiving the combination compared with those receiving single-agent docetaxel.1 Consequently, capecitabine/docetaxel has received regulatory approval worldwide, including the United States, Canada, and Europe for the treatment of anthracycline-pretreated patients with MBC. Since the initiation of this trial, third-line therapy with capecitabine has gained acceptance and interpreting the results of the study requires consideration of the benefits of adopting the combination versus extending the treatment via sequential use of the single agents. It should be emphasized that the current study was not designed to address the issue of combination versus sequential administration, and that the majority of patients receiving single-agent docetaxel did not receive capecitabine after completing a docetaxel regimen.
Health economics is a discipline concerned with resource allocation. Policy makers are concerned with balancing available medical care resources with maximizing the health gain for the population as a whole.2 Cost-effectiveness analyses (CEAs) are intended to reveal trade offs between alternative treatments and, as such, can assist the decision-making process. Cost-utility analyses (CUA) combine mortality and morbidity data into a single measure, and are commonly the primary form of CEA recommended by methodologists.3, 4
We have performed an analysis to determine the economic impact of adopting capecitabine/docetaxel combination therapy in place of docetaxel monotherapy. The model was developed to answer the following question, What are the forecasted patient benefits, costs, and cost-effectiveness of adopting capecitabine/docetaxel combination chemotherapy in a target population of patients with anthracycline-pretreated advanced breast carcinoma?
MATERIALS AND METHODS
Medical resource use and CUAs were conducted as part of the pharmacoeconomic evaluation of an open-label, multicenter, randomized Phase III trial comparing capecitabine/docetaxel combination therapy with single-agent docetaxel in patients with anthracycline-pretreated MBC. The primary objective of the current trial was to demonstrate superior TTP with the combination regimen compared with the single-agent regimen. Secondary objectives included demonstration of a superior overall response rate and overall survival with the combination regimen. Additional secondary objectives included comparison of the safety profiles, changes in HrQOL, and medical care utilization in each treatment arm.
In the current pharmacoeconomic analysis, cost analyses of medical resource use and CUAs were performed, using payer and patient perspectives and U.S. health care costs as the base case. These data were included in a health-state-transition (Markov) analytic model,5 to determine the economic impact of adopting capecitabine/docetaxel to replace single-agent docetaxel.
This Markov model incorporates the values that health care professionals attach to the different tumor-response health states described in the clinical trial results. This value, referred to as a “utility” reflects the strength of a patient's preference for a given health state or outcome. The utility approach assigns numerical values on a scale anchored at a low of 0 (representing death) to a high of 1 (representing optimal or perfect health). The most commonly used utility measure is the quality-adjusted life year (QALY), which is a measure of the quantity of life gained in each health state from a treatment, weighted by the quality of that health state.
To the extent that treatment increases the time in preferred health states, the CUA of the more effective combination therapy was expected to improve the cost-utility ratio, whereas the increased toxicity of the combination treatment was expected to reduce it. By using the original trial results as a basis for comparison, and by adding the alternative use of sequential therapy with other agents, the impact of current treatment practices was modeled.
Patient eligibility for the current trial is described in detail elsewhere.1 In total, 511 patients were enrolled from 75 centers in 16 countries. The baseline characteristics of patients were well balanced between the 2 treatment groups and the treatment histories were also similar (Table 1).
Table 1. Baseline Characteristics and Treatment Histories for all Randomized Patients
Anthracyclines were administered in both the neoadjuvant and adjuvant settings in 6% and 5% of patients in the combination and single-agent arms, respectively, and in all three settings in 0.4% of each treatment arm.
Excluding previous hormonal regimens for metastatic disease.
Two patients received study therapy as fourth-line treatment.
Patients were randomized to receive either 1) oral capecitabine 1250 mg/m2 twice daily, on Days 1–14 followed by a 7-day rest period, plus docetaxel 75 mg/m2 administered as a 1-hour intravenous (i.v.) infusion on the first day of each 3-week cycle or 2) docetaxel 100 mg/m2 administered as a 1-hour i.v. infusion on the first day of each 3-week cycle. All patients received docetaxel premedication (e.g., dexamethasone) as per the treatment center policy. Additional premedication is not required for capecitabine.
Patients achieving a complete or partial response or stable disease (SD) after 6 weeks of therapy continued to receive treatment until disease progression or unacceptable toxicity occurred. Patients with documented progressive disease (PD) were withdrawn from the study. Dose modification was undertaken according to a schedule described in detail elsewhere.1
Tumor responses were assessed based on World Health Organization criteria6 at 6-week intervals until Week 48, and then at 12-week intervals until the disease progressed. Tumor response was also assessed in each patient within 2 weeks of discontinuing study medication. Adverse events were recorded throughout the study and for 28 days after the last administration of study drug. Adverse events were graded according to National Cancer Institute of Canada Common Toxicity Criteria, and hand-foot syndrome was graded 1–3, as defined in previous capecitabine clinical studies.7, 8
HrQOL was assessed using the European Organization for Research and Treatment of Cancer quality of life questionnaire C30 (version 2.0) and the Breast Cancer module QLQ-BR23, in centers where questionnaires were available in appropriate, validated translations. The questionnaires were completed before treatment administration on the first day of each cycle, every 6 weeks (until Week 48), and when leaving the study.
Medical Resource Use
During the trial, medical resource use data were collected prospectively at all participating centers in case report forms. The data collected included hospital admissions (including admission diagnosis and length of stay), study medication (including cumulative dose, infusion duration, and frequency), treatments for adverse events (medicines and procedures), and outpatient consultations. Consultations were categorized according to type of care provider (including general practitioner, specialist, or allied health care professional), and location of care (including day care unit, emergency unit, home visit, office visit, or telephone consultation). The data collected were analyzed using standard descriptive statistical techniques.
Design and Structure of the Pharmacoeconomic Model
The Markov model was developed to assess quality-adjusted survival, health care costs, and the cost-effectiveness of capecitabine/docetaxel combination therapy. The model uses U.S. estimates for the unit costs of medical resources (this does not include other societal costs, such as time off work, productivity losses, and out-of-pocket nonmedical expenses). In oncology, these variables have a similarly low impact on cost-utility estimates compared with the impact of other variables, such as the cost of therapy or management of disease progression and adverse events. The model is structured so that it can be adapted to estimate costs in different health care systems in other countries by substituting their unit price and cost information. The first phase of the model describes the duration of study treatment in both treatment groups, which is the period of the trial with the most intensive data collection. Specifically, in the combination arm, this refers to the time patients receive capecitabine plus docetaxel and docetaxel or capecitabine alone if continued as monotherapy up to the time when disease progression was documented. In the monotherapy arm, this refers to the time patients receive single-agent docetaxel. The second phase describes the period of time after disease progression until death and incorporates survival data reported in O'Shaughnessy et al.1
Data and Assumptions in the Pharmacoeconomic Model
The data sources and base-case assumptions used in the model are detailed in Table 2. All costs are shown in U.S. dollars. The mean TTP and duration of overall survival were estimated from the clinical trial data using SPLUS statistical software, based on the methods described by Miller.9 The mean duration of survival after disease progression was estimated by subtracting the mean TTP from the mean overall survival.
Table 2. Variables in Model: Data Sources and Base-Case Assumptions
TTP: time to disease progression; AMA: American Medical Association.
Medical resource utilization data were collected prospectively during the Phase III study. Drug prices were obtained from the Red Book 2003.10 Prices for consultations and administration of chemotherapy were based on current procedural terminology (CPT™) codes developed by the American Medical Association. The mean cost per day of hospitalization was obtained from the American Hospital Association (AHA) hospital statistics 2001.;11 Alternative sources for this cost estimate were also considered to allow for variability in hospital costs.12, 13 The mean duration of treatment and the total number of docetaxel infusions in each treatment arm were calculated based on data collected during the clinical trial. Cumulative doses of capecitabine and docetaxel were also recorded during the clinical trial and, therefore, dose reductions required to manage adverse events are incorporated. The duration, frequency, and type of medications used for the treatment of adverse events were collected prospectively during the trial. Of these, the 10 most frequently used medications were included in the model. The daily cost of each treatment was estimated using recommended dosages from the drug label and cost data from published sources. The total cost of each medication was calculated by multiplying the daily cost of treatment by the total number of days the treatment was used in each treatment arm. This was then divided by the number of patients in the relevant treatment arm to provide the mean cost per patient.
The total cost of hospitalizations for adverse events in each treatment arm was calculated by multiplying the average duration of stay per patient by the estimated cost per day of hospitalization, including the fixed-bed cost and cost of medical care.
Utility assessments were not performed for patients enrolled in the trial and these values were derived from the literature. The utilities were calculated based on data collected by nurses during three Phase II and III studies evaluating docetaxel in patients with MBC.14–16 The first study, conducted in the United Kingdom, evaluated docetaxel and paclitaxel as second-line therapies for patients with anthracycline-resistant MBC.14 Utilities for various health states were established by use of the standard gamble technique and the visual analog method among 30 oncology nurses who acted as proxy patients. The utility scores reported were 0.62 for SD and 0.41 for PD. The second study, performed in France, compared docetaxel, paclitaxel, and vinorelbine as second-line chemotherapy in patients with MBC.15 Utilities, assigned by application of a standard reference lottery using 20 oncology nurses as proxies for the patients, were 0.75 for SD and 0.65 for PD. In the third study, which was conducted in the United States, docetaxel and paclitaxel were compared in patients with advanced breast carcinoma.16 Utility scores, obtained from 29 oncology nurses, were 0.81 for patients with complete or partial responses and 0.39 for patients with PD.
The utility scores identified in these 3 studies were averaged to produce mean values of 0.72 for SD and 0.48 for PD. These mean values were used in the sensitivity analyses.
The cost for each type of medical resource used was calculated by multiplying the estimated mean volume used for that type of resource (e.g., a hospital day) by its market price for the year 2001. Prices for consultations and administration of chemotherapy were based on CPT codes. The Medicare national fee17 was used to obtain a price for each of these resources. AHA hospital statistics (2001) were used for estimating the mean cost of a hospital day of $1067 (to include fixed-bed cost and cost of all medical care received during hospitalization).
Alternative sources of hospital charges per day were also reviewed that demonstrated the possibility of per day charges that may range as high as ≥ $2000.12, 13 Drug prices were obtained from the Red Book.10
Discounting and Time Horizon
In economic models, time discounting is used to compare alternative future streams of costs and benefits, by adjusting all those costs and benefits to an equivalent current value.18 In our analysis, all costs and benefits were discounted at a fixed annual rate of 3%, the standard promulgated by the U.S. Public Health Service.18 The time horizon for the model was 2.9 years, representing the last observed follow-up time. At this time point, the estimated overall survival probability was < 20% in both arms of the study. At such a low survival probability, the difference in survival is unlikely to influence estimates of the duration of survival after disease progression, and further extrapolation of the survival curves would be speculative.
Incremental Cost-Effectiveness and Cost-Utility Analyses
The incremental cost per life year gained was calculated by dividing the difference in the total costs in each arm by the increase in survival for treatment with capecitabine/docetaxel therapy compared with single-agent docetaxel. The incremental cost per QALY gained was calculated by dividing the difference in the total costs in each arm by the difference in expected survival, considering the time in each health state and adjusting for the utility of that health state.
To assess the possible effect of alternative plausible and possible variations in model parameters on the results, a series of sensitivity analyses were performed on the following parameters: 1) medical resource use cost, 2) duration of disease progression-free survival and survival after disease progression, 3) utility for each health state, and 4) discount rate. It should be noted that the Markov model approach, by projecting the disease progression-free period of the patient's disease, risks compressing treatment costs into the disease progression-free period. Sensitivity analyses assess the effect of varying estimates of resource use and effectiveness over a range of values. All assumptions in the model are tested to enable the impact of the best and worse case scenarios on the baseline findings to be investigated.
To explore the cost-effectiveness of treating patients with potentially less or more favorable prognosis than the sample mean estimated from the trial, we simultaneously varied TTP and overall survival times using the 5th percentile (poor prognosis scenario) and 95th percentiles (favorable prognosis scenario) of these distributions. All utilities were simultaneously increased and decreased by 20%. Specifically, the utility for TTP was varied between 0.59 and 0.86, and the utility for survival after disease progression was varied between 0.39 and 0.56. The discount rate was varied between 0% and 9%. Utilities were assumed to have a uniform distribution in the ranges reported above.
The sensitivity analysis also explored the possible effect of variations in hospital cost per day, as it is reasonable to expect that hospitals or other health care providers may be able to negotiate a discount over the Red Book 2003 prices used in the model. A sensitivity analysis assuming a 10% discount was therefore performed.
Finally, to provide an estimate of the distribution of cost-effectiveness, a Monte Carlo simulation (@Risk Version 4.5.1, Palisade Corporation, Newfield, NY) was performed.19 This method derives a cost-effectiveness distribution using the probability distributions of all of the input variables.
Efficacy and Safety
These efficacy results were published previously.1 In summary, patients receiving combination therapy had a 23% reduction in risk of death compared with single-agent docetaxel. The median overall survival was superior in the combination arm compared with docetaxel monotherapy (14.5 vs. 11.5 months), with a hazard ratio of 0.775 (P = 0.0126; 95% CI, 0.634–0.947).
The incidence of treatment-related adverse events was similar in the combination and single-agent arms (98% vs. 94%, respectively).1 A higher proportion of patients in the combination arm than in the single-agent docetaxel arm experienced Grade 3 treatment-related adverse events (71% vs. 49%), and 25% of patients in the combination arm experienced Grade 4 treatment-related adverse events compared with 31% in the single-agent docetaxel arm. Global health score was preselected as the primary end point of HrQOL analysis. HrQOL, as indicated by mean global health scores over time, was similar in the two treatment arms, with a trend towards less deterioration of global health score over time evident in the combination arm.1 The impact of chemotherapy-induced side effects, as measured by the systemic therapy side effects symptom scale, was similar in the two treatment arms.
Pharmacoeconomic Model Outcomes: Clinical Effectiveness
The base-case assumptions used in the model are presented in Table 2. After discounting, the mean TTP was 0.61 years for combination therapy and 0.44 years for single-agent docetaxel, and the mean survival time was 1.38 years for combination therapy and 1.16 years for docetaxel monotherapy.
Model Outcomes: Cost of Chemotherapy
Patients randomized to combination therapy received a mean of 76% of the planned dose of capecitabine, and a mean of 89% of the planned dose of docetaxel. Patients randomized to single-agent docetaxel received a mean of 95% of the planned dose. The mean cumulative doses of study medication were 272,670 mg capecitabine and 579 mg docetaxel in the combination arm, and 842 mg docetaxel in the single-agent arm.
The total cost of chemotherapy was higher in the combination therapy group ($18,252) than in the monotherapy group ($15,729) (Table 3). The number of docetaxel infusions was similar in the 2 treatment arms (4.6 in the combination group and 4.7 in the single-agent group). The reduced dose of docetaxel used in the combination arm (leading to a cost reduction of $3516) was offset by > 50% of the additional cost of capecitabine ($6039).
Table 3. Summary of Key Parameters in Cost-Effectiveness Calculationa
TTP: time to disease progression; PPS: post-disease progression survival.
The reported cost-effectiveness and cost-utility estimates are based on program outputs that use machine precision beyond two digits.
Mean duration of TTP (yrs)
Mean duration of PPS (yrs)
Mean duration of survival (yrs)
Mean duration of quality-adjusted survival (yrs)
Costs (per patient): medical resource use ($)
Hospitalizations for adverse events
Treatment for adverse events
Total costs ($)
Cost per life-year gained ($)
Cost per quality-adjusted life-year gained ($)
Model Outcomes: Costs of Managing Adverse Events
The number of treatment-related hospitalizations was similar in the combination and single-agent arms (96 vs. 91, respectively) and the predominant causes of hospitalization in both treatment groups were neutropenic fever (24% vs. 32%, respectively), neutropenia (6% vs. 8%, respectively), and neutropenic sepsis (3% vs. 6%, respectively). The mean duration of hospitalization for treatment of adverse events was shorter in the combination treatment group (4.8 days) than in the single-agent group (5.5 days). The mean per-patient cost of hospitalization was lower for patients treated with the combination regimen ($5167) than for those treated with docetaxel alone ($5861). Increased hospitalization costs for patients receiving docetaxel monotherapy occurred despite a shorter duration of treatment (median 2.8 months) compared with the combination arm (3.8 months).
The mean total concomitant medication costs (aggregated for the 10 most frequently used medications to treat adverse events in the capecitabine/docetaxel vs. docetaxel groups) were similar, having no marked impact on the outcome of the overall analysis ($501 vs. $396, respectively).
More consultations (including those by telephone) per patient were required for management of adverse events in the combination arm. Overall, a mean of 9.5 consultations per patient were required in the combination arm compared with 6.6 consultations per patient in the single-agent docetaxel arm. Overall, 805 telephone consultations were required in the combination arm versus 533 in the single-agent arm. A substantial proportion of consultations were office visits, which accounted for 36% of consultations in the combination group and for 39% of consultations in the single-agent group. This translated into an additional cost of $267 per patient for the combination arm compared with $201 for the single-agent arm.
Cost-Effectiveness of Capecitabine/Docetaxel Combination Therapy
The key parameters in the cost-effectiveness calculation are shown in Table 3. The total costs per patient were 8.9% higher with capecitabine/docetaxel compared with docetaxel alone ($24,475 vs. $22,477, respectively), resulting in a net increased cost of $1998. The cost per life year gained with capecitabine/docetaxel combination therapy compared with single-agent docetaxel was $9163. The important components contributing to the cost-effectiveness of capecitabine/docetaxel were the reduced dose of docetaxel with the combination (leading to a cost reduction of $3516), offsetting by > 50% the additional cost of capecitabine ($6039), the reduced hospitalization costs for adverse events (reduced by $694 per patient compared with single-agent docetaxel), and improved survival compared with docetaxel alone. Small additional costs for consultations (increased by $66 per patient compared with single-agent docetaxel) and medications for treatment of adverse events (increased by $105 per patient compared with single-agent docetaxel) were incurred in the combination arm.
After adjustment of TTP and survival to account for the different utility of these health states, the cost per QALY gained for capecitabine and docetaxel combination therapy compared with docetaxel alone was $13,558 (standard deviation, $6742; interquartile range, $16,432–$25,211).
Table 4 shows the impact of varying model estimates on incremental cost per life year gained. Varying docetaxel doses and, therefore, costs had a major impact on the incremental cost-effectiveness ratio. At the 5th percentile, the outcome was cost saving, whereas at the 95th percentile, the cost-effectiveness ratio is > $50,000. For other variables, including the cost of outpatient treatment for adverse events, the cost of infusions, and the cost of consultations, the cost-effectiveness ratios remained < $20,000.
Table 4. Sensitivity Analyses: Effects of Varying Model Estimates on ICER
ICER: incremental cost-effectiveness ratio; TTP: time to disease progression; OS: overall survival; CD: capecitabine/docetaxel; D: docetaxel.
Differences are capecitabine/docetaxel minus docetaxel.
Difference in cost of docetaxel (−2925)
Cost of capecitabine (4314)
Difference in cost of infusions (−2)
Difference in cost of hospitalizations for adverse events (−694)
Difference in cost of outpatient treatment for adverse events (66)
Difference in cost of consultations (105)
Survival (yrs) (TTP: CD = 0.61, D = 0.44; OS: CD = 1.38, D = 1.16)
5th percentile/poor prognosis (yrs)
TTP (CD = 0.09, D = 0.08) $55,421 OS (CD = 0.23, D = 0.16)
95th percentile/favorable prognosis (yrs)
TTP (CD = 1.98, D = 1.16) $9712 OS (CD = 2.87, D = 2.86)
Utility of progression-free survival (0.78)
Utility of post-progression survival (0.48)
Discount rate (3%)
Varying the hospital cost per day resulted in cost-effectiveness ratios ranging from $20,326 (for the 95th percentile) to a low of $6360 (for −$1755). Varying the utility of disease progression-free survival revealed cost-effectiveness ratios of $9725 at the 5th percentile and $22,383 at the 95th percentile. Varying the utility of survival after disease progression had a similar effect on the cost-effectiveness ratios.
The results of the current analysis show that the survival benefit with the combination regimen is achieved at a small incremental cost. The cost per life year gained with the capecitabine/docetaxel combination compared with single-agent docetaxel was $9163. After adjustment of TTP and survival to account for the different utility of these health states, the cost per QALY gained for capecitabine/docetaxel compared with docetaxel was calculated as $13,558. This incremental cost-effectiveness ratio is lower than that reported for many other oncologic interventions.20 The important components contributing to the cost-effectiveness of capecitabine/docetaxel are superior survival, the reduced dose of docetaxel offsetting much of the additional cost of capecitabine, and reduced hospitalization costs for adverse events compared with standard single-agent docetaxel. Minimal additional costs were incurred for consultations and medications required for the treatment of adverse events in the combination arm.
The current study is unique because it utilizes data collected prospectively during the clinical trial. A limitation of these analyses is that cost data were not collected after disease progression, as patients were no longer participating in the trial. However, additional costs associated with the subsequent care of patients do not appear to differ between the two treatment groups. The data from the clinical trial show that a similar range and incidence of treatments after the study were administered in the two treatment arms.1 Poststudy therapies for breast carcinoma in the combination and single-agent docetaxel treatment groups included surgical intervention (7% vs. 4%, respectively), radiotherapy (30% in both groups), endocrine therapy (30% vs. 27%), and chemotherapy (70% vs. 63%). The choice of poststudy chemotherapy was similar in each arm, with the most common therapies being vinorelbine (31% vs. 26%), anthracyclines (10% vs. 11%), 5-fluorouracil (20% vs. 23%), and trastuzumab (9% in both groups). After tumor progression, docetaxel alone was given to 5% of patients in the capecitabine/docetaxel arm and to 7% of the patients in the single-agent arm. The administration of poststudy capecitabine was more common in the single-agent arm (17%) than in the combination arm (3%).
Utility assessments were not derived from study participants, but rather from the breast carcinoma literature.14–16 The U.S. panel on cost-effectiveness has recommended that utilities be collected from the general population,3, 4 but there is controversy regarding whether utilities should be collected from patients, their families, health care workers, or lay people who have been given information about the health state in question. The majority of CUAs, which have been reviewed by Earle et al.,20 used preferences provided by nurses and physicians, with only a minority using patients and the community as sources. We have used data collected from oncology nurses (in three countries), who might be expected to have a good understanding of the disease and its impact on patients. There is little reason to presume that patients in France, the United Kingdom, and the United States would have significantly different preferences. therefore, we used a mean utility derived from all three trials. Collection of utility data from the general population or from patients treated in the trial may yield a different result, but it is not anticipated (nor do the sensitivity analyses suggest) that other utility values would substantially affect the conclusions of the current study. Also, although the model was constructed using U.S. cost data and a U.S. health care system perspective, it is amenable to adaptation as the treatment protocols are used internationally. Incorporation of appropriate cost data from the relevant health care system would allow conclusions to be drawn in other countries.
The estimates of total costs in the 2 treatment arms were similar, with the combination associated with a minimal net increased cost of $1998 per patient (Table 3). The addition of capecitabine was associated with an increased cost of $9163 per life year gained in the combination arm compared with the single-agent arm. The reduced dose of docetaxel in the combination arm led to a cost reduction of $3516, which offset the additional cost of capecitabine ($6039). The total cost of hospitalizations for treatment of adverse events was reduced by almost $700 on average in the combination arm compared with the single-agent arm.
Patients receiving the capecitabine/docetaxel combination therapy consulted health care professionals more frequently regarding adverse events than patients receiving docetaxel alone. Patients in the combination arm required a mean of 9.5 consultations, whereas those in the single-agent arm required a mean of 6.6 consultations. Despite the increased number of consultations in the combination therapy group, the additional cost per patient ($66) was minor. The need for more consultations in the combination arm may reflect a lower degree of clinical experience or familiarity with capecitabine and the combination therapy at the time that the clinical trial was conducted.
Improving patient education may play an important role in helping patients to recognize side effects and their severity, enabling them to take action to avoid development of more severe toxicity. In particular, this may also reduce the number of face-to-face consultations and hospitalizations related to adverse events, which incur the highest costs. Increasing the number of telephone consultations may also reduce resource utilization costs associated with treatment of adverse events.
In the metaanalysis of oncology publications performed by Earle et al.,20 the median value for cost-effectiveness among 89 assessments in 40 studies was $20,000 per QALY. Most of these studies developed models of the effectiveness of interventions from secondary sources, but 16 used primary sources. Only six of these studies analyzed data from randomized, controlled trials. Fifty-seven interventions were estimated to have cost-utility ratios of ≤ $50,000 (including 4 cost-saving interventions), and another 10 interventions cost $50,000–$100,000 per QALY. In addition, 12 interventions were dominated, being more costly but not more effective than the alternative (Fig. 1). A benchmark cost-utility ratio of $50,000 per QALY is often cited as a threshold below which interventions are considered to be cost-effective,20 but experts have recently argued for even higher thresholds.21, 22
Despite the limitations of such cross-study comparisons, the current analysis suggests that capecitabine/docetaxel in patients with advanced breast carcinoma is a cost-effective intervention ($13,558 per QALY), with a cost-effectiveness ratio more favorable than approximately 85% of the oncology cost-effectiveness ratios reported in the metaanalysis. In addition, by applying estimates of higher hospital cost per day of $2000 rather than $1067, based on certain cited sources, the cost-effectiveness ratio may be as favorable as $9438.
The sensitivity analyses in these studies were extensive, and included adjustment for utility of health state, survival, and discount. The Monte Carlo simulation revealed that the probability of the cost-effectiveness ratio > $26,000 is < 25%. The study analysis was robust to wide variations of input assumptions and it is reasonable to conclude that the capecitabine/docetaxel combination is likely to be cost-effective even when extended beyond the controlled environment of a clinical trial.
The results of this pharmacoeconomic analysis confirm that the capecitabine/docetaxel combination has a small incremental cost compared with standard single-agent docetaxel. In addition, we have demonstrated that combination therapy provides meaningful clinical benefit for patients in a cost-effective manner, and has an acceptable financial impact on health care systems.
The authors thank J. Hornberger and C. Jamieson for pharmacoeconomic analysis concept and design; C. Jamieson, J. Hornberger, and trial investigators for data acquisition; J. Hornberger and S. Verma for analysis and interpretation of pharmacoeconomic data; S. Verma, J. Hornberger, and C. Jamieson for drafting the current article; J. O'Shaughnessy, S. Jones, J. McKendrick, D. Miles, and C. Twelves for a critical revision of the current article for important intellectual content; J. Hornberger and S. Verma for their analytical expertise; and S. Verma and J. Hornberger for supervision.