|
| HOME | SUBSCRIPTIONS | CURRENT ISSUE | PAST ISSUES | CARDIOSOURCE | SEARCH | HELP | FEEDBACK |
|
|
|
|||||||||
|
|||||||||||
|
J Am Coll Cardiol, 2005; 46:2311-2321, doi:10.1016/j.jacc.2005.08.033 © 2005 by the American College of Cardiology Foundation |
,
,,
* Department of Medicine, Jefferson Medical College, Philadelphia, Pennsylvania
MEDTAP Institute at UBC, Bethesda, Maryland
Department of Health Policy Management, Johns Hopkins School of Public Health, Baltimore, Maryland
Division of Cardiology, University of Colorado, Denver, Colorado
|| Division of Cardiology, University of Southern California, Los Angeles, California
¶ Division of Cardiology, University of California, San Francisco, California
# Division of Cardiology, The Johns Hopkins University School of Medicine, Baltimore, Maryland
** Division of Cardiology, Hershey Medical Center, Hershey, Pennsylvania
Department of Medicine and Pharmacology at Georgetown University School of Medicine, Washington, DC
Veterans Affairs Medical Center, Washington, DC
Guidant Corporation, St. Paul, Minnesota
Manuscript received February 25, 2005; revised manuscript received July 7, 2005, accepted August 9, 2005.
* Reprint requests and correspondence: Dr. Arthur M. Feldman, Department of Medicine, Jefferson Medical College, 1025 Walnut Street, Philadelphia, Pennsylvania 19107 (Email: arthur.feldman{at}jefferson.edu).
A preliminary report of this work was presented at the 54th Annual Scientific Session of the American College of Cardiology, Orlando, Florida, 2005.
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionConclusionsReferences |
|---|
BACKGROUND: In the COMPANION trial, CRT-P and CRT-D reduced the combined risk of all-cause mortality or first hospitalization among patients with advanced heart failure and intraventricular conduction delays, but the cost effectiveness of the therapy remains unknown.
METHODS: In this analysis, intent-to-treat trial data were modeled to estimate the cost effectiveness of CRT-D and CRT-P relative to OPT over a base-case seven-year treatment episode. Exponential survival curves were derived from trial data and adjusted by quality-of-life trial results to yield quality-adjusted life-years (QALYs). For the first two years, follow-up hospitalizations were based on trial data. The model assumed equalized hospitalization rates beyond two years. Initial implantation and follow-up hospitalization costs were estimated using Medicare data.
RESULTS: Over two years, follow-up hospitalization costs were reduced by 29% for CRT-D and 37% for CRT-P. Extending the cost-effectiveness analysis to a seven-year base-case time period, the ICER for CRT-P was $19,600 per QALY and the ICER for CRT-D was $43,000 per QALY relative to OPT.
CONCLUSIONS: For the COMPANION trial patients, the use of CRT-P and CRT-D was associated with a cost-effectiveness ratio below generally accepted benchmarks for therapeutic interventions of $50,000 per QALY to $100,000 per QALY. This suggests that the clinical benefits of CRT-P and CRT-D can be achieved at a reasonable cost.
| |||||||||||||
Recently the Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COMPANION) trial studied a cohort of heart failure patients who received CRT via either a pacemaker (CRT-P) or pacemaker-defibrillator (CRT-D) in combination with optimal pharmacological therapy (OPT) and found a substantial reduction in the combined risk of all-cause mortality or first hospitalization relative to patients with OPT alone. In addition, CRT-D significantly reduced the risk of death from any cause (15). However, the opportunities for improved outcomes for heart failure patients provided by the salutary benefits of CRT, with or without a defibrillator, must be balanced against the initial investment in the implantation of these devices as well as costs for subsequent device management. Dollars spent on implanting devices for heart failure compete with other therapies for our finite health care resources, and it is essential to know whether the long-term health benefits produced by CRT are proportional to their costs. To better understand the relationship of the clinical benefits and health care costs related to CRT, we modeled the COMPANION trial data to estimate the cost effectiveness of CRT-P and CRT-D for patients living with heart failure.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
35%, an electrocardiographically measured QRS duration of 
120 ms, a PR interval of >150 ms, and a hospitalization for the treatment of heart failure or equivalent in the preceding 12 months. All patients received diuretics (if needed), angiotensin-converting enzyme inhibitors or an angiotensin-receptor antagonist (if tolerated), and a beta-blocker (as tolerated). Using an intention-to-treat methodology, the primary end point was the combined risk of all-cause mortality or first hospitalization analyzed from the time of randomization to the time of first event. The secondary end point was death of any cause. The median follow-up duration was 11.9 months in the OPT group, 16.2 months in the CRT-P group, and 15.7 months in the CRT-D group.
Model structure.
We constructed a model replicating the course of treatment observed in the COMPANION trial, with identical groups of 1,000 patients randomized to CRT-D, CRT-P, or OPT treatment arms. Objectives in modeling were to adjust for differential enrollment and duration of follow-up across treatment groups, to assign costs to resource utilization as documented in the trial, and to extend the period of observation. Survival, costs, and quality of life were projected based on parameters derived from trial data. The model took the economic perspective of the Center for Medicare and Medicaid Services because the majority of patients with heart failure in the U.S. are over the age of 65 years and, therefore, are Medicare beneficiaries (1). The base-case follow-up period was seven years, divided into one-month cycles. Model end points included cost of treatment, survival, and preference-weighted survival, which were used to calculate incremental cost per life-year gained and cost per quality-adjusted life-year (QALY) gained for CRT-D and CRT-P relative to OPT. Cost effectiveness is a ratio calculated as the incremental cost for CRT relative to OPT, divided by the incremental benefit (survival or QALY):
 |
Treatment costs. Initial implantation and subsequent follow-up hospitalizations were documented in case report forms. Initial implantation hospitalizations were associated with initial device implantation or re-attempt for CRT-D and CRT-P trial arm patients. Follow-up hospitalizations were defined as either care provided at a hospital for any reason, including cross-over device implantations, or emergency department admission with use of intravenous inotropes and/or vasoactive drugs for a duration >4 h. An independent adjudication committee categorized each hospital admission. Each of the initial implantation procedures and subsequent follow-up hospitalizations was mapped to an appropriate diagnostic-related group (DRG) for inpatient procedures or to ambulatory payment classification in the case of emergency department admission. Frequency and reason for follow-up hospitalization as documented in the COMPANION trial are shown in Table 1.
|
The cost of initial implantation was calculated separately for the CRT-D and CRT-P groups and factored in the average number of implantation attempts (CRT-D, 1.09; CRT-P, 1.07) and the overall success rate (CRT-D, 90.9%; CRT-P, 87.4%). All attempts and re-attempts were considered a cost. The average follow-up hospitalization cost was calculated separately for the CRT-D, CRT-P, and OPT groups, based on the observed distribution of types of admission (Table 1). Monthly risk for both all-cause and cardiac hospital admission during months 1 through 24 was based on the actual experience of each group in the clinical trial. The hospitalization rate beginning with month 25 was equalized based on observed data for months 19 through 24, pooled across the three groups. Hospitalization rates are shown in Table 2.
|
Cardiac arrhythmia devices are susceptible to battery depletion and associated pulse generator replacement costs. Battery depletion is highly dependent on device settings for individual patients. The duration of the COMPANION trial was not sufficient to observe battery longevity. Therefore, monthly probability of battery depletion was modeled based on manufacturer estimates (Table 2). These estimates incorporated 46 different device settings and had an estimated margin of error of ±10%. The battery depletion function used in this analysis established a mean battery life of 4.5 years for CRT-D and 6.0 years for CRT-P. However, in sensitivity analysis we accelerated the failure rate by approximately 20% (i.e., median battery life decreased by six months). In addition, modeled battery failure is a stochastic function, such that the first failures occur within the first year. This methodology is more conservative than recent cost-effectiveness studies of implantable cardioverter-defibrillators that assumed that all devices fail at a single specified time (e.g., five years).
Because all patients received OPT, costs for prescription medication were assumed to be equivalent across treatment groups and were not included in the analysis. No difference (p < 0.05) in the use of medications commonly used to treat cardiovascular disease was observed across the three treatment groups at the date of last follow-up, based on data from 90% of patients in each group (M.R. Bristow unpublished data, 2004). Regularly scheduled visits with primary care physicians were not documented in the COMPANION trial and, therefore, were not included in the economic analysis.
Survival. The CRT-D group achieved a statistically significant reduction in all-cause mortality (hazard ratio, 0.64; p = 0.003), whereas CRT-P was associated with a nonsignificant reduction in all-cause mortality (hazard ratio, 0.76; p = 0.059) (15). Mortality data from the COMPANION trial were used to estimate exponential survival functions for each treatment group to establish the monthly probability of death (Table 2), the number of patients remaining in each cohort, and cumulative patient-years of survival. Figure 1 shows actual survival and the fitted survival curves for each treatment group.
|
|
In addition to the stochastic sensitivity analysis, a one-way sensitivity analysis was performed to assess the impact of varying selected model parameters while holding all other variables fixed at their base-case values. These parameters included survival benefit of CRT therapy, battery depletion rate over time, Medicare overall facility payment rate (DRG dollar multiplier), Medicare payment for defibrillator system implantation (relative weight of DRG 515), duration of follow-up, and discount rate.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
At the end of seven years, patients in the CRT-D group accumulated an average 4.15 years of survival (discounted at 3%), compared with 3.87 years in the CRT-P group and 3.37 years in the OPT group. After applying preference weights to the period of survival, QALYs were 3.15 for CRT-D, 3.01 for CRT-P, and 2.30 for OPT. Costs and survival at selected time periods are summarized in Table 4.
|
|
|
|
|
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
Our results differ from those of an earlier cost-effectiveness report that analyzed the cost effectiveness of CRT in patients with heart failure (28) based on a meta-analysis of various CRT clinical trials (29). In comparison with our present analysis, the analytical methodology and statistical assumptions in their analysis contained the following differences. Six of the trials in their meta-analysis accumulated data for only three months or less, whereas two studies had a duration of only six months. The COMPANION trial, by contrast, had a treatment duration exceeding one year. Hospitalization data from the COMPANION trial were not available for their meta-analysis. Although improved survival and reduced hospitalization benefits for OPT cross-over patients were incorporated in their analysis, hospitalization costs for acquiring the device were not included because only heart failure hospitalization, not all-cause hospitalization, was evaluated. Additionally, the earlier analysis did not take into account the quality-of-life improvements observed in the COMPANION trial.
Previous analyses of CRT and implantable cardioverter-defibrillator therapy have based the duration of the treatment episode on patient lifetime. Although it is important that the time horizons used in cost-effectiveness analysis be long enough to capture all important therapeutic effects (30), this analysis has remained conservative by limiting the time period evaluated to a seven-year base case. In fact, the value of CRT may well extend beyond this time frame. However, over time, gains in cost effectiveness from amortizing initial device costs are offset by the device replacement costs and discounting.
Although important in assessing the cost effectiveness of new therapies, indirect costs such as productivity loss and caretaker burden were not included in the present analysis. Because the mean age of patients in the COMPANION trial was 67 years, the majority of patients were not in the labor force due to either retirement or health-related disability. If CRT improves functional status and ability to work for people with heart failure or reduces caretaker burden, a societal perspective that accounted for changes in productivity would yield more favorable rather than less favorable cost-effectiveness results.
Another important determinant of cost effectiveness is the Medicare payment rates for CRT-P and CRT-D system implantations. Because these are relatively new therapies, Medicare may adjust the reimbursement rate or structure as more hospital charge data are accumulated. Even with a substantial increase in payment level for this DRG, or a change in DRG assignment, CRT remains within the accepted range for effective therapy.
A conundrum that relates to economic analysis is how to best define societal costs. Whether a Center for Medicare and Medicaid Services payment to hospitals in fact represents true cost is a matter of some debate. The Center for Medicare and Medicaid Services annually revises DRG weights and multipliers in a process that relies heavily on charges and cost-to-charge ratios as reported by hospitals. Patient cost-sharing was also included, which is certainly important to patients but is often ignored. Indeed, analyses using a societal perspective typically rely on Medicare fee schedules to establish true cost for physician payments and are therefore as representative as any methodology for determining costs.
Results of probabilistic sensitivity analysis show that findings based on incremental cost per life-year gained (Fig. 2A) are more robust than the results based on quality-adjusted survival (Fig. 3). The much greater dispersion in the QALY results is attributable to variability in the underlying quality-of-life data (i.e., MLHFQ), as well as the error component in the algorithm used to map the MLHFQ scores to preference weights. Alternatively, in this population we cannot exclude the possibility that increasing survival could in some cases be associated with increased morbidity that could impair perceived quality of life.
Quality-of-life utility values, based on the MLHFQ, improved 29% for CRT-D patients, 27% for CRT-P patients, and 12% for OPT patients from baseline to month 6. In addition to MLHFQ scores, NYHA data were also recorded at three-month intervals throughout the duration of the trial. A published algorithm maps NYHA functional class to health state utilities (21). Changes in utility values from baseline to month 6 based on the MLHFQ were highly correlated with those derived from NYHA functional class. The NYHA functional class improvements measured at month 6 were typically maintained over time. We elected to rely on the MLHFQ because it is a patient-reported multidimensional quality-of-life instrument, whereas NYHA functional class is a physician rating instrument that centers on domains of physical health and functional status.
Study limitations. The COMPANION trial enrollment began in January 2000. In November 2002, the data safety monitoring and review board stopped enrollment and all efficacy follow-ups because the trial had reached the targeted number of events and had met the primary and secondary end points. Therefore, longer-term follow-up data are lacking, and the outcomes for the modeled treatment episodes beyond the observed data points have a higher level of uncertainty. For survival, exponential curves were found to be a good fit for the recorded survival data and were used instead of actual trial data. In the sensitivity analysis we tested the conservative assumption that CRT would provide no incremental survival benefit relative to OPT after two years (Table 6). Although cost effectiveness was reduced in this scenario, the incremental cost per QALY was still within the accepted range for innovative therapy.
Use of outpatient care, except as mandated by protocol, was not documented in the trial. Differences in resource use among the groups may occur in the outpatient setting. Patients with devices may require more visits as the devices are monitored. Conversely, patients without devices may require more visits because of disease progression.
For hospitalization, we used actual rates of admission as recorded during the first 24 months of the trial. This analysis was conservative and based on the assumption that neither CRT-D nor CRT-P offered benefit in the form of reduced hospitalization after month 24. After the initial device implantation hospitalization, patients in the CRT groups experienced relatively fewer admissions than the OPT group over the observed period of time. At 24 months, CRT-D was associated with a 29% ($7,400) reduction in the cost of all-cause follow-up hospitalization, whereas CRT-P showed a 37% ($9,400) reduction relative to the total $25,600 hospitalization cost in the OPT group. When only heart failure admissions were considered, CRT-D was associated with a 22% ($2,100) reduction in the cost of follow-up hospitalization, whereas CRT-P was associated with a 34% ($3,200) reduction versus the OPT group.
The COMPANION trial experienced a number of treatment group cross-overs in which CRT devices were implanted in members of the OPT group. Because our analysis follows the intent-to-treat protocol of the COMPANION study, OPT patients receiving a device remained in the OPT arm. These implantations impacted both costs and survival in the OPT group, because patients incurred the cost for a device implantation hospitalization but then experienced a reduction in follow-up hospitalization coupled with improved survival. We did not specifically project further cross-overs. However, because the modeled survival and hospital admission functions were derived from observed trial data, any cross-overs that occurred during the time period observed were implicitly modeled into the extrapolations. We replicated the analysis for an as-treated subset of the COMPANION trial patients. By this analysis, cost-effectiveness results were similar to those presented using the intention-to-treat analysis.
Some concerns might be raised regarding the interpretation of the cost-effectiveness analysis because the long-term effects of device and/or OPT on quality of life are not defined in this population. This may have limited the analysis of the data, because we cannot ensure that long-term quality of life changed in an equivalent fashion in the two groups. However, although MLHFQ scores were available for only a limited time, data on NYHA functional classification were available for two years or more and remained stable in the CRT groups. Furthermore, preference weights calculated from the MLHFQ were highly correlated with those calculated using NYHA stage. Thus, assuming that utility declines over time because of the progressive nature of heart disease, it is likely to decline at the same rate or at a more rapid rate in OPT patients when compared with those receiving a device. Assuming a decline at a similar rate, the impact on quality-adjusted cost effectiveness would be minimal.
We also cannot exclude the possibility that the cost effectiveness of resynchronization therapy as calculated in the present study, in which all patients received OPT, might differ in a group of patients who were not as effectively managed in terms of pharmacologic therapy. In addition, it is possible that future innovations may more accurately predict which patients will best respond to resynchronization therapy, thus lowering the overall treatment costs.
|
Conclusions |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionConclusionsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Ding, W. Hua, H. Niu, K. Chen, and S. Zhang Primary prevention of sudden cardiac death using implantable cardioverter defibrillators Europace, September 1, 2008; 10(9): 1034 - 1041. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. K H Lam and A. Owen Combined resynchronisation and implantable defibrillator therapy in left ventricular dysfunction: Bayesian network meta-analysis of randomised controlled trials BMJ, November 3, 2007; 335(7626): 925 - 925. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Authors/Task Force Members, P. E. Vardas, A. Auricchio, J.-J. Blanc, J.-C. Daubert, H. Drexler, H. Ector, M. Gasparini, C. Linde, F. B. Morgado, et al. Guidelines for cardiac pacing and cardiac resynchronization therapy: The Task Force for Cardiac Pacing and Cardiac Resynchronization Therapy of the European Society of Cardiology. Developed in Collaboration with the European Heart Rhythm Association Europace, October 1, 2007; 9(10): 959 - 998. [Full Text] [PDF]
|
|
|
|
 |
|
 Authors/Task Force Members, P. E. Vardas, A. Auricchio, J.-J. Blanc, J.-C. Daubert, H. Drexler, H. Ector, M. Gasparini, C. Linde, F. B. Morgado, et al. Guidelines for cardiac pacing and cardiac resynchronization therapy: The Task Force for Cardiac Pacing and Cardiac Resynchronization Therapy of the European Society of Cardiology. Developed in Collaboration with the European Heart Rhythm Association Eur. Heart J., September 2, 2007; 28(18): 2256 - 2295. [Full Text] [PDF]
|
|
|
|
 |
|
 F. A. McAlister, J. Ezekowitz, N. Hooton, B. Vandermeer, C. Spooner, D. M. Dryden, R. L. Page, M. A. Hlatky, and B. H. Rowe Cardiac Resynchronization Therapy for Patients With Left Ventricular Systolic Dysfunction: A Systematic Review JAMA, June 13, 2007; 297(22): 2502 - 2514. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Hlatky and D. B. Mark The high cost of implantable defibrillators Eur. Heart J., February 2, 2007; 28(4): 388 - 391. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Yao, N. Freemantle, M. J. Calvert, S. Bryan, J.-C. Daubert, and J. G.F. Cleland The long-term cost-effectiveness of cardiac resynchronization therapy with or without an implantable cardioverter-defibrillator Eur. Heart J., January 1, 2007; 28(1): 42 - 51. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. Kramer The Expanding Prognostic Role of Late Gadolinium Enhanced Cardiac Magnetic Resonance J. Am. Coll. Cardiol., November 21, 2006; 48(10): 1986 - 1987. [Full Text] [PDF]
|
|
|
|
|
|
H. M. Krumholz and F. A. Masoudi The Year in Epidemiology, Health Services Research, and Outcomes Research J. Am. Coll. Cardiol., November 7, 2006; 48(9): 1886 - 1895. [Full Text] [PDF]
|
|
|
|
|
|
A. Ward, S. Guo, and J. J. Caro Cost-Effectiveness of Cardiac Resynchronization Therapy J. Am. Coll. Cardiol., September 19, 2006; 48(6): 1283 - 1283. [Full Text] [PDF]
|
|
|
|
|
|
P. Kearney, G. Stokoe, G. Breithardt, C. Longson, J. Marco, J. Morgan, S. Priori, A. Ruether, R. Taylor, M. Hertog, et al. Improving patient access to novel medical technologies in Europe Eur. Heart J., April 1, 2006; 27(7): 882 - 885. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O.-A. Breithardt and G. Breithardt Quest for the Best Candidate: How Much Imaging Do We Need Before Prescribing Cardiac Resynchronization Therapy? Circulation, February 21, 2006; 113(7): 926 - 928. [Full Text] [PDF]
|
|
|
|
|
|
M. A. Hlatky Cost Effectiveness of Cardiac Resynchronization Therapy J. Am. Coll. Cardiol., December 20, 2005; 46(12): 2322 - 2324. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
