Author + information
- Received October 8, 2012
- Revision received December 12, 2012
- Accepted December 17, 2012
- Published online April 2, 2013.
- ↵⁎Reprint requests and correspondence:
Dr. Gregg C. Fonarow, Ahmanson-UCLA Cardiomyopathy Center, 10833 LeConte Avenue, Room A2-123 CHS, Los Angeles, California 90095-1679
Objectives This study sought to quantify the incremental cost-effectiveness ratios (ICER) of angiotensin-converting enzyme inhibitor (ACEI), beta-blocker (BB), and aldosterone antagonist (AldA) therapies for patients with heart failure with reduced ejection fraction (HFrEF).
Background There are evidence-based, guideline-directed medical therapies for patients with HFrEF, but the incremental cost-effectiveness of these therapies has not been well studied using contemporary data.
Methods A Markov model with lifetime horizon and two states, dead or alive, was created. We compared HFrEF patients treated with diuretic agents alone to three treatment arms: 1) ACEI therapy alone; 2) ACEI+BB; and 3) ACEI+BB+AldA. Sequential therapy was also analyzed. HF hospitalizations and mortality rates were based on representative studies. Costs of medications and inpatient and outpatient care were accounted for.
Results Treatment with ACEI and ACEI+BB strictly dominated treatment with diuretics only (cost-saving). The greatest gains in quality-adjusted life-years occurred when all 3 guideline-directed medications were provided. The incremental cost-effectiveness ratio (ICER) of ACEI+BB+AldA versus ACEI+BB and ACEI+BB versus ACEI was <$1,500 per quality-adjusted life-year. The cost-savings in the ACEI and ACEI+BB cohorts compared to that with diuretics alone were $444 and $33, respectively. Assuming lower treatment costs and lower hospitalization rates in the ACEI+BB+AldA arm resulted in greater cost-savings. Even in the most unfavorable situations, the ICER was <$10,000 per life-year gained.
Conclusions Our analysis demonstrates that medical treatment of HFrEF is highly cost-effective and may even result in cost-savings. Greater efforts to ensure optimal adherence to guideline-directed medical therapy for HFrEF are warranted.
Heart failure (HF) remains one of the leading causes of mortality, morbidity, and health care-associated costs worldwide. Approximately 6 million Americans have HF, and the lifetime risk of developing HF is 1 in 5 (1,2). After a HF hospitalization, 5-year mortality for HF is over 40% (3,4). There are also substantial costs associated with HF, including over 1 million HF admissions a year and $39.2 billion a year from direct and indirect costs (5).
Over the past 2 decades, there have been remarkable advances in medical therapy for HF with reduced ejection fraction (HFrEF). The use of angiotensin-converting enzyme inhibitors (ACEI), beta-blockers (BB), and aldosterone antagonists (AldA) has resulted in substantial reductions in mortality, morbidity, and hospitalizations in HFrEF patients (6–8). Previous studies conducted when these medications were available only as branded therapies have shown that the cost-effectiveness of these therapies in patients with mild to moderate HFrEF were $100 to $25,000 per quality-adjusted life-year (QALY) compared to that with conventional treatment (9–13).
However, now that ACEI, BB, and AldA have generic formulations available, we hypothesized that guideline-directed medical therapies for HFrEF will be of even greater value and possibly cost-saving. In addition, the incremental cost-effectiveness for each therapy has not been previously evaluated using contemporary data, including costs associated with generic formulations and use of AldA plus BB plus ACEI (AldA+BB+ACEI) therapy in chronic HFrEF. Our objective was to quantify the incremental cost-effectiveness of ACEI, BB, and AldA therapies and the cost-effectiveness of these therapies compared to that with background diuretic therapy in patients with mild to moderate HFrEF.
Markov models are commonly used to understand cost-effectiveness of therapies in chronic diseases. We created a Markov model to simulate costs, QALYs, and incremental cost-effectiveness of patients with HFrEF receiving ACEI, BB, and AldA in a cohort of hypothetical patients with mild to moderate (New York Heart Association [NYHA] class II or III) HF. Our model had a lifetime horizon, was U.S. based, and consisted of two states: dead or alive. We attempted to take the perspective of a lifetime single-payer/vertically integrated healthcare system providing full coverage for hospitalizations, office visits, laboratory tests, medications, and resources to the extent possible with available cost data. The length of follow-up in the representative studies of medical treatment for HF ranged from 12 to 41 months. We conservatively assumed that benefits for the medical therapies lasted 2 years but that the costs of treatment continued indefinitely. From year 3 until death, the cohort reverted to having the same mortality and hospitalization rates as the cohort receiving only diuretic agents. We factored in the additional mortality seen with increasing age based on U.S. lifetime tables (14). All costs and QALYs were discounted at 3% and were in 2012 U.S. dollars.
The placebo arm of the SOLVD (Study of Left Ventricular Dysfunction), a large trial that has been used as the basis for disease estimates by other studies, served as the base (diuretic-only cohort) population in this study (6). Patients included in this trial were HF patients with ejection fractions of 35% or less and were treated with oral loop diuretics. The mean age of patients in this study was 61 years of age, and 90% were in NYHA functional class II or III. This group of patients represents the overall population of patients in our cohort, defining “baseline” risk in this study. We compared our baseline cohort (treatment with diuretic-only cohort) to those in three treatment arms: patients receiving: 1) ACEI; 2) ACEI+BB; or 3) ACEI+BB+AldA therapy. We also compared each incremental therapy to the previous therapy: 1) ACEI+BB versus ACEI; and 2) ACEI+BB+AldA versus ACEI+BB therapy.
Probabilities of hospitalization for the baseline and ACEI cohorts were derived from the SOLVD, as that study was the most representative of outcomes for patients in our diuretic-only and ACEI cohorts (6). Total number of hospitalizations for HF at the end of the trial was divided by the number of people in each arm and the average follow-up (41 months) to estimate the yearly probability of hospitalization per person. The difference in deaths or hospitalizations between the 2 groups did not converge at the end of the trial (41 months), but we conservatively assumed that this decrease in hospitalization was only in the first 2 years of our model.
We derived the yearly probability of HF hospitalization for the ACEI+BB group by multiplying the yearly probability of HF hospitalization for the ACEI group by “1 − x, x = the relative risk reduction for HF hospitalization” in the MERIT-CHF (Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure) study (relative risk reduction: 31.6%), a study representative of the effects of BB+ACEI on HFrEF (7). We then derived the yearly probability of HF hospitalizations for the ACEI+BB+AldA cohort from the EMPHASIS-HF (Eplerenone in Mild Patients Hospitalization and Survival Study in Heart Failure) trial, a study representative of the effects of ACEI+BB+AldA on HFrEF (8). That study showed a hazard ratio of 58% for HF hospitalizations for patients taking ACEI+BB+AldA compared to those taking ACEI+BB. The hazard ratio from this study was calculated with a median follow-up of 21 months (we assumed this was equivalent to 24 months), so we calculated the yearly hazard ratio by taking the square root of the 2-year hazard ratio (i.e., = 1 year hazard ratio, x = 2 year hazard ratio). We then multiplied the yearly hazard ratio by the probability of HF hospitalizations in the ACEI+BB cohort. In both the EMPHASIS-HF and MERIT-HF trials, the event curves for deaths and HF hospitalizations continued to diverge at the end of follow-up (average of 21 and 12 months, respectively) (Table 1).
The costs taken into account in our model included HF-related hospitalization, medications and medication monitoring, and ambulatory care (Table 2). The average cost of one inpatient hospitalization, based on the report by Delea et al. (15), was multiplied by the probability of hospitalization each year (as derived above) to obtain the average yearly cost attributed to inpatient hospitalization per patient. The probability of hospitalization accounted for some patients requiring multiple hospitalizations in 1 year. We conservatively assumed that the benefits for the medical therapies lasted 2 years but that the costs of treatment continued until death. The cost of ambulatory care was based on the ambulatory care in the SOLVD trial (9). The costs of medications were based on the prices of 30-day prescriptions for generic medications (16). For base our models, we excluded the costs of non-HF hospitalizations. Based on our clinical experience, we assumed that the ACEI and ACEI+BB cohorts required four basic metabolic panel tests and that the ACEI+BB+AldA cohort required six basic metabolic panel tests in the first year. We assumed that our base cohort and all cohorts after year 2 required three metabolic panels a year because of diuretic use. The cost for a basic metabolic panel was based on average Centers for Medicare and Medicaid Services (CMS) reimbursement (17). All costs were in 2012 dollars and calculated using a consumer price index calculator (18). The overall average cost for 1 year in each cohort was then multiplied by the percentage of people alive in each cohort to obtain a weighted average cost for the subjects alive in each cohort (Table 2).
We calculated the annual probability of death for the baseline and ACEI cohort by calculating the percentage of subjects alive at the end of the SOLVD trial and then derived the annual probability of surviving (x = probability of surviving to 4 years, 4 = probability of surviving to 1 year) (6). We assumed the probability of death for the ACEI cohort was constant between years 1 and 2 of the model. The probability of death in the ACEI+BB cohort was derived from the MERIT-HF study, which showed a 32% relative risk reduction in death (7). We multiplied the annual probability of death in the ACEI group by 1 − x, x = relative risk reduction of mortality from the MERIT-CHF study to derive annual probability of death in the ACEI+BB cohort. We derived the annual probability of death for the ACEI+BB+AldA cohort from the EMPHASIS-HF trial. That study showed a hazard ratio of 0.76 for patients receiving ACEI+BB+AldA therapy compared to those receiving ACEI+BB therapy (8). The hazard ratio from this study was calculated with a median follow-up of 21 months (we assumed this was equivalent to 24 months), so we calculated the yearly hazard ratio by taking the square root of the 2-year hazard ratio (i.e., = 1 year hazard ratio, x = 2 year hazard ratio). We then multiplied the yearly hazard ratio by the annual probability of death in the ACEI+BB cohort. The survival curves in both the EMPHASIS-HF and MERIT-HF trials continued to diverge at the end of trials, so we assumed that the mortality rates were constant between years 1 and 2 of the model. In all three treatment arms, we assumed that in year 3 and beyond, mortality rates were equal to those in the diuretic-alone arm (Table 1).
We assumed that quality of life did not differ among the cohorts, likely underestimating the positive impact of HFrEF medical therapy. We assumed that the average health utility was 0.78 based on a literature search of QALY studies for patients with mild to moderate HF (19). We multiplied this health utility by the percentage of patients alive in each cohort to derive the average QALYs for the cohort in each year of our model.
We performed a sensitivity analysis for ACEI+BB+AldA cohort compared to the diuretic and ACEI+BB cohorts. We extended the mortality and hospitalization risk reductions in the treatment group to 5 and 15 years. We also examined the impact of assuming the total cost of care for the baseline group was 25% lower and higher, the probability of death in the treatment group was 25% lower and higher, the probability of death in the treatment group was 25% lower and 50% higher, the probability of hospitalization in the treatment group was 50% lower and higher than the base case, and included non-HF hospitalization costs. Annual non-HF hospitalization costs were derived from the SOLVD trial (9). We assumed the annual non-HF hospitalization costs, $2,650 per year in 2012 dollars, remained constant over time. All life-years and costs were discounted at 3%, and costs were in 2012 U.S. dollars. All analyses were performed using Excel software (Microsoft, Redmond, Washington).
Outcomes are reported in Table 3. Discounted total life-years (and QALYs) in the diuretic-only cohort were 4.45 years (3.47 QALYs). In the ACEI-treated cohort, the total life-years (and QALYs) were 4.60 years (3.59 QALYs); in the ACEI+BB-treated cohort, they were 4.91 years (3.83 QALYs); and in the ACEI+BB+AldA cohort, they were 5.01 years (3.90 QALYs).
Total costs are reported in Table 3. Over their lifetime, patients in the diuretic-alone cohort accumulated $12,742 (discounted at 3% per year) in healthcare costs. Compared to the diuretic-treated cohort, cost savings in those in the ACEI cohort were $444 and $33 in the ACEI+BB cohort. In the ACEI+BB+AldA cohort, there was $47 in additional spending compared to the diuretics-only cohort. There were no incremental cost-savings between the ACEI+BB versus the ACEI cohort or between the ACEI+BB+AldA versus ACEI +BB cohort.
The ACEI and ACEI+BB treatments had lower costs and higher QALYs than the diuretic-only cohort. The ICER of ACEI+BB+AldA versus ACEI+BB was $501 and $34 per life-year for the ACEI+BB+AldA versus ACEI+BB cohort.
In the first 5 years of the model, the number of HF hospitalizations per 100 people was 110 for the base cohort, 97 for the ACEI cohort, 88 for the ACEI+BB cohort, and 82 for the ACEI+BB+AldA cohort.
We performed a sensitivity analysis of the ACEI+BB+AldA cohort (Table 4). We first compared the ACEI+BB+AldA with the diuretic-only cohort. We found that when we extended the reduction in mortality and hospitalization in the treatment group to 5 years, the ACEI+BB+AldA cohort experienced better outcomes at lower health care costs (strictly dominated the diuretic-only cohort), and there was $356 in cost-savings; when benefits were extended to 15 years, ACEI+BB+AldA therapy no longer provided a cost-savings but was highly cost-effective with an ICER of $623 per life-year ($799 per QALY). When the total cost in the ACEI+BB+AldA group in the first 2 years was 25% higher, the ICER was $1,363 per life-year ($1,748 per QALY) compared to that in the diuretic-only cohort; when the total cost was 25% lower in the first 2 years, the treatment arm strictly dominated the diuretic-only cohort, and there were cost-savings of $725. When probability of death in the ACEI+BB+AldA cohort was 25% lower in the first 2 years, the ICER was $365 per life-year ($468 per QALY) compared to that in the diuretic-only cohort; when the probability of death was 50% higher in the first 2 years, the treatment arm strictly dominated the diuretic-only cohort with a cost-savings of $738. When the probability of hospitalization in the ACEI+BB+AldA cohort was 50% higher in the first 2 years, the ICER was $1,411 per life-year ($1,809 per QALY) compared to the diuretic-only cohort; when the probability of hospitalization was 50% lower in the first 2 years, the treatment arm strictly dominated the diuretic-only cohort, with a cost-savings of $756. When the cost of non-HF hospitalizations was included, the ICER of ACEI+BB+AldA versus diuretic-only cohort was $2,677 per life-year ($3,431 per QALY). An additional sensitivity analysis for ACEI+BB+AldA versus ACEI+BB is presented in Table 4.
The goal of this study was to quantify incremental cost-effectiveness of guideline-directed medical therapies for HFrEF. Our study demonstrates that not only is medical therapy of HFrEF with ACEI+BB+AldA cost-effective, but also that treatment with ACEI and BB is cost-saving compared to treatment with diuretics alone. Previous cost-effectiveness analyses showed that medical therapy in patients with mild or moderate HFrEF was very cost-effective ($100 to $25,000 per life-year) or cost-neutral (9–13). To the best of our knowledge, this is the first lifetime cost-effectiveness analysis to show cost-savings with medical therapy in patients with mild to moderate HFrEF. Only one cost-effectiveness analysis in patients with severe HF showed cost-savings in patients taking AldA+ACEI versus those taking ACEI (20).
Costs in our model were sensitive to effects of probability of death, hospitalization, total costs, and length of HF treatment's risk reduction on mortality and hospitalization. In the treatment groups, net cost was a balance between increased costs from more accrued costs from additional life-years lived and decreased costs from reduced hospitalization. The ACEI cohort and the cohort in the sensitivity analysis with higher mortality rates had more cost-savings than the base case ACEI+BB+AldA cohort compared to the diuretics-alone cohort because of the costs associated with additional life-years lived.
Many of our assumptions were conservative with regard to the impact of medical therapy on HF. We may have underestimated the positive impact on QALYs because we were unable to factor in the quality-of-life benefit of HF treatment associated with decreased hospitalizations and increased functional status. However, there may have been some adverse effects of the treatments on quality of life that we did not capture. We were unable to quantify long-term gains from HF treatment because there are no long-term studies of mortality or hospitalization reductions of HF treatment. We also did not factor in the reduction in indirect costs associated with decreased mortality and morbidity with medical therapy; the lost productivity due to early mortality of HF is estimated to be above 4 billion dollars a year in the United States (5). We also assumed that there was complete adherence to HF medical therapies in the representative studies. However, even large randomized studies have had up to 25% nonadherence rates to HF medical therapies (21). This suggests that the treatment effects of guideline-directed HF medical therapies could potentially be even greater than that reported in clinical trials with more effective strategies to ensure adherence.
Even in the most unfavorable situation, cost-effectiveness ratios for each of the guideline-directed mediations for HFrEF compare very favorably to estimates for other generally accepted medical therapies, such as dialysis. For example, some believe that Medicare coverage of patients with chronic renal failure requiring dialysis, which has an estimated cost-effectiveness of $50,000 or more per QALY, implies a societal judgment that treatments with a lower cost/QALY ratio are cost-effective and should be provided (22).
This low cost of treatment and possible cost-savings is an important finding because it suggests that further investment should be made to ensure systems such as HF disease management programs are in place to optimally implement guideline-directed medical therapies and facilitate patients adhering to the prescribed regimen (23–25). If a hypothetical program was 100% effective at ensuring all guideline-appropriate patients were receiving ACEI+BB+AldA, $13,986 could be spent per patient on such a program and maintain an ICER of $25,000 per life-year for ACEI+BB+AldA therapy versus diuretics alone. If this program was only 20% effective, $2,786 could still be spent per patient and maintain an ICER of $25,000 per life-year. This suggests that significant investments in HF treatment can be cost-effective. One such program is Get With The Guidelines, which has shown increased adherence to guideline-directed medical therapies through education of hospital staff, real-time feedback to hospitals, and performance recognition (26,27).
The direct costs of medical treatment of HF have been significantly reduced since ACEI, BB, and AldA all have generic versions; this in itself may increase adherence. Studies have shown that even a $10 increase in copayments for BB resulted in almost a 2% decrease in adherence and a 9% increase in risk of hospitalization; similar results were also found with price changes of ACEI (28). The MI FREEE (Post-Myocardial Infarction Free Rx Event and Economic Evaluation) trial demonstrated that the elimination of copayments for medications prescribed after myocardial infarction improved medication adherences, reduced rates of first major vascular events, and decreased patient spending without increasing overall health costs (29). Given the very high healthcare value provided by guideline-directed medical therapies for HFrEF, eliminating all patient costs for these medications or even providing financial incentives to promote adherence is likely to be advantageous to patients, healthcare delivery systems, and society, although these benefits may not be equally shared.
Limitations exist with any modeling process that simplifies reality by assumptions. We assumed that the therapies reduce rates of mortality and hospitalization by the same proportion as observed in clinical trials. This assumption about efficacy is commonly made in cost-effectiveness studies. We were also unable to model the increasing costs at the end of life, but assumed that the cost at the end of life in the treatment cohorts would be the same as the diuretics alone cohort. The costs used in this study are only estimates of true costs. While hospitalization costs were derived using charges and cost-to-charge ratios, this approach may not accurately reflect true costs. Further, this analysis integrated different perspectives in determining costs/cost-savings and did not account for indirect costs. We also did not include the cost of non-HF hospitalizations in our base-case analysis because we considered this cost to be unrelated to the interventions studied. We adopted the view of Garber and Phelps and Lee that future costs unrelated to the interventions studied can be excluded from cost-effectiveness models (30,31). There was no change in non-HF hospitalization rates with medical treatment of HFrEF in the SOLVD, MERIT-HF, EMPHASIS-HF, or other trial, to our knowledge (6–8). Nevertheless, we included the cost of non-HF hospitalizations in our sensitivity analysis and found that HF treatment was still highly cost-effective.
At the same time, we undertook several conservative measures that strengthen the confidence in our findings. We did not include quality of life in effectiveness, even though quality of life seems to benefit from these medications. Also, the costs of outpatient HF care are assumed to be constant. One difficulty in estimating incremental treatment effectiveness and population risk is that “background HF care,” (i.e., care other than ACEI, BB, and AldA therapy) is often poorly defined in available trials. We chose background care to be the placebo arm in the SOLVD trial. Since its publication more than a decade ago, some may argue that background HF care (e.g., use of diuretics, digoxin, diet) has changed, which may decrease the rate of hospitalizations. Despite this limitation, the SOLVD trial remains the best source of data on the long-term risks of hospitalization and death for HFrEF patients.
This analysis offers broad insight into the incremental cost-effectiveness of guideline-directed medical therapies for HFrEF. Use of ACEI, ACEI+BB, and ACEI+BB+AldA therapies were associated with significant health gains and were cost-saving or highly cost-effective, with greatest gains in QALY occurring when all three medications were provided. Even under the most unfavorable assumptions, these guideline-directed medical therapies remained highly cost-effective with cost per QALY <$10,000. Our study also suggests that $3,000 to $14,000 per patient could be spent to ensure adherence to these therapies and still maintain an attractive ICER of $25,000 per life-year gained. Further resources should be allocated to ensure full adherence to guideline-directed medical therapies for HFrEF to improve outcomes, provide high-value care, and minimize health care costs.
Dr. Fonarow holds the Eliot Corday Chair in Cardiovascular Medicine and Science and is supported by the Ahmanson Foundation; has research funding from the Agency for Healthcare Research and Quality (AHRQ); and is a consultant with Medtronic, Novartis, and Gambro. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- angiotensin-converting enzyme inhibitor
- aldosterone antagonists
- heart failure
- heart failure with reduced ejection fraction
- incremental cost-effectiveness ratio
- quality-adjusted life-year
- Received October 8, 2012.
- Revision received December 12, 2012.
- Accepted December 17, 2012.
- American College of Cardiology Foundation
- Lloyd-Jones D.M.,
- Larson M.G.,
- Leip E.P.,
- et al.
- Lloyd-Jones D.,
- Adams R.J.,
- Brown T.M.,
- et al.
- Hjalmarson A.,
- Goldstein S.,
- Fagerberg B.,
- et al.
- Arias E.
- Walmart [website]
- Centers for Medicare and Medicaid Services [website]
- Bureau of Labor Statistics, US Department of Labor
- Hirth R.A.,
- Chernew M.E.,
- Miller E.,
- Fendrick A.M.,
- Weissert W.G.
- Astrup A.,
- Dyerberg J.,
- Elwood P.,
- et al.