The management of cardiovascular diseases, especially coronary artery disease, has been revolutionized after Dr. Mason Sones performed the first coronary angiography in 1960 and Dr. Andreas Gruentzig pioneered percutaneous coronary intervention in 1977. Since 1977, percutaneous techniques for coronary angioplasty have significantly evolved. Interventional cardiology has emerged as a subspecialty of cardiology with tremendous potential of impacting patient outcomes in a broad spectrum of cardiovascular diseases including, but not limited to, acute coronary syndromes, stable coronary artery disease, heart failure syndromes, and valvular heart diseases. From the development of ever-improving coronary interventional equipements and techniques to facilitate treatment of the most complex coronary disease to the introduction of catheter-based therapy for nonoperable aortic stenosis, interventional cardiology has seen a major transformation in therapeutic possibilities over the last decade, and the continuously expanding horizon of interventional cardiology well demonstrates its potential beyond just coronary intervention as first witnessed in 1977. Nonetheless, selecting a specific therapy and offering it to the right patient become challenging in our society of limited resources. Careful evaluation of available evidence becomes vital in determining the appropriateness of a specific intervention, considering both clinical efficacy and cost-effectiveness. Cost-effectiveness of the most significant therapeutic developments in interventional cardiology over the last 2 decades is reviewed in this chapter. A basic approach in cost-effectiveness analysis is briefly reviewed here.
ICER is a widely utilized and accepted standard to evaluate cost-effectiveness of a particular intervention, and the majority of the studies that are discussed in this chapter have used an ICER as the fundamental metric. The ICER is defined as the incremental cost of providing a specific intervention or therapy divided by the incremental gain in the health benefit.1 Thus, any new intervention will be compared to a previous type of care, often the current standard of care. Often the ICER is compared to a standard willingness-to-pay threshold, eg, $50,000 per life-year gained. Values of the ICER below the threshold, also known as the ceiling ratio, would be considered a socially acceptable healthcare expenditure to gain 1 year of life. Thus, in principal, any service that falls below the threshold would be considered cost effective. However, there is no socially acceptable threshold, there are uncertainties in the measurement of cost and efficacy, and society will often spend more than the threshold for patients in acute settings (rule of rescue). This has limited the use of cost-effectiveness analysis for determining policy. However, cost-effectiveness can expose the assumptions underlying our decisions, which can add setting policy.2
Utility is a measure of overall health and functioning, generally scaled from 0 to 1.3 Thus, for a patient with coronary artery disease suffering from angina, a utility of 0 would be death and 1 would be perfect health and functioning without chest pain. Utility integrates multiple measures of health status (eg, functional status, symptoms) into a single number (utility or quality of one’s life). However, utility is difficult to measure. It may be estimated from cumbersome trade-off methods such as standard gamble or time trade-off,4,5 but survey methods such as the health utility index or EQ-5D tend to be more universally used although less exact.6–8
Survival time alone, in general, is not considered a satisfactory measure of an outcome in a patient with a utility of <1.3 Utility may be combined with survival to give QALY, an overall measure that incorporates survival and health status. Utility assessment provides the strongest base to estimate quality-adjusted survival.9 Incremental gain in QALYs is often used as the measure of overall health benefit, describing not just survival but also quality of that survival, and is often used as a more salient denominator in willingness-to-pay thresholds. Estimation of QALY can be explained with an example of severe angina. Let us consider that living 5 years of life with severe angina is equivalent to 4 years spent in good health. Here the quality adjustment would be 0.8 (4 years in good health/5 years with severe angina); 1 year with severe angina would be 0.8 QALY. Using this equation, 10 years of life with severe angina would be equivalent to 8 QALYs free from angina.
Time plays an important role in cost-effectiveness analysis. The time horizon of the analysis (determined by the analyst) is the time frame over which the effect of an intervention on cost and health is evaluated. Ideally, time horizon should cover the entire period over which the intervention may have an effect. Such time periods often span across a patient’s lifetime where the intervention affects mortality. Therefore, the selection between lifetime horizon and time horizon based on study follow-up significantly affects the analysis. Life expectancy over a fixed time horizon is similar to the area under the survival curve.3 For a specific group of patients, for example, surgery may reduce mortality from 3% to 1%, with additional perioperative mortality of 3%. On the other hand, medical therapy for the same group of patients may reduce mortality from 3% to only 2% with no upfront added mortality. Here estimation of life expectancy, with either medical or surgical therapy, as a function of time horizon (lifetime vs specific cutoff used in the study) would reveal that medical therapy is superior in short-term (in trial) follow-up, but surgical therapy provides a long-term survival benefit.3
Selection of the “perspective” is crucial in cost-effectiveness analysis because societal, payer, and patient perspectives affect specific costs and effects included in the analysis.3 Societal perspective is the most commonly utilized perspective in cost-effective analysis (CEA). Societal CEAs are often more comprehensive than those performed based on patient or payer perspectives.10 From a societal perspective, cost is not the same as price; cost is the value of the next best use of the resource.3 In determining costing, the societal perspective incorporates all resource costs associated with the use of an intervention including physician time, other healthcare resources, and the use of nonhealthcare resources such as caregiver time.11 Costing methodology also often involves indirect costs, future costs, and the valuation of costs. Indirect costs distinguishes “direct healthcare costs” from “direct nonhealthcare costs” (such as child care costs for a parent undergoing treatment, costs of transportation, and cost of the time a family member spends caring for a disabled relative).11
QALYs, cost, and the ICER are considered vital measures for CEA.3 The duration of many clinical trials is usually not long enough for an accurate estimation of the long-term course of a clinical disease. This limits the accurate estimation of the changes in life-years and QALYs with a specific therapy and thus incremental gain in life-years or QALYs. CEA based on in-trial data, in terms of cost per life-year or QALY gained, has limited utility for economic decision making due to short-term follow-up in clinical trials.3 Therefore, long-term projections of short-term results are often required to estimate cost per life-year or QALY gained. This can be done by developing a model based on short-term event rates for projecting an average life expectancy and lifetime cost of therapies to derive lifetime ICER or lifetime cost per QALY gained.3 This can be done by utilizing a Markov decision analytic model.3 The Markov model is particularly utilized when the disease process involves risk that is continuous over time, when the timing of the event is important, and when the important events (such as myocardial infarction [MI]) occur repeatedly over time.12 A Markov model is defined by a set of mutually exclusive and exhaustive health states. At any point in time, a person can reside in only 1 health state and for fixed increments of time (known as Markov cycle length). People are assumed to transition from one health state to another depending on a set of transition probabilities. Values are assigned to individual heath states, which represent the cost and utility of spending 1 Markov cycle length in that health state. Such values, when combined with time spent in individual health states, help derive estimates of average cost and effect to calculate long-term (beyond trial period) cost-effectiveness of different treatments.12,13 A comprehensive review of Markov modeling is beyond the scope of this chapter. However, it can be simplified by CEA of the PLATO (Platelet Inhibition and Patient Outcomes) trial.
The PLATO trial compared ticagrelor and clopidogrel for prevention of cardiovascular events in 18,624 patients presenting with acute coronary syndrome over 12 months.14 The composite of death from vascular cause, MI, or stroke occurred significantly less frequently in patients receiving ticagrelor compared to clopidogrel (9.8% vs 11.7%, P <.0001). Ticagrelor was also associated with significant reduction in MI alone (5.8% vs 6.9%, P = .005) and death from any cause (4.5% vs 5.9%, P <.001). Rate of major bleeding was comparable between the ticagrelor and clopidogrel groups (11.6% vs 11.2%, P = .43).14 CEA was performed using PLATO data in the Swedish health setting. CEA based on in-trial data, performed over a 12-month period, demonstrated that the total cost was €96 higher in the ticagrelor group than the clopidogrel group.15 QALYs, estimated based on EQ-5D, were 0.0006 higher in ticagrelor-treated patients compared to clopidogrel-treated patients. For the in-trial period of 12 months, ICER was €160,000/QALY gained with ticagrelor (much higher than the willingness-to-pay threshold of €20,000/QALY gained).15 Lifetime estimates of CEA, QALY, and cost were created using a Markov model based on the 1-year decision tree from PLATO. The initial start state in the long-term Markov extrapolation model was either “post MI,” “post stroke,” “no event,” or “dead” (based on the prespecified end point in PLATO). For patients with “no event” in the 12-month period, the annual risk of mortality was estimated based on the age-specific mortality rates from Swedish life tables, and the annual risk of nonfatal MI and nonfatal stroke was estimated based on the observed hazard function of clopidogrel-treated patients.15 Survival after nonfatal events was modeled by estimating the hazard ratio considering the increased hazard for death after MI or stroke relative to the standard risk of death from life tables.15 Annual costs associated with the “no event” state and costs after “nonfatal events” were estimated in the Markov model. Cumulative long-term cost was estimated based on the cost in each state of the Markov model. Long-term QALYs were estimated based on age adjustment and by applying decrements with each nonfatal MI and stroke. Lifetime CEA for acute coronary syndrome patients based on Markov model demonstrated an ICER of €2372/life-year gained and €2753/QALY gained, both well within the willingness-to-pay threshold.15 Therefore, this lifetime CEA demonstrates the limitation of the cost analysis based on in-trial data due to the trial’s short-term follow-up.
The cost-effectiveness plane is a 2-dimensional display of incremental effectiveness versus incremental cost.16 Thus, points on the plane represent a graphical display of the ICER. As described in Figure 72-1, it is divided into 4 quadrants. Quadrant B is the most cost effective since it introduces more effective newer treatment strategies at less cost than control treatment option. Conversely in quadrant D, the new therapy is less effective and costs more; it does not make sense from societal perspective to adopt treatment with an ICER in quadrant D. Quadrants A and C both introduce subjectivity in the interpretation of ICER, as the new therapy offers increased effectiveness at increased cost (or decreased effectiveness at lower cost).
FIGURE 72-1
Cost-effectiveness plane. Quadrant B demonstrates that the new treatment is dominant and should be accepted. In quadrant D, new treatment is dominated by control and should not be adopted. In quadrants A and C, adoption of new treatment is reasonable. Incremental cost-effectiveness ratio (ICER) for new therapy should be interpreted against socially acceptable threshold. Occasionally, ICER of new therapy may appear attractive for certain subgroup of population. CER, cost-effectiveness ratio.
There is uncertainty in the estimation of both effectiveness and cost. CEA must take this into account. Where patient level data from clinical trials are available, the error in calculation of the ICER due to the play of chance (stochastic error) can be evaluated. Thus, using available patient-level data from clinical trials, the 95% confidence intervals (CIs) of both cost and effectiveness can be estimated. The most often used methodology is bootstrap analysis where an empiric estimate of a sampling distribution is made by drawing a large number of samples with replacements from the original data.17 The bootstrap approach has been considered the most appropriate general method for comparing treatment costs, since it does not make any distributional assumptions in comparing differences in costs and QALYs (or other outcome measures) between the 2 treatment groups. Each estimate of cost and effectiveness derived from the dual bootstrap may then be used to form an estimation of the ICER.
The points from the conjoint bootstrap analysis may then be displayed in the cost-effectiveness plane, offering a graphical representation of the distribution of the ICER.18,19 Alternatively, uncertainty due to the play of chance may also be displayed as a cost-effectiveness acceptability curve, where for any willingness-to-pay threshold the probability of the new therapy being cost effective is plotted.3 The x-axis is the willingness-to-pay threshold, and the y-axis is the probability of being cost effective. The points on the curve represent the fraction of the ICERs below any threshold. The cost-effectiveness acceptability curve and the display of the distribution of the ICERs in the cost-effectiveness plane are complementary representations of the same data. The ICER distribution on the cost-effectiveness acceptability curve for ticagrelor in acute coronary syndrome (ACS) from the PLATO trial is demonstrated in Figure 72-2. The probability of ICER for ticagrelor compared to clopidogrel being <20,000/QALY gained approached 100% for a wide range of ACS presentations, making ticagrelor a dominant strategy compared with clopidogrel.15
FIGURE 72-2
Cost-effectiveness acceptability curves for ticagrelor in acute coronary syndrome (ACS; based on PLATO trial data). Cost-effectiveness acceptability curve for ticagrelor compared to clopidogrel in PLATO trial demonstrated ticagrelor to be a dominant antiplatelet agent with a high probability for incremental cost-effectiveness ratio (ICER) well under the willingness to pay threshold in a wide spectrum of patients with ACS. NSTEMI, non–ST-segment elevation myocardial infarction; STEMI, ST-segment elevation myocardial infarction; QALY, quality-adjusted life-years. (Reproduced from Nikolic E, Janzon M, Hauch O, Wallentin L, Henriksson M. Cost-effectiveness of treating acute coronary syndrome patients with ticagrelor for 12 months: results from the PLATO study. Eur Heart J. 2013;34(3):220-228, by permission from Oxford University Press.)
Bootstrap analysis, however, will only assess uncertainty due to the play of chance. There may also be bias, where variables used to determine the ICER have additional uncertainty. This may be assessed by sensitivity analysis in addition to the bootstrap analysis.20 When patient-level data are not available, such as in Markov model simulations, then sensitivity analysis is the only method available to consider the distribution of the ICER. Parameters used in estimating both effectiveness and cost may be varied within limits, often established from the literature. The validity of assumptions from which parameters for CEA are derived is very important. Sensitivity analysis is performed using different assumptions (reasonable variations in the parameters used in CEA) to derive cost-effectiveness estimates. If these estimates are not significantly different, it underscores the validity of CEA.3 This concept can be further understood with an example of the CEA in the PLATO trial. In the analysis discussed earlier, cost of ticagrelor was considered to be €2.21 per patient per day (as reimbursed in Sweden).15 The lifetime CEA, based on Markov model, demonstrated ICER of €2372/life-year gained and €2753/QALY gained.15 If the cost of ticagrelor is considered €3.00 per patient per day, ICER per QALY gained was €4874 (still well below the willingness-to-pay threshold).15 If the cost and QALY estimates are considered equal for clopidogrel- and ticagrelor-treated groups, cost per QALY was €5204.15 Here the cost-effectiveness results are only driven by the differences in the clinical events, as demonstrated in the PLATO trial. Similar results were observed with variations in the parameters used for the long-term Markov model and different subgroups such as patients with ST-segment elevation MI (STEMI), non–ST-segment elevation MI (NSTEMI), unstable angina (UA), and diabetes.15 Thus, the sensitivity analysis demonstrates validity of various assumptions used in the CEA by consistently demonstrating ticagrelor to be a dominant strategy compared to clopidogrel for patients presenting with ACS.
Short-term trial data are often extrapolated to lifetime estimates to arrive at cost per life-year gained or cost per QALY gained. These lifetime estimates for new treatment are often compared against other conventionally accepted treatment strategies. Such lifetime estimates have important implications on policy setting. Interpretation and generalization from clinical trials with few years of follow-up to lifetime outcomes should be done with caution. Survival, utility, cost structure, practice patterns, and resource utilization can have a wide range of variations, with uncertainty becoming greater the more that short-term data are extrapolated.
Primary percutaneous coronary intervention (PCI) is widely accepted as the preferred treatment modality for patients presenting with acute MI, with intravenous thrombolytics reserved for patients who cannot be transferred to a PCI-capable hospital within 60 minutes from presentation. Compared to thrombolytics, primary angioplasty has demonstrated less in-hospital mortality, lower in-hospital and 6-month rates of reinfarction, and less death.21 Further, primary angioplasty is associated with lower rates of fatal bleeding complications, including intracranial hemorrhage.21 The cost of either approach depends on initial hospitalization, initial cost of revascularization, subsequent readmission, subsequent investigations and interventions, adjuvant use of antiplatelet and anticoagulant agents, outpatient care, and work loss.
Five-year comparative effectiveness of angioplasty versus thrombolytic therapy was evaluated in a cohort of 395 patients enrolled between 1990 and 1993.22 All-cause mortality was significantly lower in the angioplasty group compared to the streptokinase group during the first 30 days (1% vs 7%, P = .01) and during 5-year follow up (13% vs 24%, P = .01). The streptokinase group required significantly higher revascularization procedures during the 5-year follow-up period. This study demonstrated that per patient, total medical charges at the end of the follow-up period were lower in the primary angioplasty group compared to the streptokinase group ($16,090 vs $16,813, P = .05). Total charges per patient for those who were alive at the end of follow-up trended toward being lower in the angioplasty group ($18,664 vs $21,772, P = .08). This medical cost consisted of days spent in the hospital, diagnostic and therapeutic procedures, and medications used during the follow-up period calculated as per-hospital charges in 1992.22
In a substudy of the CAPTIM trial (Comparison of Angioplasty and Prehospital Thrombolysis in Acute Myocardial Infarction), the cost efficacy of primary PCI was compared against alteplase, a more potent thrombolytic agent than streptokinase.23 In the overall study (randomized multicenter trial of 840 patients), investigators compared primary PCI to prehospital thrombolysis for patients presenting with STEMI. All prehospital thrombolysis patients were transferred to a PCI center for further management and rescue angioplasty. The trial showed that primary angioplasty was not better (for death, nonfatal MI, and stroke) than prehospital thrombolysis with rescue PCI at 30 days, but the primary PCI group had a significantly lower composite of death, nonfatal MI, stroke, revascularization, and major bleeding (34% vs 61%, P = .0001). In the cost-efficacy substudy, 299 patients were enrolled at 3 participating sites between 1997 and 2000. Cost data were prospectively collected during initial hospitalization and during 1-year follow-up. The primary PCI group had significantly lower cost of hospitalization by $883 compared to the thrombolysis group.23 Longer length of hospitalization and cost of thrombolytic agents offset the initial higher cost of angioplasty. This difference in cost persisted at 1 year ($1224 lower in the angioplasty group than the prethrombolysis group, P <.04).23
Cost and health outcomes comparing primary PCI and thrombolysis in patients presenting with STEMI have been more comprehensively evaluated in the SWEDES trial.24 The trial compared strategies of primary PCI with enoxaparin and abciximab versus thrombolysis with enoxaparin. Antiplatelet regimens were similar. In-hospital and 1-year clinical outcomes (death, nonfatal MI, stroke) were similar between the 2 groups.24 Quality-adjusted weight was obtained using EQ-5D questionnaire to calculate quality-adjusted survival. At 1 year, cost of intervention per patient in the PCI group was higher than the thrombolysis group ($4602 vs $3807, P = .047). The cost of antiplatelet agents was also higher in the PCI group ($1309 vs $1202, P = .001). The cost of hospitalization was higher in the thrombolysis group ($9278 vs $7244, P = .02). Combined cost of care per patient at 1 year was not significantly different between the 2 groups ($25,315 in PCI group vs $27,819 in thrombolysis group). Quality-adjusted survival remained marginally higher in the PCI group compared to the thrombolysis group (0.759 vs 0.728). In cost utility analysis, there was a trend toward $2504 less annual cost per patient and a 0.031 gain in quality-adjusted survival in the PCI group compared to the thrombolysis group. This trend, although statistical nonsignificant, appeared to be driven by less repeat hospitalizations in the PCI group. As depicted in Figure 72-3, PCI remained a cost-effective strategy in 88% and 89% of bootstrap replications when using threshold values of $50,000 and $100,000/QALY gained, respectively.
FIGURE 72-3
Primary percutaneous coronary intervention (PCI) is more cost effective than thrombolysis in patients presenting with ST-segment elevation myocardial infarction (STEMI) at 1 year follow-up (SWEDES trial). Graphical distribution of the incremental cost and quality-adjusted survival with primary PCI in STEMI patients compared to thrombolysis, based on 5000 bootstrap replications. Primary PCI is a favorable strategy since majority of the bootstrap replications are in right lower quadrant of cost-effectiveness plane and below the socially accepted threshold of $50,000/QALY gained. QALY, quality-adjusted life-year; USD, US dollars. (Reproduced from Aasa M, Henriksson M, Dellborg M, et al. Cost and health outcome of primary percutaneous coronary intervention versus thrombolysis in acute ST-segment elevation myocardial infarction-Results of the Swedish Early Decision reperfusion Study (SWEDES) trial. Am Heart J. 2010;160:322-328, Copyright © 2010, with pemission from Elsevier.)
In summary, primary PCI, as a first-line therapy in the management of patients with STEMI, is cost effective up to 5 years, primarily due to reduced rates of repeat hospitalization in the follow-up period.22–24
Clinical outcomes with an early invasive strategy are superior compared to the conservative management for patients presenting with UA or NSTEMI.25 Both strategies affect cost of care at several levels including hospitalization, emergency room visits, diagnostics and interventions, cardiac medications, rehabilitation and nursing home stay, and outpatient physician follow-up.
The TACTICS-TIMI-18 trial (Treat Angina With Aggrastat and Determine Cost of Therapy With an Invasive Versus Conservative Strategy–Thrombolysis in Myocardial Infarction-18) demonstrated that the composite of death, MI, or rehospitalization for ACS at 6 months was significantly lower in the early invasive group (within 48 hours) compared to the conservative group (16% vs 21%, P = .03).25 The treatment strategy in TACTICS-TIMI-18 was contemporary and inclusive of both coronary stenting and use of glycoprotein IIb/IIIa inhibition, favoring generalizability of clinical and economic findings to current clinical practice. Average initial cost of hospitalization appeared higher with an early invasive therapy compared to the conservative approach ($15,714 vs $14,047).25 As expected, initial rates of hospitalization and revascularization procedures (PCI and coronary artery bypass grafting [CABG]) were higher with an early invasive approach despite the significantly shorter length of stay especially in the high-risk populations (older, diabetic, and ischemic electrocardiogram [ECG] changes).25 This early increase in cost, however, was offset by lower 6-month follow-up costs for the early invasive arm, primarily due to reduced rates of rehospitalization. Total 6-month cost, therefore, remained comparable between the 2 treatment strategies. The ICER for death or MI prevented was $25,478 with an early invasive approach. Considering the time horizon of 6 months and CEA based on in-trial data, QALY and health utility status remained unchanged between the 2 groups.
However, applying life expectancy data from both these methods to the entire TACTICS-TIMI-18 study population demonstrated a significant advantage to the early invasive approach. The Framingham Heart Study and PURSUIT/Duke trial projections were used to estimate lifetime cost-effectiveness.25 Specifically, this analysis revealed that ICER with an early invasive approach for NSTEMI/UA patients is within society’s willingness-to-pay threshold (ICER of $16,272/life-year gained using Framingham Heart Study and ICER of $22,538/life-year gained based on the PURSUIT trial).25
Economic benefits of the early invasive approach as observed in TACTICS-TIMI-18 remained unchanged when subsequent developments in medical therapy and revascularization approaches were considered (eg, drug-eluting stents [DES]). The RITA-3 trial (Randomized Intervention Trial of Unstable Angina-3) demonstrated that high-risk patients (older, diabetic, positive biomarkers, and ST-segment changes on ECG) had the highest economic benefit of the early invasive approach at 5-year follow-up.26 The TACTICS-TIMI-18 trial provided insight into this observation, but RITA-3 (larger sample and longer follow-up) conceptualized it more definitively.26 Further, clinical efficacy of the early invasive approach was demonstrated in a United Kingdom–based population at 5-year follow-up compared to the conservative therapy in a randomized multicenter comparison of 1810 patients presenting with NSTEMI (composite of death and nonfatal MI: 16.6% vs 20%, P = .04).27 The ICER with either strategy among low-, intermediate-, and high-risk patients was compared to societal threshold of £20,000 to £30,000 per QALY gained. A Markov model was constructed to evaluate lifetime cost-effectiveness beyond the 5-year trial follow-up. This multivariate model created low-, intermediate-, and high-risk groups based on 5-year prediction of death or MI. The ICER was approximately £55,000, £22,000 and £12,000 per QALY gained for the low-, intermediate-, and high-risk groups, respectively.27
It appears that the early invasive approach compared to conservative management for patients presenting with UA/NSTEMI is cost effective, especially for the high-risk patient population,25–27
Optimal Medical Therapy (OMT) Versus Initial Revascularization in Stable Coronary Artery Disease (CAD)
The COURAGE trial (Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation) provides the most comprehensive long-term outcomes data of stable CAD patients managed with OMT compared to PCI plus OMT in 2287 patients with 4.6-year follow-up. There was no difference in the primary end point of death or MI between groups; however, the PCI group had more angina relief and improved quality of life.20 Index resource utilization was significantly higher in the PCI group compared to the OMT group ($12,162 vs $752), but subsequent resource utilization was comparable ($22,681 vs $23,996, respectively).20 The higher in-trial cost was driven by higher initial cost of PCI. CEA of COURAGE trial demonstrated that addition of PCI as an initial management strategy for symptomatic chronic stable angina patients results in an ICER between $168,000 and $300,000 per life-year or QALY gained.20 Life expectancy beyond the trial period was estimated by Framingham survival data. In-trial and lifetime cost estimates for the PCI group remained significantly higher than the medical therapy group ($11,410 and $9451, respectively).20 The ICER/QALY gained for PCI was $206,229 for the trial period and $168,019 for lifetime. Therefore, medical therapy appeared to be dominant (better outcome at lower cost) (Fig. 72-4A). The cost-effectiveness acceptability curve demonstrated that at the $50,000 ICER/QALY gained threshold, PCI was rarely cost effective for the in-trial period, and there was only a 25% probability for lifetime. At the $100,000 threshold, the probability of ICER/QALY gained with PCI was <25% for the in-trial period and 41% for lifetime (Fig. 72-4B).20 This raised a question of which patients with chronic angina may derive cost-effective benefit from PCI as an initial management approach. In another cost-effectiveness substudy of the COURAGE trial, the ICER for PCI was evaluated in 3 quartiles of patients with chronic angina.28 These quartiles were defined by identifying health status among 3 domains of Seattle Angina Questionnaire (physical limitation, angina frequency, and quality of life). The ICER for PCI ranged from $80,000 to $330,000 for patients with intermediate to severe angina (lowest and middle quartile), whereas it ranged from $520,000 to $3 million in mild (upper quartile) angina.28 Therefore, at any level of angina severity, PCI as an initial strategy did not appear to be cost effective at socially acceptable cost thresholds.
FIGURE 72-4
Joint distribution of cost and effectiveness differences in cost-effectiveness plane (COURAGE trial). A. Percutaneous coronary intervention (PCI) was dominated by medical therapy at an ICER threshold of $50,000/QALY and $100,000/QALY gained. B. CE acceptability curve shows that PCI was cost effective <25% of the time for the ICER threshold of $50,000/QALY and 41% of the time for the ICER threshold of $100,000/QALY on lifetime horizon, making medical therapy a dominant initial strategy for stable coronary artery disease. CE, cost-effectiveness; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-years. (Reproduced from Weintraub WS, Boden WE, Zhang Z, et al. Cost-effectiveness of percutaneous coronary intervention in optimally treated stable coronary patients. Circ Cardiovasc Qual Outcomes. 2008;1(1):12-20.)
The BARI-2D trial (Bypass Angioplasty Revascularization Investigation-2 Diabetes) randomly compared revascularization plus OMT (1176 patients; 378 with CABG and 798 with PCI) versus OMT alone (1192 patients) in a total of 2368 patients with type 2 diabetes mellitus and CAD over a 5-year follow-up and demonstrated comparable all-cause mortality as well as combined death/MI/cerebrovascular accident between the 2 groups.29 Four-year cost analysis showed that prompt revascularization (either catheter based or surgery based) was not favorable compared to OMT for in-trial or lifetime cost estimates. In the PCI stratum, medical therapy yielded slightly higher QALY compared to PCI (3.25 vs 3.22) at significantly lower cost ($67,800 vs $73,400, P = .02) for the in-trial period, making it the dominant strategy (quadrant B in the cost-effectiveness plan).30 Prompt revascularization with CABG was more costly and less effective at 4 years compared to medical therapy. However, lifetime estimates showed that CABG compared to medical therapy increased survival from 12.9 to 13.42 years at only marginally higher cost ($235,500 vs $210,900), yielding an ICER of $47,000/life-year gained.30 Therefore, prompt revascularization with PCI did not appear to be cost effective compared to medical therapy. However, revascularization with CABG may be cost effective for lifetime estimates.
Although CABG is still considered the revascularization method of choice for multivessel or left main coronary disease, PCI has become an increasingly utilized revascularization method for certain multivessel and left main CADs, largely due to the advent and evolution of DES. Long-term follow-up data from randomized clinical trials comparing CABG and PCI for multivessel coronary disease have suggested that both groups are comparable in death and nonfatal MI. However, the PCI group has significantly increased rates of repeat revascularization procedures during 8 to 12 years of reported follow-up.31,32
Initial cost comparison data between CABG and PCI comes from EAST (Emory Angioplasty Versus Surgery Trial). Total costs of CABG and PCI were comparable during the 8-year follow-up period ($46,548 vs $44,491).31 However, initial observations in EAST demonstrated higher revascularization costs with PCI, which were confirmed in subsequent economic and quality-of-life data analyses in the BARI trial (Bypass Angioplasty Revascularization Investigation). CEA of the BARI trial over a 12-year follow-up period suggested that the initial higher medical cost of bypass surgery was offset by lower subsequent revascularization procedures and hospitalizations during follow-up compared to PCI ($123,000 for CABG vs $120,750 for PCI, P = .55).32 Initial cost of CABG was 54% higher than PCI, but the ICER for CABG at 12 years compared to PCI was $14,300/life-year gained for the in-trial period and $13,300/life-year gained for lifetime.32 Using the Duke Activity Status Index and RAND Mental Health Inventory V, the QALYs were 6.45 years for PCI and 6.58 years for CABG, yielding an ICER of $11,300/QALY gained for CABG.32 Despite the use of balloon angioplasty as the percutaneous revascularization method in EAST and BARI, increased rates of revascularization and the associated cost persisted with the use of BMS. For instance, the ARTS trial (Arterial Revascularization Therapy Study) demonstrated comparable death and MI at 3 years between CABG and PCI, but significantly increased rates of repeat revascularization in the PCI arm. The ICER for CABG per event-free patient (death, MI, or cerebrovascular accident [CVA]) was €10,492 at 3 years (well below the ICER threshold of €20,000-30,000/QALY or life-year gained).33
Early studies that demonstrated higher rates of repeat revascularization in patients with multivessel CAD, especially with diabetes, attributed these findings largely to the use of balloon angioplasty or BMS for revascularization in the PCI group, which are known to perform less well than DES.32,33 The FREEDOM trial (Future Revascularization Evaluation in Patients With Diabetes Mellitus: Optimal Management of Multivessel Disease) compared the outcomes of multivessel coronary disease with either PCI using DES or CABG in patients with diabetes. At 5 years of follow-up, the primary composite of death, nonfatal MI, and CVA was significantly higher in the PCI group compared to CABG (26.6 vs 18.7, P = .005), again demonstrating that CABG as a revascularization strategy outperformed PCI in this population, despite the use of the most contemporary and best-performing percutaneous material.34 Index hospitalization cost was higher in the CABG group compared to DES-PCI by $8622. Five-year cumulative cost was also slightly higher in the CABG group by $3641.34 However, CABG was associated with a gain in life expectancy of 0.794 years and an overall quality-adjusted life expectancy of 0.663 QALY.34 Therefore, economic analysis demonstrated robust cost-effectiveness for CABG over DES-PCI, with lifetime estimated ICER of $8132/QALY gained (99.2% of bootstrap replications <$50,000/QALY gained) and $6791/life-year gained, and persisted across a wide range of subgroups (based on SYNTAX score, age, hemoglobin A1C, and number of vessels involved).34
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
