Cost-Effectiveness of Mitral Valve Repair Versus Replacement for Severe Ischemic Mitral Regurgitation: A Randomized Clinical Trial From the Cardiothoracic Surgical Trials Network

Abstract

Background: The CTSN (Cardiothoracic Surgical Trials Network) recently reported no difference in left ventricular end-systolic volume index or in survival at 2 years between patients with severe ischemic mitral regurgitation (MR) randomized to mitral valve repair or replacement. However, replacement provided more durable correction of MR and fewer cardiovascular readmissions. Yet, costeffectiveness outcomes have not been addressed.

Methods and results: We conducted a cost-effectiveness analysis of the surgical treatment of ischemic MR based on the CTSN trial (n=126 for repair; n=125 for replacement). Patient-level data on readmissions, survival, qualityof- life, and US hospital costs were used to estimate costs and quality-adjusted life years per patient over the trial duration and a 10-year time horizon. We performed microsimulation for extrapolation of outcomes beyond the 2 years of trial data. Bootstrap and deterministic sensitivity analyses were done to address parameter uncertainty. In-hospital cost estimates were $78 216 for replacement versus $72 761 for repair (difference: $5455; 95% uncertainty interval [UI]: −7784–21 193) while 2-year costs were $97 427 versus $96 261 (difference: $1166; 95% UI: −16 253–17 172), respectively. Quality-adjusted life years at 2 years were 1.18 for replacement versus 1.23 for repair (difference: −0.05; 95% UI: −0.17 to 0.07). Over 5 and 10 years, the benefits of reduction in cardiovascular readmission rates with replacement increased, and survival minimally improved compared with repair. At 5 years, cumulative costs and quality-adjusted life years showed no difference on average, but by 10 years, there was a small, uncertain benefit for replacement: $118 023 versus $119 837 (difference: −$1814; 95% UI: −27 144 to 22 602) and qualityadjusted life years: 4.06 versus 3.97 (difference: 0.09; 95% UI: −0.87 to 1.08). After 10 years, the incremental cost-effectiveness of replacement continued to improve.

Conclusions: Our cost-effectiveness analysis predicts potential savings in cost and gains in quality-adjusted survival at 10 years when mitral valve replacement is compared with repair for severe ischemic MR. These projected benefits, however, were small and subject to variability. Efforts to further delineate predictors of long-term outcomes in patients with severe ischemic MR are needed to optimize surgical decisions for individual patients, which should yield more cost-effective care.

Clinical trial registration: URL: https://www.clinicaltrials.gov. Unique identifier: NCT00807040.

Keywords: Cardiac surgical procedures; cost-benefit analysis; health care costs; mitral valve; quality-adjusted life years.

Conflict of interest statement

Dr Acker: consultant/advisory board for Thoratec (modest). Dr Gillinov: other research support from St. Jude Medical, Abbott (modest); consultant/advisory board for St. Jude Medical, Edwards Lifesciences (modest); consultant/advisory board for Abbott, Medtronic, CryoLife, AtriCure (significant). Dr Thourani: research grants from Edwards Lifesciences, Medtronic Corp, and Abbott Medical (modest); consultant/advisory board for Edwards Lifesciences, Abbott Medical (modest). Dr Robert L. Smith: research grant from Edwards Lifesciences (significant); speakers bureau for Abbott Medical (significant); consultant/advisory board for Abbott Medical, Edwards Lifesciences (significant); other from Abbott Medical, Edwards Lifesciences (significant). The other authors report no conflicts. This work was supported by a cooperative agreement (U01 HL088942) funded by the National Heart, Lung, and Blood Institute and the National Institute of Neurological Disorders and Stroke of the National Institute of Health and the Canadian Institutes of Health Research. Dr Ferket was supported by American Heart Association grant number 16MCPRP31030016. The views expressed in this article are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; National Institutes of Health; or the United States Department of Health and Human Services.

2018 American Heart Association, Inc.

Figures

Figure 1.. Within-trial cumulative average cost by study arm.

Shown are cumulative costs averaged across N=126 for repair and N=125 for replacement.

Figure 2.. Average SF-6D utility index by study arm.

Shown are mean SF-6D utility index scores in N=126 for repair and N=125 for replacement.

Figure 3.. Within-trial cost-effectiveness analysis bootstrap results comparing replacement vs repair.

Shown are Δs in average costs and average QALYs as measured in each bootstrap replicate of the trial data with repair as the reference strategy. The yellow and red figures represent the point estimates (Δcosts, ΔQALYs) at 1-year ($848; −0.03) and 2-year ($1,166; −0.05) respectively. The two diagonals represent commonly used cost-effectiveness thresholds of $100K/QALY and $200K/QALY. The proportion of iterations below or to the right of the selected diagonal equals the likelihood of the replacement strategy being cost-effective as compared with repair given the applicable cost-effectiveness threshold.

Figure 4.. Observed and simulated all-cause mortality estimates by study arm.

Shown are all-cause mortality estimates based on Kaplan-Meier curves of trial data with censoring at 2-year (solid lines) and simulated mortality estimates within the reference case analysis (dashed lines).

Figure 5.. Number of hospital admissions and reoperations during 1-, 2-, 5-, and 10-year follow-up.

Outcomes at 1- and 2-year are based on trial data; outcomes at 5- and 10-year are based on adding simulated outcomes occurring within 2–5 and 2–10 year time intervals within the reference case analysis.

Figure 6.. Cost-effectiveness acceptability curves for replacement according to time horizon.

These curves indicate the probability of replacement being cost-effective as compared with repair using different time horizons. Each curve equals the proportion of iterations below or to the right of the diagonal (as shown in Figure 3 for time horizons of 1- and 2-year) by changing the slope of the diagonal from 0 to infinity, i.e. increasing the cost-effectiveness threshold. The probability of repair being cost-effective equals 100% minus the depicted probability of replacement being cost-effective.