Author + information
- Received October 1, 2015
- Revision received December 11, 2015
- Accepted December 22, 2015
- Published online May 1, 2016.
- Alexander T. Sandhu, MDa,b,c,∗ (, )
- Jeremy D. Goldhaber-Fiebert, PhDb,
- Douglas K. Owens, MD, MSa,b,
- Mintu P. Turakhia, MD, MASa,c,
- Daniel W. Kaiser, MDa and
- Paul A. Heidenreich, MD, MSa,c
- aVeterans Affairs Palo Alto Health Care System, Palo Alto, California
- bCenter for Health Policy and Center for Primary Care and Outcomes Research, Department of Medicine, Stanford University, Stanford, California
- cDepartment of Medicine, Stanford University School of Medicine, Stanford, California
- ↵∗Reprint requests and correspondence:
Dr. Alexander Sandhu, Center for Health Policy and Primary Care and Outcomes Research, 117 Encina Commons, Stanford University, Stanford, California 94305.
Objectives This study aimed to evaluate the cost-effectiveness of the CardioMEMS (CardioMEMS Heart Failure System, St Jude Medical Inc, Atlanta, Georgia) device in patients with chronic heart failure.
Background The CardioMEMS device, an implantable pulmonary artery pressure monitor, was shown to reduce hospitalizations for heart failure and improve quality of life in the CHAMPION (CardioMEMS Heart Sensor Allows Monitoring of Pressure to Improve Outcomes in NYHA Class III Heart Failure Patients) trial.
Methods We developed a Markov model to determine the hospitalization, survival, quality of life, cost, and incremental cost-effectiveness ratio of CardioMEMS implantation compared with usual care among a CHAMPION trial cohort of patients with heart failure. We obtained event rates and utilities from published trial data; we used costs from literature estimates and Medicare reimbursement data. We performed subgroup analyses of preserved and reduced ejection fraction and an exploratory analysis in a lower-risk cohort on the basis of the CHARM (Candesartan in Heart failure: Reduction in Mortality and Morbidity) trials.
Results CardioMEMS reduced lifetime hospitalizations (2.18 vs. 3.12), increased quality-adjusted life-years (QALYs) (2.74 vs. 2.46), and increased costs ($176,648 vs. $156,569), thus yielding a cost of $71,462 per QALY gained and $48,054 per life-year gained. The cost per QALY gained was $82,301 in patients with reduced ejection fraction and $47,768 in those with preserved ejection fraction. In the lower-risk CHARM cohort, the device would need to reduce hospitalizations for heart failure by 41% to cost <$100,000 per QALY gained. The cost-effectiveness was most sensitive to the device’s durability.
Conclusions In populations similar to that of the CHAMPION trial, the CardioMEMS device is cost-effective if the trial effectiveness is sustained over long periods. Post-marketing surveillance data on durability will further clarify its value.
The treatment of heart failure costs more than $20.9 billion in total health care expenditures (1). Most of these costs are incurred from treating clinical decompensations of patients with heart failure that result in more than 1 million hospital admissions annually (1,2). The CHAMPION (CardioMEMS Heart Sensor Allows Monitoring of Pressure to Improve Outcomes in NYHA Class III Heart Failure Patients) trial, a randomized, single-blinded, multicenter trial, investigated the use of an implantable, wireless pulmonary artery pressure monitoring system to decrease hospitalizations related to heart failure (3). In this study, 550 patients with New York Heart Association (NYHA) functional class III heart failure and a hospitalization for heart failure within the previous year underwent pulmonary artery sensor implantation. Patients were randomized to a treatment group in which providers were given access to the pressure readings or a control group in which the provider could not access the pressure readings. The treatment group was found to have fewer hospitalizations for heart failure and improved quality of life.
Newer management strategies such as CardioMEMS (CardioMEMS Heart Failure System, St. Jude Medical Inc., Atlanta, Georgia) that reduce costly hospitalizations for heart failure may decrease the substantial clinical and economic burden of heart failure. However, the high device cost (listed as $17,750 with Medicare) raises questions regarding its value (4). We performed an independent analysis of the cost-effectiveness of this device in a cohort on the basis of the trial, as well as in subgroups defined by ejection fraction. Additionally, we performed an exploratory analysis of the device in an alternative, larger trial-based cohort of patients with heart failure by using the CHARM (Candesartan in Heart failure: Reduction in Mortality and Morbidity) trials (5).
We developed a Markov model to determine the cost-effectiveness of the CardioMEMS device compared with usual care from a societal perspective in a CHAMPION trial cohort over a lifetime horizon. This cohort included adults (average age 62 years) with NYHA functional class III heart failure who were hospitalized within 1 year with preserved ejection fraction (21.7%) or reduced ejection fraction (78.3%). We used hospitalization and mortality rates from the CHAMPION trial (3). We performed subgroup analyses of cohorts with reduced ejection fraction (average age 60 years) and with preserved ejection fraction (average age 66 years) from the CHAMPION trial by using overall trial event rates and subgroup-specific rate ratios for each event from trials with larger sample sizes than CHAMPION (6–9). Subgroup-specific device efficacy was extracted from the CHAMPION trial (3).
In the model, patients had CardioMEMS device placement at the outset, which could involve a procedural complication or device deployment failure. In subsequent monthly intervals, patients could experience hospitalizations for heart failure, hospitalizations not related to heart failure, device complications, and all-cause mortality (Online Figure 1). Patients had an increased mortality risk during hospitalization for heart failure and for 2 months post-hospitalization. The model followed all patients over their lifetimes. We matched the mortality rates over the mean duration of the trial for the control arm of the CHAMPION trial (17 months). After the trial period of 17 months, all event rates are extrapolated. We extrapolated an age-based increase in overall mortality from a previous retrospective analysis (10).
Rate of hospitalization for heart failure and efficacy of the cardiomems device
We matched the trial rates of hospitalizations secondary to heart failure for each cohort. We modeled a declining rate of hospitalization over the CHAMPION trial duration. We modeled the CardioMEMS reduction in the rate of hospitalizations for heart failure on the reduction over the entire trial (hazard ratio [HR]: 0.63). We assumed that preventing a hospitalization prevented inpatient and 2-month post-hospitalization increases in mortality (11–13). We did not model any additional CardioMEMS-associated mortality reduction in the base case. For our base case, we assumed that the benefit of the CardioMEMS device would continue lifelong and examined shorter durations in sensitivity analyses.
Cardiomems device events
We modeled periprocedural complications as a composite of the procedure-related serious adverse events and major bleeding during the 30-day post-procedure anticoagulation period (3). We additionally modeled procedural placement failure and CardioMEMS-related serious adverse events that occurred after the initial month.
Quality of life and costs
We included quality of life estimates for the patient’s baseline health, the use of the CardioMEMS device, hospitalizations, and complications by using utilities. We calculated utility values by converting the 6-month Minnesota Living with Heart Failure (MLWHF) questionnaire score for the control arm in the CHAMPION trial into EQ-5D scores (14). The difference-in-difference in EQ-5D score between groups from baseline to 6 months was applied as the quality of life benefit for the CardioMEMS device for the first year. The difference-in-difference between groups from baseline to 12 months was applied thereafter. The 6-month differences were used for the entire first year because 226 of 550 patient scores were missing at 12 months. Disutilities were applied for the initial procedure, hospitalizations, and complications. Comparisons of patient utility during and after a hospitalization for heart failure showed an 11% lower utility during hospitalization, for a decrement of approximately 3 days (15). These assumptions were tested in sensitivity analyses, including an analysis in which we alternatively assumed long-term utility change secondary to hospitalizations for heart failure (16).
We included all health care–related costs. Hospitalization costs were taken from the Agency for Healthcare Research and Quality (AHRQ) National Inpatient Sample (17), with physician costs calculated using the 2014 Medicare Professional Fees (Current Professional Technologies [CPT] codes 99212, 992213, 99222, 99223, 99254, and 99255) (18). Age-adjusted outpatient medical costs for patients with heart failure were taken from the literature (2). The cost of CardioMEMS implantation, in addition to the device cost, is the cost of the right-sided heart catheterization and angiography ($1,129) (from base Medicare 2014 Professional Fees, CPT codes 93451 and 93568) along with the fee for device placement, currently an unlisted fee, which we assumed would approximate the cost of inserting a temporary single-chamber cardiac electrode ($185, Medicare CPT 33210) (18,19). This assumption was made on the basis of selecting reimbursement for a simple intracardiac procedure, given that this is additional to catheterization and angiography reimbursement. We based the monthly cost of CardioMEMS management on the estimated time required to monitor the device by physicians and nurses and average provider wage ([20,21], Klein L, personal communication, November 2015). All costs were updated to 2014 U.S. dollars by using the medical component of the consumer price index (22).
We evaluated the effect of uncertainty in all model inputs (Table 1). We focused on device-specific uncertainties and characteristics of the patient population. Given the single clinical trial currently available, we varied estimates of the efficacy of the device extensively by adjusting the reduction in hospitalizations, the effect of the device on mortality, the effect on quality of life, and the duration of the device’s effectiveness. With its recent market introduction, we also varied the costs of the device, implantation, and monitoring substantially. Given the large heterogeneous heart failure population, we also conducted analyses on a range of hospitalization costs, baseline utility values, and baseline hospitalization and mortality rates. This included an exploratory analysis of the intervention in a cohort on the basis of the CHARM trials, which was a lower-risk cohort with greater quality of life and lower event rates. This cohort included adults (average age 62 years) with NYHA functional class II (45%), class III (52%), and class IV (3%) heart failure, as opposed to patients in CHAMPION, who exclusively had NYHA functional Class III disease. We used published CHARM trial data to estimate hospitalization and mortality rates along with utility scores (5,6,23). We adjusted the rates to match the ejection fraction subgroup composition of the CHAMPION cohort and to include only patients with a previous hospitalization for heart failure. We estimated the device effectiveness needed to meet important cost-effectiveness thresholds in this cohort.
We performed a probabilistic sensitivity analysis by performing 10,000 simulations in which we simultaneously sampled from the distributions of each input parameter with each simulation (see the Online Appendix for details).
We discounted future costs and benefits at 3% annually and adhered to best practices (24,25). The main outcome measure was cost per QALY gained. Cost-effectiveness thresholds followed American College of Cardiology/American Heart Association (ACC/AHA) guidelines, with a threshold of <$50,000 indicating highly cost-effective and than >$150,000 not cost-effective (26).
Comparison with the CHAMPION trial results
In the first 6 months, modeled rates of hospitalizations per patient for the usual care and intervention groups of 0.44 and 0.29, respectively, matched those of the CHAMPION trial 0.44 (95% confidence interval [CI]: 0.36 to 0.53) and 0.32 (95% CI: 0.26 to 0.40). The modeled annual rates over the mean trial period for the 2 groups were 0.68 and 0.45, respectively, matching trial results of 0.69 (95% CI: 0.61 to 0.78) and 0.46 (95% CI: 0.40 to 0.53). The modeled annual mortality probabilities over the mean trial period for the usual care and intervention groups were 14.8% and 13.2%, respectively, matching the trial’s 14.7% (95% CI: 11.7 to 17.9) and 11.9% (95% CI: 9.5 to 16.3). Further details are included in the Online Appendix.
In the CHAMPION trial cohort, the modeled CardioMEMS arm has a total of 2.18 hospitalizations per patient compared with 3.12 in the usual care arm of the trial, an absolute reduction of 0.94 hospitalizations over a patient’s lifetime (Table 2). The CardioMEMS arm increased life expectancy by 0.42 years and quality-adjusted life expectancy by 0.28 QALYs. The CardioMEMS arm achieved its health benefits at an increased cost of $20,079. Taken together, the CardioMEMS intervention costs $71,462 per QALY gained or $48,054 per life-year gained.
Ejection fraction subgroups
Patients receiving usual care in the preserved ejection fraction subgroup had a longer average survival than did patients in the reduced ejection fraction subgroup (Table 2). The reduction in hospitalizations with CardioMEMS was greater for patients with preserved ejection fraction compared with those with reduced ejection fraction, which also resulted in lower incremental costs. With more QALYs gained and a smaller difference in costs, CardioMEMS cost $47,768 per QALY gained in the preserved ejection fraction cohort compared with $82,301 in the reduced ejection fraction cohort.
The base case used a device cost of $17,750. The device cost <$50,000 per QALY gained if the cost is <$9,798 in patients with reduced ejection fraction and <$18,657 in patients with preserved ejection fraction (Figure 1). Use of the device would cost more than $150,000 per QALY gained if it cost more than $34,418 in the reduced ejection fraction subgroup or $59,296 in the preserved ejection fraction subgroup. The cost of a hospitalization for heart failure ($12,832) was determined on the basis of the national average; however, there is significant hospital variation. In a large, urban, public teaching hospital with a higher predicted cost of hospitalization ($16,750), the CardioMEMS device costs $62,121 per QALY gained. In a small, rural, private nonteaching hospital with lower predicted costs of hospitalization ($8,341) the device cost becomes less favorable, $82,169 per QALY gained (Figure 2).
In the base case, we assumed a monthly cost of $68 to monitor the CardioMEMS device. This cost would need to be <$190 monthly for the device to cost <$100,000 per QALY and <$403 monthly for a cost lower than $150,000 per QALY.
The efficacy of the CardioMEMS device is modeled by reduction in hospitalizations, the risk of mortality associated with hospitalizations, and the effect on baseline quality of life. Although CardioMEMS had a 0.63 HR for hospitalizations over the entire randomized period of the CHAMPIONS trial, the 95% CI was 0.52 to 0.77. Over this range, CardioMEMS costs between $52,556 and $120,143 per QALY gained. The effect of the CardioMEMS device on mortality is uncertain. In the base case, we assumed that prevented hospitalizations had an inpatient mortality risk similar to all hospitalizations for heart failure and that preventing these hospitalizations would also avert the associated increase in post-hospitalization mortality, which led to an 11% relative reduction in mortality in the CardioMEMS arm compared with usual care over the trial period. If we had instead assumed a 20% relative reduction, the value of the CardioMEMS device would improve to $55,378 per QALY gained. However, if preventing hospitalizations did not reduce mortality, the cost would be $159,984 per QALY gained. For the device to cost <$100,000 per QALY gained, the relative reduction in mortality must be at least 4%.
Although the CHAMPION trial followed patients for 17 months on average, we assumed a lifelong duration. If the effectiveness of the CardioMEMS device ceased after 17 months, its cost per QALY gained was $214,879. Cost per QALY gained declined as the duration of effectiveness increased, dropping for heart failure <$150,000 at 34 months and $100,000 at 72 months. In 2-way sensitivity analyses, we found that the device is much less cost-effective if its duration of effect is shorter in populations with lower monthly hospitalization rates (Figure 3).
Adjusting the periprocedural complication rate, chronic complication rate, or placement failure rate did not substantially alter our findings (see the Online Appendix).
The impact of the CardioMEMS device on quality of life did not substantially change our main findings (see the Online Appendix). Neither alternative assumptions of baseline utility nor duration of disutility from hospitalization substantially altered our results (see the Online Appendix). Adjusting long-term baseline utility on the basis of the number of heart failure hospitalizations also did not substantially affect our findings (see the Online Appendix).
Severity of illness
Baseline rates of hospitalization and mortality matched trial rates. However, individual health care systems will have heterogeneous patient groups with different rates of readmission and mortality. We found that decreases in the hospitalization rate and increases in the mortality rate both increased the cost per QALY of CardioMEMS, but the cost did not exceed $150,000 per QALY gained (Online Figure 2).
In the CHARM cohort, the usual care arm of the trial had greater survival (7.9 years), more QALYs (4.67), and fewer hospitalizations for heart failure (1.71) compared with the CHAMPION cohort. If the device had a similar effect on quality of life in the CHARM cohort as in the CHAMPION trial, the device would need to prevent 26.5% of hospitalizations for heart failure to cost <$150,000 per QALY gained and 41.1% to cost <$100,000 per QALY gained (Online Figure 3).
In the base case, we used a lifetime horizon. Over a 5-year horizon, which was used in the trial cost-effectiveness, we found a $15,029 difference in costs and a 0.11 difference in QALYs, yielding a cost of $138,466 per QALY gained (Online Figure 4).
Probabilistic sensitivity analysis
In the probabilistic sensitivity analysis, we found that 17.3% of simulations showed that CardioMEMS was the preferred intervention at a willingness-to-pay threshold of $50,000, 76.9% at a threshold of $100,000, and 95.1% at a threshold of $150,000 (Online Figure 5). Additional sensitivity analyses are available in the Online Appendix.
Our analysis demonstrates that the use of the CardioMEMS device is cost-effective in patients with NYHA functional class III heart failure and a history of hospitalization for heart failure in the preceding year. Our base case analysis finds a cost of $71,462 per QALY gained in a CHAMPION trial population. The device value is most sensitive to its durability and the association between the reduction in hospitalizations and survival. The device must provide benefits over at least 34 months to cost <$150,000 per QALY. With randomized data of 17 months and open-access data for an additional 13 months, the device’s effectiveness over a longer period is unknown. Although hospitalizations for heart failure are associated with an increased mortality risk, it is unclear how preventing hospitalizations with CardioMEMS affects survival. The CHAMPION trial was underpowered to detect a mortality difference; a survival analysis submitted to the Food and Drug Administration demonstrated substantial uncertainty (HR: 0.8; 95% CI: 0.55 to 1.15) (27). We demonstrate that the device would need to reduce mortality by 4% to cost <$100,000 per QALY. Although this seems likely, it is possible that prevented hospitalizations may be lower risk, and mortality may be relatively unaffected. Future studies that follow hospitalization trends and refine estimates of the effect of the device on mortality can reduce the uncertainty regarding its clinical and economic value.
Our analysis indicates that the CardioMEMS device provides better value in patients with preserved ejection fraction. There are few evidence-based treatments available for patients with preserved ejection fraction; thus, the CardioMEMS device represents a rare evidence-supported intervention for this important group. Our model predicted greater value in this group because of the longer survival and increased device effectiveness in this group; however, the estimate of effectiveness in this group is made on the basis of only 119 trial patients and should also be refined with future research.
The incremental cost-effectiveness ratio found in the CHAMPION trial was more favorable than our estimate (3). The CHAMPION study, performed over a 5-year time horizon, found a larger difference in QALYs and a much smaller cost difference between the 2 trial arms ($4,282) to lead to an incremental cost-effectiveness of $13,379 per QALY. We do not have access to their assumptions to analyze these differences. The lifetime horizon used in our model captures longer-term benefits and costs and the impact of device durability.
The importance of the CardioMEMS device is tied to the scope of the problem. Health care expenditures secondary to heart failure are expected to rise from $20.9 billion in 2012 to $53.1 billion in 2030, with 80% of these costs attributed to hospitalizations for heart failure (28). These hospitalizations not only are costly but also are markers of a worsening clinical prognosis, associated with high rates of rehospitalization and mortality. Strategies to reduce this clinical and economic burden are needed. We demonstrate that the CardioMEMS device may be a cost-effective intervention for outpatient management of heart failure. However, our analysis also shows that the savings from reduction in hospitalization costs are exceeded by the intervention costs, which could thereby still have a large budgetary impact. We also illustrate that the value decreases in lower-risk patients; with the substantial heterogeneity in morbidity of patients with heart failure, ensuring that patients fit the trial criteria will be important. Although CardioMEMS implantation is a costly intervention that should be reserved for appropriately selected patients and still requires further evaluation, the value of this device compares favorably with that of other technologies used in similar patient groups, such as left ventricular assist devices (29).
First, a single trial has evaluated the intervention; the effectiveness seen in this trial should be confirmed in post-surveillance evaluation. Second, there may be treatment benefits that are not captured in our model, such as identifying patients who need to initiate advanced therapy. Third, although no serious device-related complications occurred outside of the procedural period, long-term safety data are not currently available. Finally, we attempted to capture additional costs of using the device by using the time required at a heart failure center, but the average national cost of the monitoring program is currently unclear.
This analysis shows that the use of the CardioMEMS device is a cost-effective means of improving quality of life and reducing rehospitalizations in patients with heart failure. It is a better value in patients with preserved ejection fraction, a group with few effective therapies. The cost-effectiveness of CardioMEMS is most sensitive to the duration of effectiveness; therefore, further research on the continued hospitalization trends of patients with the device will be important for future evaluations.
COMPETENCY IN MEDICAL KNOWLEDGE: In patients with NYHA functional class III heart failure and a hospitalization for heart failure in the preceding year, the CardioMEMS device reduced readmissions for heart failure and improved quality of life at a cost lower than commonly accepted U.S. willingness-to-pay thresholds. The device was cost-effective for both patients with reduced ejection fraction and preserved ejection fraction.
TRANSLATIONAL OUTLOOK: The clinical effectiveness of the device has been demonstrated only in a single randomized clinical trial with a mean follow-up of 17 months. Evaluation of the long-term rate of hospitalization for heart failure and the relationship between averted hospitalizations and changes in mortality in patients after CardioMEMS implantation will refine estimates of the value of the device. Additional data on the cost of maintaining and monitoring the device in the community will also inform future economic evaluations.
The authors thank Iris H. Ma, MD for her thoughtful comments on the manuscript draft.
For supplemental text, figures, tables, and references, please see the online version of this article.
This study was also supported in part by the Department of Veterans Affairs Quality Enhancement and Research Initiative 04-326. Drs. Sandhu, Owens, Turakhia, Kaiser, and Heidenreich were supported by the Department of Veteran Affairs. Dr. Goldhaber-Fiebert is supported by a National Institute of Health (NIH) NIA Career Development Award (K01 AG037593-01A1: PI: Goldhaber-Fiebert). Dr. Turakhia has consulted for St. Jude Medical, Medtronic, and Precision Health Economics. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- New York Heart Association
- quality-adjusted life year
- Received October 1, 2015.
- Revision received December 11, 2015.
- Accepted December 22, 2015.
- American College of Cardiology Foundation
- Mozaffarian D.,
- Benjamin E.J.,
- Go A.S.,
- et al.
- Dunlay S.M.,
- Shah N.D.,
- Shi Q.,
- et al.
- ↵Department of Health and Human Services. 42 CFR Parts 405, 412, 413, et al. Fed Regist August 22, 2014. Available at: http://www.facs.org/∼/media/files/advocacy/regulatory/fy 2015 ipps final rule_fr.ashx. Accessed January 14, 2015.
- Bello N.A.,
- Claggett B.,
- Desai A.S.,
- et al.
- Fonarow G.C.,
- Stough W.G.,
- Abraham W.T.,
- et al.
- Solomon S.D.,
- Dobson J.,
- Pocock S.,
- et al.
- Gustafsson F.,
- Torp-Pedersen C.,
- Seibaek M.,
- Burchardt H.,
- Kober L.,
- DIAMOND Study Group
- Abraham W.T.,
- Fonarow G.C.,
- Albert N.M.,
- et al.
- Adams K.F. Jr..,
- Fonarow G.C.,
- Emerman C.L.,
- et al.
- O’Connor C.M.,
- Abraham W.T.,
- Albert N.M.,
- et al.
- Jaagosild P.,
- Dawson N.V.,
- Thomas C.,
- et al.
- ↵Center for Medicare and Medicaid. Physician Fee Schedule. 2015. Available at: http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeeSched/index.html?redirect=/PhysicianFeeSched/. Accessed January 5, 2016.
- St. Jude Medical. St. Jude Medical Reimbursement Guide. 2015. Available at: http://professional.sjm.com/resources/reimbursement/crm-ep/cardiomems. Accessed March 1, 2015.
- Medical Group Management Association
- Occupational Employment Statistics. Occupational Employment and Wages, May 2014, 29–1141 Registered Nurses. Available at: http://www.bls.gov/oes/current/oes291141.htm. Accessed November 20, 2015.
- ↵United States Department of Labor, Bureau of Labor Statistics. Consumer Price Index. Available at: www.bls.gov/cpi. Accessed January 5, 2016.
- Gold M.R.
- Caro J.J.,
- Briggs A.H.,
- Siebert U.,
- Kuntz K.M.,
- ISPOR-SMDM Modeling Good Research Practices Task Force-1
- Anderson J.L.,
- Heidenreich P.A.,
- Barnett P.G.,
- et al.
- ↵Food and Drug Administration. PMA P100045: FDA Summary of Safety and Effectiveness Data. 2015. Available at: http://www.accessdata.fda.gov/cdrh_docs/pdf10/P100045b.pdf. Accessed February 2, 2015.
- Heidenreich P.A.,
- Albert N.M.,
- Allen L.A.,
- et al.
- Miller L.W.,
- Guglin M.,
- Rogers J.