|Home | About | Journals | Submit | Contact Us | Français|
The purposes of this study are to investigate the cost-effectiveness of an implantable carotid body stimulator (Rheos®) for treating resistant hypertension and determine the range of starting systolic blood pressure (SBP) values where the device remains cost-effective.
A Markov model compared a 20 mmHg drop in SBP from an initial level of 180 with Rheos® to failed medical management in a hypothetical 50-year old cohort. Direct costs (2007$), utilities and event rates for future myocardial infarction, stroke, heart failure and end-stage renal disease were modeled. Sensitivity analyses tested the assumptions in the model.
The incremental cost-effectiveness ratio (ICER) for Rheos® was $64,400 per quality-adjusted life-year (QALY) using Framingham-derived event probabilities. The ICER was <$100,000/QALY for SBPs ≥142. A probability of device removal of <1% per year or SBP reductions of ≥24 mmHg were variables that decreased the ICER below $50,000/QALY. For cohort characteristics similar to ASCOT-BPLA trial participants, the ICER became $26,700/QALY. Two-way sensitivity analyses demonstrated that lowering SBP 12 mmHg from 220 or 21 mmHg from 140 were required.
Rheos® may be cost-effective, with an ICER between $50,000-$100,000/QALY. Cohort characteristics and efficacy are key to the cost-effectiveness of new therapies for resistant hypertension.
Elevated systolic blood pressure (SBP) is an important risk factor for cardiovascular and cerebrovascular disease; cardiovascular risk doubles for each 20 mmHg increment above 115.1 Hypertension can be effectively treated, in most cases, through lifestyle changes and medications. However, even with best medical care, some individuals develop resistant hypertension. The JNC-7 defined resistant hypertension as failing to achieve BP targets while adhering to a three-medication regimen, one of which is a diuretic.1
Additional treatment options for individuals with resistant hypertension include an implantable device and a novel agent. Rheos® (CVRx, Inc. Minneapolis, MN) is a device surgically implanted near the clavicle with leads that stimulate the carotid baroreflex system, thereby reducing blood pressure. Rheos® lowered SBP by 21-35 mmHg in several phase I trials.2-4 Aliskiren (Novartis, East Hanover, NJ) is the first medication available in a new class of anti-hypertensive agents, the direct renin inhibitors. A 300 mg daily dose lowered SBP by 5-11 mmHg.5
Despite the apparent efficacy of Rheos® or aliskiren, neither have been subjected to a formal cost-effectiveness analysis. Most cost-effectiveness analyses of treating hypertension are based on adding or comparing conventional medications.6, 7 The costs of the Rheos® device, surgical implantation and maintenance differ from the continual monthly costs of medications. The initial costs of Rheos® must be offset by the prevention of hypertension-related morbidity and mortality. Thus, the purposes of this analysis are two-fold. First, to determine the cost-effectiveness of Rheos® as new adjuvant therapy that results in 20 mmHg SBP reduction from a starting SBP of 180 mmHg compared to failed medical management. The second goal is to determine the range of SBP levels over which the device is cost-effective. The cost-effectiveness of aliskiren is presented as a point of comparison and as an additional evaluation of the cost-effectiveness model.
A Markov model incorporated future adverse events, costs and benefits of treatment with Rheos® compared to failed best medical therapy (TreeAge, Williamstown, MA). The cycle length was one year. A hypothetical cohort of 1000 people began in the healthy state. The control cohort could remain healthy or transition to death, stroke, heart failure, ESRD or myocardial infarction (MI)(Appendix Figures 1 and 2). The analysis ended when the hypothetical cohort reached the absorbing death state.
The two year Rheos® data suggest that there is no rebound in SBP.3 Thus, the SBP decrease was maintained throughout the life of the cohort. For those in the Rheos® branch, there was also the possibility of device removal from the healthy state. If the device was removed, the blood pressure was returned to the initial systolic level. The probabilities of adverse events were then adjusted to reflect the risks associated with resistant hypertension.
Cardiac events included fatal coronary heart disease or MI. Once an MI occurred, the individual was transitioned to an MI branch where the probabilities of adverse events incorporated a history of cardiovascular disease. The MI branch allowed transitions to death, stroke, heart failure, ESRD or fatal coronary heart disease. Recurrent MI was possible and the transition was back to the MI branch.
Stroke was divided into transient ischemic attack (TIA), major ischemic stroke and hemorrhagic stroke (including both subarachnoid or intracerebral hemorrhage). TIA was a temporary state where function recovered completely. Therefore, after a TIA, the transition was back to the original state. Major ischemic stroke, major hemorrhagic stroke, heart failure and ESRD were considered serious health states from which individuals could not recover. The transitions from major ischemic stroke, major hemorrhagic stroke, heart failure and ESRD were to continue in the current state or death.
The target population was an asymptomatic 50-year old cohort with uncontrolled hypertension, despite poly-pharmacological management, and no history of cardiovascular disease or stroke. The cohort age of 50 reflected phase I Rheos® trials.2, 4 Initial systolic blood pressures were varied from 140 to 220 mmHg.
The payer perspective was modeled and included direct medical costs relating to device therapy and adverse health events. The time horizon was the lifetime of the cohort.
Probabilities for heart failure, stroke, MI and fatal coronary heart disease were based on Framingham studies (Table 1, Appendix Tables 1A).8-10 The probability of ESRD and proportions for the stroke subtypes and severity were obtained from the literature.11-13 The prevalence of diabetes mellitus (DM), ESRD-related mortality and atrial fibrillation increased with age (Appendix Tables 1B-D). Please see the on-line appendix for information on prevalence of risk factors for the Framingham equations and relative risks given medical history.
To test resistant hypertension treatment in a different cohort, probabilities from the ASCOT-BPLA trial were modeled separately from the Framingham equations.14, 15 The initial age for models of the ASCOT-BPLA cohort was 63, the average age of participants in this trial.
Costs for the device, the surgical implantation or removal procedure and ongoing maintenance were estimated (Table 2). Costs of hospitalization following adverse events and of continuing care were obtained from the 2005 Nationwide Inpatient Sample of the Health Care Utilization Project and from the literature.16-32 The hospitalization cost of a hemorrhagic stroke assumed that 23% of these events were subarachnoid hemorrhage and the remaining were intracerebral hemorrhage. All costs were adjusted to 2007 dollars (US) using the Consumer Price Index for medical care. As this comparison focused on the differences between a new device and continued medical management, follow-up and medication costs shared by both branches were not included. Once Rheos® failed as a treatment, the costs of maintenance no longer accrued.
Utilities, or health state preferences, were obtained from the literature (Table 2).33-48 The occurrence of a transient ischemic attack (TIA) reduced the utility proportionally. For example, a TIA with a history of MI had a utility of 0.7832 for the year (0.88 utility of an MI *0.89 utility of a TIA). Utilities adjustments for side effects, surgical implantation or removal were not included. Quality-adjusted life years (QALYs) represent utility multiplied by the duration of the health state. For example, if someone has a post-MI utility for 1.5 years, it contributes 1.32 QALYs (utility of 0.88 for an MI * 1.5 years).
All future costs and utilities were discounted at 3%.49 The Rheos® branch, the newest and more expensive option, was the intervention. Best medical management – though failing to reach recommended blood pressure guidelines – was the comparator. The intervention was considered cost-effective if the incremental cost-effectiveness ratio (ICER) was below $50,000 per QALY. If the intervention exceeds $100,000 per QALY, the intervention was considered cost-ineffective. ICERs between $50,000 and $100,000 per QALY were in a grey area.
Lowering SBP by 20 mmHg with Rheos® in a 50-year old cohort with resistant hypertension yielded 16.868 QALYs at a cost of $72,699. In comparison, best medical management where SBP remained at 180 mmHg produced 16.584 QALYs at a cost of $54,421 (2007$) over a lifetime. The incremental cost of Rheos® was $18,278 and the incremental QALY gain was 0.284. Thus, the ICER for lowering SBP with Rheos® was $64,400 per QALY gained.
The ICER for Rheos® remained below $100,000 per QALY as long as the initial SBP was ≥142 mmHg (Figure 1A). The ICER decreased below $50,000 per QALY if the initial SBP was ≥206 mmHg.
A series of one-way sensitivity analyses tested the individual assumptions in the model (Figure 1A). Rheos® became clearly cost-effective (ICER <$50,000 per QALY) if the SBP decrease was ≥24 mmHg from an initial level of 180 mmHg. Further reductions in SBP of 30 mmHg4 and 35 mmHg3 led to ICERs of $34,380 per QALY and $26,959 per QALY, respectively. Rheos® was cost-ineffective (ICER >$100,000 per QALY) when the change in SBP was <14 mmHg. Rheos® was cost-effective in those with diabetes mellitus or a poor lipid profile (total cholesterol/high density lipoprotein ratio). As the probability of device removal dropped below 1%, the ICER for Rheos® was cost-effective.
Costs and utilities were also tested individually (Figure 1B). If the cost of maintenance exceeded $900 per year, the ICER for Rheos® was >$100,000 per QALY. As the cost of the device or the procedure dropped below $6,000, the ICER for Rheos® decreased below $50,000 per QALY. Utilities of <0.55 for MI were associated with an ICER <$50,000 per QALY. A low utility for an MI caused an incremental QALY gain for Rheos® (and in turn a lower ICER) because quality of life was preserved when MIs were avoided. The discount rate also influenced the ICER.
There was a biphasic relationship between age at the initiation of therapy and the ICER for Rheos®. The ICER decreased as the initial age increased from 40 to 60 (Figure 2A), then increased again for ages 60-70 years. However, the ICER did not exceed $100,000 per QALY. The u-shape was attributed to the underlying changes in the incremental QALYs (Figure 2B). Incremental costs decreased from $20,022 to $17,058 as the starting age increased from 40 to 70 years. However, incremental QALYs increased from 0.24 at age 40 through 0.30 at age 57. Beginning at the age of 58, incremental QALYs decreased to 0.26 at age 70. The incremental QALYs decrease (the denominator of the ICER) occurred at a greater rate than the decrease in incremental costs thus causing the ICER increase at age 60.
Gender also had a role in the cost-effectiveness of resistant hypertension treatment. The ICERs of Rheos® for women and men, using gender-specific mortalities, were $57,200 per QALY and $71,000 per QALY, respectively (Table 3).
The cost-effectiveness of adding aliskiren to reduce SBP instead of an implantable device was tested. An 11.2 point SBP reduction for aliskiren was assumed for a 300 mg dose, at a cost of $1220 per year.5 The discontinuation rate was 2.2% per year.5 The lifetime cost for aliskiren became $67,470 for 16.729 QALYs gained (Table 3). The QALYs and costs for untreated resistant hypertension remained the same. Thus, the ICER for aliskiren compared to failed medical management was $90,000 per QALY. The ICER for Rheos® compared to aliskiren was $37,600 because Rheos® had a greater QALY gain with higher costs.
To test the cost-effectiveness of treating resistant hypertension in a hypothetical cohort whose risks were different than Framingham, adverse events rates from the ASCOT-BPLA trial were incorporated into the model. The starting age was adjusted to 63 years to match the age of the ASCOT-BPLA cohort. The ICER for Rheos® therapy in an ASCOT-BPLA-like cohort was $26,700 per QALY (Table 4). Compared to a hypothetical 63-year old cohort using Framingham probabilities, the ASCOT-BPLA best medical management arm, with SBP remaining at 180, had reduced costs and increased QALYs. This reduced the ICER, compared to the Framingham-based model, reflecting both decreased incremental costs and increased incremental QALYs. With cardiovascular and cerebrovascular risks similar to the ASCOT-BPLA cohort, treating lhypertension remained cost-effective (ICER <$50,000 per QALY) down to a SBP of 158 mmHg. Even with an initial SBP of 140, the ICER for Rheos® in the ASCOT-BPLA cohort remained below $100,000 per QALY.
A two-way sensitivity analysis was performed to study the relationship between initial SBP and change in SBP. With a willingness-to-pay of $100,000 per QALY, Rheos® must lower SBP by 12 points from an initial level of 220 to remain cost-effective (Figure 3 where the initial age equals 50). Likewise, SBP must be lowered by at least 21 points if the initial SBP is 140. For cohort ages >50 years, Rheos®, with a 20-21 mmHg SBP decrease, was preferred over best medical management with resistant hypertension at all initial SBP levels tested (range: 140-220). For a 40-year old cohort with initial SBPs <160, best medical management with resistant hypertension is the favored treatment, unless the SBP decrease with Rheos® exceeds 20 mmHg.
This study reports the cost-effectiveness of treating resistant hypertension in a 50-year old cohort using a novel surgically implanted device and reports at which SBP to begin treatment. The ICER for Rheos® falls in a grey area between too expensive (>$100,000 per QALY) and an acceptable price (<$50,000 per QALY). Rheos® performs as well as, if not better, than the base case assumptions. The base case incorporated a 20 mmHg SBP decrease because most achieved this decrease at two years (average: 35 ± 8 from initial SBPs of 191 ± 32).3 Current Rheos® performance suggests that the device may lower SBP by an average of 304 to 353 mmHg. These alternatives were included in the sensitivity analyses and were cost-effective. Rheos® would also be an adoptable technology at higher initial SBPs or when individuals have a risk profile similar to the ASCOT-BPLA cohort.
The costs of the actual device or the surgery have not been definitively established. The estimated hospital cost for implantation of a carotid baroreflex activation device is $8,400 (ICD-9 procedure code 39.8).50 Our base case cost for the procedure and the device is $20,000, approximately the 2007$ hospital cost for a deep brain stimulator (18,800).50 The one-way sensitivity analysis showed that doubling the cost of the procedure generates an ICER of $99,500 (Figure 1B). Here, the costs total $30,000 for the device ($10,000) and the procedure ($20,000).
The ICER for aliskiren was explored as a comparable treatment for resistant hypertension. The ICER of $90,000 per QALY also falls within a grey area as an acceptable price for a new medication. Lastly, comparing Rheos® to aliskiren still showed that the surgically implanted device should still be adopted as a cost-effective treatment.
The cost-effectiveness of Rheos® was comparable to other implantable devices analyzed within the setting of US health care. The base case ICER of $64,400 for Rheos® was close to the ICER for deep brain stimulation for Parkinson’s Disease treatment and within the range of values for implantable cardiac defibrillators (Table 5).51, 52 A United Kingdom-based study showed that left ventricular-assist devices as destination therapy had considerably higher costs per QALY gained.53 Lastly, a Canadian study demonstrated the cost-effectiveness of losartan as an alternative to an atenolol-based regimen using data from the LIFE study.54 Adding or switching certain medications for hypertension treatment remains the first treatment option. For resistant hypertension, Rheos® had a favorable cost-effectiveness profile to other surgically implanted devices.
This analysis of resistant hypertension is consistent with other cost-effectiveness analyses of new medications to manage traditional hypertension.6, 7 Treating resistant hypertension becomes more cost-effective in those with a higher cardiovascular risk including those with diabetes. Contrary to other literature, the cost-effectiveness of Rheos® does not improve beyond 60 years or improve with male gender. Rheos® therapy was more cost-effective in hypertensive females than males. The prediction equations have a higher relative risk of stroke or MI associated with female gender, thus supporting the observation that treating hypertension in females is more cost-effective than treating hypertension in males.9, 10
The use of prediction equations for adverse events rather than event rates based on a clinical trial is a limitation to this study. Framingham-based prediction equations allow a broad look at many endpoints but may lack the accuracy of the device’s performance in its specific patient population. This limitation has been addressed, in part, by also modeling probabilities from ASCOT-BPLA trial. Another limitation is the use of direct costs and utilities from the literature rather than incorporating indirect patient costs. The strengths of this study are the inclusion of heart failure and dialysis dependent renal disease with the common disease states of MI and stroke.
One important question is whether the SBP reductions achieved in the Rheos® phase I trials are maintained throughout the Rheos® Pivotal Trial (clinicaltrials.gov id: NCT00442286). With a finite pool of health care dollars, new effective technology or medications must have reasonable cost to justify wide-spread use. The cost-effectiveness of Rheos® is dependent on the starting SBP, performance of the device and the risks of the target population. Decreases of 30 mmHg3, 4 or more support the use of Rheos® as a cost-effective treatment for resistant hypertension regardless of the initial SBP.
KCY and JCT have been supported in part by National Institutes of Health (NIH) T32HL007937 (to Thomas A Pearson MD PhD MPH, Department of Community and Preventive Medicine, University of Rochester). CGB is supported in part by NIH RO1HL080107. Some of our data were derived from USRDS estimates, therefore: The data reported here have been supplied by the United States Renal Data System (USRDS). The interpretation and reporting of these data are the responsibility of the author(s) and in no way should be seen as an official policy or interpretation of the U.S. government.
Statement of Financial Disclosures:
KCY: Portion of stipend through an NIH training grant (see above).
JCT: Portion of stipend through an NIH training grant (see above).
JDB: Scientific advisory board member for CVRx and clinical research support from CVRx.
KAI: Scientific advisory board member for CVRx and clinical research support from CVRx.