Background
Despite treatment advances over the past decade, heart failure (HF) remains a major burden on patients and healthcare systems. HF is the leading cause of hospitalization for patients who are older than 65 years, with an average length of stay of 13 days and a re-hospitalization rate of approximately 20% in the UK [
1,
2]. In Europe, HF accounts for between 1.1 and 2.0% of the total health expenditure in healthcare [
3].
The prevalence of HF in the general population is approximately 3.9% [
4] and the annual incidence varies between 1.0 and 2.5 cases per 1000 population [
5,
6]. The prognosis for HF patients remains generally poor. In the UK National Heart Failure Audit, the overall in-hospital and 1-year mortality of HF was 9.4 and 24.6%, respectively. In patients with reduced left ventricular ejection fraction (LVEF), 1-year mortality was approximately 15%, increased to 22–25% at 3 years, and reached 72% at 10 years [
7,
8]. Patients with HF have a reduced quality of life and are often limited in their activities of daily living because of symptoms, such as dyspnea and fatigue. The quality of life of patients with HF is lower than that in those with angina pectoris, breast cancer, or diabetes mellitus [
9]. The quality of life is reduced as symptoms worsen (New York Heart Association [NYHA] class) [
10].
Current treatment options include pharmacotherapy and cardiac resynchronization therapy (CRT). However, CRT is not suitable for all patients [
1]. In a UK study, only approximately 10% of patients presenting with HF were eligible for CRT [
11]. Of those, between 25 and 35% of patients do not appear to respond to CRT [
12].
Implantable cardioverter defibrillators only have a preventive effect on sudden cardiac death and do not improve symptoms or slow the progression of HF. Evidence of the beneficial use of implantable cardioverter defibrillators is less in patients with an LVEF of 30–35%, as well as in those with HF from a non-ischemic etiology [
1,
13]. The use of ventricular assist devices and heart transplants is limited in European countries and they cannot be considered as a treatment option for the broad HF group of patients [
14‐
16].
The Barostim neo™ (CVRx Inc., Minneapolis, MN, USA) is a CE-marked treatment option for HF patients. This device is indicated for patients with heart failure NYHA class III with an LVEF ≤35%. The Barostim neo™ is an implantable medical device, which elicits the body’s natural baroreflex through stimulation of the carotid baroreceptors. This therapy is expected to restore the sympatho-vagal balance, which is a central physiological mechanism and therapeutic target in HF while preserving blood pressure and renal function. This therapy reduces the workload of the heart by decreasing arterial resistance, thereby improving the heart’s ability to pump blood to the tissues. The mechanism of action is not linked to dyssynchrony. Thus, stimulation of the baroreflex can help patients who do not have an indication for or have not fully responded to CRT. Therefore, the Barostim neo™ is both compatible with and complementary to CRT.
In a randomized controlled trial (
n = 146; NCT01471860 and NCT01720160) versus guideline-directed medical therapy, BAT demonstrated improvements in the 6-min walk test (
p = 0.004), quality-of-life score (
p < 0.001), and NYHA class at 6 months (
p = 0.002). BAT was also associated with a trend towards a reduction in days of hospitalization due to HF (
p = 0.08) [
17]. In a follow-up publication by Zile et al. [
18], the clinical effect of BAT was more profound in patients without CRT. There was a significant reduction in the number of hospitalizations (reduction by 0.53 ± 0.2,
p < 0.05) and days of hospitalization due to HF (reduction by 8.89 ± 4.0 days,
p < 0.05) in the BAT arm compared with the 6 months prior to enrollment, whereas no difference was found in the control arm. However, a reduction in resource use at 6 months compared with the control arm was marginal (
p = 0.08 for hospitalizations and
p = 0.09 for days of hospitalization). Despite promising clinical data, economic consequences of the use of BAT have not been evaluated.
Therefore, the objective of this study was to determine the cost utility of BAT as a treatment in adult patients with advanced HF and reduced ejection fraction (defined as LVEF < 35% and NYHA class III) compared with the relevant pharmacological multi-drug treatment in a German population from a healthcare payer perspective over a lifetime horizon. The study focused on the population of patients without CRT, with the most profound clinical effects of the BAT.
Discussion
Our study evaluated potential long-term clinical and economic consequences of BAT compared with optimal medical treatment over a lifetime in a cohort of patients aged 63 years with NYHA class III at the start of the treatment. We showed that BAT leads to additional costs to a health care system (on average €33,185). However, BAT also could lead to additional benefits in terms of survival (an average additional 1.78 life-years) or quality-adjusted survival (an average additional 1.19 QALYs). These results indicate that BAT can be cost-effective because the resulting ICER is below the typical willingness-to-pay threshold in European countries (€35,000/QALY). In the probabilistic sensitivity analysis, the most preferred analysis to simultaneously address all data-related uncertainty in the model, BAT had an approximately 60% chance of being cost-effective at a willingness-to-pay threshold of €35,000/QALY. The potential correlation between input parameters was not captured, which might be associated with a limitation in interpreting the results of the PSA.
However, these promising findings need to be evaluated in the context of existing limitations of our analysis. First, data on the effect of BAT on mortality were obtained via application of a risk prediction model to individual patient-level data. They were not derived directly from the randomized controlled trial because these data are not yet available. Risk prediction models are used in the field of epidemiology to provide estimations of the absolute probabilities of the occurrence of a certain outcome in an individual with a specific set of characteristics or predictors. Predictors may vary from age and sex to advanced diagnostic markers, such as natriuretic peptides or the genetic profile [
26]. Models are usually developed from large datasets using multivariable regression modeling [
27]. The ultimate aim of a model is to assist clinicians and patients in decision making regarding the further management of HF, as well as to facilitate timely research [
28,
29]. To be applied correctly, risk prediction models should provide accurate and validated probability estimates for the outcomes of interest in a targeted population. This can be ensured by the internal and external validation of the model by means of statistical methods [
26].
There are several other risk scores models for predicting survival for patients with HF. These include the Seattle heart failure model [
30], the Heart Failure Survival Score [
31], the PACE Risk Score [
32], and SHOCKED Predictors [
33]. A systematic literature review identified 20 different risk prediction models [
34]. Each model uses a single cohort of patients and thus has more limited generalizability to other populations. Additionally, each model’s development is from a limited cohort size, compromising the ability to truly quantify the best risk prediction model. Because of the wide variety of different studies included, with a global representation, the findings in the MAGGIC meta-analysis appear to be inherently generalizable to a broad spectrum of current and future patients. The risk score, developed in the MAGGIC meta-analysis of a large dataset of 30 cohort studies, provides a uniquely robust and generalizable tool to quantify individual patients’ prognosis in HF.
A second limitation is that data on the effect of BAT on the rate of hospitalizations were derived from the BAT Randomized Controlled Trial, in which a trend towards a between-group difference in the number of hospitalizations was observed, but this was not statistically significant [
18]. The original study did not have sufficient power to demonstrate statistical significance in this outcome in a subgroup of patients with no CRT. This outcome shall be evaluated in a further study of BAT. However, the use of the data is justifiable in our exploratory analysis, focusing on understanding the potential of BAT as a treatment option for advanced HF. Moreover, in the sensitivity analysis, an increase in relative risk to 1.05 (indicating that BAT will lead to an increased risk of hospitalizations) only led to a marginal increase in the ICER over the willingness-to-pay threshold (ICER was €37,172/QALY).