OUP user menu

Lifetime cost-effectiveness of prophylactic implantation of a cardioverter defibrillator in patients with reduced left ventricular systolic function: results of Markov modelling in a European population

Martin R. Cowie, Deborah Marshall, Michael Drummond, Nicole Ferko, Michael Maschio, Matthias Ekman, Luc de Roy, Hein Heidbuchel, Yves Verboven, Frieder Braunschweig, Cecilia Linde, Giuseppe Boriani
DOI: http://dx.doi.org/10.1093/europace/eup068 716-726 First published online: 9 April 2009

Abstract

Aims Current European guidelines recommend prophylactic implantation of cardioverter defibrillators (ICDs) in patients with a reduced left ventricular ejection fraction (LVEF) who are not in NYHA class IV and have reasonable life expectancy. Cost and benefit implications of this recommendation have not been reported from a European perspective.

Methods and results Markov modelling estimated lifetime costs and effects [life years (LY) and quality-adjusted LY (QALY) gained] of prophylactic ICD implantation vs. conventional treatment, among patients with a reduced LVEF. Efficacy was estimated from a meta-analysis of mortality rates in the six primary prevention trials with inclusion criteria matching ACC/AHA/ESC Class I or IIa recommendations. Direct medical costs were estimated using Belgian national references. Costs and effects were discounted at 3 and 1.5% per annum, respectively. Probabilistic sensitivity and scenario analyses estimated the uncertainty around the incremental cost-effectiveness ratio. An ICD implantation increased the lifetime direct costs by €46 413. Estimated mean LY/QALY gained were 1.88/1.57, respectively. Probabilistic analysis estimated mean lifetime cost per QALY gained as €31 717 (95% CI: €19 760–€61 316). Cost-effectiveness was influenced most by ICD efficacy, time to replacement, utility, and patient age at implantation.

Conclusion In a European healthcare setting, prophylactic ICD implantation may be cost-effective if current guidelines for patients with a reduced LVEF are followed.

  • Cost-effectiveness analysis
  • Decision analytic model
  • Europe
  • Prophylactic implantable cardioverter defibrillator
  • Sudden death
  • Left ventricular ejection fraction

Introduction

An implantable cardioverter defibrillator (ICD) can terminate episodes of ventricular tachycardia or fibrillation, thus reducing the risk of sudden death in patients with a range of cardiac disease. European and North American guidelines strongly recommend the prophylactic implantation of ICDs in patients with a reduced left ventricular ejection fraction (LVEF), provided the patient is not in New York Heart Association functional class IV and has a life expectancy longer than 1 year.1,2

Such recommendations have major cost implications for healthcare systems in Europe. Previously, published reports have been based mainly on North American costs and are not directly transferable to the current situation in Europe, or have not included more recent trials and/or used data from individual trials/subgroups rather than meta-analysis of relevant studies, or did not report the incremental cost-effectiveness ratio (ICER).38

In this analysis, we examine the lifetime benefits, costs, and cost-effectiveness of prophylactic implantation of an ICD in patients with a reduced LVEF as recommended in current guidelines, adopting the perspective of the Belgian healthcare system.

Methods

We based our analysis on a model of the cost-effectiveness of ICD therapy (in addition to conventional pharmacological therapy) to prevent sudden cardiac death (SCD) compared with conventional pharmacological therapy, developed by Sanders et al.6 Our model incorporates four main modifications to the Sanders model: (i) the meta-analysis of mortality rates from the six primary prevention trials for patients fulfilling indications for ICD therapy according to current clinical guidelines1,2 (Table 1); (ii) the modelling of specific categories of deaths derived from the clinical trials: SCD, heart failure death (HFD), other cardiac death (OCD), and non-cardiac death (NCD). In those trials, where ‘non-SCDs’ were not further classified into ‘HFDs’ or ‘OCDs’, we used a pooled estimate based on the other studies to allocate such deaths to one or other of those categories; (iii) incorporating Belgian life tables to adjust the non-cardiac mortality rate as the cohort aged; and (iv) using probabilistic sensitivity analysis (PSA), allowing uncertain parameters to vary randomly within predefined distributions and to assess the overall level of uncertainty about our results. Importantly, the model uses country-specific data inputs for the Belgian population to represent the European context of this analysis.9

View this table:
Table 1

Characteristics of patients enrolled in the six studies used in the meta-analysis in this manuscript

CharacteristicsAMIOVIRT14CAT15DEFINITE13MADIT I16MADIT II12SCD-HeFT11
Patients
 Number of patients10310445819612321676
 Age [mean (SD)] (years)59 (11.5)52 (11)58636460
 Sex (Male%)708071928477
 NYHA class (%)NYHA I: 15NYHA II: 65.3NYHA I: 21.6NYHA I: 35NYHA I: 37NYHA II: 70
NYHA II: 63NYHA III: 34.6NYHA II: 57.4NYHA II or III: 65NYHA II: 35NYHA III: 30
NYHA III: 20NYHA III: 21NYHA III: 24
NYHA IV: 5
 Duration of CHF3.2 years3 months (median)2.83 years (mean)24.5 months
 Percentage with NICM (%)10010010047
 Previous MINoYesYes
 CADa7%NoNoYes
 LVEF [mean (%) (SD)]2324 (7)21.42325 (median)
LVEF ≤ 35%LVEF ≤ 30%LVEF ≤ 36%LVEF ≤ 35%LVEF ≤ 30%LVEF ≤ 35%
 NSVTYesYesYesYes
 VFNo
 History of atrial fibrillation (%)24.5915
 Previous HFYes
 Other criteriaExclusionExclusionExclusionExclusionInclusion
Age<18, NYHA IV, cardiac transplantation, syncope, pregnancy, contraindication to amiodarone or defibrillator therapy, and concomitant therapy with a class I antiarrhythmic drugNYHA I and IVPermanent pacemakers, NYHA IV, cardiac transplantation imminent, familial cardiomyopathy associated with sudden death, acute myocarditis, and congenital heart diseaseCardiac arrest or VT causing syncope that was not associated with an MI, MI within the past 3 weeks, and CABG within the past 2 monthsAge > 21
Inclusion
Age: 25–80
Medications at baseline (%)
 β-Blocker51.53.884.9177069
 ACE inhibitors/ARB8596.296.7587096
 Study designMulti-centre randomized trialMulti-centre RCTMulti-centre RCTMulti-centre RCTMulti-centre RCTMulti-centre RCT
 Site10 US centres15 Germany centresUSA30 US centres and 2 European sites71 US centres and 5 European sitesUSA
 ComparatorsbICD vs. amiodaroneICD vs. pharmacologicICD vs. pharmacologicICD vs. pharmacologicICD vs. pharmacologicICD vs. pharmacologic
 Enrolment period1996–20011991–19971998–20021997–20011997–2001
  • NYHA, New York Heart Association; VF, ventricular fibrillation; LVEF, left ventricular ejection fraction; HF, heart failure; NSVT, non-sustained ventricular tachycardia; ACE, angiotensin-converting enzyme; RCT, randomized controlled trial; CHF, chronic heart failure; ICD, implantable cardioverter defibrillator; pharmacologic, optimal pharmacological therapy; ARB, angiotensin receptor blocker; MI, myocardial infarction; NICM, non-ischaemic cardiomyopathy.

  • aCAD: coronary stenosis > 70%.

  • bAll patients were strongly encouraged to receive optimal medical therapy with ACE inhibitors, β-blockers, and potassium-sparing diuretics.

Incremental cost-effectiveness ratios were estimated based on direct medical costs, life years (LY), and quality-adjusted LY (QALY) gained per patient. All analyses were from the healthcare system perspective over the lifetime of patients. Life years gained and QALY gained were discounted at 1.5% per annum and costs at 3% per annum (as recommended by the Belgian Health Care Knowledge Centre).10 We performed both deterministic and probabilistic sensitivity analyses on input variables to test the robustness of the model.

Decision analytic model

The decision analytic Markov model, developed in Microsoft Office Excel 2003™, estimated the costs, effects, and incremental cost-effectiveness of prophylactic ICD implantation for primary prevention of SCD compared with conventional therapy (Figure 1). The model tracks a hypothetical cohort of patients over time using a 1-month cycle length. Treatment effectiveness is captured in the model by tracking progress to the following health states: well, SCD, HFD, OCD, and NCD. In each model cycle, the patient can remain in the same state or transition to another state, except for ICD complications where the patient can only remain for one model cycle. Patients receiving an ICD can experience operative death, implant-related complications, ICD-related complications, and discontinuation of ICD therapy. Pooled estimates for the probability of ICD and implant-related events are modelled. Base-case model inputs are given in Table 2.

Figure 1

Schematic diagram of the model: (A) implantable cardioverter defibrillator arm and (B) control arm.

View this table:
Table 2

Base-case model inputs

Model inputICD therapyConventional therapyData sources
One-month death probability
 Sudden cardiac death0.00150.0042Meta-analysis: AMIOVIRT,14 MADIT I,16 MADIT II,12 SCD-HeFT,11 CAT,15 and DEFINITE13
 Heart failure death0.00290.0029
 Other cardiac death0.00040.0002
 Non-cardiac death0.00240.0031
 All-cause0.00720.0105
Initial implant operative death probability0.003Not applicableAdvisory board
Pooled mean follow-up (months)28AMIOVIRT,14 MADIT I,16 MADIT II,12 SCD-HeFT11 CAT,15 and DEFINITE13
Pooled mean age (years)61.1
Gender (%)79.5 (male)
20.5 (female)
One-month probability of ICD complications (inappropriate shocks)0.0047Not applicableAdvisory board (5.5% per year)
One-month probability of discontinuing ICD after inappropriate shocks0.00065Not applicableAdvisory board (0.76% per year)
Monthly probability of a lead replacement due to failure following initial implant0.0025Not applicableAdvisory board (3% per year)
Monthly probability of a lead replacement due to failure following replacement implant0.0051Not applicableAdvisory board (6% per year)
Probability of a lead infection at initial implant0.01Not applicableAdvisory board
Probability of a lead dislodgement at initial implant0.01Not applicableAdvisory board
Probability of a lead dislodgement at replacement ICD implant0.01Not applicableAdvisory board
Probability of a lead infection at replacement ICD implant0.02Not applicableAdvisory board
ICD device cost (€)Medtronic
 Initial device + leads18 422Not applicable
 Replacement device16 650
ICD typeVVI–ICDNot applicableRIZIV/INAMI22
ICD replacement rateSee distribution in text: 100% replaced over 6.5 yearsNot applicableMedtronic performance report21,33–35,37
ICD procedure cost (€) (2006)
 Initial4650Not applicableRIZIV/INAMI22
 Replacement1436
Lead replacement cost per event (lead failure) (€) (2006)6422Not applicableRIZIV/INAMI22/advisory board
ICD complications cost (€) (2006)
 Lead infection28 410Not applicableRIZIV/INAMI22/advisory board
 Lead dislodgement2551
 Inappropriate shocks120
Monthly long-term inpatient and outpatient cost (€) (2006)244Annemans et al.23 and Muls et al.36
ICD additional monthly follow-up cost (€) (2006)20Not applicableNomenclature and advisory board
Duration of ICD benefitLifetimeNot applicableAssumption
Utility of heart failure patient annual0.85Mark et al.24
Utility of ICD complications state (annual)0.75Not applicableSanders et al.6
Discount rate (%)
 Outcomes1.5Belgian Health Care Knowledge Centre (2006)10
 Costs3.0
  • ICD, implantable cardioverter defibrillator; RIZIV/INAMI: Rijksinstituut voor Ziekte en Invaliditeitsverzekering/Institut National d’Assurance Maladie-Invalidité.

Efficacy of implantable cardioverter defibrillator therapy

Published meta-analyses included patients not eligible for ICD therapy according to current guidelines, and only yielded results for all-cause mortality and cardiac mortality. We conducted a meta-analysis, therefore, to derive pooled estimates for the efficacy of prophylactic ICD mortality compared with conventional medical therapy in patients with left ventricular systolic dysfunction. The meta-analysis included the six randomized controlled trials of primary prevention with ICD therapy compared with conventional medical therapy (AMIOVIRT, CAT, DEFINITE, MADIT I, MADIT II, and SCD-HeFT) for all model endpoints (SCD, HFD, OCD, and NCD).1116 Non-SCDs were classified as HFD or OCD in those clinical trials that did not make this distinction.

A pooled estimate of the relative risk for each type of mortality outcome was calculated using the Mantel–Haenszel method with a fixed effects model (all P-values for the test of heterogeneity >0.39) using the RevMan Version 4.2.8 software.17 The annual risk of death was 8.3 and 11.8% for the ICD and conventional therapy groups, respectively, with SCD being the main cause of cardiac death in the conventional therapy group. The relative efficacy of ICD was assumed to be constant over time. An increase in mortality with age was incorporated using life tables for the Belgium general population, which was used to adjust for the increase in the age- and sex-specific death rate from non-cardiac causes as the cohort of patients aged.9

For most of the trials, patients in the ICD and the conventional treatment arm were prescribed similar medication. An advisory board of cardiologists from across Europe, including Belgian cardiologists, confirmed that most patients who are on ICD and control therapy were likely to be following drug regimens that may include one or more of the following classes of medication: ACE inhibitors, β-blockers, and amiodarone.

Patient population

The model represents patients with chronic heart failure in NYHA class II or III, or prior myocardial infarction with or without heart failure. All patients had a reduced LVEF (≤35%). Patients with NYHA class IV symptoms were not included. The mean age of individuals recruited to the trials ranged between 58 and 64 years, and the majority (>70%) were male. The proportion of patients in NYHA class II or III ranged from 59% in MADIT II to 100% in SCD-HeFT.

Complications of implantable cardioverter defibrillator therapy

For the base-case, the effectiveness of single-chamber ICD therapy (VVI–ICD) was modelled, consistent with Belgian reimbursement policy.18 Complications associated with implantation, device technology, and the delivery of therapy were included in the model. Estimates of the proportion of patients experiencing such complications (infection, lead dislodgement, shock-related adverse events, and lead failures) were provided by the advisory board whose estimates were informed by the published literature on ICD-related complications (Table 2).11,13,19,20

Implantable cardioverter defibrillator device replacement was incorporated in the model using a 6-monthly device survival probability distribution, based on data for the GEM II VR (single-chamber ICD), reported in the 2006 Medtronic CRDM Product Performance Report.21 It was conservatively assumed that all patients would require an ICD replacement after a maximum of 6.5 years. Exact replacement time depends upon the type of defibrillator, battery technology, stimulation and shock energy requirements, the frequency of capacitor reformation, device diagnostics used, lifestyle issues, and other factors. The effect of longer ICD longevity with new generation devices was tested in sensitivity analysis.

Healthcare resource use

The advisory board of cardiologists also provided consensus advice on the model structure, the assumptions behind the pooling of the clinical trial efficacy data, and provided estimates of the frequency of resource use associated with ICD-implant and ICD-related adverse events.

Costs

The base-case analysis applies an initial device cost for a single-chamber ICD of €16 650 + €1772 for the leads and €16 650 for an ICD replacement device, based on the official reimbursement price in Belgium.

The procedural costs of an initial ICD implant, subsequent revision or replacement, and ICD-related complications (infection, dislodgement, and inappropriate shocks) were derived from the ‘Technical Cell’ of the federal public service of Public Health and of the Health Insurance [Institut National d’Assurance Maladie-Invalidité (INAMI)/Rijksinstituut voor Ziekte en Invaliditeitsverzekering (RIZIV)]22 and include physician fees, medication, hospital stay, and laboratory utilization.

Typical costs unrelated to ICD therapy include follow-up medical costs (inpatient and outpatient). For the base-case, we used cost data from a Belgian-specific heart failure cost study.23

Quality-of-life

A utility score of 0.85 was assigned to the ICD and conventional therapy groups, on the scale where 0.0 equals death and 1.0 corresponds to perfect health, based on the data collected using the time trade-off technique in the SCD-HeFT study.24 It was assumed that quality-of-life did not change as a result of the implantation of an ICD.24,25 However, patients who experience an ICD-related complication (e.g. lead infection, inappropriate shocks, and dislodgement) received a quality-of-life decrement equivalent to 3.5 days per month, making their utility 0.75 [0.85 × (26.5/30) = 0.75], based on the method adopted by Sanders et al.6

Sensitivity analyses

One-way deterministic sensitivity analyses were performed to assess whether the results were affected by changes in the assumptions. The parameters examined in sensitivity analyses were clinical (e.g. conventional therapy mortality, ICD efficacy, replacement rates, and age at the time of ICD implantation), cost (ICD device costs and long-term costs of chronic heart failure), utility parameters (ICD-related and conventional therapy-related utilities), and discount rates.

An additional analysis was conducted to investigate the potential impact on the cost-effectiveness of ICD therapy of variations in the absolute risk of SCD and variation in the competing risks of death. For this analysis, we varied the annual all-cause mortality rate and the ratio of HFD to SCD, a similar method to a previous analysis.26 The annual all-cause mortality rates considered in these analyses were: 2.5, 5.0, 10.0, 15.0, 20.0, 25.0, and 30.0%. For each of these seven mortality rates, five different ratios of HFD to SCD were considered: 1:10, 1:3, 2:3, 1:1, and 3:2.

Probabilistic sensitivity analysis with 1000 simulations was undertaken to assess the joint uncertainty of costs and effects. In the PSA, the uncertainty of the cost-effectiveness estimate was explored by defining probability distributions for the key model variables rather than using deterministic values. These variables include the transition probabilities between different states in the model as well as the health utilities associated with different states (Figure 1).

Results

Efficacy of implantable cardioverter defibrillator therapy

The meta-analysis demonstrated a significant reduction in sudden cardiac mortality (and all-cause mortality) associated with the addition of ICD therapy to conventional pharmacological therapy with an RR of 0.37 (95% CI: 0.28–0.48) for SCD and RR of 0.72 (95% CI: 0.64–0.82) for all-cause mortality. Tests for heterogeneity were non-significant.

Deterministic base-case analysis

Prophylactic implantation of an ICD improved life expectancy relative to conventional therapy alone by 2.22 years (discounted, 1.88 years) on average for each patient (Table 3). The ICD cohort experienced lower all-cause mortality throughout the model time horizon (see Supplemental material online, Figure S1). There were ∼50% fewer SCD in those implanted with an ICD device compared with those on conventional therapy, with the benefit of ICD continuing throughout the time horizon (see Supplemental material online, Figure S1). Similarly, ICD improved QALY relative to conventional therapy, with a discounted value of 1.57 years (undiscounted, 1.86 years).

View this table:
Table 3

Life years, quality-adjusted life years, costs, and incremental cost effectiveness ratios for implantable cardioverter defibrillator compared with conventional therapy

InterventionLYQALYCost (€)ICER (€/LY)ICER (€/QALY)
Discounted
 ICD8.587.2764 60024 75129 530
 Conventional therapy6.715.7018 187
 Difference1.881.5746 413
Undiscounted
 ICD9.528.0675 26224 27129 009
 Conventional therapy7.306.2021 366
 Difference2.221.8653 896
  • ICD, implantable cardioverter defibrillator; ICER, incremental cost-effectiveness ratio; LY, life years; QALY, quality-adjusted life years.

On average, 0.51 inappropriate shocks per individual were experienced over the lifetime of the cohort. There was an average of 0.38 lead failures requiring lead replacements per individual over the lifetime of the cohort, based on advisory board estimates of 3 and 6% lead failure rate per year after initial implant and replacement implant, respectively. An average of 0.04 lead dislodgement and infections per individual were predicted over the lifetime of the cohort. The average number of ICD replacements over an individual’s lifetime was estimated to be 1.03.

Prophylactic implantation of an ICD led to higher costs compared with conventional therapy (an additional €46 413). The majority of the ICD cost is the device and implant procedure cost (∼70% per person on average); the costs of ICD-related complications are relatively small.

In the deterministic base-case analysis, therefore, ICD therapy costs more and is more effective than conventional therapy alone. For patients with a mean starting age of 61 years, the estimated cost-effectiveness of ICD therapy compared with conventional therapy is €24 751 per LY gained and €29 530 per QALY gained (Table 3).

Validation of the model

We used inputs for costs, efficacy, and utility reported in the Sanders model to validate our model.6 We also compared the model-predicted mortality among patients receiving conventional therapy with that reported in the clinical trials using the same follow-up time period to ensure our extrapolations and assumptions of constant monthly mortality were validated.

Compared with the empiric trial period, our Markov model predicted mortality rates associated with conventional therapy within 5% relative to the mortality rates reported in the individual trials over the same time period. For the ICD strategy, our model matched the trial results within 5% of the reported mortality rates except for the DEFINITE trial, in which our estimated mortality rate was 11.2% at 29 months, whereas the actual rate was 12% (a 7% relative difference).

When the input values in our model were changed to mirror the values reported by Sanders et al.6 for six separate clinical trials, our Markov model predicted ICERs of ICD therapy compared with conventional therapy that were no more than 7% different from those reported by Sanders et al.6

Deterministic sensitivity analysis

Overall, the results were most sensitive to assumptions about the efficacy of ICD therapy using a one-way deterministic sensitivity analysis (Figure 2). The base-case analysis considered an ICD efficacy estimate based on pooled estimates of ICD and control mortality rates, respectively. The sensitivity analysis examined the effect of differences in the ICD efficacy using the data from the MADIT I (‘high’ efficacy; HR = 0.46)16 and SCD-HeFT trials (‘low’ efficacy; HR = 0.77).11 The other parameters that most influenced the ICERs were replacement period, utility, age at the time of ICD implantation, and discount rates. The results were least sensitive to varying the ICD device cost, conventional therapy mortality, and long-term costs. Nonetheless, for all values tested in the deterministic sensitivity analysis, the ICERs are in a range that is commonly considered economically attractive (ranging from €13 727 to €37 363).

Figure 2

Tornado diagram showing the results of deterministic one-way sensitivity analyses, comparing implantable cardioverter defibrillator and conventional therapy relative to base incremental cost-effectiveness ratio of €29 530 per quality-adjusted life year. The dotted line is a reference line, indicating the value of the incremental cost per quality-adjusted life year in the base-case. QALY, quality-adjusted life years; QOL, quality-of-life. For input variables, please see Table S1 (included in Supplemental material section).

The results of the sensitivity analyses to investigate the impact of variation on the absolute risk of SCD and competing risks of death showed that ICD therapy is more cost-effective (lower ICER) as the ratio of HFD to SCD decreases, i.e. as SCD is the main cause of death. There is a U-shaped relationship between the ICER and annual mortality for the conventional therapy group. High ICERs (less cost-effective) occur when the annual mortality rate in the conventional therapy group is low. The ICER rapidly decreases (becomes more cost-effective) as the annual rate of mortality increases, until the lowest point is reached at ∼12.5% annual mortality. The ICER then very gradually increases (less cost-effective) as the mortality rate rises further (Figure 3).

Figure 3

Cost-effectiveness (€/quality-adjusted life year, discounted) of implantable cardioverter defibrillator therapy for various combinations of annual mortality rate for the control therapy cohort and ratios of heart failure death to sudden cardiac death. HFD, heart failure death; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life year; SCD, sudden cardiac death.

Probabilistic sensitivity analysis

In the PSA, the cost-effectiveness of ICD compared with conventional therapy was €31 717 per QALY gained (95% CI: €19 760–€61 316 per QALY). The cost-effectiveness acceptability curve (Figure 4) shows the probability (on the y-axis) that the ICD therapy is cost-effective compared with conventional therapy for a range (on the x-axis) of maximum monetary values that a decision-maker might be willing to pay (WTP) per QALY.27,28 The probability that ICD is cost-effective is estimated at 85% for a WTP of €40 000 per QALY, at 93% for a WTP of €50 000 per QALY, and at 97% for a WTP of €60 000 per QALY.

Figure 4

Cost-effectiveness acceptability curve for implantable cardioverter defibrillator therapy in addition to conventional pharmacological therapy compared with conventional pharmacological therapy alone.

Discussion

We have found that under most assumptions in a European healthcare setting, the prophylactic implantation of an ICD has a cost-effectiveness ratio below €50 000 per QALY in patients with an increased risk of SCD because of a reduced LVEF (≤35%), applying Belgian healthcare resource utilization rates and costs.

Based on the relevant randomized clinical trials, we modelled the lifetime benefit of an ICD implantation and estimated that this adds 2.22 undiscounted years of life. This is substantial and compares favourably with other interventions in such patients. Also, the incremental cost-effectiveness of ICD implantation in appropriate patients is similar to that of many interventions currently considered good value for money for patients with cardiac disease.2931

A lifetime horizon is relevant to policymakers and is preferred to within-trial analyses, which are likely to lead to poor estimates of a lifetime value as incremental costs are largely the result of the initial intervention, but benefits are likely to accrue well beyond the duration of the clinical trial.

Our results show less heterogeneity than the analysis from Sanders et al.,6 due to the exclusion of studies that included patients undergoing coronary artery bypass surgery or in the acute phase of myocardial infarction. We found no evidence of meaningful variation in the efficacy of ICD therapy in the trials included in this study, but acknowledge that cost-effectiveness is likely to be different in other patient populations. Cost-effectiveness was most sensitive to ICD efficacy, replacement period, utility, age at the time of ICD implantation, and discount rates.

Our estimates are in line with the most recent results of Sanders et al., 6 although the estimates of cost-effectiveness from North America vary widely.3,6,7

The implementation of the current guidelines for prophylactic implantation of ICDs1 is far from complete across Europe, and implantation rates vary widely across Europe.32 Moreover, Europeans are much less likely to be given this therapy than patients living in North America. Although full implementation of this therapy has potentially high cost and organizational implications for European healthcare systems, this analysis indicates that such expenditure represents good value for money.

Our study has several limitations. We projected costs and benefits to a lifetime horizon, but the results were robust to most reasonable changes in the assumptions we employed. The difference between the central estimate of the ICER provided by the deterministic analysis compared with the probabilistic analysis was reassuringly small (€29 530 vs. €31 717 per QALY gained). International guidelines provide rather broad brush recommendations as to which patients should be considered for ICD therapy, with little discussion of detailed patient characteristics or specific device technology. If clinicians extend the use to patients not included in the randomized trials, then the value for money of ICD implantation is unknown. Our analysis is limited to single-chamber ICDs and cannot be extrapolated to more sophisticated devices, including those with the additional capability of cardiac resynchronization.

A newer generation of devices with improved battery capacity as well as newer lead and circuitry technology further improves the expected device longevity, and it is likely that the cost-effectiveness of this therapy will further improve.

Funding

This work was supported by Medtronic International Trading Sárl and Medtronic Bakken Research Center B.V.

Acknowledgements

In November 2006, a European advisory board met and provided guidance and consensus from clinical advisors (cardiologists) on clinical and economic model inputs. The advisory board included cardiologists from Spain, Belgium, Italy, Germany, Sweden, and the UK. The advisors were Dr M.R.C., Dr Juan Delgado, Dr F.B., Dr Uwe Wiegand, Dr Marco Metra and Dr L.R. In addition, a second Belgian advisor was consulted for expert input (Dr H.H.). Editorial assistance was provided by Dr. Gordon Brooker, Phocus Services Ltd, Basel, Switzerland. i3 Innovus Division of Ingenix Pharmaceutical Services (UK) Ltd, Maidenhead, UK and i3 Innovus, Burlington, Ontario, Canada are the institutions where the work was performed.

Conflict of interest: M.R.C. has a consultancy agreement with Medtronic and receives, on behalf of his University, unconditional research grants from Medtronic. D.M. and M.D. are consultants to i3 Innovus, a global health economics research company that receives payment from Medtronic and other device and pharmaceutical manufacturers for research. N.F. and M.M. are employed by i3 Innovus, a division of i3 Canada, Inc. Y.V. was a former employee of Medtronic Bakken Research Center, B.V. M.E.’s work on this project was funded by a grant from Medtronic. L.R. has received consultancy fees from Medtronic. H.H. receives, on behalf of his University, unconditional research grants from Medtronic and Boston Scientific; he is a Member of the Physician Advisory Board of St Jude Medical and is Co-ordinating Clinical Investigator of the Biotronik-sponsored EuroEco trial. F.B. receives research funding from Medtronic, serves as a consultant for Medtronic and Boston Scientific, and has received speaker’s honoraria from several device manufacturers. C.L. has received research grants, speaker’s honoraria, and consulting fees from Medtronic and speaker’s honoraria and consulting fees from St Jude Medical.

References

View Abstract