Editorial: The economical challenge in the treatment of chronic heart failure: is primary prophylactic implantable cardioverter defibrillator therapy cost-effective in Europe?

We read with interest the editorial commenting on our modelling of the cost-effectiveness of primary prophylactic implantable cardioverter defibrillator (ICD) implantation according to the current European guidelines in a European population, 1 claiming that our analysis overstated the likely cost-effectiveness of such therapy due to inappropriate assumptions around key determinants of cost and efficacy. Meine et al. mistakenly suggest that the results of our analysis 2 were materially different from those from the MADIT II 3 and SCD-HeFT 4 trials. Our analysis had a life-time model time horizon (as recommended in most recent national and international guidelines for health economic analyses 5–7 ), whereas the two US cost-effectiveness analyses took a (rather arbitrary) 12-year time horizon. If we had taken this time horizon, the incremental cost-effectiveness ratio (ICER) at that point in our analysis would have been €43 752 ($61 253 per QALY at exchange rate of €1: $1.40), similar to the MADIT II ($78 600–$114 000 per life year) and SCD-HeFT analyses ($58 510 per QALY). The more appropriate life-time horizon in our analysis gave an ICER of €31 717 per QALY gained ($44 404). The editorial also suggests that there is a lack of a treatment effect for ICDs in the NYHA class III, based on subgroup analysis of the SCD-HeFT trial. However, it should be remembered that DEFINITE found a greater benefit of ICD therapy among patients with NYHA class III than among patients with class II; and in MADIT-II, there were no significant differences in the survival benefit in subgroup analyses stratified according to NYHA class. 8 , 9 A meta-analysis of all primary prevention studies also failed to confirm a contribution of NYHA class (or mean age or mean left ventricular ejection fraction) to statistical heterogeneity in all-cause mortality in the relevant trials. 10 The editorial notes that there are differences between healthcare systems in Europe, and suggests that Belgium is not representative of Europe in terms of cost. We agree that it is not straightforward to compare reimbursement rates across countries and/or regions given the complexity and diversity in healthcare financing systems, including a variety of diagnosis-related group (DRG) classifications and special arrangements for funding medical devices. Some countries cluster ICD and CRT-D devices in the same DRGs, others do not; some countries have multiple DRGs corresponding to patterns of disease severity or patient co-morbidities, others do not; and some DRG systems incorporate capital equipment and facility costs, others do not. We believe that overall, the Belgium costs adequately represent most European countries and are certainly likely to be more relevant than North American models. Is the battery longevity in our model similar to real-world observed longevity? The editorial cites a study of ICD longevity in 153 patients implanted from 2000 through 2002. 11 When we examined this study, we found evidence to support the longevity in our model: eight patients had a Medtronic GEM II VR device, and seven of these patients had not had a replacement after 5.4-year follow-up; and the longevity for three Medtronic GEM devices was 6.75 years. A recent study in Belgium with a mixed sample of 143 Medtronic ICD/CRT-D devices replaced July 2007–September 2008 confirmed an average longevity of 6.19 years (P. Galloo, personal communication). Meine et al. incorrectly stated that we assumed a longevity of 6.5 years—we used a survival distribution for GEM II VR devices with a mean of 6.3 years, truncated at the right-hand side of the distribution at 6.5 years (the maximum follow-up available at that time). The editorial also suggests that we under-estimated mortality in the ICD group. Eight-year follow-up from MADIT II (recently reported at the Heart Rhythm Society 2009 meeting 12 ) suggested that the survival of ICD patients is better than previously estimated: 8-year survival was 55% in the ICD group, compared with 39% in the non-ICD group ( P < 0.001). Similarly, 8-year survival rate in a European ICD registry, perhaps a better reflection of real-life practice, has recently been reported to be 64%. 13 This compares with our modelled survival rate at 8 years of 48% in the ICD group. We also disagree that the age of the patients in our model is inappropriate. The starting age in our model was 61 years, and this is not dissimilar to the average age at implant in Belgium, Spain, Denmark, and Italy at 62.8, 61.5, 62.3, and 66.2 years respectively. 13–16 The impact of age on cost-effectiveness is, in any case, not as clear cut as the editorial suggests, with another recent cost-effectiveness analysis reporting similar ICERs for ICD therapy in patients aged above and below the much greater age of 75 years. 17 The editorial also states that the modelled mortality rate in the non-ICD treatment group is significantly higher than that seen in the two trials (DEFINITE and CAT). However, these two studies were included in our meta-analysis, so their data already are taken account of in calculating the weighted average mortality rate for conventional therapy. Lastly, the editorial states that the complication rates in our model (provided by expert opinion) may be lower than those seen in the real world where ICDs are often implanted by non-electrophysiologists, based on a recent study in the USA. 18 In that study, major complications at the time of implant occurred at rates of 0.7–1.3 and 1.3–2.7%, depending on physician speciality, for single- and dual-chamber ICDs, respectively. Although the classification of major and minor complications in this study is not directly comparable to our analysis, these complication rates are similar to those used in our analysis. We continue to contend that over a patient's life-time, the prophylactic implantation of an ICD in those with reduced left ventricular systolic function, according to the current European guidelines, is likely to be cost-effective. The assumptions in our modelling were not at all unreasonable, as suggested by Meine et al. We agree with the editorial, however, that with technicological advances and increasingly skilled implanters, this therapy is likely to become even more cost-effective with time.