The clinical and economic impact of extended battery longevity of a substernal extravascular implantable cardioverter defibrillator

We read with interest the article by Knight et al.1 evaluating extended battery longevity in implantable cardioverter defibrillators (ICDs). We congratulate the authors on an interesting study but would like to highlight areas where further clarification would be valued. This letter is supported by a completed Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist2 (Supporting Information S1: Material 1). First, the cycle length of the Markov model was not reported. Previous ICD economic models have used 1-month cycles.3-5 However, this approach could be suboptimal in the given study due to the model structure; the distinct “generator replacement” and “replacement complication” states suggest that people needing ‘generator replacement’ do not experience complications. Therefore, if a 1-month cycle was used, people in the model would be immune to complications throughout the month. Secondly, the model perspective was not justified for any country, which is unusual given it differed for each (excluding South Korea, where it was not reported). Generally, it is good practice to justify all decisions/assumptions in an economic analysis plan/protocol; such a plan was not disclosed. Thirdly, it is unclear how the fixed values for device longevity were operationalized. One approach would be to assume that all devices fail at a specific time point and, therefore, that replacement costs are only incurred in specific model cycles. This is suboptimal if longevity is a major driver of cost-effectiveness because it would introduce bias in favor of ICDs with a better life expectancy; fewer people would live to the time of replacement to incur the costs of treatment. Another approach would be to apply a cycle-specific event rate to all people alive in each cycle. This approach may be overly simplistic because it assumes that the rate of ICD replacement is time-invariant. In contrast, the most recent National Institute for Health and Care Excellence appraisal of ICDs4 supports that ICD failure rates vary over time. Moreover, the subcutaneous ICD (S-ICD) is supported by 15 years of safety data,6-12 versus 6 months for the extravascular ICD (EV-ICD).6, 13 To accommodate this, the authors have extrapolated the EV-ICD data. Consequently, there is some uncertainty in the model outputs. This has not been acknowledged by the authors or accounted for via probabilistic sensitivity analysis. Finally, there may be issues relating to the model costings. While the actual implant/explant procedures are likely similar, the assumption that S-ICDs and EV-ICDs have common costs could be problematic if there are differences in acquisition/treatment-specific costs. This is suggested in the discussion and should, therefore, have been explicitly included in the model. Furthermore, it is unclear what “ICD implant costs” encompasses. The substernal versus subcutaneous positioning of the lead may impact both short- and long-term complication rates9, 10, 12, 13: costs that were not factored into the model. Overall, the level of reporting and justification in the original article is limited. We cannot state the possible magnitude of impact but suggest a correctly parameterized model would produce different results. Boston Scientific commissioned YHEC to undertake an independent critique of the methods in the Knight et al. study and to prepare this letter to the editor. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.