Transcatheter aortic valve implantation (TAVI) has emerged as an efficacious and cost-effective treatment for inoperable patients with severe aortic stenosis. For high-risk patients, existing evidence suggests that TAVI has similar periprocedural mortality and stroke outcomes compared with surgical aortic valve replacement (AVR), with higher incidences of paravalvular leak, permanent pacemaker requirement, and vascular injuries after TAVI and higher incidences of major bleeding after AVR. Regarding the cost-effectiveness of TAVI, a recent systematic review examining its incremental cost-effectiveness ratio in relation to AVR found a paucity of data to demonstrate economic competitiveness of the transcatheter approach.

With the rapid expansion of TAVI procedures and limited health care budgets, additional focus has been directed toward the assessment of cost-effectiveness for TAVI. Minutello et al should be commended on their timely effort to compare TAVI to SAVR using propensity-matched data from the Nationwide Inpatient Sample database. Raw patient data were derived using codes for commercial cases that involved transfemoral TAVI and SAVR, and clinical and cost-related data were identified from a number of patient-level end points and billing codes. Noticeably, clinical outcomes were not graded or clearly defined according to the standardized Valve Academic Research Consortium definitions, and important end points such as paravalvular leak were not assessed. Propensity score matching was applied to match patients according to the number of baseline characteristics, such as treatment location and selected preoperative risk factors.

We have a number of concerns about the study design, analysis, and interpretation presented by Minutello and a few comments regarding some unexpected findings. Perhaps one of the main concerns was the patient selection methodology for the AVR cohort, specifically, the undisclosed proportion of patients who underwent re-do sternotomy AVR, AVR combined with other cardiac procedures, AVR for infective endocarditis, and patients who underwent urgent AVR after conversion from a TAVI procedure or other emergency setting. Such patients are known to be associated with significantly higher mortality and morbidity, and their inclusion would have affected clinical outcomes and related complications and cost analyses. Although specific proportions of such patients were not reported in Minutello’s study, at least 19.1% of patients in the propensity-matched AVR group underwent previous coronary artery bypass graft surgery. This was significantly higher than the 5.5% reported for the unmatched raw patient data and may not be representative of the general population of patients who are considered for AVR procedures. Although their inclusion maximized propensity score matching with the TAVI cohort, it may limit the generalizability of these findings to real-world settings. Similarly, it should also be noted that only transfemoral patients were included for the TAVI arm, and they have been shown to have fewer co-morbidities and superior clinical and cost-effectiveness outcomes compared with transapical patients.

Additionally, although propensity score matching is a useful statistical technique used in cases when randomized trials are not feasible, its effective application is reliant on identifying appropriate prognostic factors to match 2 treatment arms. Arguably, it may be more clinically relevant to match patients according to their EuroSCORE and STS scores than their ethnicity. Indeed, although the calculated c-statistic was high (0.96), its use in propensity score modeling does not inform whether important clinical confounders have been omitted. Additionally, the use of 3:1 propensity matching may be associated with increased bias, and previous studies have recommended 1:1 or 2:1 matching when necessary.

A number of unexpected findings from the study should be briefly discussed. The main finding of the mean cost for TAVI ($181,912 ± $123,535) versus AVR ($196,298 ± $155,033) both demonstrated relatively large standard deviation values, suggesting significant heterogeneity of costs within each cohort. The cost of AVR was significantly higher than previous reports, as noted by the investigators, including a recent study conducted in Virginia that estimated $51,145 ± $31,655 for high-risk patients who underwent isolated, elective AVR. Another unexpected finding was that a significantly higher proportion of patients experienced major bleeding after TAVI compared with AVR (10.9% vs 3.9%, p = 0.004), which contradicted with all previous studies on TAVI. Finally, it was not clear if preprocedural investigations and treatment costs for postprocedural complications, such as major bleeding, vascular complications, and permanent pacemaker insertion, all of which were significantly higher after TAVI, were fully captured in the cost analysis. Given the study methodology, TAVI cost estimations may have been underestimated if additional preprocedural investigations and further treatments and admissions for postprocedural complications were performed in smaller referring centers or rehabilitation centers.