Publications addressing the surgery versus TAVI debate for severe aortic stenosis remain frequent. The growing interest in this topic, competition to dominate specific patient subgroups, and aggregation of data, experience, and follow-up have strengthened the quality of studies—even those reliant on observational evidence from well-constructed registries. 

This particular study has sparked significant discussion on social media and in specialized press for contradicting the prevailing notion that TAVI should at least be noninferior, if not superior, to surgical alternatives for any patient group. Critical voices quickly emerged, denouncing a study of surprising robustness that demands attention. As cardiovascular professionals, we must examine its strengths and weaknesses objectively, irrespective of its conclusions. 

The study used data from the Austrian public national health system, covering 98% of the population. Data were retrieved using MEL (Medizinische Einzelleistung, German coding system) and ICD (International Classification of Diseases) codes for pre-procedural morbidity, applied treatments, and follow-up outcomes, including reintervention, pacemaker implantation, stroke, myocardial infarction, and heart failure. Patients requiring concomitant procedures (including coronary revascularization), mechanical surgical prostheses, or percutaneous revascularization within four months before or after the procedure were excluded. Follow-up extended up to 12 years, with patients included between 2010 and 2020. 

The cohort comprised 18,882 patients, 11,749 undergoing TAVI and 7,133 undergoing surgery. Surgery dominated in 2010 but trends reversed in 2019, with TAVI experiencing exponential growth thereafter. Patients were categorized into two age groups to assess the impact of age—a critical factor in treatment assignment. In the 65–75 age group (7,575 patients), 85.6% underwent surgery and 14.4% TAVI. Among those >75 years (11,307 patients), 46.6% underwent surgery and 53.4% TAVI. Analyses compared raw and propensity score-adjusted survival data, adjusting for 15 preoperative variables. 

Overall, all-cause mortality was 50% higher in the TAVI group than in the surgical group (HR = 1.5; p < .001) over a median follow-up of 5.8 years. This result held after propensity score adjustment. Survival differences became statistically significant two years post-procedure. Median estimated survival was 8.8 years for surgery versus 5 years for TAVI. Subgroup analyses showed nearly 2.5 times higher mortality in the 65–75 age group with TAVI (p < .001), regardless of propensity score adjustment. For patients >75 years, excess mortality was 1.3 times higher (p < .001). 

To explain this higher long-term mortality in the TAVI group, post-procedural morbidity was analyzed. Pacemaker implantation was required in 11.2% of TAVI patients versus 4.5% of surgical patients. Excluding those who died within one month, pacemaker implantation significantly increased follow-up mortality (HR = 1.1; p = .02). Of 287 patients requiring reintervention, 232 had surgery, and 55 had TAVI. However, reintervention rates were not significantly different between groups, nor were new myocardial infarction, stroke, or heart failure incidences. 

The authors concluded that, based on the results of their national registry, the choice of TAVI as a therapeutic alternative was associated with higher all-cause mortality rates compared to the surgical option, particularly in the subgroup of patients aged 65 to 75 years. 

COMMENTARY:

AUTHEARTVISIT has caused a true upheaval in TAVI’s aspirations to reduce the age of indication. With its strengths and weaknesses, it consolidates one of the most extensive published real-world datasets to date. 

Nevertheless, the AUTHEARTVISIT study depends on observational evidence and, like many other works of its kind, shares the same strengths and weaknesses. On one hand, more than an observational study, it is a registry that incorporates the advantages of multicenter design, real-world data presentation, and a nearly population-wide sample, including treatment candidates from almost the entire country. While the idiosyncrasy of the healthcare system and type of funding are important in such analyses, Austria’s European healthcare system has similarities to ours, making the practical outcomes—including patient allocation to one therapeutic alternative or another—more comparable to ours than, for example, the American system. Being a registry, it includes historical data from over a decade ago, where although no drastic changes occurred, older devices and less development in the TAVI sphere could unequally penalize groups. Data recovery from a health system registry, unlike institutional databases or society records like our RECC or the STS, often incurs inaccuracies and provides more limited data. Thus, the primary value of this study lies in the crude results provided, as propensity score adjustment is fundamentally flawed. Evidence of this is its inability to alter trends shown by crude data in groups likely non-comparable, as no adjustment for surgical risk—so influential during the study’s period in allocating patients to one therapeutic alternative or another—was performed. Consequently, if propensity score adjustment is already a suboptimal method for sample adjustment in retrospective studies, it becomes utterly useless when lacking sufficient representative variable volumes. Moreover, this defect compromises other subanalyses attempted in the study, but due to their deficiency, they do not warrant mention, such as the Cox survival analysis. 

In comparison, the alternative evidence lies in powerful, hyper-funded randomized clinical trials (RCTs). These paradigms of evidence-based medicine, which should guide such controversies, also possess a series of shortcomings. Their primary strength is randomization, the best method to achieve group comparability, far superior to the flawed propensity score adjustment. However, unlike the presented registry, where patients with “pure” valvular disease are studied, the aggregation of concomitant procedures unequally affects the groups, introducing confusion into the results. This is because such trials pursue, and achieve, rapid recruitments with smaller volumes than registries, but with sufficient statistical power and no impact of biases like historical cohort utilization. Being experiments, they exhibit meticulously calculated designs with inclusion criteria that limit external validity and generalizability to the population we treat. In fact, many low-risk trials randomize patients who could easily be treated with either therapeutic alternative. And while it is widely known that surgery cannot achieve everything…neither can TAVI. Thus, well-known suboptimal cases exist where the percutaneous option may have been forced, presumably in favor of less invasiveness (first-degree atrioventricular block, right bundle branch block, irregular distribution of annular or outflow tract calcification, predominant aortic insufficiency, borderline vascular access, presence of other valvular diseases, or coronary disease), while a surgical option could have been offered. Since TAVI procedures surpassed surgical ones more than five years ago, the reverse is far less common. Finally, RCTs are, by definition, multicenter, though participants are also often selected by volume, experience, and outcomes, providing less generalizable evidence compared to real-world registry results. It is desirable that the distorted presentation of results be progressively abandoned, offering transparent outcomes like those in observational studies, moving away from non-inferiority analyses and composite events. 

In summary, the TAVI versus surgery debate is reaching new horizons. It has now extended to two irreconcilable factions: the “RCT advocates,” predominant in cardiology, who champion randomized evidence; and the “registry advocates,” who praise real-world data evidence. Perhaps both alternatives, well executed, are valuable. But what the AUTHEARTVISIT study tells us is that surgery demonstrates strong outcomes when performed and that Austrian Heart Teams responsible for patient allocation are remarkably well calibrated. 

Predicting whether TAVI’s excess mortality is due to greater patient morbidity or frailty, or valve-related events during follow-up, is complex. In other words, whether patients live longer with one valve over another because they were destined to live longer or because of the valve itself, seems difficult to answer. TAVI is a powerful therapeutic alternative, perhaps the most potent available in interventional cardiology alongside acute-phase coronary angioplasty. It clearly improves patient prognosis and quality of life. Therefore, the question is whether it achieves this more effectively than surgery. Studies like AUTHEARTVISIT tell us that globally, it does not, which does not exclude it from being an excellent therapeutic option in numerous scenarios. Thus, to the words of some critics of the study I cite: “Why do patients die? … It’s not the valve’s fault,” they should be reminded of the impact of new left bundle branch block, pacemaker implantation, the role of paravalvular leakage, untreated residual heart disease, future difficulty accessing coronary ostia, mortality in subsequent valvular procedures—both surgical and percutaneous valve-in-valve—and phenomena like thrombosis and embolism, among others, assuming comparable rates of structural degeneration and endocarditis, which is a significant assumption. What is clear is that the surgical option must remain valid, and outcomes, particularly in younger age groups (<75–80 years) with adequate life expectancy, fewer comorbidities, and thus more comparability, are decidedly better. 

REFERENCE:

Auer J, Krotka P, Reichardt B, Traxler D, Wendt R, Mildner M, Ankersmit HJ, Graf A. Selection for transcatheter versus surgical aortic valve replacement and mid-term survival: results of the AUTHEARTVISIT study. Eur J Cardiothorac Surg. 2024 Jul 1;66(1):ezae214. doi: 10.1093/ejcts/ezae214.