Transcatheter Aortic Valve Implantation (TAVI) procedures are rapidly expanding, necessitating a more extensive stratification of patients with aortic stenosis. Especially in the high-risk group, some patients fail to derive optimal or any benefits from TAVI, leading to the risk of futile interventions. Despite consensus among several experts regarding the importance of recognizing and anticipating such interventions, the definition, and predictive criteria for futility in TAVI remain ambiguous.
The purpose of this study is to explore the literature addressing the definition, predictive criteria, and medical predictive models for futility in cases of TAVI.
A scoping review was conducted by two researchers and reported in accordance with the PRISMA-ScR guidelines.
Studies addressing futility in TAVI, including definitions, predictive variables, and models, were included without restrictions on study design but were excluded study only on surgical valve replacement, valve in valve or aortic stenosis causes by other pathology than calcification.
We identified 129 studies from five key sources: CINAHL, PUBMED, the Cochrane Library, ClinicalTrials.gov, and EMBASE. The literature search was conducted in two rounds-first in February 2024 and again in October 2024-using no restrictions on the year of publication or the language of the studies. Additional references were included through cross-referencing.
The definition of futility is not unanimous, although most researchers agreed on 1-year survival as a cutoff. The majority of studies focused on single variables that can predict 1-year survival, employing either prospective or retrospective designs. Frailty was the major concept studied. Numerous predictive models have been identified, but no consensus was found.
Futility concepts generate interest in the TAVI procedure. In this review, numerous articles state that 1-year mortality serves as a cutoff to define futile procedures. Some variables, cardiac or otherwise, are independent predictors of 1-year mortality. Medical predictive models showed moderate sensitivity and specificity, except for machine learning, which shows promise for the future. However, few articles delve deeply into non-quantifiable parameters such as patient goals and objectives or ethical questions. More studies should focus on these parameters.