The PROCEAS model using diabetes, atrial fibrillation, cardiac damage, sex, and age predicted 2-year mortality in severe AS with an AUC of 0.72.
A newly developed nomogram integrating clinical factors and treatment modality provides individualized 1- and 2-year mortality risk estimates for patients with severe aortic stenosis.
Absolute Event Rate: 0% vs 0%
Abstract Background Aortic stenosis (AS) is the most prevalent valvular disease requiring intervention. While therapeutic options have expanded, clinical decision-making is challenged by the diversity of patients’ profiles. Existing risk models often fail to account for population-specific factors and healthcare system variations. Developing predictive models tailored to specific populations and focused on robust endpoints may improve treatment decision-making. Objective To develop an easy-to-use predictive model for overall mortality in patients with severe AS, integrating patient characteristics and possible therapeutic approaches based on a prospective AS cohort. Methods This prospective observational study included patients with severe AS enrolled in the AS Integrated Care Process at a tertiary hospital in Spain from 2018 to 2022, with follow-up data updated to November 2024 (PROCEAS cohort). Clinical, echocardiographic and laboratory parameters were analysed alongside with treatment modality and clinical outcomes. A stratified cox proportional hazards model was elaborated to predict overall mortality (dependant variable). Patients were categorized into three subgroups based on treatment type: surgical or transcatheter (TAVR) valve replacement or conservative. Initially, only candidate variables meeting the proportional hazard’s assumption across all subgroups were included in a first stratified Cox model. The best possible model was then selected on the basis of Akaike information criterion (AIC). The performance of the model was evaluated using the area under the receiver operating characteristic curve (AUC). Results The PROCEAS cohort comprised 988 patients, of whom 984 were included in the final analysis (mean age: 77±8.4 years; 42.2% female). Treatment distribution was as follows: surgery in 432 patients, TAVR in 490 patients, and conservative in 62 patients. Overall mortality was 26.8% during a mean follow-up of 39±19.4 months, highest in the conservative group, followed by TAVR and surgery (p0.0001). Five relevant predictors were included in the treatment-stratified final cox model: diabetes mellitus, atrial fibrillation, cardiac damage stage in AS (as defined by Généreux et al.), sex, and age (Table 1). A nomogram was elaborated to facilitate clinical application, providing individualized 1-year and 2-year mortality risk estimates based on treatment strategy (Figure 2). Model performance metrics showed an AUC of 0.69 for 1-year mortality (sensitivity: 57%, specificity: 76%) and 0.72 for 2-year mortality (sensitivity: 70%, specificity: 65%). Conclusions This study presents a pilot clinically applicable predictive model for overall mortality in severe AS patients, depending on the treatment modality. The PROCEAS nomogram may aid personalised risk assessment and guide therapeutic decisions in a Spanish population. External validation is needed to confirm broader applicability. Figure 1.PROCEAS Mortality risk nomogram
Salvado et al. (Sat,) reported a other. The PROCEAS model using diabetes, atrial fibrillation, cardiac damage, sex, and age predicted 2-year mortality in severe AS with an AUC of 0.72.