Derivation and validation of risk score to predict mortality or heart failure-related hospitalisation based on clinical and echocardiographic factors in moderate or severe aortic stenosis

Abstract

Abstract Background Multiple studies from echocardiographic databases have investigated association between moderate or severe aortic stenosis (AS) and poor cardiovascular outcomes. However, outcomes and treatment response are also dependent on the clinical picture. Purpose To improve prognostication and management response of patients with moderate or severe AS by development and validation of a new risk score algorithm that combines clinical and echocardiographic parameters to predict 5-year all-cause mortality or heart failure (HF)-related hospitalisation. Methods 3,452 consecutive patients diagnosed with moderate or severe AS [aortic valvular area <1.5 cm2, peak velocity > 3 m/s or dimensionless severity index < 1.5] were included in this observational cohort study. The least absolute shrinkage and selection operator (LASSO) and logistic regression was used to develop a risk score. This score provided an estimated probability of 5-year composite outcomes of all-cause mortality or HF-related hospitalisation in moderate or severe AS patients. The risk score was then validated with a geographically separate independent test cohort (n=1,788). Results Among 3,452 patients with at least moderate AS with median follow up of 7.3 years, 41 variables were important predictive parameters to be included in risk score (Figure 1). The most significant independent predictors were non-cardiac (Table 1). Areas under the receiver operating characteristic curves for the score were 0.80 (95% CI: 0.79-0.82) in the training set and 0.76 (95% CI: 0.73-0.80) in the test dataset, demonstrating good discrimination. Conclusions A risk score algorithm based on demographics, echocardiography, and clinical features was developed and validated with over 1000 patients. This risk score can be useful to predict outcomes beyond conventional echocardiographic parameters in patients with moderate or severe AS.Fig 1.Score vs 5-yr composite outcomesTable 1.OR for predictors in risk score