Development of a 1-year mortality risk calculator in severe aortic stenosis incorporating aortic valve replacement within the first year

Background and Purpose

Severe aortic stenosis (AS) is associated with significant mortality, which is reduced with aortic valve replacement (AVR).1 Wait times for AVR have increased over the years and are associated with higher rates of mortality, making it crucial to identify patients at the highest risk who may benefit from expedited intervention.2 Individualized prediction tools to estimate 1-year mortality are imperative to improve clinical decision-making and timing of intervention.

Methods

Patients with a first diagnosis of severe AS (aortic valve area ≤ 1.0 cm², mean gradient ≥ 40 mmHg, peak velocity ≥ 4.0 m/s) were identified from an institutional echocardiography database. The outcome variable of interest was all-cause mortality within 1 year of index echocardiography. The primary explanatory variable was AVR in the first year after diagnosis. Additional explanatory variables included age group (<65, 65-74, 75-84, 85+), sex, chronic kidney disease (CKD), LVEF ≤ 50%, AVA ≤ 0.8 cm2, indexed AVA ≤ 0.5 cm²/m², transaortic mean gradient ≥ 45 mmHg, and LAVI ≥ 48 mL/m2. Univariable binary logistic regression models were fit for each candidate predictor. A multivariable model was then constructed using all predictors, followed by backwards elimination. Each regression coefficient within each selected model was divided by a common denominator to determine the points associated with each indicator variable in the final selected model. We computed a map from the minimum to the maximum number of points for each model to risk (probability of outcome) from model coefficients (including the intercept) and the inverse logit transformation.

Results

Full data comprised of N = 2559 subjects. Final variables in the multivariable analysis included receipt of AVR within 1 year, age ≥ 65 years, female sex, GFR < 60 L/min/1.73 m, LVEF ≤ 50%, aortic valve area ≤ 0.8 cm², mean gradient ≥ 45 mmHg, and left atrial volume index ≥ 48 mL/m². The risk calculator risk-stratified the population from 0.8% to 75.7%. The strongest predictor of 1-year mortality was a reduced LVEF ≤50% (OR 3.25; 95% CI 2.44, 4.33; p-value<0.001), followed by CKD (OR 2.29; 95% CI 1.73, 3.02; p-value <0.001). LAVI ≥ 48 ml/m2 was associated with a two-fold increased risk of death at 1 year (OR 2.08, 95% CI 1.59, 2.73; p-value<0.001).

AVR within 1 year was independently associated with a 77% reduction of 1-year mortality (OR 0.23; 95% CI 0.16, 0.32; p-value <0.001), offsetting the negative prognostic effects of high-risk clinical and echocardiographic features.

Conclusion

We developed a risk score predicting the likelihood of AS-related mortality within one year of diagnosis, incorporating clinical and echocardiographic factors, with consideration to receipt of AVR.Multivariable analysisRisk calculator