Lifetime prediction of incident heart failure risk in patients with established atherosclerotic cardiovascular disease: the SMART2-HF risk score

Abstract Background Heart failure (HF) is a global disease characterised by reduced quality of life, significant healthcare costs and high mortality. Patients with established atherosclerotic cardiovascular disease (ASCVD) are at high risk of developing incident HF, yet this risk varies significantly among individuals. The ESC-recommended SMART2 risk score predicts individual recurrent cardiovascular disease (CVD) risk, but does not estimate the risk of incident HF. Early identification of patients at high risk of developing HF could improve individualised prevention strategies. Purpose To develop and externally validate the SMART2-HF risk score for prediction of incident 10-year and lifetime HF risk in patients with established ASCVD. Methods The SMART2-HF risk score was derived using data from 81,400 individuals aged between 40 and 90 years with established ASCVD and free from HF at baseline from the SWEDEHEART registry in Sweden. Predictors were aligned with the existing SMART2 risk score and included: age, sex, current smoking, diabetes mellitus, type of ASCVD (coronary artery disease, cerebrovascular disease, peripheral artery disease and/or abdominal aortic aneurysm), years since first ASCVD diagnosis, systolic blood pressure, non-high density lipoprotein cholesterol, estimated glomerular filtration rate, C-reactive protein, history of myocardial infarction, atrial fibrillation, and body mass index. The SMART2-HF risk score applies sex-specific, competing risk-adjusted models with age as the underlying time scale to estimate the risk of incident non-fatal and fatal HF regardless of left ventricular ejection fraction. External validation was performed in 45,031 individuals from the HUNT3 cohort in Norway and the international REACH registry. Results The derivation population in SWEDEHEART consisted of 21,345 females and 60,055 males with established ASCVD. Mean age was 65 ± 9 years for females and 63 ± 9 years for males. During a median follow-up 5.2 years (interquartile range 2.3 – 8.7), a total of 4,076 (5.0%) incident HF events occurred in the SWEDEHEART registry. External validation C-statistic were 0.719 (95% confidence interval CI 0.697 – 0.742) in HUNT3 and 0.654 (95% CI 0.628 – 0.681) in REACH and showed good calibration (Figure 1). Model performance was similar in subgroups of sex and type of ASCVD. Additionally, SMART2-HF demonstrated meaningful risk stratification within SMART2 CVD risk categories, especially in females. This highlights the added value of using both scores to guide individualised preventive strategies (Figure 2). Conclusion The SMART2-HF risk score provides a validated tool for predicting 10-year and lifetime HF risk in patients with ASCVD, which could improve the early identification of individuals at high risk of developing incident HF.