Heart failure risk scores in advanced heart failure patients: insights from the LEVO‐D registry

Abstract

Abstract Aims The prevalence of advanced heart failure (HF) is increasing due to the growing number of patients with HF and their better treatment and survival. There is a scarcity of data on the accuracy of HF web‐based risk scores in this selected population. This study aimed to assess mortality prediction performance of the Meta‐Analysis Global Group in Chronic HF (MAGGIC‐HF) risk score and the model of the Barcelona Bio‐HF Risk Calculator (BCN‐Bio‐HF) containing N terminal pro brain natriuretic peptide in HF patients receiving intermittent inotropic support with levosimendan as destination therapy. Methods and results Four hundred and three advanced HF patients from 23 tertiary hospitals in Spain receiving intermittent inotropic support with levosimendan as destination therapy were included. Discrimination for all‐cause mortality was compared by area under the curve (AUC) and Harrell's C‐statistic at 1 year. Calibration was assessed by calibration plots comparing observed versus expected events based on estimated risk by each calculator. The included patients were predominantly men, aged 71.5 [interquartile range 64–78] years, with reduced left ventricular ejection fraction (27.5 ± 9.4%); ischaemic heart disease was the most prevalent aetiology (52.5%). Death rate at 1 year was 26.8%, while the predicted 1‐year mortality by BCN‐Bio‐HF and MAGGIC‐HF was 17.0% and 22.1%, respectively. BCN‐Bio‐HF AUC was 0.66 (Harrell's C‐statistic 0.64), and MAGGIC‐HF AUC was 0.62 (Harrell's C‐statistic 0.61). Conclusions The two evaluated risk scores showed suboptimal discrimination and calibration with an underestimation of risk in advanced HF patients receiving levosimendan as destination therapy. There is a need for specific scores for advanced HF.

Keywords

• Advanced heart failure
• Mortality
• Risk models
• Risk prediction