Aims: To investigate the prognostic value of autonomic variables in patients with symptomatic chronic heart failure (HF) treated according to current recommendations.
Methods and results: We analysed 24 h time-domain [standard deviation of all normal-to-normal RR intervals (SDNN)], frequency-domain [very low frequency and low frequency power (VFLP and LFP)], and non-linear [detrended fluctuation analysis (DFA)] heart rate variability, deceleration capacity (DC), and heart rate turbulence (HRT) in 388 sinus rhythm HF patients enrolled in the GISSI-HF Holter substudy [82% males, age 65 ±10 years, New York Heart Association (NYHA) functional class III-IV 20%, left ventricular ejection fraction (LVEF) 33 ±8%]. Cardiovascular (CV) mortality and combined sudden death + implantable cardioverter defibrillator (ICD) discharge were assessed as a function of continuous variables in the entire population and in patients with LVEF >30% in univariate and multivariable Cox proportional hazards models. After a median of 47 months, 57 patients died of CV causes and 47 experienced the arrhythmic endpoint. For CV mortality, VLFP, LFP, and turbulence slope (TS) improved predictive discrimination (c-index) and risk classification [integrated discrimination improvement (IDI)] when added to clinical variables [age ≥70 years, LVEF, non-sustained ventricular tachycardia (NSVT), serum creatinine], while for arrhythmic mortality although the c-index increased in all three autonomic markers, the results of the IDI were statistically significant only for TS when added to NSVT, serum creatinine, and ischaemic aetiology. In 194 patients with LVEF >30% (20 arrhythmic events), the hazard ratio of an impaired TS (<2.5 msper RR interval) was 3.81 (95% confidence interval 1.35-10.7,P = 0.012) after adjustment for serum creatinine.
Conclusions: Autonomic indexes still have independent predictive value on long-term outcome in HF patients. HRT may help in identifying patients with LVEF >30% at increased arrhythmic risk. Trial registration NCT00336336.