Introduction: Heart failure (HF) is a major cause of death and hospital readmission, regardless the systolic function. After its recent reclassification by the current guidelines in HF with preserved, mid-range, and re-duced ejection fraction (EF), new prognosis scores for each of these categories needs to be re-assessed.
Objective: To assess new prognosis models to predict in-hospital and long-term mortality, after an acute HF event, in patients with preserved ejection fraction (HFpEF), by combining clinical parameters and bio-markers.
Methods: We identified retrospectively all patients with a diagnosis of AHF (2013-2015), irrespective of aetiology and severity. Outcomes were in-hospital and 2-year mortality. We collected the following potenti-al prognostic markers: age; body mass index (BMI); admission systolic and diastolic blood pressure (SBP/ DBP) and mean of all measurements during hospitali-zation (SBPm/DBPm); admission heart rate (HR) and mean of all measurements (HRm); QRS duration; atri-al and ventricular arrhythmias; presence of respiratory failure and mechanical ventilation; ischaemic aetio-logy; diabetes mellitus; Charlson co-morbidity index; LVEF during the acute event; biomarkers (troponin I and NTproBNP); inflammatory response (ESR, PCR, leucocytes, fibrinogen); admission glycaemia; renal (eGRF) and hepatic function (ALT and AST); sodium and lactic acid level.
Results: 472 patients were identified, from which 109 (23%) had HFpEF (70±10 years, 57% female). In-ho-spital mortality was 1.8%, while 2-year mortality was 32%. From all mentioned parameters, only HR, SBPm, DBPm, admission glycaemia, lactic acid, and sodium level were significantly correlated with in-hospital mortality (all R >0.4, p<0.001). However, the best pre-dictor model for in-hospital mortality included admis-sion HR, DBPm, lactic acid, and sodium level (R=0.7, R2=0.5, p<0.001). HR>132 BPM, DBPm<67 mmHg, Na⁺ <132 mEq/l and lactic acid>5.4 mmol/l were all good independent predictors of in-hospital mortality (Figure). For 2-year mortality, age, BMI, respiratory failure, Tp I, NT-proBNP, eGRF, admission glycaemia, and sodium level were significantly correlated with mortality (all R >0.4, p<0.001). The best predictor mo-del included BMI, respiratory failure, sodium level, NT-proBNP, and admission glycaemia (R=0.71, R2=0.5, p=0.002). NTproBNP >4182 pg/ml (AUC=0.74, sen-sitivity=75%, specificity=75%) and BMI<24.9 kg/m² (AUC=0.83, sensitivity=80%, specificity=87%) were independent predictors of 2-year mortality.
Conclusions: Best predictor models for in-hospital and 2-year mortality are different in patients with preserved ejection fraction, after an acute decompensation. Defi-nite cut-off values can be defined for different clinical parameters and biomarkers in order to independently predict mortality. Prospective larger studies are needed in order to validate these parameters.