Results of 3-year mortality risk prediction in patients with chronic heart failure and reduced left ventricular ejection fraction

Abstract

The aim – predicting the risk of fatal outcome within 3 years in patients with chronic heart failure (CHF) and reduced left ventricular ejection fraction (LVEF) using appropriate regression logistic models.
Materials and methods. The retrospective analysis of 490 medical histories of patients who were hospitalized in the heart failure department between 2011 and 2018 years with CHF II–IV functional class according to NYHA with LVEF ≤ 40 % on the background of coronary heart disease was conducted. There were 373 (76.1 %) men and 117 (23.9 %) women among the subjects. The median age of the patients was 64 years [56.00; 69.00]. Patients with clinical signs of heart failure and II NYHA functional class – 455 (92.8 %) patients and with clinical signs of CHF and III NYHA functional class – 35 (7.2 %) patients. The analysis was conducted for 490 patients: 228 (46.5 %) patients had a fatal event within three years, 262 (53.5 %) patients survived three years. The univariate analysis of the risk of death was performed for 42 risk factors in logistic regression models. To optimize the acceptance/rejection threshold, the receiver operating characteristic curve (ROC) analysis method was used. The logistic regression model was used to analyze the relationship between the risk of fatality and factor characteristics.
Results and discussion. The analysis revealed the reliable negative relationship between the risk of a fatal outcome and the parameters of TSAT, blood pressure, LVEF, and GFR, the risk of a fatal event decreases (p<0.05) when these indicators increase. Positive relationship was found between the risk of the fatal outcome and the duration of HF, age, the size of right ventricle and left atrium, LV diastolic volume index, LV systolic volume index, PASP, LV myocardial mass index. The risk of fatal outcome increases when these indicators increase (p<0.05). Such indicators as levels of ferritin, hemoglobin, potassium, sodium, bilirubin, ALT, AST, cholesterol, blood glucose and BMI, gender, presence of COPD, arterial hypertension, diabetes did not demonstrate reliable prognostic information. The use of ACE inhibitors and beta-blockers show the tendency to the better prognosis. Multivariate logistic regression model was used to build a three-year fatality risk prediction model with better prognostic characteristics. Nine factor signs were identified: mineralocorticoid receptor antagonists, ischemic heart disease, body mass index, blood pressure, LV diastolic volume index, LV systolic volume index, PASP, LV myocardial mass index, GFR. This model is adequate (χ²=80.4, p<0.001).
Conclusions. In patients with CHF with reduced LVEF, according to the data of a univariate logistic regression model, the predictors of a fatal outcome within 3 years are age, ischemic artery disease, renal dysfunction, atrium fibrillation, duration of HF symptoms, PASP, LV ejection fraction, left atrium size, LV diastolic volume index, LV systolic volume index, right ventricle size, LV myocardial mass index, GFR, TSAT. The risk of a fatal outcome begins to increase with blood pressure ​​≤ 118 mm Hg, LVEF < 30 %, TSAT ≤ 20 %, GFR ≤ 64 ml/min/1, 73 m², patient’s age > 64 years old, right ventricle’s size > 3.66 cm, left ventricle’s size > 4.94 cm, LV diastolic volume index > 116.9 ml, LV systolic volume index > 87.88 ml, pulmonary artery systolic pressure > 57 mm Hg, LV myocardial mass index > 172.27 g/m². According to the multifactorial logistic regression model, the predictors of the occurrence of a fatal outcome within 3 years are BMI (OR 1.04, p<0.05), BP (OR 0.97, p<0.004), LV diastolic volume index (OR 0.97, р<0.022), LV systolic volume index (OR 1.05, р<0.004), pulmonary artery systolic pressure (OR 1.02, р<0.024), LV myocardial mass index (OR 1.01, р<0.005), GFR (OR 0.98, p<0.009), MRA in the treatment (OR 0.45, p<0.016). Youden Index (Y_crit=0.5044), the sensitivity of the 9-factor model is 65.4 % (95 % CI 57.6–72.7 %), the specificity of the model is 79.2 % (95 % CI 72.8–84.6 %), predictive significance +PV – 72.1 % (95 % CI 65.8–77.6 %), predictive significance –PV – 73.6 % (95 % CI 69.0–77.7 %).