Correlates of severe heart failure symptoms in patients with DCM
Parameters were tested in univariable analysis for their ability to predict severe symptoms. Variables were divided in five categories: clinical characteristics, LV systolic function, LV diastolic function, RV function and PASP. For clinical characteristics we tested parameters which differed significantly across different NYHA classes (Table 1). Since BNP serum levels were not available for all patients, this parameter was excluded from the logistic regression. For LV systolic and diastolic function, we tested well-established parameters such as LVEF, GLS-LV and, respectively, mitral E/E’ ratio and LA volume index. For RV function we chose traditional parameters of RV systolic function which differed across NYHA classes (Table 2), together with 3D RVPAC. The main correlates of severe HF are shown in Table 4. To compare the accuracy of these parameters, we performed ROC analysis and calculated the corresponding AUC (Table 5). The best result was found for RVPAC (AUC=0.712, p<0.001), with a cut-off value of 0.54 for identifying severely symptomatic patients (47.5% sensitivity, 92.3% specificity). The parameter with the highest statistical significance and highest AUC from each of the five above-mentioned categories was introduced in the multivariable logistic regression, together with age (Table 6). RVPAC emerged as the only independent correlate of severe HF symptoms in our study population (odds ratio, 0.035 [95% CI, 0.004 – 0.312], p=0.003).