Figure 3.
(A,B) ROC curve of statistically significant examined variables. Sensitivity–specificity of different variables for mortality (A) and higher NT-proBNP levels (≥75th percentile) (B). Univariable Cox regression analysis reported a significant relationship between 30-day mortality and QTeSD (hazard ratio: 1.10, 95% confidence limit: 1.03–1.18, p < 0.05), mean Te (hazard ratio 1.02, 95% confidence limit: 1.01–1.04, p < 0.001) and TeSD (hazard ratio 1.12, 95% confidence limit: 1.01–1.23, p < 0.001) (Table 3). All number value in the table of the figure should be intended as follow: .526 means 0.526.
