Table 4 The threshold effect for analysis between ferritin and clinical outcomes.
Number (%) | OR (95% CI), P-value | |
|---|---|---|
28-day mortality | ||
Model A: The linear model | 1947 (100%) | 1.13 (1.07, 1.19) < 0.0001 |
Model B: Two-segment non-linear model | ||
The turning point of ferritin (ng/ml) | ||
≤ 2340( slope 1) | 1762 (90.50%) | 1.54 (1.29, 1.85) < 0.0001 |
> 2340( slope 2) | 185 (9.50%) | 1.04 (0.97, 1.11) 0.2873 |
Slope 2 to slope 1 | 0.67 (0.54, 0.84) 0.0004 | |
Predicted at 2340 | − 0.68 (− 0.95, − 0.41) | |
P for the log-likelihood ratio test | < 0.001 | |
90-day mortality | ||
Model A: The linear model | 1947 (100%) | 1.15 (1.09, 1.21) < 0.0001 |
Model B: Two-segment non-linear model | ||
The turning point of ferritin(ng/ml) | ||
≤ 2250( slope 1) | 1754 (90.09%) | 1.54 (1.30, 1.83) < 0.0001 |
> 2250( slope 2) | 193 (9.91%) | 1.06 (0.99, 1.13) 0.0967 |
Slope 2 to slope 1 | 0.69 (0.56, 0.84) 0.0004 | |
Predicted at 2250 | − 0.31 (− 0.56, − 0.06) | |
P for the log-likelihood ratio test | < 0.001 | |
180-day mortality | ||
Model A: The linear model | 1947 (100%) | 1.16 (1.10, 1.22) < 0.0001 |
Model B: Two-segment non-linear model | ||
The turning point of ferritin(ng/ml) | ||
≤ 2280( slope 1) | 1755 (90.14%) | 1.56 (1.32, 1.84) < 0.0001 |
> 2280( slope 2) | 192 (9.86%) | 1.06 (1.00, 1.14) 0.0695 |
Slope 2 to slope 1 | 0.68 (0.56, 0.84) 0.0003 | |
Predicted at | − 0.18 (− 0.43, 0.06) | |
P for the log-likelihood ratio test | < 0.001 | |
1-year mortality | ||
Model A: The linear model | 1947 (100%) | 1.17 (1.10, 1.23) < 0.0001 |
Model B: Two-segment non-linear model | ||
The turning point of ferritin(ng/ml) | ||
≤ 2300(slope 1) | 1756 (90.19%) | 1.50 (1.27, 1.76) < 0.0001 |
> 2300(slope 2) | 191 (9.81%) | 1.08 (1.01, 1.16) 0.0281 |
Slope 2 to slope 1 | 0.72 (0.59, 0.88) 0.0017 | |
Predicted at 2300 | − 0.15 (− 0.40, 0.09) | |
P for the log-likelihood ratio test | 0.002 | |
