Table 5. Independent predictors for in-hospital mortality by multiple Cox regression analysis stratified by history of cerebrovascular disease.
| Covariate | Hazard Ratio | 95% Confidence Interval | p Value |
|---|---|---|---|
| S100B at ROSC (square root pg/mL) >31.381a | 114.786 | 15.145–869.986 | <0.0001 |
| VCAM at ROSC (ln ng/mL) <5.288 or >6.383b | 13.557 | 2.691–68.308 | 0.0016 |
| IL-6 at ROSC (ln pg/mL)c | 0.481 | 0.312–0.742 | 0.0009 |
| NT-proBNP difference (pg/mL) | 1.259 | 1.097–1.443 | 0.0010 |
| 8.011 >Cystatin-C at ROSC (ln ng/mL)>6.727d | 67.689 | 8.221–557.347 | 0.0001 |
| Cystatin-C difference (ng/mL) <−298.821e | 14.017 | 2.444–80.374 | 0.0030 |
| 7.194 > sRAGE at ROSC (ln ng/mL) >4.938f | 27.656 | 5.248–145.739 | 0.0001 |
| sRAGE difference (ng/mL) | 1.154 | 1.020–1.307 | 0.0231 |
| MDA-LDL at ROSC (ln ng/mL)g | 1.378 | 1.157–1.641 | 0.0003 |
| Thrombomodulin difference (pg/mL) >3527.921h | 23.306 | 3.913–138.828 | 0.0005 |
| PAI-1 difference (ng/mL) | 0.968 | 0.946–0.990 | 0.0049 |
| History of arrhythmia | 0.053 | 0.004–0.749 | 0.0297 |
| APACH II score × Time (days)i | 1.004 | 1.001–1.007 | 0.0183 |
aS100B at ROSC (square root pg/mL) >31.381: square root value of serum S100B level at ROSC > 31.381.
bVCAM at ROSC (ln ng/mL) <5.288 or >6.383: natural logarithm of serum VCAM level at ROSC < 5.288 and >6.383.
cIL-6 at ROSC (ln pg/mL): natural logarithm of serum IL-6 level at ROSC.
d8.011 >Cystatin-C at ROSC (ln ng/mL) >6.727: natural logarithm of serum level of Cystatin-C at ROSC <8.011 and >6.727.
eCystatin-C difference (ng/mL) <−298.821: serum level of Cystatin-C at 24 hour minus level at ROSC <298.821.
f7.194 >sRAGE at ROSC (ln ng/mL) >4.938: natural logarithm of serum sRAGE level at ROSC <7.194 and >4.938.
gMDA-LDL at ROSC (ln ng/mL): natural logarithm of serum MDA_LDL level at ROSC.
hThrombomodulin difference (pg/mL) >3527.921: serum level of Thrombomodulin at 24 hour minus level at ROSC >3527.921.
iAPACHE II score × Time (days): time-dependent interaction term in the Cox model indicated that the effect of “APACHE II score” on the hazard rate of in-hospital mortality would increase 1.004 times as survival time increased one day.
Goodness-of-fit assessment: adjusted generalized R2 = 0.7593 > 0.15 and concordance = 0.9249 (se = 0.0779), which indicated an excellent fit.