Table 3.

ROC analysis of differences between individual values for the prediction of 30-day poor outcomes. For each optimal cutoff value was calculated also a ratio as 100*(Dx-Dy)/Dx

Difference	Optimal cutoff	Sensitivity	Specificity	AUC	P
D2-D1	>5.3 mcg/L (> 31%)	72.7%	90.6%	0.824	P < 0.001
D3-D1	> 7.1 mcg/L (> 42%)	82.4%	92.3%	0.889	P < 0.001
D4-D1	> 4.0 mcg/L (> 24%)	100.0%	90.9%	0.983	P < 0.001
D3-D2	> 2.1 mcg/L (> 11%)	89.3%	86.2%	0.878	P < 0.001
D4-D2	> 11.5 mcg/L (> 63%)	77.3%	95.8%	0.867	P < 0.001
D4-D3	> 3.8 mcg/L (> 26%)	66.7%	96.4%	0.677	P < 0.043