Table 3.
Predictive values of parameters for the severity of HFRS.
| Variables | AUC | p value | Cut-off value | Sensitivity | Specificity | 95% CI for AUC | |
|---|---|---|---|---|---|---|---|
| Lower | Upper | ||||||
| Lg PCT at admission | 0.738 | <0.001 | 0.26 | 0.65 | 0.76 | 0.657 | 0.820 |
| WBC at admission | 0.706 | <0.001 | 9.30 | 0.68 | 0.63 | 0.623 | 0.789 |
| Hemorrhage | 0.640 | 0.004 | – | – | – | 0.549 | 0.731 |
| Combinationa | 0.785 | <0.001 | 0.49b | 0.74 | 0.78 | 0.710 | 0.861 |
AUC, area under the receiver operating characteristic (ROC) curve; P-value for calculated AUC in predicting severity; CI, confidence interval; Sensitivity, specificity and 95% CI are all presented as percentages; WBC, white blood counts; LgPCT, procalcitonin after log10 transformation.
a
LgPCT, WBC, and the occurrence of hemorrhage in combination.
b
Probability value of the combination was analyzed by logistic regression. The regression coefficients of these three parameters were used to set up a logit model for the progression of HFRS as follows: Logit(P|y = severe type) = −2.223+0.079LgPCT +0.934WBC+0.765Hemorrhage.