Table 3.

Experimental Sample (Validation Sample)

Risk
Value
SensitivitySpecificityPercentage of
Patients Above
Cut-Point
.37.11 (.16).97 (.99)3% (1%)
.27.20 (.24).96 (.97)5% (3%)
.21.27 (.30).95 (.96)7% (5%)
.17.42 (.37).93 (.94)10% (7%)
.15.49 (.48).92 (.93)11% (9%)
.13.56 (.53).89 (.90)15% (11%)
.10.69 (.64).83 (.84)21% (16%)
.07* .76 (.80).74 (.73)30% (28%)
.06.78 (.85).69 (.66)34% (31%)**
.05.87 (.86).62 (.57)42% (44%)
  • The sensitivity and specificity of different treatment thresholds are displayed.

  • * Value that maximizes sensitivity and specificity in derivation set data.

    An individual patient's risk value would be determined using the regression formula: risk = ew/(1

  • F3-207 ew), where w = .1673∗ANC

    .2006∗temperature

    .84343∗gender − 12.454 (gender = 1 for females; 0 for males).