Isaacman et al developed an equation for identifying the risk of bacteremia in young children with fever. This can help identify those children who would benefit from antibiotic therapy to prevent serious bacterial infection. The authors are from Eastern Virginia and Harvard Medical Schools.


NOTE: The authors refer to the antibiotic therapy in febrile children as being "prophylactic", in that it might prevent development of serious bacterial infection.


Patients: infants and young children 3 to 36 months of age



(1) absolute neutrophil count (ANC)

(2) temperature in °C

(3) gender





absolute neutrophil count (ANC)

(WBC * (percent neutrophils) / 100) per µL

(ANC) / 1000



(temperature in °C)









• The percent neutrophils used for the ANC is given in the Methods section as the sum of bands and PMNs. I've included metamyelocytes and myelocytes but not eosinophils or basophils in the implementation.


W =

= (0.1673 * (points for ANC)) + (0.2006 * (points for temperature)) + (0.8434 * (points for gender)) – 12.454


risk of bacteremia =

= EXP(W) / (1 + EXP(W))



• The risk varies from 0 to 1.

• A risk value of 0.07 gave the optimum combination of sensitivity and specificity for bacteremia. The performance of other decision points are given in Table 3 on page 980, with the ROC analysis in Figure 1.

• A child with a risk score >= 0.7 is one for whom antibiotic therapy would be considered appropriate.

• A child with a risk score < 0.7 is one for whom antibiotic therapy would not be given.



• The sensitivity and specificity varies with the cutoff point. A different cutoff point can be used to select either maximum sensitivity (0.05) or specificity (0.37).

• The presence of neutropenia, leukemia, immunosuppression or other similar disease would make the use of the equation inappropriate.


To read more or access our algorithms and calculators, please log in or register.