Occult bacteremia in the emergency department: diagnostic criteria for the young febrile child

Ann Emerg Med. 1985 Dec;14(12):1172-7. doi: 10.1016/s0196-0644(85)81024-6.


We studied a selected series of febrile infants (N = 201) in an attempt to prospectively identify risk factors for bacteremia. Infants with fever less than 39.4 C, vomiting and diarrhea, croup, or viral exanthem or enanthem were not included. Twenty-one infants (9.5%) had positive blood cultures despite the initial judgment of their physician that only viral illness or localized bacterial infection existed. WBC count of more than 15,000 correlated with bacteremia, with a sensitivity of 0.71 and a specificity of 0.73. Extensive multivariate linear regression analysis attempting to increase predictive values was completed. The combination of fever higher than 39.4 C present for more than 12 hours and absolute polymorphonuclear count of more than 9,000 cells/mm3 had a sensitivity of 0.62 and a specificity of 0.78 for bacteremia. Descriptive statistics for groups with and without bacteremia are summarized. We have defined prospectively a population of infants with a high probability of bacteremia and a lower probability of viral illness. Identification of such a group is useful to the emergency physician because early antibiotic therapy may lessen morbidity and mortality. We conclude that an easily obtained data base may be useful in the prospective identification of those at risk for bacteremia.

MeSH terms

  • Blood Cell Count
  • Body Temperature
  • Child, Preschool
  • Continental Population Groups
  • Emergency Service, Hospital
  • Female
  • Fever / etiology*
  • Humans
  • Infant
  • Male
  • Meningitis / diagnosis
  • Prospective Studies
  • Sepsis / blood
  • Sepsis / complications
  • Sepsis / diagnosis*