Am Fam Physician. 2003;67(2):389-390
Diagnosing occult bacteremia in febrile infants and children is difficult. The clinical signs tend to be nonspecific, and physicians who rely on these signs might miss the diagnosis. Laboratory tests such as complete white blood cell count and absolute neutrophil count are used to help determine which febrile infants are at risk for occult bacteremia. The C-reactive protein level is valuable as an acute measure of disease activity. This test requires only a small amount of blood, is relatively inexpensive, and can be run quickly. Isaacman and Burke conducted a prospective study to assess the usefulness of serum C-reactive protein measurements in screening for occult bacteremia in febrile infants and children.
The trial included febrile children three to 36 months of age who presented to an urban children's hospital emergency department with fever. The routine work-up for these children involved a complete blood cell count and blood cultures. If the children met the inclusion criteria for the study, an extra 1 mL of blood was drawn to evaluate C-reactive protein levels. The data collected were used to determine the optimal test values and were compared using likelihood ratio tests.
The study included 256 patients with a median age of 15.3 months and an average temperature of 40°C (104°F). Twenty-nine cases of occult bacterial infection were identified, most of which were pneumonia. The optimal cut-off point for the C-reactive protein was 4.4 mg per dL, but this measurement did not have high sensitivity and only had 81 percent specificity for detecting occult bacterial infections. The best test to predict bacterial infection was an absolute neutrophil count of 10.6 ×109 per L. Adding the C-reactive protein test did not increase the sensitivity but decreased the specificity for detection of bacterial infections.
The authors conclude that an absolute neutrophil count of 10.6 × 109 per L is the best predictor of occult bacterial infections in febrile children. The addition of C-reactive protein measurement to this test adds very little toward establishing the diagnosis of occult bacterial infections, and this measurement should not be used as a diagnostic screen in this population.