Management of infants at risk for occult bacteremia: a decision analysis.

Because febrile infants with no obvious source of bacterial infection may have bacteremia, and because bacteremia is difficult to diagnose on clinical grounds, we used decision analysis to evaluate whether such infants should be treated with antibiotics, tested further, or sent home. Using a simple decision tree, we found that the decision to give empiric antibiotic treatment is the decision of choice. The difference in quality-adjusted life expectancy between the "best" and "worst" decisions was only 11 days. However, this difference translated to prevention of death or permanent disability in 60 cases per 100,000 febrile children. Further, empiric treatment remained the best management alternative unless the probability of bacteremia was less than 1.4% (less than any published prevalence), or the efficacy of treatment was less than 21%. Our analysis demonstrated that a test with far greater sensitivity than leukocyte count or other tests currently in use is needed to justify testing rather than treating empirically. Further, an enormous patient population would be needed to find a difference of both clinical and statistical significance between treated and untreated patients in a controlled trial. In the absence of such trials, we recommend blood culture and empiric antibiotic treatment of all infants at risk for occult bacteremia.