We appreciate the kind of words of Drs Shochet and Newman. Our data were presented using discreet white blood cell cutoffs because we felt that for the purpose of establishing treatment guidelines, it would be more practical to have a single threshold value (eg, 15×109/L) above which children at risk for bacteremia should receive empiric antimicrobial therapy. We agree that in the evaluation of any individual highly febrile child, the likelihood ratio is useful in assessing the specific estimated risk for bacteremia, and that viewing the data in this way can be clinically useful.
Harper MB, Lee GM. White Blood Cell Count Likelihood Ratios for Bacteremia in Febrile Young Children. Arch Pediatr Adolesc Med. 2000;154(9):964. doi: