Objective: To determine the relationship between the duration of fever as reported by caregivers and the likelihood of occult bacteremia in highly febrile young children.
Methods: This is a prospective cohort study performed as part of a prior, multicenter, randomized, interventional trial of oral versus intramuscular antibiotics in the prevention of complications of occult bacteremia in febrile children presenting to nine urban pediatric emergency departments at eight medical centers. Participants included children three to 36 months of age with a temperature of > or = 39.0 degrees C and a nonfocal illness (or uncomplicated otitis media) managed as outpatients. The outcome measure was the presence of bacteremia.
Results: Of the 6680 randomized patients, 6619 (99.1%) had a culture of their blood and a valid reported duration of fever. The median duration of fever in patients with bacteremia (n = 192) and without bacteremia (n = 6427) was the same, one to two days, but the mean rank of patients with bacteremia was significantly lower than that of patients without bacteremia (P + 0.0009). A significantly greater proportion of patients with fever < 1 day had bacteremia than patients with fever > or = 1 day (P = 0.004), and a significantly greater proportion of patients with fever < 2 days had bacteremia than patients with fever > or = 2 days (P = 0.009). The sensitivity, specificity, positive predictive value, and negative predictive value of fever < 1 day in detecting occult bacteremia were 40.1, 69.8, 3.8, and 97.5%, respectively.