Background: Identifying febrile children with invasive bacterial infection is difficult in the absence of telltale physical findings. Urine latex agglutination (ULA) tests have been used for rapid, on-site identification of such children.
Objectives: To study the performance of ULA tests in identifying children with Haemophilus influenzae, Streptococcus pneumoniae, and group B streptococcus infection and to examine how the results of ULA tests affect patient treatment.
Design: Retrospective review.
Setting: Urban children's hospital.
Patients: All emergency department and hospital patients tested by ULA in 1990, excluding patients in the neonatal units.
Results: Of 1629 patients, 36 had positive tests (20 H influenzae, 5 S pneumoniae, and 11 group B streptococcus). Thirteen of these were false positive based on culture results. Although ULA tests demonstrated excellent specificity, their sensitivity was poor. Positive predictive values for bacteremia ranged from 0.346 to 0.600, and negative predictive values ranged from 0.972 to 0.997. The decision to treat with antibiotics was made before ULA test results were available in 19 (53%) of the 36 patients with positive test results. Of 1593 patients with negative test results, 1211 (76%) were admitted to the hospital, and approximately 81% were empirically treated with parenteral antibiotics.
Conclusions: In clinical practice, treatment and disposition decisions are frequently made without regard to ULA test results.