INFLUENCE OF BLOOD CULTURE RESULTS ON ANTIBIOTIC CHOICE IN THE TREATMENT OF BACTEREMIA

Background: It is unclear how often blood culture results influence empiric antibiotic regimens. Methods: To assess the frequency of antibiotic modification and the rates of proper documentation of blood culture results by house staff physicians, we prospectively evaluated 226 episodes of bacteremia in 199 patients. Results: Antibiotics were changed in 49.6% of episodes of true bacteremias. Physicians were more likely to change therapy if gram-negative rods (odds ratio [OR], 3.19) or Staphylococcus aureus (OR, 3.12) were isolated, if the blood culture was obtained in the first 7 days of hospitalization (OR, 1.9), or if house staff physicians properly documented the culture results in the medical chart (OR, 3.8) (all P values, <.05). Documentation of positive blood culture results by house staff physicians was absent in 26% of patients, and it was observed less often in patients on the surgical service (OR, 0.35; P=.004) or if a contaminant was recov- ered (OR, 0.24; P<.001). Eighty-three percent of ''true-positive'' blood cultures, as compared with 55% of ''contaminated'' blood cultures, were documented with a note in the medical records (P<.0001). Rates of documentation were higher for gram-negative rods, for patients who were already receiving antibiotic therapy, and for those who had a change of therapy after the culture results became available (all P values, <.05). A multivariate logistic regression model showed that documentation of the blood culture result (OR, 1.78; P=.006) or a positive culture within 7 days of hospitalization (OR, 1.49; P=.01) was independently associated with a change in therapy. Conclusions: In a significant proportion of patients with bacteremia, the blood culture result may not be the most important factor that determines antibiotic choice. Bacteremia is not adequately documented by house staff physicians in up to a quarter of patients.