Antibiotic prescription practice in an intensive care unit using twice-weekly collection of screening specimens: a prospective audit in a large UK teaching hospital.

Controversy exists regarding the optimal method of managing intensive care unit (ICU)-acquired infection. Antibiotic prescriptions in 177 sequential admissions to our ICU, which collected twice-weekly screening samples, were audited. Seventy-seven percent of patients received at least one antibiotic prescription, and 45% of patients received at least one prescription for suspected or proven sepsis. Of the 353 antibiotic prescriptions audited, 86 were prophylactic and 61 were first prescribed prior to ICU admission. One hundred and eighty-three were prescribed for sepsis; of these, 108 (59%) were empirical prescriptions and only 21% of these were subsequently changed. For the 75 prescriptions for specific organisms, 28% targeted organisms isolated at least four days previously. Clinicians in our ICU reviewed the data and reached consensus that screening was associated with decision making that did not represent current evidence-based practice, because empirical prescriptions were rarely changed or stopped on the basis of new samples, and those prescribed for confirmed infection frequently targeted organisms isolated before the septic episode. After our audit, we stopped regular collection of screening samples and used more targeted and invasive sampling, in response to clinical suspicion, to guide therapy and maintain data concerning local microbial epidemiology.