Indications for antibiotic use in ICU patients: a one-year prospective surveillance.

Abstract

The high prevalence of nosocomial infections in critically ill ICU patients is associated with high antibiotic consumption. Besides its economic impact, there is the constant threat of selection and induction of antibiotic resistance. Surveillance studies recording the incidence of infections, antibiotic use, and antimicrobial susceptibilities of pathogens supply vital information regarding infection control and prevention of antibiotic resistance. In order to analyse antibiotic consumption we recorded antibiotic use in a general ICU during one year by categorizing the indications for antibiotic use into three groups; (i) prophylaxis; (ii) therapy for a bacteriologically proven infection (BPI); (iii) therapy for a non-bacteriologically proven infection (non-BPI). Bronchoscopic techniques were used to diagnose pneumonia. In practice, BPI must be treated, but a proportion of antibiotics prescribed for non-BPI may be unnecessary. The subdivision in BPI and non-BPI may help to identify these cases. In all, 515 patients were admitted to ICU and 36% of these had at least one infection. Of all infections, 53% were ICU-acquired and 99% of these occurred in intubated patients. Antibiotics were prescribed in 61% of admissions. Of all antibiotics prescribed for therapy, 49% were for respiratory tract infections, 19% for abdominal infections and 13% for sepsis eci. Categorized by indication, 59% of all antibiotic prescriptions were for BPI, 28% for non-BPI and 13% for prophylaxis. A theoretical reduction of 25% in the number of non-BPI prescriptions would result only in a 7% decrease of total antibiotic use. We conclude that almost all antibiotics prescribed were for intubated patients and for BPI. Respiratory infections were the single most common infection and accounted for 49% of all antibiotics used. Therefore, in our setting, prevention of respiratory tract infections is probably the most effective mode to reduce antibiotic use.