Benefit of appropriate empirical antibiotic treatment: Thirty-day mortality and duration of hospital stay

PURPOSE: We evaluated the effect of inappropriate antibiotic treatment on mortality and duration of hospital stay in medical inpatients with bacterial infections. SUBJECTS AND METHODS: Two cohorts of febrile adult patients ( excluding patients with acquired immune deficiency syndrome and organ transplant recipients), hospitalized in three medical centers in Israel, Italy, and Germany, were included. Patients' data were collected prospectively. Initial empirical treatment was defined as appropriate if an antibiotic prescribed within 24 hours of the first encounter with the patient matched the in vitro susceptibility of a pathogen deemed to be the likely cause of infection. The results of cultures and serologic or direct tests, and data on outcomes were collected 30 days after initiation of empirical treatment. RESULTS: A total of 920 patients ( 26% of 3529 included patients) had microbiologically documented infections, and mortality data were available for 895 patients ( 97%). Inappropriate initial antibiotic treatment was prescribed in 36% of patients ( N = 319). All- cause 30- day mortality rates were 20.1% ( N = 64) and 11.8% ( N = 68) in patients who received inappropriate and appropriate treatment, respectively ( odds ratio = 1.88, 95% confidence interval [ CI], 1.29- 2.72, P =. 001). When adjustment was made for medical center and other variables, the association between inappropriate with mortality was significant ( odds ratio = 1.58, 95% CI, 0.99- 2.54, P =. 058). In all 3 medical centers, the mean duration of hospital stay was at least 2 days longer for patients who were prescribed inappropriate antibiotic treatment ( overall P =. 002). This association was consistent after adjusting for other variables ( P =. 006). CONCLUSION: Appropriate empirical antibiotic treatment is associated with a better survival and shortened duration of hospital stay in medical patients with bacterial infections. c 2006 Elsevier Inc. All rights reserved.