Background: In patients with severe sepsis, no randomized clinical trial has tested the concept of de-escalation of empirical antimicrobial therapy. This study aimed to compare the de-escalation strategy with the continuation of an appropriate empirical treatment in those patients.
Methods: This was a multicenter non-blinded randomized noninferiority trial of patients with severe sepsis who were randomly assigned to de-escalation or continuation of empirical antimicrobial treatment. Recruitment began in February 2012 and ended in April 2013 in nine intensive care units (ICUs) in France. Patients with severe sepsis were assigned to de-escalation (n = 59) or continuation of empirical antimicrobial treatment (n = 57). The primary outcome was to measure the duration of ICU stay. We defined a noninferiority margin of 2 days. If the lower boundary of the 95 % confidence interval (CI) for the difference in patients assigned to the de-escalation group was less than 2 days, as compared with that of patients assigned to the continuation group, de-escalation was considered to be noninferior to the continuation strategy. Secondary outcomes included mortality at 90 days, occurrence of organ failure, number of superinfections, and number of days with antibiotics during the ICU stay.
Results: The median duration of ICU stay was 9 [interquartile range (IQR) 5-22] days in the de-escalation group and 8 [IQR 4-15] days in the continuation group, respectively (P = 0.71). The mean difference was 3.4 (95 % CI -1.7 to 8.5). A superinfection occurred in 16 (27 %) patients in the de-escalation group and six (11 %) patients in the continuation group (P = 0.03). The numbers of antibiotic days were 9 [7-15] and 7.5 [6-13] in the de-escalation group and continuation group, respectively (P = 0.03). Mortality was similar in both groups.
Conclusion: As compared to the continuation of the empirical antimicrobial treatment, a strategy based on de-escalation of antibiotics resulted in prolonged duration of ICU stay. However, it did not affect the mortality rate.