De-escalation of antimicrobial treatment in neutropenic patients with severe sepsis: results from an observational study

Background In severe sepsis, guidelines recommend de-escalating the empirical antimicrobial treatment as soon as the microbiological results are available. We aimed to determine the rate of de-escalation of the empirical antimicrobial treatment in neutropenic patients with severe sepsis. The characteristics of antimicrobial treatment associated with de-escalation and its impact on short- and long-term survival were also determined. Methods In the intensive care unit (ICU) of a cancer referral center, we prospectively collected observational data related to the antimicrobial management in neutropenic patients who developed severe sepsis and were admitted to ICU for at least 48 h. De-escalation of antimicrobial therapy consisted either of deleting one of the empirical antibiotics of a combined treatment, or, whenever possible, to use a betalactam antibiotic with a narrower spectrum of activity. Multivariate logistic regression was conducted to determine the factors associated with de-escalation, while a Cox proportional hazards model with a time-dependent covariate was fitted to assess the effect of de-escalation on 30-day survival. Finally 1-year survival after ICU discharge was compared across de-escalation groups. Results Cumulative incidence of de-escalation of the empirical antimicrobial treatment among the 101 patients of the cohort was 44 %, [95 % confidence interval (CI) 38–53 %], including 30 (68 %) patients with ongoing neutropenia. A microbiological documentation was available in 63 (63 %) patients. Factors associated with de-escalation were the adequation of the empirical antimicrobial treatment in ICU [OR = 10.8 (95 % CI 1.20–96)] for adequate documented treatment versus appropriate empirical treatment, the compliance with guidelines regarding the empirical choice of the anti-pseudomonal betalactam [OR = 10.8 (95 % CI 1.3–89.5)]. De-escalation did not significantly modify the hazard of death within the first 30 days [HR = 0.51 (95 % CI 0.20–1.33)], nor within 1 year after ICU discharge [HR = 1.06 (95 % CI 0.54–2.08)]. Conclusion Our data suggest that, in ICU, de-escalation of the empirical antimicrobial treatment is frequently applied in neutropenic cancer patients with severe sepsis. No evidence of any prognostic impact of this de-escalation was found.