Implementation of a Pragmatic Biomarker-Driven Algorithm to Guide Antibiotic Use in the Pediatric Intensive Care Unit: the Optimizing Antibiotic Strategies in Sepsis (OASIS) II Study. (4th October 2017)

• Record Type: Journal Article
• Authors: Downes, Kevin; Schriver, Emily; Russo, Michael; Weiss, Scott; Fitzgerald, Julie; Balamuth, Fran; Tolomeo, Pam; Bilker, Warren; Han, Jennifer; Lautenbach, Ebbing; Coffin, Susan; Gerber, Jeffrey S
• Is Part Of: Open forum infectious diseases. Volume 4(2017)Supplement 1
• Page Start: S504
• Page End: S504
• Publication Date: 2017-10-04
• Subjects: Communicable diseases -- Periodicals; Medical microbiology -- Periodicals; Infection -- Periodicals; 616.9
• DOI: 10.1093/ofid/ofx163.1305
• Languages: English
• ISSNs: 2328-8957

Abstract

Abstract: Background: Antibiotic overuse in the pediatric intensive care unit (PICU) is common and reliable approaches are needed to promote safe antibiotic discontinuation. In prior work we developed a biomarker-based algorithm that identified children with suspected sepsis at low risk of bacterial infection. We evaluated the effectiveness of this algorithm to reduce broad-spectrum antibiotic use in the PICU. Methods: We conducted a quasi-experimental study focused on patients in whom antibiotics were initiated for presumed bacterial sepsis. Antibiotics were given per usual practice during the nonintervention period (T1: Aug 2012 – May 2015). From June 2015 – May 2016 (T2) PICU clinicians were encouraged but not required to stop antibiotics in "low-risk" patients: CRP <4 mg/dL and procalcitonin <1 ng/mL at SIRS onset (day 0) and no pathogen or signs of bacterial infection identified by day 2. The primary outcome was antibiotic length of therapy (LOT) from day 3 through 9, hospital discharge, or death. CDC NHSN definitions were used to define bacterial infections days 0–2. We reviewed all children ≤18 years with SIRS; those with immune compromise, DNR, or recent SIRS (within 30 days) were excluded. Time series analyses adjusting for significant covariates and confounders compared LOT from T1 to T2 in patients with no identified bacterial infection. We also calculated the incidence rate ratio (IRR) of LOT in the subset of patients who met our low-risk criteria. Results: 525 eligible episodes of suspected sepsis occurred during T1 and 212 during T2. Bacterial infections were detected in 34% of T1 episodes and 39% in T2 (P = 0.16). Patients in T2 had fewer cardiovascular conditions but were otherwise similar to T1 patients. Broad-spectrum LOT remained unchanged in all patients without bacterial infections following implementation of our algorithm (OR 0.72, 95% CI 0.46 – 1.13; Figure 1 ). Among the subset who met low-risk criteria, LOT decreased from T1 to T2 (197 v. 104 antibiotic days per 1000 patient-days, IRR 0.53, 95% CI 0.30–0.93). Conclusion: Implementation of a biomarker-based algorithm did not affect broad-spectrum antibiotic prescribing overall in patients without bacterial infections in our PICU, although LOT declined in those defined by our algorithm to be low-risk. Disclosures: K. Downes, Merck, Inc.: Grant Investigator, Research grant; Pfizer, Inc.: Grant Investigator, Research grant