1006. Development of an Antibiotic Spectrum Score (SpSc) to Measure Antibiotic De-escalation (AD) Based Upon Perceptions of Antibiotic Stewards (AS) Obtained Through a Modified Delphi Method
Session: Poster Abstract Session: Stewardship: Implementing Programs
Friday, October 4, 2013
Room: The Moscone Center: Poster Hall C

Antibiotic spectrum is an important concept for AS, yet is dependent on clinical context and therapeutic principles. The study aim is to develop a SpSc to characterize spectrum of antibiotic regimens, compare SpSc between regimens, and classify regimens in terms of AD.


Three iterations of a web-based modified Delphi method were conducted to develop a SpSc and operational rules to measure AD. Participant invitation was based on experience, academic affiliated practice, and involvement in AD. Delphi focus concepts included: organisms and antibiotics for SpSc development; operational aspects; application to measure AD; and SpSc calibration. Consensus was defined as >65% agreement with statements for values of 6-7 on a 7-point Likert scale. Agreement between Delphi rounds and descriptive statistics were computed. SpSc-based AD assessments were compared to expert opinion on a set of 20 therapy vignettes.


Demographics of participants completing all Delphi iterations (n=25): occupation (60% PharmD vs. MD), practice settings (24% Community, 32% VA, 52% University), 48% experience >10 years, and 96% AS program participation. Consensus was met for few organisms; however by accepting > 50% agreement for Likert values of 5-7, a SpSc with 14 organism and 28 antibiotic domains was developed. Experts agreed to award additional weight to S. aureus, E. coli, Klebsiella spp, Acinetobacter spp., E. faecium, and P. aeruginosa. In accordance with consensus, an ordinal score based on quintiles of organism susceptibility (VA microbiology data from 2008-2012) was assigned to each antibiotic. SpSc values ranged from 5.5 for nafcillin to 55 for tigecycline. Overlapping spectra were addressed by assigning points based on susceptibility results when both antibiotics were reported. Experts agreed that measurement of baseline SpSc should occur at 24 hours after start of therapy, and AD at 72 hours. Antibiotic therapy vignettes were classified by experts as to whether AD occurred. AD corresponded to a reduction in SpSc in 88% of cases. Correlation between mean Likert score of AD and ΔSpSc was significant [r=-0.50 (-0.08, -0.77), p =0.02)]. 


A SpSc and operational rules to measure AD based on deliberations of AS was developed.  In a set of vignettes, a reduction in SpSc on day 4 of therapy was predictive of expert assessments of AD.

Karl Madaras-Kelly, PharmD, M.P.H., Clinical Pharmacy, College of Pharmacy, Idaho State University and VA Medical Center, Boise, ID, Richard E. Remington, M.S., Quantified Inc. and VA Medical Center, Boise, ID, Makoto Jones, MD, MS, Internal Medicine, University of Utah School of Medicine Division of Epidemiology, Salt Lake City, UT, Nicole Hill, PhD, Division Health Sciences, Idaho State University, Pocatello, ID, Benedikt Huttner, MD, IDEAS Center, VA Salt Lake City Health Care System, Salt Lake City, UT and Matthew Samore, MD, University of Utah School of Medicine, Salt Lake City, UT


K. Madaras-Kelly, None

R. E. Remington, None

M. Jones, None

N. Hill, None

B. Huttner, None

M. Samore, None

Findings in the abstracts are embargoed until 12:01 a.m. PST, Oct. 2nd with the exception of research findings presented at the IDWeek press conferences.