226. Development of an Automated Process for Antibiogram Generation
Session: Poster Abstract Session: Clinical Practice Costs, Informatics, and Telemedicine
Thursday, October 8, 2015
Room: Poster Hall
Posters
  • Antibiogram poster final.pdf (493.0 kB)
  • Background: CLSI recommends the use of antibiograms to track the incidence of antimicrobial resistance. Antibiograms are arduous to develop and require a significant time commitment. The purpose of this study was to develop a quick and reliable process to generate institutional antibiograms.

    Methods: Antimicrobial susceptibility testing data from 1/1/2014 through 12/31/14 were obtained from the clinical microbiology laboratory for antibiogram creation. An algorithm was created to handle susceptibility data using open-source programs (RStudio v0.98 and R v3.1.2). Original raw data contain observations for multiple facilities, surveillance cultures, and multiple observations for individual isolates. Data were initially cleaned by selecting the locations of interest, eliminating surveillance cultures (i.e., axilla, groin, nares/perirectal swabs), and eliminating exact duplicate observations. Data were then transposed from “narrow” format into “wide” format with only one row for each observation. This “semi-clean” data still contains multiple observations per isolate, duplicate isolates, and many “missing” values generated by the transposition process. The final clean dataset is generated by combining multiple observations into one primary observation. This is accomplished by combining testing methods (Etest® and MIC), selecting the first isolate of the year, and removing duplicated isolates by selecting the more resistant biotype. CLSI breakpoints are applied to the current data and percent susceptible is calculated for each drug and organism to generate the antibiogram, which may undergo post-processing for optimal visual interpretation of antibiogram results.

    Results: Antibiogram generation was significantly faster after development of the algorithm. Prior to algorithm development, the antibiogram was developed manually using Microsoft Excel® taking approximately 1 month. The algorithm produces a completed antibiogram with CLSI interpretations of the MIC data in less than 60 minutes from raw data.

    Conclusion: Antibiogram development can be significantly streamlined though the use of open-source data science technology. Automated antibiogram-generation functionality significantly reduces the time and resource commitment of previous antibiogram creation methods.

    W. Cliff Rutter, PharmD and David S. Burgess, PharmD, FCCP, University of Kentucky, College of Pharmacy, Lexington, KY

    Disclosures:

    W. C. Rutter, None

    D. S. Burgess, None

    Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 7th with the exception of research findings presented at the IDWeek press conferences.