170. Automated Identification of Emerging Drug Resistance by Retrospective Mining of Electronic Medical Records
Session: Poster Abstract Session: Antimicrobial Stewardship: Current State and Future Opportunities
Thursday, October 8, 2015
Room: Poster Hall
Posters
  • 2015 - IDWeek.pdf (4.9 MB)
  • Background: Emergence of drug resistance during an infection is a worrisome finding for managing physicians, since it can cause therapy failure and poor outcomes. Additionally, genetic mechanisms for resistance can be transmitted, contributing to the worldwide epidemic of multidrug resistant organisms. Most hospitals use automated drug susceptibility testing (DST) to guide antimicrobial choice, depositing results in electronic medical records (EMRs). DST results from serial isolates can reveal changes that imply emerging resistance. We investigated whether EMR data can be mined to extract all cases of emerging resistance imputed by DST results within a large urban medical center.

    Methods: Data on collected specimens, lab results, and admit-discharge events for all patients in a five-year period at The Mount Sinai Hospital were extracted from the EMR. Lab results for cultures from all sources (including urine, blood, stool, tissue and wounds) were queried, and associated DST data for all isolates were parsed for species identifications and drug susceptibilities. All DST data were generated by Vitek2 (bioMérieux, France) during routine care. A transition from a "susceptible" to a "resistant" categorization for the same drug in serial isolates of the same species during one continuous hospital visit was considered an instance of emerging drug resistance.

    Results: We identified 1,696 instances of emerging drug resistance during a five-year period, out of 550,382 cultures and 194,753 isolated organisms involving 63,862 hospital visits. The median duration of the susceptible-to-resistant transition was 7.41 days (stddev=20.58). Top species identified included Pseudomonas aeruginosa (N=454, 27%), Klebsiella pneumoniae (N=412, 24%), Escherichia coli (N=242, 14%), Acinetobacter baumannii (N=106, 6.3%), and Staphylococcus epidermidis (N=92, 5.4%). Among involved drug classes, 363 (21%) involved cephalosporins, 263 instances (16%) involved quinolones, and notably, 245 (14%) involved carbapenems.

    Conclusion: EMR data can identify cases of antimicrobial resistance emerging during hospital visits. These cases may be analyzed further to inform future empirical therapy. If orginal isolates are available, they also may be investigated individually for genetic mechanisms underlying drug resistance emerging in the hospital setting.

    Theodore Pak, AB, Timothy O'donnell, BSc and Andrew Kasarskis, PhD, Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, NY

    Disclosures:

    T. Pak, None

    T. O'donnell, None

    A. Kasarskis, 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.