1757. Using the Desirability of Outcome Ranking for Management of Antimicrobial Therapy (DOOR-MAT) to Assess Antibiotic Therapy Guided by Rapid Molecular Diagnostics (RMD) in Bloodstream Infection (BSI) Caused by Escherichia coli (Ec) and Klebsiella pneumoniae (Kpn)
Session: Oral Abstract Session: Diagnostics Making a Difference
Saturday, October 6, 2018: 10:45 AM
Room: S 158

Background: In the setting of Ec and Kpn BSI, empiric antibiotic therapy is determined by clinical judgement and Gram stain (GS) of the positive blood culture bottle up to 2 days before antibiotic susceptibility test (AST) results become available. Within hours of GS, RMD can detect Ec and Kpn and predict the pattern of susceptibility to beta-lactams (BL) of differing spectrum. In this study, our objective was to compare “real life” empiric therapy for Ec and Kpn BSI administered between GS and AST results with simulated RMD-guided therapy.

Methods: We identified a subset of patients hospitalized within VHA between 2006 and 2015 who had a blood culture positive for Ec or Kpn, and received empiric BLs between GS and AST results. We further restricted the cohort to those with observed or implied AST results for four representative BLs: cefazolin, ceftriaxone, piperacillin-tazobactam, and imipenem. We extracted BL resistance patterns and, based on previously analyzed RMD performance on 195 Ec and Kpn, we simulated RMD results for our cohort by resampling RMD results stratified by resistance pattern at the observed frequencies of resistance patterns in our clinical isolates. We simulated therapy guided by RMDs and compared to the observed empiric BLs using a DOOR-MAT score (Figure 1).

Results: A total of 36,531 BSI cases were identified. Of these, 9,981 Ec and 4,545 Kpn met our inclusion criteria (Figure 2). Among these, susceptibility to all BLs occurred in 88% of cases; resistance to all BLs occurred in <0.5%. The isolates previously analyzed using RMD included more resistant phenotypes (Figure 3). Empiric BLs were active in 98% of BSIs, with a mean DOOR-MAT score of 66.1 and 59% of cases classified as “moderate overtreatment”. Simulated RMD-guided BL therapy would be active in 95% of cases with a mean DOOR-MAT score of 91.4 (95% CI, 91.2-91.7), and 7% of cases would be classified as overtreatment.

Conclusion: In a large cohort of BSI patients where rates of BL resistance were low, we observed that empiric BL therapy, although highly effective, is of broader spectrum than necessary. RMD-guided therapy has the potential to reduce overtreatment without compromising effective therapy. DOOR-MAT provides a flexible framework for measuring appropriateness of therapy for BSI.


Brigid Wilson, PhD1, Roberto Viau, MD1, Federico Perez, MD, MS2, Hongyu Jiang, MS3, Vance G. Fowler Jr., MD4, Henry F. Chambers, MD5, Barry N. Kreiswirth, PhD6, Robert A. Bonomo, MD1, Scott R. Evans, PhD7 and ARLG, (1)Louis Stokes Cleveland Department of Veterans Affairs Medical Center, Cleveland, OH, (2)Louis Stokes Cleveland VA Medical Center, Cleveland, OH, (3)Biostatistics, Harvard School of Public Health, Boston, MA, (4)Duke University Medical Center, Durham, NC, (5)Clinical Research Services, University of California San Francisco, Clinical and Translational Sciences Institute, San Francisco, CA, (6)Public Health Research Institute, Rutgers New Jersey Medical School, Newark, NJ, (7)Harvard School of Public Health, Boston, MA


B. Wilson, None

R. Viau, None

F. Perez, None

H. Jiang, None

V. G. Fowler Jr., None

H. F. Chambers, None

B. N. Kreiswirth, None

R. A. Bonomo, None

S. R. Evans, None

Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 3rd with the exception of research findings presented at the IDWeek press conferences.