Impact of Difficult-to-Treat Resistance in Gram-negative Bacteremia on Mortality: Retrospective Analysis of Nationwide Surveillance Data

Abstract

Abstract Background Clinically relevant categorization of antimicrobial resistance is critical to mitigating the threat it poses. Difficult-to-treat resistance (DTR) is a recently proposed category defined as nonsusceptibility to all first-line antibiotic agents. Methods A retrospective study was conducted with nonduplicate cases of gram-negative bloodstream infection (GNBSI) caused by 4 major taxa (Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, and Acinetobacter species) identified from a nationwide surveillance database. DTR was defined as nonsusceptibility to all the β-lactams and fluoroquinolones tested. Patient characteristics and mortality were compared between DTR GNBSI and GNBSI caused by carbapenem-resistant but not DTR and extended-spectrum cephalosporin–resistant but not DTR isolates using Centers for Disease Control and Prevention definitions. Adjusted odds ratios (aORs) for 30-day in-hospital mortality were examined for DTR in overall and in propensity score–matched cohorts. Results A total of 1167 episodes of monomicrobial GNBSI were identified, and 147 (12.6%) of the isolates were DTR. The majority of DTR isolates were Acinetobacter species (79.6%) and P. aeruginosa (17.7%). DTR infections were associated with previous antibiotic use, healthcare contact, ventilator use, and lower respiratory tract infection. Crude mortality for GNBSI caused by DTR was 50.3%. A multivariable model showed that only DTR, but not other categories, was significantly associated with mortality (adjusted odds ratio [aOR], 3.58 [95% confidence interval {CI}, 1.27–10.19]). DTR was also a significant predictor for mortality in the analysis of propensity score–matched cohorts (aOR, 3.48 [95% CI, 1.82–6.79]). Conclusions In patients with GNBSI, DTR was associated with higher mortality than those in other resistance categories. Our findings suggest that DTR could be useful for surveillance and prognostication.