Big Data- Epidemiology of Antimicrobial Resistance

Clinical Center

Investigators

Linked publications, trials & patents

Abstract

Co-resistance necessitating use of less effective or relatively toxic reserve antibiotics (aminoglycosides, tigecycline and colistin/polymyxin B) may worsen survival. We investigated difficult-to-treat resistance (DTR) in gram-negative bloodstream infections (GNBSIs) defined by absence of susceptibility to all first-line agents (carbapenems, beta-lactams and fluoroquinolones (FQ) using a large clinical database of US hospitals. We found that survival in antimicrobial-resistant GNBSI is highly contingent on presence of active first-line option(s); DTR limits treatment options to reserve agents, including aminoglycosides, which are far from universally active. DTR remained infrequent (1%) among GNBSI, but occurred at half of the hospitals examined and across all US regions. This work has been presented at the Annual Meetings of the Infectious Diseases Society of America and Society of Critical Care Medicine. We went on to validate our findings using the Cerner Health facts repository of electronic health records. As part of the NIH Antimicrobial Resistance Outcomes Research Initiative (NIH-ARORI), we also went on to study the landscape of emerging antibiotics to understand their real-world use and demand. We found that ceftazidime-avibactam use increased several fold replacing colistin use, however overall use was still modest. As part of an FDA-funded initiative, we determined that treatment opportunities for difficult-to-treat antibiotic resistant pathogens in US hospitals remains small, suggesting that non-revenue based strategies might be necessary to sustain antibiotic development. As part of ongoing work, we are performing national extrapolations to understand recent trends in difficult-to-treat resistance, conducting an analysis of burden and impact of inappropriate antibiotic therapy in bacteremia and sepsis. As part of the NIH Antimicrobial Resistance Outcomes Research Initiative (NIH-ARORI), we determined that one in every five patients with bloodstream infection in US hospitals receives empiric antibiotic therapy that is discordant with in vitro susceptibilities. This practice was prevalent and similar across hospital types. Bacteremic patients who received in vitro-discordant empiric therapy displayed a higher mortality risk than recipients of concordant therapy. S. aureus and Enterobacterales and their resistance phenotypes account for the overwhelming majority of burden and impact of in vitro-discordant therapy, warranting development and wide implementation of effective rapid point-of-care diagnostics, especially those targeting these pathogens and their resistance phenotypes. In collaboration with the investigators at the Harvard Department of Population Medicine, we used larger electronic health record data from over a hundred U.S. hospitals and determined that despite the extensive use of broad-spectrum empiric antibiotic therapy for community-onset sepsis, the prevalence of antibiotic resistant pathogens warranting such therapy is small. Rapid diagnostics targeting resistance phenotypes may have a role in enhancing antibiotic stewardship in sepsis.