Background. Contaminated duodenoscopes used in endoscopic retrograde cholangiopancreatography (ERCP) have been implicated in carbapenem-resistant Enterobacteriaceae infections. Multi-locus sequence typing (MLST) can identify outbreaks with rare pathogens, but it is unclear how well it can link transmission of common Enterobacteriaceae, e.g., ceftriaxone-resistant (CRO-R) Escherichia coli. Our Infection Control Unit (IC) was alerted to a suspicious number of CRO-R E. coli bloodstream infections

more »

... n patients who had undergone ERCP in the Spring of 2014. IC investigated an association with duodenoscopes and made recommendations for surveillance and reprocessing in accordance with CDC guidelines. Methods. We identified all blood cultures positive for CRO-R E. coli collected January 2014-June 2015. A case was defined as the first CRO-R E. coli positive blood culture from a unique patient who had undergone ERCP within 90 days prior. Controls were CRO-R E. coli blood cultures from unique patients who did not undergo ERCP or did so after bacteremia. Duodenoscopes used for ERCP were abstracted from the medical record. We evaluated clonal relationships by MLST and single-nucleotide polymorphism (SNP) profiles. Resistance gene and mobilizing vector analysis provided strain resolution. Results. We identified 52 blood cultures positive for CRO-R E. coli during the study period. Twenty-eight met the case definition and 11 were controls. Thirty of the 39 isolates belonged to MLST-131. Eleven of these isolates were from patients exposed to the same duodenosope within 90 days prior to bacteremia. CTXM-15 mediated CRO-R in 10 of the exposed cases. The odds ratio of bacteremia with E. coli MLST-131 if exposed to this duodenoscope was 4.6 but was not statistically significant (95% CI 0.5-42.1). Hamming distance SNP trees showed common ancestry but not clonality among the cases (separation <500 chromosomal SNPs, <1000 plasmid SNPs). Conclusion. The implicated duodenoscope was removed from use despite inability to confirm clonality. The study was underpowered to demonstrate statistical significance because our cases involved highly prevalent strains of CRO-R E. coli. While genomic analyses can identify epidemiologic associations, in outbreaks due to common pathogens a high level of suspicion should dictate intervention.