668 - Rapid Identification and Antimicrobial Susceptibility Testing of Gram-Negative Rod Bacteremia: Impact on Duration of Hospital Admission

Background: There are many emerging diagnostic technologies that provide rapid identification (ID) and antimicrobial susceptibility testing (AST) of blood cultures after initial microbial growth is detected. The clinical utility of having these results sooner than standard of care (SOC) ID/AST testing is less clear; one potential advantage would be shorter hospital admissions for those patients with less severe infections whose discharge planning is delayed while waiting for ID/AST results.

Methods: The Accelerate Pheno™ rapid ID/AST system (AP) was implemented in the clinical microbiology lab at two hospitals within our academic hospital system to run in parallel with our SOC ID/AST diagnostics for all blood cultures with Gram-negative rod (GNR) bacterial growth on initial Gram stain. Results of both diagnostic assays were uploaded to the medical record when available; our antimicrobial stewardship (AS) team discussed the AP results with clinical care teams during working hours. Length of stay (LOS) data was compared between a cohort with GNR bacteremia before and after the implementation of AP testing with AS review.

Results: There were 207 patients with GNR bacteremia admitted to the hospital during the first six months after implementation of AP testing with AS review (12/2019 – 05/2020). A historical cohort of 96 patients with GNR bacteremia (admitted 01/2018 – 08/2018) with SOC testing without AS review was used for comparison.  Median LOS was 5.28 and 5.84 days for the AP and SOC cohorts, respectively; the distribution in the two groups differed significantly (Mann-Whitney U = 8728.5, p = 0.0446). There was also an increase in the percent of admissions shorter than 2 days between the AP and SOC cohorts (12.5% v. 5.2%, p = 0.024) [Figure1].Figure 1


Conclusion: The use of a rapid ID/AST diagnostic technology with AS review may lead to shorter hospitalizations in select patient populations.