Thomas P. Lodise

Study Objective Bloodstream infections (BSIs) are a significant cause of mortality. Use of a rapid multiplex polymerase chain reaction‐based blood culture identification panel (BCID) may improve antimicrobial utilization and clinical outcomes by shortening the time to appropriate therapy and de‐escalating antibiotics among patients on overly broad‐spectrum empiric therapy. The effect of BCID on clinical outcomes across varying institutional antimicrobial stewardship program (ASP) practices is unclear. This study evaluated clinical outcomes associated with the “real‐world” implementation of BCID in a national health system with varying ASP practices. Design National, multicenter, retrospective, pre–post quasi‐experimental study of hospitalized patients admitted from 2015 to 2020 to VHA facilities, which introduced the BCID for ≥1 year. Setting United States Veterans Health Administration (VHA) hospitals with BCID. Patients Hospitalized VHA patients with ≥1 blood culture positive for bacteria featured on the BCID panel. Intervention Comparison of outcomes between the pre‐ and post‐BCID implementation groups. Measurements Outcomes evaluated included early antimicrobial de‐escalation within 48 h, defined as reduction in antimicrobial spectrum scores, time to appropriate therapy, and 30‐day mortality. Main Results A total of 4138 patients (pre‐BCID, n = 2100; post‐BCID, n = 2038) met the study criteria. Implementation of BCID was associated with significant improvements in early antimicrobial de‐escalation (34.6%: pre‐BCID vs. 38.1%: post‐BCID; p = 0.022), which persisted after adjusting for other covariates (adjusted risk ratio [aRR], 1.11; 95% confidence interval [CI], 1.02–1.20; p = 0.011). Median time to appropriate therapy was shorter in the post‐BCID implementation group relative to the pre‐BCID group (9 h: pre‐BCID vs. 8 h: post‐BCID, respectively, p = 0.005), and a greater percentage of patients received early appropriate antimicrobial therapy within 48 h in the post‐BCID implementation group (91.7%: pre‐BCID vs. 93.8%: post‐BCID; p = 0.008). In the multivariable regression analysis, BCID implementation was significantly associated with a higher likelihood of appropriate therapy within 48 h (aRR, 1.02; 95% CI, 1.01–1.08; p = 0.020). There was no difference in 30‐day mortality between groups overall (12.6% pre‐BCID vs. 11.2% post‐BCID; p = 0.211). Conclusions In a “real‐world” clinical setting, the implementation of BCID was associated with clinical improvements in antimicrobial utilization. The BCID platform may serve as a useful adjunct for BSI management in facilities with ASP.