Importance: Management of gram-negative bloodstream infections (GN-BSIs) with oral antibiotics is highly variable.
Objective: To examine the transition from intravenous (IV) to oral antibiotics, including selection, timing, and associated clinical and microbial characteristics, among hospitalized patients with GN-BSIs.
Design, setting, and participants: A retrospective cohort study was conducted of 4581 hospitalized adults with GN-BSIs at 24 US hospitals between January 1 and December 31, 2019. Patients were excluded if they died within 72 hours. Patients were excluded from the oral therapy group if transition occurred after day 7. Statistical analysis was conducted from July 2022 to October 2023.
Exposures: Administration of antibiotics for GN-BSIs.
Main outcomes and measures: Baseline characteristics and clinical parameters reflecting severity of illness were evaluated in groups receiving oral and IV therapy. The prevalence of transition from IV to oral antibiotics by day 7, median day of transition, sources of infection, and oral antibiotic selection were assessed.
Results: Of a total of 4581 episodes with GN-BSIs (median age, 67 years [IQR, 55-77 years]; 2389 men [52.2%]), 1969 patients (43.0%) receiving IV antibiotics were transitioned to oral antibiotics by day 7. Patients maintained on IV therapy were more likely than those transitioned to oral therapy to be immunosuppressed (833 of 2612 [31.9%] vs 485 of 1969 [24.6%]; P < .001), require intensive care unit admission (1033 of 2612 [39.5%] vs 334 of 1969 [17.0%]; P < .001), have fever or hypotension as of day 5 (423 of 2612 [16.2%] vs 49 of 1969 [2.5%]; P < .001), require kidney replacement therapy (280 of 2612 [10.7%] vs 63 of 1969 [3.2%]; P < .001), and less likely to have source control within 7 days (1852 of 2612 [70.9%] vs 1577 of 1969 [80.1%]; P < .001). Transitioning patients from IV to oral therapy by day 7 was highly variable across hospitals, ranging from 25.8% (66 of 256) to 65.9% (27 of 41). A total of 4109 patients (89.7%) achieved clinical stability within 5 days. For the 3429 episodes (74.9%) with successful source control by day 7, the median day of source control was day 2 (IQR, 1-3 days) for the oral group and day 2 (IQR, 1-4 days) for the IV group (P < .001). Common infection sources among patients administered oral therapy were the urinary tract (1277 of 1969 [64.9%]), hepatobiliary (239 of 1969 [12.1%]), and intra-abdominal (194 of 1969 [9.9%]). The median day of oral transition was 5 (IQR, 4-6 days). Total duration of antibiotic treatment was significantly shorter among the oral group than the IV group (median, 11 days [IQR, 9-14 days] vs median, 13 days [IQR, 8-16 days]; P < .001]. Fluoroquinolones (62.2% [1224 of 1969]), followed by β-lactams (28.3% [558 of 1969]) and trimethoprim-sulfamethoxazole (11.5% [227 of 1969]), were the most commonly prescribed oral antibiotics.
Conclusions and relevance: In this cohort study of 4581 episodes of GN-BSIs, transition to oral antibiotic therapy by day 7 occurred in fewer than half of episodes, principally with fluoroquinolones, although this practice varied significantly between hospitals. There may have been additional opportunities for earlier and more frequent oral antibiotic transitions because most patients demonstrated clinical stability by day 5.