Seven vs Fourteen Days of Antibiotics for Gram-Negative Bloodstream Infection: A Systematic Review and Noninferiority Meta-Analysis

Abstract

Importance: Gram-negative bloodstream infections are a common cause of hospitalization. A 2-week duration of antibiotic therapy has been commonly used, but shorter durations may have similar outcomes.

Objectives: To assess whether 7 days of antibiotic therapy was noninferior to 14 days.

Data sources: Starting with a 2022 individual patient data meta-analysis, PubMed, Cochrane Central Register of Controlled Trials, and Web of Science were searched to identify additional eligible randomized clinical trials (RCTs) conducted from May 1, 2022, until November 30, 2024.

Study selection: RCTs involving primarily adults who were hospitalized at the time of Gram-negative bloodstream infection and were allocated to 7 or 14 days of antibiotic therapy. Studies were independently reviewed by 2 investigators.

Data extraction and synthesis: PRISMA guidelines were followed. Data were extracted by 2 investigators. Any unpublished data were obtained directly from study authors. Risk of bias and certainty of the evidence were assessed in duplicate using the Cochrane Risk of Bias Tool, version 2, and the Grading of Recommendations Assessment, Development and Evaluation approach. Data were pooled by separate random-effects meta-analyses for the intention-to-treat (ITT) and per-protocol (PP) populations. A noninformative prior probability was used for the effect, and an evidence-based weakly informative prior probability was used for heterogeneity. Risk ratios (RRs), 95% credible intervals (CrIs), and probability of noninferiority were calculated using a prespecified upper bound of 1.25 or less.

Main outcomes and measures: Ninety-day all-cause mortality.

Results: Four eligible RCTs contributed 3729 patients in the ITT population (1912 women [51.3%]; median age range, 67-79 years) and 3126 in the PP population. In the ITT analysis, within 90 days, 226 patients (12.8%) receiving 7 days of antibiotics died compared with 253 (13.7%) receiving 14 days, corresponding to an RR for 90-day mortality of 0.91 (95% CrI, 0.69-1.22) and a 97.8% probability of noninferiority. In the PP analysis, the RR was 0.93 (95% CrI, 0.68-1.32), corresponding to a 95.1% probability of noninferiority.

Conclusions and relevance: In this systematic review and meta-analysis of patients with Gram-negative bloodstream infections and adequate source control, 7 days of antibiotic therapy had a high probability of being noninferior to 14 days. These findings support a shorter duration of antibiotic therapy for appropriately selected patients like those in the included RCTs.

Figure 1.. PRISMA Diagram

PRISMA indicates Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2.. Forest Plot of Included Studies

The top panel shows the intention-to-treat results and the lower panel the per-protocol results. The prespecified noninferiority margin (risk ratio [RR], 1.25) is indicated by the vertical dotted line. The vertical dashed lines show the point estimate for the pooled results. Size of squares indicates the relative weight of the individual study; diamonds, the pooled RR and 95% credible interval (CrI). BALANCE indicates Bacteremia Antibiotic Length Actually Needed for Clinical Effectiveness.

Figure 3.. Probability of Noninferiority (or Superiority) as a Function of the Upper Bound

The y-axis represents the probability of a result less than or equal to the x-axis value. For example, at the prespecified upper bound of 1.25, the probability that the risk ratio (RR) is 1.25 or less is 97.8% for the intention-to-treat analysis and 95.1% for the per-protocol analysis. The dashed line at 1.00 represents superiority (RR is ≤1.00); the dashed line at 1.25 represents noninferiority.