🧭 REBEL Rundown
📌 Key Points
- 🌍 Generalizability: Mostly Canadian patients with gram-negative UTIs or abdominal infections.
- 🔍 Bias: Multiple biases may have made 7 days look better than it is.
- ⚖️ Margin: Allowed up to 1 in 25 more deaths—though stricter than past trials.
- 🧫 Gram-Positive? Still not enough data for other bugs or sources.
- 🧾 Bottom Line: BALANCE showed 7 days is noninferior to 14 for sepsis.
📝 Introduction
Current IDSA guidelines for sepsis recommend individualized durations of antibiotic therapy based on source control and clinical response, but definitive guidance remains limited.1 Three small noninferiority RCTs suggested that 7 days of antibiotics may be sufficient for patients with gram-negative bacteremia—but these trials had notable limitations: they excluded ICU patients, included only clinically improving individuals, and did not assess gram-positive infections.2–4 The BALANCE trial, a large multicenter RCT, aimed to address these gaps and strengthen the evidence base for shorter-course antibiotic strategies in a broader sepsis population.5 But does The BALANCE Trial tip the scales in favor of 7 days of antibiotics for sepsis? Let’s break it down.
🧾 Paper
BALANCE Investigators, for the Canadian Critical Care Trials Group, the Association of Medical Microbiology and Infectious Disease Canada Clinical Research Network, the Australian and New Zealand Intensive Care Society Clinical Trials Group, and the Australasian Society for Infectious Diseases Clinical Research Network, Daneman N, Rishu A, et al. Antibiotic Treatment for 7 versus 14 Days in Patients with Bloodstream Infections. N Engl J Med. 2025;392(11):1065-1078. PMID: 39565030
⚙️ What They Did
📈 Results
- Screening: 36,637 patients with positive blood cultures
- Eligible: 13,597 (37.1%) met inclusion criteria
- Enrolled: 3,631 patients (26.7% of eligible)
- Analyzed: 3,608 in the intention-to-treat population
- 7-day group: 1,814 patients
- 14-day group: 1,794 patients
- Baseline characteristics: Similar across both groups
- Median age: 70 years (IQR 59–80)
- Sex: 53.3% male
- Setting at enrollment:
- ICU: 55.0%
- Community-onset: 75.4%
- Hospital ward-acquired: 13.4%
- ICU-acquired: 11.2%
- Infection sources:
- Urinary tract: 42.2%
- Intraabdominal/hepatobiliary: 18.8%
- Lung: 13.0%
- Vascular catheter: 6.3%
- Skin/soft tissue: 5.2%
- Pathogen types:
- Monomicrobial gram-negative: 71.0%
- Monomicrobial gram-positive: 17.3%
- Polymicrobial: 11.7%
💥 Critical Results
Primary Outcome: Intention-to-treat:
- Mortality at 90 Days:
- 7-day group: 261 of 1802 patients (14.5%)
- 14-day group: 286 of 1779 patients (16.1%)
- Difference: −1.6 percentage points
- 95.7% CI: −4.0 to 0.8
- Met criteria for noninferiority
- Mortality at 90 Days:
Per-Protocol Analysis:
- Mortality difference: −2.0% (95% CI: −4.5 to 0.6)
- Noninferiority preserved
Modified Intention-to-Treat (excluding patients who died before day 7):
- Mortality difference: −1.6% (95% CI: −3.9 to 0.7)
- Noninferiority preserved
Secondary Outcomes:
- No statistically significant difference in any secondary outcome
💪 Strengths
- Large sample size: Enhances statistical power and precision of estimates.
- Protocol published in advance: Reduces risk of bias through prespecified analysis plans.
- Trial registered with ClinicalTrials.gov (NCT03005145): Promotes transparency and accountability.
- Block randomization stratified by site and ICU status: Ensures balanced allocation across key subgroups.
- Well-balanced groups after randomization: Baseline characteristics were similar, reducing confounding and enhancing internal validity.
- ITT and per-protocol analyses: Provide robust, complementary assessments of the primary outcome.
- Broad inclusion criteria: Supports generalizability across a wide spectrum of patients with bloodstream infections.
- Included ICU and non-ICU patients: Reflects real-world practice across illness severity
- Patients blinded to treatment until day 7: Limits bias in early clinical management despite open-label design.
- Patient-centered primary outcome (90-day mortality): Directly relevant to patient survival and clinical decision-making.
- Objective Secondary outcomes: Use of hard endpoints like mortality minimizes measurement bias.
- Minimal loss to follow-up (0.7%): Ensures data completeness and reliability of results.
⚠️ Limitations
- Physician discretion allowed at enrollment: May have introduced selection bias despite appropriate exclusion criteria.
- High number of eligible but non-randomized patients: Nearly 10,000 eligible patients were not enrolled, potentially affecting representativeness.
- Majority of patients enrolled in Canada (75%): Limits global generalizability despite international participation.
- Open-label design: Lack of blinding after day 7 may have influenced co-interventions or clinical behavior.
- Baseline imbalance in illness severity: Slightly higher SOFA scores in the 14-day group could confound outcome comparisons.
- No granular data on co-interventions: Procedures and other supportive therapies were not uniformly reported, limiting interpretability.
- No data on route or dosing of antibiotics: Limits insight into how treatment strategies varied beyond duration alone.
- More procedural interventions in 14-day group: Suggests potential differences in clinical course that were not accounted for.
- Pathogen profile skewed toward gram-negative infections (71%): Limits applicability to gram-positive or mixed infections.
- Infection source concentrated in urinary (42%) and hepatobiliary (18.8%) tracts: May not reflect broader spectrum of bacteremia presentations.
🗣️ Discussion
BIAS BENEATH THE SURFACE
Sampling Bias: The median enrollment age was 70 years old, 75% of patients were from Canada, over 70% had gram-negative monomicrobial infections, with 60% from urinary tract or intra-abdominal infections. These factors limit the trial’s generalizability to younger patients from other countries, and those with gram positive infections from other sources.
Selection Bias: The enrollment rate was surprisingly low—one patient every two months per site. 4,506 were excluded solely based on clinician judgment. It’s possible that clinicians—consciously or unconsciously—excluded sicker patients who they believed would not do well with shorter treatment durations. If so, the trial population may have skewed toward lower-acuity cases, artificially improving outcomes and making it easier to demonstrate noninferiority.
Spectrum bias: Although 1 in 5 patients required mechanical ventilation and 1 in 3 received vasopressors, the supplemental data suggest that most patients were not considerably ill—reflected by the lower-than-expected 90-day mortality of 15.3%, compared to the anticipated 22%. When enrolling many lower-acuity patients demonstrating noninferiority becomes easier—not because the intervention is truly noninferior, but because the risk of adverse outcomes is low regardless of the treatment.
Co-Intervention Bias & Adherence Bias: ICU Patients admitted with sepsis often receive multiple therapies and interventions. Yet the trial provides limited details beyond the antibiotic choice and duration. Moreover, nonadherence to the assigned duration was common: 23.1% of patients in the 7-day group received antibiotics for longer than 9 days, while 5.8% in the 14-day group were treated for fewer than 12 days. The lack of granularity and protocol deviations make it difficult to assess whether other aspects of care influenced outcomes and may have diluted the difference between groups.
NONINFERIORITY—AT WHAT COST?
All noninferioriy trials require that clinicians accept that an experimental treatment is “not much worse” than the standard, but offers other potential advantages as a tradeoff. The noninferiority margin determines exactly how much worse the experimental treatment can be before it’s considered inferior. Investigators selected a 4% margin—stricter than the 10% used in prior studies. Noninferiority trials can raise ethical concerns, and accepting a potentially worse treatment, can feel deeply at odds with our core beliefs. Many previous trials exploring short-course antibiotics focused on clinical cure, and the consequences of treatment failure—needing a few more days of antibiotics—are more acceptable. When weighed against the risks of C. difficile, adverse drug events, lower costs or antimicrobial resistance, It’s hard to imagine that many would accept a higher chance of death.
SUBGROUPS THAT CHALLENGE THE EVIDENCE
In several prespecified subgroups, the upper bound of the 95% confidence interval for mortality exceeded the noninferiority margin, including:
- Patients with elevated frailty scores (CFS ≥5)
- Those requiring vasopressors
- Patients with high APACHE II scores (≥25)
- Vascular-catheter–associated infections
- Pulmonary infections
- Polymicrobial bacteremia
- Gram-positive organisms
- Infections of undefined or miscellaneous sources
These analyses were exploratory and underpowered, but the pattern is hard to ignore. Short-course therapy works. But maybe not for everyone.
📊 State of The Evidence
The evidence guiding antibiotic duration in sepsis is limited and largely focused on patients with gram-negative bacteremia who are clinically improving. The BALANCE trial adds to this data, showing that 7 days is noninferior to 14, while expanding to include ICU patients, diverse infection sources, and an international cohort. Though not perfect, it’s the strongest evidence we have so far. Seven days may not be right for every patient—especially those with gram-positive infections—but for most, it’s likely enough.
Start with 7 days, then individualize treatment based on source control, and clinical response. The BALANCE Trial pushes us closer to smarter, shorter therapy.
📘 Author's Conclusion
“Among hospitalized patients with bloodstream infection, antibiotic treatment for 7 days was noninferior to treatment for 14 days.”
💬 Our Conclusion
The BALANCE trial demonstrated that 7 days of antibiotics is noninferior to 14 days in patients with sepsis. However, methodological flaws and numerous biases may skew the results toward noninferiority and limit generalizability—especially to gram-positive infections. Nonetheless, clinicians should consider starting with 7 days of antibiotic therapy, then tailor care based on source control and clinical response.
🚨 Clinical Bottom Line
In patients with sepsis, the best available evidence suggests we start with 7 days of antibiotics, then individualize care and extend therapy based on source control and clinical response.
🔄 REBEL Recap
📚 References
Evans L, Rhodes A, Alhazzani W, et al.
Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock 2021.
Crit Care Med. 2021;49(11):e1063–e1143.
PMID: 34605781Yahav D, Franceschini E, Koppel F, et al.
Seven Versus 14 Days of Antibiotic Therapy for Uncomplicated Gram-negative Bacteremia: A Noninferiority Randomized Controlled Trial.
Clin Infect Dis. 2019;69(7):1091–1098.
PMID: 30535100von Dach E, Albrich WC, Brunel AS, et al.
Effect of C-Reactive Protein–Guided Antibiotic Treatment Duration, 7-Day Treatment, or 14-Day Treatment on 30-Day Clinical Failure Rate in Patients With Uncomplicated Gram-Negative Bacteremia: A Randomized Clinical Trial.
JAMA. 2020;323(21):2160–2169.
PMID: 32484534Molina J, Montero-Mateos E, Praena-Segovia J, et al.
Seven-versus 14-day course of antibiotics for the treatment of bloodstream infections by Enterobacterales: a randomized, controlled trial.
Clin Microbiol Infect. 2022;28(4):550–557.
PMID: 34508886- Daneman N, Rishu A, et al.
Antibiotic Treatment for 7 versus 14 Days in Patients with Bloodstream Infections.
N Engl J Med. 2025;392(11):1065-1078.
PMID: 39565030
👤 Co Editor-In-Chief
🔎 Your Deep-Dive Starts Here
The ADAPT-Sepsis Trial: Biomarker-Guided Antibiotic Duration for Hospitalized Patients with Suspected Sepsis
Antibiotic stewardship entails delivering the most appropriate antimicrobial therapy for ...
Clinical Conundrum: Should We Always Treat Fever in Patients with Sepsis?
Bottom Line Up Top: There doesn’t appear to be a ...
Is Pip-Tazo Harming Sepsis Patients?
Background: Patients with sepsis are routinely treated with empiric broad-spectrum ...
The ACORN Trial: Battle of the Gorilla-Cillins (Cefepime vs Piperacillin-Tazobactam)
Background: Acutely ill adults presenting to the hospital with suspected ...
Pre-Hospital Antibiotics in Sepsis?
Background: Sepsis remains one of the leading causes of morbidity ...
Congestive Heart Failure and Sepsis: A Closer Look at Fluid Management
Background: In medicine, guidelines are valuable tools that help guide ...
