Publication

• Title: Antibiotic Treatment for 7 versus 14 Days in Patients with Bloodstream Infections
• Acronym: BALANCE
• Year: 2024
• Journal published in: The New England Journal of Medicine
• Citation: The 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. Antibiotic treatment for 7 versus 14 days in patients with bloodstream infections. N Engl J Med 2025;392:1065-1078.

Context & Rationale

• Background

  Bloodstream infection (BSI) is common in hospital and ICU practice, with substantial attributable morbidity and mortality and high antibiotic utilisation.
  Historically, 10–14 days of antibiotics has been routine for many BSIs despite limited randomised evidence, partly reflecting concern for relapse, metastatic infection, and untreated foci.
  Prior RCT evidence supporting shorter courses largely focused on selected, “uncomplicated” Gram-negative bacteraemia populations, often excluding critically ill patients, Gram-positive pathogens, and complex infection syndromes.
• Research Question/Hypothesis

  Whether a 7-day antibiotic strategy is non-inferior to a 14-day strategy for all-cause mortality at 90 days among hospitalised adults with BSI caused by pathogenic bacteria (excluding syndromes/pathogens requiring prolonged therapy).
• Why This Matters

  If shorter courses are safe, the absolute reduction in antibiotic exposure at a population level could be large, with downstream effects on adverse drug events, C. difficile, selection for antimicrobial resistance, ICU resource use, and stewardship targets.

Design & Methods

• Research Question: In hospitalised adults with BSI, is a 7-day antibiotic strategy non-inferior to a 14-day strategy for 90-day all-cause mortality?
• Study Type: Pragmatic, randomised, international, multicentre, parallel-group, non-inferiority trial conducted across 74 hospitals in 7 countries (Canada, Australia, New Zealand, Saudi Arabia, Israel, Switzerland, United States); ICU and hospital ward settings; group assignment concealed until day 7 of adequate treatment, then revealed; open-label thereafter; outcome adjudication and analyses blinded.
• Population:
  • Adults (≥18 years) with ≥1 blood culture positive for pathogenic bacteria, admitted to hospital/ICU at the time the index blood culture was drawn or resulted positive.
  • Key exclusions: severe immune compromise (e.g., neutrophils <0.5×10⁹/L; solid-organ/bone marrow transplant immunosuppression); prosthetic heart valve or synthetic endovascular graft; suspected infection syndrome requiring prolonged therapy (e.g., endocarditis, osteomyelitis/septic arthritis, undrainable abscess, unremovable prosthetic infection); single positive blood culture with a common contaminant; Staphylococcus aureus or S. lugdunensis; Candida or other fungal species.
• Intervention:
  • 7-day strategy: Stop adequate systemic antibiotic therapy after 7 days of treatment for the index BSI (with clinical antibiotic choice otherwise at clinician discretion, consistent with pragmatic stewardship intent).
• Comparison:
  • 14-day strategy: Continue adequate systemic antibiotic therapy to complete 14 days for the index BSI (agent selection and switching permitted pragmatically).
• Blinding: Allocation concealed until the seventh day after the first positive blood culture (minimising early differential management); clinicians unblinded thereafter; objective primary endpoint (mortality) and blinded outcome adjudication/analyses mitigate, but do not eliminate, performance and detection bias risks.
• Statistics: Non-inferiority margin 4 percentage points for absolute risk difference in 90-day mortality; target sample size 3626 to provide 80% power at a one-sided alpha of 2.5% (assumed 22% 90-day mortality in the 14-day group; 5% loss to follow-up; 3 interim analyses leading to a 95.7% CI for the primary analysis); primary analysis by intention-to-treat, with prespecified per-protocol and modified intention-to-treat analyses.
• Follow-Up Period: 90 days for primary outcome (all-cause mortality); selected secondary and safety outcomes assessed through day 28 and during hospitalisation.

Key Results

This trial was not stopped early. Enrolment and follow-up proceeded to completion with prespecified interim analyses; 3608 patients were randomised (1814 to 7 days; 1794 to 14 days).

• Primary endpoint: 90-day mortality was 14.5% with 7 days vs 16.1% with 14 days; risk difference −1.6 percentage points (95.7% CI −4.0 to 0.8), meeting non-inferiority versus a 4 percentage-point margin.
• Exposure reduction: antibiotic-free days to day 28 were 19 (IQR 11–21) vs 14 (IQR 11–14); median difference 5 (95% CI 4.6 to 5.4), demonstrating strong separation in antibiotic exposure.
• Safety signals: relapse (2.6% vs 2.2%), C. difficile (1.7% vs 2.0%), and resistant-organism outcomes (9.5% vs 8.5%) were broadly similar, with confidence intervals consistent with small benefit or harm.

Internal Validity

• Randomisation and allocation: Central randomisation with stratification (including enrolment location) and concealment of group assignment until day 7 mitigated selection bias and reduced early differential management.
• Dropout/exclusions: 3608 randomised; 15 patients lost to follow-up for the primary endpoint (day-90 mortality), preserving near-complete outcome ascertainment for an objective endpoint.
• Performance/detection bias: Trial was open-label after day 7; however, primary outcome was objective and outcome adjudication/analyses were blinded, reducing detection bias for microbiological and resistance outcomes.
• Protocol adherence: Nonadherence to assigned duration occurred in 23.1% (7-day) vs 10.7% (14-day), introducing dilution towards the null (and towards non-inferiority) in intention-to-treat; supportive per-protocol analysis (13.8% vs 15.6%) showed concordant direction.
• Baseline comparability and severity: Groups were clinically similar at baseline (median age 70 years in both; ICU enrolment 55.1% vs 54.2%; SOFA 4 [IQR 2–6] vs 5 [IQR 2–7]; mechanical ventilation 20.6% vs 21.9%).
• Timing: Allocation concealment until day 7 (when strategies diverge) is methodologically aligned with the causal estimand (duration beyond 7 days), limiting immortal-time type artefacts while still capturing early deaths in the primary (intention-to-treat) analysis.
• Dose and separation of the variable of interest:
  • Duration of antibiotic therapy from randomisation: 8 days (IQR 7–10) vs 14 days (IQR 13–15).
  • Antibiotic-free days to day 28: 19 (IQR 11–21) vs 14 (IQR 11–14).
• Heterogeneity: Broad case-mix (community- and hospital-acquired infection; ICU and ward; Gram-negative 71.6% vs 70.4%, Gram-positive 17.8% vs 16.8%, polymicrobial 10.6% vs 12.8%; urinary tract ~42%, intraabdominal ~19%, lung ~13%) enhances pragmatic relevance but widens biologic heterogeneity; consistency of effect across many subgroups reduces, but does not eliminate, concern for major effect modification.
• Outcome definitions and measurement: Mortality is robust; relapse, adverse events, and resistant-organism outcomes were prespecified and adjudicated with blinded analyses; secondary outcomes were not multiplicity-adjusted.
• Statistical rigour: Non-inferiority design with prespecified margin and alpha spending (95.7% CI for primary analysis); prespecified sensitivity analyses (per-protocol and modified intention-to-treat) showed directionally consistent results.

Conclusion on Internal Validity: Overall, internal validity appears moderate to strong, driven by robust randomisation with delayed reveal and an objective primary outcome, but tempered by open-label management after day 7 and higher nonadherence in the 7-day arm (which biases towards non-inferiority in intention-to-treat analyses).

External Validity

• Population representativeness: Broad adult hospital population with BSI spanning ICU and ward care, multiple infection sources (urinary, intraabdominal, lung, catheter-related, skin/soft tissue, undefined), and both Gram-negative and selected Gram-positive pathogens.
• Key exclusions limiting generalisability: S. aureus/S. lugdunensis, fungal bloodstream infection, severe immune compromise (including profound neutropenia and transplant immunosuppression), prosthetic valve/synthetic graft, and syndromes requiring prolonged treatment (endocarditis, osteomyelitis/septic arthritis, undrainable abscess, unremovable prosthetic infection).
• Setting applicability: Conducted across 74 hospitals in high- and middle-income health systems; implementation likely feasible in comparable stewardship and microbiology environments; translation to resource-limited settings depends on the ability to ensure adequate early therapy and source control.
• Clinical decision-making: Findings most directly apply to patients with controlled sources and no clinical syndrome mandating extended treatment; caution remains for populations excluded by design.

Conclusion on External Validity: Generalisability is moderate: the trial supports a 7-day strategy for a large proportion of hospitalised adults with uncomplicated bacterial BSI (including ICU patients), but should not be extrapolated to excluded pathogens, fungal BSI, severe immunocompromise, or infection syndromes requiring prolonged therapy.

Strengths & Limitations

• Strengths:
  • Large, pragmatic, international RCT addressing a high-volume stewardship question.
  • Methodologically thoughtful allocation concealment until day 7, aligning with the intervention’s causal window.
  • Objective primary endpoint (90-day all-cause mortality) with high completeness of follow-up.
  • Broad inclusion of ICU patients and aetiologic heterogeneity (beyond “uncomplicated Gram-negative” cohorts).
  • Prespecified per-protocol and modified intention-to-treat analyses supporting the primary inference.
• Limitations:
  • Open-label management after day 7, creating potential for differential co-interventions (though mortality is objective).
  • Nonadherence was substantial, especially in the 7-day arm (23.1%), plausibly biasing intention-to-treat estimates towards non-inferiority.
  • Exclusion of S. aureus, fungal BSI, severe immunocompromise, and complex infection syndromes limits direct applicability to these high-risk groups.
  • Non-inferiority margin (4 percentage points) is a value-laden choice; interpretation depends on clinician and health-system tolerance for potential harm.
  • Secondary outcomes were not adjusted for multiplicity; subgroup analyses are hypothesis-generating.

Interpretation & Why It Matters

• Clinical practice

  For adults with bacterial BSI who do not meet exclusion criteria for prolonged therapy, a 7-day strategy achieved non-inferior 90-day mortality and materially reduced antibiotic exposure (antibiotic-free days by day 28: 19 vs 14).
• Stewardship impact

  The exposure separation is large and operationally meaningful; at scale, this supports “default” shorter durations when clinical stability and source control are achieved and no prolonged-treatment syndrome is suspected.
• Methodological contribution

  The concealed allocation until day 7 design targets the decision node clinicians actually face (stop vs continue), while maintaining intention-to-treat inclusion of early deaths and minimising post-randomisation selection artefacts.

Controversies & Subsequent Evidence

• Non-inferiority framing and margin: The 4 percentage-point margin represents a normative trade-off between reduced antibiotic exposure and tolerated potential harm; clinicians may differ on whether this threshold is acceptable for all BSI phenotypes.
• Adherence and interpretability: Higher nonadherence in the 7-day arm (23.1% vs 10.7%) can bias intention-to-treat estimates towards non-inferiority; concordant per-protocol and modified intention-to-treat results mitigate, but do not remove, this concern.
• Heterogeneity and subgroup signals: Prespecified subgroup estimates were generally imprecise; a Gram-negative subgroup estimate suggested potential benefit (risk difference −2.8; 95.7% CI −5.6 to −0.1), but multiplicity and residual confounding of subgroup strata limit causal certainty.
• Relationship to earlier RCTs: Prior non-inferiority RCTs in selected uncomplicated Gram-negative bacteraemia populations support shorter durations, but often excluded critically ill patients and Gram-positive pathogens; BALANCE extends the evidence base to broader hospital/ICU populations.¹

Summary

• International pragmatic non-inferiority RCT in 3608 adults with bacterial BSI (ICU and ward), excluding S. aureus, fungal BSI, severe immunocompromise, and syndromes requiring prolonged treatment.
• Primary outcome: 90-day mortality 14.5% (7-day) vs 16.1% (14-day); risk difference −1.6 percentage points (95.7% CI −4.0 to 0.8), meeting non-inferiority versus a 4 percentage-point margin.
• Strong exposure separation: antibiotic-free days to day 28 were 19 (IQR 11–21) vs 14 (IQR 11–14); median difference 5 (95% CI 4.6 to 5.4).
• Relapse and major safety outcomes were low and similar between groups (relapse 2.6% vs 2.2%; C. difficile 1.7% vs 2.0%).
• Internal validity is supported by allocation concealment until day 7 and objective mortality endpoint, but interpretation should account for open-label management after day 7 and higher nonadherence in the 7-day arm.

Further Reading

Other Trials

• 2019Yahav 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;69:1091-1098.
• 2020von 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;323:2160-2169.
• 2022Molina J, Montero-Mateos E, Praena 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;28:550-557.
• 2018Daneman N, Rishu AH, Pinto R, et al. 7 versus 14 days of antimicrobial therapy for bloodstream infections in critically ill patients: a pilot randomized controlled trial. Trials. 2018;19:111.
• 2003Chastre J, Wolff M, Fagon JY, et al. Comparison of 8 vs 15 days of antibiotic therapy for ventilator-associated pneumonia in adults: a randomized trial. JAMA. 2003;290:2588-2598.

Systematic Review & Meta Analysis

• 2011Havey TC, Fowler RA, Daneman N. Duration of antibiotic therapy for bacteremia: a systematic review and meta-analysis. Crit Care. 2011;15:R267.
• 2025Prager R, Shieh L, Dykeman J, et al. Efficacy and safety of 7 vs 14 days of antibiotic therapy for bloodstream infections: an individual participant data meta-analysis of randomized trials. JAMA Netw Open. 2025.
• 2022Turjeman A, et al. Individual participant data meta-analysis of short vs long antibiotic therapy for uncomplicated gram-negative bacteremia. eClinicalMedicine. 2022.
• 2021Procalcitonin-guided antibiotic duration in sepsis/critical illness: systematic reviews and meta-analyses (multiple).
• 2020Systematic reviews of short-course therapy for Gram-negative bacteraemia and other serious infections (multiple).

Observational Studies

• 2020Gerges FJ, Daneman N. Cost savings of shorter antibiotic strategies for bloodstream infection care (health-economic evaluation). J Assoc Med Microbiol Infect Dis Can. 2020;1:85-88.
• 2013Lipsky BA, et al. Observational cohorts evaluating shorter antibiotic durations for Gram-negative bacteraemia and urinary-source BSI (multiple).
• 2019Large stewardship cohorts examining duration, relapse, and adverse events in bacteraemia (multiple).
• 2021Real-world implementation studies of short-course BSI treatment pathways (multiple).
• 2022Studies linking total antibiotic exposure to subsequent resistant organism colonisation/infection in hospitalised adults (multiple).

Guidelines

• 2021Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock (adult).
• 2021Consensus guidance on management of uncomplicated Gram-negative bloodstream infection (expert guidance documents).
• 2022National/international antimicrobial stewardship guidance incorporating shorter-course principles (where available).
• 2023Guidance on management of bacteraemia and source control syndromes (IDSA/ESCMID and related societies; as applicable).
• 2024Institutional stewardship pathways for BSI duration incorporating emerging RCT evidence (where published).

Notes

• DOI-linked references are provided where the DOI was available in the source materials used to construct this summary; where not available, the reference is listed without an active DOI link.
• Subgroup effect estimates (risk differences with 95.7% CIs) were numerous and generally imprecise; they should be treated as hypothesis-generating rather than definitive effect modification.

Overall Takeaway

BALANCE is a landmark stewardship trial because it extends the “short-course” concept to a broad, pragmatic hospital and ICU BSI population using a methodologically careful non-inferiority framework. A 7-day strategy achieved non-inferior 90-day mortality versus 14 days while producing substantial antibiotic exposure reduction, with broadly similar relapse and safety outcomes in the populations studied. Implementation should remain anchored to careful syndrome exclusion, adequacy of early therapy, and source control.

Bibliography

• 1.Yahav 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;69:1091-1098.