Shortening antibiotic therapy duration for hospital-acquired bloodstream infections in critically ill patients: a causal inference model from the international EUROBACT-2 database.

Summary

In a causal-inference analysis of a prospective international ICU cohort, 7–10 day antibiotic courses for selected HA-BSI were associated with lower 28-day treatment failure (OR 0.64) driven by fewer subsequent infectious complications (OR 0.58), with similar mortality compared to 14–21 day courses. Results support antibiotic duration reduction when no indications for prolonged therapy exist.

Key Findings

• Among 550 eligible ICU patients, short-course therapy (7–10 days; n=213) reduced 28-day treatment failure vs long (14–21 days; n=337) (OR 0.64, 95% CI 0.44–0.93).
• Benefit was driven by fewer subsequent infectious complications (OR 0.58, 95% CI 0.37–0.91); mortality was similar (OR 0.92, p=0.70).
• Longer therapy correlated with S. aureus and difficult-to-treat pathogens and greater use of combination therapy.

Clinical Implications

For ICU patients with HA-BSI lacking indications for prolonged therapy (e.g., selected sources/microorganisms, no deterioration), clinicians may target 7–10 days of antibiotics, with careful eligibility screening and monitoring for complications.

Limitations

• Observational design with residual confounding and potential selection bias despite IPTW.
• Heterogeneity in sources/pathogens and local practices; unmeasured factors (e.g., source control quality) may influence outcomes.

Future Directions

Randomized trials to confirm optimal duration across sources/pathogens, and implementation studies embedding duration algorithms into stewardship programs.