Association of antibiotic type and timing with sepsis mortality using target trial emulation

Antibiotic therapy is essential for sepsis management, but the optimal empirical strategy remains uncertain. This study evaluated the effects of first-line antibiotic preference and initiation timing on in-hospital mortality among intensive care units (ICU) patients with sepsis. Using the MIMIC-IV database, we emulated a sequential target trial comparing patients who received antibiotics within 48 h of sepsis diagnosis versus delayed initiation. Randomization was approximated through a clone-censor-weight process to address confounding by indication. The primary outcome was in-hospital mortality. Weighted Cox regression estimated hazard ratios (HRs), and sensitivity analyses tested robustness. Among 3,669 eligible patients, 3,568 (97%) received antibiotics within 48 h. After weighting, covariate balance was achieved. Beta-lactam use was associated with lower in-hospital mortality (HR 0.88, 95% CI 0.78-0.95), with consistent reductions at 7, 14, and 60 days. Timing within the 48-hour window did not modify outcomes for either beta-lactams or glycopeptides. Empirical beta-lactam therapy was linked to improved survival among ICU sepsis patients, whereas timing of initiation showed no significant impact. These findings support prioritizing beta-lactam-based regimens as first-line empirical coverage in early sepsis management.