Comparative associations of vasopressor and antibiotic timing with mortality in emergency department patients with septic shock: a 15-year single-centre retrospective cohort study

Background The Surviving Sepsis Campaign recommends antibiotic administration within 1 h for septic shock, but the evidence is low quality, and the relative associations of antibiotic versus haemodynamic resuscitation timing with mortality remain unclear. This study aims to examine the independent and joint associations of time-to-antibiotics (TTA) and time-to-vasopressor (TTV) with mortality and their temporal changes over 15 years. Methods This single-centre retrospective cohort included adults with septic shock (vasopressor started < 72 h after arrival) presenting to a Korean tertiary hospital emergency department from 2011 to 2025. TTA and TTV were modelled as continuous and categorical exposures. Multivariable logistic and Cox regression were adjusted for pre-treatment covariates only. Robustness was assessed using restricted cubic splines, propensity-score matching, inverse-probability-of-treatment weighting, multiple imputation, competing-risk and E-value analyses. Results Among 2,086 patients (median age 74 years; 54.4% male), median TTA and TTV were 1.58 and 2.33 h, respectively; in-hospital mortality was 29.6%. In the primary model, limited to pre-treatment covariates, both timings were independently associated with in-hospital mortality (TTA aOR per 1-h delay 1.042, 95% CI 1.015–1.068, p = 0.002; TTV aOR 1.024, 95% CI 1.015–1.034, p < 0.001); a formal Wald contrast found no statistically significant difference between the two per-hour coefficients ( p = 0.25). In inverse-probability-of-treatment-weighted analyses, attaining both targets (TTA < 3 h and TTV < 3 h) was associated with lower weighted in-hospital mortality (24.1% vs. 34.1%; weighted risk difference 10.0% points, 95% CI 6.2–14.2; aOR 0.61, 95% CI 0.50–0.76; p < 0.001). Exploratory continuous calendar-year interaction analyses provided no evidence that either association changed over the 15 years (year-by-TTA p = 0.39; year-by-TTV p = 0.55). Conclusions In this single-centre emergency department cohort with septic shock, longer TTA and TTV were independently associated with higher mortality after mutual adjustment; the two per-hour associations were not statistically distinguishable, and no superiority of either timing is claimed. Attaining both timing targets was associated with lower weighted in-hospital mortality than not attaining both. Exploratory analyses provided no evidence that either association changed over the 15-year study period. These observational findings support further prospective evaluation of integrated antimicrobial and haemodynamic timing strategies.