Timing of Vasopressin Initiation in Sepsis Patients on Norepinephrine: A Target Trial Emulation

Background The optimal timing for initiating vasopressin in septic shock remains uncertain, with conflicting evidence potentially arising from variations in patient severity and methodological limitations such as immortal time bias. This study aimed to evaluate the time-varying effect of early vasopressin initiation across a spectrum of baseline norepinephrine dose requirements. Methods We included 5461 patients with septic shock from the MIMIC-IV 2.2 database. A target trial emulation framework was applied, using a clone-censor-weight approach with inverse probability of censoring weighting to address immortal time bias. The primary and secondary outcomes were 14-day and 28-day all-cause mortality. Time-varying treatment effects of early vasopressin initiation (defined by a 1-h primary grace period) were assessed via piecewise landmark Cox models, with analyses stratified by baseline norepinephrine equivalent dose thresholds (>0.1, >0.25, >0.5, and >0.75 μg/kg/min). Results The association between early vasopressin initiation and 28-day mortality varied by both baseline norepinephrine dose and time since initiation. Among patients with low-to-moderate shock severity (NE 0.1-0.5 μg/kg/min), a pattern of delayed benefit was observed. The absolute risk reduction in the broad >0.1 μg/kg/min cohort was not statistically significant, but a survival benefit emerged in the late phase for the >0.25 μg/kg/min subgroup (late-phase HR 0.82, 95% CI 0.69-0.97, P=.018). In contrast, for patients with more severe shock (NE >0.5 μg/kg/min), early initiation was associated with increased risk in the first 48 h. This early hazard was most pronounced in the >0.75 μg/kg/min subgroup, where the point estimate for short-term mortality was markedly elevated (HR 2.87, 95% CI 1.68-4.90, P<.001). Although long-term point estimates at higher thresholds favored vasopressin, the confidence intervals were extremely wide and crossed the null, precluding firm conclusions. Subgroup analyses suggested that older age and concomitant renal dysfunction might amplify susceptibility to higher early mortality. Conclusion In this target trial emulation, the benefit of early vasopressin initiation appeared dependent on baseline shock severity. The findings suggest a potential therapeutic window at norepinephrine equivalent doses between 0.1 and 0.5 μg/kg/min, with the most consistent benefit around 0.25 μg/kg/min. For patients requiring >0.5 μg/kg/min, early initiation was associated with higher early mortality, which may reflect residual confounding by shock severity rather than a direct causal harm, and the current analysis did not support a definitive compensatory long-term survival benefit.