Timing of Vasopressin Initiation and In-Hospital Mortality in Norepinephrine-Defined Shock: A Target Trial Emulation with External Validation in MIMIC-IV and eICU-CRD

Background: The optimal timing of vasopressin initiation among shock patients receiving norepinephrine remains uncertain in routine practice. Methods: We emulated a target trial using two large critical care databases (MIMIC-IV and eICU-CRD). Shock onset (time zero) was defined as the initiation of the first norepinephrine infusion. Among patients who initiated vasopressin within 24 hours of shock onset, we compared early initiation (0–6 hours) versus delayed initiation (6–24 hours). To mitigate immortal time bias, analyses were anchored at a landmark time of 6 hours after shock onset. Confounding was addressed using propensity score overlap weighting with prespecified early physiologic covariates measured within the first 6 hours. The primary outcome was all-cause in-hospital mortality. Sensitivity analyses varied landmark times (3, 6, and 12 hours) and applied weight trimming. External validation was performed across databases. Results: In MIMIC-IV, early vasopressin initiation was associated with lower in-hospital mortality compared with delayed initiation (odds ratio [OR] 0.69, 95% CI 0.61–0.79; risk difference [RD] −9.1%, 95% CI −12.4% to −5.8%). These findings were externally replicated in eICU-CRD (OR 0.81, 95% CI 0.69–0.93; RD −5.4%, 95% CI −9.0% to −2.0%), and a descriptive fixed-effect synthesis yielded a pooled OR of 0.74 (95% CI 0.67–0.81). Estimates were directionally consistent across landmark times of 3, 6, and 12 hours in both cohorts. Conclusions: Among vasopressin-treated patients with norepinephrine-defined shock initiating vasopressin within 24 hours, earlier initiation (0–6 hours) was consistently associated with lower in-hospital mortality compared with delayed initiation (6–24 hours) across two independent databases. Residual confounding cannot be excluded.