C22-05 Variation in Timing of Third Vasopressor Initiation and Associated Outcomes in Septic Shock: Insights From MIMIC-IV

Rationale The decision to add a third vasoactive agent in septic shock is a highly variable and understudied practice. Differences in the timing of this escalation, reflecting uncertainty about when refractory shock warrants additional agents. Understanding real-world patterns in third-pressor initiation—and their association with patient characteristics and outcomes—may identify opportunities to standardize escalation thresholds and improve the quality of sepsis management. Methods Using MIMIC-IV v3.1, we identified ICU admissions with sepsis who received norepinephrine, vasopressin, and a third vasoactive agent. The NE dose (µg/kg/min) at the time of third-agent initiation was recorded and categorized as “early” (≤ median) or “late” (> median) escalation. Demographics, outcomes, and third-agent selection were compared between groups. Associations of log-transformed NE dose with mortality and ICU length of stay (LOS) were evaluated using logistic and linear regression. Results Among 2,587 ICU encounters for sepsis requiring three or more vasopressors, phenylephrine (n = 1,324, 51%) and epinephrine (n = 912, 35%) were the most common third agents, followed by dopamine (n = 125, 5%), dobutamine (n = 133, 5%), and milrinone (n = 92, 4%). The median norepinephrine dose at third-pressor initiation was 0.407 µg/kg/min. There were no significant differences in age, sex, race, or insurance type between early and late escalation groups. Mortality was significantly higher in the late group (80% vs. 53%; p < 0.0001), and among survivors median ICU length of stay was longer late third-pressor initiation compared with early initiation (13.3 vs. 11.8 days, p = 0.024). Multivariable logistic regression showed log(NE dose) independently predicted mortality (OR 34.9; 95% CI 18.4-65.9; p < 0.001), controlling for age and sex. Across third-agent types, mortality increased in parallel with the mean norepinephrine dose at which the agent was initiated: Milrinone (0.21 µg/kg/min, 56%), Dobutamine (0.31 µg/kg/min, 66%), Phenylephrine (0.38 µg/kg/min, 73%), Epinephrine (0.44 µg/kg/min, 75%), and Dopamine (0.43 µg/kg/min, 75%). Receiver operating characteristic (ROC) analysis identified approximate norepinephrine dose thresholds associated with increased mortality for each third-pressor agent (AUC 0.59-0.73) ranging from 0.20 µg/kg/min for dopamine initiation to 0.48 µg/kg/min for phenylephrine initiation. Conclusion In this large retrospective ICU cohort, escalation to a third vasoactive agent typically occurred at moderate-to-high NE doses and was associated with sharply increased mortality. This study characterizes practice variation in the timing of third-vasopressor initiation in septic shock and its relationship to outcomes and underscore the potential for protocolized approaches to improve sepsis care.