A time-sensitive analysis of the prognostic utility of vasopressor dose in septic shock

It is unclear whether the association between vasopressor dose and mortality is affected by duration of administration. We examined whether prognostication in septic shock is feasible through the use of daily median vasopressor doses. We undertook a single-centre retrospective cohort study. We included patients with a diagnosis of septic shock admitted to the intensive care unit at Queen Elizabeth Hospital, Birmingham, UK, between April 2016 and July 2019. The primary outcome measure was 90-day mortality. We defined vasopressor dose as the median norepinephrine equivalent dose (equivalent infusion rates of all vasopressors and inotropes) recorded for each day, for the first four days of septic shock. We divided patients into groups by vasopressor dose quintiles and calculated their 90-day mortality rate. We examined area under the receiver operator characteristic curves for prognostic ability. In total, 844 patients were admitted with septic shock and had a 90-day mortality of 43% (n = 358). Over the first four days, median vasopressor dose decreased in 93% of survivors and increased in 56% of non-survivors. The mortality rate associated with a given vasopressor dose quintile increased on sequential days of septic shock. The area under the receiver operator characteristic curves of daily median vasopressor dose against mortality increased from day 1 to day 4 (0.67 vs. 0.86, p < 0.0001). By day 4, a median daily vasopressor dose > 0.05 μg.kg^-1 .min^-1 had an 80% sensitivity and specificity for mortality. The prognostic utility of vasopressor dose improved considerably with shock duration. Prolonged administration of small vasopressor doses was associated with a high attributable mortality.