Effect of early adjunctive vasopressin initiation for septic shock patients: a target trial emulation

Abstract

In septic shock, the optimal timing of adjunctive vasopressin initiation shock is unknown. We aimed to assess the effect of its early initiation for patients with septic shock. We conducted a multicenter target trial emulation to estimate the intensive care unit (ICU) mortality effect of early (≤ 6 h) adjunctive vasopressin compared with usual care. Eligible patients had septic shock diagnosed within 6 h of ICU admission. The primary outcome of this study was 30-day ICU mortality. Subgroup analyses were conducted to test the interaction of early vasopressin start with peak norepinephrine-equivalent dose (NED) at 6 h, APACHE score, peak lactate at 6 h and invasive mechanical ventilation. Secondary outcomes were the impact of delayed vasopressin introduction on 30-day ICU mortality and effect of NED at vasopressin start on 30-day ICU mortality. We used the parametric g-formula to emulate a target trial. Overall, 3,105 patients fulfilled the inclusion criteria. Mean age was 62 years and mean APACHE III score was 83. In the first six hours of vasopressor therapy, 1,864 (60%) patients were invasively ventilated. Estimated 30-day ICU mortality was 19.34% (95%CI, 17.0 to 21.68) in the no vasopressin group and 18.45% (95%CI, 16.26 to 20.63) in the early vasopressin group; relative risk 0.95 (95%CI, 0.93 to 0.98). The estimated 30-day ICU mortality effect of starting vasopressin was particularly strong at lower norepinephrine doses (< 0.25 µg.kg-1.min-1) and significant at lower norepinephrine doses than recommended by the Surviving Sepsis Campaign Guidelines. Vasopressin administration progressively increased over the study period, from 35.2% (95%CI, 30.0 to 40.5) in 2015 to 45.1% (95%CI, 40.7 to 49.6) in 2021 (ß = + 1.3% per year; 95%CI, + 0.46 to + 2.16, p = 0.011). Patients had progressively lower norepinephrine equivalent dose (ß = - 0.05 µg.kg-1.min-1 per year; 95%CI, - 0.09 to - 0.002, p = 0.038) and lower total SOFA score (ß = - 0.1 point per year; 95%CI, - 0.18 to - 0.07, p < 0.001) at vasopressin start. In this emulation of a hypothetical target trial, patients with septic shock benefited from early vasopressin administration. These findings can help design prospective randomised-control trials of early adjunctive vasopressin use in septic shock.