Dynamic changes in vasoactive-inotropic score and their prognostic value in severe sepsis and septic shock patients undergoing polymyxin B hemoperfusion: a real-world Asian ICU cohort study

Severe sepsis and septic shock frequently require escalating vasoactive support, but the prognostic utility of vasoactive-inotropic score (VIS) trajectories during polymyxin B hemoperfusion (PMX-HP) in Asian ICUs remains unclear. We conducted a retrospective single-center cohort study in a tertiary medical ICU in Taiwan, enrolling consecutive adults with severe sepsis or septic shock who received PMX-HP between July 2013 and December 2019. VIS was calculated before PMX-HP (VIS1), immediately after PMX-HP (VIS2), and approximately 24 h post-treatment (VIS3); the primary outcome was 28-day all-cause mortality. In 64 patients, median VIS decreased from 28.8 (IQR 13.0-47.6) at VIS1 to 18.1 (IQR 7.8–45.0) at VIS2 and was 16.9 (IQR 1.3–45.1) at VIS3 among patients with available VIS3 data. VIS1, VIS2, and VIS3 were higher in non-survivors than survivors. A mixed-effects model using log(VIS + 1) confirmed significant effects of time, survival status, and their interaction. In an exploratory time-dependent Cox model, higher time-updated log(VIS + 1) was associated with 28-day mortality. The immediate VIS change showed limited discrimination for 28-day mortality (AUC 0.57, 95% CI 0.43–0.72), whereas VIS3 demonstrated better discrimination (AUC 0.78, 95% CI 0.66–0.91). Adding VIS3 to APACHE II improved discrimination compared with APACHE II alone (AUC 0.81 vs. 0.63; Delta AUC + 0.18; p = 0.020). Kaplan-Meier analysis using the ROC-derived VIS3 cut-off of 14.56 showed significantly worse 28-day survival when VIS3 remained above this threshold. Late post-treatment VIS monitoring may support bedside risk stratification in severe sepsis and septic shock undergoing PMX-HP; multicenter validation is warranted. Clinical trial number Not applicable.