Fluid-intolerance signals as bedside safety limits during early septic shock resuscitation: a multicenter randomized trial

Fluid resuscitation in early septic shock has a narrow therapeutic index. Signals of fluid intolerance are frequently present during early resuscitation, yet current resuscitation strategies do not incorporate them to guide when to limit fluid administration. We evaluated whether integrating fluid-intolerance signals as safety limits could reduce fluid-induced harm without compromising hypoperfusion resolution.

In this multicenter, open-label randomized trial in three Chilean ICUs, mechanically ventilated adults with septic shock, persistent hypoperfusion, fluid responsiveness, and at least one high-risk criterion for fluid-induced harm were randomized 1:1 to a fluid-intolerance-limited resuscitation strategy or standard care during a 6-hour protocol window. In the intervention arm, before each bolus, fluid-intolerance signals (lung ultrasound score, VExUS, E/lateral e', and central venous pressure) were graded as traffic-light: green received a 500-mL bolus, yellow 250 mL, and red no further fluid. Standard care delivered 500-mL boluses while hypoperfusion and fluid responsiveness persisted. The primary outcome was the change in PaO₂:FiO₂ from baseline to 6 h (ΔPaO₂:FiO₂).

Recruitment was discontinued before the planned sample size was reached because continued enrollment was no longer feasible. Among 113 screened patients, 36 of a planned 62 were randomized, and 35 were analyzed (18 standard care, 17 intervention). The intervention group received a lower protocol bolus volume during the 6-hour intervention window than the standard-care group (250 [125-500] vs. 650 [500-1500] mL; between-group difference, - 500 mL [95% CI, - 750 to - 250]). Cumulative fluid exposure from pre-randomization through the end of the protocol window was similar between groups (3375 [2375-3875] vs. 3750 [1875-4851] mL; between-group difference, - 150 mL [95% CI, - 1470 to 1250]). The primary outcome favored the intervention: median ΔPaO₂:FiO₂ was 40.5 [24.0-58.5] versus - 37.9 [-69.2 to -12.5]; p < 0.001 (between-group median difference + 78.3 points; 95% CI 50.5 to 110.0), and 0-6-hour changes in lactate and capillary refill time did not differ.

Because the trial was stopped early, the effect of a fluid-intolerance-limited resuscitation strategy on oxygenation remains uncertain. A larger trial would be needed to determine whether this strategy improves oxygenation. Such a trial appears feasible.

ClinicalTrials.gov Identifier NCT06568744, registered August 21, 2024.