Effluent-Dose Response of Continuous Haemofiltration Integrated into Veno-Venous ECMO for Septic Shock: A Retrospective Cohort Study

Background and Objectives: The optimal effluent dose of continuous haemofiltration (CHF) when coupled to veno-venous extracorporeal membrane oxygenation (ECMO) for septic shock is unknown. We examined our 44-patient ECMO registry, contrasting a smaller high-dose subgroup (HDHF ≥ 45 mL kg−1 h−1; n = 13) with a larger standard-dose subgroup (SDHF 25–35 mL kg−1 h−1; n = 31). The primary endpoint was 72 h change in SOFA score (ΔSOFA). Materials and Methods: All adults cannulated for ECMO (January 2018–January 2025) and started on CHF within 2 h were eligible. Variables were abstracted at baseline, 24 h and 72 h. Continuous data were analysed by Student’s t or Mann–Whitney tests, categorical data by χ2/Fisher; and paired changes by Wilcoxon. Two-sided p < 0.05 signified significance. Results: Baseline characteristics were comparable (age 49.1 ± 15.2 vs. 50.4 ± 14.9 y; APACHE II 28.4 ± 5.3 vs. 27.5 ± 5.9). Median effluent reached 48.1 mL kg−1 h−1 (IQR 46.6–49.7) in HDHF and 29.7 mL kg−1 h−1 (27.5–31.9) in SDHF (p < 0.001). IL-6 fell by 1 061 ± 487 pg mL−1 with HDHF versus 637 ± 425 pg mL−1 with SDHF (p = 0.003). Mean arterial pressure rose 19.2 ± 8.1 vs. 12.7 ± 8.3 mmHg (p = 0.03), and norepinephrine declined 0.46 ± 0.22 vs. 0.30 ± 0.19 µg kg−1 min−1 (p = 0.04). ΔSOFA at 72 h was –4.4 ± 2.1 with HDHF and –2.6 ± 2.3 with SDHF (p = 0.01). Twenty-eight-day mortality was 38.5% (5/13) versus 45.2% (14/31), p = 0.64. Effluent dose correlated with ΔIL-6 (ρ = 0.53, p < 0.001) and ΔSOFA (ρ = 0.45, p = 0.003). Conclusions: In this ECMO cohort, high-dose haemofiltration, although applied in only 13 patients, appeared to achieve greater cytokine clearance, faster haemodynamic recovery and deeper early organ-failure improvement than standard dosing, without excess bleeding. Survival advantage was not demonstrable, underscoring the need for prospective randomised confirmation of the dose–response signal.