Abstract Background Venovenous extracorporeal membrane oxygenation (ECMO) is essential for patients with severe respiratory failure who do not respond to conventional mechanical ventilation. Adequate ECMO flow and safe circuit pressure are critical; however, cannula selection, which has a great impact on these factors, is often based on empirical judgment. This study aimed to develop a simple predictive method based on fluid dynamics for estimating ECMO flow rate and circuit pressures (P1: pre-pump, P2: pre-oxygenator, and P3: post-oxygenator). This experimental predictive model study compared the calculated and measured ECMO parameters across 36 combinations of cannula sizes, pump speeds, and bed heights. A laboratory-based ECMO circuit model was assembled with various drainage and return cannulas, an oxygenator, tubing, and a centrifugal pump. The circuit was primed with a 33% glycerin solution and tested across the 36 combinations. A four-step prediction method was applied: (1) modeling the pressure–flow relationships of ECMO components and the pump using manufacturer data; (2) identifying the expected flow rate by locating the intersection of the total circuit resistance and pump output curves; (3) sequentially calculating pressure drops across the circuit; and (4) adjusting pressures based on bed height. Results The predicted flow rate and circuit pressure values were compared to experimental measurements across the 36 combinations. The calculated and measured values showed strong agreement (R 2 = 0.96–0.97), and predictions were significant. Notably, bed height alterations were confirmed to affect circuit pressure but not flow rate. Conclusions Our newly developed method reliably predicts the ECMO flow rate and circuit pressure. Hence, it can be considered a valuable tool for preemptively selecting the optimal cannula size for ECMO, thus improving patient safety and circuit management. Furthermore, it may be a valuable educational tool, making complex hemodynamic concepts more intuitive for trainees.
Takahashi et al. (Wed,) studied this question.