<p>Calculation of O2 content, CO2 content, O2 transfer and CO2 transfer</p><p>• O2 content (mL O2/dL dl blood) = 1.34 * (Hb) * SaO2 + 0.0031 * PO2. The factor</p><p>1.34 is specified in ml*g-1 and the factor 0.0031 is specified in ml*dLl-1*mmHg-1. </p><p>- • CO2 content (mL CO2/dL dl blood) = 22.4 * (HCO3 ) + 0.030 * PCO2. The factor</p><p>22.4 (molecular volume of a gas) is specified in Ll*mol-1, and the factor 0.030 is</p><p> specified in mM*mmHg-1. </p><p>• O2 transfer (ml/min) = ((O2 content)out – (O2 content)in) * blood flow * 10;</p><p>• CO2 transfer (ml/min) = ((CO2 content)in – (CO2 content)out) * blood flow * 10.</p><p>• The respective values were collected at the inlet (in) and outlet (out) of the</p><p>MO. </p><p>3-dimensional curve fitting model</p><p>O2 transfer and CO2 elimination in relation to blood flow and sweep gas flow were put into a 3-dimensional (3d) curve-fitting model and depicted accordingly. TableCurve 3D (3D v4.0) automated surface fitting analysis software (SYSTAT Software Inc, Systat Software</p><p>Inc, San Jose, CA) was used to find equations to describe the three dimensional empirical data. Equation 531 belongs to the x-y polynomial equations and is a Cosine</p><p>Series Bivariante Polynominal where Z = a+ b*cos(x) + c*cos(y) + d*cos(2x) + e*cos(x)*col(y) + f*col(2y) with x = BF, y = GF, Z = CO2 transfer adjusted to gas exchange surface.</p><p>Statistical analysis</p><p>Intergroup differences (comparing 4 different ECMO-systems) were analysed by linear mixed models, while the correlation structure between repeated measurements was specified as unstructured. All models are controlled for type of system, type of system*time, blood flow and gas flow. These analyses were done with SAS 9.3 procedure PROC MIXED (SAS Institute, Cary, NC, USA). Post-hoc pairwise comparisons were adjusted by the Tukey-Kramer method. Statistical significance was assumed if p<0.05.</p><p>ESM tables</p><p>Table S1 Overview of cannulation modes</p><p>ECMO-system All PLS CH HL ECC</p><p>Patients (n) 317 163 59 56 39</p><p>Direction of blood flow (n, %) outflow > inflow</p><p>Femoral vein 213 (67) 106 (65) 48 (81) 40 (71) 19 (49) > jugular vein</p><p>Femoral vein 38 (12) 28 (17) 4 (7) 3 (5) 3 (8) > subclavian vein</p><p>Femoral vein 7 (2) 5 (3) 1 (2) 0 (0) 1 (3) > Femoral vein </p><p>Jugular vein, Bicaval Dual Lumen 49 (15) 22 (13) 6 (10) 11 (20) 10 (26) catheter</p><p>Femoral vein, 10 (3) 2 (1) 0 (0) 2 (4) 6 (15) Dual Lumen catheter</p><p>PLS = PLS -system, CH = Cardiohelp -system, HL = Hilite 7000 LT –system and </p><p>ECC = ECC.05 –system. Table S2: Overview of cannula sizes</p><p>ECMO-system All PLS CH HL ECC</p><p>Diameter of cannula (Fr) outflow > inflow</p><p>Femoral vein 19-25 19-25 21-23 21-23 21-23 > jugular vein >15-23 > 15-23 > 15-21 > 15-19 > 15-21</p><p>Femoral vein 21-23 21-23 21 21-22 21-23 > subclavian vein >15-21 > 15-17 > 17-21 > 17 > 17</p><p>Femoral vein 21-23 21-23 23 23 -- > Femoral vein >17-23 > 17-23 > 21 > 21</p><p>Jugular vein, 23-31 23-31 27 23-27 23-31 Dual Lumen catheter</p><p>Femoral vein, 24 24 -- 24 24 Dual Lumen catheter</p><p>PLS = PLS -system, CH = Cardiohelp -system, HL = Hilite 7000 LT –system and </p><p>ECC = ECC.05 –system. Table S3: CO2 transfer and O2 transfer during study period</p><p>Day CO2 transfer CO2 transfer norm O2 transfer O2 transfer norm</p><p>(p=0.014) (p=0.002) (p=0.885) (p=0.881)</p><p>1 180.6 (174.8, 186.3) 111.1 (107.7, 114.5) 140.5 (136.4, 144.6) 86.4 (84.0, 88.8)</p><p>2 173.6 (167.7, 179.4) 106.3 (102.9, 109.7) 141.9 (137.7, 146.0) 86.9 (84.4, 89.3)</p><p>3 170.0 (164.0, 176.1) 103.5 (100.0, 107.1) 142.4 (138.1, 146.7) 87.5 (85.0, 90.0)</p><p>4 169.2 (162.7, 175.6) 103.0 (99.2, 106.8) 142.2 (137.7, 146.8) 87.2 (84.5, 89.8)</p><p>5 170.8 (163.9, 177.7) 104.7 (100.6, 108.7) 140.9 (136.1, 145.7) 86.2 (83.5, 89.0)</p><p>Data show least squares means (95%-CI) estimated by linear mixed models. All models are controlled for type of system, type of system*time, blood flow and gas flow. CO2/O2 transfer norm: CO2/O2 transfer adjusted for exchange surface of the </p><p>MO. Table S4: CO2 transfer and O2 transfer of 4 different ECMO systems</p><p>System PLS CH HL ECC</p><p>200.6 (195.6, 167.1 (158.5, 169.7 (161.3, 154.0 (143.6, CO2 TF 205.7) a 175.6) 178.0) 164.3)</p><p>CO2 TF 111.4 (108.5, 128.7 (122.8, 92.9 (88.0, 97.8) 89.9 (85.1, 94.6) norm 114.3) c 134.6) b</p><p>149.4 (145.4, 142.6 (135.9, 141.3 (134.7, 133.0 (124.9, O2 TF 153.4) 149.3) 147.9) 141.2) d</p><p>O2 TF 110.4 (105.7, 83.0 (80.7, 85.3) f 79.2 (75.3, 83.1) 74.7 (70.8, 78.5) norm 115.2) e</p><p>CO2/O2 TF = CO2/O2 transfer. CO2/O2 TF norm: CO2/O2 transfer adjusted for exchange surface of the MO. PLS = PLS -system, CH = Cardiohelp -system, HL = </p><p>Hilite 7000 LT –system and ECC = ECC.05 –system. Data show least squares means (95%-CI) estimated by linear mixed models. All models are controlled for time, type of system*time, blood flow and gas flow. Pairwise comparisons showed: a = </p><p>PLS > ECC/CH/HL (p≤0.001 each); b = ECC > PLS/CH/HL (p<0.001); c = PLS > </p><p>CH/HL (P<0.001 each); d = ECC < PLS (p=0.002); e = ECC > PLS/CH/HL (p<0.001 each); f = PLS > HL (p=0.002); no significant differences in the remaining comparisons. Table S5: Parameters at the inlet of the oxygenator</p><p>ECMO-system All PLS CH HL ECC</p><p> b,c pvO2 (mmHg), 38 (34/41) 37 (34/41) 39 (35/42) 39 (35/42) 38 (35/42) day 1-5</p><p> b,d pvCO2 45 (40/50) 45 (40/50) 47 (42/52) 45 (40/48) 45 (41/50)</p><p>(mmHg), day1-5 vpH, 7.40 7.39 7.40 7.39 7.41 day 1-5 (7.35/7.44) (7.35/7.43) (7.37/7.45) (7.35/7.43) (7.37/7.45)</p><p> e,b</p><p>Hb (g/dl), 9.3 9.2 9.5 9.4 9.0 day 1-5 (8.6/10.2) (8.6/10.2) (8.7/10.8) a,b (8.6/10.1) (8.4/9.8)</p><p>Parameters displayed as median and interquartile range. PLS = PLS -system, CH = </p><p>Cardiohelp -system, HL = Hilite 7000 LT –system and ECC = ECC.05 –system. pvO2 </p><p>= venous oxygen tension before ECMO; pvCO2 = venous carbon dioxide tension before ECMO; vpH = venous pH before ECMO; Hb = hemoglobin. p ≤ 0.05: a = CH vs ECC; b = CH vs PLS; c = HL vs PLS; d = CH vs HL; e = ECC vs PLS. Legends for ESM figures</p><p>Figure S1</p><p>CO2 transfer adjusted to the MO exchange surface of different membrane oxygenators</p><p>(PLS-system (n = 163; 681 measurements), Cardiohelp (CH) –system (n = 58; 234 measurements), Hilite 7000 LT (HL) -system (n = 56; 255 measurements) and ECC.05- system (n = 40; 161 measurements), 1st to 5th day after ECMO implantation) as fitted</p><p>3d-curve (equation 531, Cosine Series Bivariate Order 2, see ESM). Adjusted to the exchange surface of the oxygenator the ECC.05-system was the most effective in terms of CO2 transfer elimination. </p><p>Figure S2</p><p>Platelet count for each device and the study period in days. PLS = PLS -system, CH </p><p>= Cardiohelp -system, HL = Hilite 7000 LT –system and ECC = ECC.05 –system. p < 0.05 (*, CH system; §, ECC system; #, HL system; &, PLS system) compared to values before ECMO (time 0).</p>