Results
Since 2005, a total of 951 patients received ECMO support in our institution. 572 patients under VA-ECMO with ABG available at cannulation were finally included in the present study (Figure A ). Ninety-nine patients were excluded because they had no pH at cannulation, 190 because they were veno-venous ECMO and 90 because they were duplicates in the database.
Patients characteristics are presented in Table1 . Most of the patients were male (68%) and mean age at cannulation was 54.8 (14.0) years. Mean SAPS II score was 46 (19.0).
Three groups were established considering the pH level at cannulation: Group 1 (pH<7) = 60 patients, group 2 (pH 7-7,2) = 115 patients and group 3 (pH>7,2) = 397 patients.
The three groups were significantly different for the age at cannulation (p<0.001). Patients from group 1 seemed to be younger patients (49.3 (14.0) years vs 52.5 (13.0) and 56.3 (14.0) years respectively for group 1, 2 and 3; p<0.001).
Main indication for group 1 was refractory cardiac arrest (40.7%) while predominant indications for groups 2 and 3 were post-cardiotomy (23.7 and 31.6% respectively) and refractory cardiogenic shock (22.8 and 42.2% respectively). In group 1 mean pH level was 6.86 (0.14) while mean pH level in groups 2 and 3 were 7.13 (0.06) and 7.38 (0.11) respectively (p<0.001).
Table 1 compares the main baseline characteristic between the three groups. We also compared ECMO’s complications between groups 2 and 3 (Table 4 in supplementary data).
At cannulation, doses of vasopressor amines were different (p<0.001). Patients in group 1 seemed to receive higher doses of vasopressor amines than the other 2 groups, respectively 78% vs 48 and 31%. Both groups differed in term of infectious complications (p<0.001). We described less thrombotic or hemorrhagic complications in group 1 had also: for thrombotic complications, 15 patients were affected in group 1 (25.9%), whereas 64 (55.7%) and 239 (58.1%) patients in groups 2 and 3. Considering hemorrhagic complications, it was observed in only 1 patient (4.5%) in group 1, 17 (23.9%) and 42 (18.8%) patients in groups 2 and 3.
The survival rate at 30 days for the overall population was 42.9%, 95%CI= [0.39-0.472]. Median of survival was respectively 1 day [1-2], 14 [6-27] and 33 days [18-NR] for group 1, 2 and 3. The very vast majority of deaths occurred within the 30 days after cannulation (Figure B ). Indeed, after the first month survival curves were roughly stable until 12 months. The survival rate 30 days post cannulation was significantly different (p<0.001) between the three groups; in the group 1: 15.0%, 95%CI = [0.082-0.274], in group 2 37.8%, 95%CI= [0.298-0.480] and in group 3 48.8%, 95%CI= [0.4407-0.5410]. The Kaplan Meyer estimated survival rate at 1 year were 13.3 % (95%CI = [7.0 – 25.4]) for group 1, 36.0 % (95%CI = [28.1 – 46.1]) for group 2, 43.2 % (95%CI = [38.5 – 48.6]) for group 3. These rates were significantly different (p<0.001) and are summarized in the Table 2 .
Lactate level was correlated with the blood pH rate (Figure C ). Low blood pH was associated with high rates of lactates as illustrated with -0.66 Spearman’s Rho coefficient which is in favor of a strong correlation between these 2 variables.
Figure D corresponded to the realization of subgroups according to the blood lactates level (<3; 3-7 and >7), and the curves joined those of pH with a probability of survival that decreased strongly in the first 30 days and then stayed stable. Association of low pH level (<7) and high level of lactate (>7) predicted a low chance of survival at one year (5.9% CI95 = [5.5-6.4]).
In multivariable analysis (Table 3 ), there were 5 major criteria that emerged as significant at cannulation: pH level (group 2: HR=1.42, 95% CI = [1.02 – 1.97]; group3: HR=5.07, 95% CI = [3.02 – 8.04]), age (for an increase of 10 years HR = 1.36, 95% CI = [1.22 – 1.50]), VA-ECMO implantation under external cardio pulmonary resuscitation (HR = 1.63, 95% CI = [1.13 – 2.35]), diastolic blood pressure at implantation (40mmHg or more, HR = 0.56, 95% CI = [ 0.39 – 0.81]) and kidney failure (HR = 1.56, 95% CI = [1.19 – 2.06]).
The decision curve analysis is presented in supplementary material.