Park et al., 2014 | Retrospective review /32 patients | Not a true representative sample (high) | Record review (high) | Not applicable. | Not applicable. | None. |
Single-Center Study Small sample size A retrospective review of records |
Huang et al., 2013 | Retrospective cohort study / 52 patients | Not a true representative sample (high) | Record review (high) | Yes. | Yes. | None. |
Single-Center Study Low generalizability |
Cheng et al., 2013 | Propensity-matched analysis of ECMO registry /108 matched | Not a true representative sample (high) | Record review (high) | Yes. | Yes. | None. |
Existence of occult confounders Suboptimal selection because of inappropriate comparability |
Brechot et al., 2013 | Retrospective cross-sectional survey/14 patients | Not a true representative sample (high) | Record review (high) | Not applicable. | Not applicable. | Two of the authors took payments from companies |
Single-Center Study Small sample size No controlled group was matched |
Banjas et al., 2018 | Retrospective Cohort/131 patients | Not a true representative sample (high) | Record review (high) | Yes. | Yes. | None. |
Single-center study Overall, 29% missing data Low generalizability |
*Yeo et al., 2017 | Brief communication | Case series; not representative sample (High) | Record review (high) | Yes. | Yes. | None. |
Single Center case series Small sample size No controlled group A retrospective review of records |
Lee et al., 2017 | Retrospective cohort/24 patients | Not a true representative sample (high) | Record review (high) | Not applicable | Not applicable | None. |
Small sample size • No controlled group was matched |
Ro et al., 2018 | Retrospective review/71 patients | Not a true representative sample (high) | Record review (high) | Not applicable | Not applicable | None. |
Existence of occult confounders A retrospective review of records |
Han et al., 2019 | Retrospective cohort/23 patients | Not a true representative sample (high) | Record review (high) | Not applicable | Not applicable | None. |
Existence of occult confounders Single-Center Study |
Vogel et al., 2018 | Retrospective case series/12 patients | Not a true representative sample (high) | Record review (high) | Yes. | Yes. | None. |
Small sample size • No controlled group was matched |
Falk et al., 2019 | Retrospective cohort/37 patients | Not a true representative sample (high) | Record review (high) | Yes. | Yes. | None. |
Small sample size Single-Center study |
01. | Park et al., 2014. | 32 patients (21 males) with refractory septic shock were retrospectively reviewed. Baseline: Shock-ECMO interval 30.5 hours, CPR duration (median) 23 minutes. | CPR strongly predicts in-hospital mortality after ECMO. |
Median 11.1 days (IQR 4.0–26.0). Survivors had longer stay (32.5 days) than non-survivors (7.6 days) with P=0.02. |
Not reported | Survivors had lower peak lactate levels (4.5 mmol/l) and higher peak troponin I values (32.8 ng/ml) than non-survivors. | Not reported | 7 survived, 19 died from shock/multiorgan failure. |
Successful weaning: 40.6% in all patients. Stroke: 3.1% in all patients, none in survivors. |
02. | Huang et al. | Retrospective study, 52 patients; inclusion criteria: age >18 years, V-A ECMO, positive culture/serology, exclusion criteria: ECMO primarily for respiratory support. | Not reported | Median 90.1h (IQR 28.3 – 314.7) | Median: 114.1 hours (IQ 52.3 – 404.7) | Not reported | Not reported | Survival: 15% survived, 64% died. Age <60 better survival (P=0.029). |
Duration of ECMO (n=52) 15.0 hours IQ (6.1– 29.3). Bleeding at the cannulation site & GI bleeding= 8 |
03. | Cheng et al. 2013. | Propensity-matched ECMO study: 108 septic vs. 108 non-septic patients, age 16+, VA & VV-ECMO, non-first time ECMO excluded. | Septic ECMO patients had higher mortality, especially in patients over 55 years old. | Not reported | Not reported | Not reported | IABP during ECMO: septic (n=108) 19.0%, non-septic (n=108) 25.3% (P=0.285) |
Survival to discharge: 28.7% in septic vs 37.0% without sepsis Survived beyond ECMO: 44.4% in septic vs 56.5% in non-septic patients. |
CPR during ECMO is more common in non-septic group (n=108). Post ECMO neurologic deficit higher in non-survivors (n=71). |
04. | Brechot et al. 2013 | Survey of 14 patients with refractory cardiovascular failure associated with sepsis and other criteria for VA-ECMO use, excluding certain conditions. | ICU mortality in 14 patients was 29%, with 4 deaths during ICU stay (2 during ECMO). | Shorter in non-survivors (median 10 days) compared to survivors (median 17.5 days). | Not reported | ICU patients had a peak troponin value post-ECMO of 5.8 ng/ml. | Not reported | 10 (71%) out of 14 patients. | ECMO duration: median (range) (n=10) 5.5 (2–12) days in survivors vs (n=4) 3 (1–7) days in non-survivors. |
05. | Banjas et al., 2018 | Inclusion: ECMO-treated patients. | 56% 6-year hospital mortality. | Mean: 20 (8–31) | Mean hospital stay: 27 (15–40) |
Baseline lactate levels: 13 (8–26). PaO2/FiO2 ratios: 145 (99–233) |
Not reported | Not reported | Not reported |
06. | Yeo et al., 2017 | Patients with septic shock and ARDS were included. (n=8) |
Overall survival rate: 50% Successful weaning rate: 62.5% |
Not reported | Not reported |
Baseline: -Median MAP: 40 mmHg (IQR 33–46) -Median arterial lactate: 7.8 mmol/L (IQR 6.3–16.3) |
Not reported | Not reported | Not reported |
07. | Lee et al. 2018 |
Patients: 24 patients (M:F ratio: 6:2). Inclusion: Patients 18 years or older who received ECMO for sepsis. |
6 patients died, only 2 were discharged from the hospital. | Median 4 days (1–13) | Not reported | Immediately before the start of ECMO, the median serum lactate level, CRP, and total bilirubin were higher in the survivor group. | Not reported | Survival to discharge: 25% (2 out of 8 patients). 3 patients weaned successfully, but 1 died before discharge |
Survival group: Shock to ECMO duration= 25 hours (7–43) Non-survival group: Shock to ECMO duration= 6 hours (1–75) |
08. | Ro et al. 2018 | Adults (>20 y/o) with refractory shocks who received venoarterial ECMO support. | In-hospital mortality: 93%. 90-day mortality rates: 87.3%. | Not reported | Not reported | Non-survivors had higher arterial lactate, lower platelet count, and higher total bilirubin. | Not reported | Septic shock: 5 patients (7%) survived to discharge | 11 patients (15.5%) successfully weaned off ECMO in median 7.9 days. |
09. | Han et al. 2016 |
Inclusion: Patients with persistent circulatory failure or worsened refractory septic shock. Exclude: Patients with advanced malignant tumors, or irreversible neuropathy. |
15/23 patients died; 3 died after weaning. | ICU stay was shorter for the survival group (median of 12 days) compared to the death group (median of 16.5 days). |
Survival group: Hospital stay = 19 days (range, 17.5–21). Death group: Hospital stay = 16.5 days (range, 13.0– 21.0) |
Mean lactate levels were lower in the survival group (4.4 mmol/L) than in the death group (6.8 mmol/L). | Not reported | 5 discharged alive, 15 unsuccessful weaning, 3 deaths after weaning. | Not reported |
10. | Vogel et al. 2018 | Retrospective analysis of ECMO database to identify suitable patients, followed by clinical data extraction from electronic medical records. | 3 patients (25%) died, 2 from multiorgan failure and 1 from cerebral edema with brain herniation. | Not reported | Not reported |
Baseline lactate: mean 5.0 (range 3.85–6.05). PaO2: Mean 9.1 (range 6.4–9.8). pH: mean 7.10 (range 7.08–7.22) |
5 patients (41.7%) received vasopressin and 2 patients (16.7%) received adrenaline, dobutamine, or milrinone. | 9 (75%) survived after VAV ECMO decannulation. | No deaths reported during follow-up (median 6 months). |
11. | Falk et al. 2019 | Inclusion: ECMO support received by adult patients with septic shock admitted between January 2012 and December 2017 (n=37). | 8/37 patients died. | Not reported | Not reported | Venoarterial patients had a higher Pao2-to-Fio2 ratio, higher lactate levels, and higher ECMO flow than venovenous patients. | 71.4% (5/7) of patients survived who experienced CPR before admission. | ECMO survival: 81.1%. Hospital survival: 78.4%. Long-term follow-up survival: 59.5% (median 46.1 months). | Ten (37%) started venovenous ECMO and 27 on venoarterial ECMO. Venovenous-ECMO was associated with higher risk for in-hospital death (50% vs 11%; p=0.011). |