Relationship between the Pre-ECMO and ECMO Time and Survival of Severe COVID-19 Patients: A Systematic Review and Meta-Analysis

Background: Coronavirus disease 2019 (COVID-19) is the etiology of acute respiratory distress syndrome (ARDS). Extracorporeal membrane oxygenation (ECMO) is used to support gas exchange in patients who have failed conventional mechanical ventilation. However, there is no clear consensus on the timing of ECMO use in severe COVID-19 patients. Objective: The aim of this study is to compare the differences in pre-ECMO time and ECMO duration between COVID-19 survivors and non-survivors and to explore the association between them. Methods: PubMed, the Cochrane Library, Embase, and other sources were searched until 21 October 2022. Studies reporting the relationship between ECMO-related time and COVID-19 survival were included. All available data were pooled using random-effects methods. Linear regression analysis was used to determine the correlation between pre-ECMO time and ECMO duration. The meta-analysis was registered with PROSPERO under registration number CRD42023403236. Results: Out of the initial 2473 citations, we analyzed 318 full-text articles, and 54 studies were included, involving 13,691 patients. There were significant differences between survivors and non-survivors in the time from COVID-19 diagnosis (standardized mean difference (SMD) = −0.41, 95% confidence interval (CI): [−0.53, −0.29], p < 0.00001), hospital (SMD = −0.53, 95% CI: [−0.97, −0.09], p = 0.02) and intensive care unit (ICU) admission (SMD = −0.28, 95% CI: [−0.49, −0.08], p = 0.007), intubation or mechanical ventilation to ECMO (SMD = −0.21, 95% CI: [−0.32, −0.09], p = 0.0003) and ECMO duration (SMD = −0.18, 95% CI: [−0.30, −0.06], p = 0.003). There was no statistical association between a longer time from symptom onset to ECMO (hazard ratio (HR) = 1.05, 95% CI: [0.99, 1.12], p = 0.11) or time from intubation or mechanical ventilation (MV) and the risk of mortality (highest vs. lowest time groups odds ratio (OR) = 1.18, 95% CI: [0.78, 1.78], p = 0.42; per one-day increase OR = 1.14, 95% CI: [0.86, 1.52], p = 0.36; HR = 0.99, 95% CI: [0.95, 1.02], p = 0.39). There was no linear relationship between pre-ECMO time and ECMO duration. Conclusion: There are differences in pre-ECMO time between COVID-19 survivors and non-survivors, and there is insufficient evidence to conclude that longer pre-ECMO time is responsible for reduced survival in COVID-19 patients. ECMO duration differed between survivors and non-survivors, and the timing of pre-ECMO does not have an impact on ECMO duration. Further studies are needed to explore the association between pre-ECMO and ECMO time in the survival of COVID-19 patients.

In China, the overall case-fatality rate for COVID-19 is 2.3% (1023 deaths out of 44,672 confirmed cases), with approximately 2087 patients in critical condition accounting for 5% of confirmed cases.Among critically ill COVID-19 patients, there were approximately 1023 deaths, resulting in a mortality rate of 49% [3].Based on the experience of the previous viral outbreak period [4,5], extracorporeal membrane oxygenation (ECMO) has been identified as an important form of life support for critically ill patients.It can be effective in treating severe respiratory failure due to acute respiratory distress syndrome (ARDS) and supporting gas exchange in patients who have failed conventional mechanical ventilation (MV).At the same time, Extracorporeal Life Support Organization (ELSO) guidelines suggest that ECMO can be used as a rescue treatment for critically ill COVID-19 patients who do not respond to conventional ARDS therapy [6].
Despite optimism regarding the potential role of ECMO in COVID-19 treatment, several studies still report a high mortality rate.A meta-analysis found a mortality rate of 37% in patients who received ECMO for COVID-19 in 2020 [7].Additionally, ELSO registry data reported an increase in mortality rates for the use of ECMO in COVID-19 patients, rising from 37% at the beginning of 2020 to 52% at the end of the year [8,9].Late initiation of ECMO may be an independent risk factor for increased mortality.Li et al. revealed that early initiation of ECMO was associated with decreased 60-day mortality after ECMO (50% vs. 88%, p = 0.044) [10].However, Mathilde et al. reported that late ECMO treatment in patients with refractory ARDS related to SARS-CoV-2 does not seem to be associated with an excess risk of mortality [11].The optimal timing for initiating ECMO in COVID-19 treatment is currently uncertain and controversial.There is insufficient global evidence to assess the effectiveness of ECMO timing, and no studies have shown whether ECMO duration affects mortality in severe COVID-19 patients.
To aid clinicians in accurately determining the timing of ECMO use and to improve the use and management of ECMO in severe COVID-19 patients, we conducted a systematic review and meta-analysis.This study focuses on the timing of pre-ECMO and ECMO duration in COVID-19 patients, to clarify the effects of pre-ECMO and ECMO duration on COVID-19 patient survival, and to guide current clinical practice and future research.Furthermore, we have further investigated the linear correlation between pre-ECMO time and ECMO duration.

Methods
The protocol was registered in PROSPERO (International Prospective Register of Systematic Reviews URL: https://www.york.ac.uk/crd/, accessed on 3 March 2023), with the registration number CRD42023403236.This meta-analysis was conducted according to the guidelines of the Preferred Reporting Item for Systematic Review and Meta-Analysis 2020 (PRISMA 2020) (Supplementary Table S1).

Literature Search
Three databases (PubMed, Embase, and the Cochrane Library databases) were used as our search libraries.Other sources, such as the Critical Care Medicine website (URL: http://www.ccmjournal.com,accessed on 21 October 2022), the Critical Care website (URL: http://ccforum.com,accessed on 21 October 2022), and the American Journal of Respiratory and Critical Care Medicine website (URL: http://ajrccm.atsjournals.org,accessed on 21 October 2022), were also searched.Without language restriction, the following search Medical Subject Headings (MeSHs) were used to retrieve advanced articles from inception to 21 October 2022 according to the PICOS (population, intervention/exposure, comparison, outcome, and study design) principle: (1) for patients: "COVID-19" OR "SARS-CoV-2 Infection"; (2) for intervention and comparison: "Extracorporeal Membrane Oxygenations"; (3) for outcome: "Outcome" OR "Survival".Supplementary Table S2 describes the detailed search strategy.

Study Selection
Reference management software, Endnote X9.3.3 software (Thomson Reuters, New York, NY, USA), was used to organize all studies.All titles and abstracts were reviewed after removing duplicates.Then, the full-text assessment was performed following an initial screening to consider eligibility for inclusion.
According to the PICOS principle, the inclusion criteria were as follows: (1) for population: adults (aged > 18 years) who were diagnosed with COVID-19 infection by a positive real-time reverse transcriptase-polymerase chain reaction (RT-PCR) assay and who underwent ECMO for hypoxemia; (2) for intervention, comparison, and outcome: studies on the association between differences in the duration of ECMO and patient survival, reporting corresponding risk estimates, such as odds ratios (ORs), relative risks (RRs), or hazard ratios (HRs), and their corresponding 95% confidence intervals (CIs), or providing related data; (3) for study design: random controlled trials (RCTs), post hoc analyses of RCTs, observational cohort studies, and cross-sectional studies.
Our exclusion criteria included: (1) reviews and studies with insufficient data; (2) populations with other established conditions (e.g., diabetes population) at baseline; and (3) if the same population was used in multiple studies, we excluded the less informative article.

Data Extraction and Quality Assessment
Data extraction and quality assessment of the included studies were conducted independently by two researchers.Information extracted included author, year of publication, country, study design, data source, follow-up time, sample size, mean age, gender, ECMO type, ECMO initiation, baseline comorbidities, other treatment, time period category, baseline data, estimated effect, and adjustments.
The Newcastle-Ottawa Scale (NOS) was used to assess the quality of the included observational studies.The scores range from zero to nine to evaluate the selection, comparability, and outcome of articles.Studies with NOS scores of one to three, four to six, and seven to nine are considered to be of low, medium, and high quality, respectively [12,13].

Statistical Analysis
All data (e.g., age, time) expressed as quartiles and medians were converted to the mean and standard deviation [14][15][16].To elucidate the differences in baseline pre-ECMO and ECMO time between survivors and non-survivors in severe COVID-19 patients, we respectively pooled the pre-ECMO (time from symptom onset to ECMO, time from COVID-19 diagnosis to ECMO, time from hospital admission to ECMO, time from intensive care unit (ICU) admission to ECMO, and time from MV or intubation to ECMO) and ECMO time of survivors and non-survivors using the inverse-variance method and random model to generate the effect size standardized mean difference (SMD).
OR is approximately equivalent to RR or HR when the outcome is rare [17].Therefore, to determine the relationship between time and survival, the ORs, RRs, and HRs and their 95% CIs were pooled, respectively, using a random-effects model to improve reliability.We estimated the effect size by calculating the natural logarithm of the OR, RR, or HR (log [OR], log [RR], or log [HR]) and their standard error (SElog [OR], SElog [RR], or SElog [HR]) to be pooled.Pre-ECMO and ECMO time were analyzed as a categorical variable, with the group with the longest time compared to the group with the shortest time.Time was analyzed as a continuous variable, and the units of the time (per one-day increase) were standardized.
A linear regression model was used to identify directional associations between ECMO duration and pre-ECMO time, including time from symptom onset to ECMO, time from COVID-19 diagnosis to ECMO, time from hospital admission to ECMO, time from ICU admission to ECMO, and time from MV or intubation to ECMO.SPSS version 16.0 software (SPSS Inc., Chicago, IL, USA) and Review Manager (RevMan) version 5.4 (The Cochrane Collaboration 2014; Nordic Cochrane Center Copen-hagen, Denmark) were used for statistics and analysis.A p-value of < 0.05 was considered statistically significant.

Heterogeneity Test, Publication Bias, and Sensitivity Analysis
We calculated the statistical p-value using the Q-test, with a p-value < 0.1 representing a significant difference between the two groups.To estimate the degree of heterogeneity, we applied the I 2 test between studies.Low heterogeneity, moderate heterogeneity, and high heterogeneity were defined as I 2 < 50%, 50-75%, and >75%, respectively [18].Sensitivity analyses were performed by omitting each study in turn.S3.Ultimately, our meta-analysis included a total of 54 studies with 55 cohorts .

Literature Search
A linear regression model was used to identify directional associations between ECMO duration and pre-ECMO time, including time from symptom onset to ECMO, time from COVID-19 diagnosis to ECMO, time from hospital admission to ECMO, time from ICU admission to ECMO, and time from MV or intubation to ECMO.SPSS version 16.0 software (SPSS Inc., Chicago, IL, USA) and Review Manager (RevMan) version 5.4 (The Cochrane Collaboration 2014; Nordic Cochrane Center Copenhagen, Denmark) were used for statistics and analysis.A p-value of < 0.05 was considered statistically significant.

Heterogeneity Test, Publication Bias, and Sensitivity Analysis
We calculated the statistical P-value using the Q-test, with a p-value <0.1 representing a significant difference between the two groups.To estimate the degree of heterogeneity, we applied the I 2 test between studies.Low heterogeneity, moderate heterogeneity, and high heterogeneity were defined as I 2 < 50%, 50-75%, and >75%, respectively [18].Sensitivity analyses were performed by omitting each study in turn.S3.Ultimately, our meta-analysis included a total of 54 studies with 55 cohorts .Other sources include the Critical Care Medicine website, the Critical Care website, and the American Journal of Respiratory and Critical Care Medicine website.

Sensitivity Analysis, Subgroup Analysis, and Publication Bias
We performed a sensitivity analysis for the time from intubation (MV) to ECMO.In the analysis of the OR per one-day increase group, by excluding Lebreton's study, the OR became 1.31 (95% CI: 1.14-1.51).Other sensitivity analyses by deleting one-by-one studies showed consistent results (Supplementary Figure S3).Due to the limited number of included studies (n < 10), subgroup and publication bias analyses were not performed according to the guidelines and predefined criteria.

Main Finding
Based on the meta-analysis of 54 studies with 55 cohorts and 13,691 COVID-19 patients, it was found that: (1) non-survival ECMO patients had a longer pre-ECMO time than survivors, including time from COVID-19 diagnosis to ECMO, time from hospital admission to ECMO, time from ICU admission to ECMO, time from intubation or MV use to ECMO, and there was no sufficient evidence to prove the association between pre-ECMO time and COVID-19 survival; (2) there is a longer ECMO time in non-survival COVID-19 patients than survivors; (3) there is no linear relationship between pre-ECMO time and ECMO duration.Although our analysis showed differences in pre-ECMO and ECMO time for survivors versus non-survivors, the relationship between the two needs to be further explored.
The impact of ECMO on the COVID-19 prognosis is significant.The timing and duration of ECMO are significant factors to consider when treating critically ill patients with COVID-19.Previous studies have confirmed that early ECMO intervention after MV improves survival in patients with ARDS caused by influenza A virus subtype H1N1 pneumonia [73].Most patients with ARDS and severe SARS-CoV-2 pneumonia receive delayed treatment and deteriorate rapidly.Almost all of the studies we included indicated that the pre-ECMO period was shorter in the survival group than in the non-survival group for COVID-19 patients treated with ECMO.Li et al. demonstrated that patients with COVID-19 who received early ECMO treatment had lower mortality than those who received late ECMO treatment [10].Furthermore, it has been suggested that early use of ECMO may lead to a better prognosis [74], while prolonged ECMO treatment may increase the risk of death and multi-organ failure [75].A study has shown that an invasive MV duration longer than 7 days before ECMO is a significant prognostic factor for death [76].Therefore, it is recommended to initiate ECMO as soon as possible [77].
The management of ECMO is also a major factor in the mortality of severe COVID-19 patients.ECMO is classified into three categories based on the route of blood transfusion: veno-venous ECMO (VV-ECMO), venous-arterial ECMO (VA-ECMO), and hybrid ECMO configurations.VV-ECMO provides only respiratory assistance, while VA-ECMO provides both circulatory and respiratory assistance.The choice of ECMO category may impact the patient's prognosis.In critically ill COVID-19 patients, VV-ECMO is the option when circulatory failure is not present.When circulatory failure is present, such as in the case of refractory hypoxemia associated with ARDS or shock associated with septic cardiomyopathy, VA-ECMO, or hybrid ECMO, is required.Because the use of different ECMOs is not reported in detail in the included literature, we did not perform subgroup analyses for this classification.A meta-analysis was conducted to investigate the effect of exposure to severe hyperoxemia on mortality and neurological outcomes in VA-ECMO-supported patients.The findings showed that exposure to severe hyperoxemia is associated with higher mortality (OR = 1.80, 95% CI: 1.16-2.78)and a poorer neurological outcome (OR = 1.97, 95% CI: 1.30-2.9).Therefore, it is recommended that efforts be made to avoid severe hyperoxemia during VA-ECMO support [78].In addition, the prolonged use of ECMO increases the chances of nosocomial infections due to COVID-19 infection, which leads to impaired immune function in patients [79,80].Therefore, it is crucial to improve intubation management.Also, ECMO anticoagulation management needs to pay close attention.Several anticoagulation strategies have been implemented to improve the outcome of COVID-19 patients treated with ECMO.For instance, nafamostat mesylate, a promising anticoagulant drug, could be used for systemic anticoagulation during ECMO administration.It may be able to serve as a feasible and safe option for anticoagulation during ECMO in critically ill patients with COVID-19 [81].Coagulation tests, such as activated clotting time, should be monitored regularly by healthcare professionals to avoid thrombosis or bleeding [82].In addition to this, the COVID-19 pandemic has severely strained intensive care resources in hospitals [83].Although patients may meet the ECMO to Rescue Lung Injury in Severe ARDS trial (EOLIA trial) criteria, ECMO support may not be initiated in time [84,85].
Studies investigating the association between pre-ECMO and ECMO duration and survival in COVID-19 patients have produced inconsistent results.After multifactorial adjustment, Nesserler et al. reported a higher mortality rate with longer pre-ECMO duration (HR = 1.74, 95% CI: 1.07-2.83)[49], while Saeed et al. did not reach the same conclusion (HR = 1.01, 95% CI: 0.98-1.03)[57].Our results indicated a difference in pre-ECMO time (e.g., MV or time to intubation to ECMO) between COVID-19 survivors and non-survivors, but this time is not statistically related to COVID-19 survival.It is important to note that the limited number of studies may have biased these results.Therefore, caution is advised when interpreting these findings, and further studies are needed to validate the relationship between ECMO-related time and the survival of critically ill COVID-19 patients.In addition, the inconsistency of the two statistical methods introduced some bias.As we are unable to adjust for all confounding factors, only survival and nonsurvival were considered when combining for time.Instead, in conducting research on the relationship between them, multifactor-adjusted studies were included, in which they adjusted for confounders such as age, comorbidity, type of ECMO, Respiratory ECMO Survival Prediction (RESP) score, and sequential organ failure assessment (SOFA) score, which was the main reason for the inconsistent results.In addition, we included studies for both univariate and multivariate analyses, and due to the limited number of included articles, we were unable to separately analyze studies adjusted for confounding factors.The effect of confounding factors on outcomes still needs to be elucidated.
It is also worth discussing whether the timing of pre-ECMO has an impact on the timing of ECMO.According to our results, in either survivors or non-survivors, pre-ECMO time showed no linear relationship with ECMO duration.In combination with previous relevant clinical studies and the recommendations of the ELSO, the timing of ECMO should be considered when the patient is at or above 50% risk of death in reference to any cause of hypoxic respiratory failure, and ECMO treatment should be initiated when the patient is at or above 80% risk of death.Currently, ECMO is only used as a supportive tool to allow time for primary disease treatment, rather than as a treatment in itself.Early use of ECMO can prevent cellular damage to organs and tissues caused by hypoxic metabolism and provide a favorable opportunity to treat the primary disease.Therefore, the duration of ECMO use is closely related to the improvement of the primary disease.That is, the duration of ECMO use may be shortened if earlier use of ECMO provides more adequate time to better support the treatment of the primary morbidity and if the primary morbidity improves during ECMO use.In contrast, a shorter duration of pre-ECMO does not mean a shorter duration of ECMO if the primary morbidity is not controlled.Although the early use of ECMO in severe COVID-19 patients is supported, there are no more studies that clearly show a relationship between pre-ECMO and the duration of ECMO use, and this issue still requires ongoing attention.

Underlying Mechanism
The impact of initiating ECMO early on patients with severe COVID-19 is multifactorial.Firstly, it is important to note that ARDS in COVID-19 patients aligns with the Berlin definition [86].However, Gattinoni et al. proposed an alternative perspective, suggesting that lung compliance is significantly reduced in COVID-19 patients and that severe hypoxemia is more commonly associated with ventilation/perfusion (VA/Q) mismatch.For patients with COVID-19, conventional treatments such as mechanical ventilation or prone ventilation do not improve oxygenation by recruiting collapsed areas [87].Therefore, early use of ECMO can benefit patients by minimizing ventilator-induced lung injury.Secondly, it has been found that death from COVID-19 is closely linked to hypercoagulable and thrombotic states, as supported by Yin et al.According to their report, platelet levels are higher in COVID-19 patients than in non-COVID-19 pneumonia patients [88].When administering ECMO cannulation, a systemic anticoagulation strategy is typically employed to ensure safety [89].Also, the ELSO guideline proposes to consider anticoagulation therapy targeting the higher end of normal for ven-venous ECMO in COVID-19 patients due to their known hypercoagulable state [6].This approach reduces the risk of thrombosis and subsequent death resulting from the intrinsic thrombotic state, providing an additional benefit to severe COVID-19 patients.Thirdly, hospital-acquired infections are more prevalent in hospitals than in other settings.Prolonged use of ECMO is associated with an increased risk of nosocomial infections [90], which may contribute to mortality.

Clinical Implications
Neither the clinical guidelines related to COVID-19 published by the WHO [91] nor the regularly updated guidelines of the National Institutes of Health [92] take a positive position on whether to apply ECMO to patients with severe COVID-19.However, according to the EOLIA trial, the ELSO made a standard recommendation that ECMO therapy could be used in certain patients with COVID-19 [6].Subsequently, the Korean Society for Thoracic and Cardiovascular Surgery (KSTCVS) [93] and Chinese experts [94][95][96] have also recommended ECMO as a salvage therapy for patients with severe COVID-19 who have not responded to conventional ARDS therapy.
Our results show a significant difference between COVID-19 survivors and nonsurvivors in terms of pre-ECMO time and ECMO duration.This suggests that by adjusting the timing of ECMO, there may be an impact on the survival of patients with severe COVID-19.However, our study failed to identify an association between pre-ECMO and ECMO timing and the survival of COVID-19 patients.Therefore, additional studies and more articles are required to confirm the relationship between ECMO timing and survival in patients with COVID-19, to determine the optimal timing and duration of ECMO treatment for COVID-19, and thus to improve survival in severe COVID-19 patients.In the meantime, further guideline updates or clinical trials may highlight the differences in ECMO-related time in COVID-19 patients and will still require our continued attention.
In addition, it is worth noting that prior treatment with ECMO is crucial for patients with severe COVID-19.Non-invasive respiratory support has been shown to reduce the need for intubation and invasive MV [60], but mortality may be increased in patients with COVID-19 who fail non-invasive ventilation strategies [97,98].Indeed, dysregulated spontaneous breathing, associated with wide transpulmonary pressure swings, may increase the risk of harmful "patient spontaneous induced lung injury" on non-invasive MV or high-flow nasal cannula therapy, leading to a greater susceptibility to pneumonia and fibrosis [99,100].Also, based on existing studies, emergency tracheal intubation and MV are required for severe COVID-19 patients exhibiting signs of respiratory distress, hypoxemia, or encephalopathy.A previous meta-analysis demonstrated that a longer duration of invasive MV was associated with a poor prognosis [101].Patients who remain unsuccessful after optimization of MV strategies may be considered for pulmonary resuscitation strategies.While invasive MV can benefit patients, it can also cause ventilator-associated lung injury if not used properly.This can be caused by high driving pressure, which has been linked to increased mortality in severe cases of COVID-19.
A study has shown that the prone position can relieve atelectasis even at low positive end-expiratory pressure (PEEP) levels [102].In patients with severe hypoxia, the prone position can be considered an operation to preserve PEEP.Thus, the prone-position strategy can balance the adverse effects of invasive MV [30].At the same time, the prone position can improve oxygenation in patients with prolonged hypoxemia during ECMO.When using lung-protective ventilation to reduce lung injury, the addition of prone position therapy in conjunction with ECMO can further aid and optimize alveolar recovery.This combination of strategies (ECMO and prone position) has been shown to improve overall survival [103].It is also important to consider potential complications when applying the combined strategy, such as accidental decannulation and kinking of the infusion system due to the prone position, as well as coagulation disorders and pressure ulcers.

Comparison with Prior Meta-Analysis
Previous meta-analyses have compared the effect of the presence or absence of ECMO use on COVID-19 mortality or the difference in mortality between COVID-19 and other virus-induced diseases treated with ECMO [7,77,[104][105][106].For example, Kusumawardhani's study found a significantly higher incidence of mortality in COVID-19 patients treated with ECMO compared to those not treated with ECMO (OR = 15.79,95% CI: 4.21-59.28,p < 0.0001) [106].Ramanathan's meta-analysis reported an in-hospital ECMO mortality rate of 37.1% for COVID-19 patients, which is similar to that of patients with non-COVID-19-related ARDS [7].
Currently, no studies have summarized the effect of pre-ECMO and ECMO duration on mortality in COVID-19 patients.Our meta-analysis is the first study based on this.We analyzed pre-ECMO and ECMO timing in COVID-19 survivors versus non-survivors, as well as the association between the ECMO-related time and COVID-19 survival.In addition, we sought to explore the relationship between pre-ECMO time and ECMO time in patients with COVID-19.Although no robust and definitive results were obtained, our study gives direction for future research.

Strength and Limitation
Our study is the first to investigate the relationship between ECMO duration and survival in patients with severe COVID-19.First, we conducted a comprehensive analysis of relevant literature without language restrictions, focusing on a specific and exclusive population.Second, we explored the relationship between ECMO duration and death in COVID-19 patients in two ways: by using continuous variables to explore time-specific differences and by analyzing adjusted effect size to explore associations between the two.Finally, although our study did not demonstrate a causal relationship between ECMO time and survival in ill COVID-19 patients, the difference in ECMO time between the surviving and dying populations could still suggest relevant studies for the following investigations.
Several limitations in our meta-analysis should be noted.First, relatively high heterogeneity was observed in our results.This may be due to the fact that our analysis was based on cohort studies.We included studies with both univariate and multivariate analyses, whose unadjusted confounders may have influenced our results.Second, despite the inclusion of a large number of studies, only a few articles reported effect sizes for the relationship between pre-ECMO or ECMO time and COVID-19 survival (n = 15).And the time periods varied across the studies we included, resulting in a small number of studies being included in each period (time from intubation or MV to ECMO = 3 for categorical variables and 7 for continuous variables).This may be another cause of bias and inaccurate results.In addition, the studies included in this analysis reported different effect sizes (OR, RR, and HR), making it impossible to report the pooled results due to differences in statistical methodology.It is important to continue to pay attention to relevant research and refine each analysis as needed in the future.Third, out of the 54 studies included, the majority were conducted in America.However, due to the limited number of studies included, subgroup analysis could not be conducted.Hence, the potential influence of confounding and potential intermediate factors, such as regional differences, study design, follow-up, and other clinical characteristics across studies, needs further investigation.Fourth, there were differences in the indicators for ECMO initiation in each study, with some studies following ELSO guidelines and others relying on decisions made by local experts, which may have impacted the results.Fifth, the varying definitions of death in each study and the inconsistent timing of these definitions, coupled with the short-term follow-up periods in most studies, may have led to an underestimation of reported mortality.Sixth, due to the observational nature of the analyses we included and the limited number of articles included, trial sequential analysis was not performed to assess the robustness of the findings and the need for further research.Finally, the meta-analysis is based on observational studies, so causality cannot be deduced from our study.

Conclusions
Based on current evidence, our results suggest that there are differences in pre-ECMO between COVID-19 survivors and non-survivors.We did not have sufficient evidence of a significant association between pre-ECMO time and survival in COVID-19 patients.In addition, non-survivors had a longer ECMO duration than survivors.Pre-ECMO time does not affect the timing of ECMO.Considering the limited evidence and possible bias, further studies in pre-ECMO and ECMO time on the survival of COVID-19 patients are needed to explore the association between them.Future guidelines may emphasize ECMO timing-specific risk assessment and management for severe COVID-19.

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/jcm13030868/s1,Table S1: guidelines of the Preferred Reporting Item for Systematic Review and Meta-Analysis 2020 (PRISMA 2020); Table S2: Detailed description of the search strategy; Table S3: Studies excluded (n = 150) with reasons; Table S4.Quality assessment of included studies; Figure S1 Institutional Review Board Statement: Not applicable.

Figure 1
Figure 1 shows the flowchart of the database search process.According to the preformulated search protocol, a total of 2473 publications were identified in the initial search (PubMed = 1697; Cochrane Library = 21; Embase = 532; other sources = 223).After excluding 528 duplicates and 1627 irrelevant publications after title and abstract screening, 318 articles were processed for full-text assessment.After removing 114 articles with specific publication types without data, a further 150 articles were excluded for the following reasons: (1) studies without data of interest (n = 23); (2) duplicated cohorts (n = 5); (3) studies without full text (n = 24); (4) studies not focusing on pre-ECMO or ECMO time (n = 37); (5) studies not focusing on specific populations (n = 10); (6) studies not focusing on target outcome (n = 19); and (7) case reports (n = 32).All excluded studies (n = 150) and their corresponding reasons are listed in Supplementary TableS3.Ultimately, our meta-analysis included a total of 54 studies with 55 cohorts.

Figure 1
Figure 1 shows the flowchart of the database search process.According to the preformulated search protocol, a total of 2473 publications were identified in the initial search (PubMed = 1697; Cochrane Library = 21; Embase = 532; other sources = 223).After excluding 528 duplicates and 1627 irrelevant publications after title and abstract screening, 318 articles were processed for full-text assessment.After removing 114 articles with specific publication types without data, a further 150 articles were excluded for the following reasons: (1) studies without data of interest (n = 23); (2) duplicated cohorts (n = 5); (3) studies without full text (n = 24); (4) studies not focusing on pre-ECMO or ECMO time (n = 37); (5) studies not focusing on specific populations (n = 10); (6) studies not focusing on target outcome (n = 19); and (7) case reports (n = 32).All excluded studies (n = 150) and their corresponding reasons are listed in Supplementary TableS3.Ultimately, our meta-analysis included a total of 54 studies with 55 cohorts.

Figure 1 .
Figure 1.Flow chart of study selection in the systematic review and meta-analysis of ECMO time difference in the survival of COVID-19.

Figure 2 .
Figure 2. Forest plot for the association between time from symptom onset to ECMO and survival in COVID-19 patients.(a).Forest plot showing the time differences between survivors and nonsurvivors in COVID-19 patients.(b).Forest plot for the association between time and survival, analyzed as continuous variables (per one-day increase).

Figure 2 .
Figure 2. Forest plot for the association between time from symptom onset to ECMO and survival in COVID-19 patients.(a).Forest plot showing the time differences between survivors and nonsurvivors in COVID-19 patients.(b).Forest plot for the association between time and survival, analyzed as continuous variables (per one-day increase).
: (A).Forest plot showing the differences in time from COVID-19 diagnosis to ECMO between survivors and non-survivors in COVID-19 patients; (B).Forest plot showing the differences in time from hospital admission to ECMO between survivors and non-survivors in COVID-19 patients; (C).Forest plot showing the differences in time from ICU admission to ECMO between survivors and non-survivors in COVID-19 patients; D. Forest plot for the association between time from intubation or MV to ECMO and survival in COVID-19 patients.(a).Forest plot showing the differences in time between survivors and non-survivors in COVID-19 patients.(b).Forest plot for the association between time and survival, analyzed as category variables (highest vs. lowest).(c).Forest plot for the association between time and survival, analyzed as continuous variables (per one-day increase); Figure S2.Regression analysis of pre-ECMO time and ECMO duration.(a).Survivors; (b).Non-survivors; Figure S3.Sensitivity analysis of time from MV or intubation to ECMO difference in COVID-19 for mortality by omitting one study at once.(a).Category variables; (b).Continuous OR variables; (c).Continuous HR variables.Author Contributions: Z.T., L.S. and Y.L. contributed to the study concept and design and revised the draft.Z.T. performed the search strategy and contributed to database research, the acquisition of data, and statistical analyses.Z.T., L.S., X.C., H.H. and Y.L. participated in data analysis, reviewed, and approved the final manuscript.All authors have read and agreed to the published version of the manuscript.Funding: This research was funded by National High-Level Hospital Clinical Research Funding (2022-PUMCH-B-115, 2022-PUMCH-D-005).

Table 1 .
Basic characteristics of the articles included in the meta-analysis.