Percutaneous Versus Surgical Cannulation for Veno‐Arterial Extracorporeal Life Support: A Systematic Review and Meta‐Analysis

ABSTRACT

Background and Aims

Extracorporeal life support (ECLS) is a critical intervention in the management of severe cardiac and respiratory failure. Among the commonly used cannulation techniques, percutaneous cannulation (PC) and surgical cannulation (SC) are most prevalent; however, their comparative safety and effectiveness remain under debate. The aim of this systematic review and meta‐analysis was to evaluate and compare the clinical outcomes associated with PC and SC in patients undergoing veno‐arterial ECLS.

Method

A comprehensive literature search was conducted in PubMed, Google Scholar, and the Cochrane Library to identify relevant studies. Six studies with a total of 13,744 patients (9962 PC and 3782 SC) met the inclusion criteria. Data extraction and analysis were performed using RevMan software, applying a random‐effects model. Meta‐regression and sensitivity analyses were also performed to evaluate heterogeneity. Dichotomous outcomes were assessed using relative risk (RR) with 95% confidence intervals (CIs), and continuous outcomes using mean differences (MDs) with 95% CIs. Statistical significance was set at p ≤ 0.05.

Results

PC was associated with a significantly lower risk of cannulation site infections (RR: 0.56, 95% CI: 0.41–0.78, p = 0.0005). A trend toward fewer vascular complications was also observed with PC compared to SC (RR: 0.48, 95% CI: 0.23–1.00, p = 0.05). Secondary outcomes, including in‐hospital mortality, duration of ECLS, renal replacement therapy, weaning success, limb ischemia, fasciotomy, and amputation, showed mixed or inconclusive findings.

Conclusion

Compared with SC, PC appears to reduce the risk of infections and vascular complications in veno‐arterial ECLS patients. However, further high‐quality studies are required to establish its benefits across other clinical outcomes.

List of Abbreviations

AI

Aghna Iman

AN

Ariba Nazir

BMI

body mass index

DM

diabetes mellitus

DPCs

distal perfusion catheters

ECLS

extracorporeal life support

ECMO

extracorporeal membrane oxygenation

ECPR

extracorporeal cardiopulmonary resuscitation

LK

Laiba Khurram

MD

mean difference

MI

myocardial infarction

MM

Muzna Murtaza

PAD

peripheral arterial disease

PC

percutaneous cannulation

RCTs

randomized controlled trials

RR

risk ratio

RRT

renal replacement therapy

SC

surgical cannulation

SZS

Syeda Zuha Sami

ZMH

Zainab Muhammad Hanif

1. Introduction

Extracorporeal life support (ECLS), commonly known as extracorporeal membrane oxygenation (ECMO) [1] is a modified form of cardiopulmonary bypass that provides prolonged tissue oxygen delivery in patients suffering from respiratory and cardiac failure [2]. ECLS has been effective in the treatment of respiratory failure, with survival rates of 84% in newborns, 76% in children, and 50% in adults [3]. While it can be life‐saving, risks such as limb ischemia, infections, and various vascular complications are commonly seen during ECLS cannulation [4, 5]. Hence, several cannulation techniques involving percutaneous and surgical cannulation are continuously being employed to reduce such risks [6, 7].

Percutaneous cannulation (PC) has become a more popular option for ECMO than surgical cannulation (SC) [7]. Research indicates that, in comparison to SC, PC is linked to reduced rates of in‐hospital mortality and fewer complications such as bleeding at the cannulation site and systemic infections [8]. With imaging support, intensivists can execute PC with a high success rate and low complication rate [9]. Even though PC is appropriate for femoral arterial cannulation and all venous access sites, SC is still needed for veno‐arterial (VA) support via cervical vessels, central thoracic cannulation, and failed percutaneous approaches [7].

Given this knowledge gap, we performed this rigorous meta‐analysis to assess the procedural and clinical outcomes in patients undergoing PC or SC for VA‐ECLS. This meta‐analysis aims to provide a clearer understanding of the relative benefits and risks of each approach and valuable insights into which technique will enhance the quality‐of‐care for ECLS patients within the field of interventional cardiology.

2. Methodology

This systematic review and meta‐analysis was conducted according to Preferred Reporting Items for Systematic Review and Meta‐Analysis (PRISMA) guidelines [10] with its protocol registered in PROSPERO CRD42024581340.

2.1. Data Source and Search Strategy

A comprehensive literature search was conducted on PubMed, Google Scholar, and Cochrane Library from database inception until August 2025. Bibliographies of relevant articles were also searched to make sure no studies were missed. No filters were applied on language, sample size, year of publication, author name, and institution/country of publication. The electronic search strategy was conducted using the following keywords: “Percutaneous Cannulation” OR “Surgical Cannulation” AND “ECLS” OR “Extracorporeal Life Support” OR “ECMO” OR “Extracorporeal Membrane Oxygenation” OR “ECPR” OR “Extracorporeal Cardiopulmonary Resuscitation.” A detailed description of the complete search strategy applied for each database is given in Supporting Information Table 1.

2.2. Study Selection and Eligibility Criteria

All articles retrieved from the systematic search were exported to EndNote X9 Reference Manager (Clarivate Analytics, US), where duplicates were removed among different online databases. Two independent reviewers (S.Z.S. and Z.M.H.) initially screened the remaining articles based on the title and abstract that met the study population. Finally, full texts were evaluated for relevance. Any discrepancies were resolved by discussion with a third reviewer (A.N.). The search was restricted to the following inclusion criteria: (a) studies that compared percutaneous vs. surgical ECLS in patients with cardiogenic shock or cardiac arrest, (b) patients aged 18 years or above, and (c) outcomes of interest were reported which included: in‐hospital mortality, cannulation site infection, limb ischemia, limb fasciotomy, amputation, vascular complications (perforation/dissection/bleeding), renal replacement therapy, weaning success, and ECLS duration (days). The exclusion criteria were: (a) conference papers, abstracts, case series, case reports, and duplicated studies, (b) single‐arm studies, (c) age below 18 years, (d) studies that lacked a comparator group, and (e) studies that did not report any outcome of interest.

2.3. Data Extraction and Quality Assessment

Two reviewers (S.Z.S. and A.N.) independently extracted the data from selected studies. The general characteristics of the included studies were extracted and entered on the Microsoft Excel sheet the authors created. The following data was extracted from each study: (a) study name and year, (b) study design, (c) sample size, (d) the number of patients in each group (percutaneous vs. surgical), (e) general patient characteristics (age, male sex, body mass index (BMI), diabetes mellitus, hypertension, heart failure, peripheral arterial disease, dyslipidemia, and myocardial infarction), and (g) all outcomes of interest. Continuous outcomes reported as median with interquartile ranges were converted to mean and standard deviation using Wan's method [11, 12]. Two independent reviewers (M.M. and Z.M.H.) performed a quality assessment of selected cohort studies using the Newcastle–Ottawa Quality Assessment tool [13]. The overall risk of bias for each study was evaluated and rated: as low, unclear, and high, based on the assessments detailed in Supporting Information Table 2. This thorough approach offers a meticulous evaluation of study quality and possible biases, leading to trustworthy and unbiased results.

2.4. Statistical Analysis

For statistical analysis, Review Manager (RevMan Version 5.4.1) Cochrane Collaboration Network was used. Dichotomous data were used to drive the risk ratio (RR) and for the continuous outcome, mean difference (MD) was obtained. Between‐study variance was estimated using the DerSimonian–Laird method under a random‐effects inverse variance model. Confidence intervals were calculated without additional Hartung–Knapp adjustment, in accordance with the default settings in RevMan. No variance‐stabilizing transformation (e.g., Freeman–Tukey or logit) was applied, as outcomes were synthesized using risk ratios (RRs) and mean differences (MDs). The probability value of p ≤ 0.05 was considered statistically significant. Higgins I² was used to measure heterogeneity [14, 15]. The value of I² = 25%–50% was regarded as mild heterogeneity, 50%–75% as moderate heterogeneity, and > 75% as high heterogeneity. When we noticed a lot of heterogeneity, we performed a leave‐one‐out analysis to rule out the cause. Comprehensive Meta‐Analysis (CMA) (Version 3.3.070) was used to perform meta‐regression to explore sources of heterogeneity. This rigorous approach ensured that our results were accurate and could be interpreted with confidence.

3. Results

3.1. Study Selection and Characteristics

A comprehensive literature search was conducted and yielded 2639 articles. After removing the duplicates and assessing eligibility, six cohorts were included in this meta‐analysis. The result of the literature search is summarized in the form of a PRISMA flowchart (Figure 1). A total of six studies comprising 13,744 patients (9962 in percutaneous group vs. 3782 in surgical group). The mean age of patients in the percutaneous group is 57.08 years, and 54.4 years in the surgical group. The general characteristics of included studies are presented in Table 1 and the baseline characteristics of the patients included in each study are presented in Table 2.

FIGURE 1.

PRISMA flowchart.

TABLE 1.

Characteristics of included studies.

S. no.	Author	Year	Study type	Sample size (N)	Patient population (n)		Age (y) mean ± SD/median (IQR)		Male sex (%)		BMI (kg/m2)
					PC group	SC group	PC group	SC group	PC group	SC group	PC group	SC group
1	*Danial P. et al. [16]	2018	Retrospective cohort	532	266	266	55.2 ± 14.9	54.2 ± 14.4	196 (74)	186 (70)	26.5 ± 5.0	26.4 ± 5.5
2	Goslar T. et al. [17]	2016	Retrospective cohort	48	23	25	55.0 ± 9.0	50.0 ± 13.0	18 (78)	17 (68)	—	—
3	Lee H.S. et al. [18]	2024	Retrospective cohort	99	33	66	58.8 ± 13.8	59.4 ± 13.5	29 (87.9)	55 (83.3)	24.7 ± 2.4	24.7 ± 3.4
4	Saiydoun G. et al. [19]	2021	Prospective cohort	120	61	59	59.3 ± 12.1	52.7 ± 17.5	47 (77)	39 (66)	25.3 ± 3.8	27 ± 6.1
5	**Wang L. et al. [8]	2022	Retrospective cohort	12,592	9249	3343	57 (46–65)	57 (44–65)	6366 (69)	2266 (68)	—	—
6	Wilhelm M.J. et al. [6]	2022	Retrospective cohort	353	330	23	58.2 ± 16.1	54.8 ± 15.8	248 (75.2)	12 (53.2)	—	—

Note: * Propensity‐matched data and ** missing values for some variables resulted in different denominators.

Abbreviation: BMI, body‐bass index.

TABLE 2.

Baseline characteristics of patients.

S. no.	Author	DM n (%)		Hypertension n (%)		Heart failure n (%)		PAD n (%)		Dyslipidemia n (%)		MI n (%)
		PC group	SG group	PC group	SC group	PC group	SC group	PC group	SC group	PC group	SC group	PC group	SC group
1	*Danial P. et al. 2018 [16]	55 (21)	60 (23)	83 (31)	91 (34)	151 (57)	144 (54)	6 (2)	3 (1)	98 (37)	86 (32)	55 (21)	49 (18)
2	Goslar T. et al. 2016 [17]	—	—	—	—	—	—	—	—	—	—	—	—
3	Lee H.S. et al. 2024 [18]	9 (27.3)	20 (30.3)	15 (45.5)	30 (45.5)	1 (3.0)	5 (7.6)	1 (3.0)	1(1.5)	5 (15.2)	6 (9.1)	5 (15.2)	5 (7.6)
4	Saiydoun G. et al. 2021 [19]	21 (34)	9 (15)	31 (51)	22 (37)	7 (11)	17 (29)	3 (5)	4 (7)	—	—	34 (56)	17 (29)
5	**Wang L. et al. 2021 [8]	734 (8)	236 (7)	979 (11)	324 (10)	1510 (16)	563 (17)	141 (2)	58 (2)	516 (6)	153 (5)	2724 (29)	948 (28)
6	Wilhelm M.J. et al. [6]	—	—	—	—	—	—	28 (8.5)	2(8.7)	—	—	—	—

Note: * Propensity‐matched data and ** missing values for some variables resulted in different denominators.

Abbreviations: DM, diabetes mellitus; MI, myocardial infarction; PAD, peripheral arterial disease.

3.2. Risk of Bias of Included Studies

The risk of bias in the included cohorts is assessed using the Newcastle–Ottawa Quality Assessment tool [13]. Three out of six cohorts (Danial et al., Saiydoun et al., and Wang et al.) [8, 16, 19] demonstrated a low overall risk of bias, indicating that the findings from these studies are reliable. However, other studies showed some concerns, details of which are shown by the risk of bias assessment (Supporting Information Table 2). Confirmation using Begg's and Egger's tests was also done. The p‐value of the Begg's and Egger's tests is provided in Supporting Information Table 4.

3.3. Primary Outcomes

We analyzed two primary outcomes, cannulation site infection, and vascular complications, to assess patients undergoing PC in ECLS patients.

3.4. Cannulation Site Infection

A total of three [16, 19, 20] studies reported data on cannulation site infection. The overall result shows that the percutaneous group has a significantly lower risk of on‐site infection compared to the surgical group, RR: 0.56 [95% CI (0.41, 0.78); p = 0.0005, I² = 0%] and the p‐value for the overall effect confirms the statistical significance. The results are consistent across studies, with no significant heterogeneity observed.

3.5. Vascular Complications

Our analysis of five studies [8, 16, 17, 19, 20] shows a trend toward low vascular complications with PC as opposed to the surgical group RR: 0.48 [95% CI (0.23, 1.00); p = 0.05 I² = 79%]. However, high heterogeneity is apparent for which sensitivity analysis was performed using the leave‐one‐out approach, and yet there was no significant change observed in heterogeneity.

4. Secondary Outcomes

4.1. In‐Hospital Mortality

Overall pooled results of two studies [8, 20] reporting in‐hospital mortality declared results statistically insignificant, with RR: 0.92 [95% CI (0.85, 1.01); p = 0.07, I² = 5%] suggesting a trend toward lower in‐hospital mortality in the percutaneous group. Mild heterogeneity (I² = 5%) was noted (Figure 2).

FIGURE 2.

Forest plot of in‐hospital mortality.

4.2. Weaning Success

Two studies [18, 19] that recorded weaning success showed a pooled RR of 1.20 [95% CI (0.72, 2.02); p = 0.48, I² = 64%], which showed that result is statistically insignificant with moderate heterogeneity. However, for moderate heterogeneity, we could not perform sensitivity analysis to assess the risk of bias by the leave‐one‐out method, because there were insufficient studies (Figure 3).

FIGURE 3.

Forest plot of weaning success.

4.3. Renal Replacement Therapy

Two studies [8, 18] in total reported renal replacement therapy. The results of combined findings revealed statistical insignificance [RR: 1.07, 95% CI: (0.63, 1.80); p = 0.81, I² = 71%]. However, for moderate heterogeneity, we were unable to conduct sensitivity analysis to evaluate the risk of bias using the leave‐one‐out method, because there are not enough studies (Figure 4).

FIGURE 4.

Forest plot of renal replacement therapy.

4.4. ECLS Duration (Days)

The pooled analysis of five studies [8, 16, 17, 18, 19] shows a negligible and non‐significant difference in ECLS duration between the percutaneous and surgical techniques, with a mean difference (MD) of 0.11 [95% CI (−0.05, 0.26); p = 0.19, I² = 0%]. Given that the confidence interval value and the p‐value are not significant, there is no statistically significant difference between the two treatment groups with results being consistent across studies with no heterogeneity (I² = 0%) (Figure 5).

FIGURE 5.

Forest plot of ECLS duration.

4.5. Limb Ischemia

For the outcome of limb ischemia, RR: 1.10 95% CI: [0.63, 1.91]; p = 0.74, I² = 64%) suggests that there is no statistically significant difference between the percutaneous and surgical groups in terms of the event rate. Moderate heterogeneity was found, which shows that there was some variation among the studies, for which we performed sensitivity analysis employing a leave‐one‐out approach, as a result of which a change in heterogeneity was noticed (Figure 6).

FIGURE 6.

Forest plot of limb ischemia.

4.6. Leave One Out of the Analysis

The pooled analysis of five studies [8, 16, 17, 19, 20] showed a risk ratio of 1.10 (95% CI 0.63, 1.91: p = 0.74, I² = 64%). Since moderate heterogeneity was identified, a leave‐one‐out analysis was performed by ruling out Danial et al. 2018 [16] which dropped the heterogeneity from moderate to mild, and the overall effect, which was previously deemed not significant (p = 0.74), became statistically significant (p = 0.03). The heterogeneity may be attributed to the study's low patient population and it being the only study with propensity‐matched data. Figure of forest plot of leave‐one‐out analysis displaying RR: 1.48 [95% CI (1.05, 2.09); p = 0.03, I² = 9% is given in Supplementary File (Figure 7).

FIGURE 7.

Forest plot of leave‐one‐out analysis of limb ischemia.

4.7. Limb Fasciotomy

In the outcome of limb fasciotomy, two studies [8, 16] were included. The pooled effect was [RR: 1.10, 95% CI [0.44, 2.73]; p = 0.84, I² = 70%]. However, for moderate heterogeneity, we could not perform sensitivity analysis to assess the risk of bias by the leave‐one‐out method, because sufficient studies are not present (Figure 8).

FIGURE 8.

Forest plot of limb fasciotomy.

4.8. Amputation

In the two studies [8, 16] that reported the events of amputation, the pooled result was found to be statistically insignificant (RR: 1.51, CI: [0.91, 2.49]; p = 0.11, I² = 0%) (Figure 9).

FIGURE 9.

Forest plot of amputation.

4.9. Meta‐Regression

We conducted a meta‐regression analysis to explore potential factors influencing the effect size of our primary outcome vascular complications across the included studies. The covariates examined were mean age, male percentage, and the presence of peripheral arterial disease (PAD). The analysis identified PAD as a significant predictor (coefficient: −0.3705; p = 0.0001), indicating that a higher prevalence of PAD among study populations was associated with an increased risk of cannulation‐related complications. In contrast, mean age (coefficient: −0.0318; p = 0.8812) and male percentage (coefficient: 0.1304; p = 0.3658) were not significantly associated with the primary outcome in this model.

Supporting Information Table 3 summarizes the regression coefficients and corresponding p‐values for all covariates assessed. Additionally, scatter plots illustrating the relationship between each covariate and vascular complications are presented in Supporting Information Figures S1–S3.

5. Discussion

Our systematic review and meta‐analysis which compared PC in contrast to SC, showed significantly lower risk of on‐site infections and vascular complications. Additional observations include an increased weaning success rate, with a reduction in ECLS duration and in‐hospital mortality, although not always statistically significant. Furthermore, no significant differences in limb ischemia, fasciotomy, amputation, and patients requiring renal replacement therapy were noted. Numerous clinical and procedural considerations impact the decision between percutaneous and surgical cannulation. Feasibility is frequently determined by patient anatomy, such as artery size, calcification, or prior vascular interventions [21]. Percutaneous femoral cannulation, especially when guided by ultrasonography, is often preferable in emergency settings like ECPR due to its much shorter insertion time (mean 10.3 min vs. 21.3 min for surgical cut‐down, p < 0.001) [18]. Furthermore, peripheral access is usually the sole practical method during continuous resuscitation [22] and eliminates the necessity for a sternotomy. The results of PC are greatly influenced by operator skill and institutional procedures. Intensivists who regularly used fluoroscopy (85%) and ultrasonography (93%) had a 99% success rate with little problems in one institutional series [23]. Likewise, when competent teams used image guidance [24], another location reported a 98% success rate for Veno venous ECMO insertion with low complication rates. According to the Extracorporeal Life Support Organization's (ELSO) standards, which stress customized selection based on patient characteristics, urgency, and institutional resources, outcome data must always be interpreted in the context of this clinical setting [25].

Infections are a major source of morbidity and mortality in patients receiving ECLS [26]. Studies have reported infection rates of 10.2% [27] to 40% [28] in ECLS patients. Our analysis found a lower incidence of on‐site infections in patients undergoing PC compared to SC. According to a retrospective article, the minimally invasive nature of PC, which reduces tissue exposure and disruption, facilitates easier postoperative care, and speeds up healing, is largely responsible for the lower infection rate [29]. This lowers the risk of pathogen introduction into the body. PC has been associated with less groin problems, such as fewer wound infections and lymphatic fistulae, according to many studies that examined 2483 patients in total [30, 31]. In contrast, SC, though providing rapid vessel access under visual guidance, involves greater tissue manipulation, longer procedural duration, and higher risk of hematoma, all of which contribute to elevated infection rates and prolonged hospitalization, thereby increasing susceptibility to hospital‐acquired infections [32, 33, 34]. The reduced infection risk with PC is particularly valuable in critically ill patients, where any additional infectious complication can significantly worsen the prognosis [35].

Vascular complications, including on‐site bleeding, perforations, and arterial dissection, were found to be less frequent with PC than with SC. Up to 28% of vascular problems at the femoral and axillary locations are still reported [36]. Intensivists using ultrasound and fluoroscopic guidance achieved a high success rate (98%) and low complication rate in percutaneous ECMO cannulation, suggesting that the use of ultrasound guidance in PC can aid in more accurate cannula placement, further reducing the risk of vascular injuries [9]. Infected tissues are more brittle and prone to complications like bleeding and arterial dissection [37, 38] which makes the aforementioned factor like the increased risk of infection with SC worsen the patient's condition, according to the review articles. This suggests that patients who undergo SC are at a higher risk of dying than those who undergo PC. Compared to SC, PC uses much smaller arterial cannulas, and two retrospective studies have found that using smaller cannulas is linked to a lower incidence of cannulation‐related adverse events, especially bleeding at the cannulation site [16, 39]. Hemostasis is easier to establish with PC 40 because of the smaller entry points and less tissue stress [40]. Hence, with the percutaneous approach, lower incidence of vascular complications is seen, inclining more attention to our analysis findings.

Due to the higher risk of complications and severe adverse effects associated with SC, our data revealed that patients who had SC reported a greater tendency of in‐hospital mortality, although the results are not statistically significant. Acute vascular complications or limb ischemia have been linked to a considerable rise in mortality [41, 42]. Cakici et al., a retrospective cohort analysis, also reported lower mortality in the percutaneous group (43% vs. 55%, p = 0.14) in a series of patients who received percutaneous femoral VA‐ECMO or SC of the right axillary/subclavian or the femoral artery with an arterial side‐graft. Nevertheless, the disparity failed to attain statistical significance, which aligns with our analysis as well [43]. Moreover, there seems to be a reduced need for renal replacement therapy (RRT) with PC [8, 44], due to better preservation of renal function as opposed to SC. However, our findings remained insignificant. The lack of statistical significance could be due to the variability among studies, differences in patient characteristics, or variations in ECLS management protocols. Nevertheless, the observed trends hold clinical significance and warrant additional exploration through larger and more homogeneous studies. A patient's dependence for life‐supporting procedures like ECMO or mechanical breathing eventually disappears as they wean [45]. Compared to SC, our research revealed a greater weaning success rate with PC, but one that was not statistically significant. Retrospective investigation indicates that PC is less intrusive, results in less physical trauma, and improves blood flow and oxygenation regulation, supporting respiratory and cardiovascular stability, which promotes healing and easier weaning [40, 46]. SC, on the other hand, requires a lot of tissue handling, which increases physical strain and lowers the chance of a successful weaning. Weaning patients with refractory cardiogenic shock from ECMO has a success rate ranging from 31% to 76% [47].

A slight, non‐significant difference between the groups was shown by the analysis. According to a recent single‐center retrospective study, PC [48] has a short mean VA‐ECMO commencement time of 6 min, which usually results in fewer issues and a speedier recovery. Percutaneous treatments may also help maintain patient stability and reduce the length of ECLS [49] by reducing physical stress and trauma. SC, on the other hand, is more intrusive, requires a large amount of tissue dissection, and raises inflammatory markers [50], all of which can make recovery more difficult and lengthen ECLS. Both percutaneous and surgical cannulation techniques have comparable risks associated with impaired limb blood flow, according to our meta‐analysis, which did not find any appreciable differences in the incidences of limb ischemia, fasciotomy, or amputation. Limb ischemia is still a major problem in ECLS [51], which emphasizes the necessity for preventive measures such distal perfusion catheters (DPCs), which have shown encouraging results [52, 53], and meticulous monitoring like continuous near‐infrared spectroscopy [54]. Larger cannula diameters may raise the risk of limb ischemia, according to some data [55]. Although PC has many benefits, it does not completely remove the risk of serious limb‐related problems, as indicated by the similar frequencies of all these complications.

From a theoretical perspective, these results support the growing body of literature that advocates for minimally invasive procedures in critical care settings, aligning with the broader trend toward reducing procedural invasiveness in modern medicine. The findings also suggest that with adequate training and experience, clinicians can achieve similar outcomes with PC as with SC, which traditionally has been considered the more controlled and reliable approach.

5.1. Limitations

While this novel meta‐analysis provides valuable insights into the comparative efficacy and safety of PC vs. SC in ECLS, there are still a few important limitations to be aware of. Variations in healthcare settings, patient demographics, and clinical procedures could introduce operational bias and affect the generalizability of the results. Differences in cannulation techniques and follow‐up durations across studies may influence outcomes, with only one study using propensity‐matched data to control for confounding factors, increasing the risk of selection bias. Additionally, the included studies varied widely in sample sizes and were all retrospective cohorts, lacking randomized controlled trials (RCTs), which are needed for more robust conclusions. Furthermore, a substantial level of heterogeneity was noted for a number of outcomes, which most likely reflected variations in peri‐procedural management techniques, operator expertise, patient selection, and institutional guidelines. Although we used meta‐regression to investigate possible causes of heterogeneity, the observed variability was not entirely taken into consideration by this study, and residual heterogeneity may still restrict the accuracy and applicability of our pooled estimates.

5.2. Future Research Implications

Given these constraints, a focus on conducting RCTs would be preferable for future research as the non‐randomized data is prone to confounding and selection biases. Furthermore, standardized cannulation and procedural techniques across different centers to reduce heterogeneity, longer follow‐up durations to better assess the long‐term outcomes of patients and larger and more homogeneous sample sizes are needed to reach a more conclusive decision. Since these issues have a major impact on survival and quality of life, future research should also look into neurological outcomes in relation to cannulation techniques. Furthermore, as there is currently little data on children, investigating these findings in pediatric ECLS populations would be extremely beneficial. Incorporating pediatric research could inform age‐specific guidelines and assist ascertain whether the trends in procedures and outcomes seen in adults apply to younger patients.

6. Conclusion

The overall findings suggest that PC should be strongly considered as the preferred cannulation method in ECLS, particularly in settings where rapid initiation of support and minimization of procedural complications are critical. The advantages of PC, including reduced infection and vascular complication risks and shorter ECLS duration, make it an attractive option in both emergency and elective scenarios. PC, however, is not always better. According to ELSO guidelines, the decision between percutaneous and surgical methods should be made based on established institutional norms that take into consideration the patient's anatomy, the need for immediate assistance, the operator's experience, and the resources at hand. SC remains essential in cases of complex vascular anatomy, central access requirements, or failed percutaneous attempts. Furthermore, the non‐significant rates of limb ischemia and related complications between percutaneous and surgical approaches highlight the importance of ongoing vigilance and the need for further research to refine cannulation techniques and improve patient outcomes.

Author Contributions

Syeda Zuha Sami: conceptualization. Syeda Zuha Sami, Ariba Nazir, and Muzna Murtaza: data curation, formal analysis. Zainab Muhammad Hanif: methodology, data curation. Laiba Khurram: software. Syeda Zuha Sami, Ariba Nazir, Muzna Murtaza, Zainab Muhammad Hanif, Laiba Khurram, Aghna Iman, Muhammad Ahmed, Inibehe Ime Okon: writing – original draft, writing – review and editing. Syeda Zuha Sami, Ariba Nazir: visualization, resources. Inibehe Ime Okon: supervision, critical reviewing, funding acquisition.

Funding

The authors received no specific funding for this work.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

The lead author Inibehe Ime Okon affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Supporting information

Figure S1: Scatter plot of Mean Age.

Figure S2: Scatter plot of Male (%).

Figure S3: Scatter plot of PAD.

Supplementary Table 1: Search strategy table.

Supplementary Table 2: Risk of bias assessment using Newcastle‐Otawa scale.

Supplementary Table 3: Meta‐regression Results.

Supplementary Table 4: Begg's and Egger test p values.

Data Availability Statement

The authors have nothing to report.