Guiding complexity in V-A ECMO weaning decisions for cardiogenic shock: Insights from a conjoint analysis

Abstract

Weaning and liberation from V-A ECMO for cardiogenic shock involves complex decision-making that requires balancing multiple clinical parameters. In the absence of specific weaning guidelines, ECMO centers rely on heterogeneous, institution-specific practices, which may adversely affect outcome. To gain insights into how professionals weigh multiple factors in complex weaning decisions, a survey was designed using conjoint analysis. The survey was presented to a large cohort of international ECMO clinicians via the EuroELSO network online between February 16, 2024 and May 1, 2024. Using conjoint analysis, a set of 14 hypothetical adult weaning scenarios, incorporating combinations of different attributes with varying levels across 4 cardiogenic shock etiologies was presented. Participants had to decide on liberation from ECMO (‘yes’ or ‘no’) based on a personal trade-off weighing all presented attributes for each scenario. In total, 316 participants from 49 different countries evaluated 4424 hypothetical scenarios. Pulse Pressure ≥30 mmHg (OR 2.56 (CI 2.03–3.24), cardiac index >2.1 L/min/m² (OR 2.5 (CI 1.87–3.43) and echocardiographically assessed left ventricular outflow velocity time integral (VTI) >15 cm (OR 2.53 (95% CI 201-3.2) all favored positive weaning decisions (all p < 0.001). Vasoactive-inotropic score (VIS; 11.1%), VTI (9.7%) and the presence of peripheral and/or pulmonary edema (9.8%) had the highest relative importance to overall weaning decisions. Echocardiography was found to contribute more to weaning decisions in ECMO novices. Weaning decisions were predominantly driven by parameters reflecting cardiac recovery. Leveraging these insights may set the stage for a clinical decision-support tool to benchmark individual practice against expert consensus.

Keywords

venoarterial extracorporeal membrane oxygenation (VA ECMO)cardiogenic shock clinical decision making weaning conjoint analysis

Introduction

Veno-arterial extracorporeal membrane oxygenation (V-A ECMO) is increasingly applied in severe cardiogenic shock, enabling rapid stabilization and serving as bridge to recovery, long-term support, or heart transplantation. Although highly effective in restoring systemic perfusion, the cumulative risk of ECMO-related complications can adversely impact on subsequent therapies and overall outcome.^1,2 From initiation, V-A ECMO management should prioritize individual needs towards minimally required mechanical circulatory support and timely weaning.

In current literature, varying weaning strategies propose a wealth of clinical, hemodynamic, echocardiographic and biochemical predictors for successful discontinuation of V-A ECMO.^3–7 However, mortality rates of up to 70% after initially successful weaning, often referred to as the ECMO gap, underscore the difficulty of reliably predicting long-term outcome.^8,9 Although prediction models such as SAVE, PREDICT VA ECMO, and ENCOURAGE^10–12 assess mortality risk at ECMO initiation, their static nature cannot inform cardiac recovery and weaning readiness during ECMO support.¹³ In clinical practice, weaning strategies should adhere to a structured approach, including gradual ECMO tapering and liberation.^8,14 However, the lack of robust scientific evidence and detailed international guidelines has led to considerable variability in weaning practices.¹⁵ Given the complexity of weaning decisions, shaped by clinical expertise and contextual factors, understanding how clinicians weigh multiple variables is essential for the future development of more robust and standardized weaning protocols.

To address these aspects, an international digital survey using a rating-based conjoint analysis was conducted to capture expert perspectives on weaning decisions by assessing preferences across scenarios containing multiple clinically relevant variables.

Methods

Conjoint analysis study design

This study used a clinical vignette–based rating conjoint analysis to reflect real-world weaning decision-making by incorporating dynamic clinical V-A ECMO parameters. This design was chosen to disentangle the complex trade-offs among multiple clinical and echocardiographic parameters regarding the likelihood of successful weaning.

Identification of attributes for the conjoint analysis

To represent a set of practical V-A ECMO weaning scenarios based on clinical patient profiles, the most relevant attributes influencing weaning decisions were identified. This was done by a combination of focus group interviews with ECMO specialists and a review of currently available literature^14,15. A total of 14 non-overlapping ‘key attributes’ were identified as most relevant to weaning decisions.¹⁶ A descriptive system was designed consisting of 14 attributes with 2 to 4 levels each, detailing the range of possible outcomes for that attribute. This resulted in 4*2²*3¹¹ = 2,834,352 possible scenarios. Before generating the experimental design, several constraints were specified to preclude impossible or unrealistic combinations of attribute levels, e.g., a cardiac index of <1.5 L/min/m² with an echocardiographic assessed left ventricular outflow velocity time integral (VTI) of >15 cm. Because it was impossible to elicit responses for all possible scenarios, a fractional factorial design was created to test for main effect. This consisted of 28 scenarios, of which 14 were drawn randomly for every respondent. Respondents were asked to indicate their decision after evaluating hypothetical weaning scenarios composed of multiple clinical and echocardiographic parameters. The scenarios followed an experimental design in which levels of all attributes systematically varied over the scenarios.¹⁷ The purpose was to explore how different attribute-levels influence professionals in the final decision regarding potential liberation from V-A ECMO.

To collect the data, a survey was constructed. After consent for participation and privacy guarantees, the survey started with general questions (supplement 1) to further specify the participant's clinical expertise and experience. Subsequently, participants were led through 14 randomly assigned, hypothetical weaning scenarios in which a combination of attributes with variable levels were presented with 4 different etiologies of cardiogenic shock. These attributes included clinical characteristics before initiation of a weaning trial and a set of hemodynamic and echocardiographic characteristics during the weaning trial while decreasing ECMO flow to 1 L/min (table 1). Etiologies of cardiogenic shock were: “anterior myocardial infarction at day 5; acute-on-chronic heart failure at day 7; postcardiotomy at day 5; myocarditis at day 10”. Participants were asked to decide on liberation from ECMO (‘yes’ or ‘no’) based on a personal trade-off weighing all presented attributes for each individual scenario. Additionally, they were asked how confident they felt with this decision on a scale from 1 (very certain) to 4 (very uncertain.

Table 1.

Clinical attributes and hemodynamic and echocardiographic attributes evaluated during a weaning trial (decreasing ECMO flow below 1 L/min).

Attribute	Category 1	Category 2	Category 3
Etiology of heart failure
Arterial pulse pressure (mmHg)	>60	30-60	<30
VIS	VIS <10 ^{noradrenaline 0.1 mcg/kg/min}	VIS = 15 ^{dobutamine 5 mcg/kg/min and noradrenaline = 0.10 mcg/kg/min}	VIS = 40^{dobutamine 5 mcg/kg/min and noradrenaline = 0.35 mcg/kg/min}
Resolving end organ function with lactate <2 mmol/l, restoration of hepatic function (REOF) (Antithrombin III activity <60%, prothrombin activity >50% of normal)	Yes	No
Presence of edema	Absent = Absence of peripheral and pulmonary edema	Present 1 = Presence of peripheral edema and absence of pulmonary edema	Present 2 = Presence of pulmonary edema with or without peripheral edema
Cardiac index (L/min/m^2a	>2	1.5–2	<1.5
PaWP (mmHg)	<12	12–18	>18
MAP^a (mmHg)	Stable	5–10 mmHg ↓ baseline	>10 mmHg ↓ baseline
CVP^a (mmHg)	5–10	10–15	>15
Heart rate (bpm)	Stable	<10% ↑ baseline	>10% ↑ baseline
VTI LVOT on echocardiogram (cm)	>15	10–15	<10
LVEF on echocardiogram (eyeballing, %)	Mildly impaired (>30%)	Moderately impaired (20-30%)	Severely impaired (<20%)
New LV and/or RV dilatation on echocardiogram	Absent	Present
RVEF on echocardiogram (eyeballing, %)	Mildly impaired	Moderately impaired	Severely impaired

VIS = vasoactive-inotropic score; PaWP = pulmonary arterial wedge pressure; MAP = mean arterial pressure; CVP = central venous pressure; VTI LVOT = velocity time integral left ventricular outflow tract; LVEF = left ventricular ejection fraction; LV = left ventricle; RV = right ventricle; RVEF = right ventricular ejection fraction.

^awithout increase in inotropics and vasopressors.

Study population and recruitment

The survey was approved by the Steering and Scientific Committees of the European branch of Extracorporeal Life Support Organization (EuroELSO) and distributed prospectively via the EuroELSO network, including 552 adult cardiac ECMO centers worldwide. Subscribers to the EuroELSO newsletter were sent a personal link to complete the survey via email. Additionally, the survey link was distributed via the EuroELSO social media channels. Follow-up reminder e-mails were sent biweekly. Ethical approval for the project was considered not mandatory as the study was a social science survey which did not contain personal data (completely anonymous survey) and no financial incentive was received. All responses were reviewed to exclude duplicates. The survey was built using the survey software Qualtrics (Seattle, Washington, US) and was online available from February 16, 2024 until May 1, 2024 (supplement A). All participants were subdivided in an ‘expert group’ (>5 years of experience with V-A ECMO, working in a tertiary center specialized in advanced heart failure care, exposure of >25 cases per center annually) and a less experienced and specialized ‘novice group’.

Data analysis

All data were processed anonymously using R (R Foundation for Statistical Computer, Vienna, Austria^©). A mixed-effects logistic regression model was applied to estimate the association between attribute levels (independent variables) and the weaning advice (binary outcome: “yes” vs “no”) provided by ECMO experts (dependent variable). The model included a random intercept for each individual respondent to account for clustering within experts. All categorical variables were converted to factors, with clinically relevant reference levels defined for each. These reference levels represented the most favourable clinical conditions. For each attribute level, we calculated odds ratios (ORs) with 95% confidence intervals (CIs), indicating the likelihood of a positive weaning recommendation compared to the reference category. Odds ratios >1 were interpreted as ‘in favour of weaning’, while ORs <1 indicated ‘not in favour of weaning’.

The resulting ORs and confidence intervals were visualized in a forest plot. Additionally, an analysis of variance (ANOVA) or Chi-square test was conducted to complement the regression analysis by formally assessing the significance of each attribute. The relative importance of attributes was determined using the method proposed by González et al.¹⁸ All data were reported as mean ± standard deviation (SD) where applicable.

Decision support tool

Based on the conjoint analysis and as a proof of concept, we developed a web-based V-A ECMO Weaning Decision Support Tool implemented in R Shiny. The tool provides preference-based guidance derived from the clinicians’ responses. For a user-specified combination of the 14 attributes, the application maps each chosen attribute level to the corresponding regression term and applies the fitted logistic regression coefficients (intercept + attribute-level estimates) to compute the modelled probability of a weaning recommendation. The output is reported as a probability (%) and a binary recommendation (“wean” vs “do not wean”) using a 0.5 threshold.

Results

In total, 17.875 surveys were distributed through the EuroELSO network. Responses were received from 664 ECMO experts across 49 countries, with 316 participants completing the survey (completion rate of 47.6%). The database comprised clinicians with various experience levels and background including consultants in cardiology and heart failure, cardiothoracic surgery, cardiac anesthesiology, intensivists and perfusionists. Analyses included only fully complete responses. Among participants, 79% were male and 63% were working in a tertiary center specialized in advanced heart failure care. Of all participants, 52% had a mean professional experience in clinical care for V-A ECMO supported patients of >10 years.

Participant characteristics are presented in table 2.

Table 2.

Baseline characteristics of participants and characteristics involving care for VA ECMO patients.

Baseline characteristics participants (N = 316)
Gender	n	%
Male	249	78.8
Female	64	20.3
U/S	3	9
Age
<30	1	0.3
30–40	71	22.5
40–50	137	43.4
50–60	84	26.7
>60	23	7.3
Years of experience
<1 year	3	0.9
1–5 years	46	14.6
5–10 years	102	32.3
>10 years	165	52.2
Type of hospital
Regional hospital	22	7
Teaching hospital without advanced heart failure facilities	96	30.4
Teaching hospital with advanced heart failure facilities	198	62.7
Annual VA ECMO cases per year
0–10	46	14.6
10–25	96	30.4
25–50	86	27.2
>50	88	27.8
Grading of experience level
Experts	148	46.8
Novices	168	53.2

Characteristics involving care VA ECMO patients (N = 316)
Central VA ECMO (% of all cases) n %		n	%
Never		77	24.4
<25%		165	52.2
25–50%		57	18
>50%		17	5.4
Site of ECMO insertion
Emergency room	Never	98	31
	occasionally	157	49.7
	Regularly	35	11.1
	Very often	21	6.6
	Always	5	1.6
Operating theatre	Never	11	3.5
	occasionally	96	30.4
	Regularly	88	27.8
	Very often	105	33.2
	Always	16	5.1
Cath lab	Never	20	6.3
	occasionally	144	45.6
	Regularly	76	24.1
	Very often	66	20.9
	Always	10	3.2
Intensive care unit	Never	7	2.2
	occasionally	110	34.8
	Regularly	98	31
	Very often	81	25.6
	Always	20	6.3
Involvement specialties in the care of VA ECMO supported patients in your institution
Cardiac surgeon	Not involved	16	5.1
	only occasionally involved	44	13.9
	Regularly involved	53	16.8
	often involved	61	19.3
	Always involved	142	44.9
Intensivist	Not involved	3	0.9
	only occasionally involved	15	4.7
	Regularly involved	28	8.9
	often involved	31	9.8
	Always involved	239	75.6
Heart failure cardiologist	Not involved	30	9.5
	only occasionally involved	84	26.6
	Regularly involved	73	23.1
	often involved	79	25
	Always involved	50	15.8
Cardiac anesthesiologist	Not involved	40	12.7
	only occasionally involved	75	23.7
	Regularly involved	49	15.5
	often involved	62	19.6
	Always involved	90	28.5
Interventional cardiologist	Not involved	35	11.1
	only occasionally involved	113	35.8
	Regularly involved	74	23.4
	often involved	70	22.2
	Always involved	24	7.6
How often is an echocardiogram performed during VA ECMO support?
	only pre-initiation and during weaning	25	7.9
	Daily	174	55.1
	Every 2–3 days	108	34.2
	only in case of deterioration	9	2.8
Who performs the echocardiogram?
	Exclusively cardiologist	60	19
	Exclusively intensivist	55	17.4
	Depending mainly on availability	111	35.1
	Depending mainly on complexity	90	28.5
Do you use a weaning protocol in your center?
	Yes	215	68
	No	101	32

Figure 1 presents the main results of the logistic regression analysis in the overall respondent group presented in an odds plot (Odds ratio (OR) and confidence intervals (CI). Of the 14 attributes, only pulmonary artery wedge pressure did not significantly influence weaning preferences. In terms of etiology, respondents were significantly more likely to favor weaning in ‘myocarditis’ and significantly less likely to favor weaning ‘post-cardiotomy’ compared to ‘anterior wall infarction’. The likelihood of weaning was similar in ‘acute heart failure’ and ‘anterior wall infarction’. A VTI >15 cm and cardiac index >2.1 L/min/m² were attributes that increased the likelihood of a positive decision to wean a patient from V-A ECMO (all p < 0.001; Figure 1). Unresolved organ failure, a pulse pressure (PP) < 30 mmHg or > 60 mmHg, the presence of pulmonary edema with or without peripheral edema, severely impaired left ventricular (LV) and right ventricular (RV) function, a CVD >15 mmHg and high (>15) vasoactive-inotropic score (VIS) all increased the likelihood of a negative decision to wean a patient from V-A ECMO (all p < 0.05; Figure 1). Also, both an increase or decrease of more than 10% in heart rate (HR) or the presence of LV/RV dilation during the weaning trial were found more often to be associated with a negative decision for ECMO weaning (all p < 0.001). The contribution of each attribute in the decision to liberate a patient from VA ECMO is shown in Figure 2. VIS (11.1%), resolution of organ failure (REOF: 10.1%), VTI (9.7%) and the presence of peripheral and/or pulmonary edema (9.8%) were found to have the highest relative importance in weaning decisions.

Figure 1.

Logistic regression analysis overall respondent group presented in an odds plot. Represents the likelihood (Odds ratio (OR) and confidence intervals (CI) of a positive or a negative decision to wean a patient from VA ECMO based on parameters during a weaning trial ECMO.

Figure 2.

Factors determining the decision to weaning from ECMO. Represents the contribution of each attribute in the decision to liberate a patient from VA ECMO.

A Chi-squared test (χ² = 0, df = 27, p = 1) confirmed that clinical cases were evenly distributed between expert and novice groups, indicating no bias in case assignment. All above-described results from the logistic regression analysis were equally observed in both groups, however echocardiographic parameters (LV and RV ejection fraction) were found to contribute for a larger part in weaning decisions in the novice group (13.8%/8.5% vs 2.4%/6.1% in the expert group Figure 2. The presence of peripheral and/or pulmonary edema (9.8%) was found to have the highest relative importance in weaning decisions in the expert group. The proportion of “yes” responses to weaning was similar between experts (34.5%) and novices (35.9%). Likewise, the mean confidence scores (on a scale from 1 (very certain) to 4 (very uncertain)) were identical: 1.79 for experts and 1.79 for novices. No statistically significant differences were found between the groups in either weaning decisions (p = 0.99) or confidence levels (p = 0.99).

The fitted conjoint model was translated into an interactive decision support application. The tool is publicly accessible as a proof-of-concept demonstration at https://medicaldigitaltwin.shinyapps.io/ecmoweanadvise. (Figure 3).

Figure 3.

Proof-of-concept clinical decision tool.

Discussion

Decisions surrounding V-A ECMO weaning represent one of the most complex challenges in critical care, requiring careful assessment of myocardial recovery in the context of dynamic physiological conditions. By engaging a broad cohort of international ECMO experts, a conjoint analysis comprising 14 hypothetical cases revealed that predominantly parameters representing cardiac recovery shaped these decisions.

Indicators of cardiac recovery are key determinants in guiding weaning decisions

Quantitative measures of improving cardiac performance, notably PP, CI and VTI, all emerged as indicators of weaning readiness favoured by clinicians. It underscores that clinicians consistently identify “clinical evidence of cardiac recovery” as the main prerequisite for weaning and liberation from V-A ECMO, which is consistent with previous findings.^3–5,14,19 This result aligns with established physiological principles given that native cardiac function must guarantee adequate systemic perfusion once mechanical support is withdrawn. PP values >30 mmHg were highly appreciated within the context of positive weaning decisions, reflecting the high clinical value of this readily available bedside parameter. Derived from continuous arterial pressure monitoring, PP captures the dynamic interaction between the recovering native heart and the ECMO circuit through ventriculo-arterial and ECMO-arterial coupling. The resulting pulsatility likely reflects the combined influence of contractility of the recovering heart, aortic compliance, peripheral resistance, and ECMO blood flow.²⁰ At the bedside, arterial PP may serve in this sense as a practical surrogate of the patient's own cardiac function which has previously been found to correlate with positive weaning outcome in V-A ECMO supported cardiac arrest patients.^19,21 Although guidelines recommend a minimal PP of 10 mmHg to ensure adequate LV ejection and transpulmonary flow, no specific threshold reliably defines cardiac recovery.²² It is plausible however that an increase in PP over time during ECMO support is indicative of a greater stroke volume and improving myocardial contractility. For this reason, respondents may appreciate this parameter as a meaningful surrogate marker of recovering cardiac pump function.²³ However, ECMO flow, intravascular volume status, and the use of vasoactive agents all influence the pressure waveform. Therefore, its precise clinical significance within this context requires further investigation.²⁰

A stable or rising cardiac index (>2.1 L·min/m²) during flow reduction was also strongly associated with perceived weaning readiness and highly valued by clinicians. While continuously measured via pulmonary artery catheterization, cardiac index shows good agreement with echocardiographic estimates and provides an objective, real-time reflection of the native heart’s contribution to systemic output, with increases indicating improving LV recovery during V-A ECMO support.^24,25 Although no definitive thresholds exist and its interpretation must account for dynamic loading conditions and vasoactive therapy, its use is broadly recommended during support.^22,26,27 For these reasons, cardiac index emerged as a highly valued continuous and objective measure of cardiac performance complementing intermittent echocardiographic assessment, while VTI, as a Doppler-derived surrogate for stroke volume, was also highly valued in weaning decisions by our respondents. Clinicians likely valued VTI because it can be obtained rapidly at the bedside, even with limited echocardiographic expertise, and provides direct insight into recovery of intrinsic cardiac function. VTI values >10 cm at minimal ECMO flow have been associated with successful liberation from support, reinforcing its clinical relevance.^3,8,22 Although interpretation may be challenging in the presence of variable ECMO flows or ventricular unloading devices, VTI is recommended in ECMO monitoring and weaning guidelines and is now widely adopted in clinical practice.^22,28 By complementing continuously available parameters such as PP and cardiac index, VTI meaningfully enhances the assessment of myocardial recovery. Beyond these quantitative indicators of cardiac recovery, many clinicians also highly valued qualitative echocardiographic “eyeballing” of LV and RV function when deciding on VA ECMO weaning. Originating from early postcardiotomy practice, this subjective approach still remains widely used in clinical decision-making despite its lack of formal objectivity.^4,29

Altogether, these findings underscore the need of the responding clinicians to integrate both quantitative, modern parameters like VTI and semi-quantitative, traditional parameters like “eyeballing” LV ejection fraction, cardiac index and PP. Obviously, this combined assessment provides a multidimensional view on cardiac recovery, forming a more robust basis for estimating weaning readiness at the bedside. This insight led to further exploration of whether the level of clinical experience influences preferences for specific weaning parameters.

Variation in weaning preferences between novices and experts

While novices and experts in general prioritized similar weaning attributes, novices valued echocardiography more and experts mainly emphasized clinical parameters. Although confidence levels did not differ significantly between the groups, experts appear to assess weaning readiness within a broader clinical context, whereas novices appear to adhere more to objective imaging parameters. These findings could suggest that, with increasing experience, clinicians appear to shift from analytical reasoning toward a more intuitive, holistic approach, with novices focusing on physiological details and experts on the broader clinical context.^30–32 These findings suggest that variability in weaning practices may reflect not only institutional routines and logistical factors, but also differences in individual experience, preferences, and technical expertise. They also indicate that a multidisciplinary team with mixed experience levels may be beneficial, as novices and experts can offer complementary perspectives in complex weaning decisions.³³

Practical implications and the role of decision support tools in daily weaning practice

The in-depth insights into weaning decisions gathered in this study served as a proof-of-concept development of a decision support tool that could potentially provide clinicians with patient-specific, expert-informed recommendations for ECMO withdrawal. In clinical practice, where weaning decisions are shaped by experience, logistics, ethics, and uncertainty, decision-support tools may enhance confidence when guidelines are lacking. Over the past few years, several ICU decision-support systems, including ventilator-integrated weaning tools using real-time physiological data, have been developed.³⁴ Although still preliminary, they show potential to reduce practice variation, ventilation duration, and ICU length of stay.^35–37 Whereas the application of comparable systems to ECMO weaning is not yet established, clinical decision-support tools have the potential to promote greater uniformity in the weaning process. Ultimately, the extensive availability of continuous hemodynamic data during ECMO support may foster the future development of multifaceted models which are capable of predicting weaning outcomes and identifying the optimal timing for liberation from support³⁸

Limitations and future perspectives

These study results should be interpreted with some caution. Weaning decisions were derived from hypothetical scenarios that may not fully reflect the complexity, time pressure, and contextual factors of real-world clinical decision-making. The analysis reflects stated clinician preferences rather than patient outcomes, limiting causal inference and assuming independence between physiologically interrelated attributes. Further, a challenge not addressed in this study is the increase in use of concomitant unloading LV devices, e.g., Impella, intra-aortic balloon pump, or surgical vent. By altering native cardiac loading conditions, these devices may obscure true myocardial recovery, confound interpretation of weaning parameters such as PP and VTI, and complicate decisions regarding the sequence of device removal. Although the added value of unloading devices during V-A ECMO support in general remains under debate, future studies and protocols should acknowledge and address the complexity of weaning in these scenarios.^39,40

Nevertheless, the study provides valuable insight into expert ECMO weaning trade-offs and may serve as a basis for a future clinical decision-support tool. This will require retrospective and prospective validation before clinicians can benchmark decisions against consolidated expert consensus.

Conclusion

This conjoint analysis provides detailed insights into the parameters clinicians value when making complex V-A ECMO weaning decisions. Indicators of cardiac recovery, particularly PP, cardiac index, and VTI, emerged as main appreciated determinants of weaning readiness and highlight the need for integrating hemodynamic and echocardiographic assessments. Differences between novices and experts suggest that also experience shapes the balance between clinical judgment and protocol-driven evaluation. These findings may foster the development of a clinical decision-support tool aiming at improving consistency, confidence, and uniformity in ECMO weaning. Substantial further validation and refinement are needed before any clinical implementation can be envisaged.

Supplemental material

Supplemental material - Guiding complexity in V-A ECMO weaning decisions for cardiogenic shock: Insights from a conjoint analysis

Supplemental material for Guiding complexity in V-A ECMO weaning decisions for cardiogenic shock: Insights from a conjoint analysis by J. A. J. Hermens, J. A. Van Til, L. Van Loon, N. A. Barrett, J. Swol, R. Lorusso, D. Van Dijk, C. L. Meuwese, D. W. Donker in Perfusion

Footnotes

ORCID iDs

J. A. J. Hermens

J. Swol

D. W. Donker

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: J. A. J. H. Speaker fees from Abbot, paid to the University Medical Center of Utrecht C. L. M. research funding from Charity foundation Stichting Gezondheidszorg Spaarneland and the Dutch Heart Foundation. Speaker fees from Abbot and AOP Health Institutional consultancy in a collaboration with Getinge. D. W. D. is involved in an institutional research cooperation with Maquet Critical Care AB, part of Getinge, Solna, Sweden, and Sonion BV, Hoofddorp, the Netherlands, and research and educational consultancy to HBOX Therapies, Aachen, Germany, and Abiomed, Aachen, Germany. All financial compensation is paid to the University of Twente. D. W. D. does not receive any personal fees.

Supplemental material

Supplemental material for this article is available online.

References

Brahmbhatt

Daly

Luk

, et al. Liberation from venoarterial extracorporeal membrane oxygenation: a review. Circ: Heart Fail 2021; 14: e007679. https://doi.org/10.1161/CIRCHEARTFAILURE.120.007679

Randhawa

Al-Fares

Tong

MZY

, et al. A pragmatic approach to weaning temporary mechanical circulatory support A State-of-the-Art review. Jacc Hear Fail 2021; 9: 664–673. https://doi.org/10.1016/j.jchf.2021.05.011

Aissaoui

Luyt

C-E

Leprince

, et al. Predictors of successful extracorporeal membrane oxygenation (ECMO) weaning after assistance for refractory cardiogenic shock. Intensive Care Med 2011; 37: 1738–1745. https://doi.org/10.1007/s00134-011-2358-2

Cavarocchi

Pitcher

Yang

, et al. Weaning of extracorporeal membrane oxygenation using continuous hemodynamic transesophageal echocardiography. J Thorac Cardiovasc Surg 2013; 146: 1474–1479. https://doi.org/10.1016/j.jtcvs.2013.06.055

Ortuno

Delmas

Diehl

J-L

, et al.

Weaning from veno-arterial extra-corporeal membrane oxygenation: which strategy to use?

Ann Cardiothorac Surg 2019; 8: E1–E8. https://doi.org/10.21037/acs.2018.08.05

Pappalardo

Pieri

Corada

, et al. Timing and strategy for weaning from venoarterial ECMO are complex issues. J Cardiothorac Vasc Anesth 2015; 29: 906–911.

Kim

Jang

Park

, et al. Echocardiographic predictors of successful extracorporeal membrane oxygenation weaning after refractory cardiogenic shock. J Am Soc Echocardiogr 2021; 34: 414–422.e4. https://doi.org/10.1016/j.echo.2020.12.002

Lüsebrink

Stremmel

Stark

, et al. Update on weaning from veno-arterial extracorporeal membrane oxygenation. J Clin Medicine 2020; 9: 992. https://doi.org/10.3390/jcm9040992

Makhoul

Heuts

Mansouri

, et al. Understanding the “extracorporeal membrane oxygenation gap” in veno‐arterial configuration for adult patients: timing and causes of death. Artif Organs, 2021. DOI: 10.1111/aor.14006. Epub ahead of print.

10.

Muller

Flecher

Lebreton

, et al. The ENCOURAGE mortality risk score and analysis of long-term outcomes after VA-ECMO for acute myocardial infarction with cardiogenic shock. Intensive Care Med 2016; 42: 370–378. https://doi.org/10.1007/s00134-016-4223-9

11.

Schmidt

Burrell

Roberts

, et al. Predicting survival after ECMO for refractory cardiogenic shock: the survival after veno-arterial-ECMO (SAVE)-Score. Eur Heart J 2015; 36: 2246–2256. https://doi.org/10.1093/eurheartj/ehv194

12.

Wengenmayer

Duerschmied

Graf

, et al. Development and validation of a prognostic model for survival in patients treated with venoarterial extracorporeal membrane oxygenation: the PREDICT VA-ECMO score. Eur Heart J: Acute Cardiovasc Care 2019; 8: 350–359. https://doi.org/10.1177/2048872618789052

13.

Pladet

LCA

Barten

JMM

Vernooij

, et al. Prognostic models for mortality risk in patients requiring ECMO. Intensive Care Med 2023; 49: 131–141. https://doi.org/10.1007/s00134-022-06947-z

14.

Hermens

Meuwese

Szymanski

, et al. Patient-centered weaning from venoarterial extracorporeal membrane oxygenation: “A practice-oriented narrative review of literature”. Perfusion. 2023; 38(7): 1349–1359. https://doi.org/10.1177/02676591221115938

15.

Hermens

JAJ

Til

Meuwese

, et al. Clinical decision making for VA ECMO weaning in patients with cardiogenic shock A formative qualitative study. Perfusion 2024; 39: 39S–48S. https://doi.org/10.1177/02676591241236643

16.

Mühlbacher

Bekker-Grob

Rivero-Arias

, et al. How to present a decision object in health preference research: attributes and levels, the decision model, and the descriptive framework. Patient 2025; 18: 291–302. https://doi.org/10.1007/s40271-024-00673-y

17.

Hauber

González

Groothuis-Oudshoorn

CGM

, et al. Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR conjoint analysis good research practices task force. Value Health 2016; 19: 300–315. https://doi.org/10.1016/j.jval.2016.04.004

18.

Gonzalez

. A guide to measuring and interpreting attribute importance. Patient - Patient-Centered Outcomes Res 2019; 12: 287–295. https://doi.org/10.1007/s40271-019-00360-3

19.

Park

B-W

Seo

D-C

Moon

I-K

, et al. Pulse pressure as a prognostic marker in patients receiving extracorporeal life support. Resuscitation 2013; 84: 1404–1408. https://doi.org/10.1016/j.resuscitation.2013.04.009

20.

Fresiello

Hermens

JAJ

Pladet

, et al. The physiology of venoarterial extracorporeal membrane oxygenation - a comprehensive clinical perspective. Perfusion 2024; 39: 5S–12S. https://doi.org/10.1177/02676591241237639

21.

Rilinger

Riefler

Bemtgen

, et al. Impact of pulse pressure on clinical outcome in extracorporeal cardiopulmonary resuscitation (eCPR) patients. Clin Res Cardiol 2021; 110: 1473–1483. https://doi.org/10.1007/s00392-021-01838-7

22.

Lorusso

Shekar

MacLaren

, et al. ELSO interim guidelines for venoarterial extracorporeal membrane oxygenation in adult cardiac patients. ASAIO J 2021; 67: 827–844. https://doi.org/10.1097/MAT.0000000000001510

23.

Mourad

Eliet

Zeroual

, et al. Pulse pressure and end-tidal carbon dioxide for monitoring low native cardiac output during veno-arterial ECLS: a prospective observational study. Crit Care 2020; 24: 569. https://doi.org/10.1186/s13054-020-03280-z

24.

Levy

Schmidt

Hekimian

, et al. Clinical validation of pulmonary artery catheter use for continuous thermodilution cardiac output monitoring in VA-ECMO patients. Intensive Care Med 2025; 51: 785–787. https://doi.org/10.1007/s00134-025-07841-0

25.

Norris

King

Grace

, et al. Thermodilution cardiac output-an in vitro model of low flow states. Crit Care Med 1986; 14: 57–59. https://doi.org/10.1097/00003246-198601000-00013

26.

Guglin

Zucker

Bazan

, et al. Venoarterial ECMO for adults JACC scientific expert panel. J Am Coll Cardiol 2019; 73: 698–716. https://doi.org/10.1016/j.jacc.2018.11.038

27.

Rao

Khalpey

Smith

, et al. Venoarterial extracorporeal membrane oxygenation for cardiogenic shock and cardiac arrest: cardinal considerations for initiation and management. Circ: Heart Fail; 11. DOI: 10.1161/circheartfailure.118.004905, Epub ahead of print 2018.

28.

Siriwardena

Dozois

Fan

, et al. Hemodynamic aspects of veno-arterial extracorporeal membrane oxygenation for cardiac support: a worldwide survey. Asaio J 2020; 66: 489–496. https://doi.org/10.1097/MAT.0000000000001024

29.

Goldstein

. Mechanical support for postcardiotomy cardiogenic shock. Semin Thorac Cardiovasc Surg 2000; 12: 220–228. https://doi.org/10.1053/stcs.2000.9666

30.

Pumphrey

Grenville

Colquhoun

, et al. Head, heart or checklist? How self-reported decision-making strategies change according to speciality and grade: a cross-sectional survey of doctors. Postgrad Med J 2019; 95: 148–154, postgradmedj-2018-136184. https://doi.org/10.1136/postgradmedj-2018-136184

31.

Irvine

Meer

RVD

Megiddo

. Expert decision-making in clinicians: an auto-analytic ethnographic study of operational decision-making in urgent care. PLoS One 2025; 20: e0311748. https://doi.org/10.1371/journal.pone.0311748

32.

Cristancho

Apramian

Vanstone

, et al. Thinking like an expert: surgical decision making as a cyclical process of being aware. Am J Surg 2016; 211: 64–69. https://doi.org/10.1016/j.amjsurg.2015.03.010

33.

Matsumoto

Fazzini

Malcolm

, et al. Factors influencing multidisciplinary clinical decision-making in the critical care unit: a systematic review and mixed-methods meta-synthesis. BJA Open 2025; 16: 100488. https://doi.org/10.1016/j.bjao.2025.100488

34.

Platen

Pomprapa

Lachmann

, et al. The dawn of physiological closed-loop Ventilation—a review. Crit Care 2020; 24: 121. https://doi.org/10.1186/s13054-020-2810-1

35.

Rose

Schultz

Cardwell

, et al. Automated versus non-automated weaning for reducing the duration of mechanical ventilation for critically ill adults and children. Cochrane Database Syst Rev 2025; 7: CD009235. https://doi.org/10.1002/14651858.CD009235.pub3

36.

Burns

KEA

Rizvi

Cook

, et al. Ventilator weaning and discontinuation practices for critically ill patients. JAMA 2021; 325: 1173–1184. https://doi.org/10.1001/jama.2021.2384

37.

Jordan

Rose

Dainty

, et al. Factors that impact on the use of mechanical ventilation weaning protocols in critically ill adults and children: a qualitative evidence‐synthesis. Cochrane Database Syst Rev 2015. DOI: 10.1002/14651858.cd009851.pub2. Epub ahead of print.

38.

Pladet

Luijken

Fresiello

, et al.

Clinical Decision Support for ExtraCorporeal Membrane Oxygenation: Will We Fly by Wire?

Perfusion 2023; 38: 68–81. https://doi.org/10.1177/02676591231163688

39.

Kim

Lim

Lee

, et al. Early left ventricular unloading or conventional approach after venoarterial extracorporeal membrane oxygenation: the EARLY-UNLOAD randomized clinical trial. Circulation 2023; 148: 1570–1581. https://doi.org/10.1161/CIRCULATIONAHA.123.066179

40.

Saura

Hékimian

Marmol

, et al. Effect of ECMO flow variations on pulmonary capillary wedge pressure in patients with cardiogenic shock. J Am Coll Cardiol 2025; 86: 768–778. https://doi.org/10.1016/j.jacc.2025.06.048